From bd3bff9e20f454b242d979ec2f9a4dca0d5fa06f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:41 +0200 Subject: sched: add latency tracer callbacks to the scheduler add 3 lightweight callbacks to the tracer backend. zero impact if tracing is turned off. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index cfa222a91539..463dcdb36ef8 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2467,6 +2467,7 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) out_activate: #endif /* CONFIG_SMP */ + ftrace_wake_up_task(p, rq->curr); schedstat_inc(p, se.nr_wakeups); if (sync) schedstat_inc(p, se.nr_wakeups_sync); @@ -2611,6 +2612,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } + ftrace_wake_up_new_task(p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) @@ -2783,6 +2785,7 @@ context_switch(struct rq *rq, struct task_struct *prev, struct mm_struct *mm, *oldmm; prepare_task_switch(rq, prev, next); + ftrace_ctx_switch(prev, next); mm = next->mm; oldmm = prev->active_mm; /* -- cgit v1.2.3 From 7c731e0a495e25e79dc1e9e68772a67a55721a65 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:41 +0200 Subject: ftrace: make the task state char-string visible to all The tracer wants to be able to convert the state number into a user visible character. This patch pulls that conversion string out the scheduler into the header. This way if it were to ever change, other parts of the kernel will know. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 463dcdb36ef8..73e600852365 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -5729,7 +5729,7 @@ out_unlock: return retval; } -static const char stat_nam[] = "RSDTtZX"; +static const char stat_nam[] = TASK_STATE_TO_CHAR_STR; void sched_show_task(struct task_struct *p) { -- cgit v1.2.3 From 16444a8a40d4c7b4f6de34af0cae1f76a4f6c901 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: add basic support for gcc profiler instrumentation If CONFIG_FTRACE is selected and /proc/sys/kernel/ftrace_enabled is set to a non-zero value the ftrace routine will be called everytime we enter a kernel function that is not marked with the "notrace" attribute. The ftrace routine will then call a registered function if a function happens to be registered. [ This code has been highly hacked by Steven Rostedt and Ingo Molnar, so don't blame Arnaldo for all of this ;-) ] Update: It is now possible to register more than one ftrace function. If only one ftrace function is registered, that will be the function that ftrace calls directly. If more than one function is registered, then ftrace will call a function that will loop through the functions to call. Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/Makefile | 1 + kernel/trace/Kconfig | 5 ++ kernel/trace/Makefile | 3 ++ kernel/trace/ftrace.c | 138 ++++++++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 147 insertions(+) create mode 100644 kernel/trace/Kconfig create mode 100644 kernel/trace/Makefile create mode 100644 kernel/trace/ftrace.c (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 1c9938addb9d..fa05f6d8bdbf 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -69,6 +69,7 @@ obj-$(CONFIG_TASK_DELAY_ACCT) += delayacct.o obj-$(CONFIG_TASKSTATS) += taskstats.o tsacct.o obj-$(CONFIG_MARKERS) += marker.o obj-$(CONFIG_LATENCYTOP) += latencytop.o +obj-$(CONFIG_FTRACE) += trace/ ifneq ($(CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER),y) # According to Alan Modra , the -fno-omit-frame-pointer is diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig new file mode 100644 index 000000000000..8185c91417bc --- /dev/null +++ b/kernel/trace/Kconfig @@ -0,0 +1,5 @@ +# +# Architectures that offer an FTRACE implementation should select HAVE_FTRACE: +# +config HAVE_FTRACE + bool diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile new file mode 100644 index 000000000000..bf4fd215a6a9 --- /dev/null +++ b/kernel/trace/Makefile @@ -0,0 +1,3 @@ +obj-$(CONFIG_FTRACE) += libftrace.o + +libftrace-y := ftrace.o diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c new file mode 100644 index 000000000000..b6a80b98a3fb --- /dev/null +++ b/kernel/trace/ftrace.c @@ -0,0 +1,138 @@ +/* + * Infrastructure for profiling code inserted by 'gcc -pg'. + * + * Copyright (C) 2007-2008 Steven Rostedt + * Copyright (C) 2004-2008 Ingo Molnar + * + * Originally ported from the -rt patch by: + * Copyright (C) 2007 Arnaldo Carvalho de Melo + * + * Based on code in the latency_tracer, that is: + * + * Copyright (C) 2004-2006 Ingo Molnar + * Copyright (C) 2004 William Lee Irwin III + */ + +#include +#include + +static DEFINE_SPINLOCK(ftrace_func_lock); +static struct ftrace_ops ftrace_list_end __read_mostly = +{ + .func = ftrace_stub, +}; + +static struct ftrace_ops *ftrace_list __read_mostly = &ftrace_list_end; +ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub; + +/* mcount is defined per arch in assembly */ +EXPORT_SYMBOL(mcount); + +notrace void ftrace_list_func(unsigned long ip, unsigned long parent_ip) +{ + struct ftrace_ops *op = ftrace_list; + + /* in case someone actually ports this to alpha! */ + read_barrier_depends(); + + while (op != &ftrace_list_end) { + /* silly alpha */ + read_barrier_depends(); + op->func(ip, parent_ip); + op = op->next; + }; +} + +/** + * register_ftrace_function - register a function for profiling + * @ops - ops structure that holds the function for profiling. + * + * Register a function to be called by all functions in the + * kernel. + * + * Note: @ops->func and all the functions it calls must be labeled + * with "notrace", otherwise it will go into a + * recursive loop. + */ +int register_ftrace_function(struct ftrace_ops *ops) +{ + unsigned long flags; + + spin_lock_irqsave(&ftrace_func_lock, flags); + ops->next = ftrace_list; + /* + * We are entering ops into the ftrace_list but another + * CPU might be walking that list. We need to make sure + * the ops->next pointer is valid before another CPU sees + * the ops pointer included into the ftrace_list. + */ + smp_wmb(); + ftrace_list = ops; + /* + * For one func, simply call it directly. + * For more than one func, call the chain. + */ + if (ops->next == &ftrace_list_end) + ftrace_trace_function = ops->func; + else + ftrace_trace_function = ftrace_list_func; + spin_unlock_irqrestore(&ftrace_func_lock, flags); + + return 0; +} + +/** + * unregister_ftrace_function - unresgister a function for profiling. + * @ops - ops structure that holds the function to unregister + * + * Unregister a function that was added to be called by ftrace profiling. + */ +int unregister_ftrace_function(struct ftrace_ops *ops) +{ + unsigned long flags; + struct ftrace_ops **p; + int ret = 0; + + spin_lock_irqsave(&ftrace_func_lock, flags); + + /* + * If we are the only function, then the ftrace pointer is + * pointing directly to that function. + */ + if (ftrace_list == ops && ops->next == &ftrace_list_end) { + ftrace_trace_function = ftrace_stub; + ftrace_list = &ftrace_list_end; + goto out; + } + + for (p = &ftrace_list; *p != &ftrace_list_end; p = &(*p)->next) + if (*p == ops) + break; + + if (*p != ops) { + ret = -1; + goto out; + } + + *p = (*p)->next; + + /* If we only have one func left, then call that directly */ + if (ftrace_list->next == &ftrace_list_end) + ftrace_trace_function = ftrace_list->func; + + out: + spin_unlock_irqrestore(&ftrace_func_lock, flags); + + return 0; +} + +/** + * clear_ftrace_function - reset the ftrace function + * + * This NULLs the ftrace function and in essence stops + * tracing. There may be lag + */ +void clear_ftrace_function(void) +{ + ftrace_trace_function = ftrace_stub; +} -- cgit v1.2.3 From bc0c38d139ec7fcd5c030aea16b008f3732e42ac Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: latency tracer infrastructure This patch adds the latency tracer infrastructure. This patch does not add anything that will select and turn it on, but will be used by later patches. If it were to be compiled, it would add the following files to the debugfs: The root tracing directory: /debugfs/tracing/ This patch also adds the following files: available_tracers list of available tracers. Currently no tracers are available. Looking into this file only shows "none" which is used to unregister all tracers. current_tracer The trace that is currently active. Empty on start up. To switch to a tracer simply echo one of the tracers that are listed in available_tracers: example: (used with later patches) echo function > /debugfs/tracing/current_tracer To disable the tracer: echo disable > /debugfs/tracing/current_tracer tracing_enabled echoing "1" into this file starts the ftrace function tracing (if sysctl kernel.ftrace_enabled=1) echoing "0" turns it off. latency_trace This file is readonly and holds the result of the trace. trace This file outputs a easier to read version of the trace. iter_ctrl Controls the way the output of traces look. So far there's two controls: echoing in "symonly" will only show the kallsyms variables without the addresses (if kallsyms was configured) echoing in "verbose" will change the output to show a lot more data, but not very easy to understand by humans. echoing in "nosymonly" turns off symonly. echoing in "noverbose" turns off verbose. Signed-off-by: Steven Rostedt Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/Makefile | 1 + kernel/trace/Kconfig | 5 + kernel/trace/Makefile | 2 + kernel/trace/trace.c | 1547 +++++++++++++++++++++++++++++++++++++++++++++++++ kernel/trace/trace.h | 184 ++++++ 5 files changed, 1739 insertions(+) create mode 100644 kernel/trace/trace.c create mode 100644 kernel/trace/trace.h (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index fa05f6d8bdbf..7e344e7b0cb3 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -70,6 +70,7 @@ obj-$(CONFIG_TASKSTATS) += taskstats.o tsacct.o obj-$(CONFIG_MARKERS) += marker.o obj-$(CONFIG_LATENCYTOP) += latencytop.o obj-$(CONFIG_FTRACE) += trace/ +obj-$(CONFIG_TRACING) += trace/ ifneq ($(CONFIG_SCHED_NO_NO_OMIT_FRAME_POINTER),y) # According to Alan Modra , the -fno-omit-frame-pointer is diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 8185c91417bc..ce70677afbf9 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -3,3 +3,8 @@ # config HAVE_FTRACE bool + +config TRACING + bool + select DEBUG_FS + diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index bf4fd215a6a9..7af403175255 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -1,3 +1,5 @@ obj-$(CONFIG_FTRACE) += libftrace.o +obj-$(CONFIG_TRACING) += trace.o + libftrace-y := ftrace.o diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c new file mode 100644 index 000000000000..1b8eca7650d4 --- /dev/null +++ b/kernel/trace/trace.c @@ -0,0 +1,1547 @@ +/* + * ring buffer based function tracer + * + * Copyright (C) 2007-2008 Steven Rostedt + * Copyright (C) 2008 Ingo Molnar + * + * Originally taken from the RT patch by: + * Arnaldo Carvalho de Melo + * + * Based on code from the latency_tracer, that is: + * Copyright (C) 2004-2006 Ingo Molnar + * Copyright (C) 2004 William Lee Irwin III + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "trace.h" + +unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; +unsigned long __read_mostly tracing_thresh; + +static long notrace +ns2usecs(cycle_t nsec) +{ + nsec += 500; + do_div(nsec, 1000); + return nsec; +} + +static atomic_t tracer_counter; +static struct trace_array global_trace; + +static DEFINE_PER_CPU(struct trace_array_cpu, global_trace_cpu); + +static struct trace_array max_tr; + +static DEFINE_PER_CPU(struct trace_array_cpu, max_data); + +static int tracer_enabled; +static unsigned long trace_nr_entries = 4096UL; + +static struct tracer *trace_types __read_mostly; +static struct tracer *current_trace __read_mostly; +static int max_tracer_type_len; + +static DEFINE_MUTEX(trace_types_lock); + +static int __init set_nr_entries(char *str) +{ + if (!str) + return 0; + trace_nr_entries = simple_strtoul(str, &str, 0); + return 1; +} +__setup("trace_entries=", set_nr_entries); + +enum trace_type { + __TRACE_FIRST_TYPE = 0, + + TRACE_FN, + TRACE_CTX, + + __TRACE_LAST_TYPE +}; + +enum trace_flag_type { + TRACE_FLAG_IRQS_OFF = 0x01, + TRACE_FLAG_NEED_RESCHED = 0x02, + TRACE_FLAG_HARDIRQ = 0x04, + TRACE_FLAG_SOFTIRQ = 0x08, +}; + +enum trace_iterator_flags { + TRACE_ITER_PRINT_PARENT = 0x01, + TRACE_ITER_SYM_OFFSET = 0x02, + TRACE_ITER_SYM_ADDR = 0x04, + TRACE_ITER_VERBOSE = 0x08, +}; + +#define TRACE_ITER_SYM_MASK \ + (TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR) + +/* These must match the bit postions above */ +static const char *trace_options[] = { + "print-parent", + "sym-offset", + "sym-addr", + "verbose", + NULL +}; + +static unsigned trace_flags; + + +/* + * Copy the new maximum trace into the separate maximum-trace + * structure. (this way the maximum trace is permanently saved, + * for later retrieval via /debugfs/tracing/latency_trace) + */ +static void notrace +__update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) +{ + struct trace_array_cpu *data = tr->data[cpu]; + + max_tr.cpu = cpu; + max_tr.time_start = data->preempt_timestamp; + + data = max_tr.data[cpu]; + data->saved_latency = tracing_max_latency; + + memcpy(data->comm, tsk->comm, TASK_COMM_LEN); + data->pid = tsk->pid; + data->uid = tsk->uid; + data->nice = tsk->static_prio - 20 - MAX_RT_PRIO; + data->policy = tsk->policy; + data->rt_priority = tsk->rt_priority; + + /* record this tasks comm */ + tracing_record_cmdline(current); +} + +notrace void +update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) +{ + struct trace_array_cpu *data; + void *save_trace; + int i; + + /* clear out all the previous traces */ + for_each_possible_cpu(i) { + data = tr->data[i]; + save_trace = max_tr.data[i]->trace; + memcpy(max_tr.data[i], data, sizeof(*data)); + data->trace = save_trace; + } + + __update_max_tr(tr, tsk, cpu); +} + +/** + * update_max_tr_single - only copy one trace over, and reset the rest + * @tr - tracer + * @tsk - task with the latency + * @cpu - the cpu of the buffer to copy. + */ +notrace void +update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) +{ + struct trace_array_cpu *data = tr->data[cpu]; + void *save_trace; + int i; + + for_each_possible_cpu(i) + tracing_reset(max_tr.data[i]); + + save_trace = max_tr.data[cpu]->trace; + memcpy(max_tr.data[cpu], data, sizeof(*data)); + data->trace = save_trace; + + __update_max_tr(tr, tsk, cpu); +} + +int register_tracer(struct tracer *type) +{ + struct tracer *t; + int len; + int ret = 0; + + if (!type->name) { + pr_info("Tracer must have a name\n"); + return -1; + } + + mutex_lock(&trace_types_lock); + for (t = trace_types; t; t = t->next) { + if (strcmp(type->name, t->name) == 0) { + /* already found */ + pr_info("Trace %s already registered\n", + type->name); + ret = -1; + goto out; + } + } + + type->next = trace_types; + trace_types = type; + len = strlen(type->name); + if (len > max_tracer_type_len) + max_tracer_type_len = len; + out: + mutex_unlock(&trace_types_lock); + + return ret; +} + +void unregister_tracer(struct tracer *type) +{ + struct tracer **t; + int len; + + mutex_lock(&trace_types_lock); + for (t = &trace_types; *t; t = &(*t)->next) { + if (*t == type) + goto found; + } + pr_info("Trace %s not registered\n", type->name); + goto out; + + found: + *t = (*t)->next; + if (strlen(type->name) != max_tracer_type_len) + goto out; + + max_tracer_type_len = 0; + for (t = &trace_types; *t; t = &(*t)->next) { + len = strlen((*t)->name); + if (len > max_tracer_type_len) + max_tracer_type_len = len; + } + out: + mutex_unlock(&trace_types_lock); +} + +void notrace tracing_reset(struct trace_array_cpu *data) +{ + data->trace_idx = 0; + atomic_set(&data->underrun, 0); +} + +#ifdef CONFIG_FTRACE +static void notrace +function_trace_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = &global_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (unlikely(!tracer_enabled)) + return; + + raw_local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + ftrace(tr, data, ip, parent_ip, flags); + + atomic_dec(&data->disabled); + raw_local_irq_restore(flags); +} + +static struct ftrace_ops trace_ops __read_mostly = +{ + .func = function_trace_call, +}; +#endif + +notrace void tracing_start_function_trace(void) +{ + register_ftrace_function(&trace_ops); +} + +notrace void tracing_stop_function_trace(void) +{ + unregister_ftrace_function(&trace_ops); +} + +#define SAVED_CMDLINES 128 +static unsigned map_pid_to_cmdline[PID_MAX_DEFAULT+1]; +static unsigned map_cmdline_to_pid[SAVED_CMDLINES]; +static char saved_cmdlines[SAVED_CMDLINES][TASK_COMM_LEN]; +static int cmdline_idx; +static DEFINE_SPINLOCK(trace_cmdline_lock); +atomic_t trace_record_cmdline_disabled; + +static void trace_init_cmdlines(void) +{ + memset(&map_pid_to_cmdline, -1, sizeof(map_pid_to_cmdline)); + memset(&map_cmdline_to_pid, -1, sizeof(map_cmdline_to_pid)); + cmdline_idx = 0; +} + +notrace void trace_stop_cmdline_recording(void); + +static void notrace trace_save_cmdline(struct task_struct *tsk) +{ + unsigned map; + unsigned idx; + + if (!tsk->pid || unlikely(tsk->pid > PID_MAX_DEFAULT)) + return; + + /* + * It's not the end of the world if we don't get + * the lock, but we also don't want to spin + * nor do we want to disable interrupts, + * so if we miss here, then better luck next time. + */ + if (!spin_trylock(&trace_cmdline_lock)) + return; + + idx = map_pid_to_cmdline[tsk->pid]; + if (idx >= SAVED_CMDLINES) { + idx = (cmdline_idx + 1) % SAVED_CMDLINES; + + map = map_cmdline_to_pid[idx]; + if (map <= PID_MAX_DEFAULT) + map_pid_to_cmdline[map] = (unsigned)-1; + + map_pid_to_cmdline[tsk->pid] = idx; + + cmdline_idx = idx; + } + + memcpy(&saved_cmdlines[idx], tsk->comm, TASK_COMM_LEN); + + spin_unlock(&trace_cmdline_lock); +} + +static notrace char *trace_find_cmdline(int pid) +{ + char *cmdline = "<...>"; + unsigned map; + + if (!pid) + return ""; + + if (pid > PID_MAX_DEFAULT) + goto out; + + map = map_pid_to_cmdline[pid]; + if (map >= SAVED_CMDLINES) + goto out; + + cmdline = saved_cmdlines[map]; + + out: + return cmdline; +} + +notrace void tracing_record_cmdline(struct task_struct *tsk) +{ + if (atomic_read(&trace_record_cmdline_disabled)) + return; + + trace_save_cmdline(tsk); +} + +static inline notrace struct trace_entry * +tracing_get_trace_entry(struct trace_array *tr, + struct trace_array_cpu *data) +{ + unsigned long idx, idx_next; + struct trace_entry *entry; + + idx = data->trace_idx; + idx_next = idx + 1; + + if (unlikely(idx_next >= tr->entries)) { + atomic_inc(&data->underrun); + idx_next = 0; + } + + data->trace_idx = idx_next; + + if (unlikely(idx_next != 0 && atomic_read(&data->underrun))) + atomic_inc(&data->underrun); + + entry = data->trace + idx * TRACE_ENTRY_SIZE; + + return entry; +} + +static inline notrace void +tracing_generic_entry_update(struct trace_entry *entry, + unsigned long flags) +{ + struct task_struct *tsk = current; + unsigned long pc; + + pc = preempt_count(); + + entry->idx = atomic_inc_return(&tracer_counter); + entry->preempt_count = pc & 0xff; + entry->pid = tsk->pid; + entry->t = now(raw_smp_processor_id()); + entry->flags = (irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) | + ((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) | + ((pc & SOFTIRQ_MASK) ? TRACE_FLAG_SOFTIRQ : 0) | + (need_resched() ? TRACE_FLAG_NEED_RESCHED : 0); +} + +notrace void +ftrace(struct trace_array *tr, struct trace_array_cpu *data, + unsigned long ip, unsigned long parent_ip, + unsigned long flags) +{ + struct trace_entry *entry; + + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, flags); + entry->type = TRACE_FN; + entry->fn.ip = ip; + entry->fn.parent_ip = parent_ip; +} + +notrace void +tracing_sched_switch_trace(struct trace_array *tr, + struct trace_array_cpu *data, + struct task_struct *prev, struct task_struct *next, + unsigned long flags) +{ + struct trace_entry *entry; + + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, flags); + entry->type = TRACE_CTX; + entry->ctx.prev_pid = prev->pid; + entry->ctx.prev_prio = prev->prio; + entry->ctx.prev_state = prev->state; + entry->ctx.next_pid = next->pid; + entry->ctx.next_prio = next->prio; +} + +enum trace_file_type { + TRACE_FILE_LAT_FMT = 1, +}; + +static struct trace_entry * +trace_entry_idx(struct trace_array *tr, unsigned long idx, int cpu) +{ + struct trace_entry *array = tr->data[cpu]->trace; + unsigned long underrun; + + if (idx >= tr->entries) + return NULL; + + underrun = atomic_read(&tr->data[cpu]->underrun); + if (underrun) + idx = ((underrun - 1) + idx) % tr->entries; + else if (idx >= tr->data[cpu]->trace_idx) + return NULL; + + return &array[idx]; +} + +static struct notrace trace_entry * +find_next_entry(struct trace_iterator *iter, int *ent_cpu) +{ + struct trace_array *tr = iter->tr; + struct trace_entry *ent, *next = NULL; + int next_cpu = -1; + int cpu; + + for_each_possible_cpu(cpu) { + if (!tr->data[cpu]->trace) + continue; + ent = trace_entry_idx(tr, iter->next_idx[cpu], cpu); + if (ent && + (!next || (long)(next->idx - ent->idx) > 0)) { + next = ent; + next_cpu = cpu; + } + } + + if (ent_cpu) + *ent_cpu = next_cpu; + + return next; +} + +static void *find_next_entry_inc(struct trace_iterator *iter) +{ + struct trace_entry *next; + int next_cpu = -1; + + next = find_next_entry(iter, &next_cpu); + + if (next) { + iter->next_idx[next_cpu]++; + iter->idx++; + } + iter->ent = next; + iter->cpu = next_cpu; + + return next ? iter : NULL; +} + +static void notrace * +s_next(struct seq_file *m, void *v, loff_t *pos) +{ + struct trace_iterator *iter = m->private; + void *ent; + void *last_ent = iter->ent; + int i = (int)*pos; + + (*pos)++; + + /* can't go backwards */ + if (iter->idx > i) + return NULL; + + if (iter->idx < 0) + ent = find_next_entry_inc(iter); + else + ent = iter; + + while (ent && iter->idx < i) + ent = find_next_entry_inc(iter); + + iter->pos = *pos; + + if (last_ent && !ent) + seq_puts(m, "\n\nvim:ft=help\n"); + + return ent; +} + +static void *s_start(struct seq_file *m, loff_t *pos) +{ + struct trace_iterator *iter = m->private; + void *p = NULL; + loff_t l = 0; + int i; + + mutex_lock(&trace_types_lock); + + if (!current_trace || current_trace != iter->trace) + return NULL; + + atomic_inc(&trace_record_cmdline_disabled); + + /* let the tracer grab locks here if needed */ + if (current_trace->start) + current_trace->start(iter); + + if (*pos != iter->pos) { + iter->ent = NULL; + iter->cpu = 0; + iter->idx = -1; + + for (i = 0; i < NR_CPUS; i++) + iter->next_idx[i] = 0; + + for (p = iter; p && l < *pos; p = s_next(m, p, &l)) + ; + + } else { + l = *pos; + p = s_next(m, p, &l); + } + + return p; +} + +static void s_stop(struct seq_file *m, void *p) +{ + struct trace_iterator *iter = m->private; + + atomic_dec(&trace_record_cmdline_disabled); + + /* let the tracer release locks here if needed */ + if (current_trace && current_trace == iter->trace && iter->trace->stop) + iter->trace->stop(iter); + + mutex_unlock(&trace_types_lock); +} + +static void +seq_print_sym_short(struct seq_file *m, const char *fmt, unsigned long address) +{ +#ifdef CONFIG_KALLSYMS + char str[KSYM_SYMBOL_LEN]; + + kallsyms_lookup(address, NULL, NULL, NULL, str); + + seq_printf(m, fmt, str); +#endif +} + +static void +seq_print_sym_offset(struct seq_file *m, const char *fmt, unsigned long address) +{ +#ifdef CONFIG_KALLSYMS + char str[KSYM_SYMBOL_LEN]; + + sprint_symbol(str, address); + seq_printf(m, fmt, str); +#endif +} + +#ifndef CONFIG_64BIT +# define IP_FMT "%08lx" +#else +# define IP_FMT "%016lx" +#endif + +static void notrace +seq_print_ip_sym(struct seq_file *m, unsigned long ip, unsigned long sym_flags) +{ + if (!ip) { + seq_printf(m, "0"); + return; + } + + if (sym_flags & TRACE_ITER_SYM_OFFSET) + seq_print_sym_offset(m, "%s", ip); + else + seq_print_sym_short(m, "%s", ip); + + if (sym_flags & TRACE_ITER_SYM_ADDR) + seq_printf(m, " <" IP_FMT ">", ip); +} + +static void notrace print_lat_help_header(struct seq_file *m) +{ + seq_puts(m, "# _------=> CPU# \n"); + seq_puts(m, "# / _-----=> irqs-off \n"); + seq_puts(m, "# | / _----=> need-resched \n"); + seq_puts(m, "# || / _---=> hardirq/softirq \n"); + seq_puts(m, "# ||| / _--=> preempt-depth \n"); + seq_puts(m, "# |||| / \n"); + seq_puts(m, "# ||||| delay \n"); + seq_puts(m, "# cmd pid ||||| time | caller \n"); + seq_puts(m, "# \\ / ||||| \\ | / \n"); +} + +static void notrace print_func_help_header(struct seq_file *m) +{ + seq_puts(m, "# TASK-PID CPU# TIMESTAMP FUNCTION\n"); + seq_puts(m, "# | | | | |\n"); +} + + +static void notrace +print_trace_header(struct seq_file *m, struct trace_iterator *iter) +{ + unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); + struct trace_array *tr = iter->tr; + struct trace_array_cpu *data = tr->data[tr->cpu]; + struct tracer *type = current_trace; + unsigned long underruns = 0; + unsigned long underrun; + unsigned long entries = 0; + int cpu; + const char *name = "preemption"; + + if (type) + name = type->name; + + for_each_possible_cpu(cpu) { + if (tr->data[cpu]->trace) { + underrun = atomic_read(&tr->data[cpu]->underrun); + if (underrun) { + underruns += underrun; + entries += tr->entries; + } else + entries += tr->data[cpu]->trace_idx; + } + } + + seq_printf(m, "%s latency trace v1.1.5 on %s\n", + name, UTS_RELEASE); + seq_puts(m, "-----------------------------------" + "---------------------------------\n"); + seq_printf(m, " latency: %lu us, #%lu/%lu, CPU#%d |" + " (M:%s VP:%d, KP:%d, SP:%d HP:%d", + data->saved_latency, + entries, + (entries + underruns), + tr->cpu, +#if defined(CONFIG_PREEMPT_NONE) + "server", +#elif defined(CONFIG_PREEMPT_VOLUNTARY) + "desktop", +#elif defined(CONFIG_PREEMPT_DESKTOP) + "preempt", +#else + "unknown", +#endif + /* These are reserved for later use */ + 0, 0, 0, 0); +#ifdef CONFIG_SMP + seq_printf(m, " #P:%d)\n", num_online_cpus()); +#else + seq_puts(m, ")\n"); +#endif + seq_puts(m, " -----------------\n"); + seq_printf(m, " | task: %.16s-%d " + "(uid:%d nice:%ld policy:%ld rt_prio:%ld)\n", + data->comm, data->pid, data->uid, data->nice, + data->policy, data->rt_priority); + seq_puts(m, " -----------------\n"); + + if (data->critical_start) { + seq_puts(m, " => started at: "); + seq_print_ip_sym(m, data->critical_start, sym_flags); + seq_puts(m, "\n => ended at: "); + seq_print_ip_sym(m, data->critical_end, sym_flags); + seq_puts(m, "\n"); + } + + seq_puts(m, "\n"); +} + +unsigned long nsecs_to_usecs(unsigned long nsecs) +{ + return nsecs / 1000; +} + +static void notrace +lat_print_generic(struct seq_file *m, struct trace_entry *entry, int cpu) +{ + int hardirq, softirq; + char *comm; + + comm = trace_find_cmdline(entry->pid); + + seq_printf(m, "%8.8s-%-5d ", comm, entry->pid); + seq_printf(m, "%d", cpu); + seq_printf(m, "%c%c", + (entry->flags & TRACE_FLAG_IRQS_OFF) ? 'd' : '.', + ((entry->flags & TRACE_FLAG_NEED_RESCHED) ? 'N' : '.')); + + hardirq = entry->flags & TRACE_FLAG_HARDIRQ; + softirq = entry->flags & TRACE_FLAG_SOFTIRQ; + if (hardirq && softirq) + seq_putc(m, 'H'); + else { + if (hardirq) + seq_putc(m, 'h'); + else { + if (softirq) + seq_putc(m, 's'); + else + seq_putc(m, '.'); + } + } + + if (entry->preempt_count) + seq_printf(m, "%x", entry->preempt_count); + else + seq_puts(m, "."); +} + +unsigned long preempt_mark_thresh = 100; + +static void notrace +lat_print_timestamp(struct seq_file *m, unsigned long long abs_usecs, + unsigned long rel_usecs) +{ + seq_printf(m, " %4lldus", abs_usecs); + if (rel_usecs > preempt_mark_thresh) + seq_puts(m, "!: "); + else if (rel_usecs > 1) + seq_puts(m, "+: "); + else + seq_puts(m, " : "); +} + +static const char state_to_char[] = TASK_STATE_TO_CHAR_STR; + +static void notrace +print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, + unsigned int trace_idx, int cpu) +{ + unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); + struct trace_entry *next_entry = find_next_entry(iter, NULL); + unsigned long verbose = (trace_flags & TRACE_ITER_VERBOSE); + struct trace_entry *entry = iter->ent; + unsigned long abs_usecs; + unsigned long rel_usecs; + char *comm; + int S; + + if (!next_entry) + next_entry = entry; + rel_usecs = ns2usecs(next_entry->t - entry->t); + abs_usecs = ns2usecs(entry->t - iter->tr->time_start); + + if (verbose) { + comm = trace_find_cmdline(entry->pid); + seq_printf(m, "%16s %5d %d %d %08x %08x [%08lx]" + " %ld.%03ldms (+%ld.%03ldms): ", + comm, + entry->pid, cpu, entry->flags, + entry->preempt_count, trace_idx, + ns2usecs(entry->t), + abs_usecs/1000, + abs_usecs % 1000, rel_usecs/1000, rel_usecs % 1000); + } else { + lat_print_generic(m, entry, cpu); + lat_print_timestamp(m, abs_usecs, rel_usecs); + } + switch (entry->type) { + case TRACE_FN: + seq_print_ip_sym(m, entry->fn.ip, sym_flags); + seq_puts(m, " ("); + seq_print_ip_sym(m, entry->fn.parent_ip, sym_flags); + seq_puts(m, ")\n"); + break; + case TRACE_CTX: + S = entry->ctx.prev_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.prev_state] : 'X'; + comm = trace_find_cmdline(entry->ctx.next_pid); + seq_printf(m, " %d:%d:%c --> %d:%d %s\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio, + comm); + break; + } +} + +static void notrace +print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) +{ + unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); + struct trace_entry *entry = iter->ent; + unsigned long usec_rem; + unsigned long long t; + unsigned long secs; + char *comm; + int S; + + comm = trace_find_cmdline(iter->ent->pid); + + t = ns2usecs(entry->t); + usec_rem = do_div(t, 1000000ULL); + secs = (unsigned long)t; + + seq_printf(m, "%16s-%-5d ", comm, entry->pid); + seq_printf(m, "[%02d] ", iter->cpu); + seq_printf(m, "%5lu.%06lu: ", secs, usec_rem); + + switch (entry->type) { + case TRACE_FN: + seq_print_ip_sym(m, entry->fn.ip, sym_flags); + if ((sym_flags & TRACE_ITER_PRINT_PARENT) && + entry->fn.parent_ip) { + seq_printf(m, " <-"); + seq_print_ip_sym(m, entry->fn.parent_ip, sym_flags); + } + break; + case TRACE_CTX: + S = entry->ctx.prev_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.prev_state] : 'X'; + seq_printf(m, " %d:%d:%c ==> %d:%d\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio); + break; + } + seq_printf(m, "\n"); +} + +static int trace_empty(struct trace_iterator *iter) +{ + struct trace_array_cpu *data; + int cpu; + + for_each_possible_cpu(cpu) { + data = iter->tr->data[cpu]; + + if (data->trace && + (data->trace_idx || + atomic_read(&data->underrun))) + return 0; + } + return 1; +} + +static int s_show(struct seq_file *m, void *v) +{ + struct trace_iterator *iter = v; + + if (iter->ent == NULL) { + if (iter->tr) { + seq_printf(m, "# tracer: %s\n", iter->trace->name); + seq_puts(m, "#\n"); + } + if (iter->iter_flags & TRACE_FILE_LAT_FMT) { + /* print nothing if the buffers are empty */ + if (trace_empty(iter)) + return 0; + print_trace_header(m, iter); + if (!(trace_flags & TRACE_ITER_VERBOSE)) + print_lat_help_header(m); + } else { + if (!(trace_flags & TRACE_ITER_VERBOSE)) + print_func_help_header(m); + } + } else { + if (iter->iter_flags & TRACE_FILE_LAT_FMT) + print_lat_fmt(m, iter, iter->idx, iter->cpu); + else + print_trace_fmt(m, iter); + } + + return 0; +} + +static struct seq_operations tracer_seq_ops = { + .start = s_start, + .next = s_next, + .stop = s_stop, + .show = s_show, +}; + +static struct trace_iterator notrace * +__tracing_open(struct inode *inode, struct file *file, int *ret) +{ + struct trace_iterator *iter; + + iter = kzalloc(sizeof(*iter), GFP_KERNEL); + if (!iter) { + *ret = -ENOMEM; + goto out; + } + + mutex_lock(&trace_types_lock); + if (current_trace && current_trace->print_max) + iter->tr = &max_tr; + else + iter->tr = inode->i_private; + iter->trace = current_trace; + iter->pos = -1; + + /* TODO stop tracer */ + *ret = seq_open(file, &tracer_seq_ops); + if (!*ret) { + struct seq_file *m = file->private_data; + m->private = iter; + + /* stop the trace while dumping */ + if (iter->tr->ctrl) + tracer_enabled = 0; + + if (iter->trace && iter->trace->open) + iter->trace->open(iter); + } else { + kfree(iter); + iter = NULL; + } + mutex_unlock(&trace_types_lock); + + out: + return iter; +} + +int tracing_open_generic(struct inode *inode, struct file *filp) +{ + filp->private_data = inode->i_private; + return 0; +} + +int tracing_release(struct inode *inode, struct file *file) +{ + struct seq_file *m = (struct seq_file *)file->private_data; + struct trace_iterator *iter = m->private; + + mutex_lock(&trace_types_lock); + if (iter->trace && iter->trace->close) + iter->trace->close(iter); + + /* reenable tracing if it was previously enabled */ + if (iter->tr->ctrl) + tracer_enabled = 1; + mutex_unlock(&trace_types_lock); + + seq_release(inode, file); + kfree(iter); + return 0; +} + +static int tracing_open(struct inode *inode, struct file *file) +{ + int ret; + + __tracing_open(inode, file, &ret); + + return ret; +} + +static int tracing_lt_open(struct inode *inode, struct file *file) +{ + struct trace_iterator *iter; + int ret; + + iter = __tracing_open(inode, file, &ret); + + if (!ret) + iter->iter_flags |= TRACE_FILE_LAT_FMT; + + return ret; +} + + +static void notrace * +t_next(struct seq_file *m, void *v, loff_t *pos) +{ + struct tracer *t = m->private; + + (*pos)++; + + if (t) + t = t->next; + + m->private = t; + + return t; +} + +static void *t_start(struct seq_file *m, loff_t *pos) +{ + struct tracer *t = m->private; + loff_t l = 0; + + mutex_lock(&trace_types_lock); + for (; t && l < *pos; t = t_next(m, t, &l)) + ; + + return t; +} + +static void t_stop(struct seq_file *m, void *p) +{ + mutex_unlock(&trace_types_lock); +} + +static int t_show(struct seq_file *m, void *v) +{ + struct tracer *t = v; + + if (!t) + return 0; + + seq_printf(m, "%s", t->name); + if (t->next) + seq_putc(m, ' '); + else + seq_putc(m, '\n'); + + return 0; +} + +static struct seq_operations show_traces_seq_ops = { + .start = t_start, + .next = t_next, + .stop = t_stop, + .show = t_show, +}; + +static int show_traces_open(struct inode *inode, struct file *file) +{ + int ret; + + ret = seq_open(file, &show_traces_seq_ops); + if (!ret) { + struct seq_file *m = file->private_data; + m->private = trace_types; + } + + return ret; +} + +static struct file_operations tracing_fops = { + .open = tracing_open, + .read = seq_read, + .llseek = seq_lseek, + .release = tracing_release, +}; + +static struct file_operations tracing_lt_fops = { + .open = tracing_lt_open, + .read = seq_read, + .llseek = seq_lseek, + .release = tracing_release, +}; + +static struct file_operations show_traces_fops = { + .open = show_traces_open, + .read = seq_read, + .release = seq_release, +}; + +static ssize_t +tracing_iter_ctrl_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char *buf; + int r = 0; + int len = 0; + int i; + + /* calulate max size */ + for (i = 0; trace_options[i]; i++) { + len += strlen(trace_options[i]); + len += 3; /* "no" and space */ + } + + /* +2 for \n and \0 */ + buf = kmalloc(len + 2, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + for (i = 0; trace_options[i]; i++) { + if (trace_flags & (1 << i)) + r += sprintf(buf + r, "%s ", trace_options[i]); + else + r += sprintf(buf + r, "no%s ", trace_options[i]); + } + + r += sprintf(buf + r, "\n"); + WARN_ON(r >= len + 2); + + r = simple_read_from_buffer(ubuf, cnt, ppos, + buf, r); + + kfree(buf); + + return r; +} + +static ssize_t +tracing_iter_ctrl_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char buf[64]; + char *cmp = buf; + int neg = 0; + int i; + + if (cnt > 63) + cnt = 63; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + if (strncmp(buf, "no", 2) == 0) { + neg = 1; + cmp += 2; + } + + for (i = 0; trace_options[i]; i++) { + int len = strlen(trace_options[i]); + + if (strncmp(cmp, trace_options[i], len) == 0) { + if (neg) + trace_flags &= ~(1 << i); + else + trace_flags |= (1 << i); + break; + } + } + + filp->f_pos += cnt; + + return cnt; +} + +static struct file_operations tracing_iter_fops = { + .open = tracing_open_generic, + .read = tracing_iter_ctrl_read, + .write = tracing_iter_ctrl_write, +}; + +static ssize_t +tracing_ctrl_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct trace_array *tr = filp->private_data; + char buf[64]; + int r; + + r = sprintf(buf, "%ld\n", tr->ctrl); + return simple_read_from_buffer(ubuf, cnt, ppos, + buf, r); +} + +static ssize_t +tracing_ctrl_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct trace_array *tr = filp->private_data; + long val; + char buf[64]; + + if (cnt > 63) + cnt = 63; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 10); + + val = !!val; + + mutex_lock(&trace_types_lock); + if (tr->ctrl ^ val) { + if (val) + tracer_enabled = 1; + else + tracer_enabled = 0; + + tr->ctrl = val; + + if (current_trace && current_trace->ctrl_update) + current_trace->ctrl_update(tr); + } + mutex_unlock(&trace_types_lock); + + filp->f_pos += cnt; + + return cnt; +} + +static ssize_t +tracing_set_trace_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char buf[max_tracer_type_len+2]; + int r; + + mutex_lock(&trace_types_lock); + if (current_trace) + r = sprintf(buf, "%s\n", current_trace->name); + else + r = sprintf(buf, "\n"); + mutex_unlock(&trace_types_lock); + + return simple_read_from_buffer(ubuf, cnt, ppos, + buf, r); +} + +static ssize_t +tracing_set_trace_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct trace_array *tr = &global_trace; + struct tracer *t; + char buf[max_tracer_type_len+1]; + int i; + + if (cnt > max_tracer_type_len) + cnt = max_tracer_type_len; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + /* strip ending whitespace. */ + for (i = cnt - 1; i > 0 && isspace(buf[i]); i--) + buf[i] = 0; + + mutex_lock(&trace_types_lock); + for (t = trace_types; t; t = t->next) { + if (strcmp(t->name, buf) == 0) + break; + } + if (!t || t == current_trace) + goto out; + + if (current_trace && current_trace->reset) + current_trace->reset(tr); + + current_trace = t; + if (t->init) + t->init(tr); + + out: + mutex_unlock(&trace_types_lock); + + filp->f_pos += cnt; + + return cnt; +} + +static ssize_t +tracing_max_lat_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + unsigned long *ptr = filp->private_data; + char buf[64]; + int r; + + r = snprintf(buf, 64, "%ld\n", + *ptr == (unsigned long)-1 ? -1 : nsecs_to_usecs(*ptr)); + if (r > 64) + r = 64; + return simple_read_from_buffer(ubuf, cnt, ppos, + buf, r); +} + +static ssize_t +tracing_max_lat_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + long *ptr = filp->private_data; + long val; + char buf[64]; + + if (cnt > 63) + cnt = 63; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 10); + + *ptr = val * 1000; + + return cnt; +} + +static struct file_operations tracing_max_lat_fops = { + .open = tracing_open_generic, + .read = tracing_max_lat_read, + .write = tracing_max_lat_write, +}; + +static struct file_operations tracing_ctrl_fops = { + .open = tracing_open_generic, + .read = tracing_ctrl_read, + .write = tracing_ctrl_write, +}; + +static struct file_operations set_tracer_fops = { + .open = tracing_open_generic, + .read = tracing_set_trace_read, + .write = tracing_set_trace_write, +}; + +#ifdef CONFIG_DYNAMIC_FTRACE + +static ssize_t +tracing_read_long(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + unsigned long *p = filp->private_data; + char buf[64]; + int r; + + r = sprintf(buf, "%ld\n", *p); + return simple_read_from_buffer(ubuf, cnt, ppos, + buf, r); +} + +static struct file_operations tracing_read_long_fops = { + .open = tracing_open_generic, + .read = tracing_read_long, +}; +#endif + +static struct dentry *d_tracer; + +struct dentry *tracing_init_dentry(void) +{ + static int once; + + if (d_tracer) + return d_tracer; + + d_tracer = debugfs_create_dir("tracing", NULL); + + if (!d_tracer && !once) { + once = 1; + pr_warning("Could not create debugfs directory 'tracing'\n"); + return NULL; + } + + return d_tracer; +} + +static __init void tracer_init_debugfs(void) +{ + struct dentry *d_tracer; + struct dentry *entry; + + d_tracer = tracing_init_dentry(); + + entry = debugfs_create_file("tracing_enabled", 0644, d_tracer, + &global_trace, &tracing_ctrl_fops); + if (!entry) + pr_warning("Could not create debugfs 'tracing_enabled' entry\n"); + + entry = debugfs_create_file("iter_ctrl", 0644, d_tracer, + NULL, &tracing_iter_fops); + if (!entry) + pr_warning("Could not create debugfs 'iter_ctrl' entry\n"); + + entry = debugfs_create_file("latency_trace", 0444, d_tracer, + &global_trace, &tracing_lt_fops); + if (!entry) + pr_warning("Could not create debugfs 'latency_trace' entry\n"); + + entry = debugfs_create_file("trace", 0444, d_tracer, + &global_trace, &tracing_fops); + if (!entry) + pr_warning("Could not create debugfs 'trace' entry\n"); + + entry = debugfs_create_file("available_tracers", 0444, d_tracer, + &global_trace, &show_traces_fops); + if (!entry) + pr_warning("Could not create debugfs 'trace' entry\n"); + + entry = debugfs_create_file("current_tracer", 0444, d_tracer, + &global_trace, &set_tracer_fops); + if (!entry) + pr_warning("Could not create debugfs 'trace' entry\n"); + + entry = debugfs_create_file("tracing_max_latency", 0644, d_tracer, + &tracing_max_latency, + &tracing_max_lat_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'tracing_max_latency' entry\n"); + + entry = debugfs_create_file("tracing_thresh", 0644, d_tracer, + &tracing_thresh, &tracing_max_lat_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'tracing_threash' entry\n"); + +#ifdef CONFIG_DYNAMIC_FTRACE + entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer, + &ftrace_update_tot_cnt, + &tracing_read_long_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'dyn_ftrace_total_info' entry\n"); +#endif +} + +/* dummy trace to disable tracing */ +static struct tracer no_tracer __read_mostly = +{ + .name = "none", +}; + +static inline notrace int page_order(const unsigned long size) +{ + const unsigned long nr_pages = DIV_ROUND_UP(size, PAGE_SIZE); + return ilog2(roundup_pow_of_two(nr_pages)); +} + +__init static int tracer_alloc_buffers(void) +{ + const int order = page_order(trace_nr_entries * TRACE_ENTRY_SIZE); + const unsigned long size = (1UL << order) << PAGE_SHIFT; + struct trace_entry *array; + int i; + + for_each_possible_cpu(i) { + global_trace.data[i] = &per_cpu(global_trace_cpu, i); + max_tr.data[i] = &per_cpu(max_data, i); + + array = (struct trace_entry *) + __get_free_pages(GFP_KERNEL, order); + if (array == NULL) { + printk(KERN_ERR "tracer: failed to allocate" + " %ld bytes for trace buffer!\n", size); + goto free_buffers; + } + global_trace.data[i]->trace = array; + +/* Only allocate if we are actually using the max trace */ +#ifdef CONFIG_TRACER_MAX_TRACE + array = (struct trace_entry *) + __get_free_pages(GFP_KERNEL, order); + if (array == NULL) { + printk(KERN_ERR "wakeup tracer: failed to allocate" + " %ld bytes for trace buffer!\n", size); + goto free_buffers; + } + max_tr.data[i]->trace = array; +#endif + } + + /* + * Since we allocate by orders of pages, we may be able to + * round up a bit. + */ + global_trace.entries = size / TRACE_ENTRY_SIZE; + max_tr.entries = global_trace.entries; + + pr_info("tracer: %ld bytes allocated for %ld", + size, trace_nr_entries); + pr_info(" entries of %ld bytes\n", (long)TRACE_ENTRY_SIZE); + pr_info(" actual entries %ld\n", global_trace.entries); + + tracer_init_debugfs(); + + trace_init_cmdlines(); + + register_tracer(&no_tracer); + current_trace = &no_tracer; + + return 0; + + free_buffers: + for (i-- ; i >= 0; i--) { + struct trace_array_cpu *data = global_trace.data[i]; + + if (data && data->trace) { + free_pages((unsigned long)data->trace, order); + data->trace = NULL; + } + +#ifdef CONFIG_TRACER_MAX_TRACE + data = max_tr.data[i]; + if (data && data->trace) { + free_pages((unsigned long)data->trace, order); + data->trace = NULL; + } +#endif + } + return -ENOMEM; +} + +device_initcall(tracer_alloc_buffers); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h new file mode 100644 index 000000000000..3173a93561d4 --- /dev/null +++ b/kernel/trace/trace.h @@ -0,0 +1,184 @@ +#ifndef _LINUX_KERNEL_TRACE_H +#define _LINUX_KERNEL_TRACE_H + +#include +#include +#include +#include + +/* + * Function trace entry - function address and parent function addres: + */ +struct ftrace_entry { + unsigned long ip; + unsigned long parent_ip; +}; + +/* + * Context switch trace entry - which task (and prio) we switched from/to: + */ +struct ctx_switch_entry { + unsigned int prev_pid; + unsigned char prev_prio; + unsigned char prev_state; + unsigned int next_pid; + unsigned char next_prio; +}; + +/* + * The trace entry - the most basic unit of tracing. This is what + * is printed in the end as a single line in the trace output, such as: + * + * bash-15816 [01] 235.197585: idle_cpu <- irq_enter + */ +struct trace_entry { + char type; + char cpu; + char flags; + char preempt_count; + int pid; + cycle_t t; + unsigned long idx; + union { + struct ftrace_entry fn; + struct ctx_switch_entry ctx; + }; +}; + +#define TRACE_ENTRY_SIZE sizeof(struct trace_entry) + +/* + * The CPU trace array - it consists of thousands of trace entries + * plus some other descriptor data: (for example which task started + * the trace, etc.) + */ +struct trace_array_cpu { + void *trace; + unsigned long trace_idx; + atomic_t disabled; + atomic_t underrun; + unsigned long saved_latency; + unsigned long critical_start; + unsigned long critical_end; + unsigned long critical_sequence; + unsigned long nice; + unsigned long policy; + unsigned long rt_priority; + cycle_t preempt_timestamp; + pid_t pid; + uid_t uid; + char comm[TASK_COMM_LEN]; +}; + +struct trace_iterator; + +/* + * The trace array - an array of per-CPU trace arrays. This is the + * highest level data structure that individual tracers deal with. + * They have on/off state as well: + */ +struct trace_array { + unsigned long entries; + long ctrl; + int cpu; + cycle_t time_start; + struct trace_array_cpu *data[NR_CPUS]; +}; + +/* + * A specific tracer, represented by methods that operate on a trace array: + */ +struct tracer { + const char *name; + void (*init)(struct trace_array *tr); + void (*reset)(struct trace_array *tr); + void (*open)(struct trace_iterator *iter); + void (*close)(struct trace_iterator *iter); + void (*start)(struct trace_iterator *iter); + void (*stop)(struct trace_iterator *iter); + void (*ctrl_update)(struct trace_array *tr); + struct tracer *next; + int print_max; +}; + +/* + * Trace iterator - used by printout routines who present trace + * results to users and which routines might sleep, etc: + */ +struct trace_iterator { + struct trace_array *tr; + struct tracer *trace; + struct trace_entry *ent; + unsigned long iter_flags; + loff_t pos; + unsigned long next_idx[NR_CPUS]; + int cpu; + int idx; +}; + +void notrace tracing_reset(struct trace_array_cpu *data); +int tracing_open_generic(struct inode *inode, struct file *filp); +struct dentry *tracing_init_dentry(void); +void ftrace(struct trace_array *tr, + struct trace_array_cpu *data, + unsigned long ip, + unsigned long parent_ip, + unsigned long flags); +void tracing_sched_switch_trace(struct trace_array *tr, + struct trace_array_cpu *data, + struct task_struct *prev, + struct task_struct *next, + unsigned long flags); +void tracing_record_cmdline(struct task_struct *tsk); + +void tracing_start_function_trace(void); +void tracing_stop_function_trace(void); +int register_tracer(struct tracer *type); +void unregister_tracer(struct tracer *type); + +extern unsigned long nsecs_to_usecs(unsigned long nsecs); + +extern unsigned long tracing_max_latency; +extern unsigned long tracing_thresh; + +void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); +void update_max_tr_single(struct trace_array *tr, + struct task_struct *tsk, int cpu); + +static inline notrace cycle_t now(int cpu) +{ + return cpu_clock(cpu); +} + +#ifdef CONFIG_SCHED_TRACER +extern void notrace +wakeup_sched_switch(struct task_struct *prev, struct task_struct *next); +#else +static inline void +wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) +{ +} +#endif + +#ifdef CONFIG_CONTEXT_SWITCH_TRACER +typedef void +(*tracer_switch_func_t)(void *private, + struct task_struct *prev, + struct task_struct *next); + +struct tracer_switch_ops { + tracer_switch_func_t func; + void *private; + struct tracer_switch_ops *next; +}; + +extern int register_tracer_switch(struct tracer_switch_ops *ops); +extern int unregister_tracer_switch(struct tracer_switch_ops *ops); + +#endif /* CONFIG_CONTEXT_SWITCH_TRACER */ + +#ifdef CONFIG_DYNAMIC_FTRACE +extern unsigned long ftrace_update_tot_cnt; +#endif + +#endif /* _LINUX_KERNEL_TRACE_H */ -- cgit v1.2.3 From 1b29b01887e6032dcaf818c14999c7a39593b4e7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: function tracer This is a simple trace that uses the ftrace infrastructure. It is designed to be fast and small, and easy to use. It is useful to record things that happen over a very short period of time, and not to analyze the system in general. Updates: available_tracers "function" is added to this file. current_tracer To enable the function tracer: echo function > /debugfs/tracing/current_tracer To disable the tracer: echo disable > /debugfs/tracing/current_tracer The output of the function_trace file is as follows "echo noverbose > /debugfs/tracing/iter_ctrl" preemption latency trace v1.1.5 on 2.6.24-rc7-tst Signed-off-by: Ingo Molnar -------------------------------------------------------------------- latency: 0 us, #419428/4361791, CPU#1 | (M:desktop VP:0, KP:0, SP:0 HP:0 #P:4) ----------------- | task: -0 (uid:0 nice:0 policy:0 rt_prio:0) ----------------- _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / swapper-0 0d.h. 1595128us+: set_normalized_timespec+0x8/0x2d (ktime_get_ts+0x4a/0x4e ) swapper-0 0d.h. 1595131us+: _spin_lock+0x8/0x18 (hrtimer_interrupt+0x6e/0x1b0 ) Or with verbose turned on: "echo verbose > /debugfs/tracing/iter_ctrl" preemption latency trace v1.1.5 on 2.6.24-rc7-tst -------------------------------------------------------------------- latency: 0 us, #419428/4361791, CPU#1 | (M:desktop VP:0, KP:0, SP:0 HP:0 #P:4) ----------------- | task: -0 (uid:0 nice:0 policy:0 rt_prio:0) ----------------- swapper 0 0 9 00000000 00000000 [f3675f41] 1595.128ms (+0.003ms): set_normalized_timespec+0x8/0x2d (ktime_get_ts+0x4a/0x4e ) swapper 0 0 9 00000000 00000001 [f3675f45] 1595.131ms (+0.003ms): _spin_lock+0x8/0x18 (hrtimer_interrupt+0x6e/0x1b0 ) swapper 0 0 9 00000000 00000002 [f3675f48] 1595.135ms (+0.003ms): _spin_lock+0x8/0x18 (hrtimer_interrupt+0x6e/0x1b0 ) The "trace" file is not affected by the verbose mode, but is by the symonly. echo "nosymonly" > /debugfs/tracing/iter_ctrl tracer: [ 81.479967] CPU 0: bash:3154 register_ftrace_function+0x5f/0x66 <-- _spin_unlock_irqrestore+0xe/0x5a [ 81.479967] CPU 0: bash:3154 _spin_unlock_irqrestore+0x3e/0x5a <-- sub_preempt_count+0xc/0x7a [ 81.479968] CPU 0: bash:3154 sub_preempt_count+0x30/0x7a <-- in_lock_functions+0x9/0x24 [ 81.479968] CPU 0: bash:3154 vfs_write+0x11d/0x155 <-- dnotify_parent+0x12/0x78 [ 81.479968] CPU 0: bash:3154 dnotify_parent+0x2d/0x78 <-- _spin_lock+0xe/0x70 [ 81.479969] CPU 0: bash:3154 _spin_lock+0x1b/0x70 <-- add_preempt_count+0xe/0x77 [ 81.479969] CPU 0: bash:3154 add_preempt_count+0x3e/0x77 <-- in_lock_functions+0x9/0x24 echo "symonly" > /debugfs/tracing/iter_ctrl tracer: [ 81.479913] CPU 0: bash:3154 register_ftrace_function+0x5f/0x66 <-- _spin_unlock_irqrestore+0xe/0x5a [ 81.479913] CPU 0: bash:3154 _spin_unlock_irqrestore+0x3e/0x5a <-- sub_preempt_count+0xc/0x7a [ 81.479913] CPU 0: bash:3154 sub_preempt_count+0x30/0x7a <-- in_lock_functions+0x9/0x24 [ 81.479914] CPU 0: bash:3154 vfs_write+0x11d/0x155 <-- dnotify_parent+0x12/0x78 [ 81.479914] CPU 0: bash:3154 dnotify_parent+0x2d/0x78 <-- _spin_lock+0xe/0x70 [ 81.479914] CPU 0: bash:3154 _spin_lock+0x1b/0x70 <-- add_preempt_count+0xe/0x77 [ 81.479914] CPU 0: bash:3154 add_preempt_count+0x3e/0x77 <-- in_lock_functions+0x9/0x24 Signed-off-by: Steven Rostedt Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 13 ++++++++ kernel/trace/Makefile | 1 + kernel/trace/trace_functions.c | 73 ++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 87 insertions(+) create mode 100644 kernel/trace/trace_functions.c (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index ce70677afbf9..1399f372b5dc 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -8,3 +8,16 @@ config TRACING bool select DEBUG_FS +config FTRACE + bool "Kernel Function Tracer" + depends on DEBUG_KERNEL && HAVE_FTRACE + select FRAME_POINTER + select TRACING + help + Enable the kernel to trace every kernel function. This is done + by using a compiler feature to insert a small, 5-byte No-Operation + instruction to the beginning of every kernel function, which NOP + sequence is then dynamically patched into a tracer call when + tracing is enabled by the administrator. If it's runtime disabled + (the bootup default), then the overhead of the instructions is very + small and not measurable even in micro-benchmarks. diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 7af403175255..6bb5e50b4a40 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -1,5 +1,6 @@ obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o +obj-$(CONFIG_FTRACE) += trace_functions.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c new file mode 100644 index 000000000000..82988c5336e0 --- /dev/null +++ b/kernel/trace/trace_functions.c @@ -0,0 +1,73 @@ +/* + * ring buffer based function tracer + * + * Copyright (C) 2007-2008 Steven Rostedt + * Copyright (C) 2008 Ingo Molnar + * + * Based on code from the latency_tracer, that is: + * + * Copyright (C) 2004-2006 Ingo Molnar + * Copyright (C) 2004 William Lee Irwin III + */ +#include +#include +#include +#include + +#include "trace.h" + +static notrace void function_reset(struct trace_array *tr) +{ + int cpu; + + tr->time_start = now(tr->cpu); + + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); +} + +static notrace void start_function_trace(struct trace_array *tr) +{ + function_reset(tr); + tracing_start_function_trace(); +} + +static notrace void stop_function_trace(struct trace_array *tr) +{ + tracing_stop_function_trace(); +} + +static notrace void function_trace_init(struct trace_array *tr) +{ + if (tr->ctrl) + start_function_trace(tr); +} + +static notrace void function_trace_reset(struct trace_array *tr) +{ + if (tr->ctrl) + stop_function_trace(tr); +} + +static notrace void function_trace_ctrl_update(struct trace_array *tr) +{ + if (tr->ctrl) + start_function_trace(tr); + else + stop_function_trace(tr); +} + +static struct tracer function_trace __read_mostly = +{ + .name = "ftrace", + .init = function_trace_init, + .reset = function_trace_reset, + .ctrl_update = function_trace_ctrl_update, +}; + +static __init int init_function_trace(void) +{ + return register_tracer(&function_trace); +} + +device_initcall(init_function_trace); -- cgit v1.2.3 From 35e8e302e5d6e32675df2fc1dd3a53dfa6630dc1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: add tracing of context switches This patch adds context switch tracing, of the format of: _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | pid:prio:state \ / ||||| \ | / swapper-0 1d..3 137us+: 0:140:R --> 2912:120 sshd-2912 1d..3 216us+: 2912:120:S --> 0:140 swapper-0 1d..3 261us+: 0:140:R --> 2912:120 bash-2920 0d..3 267us+: 2920:120:S --> 0:140 sshd-2912 1d..3 330us!: 2912:120:S --> 0:140 swapper-0 1d..3 2389us+: 0:140:R --> 2847:120 yum-upda-2847 1d..3 2411us!: 2847:120:S --> 0:140 swapper-0 0d..3 11089us+: 0:140:R --> 3139:120 gdm-bina-3139 0d..3 11113us!: 3139:120:S --> 0:140 swapper-0 1d..3 102328us+: 0:140:R --> 2847:120 yum-upda-2847 1d..3 102348us!: 2847:120:S --> 0:140 "sched_switch" is added to /debugfs/tracing/available_tracers [ Eugene Teo Cc: Mathieu Desnoyers Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 11 ++++ kernel/trace/Makefile | 1 + kernel/trace/trace_sched_switch.c | 116 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 128 insertions(+) create mode 100644 kernel/trace/trace_sched_switch.c (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 1399f372b5dc..5d6aa92866cd 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -13,6 +13,7 @@ config FTRACE depends on DEBUG_KERNEL && HAVE_FTRACE select FRAME_POINTER select TRACING + select CONTEXT_SWITCH_TRACER help Enable the kernel to trace every kernel function. This is done by using a compiler feature to insert a small, 5-byte No-Operation @@ -21,3 +22,13 @@ config FTRACE tracing is enabled by the administrator. If it's runtime disabled (the bootup default), then the overhead of the instructions is very small and not measurable even in micro-benchmarks. + +config CONTEXT_SWITCH_TRACER + bool "Trace process context switches" + depends on DEBUG_KERNEL + select TRACING + select MARKERS + help + This tracer gets called from the context switch and records + all switching of tasks. + diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 6bb5e50b4a40..6b54ceb7f16e 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -1,6 +1,7 @@ obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o +obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o obj-$(CONFIG_FTRACE) += trace_functions.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c new file mode 100644 index 000000000000..3e4771d3b890 --- /dev/null +++ b/kernel/trace/trace_sched_switch.c @@ -0,0 +1,116 @@ +/* + * trace context switch + * + * Copyright (C) 2007 Steven Rostedt + * + */ +#include +#include +#include +#include +#include +#include +#include + +#include "trace.h" + +static struct trace_array *ctx_trace; +static int __read_mostly tracer_enabled; + +static void notrace +ctx_switch_func(struct task_struct *prev, struct task_struct *next) +{ + struct trace_array *tr = ctx_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (!tracer_enabled) + return; + + raw_local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + tracing_sched_switch_trace(tr, data, prev, next, flags); + + atomic_dec(&data->disabled); + raw_local_irq_restore(flags); +} + +void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) +{ + tracing_record_cmdline(prev); + + /* + * If tracer_switch_func only points to the local + * switch func, it still needs the ptr passed to it. + */ + ctx_switch_func(prev, next); + + /* + * Chain to the wakeup tracer (this is a NOP if disabled): + */ + wakeup_sched_switch(prev, next); +} + +static notrace void sched_switch_reset(struct trace_array *tr) +{ + int cpu; + + tr->time_start = now(tr->cpu); + + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); +} + +static notrace void start_sched_trace(struct trace_array *tr) +{ + sched_switch_reset(tr); + tracer_enabled = 1; +} + +static notrace void stop_sched_trace(struct trace_array *tr) +{ + tracer_enabled = 0; +} + +static notrace void sched_switch_trace_init(struct trace_array *tr) +{ + ctx_trace = tr; + + if (tr->ctrl) + start_sched_trace(tr); +} + +static notrace void sched_switch_trace_reset(struct trace_array *tr) +{ + if (tr->ctrl) + stop_sched_trace(tr); +} + +static void sched_switch_trace_ctrl_update(struct trace_array *tr) +{ + /* When starting a new trace, reset the buffers */ + if (tr->ctrl) + start_sched_trace(tr); + else + stop_sched_trace(tr); +} + +static struct tracer sched_switch_trace __read_mostly = +{ + .name = "sched_switch", + .init = sched_switch_trace_init, + .reset = sched_switch_trace_reset, + .ctrl_update = sched_switch_trace_ctrl_update, +}; + +__init static int init_sched_switch_trace(void) +{ + return register_tracer(&sched_switch_trace); +} +device_initcall(init_sched_switch_trace); -- cgit v1.2.3 From 352ad25aa4a189c667cb2af333948d34692a2d27 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: tracer for scheduler wakeup latency This patch adds the tracer that tracks the wakeup latency of the highest priority waking task. "wakeup" is added to /debugfs/tracing/available_tracers Also added to /debugfs/tracing tracing_max_latency holds the current max latency for the wakeup wakeup_thresh if set to other than zero, a log will be recorded for every wakeup that takes longer than the number entered in here (usecs for all counters) (deletes previous trace) Examples: (with ftrace_enabled = 0) ============ preemption latency trace v1.1.5 on 2.6.24-rc8 Signed-off-by: Ingo Molnar -------------------------------------------------------------------- latency: 26 us, #2/2, CPU#1 | (M:rt VP:0, KP:0, SP:0 HP:0 #P:2) ----------------- | task: migration/0-3 (uid:0 nice:-5 policy:1 rt_prio:99) ----------------- _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / quilt-8551 0d..3 0us+: wake_up_process+0x15/0x17 (sched_exec+0xc9/0x100 ) quilt-8551 0d..4 26us : sched_switch_callback+0x73/0x81 (schedule+0x483/0x6d5 ) vim:ft=help ============ (with ftrace_enabled = 1) ============ preemption latency trace v1.1.5 on 2.6.24-rc8 -------------------------------------------------------------------- latency: 36 us, #45/45, CPU#0 | (M:rt VP:0, KP:0, SP:0 HP:0 #P:2) ----------------- | task: migration/1-5 (uid:0 nice:-5 policy:1 rt_prio:99) ----------------- _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / bash-10653 1d..3 0us : wake_up_process+0x15/0x17 (sched_exec+0xc9/0x100 ) bash-10653 1d..3 1us : try_to_wake_up+0x271/0x2e7 (sub_preempt_count+0xc/0x7a ) bash-10653 1d..2 2us : try_to_wake_up+0x296/0x2e7 (update_rq_clock+0x9/0x20 ) bash-10653 1d..2 2us : update_rq_clock+0x1e/0x20 (__update_rq_clock+0xc/0x90 ) bash-10653 1d..2 3us : __update_rq_clock+0x1b/0x90 (sched_clock+0x9/0x29 ) bash-10653 1d..2 4us : try_to_wake_up+0x2a6/0x2e7 (activate_task+0xc/0x3f ) bash-10653 1d..2 4us : activate_task+0x2d/0x3f (enqueue_task+0xe/0x66 ) bash-10653 1d..2 5us : enqueue_task+0x5b/0x66 (enqueue_task_rt+0x9/0x3c ) bash-10653 1d..2 6us : try_to_wake_up+0x2ba/0x2e7 (check_preempt_wakeup+0x12/0x99 ) [...] bash-10653 1d..5 33us : tracing_record_cmdline+0xcf/0xd4 (_spin_unlock+0x9/0x33 ) bash-10653 1d..5 34us : _spin_unlock+0x19/0x33 (sub_preempt_count+0xc/0x7a ) bash-10653 1d..4 35us : wakeup_sched_switch+0x65/0x2ff (_spin_lock_irqsave+0xc/0xa9 ) bash-10653 1d..4 35us : _spin_lock_irqsave+0x19/0xa9 (add_preempt_count+0xe/0x77 ) bash-10653 1d..4 36us : sched_switch_callback+0x73/0x81 (schedule+0x483/0x6d5 ) vim:ft=help ============ The [...] was added here to not waste your email box space. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 13 ++ kernel/trace/Makefile | 1 + kernel/trace/trace_sched_wakeup.c | 310 ++++++++++++++++++++++++++++++++++++++ 3 files changed, 324 insertions(+) create mode 100644 kernel/trace/trace_sched_wakeup.c (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 5d6aa92866cd..892ecc94a82b 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -4,6 +4,9 @@ config HAVE_FTRACE bool +config TRACER_MAX_TRACE + bool + config TRACING bool select DEBUG_FS @@ -23,6 +26,16 @@ config FTRACE (the bootup default), then the overhead of the instructions is very small and not measurable even in micro-benchmarks. +config SCHED_TRACER + bool "Scheduling Latency Tracer" + depends on DEBUG_KERNEL + select TRACING + select CONTEXT_SWITCH_TRACER + select TRACER_MAX_TRACE + help + This tracer tracks the latency of the highest priority task + to be scheduled in, starting from the point it has woken up. + config CONTEXT_SWITCH_TRACER bool "Trace process context switches" depends on DEBUG_KERNEL diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 6b54ceb7f16e..5508cdb19aea 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -3,5 +3,6 @@ obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o obj-$(CONFIG_FTRACE) += trace_functions.o +obj-$(CONFIG_SCHED_TRACER) += trace_sched_wakeup.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c new file mode 100644 index 000000000000..7c3ccefcf4c3 --- /dev/null +++ b/kernel/trace/trace_sched_wakeup.c @@ -0,0 +1,310 @@ +/* + * trace task wakeup timings + * + * Copyright (C) 2007-2008 Steven Rostedt + * Copyright (C) 2008 Ingo Molnar + * + * Based on code from the latency_tracer, that is: + * + * Copyright (C) 2004-2006 Ingo Molnar + * Copyright (C) 2004 William Lee Irwin III + */ +#include +#include +#include +#include +#include +#include + +#include "trace.h" + +static struct trace_array *wakeup_trace; +static int __read_mostly tracer_enabled; + +static struct task_struct *wakeup_task; +static int wakeup_cpu; +static unsigned wakeup_prio = -1; + +static DEFINE_SPINLOCK(wakeup_lock); + +static void notrace __wakeup_reset(struct trace_array *tr); + +/* + * Should this new latency be reported/recorded? + */ +static int notrace report_latency(cycle_t delta) +{ + if (tracing_thresh) { + if (delta < tracing_thresh) + return 0; + } else { + if (delta <= tracing_max_latency) + return 0; + } + return 1; +} + +void notrace +wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) +{ + unsigned long latency = 0, t0 = 0, t1 = 0; + struct trace_array *tr = wakeup_trace; + struct trace_array_cpu *data; + cycle_t T0, T1, delta; + unsigned long flags; + long disabled; + int cpu; + + if (unlikely(!tracer_enabled)) + return; + + /* + * When we start a new trace, we set wakeup_task to NULL + * and then set tracer_enabled = 1. We want to make sure + * that another CPU does not see the tracer_enabled = 1 + * and the wakeup_task with an older task, that might + * actually be the same as next. + */ + smp_rmb(); + + if (next != wakeup_task) + return; + + /* The task we are waitng for is waking up */ + data = tr->data[wakeup_cpu]; + + /* disable local data, not wakeup_cpu data */ + cpu = raw_smp_processor_id(); + disabled = atomic_inc_return(&tr->data[cpu]->disabled); + if (likely(disabled != 1)) + goto out; + + spin_lock_irqsave(&wakeup_lock, flags); + + /* We could race with grabbing wakeup_lock */ + if (unlikely(!tracer_enabled || next != wakeup_task)) + goto out_unlock; + + ftrace(tr, data, CALLER_ADDR1, CALLER_ADDR2, flags); + + /* + * usecs conversion is slow so we try to delay the conversion + * as long as possible: + */ + T0 = data->preempt_timestamp; + T1 = now(cpu); + delta = T1-T0; + + if (!report_latency(delta)) + goto out_unlock; + + latency = nsecs_to_usecs(delta); + + tracing_max_latency = delta; + t0 = nsecs_to_usecs(T0); + t1 = nsecs_to_usecs(T1); + + update_max_tr(tr, wakeup_task, wakeup_cpu); + + if (tracing_thresh) { + printk(KERN_INFO "(%16s-%-5d|#%d): %lu us wakeup latency " + "violates %lu us threshold.\n" + " => started at timestamp %lu: ", + wakeup_task->comm, wakeup_task->pid, + raw_smp_processor_id(), + latency, nsecs_to_usecs(tracing_thresh), t0); + } else { + printk(KERN_INFO "(%16s-%-5d|#%d): new %lu us maximum " + "wakeup latency.\n => started at timestamp %lu: ", + wakeup_task->comm, wakeup_task->pid, + cpu, latency, t0); + } + + printk(KERN_CONT " ended at timestamp %lu: ", t1); + dump_stack(); + t1 = nsecs_to_usecs(now(cpu)); + printk(KERN_CONT " dump-end timestamp %lu\n\n", t1); + +out_unlock: + __wakeup_reset(tr); + spin_unlock_irqrestore(&wakeup_lock, flags); +out: + atomic_dec(&tr->data[cpu]->disabled); +} + +static void notrace __wakeup_reset(struct trace_array *tr) +{ + struct trace_array_cpu *data; + int cpu; + + assert_spin_locked(&wakeup_lock); + + for_each_possible_cpu(cpu) { + data = tr->data[cpu]; + tracing_reset(data); + } + + wakeup_cpu = -1; + wakeup_prio = -1; + + if (wakeup_task) + put_task_struct(wakeup_task); + + wakeup_task = NULL; +} + +static void notrace wakeup_reset(struct trace_array *tr) +{ + unsigned long flags; + + spin_lock_irqsave(&wakeup_lock, flags); + __wakeup_reset(tr); + spin_unlock_irqrestore(&wakeup_lock, flags); +} + +static notrace void +wakeup_check_start(struct trace_array *tr, struct task_struct *p, + struct task_struct *curr) +{ + int cpu = smp_processor_id(); + unsigned long flags; + long disabled; + + if (likely(!rt_task(p)) || + p->prio >= wakeup_prio || + p->prio >= curr->prio) + return; + + disabled = atomic_inc_return(&tr->data[cpu]->disabled); + if (unlikely(disabled != 1)) + goto out; + + /* interrupts should be off from try_to_wake_up */ + spin_lock(&wakeup_lock); + + /* check for races. */ + if (!tracer_enabled || p->prio >= wakeup_prio) + goto out_locked; + + /* reset the trace */ + __wakeup_reset(tr); + + wakeup_cpu = task_cpu(p); + wakeup_prio = p->prio; + + wakeup_task = p; + get_task_struct(wakeup_task); + + local_save_flags(flags); + + tr->data[wakeup_cpu]->preempt_timestamp = now(cpu); + ftrace(tr, tr->data[wakeup_cpu], CALLER_ADDR1, CALLER_ADDR2, flags); + +out_locked: + spin_unlock(&wakeup_lock); +out: + atomic_dec(&tr->data[cpu]->disabled); +} + +notrace void +ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +{ + if (likely(!tracer_enabled)) + return; + + wakeup_check_start(wakeup_trace, wakee, curr); +} + +notrace void +ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) +{ + if (likely(!tracer_enabled)) + return; + + wakeup_check_start(wakeup_trace, wakee, curr); +} + +static notrace void start_wakeup_tracer(struct trace_array *tr) +{ + wakeup_reset(tr); + + /* + * Don't let the tracer_enabled = 1 show up before + * the wakeup_task is reset. This may be overkill since + * wakeup_reset does a spin_unlock after setting the + * wakeup_task to NULL, but I want to be safe. + * This is a slow path anyway. + */ + smp_wmb(); + + tracer_enabled = 1; + + return; +} + +static notrace void stop_wakeup_tracer(struct trace_array *tr) +{ + tracer_enabled = 0; +} + +static notrace void wakeup_tracer_init(struct trace_array *tr) +{ + wakeup_trace = tr; + + if (tr->ctrl) + start_wakeup_tracer(tr); +} + +static notrace void wakeup_tracer_reset(struct trace_array *tr) +{ + if (tr->ctrl) { + stop_wakeup_tracer(tr); + /* make sure we put back any tasks we are tracing */ + wakeup_reset(tr); + } +} + +static void wakeup_tracer_ctrl_update(struct trace_array *tr) +{ + if (tr->ctrl) + start_wakeup_tracer(tr); + else + stop_wakeup_tracer(tr); +} + +static void notrace wakeup_tracer_open(struct trace_iterator *iter) +{ + /* stop the trace while dumping */ + if (iter->tr->ctrl) + stop_wakeup_tracer(iter->tr); +} + +static void notrace wakeup_tracer_close(struct trace_iterator *iter) +{ + /* forget about any processes we were recording */ + if (iter->tr->ctrl) + start_wakeup_tracer(iter->tr); +} + +static struct tracer wakeup_tracer __read_mostly = +{ + .name = "wakeup", + .init = wakeup_tracer_init, + .reset = wakeup_tracer_reset, + .open = wakeup_tracer_open, + .close = wakeup_tracer_close, + .ctrl_update = wakeup_tracer_ctrl_update, + .print_max = 1, +}; + +__init static int init_wakeup_tracer(void) +{ + int ret; + + ret = register_tracer(&wakeup_tracer); + if (ret) + return ret; + + return 0; +} +device_initcall(init_wakeup_tracer); -- cgit v1.2.3 From 81d68a96a39844853b37f20cc8282d9b65b78ef3 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: trace irq disabled critical timings This patch adds latency tracing for critical timings (how long interrupts are disabled for). "irqsoff" is added to /debugfs/tracing/available_tracers Note: tracing_max_latency also holds the max latency for irqsoff (in usecs). (default to large number so one must start latency tracing) tracing_thresh threshold (in usecs) to always print out if irqs off is detected to be longer than stated here. If irq_thresh is non-zero, then max_irq_latency is ignored. Here's an example of a trace with ftrace_enabled = 0 ======= preemption latency trace v1.1.5 on 2.6.24-rc7 Signed-off-by: Ingo Molnar -------------------------------------------------------------------- latency: 100 us, #3/3, CPU#1 | (M:rt VP:0, KP:0, SP:0 HP:0 #P:2) ----------------- | task: swapper-0 (uid:0 nice:0 policy:0 rt_prio:0) ----------------- => started at: _spin_lock_irqsave+0x2a/0xb7 => ended at: _spin_unlock_irqrestore+0x32/0x5f _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / swapper-0 1d.s3 0us+: _spin_lock_irqsave+0x2a/0xb7 (e1000_update_stats+0x47/0x64c [e1000]) swapper-0 1d.s3 100us : _spin_unlock_irqrestore+0x32/0x5f (e1000_update_stats+0x641/0x64c [e1000]) swapper-0 1d.s3 100us : trace_hardirqs_on_caller+0x75/0x89 (_spin_unlock_irqrestore+0x32/0x5f) vim:ft=help ======= And this is a trace with ftrace_enabled == 1 ======= preemption latency trace v1.1.5 on 2.6.24-rc7 -------------------------------------------------------------------- latency: 102 us, #12/12, CPU#1 | (M:rt VP:0, KP:0, SP:0 HP:0 #P:2) ----------------- | task: swapper-0 (uid:0 nice:0 policy:0 rt_prio:0) ----------------- => started at: _spin_lock_irqsave+0x2a/0xb7 => ended at: _spin_unlock_irqrestore+0x32/0x5f _------=> CPU# / _-----=> irqs-off | / _----=> need-resched || / _---=> hardirq/softirq ||| / _--=> preempt-depth |||| / ||||| delay cmd pid ||||| time | caller \ / ||||| \ | / swapper-0 1dNs3 0us+: _spin_lock_irqsave+0x2a/0xb7 (e1000_update_stats+0x47/0x64c [e1000]) swapper-0 1dNs3 46us : e1000_read_phy_reg+0x16/0x225 [e1000] (e1000_update_stats+0x5e2/0x64c [e1000]) swapper-0 1dNs3 46us : e1000_swfw_sync_acquire+0x10/0x99 [e1000] (e1000_read_phy_reg+0x49/0x225 [e1000]) swapper-0 1dNs3 46us : e1000_get_hw_eeprom_semaphore+0x12/0xa6 [e1000] (e1000_swfw_sync_acquire+0x36/0x99 [e1000]) swapper-0 1dNs3 47us : __const_udelay+0x9/0x47 (e1000_read_phy_reg+0x116/0x225 [e1000]) swapper-0 1dNs3 47us+: __delay+0x9/0x50 (__const_udelay+0x45/0x47) swapper-0 1dNs3 97us : preempt_schedule+0xc/0x84 (__delay+0x4e/0x50) swapper-0 1dNs3 98us : e1000_swfw_sync_release+0xc/0x55 [e1000] (e1000_read_phy_reg+0x211/0x225 [e1000]) swapper-0 1dNs3 99us+: e1000_put_hw_eeprom_semaphore+0x9/0x35 [e1000] (e1000_swfw_sync_release+0x50/0x55 [e1000]) swapper-0 1dNs3 101us : _spin_unlock_irqrestore+0xe/0x5f (e1000_update_stats+0x641/0x64c [e1000]) swapper-0 1dNs3 102us : _spin_unlock_irqrestore+0x32/0x5f (e1000_update_stats+0x641/0x64c [e1000]) swapper-0 1dNs3 102us : trace_hardirqs_on_caller+0x75/0x89 (_spin_unlock_irqrestore+0x32/0x5f) vim:ft=help ======= Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/fork.c | 2 +- kernel/lockdep.c | 23 ++- kernel/printk.c | 2 + kernel/trace/Kconfig | 18 ++ kernel/trace/Makefile | 1 + kernel/trace/trace_irqsoff.c | 402 +++++++++++++++++++++++++++++++++++++++++++ 6 files changed, 443 insertions(+), 5 deletions(-) create mode 100644 kernel/trace/trace_irqsoff.c (limited to 'kernel') diff --git a/kernel/fork.c b/kernel/fork.c index 19908b26cf80..d66d676dc362 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -909,7 +909,7 @@ static struct task_struct *copy_process(unsigned long clone_flags, rt_mutex_init_task(p); -#ifdef CONFIG_TRACE_IRQFLAGS +#if defined(CONFIG_TRACE_IRQFLAGS) && defined(CONFIG_LOCKDEP) DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled); DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled); #endif diff --git a/kernel/lockdep.c b/kernel/lockdep.c index 81a4e4a3f087..e21924365ea3 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -39,6 +39,7 @@ #include #include #include +#include #include @@ -982,7 +983,7 @@ check_noncircular(struct lock_class *source, unsigned int depth) return 1; } -#ifdef CONFIG_TRACE_IRQFLAGS +#if defined(CONFIG_TRACE_IRQFLAGS) && defined(CONFIG_PROVE_LOCKING) /* * Forwards and backwards subgraph searching, for the purposes of * proving that two subgraphs can be connected by a new dependency @@ -1680,7 +1681,7 @@ valid_state(struct task_struct *curr, struct held_lock *this, static int mark_lock(struct task_struct *curr, struct held_lock *this, enum lock_usage_bit new_bit); -#ifdef CONFIG_TRACE_IRQFLAGS +#if defined(CONFIG_TRACE_IRQFLAGS) && defined(CONFIG_PROVE_LOCKING) /* * print irq inversion bug: @@ -2013,11 +2014,13 @@ void early_boot_irqs_on(void) /* * Hardirqs will be enabled: */ -void trace_hardirqs_on(void) +void notrace trace_hardirqs_on_caller(unsigned long a0) { struct task_struct *curr = current; unsigned long ip; + time_hardirqs_on(CALLER_ADDR0, a0); + if (unlikely(!debug_locks || current->lockdep_recursion)) return; @@ -2055,16 +2058,23 @@ void trace_hardirqs_on(void) curr->hardirq_enable_event = ++curr->irq_events; debug_atomic_inc(&hardirqs_on_events); } +EXPORT_SYMBOL(trace_hardirqs_on_caller); +void notrace trace_hardirqs_on(void) +{ + trace_hardirqs_on_caller(CALLER_ADDR0); +} EXPORT_SYMBOL(trace_hardirqs_on); /* * Hardirqs were disabled: */ -void trace_hardirqs_off(void) +void notrace trace_hardirqs_off_caller(unsigned long a0) { struct task_struct *curr = current; + time_hardirqs_off(CALLER_ADDR0, a0); + if (unlikely(!debug_locks || current->lockdep_recursion)) return; @@ -2082,7 +2092,12 @@ void trace_hardirqs_off(void) } else debug_atomic_inc(&redundant_hardirqs_off); } +EXPORT_SYMBOL(trace_hardirqs_off_caller); +void notrace trace_hardirqs_off(void) +{ + trace_hardirqs_off_caller(CALLER_ADDR0); +} EXPORT_SYMBOL(trace_hardirqs_off); /* diff --git a/kernel/printk.c b/kernel/printk.c index 8fb01c32aa3b..ae7d5b9e535d 100644 --- a/kernel/printk.c +++ b/kernel/printk.c @@ -1041,7 +1041,9 @@ void release_console_sem(void) _log_end = log_end; con_start = log_end; /* Flush */ spin_unlock(&logbuf_lock); + stop_critical_timings(); /* don't trace print latency */ call_console_drivers(_con_start, _log_end); + start_critical_timings(); local_irq_restore(flags); } console_locked = 0; diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 892ecc94a82b..896df1cf6adc 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -26,6 +26,24 @@ config FTRACE (the bootup default), then the overhead of the instructions is very small and not measurable even in micro-benchmarks. +config IRQSOFF_TRACER + bool "Interrupts-off Latency Tracer" + default n + depends on TRACE_IRQFLAGS_SUPPORT + depends on GENERIC_TIME + select TRACE_IRQFLAGS + select TRACING + select TRACER_MAX_TRACE + help + This option measures the time spent in irqs-off critical + sections, with microsecond accuracy. + + The default measurement method is a maximum search, which is + disabled by default and can be runtime (re-)started + via: + + echo 0 > /debugfs/tracing/tracing_max_latency + config SCHED_TRACER bool "Scheduling Latency Tracer" depends on DEBUG_KERNEL diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 5508cdb19aea..46be8647fb65 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -3,6 +3,7 @@ obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o obj-$(CONFIG_FTRACE) += trace_functions.o +obj-$(CONFIG_IRQSOFF_TRACER) += trace_irqsoff.o obj-$(CONFIG_SCHED_TRACER) += trace_sched_wakeup.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c new file mode 100644 index 000000000000..a9131b0cf1a5 --- /dev/null +++ b/kernel/trace/trace_irqsoff.c @@ -0,0 +1,402 @@ +/* + * trace irqs off criticall timings + * + * Copyright (C) 2007-2008 Steven Rostedt + * Copyright (C) 2008 Ingo Molnar + * + * From code in the latency_tracer, that is: + * + * Copyright (C) 2004-2006 Ingo Molnar + * Copyright (C) 2004 William Lee Irwin III + */ +#include +#include +#include +#include +#include +#include + +#include "trace.h" + +static struct trace_array *irqsoff_trace __read_mostly; +static int tracer_enabled __read_mostly; + +/* + * Sequence count - we record it when starting a measurement and + * skip the latency if the sequence has changed - some other section + * did a maximum and could disturb our measurement with serial console + * printouts, etc. Truly coinciding maximum latencies should be rare + * and what happens together happens separately as well, so this doesnt + * decrease the validity of the maximum found: + */ +static __cacheline_aligned_in_smp unsigned long max_sequence; + +#ifdef CONFIG_FTRACE +/* + * irqsoff uses its own tracer function to keep the overhead down: + */ +static void notrace +irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = irqsoff_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (likely(!tracer_enabled)) + return; + + local_save_flags(flags); + + if (!irqs_disabled_flags(flags)) + return; + + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + ftrace(tr, data, ip, parent_ip, flags); + + atomic_dec(&data->disabled); +} + +static struct ftrace_ops trace_ops __read_mostly = +{ + .func = irqsoff_tracer_call, +}; +#endif /* CONFIG_FTRACE */ + +/* + * Should this new latency be reported/recorded? + */ +static int notrace report_latency(cycle_t delta) +{ + if (tracing_thresh) { + if (delta < tracing_thresh) + return 0; + } else { + if (delta <= tracing_max_latency) + return 0; + } + return 1; +} + +static void notrace +check_critical_timing(struct trace_array *tr, + struct trace_array_cpu *data, + unsigned long parent_ip, + int cpu) +{ + unsigned long latency, t0, t1; + cycle_t T0, T1, T2, delta; + unsigned long flags; + + /* + * usecs conversion is slow so we try to delay the conversion + * as long as possible: + */ + T0 = data->preempt_timestamp; + T1 = now(cpu); + delta = T1-T0; + + local_save_flags(flags); + + if (!report_latency(delta)) + goto out; + + ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); + /* + * Update the timestamp, because the trace entry above + * might change it (it can only get larger so the latency + * is fair to be reported): + */ + T2 = now(cpu); + + delta = T2-T0; + + latency = nsecs_to_usecs(delta); + + if (data->critical_sequence != max_sequence) + goto out; + + tracing_max_latency = delta; + t0 = nsecs_to_usecs(T0); + t1 = nsecs_to_usecs(T1); + + data->critical_end = parent_ip; + + update_max_tr_single(tr, current, cpu); + + if (tracing_thresh) + printk(KERN_INFO "(%16s-%-5d|#%d): %lu us critical section " + "violates %lu us threshold.\n" + " => started at timestamp %lu: ", + current->comm, current->pid, + raw_smp_processor_id(), + latency, nsecs_to_usecs(tracing_thresh), t0); + else + printk(KERN_INFO "(%16s-%-5d|#%d):" + " new %lu us maximum-latency " + "critical section.\n => started at timestamp %lu: ", + current->comm, current->pid, + raw_smp_processor_id(), + latency, t0); + + print_symbol(KERN_CONT "<%s>\n", data->critical_start); + printk(KERN_CONT " => ended at timestamp %lu: ", t1); + print_symbol(KERN_CONT "<%s>\n", data->critical_end); + dump_stack(); + t1 = nsecs_to_usecs(now(cpu)); + printk(KERN_CONT " => dump-end timestamp %lu\n\n", t1); + + max_sequence++; + +out: + data->critical_sequence = max_sequence; + data->preempt_timestamp = now(cpu); + tracing_reset(data); + ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); +} + +static inline void notrace +start_critical_timing(unsigned long ip, unsigned long parent_ip) +{ + int cpu; + struct trace_array *tr = irqsoff_trace; + struct trace_array_cpu *data; + unsigned long flags; + + if (likely(!tracer_enabled)) + return; + + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + + if (unlikely(!data) || unlikely(!data->trace) || + data->critical_start || atomic_read(&data->disabled)) + return; + + atomic_inc(&data->disabled); + + data->critical_sequence = max_sequence; + data->preempt_timestamp = now(cpu); + data->critical_start = parent_ip; + tracing_reset(data); + + local_save_flags(flags); + ftrace(tr, data, ip, parent_ip, flags); + + atomic_dec(&data->disabled); +} + +static inline void notrace +stop_critical_timing(unsigned long ip, unsigned long parent_ip) +{ + int cpu; + struct trace_array *tr = irqsoff_trace; + struct trace_array_cpu *data; + unsigned long flags; + + if (likely(!tracer_enabled)) + return; + + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + + if (unlikely(!data) || unlikely(!data->trace) || + !data->critical_start || atomic_read(&data->disabled)) + return; + + atomic_inc(&data->disabled); + local_save_flags(flags); + ftrace(tr, data, ip, parent_ip, flags); + check_critical_timing(tr, data, parent_ip, cpu); + data->critical_start = 0; + atomic_dec(&data->disabled); +} + +void notrace start_critical_timings(void) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); +} + +void notrace stop_critical_timings(void) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); +} + +#ifdef CONFIG_PROVE_LOCKING +void notrace time_hardirqs_on(unsigned long a0, unsigned long a1) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + stop_critical_timing(a0, a1); +} + +void notrace time_hardirqs_off(unsigned long a0, unsigned long a1) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + start_critical_timing(a0, a1); +} + +#else /* !CONFIG_PROVE_LOCKING */ + +/* + * Stubs: + */ + +void early_boot_irqs_off(void) +{ +} + +void early_boot_irqs_on(void) +{ +} + +void trace_softirqs_on(unsigned long ip) +{ +} + +void trace_softirqs_off(unsigned long ip) +{ +} + +inline void print_irqtrace_events(struct task_struct *curr) +{ +} + +/* + * We are only interested in hardirq on/off events: + */ +void notrace trace_hardirqs_on(void) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); +} +EXPORT_SYMBOL(trace_hardirqs_on); + +void notrace trace_hardirqs_off(void) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); +} +EXPORT_SYMBOL(trace_hardirqs_off); + +void notrace trace_hardirqs_on_caller(unsigned long caller_addr) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + stop_critical_timing(CALLER_ADDR0, caller_addr); +} +EXPORT_SYMBOL(trace_hardirqs_on_caller); + +void notrace trace_hardirqs_off_caller(unsigned long caller_addr) +{ + unsigned long flags; + + local_save_flags(flags); + + if (irqs_disabled_flags(flags)) + start_critical_timing(CALLER_ADDR0, caller_addr); +} +EXPORT_SYMBOL(trace_hardirqs_off_caller); + +#endif /* CONFIG_PROVE_LOCKING */ + +static void start_irqsoff_tracer(struct trace_array *tr) +{ + tracer_enabled = 1; + register_ftrace_function(&trace_ops); +} + +static void stop_irqsoff_tracer(struct trace_array *tr) +{ + unregister_ftrace_function(&trace_ops); + tracer_enabled = 0; +} + +static void irqsoff_tracer_init(struct trace_array *tr) +{ + irqsoff_trace = tr; + /* make sure that the tracer is visibel */ + smp_wmb(); + + if (tr->ctrl) + start_irqsoff_tracer(tr); +} + +static void irqsoff_tracer_reset(struct trace_array *tr) +{ + if (tr->ctrl) + stop_irqsoff_tracer(tr); +} + +static void irqsoff_tracer_ctrl_update(struct trace_array *tr) +{ + if (tr->ctrl) + start_irqsoff_tracer(tr); + else + stop_irqsoff_tracer(tr); +} + +static void notrace irqsoff_tracer_open(struct trace_iterator *iter) +{ + /* stop the trace while dumping */ + if (iter->tr->ctrl) + stop_irqsoff_tracer(iter->tr); +} + +static void notrace irqsoff_tracer_close(struct trace_iterator *iter) +{ + if (iter->tr->ctrl) + start_irqsoff_tracer(iter->tr); +} + +static struct tracer irqsoff_tracer __read_mostly = +{ + .name = "irqsoff", + .init = irqsoff_tracer_init, + .reset = irqsoff_tracer_reset, + .open = irqsoff_tracer_open, + .close = irqsoff_tracer_close, + .ctrl_update = irqsoff_tracer_ctrl_update, + .print_max = 1, +}; + +__init static int init_irqsoff_tracer(void) +{ + register_tracer(&irqsoff_tracer); + + return 0; +} +device_initcall(init_irqsoff_tracer); -- cgit v1.2.3 From 6cd8a4bb2f97527a9ceb30bc77ea4e959c6a95e3 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: trace preempt off critical timings Add preempt off timings. A lot of kernel core code is taken from the RT patch latency trace that was written by Ingo Molnar. This adds "preemptoff" and "preemptirqsoff" to /debugfs/tracing/available_tracers Now instead of just tracing irqs off, preemption off can be selected to be recorded. When this is selected, it shares the same files as irqs off timings. One can either trace preemption off, irqs off, or one or the other off. By echoing "preemptoff" into /debugfs/tracing/current_tracer, recording of preempt off only is performed. "irqsoff" will only record the time irqs are disabled, but "preemptirqsoff" will take the total time irqs or preemption are disabled. Runtime switching of these options is now supported by simpling echoing in the appropriate trace name into /debugfs/tracing/current_tracer. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 24 +++++- kernel/trace/Kconfig | 25 ++++++ kernel/trace/Makefile | 1 + kernel/trace/trace_irqsoff.c | 184 +++++++++++++++++++++++++++++++------------ 4 files changed, 183 insertions(+), 51 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 73e600852365..328494e28df2 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -70,6 +70,7 @@ #include #include #include +#include #include #include @@ -4365,26 +4366,44 @@ void scheduler_tick(void) #endif } -#if defined(CONFIG_PREEMPT) && defined(CONFIG_DEBUG_PREEMPT) +#if defined(CONFIG_PREEMPT) && (defined(CONFIG_DEBUG_PREEMPT) || \ + defined(CONFIG_PREEMPT_TRACER)) + +static inline unsigned long get_parent_ip(unsigned long addr) +{ + if (in_lock_functions(addr)) { + addr = CALLER_ADDR2; + if (in_lock_functions(addr)) + addr = CALLER_ADDR3; + } + return addr; +} void __kprobes add_preempt_count(int val) { +#ifdef CONFIG_DEBUG_PREEMPT /* * Underflow? */ if (DEBUG_LOCKS_WARN_ON((preempt_count() < 0))) return; +#endif preempt_count() += val; +#ifdef CONFIG_DEBUG_PREEMPT /* * Spinlock count overflowing soon? */ DEBUG_LOCKS_WARN_ON((preempt_count() & PREEMPT_MASK) >= PREEMPT_MASK - 10); +#endif + if (preempt_count() == val) + trace_preempt_off(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1)); } EXPORT_SYMBOL(add_preempt_count); void __kprobes sub_preempt_count(int val) { +#ifdef CONFIG_DEBUG_PREEMPT /* * Underflow? */ @@ -4396,7 +4415,10 @@ void __kprobes sub_preempt_count(int val) if (DEBUG_LOCKS_WARN_ON((val < PREEMPT_MASK) && !(preempt_count() & PREEMPT_MASK))) return; +#endif + if (preempt_count() == val) + trace_preempt_on(CALLER_ADDR0, get_parent_ip(CALLER_ADDR1)); preempt_count() -= val; } EXPORT_SYMBOL(sub_preempt_count); diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 896df1cf6adc..6430016b98e8 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -44,6 +44,31 @@ config IRQSOFF_TRACER echo 0 > /debugfs/tracing/tracing_max_latency + (Note that kernel size and overhead increases with this option + enabled. This option and the preempt-off timing option can be + used together or separately.) + +config PREEMPT_TRACER + bool "Preemption-off Latency Tracer" + default n + depends on GENERIC_TIME + depends on PREEMPT + select TRACING + select TRACER_MAX_TRACE + help + This option measures the time spent in preemption off critical + sections, with microsecond accuracy. + + The default measurement method is a maximum search, which is + disabled by default and can be runtime (re-)started + via: + + echo 0 > /debugfs/tracing/tracing_max_latency + + (Note that kernel size and overhead increases with this option + enabled. This option and the irqs-off timing option can be + used together or separately.) + config SCHED_TRACER bool "Scheduling Latency Tracer" depends on DEBUG_KERNEL diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 46be8647fb65..3fec653d6533 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -4,6 +4,7 @@ obj-$(CONFIG_TRACING) += trace.o obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o obj-$(CONFIG_FTRACE) += trace_functions.o obj-$(CONFIG_IRQSOFF_TRACER) += trace_irqsoff.o +obj-$(CONFIG_PREEMPT_TRACER) += trace_irqsoff.o obj-$(CONFIG_SCHED_TRACER) += trace_sched_wakeup.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index a9131b0cf1a5..8b1231633dc5 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -21,6 +21,36 @@ static struct trace_array *irqsoff_trace __read_mostly; static int tracer_enabled __read_mostly; +static DEFINE_PER_CPU(int, tracing_cpu); + +enum { + TRACER_IRQS_OFF = (1 << 1), + TRACER_PREEMPT_OFF = (1 << 2), +}; + +static int trace_type __read_mostly; + +#ifdef CONFIG_PREEMPT_TRACER +static inline int notrace +preempt_trace(void) +{ + return ((trace_type & TRACER_PREEMPT_OFF) && preempt_count()); +} +#else +# define preempt_trace() (0) +#endif + +#ifdef CONFIG_IRQSOFF_TRACER +static inline int notrace +irq_trace(void) +{ + return ((trace_type & TRACER_IRQS_OFF) && + irqs_disabled()); +} +#else +# define irq_trace() (0) +#endif + /* * Sequence count - we record it when starting a measurement and * skip the latency if the sequence has changed - some other section @@ -44,14 +74,11 @@ irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) long disabled; int cpu; - if (likely(!tracer_enabled)) + if (likely(!__get_cpu_var(tracing_cpu))) return; local_save_flags(flags); - if (!irqs_disabled_flags(flags)) - return; - cpu = raw_smp_processor_id(); data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); @@ -171,23 +198,29 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) if (likely(!tracer_enabled)) return; + if (__get_cpu_var(tracing_cpu)) + return; + cpu = raw_smp_processor_id(); data = tr->data[cpu]; if (unlikely(!data) || unlikely(!data->trace) || - data->critical_start || atomic_read(&data->disabled)) + atomic_read(&data->disabled)) return; atomic_inc(&data->disabled); data->critical_sequence = max_sequence; data->preempt_timestamp = now(cpu); - data->critical_start = parent_ip; + data->critical_start = parent_ip ? : ip; tracing_reset(data); local_save_flags(flags); + ftrace(tr, data, ip, parent_ip, flags); + __get_cpu_var(tracing_cpu) = 1; + atomic_dec(&data->disabled); } @@ -199,7 +232,13 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) struct trace_array_cpu *data; unsigned long flags; - if (likely(!tracer_enabled)) + /* Always clear the tracing cpu on stopping the trace */ + if (unlikely(__get_cpu_var(tracing_cpu))) + __get_cpu_var(tracing_cpu) = 0; + else + return; + + if (!tracer_enabled) return; cpu = raw_smp_processor_id(); @@ -212,49 +251,35 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) atomic_inc(&data->disabled); local_save_flags(flags); ftrace(tr, data, ip, parent_ip, flags); - check_critical_timing(tr, data, parent_ip, cpu); + check_critical_timing(tr, data, parent_ip ? : ip, cpu); data->critical_start = 0; atomic_dec(&data->disabled); } +/* start and stop critical timings used to for stoppage (in idle) */ void notrace start_critical_timings(void) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (preempt_trace() || irq_trace()) start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } void notrace stop_critical_timings(void) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (preempt_trace() || irq_trace()) stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } +#ifdef CONFIG_IRQSOFF_TRACER #ifdef CONFIG_PROVE_LOCKING void notrace time_hardirqs_on(unsigned long a0, unsigned long a1) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) stop_critical_timing(a0, a1); } void notrace time_hardirqs_off(unsigned long a0, unsigned long a1) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) start_critical_timing(a0, a1); } @@ -289,49 +314,46 @@ inline void print_irqtrace_events(struct task_struct *curr) */ void notrace trace_hardirqs_on(void) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } EXPORT_SYMBOL(trace_hardirqs_on); void notrace trace_hardirqs_off(void) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } EXPORT_SYMBOL(trace_hardirqs_off); void notrace trace_hardirqs_on_caller(unsigned long caller_addr) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) stop_critical_timing(CALLER_ADDR0, caller_addr); } EXPORT_SYMBOL(trace_hardirqs_on_caller); void notrace trace_hardirqs_off_caller(unsigned long caller_addr) { - unsigned long flags; - - local_save_flags(flags); - - if (irqs_disabled_flags(flags)) + if (!preempt_trace() && irq_trace()) start_critical_timing(CALLER_ADDR0, caller_addr); } EXPORT_SYMBOL(trace_hardirqs_off_caller); #endif /* CONFIG_PROVE_LOCKING */ +#endif /* CONFIG_IRQSOFF_TRACER */ + +#ifdef CONFIG_PREEMPT_TRACER +void notrace trace_preempt_on(unsigned long a0, unsigned long a1) +{ + stop_critical_timing(a0, a1); +} + +void notrace trace_preempt_off(unsigned long a0, unsigned long a1) +{ + start_critical_timing(a0, a1); +} +#endif /* CONFIG_PREEMPT_TRACER */ static void start_irqsoff_tracer(struct trace_array *tr) { @@ -345,7 +367,7 @@ static void stop_irqsoff_tracer(struct trace_array *tr) tracer_enabled = 0; } -static void irqsoff_tracer_init(struct trace_array *tr) +static void __irqsoff_tracer_init(struct trace_array *tr) { irqsoff_trace = tr; /* make sure that the tracer is visibel */ @@ -382,6 +404,13 @@ static void notrace irqsoff_tracer_close(struct trace_iterator *iter) start_irqsoff_tracer(iter->tr); } +#ifdef CONFIG_IRQSOFF_TRACER +static void irqsoff_tracer_init(struct trace_array *tr) +{ + trace_type = TRACER_IRQS_OFF; + + __irqsoff_tracer_init(tr); +} static struct tracer irqsoff_tracer __read_mostly = { .name = "irqsoff", @@ -392,10 +421,65 @@ static struct tracer irqsoff_tracer __read_mostly = .ctrl_update = irqsoff_tracer_ctrl_update, .print_max = 1, }; +# define register_irqsoff(trace) register_tracer(&trace) +#else +# define register_irqsoff(trace) do { } while (0) +#endif + +#ifdef CONFIG_PREEMPT_TRACER +static void preemptoff_tracer_init(struct trace_array *tr) +{ + trace_type = TRACER_PREEMPT_OFF; + + __irqsoff_tracer_init(tr); +} + +static struct tracer preemptoff_tracer __read_mostly = +{ + .name = "preemptoff", + .init = preemptoff_tracer_init, + .reset = irqsoff_tracer_reset, + .open = irqsoff_tracer_open, + .close = irqsoff_tracer_close, + .ctrl_update = irqsoff_tracer_ctrl_update, + .print_max = 1, +}; +# define register_preemptoff(trace) register_tracer(&trace) +#else +# define register_preemptoff(trace) do { } while (0) +#endif + +#if defined(CONFIG_IRQSOFF_TRACER) && \ + defined(CONFIG_PREEMPT_TRACER) + +static void preemptirqsoff_tracer_init(struct trace_array *tr) +{ + trace_type = TRACER_IRQS_OFF | TRACER_PREEMPT_OFF; + + __irqsoff_tracer_init(tr); +} + +static struct tracer preemptirqsoff_tracer __read_mostly = +{ + .name = "preemptirqsoff", + .init = preemptirqsoff_tracer_init, + .reset = irqsoff_tracer_reset, + .open = irqsoff_tracer_open, + .close = irqsoff_tracer_close, + .ctrl_update = irqsoff_tracer_ctrl_update, + .print_max = 1, +}; + +# define register_preemptirqsoff(trace) register_tracer(&trace) +#else +# define register_preemptirqsoff(trace) do { } while (0) +#endif __init static int init_irqsoff_tracer(void) { - register_tracer(&irqsoff_tracer); + register_irqsoff(irqsoff_tracer); + register_preemptoff(preemptoff_tracer); + register_preemptirqsoff(preemptirqsoff_tracer); return 0; } -- cgit v1.2.3 From 3d0833953e1b98b79ddf491dd49229eef9baeac1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:42 +0200 Subject: ftrace: dynamic enabling/disabling of function calls This patch adds a feature to dynamically replace the ftrace code with the jmps to allow a kernel with ftrace configured to run as fast as it can without it configured. The way this works, is on bootup (if ftrace is enabled), a ftrace function is registered to record the instruction pointer of all places that call the function. Later, if there's still any code to patch, a kthread is awoken (rate limited to at most once a second) that performs a stop_machine, and replaces all the code that was called with a jmp over the call to ftrace. It only replaces what was found the previous time. Typically the system reaches equilibrium quickly after bootup and there's no code patching needed at all. e.g. call ftrace /* 5 bytes */ is replaced with jmp 3f /* jmp is 2 bytes and we jump 3 forward */ 3: When we want to enable ftrace for function tracing, the IP recording is removed, and stop_machine is called again to replace all the locations of that were recorded back to the call of ftrace. When it is disabled, we replace the code back to the jmp. Allocation is done by the kthread. If the ftrace recording function is called, and we don't have any record slots available, then we simply skip that call. Once a second a new page (if needed) is allocated for recording new ftrace function calls. A large batch is allocated at boot up to get most of the calls there. Because we do this via stop_machine, we don't have to worry about another CPU executing a ftrace call as we modify it. But we do need to worry about NMI's so all functions that might be called via nmi must be annotated with notrace_nmi. When this code is configured in, the NMI code will not call notrace. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 17 +++ kernel/trace/ftrace.c | 356 +++++++++++++++++++++++++++++++++++++++++++++----- 2 files changed, 341 insertions(+), 32 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 6430016b98e8..cad9db1dee02 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -88,3 +88,20 @@ config CONTEXT_SWITCH_TRACER This tracer gets called from the context switch and records all switching of tasks. +config DYNAMIC_FTRACE + bool "enable/disable ftrace tracepoints dynamically" + depends on FTRACE + default y + help + This option will modify all the calls to ftrace dynamically + (will patch them out of the binary image and replaces them + with a No-Op instruction) as they are called. A table is + created to dynamically enable them again. + + This way a CONFIG_FTRACE kernel is slightly larger, but otherwise + has native performance as long as no tracing is active. + + The changes to the code are done by a kernel thread that + wakes up once a second and checks to see if any ftrace calls + were made. If so, it runs stop_machine (stops all CPUS) + and modifies the code to jump over the call to ftrace. diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index b6a80b98a3fb..d1ae2ba25274 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -13,10 +13,19 @@ * Copyright (C) 2004 William Lee Irwin III */ -#include +#include +#include +#include +#include +#include #include +#include +#include +#include + +#include "trace.h" -static DEFINE_SPINLOCK(ftrace_func_lock); +static DEFINE_SPINLOCK(ftrace_lock); static struct ftrace_ops ftrace_list_end __read_mostly = { .func = ftrace_stub, @@ -44,21 +53,21 @@ notrace void ftrace_list_func(unsigned long ip, unsigned long parent_ip) } /** - * register_ftrace_function - register a function for profiling - * @ops - ops structure that holds the function for profiling. - * - * Register a function to be called by all functions in the - * kernel. + * clear_ftrace_function - reset the ftrace function * - * Note: @ops->func and all the functions it calls must be labeled - * with "notrace", otherwise it will go into a - * recursive loop. + * This NULLs the ftrace function and in essence stops + * tracing. There may be lag */ -int register_ftrace_function(struct ftrace_ops *ops) +void clear_ftrace_function(void) { - unsigned long flags; + ftrace_trace_function = ftrace_stub; +} + +static int notrace __register_ftrace_function(struct ftrace_ops *ops) +{ + /* Should never be called by interrupts */ + spin_lock(&ftrace_lock); - spin_lock_irqsave(&ftrace_func_lock, flags); ops->next = ftrace_list; /* * We are entering ops into the ftrace_list but another @@ -68,6 +77,7 @@ int register_ftrace_function(struct ftrace_ops *ops) */ smp_wmb(); ftrace_list = ops; + /* * For one func, simply call it directly. * For more than one func, call the chain. @@ -76,28 +86,22 @@ int register_ftrace_function(struct ftrace_ops *ops) ftrace_trace_function = ops->func; else ftrace_trace_function = ftrace_list_func; - spin_unlock_irqrestore(&ftrace_func_lock, flags); + + spin_unlock(&ftrace_lock); return 0; } -/** - * unregister_ftrace_function - unresgister a function for profiling. - * @ops - ops structure that holds the function to unregister - * - * Unregister a function that was added to be called by ftrace profiling. - */ -int unregister_ftrace_function(struct ftrace_ops *ops) +static int notrace __unregister_ftrace_function(struct ftrace_ops *ops) { - unsigned long flags; struct ftrace_ops **p; int ret = 0; - spin_lock_irqsave(&ftrace_func_lock, flags); + spin_lock(&ftrace_lock); /* - * If we are the only function, then the ftrace pointer is - * pointing directly to that function. + * If we are removing the last function, then simply point + * to the ftrace_stub. */ if (ftrace_list == ops && ops->next == &ftrace_list_end) { ftrace_trace_function = ftrace_stub; @@ -117,22 +121,310 @@ int unregister_ftrace_function(struct ftrace_ops *ops) *p = (*p)->next; /* If we only have one func left, then call that directly */ - if (ftrace_list->next == &ftrace_list_end) + if (ftrace_list == &ftrace_list_end || + ftrace_list->next == &ftrace_list_end) ftrace_trace_function = ftrace_list->func; out: - spin_unlock_irqrestore(&ftrace_func_lock, flags); + spin_unlock(&ftrace_lock); + + return ret; +} + +#ifdef CONFIG_DYNAMIC_FTRACE + +static struct hlist_head ftrace_hash[FTRACE_HASHSIZE]; + +static DEFINE_PER_CPU(int, ftrace_shutdown_disable_cpu); + +static DEFINE_SPINLOCK(ftrace_shutdown_lock); +static DEFINE_MUTEX(ftraced_lock); + +static int ftraced_trigger; +static int ftraced_suspend; + +static int ftrace_record_suspend; + +static inline int +notrace ftrace_ip_in_hash(unsigned long ip, unsigned long key) +{ + struct dyn_ftrace *p; + struct hlist_node *t; + int found = 0; + + hlist_for_each_entry(p, t, &ftrace_hash[key], node) { + if (p->ip == ip) { + found = 1; + break; + } + } + + return found; +} + +static inline void notrace +ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) +{ + hlist_add_head(&node->node, &ftrace_hash[key]); +} + +static void notrace +ftrace_record_ip(unsigned long ip, unsigned long parent_ip) +{ + struct dyn_ftrace *node; + unsigned long flags; + unsigned long key; + int resched; + int atomic; + + resched = need_resched(); + preempt_disable_notrace(); + + /* We simply need to protect against recursion */ + __get_cpu_var(ftrace_shutdown_disable_cpu)++; + if (__get_cpu_var(ftrace_shutdown_disable_cpu) != 1) + goto out; + + if (unlikely(ftrace_record_suspend)) + goto out; + + key = hash_long(ip, FTRACE_HASHBITS); + + WARN_ON_ONCE(key >= FTRACE_HASHSIZE); + + if (ftrace_ip_in_hash(ip, key)) + goto out; + + atomic = irqs_disabled(); + + spin_lock_irqsave(&ftrace_shutdown_lock, flags); + + /* This ip may have hit the hash before the lock */ + if (ftrace_ip_in_hash(ip, key)) + goto out_unlock; + + /* + * There's a slight race that the ftraced will update the + * hash and reset here. The arch alloc is responsible + * for seeing if the IP has already changed, and if + * it has, the alloc will fail. + */ + node = ftrace_alloc_shutdown_node(ip); + if (!node) + goto out_unlock; + + node->ip = ip; + + ftrace_add_hash(node, key); + + ftraced_trigger = 1; + + out_unlock: + spin_unlock_irqrestore(&ftrace_shutdown_lock, flags); + out: + __get_cpu_var(ftrace_shutdown_disable_cpu)--; + + /* prevent recursion with scheduler */ + if (resched) + preempt_enable_no_resched_notrace(); + else + preempt_enable_notrace(); +} + +static struct ftrace_ops ftrace_shutdown_ops __read_mostly = +{ + .func = ftrace_record_ip, +}; + + +static int notrace __ftrace_modify_code(void *data) +{ + void (*func)(void) = data; + + func(); + return 0; +} + +static void notrace ftrace_run_startup_code(void) +{ + stop_machine_run(__ftrace_modify_code, ftrace_startup_code, NR_CPUS); +} + +static void notrace ftrace_run_shutdown_code(void) +{ + stop_machine_run(__ftrace_modify_code, ftrace_shutdown_code, NR_CPUS); +} + +static void notrace ftrace_startup(void) +{ + mutex_lock(&ftraced_lock); + ftraced_suspend++; + if (ftraced_suspend != 1) + goto out; + __unregister_ftrace_function(&ftrace_shutdown_ops); + + ftrace_run_startup_code(); + out: + mutex_unlock(&ftraced_lock); +} + +static void notrace ftrace_shutdown(void) +{ + mutex_lock(&ftraced_lock); + ftraced_suspend--; + if (ftraced_suspend) + goto out; + + ftrace_run_shutdown_code(); + + __register_ftrace_function(&ftrace_shutdown_ops); + out: + mutex_unlock(&ftraced_lock); +} + +static cycle_t ftrace_update_time; +static unsigned long ftrace_update_cnt; +unsigned long ftrace_update_tot_cnt; + +static int notrace __ftrace_update_code(void *ignore) +{ + struct dyn_ftrace *p; + struct hlist_head head; + struct hlist_node *t; + cycle_t start, stop; + int i; + + /* Don't be calling ftrace ops now */ + __unregister_ftrace_function(&ftrace_shutdown_ops); + + start = now(raw_smp_processor_id()); + ftrace_update_cnt = 0; + + /* No locks needed, the machine is stopped! */ + for (i = 0; i < FTRACE_HASHSIZE; i++) { + if (hlist_empty(&ftrace_hash[i])) + continue; + + head = ftrace_hash[i]; + INIT_HLIST_HEAD(&ftrace_hash[i]); + + /* all CPUS are stopped, we are safe to modify code */ + hlist_for_each_entry(p, t, &head, node) { + ftrace_code_disable(p); + ftrace_update_cnt++; + } + + } + + stop = now(raw_smp_processor_id()); + ftrace_update_time = stop - start; + ftrace_update_tot_cnt += ftrace_update_cnt; + + __register_ftrace_function(&ftrace_shutdown_ops); return 0; } +static void notrace ftrace_update_code(void) +{ + stop_machine_run(__ftrace_update_code, NULL, NR_CPUS); +} + +static int notrace ftraced(void *ignore) +{ + unsigned long usecs; + + set_current_state(TASK_INTERRUPTIBLE); + + while (!kthread_should_stop()) { + + /* check once a second */ + schedule_timeout(HZ); + + mutex_lock(&ftraced_lock); + if (ftraced_trigger && !ftraced_suspend) { + ftrace_record_suspend++; + ftrace_update_code(); + usecs = nsecs_to_usecs(ftrace_update_time); + if (ftrace_update_tot_cnt > 100000) { + ftrace_update_tot_cnt = 0; + pr_info("hm, dftrace overflow: %lu change%s" + " (%lu total) in %lu usec%s\n", + ftrace_update_cnt, + ftrace_update_cnt != 1 ? "s" : "", + ftrace_update_tot_cnt, + usecs, usecs != 1 ? "s" : ""); + WARN_ON_ONCE(1); + } + ftraced_trigger = 0; + ftrace_record_suspend--; + } + mutex_unlock(&ftraced_lock); + + ftrace_shutdown_replenish(); + + set_current_state(TASK_INTERRUPTIBLE); + } + __set_current_state(TASK_RUNNING); + return 0; +} + +static int __init notrace ftrace_shutdown_init(void) +{ + struct task_struct *p; + int ret; + + ret = ftrace_shutdown_arch_init(); + if (ret) + return ret; + + p = kthread_run(ftraced, NULL, "ftraced"); + if (IS_ERR(p)) + return -1; + + __register_ftrace_function(&ftrace_shutdown_ops); + + return 0; +} + +core_initcall(ftrace_shutdown_init); +#else +# define ftrace_startup() do { } while (0) +# define ftrace_shutdown() do { } while (0) +#endif /* CONFIG_DYNAMIC_FTRACE */ + /** - * clear_ftrace_function - reset the ftrace function + * register_ftrace_function - register a function for profiling + * @ops - ops structure that holds the function for profiling. * - * This NULLs the ftrace function and in essence stops - * tracing. There may be lag + * Register a function to be called by all functions in the + * kernel. + * + * Note: @ops->func and all the functions it calls must be labeled + * with "notrace", otherwise it will go into a + * recursive loop. */ -void clear_ftrace_function(void) +int register_ftrace_function(struct ftrace_ops *ops) { - ftrace_trace_function = ftrace_stub; + ftrace_startup(); + + return __register_ftrace_function(ops); +} + +/** + * unregister_ftrace_function - unresgister a function for profiling. + * @ops - ops structure that holds the function to unregister + * + * Unregister a function that was added to be called by ftrace profiling. + */ +int unregister_ftrace_function(struct ftrace_ops *ops) +{ + int ret; + + ret = __unregister_ftrace_function(ops); + + if (ftrace_list == &ftrace_list_end) + ftrace_shutdown(); + + return ret; } -- cgit v1.2.3 From b0fc494fae96a7089f3651cb451f461c7291244c Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:43 +0200 Subject: ftrace: add ftrace_enabled sysctl to disable mcount function This patch adds back the sysctl ftrace_enabled. This time it is defaulted to on, if DYNAMIC_FTRACE is configured. When ftrace_enabled is disabled, the ftrace function is set to the stub return. If DYNAMIC_FTRACE is also configured, on ftrace_enabled = 0, the registered ftrace functions will all be set to jmps, but no more new calls to ftrace recording (used to find the ftrace calling sites) will be called. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sysctl.c | 11 +++++ kernel/trace/ftrace.c | 125 ++++++++++++++++++++++++++++++++++++++++++-------- 2 files changed, 118 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 29116652dca8..efaf7c5500e9 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -46,6 +46,7 @@ #include #include #include +#include #include #include @@ -455,6 +456,16 @@ static struct ctl_table kern_table[] = { .mode = 0644, .proc_handler = &proc_dointvec, }, +#ifdef CONFIG_FTRACE + { + .ctl_name = CTL_UNNUMBERED, + .procname = "ftrace_enabled", + .data = &ftrace_enabled, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = &ftrace_enable_sysctl, + }, +#endif #ifdef CONFIG_KMOD { .ctl_name = KERN_MODPROBE, diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index d1ae2ba25274..d3de37299ba4 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -20,12 +20,24 @@ #include #include #include +#include #include #include #include "trace.h" +#ifdef CONFIG_DYNAMIC_FTRACE +# define FTRACE_ENABLED_INIT 1 +#else +# define FTRACE_ENABLED_INIT 0 +#endif + +int ftrace_enabled = FTRACE_ENABLED_INIT; +static int last_ftrace_enabled = FTRACE_ENABLED_INIT; + static DEFINE_SPINLOCK(ftrace_lock); +static DEFINE_MUTEX(ftrace_sysctl_lock); + static struct ftrace_ops ftrace_list_end __read_mostly = { .func = ftrace_stub, @@ -78,14 +90,16 @@ static int notrace __register_ftrace_function(struct ftrace_ops *ops) smp_wmb(); ftrace_list = ops; - /* - * For one func, simply call it directly. - * For more than one func, call the chain. - */ - if (ops->next == &ftrace_list_end) - ftrace_trace_function = ops->func; - else - ftrace_trace_function = ftrace_list_func; + if (ftrace_enabled) { + /* + * For one func, simply call it directly. + * For more than one func, call the chain. + */ + if (ops->next == &ftrace_list_end) + ftrace_trace_function = ops->func; + else + ftrace_trace_function = ftrace_list_func; + } spin_unlock(&ftrace_lock); @@ -120,10 +134,12 @@ static int notrace __unregister_ftrace_function(struct ftrace_ops *ops) *p = (*p)->next; - /* If we only have one func left, then call that directly */ - if (ftrace_list == &ftrace_list_end || - ftrace_list->next == &ftrace_list_end) - ftrace_trace_function = ftrace_list->func; + if (ftrace_enabled) { + /* If we only have one func left, then call that directly */ + if (ftrace_list == &ftrace_list_end || + ftrace_list->next == &ftrace_list_end) + ftrace_trace_function = ftrace_list->func; + } out: spin_unlock(&ftrace_lock); @@ -263,7 +279,8 @@ static void notrace ftrace_startup(void) goto out; __unregister_ftrace_function(&ftrace_shutdown_ops); - ftrace_run_startup_code(); + if (ftrace_enabled) + ftrace_run_startup_code(); out: mutex_unlock(&ftraced_lock); } @@ -275,13 +292,32 @@ static void notrace ftrace_shutdown(void) if (ftraced_suspend) goto out; - ftrace_run_shutdown_code(); + if (ftrace_enabled) + ftrace_run_shutdown_code(); __register_ftrace_function(&ftrace_shutdown_ops); out: mutex_unlock(&ftraced_lock); } +static void notrace ftrace_startup_sysctl(void) +{ + mutex_lock(&ftraced_lock); + /* ftraced_suspend is true if we want ftrace running */ + if (ftraced_suspend) + ftrace_run_startup_code(); + mutex_unlock(&ftraced_lock); +} + +static void notrace ftrace_shutdown_sysctl(void) +{ + mutex_lock(&ftraced_lock); + /* ftraced_suspend is true if ftrace is running */ + if (ftraced_suspend) + ftrace_run_shutdown_code(); + mutex_unlock(&ftraced_lock); +} + static cycle_t ftrace_update_time; static unsigned long ftrace_update_cnt; unsigned long ftrace_update_tot_cnt; @@ -341,8 +377,9 @@ static int notrace ftraced(void *ignore) /* check once a second */ schedule_timeout(HZ); + mutex_lock(&ftrace_sysctl_lock); mutex_lock(&ftraced_lock); - if (ftraced_trigger && !ftraced_suspend) { + if (ftrace_enabled && ftraced_trigger && !ftraced_suspend) { ftrace_record_suspend++; ftrace_update_code(); usecs = nsecs_to_usecs(ftrace_update_time); @@ -360,6 +397,7 @@ static int notrace ftraced(void *ignore) ftrace_record_suspend--; } mutex_unlock(&ftraced_lock); + mutex_unlock(&ftrace_sysctl_lock); ftrace_shutdown_replenish(); @@ -389,8 +427,10 @@ static int __init notrace ftrace_shutdown_init(void) core_initcall(ftrace_shutdown_init); #else -# define ftrace_startup() do { } while (0) -# define ftrace_shutdown() do { } while (0) +# define ftrace_startup() do { } while (0) +# define ftrace_shutdown() do { } while (0) +# define ftrace_startup_sysctl() do { } while (0) +# define ftrace_shutdown_sysctl() do { } while (0) #endif /* CONFIG_DYNAMIC_FTRACE */ /** @@ -406,9 +446,15 @@ core_initcall(ftrace_shutdown_init); */ int register_ftrace_function(struct ftrace_ops *ops) { + int ret; + + mutex_lock(&ftrace_sysctl_lock); ftrace_startup(); - return __register_ftrace_function(ops); + ret = __register_ftrace_function(ops); + mutex_unlock(&ftrace_sysctl_lock); + + return ret; } /** @@ -421,10 +467,53 @@ int unregister_ftrace_function(struct ftrace_ops *ops) { int ret; + mutex_lock(&ftrace_sysctl_lock); ret = __unregister_ftrace_function(ops); if (ftrace_list == &ftrace_list_end) ftrace_shutdown(); + mutex_unlock(&ftrace_sysctl_lock); + + return ret; +} + +notrace int +ftrace_enable_sysctl(struct ctl_table *table, int write, + struct file *filp, void __user *buffer, size_t *lenp, + loff_t *ppos) +{ + int ret; + + mutex_lock(&ftrace_sysctl_lock); + + ret = proc_dointvec(table, write, filp, buffer, lenp, ppos); + + if (ret || !write || (last_ftrace_enabled == ftrace_enabled)) + goto out; + + last_ftrace_enabled = ftrace_enabled; + + if (ftrace_enabled) { + + ftrace_startup_sysctl(); + + /* we are starting ftrace again */ + if (ftrace_list != &ftrace_list_end) { + if (ftrace_list->next == &ftrace_list_end) + ftrace_trace_function = ftrace_list->func; + else + ftrace_trace_function = ftrace_list_func; + } + + } else { + /* stopping ftrace calls (just send to ftrace_stub) */ + ftrace_trace_function = ftrace_stub; + + ftrace_shutdown_sysctl(); + } + + out: + mutex_unlock(&ftrace_sysctl_lock); return ret; } -- cgit v1.2.3 From 3c1720f00bb619302ba19d55986ab565e74d06db Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:43 +0200 Subject: ftrace: move memory management out of arch code This patch moves the memory management of the ftrace records out of the arch code and into the generic code making the arch code simpler. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 154 +++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 152 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index d3de37299ba4..f6d9af3bf66b 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -156,6 +156,21 @@ static DEFINE_PER_CPU(int, ftrace_shutdown_disable_cpu); static DEFINE_SPINLOCK(ftrace_shutdown_lock); static DEFINE_MUTEX(ftraced_lock); +struct ftrace_page { + struct ftrace_page *next; + int index; + struct dyn_ftrace records[]; +} __attribute__((packed)); + +#define ENTRIES_PER_PAGE \ + ((PAGE_SIZE - sizeof(struct ftrace_page)) / sizeof(struct dyn_ftrace)) + +/* estimate from running different kernels */ +#define NR_TO_INIT 10000 + +static struct ftrace_page *ftrace_pages_start; +static struct ftrace_page *ftrace_pages; + static int ftraced_trigger; static int ftraced_suspend; @@ -184,6 +199,21 @@ ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) hlist_add_head(&node->node, &ftrace_hash[key]); } +static notrace struct dyn_ftrace *ftrace_alloc_shutdown_node(unsigned long ip) +{ + /* If this was already converted, skip it */ + if (ftrace_ip_converted(ip)) + return NULL; + + if (ftrace_pages->index == ENTRIES_PER_PAGE) { + if (!ftrace_pages->next) + return NULL; + ftrace_pages = ftrace_pages->next; + } + + return &ftrace_pages->records[ftrace_pages->index++]; +} + static void notrace ftrace_record_ip(unsigned long ip, unsigned long parent_ip) { @@ -252,6 +282,62 @@ static struct ftrace_ops ftrace_shutdown_ops __read_mostly = .func = ftrace_record_ip, }; +#define MCOUNT_ADDR ((long)(&mcount)) + +static void notrace ftrace_replace_code(int saved) +{ + unsigned char *new = NULL, *old = NULL; + struct dyn_ftrace *rec; + struct ftrace_page *pg; + unsigned long ip; + int failed; + int i; + + if (saved) + old = ftrace_nop_replace(); + else + new = ftrace_nop_replace(); + + for (pg = ftrace_pages_start; pg; pg = pg->next) { + for (i = 0; i < pg->index; i++) { + rec = &pg->records[i]; + + /* don't modify code that has already faulted */ + if (rec->flags & FTRACE_FL_FAILED) + continue; + + ip = rec->ip; + + if (saved) + new = ftrace_call_replace(ip, MCOUNT_ADDR); + else + old = ftrace_call_replace(ip, MCOUNT_ADDR); + + failed = ftrace_modify_code(ip, old, new); + if (failed) + rec->flags |= FTRACE_FL_FAILED; + } + } +} + +static notrace void ftrace_startup_code(void) +{ + ftrace_replace_code(1); +} + +static notrace void ftrace_shutdown_code(void) +{ + ftrace_replace_code(0); +} + +static notrace void ftrace_shutdown_replenish(void) +{ + if (ftrace_pages->next) + return; + + /* allocate another page */ + ftrace_pages->next = (void *)get_zeroed_page(GFP_KERNEL); +} static int notrace __ftrace_modify_code(void *data) { @@ -261,6 +347,23 @@ static int notrace __ftrace_modify_code(void *data) return 0; } +static notrace void +ftrace_code_disable(struct dyn_ftrace *rec, unsigned long addr) +{ + unsigned long ip; + unsigned char *nop, *call; + int failed; + + ip = rec->ip; + + nop = ftrace_nop_replace(); + call = ftrace_call_replace(ip, addr); + + failed = ftrace_modify_code(ip, call, nop); + if (failed) + rec->flags |= FTRACE_FL_FAILED; +} + static void notrace ftrace_run_startup_code(void) { stop_machine_run(__ftrace_modify_code, ftrace_startup_code, NR_CPUS); @@ -346,7 +449,7 @@ static int notrace __ftrace_update_code(void *ignore) /* all CPUS are stopped, we are safe to modify code */ hlist_for_each_entry(p, t, &head, node) { - ftrace_code_disable(p); + ftrace_code_disable(p, MCOUNT_ADDR); ftrace_update_cnt++; } @@ -407,12 +510,59 @@ static int notrace ftraced(void *ignore) return 0; } +static int __init ftrace_dyn_table_alloc(void) +{ + struct ftrace_page *pg; + int cnt; + int i; + int ret; + + ret = ftrace_dyn_arch_init(); + if (ret) + return ret; + + /* allocate a few pages */ + ftrace_pages_start = (void *)get_zeroed_page(GFP_KERNEL); + if (!ftrace_pages_start) + return -1; + + /* + * Allocate a few more pages. + * + * TODO: have some parser search vmlinux before + * final linking to find all calls to ftrace. + * Then we can: + * a) know how many pages to allocate. + * and/or + * b) set up the table then. + * + * The dynamic code is still necessary for + * modules. + */ + + pg = ftrace_pages = ftrace_pages_start; + + cnt = NR_TO_INIT / ENTRIES_PER_PAGE; + + for (i = 0; i < cnt; i++) { + pg->next = (void *)get_zeroed_page(GFP_KERNEL); + + /* If we fail, we'll try later anyway */ + if (!pg->next) + break; + + pg = pg->next; + } + + return 0; +} + static int __init notrace ftrace_shutdown_init(void) { struct task_struct *p; int ret; - ret = ftrace_shutdown_arch_init(); + ret = ftrace_dyn_table_alloc(); if (ret) return ret; -- cgit v1.2.3 From d61f82d06672f57fca410da6f7fffd15867db622 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:43 +0200 Subject: ftrace: use dynamic patching for updating mcount calls This patch replaces the indirect call to the mcount function pointer with a direct call that will be patched by the dynamic ftrace routines. On boot up, the mcount function calls the ftace_stub function. When the dynamic ftrace code is initialized, the ftrace_stub is replaced with a call to the ftrace_record_ip, which records the instruction pointers of the locations that call it. Later, the ftraced daemon will call kstop_machine and patch all the locations to nops. When a ftrace is enabled, the original calls to mcount will now be set top call ftrace_caller, which will do a direct call to the registered ftrace function. This direct call is also patched when the function that should be called is updated. All patching is performed by a kstop_machine routine to prevent any type of race conditions that is associated with modifying code on the fly. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 183 +++++++++++++++++++++++++++++--------------------- 1 file changed, 105 insertions(+), 78 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index f6d9af3bf66b..88544f9bc0ed 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -26,14 +26,8 @@ #include "trace.h" -#ifdef CONFIG_DYNAMIC_FTRACE -# define FTRACE_ENABLED_INIT 1 -#else -# define FTRACE_ENABLED_INIT 0 -#endif - -int ftrace_enabled = FTRACE_ENABLED_INIT; -static int last_ftrace_enabled = FTRACE_ENABLED_INIT; +int ftrace_enabled; +static int last_ftrace_enabled; static DEFINE_SPINLOCK(ftrace_lock); static DEFINE_MUTEX(ftrace_sysctl_lock); @@ -149,6 +143,14 @@ static int notrace __unregister_ftrace_function(struct ftrace_ops *ops) #ifdef CONFIG_DYNAMIC_FTRACE +enum { + FTRACE_ENABLE_CALLS = (1 << 0), + FTRACE_DISABLE_CALLS = (1 << 1), + FTRACE_UPDATE_TRACE_FUNC = (1 << 2), + FTRACE_ENABLE_MCOUNT = (1 << 3), + FTRACE_DISABLE_MCOUNT = (1 << 4), +}; + static struct hlist_head ftrace_hash[FTRACE_HASHSIZE]; static DEFINE_PER_CPU(int, ftrace_shutdown_disable_cpu); @@ -199,12 +201,8 @@ ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) hlist_add_head(&node->node, &ftrace_hash[key]); } -static notrace struct dyn_ftrace *ftrace_alloc_shutdown_node(unsigned long ip) +static notrace struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) { - /* If this was already converted, skip it */ - if (ftrace_ip_converted(ip)) - return NULL; - if (ftrace_pages->index == ENTRIES_PER_PAGE) { if (!ftrace_pages->next) return NULL; @@ -215,7 +213,7 @@ static notrace struct dyn_ftrace *ftrace_alloc_shutdown_node(unsigned long ip) } static void notrace -ftrace_record_ip(unsigned long ip, unsigned long parent_ip) +ftrace_record_ip(unsigned long ip) { struct dyn_ftrace *node; unsigned long flags; @@ -223,6 +221,9 @@ ftrace_record_ip(unsigned long ip, unsigned long parent_ip) int resched; int atomic; + if (!ftrace_enabled) + return; + resched = need_resched(); preempt_disable_notrace(); @@ -251,11 +252,12 @@ ftrace_record_ip(unsigned long ip, unsigned long parent_ip) /* * There's a slight race that the ftraced will update the - * hash and reset here. The arch alloc is responsible - * for seeing if the IP has already changed, and if - * it has, the alloc will fail. + * hash and reset here. If it is already converted, skip it. */ - node = ftrace_alloc_shutdown_node(ip); + if (ftrace_ip_converted(ip)) + goto out_unlock; + + node = ftrace_alloc_dyn_node(ip); if (!node) goto out_unlock; @@ -277,11 +279,7 @@ ftrace_record_ip(unsigned long ip, unsigned long parent_ip) preempt_enable_notrace(); } -static struct ftrace_ops ftrace_shutdown_ops __read_mostly = -{ - .func = ftrace_record_ip, -}; - +#define FTRACE_ADDR ((long)(&ftrace_caller)) #define MCOUNT_ADDR ((long)(&mcount)) static void notrace ftrace_replace_code(int saved) @@ -309,9 +307,9 @@ static void notrace ftrace_replace_code(int saved) ip = rec->ip; if (saved) - new = ftrace_call_replace(ip, MCOUNT_ADDR); + new = ftrace_call_replace(ip, FTRACE_ADDR); else - old = ftrace_call_replace(ip, MCOUNT_ADDR); + old = ftrace_call_replace(ip, FTRACE_ADDR); failed = ftrace_modify_code(ip, old, new); if (failed) @@ -320,16 +318,6 @@ static void notrace ftrace_replace_code(int saved) } } -static notrace void ftrace_startup_code(void) -{ - ftrace_replace_code(1); -} - -static notrace void ftrace_shutdown_code(void) -{ - ftrace_replace_code(0); -} - static notrace void ftrace_shutdown_replenish(void) { if (ftrace_pages->next) @@ -339,16 +327,8 @@ static notrace void ftrace_shutdown_replenish(void) ftrace_pages->next = (void *)get_zeroed_page(GFP_KERNEL); } -static int notrace __ftrace_modify_code(void *data) -{ - void (*func)(void) = data; - - func(); - return 0; -} - static notrace void -ftrace_code_disable(struct dyn_ftrace *rec, unsigned long addr) +ftrace_code_disable(struct dyn_ftrace *rec) { unsigned long ip; unsigned char *nop, *call; @@ -357,67 +337,113 @@ ftrace_code_disable(struct dyn_ftrace *rec, unsigned long addr) ip = rec->ip; nop = ftrace_nop_replace(); - call = ftrace_call_replace(ip, addr); + call = ftrace_call_replace(ip, MCOUNT_ADDR); failed = ftrace_modify_code(ip, call, nop); if (failed) rec->flags |= FTRACE_FL_FAILED; } -static void notrace ftrace_run_startup_code(void) +static int notrace __ftrace_modify_code(void *data) { - stop_machine_run(__ftrace_modify_code, ftrace_startup_code, NR_CPUS); + unsigned long addr; + int *command = data; + + if (*command & FTRACE_ENABLE_CALLS) + ftrace_replace_code(1); + else if (*command & FTRACE_DISABLE_CALLS) + ftrace_replace_code(0); + + if (*command & FTRACE_UPDATE_TRACE_FUNC) + ftrace_update_ftrace_func(ftrace_trace_function); + + if (*command & FTRACE_ENABLE_MCOUNT) { + addr = (unsigned long)ftrace_record_ip; + ftrace_mcount_set(&addr); + } else if (*command & FTRACE_DISABLE_MCOUNT) { + addr = (unsigned long)ftrace_stub; + ftrace_mcount_set(&addr); + } + + return 0; } -static void notrace ftrace_run_shutdown_code(void) +static void notrace ftrace_run_update_code(int command) { - stop_machine_run(__ftrace_modify_code, ftrace_shutdown_code, NR_CPUS); + stop_machine_run(__ftrace_modify_code, &command, NR_CPUS); } +static ftrace_func_t saved_ftrace_func; + static void notrace ftrace_startup(void) { + int command = 0; + mutex_lock(&ftraced_lock); ftraced_suspend++; - if (ftraced_suspend != 1) + if (ftraced_suspend == 1) + command |= FTRACE_ENABLE_CALLS; + + if (saved_ftrace_func != ftrace_trace_function) { + saved_ftrace_func = ftrace_trace_function; + command |= FTRACE_UPDATE_TRACE_FUNC; + } + + if (!command || !ftrace_enabled) goto out; - __unregister_ftrace_function(&ftrace_shutdown_ops); - if (ftrace_enabled) - ftrace_run_startup_code(); + ftrace_run_update_code(command); out: mutex_unlock(&ftraced_lock); } static void notrace ftrace_shutdown(void) { + int command = 0; + mutex_lock(&ftraced_lock); ftraced_suspend--; - if (ftraced_suspend) - goto out; + if (!ftraced_suspend) + command |= FTRACE_DISABLE_CALLS; - if (ftrace_enabled) - ftrace_run_shutdown_code(); + if (saved_ftrace_func != ftrace_trace_function) { + saved_ftrace_func = ftrace_trace_function; + command |= FTRACE_UPDATE_TRACE_FUNC; + } - __register_ftrace_function(&ftrace_shutdown_ops); + if (!command || !ftrace_enabled) + goto out; + + ftrace_run_update_code(command); out: mutex_unlock(&ftraced_lock); } static void notrace ftrace_startup_sysctl(void) { + int command = FTRACE_ENABLE_MCOUNT; + mutex_lock(&ftraced_lock); + /* Force update next time */ + saved_ftrace_func = NULL; /* ftraced_suspend is true if we want ftrace running */ if (ftraced_suspend) - ftrace_run_startup_code(); + command |= FTRACE_ENABLE_CALLS; + + ftrace_run_update_code(command); mutex_unlock(&ftraced_lock); } static void notrace ftrace_shutdown_sysctl(void) { + int command = FTRACE_DISABLE_MCOUNT; + mutex_lock(&ftraced_lock); /* ftraced_suspend is true if ftrace is running */ if (ftraced_suspend) - ftrace_run_shutdown_code(); + command |= FTRACE_DISABLE_CALLS; + + ftrace_run_update_code(command); mutex_unlock(&ftraced_lock); } @@ -430,11 +456,13 @@ static int notrace __ftrace_update_code(void *ignore) struct dyn_ftrace *p; struct hlist_head head; struct hlist_node *t; + int save_ftrace_enabled; cycle_t start, stop; int i; - /* Don't be calling ftrace ops now */ - __unregister_ftrace_function(&ftrace_shutdown_ops); + /* Don't be recording funcs now */ + save_ftrace_enabled = ftrace_enabled; + ftrace_enabled = 0; start = now(raw_smp_processor_id()); ftrace_update_cnt = 0; @@ -449,7 +477,7 @@ static int notrace __ftrace_update_code(void *ignore) /* all CPUS are stopped, we are safe to modify code */ hlist_for_each_entry(p, t, &head, node) { - ftrace_code_disable(p, MCOUNT_ADDR); + ftrace_code_disable(p); ftrace_update_cnt++; } @@ -459,7 +487,7 @@ static int notrace __ftrace_update_code(void *ignore) ftrace_update_time = stop - start; ftrace_update_tot_cnt += ftrace_update_cnt; - __register_ftrace_function(&ftrace_shutdown_ops); + ftrace_enabled = save_ftrace_enabled; return 0; } @@ -515,11 +543,6 @@ static int __init ftrace_dyn_table_alloc(void) struct ftrace_page *pg; int cnt; int i; - int ret; - - ret = ftrace_dyn_arch_init(); - if (ret) - return ret; /* allocate a few pages */ ftrace_pages_start = (void *)get_zeroed_page(GFP_KERNEL); @@ -557,11 +580,19 @@ static int __init ftrace_dyn_table_alloc(void) return 0; } -static int __init notrace ftrace_shutdown_init(void) +static int __init notrace ftrace_dynamic_init(void) { struct task_struct *p; + unsigned long addr; int ret; + addr = (unsigned long)ftrace_record_ip; + stop_machine_run(ftrace_dyn_arch_init, &addr, NR_CPUS); + + /* ftrace_dyn_arch_init places the return code in addr */ + if (addr) + return addr; + ret = ftrace_dyn_table_alloc(); if (ret) return ret; @@ -570,12 +601,12 @@ static int __init notrace ftrace_shutdown_init(void) if (IS_ERR(p)) return -1; - __register_ftrace_function(&ftrace_shutdown_ops); + last_ftrace_enabled = ftrace_enabled = 1; return 0; } -core_initcall(ftrace_shutdown_init); +core_initcall(ftrace_dynamic_init); #else # define ftrace_startup() do { } while (0) # define ftrace_shutdown() do { } while (0) @@ -599,9 +630,8 @@ int register_ftrace_function(struct ftrace_ops *ops) int ret; mutex_lock(&ftrace_sysctl_lock); - ftrace_startup(); - ret = __register_ftrace_function(ops); + ftrace_startup(); mutex_unlock(&ftrace_sysctl_lock); return ret; @@ -619,10 +649,7 @@ int unregister_ftrace_function(struct ftrace_ops *ops) mutex_lock(&ftrace_sysctl_lock); ret = __unregister_ftrace_function(ops); - - if (ftrace_list == &ftrace_list_end) - ftrace_shutdown(); - + ftrace_shutdown(); mutex_unlock(&ftrace_sysctl_lock); return ret; -- cgit v1.2.3 From 5072c59fd45e9976d02ee6f18c7336ef97623cbc Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:43 +0200 Subject: ftrace: add filter select functions to trace This patch adds two files to the debugfs system: /debugfs/tracing/available_filter_functions and /debugfs/tracing/set_ftrace_filter The available_filter_functions lists all functions that has been recorded by the ftraced that has called the ftrace_record_ip function. This is to allow users to see what functions have been converted to nops and can be enabled for tracing. To enable functions, simply echo the names (whitespace delimited) into set_ftrace_filter. Simple wildcards are also allowed. echo 'scheduler' > /debugfs/tracing/set_ftrace_filter Will have only the scheduler be activated when tracing is enabled. echo 'sched_*' > /debugfs/tracing/set_ftrace_filter Will have only the functions starting with 'sched_' be activated. echo '*lock' > /debugfs/tracing/set_ftrace_filter Will have only functions ending with 'lock' be activated. echo '*lock*' > /debugfs/tracing/set_ftrace_filter Will have only functions with 'lock' in its name be activated. Note: 'sched*lock' will not work. The only wildcards that are allowed is an asterisk and the beginning and or end of the string passed in. Multiple names can be passed in with whitespace delimited: echo 'scheduler *lock *acpi*' > /debugfs/tracing/set_ftrace_filter is also the same as: echo 'scheduler' > /debugfs/tracing/set_ftrace_filter echo '*lock' >> /debugfs/tracing/set_ftrace_filter echo '*acpi*' >> /debugfs/tracing/set_ftrace_filter Appending does just that. It appends to the list. To disable all filters simply echo an empty line in: echo > /debugfs/tracing/set_ftrace_filter Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 527 ++++++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 510 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 88544f9bc0ed..97d5cb7b7e75 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -16,12 +16,15 @@ #include #include #include +#include +#include #include #include #include -#include +#include #include #include +#include #include #include "trace.h" @@ -151,12 +154,15 @@ enum { FTRACE_DISABLE_MCOUNT = (1 << 4), }; +static int ftrace_filtered; + static struct hlist_head ftrace_hash[FTRACE_HASHSIZE]; static DEFINE_PER_CPU(int, ftrace_shutdown_disable_cpu); static DEFINE_SPINLOCK(ftrace_shutdown_lock); static DEFINE_MUTEX(ftraced_lock); +static DEFINE_MUTEX(ftrace_filter_lock); struct ftrace_page { struct ftrace_page *next; @@ -282,16 +288,82 @@ ftrace_record_ip(unsigned long ip) #define FTRACE_ADDR ((long)(&ftrace_caller)) #define MCOUNT_ADDR ((long)(&mcount)) -static void notrace ftrace_replace_code(int saved) +static void notrace +__ftrace_replace_code(struct dyn_ftrace *rec, + unsigned char *old, unsigned char *new, int enable) +{ + unsigned long ip; + int failed; + + ip = rec->ip; + + if (ftrace_filtered && enable) { + unsigned long fl; + /* + * If filtering is on: + * + * If this record is set to be filtered and + * is enabled then do nothing. + * + * If this record is set to be filtered and + * it is not enabled, enable it. + * + * If this record is not set to be filtered + * and it is not enabled do nothing. + * + * If this record is not set to be filtered and + * it is enabled, disable it. + */ + fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_ENABLED); + + if ((fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) || + (fl == 0)) + return; + + /* + * If it is enabled disable it, + * otherwise enable it! + */ + if (fl == FTRACE_FL_ENABLED) { + /* swap new and old */ + new = old; + old = ftrace_call_replace(ip, FTRACE_ADDR); + rec->flags &= ~FTRACE_FL_ENABLED; + } else { + new = ftrace_call_replace(ip, FTRACE_ADDR); + rec->flags |= FTRACE_FL_ENABLED; + } + } else { + + if (enable) + new = ftrace_call_replace(ip, FTRACE_ADDR); + else + old = ftrace_call_replace(ip, FTRACE_ADDR); + + if (enable) { + if (rec->flags & FTRACE_FL_ENABLED) + return; + rec->flags |= FTRACE_FL_ENABLED; + } else { + if (!(rec->flags & FTRACE_FL_ENABLED)) + return; + rec->flags &= ~FTRACE_FL_ENABLED; + } + } + + failed = ftrace_modify_code(ip, old, new); + if (failed) + rec->flags |= FTRACE_FL_FAILED; +} + +static void notrace ftrace_replace_code(int enable) { unsigned char *new = NULL, *old = NULL; struct dyn_ftrace *rec; struct ftrace_page *pg; - unsigned long ip; - int failed; int i; - if (saved) + if (enable) old = ftrace_nop_replace(); else new = ftrace_nop_replace(); @@ -304,16 +376,7 @@ static void notrace ftrace_replace_code(int saved) if (rec->flags & FTRACE_FL_FAILED) continue; - ip = rec->ip; - - if (saved) - new = ftrace_call_replace(ip, FTRACE_ADDR); - else - old = ftrace_call_replace(ip, FTRACE_ADDR); - - failed = ftrace_modify_code(ip, old, new); - if (failed) - rec->flags |= FTRACE_FL_FAILED; + __ftrace_replace_code(rec, old, new, enable); } } } @@ -580,6 +643,436 @@ static int __init ftrace_dyn_table_alloc(void) return 0; } +enum { + FTRACE_ITER_FILTER = (1 << 0), + FTRACE_ITER_CONT = (1 << 1), +}; + +#define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ + +struct ftrace_iterator { + loff_t pos; + struct ftrace_page *pg; + unsigned idx; + unsigned flags; + unsigned char buffer[FTRACE_BUFF_MAX+1]; + unsigned buffer_idx; + unsigned filtered; +}; + +static void notrace * +t_next(struct seq_file *m, void *v, loff_t *pos) +{ + struct ftrace_iterator *iter = m->private; + struct dyn_ftrace *rec = NULL; + + (*pos)++; + + retry: + if (iter->idx >= iter->pg->index) { + if (iter->pg->next) { + iter->pg = iter->pg->next; + iter->idx = 0; + goto retry; + } + } else { + rec = &iter->pg->records[iter->idx++]; + if ((rec->flags & FTRACE_FL_FAILED) || + ((iter->flags & FTRACE_ITER_FILTER) && + !(rec->flags & FTRACE_FL_FILTER))) { + rec = NULL; + goto retry; + } + } + + iter->pos = *pos; + + return rec; +} + +static void *t_start(struct seq_file *m, loff_t *pos) +{ + struct ftrace_iterator *iter = m->private; + void *p = NULL; + loff_t l = -1; + + if (*pos != iter->pos) { + for (p = t_next(m, p, &l); p && l < *pos; p = t_next(m, p, &l)) + ; + } else { + l = *pos; + p = t_next(m, p, &l); + } + + return p; +} + +static void t_stop(struct seq_file *m, void *p) +{ +} + +static int t_show(struct seq_file *m, void *v) +{ + struct dyn_ftrace *rec = v; + char str[KSYM_SYMBOL_LEN]; + + if (!rec) + return 0; + + kallsyms_lookup(rec->ip, NULL, NULL, NULL, str); + + seq_printf(m, "%s\n", str); + + return 0; +} + +static struct seq_operations show_ftrace_seq_ops = { + .start = t_start, + .next = t_next, + .stop = t_stop, + .show = t_show, +}; + +static int notrace +ftrace_avail_open(struct inode *inode, struct file *file) +{ + struct ftrace_iterator *iter; + int ret; + + iter = kzalloc(sizeof(*iter), GFP_KERNEL); + if (!iter) + return -ENOMEM; + + iter->pg = ftrace_pages_start; + iter->pos = -1; + + ret = seq_open(file, &show_ftrace_seq_ops); + if (!ret) { + struct seq_file *m = file->private_data; + m->private = iter; + } else + kfree(iter); + + return ret; +} + +int ftrace_avail_release(struct inode *inode, struct file *file) +{ + struct seq_file *m = (struct seq_file *)file->private_data; + struct ftrace_iterator *iter = m->private; + + seq_release(inode, file); + kfree(iter); + return 0; +} + +static void notrace ftrace_filter_reset(void) +{ + struct ftrace_page *pg; + struct dyn_ftrace *rec; + unsigned i; + + /* keep kstop machine from running */ + preempt_disable(); + ftrace_filtered = 0; + pg = ftrace_pages_start; + while (pg) { + for (i = 0; i < pg->index; i++) { + rec = &pg->records[i]; + if (rec->flags & FTRACE_FL_FAILED) + continue; + rec->flags &= ~FTRACE_FL_FILTER; + } + pg = pg->next; + } + preempt_enable(); +} + +static int notrace +ftrace_filter_open(struct inode *inode, struct file *file) +{ + struct ftrace_iterator *iter; + int ret = 0; + + iter = kzalloc(sizeof(*iter), GFP_KERNEL); + if (!iter) + return -ENOMEM; + + mutex_lock(&ftrace_filter_lock); + if ((file->f_mode & FMODE_WRITE) && + !(file->f_flags & O_APPEND)) + ftrace_filter_reset(); + + if (file->f_mode & FMODE_READ) { + iter->pg = ftrace_pages_start; + iter->pos = -1; + iter->flags = FTRACE_ITER_FILTER; + + ret = seq_open(file, &show_ftrace_seq_ops); + if (!ret) { + struct seq_file *m = file->private_data; + m->private = iter; + } else + kfree(iter); + } else + file->private_data = iter; + mutex_unlock(&ftrace_filter_lock); + + return ret; +} + +static ssize_t notrace +ftrace_filter_read(struct file *file, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + if (file->f_mode & FMODE_READ) + return seq_read(file, ubuf, cnt, ppos); + else + return -EPERM; +} + +static loff_t notrace +ftrace_filter_lseek(struct file *file, loff_t offset, int origin) +{ + loff_t ret; + + if (file->f_mode & FMODE_READ) + ret = seq_lseek(file, offset, origin); + else + file->f_pos = ret = 1; + + return ret; +} + +enum { + MATCH_FULL, + MATCH_FRONT_ONLY, + MATCH_MIDDLE_ONLY, + MATCH_END_ONLY, +}; + +static void notrace +ftrace_match(unsigned char *buff, int len) +{ + char str[KSYM_SYMBOL_LEN]; + char *search = NULL; + struct ftrace_page *pg; + struct dyn_ftrace *rec; + int type = MATCH_FULL; + unsigned i, match = 0, search_len = 0; + + for (i = 0; i < len; i++) { + if (buff[i] == '*') { + if (!i) { + search = buff + i + 1; + type = MATCH_END_ONLY; + search_len = len - (i + 1); + } else { + if (type == MATCH_END_ONLY) { + type = MATCH_MIDDLE_ONLY; + } else { + match = i; + type = MATCH_FRONT_ONLY; + } + buff[i] = 0; + break; + } + } + } + + /* keep kstop machine from running */ + preempt_disable(); + ftrace_filtered = 1; + pg = ftrace_pages_start; + while (pg) { + for (i = 0; i < pg->index; i++) { + int matched = 0; + char *ptr; + + rec = &pg->records[i]; + if (rec->flags & FTRACE_FL_FAILED) + continue; + kallsyms_lookup(rec->ip, NULL, NULL, NULL, str); + switch (type) { + case MATCH_FULL: + if (strcmp(str, buff) == 0) + matched = 1; + break; + case MATCH_FRONT_ONLY: + if (memcmp(str, buff, match) == 0) + matched = 1; + break; + case MATCH_MIDDLE_ONLY: + if (strstr(str, search)) + matched = 1; + break; + case MATCH_END_ONLY: + ptr = strstr(str, search); + if (ptr && (ptr[search_len] == 0)) + matched = 1; + break; + } + if (matched) + rec->flags |= FTRACE_FL_FILTER; + } + pg = pg->next; + } + preempt_enable(); +} + +static ssize_t notrace +ftrace_filter_write(struct file *file, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct ftrace_iterator *iter; + char ch; + size_t read = 0; + ssize_t ret; + + if (!cnt || cnt < 0) + return 0; + + mutex_lock(&ftrace_filter_lock); + + if (file->f_mode & FMODE_READ) { + struct seq_file *m = file->private_data; + iter = m->private; + } else + iter = file->private_data; + + if (!*ppos) { + iter->flags &= ~FTRACE_ITER_CONT; + iter->buffer_idx = 0; + } + + ret = get_user(ch, ubuf++); + if (ret) + goto out; + read++; + cnt--; + + if (!(iter->flags & ~FTRACE_ITER_CONT)) { + /* skip white space */ + while (cnt && isspace(ch)) { + ret = get_user(ch, ubuf++); + if (ret) + goto out; + read++; + cnt--; + } + + + if (isspace(ch)) { + file->f_pos += read; + ret = read; + goto out; + } + + iter->buffer_idx = 0; + } + + while (cnt && !isspace(ch)) { + if (iter->buffer_idx < FTRACE_BUFF_MAX) + iter->buffer[iter->buffer_idx++] = ch; + else { + ret = -EINVAL; + goto out; + } + ret = get_user(ch, ubuf++); + if (ret) + goto out; + read++; + cnt--; + } + + if (isspace(ch)) { + iter->filtered++; + iter->buffer[iter->buffer_idx] = 0; + ftrace_match(iter->buffer, iter->buffer_idx); + iter->buffer_idx = 0; + } else + iter->flags |= FTRACE_ITER_CONT; + + + file->f_pos += read; + + ret = read; + out: + mutex_unlock(&ftrace_filter_lock); + + return ret; +} + +static int notrace +ftrace_filter_release(struct inode *inode, struct file *file) +{ + struct seq_file *m = (struct seq_file *)file->private_data; + struct ftrace_iterator *iter; + + mutex_lock(&ftrace_filter_lock); + if (file->f_mode & FMODE_READ) { + iter = m->private; + + seq_release(inode, file); + } else + iter = file->private_data; + + if (iter->buffer_idx) { + iter->filtered++; + iter->buffer[iter->buffer_idx] = 0; + ftrace_match(iter->buffer, iter->buffer_idx); + } + + mutex_lock(&ftrace_sysctl_lock); + mutex_lock(&ftraced_lock); + if (iter->filtered && ftraced_suspend && ftrace_enabled) + ftrace_run_update_code(FTRACE_ENABLE_CALLS); + mutex_unlock(&ftraced_lock); + mutex_unlock(&ftrace_sysctl_lock); + + kfree(iter); + mutex_unlock(&ftrace_filter_lock); + return 0; +} + +static struct file_operations ftrace_avail_fops = { + .open = ftrace_avail_open, + .read = seq_read, + .llseek = seq_lseek, + .release = ftrace_avail_release, +}; + +static struct file_operations ftrace_filter_fops = { + .open = ftrace_filter_open, + .read = ftrace_filter_read, + .write = ftrace_filter_write, + .llseek = ftrace_filter_lseek, + .release = ftrace_filter_release, +}; + +static __init int ftrace_init_debugfs(void) +{ + struct dentry *d_tracer; + struct dentry *entry; + + d_tracer = tracing_init_dentry(); + + entry = debugfs_create_file("available_filter_functions", 0444, + d_tracer, NULL, &ftrace_avail_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'available_filter_functions' entry\n"); + + entry = debugfs_create_file("set_ftrace_filter", 0644, d_tracer, + NULL, &ftrace_filter_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'set_ftrace_filter' entry\n"); + return 0; +} + +fs_initcall(ftrace_init_debugfs); + static int __init notrace ftrace_dynamic_init(void) { struct task_struct *p; @@ -657,14 +1150,14 @@ int unregister_ftrace_function(struct ftrace_ops *ops) notrace int ftrace_enable_sysctl(struct ctl_table *table, int write, - struct file *filp, void __user *buffer, size_t *lenp, + struct file *file, void __user *buffer, size_t *lenp, loff_t *ppos) { int ret; mutex_lock(&ftrace_sysctl_lock); - ret = proc_dointvec(table, write, filp, buffer, lenp, ppos); + ret = proc_dointvec(table, write, file, buffer, lenp, ppos); if (ret || !write || (last_ftrace_enabled == ftrace_enabled)) goto out; -- cgit v1.2.3 From 4c11d7aed389375253b59e2b1865eec96663c65d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:43 +0200 Subject: ftrace: convert single large buffer into single pages. Allocating large buffers for the tracer may fail easily. This patch converts the buffer from a large ordered allocation to single pages. It uses the struct page LRU field to link the pages together. Later patches may also implement dynamic increasing and decreasing of the trace buffers. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 245 +++++++++++++++++++++++++++++++++++++++------------ kernel/trace/trace.h | 8 +- 2 files changed, 195 insertions(+), 58 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 1b8eca7650d4..d7ad030a4c49 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -15,6 +15,7 @@ #include #include #include +#include #include #include #include @@ -49,7 +50,7 @@ static struct trace_array max_tr; static DEFINE_PER_CPU(struct trace_array_cpu, max_data); static int tracer_enabled; -static unsigned long trace_nr_entries = 4096UL; +static unsigned long trace_nr_entries = 16384UL; static struct tracer *trace_types __read_mostly; static struct tracer *current_trace __read_mostly; @@ -57,6 +58,8 @@ static int max_tracer_type_len; static DEFINE_MUTEX(trace_types_lock); +#define ENTRIES_PER_PAGE (PAGE_SIZE / sizeof(struct trace_entry)) + static int __init set_nr_entries(char *str) { if (!str) @@ -103,6 +106,7 @@ static const char *trace_options[] = { static unsigned trace_flags; +static DEFINE_SPINLOCK(ftrace_max_lock); /* * Copy the new maximum trace into the separate maximum-trace @@ -136,17 +140,23 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data; void *save_trace; + struct list_head save_pages; int i; + WARN_ON_ONCE(!irqs_disabled()); + spin_lock(&ftrace_max_lock); /* clear out all the previous traces */ for_each_possible_cpu(i) { data = tr->data[i]; save_trace = max_tr.data[i]->trace; + save_pages = max_tr.data[i]->trace_pages; memcpy(max_tr.data[i], data, sizeof(*data)); data->trace = save_trace; + data->trace_pages = save_pages; } __update_max_tr(tr, tsk, cpu); + spin_unlock(&ftrace_max_lock); } /** @@ -160,16 +170,22 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data = tr->data[cpu]; void *save_trace; + struct list_head save_pages; int i; + WARN_ON_ONCE(!irqs_disabled()); + spin_lock(&ftrace_max_lock); for_each_possible_cpu(i) tracing_reset(max_tr.data[i]); save_trace = max_tr.data[cpu]->trace; + save_pages = max_tr.data[cpu]->trace_pages; memcpy(max_tr.data[cpu], data, sizeof(*data)); data->trace = save_trace; + data->trace_pages = save_pages; __update_max_tr(tr, tsk, cpu); + spin_unlock(&ftrace_max_lock); } int register_tracer(struct tracer *type) @@ -236,7 +252,8 @@ void unregister_tracer(struct tracer *type) void notrace tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; - atomic_set(&data->underrun, 0); + data->trace_current = data->trace; + data->trace_current_idx = 0; } #ifdef CONFIG_FTRACE @@ -367,21 +384,27 @@ tracing_get_trace_entry(struct trace_array *tr, { unsigned long idx, idx_next; struct trace_entry *entry; + struct page *page; + struct list_head *next; - idx = data->trace_idx; + data->trace_idx++; + idx = data->trace_current_idx; idx_next = idx + 1; - if (unlikely(idx_next >= tr->entries)) { - atomic_inc(&data->underrun); + entry = data->trace_current + idx * TRACE_ENTRY_SIZE; + + if (unlikely(idx_next >= ENTRIES_PER_PAGE)) { + page = virt_to_page(data->trace_current); + if (unlikely(&page->lru == data->trace_pages.prev)) + next = data->trace_pages.next; + else + next = page->lru.next; + page = list_entry(next, struct page, lru); + data->trace_current = page_address(page); idx_next = 0; } - data->trace_idx = idx_next; - - if (unlikely(idx_next != 0 && atomic_read(&data->underrun))) - atomic_inc(&data->underrun); - - entry = data->trace + idx * TRACE_ENTRY_SIZE; + data->trace_current_idx = idx_next; return entry; } @@ -442,21 +465,38 @@ enum trace_file_type { }; static struct trace_entry * -trace_entry_idx(struct trace_array *tr, unsigned long idx, int cpu) +trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, + struct trace_iterator *iter, int cpu) { - struct trace_entry *array = tr->data[cpu]->trace; - unsigned long underrun; + struct page *page; + struct trace_entry *array; - if (idx >= tr->entries) + if (iter->next_idx[cpu] >= tr->entries || + iter->next_idx[cpu] >= data->trace_idx) return NULL; - underrun = atomic_read(&tr->data[cpu]->underrun); - if (underrun) - idx = ((underrun - 1) + idx) % tr->entries; - else if (idx >= tr->data[cpu]->trace_idx) - return NULL; + if (!iter->next_page[cpu]) { + /* + * Initialize. If the count of elements in + * this buffer is greater than the max entries + * we had an underrun. Which means we looped around. + * We can simply use the current pointer as our + * starting point. + */ + if (data->trace_idx >= tr->entries) { + page = virt_to_page(data->trace_current); + iter->next_page[cpu] = &page->lru; + iter->next_page_idx[cpu] = data->trace_current_idx; + } else { + iter->next_page[cpu] = data->trace_pages.next; + iter->next_page_idx[cpu] = 0; + } + } - return &array[idx]; + page = list_entry(iter->next_page[cpu], struct page, lru); + array = page_address(page); + + return &array[iter->next_page_idx[cpu]]; } static struct notrace trace_entry * @@ -470,7 +510,7 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) for_each_possible_cpu(cpu) { if (!tr->data[cpu]->trace) continue; - ent = trace_entry_idx(tr, iter->next_idx[cpu], cpu); + ent = trace_entry_idx(tr, tr->data[cpu], iter, cpu); if (ent && (!next || (long)(next->idx - ent->idx) > 0)) { next = ent; @@ -492,8 +532,19 @@ static void *find_next_entry_inc(struct trace_iterator *iter) next = find_next_entry(iter, &next_cpu); if (next) { - iter->next_idx[next_cpu]++; iter->idx++; + iter->next_idx[next_cpu]++; + iter->next_page_idx[next_cpu]++; + if (iter->next_page_idx[next_cpu] >= ENTRIES_PER_PAGE) { + struct trace_array_cpu *data = iter->tr->data[next_cpu]; + + iter->next_page_idx[next_cpu] = 0; + iter->next_page[next_cpu] = + iter->next_page[next_cpu]->next; + if (iter->next_page[next_cpu] == &data->trace_pages) + iter->next_page[next_cpu] = + data->trace_pages.next; + } } iter->ent = next; iter->cpu = next_cpu; @@ -554,14 +605,16 @@ static void *s_start(struct seq_file *m, loff_t *pos) iter->cpu = 0; iter->idx = -1; - for (i = 0; i < NR_CPUS; i++) + for_each_possible_cpu(i) { iter->next_idx[i] = 0; + iter->next_page[i] = NULL; + } for (p = iter; p && l < *pos; p = s_next(m, p, &l)) ; } else { - l = *pos; + l = *pos - 1; p = s_next(m, p, &l); } @@ -654,9 +707,8 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) struct trace_array *tr = iter->tr; struct trace_array_cpu *data = tr->data[tr->cpu]; struct tracer *type = current_trace; - unsigned long underruns = 0; - unsigned long underrun; - unsigned long entries = 0; + unsigned long total = 0; + unsigned long entries = 0; int cpu; const char *name = "preemption"; @@ -665,11 +717,10 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) for_each_possible_cpu(cpu) { if (tr->data[cpu]->trace) { - underrun = atomic_read(&tr->data[cpu]->underrun); - if (underrun) { - underruns += underrun; + total += tr->data[cpu]->trace_idx; + if (tr->data[cpu]->trace_idx > tr->entries) entries += tr->entries; - } else + else entries += tr->data[cpu]->trace_idx; } } @@ -682,7 +733,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) " (M:%s VP:%d, KP:%d, SP:%d HP:%d", data->saved_latency, entries, - (entries + underruns), + total, tr->cpu, #if defined(CONFIG_PREEMPT_NONE) "server", @@ -882,8 +933,7 @@ static int trace_empty(struct trace_iterator *iter) data = iter->tr->data[cpu]; if (data->trace && - (data->trace_idx || - atomic_read(&data->underrun))) + data->trace_idx) return 0; } return 1; @@ -1464,42 +1514,109 @@ static struct tracer no_tracer __read_mostly = .name = "none", }; -static inline notrace int page_order(const unsigned long size) +static int trace_alloc_page(void) { - const unsigned long nr_pages = DIV_ROUND_UP(size, PAGE_SIZE); - return ilog2(roundup_pow_of_two(nr_pages)); + struct trace_array_cpu *data; + void *array; + struct page *page, *tmp; + LIST_HEAD(pages); + int i; + + /* first allocate a page for each CPU */ + for_each_possible_cpu(i) { + array = (void *)__get_free_page(GFP_KERNEL); + if (array == NULL) { + printk(KERN_ERR "tracer: failed to allocate page" + "for trace buffer!\n"); + goto free_pages; + } + + page = virt_to_page(array); + list_add(&page->lru, &pages); + +/* Only allocate if we are actually using the max trace */ +#ifdef CONFIG_TRACER_MAX_TRACE + array = (void *)__get_free_page(GFP_KERNEL); + if (array == NULL) { + printk(KERN_ERR "tracer: failed to allocate page" + "for trace buffer!\n"); + goto free_pages; + } + page = virt_to_page(array); + list_add(&page->lru, &pages); +#endif + } + + /* Now that we successfully allocate a page per CPU, add them */ + for_each_possible_cpu(i) { + data = global_trace.data[i]; + page = list_entry(pages.next, struct page, lru); + list_del(&page->lru); + list_add_tail(&page->lru, &data->trace_pages); + ClearPageLRU(page); + +#ifdef CONFIG_TRACER_MAX_TRACE + data = max_tr.data[i]; + page = list_entry(pages.next, struct page, lru); + list_del(&page->lru); + list_add_tail(&page->lru, &data->trace_pages); + SetPageLRU(page); +#endif + } + global_trace.entries += ENTRIES_PER_PAGE; + + return 0; + + free_pages: + list_for_each_entry_safe(page, tmp, &pages, lru) { + list_del(&page->lru); + __free_page(page); + } + return -ENOMEM; } __init static int tracer_alloc_buffers(void) { - const int order = page_order(trace_nr_entries * TRACE_ENTRY_SIZE); - const unsigned long size = (1UL << order) << PAGE_SHIFT; - struct trace_entry *array; + struct trace_array_cpu *data; + void *array; + struct page *page; + int pages = 0; int i; + /* Allocate the first page for all buffers */ for_each_possible_cpu(i) { - global_trace.data[i] = &per_cpu(global_trace_cpu, i); + data = global_trace.data[i] = &per_cpu(global_trace_cpu, i); max_tr.data[i] = &per_cpu(max_data, i); - array = (struct trace_entry *) - __get_free_pages(GFP_KERNEL, order); + array = (void *)__get_free_page(GFP_KERNEL); if (array == NULL) { - printk(KERN_ERR "tracer: failed to allocate" - " %ld bytes for trace buffer!\n", size); + printk(KERN_ERR "tracer: failed to allocate page" + "for trace buffer!\n"); goto free_buffers; } - global_trace.data[i]->trace = array; + data->trace = array; + + /* set the array to the list */ + INIT_LIST_HEAD(&data->trace_pages); + page = virt_to_page(array); + list_add(&page->lru, &data->trace_pages); + /* use the LRU flag to differentiate the two buffers */ + ClearPageLRU(page); /* Only allocate if we are actually using the max trace */ #ifdef CONFIG_TRACER_MAX_TRACE - array = (struct trace_entry *) - __get_free_pages(GFP_KERNEL, order); + array = (void *)__get_free_page(GFP_KERNEL); if (array == NULL) { - printk(KERN_ERR "wakeup tracer: failed to allocate" - " %ld bytes for trace buffer!\n", size); + printk(KERN_ERR "tracer: failed to allocate page" + "for trace buffer!\n"); goto free_buffers; } max_tr.data[i]->trace = array; + + INIT_LIST_HEAD(&max_tr.data[i]->trace_pages); + page = virt_to_page(array); + list_add(&page->lru, &max_tr.data[i]->trace_pages); + SetPageLRU(page); #endif } @@ -1507,11 +1624,18 @@ __init static int tracer_alloc_buffers(void) * Since we allocate by orders of pages, we may be able to * round up a bit. */ - global_trace.entries = size / TRACE_ENTRY_SIZE; + global_trace.entries = ENTRIES_PER_PAGE; max_tr.entries = global_trace.entries; + pages++; + + while (global_trace.entries < trace_nr_entries) { + if (trace_alloc_page()) + break; + pages++; + } - pr_info("tracer: %ld bytes allocated for %ld", - size, trace_nr_entries); + pr_info("tracer: %d pages allocated for %ld", + pages, trace_nr_entries); pr_info(" entries of %ld bytes\n", (long)TRACE_ENTRY_SIZE); pr_info(" actual entries %ld\n", global_trace.entries); @@ -1526,17 +1650,26 @@ __init static int tracer_alloc_buffers(void) free_buffers: for (i-- ; i >= 0; i--) { + struct page *page, *tmp; struct trace_array_cpu *data = global_trace.data[i]; if (data && data->trace) { - free_pages((unsigned long)data->trace, order); + list_for_each_entry_safe(page, tmp, + &data->trace_pages, lru) { + list_del(&page->lru); + __free_page(page); + } data->trace = NULL; } #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; if (data && data->trace) { - free_pages((unsigned long)data->trace, order); + list_for_each_entry_safe(page, tmp, + &data->trace_pages, lru) { + list_del(&page->lru); + __free_page(page); + } data->trace = NULL; } #endif diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 3173a93561d4..83e257e38084 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -54,9 +54,11 @@ struct trace_entry { */ struct trace_array_cpu { void *trace; + void *trace_current; + unsigned trace_current_idx; + struct list_head trace_pages; unsigned long trace_idx; atomic_t disabled; - atomic_t underrun; unsigned long saved_latency; unsigned long critical_start; unsigned long critical_end; @@ -112,8 +114,10 @@ struct trace_iterator { unsigned long iter_flags; loff_t pos; unsigned long next_idx[NR_CPUS]; + struct list_head *next_page[NR_CPUS]; + unsigned next_page_idx[NR_CPUS]; + long idx; int cpu; - int idx; }; void notrace tracing_reset(struct trace_array_cpu *data); -- cgit v1.2.3 From 361943ad0ba3f16e66859e30a408915e008ba91e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: irqs off smp_processor_id() fix The irqsoff function tracer did a __get_cpu_var to determine if it should trace the function or not. The problem is that __get_cpu_var can preempt between getting the CPU and reading the cpu variable. This means that the cpu variable that is being read is not from the cpu being run on. At worst, this can give a false positive, where we trace the function when we should not. It will never give a false negative since we only want to trace when interrupts are disabled and we never preempt when they are. This fix adds a check after reading the irq flags to only trace if the interrupts are actually disabled. It also changes the reading of the cpu variable to use a raw_smp_processor_id since we now don't care if we preempt. We still catch that fact. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_irqsoff.c | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 8b1231633dc5..bd3f88198308 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -74,12 +74,21 @@ irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) long disabled; int cpu; - if (likely(!__get_cpu_var(tracing_cpu))) + /* + * Does not matter if we preempt. We test the flags + * afterward, to see if irqs are disabled or not. + * If we preempt and get a false positive, the flags + * test will fail. + */ + cpu = raw_smp_processor_id(); + if (likely(!per_cpu(tracing_cpu, cpu))) return; local_save_flags(flags); + /* slight chance to get a false positive on tracing_cpu */ + if (!irqs_disabled_flags(flags)) + return; - cpu = raw_smp_processor_id(); data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); -- cgit v1.2.3 From 0764d23cf066c52de42b653144605b481d3fbdbc Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: lockdep notrace annotations Add notrace annotations to lockdep to keep ftrace from causing recursive problems with lock tracing and debugging. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/lockdep.c | 23 ++++++++++++----------- kernel/spinlock.c | 2 +- 2 files changed, 13 insertions(+), 12 deletions(-) (limited to 'kernel') diff --git a/kernel/lockdep.c b/kernel/lockdep.c index e21924365ea3..ac46847ba0c9 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -271,14 +271,14 @@ static struct list_head chainhash_table[CHAINHASH_SIZE]; ((key1) >> (64-MAX_LOCKDEP_KEYS_BITS)) ^ \ (key2)) -void lockdep_off(void) +notrace void lockdep_off(void) { current->lockdep_recursion++; } EXPORT_SYMBOL(lockdep_off); -void lockdep_on(void) +notrace void lockdep_on(void) { current->lockdep_recursion--; } @@ -1041,7 +1041,7 @@ find_usage_forwards(struct lock_class *source, unsigned int depth) * Return 1 otherwise and keep unchanged. * Return 0 on error. */ -static noinline int +static noinline notrace int find_usage_backwards(struct lock_class *source, unsigned int depth) { struct lock_list *entry; @@ -1591,7 +1591,7 @@ static inline int validate_chain(struct task_struct *curr, * We are building curr_chain_key incrementally, so double-check * it from scratch, to make sure that it's done correctly: */ -static void check_chain_key(struct task_struct *curr) +static notrace void check_chain_key(struct task_struct *curr) { #ifdef CONFIG_DEBUG_LOCKDEP struct held_lock *hlock, *prev_hlock = NULL; @@ -1967,7 +1967,7 @@ static int mark_lock_irq(struct task_struct *curr, struct held_lock *this, /* * Mark all held locks with a usage bit: */ -static int +static notrace int mark_held_locks(struct task_struct *curr, int hardirq) { enum lock_usage_bit usage_bit; @@ -2260,8 +2260,8 @@ static inline int separate_irq_context(struct task_struct *curr, /* * Mark a lock with a usage bit, and validate the state transition: */ -static int mark_lock(struct task_struct *curr, struct held_lock *this, - enum lock_usage_bit new_bit) +static notrace int mark_lock(struct task_struct *curr, struct held_lock *this, + enum lock_usage_bit new_bit) { unsigned int new_mask = 1 << new_bit, ret = 1; @@ -2663,7 +2663,7 @@ __lock_release(struct lockdep_map *lock, int nested, unsigned long ip) /* * Check whether we follow the irq-flags state precisely: */ -static void check_flags(unsigned long flags) +static notrace void check_flags(unsigned long flags) { #if defined(CONFIG_DEBUG_LOCKDEP) && defined(CONFIG_TRACE_IRQFLAGS) if (!debug_locks) @@ -2700,8 +2700,8 @@ static void check_flags(unsigned long flags) * We are not always called with irqs disabled - do that here, * and also avoid lockdep recursion: */ -void lock_acquire(struct lockdep_map *lock, unsigned int subclass, - int trylock, int read, int check, unsigned long ip) +notrace void lock_acquire(struct lockdep_map *lock, unsigned int subclass, + int trylock, int read, int check, unsigned long ip) { unsigned long flags; @@ -2723,7 +2723,8 @@ void lock_acquire(struct lockdep_map *lock, unsigned int subclass, EXPORT_SYMBOL_GPL(lock_acquire); -void lock_release(struct lockdep_map *lock, int nested, unsigned long ip) +notrace void lock_release(struct lockdep_map *lock, int nested, + unsigned long ip) { unsigned long flags; diff --git a/kernel/spinlock.c b/kernel/spinlock.c index ae28c8245123..a1fb54c93cdd 100644 --- a/kernel/spinlock.c +++ b/kernel/spinlock.c @@ -436,7 +436,7 @@ int __lockfunc _spin_trylock_bh(spinlock_t *lock) } EXPORT_SYMBOL(_spin_trylock_bh); -int in_lock_functions(unsigned long addr) +notrace int in_lock_functions(unsigned long addr) { /* Linker adds these: start and end of __lockfunc functions */ extern char __lock_text_start[], __lock_text_end[]; -- cgit v1.2.3 From 18cef379d30f5ded20cc31d7f2d342639d39919d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: don't use raw_local_irq_save/restore Using raw_local_irq_save/restore confuses lockdep. It's fine to use the normal ones. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 ++-- kernel/trace/trace_sched_switch.c | 4 ++-- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d7ad030a4c49..9175ce91b8f6 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -269,7 +269,7 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) if (unlikely(!tracer_enabled)) return; - raw_local_irq_save(flags); + local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); @@ -278,7 +278,7 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) ftrace(tr, data, ip, parent_ip, flags); atomic_dec(&data->disabled); - raw_local_irq_restore(flags); + local_irq_restore(flags); } static struct ftrace_ops trace_ops __read_mostly = diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 3e4771d3b890..2715267be469 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -29,7 +29,7 @@ ctx_switch_func(struct task_struct *prev, struct task_struct *next) if (!tracer_enabled) return; - raw_local_irq_save(flags); + local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); @@ -38,7 +38,7 @@ ctx_switch_func(struct task_struct *prev, struct task_struct *next) tracing_sched_switch_trace(tr, data, prev, next, flags); atomic_dec(&data->disabled); - raw_local_irq_restore(flags); + local_irq_restore(flags); } void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) -- cgit v1.2.3 From 89b2f97819dd074297bbe3e19eaa4afcc98845ad Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: fix updates to max trace This patch fixes some bugs to the updating of the max trace that was caused by implementing the new buffering. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 6 +++++- kernel/trace/trace_irqsoff.c | 27 +++++++++++++++------------ 2 files changed, 20 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9175ce91b8f6..95966561ba3d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -153,6 +153,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) memcpy(max_tr.data[i], data, sizeof(*data)); data->trace = save_trace; data->trace_pages = save_pages; + tracing_reset(data); } __update_max_tr(tr, tsk, cpu); @@ -183,6 +184,7 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) memcpy(max_tr.data[cpu], data, sizeof(*data)); data->trace = save_trace; data->trace_pages = save_pages; + tracing_reset(data); __update_max_tr(tr, tsk, cpu); spin_unlock(&ftrace_max_lock); @@ -877,6 +879,8 @@ print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, entry->ctx.next_prio, comm); break; + default: + seq_printf(m, "Unknown type %d\n", entry->type); } } @@ -1625,7 +1629,6 @@ __init static int tracer_alloc_buffers(void) * round up a bit. */ global_trace.entries = ENTRIES_PER_PAGE; - max_tr.entries = global_trace.entries; pages++; while (global_trace.entries < trace_nr_entries) { @@ -1633,6 +1636,7 @@ __init static int tracer_alloc_buffers(void) break; pages++; } + max_tr.entries = global_trace.entries; pr_info("tracer: %d pages allocated for %ld", pages, trace_nr_entries); diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index bd3f88198308..74165f611f36 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -23,6 +23,8 @@ static int tracer_enabled __read_mostly; static DEFINE_PER_CPU(int, tracing_cpu); +static DEFINE_SPINLOCK(max_trace_lock); + enum { TRACER_IRQS_OFF = (1 << 1), TRACER_PREEMPT_OFF = (1 << 2), @@ -126,7 +128,7 @@ check_critical_timing(struct trace_array *tr, int cpu) { unsigned long latency, t0, t1; - cycle_t T0, T1, T2, delta; + cycle_t T0, T1, delta; unsigned long flags; /* @@ -142,20 +144,18 @@ check_critical_timing(struct trace_array *tr, if (!report_latency(delta)) goto out; - ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); - /* - * Update the timestamp, because the trace entry above - * might change it (it can only get larger so the latency - * is fair to be reported): - */ - T2 = now(cpu); + spin_lock(&max_trace_lock); - delta = T2-T0; + /* check if we are still the max latency */ + if (!report_latency(delta)) + goto out_unlock; + + ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); latency = nsecs_to_usecs(delta); if (data->critical_sequence != max_sequence) - goto out; + goto out_unlock; tracing_max_latency = delta; t0 = nsecs_to_usecs(T0); @@ -189,6 +189,9 @@ check_critical_timing(struct trace_array *tr, max_sequence++; +out_unlock: + spin_unlock(&max_trace_lock); + out: data->critical_sequence = max_sequence; data->preempt_timestamp = now(cpu); @@ -366,14 +369,14 @@ void notrace trace_preempt_off(unsigned long a0, unsigned long a1) static void start_irqsoff_tracer(struct trace_array *tr) { - tracer_enabled = 1; register_ftrace_function(&trace_ops); + tracer_enabled = 1; } static void stop_irqsoff_tracer(struct trace_array *tr) { - unregister_ftrace_function(&trace_ops); tracer_enabled = 0; + unregister_ftrace_function(&trace_ops); } static void __irqsoff_tracer_init(struct trace_array *tr) -- cgit v1.2.3 From 57f50be14d57b0dbf88dd019e7bb0ff3a3dc7b81 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: fix max latency Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 95966561ba3d..9bad2379115a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -69,6 +69,11 @@ static int __init set_nr_entries(char *str) } __setup("trace_entries=", set_nr_entries); +unsigned long nsecs_to_usecs(unsigned long nsecs) +{ + return nsecs / 1000; +} + enum trace_type { __TRACE_FIRST_TYPE = 0, @@ -733,7 +738,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) "---------------------------------\n"); seq_printf(m, " latency: %lu us, #%lu/%lu, CPU#%d |" " (M:%s VP:%d, KP:%d, SP:%d HP:%d", - data->saved_latency, + nsecs_to_usecs(data->saved_latency), entries, total, tr->cpu, @@ -771,11 +776,6 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) seq_puts(m, "\n"); } -unsigned long nsecs_to_usecs(unsigned long nsecs) -{ - return nsecs / 1000; -} - static void notrace lat_print_generic(struct seq_file *m, struct trace_entry *entry, int cpu) { -- cgit v1.2.3 From e1c08bdd9fa73e44096e5a82c0d5928b04ab02c8 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: force recording Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 51 +++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 51 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 97d5cb7b7e75..4facf5ceeb86 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -146,6 +146,10 @@ static int notrace __unregister_ftrace_function(struct ftrace_ops *ops) #ifdef CONFIG_DYNAMIC_FTRACE +static struct task_struct *ftraced_task; +static DECLARE_WAIT_QUEUE_HEAD(ftraced_waiters); +static unsigned long ftraced_iteration_counter; + enum { FTRACE_ENABLE_CALLS = (1 << 0), FTRACE_DISABLE_CALLS = (1 << 1), @@ -590,9 +594,12 @@ static int notrace ftraced(void *ignore) ftraced_trigger = 0; ftrace_record_suspend--; } + ftraced_iteration_counter++; mutex_unlock(&ftraced_lock); mutex_unlock(&ftrace_sysctl_lock); + wake_up_interruptible(&ftraced_waiters); + ftrace_shutdown_replenish(); set_current_state(TASK_INTERRUPTIBLE); @@ -1050,6 +1057,49 @@ static struct file_operations ftrace_filter_fops = { .release = ftrace_filter_release, }; +/** + * ftrace_force_update - force an update to all recording ftrace functions + * + * The ftrace dynamic update daemon only wakes up once a second. + * There may be cases where an update needs to be done immediately + * for tests or internal kernel tracing to begin. This function + * wakes the daemon to do an update and will not return until the + * update is complete. + */ +int ftrace_force_update(void) +{ + unsigned long last_counter; + DECLARE_WAITQUEUE(wait, current); + int ret = 0; + + if (!ftraced_task) + return -ENODEV; + + mutex_lock(&ftraced_lock); + last_counter = ftraced_iteration_counter; + + set_current_state(TASK_INTERRUPTIBLE); + add_wait_queue(&ftraced_waiters, &wait); + + do { + mutex_unlock(&ftraced_lock); + wake_up_process(ftraced_task); + schedule(); + mutex_lock(&ftraced_lock); + if (signal_pending(current)) { + ret = -EINTR; + break; + } + set_current_state(TASK_INTERRUPTIBLE); + } while (last_counter == ftraced_iteration_counter); + + mutex_unlock(&ftraced_lock); + remove_wait_queue(&ftraced_waiters, &wait); + set_current_state(TASK_RUNNING); + + return ret; +} + static __init int ftrace_init_debugfs(void) { struct dentry *d_tracer; @@ -1095,6 +1145,7 @@ static int __init notrace ftrace_dynamic_init(void) return -1; last_ftrace_enabled = ftrace_enabled = 1; + ftraced_task = p; return 0; } -- cgit v1.2.3 From 60a11774b38fef1ab90b18c5353bd1c7c4d311c8 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:44 +0200 Subject: ftrace: add self-tests Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 13 ++ kernel/trace/trace.c | 63 +++++- kernel/trace/trace.h | 31 +++ kernel/trace/trace_functions.c | 3 + kernel/trace/trace_irqsoff.c | 9 + kernel/trace/trace_sched_switch.c | 3 + kernel/trace/trace_sched_wakeup.c | 3 + kernel/trace/trace_selftest.c | 415 ++++++++++++++++++++++++++++++++++++++ 8 files changed, 538 insertions(+), 2 deletions(-) create mode 100644 kernel/trace/trace_selftest.c (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index cad9db1dee02..3f73a1710242 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -105,3 +105,16 @@ config DYNAMIC_FTRACE wakes up once a second and checks to see if any ftrace calls were made. If so, it runs stop_machine (stops all CPUS) and modifies the code to jump over the call to ftrace. + +config FTRACE_SELFTEST + bool + +config FTRACE_STARTUP_TEST + bool "Perform a startup test on ftrace" + depends on TRACING + select FTRACE_SELFTEST + help + This option performs a series of startup tests on ftrace. On bootup + a series of tests are made to verify that the tracer is + functioning properly. It will do tests on all the configured + tracers of ftrace. diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9bad2379115a..f6d026f17dbb 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -32,6 +32,8 @@ unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; unsigned long __read_mostly tracing_thresh; +static int tracing_disabled = 1; + static long notrace ns2usecs(cycle_t nsec) { @@ -217,11 +219,48 @@ int register_tracer(struct tracer *type) } } +#ifdef CONFIG_FTRACE_STARTUP_TEST + if (type->selftest) { + struct tracer *saved_tracer = current_trace; + struct trace_array_cpu *data; + struct trace_array *tr = &global_trace; + int saved_ctrl = tr->ctrl; + int i; + /* + * Run a selftest on this tracer. + * Here we reset the trace buffer, and set the current + * tracer to be this tracer. The tracer can then run some + * internal tracing to verify that everything is in order. + * If we fail, we do not register this tracer. + */ + for_each_possible_cpu(i) { + if (!data->trace) + continue; + data = tr->data[i]; + tracing_reset(data); + } + current_trace = type; + tr->ctrl = 0; + /* the test is responsible for initializing and enabling */ + pr_info("Testing tracer %s: ", type->name); + ret = type->selftest(type, tr); + /* the test is responsible for resetting too */ + current_trace = saved_tracer; + tr->ctrl = saved_ctrl; + if (ret) { + printk(KERN_CONT "FAILED!\n"); + goto out; + } + printk(KERN_CONT "PASSED\n"); + } +#endif + type->next = trace_types; trace_types = type; len = strlen(type->name); if (len > max_tracer_type_len) max_tracer_type_len = len; + out: mutex_unlock(&trace_types_lock); @@ -985,6 +1024,11 @@ __tracing_open(struct inode *inode, struct file *file, int *ret) { struct trace_iterator *iter; + if (tracing_disabled) { + *ret = -ENODEV; + return NULL; + } + iter = kzalloc(sizeof(*iter), GFP_KERNEL); if (!iter) { *ret = -ENOMEM; @@ -1023,6 +1067,9 @@ __tracing_open(struct inode *inode, struct file *file, int *ret) int tracing_open_generic(struct inode *inode, struct file *filp) { + if (tracing_disabled) + return -ENODEV; + filp->private_data = inode->i_private; return 0; } @@ -1128,6 +1175,9 @@ static int show_traces_open(struct inode *inode, struct file *file) { int ret; + if (tracing_disabled) + return -ENODEV; + ret = seq_open(file, &show_traces_seq_ops); if (!ret) { struct seq_file *m = file->private_data; @@ -1452,6 +1502,11 @@ struct dentry *tracing_init_dentry(void) return d_tracer; } +#ifdef CONFIG_FTRACE_SELFTEST +/* Let selftest have access to static functions in this file */ +#include "trace_selftest.c" +#endif + static __init void tracer_init_debugfs(void) { struct dentry *d_tracer; @@ -1585,6 +1640,7 @@ __init static int tracer_alloc_buffers(void) void *array; struct page *page; int pages = 0; + int ret = -ENOMEM; int i; /* Allocate the first page for all buffers */ @@ -1650,6 +1706,9 @@ __init static int tracer_alloc_buffers(void) register_tracer(&no_tracer); current_trace = &no_tracer; + /* All seems OK, enable tracing */ + tracing_disabled = 0; + return 0; free_buffers: @@ -1678,7 +1737,7 @@ __init static int tracer_alloc_buffers(void) } #endif } - return -ENOMEM; + return ret; } -device_initcall(tracer_alloc_buffers); +fs_initcall(tracer_alloc_buffers); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 83e257e38084..88edbf1f6788 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -99,6 +99,10 @@ struct tracer { void (*start)(struct trace_iterator *iter); void (*stop)(struct trace_iterator *iter); void (*ctrl_update)(struct trace_array *tr); +#ifdef CONFIG_FTRACE_STARTUP_TEST + int (*selftest)(struct tracer *trace, + struct trace_array *tr); +#endif struct tracer *next; int print_max; }; @@ -185,4 +189,31 @@ extern int unregister_tracer_switch(struct tracer_switch_ops *ops); extern unsigned long ftrace_update_tot_cnt; #endif +#ifdef CONFIG_FTRACE_STARTUP_TEST +#ifdef CONFIG_FTRACE +extern int trace_selftest_startup_function(struct tracer *trace, + struct trace_array *tr); +#endif +#ifdef CONFIG_IRQSOFF_TRACER +extern int trace_selftest_startup_irqsoff(struct tracer *trace, + struct trace_array *tr); +#endif +#ifdef CONFIG_PREEMPT_TRACER +extern int trace_selftest_startup_preemptoff(struct tracer *trace, + struct trace_array *tr); +#endif +#if defined(CONFIG_IRQSOFF_TRACER) && defined(CONFIG_PREEMPT_TRACER) +extern int trace_selftest_startup_preemptirqsoff(struct tracer *trace, + struct trace_array *tr); +#endif +#ifdef CONFIG_SCHED_TRACER +extern int trace_selftest_startup_wakeup(struct tracer *trace, + struct trace_array *tr); +#endif +#ifdef CONFIG_CONTEXT_SWITCH_TRACER +extern int trace_selftest_startup_sched_switch(struct tracer *trace, + struct trace_array *tr); +#endif +#endif /* CONFIG_FTRACE_STARTUP_TEST */ + #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 82988c5336e0..5d8ad7a09605 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -63,6 +63,9 @@ static struct tracer function_trace __read_mostly = .init = function_trace_init, .reset = function_trace_reset, .ctrl_update = function_trace_ctrl_update, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_function, +#endif }; static __init int init_function_trace(void) diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 74165f611f36..14183b8f79c5 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -432,6 +432,9 @@ static struct tracer irqsoff_tracer __read_mostly = .close = irqsoff_tracer_close, .ctrl_update = irqsoff_tracer_ctrl_update, .print_max = 1, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_irqsoff, +#endif }; # define register_irqsoff(trace) register_tracer(&trace) #else @@ -455,6 +458,9 @@ static struct tracer preemptoff_tracer __read_mostly = .close = irqsoff_tracer_close, .ctrl_update = irqsoff_tracer_ctrl_update, .print_max = 1, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_preemptoff, +#endif }; # define register_preemptoff(trace) register_tracer(&trace) #else @@ -480,6 +486,9 @@ static struct tracer preemptirqsoff_tracer __read_mostly = .close = irqsoff_tracer_close, .ctrl_update = irqsoff_tracer_ctrl_update, .print_max = 1, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_preemptirqsoff, +#endif }; # define register_preemptirqsoff(trace) register_tracer(&trace) diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 2715267be469..6c9284103a62 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -107,6 +107,9 @@ static struct tracer sched_switch_trace __read_mostly = .init = sched_switch_trace_init, .reset = sched_switch_trace_reset, .ctrl_update = sched_switch_trace_ctrl_update, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_sched_switch, +#endif }; __init static int init_sched_switch_trace(void) diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 7c3ccefcf4c3..3d10ff01f805 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -295,6 +295,9 @@ static struct tracer wakeup_tracer __read_mostly = .close = wakeup_tracer_close, .ctrl_update = wakeup_tracer_ctrl_update, .print_max = 1, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_wakeup, +#endif }; __init static int init_wakeup_tracer(void) diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c new file mode 100644 index 000000000000..ef4d3cc009f5 --- /dev/null +++ b/kernel/trace/trace_selftest.c @@ -0,0 +1,415 @@ +/* Include in trace.c */ + +#include + +static inline int trace_valid_entry(struct trace_entry *entry) +{ + switch (entry->type) { + case TRACE_FN: + case TRACE_CTX: + return 1; + } + return 0; +} + +static int +trace_test_buffer_cpu(struct trace_array *tr, struct trace_array_cpu *data) +{ + struct page *page; + struct trace_entry *entries; + int idx = 0; + int i; + + page = list_entry(data->trace_pages.next, struct page, lru); + entries = page_address(page); + + if (data->trace != entries) + goto failed; + + /* + * The starting trace buffer always has valid elements, + * if any element exits. + */ + entries = data->trace; + + for (i = 0; i < tr->entries; i++) { + + if (i < data->trace_idx && + !trace_valid_entry(&entries[idx])) { + printk(KERN_CONT ".. invalid entry %d ", entries[idx].type); + goto failed; + } + + idx++; + if (idx >= ENTRIES_PER_PAGE) { + page = virt_to_page(entries); + if (page->lru.next == &data->trace_pages) { + if (i != tr->entries - 1) { + printk(KERN_CONT ".. entries buffer mismatch"); + goto failed; + } + } else { + page = list_entry(page->lru.next, struct page, lru); + entries = page_address(page); + } + idx = 0; + } + } + + page = virt_to_page(entries); + if (page->lru.next != &data->trace_pages) { + printk(KERN_CONT ".. too many entries"); + goto failed; + } + + return 0; + + failed: + printk(KERN_CONT ".. corrupted trace buffer .. "); + return -1; +} + +/* + * Test the trace buffer to see if all the elements + * are still sane. + */ +static int trace_test_buffer(struct trace_array *tr, unsigned long *count) +{ + unsigned long cnt = 0; + int cpu; + int ret = 0; + + for_each_possible_cpu(cpu) { + if (!tr->data[cpu]->trace) + continue; + + cnt += tr->data[cpu]->trace_idx; + printk("%d: count = %ld\n", cpu, cnt); + + ret = trace_test_buffer_cpu(tr, tr->data[cpu]); + if (ret) + break; + } + + if (count) + *count = cnt; + + return ret; +} + +#ifdef CONFIG_FTRACE +/* + * Simple verification test of ftrace function tracer. + * Enable ftrace, sleep 1/10 second, and then read the trace + * buffer to see if all is in order. + */ +int +trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) +{ + unsigned long count; + int ret; + + /* make sure functions have been recorded */ + ret = ftrace_force_update(); + if (ret) { + printk(KERN_CONT ".. ftraced failed .. "); + return ret; + } + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* Sleep for a 1/10 of a second */ + msleep(100); + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check the trace buffer */ + ret = trace_test_buffer(tr, &count); + trace->reset(tr); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + return ret; +} +#endif /* CONFIG_FTRACE */ + +#ifdef CONFIG_IRQSOFF_TRACER +int +trace_selftest_startup_irqsoff(struct tracer *trace, struct trace_array *tr) +{ + unsigned long save_max = tracing_max_latency; + unsigned long count; + int ret; + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* reset the max latency */ + tracing_max_latency = 0; + /* disable interrupts for a bit */ + local_irq_disable(); + udelay(100); + local_irq_enable(); + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check both trace buffers */ + ret = trace_test_buffer(tr, NULL); + if (!ret) + ret = trace_test_buffer(&max_tr, &count); + trace->reset(tr); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + tracing_max_latency = save_max; + + return ret; +} +#endif /* CONFIG_IRQSOFF_TRACER */ + +#ifdef CONFIG_PREEMPT_TRACER +int +trace_selftest_startup_preemptoff(struct tracer *trace, struct trace_array *tr) +{ + unsigned long save_max = tracing_max_latency; + unsigned long count; + int ret; + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* reset the max latency */ + tracing_max_latency = 0; + /* disable preemption for a bit */ + preempt_disable(); + udelay(100); + preempt_enable(); + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check both trace buffers */ + ret = trace_test_buffer(tr, NULL); + if (!ret) + ret = trace_test_buffer(&max_tr, &count); + trace->reset(tr); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + tracing_max_latency = save_max; + + return ret; +} +#endif /* CONFIG_PREEMPT_TRACER */ + +#if defined(CONFIG_IRQSOFF_TRACER) && defined(CONFIG_PREEMPT_TRACER) +int +trace_selftest_startup_preemptirqsoff(struct tracer *trace, struct trace_array *tr) +{ + unsigned long save_max = tracing_max_latency; + unsigned long count; + int ret; + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + + /* reset the max latency */ + tracing_max_latency = 0; + + /* disable preemption and interrupts for a bit */ + preempt_disable(); + local_irq_disable(); + udelay(100); + preempt_enable(); + /* reverse the order of preempt vs irqs */ + local_irq_enable(); + + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check both trace buffers */ + ret = trace_test_buffer(tr, NULL); + if (ret) + goto out; + + ret = trace_test_buffer(&max_tr, &count); + if (ret) + goto out; + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + goto out; + } + + /* do the test by disabling interrupts first this time */ + tracing_max_latency = 0; + tr->ctrl = 1; + trace->ctrl_update(tr); + preempt_disable(); + local_irq_disable(); + udelay(100); + preempt_enable(); + /* reverse the order of preempt vs irqs */ + local_irq_enable(); + + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check both trace buffers */ + ret = trace_test_buffer(tr, NULL); + if (ret) + goto out; + + ret = trace_test_buffer(&max_tr, &count); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + goto out; + } + + out: + trace->reset(tr); + tracing_max_latency = save_max; + + return ret; +} +#endif /* CONFIG_IRQSOFF_TRACER && CONFIG_PREEMPT_TRACER */ + +#ifdef CONFIG_SCHED_TRACER +static int trace_wakeup_test_thread(void *data) +{ + struct completion *x = data; + + /* Make this a RT thread, doesn't need to be too high */ + + rt_mutex_setprio(current, MAX_RT_PRIO - 5); + + /* Make it know we have a new prio */ + complete(x); + + /* now go to sleep and let the test wake us up */ + set_current_state(TASK_INTERRUPTIBLE); + schedule(); + + /* we are awake, now wait to disappear */ + while (!kthread_should_stop()) { + /* + * This is an RT task, do short sleeps to let + * others run. + */ + msleep(100); + } + + return 0; +} + +int +trace_selftest_startup_wakeup(struct tracer *trace, struct trace_array *tr) +{ + unsigned long save_max = tracing_max_latency; + struct task_struct *p; + struct completion isrt; + unsigned long count; + int ret; + + init_completion(&isrt); + + /* create a high prio thread */ + p = kthread_run(trace_wakeup_test_thread, &isrt, "ftrace-test"); + if (!IS_ERR(p)) { + printk(KERN_CONT "Failed to create ftrace wakeup test thread "); + return -1; + } + + /* make sure the thread is running at an RT prio */ + wait_for_completion(&isrt); + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* reset the max latency */ + tracing_max_latency = 0; + + /* sleep to let the RT thread sleep too */ + msleep(100); + + /* + * Yes this is slightly racy. It is possible that for some + * strange reason that the RT thread we created, did not + * call schedule for 100ms after doing the completion, + * and we do a wakeup on a task that already is awake. + * But that is extremely unlikely, and the worst thing that + * happens in such a case, is that we disable tracing. + * Honestly, if this race does happen something is horrible + * wrong with the system. + */ + + wake_up_process(p); + + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check both trace buffers */ + ret = trace_test_buffer(tr, NULL); + if (!ret) + ret = trace_test_buffer(&max_tr, &count); + + + trace->reset(tr); + + tracing_max_latency = save_max; + + /* kill the thread */ + kthread_stop(p); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + return ret; +} +#endif /* CONFIG_SCHED_TRACER */ + +#ifdef CONFIG_CONTEXT_SWITCH_TRACER +int +trace_selftest_startup_sched_switch(struct tracer *trace, struct trace_array *tr) +{ + unsigned long count; + int ret; + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* Sleep for a 1/10 of a second */ + msleep(100); + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check the trace buffer */ + ret = trace_test_buffer(tr, &count); + trace->reset(tr); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + return ret; +} +#endif /* CONFIG_CONTEXT_SWITCH_TRACER */ + +#ifdef CONFIG_DYNAMIC_FTRACE +#endif /* CONFIG_DYNAMIC_FTRACE */ -- cgit v1.2.3 From c7aafc549766b87819285d3480648fc652a47bc4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: cleanups factor out code and clean it up. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 8 +-- kernel/trace/trace.c | 144 ++++++++++++++++++++++++-------------- kernel/trace/trace.h | 8 ++- kernel/trace/trace_irqsoff.c | 32 ++++----- kernel/trace/trace_sched_wakeup.c | 18 ++--- kernel/trace/trace_selftest.c | 25 ++++--- 6 files changed, 133 insertions(+), 102 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 4facf5ceeb86..6d4d2e86debc 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1152,10 +1152,10 @@ static int __init notrace ftrace_dynamic_init(void) core_initcall(ftrace_dynamic_init); #else -# define ftrace_startup() do { } while (0) -# define ftrace_shutdown() do { } while (0) -# define ftrace_startup_sysctl() do { } while (0) -# define ftrace_shutdown_sysctl() do { } while (0) +# define ftrace_startup() do { } while (0) +# define ftrace_shutdown() do { } while (0) +# define ftrace_startup_sysctl() do { } while (0) +# define ftrace_shutdown_sysctl() do { } while (0) #endif /* CONFIG_DYNAMIC_FTRACE */ /** diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f6d026f17dbb..61d2f0228866 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -142,12 +142,59 @@ __update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) tracing_record_cmdline(current); } +void check_pages(struct trace_array_cpu *data) +{ + struct page *page, *tmp; + + BUG_ON(data->trace_pages.next->prev != &data->trace_pages); + BUG_ON(data->trace_pages.prev->next != &data->trace_pages); + + list_for_each_entry_safe(page, tmp, &data->trace_pages, lru) { + BUG_ON(page->lru.next->prev != &page->lru); + BUG_ON(page->lru.prev->next != &page->lru); + } +} + +void *head_page(struct trace_array_cpu *data) +{ + struct page *page; + + check_pages(data); + if (list_empty(&data->trace_pages)) + return NULL; + + page = list_entry(data->trace_pages.next, struct page, lru); + BUG_ON(&page->lru == &data->trace_pages); + + return page_address(page); +} + +notrace static void +flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) +{ + struct list_head flip_pages; + + INIT_LIST_HEAD(&flip_pages); + + tr1->trace_current = NULL; + memcpy(&tr1->trace_current_idx, &tr2->trace_current_idx, + sizeof(struct trace_array_cpu) - + offsetof(struct trace_array_cpu, trace_current_idx)); + + check_pages(tr1); + check_pages(tr2); + list_splice_init(&tr1->trace_pages, &flip_pages); + list_splice_init(&tr2->trace_pages, &tr1->trace_pages); + list_splice_init(&flip_pages, &tr2->trace_pages); + BUG_ON(!list_empty(&flip_pages)); + check_pages(tr1); + check_pages(tr2); +} + notrace void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data; - void *save_trace; - struct list_head save_pages; int i; WARN_ON_ONCE(!irqs_disabled()); @@ -155,11 +202,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) /* clear out all the previous traces */ for_each_possible_cpu(i) { data = tr->data[i]; - save_trace = max_tr.data[i]->trace; - save_pages = max_tr.data[i]->trace_pages; - memcpy(max_tr.data[i], data, sizeof(*data)); - data->trace = save_trace; - data->trace_pages = save_pages; + flip_trace(max_tr.data[i], data); tracing_reset(data); } @@ -177,8 +220,6 @@ notrace void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data = tr->data[cpu]; - void *save_trace; - struct list_head save_pages; int i; WARN_ON_ONCE(!irqs_disabled()); @@ -186,11 +227,8 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) for_each_possible_cpu(i) tracing_reset(max_tr.data[i]); - save_trace = max_tr.data[cpu]->trace; - save_pages = max_tr.data[cpu]->trace_pages; - memcpy(max_tr.data[cpu], data, sizeof(*data)); - data->trace = save_trace; - data->trace_pages = save_pages; + flip_trace(max_tr.data[cpu], data); + tracing_reset(data); __update_max_tr(tr, tsk, cpu); @@ -234,9 +272,9 @@ int register_tracer(struct tracer *type) * If we fail, we do not register this tracer. */ for_each_possible_cpu(i) { - if (!data->trace) - continue; data = tr->data[i]; + if (!head_page(data)) + continue; tracing_reset(data); } current_trace = type; @@ -298,7 +336,7 @@ void unregister_tracer(struct tracer *type) void notrace tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; - data->trace_current = data->trace; + data->trace_current = head_page(data); data->trace_current_idx = 0; } @@ -425,26 +463,31 @@ notrace void tracing_record_cmdline(struct task_struct *tsk) } static inline notrace struct trace_entry * -tracing_get_trace_entry(struct trace_array *tr, - struct trace_array_cpu *data) +tracing_get_trace_entry(struct trace_array *tr, struct trace_array_cpu *data) { unsigned long idx, idx_next; struct trace_entry *entry; - struct page *page; struct list_head *next; + struct page *page; data->trace_idx++; idx = data->trace_current_idx; idx_next = idx + 1; + BUG_ON(idx * TRACE_ENTRY_SIZE >= PAGE_SIZE); + entry = data->trace_current + idx * TRACE_ENTRY_SIZE; if (unlikely(idx_next >= ENTRIES_PER_PAGE)) { page = virt_to_page(data->trace_current); - if (unlikely(&page->lru == data->trace_pages.prev)) - next = data->trace_pages.next; - else - next = page->lru.next; + /* + * Roundrobin - but skip the head (which is not a real page): + */ + next = page->lru.next; + if (unlikely(next == &data->trace_pages)) + next = next->next; + BUG_ON(next == &data->trace_pages); + page = list_entry(next, struct page, lru); data->trace_current = page_address(page); idx_next = 0; @@ -456,18 +499,17 @@ tracing_get_trace_entry(struct trace_array *tr, } static inline notrace void -tracing_generic_entry_update(struct trace_entry *entry, - unsigned long flags) +tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) { struct task_struct *tsk = current; unsigned long pc; pc = preempt_count(); - entry->idx = atomic_inc_return(&tracer_counter); - entry->preempt_count = pc & 0xff; - entry->pid = tsk->pid; - entry->t = now(raw_smp_processor_id()); + entry->idx = atomic_inc_return(&tracer_counter); + entry->preempt_count = pc & 0xff; + entry->pid = tsk->pid; + entry->t = now(raw_smp_processor_id()); entry->flags = (irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) | ((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) | ((pc & SOFTIRQ_MASK) ? TRACE_FLAG_SOFTIRQ : 0) | @@ -476,16 +518,15 @@ tracing_generic_entry_update(struct trace_entry *entry, notrace void ftrace(struct trace_array *tr, struct trace_array_cpu *data, - unsigned long ip, unsigned long parent_ip, - unsigned long flags) + unsigned long ip, unsigned long parent_ip, unsigned long flags) { struct trace_entry *entry; - entry = tracing_get_trace_entry(tr, data); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); - entry->type = TRACE_FN; - entry->fn.ip = ip; - entry->fn.parent_ip = parent_ip; + entry->type = TRACE_FN; + entry->fn.ip = ip; + entry->fn.parent_ip = parent_ip; } notrace void @@ -496,7 +537,7 @@ tracing_sched_switch_trace(struct trace_array *tr, { struct trace_entry *entry; - entry = tracing_get_trace_entry(tr, data); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_CTX; entry->ctx.prev_pid = prev->pid; @@ -540,6 +581,8 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, } page = list_entry(iter->next_page[cpu], struct page, lru); + BUG_ON(&data->trace_pages == &page->lru); + array = page_address(page); return &array[iter->next_page_idx[cpu]]; @@ -554,7 +597,7 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) int cpu; for_each_possible_cpu(cpu) { - if (!tr->data[cpu]->trace) + if (!head_page(tr->data[cpu])) continue; ent = trace_entry_idx(tr, tr->data[cpu], iter, cpu); if (ent && @@ -762,7 +805,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) name = type->name; for_each_possible_cpu(cpu) { - if (tr->data[cpu]->trace) { + if (head_page(tr->data[cpu])) { total += tr->data[cpu]->trace_idx; if (tr->data[cpu]->trace_idx > tr->entries) entries += tr->entries; @@ -975,8 +1018,7 @@ static int trace_empty(struct trace_iterator *iter) for_each_possible_cpu(cpu) { data = iter->tr->data[cpu]; - if (data->trace && - data->trace_idx) + if (head_page(data) && data->trace_idx) return 0; } return 1; @@ -1576,9 +1618,9 @@ static struct tracer no_tracer __read_mostly = static int trace_alloc_page(void) { struct trace_array_cpu *data; - void *array; struct page *page, *tmp; LIST_HEAD(pages); + void *array; int i; /* first allocate a page for each CPU */ @@ -1610,14 +1652,14 @@ static int trace_alloc_page(void) for_each_possible_cpu(i) { data = global_trace.data[i]; page = list_entry(pages.next, struct page, lru); - list_del(&page->lru); + list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); ClearPageLRU(page); #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; page = list_entry(pages.next, struct page, lru); - list_del(&page->lru); + list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); SetPageLRU(page); #endif @@ -1628,7 +1670,7 @@ static int trace_alloc_page(void) free_pages: list_for_each_entry_safe(page, tmp, &pages, lru) { - list_del(&page->lru); + list_del_init(&page->lru); __free_page(page); } return -ENOMEM; @@ -1654,7 +1696,6 @@ __init static int tracer_alloc_buffers(void) "for trace buffer!\n"); goto free_buffers; } - data->trace = array; /* set the array to the list */ INIT_LIST_HEAD(&data->trace_pages); @@ -1671,7 +1712,6 @@ __init static int tracer_alloc_buffers(void) "for trace buffer!\n"); goto free_buffers; } - max_tr.data[i]->trace = array; INIT_LIST_HEAD(&max_tr.data[i]->trace_pages); page = virt_to_page(array); @@ -1716,24 +1756,22 @@ __init static int tracer_alloc_buffers(void) struct page *page, *tmp; struct trace_array_cpu *data = global_trace.data[i]; - if (data && data->trace) { + if (data) { list_for_each_entry_safe(page, tmp, &data->trace_pages, lru) { - list_del(&page->lru); + list_del_init(&page->lru); __free_page(page); } - data->trace = NULL; } #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; - if (data && data->trace) { + if (data) { list_for_each_entry_safe(page, tmp, &data->trace_pages, lru) { - list_del(&page->lru); + list_del_init(&page->lru); __free_page(page); } - data->trace = NULL; } #endif } diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 88edbf1f6788..cc1d34b8b771 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -53,12 +53,12 @@ struct trace_entry { * the trace, etc.) */ struct trace_array_cpu { - void *trace; void *trace_current; - unsigned trace_current_idx; struct list_head trace_pages; - unsigned long trace_idx; atomic_t disabled; + /* these fields get copied into max-trace: */ + unsigned trace_current_idx; + unsigned long trace_idx; unsigned long saved_latency; unsigned long critical_start; unsigned long critical_end; @@ -216,4 +216,6 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace, #endif #endif /* CONFIG_FTRACE_STARTUP_TEST */ +extern void *head_page(struct trace_array_cpu *data); + #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 14183b8f79c5..2dfebb67fdfb 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -144,7 +144,7 @@ check_critical_timing(struct trace_array *tr, if (!report_latency(delta)) goto out; - spin_lock(&max_trace_lock); + spin_lock_irqsave(&max_trace_lock, flags); /* check if we are still the max latency */ if (!report_latency(delta)) @@ -165,32 +165,24 @@ check_critical_timing(struct trace_array *tr, update_max_tr_single(tr, current, cpu); - if (tracing_thresh) - printk(KERN_INFO "(%16s-%-5d|#%d): %lu us critical section " - "violates %lu us threshold.\n" - " => started at timestamp %lu: ", + if (tracing_thresh) { + printk(KERN_INFO "(%16s-%-5d|#%d):" + " %lu us critical section violates %lu us threshold.\n", current->comm, current->pid, raw_smp_processor_id(), - latency, nsecs_to_usecs(tracing_thresh), t0); - else + latency, nsecs_to_usecs(tracing_thresh)); + } else { printk(KERN_INFO "(%16s-%-5d|#%d):" - " new %lu us maximum-latency " - "critical section.\n => started at timestamp %lu: ", + " new %lu us maximum-latency critical section.\n", current->comm, current->pid, raw_smp_processor_id(), - latency, t0); - - print_symbol(KERN_CONT "<%s>\n", data->critical_start); - printk(KERN_CONT " => ended at timestamp %lu: ", t1); - print_symbol(KERN_CONT "<%s>\n", data->critical_end); - dump_stack(); - t1 = nsecs_to_usecs(now(cpu)); - printk(KERN_CONT " => dump-end timestamp %lu\n\n", t1); + latency); + } max_sequence++; out_unlock: - spin_unlock(&max_trace_lock); + spin_unlock_irqrestore(&max_trace_lock, flags); out: data->critical_sequence = max_sequence; @@ -216,7 +208,7 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) cpu = raw_smp_processor_id(); data = tr->data[cpu]; - if (unlikely(!data) || unlikely(!data->trace) || + if (unlikely(!data) || unlikely(!head_page(data)) || atomic_read(&data->disabled)) return; @@ -256,7 +248,7 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) cpu = raw_smp_processor_id(); data = tr->data[cpu]; - if (unlikely(!data) || unlikely(!data->trace) || + if (unlikely(!data) || unlikely(!head_page(data)) || !data->critical_start || atomic_read(&data->disabled)) return; diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 3d10ff01f805..688df965f3f2 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -107,24 +107,18 @@ wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) update_max_tr(tr, wakeup_task, wakeup_cpu); if (tracing_thresh) { - printk(KERN_INFO "(%16s-%-5d|#%d): %lu us wakeup latency " - "violates %lu us threshold.\n" - " => started at timestamp %lu: ", + printk(KERN_INFO "(%16s-%-5d|#%d):" + " %lu us wakeup latency violates %lu us threshold.\n", wakeup_task->comm, wakeup_task->pid, raw_smp_processor_id(), - latency, nsecs_to_usecs(tracing_thresh), t0); + latency, nsecs_to_usecs(tracing_thresh)); } else { - printk(KERN_INFO "(%16s-%-5d|#%d): new %lu us maximum " - "wakeup latency.\n => started at timestamp %lu: ", + printk(KERN_INFO "(%16s-%-5d|#%d):" + " new %lu us maximum wakeup latency.\n", wakeup_task->comm, wakeup_task->pid, - cpu, latency, t0); + cpu, latency); } - printk(KERN_CONT " ended at timestamp %lu: ", t1); - dump_stack(); - t1 = nsecs_to_usecs(now(cpu)); - printk(KERN_CONT " dump-end timestamp %lu\n\n", t1); - out_unlock: __wakeup_reset(tr); spin_unlock_irqrestore(&wakeup_lock, flags); diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index ef4d3cc009f5..c01874c3b1f9 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -1,6 +1,7 @@ /* Include in trace.c */ #include +#include static inline int trace_valid_entry(struct trace_entry *entry) { @@ -15,28 +16,29 @@ static inline int trace_valid_entry(struct trace_entry *entry) static int trace_test_buffer_cpu(struct trace_array *tr, struct trace_array_cpu *data) { - struct page *page; struct trace_entry *entries; + struct page *page; int idx = 0; int i; + BUG_ON(list_empty(&data->trace_pages)); page = list_entry(data->trace_pages.next, struct page, lru); entries = page_address(page); - if (data->trace != entries) + if (head_page(data) != entries) goto failed; /* * The starting trace buffer always has valid elements, - * if any element exits. + * if any element exists. */ - entries = data->trace; + entries = head_page(data); for (i = 0; i < tr->entries; i++) { - if (i < data->trace_idx && - !trace_valid_entry(&entries[idx])) { - printk(KERN_CONT ".. invalid entry %d ", entries[idx].type); + if (i < data->trace_idx && !trace_valid_entry(&entries[idx])) { + printk(KERN_CONT ".. invalid entry %d ", + entries[idx].type); goto failed; } @@ -80,11 +82,10 @@ static int trace_test_buffer(struct trace_array *tr, unsigned long *count) int ret = 0; for_each_possible_cpu(cpu) { - if (!tr->data[cpu]->trace) + if (!head_page(tr->data[cpu])) continue; cnt += tr->data[cpu]->trace_idx; - printk("%d: count = %ld\n", cpu, cnt); ret = trace_test_buffer_cpu(tr, tr->data[cpu]); if (ret) @@ -117,6 +118,8 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) } /* start the tracing */ + ftrace_enabled = 1; + tr->ctrl = 1; trace->init(tr); /* Sleep for a 1/10 of a second */ @@ -124,6 +127,8 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) /* stop the tracing. */ tr->ctrl = 0; trace->ctrl_update(tr); + ftrace_enabled = 0; + /* check the trace buffer */ ret = trace_test_buffer(tr, &count); trace->reset(tr); @@ -328,7 +333,7 @@ trace_selftest_startup_wakeup(struct tracer *trace, struct trace_array *tr) /* create a high prio thread */ p = kthread_run(trace_wakeup_test_thread, &isrt, "ftrace-test"); - if (!IS_ERR(p)) { + if (IS_ERR(p)) { printk(KERN_CONT "Failed to create ftrace wakeup test thread "); return -1; } -- cgit v1.2.3 From 7bd2f24c2f769e3f8f1d4fc8b9fddf689825f6a7 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: add README make it easier for newbies to find their way around. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 40 +++++++++++++++++++++++++++++++++++++++- 1 file changed, 39 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 61d2f0228866..c736dd2e068d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -995,6 +995,7 @@ print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) seq_printf(m, " <-"); seq_print_ip_sym(m, entry->fn.parent_ip, sym_flags); } + seq_printf(m, "\n"); break; case TRACE_CTX: S = entry->ctx.prev_state < sizeof(state_to_char) ? @@ -1007,7 +1008,6 @@ print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) entry->ctx.next_prio); break; } - seq_printf(m, "\n"); } static int trace_empty(struct trace_iterator *iter) @@ -1332,6 +1332,39 @@ static struct file_operations tracing_iter_fops = { .write = tracing_iter_ctrl_write, }; +static const char readme_msg[] = + "tracing mini-HOWTO:\n\n" + "# mkdir /debug\n" + "# mount -t debugfs nodev /debug\n\n" + "# cat /debug/tracing/available_tracers\n" + "wakeup preemptirqsoff preemptoff irqsoff ftrace sched_switch none\n\n" + "# cat /debug/tracing/current_tracer\n" + "none\n" + "# echo sched_switch > /debug/tracing/current_tracer\n" + "# cat /debug/tracing/current_tracer\n" + "sched_switch\n" + "# cat /debug/tracing/iter_ctrl\n" + "noprint-parent nosym-offset nosym-addr noverbose\n" + "# echo print-parent > /debug/tracing/iter_ctrl\n" + "# echo 1 > /debug/tracing/tracing_enabled\n" + "# cat /debug/tracing/trace > /tmp/trace.txt\n" + "echo 0 > /debug/tracing/tracing_enabled\n" +; + +static ssize_t +tracing_readme_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + return simple_read_from_buffer(ubuf, cnt, ppos, + readme_msg, strlen(readme_msg)); +} + +static struct file_operations tracing_readme_fops = { + .open = tracing_open_generic, + .read = tracing_readme_read, +}; + + static ssize_t tracing_ctrl_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) @@ -1598,6 +1631,11 @@ static __init void tracer_init_debugfs(void) if (!entry) pr_warning("Could not create debugfs " "'tracing_threash' entry\n"); + entry = debugfs_create_file("README", 0644, d_tracer, + NULL, &tracing_readme_fops); + if (!entry) + pr_warning("Could not create debugfs 'README' entry\n"); + #ifdef CONFIG_DYNAMIC_FTRACE entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer, -- cgit v1.2.3 From 77a2b37d227483fe52aead242652aee406c25bf0 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: startup tester on dynamic tracing. This patch adds a startup self test on dynamic code modification and filters. The test filters on a specific function, makes sure that no other function is traced, exectutes the function, then makes sure that the function is traced. This patch also fixes a slight bug with the ftrace selftest, where tracer_enabled was not being set. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 19 +++++++ kernel/trace/trace_selftest.c | 113 ++++++++++++++++++++++++++++++++++++++++-- 2 files changed, 128 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 6d4d2e86debc..5e9389faaf75 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1010,6 +1010,25 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, return ret; } +/** + * ftrace_set_filter - set a function to filter on in ftrace + * @buf - the string that holds the function filter text. + * @len - the length of the string. + * @reset - non zero to reset all filters before applying this filter. + * + * Filters denote which functions should be enabled when tracing is enabled. + * If @buf is NULL and reset is set, all functions will be enabled for tracing. + */ +notrace void ftrace_set_filter(unsigned char *buf, int len, int reset) +{ + mutex_lock(&ftrace_filter_lock); + if (reset) + ftrace_filter_reset(); + if (buf) + ftrace_match(buf, len); + mutex_unlock(&ftrace_filter_lock); +} + static int notrace ftrace_filter_release(struct inode *inode, struct file *file) { diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index c01874c3b1f9..4c8a1b2d8231 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -99,6 +99,100 @@ static int trace_test_buffer(struct trace_array *tr, unsigned long *count) } #ifdef CONFIG_FTRACE + +#ifdef CONFIG_DYNAMIC_FTRACE + +#define DYN_FTRACE_TEST_NAME trace_selftest_dynamic_test_func +#define __STR(x) #x +#define STR(x) __STR(x) +static int DYN_FTRACE_TEST_NAME(void) +{ + /* used to call mcount */ + return 0; +} + +/* Test dynamic code modification and ftrace filters */ +int trace_selftest_startup_dynamic_tracing(struct tracer *trace, + struct trace_array *tr, + int (*func)(void)) +{ + unsigned long count; + int ret; + int save_ftrace_enabled = ftrace_enabled; + int save_tracer_enabled = tracer_enabled; + + /* The ftrace test PASSED */ + printk(KERN_CONT "PASSED\n"); + pr_info("Testing dynamic ftrace: "); + + /* enable tracing, and record the filter function */ + ftrace_enabled = 1; + tracer_enabled = 1; + + /* passed in by parameter to fool gcc from optimizing */ + func(); + + /* update the records */ + ret = ftrace_force_update(); + if (ret) { + printk(KERN_CONT ".. ftraced failed .. "); + return ret; + } + + /* filter only on our function */ + ftrace_set_filter(STR(DYN_FTRACE_TEST_NAME), + sizeof(STR(DYN_FTRACE_TEST_NAME)), 1); + + /* enable tracing */ + tr->ctrl = 1; + trace->init(tr); + /* Sleep for a 1/10 of a second */ + msleep(100); + + /* we should have nothing in the buffer */ + ret = trace_test_buffer(tr, &count); + if (ret) + goto out; + + if (count) { + ret = -1; + printk(KERN_CONT ".. filter did not filter .. "); + goto out; + } + + /* call our function again */ + func(); + + /* sleep again */ + msleep(100); + + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + ftrace_enabled = 0; + + /* check the trace buffer */ + ret = trace_test_buffer(tr, &count); + trace->reset(tr); + + /* we should only have one item */ + if (!ret && count != 1) { + printk(KERN_CONT ".. filter failed .."); + ret = -1; + goto out; + } + out: + ftrace_enabled = save_ftrace_enabled; + tracer_enabled = save_tracer_enabled; + + /* Enable tracing on all functions again */ + ftrace_set_filter(NULL, 0, 1); + + return ret; +} +#else +# define trace_selftest_startup_dynamic_tracing(trace, tr, func) ({ 0; }) +#endif /* CONFIG_DYNAMIC_FTRACE */ /* * Simple verification test of ftrace function tracer. * Enable ftrace, sleep 1/10 second, and then read the trace @@ -109,8 +203,13 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) { unsigned long count; int ret; + int save_ftrace_enabled = ftrace_enabled; + int save_tracer_enabled = tracer_enabled; - /* make sure functions have been recorded */ + /* make sure msleep has been recorded */ + msleep(1); + + /* force the recorded functions to be traced */ ret = ftrace_force_update(); if (ret) { printk(KERN_CONT ".. ftraced failed .. "); @@ -119,6 +218,7 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) /* start the tracing */ ftrace_enabled = 1; + tracer_enabled = 1; tr->ctrl = 1; trace->init(tr); @@ -136,8 +236,16 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) if (!ret && !count) { printk(KERN_CONT ".. no entries found .."); ret = -1; + goto out; } + ret = trace_selftest_startup_dynamic_tracing(trace, tr, + DYN_FTRACE_TEST_NAME); + + out: + ftrace_enabled = save_ftrace_enabled; + tracer_enabled = save_tracer_enabled; + return ret; } #endif /* CONFIG_FTRACE */ @@ -415,6 +523,3 @@ trace_selftest_startup_sched_switch(struct tracer *trace, struct trace_array *tr return ret; } #endif /* CONFIG_CONTEXT_SWITCH_TRACER */ - -#ifdef CONFIG_DYNAMIC_FTRACE -#endif /* CONFIG_DYNAMIC_FTRACE */ -- cgit v1.2.3 From 4e3c3333f3bd7eedfd21b1155b3c7cd24fc7f754 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: fix time offset fix time offset calculations and ordering, plus make code more consistent. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 75 ++++++++++++++++++++++++++++++++++++++-------------- kernel/trace/trace.h | 9 ++++++- 2 files changed, 63 insertions(+), 21 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index c736dd2e068d..8755a4370484 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -120,7 +120,7 @@ static DEFINE_SPINLOCK(ftrace_max_lock); * structure. (this way the maximum trace is permanently saved, * for later retrieval via /debugfs/tracing/latency_trace) */ -static void notrace +static notrace void __update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data = tr->data[cpu]; @@ -333,15 +333,16 @@ void unregister_tracer(struct tracer *type) mutex_unlock(&trace_types_lock); } -void notrace tracing_reset(struct trace_array_cpu *data) +notrace void tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; data->trace_current = head_page(data); data->trace_current_idx = 0; + data->time_offset = 0; } #ifdef CONFIG_FTRACE -static void notrace +static notrace void function_trace_call(unsigned long ip, unsigned long parent_ip) { struct trace_array *tr = &global_trace; @@ -398,7 +399,7 @@ static void trace_init_cmdlines(void) notrace void trace_stop_cmdline_recording(void); -static void notrace trace_save_cmdline(struct task_struct *tsk) +static notrace void trace_save_cmdline(struct task_struct *tsk) { unsigned map; unsigned idx; @@ -624,6 +625,7 @@ static void *find_next_entry_inc(struct trace_iterator *iter) iter->idx++; iter->next_idx[next_cpu]++; iter->next_page_idx[next_cpu]++; + if (iter->next_page_idx[next_cpu] >= ENTRIES_PER_PAGE) { struct trace_array_cpu *data = iter->tr->data[next_cpu]; @@ -635,19 +637,21 @@ static void *find_next_entry_inc(struct trace_iterator *iter) data->trace_pages.next; } } + iter->prev_ent = iter->ent; + iter->prev_cpu = iter->cpu; + iter->ent = next; iter->cpu = next_cpu; return next ? iter : NULL; } -static void notrace * -s_next(struct seq_file *m, void *v, loff_t *pos) +static notrace void *s_next(struct seq_file *m, void *v, loff_t *pos) { struct trace_iterator *iter = m->private; - void *ent; void *last_ent = iter->ent; int i = (int)*pos; + void *ent; (*pos)++; @@ -693,6 +697,8 @@ static void *s_start(struct seq_file *m, loff_t *pos) iter->ent = NULL; iter->cpu = 0; iter->idx = -1; + iter->prev_ent = NULL; + iter->prev_cpu = -1; for_each_possible_cpu(i) { iter->next_idx[i] = 0; @@ -752,7 +758,7 @@ seq_print_sym_offset(struct seq_file *m, const char *fmt, unsigned long address) # define IP_FMT "%016lx" #endif -static void notrace +static notrace void seq_print_ip_sym(struct seq_file *m, unsigned long ip, unsigned long sym_flags) { if (!ip) { @@ -769,7 +775,7 @@ seq_print_ip_sym(struct seq_file *m, unsigned long ip, unsigned long sym_flags) seq_printf(m, " <" IP_FMT ">", ip); } -static void notrace print_lat_help_header(struct seq_file *m) +static notrace void print_lat_help_header(struct seq_file *m) { seq_puts(m, "# _------=> CPU# \n"); seq_puts(m, "# / _-----=> irqs-off \n"); @@ -782,14 +788,14 @@ static void notrace print_lat_help_header(struct seq_file *m) seq_puts(m, "# \\ / ||||| \\ | / \n"); } -static void notrace print_func_help_header(struct seq_file *m) +static notrace void print_func_help_header(struct seq_file *m) { seq_puts(m, "# TASK-PID CPU# TIMESTAMP FUNCTION\n"); seq_puts(m, "# | | | | |\n"); } -static void notrace +static notrace void print_trace_header(struct seq_file *m, struct trace_iterator *iter) { unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); @@ -858,7 +864,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) seq_puts(m, "\n"); } -static void notrace +static notrace void lat_print_generic(struct seq_file *m, struct trace_entry *entry, int cpu) { int hardirq, softirq; @@ -895,7 +901,7 @@ lat_print_generic(struct seq_file *m, struct trace_entry *entry, int cpu) unsigned long preempt_mark_thresh = 100; -static void notrace +static notrace void lat_print_timestamp(struct seq_file *m, unsigned long long abs_usecs, unsigned long rel_usecs) { @@ -910,7 +916,7 @@ lat_print_timestamp(struct seq_file *m, unsigned long long abs_usecs, static const char state_to_char[] = TASK_STATE_TO_CHAR_STR; -static void notrace +static notrace void print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, unsigned int trace_idx, int cpu) { @@ -966,20 +972,50 @@ print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, } } -static void notrace +static notrace void sync_time_offset(struct trace_iterator *iter) +{ + struct trace_array_cpu *prev_array, *array; + struct trace_entry *prev_entry, *entry; + cycle_t prev_t, t; + + entry = iter->ent; + prev_entry = iter->prev_ent; + if (!prev_entry) + return; + + prev_array = iter->tr->data[iter->prev_cpu]; + array = iter->tr->data[iter->cpu]; + + prev_t = prev_entry->t + prev_array->time_offset; + t = entry->t + array->time_offset; + + /* + * If time goes backwards we increase the offset of + * the current array, to not have observable time warps. + * This will quickly synchronize the time offsets of + * multiple CPUs: + */ + if (t < prev_t) + array->time_offset += prev_t - t; +} + +static notrace void print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) { unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); - struct trace_entry *entry = iter->ent; + struct trace_entry *entry; unsigned long usec_rem; unsigned long long t; unsigned long secs; char *comm; int S; + sync_time_offset(iter); + entry = iter->ent; + comm = trace_find_cmdline(iter->ent->pid); - t = ns2usecs(entry->t); + t = ns2usecs(entry->t + iter->tr->data[iter->cpu]->time_offset); usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; @@ -1158,7 +1194,7 @@ static int tracing_lt_open(struct inode *inode, struct file *file) } -static void notrace * +static notrace void * t_next(struct seq_file *m, void *v, loff_t *pos) { struct tracer *t = m->private; @@ -1374,8 +1410,7 @@ tracing_ctrl_read(struct file *filp, char __user *ubuf, int r; r = sprintf(buf, "%ld\n", tr->ctrl); - return simple_read_from_buffer(ubuf, cnt, ppos, - buf, r); + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } static ssize_t diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index cc1d34b8b771..5df8ff2b84a7 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -56,6 +56,8 @@ struct trace_array_cpu { void *trace_current; struct list_head trace_pages; atomic_t disabled; + cycle_t time_offset; + /* these fields get copied into max-trace: */ unsigned trace_current_idx; unsigned long trace_idx; @@ -114,14 +116,19 @@ struct tracer { struct trace_iterator { struct trace_array *tr; struct tracer *trace; + struct trace_entry *ent; + int cpu; + + struct trace_entry *prev_ent; + int prev_cpu; + unsigned long iter_flags; loff_t pos; unsigned long next_idx[NR_CPUS]; struct list_head *next_page[NR_CPUS]; unsigned next_page_idx[NR_CPUS]; long idx; - int cpu; }; void notrace tracing_reset(struct trace_array_cpu *data); -- cgit v1.2.3 From 08bafa0efcf29fe18ec39c2147077b597368b018 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: disable all tracers on corrupted buffer If the trace buffer is detected to be corrupted, then we disable all tracers. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_selftest.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 4c8a1b2d8231..a6f1ed75f836 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -67,6 +67,8 @@ trace_test_buffer_cpu(struct trace_array *tr, struct trace_array_cpu *data) return 0; failed: + /* disable tracing */ + tracing_disabled = 1; printk(KERN_CONT ".. corrupted trace buffer .. "); return -1; } -- cgit v1.2.3 From 1d4db00a5e30c7b8f8dc2a1b19e886fd942be143 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: reset selftests The tests may leave stuff in the buffers. This resets the buffers after each test is run. If a test fails, it does not reset the buffer to avoid touching a buffer that is corrupted. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 8755a4370484..6580e7ed04be 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -289,6 +289,13 @@ int register_tracer(struct tracer *type) printk(KERN_CONT "FAILED!\n"); goto out; } + /* Only reset on passing, to avoid touching corrupted buffers */ + for_each_possible_cpu(i) { + data = tr->data[i]; + if (!head_page(data)) + continue; + tracing_reset(data); + } printk(KERN_CONT "PASSED\n"); } #endif -- cgit v1.2.3 From 93a588f459da134be6ab17c4104e28441beb0d22 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:45 +0200 Subject: ftrace: change buffers to producer consumer This patch changes the way the CPU trace buffers are handled. Instead of always starting from the trace page head, the logic is changed to a producer consumer logic. This allows for the buffers to be drained while they are alive. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 103 ++++++++++++++++++++++++++++++--------------------- kernel/trace/trace.h | 6 ++- 2 files changed, 65 insertions(+), 44 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 6580e7ed04be..777b859e1c2e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -176,10 +176,9 @@ flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) INIT_LIST_HEAD(&flip_pages); - tr1->trace_current = NULL; - memcpy(&tr1->trace_current_idx, &tr2->trace_current_idx, + memcpy(&tr1->trace_head_idx, &tr2->trace_head_idx, sizeof(struct trace_array_cpu) - - offsetof(struct trace_array_cpu, trace_current_idx)); + offsetof(struct trace_array_cpu, trace_head_idx)); check_pages(tr1); check_pages(tr2); @@ -228,7 +227,6 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) tracing_reset(max_tr.data[i]); flip_trace(max_tr.data[cpu], data); - tracing_reset(data); __update_max_tr(tr, tsk, cpu); @@ -343,9 +341,9 @@ void unregister_tracer(struct tracer *type) notrace void tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; - data->trace_current = head_page(data); - data->trace_current_idx = 0; - data->time_offset = 0; + data->trace_head = data->trace_tail = head_page(data); + data->trace_head_idx = 0; + data->trace_tail_idx = 0; } #ifdef CONFIG_FTRACE @@ -470,38 +468,65 @@ notrace void tracing_record_cmdline(struct task_struct *tsk) trace_save_cmdline(tsk); } +static inline notrace struct list_head * +trace_next_list(struct trace_array_cpu *data, struct list_head *next) +{ + /* + * Roundrobin - but skip the head (which is not a real page): + */ + next = next->next; + if (unlikely(next == &data->trace_pages)) + next = next->next; + BUG_ON(next == &data->trace_pages); + + return next; +} + +static inline notrace void * +trace_next_page(struct trace_array_cpu *data, void *addr) +{ + struct list_head *next; + struct page *page; + + page = virt_to_page(addr); + + next = trace_next_list(data, &page->lru); + page = list_entry(next, struct page, lru); + + return page_address(page); +} + static inline notrace struct trace_entry * tracing_get_trace_entry(struct trace_array *tr, struct trace_array_cpu *data) { unsigned long idx, idx_next; struct trace_entry *entry; - struct list_head *next; - struct page *page; data->trace_idx++; - idx = data->trace_current_idx; + idx = data->trace_head_idx; idx_next = idx + 1; BUG_ON(idx * TRACE_ENTRY_SIZE >= PAGE_SIZE); - entry = data->trace_current + idx * TRACE_ENTRY_SIZE; + entry = data->trace_head + idx * TRACE_ENTRY_SIZE; if (unlikely(idx_next >= ENTRIES_PER_PAGE)) { - page = virt_to_page(data->trace_current); - /* - * Roundrobin - but skip the head (which is not a real page): - */ - next = page->lru.next; - if (unlikely(next == &data->trace_pages)) - next = next->next; - BUG_ON(next == &data->trace_pages); - - page = list_entry(next, struct page, lru); - data->trace_current = page_address(page); + data->trace_head = trace_next_page(data, data->trace_head); idx_next = 0; } - data->trace_current_idx = idx_next; + if (data->trace_head == data->trace_tail && + idx_next == data->trace_tail_idx) { + /* overrun */ + data->trace_tail_idx++; + if (data->trace_tail_idx >= ENTRIES_PER_PAGE) { + data->trace_tail = + trace_next_page(data, data->trace_tail); + data->trace_tail_idx = 0; + } + } + + data->trace_head_idx = idx_next; return entry; } @@ -571,21 +596,11 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, return NULL; if (!iter->next_page[cpu]) { - /* - * Initialize. If the count of elements in - * this buffer is greater than the max entries - * we had an underrun. Which means we looped around. - * We can simply use the current pointer as our - * starting point. - */ - if (data->trace_idx >= tr->entries) { - page = virt_to_page(data->trace_current); - iter->next_page[cpu] = &page->lru; - iter->next_page_idx[cpu] = data->trace_current_idx; - } else { - iter->next_page[cpu] = data->trace_pages.next; - iter->next_page_idx[cpu] = 0; - } + /* Initialize the iterator for this cpu trace buffer */ + WARN_ON(!data->trace_tail); + page = virt_to_page(data->trace_tail); + iter->next_page[cpu] = &page->lru; + iter->next_page_idx[cpu] = data->trace_tail_idx; } page = list_entry(iter->next_page[cpu], struct page, lru); @@ -593,6 +608,12 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, array = page_address(page); + /* Still possible to catch up to the tail */ + if (iter->next_idx[cpu] && array == data->trace_tail && + iter->next_page_idx[cpu] == data->trace_tail_idx) + return NULL; + + WARN_ON(iter->next_page_idx[cpu] >= ENTRIES_PER_PAGE); return &array[iter->next_page_idx[cpu]]; } @@ -638,10 +659,8 @@ static void *find_next_entry_inc(struct trace_iterator *iter) iter->next_page_idx[next_cpu] = 0; iter->next_page[next_cpu] = - iter->next_page[next_cpu]->next; - if (iter->next_page[next_cpu] == &data->trace_pages) - iter->next_page[next_cpu] = - data->trace_pages.next; + trace_next_list(data, iter->next_page[next_cpu]); + } } iter->prev_ent = iter->ent; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 5df8ff2b84a7..0ce127455b4b 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -53,13 +53,15 @@ struct trace_entry { * the trace, etc.) */ struct trace_array_cpu { - void *trace_current; struct list_head trace_pages; atomic_t disabled; cycle_t time_offset; /* these fields get copied into max-trace: */ - unsigned trace_current_idx; + unsigned trace_head_idx; + unsigned trace_tail_idx; + void *trace_head; /* producer */ + void *trace_tail; /* consumer */ unsigned long trace_idx; unsigned long saved_latency; unsigned long critical_start; -- cgit v1.2.3 From 214023c3d13a71525e463b5e54e360b926b4dc90 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: add a buffer for output Later patches will need to print the same things as the seq output does. But those outputs will not use the seq utility. This patch adds a buffer to the iterator, that can be used by either the seq utility or other output. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 202 ++++++++++++++++++++++++++++++++++----------------- kernel/trace/trace.h | 6 ++ 2 files changed, 140 insertions(+), 68 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 777b859e1c2e..d39f4faec7c3 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -169,6 +169,66 @@ void *head_page(struct trace_array_cpu *data) return page_address(page); } +static notrace int +trace_seq_printf(struct trace_seq *s, const char *fmt, ...) +{ + int len = (PAGE_SIZE - 1) - s->len; + va_list ap; + + if (!len) + return 0; + + va_start(ap, fmt); + len = vsnprintf(s->buffer + s->len, len, fmt, ap); + va_end(ap); + + s->len += len; + + return len; +} + +static notrace int +trace_seq_puts(struct trace_seq *s, const char *str) +{ + int len = strlen(str); + + if (len > ((PAGE_SIZE - 1) - s->len)) + len = (PAGE_SIZE - 1) - s->len; + + memcpy(s->buffer + s->len, str, len); + s->len += len; + + return len; +} + +static notrace int +trace_seq_putc(struct trace_seq *s, unsigned char c) +{ + if (s->len >= (PAGE_SIZE - 1)) + return 0; + + s->buffer[s->len++] = c; + + return 1; +} + +static notrace void +trace_seq_reset(struct trace_seq *s) +{ + s->len = 0; +} + +static notrace void +trace_print_seq(struct seq_file *m, struct trace_seq *s) +{ + int len = s->len >= PAGE_SIZE ? PAGE_SIZE - 1 : s->len; + + s->buffer[len] = 0; + seq_puts(m, s->buffer); + + trace_seq_reset(s); +} + notrace static void flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) { @@ -756,25 +816,26 @@ static void s_stop(struct seq_file *m, void *p) } static void -seq_print_sym_short(struct seq_file *m, const char *fmt, unsigned long address) +seq_print_sym_short(struct trace_seq *s, const char *fmt, unsigned long address) { #ifdef CONFIG_KALLSYMS char str[KSYM_SYMBOL_LEN]; kallsyms_lookup(address, NULL, NULL, NULL, str); - seq_printf(m, fmt, str); + trace_seq_printf(s, fmt, str); #endif } static void -seq_print_sym_offset(struct seq_file *m, const char *fmt, unsigned long address) +seq_print_sym_offset(struct trace_seq *s, const char *fmt, + unsigned long address) { #ifdef CONFIG_KALLSYMS char str[KSYM_SYMBOL_LEN]; sprint_symbol(str, address); - seq_printf(m, fmt, str); + trace_seq_printf(s, fmt, str); #endif } @@ -785,20 +846,20 @@ seq_print_sym_offset(struct seq_file *m, const char *fmt, unsigned long address) #endif static notrace void -seq_print_ip_sym(struct seq_file *m, unsigned long ip, unsigned long sym_flags) +seq_print_ip_sym(struct trace_seq *s, unsigned long ip, unsigned long sym_flags) { if (!ip) { - seq_printf(m, "0"); + trace_seq_printf(s, "0"); return; } if (sym_flags & TRACE_ITER_SYM_OFFSET) - seq_print_sym_offset(m, "%s", ip); + seq_print_sym_offset(s, "%s", ip); else - seq_print_sym_short(m, "%s", ip); + seq_print_sym_short(s, "%s", ip); if (sym_flags & TRACE_ITER_SYM_ADDR) - seq_printf(m, " <" IP_FMT ">", ip); + trace_seq_printf(s, " <" IP_FMT ">", ip); } static notrace void print_lat_help_header(struct seq_file *m) @@ -881,9 +942,11 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) if (data->critical_start) { seq_puts(m, " => started at: "); - seq_print_ip_sym(m, data->critical_start, sym_flags); + seq_print_ip_sym(&iter->seq, data->critical_start, sym_flags); + trace_print_seq(m, &iter->seq); seq_puts(m, "\n => ended at: "); - seq_print_ip_sym(m, data->critical_end, sym_flags); + seq_print_ip_sym(&iter->seq, data->critical_end, sym_flags); + trace_print_seq(m, &iter->seq); seq_puts(m, "\n"); } @@ -891,61 +954,61 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) } static notrace void -lat_print_generic(struct seq_file *m, struct trace_entry *entry, int cpu) +lat_print_generic(struct trace_seq *s, struct trace_entry *entry, int cpu) { int hardirq, softirq; char *comm; comm = trace_find_cmdline(entry->pid); - seq_printf(m, "%8.8s-%-5d ", comm, entry->pid); - seq_printf(m, "%d", cpu); - seq_printf(m, "%c%c", - (entry->flags & TRACE_FLAG_IRQS_OFF) ? 'd' : '.', - ((entry->flags & TRACE_FLAG_NEED_RESCHED) ? 'N' : '.')); + trace_seq_printf(s, "%8.8s-%-5d ", comm, entry->pid); + trace_seq_printf(s, "%d", cpu); + trace_seq_printf(s, "%c%c", + (entry->flags & TRACE_FLAG_IRQS_OFF) ? 'd' : '.', + ((entry->flags & TRACE_FLAG_NEED_RESCHED) ? 'N' : '.')); hardirq = entry->flags & TRACE_FLAG_HARDIRQ; softirq = entry->flags & TRACE_FLAG_SOFTIRQ; if (hardirq && softirq) - seq_putc(m, 'H'); + trace_seq_putc(s, 'H'); else { if (hardirq) - seq_putc(m, 'h'); + trace_seq_putc(s, 'h'); else { if (softirq) - seq_putc(m, 's'); + trace_seq_putc(s, 's'); else - seq_putc(m, '.'); + trace_seq_putc(s, '.'); } } if (entry->preempt_count) - seq_printf(m, "%x", entry->preempt_count); + trace_seq_printf(s, "%x", entry->preempt_count); else - seq_puts(m, "."); + trace_seq_puts(s, "."); } unsigned long preempt_mark_thresh = 100; static notrace void -lat_print_timestamp(struct seq_file *m, unsigned long long abs_usecs, +lat_print_timestamp(struct trace_seq *s, unsigned long long abs_usecs, unsigned long rel_usecs) { - seq_printf(m, " %4lldus", abs_usecs); + trace_seq_printf(s, " %4lldus", abs_usecs); if (rel_usecs > preempt_mark_thresh) - seq_puts(m, "!: "); + trace_seq_puts(s, "!: "); else if (rel_usecs > 1) - seq_puts(m, "+: "); + trace_seq_puts(s, "+: "); else - seq_puts(m, " : "); + trace_seq_puts(s, " : "); } static const char state_to_char[] = TASK_STATE_TO_CHAR_STR; static notrace void -print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, - unsigned int trace_idx, int cpu) +print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) { + struct trace_seq *s = &iter->seq; unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); struct trace_entry *next_entry = find_next_entry(iter, NULL); unsigned long verbose = (trace_flags & TRACE_ITER_VERBOSE); @@ -962,39 +1025,40 @@ print_lat_fmt(struct seq_file *m, struct trace_iterator *iter, if (verbose) { comm = trace_find_cmdline(entry->pid); - seq_printf(m, "%16s %5d %d %d %08x %08x [%08lx]" - " %ld.%03ldms (+%ld.%03ldms): ", - comm, - entry->pid, cpu, entry->flags, - entry->preempt_count, trace_idx, - ns2usecs(entry->t), - abs_usecs/1000, - abs_usecs % 1000, rel_usecs/1000, rel_usecs % 1000); + trace_seq_printf(s, "%16s %5d %d %d %08x %08x [%08lx]" + " %ld.%03ldms (+%ld.%03ldms): ", + comm, + entry->pid, cpu, entry->flags, + entry->preempt_count, trace_idx, + ns2usecs(entry->t), + abs_usecs/1000, + abs_usecs % 1000, rel_usecs/1000, + rel_usecs % 1000); } else { - lat_print_generic(m, entry, cpu); - lat_print_timestamp(m, abs_usecs, rel_usecs); + lat_print_generic(s, entry, cpu); + lat_print_timestamp(s, abs_usecs, rel_usecs); } switch (entry->type) { case TRACE_FN: - seq_print_ip_sym(m, entry->fn.ip, sym_flags); - seq_puts(m, " ("); - seq_print_ip_sym(m, entry->fn.parent_ip, sym_flags); - seq_puts(m, ")\n"); + seq_print_ip_sym(s, entry->fn.ip, sym_flags); + trace_seq_puts(s, " ("); + seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags); + trace_seq_puts(s, ")\n"); break; case TRACE_CTX: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; comm = trace_find_cmdline(entry->ctx.next_pid); - seq_printf(m, " %d:%d:%c --> %d:%d %s\n", - entry->ctx.prev_pid, - entry->ctx.prev_prio, - S, - entry->ctx.next_pid, - entry->ctx.next_prio, - comm); + trace_seq_printf(s, " %d:%d:%c --> %d:%d %s\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio, + comm); break; default: - seq_printf(m, "Unknown type %d\n", entry->type); + trace_seq_printf(s, "Unknown type %d\n", entry->type); } } @@ -1026,8 +1090,9 @@ static notrace void sync_time_offset(struct trace_iterator *iter) } static notrace void -print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) +print_trace_fmt(struct trace_iterator *iter) { + struct trace_seq *s = &iter->seq; unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); struct trace_entry *entry; unsigned long usec_rem; @@ -1045,29 +1110,29 @@ print_trace_fmt(struct seq_file *m, struct trace_iterator *iter) usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; - seq_printf(m, "%16s-%-5d ", comm, entry->pid); - seq_printf(m, "[%02d] ", iter->cpu); - seq_printf(m, "%5lu.%06lu: ", secs, usec_rem); + trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); + trace_seq_printf(s, "[%02d] ", iter->cpu); + trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); switch (entry->type) { case TRACE_FN: - seq_print_ip_sym(m, entry->fn.ip, sym_flags); + seq_print_ip_sym(s, entry->fn.ip, sym_flags); if ((sym_flags & TRACE_ITER_PRINT_PARENT) && entry->fn.parent_ip) { - seq_printf(m, " <-"); - seq_print_ip_sym(m, entry->fn.parent_ip, sym_flags); + trace_seq_printf(s, " <-"); + seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags); } - seq_printf(m, "\n"); + trace_seq_printf(s, "\n"); break; case TRACE_CTX: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; - seq_printf(m, " %d:%d:%c ==> %d:%d\n", - entry->ctx.prev_pid, - entry->ctx.prev_prio, - S, - entry->ctx.next_pid, - entry->ctx.next_prio); + trace_seq_printf(s, " %d:%d:%c ==> %d:%d\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio); break; } } @@ -1108,9 +1173,10 @@ static int s_show(struct seq_file *m, void *v) } } else { if (iter->iter_flags & TRACE_FILE_LAT_FMT) - print_lat_fmt(m, iter, iter->idx, iter->cpu); + print_lat_fmt(iter, iter->idx, iter->cpu); else - print_trace_fmt(m, iter); + print_trace_fmt(iter); + trace_print_seq(m, &iter->seq); } return 0; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 0ce127455b4b..f5b32ca0b457 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -111,11 +111,17 @@ struct tracer { int print_max; }; +struct trace_seq { + unsigned char buffer[PAGE_SIZE]; + unsigned int len; +}; + /* * Trace iterator - used by printout routines who present trace * results to users and which routines might sleep, etc: */ struct trace_iterator { + struct trace_seq seq; struct trace_array *tr; struct tracer *trace; -- cgit v1.2.3 From b3806b4316306dc9c542eff6c23d7d42918f3504 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: user run time file reading This patch creates a file called trace_pipe in the tracing debug directory. This file is a consumer of the trace buffers. This means that reads of this file consumes the entries from the trace buffers so that they will not be read a second time, as contrast to the static buffers latency_trace and trace. Reading from the trace_pipe will remove the entries from trace and latency_trace too. The advantage that trace_pipe has is that it can record live traces. It will block when there is nothing in the buffer, and read the entries as they are entered. An EOF happens when tracing is disabled (tracing_enabled = 0). Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 353 ++++++++++++++++++++++++++++++++++++++++++++------- kernel/trace/trace.h | 2 + 2 files changed, 309 insertions(+), 46 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d39f4faec7c3..a40687a4413a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -174,15 +174,20 @@ trace_seq_printf(struct trace_seq *s, const char *fmt, ...) { int len = (PAGE_SIZE - 1) - s->len; va_list ap; + int ret; if (!len) return 0; va_start(ap, fmt); - len = vsnprintf(s->buffer + s->len, len, fmt, ap); + ret = vsnprintf(s->buffer + s->len, len, fmt, ap); va_end(ap); - s->len += len; + /* If we can't write it all, don't bother writing anything */ + if (ret > len) + return 0; + + s->len += ret; return len; } @@ -193,7 +198,7 @@ trace_seq_puts(struct trace_seq *s, const char *str) int len = strlen(str); if (len > ((PAGE_SIZE - 1) - s->len)) - len = (PAGE_SIZE - 1) - s->len; + return 0; memcpy(s->buffer + s->len, str, len); s->len += len; @@ -615,11 +620,13 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, { struct trace_entry *entry; + spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_FN; entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; + spin_unlock(&data->lock); } notrace void @@ -630,6 +637,7 @@ tracing_sched_switch_trace(struct trace_array *tr, { struct trace_entry *entry; + spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_CTX; @@ -638,6 +646,7 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.prev_state = prev->state; entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; + spin_unlock(&data->lock); } enum trace_file_type { @@ -652,7 +661,9 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *array; if (iter->next_idx[cpu] >= tr->entries || - iter->next_idx[cpu] >= data->trace_idx) + iter->next_idx[cpu] >= data->trace_idx || + (data->trace_head == data->trace_tail && + data->trace_head_idx == data->trace_tail_idx)) return NULL; if (!iter->next_page[cpu]) { @@ -702,33 +713,57 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) return next; } -static void *find_next_entry_inc(struct trace_iterator *iter) +static notrace void +trace_iterator_increment(struct trace_iterator *iter) { - struct trace_entry *next; - int next_cpu = -1; + iter->idx++; + iter->next_idx[iter->cpu]++; + iter->next_page_idx[iter->cpu]++; + if (iter->next_page_idx[iter->cpu] >= ENTRIES_PER_PAGE) { + struct trace_array_cpu *data = iter->tr->data[iter->cpu]; - next = find_next_entry(iter, &next_cpu); + iter->next_page_idx[iter->cpu] = 0; + iter->next_page[iter->cpu] = + trace_next_list(data, iter->next_page[iter->cpu]); + } +} - if (next) { - iter->idx++; - iter->next_idx[next_cpu]++; - iter->next_page_idx[next_cpu]++; +static notrace void +trace_consume(struct trace_iterator *iter) +{ + struct trace_array_cpu *data = iter->tr->data[iter->cpu]; + + data->trace_tail_idx++; + if (data->trace_tail_idx >= ENTRIES_PER_PAGE) { + data->trace_tail = trace_next_page(data, data->trace_tail); + data->trace_tail_idx = 0; + } - if (iter->next_page_idx[next_cpu] >= ENTRIES_PER_PAGE) { - struct trace_array_cpu *data = iter->tr->data[next_cpu]; + /* Check if we empty it, then reset the index */ + if (data->trace_head == data->trace_tail && + data->trace_head_idx == data->trace_tail_idx) + data->trace_idx = 0; - iter->next_page_idx[next_cpu] = 0; - iter->next_page[next_cpu] = - trace_next_list(data, iter->next_page[next_cpu]); + trace_iterator_increment(iter); +} + +static notrace void * +find_next_entry_inc(struct trace_iterator *iter) +{ + struct trace_entry *next; + int next_cpu = -1; + + next = find_next_entry(iter, &next_cpu); - } - } iter->prev_ent = iter->ent; iter->prev_cpu = iter->cpu; iter->ent = next; iter->cpu = next_cpu; + if (next) + trace_iterator_increment(iter); + return next ? iter : NULL; } @@ -815,7 +850,7 @@ static void s_stop(struct seq_file *m, void *p) mutex_unlock(&trace_types_lock); } -static void +static int seq_print_sym_short(struct trace_seq *s, const char *fmt, unsigned long address) { #ifdef CONFIG_KALLSYMS @@ -823,11 +858,12 @@ seq_print_sym_short(struct trace_seq *s, const char *fmt, unsigned long address) kallsyms_lookup(address, NULL, NULL, NULL, str); - trace_seq_printf(s, fmt, str); + return trace_seq_printf(s, fmt, str); #endif + return 1; } -static void +static int seq_print_sym_offset(struct trace_seq *s, const char *fmt, unsigned long address) { @@ -835,8 +871,9 @@ seq_print_sym_offset(struct trace_seq *s, const char *fmt, char str[KSYM_SYMBOL_LEN]; sprint_symbol(str, address); - trace_seq_printf(s, fmt, str); + return trace_seq_printf(s, fmt, str); #endif + return 1; } #ifndef CONFIG_64BIT @@ -845,21 +882,25 @@ seq_print_sym_offset(struct trace_seq *s, const char *fmt, # define IP_FMT "%016lx" #endif -static notrace void +static notrace int seq_print_ip_sym(struct trace_seq *s, unsigned long ip, unsigned long sym_flags) { - if (!ip) { - trace_seq_printf(s, "0"); - return; - } + int ret; + + if (!ip) + return trace_seq_printf(s, "0"); if (sym_flags & TRACE_ITER_SYM_OFFSET) - seq_print_sym_offset(s, "%s", ip); + ret = seq_print_sym_offset(s, "%s", ip); else - seq_print_sym_short(s, "%s", ip); + ret = seq_print_sym_short(s, "%s", ip); + + if (!ret) + return 0; if (sym_flags & TRACE_ITER_SYM_ADDR) - trace_seq_printf(s, " <" IP_FMT ">", ip); + ret = trace_seq_printf(s, " <" IP_FMT ">", ip); + return ret; } static notrace void print_lat_help_header(struct seq_file *m) @@ -1089,7 +1130,7 @@ static notrace void sync_time_offset(struct trace_iterator *iter) array->time_offset += prev_t - t; } -static notrace void +static notrace int print_trace_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; @@ -1100,6 +1141,7 @@ print_trace_fmt(struct trace_iterator *iter) unsigned long secs; char *comm; int S; + int ret; sync_time_offset(iter); entry = iter->ent; @@ -1110,31 +1152,49 @@ print_trace_fmt(struct trace_iterator *iter) usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; - trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); - trace_seq_printf(s, "[%02d] ", iter->cpu); - trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); + ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); + if (!ret) + return 0; + ret = trace_seq_printf(s, "[%02d] ", iter->cpu); + if (!ret) + return 0; + ret = trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); + if (!ret) + return 0; switch (entry->type) { case TRACE_FN: - seq_print_ip_sym(s, entry->fn.ip, sym_flags); + ret = seq_print_ip_sym(s, entry->fn.ip, sym_flags); + if (!ret) + return 0; if ((sym_flags & TRACE_ITER_PRINT_PARENT) && entry->fn.parent_ip) { - trace_seq_printf(s, " <-"); - seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags); + ret = trace_seq_printf(s, " <-"); + if (!ret) + return 0; + ret = seq_print_ip_sym(s, entry->fn.parent_ip, + sym_flags); + if (!ret) + return 0; } - trace_seq_printf(s, "\n"); + ret = trace_seq_printf(s, "\n"); + if (!ret) + return 0; break; case TRACE_CTX: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; - trace_seq_printf(s, " %d:%d:%c ==> %d:%d\n", - entry->ctx.prev_pid, - entry->ctx.prev_prio, - S, - entry->ctx.next_pid, - entry->ctx.next_prio); + ret = trace_seq_printf(s, " %d:%d:%c ==> %d:%d\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio); + if (!ret) + return 0; break; } + return 1; } static int trace_empty(struct trace_iterator *iter) @@ -1145,7 +1205,9 @@ static int trace_empty(struct trace_iterator *iter) for_each_possible_cpu(cpu) { data = iter->tr->data[cpu]; - if (head_page(data) && data->trace_idx) + if (head_page(data) && data->trace_idx && + (data->trace_tail != data->trace_head || + data->trace_tail_idx != data->trace_head_idx)) return 0; } return 1; @@ -1645,6 +1707,192 @@ tracing_max_lat_write(struct file *filp, const char __user *ubuf, return cnt; } +static atomic_t tracing_reader; + +static int tracing_open_pipe(struct inode *inode, struct file *filp) +{ + struct trace_iterator *iter; + + if (tracing_disabled) + return -ENODEV; + + /* We only allow for reader of the pipe */ + if (atomic_inc_return(&tracing_reader) != 1) { + atomic_dec(&tracing_reader); + return -EBUSY; + } + + /* create a buffer to store the information to pass to userspace */ + iter = kzalloc(sizeof(*iter), GFP_KERNEL); + if (!iter) + return -ENOMEM; + + iter->tr = &global_trace; + + filp->private_data = iter; + + return 0; +} + +static int tracing_release_pipe(struct inode *inode, struct file *file) +{ + struct trace_iterator *iter = file->private_data; + + kfree(iter); + atomic_dec(&tracing_reader); + + return 0; +} + +/* + * Consumer reader. + */ +static ssize_t +tracing_read_pipe(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct trace_iterator *iter = filp->private_data; + struct trace_array_cpu *data; + static cpumask_t mask; + struct trace_entry *entry; + static int start; + unsigned long flags; + int read = 0; + int cpu; + int len; + int ret; + + /* return any leftover data */ + if (iter->seq.len > start) { + len = iter->seq.len - start; + if (cnt > len) + cnt = len; + ret = copy_to_user(ubuf, iter->seq.buffer + start, cnt); + if (ret) + cnt = -EFAULT; + + start += len; + + return cnt; + } + + trace_seq_reset(&iter->seq); + start = 0; + + while (trace_empty(iter)) { + /* + * This is a make-shift waitqueue. The reason we don't use + * an actual wait queue is because: + * 1) we only ever have one waiter + * 2) the tracing, traces all functions, we don't want + * the overhead of calling wake_up and friends + * (and tracing them too) + * Anyway, this is really very primitive wakeup. + */ + set_current_state(TASK_INTERRUPTIBLE); + iter->tr->waiter = current; + + /* sleep for one second, and try again. */ + schedule_timeout(HZ); + + iter->tr->waiter = NULL; + + if (signal_pending(current)) + return -EINTR; + + /* + * We block until we read something and tracing is disabled. + * We still block if tracing is disabled, but we have never + * read anything. This allows a user to cat this file, and + * then enable tracing. But after we have read something, + * we give an EOF when tracing is again disabled. + * + * iter->pos will be 0 if we haven't read anything. + */ + if (!tracer_enabled && iter->pos) + break; + + continue; + } + + /* stop when tracing is finished */ + if (trace_empty(iter)) + return 0; + + if (cnt >= PAGE_SIZE) + cnt = PAGE_SIZE - 1; + + memset(iter, 0, sizeof(*iter)); + iter->tr = &global_trace; + iter->pos = -1; + + /* + * We need to stop all tracing on all CPUS to read the + * the next buffer. This is a bit expensive, but is + * not done often. We fill all what we can read, + * and then release the locks again. + */ + + cpus_clear(mask); + local_irq_save(flags); + for_each_possible_cpu(cpu) { + data = iter->tr->data[cpu]; + + if (!head_page(data) || !data->trace_idx) + continue; + + atomic_inc(&data->disabled); + spin_lock(&data->lock); + cpu_set(cpu, mask); + } + + while ((entry = find_next_entry(iter, &cpu))) { + + if (!entry) + break; + + iter->ent = entry; + iter->cpu = cpu; + + ret = print_trace_fmt(iter); + if (!ret) + break; + + trace_consume(iter); + + if (iter->seq.len >= cnt) + break; + + } + + for_each_possible_cpu(cpu) { + data = iter->tr->data[cpu]; + + if (!cpu_isset(cpu, mask)) + continue; + spin_unlock(&data->lock); + atomic_dec(&data->disabled); + } + local_irq_restore(flags); + + /* Now copy what we have to the user */ + read = iter->seq.len; + if (read > cnt) + read = cnt; + + ret = copy_to_user(ubuf, iter->seq.buffer, read); + + if (read < iter->seq.len) + start = read; + else + trace_seq_reset(&iter->seq); + + if (ret) + read = -EFAULT; + + return read; +} + static struct file_operations tracing_max_lat_fops = { .open = tracing_open_generic, .read = tracing_max_lat_read, @@ -1663,6 +1911,12 @@ static struct file_operations set_tracer_fops = { .write = tracing_set_trace_write, }; +static struct file_operations tracing_pipe_fops = { + .open = tracing_open_pipe, + .read = tracing_read_pipe, + .release = tracing_release_pipe, +}; + #ifdef CONFIG_DYNAMIC_FTRACE static ssize_t @@ -1763,6 +2017,11 @@ static __init void tracer_init_debugfs(void) if (!entry) pr_warning("Could not create debugfs 'README' entry\n"); + entry = debugfs_create_file("trace_pipe", 0644, d_tracer, + NULL, &tracing_pipe_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'tracing_threash' entry\n"); #ifdef CONFIG_DYNAMIC_FTRACE entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer, @@ -1816,6 +2075,7 @@ static int trace_alloc_page(void) /* Now that we successfully allocate a page per CPU, add them */ for_each_possible_cpu(i) { data = global_trace.data[i]; + spin_lock_init(&data->lock); page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -1823,6 +2083,7 @@ static int trace_alloc_page(void) #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; + spin_lock_init(&data->lock); page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index f5b32ca0b457..29a7ea59de50 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -55,6 +55,7 @@ struct trace_entry { struct trace_array_cpu { struct list_head trace_pages; atomic_t disabled; + spinlock_t lock; cycle_t time_offset; /* these fields get copied into max-trace: */ @@ -88,6 +89,7 @@ struct trace_array { long ctrl; int cpu; cycle_t time_start; + struct task_struct *waiter; struct trace_array_cpu *data[NR_CPUS]; }; -- cgit v1.2.3 From d4c5a2f5870939d837293de87b41dda0012a4572 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: fix locking we can hold all cpu trace buffer locks at once - put each into a separate lock class. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 8 +++----- kernel/trace/trace.h | 1 + 2 files changed, 4 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index a40687a4413a..b3811ca74071 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1865,11 +1865,8 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, } - for_each_possible_cpu(cpu) { + for_each_cpu_mask(cpu, mask) { data = iter->tr->data[cpu]; - - if (!cpu_isset(cpu, mask)) - continue; spin_unlock(&data->lock); atomic_dec(&data->disabled); } @@ -2076,6 +2073,7 @@ static int trace_alloc_page(void) for_each_possible_cpu(i) { data = global_trace.data[i]; spin_lock_init(&data->lock); + lockdep_set_class(&data->lock, &data->lock_key); page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -2084,6 +2082,7 @@ static int trace_alloc_page(void) #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; spin_lock_init(&data->lock); + lockdep_set_class(&data->lock, &data->lock_key); page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -2203,5 +2202,4 @@ __init static int tracer_alloc_buffers(void) } return ret; } - fs_initcall(tracer_alloc_buffers); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 29a7ea59de50..b0408356f0e0 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -56,6 +56,7 @@ struct trace_array_cpu { struct list_head trace_pages; atomic_t disabled; spinlock_t lock; + struct lock_class_key lock_key; cycle_t time_offset; /* these fields get copied into max-trace: */ -- cgit v1.2.3 From 4bf39a9411a4ce8712954e03a9bd1592ee345919 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: cleanups no code changed. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 5 +++- kernel/trace/trace.c | 72 +++++++++++++++++++++++++-------------------------- 2 files changed, 39 insertions(+), 38 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 5e9389faaf75..97c40865a93e 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -756,9 +756,11 @@ ftrace_avail_open(struct inode *inode, struct file *file) ret = seq_open(file, &show_ftrace_seq_ops); if (!ret) { struct seq_file *m = file->private_data; + m->private = iter; - } else + } else { kfree(iter); + } return ret; } @@ -770,6 +772,7 @@ int ftrace_avail_release(struct inode *inode, struct file *file) seq_release(inode, file); kfree(iter); + return 0; } diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index b3811ca74071..4550afda9607 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1245,10 +1245,10 @@ static int s_show(struct seq_file *m, void *v) } static struct seq_operations tracer_seq_ops = { - .start = s_start, - .next = s_next, - .stop = s_stop, - .show = s_show, + .start = s_start, + .next = s_next, + .stop = s_stop, + .show = s_show, }; static struct trace_iterator notrace * @@ -1397,10 +1397,10 @@ static int t_show(struct seq_file *m, void *v) } static struct seq_operations show_traces_seq_ops = { - .start = t_start, - .next = t_next, - .stop = t_stop, - .show = t_show, + .start = t_start, + .next = t_next, + .stop = t_stop, + .show = t_show, }; static int show_traces_open(struct inode *inode, struct file *file) @@ -1420,17 +1420,17 @@ static int show_traces_open(struct inode *inode, struct file *file) } static struct file_operations tracing_fops = { - .open = tracing_open, - .read = seq_read, - .llseek = seq_lseek, - .release = tracing_release, + .open = tracing_open, + .read = seq_read, + .llseek = seq_lseek, + .release = tracing_release, }; static struct file_operations tracing_lt_fops = { - .open = tracing_lt_open, - .read = seq_read, - .llseek = seq_lseek, - .release = tracing_release, + .open = tracing_lt_open, + .read = seq_read, + .llseek = seq_lseek, + .release = tracing_release, }; static struct file_operations show_traces_fops = { @@ -1620,8 +1620,7 @@ tracing_set_trace_read(struct file *filp, char __user *ubuf, r = sprintf(buf, "\n"); mutex_unlock(&trace_types_lock); - return simple_read_from_buffer(ubuf, cnt, ppos, - buf, r); + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } static ssize_t @@ -1680,8 +1679,7 @@ tracing_max_lat_read(struct file *filp, char __user *ubuf, *ptr == (unsigned long)-1 ? -1 : nsecs_to_usecs(*ptr)); if (r > 64) r = 64; - return simple_read_from_buffer(ubuf, cnt, ppos, - buf, r); + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } static ssize_t @@ -1891,27 +1889,27 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, } static struct file_operations tracing_max_lat_fops = { - .open = tracing_open_generic, - .read = tracing_max_lat_read, - .write = tracing_max_lat_write, + .open = tracing_open_generic, + .read = tracing_max_lat_read, + .write = tracing_max_lat_write, }; static struct file_operations tracing_ctrl_fops = { - .open = tracing_open_generic, - .read = tracing_ctrl_read, - .write = tracing_ctrl_write, + .open = tracing_open_generic, + .read = tracing_ctrl_read, + .write = tracing_ctrl_write, }; static struct file_operations set_tracer_fops = { - .open = tracing_open_generic, - .read = tracing_set_trace_read, - .write = tracing_set_trace_write, + .open = tracing_open_generic, + .read = tracing_set_trace_read, + .write = tracing_set_trace_write, }; static struct file_operations tracing_pipe_fops = { - .open = tracing_open_pipe, - .read = tracing_read_pipe, - .release = tracing_release_pipe, + .open = tracing_open_pipe, + .read = tracing_read_pipe, + .release = tracing_release_pipe, }; #ifdef CONFIG_DYNAMIC_FTRACE @@ -1925,13 +1923,13 @@ tracing_read_long(struct file *filp, char __user *ubuf, int r; r = sprintf(buf, "%ld\n", *p); - return simple_read_from_buffer(ubuf, cnt, ppos, - buf, r); + + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } static struct file_operations tracing_read_long_fops = { - .open = tracing_open_generic, - .read = tracing_read_long, + .open = tracing_open_generic, + .read = tracing_read_long, }; #endif @@ -2033,7 +2031,7 @@ static __init void tracer_init_debugfs(void) /* dummy trace to disable tracing */ static struct tracer no_tracer __read_mostly = { - .name = "none", + .name = "none", }; static int trace_alloc_page(void) -- cgit v1.2.3 From 750ed1a40783432d0dcb0e6c2e813a12615d7664 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: timestamp syncing, prepare rename and uninline now() to ftrace_now(). Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 4 ++-- kernel/trace/trace.c | 7 ++++++- kernel/trace/trace.h | 5 +---- kernel/trace/trace_functions.c | 2 +- kernel/trace/trace_irqsoff.c | 6 +++--- kernel/trace/trace_sched_switch.c | 2 +- kernel/trace/trace_sched_wakeup.c | 4 ++-- 7 files changed, 16 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 97c40865a93e..a15e068535f8 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -531,7 +531,7 @@ static int notrace __ftrace_update_code(void *ignore) save_ftrace_enabled = ftrace_enabled; ftrace_enabled = 0; - start = now(raw_smp_processor_id()); + start = ftrace_now(raw_smp_processor_id()); ftrace_update_cnt = 0; /* No locks needed, the machine is stopped! */ @@ -550,7 +550,7 @@ static int notrace __ftrace_update_code(void *ignore) } - stop = now(raw_smp_processor_id()); + stop = ftrace_now(raw_smp_processor_id()); ftrace_update_time = stop - start; ftrace_update_tot_cnt += ftrace_update_cnt; diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 4550afda9607..e3778ab0d3f7 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -42,6 +42,11 @@ ns2usecs(cycle_t nsec) return nsec; } +notrace cycle_t ftrace_now(int cpu) +{ + return cpu_clock(cpu); +} + static atomic_t tracer_counter; static struct trace_array global_trace; @@ -607,7 +612,7 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) entry->idx = atomic_inc_return(&tracer_counter); entry->preempt_count = pc & 0xff; entry->pid = tsk->pid; - entry->t = now(raw_smp_processor_id()); + entry->t = ftrace_now(raw_smp_processor_id()); entry->flags = (irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) | ((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) | ((pc & SOFTIRQ_MASK) ? TRACE_FLAG_SOFTIRQ : 0) | diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index b0408356f0e0..30cad677e9d0 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -171,10 +171,7 @@ void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu); -static inline notrace cycle_t now(int cpu) -{ - return cpu_clock(cpu); -} +extern notrace cycle_t ftrace_now(int cpu); #ifdef CONFIG_SCHED_TRACER extern void notrace diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 5d8ad7a09605..e5d34b78fc99 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -20,7 +20,7 @@ static notrace void function_reset(struct trace_array *tr) { int cpu; - tr->time_start = now(tr->cpu); + tr->time_start = ftrace_now(tr->cpu); for_each_online_cpu(cpu) tracing_reset(tr->data[cpu]); diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 2dfebb67fdfb..d2a6e6f1ad2d 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -136,7 +136,7 @@ check_critical_timing(struct trace_array *tr, * as long as possible: */ T0 = data->preempt_timestamp; - T1 = now(cpu); + T1 = ftrace_now(cpu); delta = T1-T0; local_save_flags(flags); @@ -186,7 +186,7 @@ out_unlock: out: data->critical_sequence = max_sequence; - data->preempt_timestamp = now(cpu); + data->preempt_timestamp = ftrace_now(cpu); tracing_reset(data); ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); } @@ -215,7 +215,7 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) atomic_inc(&data->disabled); data->critical_sequence = max_sequence; - data->preempt_timestamp = now(cpu); + data->preempt_timestamp = ftrace_now(cpu); data->critical_start = parent_ip ? : ip; tracing_reset(data); diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 6c9284103a62..8d656672da93 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -61,7 +61,7 @@ static notrace void sched_switch_reset(struct trace_array *tr) { int cpu; - tr->time_start = now(tr->cpu); + tr->time_start = ftrace_now(tr->cpu); for_each_online_cpu(cpu) tracing_reset(tr->data[cpu]); diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 688df965f3f2..b7df825c3af9 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -92,7 +92,7 @@ wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) * as long as possible: */ T0 = data->preempt_timestamp; - T1 = now(cpu); + T1 = ftrace_now(cpu); delta = T1-T0; if (!report_latency(delta)) @@ -191,7 +191,7 @@ wakeup_check_start(struct trace_array *tr, struct task_struct *p, local_save_flags(flags); - tr->data[wakeup_cpu]->preempt_timestamp = now(cpu); + tr->data[wakeup_cpu]->preempt_timestamp = ftrace_now(cpu); ftrace(tr, tr->data[wakeup_cpu], CALLER_ADDR1, CALLER_ADDR2, flags); out_locked: -- cgit v1.2.3 From 53c37c17aafcf50f7c6fddaf01dda8f9d7e31ddf Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: fast, scalable, synchronized timestamps implement globally synchronized, fast and scalable time source for tracing. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 53 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index e3778ab0d3f7..9a931c7c2da3 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -42,9 +42,61 @@ ns2usecs(cycle_t nsec) return nsec; } +static const int time_sync_freq_max = 128; +static const cycle_t time_sync_thresh = 100000; + +static DEFINE_PER_CPU(cycle_t, time_offset); +static DEFINE_PER_CPU(cycle_t, prev_cpu_time); +static DEFINE_PER_CPU(int, time_sync_count); +static DEFINE_PER_CPU(int, time_sync_freq); + +/* + * Global lock which we take every now and then to synchronize + * the CPUs time. This method is not warp-safe, but it's good + * enough to synchronize slowly diverging time sources and thus + * it's good enough for tracing: + */ +static DEFINE_SPINLOCK(time_sync_lock); +static cycle_t prev_global_time; + +static notrace cycle_t __ftrace_now_sync(cycles_t time, int cpu) +{ + unsigned long flags; + + spin_lock_irqsave(&time_sync_lock, flags); + + /* + * Update the synchronization frequency: + */ + if (per_cpu(time_sync_freq, cpu) < time_sync_freq_max) + per_cpu(time_sync_freq, cpu) *= 2; + per_cpu(time_sync_count, cpu) = per_cpu(time_sync_freq, cpu); + + if (time < prev_global_time) { + per_cpu(time_offset, cpu) += prev_global_time - time; + time = prev_global_time; + } else { + prev_global_time = time; + } + + spin_unlock_irqrestore(&time_sync_lock, flags); + + return time; +} + notrace cycle_t ftrace_now(int cpu) { - return cpu_clock(cpu); + cycle_t prev_cpu_time, time, delta_time; + + prev_cpu_time = per_cpu(prev_cpu_time, cpu); + time = sched_clock() + per_cpu(time_offset, cpu); + delta_time = time-prev_cpu_time; + + if (unlikely(delta_time > time_sync_thresh || + --per_cpu(time_sync_count, cpu) <= 0)) + time = __ftrace_now_sync(time, cpu); + + return time; } static atomic_t tracer_counter; -- cgit v1.2.3 From cdd31cd2d7a0dcbec2cce3974f7129dd4fc8c879 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: remove-idx-sync remove idx syncing - it's expensive on SMP. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 40 +++++----------------------------------- kernel/trace/trace.h | 2 -- 2 files changed, 5 insertions(+), 37 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9a931c7c2da3..ce8ceb8aea6a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -99,7 +99,6 @@ notrace cycle_t ftrace_now(int cpu) return time; } -static atomic_t tracer_counter; static struct trace_array global_trace; static DEFINE_PER_CPU(struct trace_array_cpu, global_trace_cpu); @@ -661,7 +660,6 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) pc = preempt_count(); - entry->idx = atomic_inc_return(&tracer_counter); entry->preempt_count = pc & 0xff; entry->pid = tsk->pid; entry->t = ftrace_now(raw_smp_processor_id()); @@ -757,8 +755,10 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) if (!head_page(tr->data[cpu])) continue; ent = trace_entry_idx(tr, tr->data[cpu], iter, cpu); - if (ent && - (!next || (long)(next->idx - ent->idx) > 0)) { + /* + * Pick the entry with the smallest timestamp: + */ + if (ent && (!next || ent->t < next->t)) { next = ent; next_cpu = cpu; } @@ -800,8 +800,6 @@ trace_consume(struct trace_iterator *iter) if (data->trace_head == data->trace_tail && data->trace_head_idx == data->trace_tail_idx) data->trace_idx = 0; - - trace_iterator_increment(iter); } static notrace void * @@ -1160,33 +1158,6 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) } } -static notrace void sync_time_offset(struct trace_iterator *iter) -{ - struct trace_array_cpu *prev_array, *array; - struct trace_entry *prev_entry, *entry; - cycle_t prev_t, t; - - entry = iter->ent; - prev_entry = iter->prev_ent; - if (!prev_entry) - return; - - prev_array = iter->tr->data[iter->prev_cpu]; - array = iter->tr->data[iter->cpu]; - - prev_t = prev_entry->t + prev_array->time_offset; - t = entry->t + array->time_offset; - - /* - * If time goes backwards we increase the offset of - * the current array, to not have observable time warps. - * This will quickly synchronize the time offsets of - * multiple CPUs: - */ - if (t < prev_t) - array->time_offset += prev_t - t; -} - static notrace int print_trace_fmt(struct trace_iterator *iter) { @@ -1200,12 +1171,11 @@ print_trace_fmt(struct trace_iterator *iter) int S; int ret; - sync_time_offset(iter); entry = iter->ent; comm = trace_find_cmdline(iter->ent->pid); - t = ns2usecs(entry->t + iter->tr->data[iter->cpu]->time_offset); + t = ns2usecs(entry->t); usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 30cad677e9d0..27fa2d06f499 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -38,7 +38,6 @@ struct trace_entry { char preempt_count; int pid; cycle_t t; - unsigned long idx; union { struct ftrace_entry fn; struct ctx_switch_entry ctx; @@ -57,7 +56,6 @@ struct trace_array_cpu { atomic_t disabled; spinlock_t lock; struct lock_class_key lock_key; - cycle_t time_offset; /* these fields get copied into max-trace: */ unsigned trace_head_idx; -- cgit v1.2.3 From 8c523a9d82dbc4f3f7d972df8c0f1eacd83d0d55 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:46 +0200 Subject: ftrace: clean-up-pipe-iteration Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 20 +++++--------------- 1 file changed, 5 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index ce8ceb8aea6a..42f1926acf73 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -770,12 +770,12 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) return next; } -static notrace void -trace_iterator_increment(struct trace_iterator *iter) +static notrace void trace_iterator_increment(struct trace_iterator *iter) { iter->idx++; iter->next_idx[iter->cpu]++; iter->next_page_idx[iter->cpu]++; + if (iter->next_page_idx[iter->cpu] >= ENTRIES_PER_PAGE) { struct trace_array_cpu *data = iter->tr->data[iter->cpu]; @@ -785,8 +785,7 @@ trace_iterator_increment(struct trace_iterator *iter) } } -static notrace void -trace_consume(struct trace_iterator *iter) +static notrace void trace_consume(struct trace_iterator *iter) { struct trace_array_cpu *data = iter->tr->data[iter->cpu]; @@ -802,8 +801,7 @@ trace_consume(struct trace_iterator *iter) data->trace_idx = 0; } -static notrace void * -find_next_entry_inc(struct trace_iterator *iter) +static notrace void *find_next_entry_inc(struct trace_iterator *iter) { struct trace_entry *next; int next_cpu = -1; @@ -1871,14 +1869,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cpu_set(cpu, mask); } - while ((entry = find_next_entry(iter, &cpu))) { - - if (!entry) - break; - - iter->ent = entry; - iter->cpu = cpu; - + while ((entry = find_next_entry_inc(iter)) != NULL) { ret = print_trace_fmt(iter); if (!ret) break; @@ -1887,7 +1878,6 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, if (iter->seq.len >= cnt) break; - } for_each_cpu_mask(cpu, mask) { -- cgit v1.2.3 From f9896bf30928922a3913a3810a4ab7908da6cfe7 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: add raw output Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 63 +++++++++++++++++++++++++++++++++++++++++++++------- 1 file changed, 55 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 42f1926acf73..bebd263f582f 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -153,6 +153,7 @@ enum trace_iterator_flags { TRACE_ITER_SYM_OFFSET = 0x02, TRACE_ITER_SYM_ADDR = 0x04, TRACE_ITER_VERBOSE = 0x08, + TRACE_ITER_RAW = 0x10, }; #define TRACE_ITER_SYM_MASK \ @@ -164,6 +165,7 @@ static const char *trace_options[] = { "sym-offset", "sym-addr", "verbose", + "raw", NULL }; @@ -1099,7 +1101,7 @@ lat_print_timestamp(struct trace_seq *s, unsigned long long abs_usecs, static const char state_to_char[] = TASK_STATE_TO_CHAR_STR; -static notrace void +static notrace int print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) { struct trace_seq *s = &iter->seq; @@ -1154,10 +1156,10 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) default: trace_seq_printf(s, "Unknown type %d\n", entry->type); } + return 1; } -static notrace int -print_trace_fmt(struct trace_iterator *iter) +static notrace int print_trace_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); @@ -1222,6 +1224,43 @@ print_trace_fmt(struct trace_iterator *iter) return 1; } +static notrace int print_raw_fmt(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + struct trace_entry *entry; + int ret; + int S; + + entry = iter->ent; + + ret = trace_seq_printf(s, "%d %d %llu ", + entry->pid, iter->cpu, entry->t); + if (!ret) + return 0; + + switch (entry->type) { + case TRACE_FN: + ret = trace_seq_printf(s, "%x %x\n", + entry->fn.ip, entry->fn.parent_ip); + if (!ret) + return 0; + break; + case TRACE_CTX: + S = entry->ctx.prev_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.prev_state] : 'X'; + ret = trace_seq_printf(s, "%d %d %c %d %d\n", + entry->ctx.prev_pid, + entry->ctx.prev_prio, + S, + entry->ctx.next_pid, + entry->ctx.next_prio); + if (!ret) + return 0; + break; + } + return 1; +} + static int trace_empty(struct trace_iterator *iter) { struct trace_array_cpu *data; @@ -1238,6 +1277,17 @@ static int trace_empty(struct trace_iterator *iter) return 1; } +static int print_trace_line(struct trace_iterator *iter) +{ + if (trace_flags & TRACE_ITER_RAW) + return print_raw_fmt(iter); + + if (iter->iter_flags & TRACE_FILE_LAT_FMT) + return print_lat_fmt(iter, iter->idx, iter->cpu); + + return print_trace_fmt(iter); +} + static int s_show(struct seq_file *m, void *v) { struct trace_iterator *iter = v; @@ -1259,10 +1309,7 @@ static int s_show(struct seq_file *m, void *v) print_func_help_header(m); } } else { - if (iter->iter_flags & TRACE_FILE_LAT_FMT) - print_lat_fmt(iter, iter->idx, iter->cpu); - else - print_trace_fmt(iter); + print_trace_line(iter); trace_print_seq(m, &iter->seq); } @@ -1870,7 +1917,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, } while ((entry = find_next_entry_inc(iter)) != NULL) { - ret = print_trace_fmt(iter); + ret = print_trace_line(iter); if (!ret) break; -- cgit v1.2.3 From cb0f12aae8d085140d37ada351aa5a8e76c3f9b0 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: bin-output Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 50 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index bebd263f582f..d78cbc4fc519 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -154,6 +154,7 @@ enum trace_iterator_flags { TRACE_ITER_SYM_ADDR = 0x04, TRACE_ITER_VERBOSE = 0x08, TRACE_ITER_RAW = 0x10, + TRACE_ITER_BIN = 0x20, }; #define TRACE_ITER_SYM_MASK \ @@ -166,6 +167,7 @@ static const char *trace_options[] = { "sym-addr", "verbose", "raw", + "bin", NULL }; @@ -275,6 +277,18 @@ trace_seq_putc(struct trace_seq *s, unsigned char c) return 1; } +static notrace int +trace_seq_putmem(struct trace_seq *s, void *mem, size_t len) +{ + if (len > ((PAGE_SIZE - 1) - s->len)) + return 0; + + memcpy(s->buffer + s->len, mem, len); + s->len += len; + + return len; +} + static notrace void trace_seq_reset(struct trace_seq *s) { @@ -1261,6 +1275,39 @@ static notrace int print_raw_fmt(struct trace_iterator *iter) return 1; } +#define SEQ_PUT_FIELD_RET(s, x) \ +do { \ + if (!trace_seq_putmem(s, &(x), sizeof(x))) \ + return 0; \ +} while (0) + +static notrace int print_bin_fmt(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + struct trace_entry *entry; + + entry = iter->ent; + + SEQ_PUT_FIELD_RET(s, entry->pid); + SEQ_PUT_FIELD_RET(s, entry->cpu); + SEQ_PUT_FIELD_RET(s, entry->t); + + switch (entry->type) { + case TRACE_FN: + SEQ_PUT_FIELD_RET(s, entry->fn.ip); + SEQ_PUT_FIELD_RET(s, entry->fn.parent_ip); + break; + case TRACE_CTX: + SEQ_PUT_FIELD_RET(s, entry->ctx.prev_pid); + SEQ_PUT_FIELD_RET(s, entry->ctx.prev_prio); + SEQ_PUT_FIELD_RET(s, entry->ctx.prev_state); + SEQ_PUT_FIELD_RET(s, entry->ctx.next_pid); + SEQ_PUT_FIELD_RET(s, entry->ctx.next_prio); + break; + } + return 1; +} + static int trace_empty(struct trace_iterator *iter) { struct trace_array_cpu *data; @@ -1279,6 +1326,9 @@ static int trace_empty(struct trace_iterator *iter) static int print_trace_line(struct trace_iterator *iter) { + if (trace_flags & TRACE_ITER_BIN) + return print_bin_fmt(iter); + if (trace_flags & TRACE_ITER_RAW) return print_raw_fmt(iter); -- cgit v1.2.3 From f0a920d5752e1788c0cba2add103076bcc0f7a49 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: add trace_special() for ad-hoc tracing. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 44 ++++++++++++++++++++++++++++++++++++++++++++ kernel/trace/trace.h | 15 +++++++++++++++ 2 files changed, 59 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d78cbc4fc519..fa13059eb462 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -137,6 +137,7 @@ enum trace_type { TRACE_FN, TRACE_CTX, + TRACE_SPECIAL, __TRACE_LAST_TYPE }; @@ -700,6 +701,22 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, spin_unlock(&data->lock); } +notrace void +trace_special(struct trace_array *tr, struct trace_array_cpu *data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ + struct trace_entry *entry; + + spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_SPECIAL; + entry->special.arg1 = arg1; + entry->special.arg2 = arg2; + entry->special.arg3 = arg3; + spin_unlock(&data->lock); +} + notrace void tracing_sched_switch_trace(struct trace_array *tr, struct trace_array_cpu *data, @@ -1167,6 +1184,12 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) entry->ctx.next_prio, comm); break; + case TRACE_SPECIAL: + trace_seq_printf(s, " %lx %lx %lx\n", + entry->special.arg1, + entry->special.arg2, + entry->special.arg3); + break; default: trace_seq_printf(s, "Unknown type %d\n", entry->type); } @@ -1234,6 +1257,14 @@ static notrace int print_trace_fmt(struct trace_iterator *iter) if (!ret) return 0; break; + case TRACE_SPECIAL: + ret = trace_seq_printf(s, " %lx %lx %lx\n", + entry->special.arg1, + entry->special.arg2, + entry->special.arg3); + if (!ret) + return 0; + break; } return 1; } @@ -1271,6 +1302,14 @@ static notrace int print_raw_fmt(struct trace_iterator *iter) if (!ret) return 0; break; + case TRACE_SPECIAL: + ret = trace_seq_printf(s, " %lx %lx %lx\n", + entry->special.arg1, + entry->special.arg2, + entry->special.arg3); + if (!ret) + return 0; + break; } return 1; } @@ -1304,6 +1343,11 @@ static notrace int print_bin_fmt(struct trace_iterator *iter) SEQ_PUT_FIELD_RET(s, entry->ctx.next_pid); SEQ_PUT_FIELD_RET(s, entry->ctx.next_prio); break; + case TRACE_SPECIAL: + SEQ_PUT_FIELD_RET(s, entry->special.arg1); + SEQ_PUT_FIELD_RET(s, entry->special.arg2); + SEQ_PUT_FIELD_RET(s, entry->special.arg3); + break; } return 1; } diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 27fa2d06f499..7bdfef35c05a 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -25,6 +25,15 @@ struct ctx_switch_entry { unsigned char next_prio; }; +/* + * Special (free-form) trace entry: + */ +struct special_entry { + unsigned long arg1; + unsigned long arg2; + unsigned long arg3; +}; + /* * The trace entry - the most basic unit of tracing. This is what * is printed in the end as a single line in the trace output, such as: @@ -41,6 +50,7 @@ struct trace_entry { union { struct ftrace_entry fn; struct ctx_switch_entry ctx; + struct special_entry special; }; }; @@ -154,6 +164,11 @@ void tracing_sched_switch_trace(struct trace_array *tr, struct task_struct *next, unsigned long flags); void tracing_record_cmdline(struct task_struct *tsk); +void trace_special(struct trace_array *tr, + struct trace_array_cpu *data, + unsigned long arg1, + unsigned long arg2, + unsigned long arg3); void tracing_start_function_trace(void); void tracing_stop_function_trace(void); -- cgit v1.2.3 From dcb6308f2b56720599f6b9d5a01c33e67e69bde4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace, locking fix should be an irq-safe lock ... Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 15 +++++++++------ 1 file changed, 9 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index fa13059eb462..70f94fa92c10 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -691,14 +691,15 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, unsigned long parent_ip, unsigned long flags) { struct trace_entry *entry; + unsigned long irq_flags; - spin_lock(&data->lock); + spin_lock_irqsave(&data->lock, irq_flags); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_FN; entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; - spin_unlock(&data->lock); + spin_unlock_irqrestore(&data->lock, irq_flags); } notrace void @@ -706,15 +707,16 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, unsigned long arg1, unsigned long arg2, unsigned long arg3) { struct trace_entry *entry; + unsigned long irq_flags; - spin_lock(&data->lock); + spin_lock_irqsave(&data->lock, irq_flags); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_SPECIAL; entry->special.arg1 = arg1; entry->special.arg2 = arg2; entry->special.arg3 = arg3; - spin_unlock(&data->lock); + spin_unlock_irqrestore(&data->lock, irq_flags); } notrace void @@ -724,8 +726,9 @@ tracing_sched_switch_trace(struct trace_array *tr, unsigned long flags) { struct trace_entry *entry; + unsigned long irq_flags; - spin_lock(&data->lock); + spin_lock_irqsave(&data->lock, irq_flags); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_CTX; @@ -734,7 +737,7 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.prev_state = prev->state; entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; - spin_unlock(&data->lock); + spin_unlock_irqrestore(&data->lock, irq_flags); } enum trace_file_type { -- cgit v1.2.3 From 088b1e427dbba2af93cb6a7d39258c10ff58dd27 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: pipe fixes Some fixes for better output with the trace pipe. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 15 +++++++-------- 1 file changed, 7 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 70f94fa92c10..c56fc5e60133 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -770,11 +770,6 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, array = page_address(page); - /* Still possible to catch up to the tail */ - if (iter->next_idx[cpu] && array == data->trace_tail && - iter->next_page_idx[cpu] == data->trace_tail_idx) - return NULL; - WARN_ON(iter->next_page_idx[cpu] >= ENTRIES_PER_PAGE); return &array[iter->next_page_idx[cpu]]; } @@ -1921,7 +1916,6 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, struct trace_iterator *iter = filp->private_data; struct trace_array_cpu *data; static cpumask_t mask; - struct trace_entry *entry; static int start; unsigned long flags; int read = 0; @@ -2013,10 +2007,15 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cpu_set(cpu, mask); } - while ((entry = find_next_entry_inc(iter)) != NULL) { + while (find_next_entry_inc(iter) != NULL) { + int len = iter->seq.len; + ret = print_trace_line(iter); - if (!ret) + if (!ret) { + /* don't print partial lines */ + iter->seq.len = len; break; + } trace_consume(iter); -- cgit v1.2.3 From 37ad508419f0fdfda7b378756eb1f35cfd26d96d Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace - fix dynamic ftrace memory leak The ftrace dynamic function update allocates a record to store the instruction pointers that are being modified. If the modified instruction pointer fails to update, then the record is marked as failed and nothing more is done. Worse, if the modification fails, but the record ip function is still called, it will allocate a new record and try again. In just a matter of time, will this cause a serious memory leak and crash the system. This patch plugs this memory leak. When a record fails, it is included back into the pool of records to be used. Now a record may fail over and over again, but the number of allocated records will not increase. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 45 ++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 42 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index a15e068535f8..8e02aa690b2b 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -188,6 +188,8 @@ static int ftraced_suspend; static int ftrace_record_suspend; +static struct dyn_ftrace *ftrace_free_records; + static inline int notrace ftrace_ip_in_hash(unsigned long ip, unsigned long key) { @@ -211,8 +213,35 @@ ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) hlist_add_head(&node->node, &ftrace_hash[key]); } +static notrace void ftrace_free_rec(struct dyn_ftrace *rec) +{ + /* no locking, only called from kstop_machine */ + + rec->ip = (unsigned long)ftrace_free_records; + ftrace_free_records = rec; + rec->flags |= FTRACE_FL_FREE; +} + static notrace struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) { + struct dyn_ftrace *rec; + + /* First check for freed records */ + if (ftrace_free_records) { + rec = ftrace_free_records; + + /* todo, disable tracing altogether on this warning */ + if (unlikely(!(rec->flags & FTRACE_FL_FREE))) { + WARN_ON_ONCE(1); + ftrace_free_records = NULL; + return NULL; + } + + ftrace_free_records = (void *)rec->ip; + memset(rec, 0, sizeof(*rec)); + return rec; + } + if (ftrace_pages->index == ENTRIES_PER_PAGE) { if (!ftrace_pages->next) return NULL; @@ -356,8 +385,16 @@ __ftrace_replace_code(struct dyn_ftrace *rec, } failed = ftrace_modify_code(ip, old, new); - if (failed) - rec->flags |= FTRACE_FL_FAILED; + if (failed) { + unsigned long key; + /* It is possible that the function hasn't been converted yet */ + key = hash_long(ip, FTRACE_HASHBITS); + if (!ftrace_ip_in_hash(ip, key)) { + rec->flags |= FTRACE_FL_FAILED; + ftrace_free_rec(rec); + } + + } } static void notrace ftrace_replace_code(int enable) @@ -407,8 +444,10 @@ ftrace_code_disable(struct dyn_ftrace *rec) call = ftrace_call_replace(ip, MCOUNT_ADDR); failed = ftrace_modify_code(ip, call, nop); - if (failed) + if (failed) { rec->flags |= FTRACE_FL_FAILED; + ftrace_free_rec(rec); + } } static int notrace __ftrace_modify_code(void *data) -- cgit v1.2.3 From 4eebcc81a33fbc45e28542b50197ed7b3c486d90 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: disable tracing on failure Since ftrace touches practically every function. If we detect any anomaly, we want to fully disable ftrace. This patch adds code to try shutdown ftrace as much as possible without doing any more harm is something is detected not quite correct. This only kills ftrace, this patch does have checks for other parts of the tracer (irqsoff, wakeup, etc.). Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 112 ++++++++++++++++++++++++++++++++++++++---- kernel/trace/trace_selftest.c | 4 ++ 2 files changed, 107 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 8e02aa690b2b..ff42345dd78e 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -29,9 +29,16 @@ #include "trace.h" -int ftrace_enabled; +/* ftrace_enabled is a method to turn ftrace on or off */ +int ftrace_enabled __read_mostly; static int last_ftrace_enabled; +/* + * ftrace_disabled is set when an anomaly is discovered. + * ftrace_disabled is much stronger than ftrace_enabled. + */ +static int ftrace_disabled __read_mostly; + static DEFINE_SPINLOCK(ftrace_lock); static DEFINE_MUTEX(ftrace_sysctl_lock); @@ -230,10 +237,11 @@ static notrace struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) if (ftrace_free_records) { rec = ftrace_free_records; - /* todo, disable tracing altogether on this warning */ if (unlikely(!(rec->flags & FTRACE_FL_FREE))) { WARN_ON_ONCE(1); ftrace_free_records = NULL; + ftrace_disabled = 1; + ftrace_enabled = 0; return NULL; } @@ -260,7 +268,7 @@ ftrace_record_ip(unsigned long ip) int resched; int atomic; - if (!ftrace_enabled) + if (!ftrace_enabled || ftrace_disabled) return; resched = need_resched(); @@ -485,6 +493,9 @@ static void notrace ftrace_startup(void) { int command = 0; + if (unlikely(ftrace_disabled)) + return; + mutex_lock(&ftraced_lock); ftraced_suspend++; if (ftraced_suspend == 1) @@ -507,6 +518,9 @@ static void notrace ftrace_shutdown(void) { int command = 0; + if (unlikely(ftrace_disabled)) + return; + mutex_lock(&ftraced_lock); ftraced_suspend--; if (!ftraced_suspend) @@ -529,6 +543,9 @@ static void notrace ftrace_startup_sysctl(void) { int command = FTRACE_ENABLE_MCOUNT; + if (unlikely(ftrace_disabled)) + return; + mutex_lock(&ftraced_lock); /* Force update next time */ saved_ftrace_func = NULL; @@ -544,6 +561,9 @@ static void notrace ftrace_shutdown_sysctl(void) { int command = FTRACE_DISABLE_MCOUNT; + if (unlikely(ftrace_disabled)) + return; + mutex_lock(&ftraced_lock); /* ftraced_suspend is true if ftrace is running */ if (ftraced_suspend) @@ -600,6 +620,9 @@ static int notrace __ftrace_update_code(void *ignore) static void notrace ftrace_update_code(void) { + if (unlikely(ftrace_disabled)) + return; + stop_machine_run(__ftrace_update_code, NULL, NR_CPUS); } @@ -614,6 +637,9 @@ static int notrace ftraced(void *ignore) /* check once a second */ schedule_timeout(HZ); + if (unlikely(ftrace_disabled)) + continue; + mutex_lock(&ftrace_sysctl_lock); mutex_lock(&ftraced_lock); if (ftrace_enabled && ftraced_trigger && !ftraced_suspend) { @@ -628,6 +654,7 @@ static int notrace ftraced(void *ignore) ftrace_update_cnt != 1 ? "s" : "", ftrace_update_tot_cnt, usecs, usecs != 1 ? "s" : ""); + ftrace_disabled = 1; WARN_ON_ONCE(1); } ftraced_trigger = 0; @@ -785,6 +812,9 @@ ftrace_avail_open(struct inode *inode, struct file *file) struct ftrace_iterator *iter; int ret; + if (unlikely(ftrace_disabled)) + return -ENODEV; + iter = kzalloc(sizeof(*iter), GFP_KERNEL); if (!iter) return -ENOMEM; @@ -843,6 +873,9 @@ ftrace_filter_open(struct inode *inode, struct file *file) struct ftrace_iterator *iter; int ret = 0; + if (unlikely(ftrace_disabled)) + return -ENODEV; + iter = kzalloc(sizeof(*iter), GFP_KERNEL); if (!iter) return -ENOMEM; @@ -1063,6 +1096,9 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, */ notrace void ftrace_set_filter(unsigned char *buf, int len, int reset) { + if (unlikely(ftrace_disabled)) + return; + mutex_lock(&ftrace_filter_lock); if (reset) ftrace_filter_reset(); @@ -1133,7 +1169,7 @@ int ftrace_force_update(void) DECLARE_WAITQUEUE(wait, current); int ret = 0; - if (!ftraced_task) + if (unlikely(ftrace_disabled)) return -ENODEV; mutex_lock(&ftraced_lock); @@ -1142,6 +1178,11 @@ int ftrace_force_update(void) set_current_state(TASK_INTERRUPTIBLE); add_wait_queue(&ftraced_waiters, &wait); + if (unlikely(!ftraced_task)) { + ret = -ENODEV; + goto out; + } + do { mutex_unlock(&ftraced_lock); wake_up_process(ftraced_task); @@ -1154,6 +1195,7 @@ int ftrace_force_update(void) set_current_state(TASK_INTERRUPTIBLE); } while (last_counter == ftraced_iteration_counter); + out: mutex_unlock(&ftraced_lock); remove_wait_queue(&ftraced_waiters, &wait); set_current_state(TASK_RUNNING); @@ -1161,6 +1203,22 @@ int ftrace_force_update(void) return ret; } +static void ftrace_force_shutdown(void) +{ + struct task_struct *task; + int command = FTRACE_DISABLE_CALLS | FTRACE_UPDATE_TRACE_FUNC; + + mutex_lock(&ftraced_lock); + task = ftraced_task; + ftraced_task = NULL; + ftraced_suspend = -1; + ftrace_run_update_code(command); + mutex_unlock(&ftraced_lock); + + if (task) + kthread_stop(task); +} + static __init int ftrace_init_debugfs(void) { struct dentry *d_tracer; @@ -1194,21 +1252,29 @@ static int __init notrace ftrace_dynamic_init(void) stop_machine_run(ftrace_dyn_arch_init, &addr, NR_CPUS); /* ftrace_dyn_arch_init places the return code in addr */ - if (addr) - return addr; + if (addr) { + ret = (int)addr; + goto failed; + } ret = ftrace_dyn_table_alloc(); if (ret) - return ret; + goto failed; p = kthread_run(ftraced, NULL, "ftraced"); - if (IS_ERR(p)) - return -1; + if (IS_ERR(p)) { + ret = -1; + goto failed; + } last_ftrace_enabled = ftrace_enabled = 1; ftraced_task = p; return 0; + + failed: + ftrace_disabled = 1; + return ret; } core_initcall(ftrace_dynamic_init); @@ -1217,8 +1283,30 @@ core_initcall(ftrace_dynamic_init); # define ftrace_shutdown() do { } while (0) # define ftrace_startup_sysctl() do { } while (0) # define ftrace_shutdown_sysctl() do { } while (0) +# define ftrace_force_shutdown() do { } while (0) #endif /* CONFIG_DYNAMIC_FTRACE */ +/** + * ftrace_kill - totally shutdown ftrace + * + * This is a safety measure. If something was detected that seems + * wrong, calling this function will keep ftrace from doing + * any more modifications, and updates. + * used when something went wrong. + */ +void ftrace_kill(void) +{ + mutex_lock(&ftrace_sysctl_lock); + ftrace_disabled = 1; + ftrace_enabled = 0; + + clear_ftrace_function(); + mutex_unlock(&ftrace_sysctl_lock); + + /* Try to totally disable ftrace */ + ftrace_force_shutdown(); +} + /** * register_ftrace_function - register a function for profiling * @ops - ops structure that holds the function for profiling. @@ -1234,6 +1322,9 @@ int register_ftrace_function(struct ftrace_ops *ops) { int ret; + if (unlikely(ftrace_disabled)) + return -1; + mutex_lock(&ftrace_sysctl_lock); ret = __register_ftrace_function(ops); ftrace_startup(); @@ -1267,6 +1358,9 @@ ftrace_enable_sysctl(struct ctl_table *table, int write, { int ret; + if (unlikely(ftrace_disabled)) + return -ENODEV; + mutex_lock(&ftrace_sysctl_lock); ret = proc_dointvec(table, write, file, buffer, lenp, ppos); diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index a6f1ed75f836..85715b86a342 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -248,6 +248,10 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr) ftrace_enabled = save_ftrace_enabled; tracer_enabled = save_tracer_enabled; + /* kill ftrace totally if we failed */ + if (ret) + ftrace_kill(); + return ret; } #endif /* CONFIG_FTRACE */ -- cgit v1.2.3 From 26994ead1fc8cced63f17e9848edc1771036664e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: enabled tracing by default This patch is the correct way to have tracing enabled by default. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index c56fc5e60133..3dc6eac85b11 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -107,7 +107,7 @@ static struct trace_array max_tr; static DEFINE_PER_CPU(struct trace_array_cpu, max_data); -static int tracer_enabled; +static int tracer_enabled = 1; static unsigned long trace_nr_entries = 16384UL; static struct tracer *trace_types __read_mostly; @@ -2268,6 +2268,8 @@ __init static int tracer_alloc_buffers(void) int ret = -ENOMEM; int i; + global_trace.ctrl = tracer_enabled; + /* Allocate the first page for all buffers */ for_each_possible_cpu(i) { data = global_trace.data[i] = &per_cpu(global_trace_cpu, i); -- cgit v1.2.3 From 0fd9e0dac9026df09986a4b201518ae015814aef Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: use cpu clock again Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 54 +--------------------------------------------------- 1 file changed, 1 insertion(+), 53 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3dc6eac85b11..d74c039305ad 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -42,61 +42,9 @@ ns2usecs(cycle_t nsec) return nsec; } -static const int time_sync_freq_max = 128; -static const cycle_t time_sync_thresh = 100000; - -static DEFINE_PER_CPU(cycle_t, time_offset); -static DEFINE_PER_CPU(cycle_t, prev_cpu_time); -static DEFINE_PER_CPU(int, time_sync_count); -static DEFINE_PER_CPU(int, time_sync_freq); - -/* - * Global lock which we take every now and then to synchronize - * the CPUs time. This method is not warp-safe, but it's good - * enough to synchronize slowly diverging time sources and thus - * it's good enough for tracing: - */ -static DEFINE_SPINLOCK(time_sync_lock); -static cycle_t prev_global_time; - -static notrace cycle_t __ftrace_now_sync(cycles_t time, int cpu) -{ - unsigned long flags; - - spin_lock_irqsave(&time_sync_lock, flags); - - /* - * Update the synchronization frequency: - */ - if (per_cpu(time_sync_freq, cpu) < time_sync_freq_max) - per_cpu(time_sync_freq, cpu) *= 2; - per_cpu(time_sync_count, cpu) = per_cpu(time_sync_freq, cpu); - - if (time < prev_global_time) { - per_cpu(time_offset, cpu) += prev_global_time - time; - time = prev_global_time; - } else { - prev_global_time = time; - } - - spin_unlock_irqrestore(&time_sync_lock, flags); - - return time; -} - notrace cycle_t ftrace_now(int cpu) { - cycle_t prev_cpu_time, time, delta_time; - - prev_cpu_time = per_cpu(prev_cpu_time, cpu); - time = sched_clock() + per_cpu(time_offset, cpu); - delta_time = time-prev_cpu_time; - - if (unlikely(delta_time > time_sync_thresh || - --per_cpu(time_sync_count, cpu) <= 0)) - time = __ftrace_now_sync(time, cpu); - - return time; + return cpu_clock(cpu); } static struct trace_array global_trace; -- cgit v1.2.3 From 2e0f57618529a2739a5e1570e6c445c9c966b595 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: build fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 111 ++++++++++++++++++++++++----------------- kernel/trace/trace_functions.c | 2 +- 2 files changed, 67 insertions(+), 46 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d74c039305ad..71b25b79b3de 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -432,47 +432,6 @@ notrace void tracing_reset(struct trace_array_cpu *data) data->trace_tail_idx = 0; } -#ifdef CONFIG_FTRACE -static notrace void -function_trace_call(unsigned long ip, unsigned long parent_ip) -{ - struct trace_array *tr = &global_trace; - struct trace_array_cpu *data; - unsigned long flags; - long disabled; - int cpu; - - if (unlikely(!tracer_enabled)) - return; - - local_irq_save(flags); - cpu = raw_smp_processor_id(); - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - - if (likely(disabled == 1)) - ftrace(tr, data, ip, parent_ip, flags); - - atomic_dec(&data->disabled); - local_irq_restore(flags); -} - -static struct ftrace_ops trace_ops __read_mostly = -{ - .func = function_trace_call, -}; -#endif - -notrace void tracing_start_function_trace(void) -{ - register_ftrace_function(&trace_ops); -} - -notrace void tracing_stop_function_trace(void) -{ - unregister_ftrace_function(&trace_ops); -} - #define SAVED_CMDLINES 128 static unsigned map_pid_to_cmdline[PID_MAX_DEFAULT+1]; static unsigned map_cmdline_to_pid[SAVED_CMDLINES]; @@ -635,8 +594,8 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) } notrace void -ftrace(struct trace_array *tr, struct trace_array_cpu *data, - unsigned long ip, unsigned long parent_ip, unsigned long flags) +__ftrace(struct trace_array *tr, struct trace_array_cpu *data, + unsigned long ip, unsigned long parent_ip, unsigned long flags) { struct trace_entry *entry; unsigned long irq_flags; @@ -650,6 +609,14 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, spin_unlock_irqrestore(&data->lock, irq_flags); } +notrace void +ftrace(struct trace_array *tr, struct trace_array_cpu *data, + unsigned long ip, unsigned long parent_ip, unsigned long flags) +{ + if (likely(!atomic_read(&data->disabled))) + __ftrace(tr, data, ip, parent_ip, flags); +} + notrace void trace_special(struct trace_array *tr, struct trace_array_cpu *data, unsigned long arg1, unsigned long arg2, unsigned long arg3) @@ -688,6 +655,47 @@ tracing_sched_switch_trace(struct trace_array *tr, spin_unlock_irqrestore(&data->lock, irq_flags); } +#ifdef CONFIG_FTRACE +static notrace void +function_trace_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = &global_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (unlikely(!tracer_enabled)) + return; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + __ftrace(tr, data, ip, parent_ip, flags); + + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + +static struct ftrace_ops trace_ops __read_mostly = +{ + .func = function_trace_call, +}; + +notrace void tracing_start_function_trace(void) +{ + register_ftrace_function(&trace_ops); +} + +notrace void tracing_stop_function_trace(void) +{ + unregister_ftrace_function(&trace_ops); +} +#endif + enum trace_file_type { TRACE_FILE_LAT_FMT = 1, }; @@ -722,7 +730,7 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, return &array[iter->next_page_idx[cpu]]; } -static struct notrace trace_entry * +static struct trace_entry * notrace find_next_entry(struct trace_iterator *iter, int *ent_cpu) { struct trace_array *tr = iter->tr; @@ -1866,6 +1874,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, static cpumask_t mask; static int start; unsigned long flags; + int ftrace_save; int read = 0; int cpu; int len; @@ -1944,6 +1953,9 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cpus_clear(mask); local_irq_save(flags); + ftrace_save = ftrace_enabled; + ftrace_enabled = 0; + smp_wmb(); for_each_possible_cpu(cpu) { data = iter->tr->data[cpu]; @@ -1951,10 +1963,14 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, continue; atomic_inc(&data->disabled); - spin_lock(&data->lock); cpu_set(cpu, mask); } + for_each_cpu_mask(cpu, mask) { + data = iter->tr->data[cpu]; + spin_lock(&data->lock); + } + while (find_next_entry_inc(iter) != NULL) { int len = iter->seq.len; @@ -1974,8 +1990,13 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, for_each_cpu_mask(cpu, mask) { data = iter->tr->data[cpu]; spin_unlock(&data->lock); + } + + for_each_cpu_mask(cpu, mask) { + data = iter->tr->data[cpu]; atomic_dec(&data->disabled); } + ftrace_enabled = ftrace_save; local_irq_restore(flags); /* Now copy what we have to the user */ diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index e5d34b78fc99..69a0eb00a0a5 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -9,10 +9,10 @@ * Copyright (C) 2004-2006 Ingo Molnar * Copyright (C) 2004 William Lee Irwin III */ -#include #include #include #include +#include #include "trace.h" -- cgit v1.2.3 From 5e3ca0ec76fce92daa4eed0d02de9c79b1fe3920 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: introduce the "hex" output method Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 93 +++++++++++++++++++++++++++++++++++++++++++++++++++- 1 file changed, 92 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 71b25b79b3de..6974b212e938 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -103,7 +103,8 @@ enum trace_iterator_flags { TRACE_ITER_SYM_ADDR = 0x04, TRACE_ITER_VERBOSE = 0x08, TRACE_ITER_RAW = 0x10, - TRACE_ITER_BIN = 0x20, + TRACE_ITER_HEX = 0x20, + TRACE_ITER_BIN = 0x40, }; #define TRACE_ITER_SYM_MASK \ @@ -116,6 +117,7 @@ static const char *trace_options[] = { "sym-addr", "verbose", "raw", + "hex", "bin", NULL }; @@ -238,6 +240,47 @@ trace_seq_putmem(struct trace_seq *s, void *mem, size_t len) return len; } +#define HEX_CHARS 17 + +static notrace int +trace_seq_putmem_hex(struct trace_seq *s, void *mem, size_t len) +{ + unsigned char hex[HEX_CHARS]; + unsigned char *data; + unsigned char byte; + int i, j; + + BUG_ON(len >= HEX_CHARS); + + data = mem; + +#ifdef __BIG_ENDIAN + for (i = 0, j = 0; i < len; i++) { +#else + for (i = len-1, j = 0; i >= 0; i--) { +#endif + byte = data[i]; + + hex[j] = byte & 0x0f; + if (hex[j] >= 10) + hex[j] += 'a' - 10; + else + hex[j] += '0'; + j++; + + hex[j] = byte >> 4; + if (hex[j] >= 10) + hex[j] += 'a' - 10; + else + hex[j] += '0'; + j++; + } + hex[j] = ' '; + j++; + + return trace_seq_putmem(s, hex, j); +} + static notrace void trace_seq_reset(struct trace_seq *s) { @@ -1274,6 +1317,51 @@ do { \ return 0; \ } while (0) +#define SEQ_PUT_HEX_FIELD_RET(s, x) \ +do { \ + if (!trace_seq_putmem_hex(s, &(x), sizeof(x))) \ + return 0; \ +} while (0) + +static notrace int print_hex_fmt(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + unsigned char newline = '\n'; + struct trace_entry *entry; + int S; + + entry = iter->ent; + + SEQ_PUT_HEX_FIELD_RET(s, entry->pid); + SEQ_PUT_HEX_FIELD_RET(s, iter->cpu); + SEQ_PUT_HEX_FIELD_RET(s, entry->t); + + switch (entry->type) { + case TRACE_FN: + SEQ_PUT_HEX_FIELD_RET(s, entry->fn.ip); + SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip); + break; + case TRACE_CTX: + S = entry->ctx.prev_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.prev_state] : 'X'; + SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_pid); + SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_prio); + SEQ_PUT_HEX_FIELD_RET(s, S); + SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_pid); + SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_prio); + SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip); + break; + case TRACE_SPECIAL: + SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg1); + SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg2); + SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg3); + break; + } + SEQ_PUT_FIELD_RET(s, newline); + + return 1; +} + static notrace int print_bin_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; @@ -1327,6 +1415,9 @@ static int print_trace_line(struct trace_iterator *iter) if (trace_flags & TRACE_ITER_BIN) return print_bin_fmt(iter); + if (trace_flags & TRACE_ITER_HEX) + return print_hex_fmt(iter); + if (trace_flags & TRACE_ITER_RAW) return print_raw_fmt(iter); -- cgit v1.2.3 From 2577046740fe6d77864128c6187c11125c2449ea Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: build fix no need to backmerge, only affects ftrace-enabled kernels. (which is not the default) Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 6974b212e938..958c4d77a67b 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1965,7 +1965,9 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, static cpumask_t mask; static int start; unsigned long flags; +#ifdef CONFIG_FTRACE int ftrace_save; +#endif int read = 0; int cpu; int len; @@ -2044,8 +2046,10 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cpus_clear(mask); local_irq_save(flags); +#ifdef CONFIG_FTRACE ftrace_save = ftrace_enabled; ftrace_enabled = 0; +#endif smp_wmb(); for_each_possible_cpu(cpu) { data = iter->tr->data[cpu]; @@ -2087,7 +2091,9 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, data = iter->tr->data[cpu]; atomic_dec(&data->disabled); } +#ifdef CONFIG_FTRACE ftrace_enabled = ftrace_save; +#endif local_irq_restore(flags); /* Now copy what we have to the user */ -- cgit v1.2.3 From 2a2cc8f7c4d0dfd75720867f7dc58d24f075edfc Mon Sep 17 00:00:00 2001 From: Soeren Sandmann Pedersen Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: allow the event pipe to be polled Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 39 ++++++++++++++++++++++++++++++++++++++- 1 file changed, 38 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 958c4d77a67b..d041578affd0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -24,6 +24,7 @@ #include #include #include +#include #include #include @@ -63,6 +64,7 @@ static struct tracer *current_trace __read_mostly; static int max_tracer_type_len; static DEFINE_MUTEX(trace_types_lock); +static DECLARE_WAIT_QUEUE_HEAD (trace_wait); #define ENTRIES_PER_PAGE (PAGE_SIZE / sizeof(struct trace_entry)) @@ -105,6 +107,7 @@ enum trace_iterator_flags { TRACE_ITER_RAW = 0x10, TRACE_ITER_HEX = 0x20, TRACE_ITER_BIN = 0x40, + TRACE_ITER_BLOCK = 0x80, }; #define TRACE_ITER_SYM_MASK \ @@ -119,6 +122,7 @@ static const char *trace_options[] = { "raw", "hex", "bin", + "block", NULL }; @@ -650,6 +654,9 @@ __ftrace(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); + + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up (&trace_wait); } notrace void @@ -675,6 +682,9 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, entry->special.arg2 = arg2; entry->special.arg3 = arg3; spin_unlock_irqrestore(&data->lock, irq_flags); + + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up (&trace_wait); } notrace void @@ -696,6 +706,9 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; spin_unlock_irqrestore(&data->lock, irq_flags); + + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up (&trace_wait); } #ifdef CONFIG_FTRACE @@ -1765,7 +1778,6 @@ static struct file_operations tracing_readme_fops = { .read = tracing_readme_read, }; - static ssize_t tracing_ctrl_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) @@ -1953,6 +1965,28 @@ static int tracing_release_pipe(struct inode *inode, struct file *file) return 0; } +static unsigned int +tracing_poll_pipe(struct file *filp, poll_table *poll_table) +{ + struct trace_iterator *iter = filp->private_data; + + if (trace_flags & TRACE_ITER_BLOCK) { + /* + * Always select as readable when in blocking mode + */ + return POLLIN | POLLRDNORM; + } + else { + if (!trace_empty(iter)) + return POLLIN | POLLRDNORM; + poll_wait(filp, &trace_wait, poll_table); + if (!trace_empty(iter)) + return POLLIN | POLLRDNORM; + + return 0; + } +} + /* * Consumer reader. */ @@ -1991,6 +2025,8 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, start = 0; while (trace_empty(iter)) { + if (!(trace_flags & TRACE_ITER_BLOCK)) + return -EWOULDBLOCK; /* * This is a make-shift waitqueue. The reason we don't use * an actual wait queue is because: @@ -2134,6 +2170,7 @@ static struct file_operations set_tracer_fops = { static struct file_operations tracing_pipe_fops = { .open = tracing_open_pipe, + .poll = tracing_poll_pipe, .read = tracing_read_pipe, .release = tracing_release_pipe, }; -- cgit v1.2.3 From 6fb44b717c10ecf37beaaebd312f3afa93fed714 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: add trace_function api for other tracers to use A new check was added in the ftrace function that wont trace if the CPU trace buffer is disabled. Unfortunately, other tracers used ftrace() to write to the buffer after they disabled it. The new disable check makes these calls into a nop. This patch changes the __ftrace that is called without the check into a new api for the other tracers to use, called "trace_function". The other tracers use this interface instead when the trace CPU buffer is already disabled. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 8 ++++---- kernel/trace/trace.h | 5 +++++ kernel/trace/trace_irqsoff.c | 10 +++++----- kernel/trace/trace_sched_wakeup.c | 5 +++-- 4 files changed, 17 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d041578affd0..9022c357032a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -641,8 +641,8 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) } notrace void -__ftrace(struct trace_array *tr, struct trace_array_cpu *data, - unsigned long ip, unsigned long parent_ip, unsigned long flags) +trace_function(struct trace_array *tr, struct trace_array_cpu *data, + unsigned long ip, unsigned long parent_ip, unsigned long flags) { struct trace_entry *entry; unsigned long irq_flags; @@ -664,7 +664,7 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, unsigned long parent_ip, unsigned long flags) { if (likely(!atomic_read(&data->disabled))) - __ftrace(tr, data, ip, parent_ip, flags); + trace_function(tr, data, ip, parent_ip, flags); } notrace void @@ -730,7 +730,7 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) disabled = atomic_inc_return(&data->disabled); if (likely(disabled == 1)) - __ftrace(tr, data, ip, parent_ip, flags); + trace_function(tr, data, ip, parent_ip, flags); atomic_dec(&data->disabled); local_irq_restore(flags); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 7bdfef35c05a..faf9f67246ac 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -169,6 +169,11 @@ void trace_special(struct trace_array *tr, unsigned long arg1, unsigned long arg2, unsigned long arg3); +void trace_function(struct trace_array *tr, + struct trace_array_cpu *data, + unsigned long ip, + unsigned long parent_ip, + unsigned long flags); void tracing_start_function_trace(void); void tracing_stop_function_trace(void); diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index d2a6e6f1ad2d..3269f4ff5172 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -95,7 +95,7 @@ irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) disabled = atomic_inc_return(&data->disabled); if (likely(disabled == 1)) - ftrace(tr, data, ip, parent_ip, flags); + trace_function(tr, data, ip, parent_ip, flags); atomic_dec(&data->disabled); } @@ -150,7 +150,7 @@ check_critical_timing(struct trace_array *tr, if (!report_latency(delta)) goto out_unlock; - ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); + trace_function(tr, data, CALLER_ADDR0, parent_ip, flags); latency = nsecs_to_usecs(delta); @@ -188,7 +188,7 @@ out: data->critical_sequence = max_sequence; data->preempt_timestamp = ftrace_now(cpu); tracing_reset(data); - ftrace(tr, data, CALLER_ADDR0, parent_ip, flags); + trace_function(tr, data, CALLER_ADDR0, parent_ip, flags); } static inline void notrace @@ -221,7 +221,7 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) local_save_flags(flags); - ftrace(tr, data, ip, parent_ip, flags); + trace_function(tr, data, ip, parent_ip, flags); __get_cpu_var(tracing_cpu) = 1; @@ -254,7 +254,7 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) atomic_inc(&data->disabled); local_save_flags(flags); - ftrace(tr, data, ip, parent_ip, flags); + trace_function(tr, data, ip, parent_ip, flags); check_critical_timing(tr, data, parent_ip ? : ip, cpu); data->critical_start = 0; atomic_dec(&data->disabled); diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index b7df825c3af9..3549e4154f1f 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -85,7 +85,7 @@ wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) if (unlikely(!tracer_enabled || next != wakeup_task)) goto out_unlock; - ftrace(tr, data, CALLER_ADDR1, CALLER_ADDR2, flags); + trace_function(tr, data, CALLER_ADDR1, CALLER_ADDR2, flags); /* * usecs conversion is slow so we try to delay the conversion @@ -192,7 +192,8 @@ wakeup_check_start(struct trace_array *tr, struct task_struct *p, local_save_flags(flags); tr->data[wakeup_cpu]->preempt_timestamp = ftrace_now(cpu); - ftrace(tr, tr->data[wakeup_cpu], CALLER_ADDR1, CALLER_ADDR2, flags); + trace_function(tr, tr->data[wakeup_cpu], + CALLER_ADDR1, CALLER_ADDR2, flags); out_locked: spin_unlock(&wakeup_lock); -- cgit v1.2.3 From 9ff9cdb2d3b0971f89e899b3420aadd91bddc215 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:50 +0200 Subject: ftrace: cleanups clean up recent code. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 5 +++-- kernel/trace/trace_irqsoff.c | 2 +- kernel/trace/trace_selftest.c | 2 +- 3 files changed, 5 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index ff42345dd78e..425b1fec3d83 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -197,8 +197,8 @@ static int ftrace_record_suspend; static struct dyn_ftrace *ftrace_free_records; -static inline int -notrace ftrace_ip_in_hash(unsigned long ip, unsigned long key) +static inline int notrace +ftrace_ip_in_hash(unsigned long ip, unsigned long key) { struct dyn_ftrace *p; struct hlist_node *t; @@ -1249,6 +1249,7 @@ static int __init notrace ftrace_dynamic_init(void) int ret; addr = (unsigned long)ftrace_record_ip; + stop_machine_run(ftrace_dyn_arch_init, &addr, NR_CPUS); /* ftrace_dyn_arch_init places the return code in addr */ diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 3269f4ff5172..2ac0d09db6fb 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -309,7 +309,7 @@ void trace_softirqs_off(unsigned long ip) { } -inline void print_irqtrace_events(struct task_struct *curr) +inline notrace void print_irqtrace_events(struct task_struct *curr) { } diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 85715b86a342..546307de6e3d 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -3,7 +3,7 @@ #include #include -static inline int trace_valid_entry(struct trace_entry *entry) +static notrace inline int trace_valid_entry(struct trace_entry *entry) { switch (entry->type) { case TRACE_FN: -- cgit v1.2.3 From caf8cdebfb6c1cff50ea8077f1a07c2333d6d1fd Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:50 +0200 Subject: ftrace: remove address of function names PowerPC is very fragile when it comes to use of function names and function addresses. ftrace needs to either use all function addresses or function names (i.e. my_func as suppose to &my_func). This patch chooses to use the names and not the addresses, and makes ftrace consistent. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 425b1fec3d83..57350cbd1f61 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -326,8 +326,8 @@ ftrace_record_ip(unsigned long ip) preempt_enable_notrace(); } -#define FTRACE_ADDR ((long)(&ftrace_caller)) -#define MCOUNT_ADDR ((long)(&mcount)) +#define FTRACE_ADDR ((long)(ftrace_caller)) +#define MCOUNT_ADDR ((long)(mcount)) static void notrace __ftrace_replace_code(struct dyn_ftrace *rec, -- cgit v1.2.3 From b53dde9d34f2df396540988ebc65c33400f57b04 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: disable -pg for the tracer itself Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Makefile | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index 3fec653d6533..c25a6cd6a529 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -1,3 +1,11 @@ + +# Do not instrument the tracer itself: + +ifdef CONFIG_FTRACE +ORIG_CFLAGS := $(KBUILD_CFLAGS) +KBUILD_CFLAGS = $(subst -pg,,$(ORIG_CFLAGS)) +endif + obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o -- cgit v1.2.3 From e309b41dd65aa953f86765eeeecc941d8e1e8b8f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: remove notrace now that we have a kbuild method for notrace, no need to pollute the C code with the annotations. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 66 +++++++++++++-------------- kernel/trace/trace.c | 94 +++++++++++++++++++-------------------- kernel/trace/trace.h | 6 +-- kernel/trace/trace_functions.c | 12 ++--- kernel/trace/trace_irqsoff.c | 40 ++++++++--------- kernel/trace/trace_sched_switch.c | 12 ++--- kernel/trace/trace_sched_wakeup.c | 28 ++++++------ kernel/trace/trace_selftest.c | 2 +- 8 files changed, 130 insertions(+), 130 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 57350cbd1f61..281d97a3208c 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -53,7 +53,7 @@ ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub; /* mcount is defined per arch in assembly */ EXPORT_SYMBOL(mcount); -notrace void ftrace_list_func(unsigned long ip, unsigned long parent_ip) +void ftrace_list_func(unsigned long ip, unsigned long parent_ip) { struct ftrace_ops *op = ftrace_list; @@ -79,7 +79,7 @@ void clear_ftrace_function(void) ftrace_trace_function = ftrace_stub; } -static int notrace __register_ftrace_function(struct ftrace_ops *ops) +static int __register_ftrace_function(struct ftrace_ops *ops) { /* Should never be called by interrupts */ spin_lock(&ftrace_lock); @@ -110,7 +110,7 @@ static int notrace __register_ftrace_function(struct ftrace_ops *ops) return 0; } -static int notrace __unregister_ftrace_function(struct ftrace_ops *ops) +static int __unregister_ftrace_function(struct ftrace_ops *ops) { struct ftrace_ops **p; int ret = 0; @@ -197,7 +197,7 @@ static int ftrace_record_suspend; static struct dyn_ftrace *ftrace_free_records; -static inline int notrace +static inline int ftrace_ip_in_hash(unsigned long ip, unsigned long key) { struct dyn_ftrace *p; @@ -214,13 +214,13 @@ ftrace_ip_in_hash(unsigned long ip, unsigned long key) return found; } -static inline void notrace +static inline void ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) { hlist_add_head(&node->node, &ftrace_hash[key]); } -static notrace void ftrace_free_rec(struct dyn_ftrace *rec) +static void ftrace_free_rec(struct dyn_ftrace *rec) { /* no locking, only called from kstop_machine */ @@ -229,7 +229,7 @@ static notrace void ftrace_free_rec(struct dyn_ftrace *rec) rec->flags |= FTRACE_FL_FREE; } -static notrace struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) +static struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) { struct dyn_ftrace *rec; @@ -259,7 +259,7 @@ static notrace struct dyn_ftrace *ftrace_alloc_dyn_node(unsigned long ip) return &ftrace_pages->records[ftrace_pages->index++]; } -static void notrace +static void ftrace_record_ip(unsigned long ip) { struct dyn_ftrace *node; @@ -329,7 +329,7 @@ ftrace_record_ip(unsigned long ip) #define FTRACE_ADDR ((long)(ftrace_caller)) #define MCOUNT_ADDR ((long)(mcount)) -static void notrace +static void __ftrace_replace_code(struct dyn_ftrace *rec, unsigned char *old, unsigned char *new, int enable) { @@ -405,7 +405,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, } } -static void notrace ftrace_replace_code(int enable) +static void ftrace_replace_code(int enable) { unsigned char *new = NULL, *old = NULL; struct dyn_ftrace *rec; @@ -430,7 +430,7 @@ static void notrace ftrace_replace_code(int enable) } } -static notrace void ftrace_shutdown_replenish(void) +static void ftrace_shutdown_replenish(void) { if (ftrace_pages->next) return; @@ -439,7 +439,7 @@ static notrace void ftrace_shutdown_replenish(void) ftrace_pages->next = (void *)get_zeroed_page(GFP_KERNEL); } -static notrace void +static void ftrace_code_disable(struct dyn_ftrace *rec) { unsigned long ip; @@ -458,7 +458,7 @@ ftrace_code_disable(struct dyn_ftrace *rec) } } -static int notrace __ftrace_modify_code(void *data) +static int __ftrace_modify_code(void *data) { unsigned long addr; int *command = data; @@ -482,14 +482,14 @@ static int notrace __ftrace_modify_code(void *data) return 0; } -static void notrace ftrace_run_update_code(int command) +static void ftrace_run_update_code(int command) { stop_machine_run(__ftrace_modify_code, &command, NR_CPUS); } static ftrace_func_t saved_ftrace_func; -static void notrace ftrace_startup(void) +static void ftrace_startup(void) { int command = 0; @@ -514,7 +514,7 @@ static void notrace ftrace_startup(void) mutex_unlock(&ftraced_lock); } -static void notrace ftrace_shutdown(void) +static void ftrace_shutdown(void) { int command = 0; @@ -539,7 +539,7 @@ static void notrace ftrace_shutdown(void) mutex_unlock(&ftraced_lock); } -static void notrace ftrace_startup_sysctl(void) +static void ftrace_startup_sysctl(void) { int command = FTRACE_ENABLE_MCOUNT; @@ -557,7 +557,7 @@ static void notrace ftrace_startup_sysctl(void) mutex_unlock(&ftraced_lock); } -static void notrace ftrace_shutdown_sysctl(void) +static void ftrace_shutdown_sysctl(void) { int command = FTRACE_DISABLE_MCOUNT; @@ -577,7 +577,7 @@ static cycle_t ftrace_update_time; static unsigned long ftrace_update_cnt; unsigned long ftrace_update_tot_cnt; -static int notrace __ftrace_update_code(void *ignore) +static int __ftrace_update_code(void *ignore) { struct dyn_ftrace *p; struct hlist_head head; @@ -618,7 +618,7 @@ static int notrace __ftrace_update_code(void *ignore) return 0; } -static void notrace ftrace_update_code(void) +static void ftrace_update_code(void) { if (unlikely(ftrace_disabled)) return; @@ -626,7 +626,7 @@ static void notrace ftrace_update_code(void) stop_machine_run(__ftrace_update_code, NULL, NR_CPUS); } -static int notrace ftraced(void *ignore) +static int ftraced(void *ignore) { unsigned long usecs; @@ -733,7 +733,7 @@ struct ftrace_iterator { unsigned filtered; }; -static void notrace * +static void * t_next(struct seq_file *m, void *v, loff_t *pos) { struct ftrace_iterator *iter = m->private; @@ -806,7 +806,7 @@ static struct seq_operations show_ftrace_seq_ops = { .show = t_show, }; -static int notrace +static int ftrace_avail_open(struct inode *inode, struct file *file) { struct ftrace_iterator *iter; @@ -845,7 +845,7 @@ int ftrace_avail_release(struct inode *inode, struct file *file) return 0; } -static void notrace ftrace_filter_reset(void) +static void ftrace_filter_reset(void) { struct ftrace_page *pg; struct dyn_ftrace *rec; @@ -867,7 +867,7 @@ static void notrace ftrace_filter_reset(void) preempt_enable(); } -static int notrace +static int ftrace_filter_open(struct inode *inode, struct file *file) { struct ftrace_iterator *iter; @@ -903,7 +903,7 @@ ftrace_filter_open(struct inode *inode, struct file *file) return ret; } -static ssize_t notrace +static ssize_t ftrace_filter_read(struct file *file, char __user *ubuf, size_t cnt, loff_t *ppos) { @@ -913,7 +913,7 @@ ftrace_filter_read(struct file *file, char __user *ubuf, return -EPERM; } -static loff_t notrace +static loff_t ftrace_filter_lseek(struct file *file, loff_t offset, int origin) { loff_t ret; @@ -933,7 +933,7 @@ enum { MATCH_END_ONLY, }; -static void notrace +static void ftrace_match(unsigned char *buff, int len) { char str[KSYM_SYMBOL_LEN]; @@ -1002,7 +1002,7 @@ ftrace_match(unsigned char *buff, int len) preempt_enable(); } -static ssize_t notrace +static ssize_t ftrace_filter_write(struct file *file, const char __user *ubuf, size_t cnt, loff_t *ppos) { @@ -1094,7 +1094,7 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, * Filters denote which functions should be enabled when tracing is enabled. * If @buf is NULL and reset is set, all functions will be enabled for tracing. */ -notrace void ftrace_set_filter(unsigned char *buf, int len, int reset) +void ftrace_set_filter(unsigned char *buf, int len, int reset) { if (unlikely(ftrace_disabled)) return; @@ -1107,7 +1107,7 @@ notrace void ftrace_set_filter(unsigned char *buf, int len, int reset) mutex_unlock(&ftrace_filter_lock); } -static int notrace +static int ftrace_filter_release(struct inode *inode, struct file *file) { struct seq_file *m = (struct seq_file *)file->private_data; @@ -1242,7 +1242,7 @@ static __init int ftrace_init_debugfs(void) fs_initcall(ftrace_init_debugfs); -static int __init notrace ftrace_dynamic_init(void) +static int __init ftrace_dynamic_init(void) { struct task_struct *p; unsigned long addr; @@ -1352,7 +1352,7 @@ int unregister_ftrace_function(struct ftrace_ops *ops) return ret; } -notrace int +int ftrace_enable_sysctl(struct ctl_table *table, int write, struct file *file, void __user *buffer, size_t *lenp, loff_t *ppos) diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9022c357032a..f5898051fdd9 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -35,7 +35,7 @@ unsigned long __read_mostly tracing_thresh; static int tracing_disabled = 1; -static long notrace +static long ns2usecs(cycle_t nsec) { nsec += 500; @@ -43,7 +43,7 @@ ns2usecs(cycle_t nsec) return nsec; } -notrace cycle_t ftrace_now(int cpu) +cycle_t ftrace_now(int cpu) { return cpu_clock(cpu); } @@ -135,7 +135,7 @@ static DEFINE_SPINLOCK(ftrace_max_lock); * structure. (this way the maximum trace is permanently saved, * for later retrieval via /debugfs/tracing/latency_trace) */ -static notrace void +static void __update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data = tr->data[cpu]; @@ -184,7 +184,7 @@ void *head_page(struct trace_array_cpu *data) return page_address(page); } -static notrace int +static int trace_seq_printf(struct trace_seq *s, const char *fmt, ...) { int len = (PAGE_SIZE - 1) - s->len; @@ -207,7 +207,7 @@ trace_seq_printf(struct trace_seq *s, const char *fmt, ...) return len; } -static notrace int +static int trace_seq_puts(struct trace_seq *s, const char *str) { int len = strlen(str); @@ -221,7 +221,7 @@ trace_seq_puts(struct trace_seq *s, const char *str) return len; } -static notrace int +static int trace_seq_putc(struct trace_seq *s, unsigned char c) { if (s->len >= (PAGE_SIZE - 1)) @@ -232,7 +232,7 @@ trace_seq_putc(struct trace_seq *s, unsigned char c) return 1; } -static notrace int +static int trace_seq_putmem(struct trace_seq *s, void *mem, size_t len) { if (len > ((PAGE_SIZE - 1) - s->len)) @@ -246,7 +246,7 @@ trace_seq_putmem(struct trace_seq *s, void *mem, size_t len) #define HEX_CHARS 17 -static notrace int +static int trace_seq_putmem_hex(struct trace_seq *s, void *mem, size_t len) { unsigned char hex[HEX_CHARS]; @@ -285,13 +285,13 @@ trace_seq_putmem_hex(struct trace_seq *s, void *mem, size_t len) return trace_seq_putmem(s, hex, j); } -static notrace void +static void trace_seq_reset(struct trace_seq *s) { s->len = 0; } -static notrace void +static void trace_print_seq(struct seq_file *m, struct trace_seq *s) { int len = s->len >= PAGE_SIZE ? PAGE_SIZE - 1 : s->len; @@ -302,7 +302,7 @@ trace_print_seq(struct seq_file *m, struct trace_seq *s) trace_seq_reset(s); } -notrace static void +static void flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) { struct list_head flip_pages; @@ -323,7 +323,7 @@ flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) check_pages(tr2); } -notrace void +void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data; @@ -348,7 +348,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) * @tsk - task with the latency * @cpu - the cpu of the buffer to copy. */ -notrace void +void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) { struct trace_array_cpu *data = tr->data[cpu]; @@ -471,7 +471,7 @@ void unregister_tracer(struct tracer *type) mutex_unlock(&trace_types_lock); } -notrace void tracing_reset(struct trace_array_cpu *data) +void tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; data->trace_head = data->trace_tail = head_page(data); @@ -494,9 +494,9 @@ static void trace_init_cmdlines(void) cmdline_idx = 0; } -notrace void trace_stop_cmdline_recording(void); +void trace_stop_cmdline_recording(void); -static notrace void trace_save_cmdline(struct task_struct *tsk) +static void trace_save_cmdline(struct task_struct *tsk) { unsigned map; unsigned idx; @@ -531,7 +531,7 @@ static notrace void trace_save_cmdline(struct task_struct *tsk) spin_unlock(&trace_cmdline_lock); } -static notrace char *trace_find_cmdline(int pid) +static char *trace_find_cmdline(int pid) { char *cmdline = "<...>"; unsigned map; @@ -552,7 +552,7 @@ static notrace char *trace_find_cmdline(int pid) return cmdline; } -notrace void tracing_record_cmdline(struct task_struct *tsk) +void tracing_record_cmdline(struct task_struct *tsk) { if (atomic_read(&trace_record_cmdline_disabled)) return; @@ -560,7 +560,7 @@ notrace void tracing_record_cmdline(struct task_struct *tsk) trace_save_cmdline(tsk); } -static inline notrace struct list_head * +static inline struct list_head * trace_next_list(struct trace_array_cpu *data, struct list_head *next) { /* @@ -574,7 +574,7 @@ trace_next_list(struct trace_array_cpu *data, struct list_head *next) return next; } -static inline notrace void * +static inline void * trace_next_page(struct trace_array_cpu *data, void *addr) { struct list_head *next; @@ -588,7 +588,7 @@ trace_next_page(struct trace_array_cpu *data, void *addr) return page_address(page); } -static inline notrace struct trace_entry * +static inline struct trace_entry * tracing_get_trace_entry(struct trace_array *tr, struct trace_array_cpu *data) { unsigned long idx, idx_next; @@ -623,7 +623,7 @@ tracing_get_trace_entry(struct trace_array *tr, struct trace_array_cpu *data) return entry; } -static inline notrace void +static inline void tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) { struct task_struct *tsk = current; @@ -640,7 +640,7 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) (need_resched() ? TRACE_FLAG_NEED_RESCHED : 0); } -notrace void +void trace_function(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, unsigned long parent_ip, unsigned long flags) { @@ -659,7 +659,7 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, wake_up (&trace_wait); } -notrace void +void ftrace(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, unsigned long parent_ip, unsigned long flags) { @@ -667,7 +667,7 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } -notrace void +void trace_special(struct trace_array *tr, struct trace_array_cpu *data, unsigned long arg1, unsigned long arg2, unsigned long arg3) { @@ -687,7 +687,7 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, wake_up (&trace_wait); } -notrace void +void tracing_sched_switch_trace(struct trace_array *tr, struct trace_array_cpu *data, struct task_struct *prev, struct task_struct *next, @@ -712,7 +712,7 @@ tracing_sched_switch_trace(struct trace_array *tr, } #ifdef CONFIG_FTRACE -static notrace void +static void function_trace_call(unsigned long ip, unsigned long parent_ip) { struct trace_array *tr = &global_trace; @@ -741,12 +741,12 @@ static struct ftrace_ops trace_ops __read_mostly = .func = function_trace_call, }; -notrace void tracing_start_function_trace(void) +void tracing_start_function_trace(void) { register_ftrace_function(&trace_ops); } -notrace void tracing_stop_function_trace(void) +void tracing_stop_function_trace(void) { unregister_ftrace_function(&trace_ops); } @@ -786,7 +786,7 @@ trace_entry_idx(struct trace_array *tr, struct trace_array_cpu *data, return &array[iter->next_page_idx[cpu]]; } -static struct trace_entry * notrace +static struct trace_entry * find_next_entry(struct trace_iterator *iter, int *ent_cpu) { struct trace_array *tr = iter->tr; @@ -813,7 +813,7 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) return next; } -static notrace void trace_iterator_increment(struct trace_iterator *iter) +static void trace_iterator_increment(struct trace_iterator *iter) { iter->idx++; iter->next_idx[iter->cpu]++; @@ -828,7 +828,7 @@ static notrace void trace_iterator_increment(struct trace_iterator *iter) } } -static notrace void trace_consume(struct trace_iterator *iter) +static void trace_consume(struct trace_iterator *iter) { struct trace_array_cpu *data = iter->tr->data[iter->cpu]; @@ -844,7 +844,7 @@ static notrace void trace_consume(struct trace_iterator *iter) data->trace_idx = 0; } -static notrace void *find_next_entry_inc(struct trace_iterator *iter) +static void *find_next_entry_inc(struct trace_iterator *iter) { struct trace_entry *next; int next_cpu = -1; @@ -863,7 +863,7 @@ static notrace void *find_next_entry_inc(struct trace_iterator *iter) return next ? iter : NULL; } -static notrace void *s_next(struct seq_file *m, void *v, loff_t *pos) +static void *s_next(struct seq_file *m, void *v, loff_t *pos) { struct trace_iterator *iter = m->private; void *last_ent = iter->ent; @@ -978,7 +978,7 @@ seq_print_sym_offset(struct trace_seq *s, const char *fmt, # define IP_FMT "%016lx" #endif -static notrace int +static int seq_print_ip_sym(struct trace_seq *s, unsigned long ip, unsigned long sym_flags) { int ret; @@ -999,7 +999,7 @@ seq_print_ip_sym(struct trace_seq *s, unsigned long ip, unsigned long sym_flags) return ret; } -static notrace void print_lat_help_header(struct seq_file *m) +static void print_lat_help_header(struct seq_file *m) { seq_puts(m, "# _------=> CPU# \n"); seq_puts(m, "# / _-----=> irqs-off \n"); @@ -1012,14 +1012,14 @@ static notrace void print_lat_help_header(struct seq_file *m) seq_puts(m, "# \\ / ||||| \\ | / \n"); } -static notrace void print_func_help_header(struct seq_file *m) +static void print_func_help_header(struct seq_file *m) { seq_puts(m, "# TASK-PID CPU# TIMESTAMP FUNCTION\n"); seq_puts(m, "# | | | | |\n"); } -static notrace void +static void print_trace_header(struct seq_file *m, struct trace_iterator *iter) { unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); @@ -1090,7 +1090,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) seq_puts(m, "\n"); } -static notrace void +static void lat_print_generic(struct trace_seq *s, struct trace_entry *entry, int cpu) { int hardirq, softirq; @@ -1127,7 +1127,7 @@ lat_print_generic(struct trace_seq *s, struct trace_entry *entry, int cpu) unsigned long preempt_mark_thresh = 100; -static notrace void +static void lat_print_timestamp(struct trace_seq *s, unsigned long long abs_usecs, unsigned long rel_usecs) { @@ -1142,7 +1142,7 @@ lat_print_timestamp(struct trace_seq *s, unsigned long long abs_usecs, static const char state_to_char[] = TASK_STATE_TO_CHAR_STR; -static notrace int +static int print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) { struct trace_seq *s = &iter->seq; @@ -1206,7 +1206,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) return 1; } -static notrace int print_trace_fmt(struct trace_iterator *iter) +static int print_trace_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; unsigned long sym_flags = (trace_flags & TRACE_ITER_SYM_MASK); @@ -1279,7 +1279,7 @@ static notrace int print_trace_fmt(struct trace_iterator *iter) return 1; } -static notrace int print_raw_fmt(struct trace_iterator *iter) +static int print_raw_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; struct trace_entry *entry; @@ -1336,7 +1336,7 @@ do { \ return 0; \ } while (0) -static notrace int print_hex_fmt(struct trace_iterator *iter) +static int print_hex_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; unsigned char newline = '\n'; @@ -1375,7 +1375,7 @@ static notrace int print_hex_fmt(struct trace_iterator *iter) return 1; } -static notrace int print_bin_fmt(struct trace_iterator *iter) +static int print_bin_fmt(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; struct trace_entry *entry; @@ -1475,7 +1475,7 @@ static struct seq_operations tracer_seq_ops = { .show = s_show, }; -static struct trace_iterator notrace * +static struct trace_iterator * __tracing_open(struct inode *inode, struct file *file, int *ret) { struct trace_iterator *iter; @@ -1572,7 +1572,7 @@ static int tracing_lt_open(struct inode *inode, struct file *file) } -static notrace void * +static void * t_next(struct seq_file *m, void *v, loff_t *pos) { struct tracer *t = m->private; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index faf9f67246ac..2b7352bf1ce6 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -150,7 +150,7 @@ struct trace_iterator { long idx; }; -void notrace tracing_reset(struct trace_array_cpu *data); +void tracing_reset(struct trace_array_cpu *data); int tracing_open_generic(struct inode *inode, struct file *filp); struct dentry *tracing_init_dentry(void); void ftrace(struct trace_array *tr, @@ -189,10 +189,10 @@ void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu); -extern notrace cycle_t ftrace_now(int cpu); +extern cycle_t ftrace_now(int cpu); #ifdef CONFIG_SCHED_TRACER -extern void notrace +extern void wakeup_sched_switch(struct task_struct *prev, struct task_struct *next); #else static inline void diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 69a0eb00a0a5..4165d34bd28a 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -16,7 +16,7 @@ #include "trace.h" -static notrace void function_reset(struct trace_array *tr) +static void function_reset(struct trace_array *tr) { int cpu; @@ -26,30 +26,30 @@ static notrace void function_reset(struct trace_array *tr) tracing_reset(tr->data[cpu]); } -static notrace void start_function_trace(struct trace_array *tr) +static void start_function_trace(struct trace_array *tr) { function_reset(tr); tracing_start_function_trace(); } -static notrace void stop_function_trace(struct trace_array *tr) +static void stop_function_trace(struct trace_array *tr) { tracing_stop_function_trace(); } -static notrace void function_trace_init(struct trace_array *tr) +static void function_trace_init(struct trace_array *tr) { if (tr->ctrl) start_function_trace(tr); } -static notrace void function_trace_reset(struct trace_array *tr) +static void function_trace_reset(struct trace_array *tr) { if (tr->ctrl) stop_function_trace(tr); } -static notrace void function_trace_ctrl_update(struct trace_array *tr) +static void function_trace_ctrl_update(struct trace_array *tr) { if (tr->ctrl) start_function_trace(tr); diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 2ac0d09db6fb..7a4dc014b8ab 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -33,7 +33,7 @@ enum { static int trace_type __read_mostly; #ifdef CONFIG_PREEMPT_TRACER -static inline int notrace +static inline int preempt_trace(void) { return ((trace_type & TRACER_PREEMPT_OFF) && preempt_count()); @@ -43,7 +43,7 @@ preempt_trace(void) #endif #ifdef CONFIG_IRQSOFF_TRACER -static inline int notrace +static inline int irq_trace(void) { return ((trace_type & TRACER_IRQS_OFF) && @@ -67,7 +67,7 @@ static __cacheline_aligned_in_smp unsigned long max_sequence; /* * irqsoff uses its own tracer function to keep the overhead down: */ -static void notrace +static void irqsoff_tracer_call(unsigned long ip, unsigned long parent_ip) { struct trace_array *tr = irqsoff_trace; @@ -109,7 +109,7 @@ static struct ftrace_ops trace_ops __read_mostly = /* * Should this new latency be reported/recorded? */ -static int notrace report_latency(cycle_t delta) +static int report_latency(cycle_t delta) { if (tracing_thresh) { if (delta < tracing_thresh) @@ -121,7 +121,7 @@ static int notrace report_latency(cycle_t delta) return 1; } -static void notrace +static void check_critical_timing(struct trace_array *tr, struct trace_array_cpu *data, unsigned long parent_ip, @@ -191,7 +191,7 @@ out: trace_function(tr, data, CALLER_ADDR0, parent_ip, flags); } -static inline void notrace +static inline void start_critical_timing(unsigned long ip, unsigned long parent_ip) { int cpu; @@ -228,7 +228,7 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) atomic_dec(&data->disabled); } -static inline void notrace +static inline void stop_critical_timing(unsigned long ip, unsigned long parent_ip) { int cpu; @@ -261,13 +261,13 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) } /* start and stop critical timings used to for stoppage (in idle) */ -void notrace start_critical_timings(void) +void start_critical_timings(void) { if (preempt_trace() || irq_trace()) start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } -void notrace stop_critical_timings(void) +void stop_critical_timings(void) { if (preempt_trace() || irq_trace()) stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); @@ -275,13 +275,13 @@ void notrace stop_critical_timings(void) #ifdef CONFIG_IRQSOFF_TRACER #ifdef CONFIG_PROVE_LOCKING -void notrace time_hardirqs_on(unsigned long a0, unsigned long a1) +void time_hardirqs_on(unsigned long a0, unsigned long a1) { if (!preempt_trace() && irq_trace()) stop_critical_timing(a0, a1); } -void notrace time_hardirqs_off(unsigned long a0, unsigned long a1) +void time_hardirqs_off(unsigned long a0, unsigned long a1) { if (!preempt_trace() && irq_trace()) start_critical_timing(a0, a1); @@ -309,35 +309,35 @@ void trace_softirqs_off(unsigned long ip) { } -inline notrace void print_irqtrace_events(struct task_struct *curr) +inline void print_irqtrace_events(struct task_struct *curr) { } /* * We are only interested in hardirq on/off events: */ -void notrace trace_hardirqs_on(void) +void trace_hardirqs_on(void) { if (!preempt_trace() && irq_trace()) stop_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } EXPORT_SYMBOL(trace_hardirqs_on); -void notrace trace_hardirqs_off(void) +void trace_hardirqs_off(void) { if (!preempt_trace() && irq_trace()) start_critical_timing(CALLER_ADDR0, CALLER_ADDR1); } EXPORT_SYMBOL(trace_hardirqs_off); -void notrace trace_hardirqs_on_caller(unsigned long caller_addr) +void trace_hardirqs_on_caller(unsigned long caller_addr) { if (!preempt_trace() && irq_trace()) stop_critical_timing(CALLER_ADDR0, caller_addr); } EXPORT_SYMBOL(trace_hardirqs_on_caller); -void notrace trace_hardirqs_off_caller(unsigned long caller_addr) +void trace_hardirqs_off_caller(unsigned long caller_addr) { if (!preempt_trace() && irq_trace()) start_critical_timing(CALLER_ADDR0, caller_addr); @@ -348,12 +348,12 @@ EXPORT_SYMBOL(trace_hardirqs_off_caller); #endif /* CONFIG_IRQSOFF_TRACER */ #ifdef CONFIG_PREEMPT_TRACER -void notrace trace_preempt_on(unsigned long a0, unsigned long a1) +void trace_preempt_on(unsigned long a0, unsigned long a1) { stop_critical_timing(a0, a1); } -void notrace trace_preempt_off(unsigned long a0, unsigned long a1) +void trace_preempt_off(unsigned long a0, unsigned long a1) { start_critical_timing(a0, a1); } @@ -395,14 +395,14 @@ static void irqsoff_tracer_ctrl_update(struct trace_array *tr) stop_irqsoff_tracer(tr); } -static void notrace irqsoff_tracer_open(struct trace_iterator *iter) +static void irqsoff_tracer_open(struct trace_iterator *iter) { /* stop the trace while dumping */ if (iter->tr->ctrl) stop_irqsoff_tracer(iter->tr); } -static void notrace irqsoff_tracer_close(struct trace_iterator *iter) +static void irqsoff_tracer_close(struct trace_iterator *iter) { if (iter->tr->ctrl) start_irqsoff_tracer(iter->tr); diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 8d656672da93..b738eaca1dbe 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -17,7 +17,7 @@ static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; -static void notrace +static void ctx_switch_func(struct task_struct *prev, struct task_struct *next) { struct trace_array *tr = ctx_trace; @@ -57,7 +57,7 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) wakeup_sched_switch(prev, next); } -static notrace void sched_switch_reset(struct trace_array *tr) +static void sched_switch_reset(struct trace_array *tr) { int cpu; @@ -67,18 +67,18 @@ static notrace void sched_switch_reset(struct trace_array *tr) tracing_reset(tr->data[cpu]); } -static notrace void start_sched_trace(struct trace_array *tr) +static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); tracer_enabled = 1; } -static notrace void stop_sched_trace(struct trace_array *tr) +static void stop_sched_trace(struct trace_array *tr) { tracer_enabled = 0; } -static notrace void sched_switch_trace_init(struct trace_array *tr) +static void sched_switch_trace_init(struct trace_array *tr) { ctx_trace = tr; @@ -86,7 +86,7 @@ static notrace void sched_switch_trace_init(struct trace_array *tr) start_sched_trace(tr); } -static notrace void sched_switch_trace_reset(struct trace_array *tr) +static void sched_switch_trace_reset(struct trace_array *tr) { if (tr->ctrl) stop_sched_trace(tr); diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 3549e4154f1f..662679c78b66 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -27,12 +27,12 @@ static unsigned wakeup_prio = -1; static DEFINE_SPINLOCK(wakeup_lock); -static void notrace __wakeup_reset(struct trace_array *tr); +static void __wakeup_reset(struct trace_array *tr); /* * Should this new latency be reported/recorded? */ -static int notrace report_latency(cycle_t delta) +static int report_latency(cycle_t delta) { if (tracing_thresh) { if (delta < tracing_thresh) @@ -44,7 +44,7 @@ static int notrace report_latency(cycle_t delta) return 1; } -void notrace +void wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) { unsigned long latency = 0, t0 = 0, t1 = 0; @@ -126,7 +126,7 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -static void notrace __wakeup_reset(struct trace_array *tr) +static void __wakeup_reset(struct trace_array *tr) { struct trace_array_cpu *data; int cpu; @@ -147,7 +147,7 @@ static void notrace __wakeup_reset(struct trace_array *tr) wakeup_task = NULL; } -static void notrace wakeup_reset(struct trace_array *tr) +static void wakeup_reset(struct trace_array *tr) { unsigned long flags; @@ -156,7 +156,7 @@ static void notrace wakeup_reset(struct trace_array *tr) spin_unlock_irqrestore(&wakeup_lock, flags); } -static notrace void +static void wakeup_check_start(struct trace_array *tr, struct task_struct *p, struct task_struct *curr) { @@ -201,7 +201,7 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -notrace void +void ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) { if (likely(!tracer_enabled)) @@ -210,7 +210,7 @@ ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) wakeup_check_start(wakeup_trace, wakee, curr); } -notrace void +void ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) { if (likely(!tracer_enabled)) @@ -219,7 +219,7 @@ ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) wakeup_check_start(wakeup_trace, wakee, curr); } -static notrace void start_wakeup_tracer(struct trace_array *tr) +static void start_wakeup_tracer(struct trace_array *tr) { wakeup_reset(tr); @@ -237,12 +237,12 @@ static notrace void start_wakeup_tracer(struct trace_array *tr) return; } -static notrace void stop_wakeup_tracer(struct trace_array *tr) +static void stop_wakeup_tracer(struct trace_array *tr) { tracer_enabled = 0; } -static notrace void wakeup_tracer_init(struct trace_array *tr) +static void wakeup_tracer_init(struct trace_array *tr) { wakeup_trace = tr; @@ -250,7 +250,7 @@ static notrace void wakeup_tracer_init(struct trace_array *tr) start_wakeup_tracer(tr); } -static notrace void wakeup_tracer_reset(struct trace_array *tr) +static void wakeup_tracer_reset(struct trace_array *tr) { if (tr->ctrl) { stop_wakeup_tracer(tr); @@ -267,14 +267,14 @@ static void wakeup_tracer_ctrl_update(struct trace_array *tr) stop_wakeup_tracer(tr); } -static void notrace wakeup_tracer_open(struct trace_iterator *iter) +static void wakeup_tracer_open(struct trace_iterator *iter) { /* stop the trace while dumping */ if (iter->tr->ctrl) stop_wakeup_tracer(iter->tr); } -static void notrace wakeup_tracer_close(struct trace_iterator *iter) +static void wakeup_tracer_close(struct trace_iterator *iter) { /* forget about any processes we were recording */ if (iter->tr->ctrl) diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 546307de6e3d..85715b86a342 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -3,7 +3,7 @@ #include #include -static notrace inline int trace_valid_entry(struct trace_entry *entry) +static inline int trace_valid_entry(struct trace_entry *entry) { switch (entry->type) { case TRACE_FN: -- cgit v1.2.3 From 57422797dc009fc83766bcf230d29dbe6e08e21e Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: add wakeup events to sched tracer Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 42 +++++++++++++++++++++++++++++++++++---- kernel/trace/trace.h | 12 +++++++++++ kernel/trace/trace_sched_switch.c | 36 +++++++++++++++++++++++++++++++++ kernel/trace/trace_sched_wakeup.c | 2 +- kernel/trace/trace_selftest.c | 1 + 5 files changed, 88 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f5898051fdd9..192c1354a7e0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -57,7 +57,7 @@ static struct trace_array max_tr; static DEFINE_PER_CPU(struct trace_array_cpu, max_data); static int tracer_enabled = 1; -static unsigned long trace_nr_entries = 16384UL; +static unsigned long trace_nr_entries = 65536UL; static struct tracer *trace_types __read_mostly; static struct tracer *current_trace __read_mostly; @@ -87,6 +87,7 @@ enum trace_type { TRACE_FN, TRACE_CTX, + TRACE_WAKE, TRACE_SPECIAL, __TRACE_LAST_TYPE @@ -711,6 +712,30 @@ tracing_sched_switch_trace(struct trace_array *tr, wake_up (&trace_wait); } +void +tracing_sched_wakeup_trace(struct trace_array *tr, + struct trace_array_cpu *data, + struct task_struct *wakee, struct task_struct *curr, + unsigned long flags) +{ + struct trace_entry *entry; + unsigned long irq_flags; + + spin_lock_irqsave(&data->lock, irq_flags); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, flags); + entry->type = TRACE_WAKE; + entry->ctx.prev_pid = curr->pid; + entry->ctx.prev_prio = curr->prio; + entry->ctx.prev_state = curr->state; + entry->ctx.next_pid = wakee->pid; + entry->ctx.next_prio = wakee->prio; + spin_unlock_irqrestore(&data->lock, irq_flags); + + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up(&trace_wait); +} + #ifdef CONFIG_FTRACE static void function_trace_call(unsigned long ip, unsigned long parent_ip) @@ -1183,13 +1208,14 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) trace_seq_puts(s, ")\n"); break; case TRACE_CTX: + case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; comm = trace_find_cmdline(entry->ctx.next_pid); - trace_seq_printf(s, " %d:%d:%c --> %d:%d %s\n", + trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d %s\n", entry->ctx.prev_pid, entry->ctx.prev_prio, - S, + S, entry->type == TRACE_CTX ? "==>" : " +", entry->ctx.next_pid, entry->ctx.next_prio, comm); @@ -1256,12 +1282,14 @@ static int print_trace_fmt(struct trace_iterator *iter) return 0; break; case TRACE_CTX: + case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; - ret = trace_seq_printf(s, " %d:%d:%c ==> %d:%d\n", + ret = trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d\n", entry->ctx.prev_pid, entry->ctx.prev_prio, S, + entry->type == TRACE_CTX ? "==>" : " +", entry->ctx.next_pid, entry->ctx.next_prio); if (!ret) @@ -1301,8 +1329,11 @@ static int print_raw_fmt(struct trace_iterator *iter) return 0; break; case TRACE_CTX: + case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; + if (entry->type == TRACE_WAKE) + S = '+'; ret = trace_seq_printf(s, "%d %d %c %d %d\n", entry->ctx.prev_pid, entry->ctx.prev_prio, @@ -1355,8 +1386,11 @@ static int print_hex_fmt(struct trace_iterator *iter) SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip); break; case TRACE_CTX: + case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; + if (entry->type == TRACE_WAKE) + S = '+'; SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_pid); SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_prio); SEQ_PUT_HEX_FIELD_RET(s, S); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 2b7352bf1ce6..90e0ba0f6eba 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -164,6 +164,12 @@ void tracing_sched_switch_trace(struct trace_array *tr, struct task_struct *next, unsigned long flags); void tracing_record_cmdline(struct task_struct *tsk); + +void tracing_sched_wakeup_trace(struct trace_array *tr, + struct trace_array_cpu *data, + struct task_struct *wakee, + struct task_struct *cur, + unsigned long flags); void trace_special(struct trace_array *tr, struct trace_array_cpu *data, unsigned long arg1, @@ -194,11 +200,17 @@ extern cycle_t ftrace_now(int cpu); #ifdef CONFIG_SCHED_TRACER extern void wakeup_sched_switch(struct task_struct *prev, struct task_struct *next); +extern void +wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr); #else static inline void wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) { } +static inline void +wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) +{ +} #endif #ifdef CONFIG_CONTEXT_SWITCH_TRACER diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index b738eaca1dbe..8b1cf1a3aee0 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -41,6 +41,29 @@ ctx_switch_func(struct task_struct *prev, struct task_struct *next) local_irq_restore(flags); } +static void wakeup_func(struct task_struct *wakee, struct task_struct *curr) +{ + struct trace_array *tr = ctx_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (!tracer_enabled) + return; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + tracing_sched_wakeup_trace(tr, data, wakee, curr, flags); + + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) { tracing_record_cmdline(prev); @@ -57,6 +80,19 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) wakeup_sched_switch(prev, next); } +void +ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +{ + tracing_record_cmdline(curr); + + wakeup_func(wakee, curr); + + /* + * Chain to the wakeup tracer (this is a NOP if disabled): + */ + wakeup_sched_wakeup(wakee, curr); +} + static void sched_switch_reset(struct trace_array *tr) { int cpu; diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 662679c78b66..87fa7b253b57 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -202,7 +202,7 @@ out: } void -ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) { if (likely(!tracer_enabled)) return; diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 85715b86a342..39dd452647da 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -8,6 +8,7 @@ static inline int trace_valid_entry(struct trace_entry *entry) switch (entry->type) { case TRACE_FN: case TRACE_CTX: + case TRACE_WAKE: return 1; } return 0; -- cgit v1.2.3 From 86387f7ee5d3273ff4859e2c64ce656639b6ca65 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: add stack tracing Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 1 + kernel/trace/trace.c | 103 ++++++++++++++++++++++++++++++++++++++++++--------- kernel/trace/trace.h | 11 ++++++ 3 files changed, 97 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 3f73a1710242..eb1988ed84b7 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -10,6 +10,7 @@ config TRACER_MAX_TRACE config TRACING bool select DEBUG_FS + select STACKTRACE config FTRACE bool "Kernel Function Tracer" diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 192c1354a7e0..b4b1b4fe99fd 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -28,6 +28,8 @@ #include #include +#include + #include "trace.h" unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; @@ -88,6 +90,7 @@ enum trace_type { TRACE_FN, TRACE_CTX, TRACE_WAKE, + TRACE_STACK, TRACE_SPECIAL, __TRACE_LAST_TYPE @@ -109,6 +112,7 @@ enum trace_iterator_flags { TRACE_ITER_HEX = 0x20, TRACE_ITER_BIN = 0x40, TRACE_ITER_BLOCK = 0x80, + TRACE_ITER_STACKTRACE = 0x100, }; #define TRACE_ITER_SYM_MASK \ @@ -124,10 +128,11 @@ static const char *trace_options[] = { "hex", "bin", "block", + "stacktrace", NULL }; -static unsigned trace_flags; +static unsigned trace_flags = TRACE_ITER_PRINT_PARENT | TRACE_ITER_STACKTRACE; static DEFINE_SPINLOCK(ftrace_max_lock); @@ -657,7 +662,7 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, spin_unlock_irqrestore(&data->lock, irq_flags); if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up (&trace_wait); + wake_up(&trace_wait); } void @@ -685,13 +690,39 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, spin_unlock_irqrestore(&data->lock, irq_flags); if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up (&trace_wait); + wake_up(&trace_wait); +} + +void __trace_stack(struct trace_array *tr, + struct trace_array_cpu *data, + unsigned long flags, + int skip) +{ + struct trace_entry *entry; + struct stack_trace trace; + + if (!(trace_flags & TRACE_ITER_STACKTRACE)) + return; + + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, flags); + entry->type = TRACE_STACK; + + memset(&entry->stack, 0, sizeof(entry->stack)); + + trace.nr_entries = 0; + trace.max_entries = FTRACE_STACK_ENTRIES; + trace.skip = skip; + trace.entries = entry->stack.caller; + + save_stack_trace(&trace); } void tracing_sched_switch_trace(struct trace_array *tr, struct trace_array_cpu *data, - struct task_struct *prev, struct task_struct *next, + struct task_struct *prev, + struct task_struct *next, unsigned long flags) { struct trace_entry *entry; @@ -706,16 +737,18 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.prev_state = prev->state; entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; + __trace_stack(tr, data, flags, 4); spin_unlock_irqrestore(&data->lock, irq_flags); if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up (&trace_wait); + wake_up(&trace_wait); } void tracing_sched_wakeup_trace(struct trace_array *tr, struct trace_array_cpu *data, - struct task_struct *wakee, struct task_struct *curr, + struct task_struct *wakee, + struct task_struct *curr, unsigned long flags) { struct trace_entry *entry; @@ -730,6 +763,7 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.prev_state = curr->state; entry->ctx.next_pid = wakee->pid; entry->ctx.next_prio = wakee->prio; + __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); if (!(trace_flags & TRACE_ITER_BLOCK)) @@ -1179,6 +1213,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) unsigned long rel_usecs; char *comm; int S; + int i; if (!next_entry) next_entry = entry; @@ -1197,8 +1232,10 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) abs_usecs % 1000, rel_usecs/1000, rel_usecs % 1000); } else { - lat_print_generic(s, entry, cpu); - lat_print_timestamp(s, abs_usecs, rel_usecs); + if (entry->type != TRACE_STACK) { + lat_print_generic(s, entry, cpu); + lat_print_timestamp(s, abs_usecs, rel_usecs); + } } switch (entry->type) { case TRACE_FN: @@ -1226,6 +1263,14 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) entry->special.arg2, entry->special.arg3); break; + case TRACE_STACK: + for (i = 0; i < FTRACE_STACK_ENTRIES; i++) { + if (i) + trace_seq_puts(s, " <= "); + seq_print_ip_sym(s, entry->stack.caller[i], sym_flags); + } + trace_seq_puts(s, "\n"); + break; default: trace_seq_printf(s, "Unknown type %d\n", entry->type); } @@ -1241,8 +1286,9 @@ static int print_trace_fmt(struct trace_iterator *iter) unsigned long long t; unsigned long secs; char *comm; - int S; int ret; + int S; + int i; entry = iter->ent; @@ -1252,15 +1298,17 @@ static int print_trace_fmt(struct trace_iterator *iter) usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; - ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); - if (!ret) - return 0; - ret = trace_seq_printf(s, "[%02d] ", iter->cpu); - if (!ret) - return 0; - ret = trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); - if (!ret) - return 0; + if (entry->type != TRACE_STACK) { + ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); + if (!ret) + return 0; + ret = trace_seq_printf(s, "[%02d] ", iter->cpu); + if (!ret) + return 0; + ret = trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); + if (!ret) + return 0; + } switch (entry->type) { case TRACE_FN: @@ -1303,6 +1351,22 @@ static int print_trace_fmt(struct trace_iterator *iter) if (!ret) return 0; break; + case TRACE_STACK: + for (i = 0; i < FTRACE_STACK_ENTRIES; i++) { + if (i) { + ret = trace_seq_puts(s, " <= "); + if (!ret) + return 0; + } + ret = seq_print_ip_sym(s, entry->stack.caller[i], + sym_flags); + if (!ret) + return 0; + } + ret = trace_seq_puts(s, "\n"); + if (!ret) + return 0; + break; } return 1; } @@ -1344,6 +1408,7 @@ static int print_raw_fmt(struct trace_iterator *iter) return 0; break; case TRACE_SPECIAL: + case TRACE_STACK: ret = trace_seq_printf(s, " %lx %lx %lx\n", entry->special.arg1, entry->special.arg2, @@ -1399,6 +1464,7 @@ static int print_hex_fmt(struct trace_iterator *iter) SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip); break; case TRACE_SPECIAL: + case TRACE_STACK: SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg1); SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg2); SEQ_PUT_HEX_FIELD_RET(s, entry->special.arg3); @@ -1433,6 +1499,7 @@ static int print_bin_fmt(struct trace_iterator *iter) SEQ_PUT_FIELD_RET(s, entry->ctx.next_prio); break; case TRACE_SPECIAL: + case TRACE_STACK: SEQ_PUT_FIELD_RET(s, entry->special.arg1); SEQ_PUT_FIELD_RET(s, entry->special.arg2); SEQ_PUT_FIELD_RET(s, entry->special.arg3); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 90e0ba0f6eba..387bdcf45e28 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -34,6 +34,16 @@ struct special_entry { unsigned long arg3; }; +/* + * Stack-trace entry: + */ + +#define FTRACE_STACK_ENTRIES 5 + +struct stack_entry { + unsigned long caller[FTRACE_STACK_ENTRIES]; +}; + /* * The trace entry - the most basic unit of tracing. This is what * is printed in the end as a single line in the trace output, such as: @@ -51,6 +61,7 @@ struct trace_entry { struct ftrace_entry fn; struct ctx_switch_entry ctx; struct special_entry special; + struct stack_entry stack; }; }; -- cgit v1.2.3 From 8ac0fca4ccb355ce50471d7aa3f10f5900b28b95 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: sched tracer fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 2 +- kernel/trace/trace_sched_wakeup.c | 13 +++---------- 2 files changed, 4 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 328494e28df2..53ab1174664f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2613,7 +2613,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_new_task(p, rq->curr); + ftrace_wake_up_task(p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 87fa7b253b57..2a012423f9d0 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -201,20 +201,13 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) +void wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) { if (likely(!tracer_enabled)) return; - wakeup_check_start(wakeup_trace, wakee, curr); -} - -void -ftrace_wake_up_new_task(struct task_struct *wakee, struct task_struct *curr) -{ - if (likely(!tracer_enabled)) - return; + tracing_record_cmdline(curr); + tracing_record_cmdline(wakee); wakeup_check_start(wakeup_trace, wakee, curr); } -- cgit v1.2.3 From 4c1f4d4f0175129934d5dbc19a39296430937a05 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:51 +0200 Subject: ftrace: make nostacktrace the default Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index b4b1b4fe99fd..0e4b7119e263 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -132,7 +132,7 @@ static const char *trace_options[] = { NULL }; -static unsigned trace_flags = TRACE_ITER_PRINT_PARENT | TRACE_ITER_STACKTRACE; +static unsigned trace_flags = TRACE_ITER_PRINT_PARENT; static DEFINE_SPINLOCK(ftrace_max_lock); -- cgit v1.2.3 From 4e65551905fb0300ae7e667cbaa41ee2e3f29a13 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: sched tracer, trace full rbtree Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 35 ++++++++++++++++++++++--- kernel/trace/trace.c | 55 ++++++++++++++++----------------------- kernel/trace/trace.h | 14 ++++++++++ kernel/trace/trace_sched_switch.c | 24 +++++++++++------ 4 files changed, 85 insertions(+), 43 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 53ab1174664f..b9208a0e33a0 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2394,6 +2394,35 @@ static int sched_balance_self(int cpu, int flag) #endif /* CONFIG_SMP */ +#ifdef CONFIG_CONTEXT_SWITCH_TRACER + +void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) +{ + struct sched_entity *se; + struct task_struct *p; + struct rb_node *curr; + struct rq *rq = __rq; + + curr = first_fair(&rq->cfs); + if (!curr) + return; + + while (curr) { + se = rb_entry(curr, struct sched_entity, run_node); + if (!entity_is_task(se)) + continue; + + p = task_of(se); + + __trace_special(__tr, __data, + p->pid, p->se.vruntime, p->se.sum_exec_runtime); + + curr = rb_next(curr); + } +} + +#endif + /*** * try_to_wake_up - wake up a thread * @p: the to-be-woken-up thread @@ -2468,7 +2497,7 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) out_activate: #endif /* CONFIG_SMP */ - ftrace_wake_up_task(p, rq->curr); + ftrace_wake_up_task(rq, p, rq->curr); schedstat_inc(p, se.nr_wakeups); if (sync) schedstat_inc(p, se.nr_wakeups_sync); @@ -2613,7 +2642,7 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_task(p, rq->curr); + ftrace_wake_up_task(rq, p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) @@ -2786,7 +2815,7 @@ context_switch(struct rq *rq, struct task_struct *prev, struct mm_struct *mm, *oldmm; prepare_task_switch(rq, prev, next); - ftrace_ctx_switch(prev, next); + ftrace_ctx_switch(rq, prev, next); mm = next->mm; oldmm = prev->active_mm; /* diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0e4b7119e263..65173b14b914 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -66,7 +66,18 @@ static struct tracer *current_trace __read_mostly; static int max_tracer_type_len; static DEFINE_MUTEX(trace_types_lock); -static DECLARE_WAIT_QUEUE_HEAD (trace_wait); +static DECLARE_WAIT_QUEUE_HEAD(trace_wait); + +unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; + +/* + * FIXME: where should this be called? + */ +void trace_wake_up(void) +{ + if (!(trace_flags & TRACE_ITER_BLOCK)) + wake_up(&trace_wait); +} #define ENTRIES_PER_PAGE (PAGE_SIZE / sizeof(struct trace_entry)) @@ -103,18 +114,6 @@ enum trace_flag_type { TRACE_FLAG_SOFTIRQ = 0x08, }; -enum trace_iterator_flags { - TRACE_ITER_PRINT_PARENT = 0x01, - TRACE_ITER_SYM_OFFSET = 0x02, - TRACE_ITER_SYM_ADDR = 0x04, - TRACE_ITER_VERBOSE = 0x08, - TRACE_ITER_RAW = 0x10, - TRACE_ITER_HEX = 0x20, - TRACE_ITER_BIN = 0x40, - TRACE_ITER_BLOCK = 0x80, - TRACE_ITER_STACKTRACE = 0x100, -}; - #define TRACE_ITER_SYM_MASK \ (TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR) @@ -132,8 +131,6 @@ static const char *trace_options[] = { NULL }; -static unsigned trace_flags = TRACE_ITER_PRINT_PARENT; - static DEFINE_SPINLOCK(ftrace_max_lock); /* @@ -660,9 +657,6 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } void @@ -673,10 +667,14 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } +#ifdef CONFIG_CONTEXT_SWITCH_TRACER + void -trace_special(struct trace_array *tr, struct trace_array_cpu *data, - unsigned long arg1, unsigned long arg2, unsigned long arg3) +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) { + struct trace_array_cpu *data = __data; + struct trace_array *tr = __tr; struct trace_entry *entry; unsigned long irq_flags; @@ -688,11 +686,10 @@ trace_special(struct trace_array *tr, struct trace_array_cpu *data, entry->special.arg2 = arg2; entry->special.arg3 = arg3; spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } +#endif + void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, @@ -739,9 +736,6 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.next_prio = next->prio; __trace_stack(tr, data, flags, 4); spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } void @@ -765,9 +759,6 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_prio = wakee->prio; __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); - - if (!(trace_flags & TRACE_ITER_BLOCK)) - wake_up(&trace_wait); } #ifdef CONFIG_FTRACE @@ -1258,7 +1249,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) comm); break; case TRACE_SPECIAL: - trace_seq_printf(s, " %lx %lx %lx\n", + trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1344,7 +1335,7 @@ static int print_trace_fmt(struct trace_iterator *iter) return 0; break; case TRACE_SPECIAL: - ret = trace_seq_printf(s, " %lx %lx %lx\n", + ret = trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1409,7 +1400,7 @@ static int print_raw_fmt(struct trace_iterator *iter) break; case TRACE_SPECIAL: case TRACE_STACK: - ret = trace_seq_printf(s, " %lx %lx %lx\n", + ret = trace_seq_printf(s, " %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 387bdcf45e28..75e237475674 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -274,4 +274,18 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace, extern void *head_page(struct trace_array_cpu *data); +extern unsigned long trace_flags; + +enum trace_iterator_flags { + TRACE_ITER_PRINT_PARENT = 0x01, + TRACE_ITER_SYM_OFFSET = 0x02, + TRACE_ITER_SYM_ADDR = 0x04, + TRACE_ITER_VERBOSE = 0x08, + TRACE_ITER_RAW = 0x10, + TRACE_ITER_HEX = 0x20, + TRACE_ITER_BIN = 0x40, + TRACE_ITER_BLOCK = 0x80, + TRACE_ITER_STACKTRACE = 0x100, +}; + #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 8b1cf1a3aee0..12658b3f2b28 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -18,7 +18,7 @@ static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; static void -ctx_switch_func(struct task_struct *prev, struct task_struct *next) +ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) { struct trace_array *tr = ctx_trace; struct trace_array_cpu *data; @@ -34,14 +34,17 @@ ctx_switch_func(struct task_struct *prev, struct task_struct *next) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) + if (likely(disabled == 1)) { tracing_sched_switch_trace(tr, data, prev, next, flags); + ftrace_all_fair_tasks(__rq, tr, data); + } atomic_dec(&data->disabled); local_irq_restore(flags); } -static void wakeup_func(struct task_struct *wakee, struct task_struct *curr) +static void +wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) { struct trace_array *tr = ctx_trace; struct trace_array_cpu *data; @@ -57,14 +60,18 @@ static void wakeup_func(struct task_struct *wakee, struct task_struct *curr) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) + if (likely(disabled == 1)) { tracing_sched_wakeup_trace(tr, data, wakee, curr, flags); + ftrace_all_fair_tasks(__rq, tr, data); + } atomic_dec(&data->disabled); local_irq_restore(flags); } -void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) +void +ftrace_ctx_switch(void *__rq, struct task_struct *prev, + struct task_struct *next) { tracing_record_cmdline(prev); @@ -72,7 +79,7 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) * If tracer_switch_func only points to the local * switch func, it still needs the ptr passed to it. */ - ctx_switch_func(prev, next); + ctx_switch_func(__rq, prev, next); /* * Chain to the wakeup tracer (this is a NOP if disabled): @@ -81,11 +88,12 @@ void ftrace_ctx_switch(struct task_struct *prev, struct task_struct *next) } void -ftrace_wake_up_task(struct task_struct *wakee, struct task_struct *curr) +ftrace_wake_up_task(void *__rq, struct task_struct *wakee, + struct task_struct *curr) { tracing_record_cmdline(curr); - wakeup_func(wakee, curr); + wakeup_func(__rq, wakee, curr); /* * Chain to the wakeup tracer (this is a NOP if disabled): -- cgit v1.2.3 From 24cd5d111e8c713e62cda7ca1d01232402e3d3c9 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: trace curr/next tasks Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 13 ++++++++++++- 1 file changed, 12 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index b9208a0e33a0..673b588b713b 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2398,8 +2398,8 @@ static int sched_balance_self(int cpu, int flag) void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) { - struct sched_entity *se; struct task_struct *p; + struct sched_entity *se; struct rb_node *curr; struct rq *rq = __rq; @@ -2407,6 +2407,17 @@ void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) if (!curr) return; + if (rq->cfs.curr) { + p = task_of(rq->cfs.curr); + __trace_special(__tr, __data, + p->pid, p->se.vruntime, p->se.sum_exec_runtime); + } + if (rq->cfs.next) { + p = task_of(rq->cfs.next); + __trace_special(__tr, __data, + p->pid, p->se.vruntime, p->se.sum_exec_runtime); + } + while (curr) { se = rb_entry(curr, struct sched_entity, run_node); if (!entity_is_task(se)) -- cgit v1.2.3 From 017730c11241e26577673eb9d957cfc66172ea91 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: fix wakeups Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 18 ++++++++++++++++++ kernel/trace/trace.c | 15 +++++++++++---- 2 files changed, 29 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 673b588b713b..9ca4a2e6a236 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -642,6 +642,24 @@ static inline void update_rq_clock(struct rq *rq) # define const_debug static const #endif +/** + * runqueue_is_locked + * + * Returns true if the current cpu runqueue is locked. + * This interface allows printk to be called with the runqueue lock + * held and know whether or not it is OK to wake up the klogd. + */ +int runqueue_is_locked(void) +{ + int cpu = get_cpu(); + struct rq *rq = cpu_rq(cpu); + int ret; + + ret = spin_is_locked(&rq->lock); + put_cpu(); + return ret; +} + /* * Debugging: various feature bits */ diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 65173b14b914..2ca9d66aa74e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -70,12 +70,13 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; -/* - * FIXME: where should this be called? - */ void trace_wake_up(void) { - if (!(trace_flags & TRACE_ITER_BLOCK)) + /* + * The runqueue_is_locked() can fail, but this is the best we + * have for now: + */ + if (!(trace_flags & TRACE_ITER_BLOCK) && !runqueue_is_locked()) wake_up(&trace_wait); } @@ -657,6 +658,8 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } void @@ -686,6 +689,8 @@ __trace_special(void *__tr, void *__data, entry->special.arg2 = arg2; entry->special.arg3 = arg3; spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } #endif @@ -759,6 +764,8 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_prio = wakee->prio; __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); } #ifdef CONFIG_FTRACE -- cgit v1.2.3 From 1a3c3034336320554a3342572dae98d69e054fc7 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: fix __trace_special() Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 2ca9d66aa74e..65d2c0a61ed4 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -670,8 +670,6 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } -#ifdef CONFIG_CONTEXT_SWITCH_TRACER - void __trace_special(void *__tr, void *__data, unsigned long arg1, unsigned long arg2, unsigned long arg3) @@ -693,8 +691,6 @@ __trace_special(void *__tr, void *__data, trace_wake_up(); } -#endif - void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, -- cgit v1.2.3 From 4ac3ba41d372e3a9e420b36bc43589662b188a14 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: trace scheduler rbtree Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 1 + kernel/trace/trace.h | 1 + kernel/trace/trace_sched_switch.c | 6 ++++-- 3 files changed, 6 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 65d2c0a61ed4..06380dc1ebe0 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -129,6 +129,7 @@ static const char *trace_options[] = { "bin", "block", "stacktrace", + "sched-tree", NULL }; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 75e237475674..a52015702a28 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -286,6 +286,7 @@ enum trace_iterator_flags { TRACE_ITER_BIN = 0x40, TRACE_ITER_BLOCK = 0x80, TRACE_ITER_STACKTRACE = 0x100, + TRACE_ITER_SCHED_TREE = 0x200, }; #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 12658b3f2b28..5555b906a666 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -36,7 +36,8 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) if (likely(disabled == 1)) { tracing_sched_switch_trace(tr, data, prev, next, flags); - ftrace_all_fair_tasks(__rq, tr, data); + if (trace_flags & TRACE_ITER_SCHED_TREE) + ftrace_all_fair_tasks(__rq, tr, data); } atomic_dec(&data->disabled); @@ -62,7 +63,8 @@ wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) if (likely(disabled == 1)) { tracing_sched_wakeup_trace(tr, data, wakee, curr, flags); - ftrace_all_fair_tasks(__rq, tr, data); + if (trace_flags & TRACE_ITER_SCHED_TREE) + ftrace_all_fair_tasks(__rq, tr, data); } atomic_dec(&data->disabled); -- cgit v1.2.3 From c7078de1aaf562a31b20984409c38cc1b40fa8a3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: add tracing_cpumask Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 78 ++++++++++++++++++++++++++++++++++++++++++++++------ 1 file changed, 70 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 06380dc1ebe0..18f0ab88ca85 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -70,6 +70,23 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; +/* + * Only trace on a CPU if the bitmask is set: + */ +static cpumask_t tracing_cpumask __read_mostly = CPU_MASK_ALL; + +/* + * The tracer itself will not take this lock, but still we want + * to provide a consistent cpumask to user-space: + */ +static DEFINE_MUTEX(tracing_cpumask_update_lock); + +/* + * Temporary storage for the character representation of the + * CPU bitmask: + */ +static char mask_str[NR_CPUS]; + void trace_wake_up(void) { /* @@ -1754,9 +1771,49 @@ static struct file_operations tracing_lt_fops = { }; static struct file_operations show_traces_fops = { - .open = show_traces_open, - .read = seq_read, - .release = seq_release, + .open = show_traces_open, + .read = seq_read, + .release = seq_release, +}; + +static ssize_t +tracing_cpumask_read(struct file *filp, char __user *ubuf, + size_t count, loff_t *ppos) +{ + int err; + + count = min(count, (size_t)NR_CPUS); + + mutex_lock(&tracing_cpumask_update_lock); + cpumask_scnprintf(mask_str, NR_CPUS, tracing_cpumask); + err = copy_to_user(ubuf, mask_str, count); + if (err) + count = -EFAULT; + mutex_unlock(&tracing_cpumask_update_lock); + + return count; +} + +static ssize_t +tracing_cpumask_write(struct file *filp, const char __user *ubuf, + size_t count, loff_t *ppos) +{ + int err; + + mutex_lock(&tracing_cpumask_update_lock); + err = cpumask_parse_user(ubuf, count, tracing_cpumask); + mutex_unlock(&tracing_cpumask_update_lock); + + if (err) + return err; + + return count; +} + +static struct file_operations tracing_cpumask_fops = { + .open = tracing_open_generic, + .read = tracing_cpumask_read, + .write = tracing_cpumask_write, }; static ssize_t @@ -1837,9 +1894,9 @@ tracing_iter_ctrl_write(struct file *filp, const char __user *ubuf, } static struct file_operations tracing_iter_fops = { - .open = tracing_open_generic, - .read = tracing_iter_ctrl_read, - .write = tracing_iter_ctrl_write, + .open = tracing_open_generic, + .read = tracing_iter_ctrl_read, + .write = tracing_iter_ctrl_write, }; static const char readme_msg[] = @@ -1870,8 +1927,8 @@ tracing_readme_read(struct file *filp, char __user *ubuf, } static struct file_operations tracing_readme_fops = { - .open = tracing_open_generic, - .read = tracing_readme_read, + .open = tracing_open_generic, + .read = tracing_readme_read, }; static ssize_t @@ -2334,6 +2391,11 @@ static __init void tracer_init_debugfs(void) if (!entry) pr_warning("Could not create debugfs 'iter_ctrl' entry\n"); + entry = debugfs_create_file("tracing_cpumask", 0644, d_tracer, + NULL, &tracing_cpumask_fops); + if (!entry) + pr_warning("Could not create debugfs 'tracing_cpumask' entry\n"); + entry = debugfs_create_file("latency_trace", 0444, d_tracer, &global_trace, &tracing_lt_fops); if (!entry) -- cgit v1.2.3 From 36dfe9252bd4c9b55e8454363fb7e444c92c5030 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:52 +0200 Subject: ftrace: make use of tracing_cpumask Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 95 +++++++++++++++++++++++++++++++++++----------------- 1 file changed, 64 insertions(+), 31 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 18f0ab88ca85..3ae17e254fc3 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -70,23 +70,6 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; -/* - * Only trace on a CPU if the bitmask is set: - */ -static cpumask_t tracing_cpumask __read_mostly = CPU_MASK_ALL; - -/* - * The tracer itself will not take this lock, but still we want - * to provide a consistent cpumask to user-space: - */ -static DEFINE_MUTEX(tracing_cpumask_update_lock); - -/* - * Temporary storage for the character representation of the - * CPU bitmask: - */ -static char mask_str[NR_CPUS]; - void trace_wake_up(void) { /* @@ -1776,19 +1759,46 @@ static struct file_operations show_traces_fops = { .release = seq_release, }; +/* + * Only trace on a CPU if the bitmask is set: + */ +static cpumask_t tracing_cpumask = CPU_MASK_ALL; + +/* + * When tracing/tracing_cpu_mask is modified then this holds + * the new bitmask we are about to install: + */ +static cpumask_t tracing_cpumask_new; + +/* + * The tracer itself will not take this lock, but still we want + * to provide a consistent cpumask to user-space: + */ +static DEFINE_MUTEX(tracing_cpumask_update_lock); + +/* + * Temporary storage for the character representation of the + * CPU bitmask (and one more byte for the newline): + */ +static char mask_str[NR_CPUS + 1]; + static ssize_t tracing_cpumask_read(struct file *filp, char __user *ubuf, size_t count, loff_t *ppos) { - int err; - - count = min(count, (size_t)NR_CPUS); + int len; mutex_lock(&tracing_cpumask_update_lock); - cpumask_scnprintf(mask_str, NR_CPUS, tracing_cpumask); - err = copy_to_user(ubuf, mask_str, count); - if (err) - count = -EFAULT; + + len = cpumask_scnprintf(mask_str, count, tracing_cpumask); + if (count - len < 2) { + count = -EINVAL; + goto out_err; + } + len += sprintf(mask_str + len, "\n"); + count = simple_read_from_buffer(ubuf, count, ppos, mask_str, NR_CPUS+1); + +out_err: mutex_unlock(&tracing_cpumask_update_lock); return count; @@ -1798,16 +1808,40 @@ static ssize_t tracing_cpumask_write(struct file *filp, const char __user *ubuf, size_t count, loff_t *ppos) { - int err; + int err, cpu; mutex_lock(&tracing_cpumask_update_lock); - err = cpumask_parse_user(ubuf, count, tracing_cpumask); - mutex_unlock(&tracing_cpumask_update_lock); - + err = cpumask_parse_user(ubuf, count, tracing_cpumask_new); if (err) - return err; + goto err_unlock; + + spin_lock_irq(&ftrace_max_lock); + for_each_possible_cpu(cpu) { + /* + * Increase/decrease the disabled counter if we are + * about to flip a bit in the cpumask: + */ + if (cpu_isset(cpu, tracing_cpumask) && + !cpu_isset(cpu, tracing_cpumask_new)) { + atomic_inc(&global_trace.data[cpu]->disabled); + } + if (!cpu_isset(cpu, tracing_cpumask) && + cpu_isset(cpu, tracing_cpumask_new)) { + atomic_dec(&global_trace.data[cpu]->disabled); + } + } + spin_unlock_irq(&ftrace_max_lock); + + tracing_cpumask = tracing_cpumask_new; + + mutex_unlock(&tracing_cpumask_update_lock); return count; + +err_unlock: + mutex_unlock(&tracing_cpumask_update_lock); + + return err; } static struct file_operations tracing_cpumask_fops = { @@ -1846,8 +1880,7 @@ tracing_iter_ctrl_read(struct file *filp, char __user *ubuf, r += sprintf(buf + r, "\n"); WARN_ON(r >= len + 2); - r = simple_read_from_buffer(ubuf, cnt, ppos, - buf, r); + r = simple_read_from_buffer(ubuf, cnt, ppos, buf, r); kfree(buf); -- cgit v1.2.3 From d9af56fbd8b37181bffaad060f74aa2e17382874 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: fix cmdline tracing Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sched_switch.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 5555b906a666..5a217e863586 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -29,6 +29,8 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) if (!tracer_enabled) return; + tracing_record_cmdline(prev); + local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; @@ -56,6 +58,8 @@ wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) if (!tracer_enabled) return; + tracing_record_cmdline(curr); + local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; @@ -75,8 +79,6 @@ void ftrace_ctx_switch(void *__rq, struct task_struct *prev, struct task_struct *next) { - tracing_record_cmdline(prev); - /* * If tracer_switch_func only points to the local * switch func, it still needs the ptr passed to it. @@ -93,8 +95,6 @@ void ftrace_wake_up_task(void *__rq, struct task_struct *wakee, struct task_struct *curr) { - tracing_record_cmdline(curr); - wakeup_func(__rq, wakee, curr); /* -- cgit v1.2.3 From 442e544ce52d4415a024920b84fb95c5f9aa0855 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: iter ctrl fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3ae17e254fc3..688b4cf72d99 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1920,6 +1920,11 @@ tracing_iter_ctrl_write(struct file *filp, const char __user *ubuf, break; } } + /* + * If no option could be set, return an error: + */ + if (!trace_options[i]) + return -EINVAL; filp->f_pos += cnt; -- cgit v1.2.3 From f29c73fe3404f8799ed57aaf48859e0b55fc071f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: include cpu in stacktrace Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 26 +++++++++++--------------- 1 file changed, 11 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 688b4cf72d99..3a4032492fcb 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1227,10 +1227,8 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) abs_usecs % 1000, rel_usecs/1000, rel_usecs % 1000); } else { - if (entry->type != TRACE_STACK) { - lat_print_generic(s, entry, cpu); - lat_print_timestamp(s, abs_usecs, rel_usecs); - } + lat_print_generic(s, entry, cpu); + lat_print_timestamp(s, abs_usecs, rel_usecs); } switch (entry->type) { case TRACE_FN: @@ -1293,17 +1291,15 @@ static int print_trace_fmt(struct trace_iterator *iter) usec_rem = do_div(t, 1000000ULL); secs = (unsigned long)t; - if (entry->type != TRACE_STACK) { - ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); - if (!ret) - return 0; - ret = trace_seq_printf(s, "[%02d] ", iter->cpu); - if (!ret) - return 0; - ret = trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); - if (!ret) - return 0; - } + ret = trace_seq_printf(s, "%16s-%-5d ", comm, entry->pid); + if (!ret) + return 0; + ret = trace_seq_printf(s, "[%02d] ", iter->cpu); + if (!ret) + return 0; + ret = trace_seq_printf(s, "%5lu.%06lu: ", secs, usec_rem); + if (!ret) + return 0; switch (entry->type) { case TRACE_FN: -- cgit v1.2.3 From 36fc25a9f48deacd8aa08cd2d1c186a4e412604f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: sched tree fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 35 +++++++++++++++++++++-------------- 1 file changed, 21 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 9ca4a2e6a236..3bc7c5362998 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2414,6 +2414,23 @@ static int sched_balance_self(int cpu, int flag) #ifdef CONFIG_CONTEXT_SWITCH_TRACER +void ftrace_task(struct task_struct *p, void *__tr, void *__data) +{ +#if 0 + /* + * trace timeline tree + */ + __trace_special(__tr, __data, + p->pid, p->se.vruntime, p->se.sum_exec_runtime); +#else + /* + * trace balance metrics + */ + __trace_special(__tr, __data, + p->pid, p->se.avg_overlap, 0); +#endif +} + void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) { struct task_struct *p; @@ -2421,32 +2438,22 @@ void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) struct rb_node *curr; struct rq *rq = __rq; - curr = first_fair(&rq->cfs); - if (!curr) - return; - if (rq->cfs.curr) { p = task_of(rq->cfs.curr); - __trace_special(__tr, __data, - p->pid, p->se.vruntime, p->se.sum_exec_runtime); + ftrace_task(p, __tr, __data); } if (rq->cfs.next) { p = task_of(rq->cfs.next); - __trace_special(__tr, __data, - p->pid, p->se.vruntime, p->se.sum_exec_runtime); + ftrace_task(p, __tr, __data); } - while (curr) { + for (curr = first_fair(&rq->cfs); curr; curr = rb_next(curr)) { se = rb_entry(curr, struct sched_entity, run_node); if (!entity_is_task(se)) continue; p = task_of(se); - - __trace_special(__tr, __data, - p->pid, p->se.vruntime, p->se.sum_exec_runtime); - - curr = rb_next(curr); + ftrace_task(p, __tr, __data); } } -- cgit v1.2.3 From 88a4216c3ec4281fc7e6725cc3a3ccd01fb1aa14 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: sched special Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched_fair.c | 3 +++ kernel/trace/trace.c | 6 +++--- kernel/trace/trace_sched_switch.c | 24 ++++++++++++++++++++++++ 3 files changed, 30 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index e24ecd39c4b8..dc1856f10795 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1061,6 +1061,8 @@ wake_affine(struct rq *rq, struct sched_domain *this_sd, struct rq *this_rq, if (!(this_sd->flags & SD_WAKE_AFFINE)) return 0; + ftrace_special(__LINE__, curr->se.avg_overlap, sync); + ftrace_special(__LINE__, p->se.avg_overlap, -1); /* * If the currently running task will sleep within * a reasonable amount of time then attract this newly @@ -1238,6 +1240,7 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p) if (unlikely(se == pse)) return; + ftrace_special(__LINE__, p->pid, se->last_wakeup); cfs_rq_of(pse)->next = pse; /* diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3a4032492fcb..b87a26414892 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1251,7 +1251,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) comm); break; case TRACE_SPECIAL: - trace_seq_printf(s, " %ld %ld %ld\n", + trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1335,7 +1335,7 @@ static int print_trace_fmt(struct trace_iterator *iter) return 0; break; case TRACE_SPECIAL: - ret = trace_seq_printf(s, " %ld %ld %ld\n", + ret = trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); @@ -1400,7 +1400,7 @@ static int print_raw_fmt(struct trace_iterator *iter) break; case TRACE_SPECIAL: case TRACE_STACK: - ret = trace_seq_printf(s, " %ld %ld %ld\n", + ret = trace_seq_printf(s, "# %ld %ld %ld\n", entry->special.arg1, entry->special.arg2, entry->special.arg3); diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 5a217e863586..bddf676914ed 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -103,6 +103,30 @@ ftrace_wake_up_task(void *__rq, struct task_struct *wakee, wakeup_sched_wakeup(wakee, curr); } +void +ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ + struct trace_array *tr = ctx_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (!tracer_enabled) + return; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + __trace_special(tr, data, arg1, arg2, arg3); + + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + static void sched_switch_reset(struct trace_array *tr) { int cpu; -- cgit v1.2.3 From bac524d3f3dfeffa3a9d44f2c64035b88bcaacb4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: trace next state Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 35 +++++++++++++++++++++++++---------- kernel/trace/trace.h | 1 + 2 files changed, 26 insertions(+), 10 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index b87a26414892..b63fe909f87b 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -736,6 +736,7 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.prev_state = prev->state; entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; + entry->ctx.next_state = next->state; __trace_stack(tr, data, flags, 4); spin_unlock_irqrestore(&data->lock, irq_flags); } @@ -759,6 +760,7 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.prev_state = curr->state; entry->ctx.next_pid = wakee->pid; entry->ctx.next_prio = wakee->prio; + entry->ctx.next_state = wakee->state; __trace_stack(tr, data, flags, 5); spin_unlock_irqrestore(&data->lock, irq_flags); @@ -1207,7 +1209,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) unsigned long abs_usecs; unsigned long rel_usecs; char *comm; - int S; + int S, T; int i; if (!next_entry) @@ -1241,14 +1243,17 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; + T = entry->ctx.next_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.next_state] : 'X'; + comm = trace_find_cmdline(entry->ctx.next_pid); - trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d %s\n", + trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d:%c %s\n", entry->ctx.prev_pid, entry->ctx.prev_prio, S, entry->type == TRACE_CTX ? "==>" : " +", entry->ctx.next_pid, entry->ctx.next_prio, - comm); + T, comm); break; case TRACE_SPECIAL: trace_seq_printf(s, "# %ld %ld %ld\n", @@ -1280,7 +1285,7 @@ static int print_trace_fmt(struct trace_iterator *iter) unsigned long secs; char *comm; int ret; - int S; + int S, T; int i; entry = iter->ent; @@ -1324,13 +1329,16 @@ static int print_trace_fmt(struct trace_iterator *iter) case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; - ret = trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d\n", + T = entry->ctx.next_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.next_state] : 'X'; + ret = trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d:%c\n", entry->ctx.prev_pid, entry->ctx.prev_prio, S, entry->type == TRACE_CTX ? "==>" : " +", entry->ctx.next_pid, - entry->ctx.next_prio); + entry->ctx.next_prio, + T); if (!ret) return 0; break; @@ -1367,7 +1375,7 @@ static int print_raw_fmt(struct trace_iterator *iter) struct trace_seq *s = &iter->seq; struct trace_entry *entry; int ret; - int S; + int S, T; entry = iter->ent; @@ -1387,14 +1395,17 @@ static int print_raw_fmt(struct trace_iterator *iter) case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; + T = entry->ctx.next_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.next_state] : 'X'; if (entry->type == TRACE_WAKE) S = '+'; - ret = trace_seq_printf(s, "%d %d %c %d %d\n", + ret = trace_seq_printf(s, "%d %d %c %d %d %c\n", entry->ctx.prev_pid, entry->ctx.prev_prio, S, entry->ctx.next_pid, - entry->ctx.next_prio); + entry->ctx.next_prio, + T); if (!ret) return 0; break; @@ -1428,7 +1439,7 @@ static int print_hex_fmt(struct trace_iterator *iter) struct trace_seq *s = &iter->seq; unsigned char newline = '\n'; struct trace_entry *entry; - int S; + int S, T; entry = iter->ent; @@ -1445,6 +1456,8 @@ static int print_hex_fmt(struct trace_iterator *iter) case TRACE_WAKE: S = entry->ctx.prev_state < sizeof(state_to_char) ? state_to_char[entry->ctx.prev_state] : 'X'; + T = entry->ctx.next_state < sizeof(state_to_char) ? + state_to_char[entry->ctx.next_state] : 'X'; if (entry->type == TRACE_WAKE) S = '+'; SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.prev_pid); @@ -1453,6 +1466,7 @@ static int print_hex_fmt(struct trace_iterator *iter) SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_pid); SEQ_PUT_HEX_FIELD_RET(s, entry->ctx.next_prio); SEQ_PUT_HEX_FIELD_RET(s, entry->fn.parent_ip); + SEQ_PUT_HEX_FIELD_RET(s, T); break; case TRACE_SPECIAL: case TRACE_STACK: @@ -1488,6 +1502,7 @@ static int print_bin_fmt(struct trace_iterator *iter) SEQ_PUT_FIELD_RET(s, entry->ctx.prev_state); SEQ_PUT_FIELD_RET(s, entry->ctx.next_pid); SEQ_PUT_FIELD_RET(s, entry->ctx.next_prio); + SEQ_PUT_FIELD_RET(s, entry->ctx.next_state); break; case TRACE_SPECIAL: case TRACE_STACK: diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index a52015702a28..96951a8d09a4 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -23,6 +23,7 @@ struct ctx_switch_entry { unsigned char prev_state; unsigned int next_pid; unsigned char next_prio; + unsigned char next_state; }; /* -- cgit v1.2.3 From 5429db2d26a59903c81a4f6c6dae7eb9daaea5fc Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Mon, 12 May 2008 21:20:53 +0200 Subject: ftrace: fix wakeup callback Signed-off-by: Peter Zijlstra Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 3bc7c5362998..1ec3fb2efee6 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2533,7 +2533,6 @@ static int try_to_wake_up(struct task_struct *p, unsigned int state, int sync) out_activate: #endif /* CONFIG_SMP */ - ftrace_wake_up_task(rq, p, rq->curr); schedstat_inc(p, se.nr_wakeups); if (sync) schedstat_inc(p, se.nr_wakeups_sync); @@ -2548,6 +2547,7 @@ out_activate: success = 1; out_running: + ftrace_wake_up_task(rq, p, rq->curr); check_preempt_curr(rq, p); p->state = TASK_RUNNING; -- cgit v1.2.3 From 694379e9ed4f2f6babe111bf001c69e2e263338b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: make it more available in the Kconfig Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index eb1988ed84b7..ebc158e6d59a 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -14,7 +14,7 @@ config TRACING config FTRACE bool "Kernel Function Tracer" - depends on DEBUG_KERNEL && HAVE_FTRACE + depends on HAVE_FTRACE select FRAME_POINTER select TRACING select CONTEXT_SWITCH_TRACER @@ -72,7 +72,6 @@ config PREEMPT_TRACER config SCHED_TRACER bool "Scheduling Latency Tracer" - depends on DEBUG_KERNEL select TRACING select CONTEXT_SWITCH_TRACER select TRACER_MAX_TRACE @@ -82,7 +81,6 @@ config SCHED_TRACER config CONTEXT_SWITCH_TRACER bool "Trace process context switches" - depends on DEBUG_KERNEL select TRACING select MARKERS help -- cgit v1.2.3 From 8f96da02c14d722ad9a3713cd7273ce28c9036ad Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: remove wakeup from function trace trace_function is called by mcount and calling wake_up from that can have unpredictable results. This patch removes the wakeup from trace_function. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index b63fe909f87b..736dcfb3ed01 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -659,8 +659,6 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; spin_unlock_irqrestore(&data->lock, irq_flags); - - trace_wake_up(); } void -- cgit v1.2.3 From 4fe8c3048cd8280a54256bca9cac2007bd546c33 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: printk and trace irqsoff and wakeups printk called from wakeup critical timings and irqs off can cause deadlocks since printk might do a wakeup itself. If the call to printk happens with the runqueue lock held, it can deadlock. This patch protects the printk from being called in trace irqs off with a test to see if the runqueue for the current CPU is locked. If it is locked, the printk is skipped. The wakeup always holds the runqueue lock, so the printk is simply removed. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_irqsoff.c | 26 ++++++++++++++------------ kernel/trace/trace_sched_wakeup.c | 13 ------------- 2 files changed, 14 insertions(+), 25 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 7a4dc014b8ab..d0c1748b1e2c 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -165,18 +165,20 @@ check_critical_timing(struct trace_array *tr, update_max_tr_single(tr, current, cpu); - if (tracing_thresh) { - printk(KERN_INFO "(%16s-%-5d|#%d):" - " %lu us critical section violates %lu us threshold.\n", - current->comm, current->pid, - raw_smp_processor_id(), - latency, nsecs_to_usecs(tracing_thresh)); - } else { - printk(KERN_INFO "(%16s-%-5d|#%d):" - " new %lu us maximum-latency critical section.\n", - current->comm, current->pid, - raw_smp_processor_id(), - latency); + if (!runqueue_is_locked()) { + if (tracing_thresh) { + printk(KERN_INFO "(%16s-%-5d|#%d): %lu us critical" + " section violates %lu us threshold.\n", + current->comm, current->pid, + raw_smp_processor_id(), + latency, nsecs_to_usecs(tracing_thresh)); + } else { + printk(KERN_INFO "(%16s-%-5d|#%d): new %lu us" + " maximum-latency critical section.\n", + current->comm, current->pid, + raw_smp_processor_id(), + latency); + } } max_sequence++; diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 2a012423f9d0..5948011006bc 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -106,19 +106,6 @@ wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) update_max_tr(tr, wakeup_task, wakeup_cpu); - if (tracing_thresh) { - printk(KERN_INFO "(%16s-%-5d|#%d):" - " %lu us wakeup latency violates %lu us threshold.\n", - wakeup_task->comm, wakeup_task->pid, - raw_smp_processor_id(), - latency, nsecs_to_usecs(tracing_thresh)); - } else { - printk(KERN_INFO "(%16s-%-5d|#%d):" - " new %lu us maximum wakeup latency.\n", - wakeup_task->comm, wakeup_task->pid, - cpu, latency); - } - out_unlock: __wakeup_reset(tr); spin_unlock_irqrestore(&wakeup_lock, flags); -- cgit v1.2.3 From 06fa75ab566c50e01bfd7b055bde85cf9b1bc98a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: add TRACE_STACK and TRACE_SPECIAL to selftest validation The selftest validation code checks for valid entries in the trace buffer. TRACE_STACK and TRACE_SPECIAL have been added to the code but not to the validator. This patch adds the two to prevent them from flagging a failure in the selftest. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_selftest.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 39dd452647da..92f4acb7740c 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -9,6 +9,8 @@ static inline int trace_valid_entry(struct trace_entry *entry) case TRACE_FN: case TRACE_CTX: case TRACE_WAKE: + case TRACE_STACK: + case TRACE_SPECIAL: return 1; } return 0; @@ -180,7 +182,7 @@ int trace_selftest_startup_dynamic_tracing(struct tracer *trace, /* we should only have one item */ if (!ret && count != 1) { - printk(KERN_CONT ".. filter failed .."); + printk(KERN_CONT ".. filter failed count=%ld ..", count); ret = -1; goto out; } -- cgit v1.2.3 From d05cdb25d80f06f77aa6bddb53cd1390d4d91a0b Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: fix dynamic ftrace selftest With the adding of the configuration changes in the Makefile to prevent tracing of functions in the ftrace code, all tracing of all the ftrace code has been removed. Unfortunately, one of the selftests, relied on a function to be traced. With the new change, the function was no longer traced and the test failed. This patch separates out the test function into its own file so that we can add the "-pg" flag to the compilation of that function and the adding of the mcount call to that function. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Makefile | 4 ++++ kernel/trace/trace.h | 2 ++ kernel/trace/trace_selftest.c | 6 ------ kernel/trace/trace_selftest_dynamic.c | 7 +++++++ 4 files changed, 13 insertions(+), 6 deletions(-) create mode 100644 kernel/trace/trace_selftest_dynamic.c (limited to 'kernel') diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index c25a6cd6a529..d9efbbfa2bdf 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -4,6 +4,10 @@ ifdef CONFIG_FTRACE ORIG_CFLAGS := $(KBUILD_CFLAGS) KBUILD_CFLAGS = $(subst -pg,,$(ORIG_CFLAGS)) + +# selftest needs instrumentation +CFLAGS_trace_selftest_dynamic.o = -pg +obj-y += trace_selftest_dynamic.o endif obj-$(CONFIG_FTRACE) += libftrace.o diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 96951a8d09a4..98cbfd05d754 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -244,6 +244,8 @@ extern int unregister_tracer_switch(struct tracer_switch_ops *ops); #ifdef CONFIG_DYNAMIC_FTRACE extern unsigned long ftrace_update_tot_cnt; +#define DYN_FTRACE_TEST_NAME trace_selftest_dynamic_test_func +extern int DYN_FTRACE_TEST_NAME(void); #endif #ifdef CONFIG_FTRACE_STARTUP_TEST diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 92f4acb7740c..83e55a2000cc 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -107,14 +107,8 @@ static int trace_test_buffer(struct trace_array *tr, unsigned long *count) #ifdef CONFIG_DYNAMIC_FTRACE -#define DYN_FTRACE_TEST_NAME trace_selftest_dynamic_test_func #define __STR(x) #x #define STR(x) __STR(x) -static int DYN_FTRACE_TEST_NAME(void) -{ - /* used to call mcount */ - return 0; -} /* Test dynamic code modification and ftrace filters */ int trace_selftest_startup_dynamic_tracing(struct tracer *trace, diff --git a/kernel/trace/trace_selftest_dynamic.c b/kernel/trace/trace_selftest_dynamic.c new file mode 100644 index 000000000000..54dd77cce5bf --- /dev/null +++ b/kernel/trace/trace_selftest_dynamic.c @@ -0,0 +1,7 @@ +#include "trace.h" + +int DYN_FTRACE_TEST_NAME(void) +{ + /* used to call mcount */ + return 0; +} -- cgit v1.2.3 From 4d9493c90f8e6e1b164aede3814010a290161abb Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:54 +0200 Subject: ftrace: remove add-hoc code Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 47 --------------------------------------- kernel/sched_fair.c | 3 --- kernel/trace/trace_sched_switch.c | 10 ++------- 3 files changed, 2 insertions(+), 58 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 1ec3fb2efee6..ad95cca4e42e 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2412,53 +2412,6 @@ static int sched_balance_self(int cpu, int flag) #endif /* CONFIG_SMP */ -#ifdef CONFIG_CONTEXT_SWITCH_TRACER - -void ftrace_task(struct task_struct *p, void *__tr, void *__data) -{ -#if 0 - /* - * trace timeline tree - */ - __trace_special(__tr, __data, - p->pid, p->se.vruntime, p->se.sum_exec_runtime); -#else - /* - * trace balance metrics - */ - __trace_special(__tr, __data, - p->pid, p->se.avg_overlap, 0); -#endif -} - -void ftrace_all_fair_tasks(void *__rq, void *__tr, void *__data) -{ - struct task_struct *p; - struct sched_entity *se; - struct rb_node *curr; - struct rq *rq = __rq; - - if (rq->cfs.curr) { - p = task_of(rq->cfs.curr); - ftrace_task(p, __tr, __data); - } - if (rq->cfs.next) { - p = task_of(rq->cfs.next); - ftrace_task(p, __tr, __data); - } - - for (curr = first_fair(&rq->cfs); curr; curr = rb_next(curr)) { - se = rb_entry(curr, struct sched_entity, run_node); - if (!entity_is_task(se)) - continue; - - p = task_of(se); - ftrace_task(p, __tr, __data); - } -} - -#endif - /*** * try_to_wake_up - wake up a thread * @p: the to-be-woken-up thread diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index dc1856f10795..e24ecd39c4b8 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -1061,8 +1061,6 @@ wake_affine(struct rq *rq, struct sched_domain *this_sd, struct rq *this_rq, if (!(this_sd->flags & SD_WAKE_AFFINE)) return 0; - ftrace_special(__LINE__, curr->se.avg_overlap, sync); - ftrace_special(__LINE__, p->se.avg_overlap, -1); /* * If the currently running task will sleep within * a reasonable amount of time then attract this newly @@ -1240,7 +1238,6 @@ static void check_preempt_wakeup(struct rq *rq, struct task_struct *p) if (unlikely(se == pse)) return; - ftrace_special(__LINE__, p->pid, se->last_wakeup); cfs_rq_of(pse)->next = pse; /* diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index bddf676914ed..5671db0e1827 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -36,11 +36,8 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) { + if (likely(disabled == 1)) tracing_sched_switch_trace(tr, data, prev, next, flags); - if (trace_flags & TRACE_ITER_SCHED_TREE) - ftrace_all_fair_tasks(__rq, tr, data); - } atomic_dec(&data->disabled); local_irq_restore(flags); @@ -65,11 +62,8 @@ wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) data = tr->data[cpu]; disabled = atomic_inc_return(&data->disabled); - if (likely(disabled == 1)) { + if (likely(disabled == 1)) tracing_sched_wakeup_trace(tr, data, wakee, curr, flags); - if (trace_flags & TRACE_ITER_SCHED_TREE) - ftrace_all_fair_tasks(__rq, tr, data); - } atomic_dec(&data->disabled); local_irq_restore(flags); -- cgit v1.2.3 From c5f888cae49dfe3e86d9d1e0dab2b63ceb057be3 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:55 +0200 Subject: ftrace: irqsoff use raw_smp_processor_id This patch changes the use of __get_cpu_var to explicitly calling raw_smp_processor_id and using the per_cpu() macro. On some debug configurations, the use of __get_cpu_var may cause ftrace to trigger and this can cause problems with the irqsoff tracing. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_irqsoff.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index d0c1748b1e2c..761f3ec66c50 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -204,14 +204,14 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) if (likely(!tracer_enabled)) return; - if (__get_cpu_var(tracing_cpu)) + cpu = raw_smp_processor_id(); + + if (per_cpu(tracing_cpu, cpu)) return; - cpu = raw_smp_processor_id(); data = tr->data[cpu]; - if (unlikely(!data) || unlikely(!head_page(data)) || - atomic_read(&data->disabled)) + if (unlikely(!data) || atomic_read(&data->disabled)) return; atomic_inc(&data->disabled); @@ -225,7 +225,7 @@ start_critical_timing(unsigned long ip, unsigned long parent_ip) trace_function(tr, data, ip, parent_ip, flags); - __get_cpu_var(tracing_cpu) = 1; + per_cpu(tracing_cpu, cpu) = 1; atomic_dec(&data->disabled); } @@ -238,16 +238,16 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) struct trace_array_cpu *data; unsigned long flags; + cpu = raw_smp_processor_id(); /* Always clear the tracing cpu on stopping the trace */ - if (unlikely(__get_cpu_var(tracing_cpu))) - __get_cpu_var(tracing_cpu) = 0; + if (unlikely(per_cpu(tracing_cpu, cpu))) + per_cpu(tracing_cpu, cpu) = 0; else return; if (!tracer_enabled) return; - cpu = raw_smp_processor_id(); data = tr->data[cpu]; if (unlikely(!data) || unlikely(!head_page(data)) || @@ -255,6 +255,7 @@ stop_critical_timing(unsigned long ip, unsigned long parent_ip) return; atomic_inc(&data->disabled); + local_save_flags(flags); trace_function(tr, data, ip, parent_ip, flags); check_critical_timing(tr, data, parent_ip ? : ip, cpu); @@ -376,7 +377,7 @@ static void stop_irqsoff_tracer(struct trace_array *tr) static void __irqsoff_tracer_init(struct trace_array *tr) { irqsoff_trace = tr; - /* make sure that the tracer is visibel */ + /* make sure that the tracer is visible */ smp_wmb(); if (tr->ctrl) -- cgit v1.2.3 From 92205c2343527a863d660360599a4bf8cede77b0 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:55 +0200 Subject: ftrace: user raw_spin_lock in tracing Lock debugging enabled cause huge performance problems for tracing. Having the lock verification happening for every function that is called because mcount calls spin_lock can cripple the system. This patch converts the spin_locks used by ftrace into raw_spin_locks. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 51 ++++++++++++++++++++++++++++++--------------------- kernel/trace/trace.h | 2 +- 2 files changed, 31 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 736dcfb3ed01..3009aafa4dde 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -133,7 +133,8 @@ static const char *trace_options[] = { NULL }; -static DEFINE_SPINLOCK(ftrace_max_lock); +static raw_spinlock_t ftrace_max_lock = + (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; /* * Copy the new maximum trace into the separate maximum-trace @@ -335,7 +336,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) int i; WARN_ON_ONCE(!irqs_disabled()); - spin_lock(&ftrace_max_lock); + __raw_spin_lock(&ftrace_max_lock); /* clear out all the previous traces */ for_each_possible_cpu(i) { data = tr->data[i]; @@ -344,7 +345,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) } __update_max_tr(tr, tsk, cpu); - spin_unlock(&ftrace_max_lock); + __raw_spin_unlock(&ftrace_max_lock); } /** @@ -360,7 +361,7 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) int i; WARN_ON_ONCE(!irqs_disabled()); - spin_lock(&ftrace_max_lock); + __raw_spin_lock(&ftrace_max_lock); for_each_possible_cpu(i) tracing_reset(max_tr.data[i]); @@ -368,7 +369,7 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) tracing_reset(data); __update_max_tr(tr, tsk, cpu); - spin_unlock(&ftrace_max_lock); + __raw_spin_unlock(&ftrace_max_lock); } int register_tracer(struct tracer *type) @@ -652,13 +653,15 @@ trace_function(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_FN; entry->fn.ip = ip; entry->fn.parent_ip = parent_ip; - spin_unlock_irqrestore(&data->lock, irq_flags); + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); } void @@ -678,14 +681,16 @@ __trace_special(void *__tr, void *__data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_SPECIAL; entry->special.arg1 = arg1; entry->special.arg2 = arg2; entry->special.arg3 = arg3; - spin_unlock_irqrestore(&data->lock, irq_flags); + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } @@ -725,7 +730,8 @@ tracing_sched_switch_trace(struct trace_array *tr, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_CTX; @@ -736,7 +742,8 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.next_prio = next->prio; entry->ctx.next_state = next->state; __trace_stack(tr, data, flags, 4); - spin_unlock_irqrestore(&data->lock, irq_flags); + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); } void @@ -749,7 +756,8 @@ tracing_sched_wakeup_trace(struct trace_array *tr, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, flags); entry->type = TRACE_WAKE; @@ -760,7 +768,8 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_prio = wakee->prio; entry->ctx.next_state = wakee->state; __trace_stack(tr, data, flags, 5); - spin_unlock_irqrestore(&data->lock, irq_flags); + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } @@ -1824,7 +1833,8 @@ tracing_cpumask_write(struct file *filp, const char __user *ubuf, if (err) goto err_unlock; - spin_lock_irq(&ftrace_max_lock); + raw_local_irq_disable(); + __raw_spin_lock(&ftrace_max_lock); for_each_possible_cpu(cpu) { /* * Increase/decrease the disabled counter if we are @@ -1839,7 +1849,8 @@ tracing_cpumask_write(struct file *filp, const char __user *ubuf, atomic_dec(&global_trace.data[cpu]->disabled); } } - spin_unlock_irq(&ftrace_max_lock); + __raw_spin_unlock(&ftrace_max_lock); + raw_local_irq_enable(); tracing_cpumask = tracing_cpumask_new; @@ -2299,7 +2310,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, for_each_cpu_mask(cpu, mask) { data = iter->tr->data[cpu]; - spin_lock(&data->lock); + __raw_spin_lock(&data->lock); } while (find_next_entry_inc(iter) != NULL) { @@ -2320,7 +2331,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, for_each_cpu_mask(cpu, mask) { data = iter->tr->data[cpu]; - spin_unlock(&data->lock); + __raw_spin_unlock(&data->lock); } for_each_cpu_mask(cpu, mask) { @@ -2538,8 +2549,7 @@ static int trace_alloc_page(void) /* Now that we successfully allocate a page per CPU, add them */ for_each_possible_cpu(i) { data = global_trace.data[i]; - spin_lock_init(&data->lock); - lockdep_set_class(&data->lock, &data->lock_key); + data->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -2547,8 +2557,7 @@ static int trace_alloc_page(void) #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; - spin_lock_init(&data->lock); - lockdep_set_class(&data->lock, &data->lock_key); + data->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 98cbfd05d754..25cba28eb9ba 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -76,7 +76,7 @@ struct trace_entry { struct trace_array_cpu { struct list_head trace_pages; atomic_t disabled; - spinlock_t lock; + raw_spinlock_t lock; struct lock_class_key lock_key; /* these fields get copied into max-trace: */ -- cgit v1.2.3 From 1d09daa55d2e9bab7e7d30f0d05e5a7bc60b2a4a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:55 +0200 Subject: ftrace: use Makefile to remove tracing from lockdep This patch removes the "notrace" annotation from lockdep and adds the debugging files in the kernel director to those that should not be compiled with "-pg" mcount tracing. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/Makefile | 8 ++++++++ kernel/lockdep.c | 26 +++++++++++++------------- 2 files changed, 21 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index 7e344e7b0cb3..d2f80ea4cd9a 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -11,6 +11,14 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o profile.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o +ifdef CONFIG_FTRACE +# Do not profile debug utilities +ORIG_CFLAGS := $(KBUILD_CFLAGS) +KBUILD_CFLAGS = $(if $(filter-out lockdep% %debug,$(basename $(notdir $@))), \ + $(ORIG_CFLAGS), \ + $(subst -pg,,$(ORIG_CFLAGS))) +endif + obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o obj-$(CONFIG_STACKTRACE) += stacktrace.o obj-y += time/ diff --git a/kernel/lockdep.c b/kernel/lockdep.c index ac46847ba0c9..90a440cbd6de 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -271,14 +271,14 @@ static struct list_head chainhash_table[CHAINHASH_SIZE]; ((key1) >> (64-MAX_LOCKDEP_KEYS_BITS)) ^ \ (key2)) -notrace void lockdep_off(void) +void lockdep_off(void) { current->lockdep_recursion++; } EXPORT_SYMBOL(lockdep_off); -notrace void lockdep_on(void) +void lockdep_on(void) { current->lockdep_recursion--; } @@ -1041,7 +1041,7 @@ find_usage_forwards(struct lock_class *source, unsigned int depth) * Return 1 otherwise and keep unchanged. * Return 0 on error. */ -static noinline notrace int +static noinline int find_usage_backwards(struct lock_class *source, unsigned int depth) { struct lock_list *entry; @@ -1591,7 +1591,7 @@ static inline int validate_chain(struct task_struct *curr, * We are building curr_chain_key incrementally, so double-check * it from scratch, to make sure that it's done correctly: */ -static notrace void check_chain_key(struct task_struct *curr) +static void check_chain_key(struct task_struct *curr) { #ifdef CONFIG_DEBUG_LOCKDEP struct held_lock *hlock, *prev_hlock = NULL; @@ -1967,7 +1967,7 @@ static int mark_lock_irq(struct task_struct *curr, struct held_lock *this, /* * Mark all held locks with a usage bit: */ -static notrace int +static int mark_held_locks(struct task_struct *curr, int hardirq) { enum lock_usage_bit usage_bit; @@ -2014,7 +2014,7 @@ void early_boot_irqs_on(void) /* * Hardirqs will be enabled: */ -void notrace trace_hardirqs_on_caller(unsigned long a0) +void trace_hardirqs_on_caller(unsigned long a0) { struct task_struct *curr = current; unsigned long ip; @@ -2060,7 +2060,7 @@ void notrace trace_hardirqs_on_caller(unsigned long a0) } EXPORT_SYMBOL(trace_hardirqs_on_caller); -void notrace trace_hardirqs_on(void) +void trace_hardirqs_on(void) { trace_hardirqs_on_caller(CALLER_ADDR0); } @@ -2069,7 +2069,7 @@ EXPORT_SYMBOL(trace_hardirqs_on); /* * Hardirqs were disabled: */ -void notrace trace_hardirqs_off_caller(unsigned long a0) +void trace_hardirqs_off_caller(unsigned long a0) { struct task_struct *curr = current; @@ -2094,7 +2094,7 @@ void notrace trace_hardirqs_off_caller(unsigned long a0) } EXPORT_SYMBOL(trace_hardirqs_off_caller); -void notrace trace_hardirqs_off(void) +void trace_hardirqs_off(void) { trace_hardirqs_off_caller(CALLER_ADDR0); } @@ -2260,7 +2260,7 @@ static inline int separate_irq_context(struct task_struct *curr, /* * Mark a lock with a usage bit, and validate the state transition: */ -static notrace int mark_lock(struct task_struct *curr, struct held_lock *this, +static int mark_lock(struct task_struct *curr, struct held_lock *this, enum lock_usage_bit new_bit) { unsigned int new_mask = 1 << new_bit, ret = 1; @@ -2663,7 +2663,7 @@ __lock_release(struct lockdep_map *lock, int nested, unsigned long ip) /* * Check whether we follow the irq-flags state precisely: */ -static notrace void check_flags(unsigned long flags) +static void check_flags(unsigned long flags) { #if defined(CONFIG_DEBUG_LOCKDEP) && defined(CONFIG_TRACE_IRQFLAGS) if (!debug_locks) @@ -2700,7 +2700,7 @@ static notrace void check_flags(unsigned long flags) * We are not always called with irqs disabled - do that here, * and also avoid lockdep recursion: */ -notrace void lock_acquire(struct lockdep_map *lock, unsigned int subclass, +void lock_acquire(struct lockdep_map *lock, unsigned int subclass, int trylock, int read, int check, unsigned long ip) { unsigned long flags; @@ -2723,7 +2723,7 @@ notrace void lock_acquire(struct lockdep_map *lock, unsigned int subclass, EXPORT_SYMBOL_GPL(lock_acquire); -notrace void lock_release(struct lockdep_map *lock, int nested, +void lock_release(struct lockdep_map *lock, int nested, unsigned long ip) { unsigned long flags; -- cgit v1.2.3 From c1d2327b36f2261ffa8ff7227321ba900c7eee7f Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:55 +0200 Subject: ftrace: restrict tracing to HAVE_FTRACE architectures Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index ebc158e6d59a..f3005717bcd0 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -32,6 +32,7 @@ config IRQSOFF_TRACER default n depends on TRACE_IRQFLAGS_SUPPORT depends on GENERIC_TIME + depends on HAVE_FTRACE select TRACE_IRQFLAGS select TRACING select TRACER_MAX_TRACE @@ -54,6 +55,7 @@ config PREEMPT_TRACER default n depends on GENERIC_TIME depends on PREEMPT + depends on HAVE_FTRACE select TRACING select TRACER_MAX_TRACE help @@ -72,6 +74,7 @@ config PREEMPT_TRACER config SCHED_TRACER bool "Scheduling Latency Tracer" + depends on HAVE_FTRACE select TRACING select CONTEXT_SWITCH_TRACER select TRACER_MAX_TRACE @@ -81,6 +84,7 @@ config SCHED_TRACER config CONTEXT_SWITCH_TRACER bool "Trace process context switches" + depends on HAVE_FTRACE select TRACING select MARKERS help -- cgit v1.2.3 From 07a267cdd2fd7d1de9455b1e36a1635ace7276c7 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:55 +0200 Subject: ftrace: add UNINTERRUPTIBLE state for kftraced on disable When dynamic ftrace fails and sets itself disabled, the ftraced daemon will go back to sleep everytime it wakes up. The setting of the ftraced state to UNINTERRUPTIBLE is skipped in this process, and the daemon takes up 100% of the CPU. This patch makes sure the ftraced daemon sets itself to UNINTERRUPTIBLE in that loop. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 281d97a3208c..40f64f7cd850 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -630,10 +630,10 @@ static int ftraced(void *ignore) { unsigned long usecs; - set_current_state(TASK_INTERRUPTIBLE); - while (!kthread_should_stop()) { + set_current_state(TASK_INTERRUPTIBLE); + /* check once a second */ schedule_timeout(HZ); @@ -667,8 +667,6 @@ static int ftraced(void *ignore) wake_up_interruptible(&ftraced_waiters); ftrace_shutdown_replenish(); - - set_current_state(TASK_INTERRUPTIBLE); } __set_current_state(TASK_RUNNING); return 0; -- cgit v1.2.3 From d15f57f23eaba975309a153b23699cd0c0236974 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:56 +0200 Subject: ftrace: fix mutex unlock in trace output If the trace output changes on reading the trace files, there is a chance that the start function will return NULL. If the start function of a sequence returns NULL the stop equivalent is not called. In this case, all locks that are taken must be released even if they are released in the stop function. This patch fixes a case that a mutex was not released on return of NULL in the start sequence function. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3009aafa4dde..ea11f4ebfae1 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -964,8 +964,10 @@ static void *s_start(struct seq_file *m, loff_t *pos) mutex_lock(&trace_types_lock); - if (!current_trace || current_trace != iter->trace) + if (!current_trace || current_trace != iter->trace) { + mutex_unlock(&trace_types_lock); return NULL; + } atomic_inc(&trace_record_cmdline_disabled); -- cgit v1.2.3 From 30afdcb1de0a37a2086145a82ca3febebe47d019 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:56 +0200 Subject: ftrace: selftest protect againt max flip There is a slight race condition in the selftest where the max update of the wakeup and irqs/preemption off tests can be doing a max update as the buffers are being tested. If this happens the system can crash with a GPF. This patch adds the max update spinlock around the checking of the buffers to prevent such a race. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_selftest.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 83e55a2000cc..395274e783b3 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -82,10 +82,12 @@ trace_test_buffer_cpu(struct trace_array *tr, struct trace_array_cpu *data) */ static int trace_test_buffer(struct trace_array *tr, unsigned long *count) { - unsigned long cnt = 0; - int cpu; - int ret = 0; + unsigned long flags, cnt = 0; + int cpu, ret = 0; + /* Don't allow flipping of max traces now */ + raw_local_irq_save(flags); + __raw_spin_lock(&ftrace_max_lock); for_each_possible_cpu(cpu) { if (!head_page(tr->data[cpu])) continue; @@ -96,6 +98,8 @@ static int trace_test_buffer(struct trace_array *tr, unsigned long *count) if (ret) break; } + __raw_spin_unlock(&ftrace_max_lock); + raw_local_irq_restore(flags); if (count) *count = cnt; -- cgit v1.2.3 From 72829bc3d63cdc592d8f7dd86ad3b3fe8900fb74 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 23 May 2008 21:37:28 +0200 Subject: ftrace: move enums to ftrace.h and make helper function global picked from the mmiotracer patches to distangle the patch queues. Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 24 ++++++++---------------- kernel/trace/trace.h | 15 +++++++++++++++ 2 files changed, 23 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index ea11f4ebfae1..0eef0503febd 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -37,7 +37,7 @@ unsigned long __read_mostly tracing_thresh; static int tracing_disabled = 1; -static long +long ns2usecs(cycle_t nsec) { nsec += 500; @@ -96,18 +96,6 @@ unsigned long nsecs_to_usecs(unsigned long nsecs) return nsecs / 1000; } -enum trace_type { - __TRACE_FIRST_TYPE = 0, - - TRACE_FN, - TRACE_CTX, - TRACE_WAKE, - TRACE_STACK, - TRACE_SPECIAL, - - __TRACE_LAST_TYPE -}; - enum trace_flag_type { TRACE_FLAG_IRQS_OFF = 0x01, TRACE_FLAG_NEED_RESCHED = 0x02, @@ -190,7 +178,7 @@ void *head_page(struct trace_array_cpu *data) return page_address(page); } -static int +int trace_seq_printf(struct trace_seq *s, const char *fmt, ...) { int len = (PAGE_SIZE - 1) - s->len; @@ -205,7 +193,7 @@ trace_seq_printf(struct trace_seq *s, const char *fmt, ...) va_end(ap); /* If we can't write it all, don't bother writing anything */ - if (ret > len) + if (ret >= len) return 0; s->len += ret; @@ -638,7 +626,7 @@ tracing_generic_entry_update(struct trace_entry *entry, unsigned long flags) pc = preempt_count(); entry->preempt_count = pc & 0xff; - entry->pid = tsk->pid; + entry->pid = (tsk) ? tsk->pid : 0; entry->t = ftrace_now(raw_smp_processor_id()); entry->flags = (irqs_disabled_flags(flags) ? TRACE_FLAG_IRQS_OFF : 0) | ((pc & HARDIRQ_MASK) ? TRACE_FLAG_HARDIRQ : 0) | @@ -1541,6 +1529,9 @@ static int trace_empty(struct trace_iterator *iter) static int print_trace_line(struct trace_iterator *iter) { + if (iter->trace && iter->trace->print_line) + return iter->trace->print_line(iter); + if (trace_flags & TRACE_ITER_BIN) return print_bin_fmt(iter); @@ -2162,6 +2153,7 @@ static int tracing_open_pipe(struct inode *inode, struct file *filp) return -ENOMEM; iter->tr = &global_trace; + iter->trace = current_trace; filp->private_data = iter; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 25cba28eb9ba..b0ca7473671b 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -6,6 +6,18 @@ #include #include +enum trace_type { + __TRACE_FIRST_TYPE = 0, + + TRACE_FN, + TRACE_CTX, + TRACE_WAKE, + TRACE_STACK, + TRACE_SPECIAL, + + __TRACE_LAST_TYPE +}; + /* * Function trace entry - function address and parent function addres: */ @@ -130,6 +142,7 @@ struct tracer { int (*selftest)(struct tracer *trace, struct trace_array *tr); #endif + int (*print_line)(struct trace_iterator *iter); struct tracer *next; int print_max; }; @@ -276,6 +289,8 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace, #endif /* CONFIG_FTRACE_STARTUP_TEST */ extern void *head_page(struct trace_array_cpu *data); +extern int trace_seq_printf(struct trace_seq *s, const char *fmt, ...); +extern long ns2usecs(cycle_t nsec); extern unsigned long trace_flags; -- cgit v1.2.3 From d17d969160c18b631a19c2b34d260691402650f8 Mon Sep 17 00:00:00 2001 From: Ankita Garg Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: fix conversion of task state to char in latency tracer The conversion of task states to a character in the sched_switch tracer (part of latency tracer infrastructure), seems to be incorrect. We currently do it by indexing into the state_to_char array using the state value. The state values do not map directly into the array index and are thus incorrect. The following patch addresses this issue. This is also what is being done even in the show_task() routine in kernel/sched.c The patch has been compile and run tested. Signed-off-by: Ankita Garg Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0eef0503febd..9197782d15c9 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1208,6 +1208,7 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) char *comm; int S, T; int i; + unsigned state; if (!next_entry) next_entry = entry; @@ -1238,11 +1239,11 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) break; case TRACE_CTX: case TRACE_WAKE: - S = entry->ctx.prev_state < sizeof(state_to_char) ? - state_to_char[entry->ctx.prev_state] : 'X'; T = entry->ctx.next_state < sizeof(state_to_char) ? state_to_char[entry->ctx.next_state] : 'X'; + state = entry->ctx.prev_state ? __ffs(entry->ctx.prev_state) + 1 : 0; + S = state < sizeof(state_to_char) - 1 ? state_to_char[state] : 'X'; comm = trace_find_cmdline(entry->ctx.next_pid); trace_seq_printf(s, " %5d:%3d:%c %s %5d:%3d:%c %s\n", entry->ctx.prev_pid, -- cgit v1.2.3 From 8487c23765b6e0444ec6b5f1530766d63fe68e35 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: allow trace_pipe to block on all reads We expect things like "cat" to block on reads to trace_pipe. That's what trace_pipe is for. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9197782d15c9..fd4ecc292005 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2231,8 +2231,6 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, start = 0; while (trace_empty(iter)) { - if (!(trace_flags & TRACE_ITER_BLOCK)) - return -EWOULDBLOCK; /* * This is a make-shift waitqueue. The reason we don't use * an actual wait queue is because: -- cgit v1.2.3 From b5685aede3b7b65e72ddc73b951aa1f70798a614 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: restore iterator trace in pipe read The trace iterator is reset in the read. We still need to restore the tracer that the trace_pipe was opened with. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index fd4ecc292005..d141fc98f3a8 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2202,6 +2202,8 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, { struct trace_iterator *iter = filp->private_data; struct trace_array_cpu *data; + struct trace_array *tr = iter->tr; + struct tracer *tracer = iter->trace; static cpumask_t mask; static int start; unsigned long flags; @@ -2274,7 +2276,8 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cnt = PAGE_SIZE - 1; memset(iter, 0, sizeof(*iter)); - iter->tr = &global_trace; + iter->tr = tr; + iter->trace = tracer; iter->pos = -1; /* -- cgit v1.2.3 From 845279972f1736c3463c9cebd1bad92a0a347176 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: return EOF in trace_pipe on change of tracer Break out of while loop with EOF when the current_trace changes. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d141fc98f3a8..2af940433e96 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2253,6 +2253,9 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, if (signal_pending(current)) return -EINTR; + if (iter->trace != current_trace) + return 0; + /* * We block until we read something and tracing is disabled. * We still block if tracing is disabled, but we have never -- cgit v1.2.3 From 2dc8f09571a61d9cb3dc47bba6608689dfe15d16 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: trace_pipe implement NONBLOCK This patch implements "NONBLOCK" for trace_pipe. If the trace_pipe is opened with O_NONBLOCK, then the trace_pipe read will not block when buffer is empty. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 2af940433e96..3b4eaf36ed5d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2233,6 +2233,10 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, start = 0; while (trace_empty(iter)) { + + if ((filp->f_flags & O_NONBLOCK)) + return -EAGAIN; + /* * This is a make-shift waitqueue. The reason we don't use * an actual wait queue is because: -- cgit v1.2.3 From 05bd68c514579e007b46e4fa0461b78416a3f4c2 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:59 +0200 Subject: ftrace: user proper API for setting RT prios in selftest The wakeup selftest used an internal API for setting the test task priority. This patch fixes it to use the proper API for performing such a task. Thanks goes to Randy Dunlap for pointing out this build failure. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_selftest.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 395274e783b3..a5f6001c3332 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -410,11 +410,11 @@ trace_selftest_startup_preemptirqsoff(struct tracer *trace, struct trace_array * #ifdef CONFIG_SCHED_TRACER static int trace_wakeup_test_thread(void *data) { - struct completion *x = data; - /* Make this a RT thread, doesn't need to be too high */ + struct sched_param param = { .sched_priority = 5 }; + struct completion *x = data; - rt_mutex_setprio(current, MAX_RT_PRIO - 5); + sched_setscheduler(current, SCHED_FIFO, ¶m); /* Make it know we have a new prio */ complete(x); -- cgit v1.2.3 From a98a3c3fde3ae7614f19758a043691b6f59dac53 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:20:59 +0200 Subject: ftrace: trace_entries to dynamically change trace buffer size This patch adds /debug/tracing/trace_entries that allows users to see as well as modify the number of trace entries the buffers hold. The number of entries only increments in ENTRIES_PER_PAGE which is calculated by the size of an entry with the number of entries that can fit in a page. The user does not need to use an exact size, but the entries will be rounded to one of the increments. Trying to set the entries to 0 will return with -EINVAL. To avoid race conditions, the modification of the buffer size can only be done when tracing is completely disabled (current_tracer == none). A info message will be printed if a user tries to modify the buffer size when not set to none. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 145 ++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 137 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3b4eaf36ed5d..4723e012151d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -35,6 +35,15 @@ unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; unsigned long __read_mostly tracing_thresh; +/* dummy trace to disable tracing */ +static struct tracer no_tracer __read_mostly = +{ + .name = "none", +}; + +static int trace_alloc_page(void); +static int trace_free_page(void); + static int tracing_disabled = 1; long @@ -2364,6 +2373,70 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, return read; } +static ssize_t +tracing_entries_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + struct trace_array *tr = filp->private_data; + char buf[64]; + int r; + + r = sprintf(buf, "%lu\n", tr->entries); + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); +} + +static ssize_t +tracing_entries_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + unsigned long val; + char buf[64]; + + if (cnt > 63) + cnt = 63; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 10); + + /* must have at least 1 entry */ + if (!val) + return -EINVAL; + + mutex_lock(&trace_types_lock); + + if (current_trace != &no_tracer) { + cnt = -EBUSY; + pr_info("ftrace: set current_tracer to none" + " before modifying buffer size\n"); + goto out; + } + + if (val > global_trace.entries) { + while (global_trace.entries < val) { + if (trace_alloc_page()) { + cnt = -ENOMEM; + goto out; + } + } + } else { + /* include the number of entries in val (inc of page entries) */ + while (global_trace.entries > val + (ENTRIES_PER_PAGE - 1)) + trace_free_page(); + } + + filp->f_pos += cnt; + + out: + max_tr.entries = global_trace.entries; + mutex_unlock(&trace_types_lock); + + return cnt; +} + static struct file_operations tracing_max_lat_fops = { .open = tracing_open_generic, .read = tracing_max_lat_read, @@ -2389,6 +2462,12 @@ static struct file_operations tracing_pipe_fops = { .release = tracing_release_pipe, }; +static struct file_operations tracing_entries_fops = { + .open = tracing_open_generic, + .read = tracing_entries_read, + .write = tracing_entries_write, +}; + #ifdef CONFIG_DYNAMIC_FTRACE static ssize_t @@ -2500,6 +2579,12 @@ static __init void tracer_init_debugfs(void) pr_warning("Could not create debugfs " "'tracing_threash' entry\n"); + entry = debugfs_create_file("trace_entries", 0644, d_tracer, + &global_trace, &tracing_entries_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'tracing_threash' entry\n"); + #ifdef CONFIG_DYNAMIC_FTRACE entry = debugfs_create_file("dyn_ftrace_total_info", 0444, d_tracer, &ftrace_update_tot_cnt, @@ -2510,12 +2595,6 @@ static __init void tracer_init_debugfs(void) #endif } -/* dummy trace to disable tracing */ -static struct tracer no_tracer __read_mostly = -{ - .name = "none", -}; - static int trace_alloc_page(void) { struct trace_array_cpu *data; @@ -2552,7 +2631,6 @@ static int trace_alloc_page(void) /* Now that we successfully allocate a page per CPU, add them */ for_each_possible_cpu(i) { data = global_trace.data[i]; - data->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -2560,7 +2638,6 @@ static int trace_alloc_page(void) #ifdef CONFIG_TRACER_MAX_TRACE data = max_tr.data[i]; - data->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); list_add_tail(&page->lru, &data->trace_pages); @@ -2579,6 +2656,55 @@ static int trace_alloc_page(void) return -ENOMEM; } +static int trace_free_page(void) +{ + struct trace_array_cpu *data; + struct page *page; + struct list_head *p; + int i; + int ret = 0; + + /* free one page from each buffer */ + for_each_possible_cpu(i) { + data = global_trace.data[i]; + p = data->trace_pages.next; + if (p == &data->trace_pages) { + /* should never happen */ + WARN_ON(1); + tracing_disabled = 1; + ret = -1; + break; + } + page = list_entry(p, struct page, lru); + ClearPageLRU(page); + list_del(&page->lru); + __free_page(page); + + tracing_reset(data); + +#ifdef CONFIG_TRACER_MAX_TRACE + data = max_tr.data[i]; + p = data->trace_pages.next; + if (p == &data->trace_pages) { + /* should never happen */ + WARN_ON(1); + tracing_disabled = 1; + ret = -1; + break; + } + page = list_entry(p, struct page, lru); + ClearPageLRU(page); + list_del(&page->lru); + __free_page(page); + + tracing_reset(data); +#endif + } + global_trace.entries -= ENTRIES_PER_PAGE; + + return ret; +} + __init static int tracer_alloc_buffers(void) { struct trace_array_cpu *data; @@ -2609,6 +2735,9 @@ __init static int tracer_alloc_buffers(void) /* use the LRU flag to differentiate the two buffers */ ClearPageLRU(page); + data->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; + max_tr.data[i]->lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; + /* Only allocate if we are actually using the max trace */ #ifdef CONFIG_TRACER_MAX_TRACE array = (void *)__get_free_page(GFP_KERNEL); -- cgit v1.2.3 From bb065afb8ebd07a03155502dba29ebf0f6fe67e8 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: lockdep: update lockdep_recursion on graph_lock With the introduction of ftrace, it is possible to recurse into the lockdep functions via the mcount call. To prevent possible lockups, updating the lockdep_recursion counter on grabbing the internal lockdep_lock should prevent deadlocks. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/lockdep.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/lockdep.c b/kernel/lockdep.c index 90a440cbd6de..65548eff029e 100644 --- a/kernel/lockdep.c +++ b/kernel/lockdep.c @@ -82,6 +82,8 @@ static int graph_lock(void) __raw_spin_unlock(&lockdep_lock); return 0; } + /* prevent any recursions within lockdep from causing deadlocks */ + current->lockdep_recursion++; return 1; } @@ -90,6 +92,7 @@ static inline int graph_unlock(void) if (debug_locks && !__raw_spin_is_locked(&lockdep_lock)) return DEBUG_LOCKS_WARN_ON(1); + current->lockdep_recursion--; __raw_spin_unlock(&lockdep_lock); return 0; } -- cgit v1.2.3 From 93dcc6ea096c51cc30ad0f6647ed05fead2439bf Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: simplify hexprint simplify hex to ascii conversion. Signed-off-by: Thomas Gleixner Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 23 +++++------------------ 1 file changed, 5 insertions(+), 18 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 4723e012151d..627e39936ea7 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -248,19 +248,18 @@ trace_seq_putmem(struct trace_seq *s, void *mem, size_t len) } #define HEX_CHARS 17 +static const char hex2asc[] = "0123456789abcdef"; static int trace_seq_putmem_hex(struct trace_seq *s, void *mem, size_t len) { unsigned char hex[HEX_CHARS]; - unsigned char *data; + unsigned char *data = mem; unsigned char byte; int i, j; BUG_ON(len >= HEX_CHARS); - data = mem; - #ifdef __BIG_ENDIAN for (i = 0, j = 0; i < len; i++) { #else @@ -268,22 +267,10 @@ trace_seq_putmem_hex(struct trace_seq *s, void *mem, size_t len) #endif byte = data[i]; - hex[j] = byte & 0x0f; - if (hex[j] >= 10) - hex[j] += 'a' - 10; - else - hex[j] += '0'; - j++; - - hex[j] = byte >> 4; - if (hex[j] >= 10) - hex[j] += 'a' - 10; - else - hex[j] += '0'; - j++; + hex[j++] = hex2asc[byte & 0x0f]; + hex[j++] = hex2asc[byte >> 4]; } - hex[j] = ' '; - j++; + hex[j++] = ' '; return trace_seq_putmem(s, hex, j); } -- cgit v1.2.3 From afc2abc0ae4265730a0fc48618012193a09a1d10 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: cleanups no code changed. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 11 +++++------ 1 file changed, 5 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 627e39936ea7..d6b60576f991 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1155,12 +1155,12 @@ lat_print_generic(struct trace_seq *s, struct trace_entry *entry, int cpu) hardirq = entry->flags & TRACE_FLAG_HARDIRQ; softirq = entry->flags & TRACE_FLAG_SOFTIRQ; - if (hardirq && softirq) + if (hardirq && softirq) { trace_seq_putc(s, 'H'); - else { - if (hardirq) + } else { + if (hardirq) { trace_seq_putc(s, 'h'); - else { + } else { if (softirq) trace_seq_putc(s, 's'); else @@ -2177,8 +2177,7 @@ tracing_poll_pipe(struct file *filp, poll_table *poll_table) * Always select as readable when in blocking mode */ return POLLIN | POLLRDNORM; - } - else { + } else { if (!trace_empty(iter)) return POLLIN | POLLRDNORM; poll_wait(filp, &trace_wait, poll_table); -- cgit v1.2.3 From cffae437cdfbc8a5370d38cefbff1dfe34dad6ca Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: simple clean ups Andrew Morton mentioned some clean ups that should be done to ftrace. This patch does some of the simple clean ups. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 25 ++++++++++++------------- 1 file changed, 12 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d6b60576f991..0a367362ba2f 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -36,8 +36,7 @@ unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; unsigned long __read_mostly tracing_thresh; /* dummy trace to disable tracing */ -static struct tracer no_tracer __read_mostly = -{ +static struct tracer no_tracer __read_mostly = { .name = "none", }; @@ -1906,8 +1905,8 @@ tracing_iter_ctrl_write(struct file *filp, const char __user *ubuf, int neg = 0; int i; - if (cnt > 63) - cnt = 63; + if (cnt >= sizeof(buf)) + return -EINVAL; if (copy_from_user(&buf, ubuf, cnt)) return -EFAULT; @@ -1999,8 +1998,8 @@ tracing_ctrl_write(struct file *filp, const char __user *ubuf, long val; char buf[64]; - if (cnt > 63) - cnt = 63; + if (cnt >= sizeof(buf)) + return -EINVAL; if (copy_from_user(&buf, ubuf, cnt)) return -EFAULT; @@ -2099,10 +2098,10 @@ tracing_max_lat_read(struct file *filp, char __user *ubuf, char buf[64]; int r; - r = snprintf(buf, 64, "%ld\n", + r = snprintf(buf, sizeof(buf), "%ld\n", *ptr == (unsigned long)-1 ? -1 : nsecs_to_usecs(*ptr)); - if (r > 64) - r = 64; + if (r > sizeof(buf)) + r = sizeof(buf); return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); } @@ -2114,8 +2113,8 @@ tracing_max_lat_write(struct file *filp, const char __user *ubuf, long val; char buf[64]; - if (cnt > 63) - cnt = 63; + if (cnt >= sizeof(buf)) + return -EINVAL; if (copy_from_user(&buf, ubuf, cnt)) return -EFAULT; @@ -2378,8 +2377,8 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, unsigned long val; char buf[64]; - if (cnt > 63) - cnt = 63; + if (cnt >= sizeof(buf)) + return -EINVAL; if (copy_from_user(&buf, ubuf, cnt)) return -EFAULT; -- cgit v1.2.3 From c6caeeb142cd3a03c46107aac45c8effc02bbadb Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: replace simple_strtoul with strict_strtoul Andrew Morton suggested using strict_strtoul over simple_strtoul. This patch replaces them in ftrace. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 28 ++++++++++++++++++++++------ 1 file changed, 22 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0a367362ba2f..290e9da7aa9a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -92,9 +92,16 @@ void trace_wake_up(void) static int __init set_nr_entries(char *str) { + unsigned long nr_entries; + int ret; + if (!str) return 0; - trace_nr_entries = simple_strtoul(str, &str, 0); + ret = strict_strtoul(str, 0, &nr_entries); + /* nr_entries can not be zero */ + if (ret < 0 || nr_entries == 0) + return 0; + trace_nr_entries = nr_entries; return 1; } __setup("trace_entries=", set_nr_entries); @@ -1995,8 +2002,9 @@ tracing_ctrl_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { struct trace_array *tr = filp->private_data; - long val; char buf[64]; + long val; + int ret; if (cnt >= sizeof(buf)) return -EINVAL; @@ -2006,7 +2014,9 @@ tracing_ctrl_write(struct file *filp, const char __user *ubuf, buf[cnt] = 0; - val = simple_strtoul(buf, NULL, 10); + ret = strict_strtoul(buf, 10, &val); + if (ret < 0) + return ret; val = !!val; @@ -2110,8 +2120,9 @@ tracing_max_lat_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { long *ptr = filp->private_data; - long val; char buf[64]; + long val; + int ret; if (cnt >= sizeof(buf)) return -EINVAL; @@ -2121,7 +2132,9 @@ tracing_max_lat_write(struct file *filp, const char __user *ubuf, buf[cnt] = 0; - val = simple_strtoul(buf, NULL, 10); + ret = strict_strtoul(buf, 10, &val); + if (ret < 0) + return ret; *ptr = val * 1000; @@ -2376,6 +2389,7 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, { unsigned long val; char buf[64]; + int ret; if (cnt >= sizeof(buf)) return -EINVAL; @@ -2385,7 +2399,9 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, buf[cnt] = 0; - val = simple_strtoul(buf, NULL, 10); + ret = strict_strtoul(buf, 10, &val); + if (ret < 0) + return ret; /* must have at least 1 entry */ if (!val) -- cgit v1.2.3 From ab46428c6969d50ecf6f6e97b7a84abba6274368 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: modulize the number of CPU buffers Currently ftrace allocates a trace buffer for every possible CPU. Work is being done to change it to only online CPUs and add hooks to hotplug CPUS. This patch lays out the infrastructure for such a change. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 38 ++++++++++++++++++++++++-------------- 1 file changed, 24 insertions(+), 14 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 290e9da7aa9a..5da391c5fb0d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -35,6 +35,12 @@ unsigned long __read_mostly tracing_max_latency = (cycle_t)ULONG_MAX; unsigned long __read_mostly tracing_thresh; +static unsigned long __read_mostly tracing_nr_buffers; +static cpumask_t __read_mostly tracing_buffer_mask; + +#define for_each_tracing_cpu(cpu) \ + for_each_cpu_mask(cpu, tracing_buffer_mask) + /* dummy trace to disable tracing */ static struct tracer no_tracer __read_mostly = { .name = "none", @@ -328,7 +334,7 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) WARN_ON_ONCE(!irqs_disabled()); __raw_spin_lock(&ftrace_max_lock); /* clear out all the previous traces */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = tr->data[i]; flip_trace(max_tr.data[i], data); tracing_reset(data); @@ -352,7 +358,7 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) WARN_ON_ONCE(!irqs_disabled()); __raw_spin_lock(&ftrace_max_lock); - for_each_possible_cpu(i) + for_each_tracing_cpu(i) tracing_reset(max_tr.data[i]); flip_trace(max_tr.data[cpu], data); @@ -398,7 +404,7 @@ int register_tracer(struct tracer *type) * internal tracing to verify that everything is in order. * If we fail, we do not register this tracer. */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = tr->data[i]; if (!head_page(data)) continue; @@ -417,7 +423,7 @@ int register_tracer(struct tracer *type) goto out; } /* Only reset on passing, to avoid touching corrupted buffers */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = tr->data[i]; if (!head_page(data)) continue; @@ -847,7 +853,7 @@ find_next_entry(struct trace_iterator *iter, int *ent_cpu) int next_cpu = -1; int cpu; - for_each_possible_cpu(cpu) { + for_each_tracing_cpu(cpu) { if (!head_page(tr->data[cpu])) continue; ent = trace_entry_idx(tr, tr->data[cpu], iter, cpu); @@ -972,7 +978,7 @@ static void *s_start(struct seq_file *m, loff_t *pos) iter->prev_ent = NULL; iter->prev_cpu = -1; - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { iter->next_idx[i] = 0; iter->next_page[i] = NULL; } @@ -1089,7 +1095,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) if (type) name = type->name; - for_each_possible_cpu(cpu) { + for_each_tracing_cpu(cpu) { if (head_page(tr->data[cpu])) { total += tr->data[cpu]->trace_idx; if (tr->data[cpu]->trace_idx > tr->entries) @@ -1519,7 +1525,7 @@ static int trace_empty(struct trace_iterator *iter) struct trace_array_cpu *data; int cpu; - for_each_possible_cpu(cpu) { + for_each_tracing_cpu(cpu) { data = iter->tr->data[cpu]; if (head_page(data) && data->trace_idx && @@ -1831,7 +1837,7 @@ tracing_cpumask_write(struct file *filp, const char __user *ubuf, raw_local_irq_disable(); __raw_spin_lock(&ftrace_max_lock); - for_each_possible_cpu(cpu) { + for_each_tracing_cpu(cpu) { /* * Increase/decrease the disabled counter if we are * about to flip a bit in the cpumask: @@ -2308,7 +2314,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, ftrace_enabled = 0; #endif smp_wmb(); - for_each_possible_cpu(cpu) { + for_each_tracing_cpu(cpu) { data = iter->tr->data[cpu]; if (!head_page(data) || !data->trace_idx) @@ -2605,7 +2611,7 @@ static int trace_alloc_page(void) int i; /* first allocate a page for each CPU */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { array = (void *)__get_free_page(GFP_KERNEL); if (array == NULL) { printk(KERN_ERR "tracer: failed to allocate page" @@ -2630,7 +2636,7 @@ static int trace_alloc_page(void) } /* Now that we successfully allocate a page per CPU, add them */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = global_trace.data[i]; page = list_entry(pages.next, struct page, lru); list_del_init(&page->lru); @@ -2666,7 +2672,7 @@ static int trace_free_page(void) int ret = 0; /* free one page from each buffer */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = global_trace.data[i]; p = data->trace_pages.next; if (p == &data->trace_pages) { @@ -2717,8 +2723,12 @@ __init static int tracer_alloc_buffers(void) global_trace.ctrl = tracer_enabled; + /* TODO: make the number of buffers hot pluggable with CPUS */ + tracing_nr_buffers = num_possible_cpus(); + tracing_buffer_mask = cpu_possible_map; + /* Allocate the first page for all buffers */ - for_each_possible_cpu(i) { + for_each_tracing_cpu(i) { data = global_trace.data[i] = &per_cpu(global_trace_cpu, i); max_tr.data[i] = &per_cpu(max_data, i); -- cgit v1.2.3 From 4fcdae83cebda24b519a89d3dd976081fff1ca80 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: comment code This is first installment of adding documentation to the ftrace. Expect many more patches of this kind in the near future. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 135 ++++++++++++++++++++++++++++++++++++++++++++++++++- kernel/trace/trace.h | 7 +++ 2 files changed, 141 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 5da391c5fb0d..a102b11eacf2 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -64,26 +64,79 @@ cycle_t ftrace_now(int cpu) return cpu_clock(cpu); } +/* + * The global_trace is the descriptor that holds the tracing + * buffers for the live tracing. For each CPU, it contains + * a link list of pages that will store trace entries. The + * page descriptor of the pages in the memory is used to hold + * the link list by linking the lru item in the page descriptor + * to each of the pages in the buffer per CPU. + * + * For each active CPU there is a data field that holds the + * pages for the buffer for that CPU. Each CPU has the same number + * of pages allocated for its buffer. + */ static struct trace_array global_trace; static DEFINE_PER_CPU(struct trace_array_cpu, global_trace_cpu); +/* + * The max_tr is used to snapshot the global_trace when a maximum + * latency is reached. Some tracers will use this to store a maximum + * trace while it continues examining live traces. + * + * The buffers for the max_tr are set up the same as the global_trace. + * When a snapshot is taken, the link list of the max_tr is swapped + * with the link list of the global_trace and the buffers are reset for + * the global_trace so the tracing can continue. + */ static struct trace_array max_tr; static DEFINE_PER_CPU(struct trace_array_cpu, max_data); +/* tracer_enabled is used to toggle activation of a tracer */ static int tracer_enabled = 1; + +/* + * trace_nr_entries is the number of entries that is allocated + * for a buffer. Note, the number of entries is always rounded + * to ENTRIES_PER_PAGE. + */ static unsigned long trace_nr_entries = 65536UL; +/* trace_types holds a link list of available tracers. */ static struct tracer *trace_types __read_mostly; + +/* current_trace points to the tracer that is currently active */ static struct tracer *current_trace __read_mostly; + +/* + * max_tracer_type_len is used to simplify the allocating of + * buffers to read userspace tracer names. We keep track of + * the longest tracer name registered. + */ static int max_tracer_type_len; +/* + * trace_types_lock is used to protect the trace_types list. + * This lock is also used to keep user access serialized. + * Accesses from userspace will grab this lock while userspace + * activities happen inside the kernel. + */ static DEFINE_MUTEX(trace_types_lock); + +/* trace_wait is a waitqueue for tasks blocked on trace_poll */ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); +/* trace_flags holds iter_ctrl options */ unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; +/** + * trace_wake_up - wake up tasks waiting for trace input + * + * Simply wakes up any task that is blocked on the trace_wait + * queue. These is used with trace_poll for tasks polling the trace. + */ void trace_wake_up(void) { /* @@ -117,6 +170,14 @@ unsigned long nsecs_to_usecs(unsigned long nsecs) return nsecs / 1000; } +/* + * trace_flag_type is an enumeration that holds different + * states when a trace occurs. These are: + * IRQS_OFF - interrupts were disabled + * NEED_RESCED - reschedule is requested + * HARDIRQ - inside an interrupt handler + * SOFTIRQ - inside a softirq handler + */ enum trace_flag_type { TRACE_FLAG_IRQS_OFF = 0x01, TRACE_FLAG_NEED_RESCHED = 0x02, @@ -124,10 +185,14 @@ enum trace_flag_type { TRACE_FLAG_SOFTIRQ = 0x08, }; +/* + * TRACE_ITER_SYM_MASK masks the options in trace_flags that + * control the output of kernel symbols. + */ #define TRACE_ITER_SYM_MASK \ (TRACE_ITER_PRINT_PARENT|TRACE_ITER_SYM_OFFSET|TRACE_ITER_SYM_ADDR) -/* These must match the bit postions above */ +/* These must match the bit postions in trace_iterator_flags */ static const char *trace_options[] = { "print-parent", "sym-offset", @@ -142,6 +207,15 @@ static const char *trace_options[] = { NULL }; +/* + * ftrace_max_lock is used to protect the swapping of buffers + * when taking a max snapshot. The buffers themselves are + * protected by per_cpu spinlocks. But the action of the swap + * needs its own lock. + * + * This is defined as a raw_spinlock_t in order to help + * with performance when lockdep debugging is enabled. + */ static raw_spinlock_t ftrace_max_lock = (raw_spinlock_t)__RAW_SPIN_LOCK_UNLOCKED; @@ -172,6 +246,13 @@ __update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) tracing_record_cmdline(current); } +/** + * check_pages - integrity check of trace buffers + * + * As a safty measure we check to make sure the data pages have not + * been corrupted. TODO: configure to disable this because it adds + * a bit of overhead. + */ void check_pages(struct trace_array_cpu *data) { struct page *page, *tmp; @@ -185,6 +266,13 @@ void check_pages(struct trace_array_cpu *data) } } +/** + * head_page - page address of the first page in per_cpu buffer. + * + * head_page returns the page address of the first page in + * a per_cpu buffer. This also preforms various consistency + * checks to make sure the buffer has not been corrupted. + */ void *head_page(struct trace_array_cpu *data) { struct page *page; @@ -199,6 +287,17 @@ void *head_page(struct trace_array_cpu *data) return page_address(page); } +/** + * trace_seq_printf - sequence printing of trace information + * @s: trace sequence descriptor + * @fmt: printf format string + * + * The tracer may use either sequence operations or its own + * copy to user routines. To simplify formating of a trace + * trace_seq_printf is used to store strings into a special + * buffer (@s). Then the output may be either used by + * the sequencer or pulled into another buffer. + */ int trace_seq_printf(struct trace_seq *s, const char *fmt, ...) { @@ -222,6 +321,16 @@ trace_seq_printf(struct trace_seq *s, const char *fmt, ...) return len; } +/** + * trace_seq_puts - trace sequence printing of simple string + * @s: trace sequence descriptor + * @str: simple string to record + * + * The tracer may use either the sequence operations or its own + * copy to user routines. This function records a simple string + * into a special buffer (@s) for later retrieval by a sequencer + * or other mechanism. + */ static int trace_seq_puts(struct trace_seq *s, const char *str) { @@ -304,6 +413,13 @@ trace_print_seq(struct seq_file *m, struct trace_seq *s) trace_seq_reset(s); } +/* + * flip the trace buffers between two trace descriptors. + * This usually is the buffers between the global_trace and + * the max_tr to record a snapshot of a current trace. + * + * The ftrace_max_lock must be held. + */ static void flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) { @@ -325,6 +441,15 @@ flip_trace(struct trace_array_cpu *tr1, struct trace_array_cpu *tr2) check_pages(tr2); } +/** + * update_max_tr - snapshot all trace buffers from global_trace to max_tr + * @tr: tracer + * @tsk: the task with the latency + * @cpu: The cpu that initiated the trace. + * + * Flip the buffers between the @tr and the max_tr and record information + * about which task was the cause of this latency. + */ void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) { @@ -349,6 +474,8 @@ update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) * @tr - tracer * @tsk - task with the latency * @cpu - the cpu of the buffer to copy. + * + * Flip the trace of a single CPU buffer between the @tr and the max_tr. */ void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) @@ -368,6 +495,12 @@ update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu) __raw_spin_unlock(&ftrace_max_lock); } +/** + * register_tracer - register a tracer with the ftrace system. + * @type - the plugin for the tracer + * + * Register a new plugin tracer. + */ int register_tracer(struct tracer *type) { struct tracer *t; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index b0ca7473671b..21c29ee13e53 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -294,6 +294,13 @@ extern long ns2usecs(cycle_t nsec); extern unsigned long trace_flags; +/* + * trace_iterator_flags is an enumeration that defines bit + * positions into trace_flags that controls the output. + * + * NOTE: These bits must match the trace_options array in + * trace.c. + */ enum trace_iterator_flags { TRACE_ITER_PRINT_PARENT = 0x01, TRACE_ITER_SYM_OFFSET = 0x02, -- cgit v1.2.3 From 25b0b44a1c732ccfc58095cdd8438955a0a19fff Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:00 +0200 Subject: ftrace: fix comm on function trace output In cleaning up of the sched_switch code, the function trace recording of task comms was removed. This patch adds back the recording of comms for function trace. The output of ftrace now has the task comm instead of <...>. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 7 ++++++- kernel/trace/trace.h | 2 ++ kernel/trace/trace_functions.c | 2 ++ kernel/trace/trace_sched_switch.c | 7 +++++-- 4 files changed, 15 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index a102b11eacf2..1281969103b8 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -620,7 +620,12 @@ static unsigned map_cmdline_to_pid[SAVED_CMDLINES]; static char saved_cmdlines[SAVED_CMDLINES][TASK_COMM_LEN]; static int cmdline_idx; static DEFINE_SPINLOCK(trace_cmdline_lock); -atomic_t trace_record_cmdline_disabled; + +/* trace in all context switches */ +atomic_t trace_record_cmdline_enabled __read_mostly; + +/* temporary disable recording */ +atomic_t trace_record_cmdline_disabled __read_mostly; static void trace_init_cmdlines(void) { diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 21c29ee13e53..8991c5efcc74 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -216,6 +216,8 @@ extern unsigned long nsecs_to_usecs(unsigned long nsecs); extern unsigned long tracing_max_latency; extern unsigned long tracing_thresh; +extern atomic_t trace_record_cmdline_enabled; + void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu); diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 4165d34bd28a..0a084656d7cf 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -29,12 +29,14 @@ static void function_reset(struct trace_array *tr) static void start_function_trace(struct trace_array *tr) { function_reset(tr); + atomic_inc(&trace_record_cmdline_enabled); tracing_start_function_trace(); } static void stop_function_trace(struct trace_array *tr) { tracing_stop_function_trace(); + atomic_dec(&trace_record_cmdline_enabled); } static void function_trace_init(struct trace_array *tr) diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 5671db0e1827..a3376478fc2c 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -29,8 +29,6 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) if (!tracer_enabled) return; - tracing_record_cmdline(prev); - local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; @@ -73,6 +71,9 @@ void ftrace_ctx_switch(void *__rq, struct task_struct *prev, struct task_struct *next) { + if (unlikely(atomic_read(&trace_record_cmdline_enabled))) + tracing_record_cmdline(prev); + /* * If tracer_switch_func only points to the local * switch func, it still needs the ptr passed to it. @@ -134,11 +135,13 @@ static void sched_switch_reset(struct trace_array *tr) static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); + atomic_inc(&trace_record_cmdline_enabled); tracer_enabled = 1; } static void stop_sched_trace(struct trace_array *tr) { + atomic_dec(&trace_record_cmdline_enabled); tracer_enabled = 0; } -- cgit v1.2.3 From 53d0aa773053ab18287781e25d52c5faf9e0e09e Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:01 +0200 Subject: ftrace: add logic to record overruns This patch sets up the infrastructure to record overruns of the tracing buffer. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 16 +++++++++++----- kernel/trace/trace.h | 6 +++++- 2 files changed, 16 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 1281969103b8..b9126ef46a9e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -609,6 +609,7 @@ void unregister_tracer(struct tracer *type) void tracing_reset(struct trace_array_cpu *data) { data->trace_idx = 0; + data->overrun = 0; data->trace_head = data->trace_tail = head_page(data); data->trace_head_idx = 0; data->trace_tail_idx = 0; @@ -750,6 +751,7 @@ tracing_get_trace_entry(struct trace_array *tr, struct trace_array_cpu *data) if (data->trace_head == data->trace_tail && idx_next == data->trace_tail_idx) { /* overrun */ + data->overrun++; data->trace_tail_idx++; if (data->trace_tail_idx >= ENTRIES_PER_PAGE) { data->trace_tail = @@ -2353,8 +2355,6 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, { struct trace_iterator *iter = filp->private_data; struct trace_array_cpu *data; - struct trace_array *tr = iter->tr; - struct tracer *tracer = iter->trace; static cpumask_t mask; static int start; unsigned long flags; @@ -2433,10 +2433,11 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, if (cnt >= PAGE_SIZE) cnt = PAGE_SIZE - 1; - memset(iter, 0, sizeof(*iter)); - iter->tr = tr; - iter->trace = tracer; + /* reset all but tr, trace, and overruns */ iter->pos = -1; + memset(&iter->seq, 0, + sizeof(struct trace_iterator) - + offsetof(struct trace_iterator, seq)); /* * We need to stop all tracing on all CPUS to read the @@ -2465,6 +2466,11 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, for_each_cpu_mask(cpu, mask) { data = iter->tr->data[cpu]; __raw_spin_lock(&data->lock); + + if (data->overrun > iter->last_overrun[cpu]) + iter->overrun[cpu] += + data->overrun - iter->last_overrun[cpu]; + iter->last_overrun[cpu] = data->overrun; } while (find_next_entry_inc(iter) != NULL) { diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 8991c5efcc74..c1ec134ac356 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -97,6 +97,7 @@ struct trace_array_cpu { void *trace_head; /* producer */ void *trace_tail; /* consumer */ unsigned long trace_idx; + unsigned long overrun; unsigned long saved_latency; unsigned long critical_start; unsigned long critical_end; @@ -157,10 +158,13 @@ struct trace_seq { * results to users and which routines might sleep, etc: */ struct trace_iterator { - struct trace_seq seq; struct trace_array *tr; struct tracer *trace; + long last_overrun[NR_CPUS]; + long overrun[NR_CPUS]; + /* The below is zeroed out in pipe_read */ + struct trace_seq seq; struct trace_entry *ent; int cpu; -- cgit v1.2.3 From 107bad8bef5ab2c3a3bff7648c18c9dc3abdc13b Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:01 +0200 Subject: ftrace: add trace pipe header pluggin This patch adds a method for open_pipe and open_read to the pluggins so that they can add a header to the trace pipe call. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 38 +++++++++++++++++++++++++++++++------- kernel/trace/trace.h | 5 +++++ 2 files changed, 36 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index b9126ef46a9e..32f9106d612c 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2307,11 +2307,15 @@ static int tracing_open_pipe(struct inode *inode, struct file *filp) if (!iter) return -ENOMEM; + mutex_lock(&trace_types_lock); iter->tr = &global_trace; iter->trace = current_trace; - filp->private_data = iter; + if (iter->trace->pipe_open) + iter->trace->pipe_open(iter); + mutex_unlock(&trace_types_lock); + return 0; } @@ -2380,13 +2384,24 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, return cnt; } + mutex_lock(&trace_types_lock); + if (iter->trace->read) { + ret = iter->trace->read(iter, filp, ubuf, cnt, ppos); + if (ret) { + read = ret; + goto out; + } + } + trace_seq_reset(&iter->seq); start = 0; while (trace_empty(iter)) { - if ((filp->f_flags & O_NONBLOCK)) - return -EAGAIN; + if ((filp->f_flags & O_NONBLOCK)) { + read = -EAGAIN; + goto out; + } /* * This is a make-shift waitqueue. The reason we don't use @@ -2400,16 +2415,22 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, set_current_state(TASK_INTERRUPTIBLE); iter->tr->waiter = current; + mutex_unlock(&trace_types_lock); + /* sleep for one second, and try again. */ schedule_timeout(HZ); + mutex_lock(&trace_types_lock); + iter->tr->waiter = NULL; - if (signal_pending(current)) - return -EINTR; + if (signal_pending(current)) { + read = -EINTR; + goto out; + } if (iter->trace != current_trace) - return 0; + goto out; /* * We block until we read something and tracing is disabled. @@ -2428,7 +2449,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, /* stop when tracing is finished */ if (trace_empty(iter)) - return 0; + goto out; if (cnt >= PAGE_SIZE) cnt = PAGE_SIZE - 1; @@ -2518,6 +2539,9 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, if (ret) read = -EFAULT; +out: + mutex_unlock(&trace_types_lock); + return read; } diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index c1ec134ac356..ee53d706066f 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -135,9 +135,13 @@ struct tracer { void (*init)(struct trace_array *tr); void (*reset)(struct trace_array *tr); void (*open)(struct trace_iterator *iter); + void (*pipe_open)(struct trace_iterator *iter); void (*close)(struct trace_iterator *iter); void (*start)(struct trace_iterator *iter); void (*stop)(struct trace_iterator *iter); + ssize_t (*read)(struct trace_iterator *iter, + struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos); void (*ctrl_update)(struct trace_array *tr); #ifdef CONFIG_FTRACE_STARTUP_TEST int (*selftest)(struct tracer *trace, @@ -160,6 +164,7 @@ struct trace_seq { struct trace_iterator { struct trace_array *tr; struct tracer *trace; + void *private; long last_overrun[NR_CPUS]; long overrun[NR_CPUS]; -- cgit v1.2.3 From 4823ed7eadf35e4b57ce581327e21d39585f1f32 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:01 +0200 Subject: ftrace: fix setting of pos in read_pipe In resetting the iterator in read_pipe, the reset of pos was postitioned in the wrong location with respect to the memset operation. The current code sets pos, incorrectly, to zero. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 32f9106d612c..49e16630628a 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2455,10 +2455,10 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, cnt = PAGE_SIZE - 1; /* reset all but tr, trace, and overruns */ - iter->pos = -1; memset(&iter->seq, 0, sizeof(struct trace_iterator) - offsetof(struct trace_iterator, seq)); + iter->pos = -1; /* * We need to stop all tracing on all CPUS to read the -- cgit v1.2.3 From 9fe068e92f6290e89e19adc521441661a1229f00 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:02 +0200 Subject: ftrace: trace faster Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 49e16630628a..ca0d6ff74c11 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2417,8 +2417,8 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, mutex_unlock(&trace_types_lock); - /* sleep for one second, and try again. */ - schedule_timeout(HZ); + /* sleep for 100 msecs, and try again. */ + schedule_timeout(HZ/10); mutex_lock(&trace_types_lock); -- cgit v1.2.3 From a4feb8348b62fe76a63cdb5569f5c920f5283c06 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:02 +0200 Subject: ftrace: special stacktrace Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 47 ++++++++++++++++++++++++----------------------- 1 file changed, 24 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index ca0d6ff74c11..c232d8248a09 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -808,29 +808,6 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } -void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ - struct trace_array_cpu *data = __data; - struct trace_array *tr = __tr; - struct trace_entry *entry; - unsigned long irq_flags; - - raw_local_irq_save(irq_flags); - __raw_spin_lock(&data->lock); - entry = tracing_get_trace_entry(tr, data); - tracing_generic_entry_update(entry, 0); - entry->type = TRACE_SPECIAL; - entry->special.arg1 = arg1; - entry->special.arg2 = arg2; - entry->special.arg3 = arg3; - __raw_spin_unlock(&data->lock); - raw_local_irq_restore(irq_flags); - - trace_wake_up(); -} - void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, @@ -856,6 +833,30 @@ void __trace_stack(struct trace_array *tr, save_stack_trace(&trace); } +void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ + struct trace_array_cpu *data = __data; + struct trace_array *tr = __tr; + struct trace_entry *entry; + unsigned long irq_flags; + + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_SPECIAL; + entry->special.arg1 = arg1; + entry->special.arg2 = arg2; + entry->special.arg3 = arg3; + __trace_stack(tr, data, irq_flags, 4); + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); + + trace_wake_up(); +} + void tracing_sched_switch_trace(struct trace_array *tr, struct trace_array_cpu *data, -- cgit v1.2.3 From 2bb6f8d6389cbfadd657e7dc069f6986abf35e4f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:02 +0200 Subject: ftrace: use raw_smp_processor_id for mcount functions Due to debug hooks in the kernel that can change the way smp_processor_id works, use raw_smp_processor_id in mcount called functions (namely ftrace_record_ip). Currently we annotate most debug functions from calling mcount, but we should not rely on that to prevent kernel lockups. This patch uses the raw_smp_processor_id to prevent a recusive crash that can happen if a debug hook in smp_processor_id calls mcount. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 40f64f7cd850..af5ad8949abb 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -267,6 +267,7 @@ ftrace_record_ip(unsigned long ip) unsigned long key; int resched; int atomic; + int cpu; if (!ftrace_enabled || ftrace_disabled) return; @@ -274,9 +275,15 @@ ftrace_record_ip(unsigned long ip) resched = need_resched(); preempt_disable_notrace(); - /* We simply need to protect against recursion */ - __get_cpu_var(ftrace_shutdown_disable_cpu)++; - if (__get_cpu_var(ftrace_shutdown_disable_cpu) != 1) + /* + * We simply need to protect against recursion. + * Use the the raw version of smp_processor_id and not + * __get_cpu_var which can call debug hooks that can + * cause a recursive crash here. + */ + cpu = raw_smp_processor_id(); + per_cpu(ftrace_shutdown_disable_cpu, cpu)++; + if (per_cpu(ftrace_shutdown_disable_cpu, cpu) != 1) goto out; if (unlikely(ftrace_record_suspend)) @@ -317,7 +324,7 @@ ftrace_record_ip(unsigned long ip) out_unlock: spin_unlock_irqrestore(&ftrace_shutdown_lock, flags); out: - __get_cpu_var(ftrace_shutdown_disable_cpu)--; + per_cpu(ftrace_shutdown_disable_cpu, cpu)--; /* prevent recursion with scheduler */ if (resched) -- cgit v1.2.3 From 6c6c27969a4c6024e6c8838829546c02aaddca18 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: ftrace: add readpos to struct trace_seq; add trace_seq_to_user() Refactor code from tracing_read_pipe() and create trace_seq_to_user(). Moved trace_seq_reset() call before iter->trace->read() call so that when all leftover data is returned, trace_seq is reset automatically. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 73 ++++++++++++++++++++++++++-------------------------- kernel/trace/trace.h | 3 +++ 2 files changed, 39 insertions(+), 37 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index c232d8248a09..82ced406aacf 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -400,6 +400,26 @@ static void trace_seq_reset(struct trace_seq *s) { s->len = 0; + s->readpos = 0; +} + +ssize_t trace_seq_to_user(struct trace_seq *s, char __user *ubuf, size_t cnt) +{ + int len; + int ret; + + if (s->len <= s->readpos) + return -EBUSY; + + len = s->len - s->readpos; + if (cnt > len) + cnt = len; + ret = copy_to_user(ubuf, s->buffer + s->readpos, cnt); + if (ret) + return -EFAULT; + + s->readpos += len; + return cnt; } static void @@ -2361,46 +2381,32 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, struct trace_iterator *iter = filp->private_data; struct trace_array_cpu *data; static cpumask_t mask; - static int start; unsigned long flags; #ifdef CONFIG_FTRACE int ftrace_save; #endif - int read = 0; int cpu; - int len; - int ret; + ssize_t sret; /* return any leftover data */ - if (iter->seq.len > start) { - len = iter->seq.len - start; - if (cnt > len) - cnt = len; - ret = copy_to_user(ubuf, iter->seq.buffer + start, cnt); - if (ret) - cnt = -EFAULT; - - start += len; + sret = trace_seq_to_user(&iter->seq, ubuf, cnt); + if (sret != -EBUSY) + return sret; + sret = 0; - return cnt; - } + trace_seq_reset(&iter->seq); mutex_lock(&trace_types_lock); if (iter->trace->read) { - ret = iter->trace->read(iter, filp, ubuf, cnt, ppos); - if (ret) { - read = ret; + sret = iter->trace->read(iter, filp, ubuf, cnt, ppos); + if (sret) goto out; - } } - trace_seq_reset(&iter->seq); - start = 0; - while (trace_empty(iter)) { if ((filp->f_flags & O_NONBLOCK)) { - read = -EAGAIN; + sret = -EAGAIN; goto out; } @@ -2426,7 +2432,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, iter->tr->waiter = NULL; if (signal_pending(current)) { - read = -EINTR; + sret = -EINTR; goto out; } @@ -2496,6 +2502,7 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, } while (find_next_entry_inc(iter) != NULL) { + int ret; int len = iter->seq.len; ret = print_trace_line(iter); @@ -2526,24 +2533,16 @@ tracing_read_pipe(struct file *filp, char __user *ubuf, local_irq_restore(flags); /* Now copy what we have to the user */ - read = iter->seq.len; - if (read > cnt) - read = cnt; - - ret = copy_to_user(ubuf, iter->seq.buffer, read); - - if (read < iter->seq.len) - start = read; - else + sret = trace_seq_to_user(&iter->seq, ubuf, cnt); + if (iter->seq.readpos >= iter->seq.len) trace_seq_reset(&iter->seq); - - if (ret) - read = -EFAULT; + if (sret == -EBUSY) + sret = 0; out: mutex_unlock(&trace_types_lock); - return read; + return sret; } static ssize_t diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index ee53d706066f..8845033ab49d 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -155,6 +155,7 @@ struct tracer { struct trace_seq { unsigned char buffer[PAGE_SIZE]; unsigned int len; + unsigned int readpos; }; /* @@ -301,6 +302,8 @@ extern int trace_selftest_startup_sched_switch(struct tracer *trace, extern void *head_page(struct trace_array_cpu *data); extern int trace_seq_printf(struct trace_seq *s, const char *fmt, ...); +extern ssize_t trace_seq_to_user(struct trace_seq *s, char __user *ubuf, + size_t cnt); extern long ns2usecs(cycle_t nsec); extern unsigned long trace_flags; -- cgit v1.2.3 From 3eefae994d9224fb7771a3ddb683868363c23510 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Mon, 12 May 2008 21:21:04 +0200 Subject: ftrace: limit trace entries Currently there is no protection from the root user to use up all of memory for trace buffers. If the root user allocates too many entries, the OOM killer might start kill off all tasks. This patch adds an algorith to check the following condition: pages_requested > (freeable_memory + current_trace_buffer_pages) / 4 If the above is met then the allocation fails. The above prevents more than 1/4th of freeable memory from being used by trace buffers. To determine the freeable_memory, I made determine_dirtyable_memory in mm/page-writeback.c global. Special thanks goes to Peter Zijlstra for suggesting the above calculation. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 38 ++++++++++++++++++++++++++++++++++++++ 1 file changed, 38 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 82ced406aacf..2824cf48cdca 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -27,6 +27,7 @@ #include #include #include +#include #include @@ -51,6 +52,8 @@ static int trace_free_page(void); static int tracing_disabled = 1; +static unsigned long tracing_pages_allocated; + long ns2usecs(cycle_t nsec) { @@ -2591,12 +2594,41 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, } if (val > global_trace.entries) { + long pages_requested; + unsigned long freeable_pages; + + /* make sure we have enough memory before mapping */ + pages_requested = + (val + (ENTRIES_PER_PAGE-1)) / ENTRIES_PER_PAGE; + + /* account for each buffer (and max_tr) */ + pages_requested *= tracing_nr_buffers * 2; + + /* Check for overflow */ + if (pages_requested < 0) { + cnt = -ENOMEM; + goto out; + } + + freeable_pages = determine_dirtyable_memory(); + + /* we only allow to request 1/4 of useable memory */ + if (pages_requested > + ((freeable_pages + tracing_pages_allocated) / 4)) { + cnt = -ENOMEM; + goto out; + } + while (global_trace.entries < val) { if (trace_alloc_page()) { cnt = -ENOMEM; goto out; } + /* double check that we don't go over the known pages */ + if (tracing_pages_allocated > pages_requested) + break; } + } else { /* include the number of entries in val (inc of page entries) */ while (global_trace.entries > val + (ENTRIES_PER_PAGE - 1)) @@ -2776,6 +2808,7 @@ static int trace_alloc_page(void) struct page *page, *tmp; LIST_HEAD(pages); void *array; + unsigned pages_allocated = 0; int i; /* first allocate a page for each CPU */ @@ -2787,6 +2820,7 @@ static int trace_alloc_page(void) goto free_pages; } + pages_allocated++; page = virt_to_page(array); list_add(&page->lru, &pages); @@ -2798,6 +2832,7 @@ static int trace_alloc_page(void) "for trace buffer!\n"); goto free_pages; } + pages_allocated++; page = virt_to_page(array); list_add(&page->lru, &pages); #endif @@ -2819,6 +2854,7 @@ static int trace_alloc_page(void) SetPageLRU(page); #endif } + tracing_pages_allocated += pages_allocated; global_trace.entries += ENTRIES_PER_PAGE; return 0; @@ -2853,6 +2889,8 @@ static int trace_free_page(void) page = list_entry(p, struct page, lru); ClearPageLRU(page); list_del(&page->lru); + tracing_pages_allocated--; + tracing_pages_allocated--; __free_page(page); tracing_reset(data); -- cgit v1.2.3 From dc102a8fae2d0d6bf5223fc549247f2e23959ae6 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 12 May 2008 21:21:09 +0200 Subject: Markers - remove extra format argument Denys Vlasenko : > Not in this patch, but I noticed: > > #define __trace_mark(name, call_private, format, args...) \ > do { \ > static const char __mstrtab_##name[] \ > __attribute__((section("__markers_strings"))) \ > = #name "\0" format; \ > static struct marker __mark_##name \ > __attribute__((section("__markers"), aligned(8))) = \ > { __mstrtab_##name, &__mstrtab_##name[sizeof(#name)], \ > 0, 0, marker_probe_cb, \ > { __mark_empty_function, NULL}, NULL }; \ > __mark_check_format(format, ## args); \ > if (unlikely(__mark_##name.state)) { \ > (*__mark_##name.call) \ > (&__mark_##name, call_private, \ > format, ## args); \ > } \ > } while (0) > > In this call: > > (*__mark_##name.call) \ > (&__mark_##name, call_private, \ > format, ## args); \ > > you make gcc allocate duplicate format string. You can use > &__mstrtab_##name[sizeof(#name)] instead since it holds the same string, > or drop ", format," above and "const char *fmt" from here: > > void (*call)(const struct marker *mdata, /* Probe wrapper */ > void *call_private, const char *fmt, ...); > > since mdata->format is the same and all callees which need it can take it there. Very good point. I actually thought about dropping it, since it would remove an unnecessary argument from the stack. And actually, since I now have the marker_probe_cb sitting between the marker site and the callbacks, there is no API change required. Thanks :) Mathieu Signed-off-by: Mathieu Desnoyers CC: Denys Vlasenko Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/marker.c | 30 ++++++++++++++---------------- 1 file changed, 14 insertions(+), 16 deletions(-) (limited to 'kernel') diff --git a/kernel/marker.c b/kernel/marker.c index b5a9fe1d50d5..1abfb923b761 100644 --- a/kernel/marker.c +++ b/kernel/marker.c @@ -55,8 +55,8 @@ static DEFINE_MUTEX(markers_mutex); struct marker_entry { struct hlist_node hlist; char *format; - void (*call)(const struct marker *mdata, /* Probe wrapper */ - void *call_private, const char *fmt, ...); + /* Probe wrapper */ + void (*call)(const struct marker *mdata, void *call_private, ...); struct marker_probe_closure single; struct marker_probe_closure *multi; int refcount; /* Number of times armed. 0 if disarmed. */ @@ -91,15 +91,13 @@ EXPORT_SYMBOL_GPL(__mark_empty_function); * marker_probe_cb Callback that prepares the variable argument list for probes. * @mdata: pointer of type struct marker * @call_private: caller site private data - * @fmt: format string * @...: Variable argument list. * * Since we do not use "typical" pointer based RCU in the 1 argument case, we * need to put a full smp_rmb() in this branch. This is why we do not use * rcu_dereference() for the pointer read. */ -void marker_probe_cb(const struct marker *mdata, void *call_private, - const char *fmt, ...) +void marker_probe_cb(const struct marker *mdata, void *call_private, ...) { va_list args; char ptype; @@ -120,8 +118,9 @@ void marker_probe_cb(const struct marker *mdata, void *call_private, /* Must read the ptr before private data. They are not data * dependant, so we put an explicit smp_rmb() here. */ smp_rmb(); - va_start(args, fmt); - func(mdata->single.probe_private, call_private, fmt, &args); + va_start(args, call_private); + func(mdata->single.probe_private, call_private, mdata->format, + &args); va_end(args); } else { struct marker_probe_closure *multi; @@ -136,9 +135,9 @@ void marker_probe_cb(const struct marker *mdata, void *call_private, smp_read_barrier_depends(); multi = mdata->multi; for (i = 0; multi[i].func; i++) { - va_start(args, fmt); - multi[i].func(multi[i].probe_private, call_private, fmt, - &args); + va_start(args, call_private); + multi[i].func(multi[i].probe_private, call_private, + mdata->format, &args); va_end(args); } } @@ -150,13 +149,11 @@ EXPORT_SYMBOL_GPL(marker_probe_cb); * marker_probe_cb Callback that does not prepare the variable argument list. * @mdata: pointer of type struct marker * @call_private: caller site private data - * @fmt: format string * @...: Variable argument list. * * Should be connected to markers "MARK_NOARGS". */ -void marker_probe_cb_noarg(const struct marker *mdata, - void *call_private, const char *fmt, ...) +void marker_probe_cb_noarg(const struct marker *mdata, void *call_private, ...) { va_list args; /* not initialized */ char ptype; @@ -172,7 +169,8 @@ void marker_probe_cb_noarg(const struct marker *mdata, /* Must read the ptr before private data. They are not data * dependant, so we put an explicit smp_rmb() here. */ smp_rmb(); - func(mdata->single.probe_private, call_private, fmt, &args); + func(mdata->single.probe_private, call_private, mdata->format, + &args); } else { struct marker_probe_closure *multi; int i; @@ -186,8 +184,8 @@ void marker_probe_cb_noarg(const struct marker *mdata, smp_read_barrier_depends(); multi = mdata->multi; for (i = 0; multi[i].func; i++) - multi[i].func(multi[i].probe_private, call_private, fmt, - &args); + multi[i].func(multi[i].probe_private, call_private, + mdata->format, &args); } preempt_enable(); } -- cgit v1.2.3 From 5b82a1b08a00b2adca3d9dd9777efff40b7aaaa1 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 12 May 2008 21:21:10 +0200 Subject: Port ftrace to markers Porting ftrace to the marker infrastructure. Don't need to chain to the wakeup tracer from the sched tracer, because markers support multiple probes connected. Signed-off-by: Mathieu Desnoyers CC: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 14 +++- kernel/trace/trace.h | 20 +---- kernel/trace/trace_sched_switch.c | 171 +++++++++++++++++++++++++++++++------- kernel/trace/trace_sched_wakeup.c | 106 +++++++++++++++++++++-- 4 files changed, 255 insertions(+), 56 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index ad95cca4e42e..e2e985eeee78 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2500,7 +2500,9 @@ out_activate: success = 1; out_running: - ftrace_wake_up_task(rq, p, rq->curr); + trace_mark(kernel_sched_wakeup, + "pid %d state %ld ## rq %p task %p rq->curr %p", + p->pid, p->state, rq, p, rq->curr); check_preempt_curr(rq, p); p->state = TASK_RUNNING; @@ -2631,7 +2633,9 @@ void wake_up_new_task(struct task_struct *p, unsigned long clone_flags) p->sched_class->task_new(rq, p); inc_nr_running(rq); } - ftrace_wake_up_task(rq, p, rq->curr); + trace_mark(kernel_sched_wakeup_new, + "pid %d state %ld ## rq %p task %p rq->curr %p", + p->pid, p->state, rq, p, rq->curr); check_preempt_curr(rq, p); #ifdef CONFIG_SMP if (p->sched_class->task_wake_up) @@ -2804,7 +2808,11 @@ context_switch(struct rq *rq, struct task_struct *prev, struct mm_struct *mm, *oldmm; prepare_task_switch(rq, prev, next); - ftrace_ctx_switch(rq, prev, next); + trace_mark(kernel_sched_schedule, + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + prev->pid, next->pid, prev->state, + rq, prev, next); mm = next->mm; oldmm = prev->active_mm; /* diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 8845033ab49d..f5de0601b408 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -234,25 +234,10 @@ void update_max_tr_single(struct trace_array *tr, extern cycle_t ftrace_now(int cpu); -#ifdef CONFIG_SCHED_TRACER -extern void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next); -extern void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr); -#else -static inline void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) -{ -} -static inline void -wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) -{ -} -#endif - #ifdef CONFIG_CONTEXT_SWITCH_TRACER typedef void (*tracer_switch_func_t)(void *private, + void *__rq, struct task_struct *prev, struct task_struct *next); @@ -262,9 +247,6 @@ struct tracer_switch_ops { struct tracer_switch_ops *next; }; -extern int register_tracer_switch(struct tracer_switch_ops *ops); -extern int unregister_tracer_switch(struct tracer_switch_ops *ops); - #endif /* CONFIG_CONTEXT_SWITCH_TRACER */ #ifdef CONFIG_DYNAMIC_FTRACE diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index a3376478fc2c..d25ffa5eaf2b 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -16,11 +16,14 @@ static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; +static atomic_t sched_ref; static void -ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) +sched_switch_func(void *private, void *__rq, struct task_struct *prev, + struct task_struct *next) { - struct trace_array *tr = ctx_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; unsigned long flags; long disabled; @@ -41,10 +44,40 @@ ctx_switch_func(void *__rq, struct task_struct *prev, struct task_struct *next) local_irq_restore(flags); } +static notrace void +sched_switch_callback(void *probe_data, void *call_data, + const char *format, va_list *args) +{ + struct task_struct *prev; + struct task_struct *next; + struct rq *__rq; + + if (!atomic_read(&sched_ref)) + return; + + /* skip prev_pid %d next_pid %d prev_state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, int); + (void)va_arg(*args, long); + __rq = va_arg(*args, typeof(__rq)); + prev = va_arg(*args, typeof(prev)); + next = va_arg(*args, typeof(next)); + + tracing_record_cmdline(prev); + + /* + * If tracer_switch_func only points to the local + * switch func, it still needs the ptr passed to it. + */ + sched_switch_func(probe_data, __rq, prev, next); +} + static void -wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) +wakeup_func(void *private, void *__rq, struct task_struct *wakee, struct + task_struct *curr) { - struct trace_array *tr = ctx_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; unsigned long flags; long disabled; @@ -67,35 +100,29 @@ wakeup_func(void *__rq, struct task_struct *wakee, struct task_struct *curr) local_irq_restore(flags); } -void -ftrace_ctx_switch(void *__rq, struct task_struct *prev, - struct task_struct *next) +static notrace void +wake_up_callback(void *probe_data, void *call_data, + const char *format, va_list *args) { - if (unlikely(atomic_read(&trace_record_cmdline_enabled))) - tracing_record_cmdline(prev); + struct task_struct *curr; + struct task_struct *task; + struct rq *__rq; - /* - * If tracer_switch_func only points to the local - * switch func, it still needs the ptr passed to it. - */ - ctx_switch_func(__rq, prev, next); + if (likely(!tracer_enabled)) + return; - /* - * Chain to the wakeup tracer (this is a NOP if disabled): - */ - wakeup_sched_switch(prev, next); -} + /* Skip pid %d state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, long); + /* now get the meat: "rq %p task %p rq->curr %p" */ + __rq = va_arg(*args, typeof(__rq)); + task = va_arg(*args, typeof(task)); + curr = va_arg(*args, typeof(curr)); -void -ftrace_wake_up_task(void *__rq, struct task_struct *wakee, - struct task_struct *curr) -{ - wakeup_func(__rq, wakee, curr); + tracing_record_cmdline(task); + tracing_record_cmdline(curr); - /* - * Chain to the wakeup tracer (this is a NOP if disabled): - */ - wakeup_sched_wakeup(wakee, curr); + wakeup_func(probe_data, __rq, task, curr); } void @@ -132,15 +159,95 @@ static void sched_switch_reset(struct trace_array *tr) tracing_reset(tr->data[cpu]); } +static int tracing_sched_register(void) +{ + int ret; + + ret = marker_probe_register("kernel_sched_wakeup", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &ctx_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup\n"); + return ret; + } + + ret = marker_probe_register("kernel_sched_wakeup_new", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &ctx_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup_new\n"); + goto fail_deprobe; + } + + ret = marker_probe_register("kernel_sched_schedule", + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + sched_switch_callback, + &ctx_trace); + if (ret) { + pr_info("sched trace: Couldn't add marker" + " probe to kernel_sched_schedule\n"); + goto fail_deprobe_wake_new; + } + + return ret; +fail_deprobe_wake_new: + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &ctx_trace); +fail_deprobe: + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &ctx_trace); + return ret; +} + +static void tracing_sched_unregister(void) +{ + marker_probe_unregister("kernel_sched_schedule", + sched_switch_callback, + &ctx_trace); + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &ctx_trace); + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &ctx_trace); +} + +void tracing_start_sched_switch(void) +{ + long ref; + + ref = atomic_inc_return(&sched_ref); + if (ref == 1) + tracing_sched_register(); +} + +void tracing_stop_sched_switch(void) +{ + long ref; + + ref = atomic_dec_and_test(&sched_ref); + if (ref) + tracing_sched_unregister(); +} + static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); atomic_inc(&trace_record_cmdline_enabled); tracer_enabled = 1; + tracing_start_sched_switch(); } static void stop_sched_trace(struct trace_array *tr) { + tracing_stop_sched_switch(); atomic_dec(&trace_record_cmdline_enabled); tracer_enabled = 0; } @@ -181,6 +288,14 @@ static struct tracer sched_switch_trace __read_mostly = __init static int init_sched_switch_trace(void) { + int ret = 0; + + if (atomic_read(&sched_ref)) + ret = tracing_sched_register(); + if (ret) { + pr_info("error registering scheduler trace\n"); + return ret; + } return register_tracer(&sched_switch_trace); } device_initcall(init_sched_switch_trace); diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 5948011006bc..5d2fb48e47f8 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -15,6 +15,7 @@ #include #include #include +#include #include "trace.h" @@ -44,11 +45,13 @@ static int report_latency(cycle_t delta) return 1; } -void -wakeup_sched_switch(struct task_struct *prev, struct task_struct *next) +static void notrace +wakeup_sched_switch(void *private, void *rq, struct task_struct *prev, + struct task_struct *next) { unsigned long latency = 0, t0 = 0, t1 = 0; - struct trace_array *tr = wakeup_trace; + struct trace_array **ptr = private; + struct trace_array *tr = *ptr; struct trace_array_cpu *data; cycle_t T0, T1, delta; unsigned long flags; @@ -113,6 +116,31 @@ out: atomic_dec(&tr->data[cpu]->disabled); } +static notrace void +sched_switch_callback(void *probe_data, void *call_data, + const char *format, va_list *args) +{ + struct task_struct *prev; + struct task_struct *next; + struct rq *__rq; + + /* skip prev_pid %d next_pid %d prev_state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, int); + (void)va_arg(*args, long); + __rq = va_arg(*args, typeof(__rq)); + prev = va_arg(*args, typeof(prev)); + next = va_arg(*args, typeof(next)); + + tracing_record_cmdline(prev); + + /* + * If tracer_switch_func only points to the local + * switch func, it still needs the ptr passed to it. + */ + wakeup_sched_switch(probe_data, __rq, prev, next); +} + static void __wakeup_reset(struct trace_array *tr) { struct trace_array_cpu *data; @@ -188,19 +216,68 @@ out: atomic_dec(&tr->data[cpu]->disabled); } -void wakeup_sched_wakeup(struct task_struct *wakee, struct task_struct *curr) +static notrace void +wake_up_callback(void *probe_data, void *call_data, + const char *format, va_list *args) { + struct trace_array **ptr = probe_data; + struct trace_array *tr = *ptr; + struct task_struct *curr; + struct task_struct *task; + struct rq *__rq; + if (likely(!tracer_enabled)) return; + /* Skip pid %d state %ld */ + (void)va_arg(*args, int); + (void)va_arg(*args, long); + /* now get the meat: "rq %p task %p rq->curr %p" */ + __rq = va_arg(*args, typeof(__rq)); + task = va_arg(*args, typeof(task)); + curr = va_arg(*args, typeof(curr)); + + tracing_record_cmdline(task); tracing_record_cmdline(curr); - tracing_record_cmdline(wakee); - wakeup_check_start(wakeup_trace, wakee, curr); + wakeup_check_start(tr, task, curr); } static void start_wakeup_tracer(struct trace_array *tr) { + int ret; + + ret = marker_probe_register("kernel_sched_wakeup", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &wakeup_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup\n"); + return; + } + + ret = marker_probe_register("kernel_sched_wakeup_new", + "pid %d state %ld ## rq %p task %p rq->curr %p", + wake_up_callback, + &wakeup_trace); + if (ret) { + pr_info("wakeup trace: Couldn't add marker" + " probe to kernel_sched_wakeup_new\n"); + goto fail_deprobe; + } + + ret = marker_probe_register("kernel_sched_schedule", + "prev_pid %d next_pid %d prev_state %ld " + "## rq %p prev %p next %p", + sched_switch_callback, + &wakeup_trace); + if (ret) { + pr_info("sched trace: Couldn't add marker" + " probe to kernel_sched_schedule\n"); + goto fail_deprobe_wake_new; + } + wakeup_reset(tr); /* @@ -215,11 +292,28 @@ static void start_wakeup_tracer(struct trace_array *tr) tracer_enabled = 1; return; +fail_deprobe_wake_new: + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &wakeup_trace); +fail_deprobe: + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &wakeup_trace); } static void stop_wakeup_tracer(struct trace_array *tr) { tracer_enabled = 0; + marker_probe_unregister("kernel_sched_schedule", + sched_switch_callback, + &wakeup_trace); + marker_probe_unregister("kernel_sched_wakeup_new", + wake_up_callback, + &wakeup_trace); + marker_probe_unregister("kernel_sched_wakeup", + wake_up_callback, + &wakeup_trace); } static void wakeup_tracer_init(struct trace_array *tr) -- cgit v1.2.3 From 74f4e369fc5b52433ad824cef32d3bf1304549be Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:15 +0200 Subject: ftrace: stacktrace fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/semaphore.c | 2 ++ kernel/trace/trace.c | 4 ++-- kernel/trace/trace.h | 2 +- 3 files changed, 5 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/semaphore.c b/kernel/semaphore.c index 5c2942e768cd..1a064adab658 100644 --- a/kernel/semaphore.c +++ b/kernel/semaphore.c @@ -31,6 +31,7 @@ #include #include #include +#include static noinline void __down(struct semaphore *sem); static noinline int __down_interruptible(struct semaphore *sem); @@ -53,6 +54,7 @@ void down(struct semaphore *sem) { unsigned long flags; + ftrace_special(sem->count, 0, __LINE__); spin_lock_irqsave(&sem->lock, flags); if (likely(sem->count > 0)) sem->count--; diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 2824cf48cdca..3271916ff033 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -901,7 +901,7 @@ tracing_sched_switch_trace(struct trace_array *tr, entry->ctx.next_pid = next->pid; entry->ctx.next_prio = next->prio; entry->ctx.next_state = next->state; - __trace_stack(tr, data, flags, 4); + __trace_stack(tr, data, flags, 5); __raw_spin_unlock(&data->lock); raw_local_irq_restore(irq_flags); } @@ -927,7 +927,7 @@ tracing_sched_wakeup_trace(struct trace_array *tr, entry->ctx.next_pid = wakee->pid; entry->ctx.next_prio = wakee->prio; entry->ctx.next_state = wakee->state; - __trace_stack(tr, data, flags, 5); + __trace_stack(tr, data, flags, 6); __raw_spin_unlock(&data->lock); raw_local_irq_restore(irq_flags); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index f5de0601b408..c460e85e94ed 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -51,7 +51,7 @@ struct special_entry { * Stack-trace entry: */ -#define FTRACE_STACK_ENTRIES 5 +#define FTRACE_STACK_ENTRIES 8 struct stack_entry { unsigned long caller[FTRACE_STACK_ENTRIES]; -- cgit v1.2.3 From aa5e5ceaf52a882a29d9b86531a20733f5116066 Mon Sep 17 00:00:00 2001 From: David Miller Date: Tue, 13 May 2008 22:06:56 -0700 Subject: ftrace: remove packed attribute on ftrace_page. It causes unaligned access traps on platforms like sparc (ftrace_page may be marked packed, but once we return a dyn_ftrace sub-object from this array to another piece of code, the "packed" part of the typing information doesn't propagate). But also, it didn't serve any purpose either. Even if packed, on 64-bit or 32-bit, it didn't give us any more dyn_ftrace entries per-page. Signed-off-by: David S. Miller Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index af5ad8949abb..07b2a14943f8 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -177,9 +177,9 @@ static DEFINE_MUTEX(ftrace_filter_lock); struct ftrace_page { struct ftrace_page *next; - int index; + unsigned long index; struct dyn_ftrace records[]; -} __attribute__((packed)); +}; #define ENTRIES_PER_PAGE \ ((PAGE_SIZE - sizeof(struct ftrace_page)) / sizeof(struct dyn_ftrace)) -- cgit v1.2.3 From 37135677e653537ffc6e7def679443272a1c03c3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Wed, 14 May 2008 08:10:31 +0200 Subject: ftrace: fix mcount export bug David S. Miller noticed the following bug: the -pg instrumentation function callback is named differently on each platform. On x86 it is mcount, on sparc it is _mcount. So the export does not make sense in kernel/trace/ftrace.c - move it to x86. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 07b2a14943f8..a3e47f43f8a0 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -50,9 +50,6 @@ static struct ftrace_ops ftrace_list_end __read_mostly = static struct ftrace_ops *ftrace_list __read_mostly = &ftrace_list_end; ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub; -/* mcount is defined per arch in assembly */ -EXPORT_SYMBOL(mcount); - void ftrace_list_func(unsigned long ip, unsigned long parent_ip) { struct ftrace_ops *op = ftrace_list; -- cgit v1.2.3 From 2d8b820b2e81954754277723379ae9ed5de316fa Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sat, 23 Feb 2008 16:55:50 +0100 Subject: ftrace: cleanups factor out code and clean it up. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index a3e47f43f8a0..89bd9a6f52ec 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -18,13 +18,13 @@ #include #include #include -#include #include -#include +#include #include +#include #include -#include #include +#include #include #include "trace.h" -- cgit v1.2.3 From 4e491d14f2506b218d678935c25a7027b79178b1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 14 May 2008 23:49:44 -0400 Subject: ftrace: support for PowerPC This patch adds full support for ftrace for PowerPC (both 64 and 32 bit). This includes dynamic tracing and function filtering. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_selftest.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index a5f6001c3332..3877dd9102f1 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -123,6 +123,7 @@ int trace_selftest_startup_dynamic_tracing(struct tracer *trace, int ret; int save_ftrace_enabled = ftrace_enabled; int save_tracer_enabled = tracer_enabled; + char *func_name; /* The ftrace test PASSED */ printk(KERN_CONT "PASSED\n"); @@ -142,9 +143,15 @@ int trace_selftest_startup_dynamic_tracing(struct tracer *trace, return ret; } + /* + * Some archs *cough*PowerPC*cough* add charachters to the + * start of the function names. We simply put a '*' to + * accomodate them. + */ + func_name = "*" STR(DYN_FTRACE_TEST_NAME); + /* filter only on our function */ - ftrace_set_filter(STR(DYN_FTRACE_TEST_NAME), - sizeof(STR(DYN_FTRACE_TEST_NAME)), 1); + ftrace_set_filter(func_name, strlen(func_name), 1); /* enable tracing */ tr->ctrl = 1; -- cgit v1.2.3 From 6ec562328fda585be2d7f472cfac99d3b44d362a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Wed, 14 May 2008 21:30:30 -0400 Subject: ftrace: use the new kbuild CFLAGS_REMOVE for kernel directory This patch removes the Makefile turd and uses the nice CFLAGS_REMOVE macro in the kernel directory. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/Makefile | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index d2f80ea4cd9a..ca2433e84873 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -11,12 +11,16 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o profile.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o +CFLAGS_REMOVE_sched.o = -pg -mno-spe + ifdef CONFIG_FTRACE -# Do not profile debug utilities -ORIG_CFLAGS := $(KBUILD_CFLAGS) -KBUILD_CFLAGS = $(if $(filter-out lockdep% %debug,$(basename $(notdir $@))), \ - $(ORIG_CFLAGS), \ - $(subst -pg,,$(ORIG_CFLAGS))) +# Do not trace debug files and internal ftrace files +CFLAGS_REMOVE_lockdep.o = -pg +CFLAGS_REMOVE_lockdep_proc.o = -pg +CFLAGS_REMOVE_mutex-debug.o = -pg +CFLAGS_REMOVE_rtmutex-debug.o = -pg +CFLAGS_REMOVE_cgroup-debug.o = -pg +CFLAGS_REMOVE_sched_clock.o = -pg endif obj-$(CONFIG_SYSCTL_SYSCALL_CHECK) += sysctl_check.o -- cgit v1.2.3 From 677aa9f77e8de3791b481a0cec6c8b84d1eec626 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Sat, 17 May 2008 00:01:36 -0400 Subject: ftrace: add have dynamic ftrace config for archs Now that ftrace is being ported to other architectures, it has become apparent that DYNAMIC_FTRACE is dependent on whether or not that architecture implements dynamic ftrace. FTRACE itself may be ported to an architecture without porting dynamic ftrace. This patch adds HAVE_DYNAMIC_FTRACE to allow architectures to port ftrace without having to also port the dynamic aspect as well. Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index f3005717bcd0..5c2295b29f2c 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -4,6 +4,9 @@ config HAVE_FTRACE bool +config HAVE_DYNAMIC_FTRACE + bool + config TRACER_MAX_TRACE bool @@ -94,6 +97,7 @@ config CONTEXT_SWITCH_TRACER config DYNAMIC_FTRACE bool "enable/disable ftrace tracepoints dynamically" depends on FTRACE + depends on HAVE_DYNAMIC_FTRACE default y help This option will modify all the calls to ftrace dynamically -- cgit v1.2.3 From f06c38103ea9dbca27c3f4d77f444ddefb5477cd Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: add sysprof plugin very first baby version. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 8 +++++ kernel/trace/Makefile | 1 + kernel/trace/trace_sysprof.c | 80 ++++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 89 insertions(+) create mode 100644 kernel/trace/trace_sysprof.c (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 5c2295b29f2c..e101c9a85f0f 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -75,6 +75,14 @@ config PREEMPT_TRACER enabled. This option and the irqs-off timing option can be used together or separately.) +config SYSPROF_TRACER + bool "Sysprof Tracer" + depends on DEBUG_KERNEL + select TRACING + help + This tracer provides the trace needed by the 'Sysprof' userspace + tool. + config SCHED_TRACER bool "Scheduling Latency Tracer" depends on HAVE_FTRACE diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index d9efbbfa2bdf..7aec123ec1d8 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -14,6 +14,7 @@ obj-$(CONFIG_FTRACE) += libftrace.o obj-$(CONFIG_TRACING) += trace.o obj-$(CONFIG_CONTEXT_SWITCH_TRACER) += trace_sched_switch.o +obj-$(CONFIG_SYSPROF_TRACER) += trace_sysprof.o obj-$(CONFIG_FTRACE) += trace_functions.o obj-$(CONFIG_IRQSOFF_TRACER) += trace_irqsoff.o obj-$(CONFIG_PREEMPT_TRACER) += trace_irqsoff.o diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c new file mode 100644 index 000000000000..6c139bc1be7e --- /dev/null +++ b/kernel/trace/trace_sysprof.c @@ -0,0 +1,80 @@ +/* + * trace stack traces + * + * Copyright (C) 2007 Steven Rostedt + * Copyright (C) 2008 Ingo Molnar + * + */ +#include +#include +#include +#include +#include +#include +#include + +#include "trace.h" + +static struct trace_array *ctx_trace; +static int __read_mostly tracer_enabled; + +static notrace void stack_reset(struct trace_array *tr) +{ + int cpu; + + tr->time_start = ftrace_now(tr->cpu); + + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); +} + +static notrace void start_stack_trace(struct trace_array *tr) +{ + stack_reset(tr); + tracer_enabled = 1; +} + +static notrace void stop_stack_trace(struct trace_array *tr) +{ + tracer_enabled = 0; +} + +static notrace void stack_trace_init(struct trace_array *tr) +{ + ctx_trace = tr; + + if (tr->ctrl) + start_stack_trace(tr); +} + +static notrace void stack_trace_reset(struct trace_array *tr) +{ + if (tr->ctrl) + stop_stack_trace(tr); +} + +static void stack_trace_ctrl_update(struct trace_array *tr) +{ + /* When starting a new trace, reset the buffers */ + if (tr->ctrl) + start_stack_trace(tr); + else + stop_stack_trace(tr); +} + +static struct tracer stack_trace __read_mostly = +{ + .name = "sysprof", + .init = stack_trace_init, + .reset = stack_trace_reset, + .ctrl_update = stack_trace_ctrl_update, +#ifdef CONFIG_FTRACE_SELFTEST + .selftest = trace_selftest_startup_stack, +#endif +}; + +__init static int init_stack_trace(void) +{ + return register_tracer(&stack_trace); +} +device_initcall(init_stack_trace); -- cgit v1.2.3 From 0075fa80305f3231a2d5df97b00d7f55a48ea27e Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: extend sysprof plugin add per CPU hrtimers. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 67 +++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 63 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index 6c139bc1be7e..ba55b871b3d9 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -5,19 +5,76 @@ * Copyright (C) 2008 Ingo Molnar * */ -#include -#include -#include #include +#include +#include #include -#include #include +#include +#include #include "trace.h" static struct trace_array *ctx_trace; static int __read_mostly tracer_enabled; +static const unsigned long sample_period = 1000000; + +/* + * Per CPU hrtimers that do the profiling: + */ +static DEFINE_PER_CPU(struct hrtimer, stack_trace_hrtimer); + +static enum hrtimer_restart stack_trace_timer_fn(struct hrtimer *hrtimer) +{ + /* trace here */ + panic_timeout++; + + hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period)); + + return HRTIMER_RESTART; +} + +static void start_stack_timer(int cpu) +{ + struct hrtimer *hrtimer = &per_cpu(stack_trace_hrtimer, cpu); + + hrtimer_init(hrtimer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); + hrtimer->function = stack_trace_timer_fn; + hrtimer->cb_mode = HRTIMER_CB_IRQSAFE_NO_SOFTIRQ; + + hrtimer_start(hrtimer, ns_to_ktime(sample_period), HRTIMER_MODE_REL); +} + +static void start_stack_timers(void) +{ + cpumask_t saved_mask = current->cpus_allowed; + int cpu; + + for_each_online_cpu(cpu) { + set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu)); + start_stack_timer(cpu); + printk("started timer on cpu%d\n", cpu); + } + set_cpus_allowed_ptr(current, &saved_mask); +} + +static void stop_stack_timer(int cpu) +{ + struct hrtimer *hrtimer = &per_cpu(stack_trace_hrtimer, cpu); + + hrtimer_cancel(hrtimer); + printk("cancelled timer on cpu%d\n", cpu); +} + +static void stop_stack_timers(void) +{ + int cpu; + + for_each_online_cpu(cpu) + stop_stack_timer(cpu); +} + static notrace void stack_reset(struct trace_array *tr) { int cpu; @@ -31,11 +88,13 @@ static notrace void stack_reset(struct trace_array *tr) static notrace void start_stack_trace(struct trace_array *tr) { stack_reset(tr); + start_stack_timers(); tracer_enabled = 1; } static notrace void stop_stack_trace(struct trace_array *tr) { + stop_stack_timers(); tracer_enabled = 0; } -- cgit v1.2.3 From 56a08bdcff20f0022bd9160c1093e56f763499aa Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: extend sysprof plugin some more Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 80 +++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 76 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index ba55b871b3d9..b1137c11ef8b 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -3,7 +3,7 @@ * * Copyright (C) 2007 Steven Rostedt * Copyright (C) 2008 Ingo Molnar - * + * Copyright (C) 2004, 2005, Soeren Sandmann */ #include #include @@ -11,13 +11,17 @@ #include #include #include +#include #include #include "trace.h" -static struct trace_array *ctx_trace; +static struct trace_array *sysprof_trace; static int __read_mostly tracer_enabled; +/* + * 10 msecs for now: + */ static const unsigned long sample_period = 1000000; /* @@ -25,10 +29,78 @@ static const unsigned long sample_period = 1000000; */ static DEFINE_PER_CPU(struct hrtimer, stack_trace_hrtimer); +struct stack_frame { + const void __user *next_fp; + unsigned long return_address; +}; + +static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) +{ + if (!access_ok(VERIFY_READ, fp, sizeof(*frame))) + return 0; + + if (__copy_from_user_inatomic(frame, frame_pointer, sizeof(*frame))) + return 0; + + return 1; +} + +#define SYSPROF_MAX_ADDRESSES 512 + +static void timer_notify(struct pt_regs *regs, int cpu) +{ + const void __user *frame_pointer; + struct trace_array_cpu *data; + struct stack_frame frame; + struct trace_array *tr; + int is_user; + int i; + + if (!regs) + return; + + tr = sysprof_trace; + data = tr->data[cpu]; + is_user = user_mode(regs); + + if (!current || current->pid == 0) + return; + + if (is_user && current->state != TASK_RUNNING) + return; + + if (!is_user) { + /* kernel */ + ftrace(tr, data, current->pid, 1, 0); + return; + + } + + trace_special(tr, data, 0, current->pid, regs->ip); + + frame_pointer = (void __user *)regs->bp; + + for (i = 0; i < SYSPROF_MAX_ADDRESSES; i++) { + if (!copy_stack_frame(frame_pointer, &frame)) + break; + if ((unsigned long)frame_pointer < regs->sp) + break; + + trace_special(tr, data, 1, frame.return_address, + (unsigned long)frame_pointer); + frame_pointer = frame.next_fp; + } + + trace_special(tr, data, 2, current->pid, i); + + if (i == SYSPROF_MAX_ADDRESSES) + trace_special(tr, data, -1, -1, -1); +} + static enum hrtimer_restart stack_trace_timer_fn(struct hrtimer *hrtimer) { /* trace here */ - panic_timeout++; + timer_notify(get_irq_regs(), smp_processor_id()); hrtimer_forward_now(hrtimer, ns_to_ktime(sample_period)); @@ -100,7 +172,7 @@ static notrace void stop_stack_trace(struct trace_array *tr) static notrace void stack_trace_init(struct trace_array *tr) { - ctx_trace = tr; + sysprof_trace = tr; if (tr->ctrl) start_stack_trace(tr); -- cgit v1.2.3 From a6dd24f8d00cbccb560b19a723e6fb9bdfb20799 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: sysprof-plugin, add self-tests Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.h | 4 ++++ kernel/trace/trace_selftest.c | 28 ++++++++++++++++++++++++++++ kernel/trace/trace_sysprof.c | 6 +++--- 3 files changed, 35 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index c460e85e94ed..b2198bc830ae 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -280,6 +280,10 @@ extern int trace_selftest_startup_wakeup(struct tracer *trace, extern int trace_selftest_startup_sched_switch(struct tracer *trace, struct trace_array *tr); #endif +#ifdef CONFIG_SYSPROF_TRACER +extern int trace_selftest_startup_sysprof(struct tracer *trace, + struct trace_array *tr); +#endif #endif /* CONFIG_FTRACE_STARTUP_TEST */ extern void *head_page(struct trace_array_cpu *data); diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 3877dd9102f1..033a6fb2e5ff 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -537,3 +537,31 @@ trace_selftest_startup_sched_switch(struct tracer *trace, struct trace_array *tr return ret; } #endif /* CONFIG_CONTEXT_SWITCH_TRACER */ + +#ifdef CONFIG_SYSPROF_TRACER +int +trace_selftest_startup_sysprof(struct tracer *trace, struct trace_array *tr) +{ + unsigned long count; + int ret; + + /* start the tracing */ + tr->ctrl = 1; + trace->init(tr); + /* Sleep for a 1/10 of a second */ + msleep(100); + /* stop the tracing. */ + tr->ctrl = 0; + trace->ctrl_update(tr); + /* check the trace buffer */ + ret = trace_test_buffer(tr, &count); + trace->reset(tr); + + if (!ret && !count) { + printk(KERN_CONT ".. no entries found .."); + ret = -1; + } + + return ret; +} +#endif /* CONFIG_SYSPROF_TRACER */ diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index b1137c11ef8b..b78f12f77fca 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -126,7 +126,7 @@ static void start_stack_timers(void) for_each_online_cpu(cpu) { set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu)); start_stack_timer(cpu); - printk("started timer on cpu%d\n", cpu); + printk(KERN_INFO "started sysprof timer on cpu%d\n", cpu); } set_cpus_allowed_ptr(current, &saved_mask); } @@ -136,7 +136,7 @@ static void stop_stack_timer(int cpu) struct hrtimer *hrtimer = &per_cpu(stack_trace_hrtimer, cpu); hrtimer_cancel(hrtimer); - printk("cancelled timer on cpu%d\n", cpu); + printk(KERN_INFO "cancelled sysprof timer on cpu%d\n", cpu); } static void stop_stack_timers(void) @@ -200,7 +200,7 @@ static struct tracer stack_trace __read_mostly = .reset = stack_trace_reset, .ctrl_update = stack_trace_ctrl_update, #ifdef CONFIG_FTRACE_SELFTEST - .selftest = trace_selftest_startup_stack, + .selftest = trace_selftest_startup_sysprof, #endif }; -- cgit v1.2.3 From 842af315e8b0adad58fc642eaa5e6f53525e0534 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:47 +0200 Subject: ftrace: sysprof plugin improvement add sample maximum depth. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index b78f12f77fca..7f6fcccffb88 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -23,6 +23,7 @@ static int __read_mostly tracer_enabled; * 10 msecs for now: */ static const unsigned long sample_period = 1000000; +static const unsigned int sample_max_depth = 512; /* * Per CPU hrtimers that do the profiling: @@ -45,8 +46,6 @@ static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) return 1; } -#define SYSPROF_MAX_ADDRESSES 512 - static void timer_notify(struct pt_regs *regs, int cpu) { const void __user *frame_pointer; @@ -80,7 +79,7 @@ static void timer_notify(struct pt_regs *regs, int cpu) frame_pointer = (void __user *)regs->bp; - for (i = 0; i < SYSPROF_MAX_ADDRESSES; i++) { + for (i = 0; i < sample_max_depth; i++) { if (!copy_stack_frame(frame_pointer, &frame)) break; if ((unsigned long)frame_pointer < regs->sp) @@ -93,7 +92,7 @@ static void timer_notify(struct pt_regs *regs, int cpu) trace_special(tr, data, 2, current->pid, i); - if (i == SYSPROF_MAX_ADDRESSES) + if (i == sample_max_depth) trace_special(tr, data, -1, -1, -1); } @@ -126,7 +125,6 @@ static void start_stack_timers(void) for_each_online_cpu(cpu) { set_cpus_allowed_ptr(current, &cpumask_of_cpu(cpu)); start_stack_timer(cpu); - printk(KERN_INFO "started sysprof timer on cpu%d\n", cpu); } set_cpus_allowed_ptr(current, &saved_mask); } @@ -136,7 +134,6 @@ static void stop_stack_timer(int cpu) struct hrtimer *hrtimer = &per_cpu(stack_trace_hrtimer, cpu); hrtimer_cancel(hrtimer); - printk(KERN_INFO "cancelled sysprof timer on cpu%d\n", cpu); } static void stop_stack_timers(void) -- cgit v1.2.3 From ef4ab15ff34fd9c65e92bee70f58e7179da881c5 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: make sysprof dependent on x86 for now that's the only tested platform for now. If there's interest we can make it generic easily. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 2 +- kernel/trace/trace_selftest.c | 5 ----- 2 files changed, 1 insertion(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index e101c9a85f0f..9b49526ac0b5 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -77,7 +77,7 @@ config PREEMPT_TRACER config SYSPROF_TRACER bool "Sysprof Tracer" - depends on DEBUG_KERNEL + depends on DEBUG_KERNEL && X86 select TRACING help This tracer provides the trace needed by the 'Sysprof' userspace diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 033a6fb2e5ff..5588ecc40985 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -557,11 +557,6 @@ trace_selftest_startup_sysprof(struct tracer *trace, struct trace_array *tr) ret = trace_test_buffer(tr, &count); trace->reset(tr); - if (!ret && !count) { - printk(KERN_CONT ".. no entries found .."); - ret = -1; - } - return ret; } #endif /* CONFIG_SYSPROF_TRACER */ -- cgit v1.2.3 From 9f6b4e3f4a24f2590f1c96f117fc45fbea9b0fa4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:48 +0200 Subject: ftrace: sysprof fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 28 +++++++++++++++++++--------- 1 file changed, 19 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index 7f6fcccffb88..f9a09fe705b0 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -37,21 +37,26 @@ struct stack_frame { static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) { + int ret; + if (!access_ok(VERIFY_READ, fp, sizeof(*frame))) return 0; - if (__copy_from_user_inatomic(frame, frame_pointer, sizeof(*frame))) - return 0; + ret = 1; + pagefault_disable(); + if (__copy_from_user_inatomic(frame, fp, sizeof(*frame))) + ret = 0; + pagefault_enable(); - return 1; + return ret; } static void timer_notify(struct pt_regs *regs, int cpu) { - const void __user *frame_pointer; struct trace_array_cpu *data; struct stack_frame frame; struct trace_array *tr; + const void __user *fp; int is_user; int i; @@ -77,21 +82,26 @@ static void timer_notify(struct pt_regs *regs, int cpu) trace_special(tr, data, 0, current->pid, regs->ip); - frame_pointer = (void __user *)regs->bp; + fp = (void __user *)regs->bp; for (i = 0; i < sample_max_depth; i++) { - if (!copy_stack_frame(frame_pointer, &frame)) + frame.next_fp = 0; + frame.return_address = 0; + if (!copy_stack_frame(fp, &frame)) break; - if ((unsigned long)frame_pointer < regs->sp) + if ((unsigned long)fp < regs->sp) break; trace_special(tr, data, 1, frame.return_address, - (unsigned long)frame_pointer); - frame_pointer = frame.next_fp; + (unsigned long)fp); + fp = frame.next_fp; } trace_special(tr, data, 2, current->pid, i); + /* + * Special trace entry if we overflow the max depth: + */ if (i == sample_max_depth) trace_special(tr, data, -1, -1, -1); } -- cgit v1.2.3 From d618b3e6e50970a6248ac857653fdd49bcd3c045 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:49 +0200 Subject: ftrace: sysprof updates make the sample period configurable. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 3 ++ kernel/trace/trace.h | 2 ++ kernel/trace/trace_sysprof.c | 70 ++++++++++++++++++++++++++++++++++++++++++-- 3 files changed, 73 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3271916ff033..95b7c48a9a1d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -2800,6 +2800,9 @@ static __init void tracer_init_debugfs(void) pr_warning("Could not create debugfs " "'dyn_ftrace_total_info' entry\n"); #endif +#ifdef CONFIG_SYSPROF_TRACER + init_tracer_sysprof_debugfs(d_tracer); +#endif } static int trace_alloc_page(void) diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index b2198bc830ae..b7f85d9c80d7 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -188,6 +188,8 @@ struct trace_iterator { void tracing_reset(struct trace_array_cpu *data); int tracing_open_generic(struct inode *inode, struct file *filp); struct dentry *tracing_init_dentry(void); +void init_tracer_sysprof_debugfs(struct dentry *d_tracer); + void ftrace(struct trace_array *tr, struct trace_array_cpu *data, unsigned long ip, diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index f9a09fe705b0..19406236b67b 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -20,11 +20,12 @@ static struct trace_array *sysprof_trace; static int __read_mostly tracer_enabled; /* - * 10 msecs for now: + * 1 msec sample interval by default: */ -static const unsigned long sample_period = 1000000; +static unsigned long sample_period = 1000000; static const unsigned int sample_max_depth = 512; +static DEFINE_MUTEX(sample_timer_lock); /* * Per CPU hrtimers that do the profiling: */ @@ -166,15 +167,19 @@ static notrace void stack_reset(struct trace_array *tr) static notrace void start_stack_trace(struct trace_array *tr) { + mutex_lock(&sample_timer_lock); stack_reset(tr); start_stack_timers(); tracer_enabled = 1; + mutex_unlock(&sample_timer_lock); } static notrace void stop_stack_trace(struct trace_array *tr) { + mutex_lock(&sample_timer_lock); stop_stack_timers(); tracer_enabled = 0; + mutex_unlock(&sample_timer_lock); } static notrace void stack_trace_init(struct trace_array *tr) @@ -216,3 +221,64 @@ __init static int init_stack_trace(void) return register_tracer(&stack_trace); } device_initcall(init_stack_trace); + +#define MAX_LONG_DIGITS 22 + +static ssize_t +sysprof_sample_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char buf[MAX_LONG_DIGITS]; + int r; + + r = sprintf(buf, "%ld\n", nsecs_to_usecs(sample_period)); + + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); +} + +static ssize_t +sysprof_sample_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char buf[MAX_LONG_DIGITS]; + unsigned long val; + + if (cnt > MAX_LONG_DIGITS-1) + cnt = MAX_LONG_DIGITS-1; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + buf[cnt] = 0; + + val = simple_strtoul(buf, NULL, 10); + /* + * Enforce a minimum sample period of 100 usecs: + */ + if (val < 100) + val = 100; + + mutex_lock(&sample_timer_lock); + stop_stack_timers(); + sample_period = val * 1000; + start_stack_timers(); + mutex_unlock(&sample_timer_lock); + + return cnt; +} + +static struct file_operations sysprof_sample_fops = { + .read = sysprof_sample_read, + .write = sysprof_sample_write, +}; + +void init_tracer_sysprof_debugfs(struct dentry *d_tracer) +{ + struct dentry *entry; + + entry = debugfs_create_file("sysprof_sample_period", 0644, + d_tracer, NULL, &sysprof_sample_fops); + if (entry) + return; + pr_warning("Could not create debugfs 'dyn_ftrace_total_info' entry\n"); +} -- cgit v1.2.3 From ada6b835067dc022f11cdae1c313a3710d3d977c Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 23 May 2008 23:50:41 +0200 Subject: ftrace: remove notrace Remove the notrace annotations. The build logic takes care of that. Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 10 +++++----- 1 file changed, 5 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index 19406236b67b..3b1e4ba9180d 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -155,7 +155,7 @@ static void stop_stack_timers(void) stop_stack_timer(cpu); } -static notrace void stack_reset(struct trace_array *tr) +static void stack_reset(struct trace_array *tr) { int cpu; @@ -165,7 +165,7 @@ static notrace void stack_reset(struct trace_array *tr) tracing_reset(tr->data[cpu]); } -static notrace void start_stack_trace(struct trace_array *tr) +static void start_stack_trace(struct trace_array *tr) { mutex_lock(&sample_timer_lock); stack_reset(tr); @@ -174,7 +174,7 @@ static notrace void start_stack_trace(struct trace_array *tr) mutex_unlock(&sample_timer_lock); } -static notrace void stop_stack_trace(struct trace_array *tr) +static void stop_stack_trace(struct trace_array *tr) { mutex_lock(&sample_timer_lock); stop_stack_timers(); @@ -182,7 +182,7 @@ static notrace void stop_stack_trace(struct trace_array *tr) mutex_unlock(&sample_timer_lock); } -static notrace void stack_trace_init(struct trace_array *tr) +static void stack_trace_init(struct trace_array *tr) { sysprof_trace = tr; @@ -190,7 +190,7 @@ static notrace void stack_trace_init(struct trace_array *tr) start_stack_trace(tr); } -static notrace void stack_trace_reset(struct trace_array *tr) +static void stack_trace_reset(struct trace_array *tr) { if (tr->ctrl) stop_stack_trace(tr); -- cgit v1.2.3 From 9caee613d3b860ae81b79370eeae9ac967c07536 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 23 May 2008 23:55:54 +0200 Subject: ftrace: fix __trace_special() Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index 3b1e4ba9180d..76dd953eeccd 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -81,7 +81,7 @@ static void timer_notify(struct pt_regs *regs, int cpu) } - trace_special(tr, data, 0, current->pid, regs->ip); + __trace_special(tr, data, 0, current->pid, regs->ip); fp = (void __user *)regs->bp; @@ -93,18 +93,18 @@ static void timer_notify(struct pt_regs *regs, int cpu) if ((unsigned long)fp < regs->sp) break; - trace_special(tr, data, 1, frame.return_address, + __trace_special(tr, data, 1, frame.return_address, (unsigned long)fp); fp = frame.next_fp; } - trace_special(tr, data, 2, current->pid, i); + __trace_special(tr, data, 2, current->pid, i); /* * Special trace entry if we overflow the max depth: */ if (i == sample_max_depth) - trace_special(tr, data, -1, -1, -1); + __trace_special(tr, data, -1, -1, -1); } static enum hrtimer_restart stack_trace_timer_fn(struct hrtimer *hrtimer) -- cgit v1.2.3 From 5fc4511c756860149b81aead6eca5bdf5c438ea7 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 23 May 2008 23:58:21 +0200 Subject: ftrace: make it more available in the Kconfig Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index 9b49526ac0b5..e101c9a85f0f 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -77,7 +77,7 @@ config PREEMPT_TRACER config SYSPROF_TRACER bool "Sysprof Tracer" - depends on DEBUG_KERNEL && X86 + depends on DEBUG_KERNEL select TRACING help This tracer provides the trace needed by the 'Sysprof' userspace -- cgit v1.2.3 From cd2134b1dda92fd450e6a1e12b1c7960dd6a2178 Mon Sep 17 00:00:00 2001 From: Soeren Sandmann Pedersen Date: Mon, 12 May 2008 21:20:54 +0200 Subject: sysprof: kernel trace add kernel backtracing to the sysprof tracer. change the format of the data, so that type=0 means beginning of stack trace, 1 means kernel address, 2 means user address, and 3 means end of trace. EIP addresses are no longer distinguished from return addresses, mostly because sysprof userspace doesn't make use of it. It may be worthwhile adding this back in though, just in case it becomes interesting. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 89 +++++++++++++++++++++++++++++++++++++++----- 1 file changed, 80 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index 76dd953eeccd..ebcb66d054cc 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -14,6 +14,8 @@ #include #include +#include + #include "trace.h" static struct trace_array *sysprof_trace; @@ -52,6 +54,77 @@ static int copy_stack_frame(const void __user *fp, struct stack_frame *frame) return ret; } +struct backtrace_info { + struct trace_array_cpu *data; + struct trace_array *tr; + int pos; +}; + +static void +backtrace_warning_symbol(void *data, char *msg, unsigned long symbol) +{ + /* Ignore warnings */ +} + +static void backtrace_warning(void *data, char *msg) +{ + /* Ignore warnings */ +} + +static int backtrace_stack(void *data, char *name) +{ + /* Don't bother with IRQ stacks for now */ + return -1; +} + +static void backtrace_address(void *data, unsigned long addr, int reliable) +{ + struct backtrace_info *info = data; + + if (info->pos < sample_max_depth && reliable) { + __trace_special(info->tr, info->data, 1, addr, 0); + + info->pos++; + } +} + +const static struct stacktrace_ops backtrace_ops = { + .warning = backtrace_warning, + .warning_symbol = backtrace_warning_symbol, + .stack = backtrace_stack, + .address = backtrace_address, +}; + +static struct pt_regs * +trace_kernel(struct pt_regs *regs, struct trace_array *tr, + struct trace_array_cpu *data) +{ + struct backtrace_info info; + unsigned long bp; + char *user_stack; + char *stack; + + info.tr = tr; + info.data = data; + info.pos = 1; + + __trace_special(info.tr, info.data, 1, regs->ip, 0); + + stack = ((char *)regs + sizeof(struct pt_regs)); +#ifdef CONFIG_FRAME_POINTER + bp = regs->bp; +#else + bp = 0; +#endif + + dump_trace(NULL, regs, (void *)stack, bp, &backtrace_ops, &info); + + /* Now trace the user stack */ + user_stack = ((char *)current->thread.sp0 - sizeof(struct pt_regs)); + + return (struct pt_regs *)user_stack; +} + static void timer_notify(struct pt_regs *regs, int cpu) { struct trace_array_cpu *data; @@ -74,17 +147,15 @@ static void timer_notify(struct pt_regs *regs, int cpu) if (is_user && current->state != TASK_RUNNING) return; - if (!is_user) { - /* kernel */ - ftrace(tr, data, current->pid, 1, 0); - return; + __trace_special(tr, data, 0, 0, current->pid); - } - - __trace_special(tr, data, 0, current->pid, regs->ip); + if (!is_user) + regs = trace_kernel(regs, tr, data); fp = (void __user *)regs->bp; + __trace_special(tr, data, 2, regs->ip, 0); + for (i = 0; i < sample_max_depth; i++) { frame.next_fp = 0; frame.return_address = 0; @@ -93,12 +164,12 @@ static void timer_notify(struct pt_regs *regs, int cpu) if ((unsigned long)fp < regs->sp) break; - __trace_special(tr, data, 1, frame.return_address, + __trace_special(tr, data, 2, frame.return_address, (unsigned long)fp); fp = frame.next_fp; } - __trace_special(tr, data, 2, current->pid, i); + __trace_special(tr, data, 3, current->pid, i); /* * Special trace entry if we overflow the max depth: -- cgit v1.2.3 From 8a9e94c1fbfdac45a3b6811b880777c4116aa309 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:54 +0200 Subject: sysprof: update copyrights Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index ebcb66d054cc..fe23d6dba7f1 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -1,9 +1,9 @@ /* * trace stack traces * + * Copyright (C) 2004-2008, Soeren Sandmann * Copyright (C) 2007 Steven Rostedt * Copyright (C) 2008 Ingo Molnar - * Copyright (C) 2004, 2005, Soeren Sandmann */ #include #include -- cgit v1.2.3 From cf3271a73b612a03da00681ecd9bfefab37c74c9 Mon Sep 17 00:00:00 2001 From: Soeren Sandmann Date: Mon, 12 May 2008 05:28:50 +0200 Subject: ftrace/sysprof: don't trace the user stack if we are a kernel thread. Check that current->mm is non-NULL before attempting to trace the user stack. Also take depth of the kernel stack into account when comparing against sample_max_depth. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sysprof.c | 50 +++++++++++++++++++++++++------------------- 1 file changed, 29 insertions(+), 21 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sysprof.c b/kernel/trace/trace_sysprof.c index fe23d6dba7f1..2301e1e7c606 100644 --- a/kernel/trace/trace_sysprof.c +++ b/kernel/trace/trace_sysprof.c @@ -95,13 +95,12 @@ const static struct stacktrace_ops backtrace_ops = { .address = backtrace_address, }; -static struct pt_regs * +static int trace_kernel(struct pt_regs *regs, struct trace_array *tr, struct trace_array_cpu *data) { struct backtrace_info info; unsigned long bp; - char *user_stack; char *stack; info.tr = tr; @@ -119,10 +118,7 @@ trace_kernel(struct pt_regs *regs, struct trace_array *tr, dump_trace(NULL, regs, (void *)stack, bp, &backtrace_ops, &info); - /* Now trace the user stack */ - user_stack = ((char *)current->thread.sp0 - sizeof(struct pt_regs)); - - return (struct pt_regs *)user_stack; + return info.pos; } static void timer_notify(struct pt_regs *regs, int cpu) @@ -150,32 +146,44 @@ static void timer_notify(struct pt_regs *regs, int cpu) __trace_special(tr, data, 0, 0, current->pid); if (!is_user) - regs = trace_kernel(regs, tr, data); + i = trace_kernel(regs, tr, data); + else + i = 0; - fp = (void __user *)regs->bp; + /* + * Trace user stack if we are not a kernel thread + */ + if (current->mm && i < sample_max_depth) { + regs = (struct pt_regs *)current->thread.sp0 - 1; - __trace_special(tr, data, 2, regs->ip, 0); + fp = (void __user *)regs->bp; - for (i = 0; i < sample_max_depth; i++) { - frame.next_fp = 0; - frame.return_address = 0; - if (!copy_stack_frame(fp, &frame)) - break; - if ((unsigned long)fp < regs->sp) - break; + __trace_special(tr, data, 2, regs->ip, 0); - __trace_special(tr, data, 2, frame.return_address, - (unsigned long)fp); - fp = frame.next_fp; - } + while (i < sample_max_depth) { + frame.next_fp = 0; + frame.return_address = 0; + if (!copy_stack_frame(fp, &frame)) + break; + if ((unsigned long)fp < regs->sp) + break; - __trace_special(tr, data, 3, current->pid, i); + __trace_special(tr, data, 2, frame.return_address, + (unsigned long)fp); + fp = frame.next_fp; + + i++; + } + + } /* * Special trace entry if we overflow the max depth: */ if (i == sample_max_depth) __trace_special(tr, data, -1, -1, -1); + + __trace_special(tr, data, 3, current->pid, i); } static enum hrtimer_restart stack_trace_timer_fn(struct hrtimer *hrtimer) -- cgit v1.2.3 From f984b51e0779a6dd30feedc41404013ca54e5d05 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:57 +0200 Subject: ftrace: add mmiotrace plugin On Sat, 22 Mar 2008 13:07:47 +0100 Ingo Molnar wrote: > > > i'd suggest the following: pull x86.git and sched-devel.git into a > > > single tree [the two will combine without rejects]. Then try to add a > > > kernel/tracing/trace_mmiotrace.c ftrace plugin. The trace_sysprof.c > > > plugin might be a good example. > > > > I did this and now I have mmiotrace enabled/disabled via the tracing > > framework (what do we call this, since ftrace is one of the tracers?). > > cool! could you send the patches for that? (even if they are not fully > functional yet) Patch attached in the end. Nice to see how much code disappeared. I tried to mark all the features I had to break with XXX-comments. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/Makefile | 1 + kernel/trace/trace_mmiotrace.c | 84 ++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 85 insertions(+) create mode 100644 kernel/trace/trace_mmiotrace.c (limited to 'kernel') diff --git a/kernel/trace/Makefile b/kernel/trace/Makefile index d9efbbfa2bdf..c44a7dce9086 100644 --- a/kernel/trace/Makefile +++ b/kernel/trace/Makefile @@ -18,5 +18,6 @@ obj-$(CONFIG_FTRACE) += trace_functions.o obj-$(CONFIG_IRQSOFF_TRACER) += trace_irqsoff.o obj-$(CONFIG_PREEMPT_TRACER) += trace_irqsoff.o obj-$(CONFIG_SCHED_TRACER) += trace_sched_wakeup.o +obj-$(CONFIG_MMIOTRACE) += trace_mmiotrace.o libftrace-y := ftrace.o diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c new file mode 100644 index 000000000000..e4dd03cc5aa6 --- /dev/null +++ b/kernel/trace/trace_mmiotrace.c @@ -0,0 +1,84 @@ +/* + * Memory mapped I/O tracing + * + * Copyright (C) 2008 Pekka Paalanen + */ + +#define DEBUG 1 + +#include +#include + +#include "trace.h" + +extern void +__trace_special(void *__tr, void *__data, + unsigned long arg1, unsigned long arg2, unsigned long arg3); + +static struct trace_array *mmio_trace_array; + + +static void mmio_trace_init(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + mmio_trace_array = tr; + if (tr->ctrl) + enable_mmiotrace(); +} + +static void mmio_trace_reset(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + if (tr->ctrl) + disable_mmiotrace(); +} + +static void mmio_trace_ctrl_update(struct trace_array *tr) +{ + pr_debug("in %s\n", __func__); + if (tr->ctrl) + enable_mmiotrace(); + else + disable_mmiotrace(); +} + +static struct tracer mmio_tracer __read_mostly = +{ + .name = "mmiotrace", + .init = mmio_trace_init, + .reset = mmio_trace_reset, + .ctrl_update = mmio_trace_ctrl_update, +}; + +__init static int init_mmio_trace(void) +{ + int ret = init_mmiotrace(); + if (ret) + return ret; + return register_tracer(&mmio_tracer); +} +device_initcall(init_mmio_trace); + +void mmio_trace_record(u32 type, unsigned long addr, unsigned long arg) +{ + struct trace_array *tr = mmio_trace_array; + struct trace_array_cpu *data = tr->data[smp_processor_id()]; + + if (!current || current->pid == 0) { + /* + * XXX: This is a problem. We need to able to record, no + * matter what. tracing_generic_entry_update() would crash. + */ + static unsigned limit; + if (limit++ < 12) + pr_err("Error in %s: no current.\n", __func__); + return; + } + if (!tr || !data) { + static unsigned limit; + if (limit++ < 12) + pr_err("%s: no tr or data\n", __func__); + return; + } + __trace_special(tr, data, type, addr, arg); +} -- cgit v1.2.3 From bd8ac686c73c7e925fcfe0b02dc4e7b947127864 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:57 +0200 Subject: ftrace: mmiotrace, updates here is a patch that makes mmiotrace work almost well within the tracing framework. The patch applies on top of my previous patch. I have my own output formatting in place now. Summary of changes: - fix the NULL dereference that was due to not calling tracing_reset() - add print_line() callback into struct tracer - implement print_line() for mmiotrace, producing up-to-spec text - add my output header, but that is not really called in the right place - rewrote the main structs in mmiotrace - added two new trace entry types: TRACE_MMIO_RW and TRACE_MMIO_MAP - made some functions in trace.c non-static - check current==NULL in tracing_generic_entry_update() - fix(?) comparison in trace_seq_printf() Things seem to work fine except a few issues. Markers (text lines injected into mmiotrace log) are missing, I did not feel hacking them in before we have variable length entries. My output header is printed only for 'trace' file, but not 'trace_pipe'. For some reason, despite my quick fix, iter->trace is NULL in print_trace_line() when called from 'trace_pipe' file, which means I don't get proper output formatting. I only tried by loading nouveau.ko, which just detects the card, and that is traced fine. I didn't try further. Map, two reads and unmap. Works perfectly. I am missing the information about overflows, I'd prefer to have a counter for lost events. I didn't try, but I guess currently there is no way of knowning when it overflows? So, not too far from being fully operational, it seems :-) And looking at the diffstat, there also is some 700-900 lines of user space code that just became obsolete. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 34 ++++++++++ kernel/trace/trace.h | 14 ++++ kernel/trace/trace_mmiotrace.c | 151 +++++++++++++++++++++++++++++++++-------- 3 files changed, 171 insertions(+), 28 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3271916ff033..d14fe49e9638 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -831,6 +831,40 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, trace_function(tr, data, ip, parent_ip, flags); } +#ifdef CONFIG_MMIOTRACE +void __trace_mmiotrace_rw(struct trace_array *tr, struct trace_array_cpu *data, + struct mmiotrace_rw *rw) +{ + struct trace_entry *entry; + unsigned long irq_flags; + + spin_lock_irqsave(&data->lock, irq_flags); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_MMIO_RW; + entry->mmiorw = *rw; + spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); +} + +void __trace_mmiotrace_map(struct trace_array *tr, struct trace_array_cpu *data, + struct mmiotrace_map *map) +{ + struct trace_entry *entry; + unsigned long irq_flags; + + spin_lock_irqsave(&data->lock, irq_flags); + entry = tracing_get_trace_entry(tr, data); + tracing_generic_entry_update(entry, 0); + entry->type = TRACE_MMIO_MAP; + entry->mmiomap = *map; + spin_unlock_irqrestore(&data->lock, irq_flags); + + trace_wake_up(); +} +#endif + void __trace_stack(struct trace_array *tr, struct trace_array_cpu *data, unsigned long flags, diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index c460e85e94ed..0ef9ef74c806 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -5,6 +5,7 @@ #include #include #include +#include enum trace_type { __TRACE_FIRST_TYPE = 0, @@ -14,6 +15,8 @@ enum trace_type { TRACE_WAKE, TRACE_STACK, TRACE_SPECIAL, + TRACE_MMIO_RW, + TRACE_MMIO_MAP, __TRACE_LAST_TYPE }; @@ -75,6 +78,8 @@ struct trace_entry { struct ctx_switch_entry ctx; struct special_entry special; struct stack_entry stack; + struct mmiotrace_rw mmiorw; + struct mmiotrace_map mmiomap; }; }; @@ -255,6 +260,15 @@ extern unsigned long ftrace_update_tot_cnt; extern int DYN_FTRACE_TEST_NAME(void); #endif +#ifdef CONFIG_MMIOTRACE +extern void __trace_mmiotrace_rw(struct trace_array *tr, + struct trace_array_cpu *data, + struct mmiotrace_rw *rw); +extern void __trace_mmiotrace_map(struct trace_array *tr, + struct trace_array_cpu *data, + struct mmiotrace_map *map); +#endif + #ifdef CONFIG_FTRACE_STARTUP_TEST #ifdef CONFIG_FTRACE extern int trace_selftest_startup_function(struct tracer *trace, diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index e4dd03cc5aa6..3a12b1ad0c63 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -11,19 +11,26 @@ #include "trace.h" -extern void -__trace_special(void *__tr, void *__data, - unsigned long arg1, unsigned long arg2, unsigned long arg3); - static struct trace_array *mmio_trace_array; +static void mmio_reset_data(struct trace_array *tr) +{ + int cpu; + + tr->time_start = ftrace_now(tr->cpu); + + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); +} static void mmio_trace_init(struct trace_array *tr) { pr_debug("in %s\n", __func__); mmio_trace_array = tr; - if (tr->ctrl) + if (tr->ctrl) { + mmio_reset_data(tr); enable_mmiotrace(); + } } static void mmio_trace_reset(struct trace_array *tr) @@ -31,15 +38,110 @@ static void mmio_trace_reset(struct trace_array *tr) pr_debug("in %s\n", __func__); if (tr->ctrl) disable_mmiotrace(); + mmio_reset_data(tr); + mmio_trace_array = NULL; } static void mmio_trace_ctrl_update(struct trace_array *tr) { pr_debug("in %s\n", __func__); - if (tr->ctrl) + if (tr->ctrl) { + mmio_reset_data(tr); enable_mmiotrace(); - else + } else { disable_mmiotrace(); + } +} + +/* XXX: This is not called for trace_pipe file! */ +void mmio_print_header(struct trace_iterator *iter) +{ + struct trace_seq *s = &iter->seq; + trace_seq_printf(s, "VERSION broken 20070824\n"); + /* TODO: print /proc/bus/pci/devices contents as PCIDEV lines */ +} + +static int mmio_print_rw(struct trace_iterator *iter) +{ + struct trace_entry *entry = iter->ent; + struct mmiotrace_rw *rw = &entry->mmiorw; + struct trace_seq *s = &iter->seq; + unsigned long long t = ns2usecs(entry->t); + unsigned long usec_rem = do_div(t, 1000000ULL); + unsigned secs = (unsigned long)t; + int ret = 1; + + switch (entry->mmiorw.opcode) { + case MMIO_READ: + ret = trace_seq_printf(s, + "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, rw->phys, + rw->value, rw->pc, entry->pid); + break; + case MMIO_WRITE: + ret = trace_seq_printf(s, + "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, rw->phys, + rw->value, rw->pc, entry->pid); + break; + case MMIO_UNKNOWN_OP: + ret = trace_seq_printf(s, + "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", + secs, usec_rem, rw->map_id, rw->phys, + (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, + (rw->value >> 0) & 0xff, rw->pc, entry->pid); + break; + default: + ret = trace_seq_printf(s, "rw what?\n"); + break; + } + if (ret) + return 1; + return 0; +} + +static int mmio_print_map(struct trace_iterator *iter) +{ + struct trace_entry *entry = iter->ent; + struct mmiotrace_map *m = &entry->mmiomap; + struct trace_seq *s = &iter->seq; + unsigned long long t = ns2usecs(entry->t); + unsigned long usec_rem = do_div(t, 1000000ULL); + unsigned secs = (unsigned long)t; + int ret = 1; + + switch (entry->mmiorw.opcode) { + case MMIO_PROBE: + ret = trace_seq_printf(s, + "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", + secs, usec_rem, m->map_id, m->phys, m->virt, m->len, + 0UL, entry->pid); + break; + case MMIO_UNPROBE: + ret = trace_seq_printf(s, + "UNMAP %lu.%06lu %d 0x%lx %d\n", + secs, usec_rem, m->map_id, 0UL, entry->pid); + break; + default: + ret = trace_seq_printf(s, "map what?\n"); + break; + } + if (ret) + return 1; + return 0; +} + +/* return 0 to abort printing without consuming current entry in pipe mode */ +static int mmio_print_line(struct trace_iterator *iter) +{ + switch (iter->ent->type) { + case TRACE_MMIO_RW: + return mmio_print_rw(iter); + case TRACE_MMIO_MAP: + return mmio_print_map(iter); + default: + return 1; /* ignore unknown entries */ + } } static struct tracer mmio_tracer __read_mostly = @@ -47,38 +149,31 @@ static struct tracer mmio_tracer __read_mostly = .name = "mmiotrace", .init = mmio_trace_init, .reset = mmio_trace_reset, + .open = mmio_print_header, .ctrl_update = mmio_trace_ctrl_update, + .print_line = mmio_print_line, }; __init static int init_mmio_trace(void) { - int ret = init_mmiotrace(); - if (ret) - return ret; return register_tracer(&mmio_tracer); } device_initcall(init_mmio_trace); -void mmio_trace_record(u32 type, unsigned long addr, unsigned long arg) +void mmio_trace_rw(struct mmiotrace_rw *rw) { struct trace_array *tr = mmio_trace_array; struct trace_array_cpu *data = tr->data[smp_processor_id()]; + __trace_mmiotrace_rw(tr, data, rw); +} - if (!current || current->pid == 0) { - /* - * XXX: This is a problem. We need to able to record, no - * matter what. tracing_generic_entry_update() would crash. - */ - static unsigned limit; - if (limit++ < 12) - pr_err("Error in %s: no current.\n", __func__); - return; - } - if (!tr || !data) { - static unsigned limit; - if (limit++ < 12) - pr_err("%s: no tr or data\n", __func__); - return; - } - __trace_special(tr, data, type, addr, arg); +void mmio_trace_mapping(struct mmiotrace_map *map) +{ + struct trace_array *tr = mmio_trace_array; + struct trace_array_cpu *data; + + preempt_disable(); + data = tr->data[smp_processor_id()]; + __trace_mmiotrace_map(tr, data, map); + preempt_enable(); } -- cgit v1.2.3 From 138295373ccf7625fcb0218dfea114837983bc39 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:58 +0200 Subject: ftrace: mmiotrace update, #2 another weekend, another patch. This should apply on top of my previous patch from March 23rd. Summary of changes: - Print PCI device list in output header - work around recursive probe hits on SMP - refactor dis/arm_kmmio_fault_page() and add check for page levels - remove un/reference_kmmio(), the die notifier hook is registered permanently into the list - explicitly check for single stepping in die notifier callback I have tested this version on my UP Athlon64 desktop with Nouveau, and SMP Core 2 Duo laptop with the proprietary nvidia driver. Both systems are 64-bit. One previously unknown bug crept into daylight: the ftrace framework's output routines print the first entry last after buffer has wrapped around. The most important regressions compared to non-ftrace mmiotrace at this time are: - failure of trace_pipe file - illegal lines in output file - unaware of losing data due to buffer full Personally I'd like to see these three solved before submitting to mainline. Other issues may come up once we know when we lose events. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_mmiotrace.c | 47 +++++++++++++++++++++++++++++++++++++++--- 1 file changed, 44 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 3a12b1ad0c63..361472b5788c 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -8,6 +8,7 @@ #include #include +#include #include "trace.h" @@ -53,12 +54,52 @@ static void mmio_trace_ctrl_update(struct trace_array *tr) } } +static int mmio_print_pcidev(struct trace_seq *s, const struct pci_dev *dev) +{ + int ret = 0; + int i; + resource_size_t start, end; + const struct pci_driver *drv = pci_dev_driver(dev); + + /* XXX: incomplete checks for trace_seq_printf() return value */ + ret += trace_seq_printf(s, "PCIDEV %02x%02x %04x%04x %x", + dev->bus->number, dev->devfn, + dev->vendor, dev->device, dev->irq); + /* + * XXX: is pci_resource_to_user() appropriate, since we are + * supposed to interpret the __ioremap() phys_addr argument based on + * these printed values? + */ + for (i = 0; i < 7; i++) { + pci_resource_to_user(dev, i, &dev->resource[i], &start, &end); + ret += trace_seq_printf(s, " %llx", + (unsigned long long)(start | + (dev->resource[i].flags & PCI_REGION_FLAG_MASK))); + } + for (i = 0; i < 7; i++) { + pci_resource_to_user(dev, i, &dev->resource[i], &start, &end); + ret += trace_seq_printf(s, " %llx", + dev->resource[i].start < dev->resource[i].end ? + (unsigned long long)(end - start) + 1 : 0); + } + if (drv) + ret += trace_seq_printf(s, " %s\n", drv->name); + else + ret += trace_seq_printf(s, " \n"); + return ret; +} + /* XXX: This is not called for trace_pipe file! */ -void mmio_print_header(struct trace_iterator *iter) +static void mmio_print_header(struct trace_iterator *iter) { struct trace_seq *s = &iter->seq; - trace_seq_printf(s, "VERSION broken 20070824\n"); - /* TODO: print /proc/bus/pci/devices contents as PCIDEV lines */ + struct pci_dev *dev = NULL; + + trace_seq_printf(s, "VERSION 20070824\n"); + + for_each_pci_dev(dev) + mmio_print_pcidev(s, dev); + /* XXX: return value? What if header is very long? */ } static int mmio_print_rw(struct trace_iterator *iter) -- cgit v1.2.3 From 801a175bf601f9a9d5e86e92dee9adeeb6625da8 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:20:58 +0200 Subject: mmiotrace: ftrace fix Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index d14fe49e9638..4dcc4e85c5d6 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -838,12 +838,16 @@ void __trace_mmiotrace_rw(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_MMIO_RW; entry->mmiorw = *rw; - spin_unlock_irqrestore(&data->lock, irq_flags); + + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } @@ -854,12 +858,16 @@ void __trace_mmiotrace_map(struct trace_array *tr, struct trace_array_cpu *data, struct trace_entry *entry; unsigned long irq_flags; - spin_lock_irqsave(&data->lock, irq_flags); + raw_local_irq_save(irq_flags); + __raw_spin_lock(&data->lock); + entry = tracing_get_trace_entry(tr, data); tracing_generic_entry_update(entry, 0); entry->type = TRACE_MMIO_MAP; entry->mmiomap = *map; - spin_unlock_irqrestore(&data->lock, irq_flags); + + __raw_spin_unlock(&data->lock); + raw_local_irq_restore(irq_flags); trace_wake_up(); } -- cgit v1.2.3 From 736ca61fa81874b3fee205a593251b1869d0bcf1 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:20:59 +0200 Subject: x86 mmiotrace: Do not print bogus pid Non-zero pid indicates the MMIO access originated in user space. We do not catch that kind of accesses yet, so always print zero for now. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/trace/trace_mmiotrace.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 361472b5788c..79be4a18ec1e 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -117,20 +117,20 @@ static int mmio_print_rw(struct trace_iterator *iter) ret = trace_seq_printf(s, "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", rw->width, secs, usec_rem, rw->map_id, rw->phys, - rw->value, rw->pc, entry->pid); + rw->value, rw->pc, 0); break; case MMIO_WRITE: ret = trace_seq_printf(s, "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", rw->width, secs, usec_rem, rw->map_id, rw->phys, - rw->value, rw->pc, entry->pid); + rw->value, rw->pc, 0); break; case MMIO_UNKNOWN_OP: ret = trace_seq_printf(s, "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", secs, usec_rem, rw->map_id, rw->phys, (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, - (rw->value >> 0) & 0xff, rw->pc, entry->pid); + (rw->value >> 0) & 0xff, rw->pc, 0); break; default: ret = trace_seq_printf(s, "rw what?\n"); -- cgit v1.2.3 From d0a7e8ca5b996d36219e6fc002907291c8ee677b Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: print header using the read hook. Now the header is printed only for `trace_pipe' file. Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 60 +++++++++++++++++++++++++++++++++++++----- 1 file changed, 53 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 79be4a18ec1e..6d2edbdde939 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -12,6 +12,10 @@ #include "trace.h" +struct header_iter { + struct pci_dev *dev; +}; + static struct trace_array *mmio_trace_array; static void mmio_reset_data(struct trace_array *tr) @@ -89,17 +93,57 @@ static int mmio_print_pcidev(struct trace_seq *s, const struct pci_dev *dev) return ret; } -/* XXX: This is not called for trace_pipe file! */ -static void mmio_print_header(struct trace_iterator *iter) +static void destroy_header_iter(struct header_iter *hiter) +{ + if (!hiter) + return; + pci_dev_put(hiter->dev); + kfree(hiter); +} + +static void mmio_pipe_open(struct trace_iterator *iter) { + struct header_iter *hiter; struct trace_seq *s = &iter->seq; - struct pci_dev *dev = NULL; trace_seq_printf(s, "VERSION 20070824\n"); - for_each_pci_dev(dev) - mmio_print_pcidev(s, dev); - /* XXX: return value? What if header is very long? */ + hiter = kzalloc(sizeof(*hiter), GFP_KERNEL); + if (!hiter) + return; + + hiter->dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, NULL); + iter->private = hiter; +} + +/* XXX: This is not called when the pipe is closed! */ +static void mmio_close(struct trace_iterator *iter) +{ + struct header_iter *hiter = iter->private; + destroy_header_iter(hiter); + iter->private = NULL; +} + +static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, + char __user *ubuf, size_t cnt, loff_t *ppos) +{ + ssize_t ret; + struct header_iter *hiter = iter->private; + struct trace_seq *s = &iter->seq; + + if (!hiter) + return 0; + + mmio_print_pcidev(s, hiter->dev); + hiter->dev = pci_get_device(PCI_ANY_ID, PCI_ANY_ID, hiter->dev); + + if (!hiter->dev) { + destroy_header_iter(hiter); + iter->private = NULL; + } + + ret = trace_seq_to_user(s, ubuf, cnt); + return (ret == -EBUSY) ? 0 : ret; } static int mmio_print_rw(struct trace_iterator *iter) @@ -190,7 +234,9 @@ static struct tracer mmio_tracer __read_mostly = .name = "mmiotrace", .init = mmio_trace_init, .reset = mmio_trace_reset, - .open = mmio_print_header, + .pipe_open = mmio_pipe_open, + .close = mmio_close, + .read = mmio_read, .ctrl_update = mmio_trace_ctrl_update, .print_line = mmio_print_line, }; -- cgit v1.2.3 From 2039238b79b51a50f8477f53f33750e1c3fc146a Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: print overrun counts Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 6d2edbdde939..d0f649a2203d 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -17,11 +17,13 @@ struct header_iter { }; static struct trace_array *mmio_trace_array; +static bool overrun_detected; static void mmio_reset_data(struct trace_array *tr) { int cpu; + overrun_detected = false; tr->time_start = ftrace_now(tr->cpu); for_each_online_cpu(cpu) @@ -124,12 +126,34 @@ static void mmio_close(struct trace_iterator *iter) iter->private = NULL; } +static unsigned long count_overruns(struct trace_iterator *iter) +{ + int cpu; + unsigned long cnt = 0; + for_each_online_cpu(cpu) { + cnt += iter->overrun[cpu]; + iter->overrun[cpu] = 0; + } + return cnt; +} + static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { ssize_t ret; struct header_iter *hiter = iter->private; struct trace_seq *s = &iter->seq; + unsigned long n; + + n = count_overruns(iter); + if (n) { + /* XXX: This is later than where events were lost. */ + trace_seq_printf(s, "MARK 0.000000 Lost %lu events.\n", n); + if (!overrun_detected) + pr_warning("mmiotrace has lost events.\n"); + overrun_detected = true; + goto print_out; + } if (!hiter) return 0; @@ -142,6 +166,7 @@ static ssize_t mmio_read(struct trace_iterator *iter, struct file *filp, iter->private = NULL; } +print_out: ret = trace_seq_to_user(s, ubuf, cnt); return (ret == -EBUSY) ? 0 : ret; } -- cgit v1.2.3 From e0fd5c2fa188311667267c02a702ae699a9fc2bd Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:02 +0200 Subject: mmiotrace: do not print bogus pid for maps either Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index d0f649a2203d..3c1dacdc2d85 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -225,12 +225,12 @@ static int mmio_print_map(struct trace_iterator *iter) ret = trace_seq_printf(s, "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", secs, usec_rem, m->map_id, m->phys, m->virt, m->len, - 0UL, entry->pid); + 0UL, 0); break; case MMIO_UNPROBE: ret = trace_seq_printf(s, "UNMAP %lu.%06lu %d 0x%lx %d\n", - secs, usec_rem, m->map_id, 0UL, entry->pid); + secs, usec_rem, m->map_id, 0UL, 0); break; default: ret = trace_seq_printf(s, "map what?\n"); -- cgit v1.2.3 From dee310d0adf41019aca476052ac3085ff286d9be Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 12 May 2008 21:21:03 +0200 Subject: x86 mmiotrace: use resource_size_t for phys addresses Signed-off-by: Pekka Paalanen Signed-off-by: Ingo Molnar --- kernel/trace/trace_mmiotrace.c | 20 ++++++++++++-------- 1 file changed, 12 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_mmiotrace.c b/kernel/trace/trace_mmiotrace.c index 3c1dacdc2d85..b13dc19dcbb4 100644 --- a/kernel/trace/trace_mmiotrace.c +++ b/kernel/trace/trace_mmiotrace.c @@ -184,20 +184,23 @@ static int mmio_print_rw(struct trace_iterator *iter) switch (entry->mmiorw.opcode) { case MMIO_READ: ret = trace_seq_printf(s, - "R %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", - rw->width, secs, usec_rem, rw->map_id, rw->phys, + "R %d %lu.%06lu %d 0x%llx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, rw->value, rw->pc, 0); break; case MMIO_WRITE: ret = trace_seq_printf(s, - "W %d %lu.%06lu %d 0x%lx 0x%lx 0x%lx %d\n", - rw->width, secs, usec_rem, rw->map_id, rw->phys, + "W %d %lu.%06lu %d 0x%llx 0x%lx 0x%lx %d\n", + rw->width, secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, rw->value, rw->pc, 0); break; case MMIO_UNKNOWN_OP: ret = trace_seq_printf(s, - "UNKNOWN %lu.%06lu %d 0x%lx %02x,%02x,%02x 0x%lx %d\n", - secs, usec_rem, rw->map_id, rw->phys, + "UNKNOWN %lu.%06lu %d 0x%llx %02x,%02x,%02x 0x%lx %d\n", + secs, usec_rem, rw->map_id, + (unsigned long long)rw->phys, (rw->value >> 16) & 0xff, (rw->value >> 8) & 0xff, (rw->value >> 0) & 0xff, rw->pc, 0); break; @@ -223,8 +226,9 @@ static int mmio_print_map(struct trace_iterator *iter) switch (entry->mmiorw.opcode) { case MMIO_PROBE: ret = trace_seq_printf(s, - "MAP %lu.%06lu %d 0x%lx 0x%lx 0x%lx 0x%lx %d\n", - secs, usec_rem, m->map_id, m->phys, m->virt, m->len, + "MAP %lu.%06lu %d 0x%llx 0x%lx 0x%lx 0x%lx %d\n", + secs, usec_rem, m->map_id, + (unsigned long long)m->phys, m->virt, m->len, 0UL, 0); break; case MMIO_UNPROBE: -- cgit v1.2.3 From 4d2df795f0c3eb91f97a666f47716121a2f166ed Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Sat, 24 May 2008 15:00:46 +0200 Subject: sysprof: make it depend on X86 Signed-off-by: Thomas Gleixner --- kernel/trace/Kconfig | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig index e101c9a85f0f..263e9e6bbd60 100644 --- a/kernel/trace/Kconfig +++ b/kernel/trace/Kconfig @@ -77,7 +77,7 @@ config PREEMPT_TRACER config SYSPROF_TRACER bool "Sysprof Tracer" - depends on DEBUG_KERNEL + depends on X86 select TRACING help This tracer provides the trace needed by the 'Sysprof' userspace -- cgit v1.2.3 From c6531cce6e6e4b99bcda46b6268d6f2d9e30aea4 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 12 May 2008 21:21:14 +0200 Subject: sched: do not trace sched_clock The tracer uses sched_clock, so do not trace it. Signed-off-by: Ingo Molnar Signed-off-by: Thomas Gleixner --- kernel/sched.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index e2e985eeee78..6590a828138f 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -884,12 +884,12 @@ static unsigned long long __cpu_clock(int cpu) * For kernel-internal use: high-speed (but slightly incorrect) per-cpu * clock constructed from sched_clock(): */ -unsigned long long cpu_clock(int cpu) +unsigned long long notrace cpu_clock(int cpu) { unsigned long long prev_cpu_time, time, delta_time; unsigned long flags; - local_irq_save(flags); + raw_local_irq_save(flags); prev_cpu_time = per_cpu(prev_cpu_time, cpu); time = __cpu_clock(cpu) + per_cpu(time_offset, cpu); delta_time = time-prev_cpu_time; @@ -898,7 +898,7 @@ unsigned long long cpu_clock(int cpu) time = __sync_cpu_clock(time, cpu); per_cpu(prev_cpu_time, cpu) = time; } - local_irq_restore(flags); + raw_local_irq_restore(flags); return time; } -- cgit v1.2.3 From 19384c0314342222b18d4c7f09cdce1ca74dfd2a Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 00:22:16 -0400 Subject: ftrace: limit use of check pages The check_pages function is called often enough that it can cause problems with trace outputs or even bringing the system to a halt. This patch limits the check_pages to the places that are most likely to have problems. The check is made at the flip between the global array and the max save array, as well as when the size of the buffers changes and the self tests. This patch also removes the BUG_ON from check_pages and replaces it with a WARN_ON and disabling of the tracer. Signed-off-by: Steven Rostedt Cc: pq@iki.fi Cc: proski@gnu.org Cc: sandmann@redhat.com Cc: a.p.zijlstra@chello.nl Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 32 +++++++++++++++++++++++--------- kernel/trace/trace_selftest.c | 1 + 2 files changed, 24 insertions(+), 9 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 3271916ff033..0567f51bbea4 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -249,24 +249,32 @@ __update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu) tracing_record_cmdline(current); } +#define CHECK_COND(cond) \ + if (unlikely(cond)) { \ + tracing_disabled = 1; \ + WARN_ON(1); \ + return -1; \ + } + /** * check_pages - integrity check of trace buffers * * As a safty measure we check to make sure the data pages have not - * been corrupted. TODO: configure to disable this because it adds - * a bit of overhead. + * been corrupted. */ -void check_pages(struct trace_array_cpu *data) +int check_pages(struct trace_array_cpu *data) { struct page *page, *tmp; - BUG_ON(data->trace_pages.next->prev != &data->trace_pages); - BUG_ON(data->trace_pages.prev->next != &data->trace_pages); + CHECK_COND(data->trace_pages.next->prev != &data->trace_pages); + CHECK_COND(data->trace_pages.prev->next != &data->trace_pages); list_for_each_entry_safe(page, tmp, &data->trace_pages, lru) { - BUG_ON(page->lru.next->prev != &page->lru); - BUG_ON(page->lru.prev->next != &page->lru); + CHECK_COND(page->lru.next->prev != &page->lru); + CHECK_COND(page->lru.prev->next != &page->lru); } + + return 0; } /** @@ -280,7 +288,6 @@ void *head_page(struct trace_array_cpu *data) { struct page *page; - check_pages(data); if (list_empty(&data->trace_pages)) return NULL; @@ -2566,7 +2573,7 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, { unsigned long val; char buf[64]; - int ret; + int i, ret; if (cnt >= sizeof(buf)) return -EINVAL; @@ -2635,8 +2642,15 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, trace_free_page(); } + /* check integrity */ + for_each_tracing_cpu(i) + check_pages(global_trace.data[i]); + filp->f_pos += cnt; + /* If check pages failed, return ENOMEM */ + if (tracing_disabled) + cnt = -ENOMEM; out: max_tr.entries = global_trace.entries; mutex_unlock(&trace_types_lock); diff --git a/kernel/trace/trace_selftest.c b/kernel/trace/trace_selftest.c index 3877dd9102f1..18c5423bc977 100644 --- a/kernel/trace/trace_selftest.c +++ b/kernel/trace/trace_selftest.c @@ -28,6 +28,7 @@ trace_test_buffer_cpu(struct trace_array *tr, struct trace_array_cpu *data) page = list_entry(data->trace_pages.next, struct page, lru); entries = page_address(page); + check_pages(data); if (head_page(data) != entries) goto failed; -- cgit v1.2.3 From 4902f8849da6d2805bd291551a6dfd48f1b4f604 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 00:22:18 -0400 Subject: ftrace: move ftrace_special to trace.c Move the ftrace_special out of sched_switch to trace.c. Signed-off-by: Steven Rostedt Cc: pq@iki.fi Cc: proski@gnu.org Cc: sandmann@redhat.com Cc: a.p.zijlstra@chello.nl Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 27 +++++++++++++++++++++++++-- kernel/trace/trace_sched_switch.c | 24 ------------------------ 2 files changed, 25 insertions(+), 26 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0567f51bbea4..583fe24903d3 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -941,6 +941,30 @@ tracing_sched_wakeup_trace(struct trace_array *tr, trace_wake_up(); } +void +ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) +{ + struct trace_array *tr = &global_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int cpu; + + if (tracing_disabled || current_trace == &no_tracer || !tr->ctrl) + return; + + local_irq_save(flags); + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + + if (likely(disabled == 1)) + __trace_special(tr, data, arg1, arg2, arg3); + + atomic_dec(&data->disabled); + local_irq_restore(flags); +} + #ifdef CONFIG_FTRACE static void function_trace_call(unsigned long ip, unsigned long parent_ip) @@ -2941,8 +2965,6 @@ __init static int tracer_alloc_buffers(void) int ret = -ENOMEM; int i; - global_trace.ctrl = tracer_enabled; - /* TODO: make the number of buffers hot pluggable with CPUS */ tracing_nr_buffers = num_possible_cpus(); tracing_buffer_mask = cpu_possible_map; @@ -3012,6 +3034,7 @@ __init static int tracer_alloc_buffers(void) current_trace = &no_tracer; /* All seems OK, enable tracing */ + global_trace.ctrl = tracer_enabled; tracing_disabled = 0; return 0; diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index d25ffa5eaf2b..798ec0dc863c 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -125,30 +125,6 @@ wake_up_callback(void *probe_data, void *call_data, wakeup_func(probe_data, __rq, task, curr); } -void -ftrace_special(unsigned long arg1, unsigned long arg2, unsigned long arg3) -{ - struct trace_array *tr = ctx_trace; - struct trace_array_cpu *data; - unsigned long flags; - long disabled; - int cpu; - - if (!tracer_enabled) - return; - - local_irq_save(flags); - cpu = raw_smp_processor_id(); - data = tr->data[cpu]; - disabled = atomic_inc_return(&data->disabled); - - if (likely(disabled == 1)) - __trace_special(tr, data, arg1, arg2, arg3); - - atomic_dec(&data->disabled); - local_irq_restore(flags); -} - static void sched_switch_reset(struct trace_array *tr) { int cpu; -- cgit v1.2.3 From 7e18d8e701b6798a5df11e0a16881a60ab1018b6 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 00:22:19 -0400 Subject: ftrace: add function tracing to wake up tracing This patch adds function tracing to the functions that are called on the CPU of the task being traced. Signed-off-by: Steven Rostedt Cc: pq@iki.fi Cc: proski@gnu.org Cc: sandmann@redhat.com Cc: a.p.zijlstra@chello.nl Signed-off-by: Thomas Gleixner --- kernel/trace/trace_sched_wakeup.c | 67 ++++++++++++++++++++++++++++++++++++++- 1 file changed, 66 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index 5d2fb48e47f8..bf7e91caef57 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -30,6 +30,69 @@ static DEFINE_SPINLOCK(wakeup_lock); static void __wakeup_reset(struct trace_array *tr); +#ifdef CONFIG_FTRACE +/* + * irqsoff uses its own tracer function to keep the overhead down: + */ +static void +wakeup_tracer_call(unsigned long ip, unsigned long parent_ip) +{ + struct trace_array *tr = wakeup_trace; + struct trace_array_cpu *data; + unsigned long flags; + long disabled; + int resched; + int cpu; + + if (likely(!wakeup_task)) + return; + + resched = need_resched(); + preempt_disable_notrace(); + + cpu = raw_smp_processor_id(); + data = tr->data[cpu]; + disabled = atomic_inc_return(&data->disabled); + if (unlikely(disabled != 1)) + goto out; + + spin_lock_irqsave(&wakeup_lock, flags); + + if (unlikely(!wakeup_task)) + goto unlock; + + /* + * The task can't disappear because it needs to + * wake up first, and we have the wakeup_lock. + */ + if (task_cpu(wakeup_task) != cpu) + goto unlock; + + trace_function(tr, data, ip, parent_ip, flags); + + unlock: + spin_unlock_irqrestore(&wakeup_lock, flags); + + out: + atomic_dec(&data->disabled); + + /* + * To prevent recursion from the scheduler, if the + * resched flag was set before we entered, then + * don't reschedule. + */ + if (resched) + preempt_enable_no_resched_notrace(); + else + preempt_enable_notrace(); +} + +static struct ftrace_ops trace_ops __read_mostly = +{ + .func = wakeup_tracer_call, +}; +#endif /* CONFIG_FTRACE */ + /* * Should this new latency be reported/recorded? */ @@ -73,7 +136,7 @@ wakeup_sched_switch(void *private, void *rq, struct task_struct *prev, if (next != wakeup_task) return; - /* The task we are waitng for is waking up */ + /* The task we are waiting for is waking up */ data = tr->data[wakeup_cpu]; /* disable local data, not wakeup_cpu data */ @@ -290,6 +353,7 @@ static void start_wakeup_tracer(struct trace_array *tr) smp_wmb(); tracer_enabled = 1; + register_ftrace_function(&trace_ops); return; fail_deprobe_wake_new: @@ -305,6 +369,7 @@ fail_deprobe: static void stop_wakeup_tracer(struct trace_array *tr) { tracer_enabled = 0; + unregister_ftrace_function(&trace_ops); marker_probe_unregister("kernel_sched_schedule", sched_switch_callback, &wakeup_trace); -- cgit v1.2.3 From da89a7a2536c46e76a1a4351a70a8b8417e5fed1 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 00:22:20 -0400 Subject: ftrace: remove printks from irqsoff trace Printing out new max latencies was fine for the old RT tracer. But for mainline it is a bit messy. We also need to test if the run queue is locked before we can do the print. This means that we may not be printing out latencies if the run queue is locked on another CPU. This produces inconsistencies in the output. This patch simply removes the print altogether. Signed-off-by: Steven Rostedt Cc: pq@iki.fi Cc: proski@gnu.org Cc: sandmann@redhat.com Cc: a.p.zijlstra@chello.nl Signed-off-by: Thomas Gleixner --- kernel/trace/trace_irqsoff.c | 16 ---------------- 1 file changed, 16 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index 761f3ec66c50..421d6fe3650e 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -165,22 +165,6 @@ check_critical_timing(struct trace_array *tr, update_max_tr_single(tr, current, cpu); - if (!runqueue_is_locked()) { - if (tracing_thresh) { - printk(KERN_INFO "(%16s-%-5d|#%d): %lu us critical" - " section violates %lu us threshold.\n", - current->comm, current->pid, - raw_smp_processor_id(), - latency, nsecs_to_usecs(tracing_thresh)); - } else { - printk(KERN_INFO "(%16s-%-5d|#%d): new %lu us" - " maximum-latency critical section.\n", - current->comm, current->pid, - raw_smp_processor_id(), - latency); - } - } - max_sequence++; out_unlock: -- cgit v1.2.3 From 41c52c0db9607e59f90da7da5309489fa06e887f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 11:46:33 -0400 Subject: ftrace: set_ftrace_notrace feature While debugging latencies in the RT kernel, I found that it would be nice to be able to filter away functions from the trace than just to filter on functions. I added a new interface to the debugfs tracing directory called set_ftrace_notrace When dynamic frace is enabled, this lets you filter away functions that will not be recorded in the trace. It is similar to adding 'notrace' to those functions but by doing it without recompiling the kernel. Here's how set_ftrace_filter and set_ftrace_notrace interact. Remember, if set_ftrace_filter is set, it removes all functions from the trace execpt for those listed in the set_ftrace_filter. set_ftrace_notrace will prevent those functions from being traced. If you were to set one function in both set_ftrace_filter and set_ftrace_notrace and that function was the same, then you would end up with an empty trace. the set of functions to trace is: set_ftrace_filter == empty then all functions not in set_ftrace_notrace else set of the set_ftrace_filter and not in set of set_ftrace_notrace. Signed-off-by: Steven Rostedt Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 170 ++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 130 insertions(+), 40 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 89bd9a6f52ec..2552454609cf 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -170,7 +170,7 @@ static DEFINE_PER_CPU(int, ftrace_shutdown_disable_cpu); static DEFINE_SPINLOCK(ftrace_shutdown_lock); static DEFINE_MUTEX(ftraced_lock); -static DEFINE_MUTEX(ftrace_filter_lock); +static DEFINE_MUTEX(ftrace_regex_lock); struct ftrace_page { struct ftrace_page *next; @@ -337,13 +337,12 @@ static void __ftrace_replace_code(struct dyn_ftrace *rec, unsigned char *old, unsigned char *new, int enable) { - unsigned long ip; + unsigned long ip, fl; int failed; ip = rec->ip; if (ftrace_filtered && enable) { - unsigned long fl; /* * If filtering is on: * @@ -356,13 +355,16 @@ __ftrace_replace_code(struct dyn_ftrace *rec, * If this record is not set to be filtered * and it is not enabled do nothing. * + * If this record is set not to trace then + * do nothing. + * * If this record is not set to be filtered and * it is enabled, disable it. */ fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_ENABLED); if ((fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) || - (fl == 0)) + (fl == 0) || (rec->flags & FTRACE_FL_NOTRACE)) return; /* @@ -380,9 +382,17 @@ __ftrace_replace_code(struct dyn_ftrace *rec, } } else { - if (enable) + if (enable) { + /* + * If this record is set not to trace and is + * not enabled, do nothing. + */ + fl = rec->flags & (FTRACE_FL_NOTRACE | FTRACE_FL_ENABLED); + if (fl == FTRACE_FL_NOTRACE) + return; + new = ftrace_call_replace(ip, FTRACE_ADDR); - else + } else old = ftrace_call_replace(ip, FTRACE_ADDR); if (enable) { @@ -721,6 +731,7 @@ static int __init ftrace_dyn_table_alloc(void) enum { FTRACE_ITER_FILTER = (1 << 0), FTRACE_ITER_CONT = (1 << 1), + FTRACE_ITER_NOTRACE = (1 << 2), }; #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ @@ -754,7 +765,9 @@ t_next(struct seq_file *m, void *v, loff_t *pos) rec = &iter->pg->records[iter->idx++]; if ((rec->flags & FTRACE_FL_FAILED) || ((iter->flags & FTRACE_ITER_FILTER) && - !(rec->flags & FTRACE_FL_FILTER))) { + !(rec->flags & FTRACE_FL_FILTER)) || + ((iter->flags & FTRACE_ITER_NOTRACE) && + !(rec->flags & FTRACE_FL_NOTRACE))) { rec = NULL; goto retry; } @@ -847,22 +860,24 @@ int ftrace_avail_release(struct inode *inode, struct file *file) return 0; } -static void ftrace_filter_reset(void) +static void ftrace_filter_reset(int enable) { struct ftrace_page *pg; struct dyn_ftrace *rec; + unsigned long type = enable ? FTRACE_FL_FILTER : FTRACE_FL_NOTRACE; unsigned i; /* keep kstop machine from running */ preempt_disable(); - ftrace_filtered = 0; + if (enable) + ftrace_filtered = 0; pg = ftrace_pages_start; while (pg) { for (i = 0; i < pg->index; i++) { rec = &pg->records[i]; if (rec->flags & FTRACE_FL_FAILED) continue; - rec->flags &= ~FTRACE_FL_FILTER; + rec->flags &= ~type; } pg = pg->next; } @@ -870,7 +885,7 @@ static void ftrace_filter_reset(void) } static int -ftrace_filter_open(struct inode *inode, struct file *file) +ftrace_regex_open(struct inode *inode, struct file *file, int enable) { struct ftrace_iterator *iter; int ret = 0; @@ -882,15 +897,16 @@ ftrace_filter_open(struct inode *inode, struct file *file) if (!iter) return -ENOMEM; - mutex_lock(&ftrace_filter_lock); + mutex_lock(&ftrace_regex_lock); if ((file->f_mode & FMODE_WRITE) && !(file->f_flags & O_APPEND)) - ftrace_filter_reset(); + ftrace_filter_reset(enable); if (file->f_mode & FMODE_READ) { iter->pg = ftrace_pages_start; iter->pos = -1; - iter->flags = FTRACE_ITER_FILTER; + iter->flags = enable ? FTRACE_ITER_FILTER : + FTRACE_ITER_NOTRACE; ret = seq_open(file, &show_ftrace_seq_ops); if (!ret) { @@ -900,13 +916,25 @@ ftrace_filter_open(struct inode *inode, struct file *file) kfree(iter); } else file->private_data = iter; - mutex_unlock(&ftrace_filter_lock); + mutex_unlock(&ftrace_regex_lock); return ret; } +static int +ftrace_filter_open(struct inode *inode, struct file *file) +{ + return ftrace_regex_open(inode, file, 1); +} + +static int +ftrace_notrace_open(struct inode *inode, struct file *file) +{ + return ftrace_regex_open(inode, file, 0); +} + static ssize_t -ftrace_filter_read(struct file *file, char __user *ubuf, +ftrace_regex_read(struct file *file, char __user *ubuf, size_t cnt, loff_t *ppos) { if (file->f_mode & FMODE_READ) @@ -916,7 +944,7 @@ ftrace_filter_read(struct file *file, char __user *ubuf, } static loff_t -ftrace_filter_lseek(struct file *file, loff_t offset, int origin) +ftrace_regex_lseek(struct file *file, loff_t offset, int origin) { loff_t ret; @@ -936,13 +964,14 @@ enum { }; static void -ftrace_match(unsigned char *buff, int len) +ftrace_match(unsigned char *buff, int len, int enable) { char str[KSYM_SYMBOL_LEN]; char *search = NULL; struct ftrace_page *pg; struct dyn_ftrace *rec; int type = MATCH_FULL; + unsigned long flag = enable ? FTRACE_FL_FILTER : FTRACE_FL_NOTRACE; unsigned i, match = 0, search_len = 0; for (i = 0; i < len; i++) { @@ -966,7 +995,8 @@ ftrace_match(unsigned char *buff, int len) /* keep kstop machine from running */ preempt_disable(); - ftrace_filtered = 1; + if (enable) + ftrace_filtered = 1; pg = ftrace_pages_start; while (pg) { for (i = 0; i < pg->index; i++) { @@ -997,7 +1027,7 @@ ftrace_match(unsigned char *buff, int len) break; } if (matched) - rec->flags |= FTRACE_FL_FILTER; + rec->flags |= flag; } pg = pg->next; } @@ -1005,8 +1035,8 @@ ftrace_match(unsigned char *buff, int len) } static ssize_t -ftrace_filter_write(struct file *file, const char __user *ubuf, - size_t cnt, loff_t *ppos) +ftrace_regex_write(struct file *file, const char __user *ubuf, + size_t cnt, loff_t *ppos, int enable) { struct ftrace_iterator *iter; char ch; @@ -1016,7 +1046,7 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, if (!cnt || cnt < 0) return 0; - mutex_lock(&ftrace_filter_lock); + mutex_lock(&ftrace_regex_lock); if (file->f_mode & FMODE_READ) { struct seq_file *m = file->private_data; @@ -1045,7 +1075,6 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, cnt--; } - if (isspace(ch)) { file->f_pos += read; ret = read; @@ -1072,7 +1101,7 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, if (isspace(ch)) { iter->filtered++; iter->buffer[iter->buffer_idx] = 0; - ftrace_match(iter->buffer, iter->buffer_idx); + ftrace_match(iter->buffer, iter->buffer_idx, enable); iter->buffer_idx = 0; } else iter->flags |= FTRACE_ITER_CONT; @@ -1082,11 +1111,39 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, ret = read; out: - mutex_unlock(&ftrace_filter_lock); + mutex_unlock(&ftrace_regex_lock); return ret; } +static ssize_t +ftrace_filter_write(struct file *file, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + return ftrace_regex_write(file, ubuf, cnt, ppos, 1); +} + +static ssize_t +ftrace_notrace_write(struct file *file, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + return ftrace_regex_write(file, ubuf, cnt, ppos, 0); +} + +static void +ftrace_set_regex(unsigned char *buf, int len, int reset, int enable) +{ + if (unlikely(ftrace_disabled)) + return; + + mutex_lock(&ftrace_regex_lock); + if (reset) + ftrace_filter_reset(enable); + if (buf) + ftrace_match(buf, len, enable); + mutex_unlock(&ftrace_regex_lock); +} + /** * ftrace_set_filter - set a function to filter on in ftrace * @buf - the string that holds the function filter text. @@ -1098,24 +1155,31 @@ ftrace_filter_write(struct file *file, const char __user *ubuf, */ void ftrace_set_filter(unsigned char *buf, int len, int reset) { - if (unlikely(ftrace_disabled)) - return; + ftrace_set_regex(buf, len, reset, 1); +} - mutex_lock(&ftrace_filter_lock); - if (reset) - ftrace_filter_reset(); - if (buf) - ftrace_match(buf, len); - mutex_unlock(&ftrace_filter_lock); +/** + * ftrace_set_notrace - set a function to not trace in ftrace + * @buf - the string that holds the function notrace text. + * @len - the length of the string. + * @reset - non zero to reset all filters before applying this filter. + * + * Notrace Filters denote which functions should not be enabled when tracing + * is enabled. If @buf is NULL and reset is set, all functions will be enabled + * for tracing. + */ +void ftrace_set_notrace(unsigned char *buf, int len, int reset) +{ + ftrace_set_regex(buf, len, reset, 0); } static int -ftrace_filter_release(struct inode *inode, struct file *file) +ftrace_regex_release(struct inode *inode, struct file *file, int enable) { struct seq_file *m = (struct seq_file *)file->private_data; struct ftrace_iterator *iter; - mutex_lock(&ftrace_filter_lock); + mutex_lock(&ftrace_regex_lock); if (file->f_mode & FMODE_READ) { iter = m->private; @@ -1126,7 +1190,7 @@ ftrace_filter_release(struct inode *inode, struct file *file) if (iter->buffer_idx) { iter->filtered++; iter->buffer[iter->buffer_idx] = 0; - ftrace_match(iter->buffer, iter->buffer_idx); + ftrace_match(iter->buffer, iter->buffer_idx, enable); } mutex_lock(&ftrace_sysctl_lock); @@ -1137,10 +1201,22 @@ ftrace_filter_release(struct inode *inode, struct file *file) mutex_unlock(&ftrace_sysctl_lock); kfree(iter); - mutex_unlock(&ftrace_filter_lock); + mutex_unlock(&ftrace_regex_lock); return 0; } +static int +ftrace_filter_release(struct inode *inode, struct file *file) +{ + return ftrace_regex_release(inode, file, 1); +} + +static int +ftrace_notrace_release(struct inode *inode, struct file *file) +{ + return ftrace_regex_release(inode, file, 0); +} + static struct file_operations ftrace_avail_fops = { .open = ftrace_avail_open, .read = seq_read, @@ -1150,12 +1226,20 @@ static struct file_operations ftrace_avail_fops = { static struct file_operations ftrace_filter_fops = { .open = ftrace_filter_open, - .read = ftrace_filter_read, + .read = ftrace_regex_read, .write = ftrace_filter_write, - .llseek = ftrace_filter_lseek, + .llseek = ftrace_regex_lseek, .release = ftrace_filter_release, }; +static struct file_operations ftrace_notrace_fops = { + .open = ftrace_notrace_open, + .read = ftrace_regex_read, + .write = ftrace_notrace_write, + .llseek = ftrace_regex_lseek, + .release = ftrace_notrace_release, +}; + /** * ftrace_force_update - force an update to all recording ftrace functions * @@ -1239,6 +1323,12 @@ static __init int ftrace_init_debugfs(void) if (!entry) pr_warning("Could not create debugfs " "'set_ftrace_filter' entry\n"); + + entry = debugfs_create_file("set_ftrace_notrace", 0644, d_tracer, + NULL, &ftrace_notrace_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'set_ftrace_notrace' entry\n"); return 0; } -- cgit v1.2.3 From 41bc8144d02028133bcd1d545023c6f49e8b2411 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 22 May 2008 11:49:22 -0400 Subject: ftrace: fix up cmdline recording The new work with converting the trace hooks over to markers broke the command line recording of ftrace. This patch fixes it again. Signed-off-by: Steven Rostedt Cc: Peter Zijlstra Cc: Mathieu Desnoyers Signed-off-by: Thomas Gleixner --- kernel/trace/trace.c | 3 --- kernel/trace/trace.h | 4 ++-- kernel/trace/trace_functions.c | 4 ++-- kernel/trace/trace_sched_switch.c | 21 +++++++++++++++------ 4 files changed, 19 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 583fe24903d3..0feae23d9893 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -652,9 +652,6 @@ static char saved_cmdlines[SAVED_CMDLINES][TASK_COMM_LEN]; static int cmdline_idx; static DEFINE_SPINLOCK(trace_cmdline_lock); -/* trace in all context switches */ -atomic_t trace_record_cmdline_enabled __read_mostly; - /* temporary disable recording */ atomic_t trace_record_cmdline_disabled __read_mostly; diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index c460e85e94ed..6b8bd8800d04 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -218,6 +218,8 @@ void trace_function(struct trace_array *tr, void tracing_start_function_trace(void); void tracing_stop_function_trace(void); +void tracing_start_cmdline_record(void); +void tracing_stop_cmdline_record(void); int register_tracer(struct tracer *type); void unregister_tracer(struct tracer *type); @@ -226,8 +228,6 @@ extern unsigned long nsecs_to_usecs(unsigned long nsecs); extern unsigned long tracing_max_latency; extern unsigned long tracing_thresh; -extern atomic_t trace_record_cmdline_enabled; - void update_max_tr(struct trace_array *tr, struct task_struct *tsk, int cpu); void update_max_tr_single(struct trace_array *tr, struct task_struct *tsk, int cpu); diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 0a084656d7cf..7ee7dcd76b7d 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -29,14 +29,14 @@ static void function_reset(struct trace_array *tr) static void start_function_trace(struct trace_array *tr) { function_reset(tr); - atomic_inc(&trace_record_cmdline_enabled); + tracing_start_cmdline_record(); tracing_start_function_trace(); } static void stop_function_trace(struct trace_array *tr) { tracing_stop_function_trace(); - atomic_dec(&trace_record_cmdline_enabled); + tracing_stop_cmdline_record(); } static void function_trace_init(struct trace_array *tr) diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 798ec0dc863c..c16935d3bc5c 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -29,6 +29,9 @@ sched_switch_func(void *private, void *__rq, struct task_struct *prev, long disabled; int cpu; + tracing_record_cmdline(prev); + tracing_record_cmdline(next); + if (!tracer_enabled) return; @@ -63,8 +66,6 @@ sched_switch_callback(void *probe_data, void *call_data, prev = va_arg(*args, typeof(prev)); next = va_arg(*args, typeof(next)); - tracing_record_cmdline(prev); - /* * If tracer_switch_func only points to the local * switch func, it still needs the ptr passed to it. @@ -213,18 +214,26 @@ void tracing_stop_sched_switch(void) tracing_sched_unregister(); } +void tracing_start_cmdline_record(void) +{ + tracing_start_sched_switch(); +} + +void tracing_stop_cmdline_record(void) +{ + tracing_stop_sched_switch(); +} + static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); - atomic_inc(&trace_record_cmdline_enabled); tracer_enabled = 1; - tracing_start_sched_switch(); + tracing_start_cmdline_record(); } static void stop_sched_trace(struct trace_array *tr) { - tracing_stop_sched_switch(); - atomic_dec(&trace_record_cmdline_enabled); + tracing_stop_cmdline_record(); tracer_enabled = 0; } -- cgit v1.2.3 From ffdaa3582b6b39d625d585d07e329ffdc925e971 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sat, 24 May 2008 23:45:02 +0530 Subject: ftrace: safe traversal of ftrace_hash hlist Hi Steven, I noticed that concurrent instances of ftrace_record_ip() have a race between ftrace_hash list traversal during ftrace_ip_in_hash() (before acquiring ftrace_shutdown_lock) and ftrace_add_hash(). If it's so then this should fix it. Signed-off-by: Abhishek Sagar Cc: rostedt@goodmis.org Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 2552454609cf..9b7c54f8a62f 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -201,7 +201,7 @@ ftrace_ip_in_hash(unsigned long ip, unsigned long key) struct hlist_node *t; int found = 0; - hlist_for_each_entry(p, t, &ftrace_hash[key], node) { + hlist_for_each_entry_rcu(p, t, &ftrace_hash[key], node) { if (p->ip == ip) { found = 1; break; @@ -214,7 +214,7 @@ ftrace_ip_in_hash(unsigned long ip, unsigned long key) static inline void ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) { - hlist_add_head(&node->node, &ftrace_hash[key]); + hlist_add_head_rcu(&node->node, &ftrace_hash[key]); } static void ftrace_free_rec(struct dyn_ftrace *rec) -- cgit v1.2.3 From 492a7ea5bcf263ee02a9eb6a3ab0222a1946fade Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sun, 25 May 2008 00:10:04 +0530 Subject: ftrace: fix updating of ftrace_update_cnt Hi Ingo/Steven, Ftrace currently maintains an update count which includes false updates, i.e, updates which failed. If anything, such failures should be tracked by some separate variable, but this patch provides a minimal fix. Signed-off-by: Abhishek Sagar Cc: rostedt@goodmis.org Signed-off-by: Thomas Gleixner --- kernel/trace/ftrace.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 9b7c54f8a62f..1843edc098a6 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -453,7 +453,7 @@ static void ftrace_shutdown_replenish(void) ftrace_pages->next = (void *)get_zeroed_page(GFP_KERNEL); } -static void +static int ftrace_code_disable(struct dyn_ftrace *rec) { unsigned long ip; @@ -469,7 +469,9 @@ ftrace_code_disable(struct dyn_ftrace *rec) if (failed) { rec->flags |= FTRACE_FL_FAILED; ftrace_free_rec(rec); + return 0; } + return 1; } static int __ftrace_modify_code(void *data) @@ -617,8 +619,8 @@ static int __ftrace_update_code(void *ignore) /* all CPUS are stopped, we are safe to modify code */ hlist_for_each_entry(p, t, &head, node) { - ftrace_code_disable(p); - ftrace_update_cnt++; + if (ftrace_code_disable(p)) + ftrace_update_cnt++; } } -- cgit v1.2.3 From 76094a2cf46e4ab776055d4086615b884408568c Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Wed, 28 May 2008 00:03:18 +0530 Subject: ftrace: distinguish kretprobe'd functions in trace logs Tracing functions via ftrace which have a kretprobe installed on them, can produce misleading output in their trace logs. E.g, consider the correct trace of the following sequence: do_IRQ() { ~ irq_enter(); ~ } Trace log (sample): -0 [00] 4154504455.781616: irq_enter <- do_IRQ But if irq_enter() has a kretprobe installed on it, the return value stored on the stack at each invocation is modified to divert the return to a kprobe trampoline function called kretprobe_trampoline(). So with this the trace would (currently) look like: -0 [00] 4154504455.781616: irq_enter <- kretprobe_trampoline Now this is quite misleading to the end user, as it suggests something that didn't actually happen. So just to avoid such misinterpretations, the inlined patch aims to output such a log as: -0 [00] 4154504455.781616: irq_enter <- [unknown/kretprobe'd] Signed-off-by: Abhishek Sagar Acked-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 27 ++++++++++++++++++++++++--- 1 file changed, 24 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 0feae23d9893..12f5e817380e 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -27,6 +27,7 @@ #include #include #include +#include #include #include @@ -1199,6 +1200,20 @@ static void s_stop(struct seq_file *m, void *p) mutex_unlock(&trace_types_lock); } +#define KRETPROBE_MSG "[unknown/kretprobe'd]" + +#ifdef CONFIG_KRETPROBES +static inline int kretprobed(unsigned long addr) +{ + return addr == (unsigned long)kretprobe_trampoline; +} +#else +static inline int kretprobed(unsigned long addr) +{ + return 0; +} +#endif /* CONFIG_KRETPROBES */ + static int seq_print_sym_short(struct trace_seq *s, const char *fmt, unsigned long address) { @@ -1434,7 +1449,10 @@ print_lat_fmt(struct trace_iterator *iter, unsigned int trace_idx, int cpu) case TRACE_FN: seq_print_ip_sym(s, entry->fn.ip, sym_flags); trace_seq_puts(s, " ("); - seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags); + if (kretprobed(entry->fn.parent_ip)) + trace_seq_puts(s, KRETPROBE_MSG); + else + seq_print_ip_sym(s, entry->fn.parent_ip, sym_flags); trace_seq_puts(s, ")\n"); break; case TRACE_CTX: @@ -1514,8 +1532,11 @@ static int print_trace_fmt(struct trace_iterator *iter) ret = trace_seq_printf(s, " <-"); if (!ret) return 0; - ret = seq_print_ip_sym(s, entry->fn.parent_ip, - sym_flags); + if (kretprobed(entry->fn.parent_ip)) + ret = trace_seq_puts(s, KRETPROBE_MSG); + else + ret = seq_print_ip_sym(s, entry->fn.parent_ip, + sym_flags); if (!ret) return 0; } -- cgit v1.2.3 From ad90c0e3ce8d20d6873b57e36181ef6d7a0097fe Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Tue, 27 May 2008 20:48:37 -0400 Subject: ftrace: user update and disable dynamic ftrace daemon In dynamic ftrace, the mcount function starts off pointing to a stub function that just returns. On start up, the call to the stub is modified to point to a "record_ip" function. The job of the record_ip function is to add the function to a pre-allocated hash list. If the function is already there, it simply is ignored, otherwise it is added to the list. Later, a ftraced daemon wakes up and calls kstop_machine if any functions have been recorded, and changes the calls to the recorded functions to a simple nop. If no functions were recorded, the daemon goes back to sleep. The daemon wakes up once a second to see if it needs to update any newly recorded functions into nops. Usually it does not, but if a lot of code has been executed for the first time in the kernel, the ftraced daemon will call kstop_machine to update those into nops. The problem currently is that there's no way to stop the daemon from doing this, and it can cause unneeded latencies (800us which for some is bothersome). This patch adds a new file /debugfs/tracing/ftraced_enabled. If the daemon is active, reading this will return "enabled\n" and "disabled\n" when the daemon is not running. To disable the daemon, the user can echo "0" or "disable" into this file, and "1" or "enable" to re-enable the daemon. Since the daemon is used to convert the functions into nops to increase the performance of the system, I also added that anytime something is written into the ftraced_enabled file, kstop_machine will run if there are new functions that have been detected that need to be converted. This way the user can disable the daemon but still be able to control the conversion of the mcount calls to nops by simply, "echo 0 > /debugfs/tracing/ftraced_enabled" when they need to do more conversions. To see the number of converted functions: "cat /debugfs/tracing/dyn_ftrace_total_info" Signed-off-by: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 157 +++++++++++++++++++++++++++++++++++--------------- 1 file changed, 110 insertions(+), 47 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 1843edc098a6..f762f5a2d331 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -151,8 +151,6 @@ static int __unregister_ftrace_function(struct ftrace_ops *ops) #ifdef CONFIG_DYNAMIC_FTRACE static struct task_struct *ftraced_task; -static DECLARE_WAIT_QUEUE_HEAD(ftraced_waiters); -static unsigned long ftraced_iteration_counter; enum { FTRACE_ENABLE_CALLS = (1 << 0), @@ -189,6 +187,7 @@ static struct ftrace_page *ftrace_pages; static int ftraced_trigger; static int ftraced_suspend; +static int ftraced_stop; static int ftrace_record_suspend; @@ -474,14 +473,21 @@ ftrace_code_disable(struct dyn_ftrace *rec) return 1; } +static int __ftrace_update_code(void *ignore); + static int __ftrace_modify_code(void *data) { unsigned long addr; int *command = data; - if (*command & FTRACE_ENABLE_CALLS) + if (*command & FTRACE_ENABLE_CALLS) { + /* + * Update any recorded ips now that we have the + * machine stopped + */ + __ftrace_update_code(NULL); ftrace_replace_code(1); - else if (*command & FTRACE_DISABLE_CALLS) + } else if (*command & FTRACE_DISABLE_CALLS) ftrace_replace_code(0); if (*command & FTRACE_UPDATE_TRACE_FUNC) @@ -503,6 +509,25 @@ static void ftrace_run_update_code(int command) stop_machine_run(__ftrace_modify_code, &command, NR_CPUS); } +void ftrace_disable_daemon(void) +{ + /* Stop the daemon from calling kstop_machine */ + mutex_lock(&ftraced_lock); + ftraced_stop = 1; + mutex_unlock(&ftraced_lock); + + ftrace_force_update(); +} + +void ftrace_enable_daemon(void) +{ + mutex_lock(&ftraced_lock); + ftraced_stop = 0; + mutex_unlock(&ftraced_lock); + + ftrace_force_update(); +} + static ftrace_func_t saved_ftrace_func; static void ftrace_startup(void) @@ -603,6 +628,7 @@ static int __ftrace_update_code(void *ignore) int i; /* Don't be recording funcs now */ + ftrace_record_suspend++; save_ftrace_enabled = ftrace_enabled; ftrace_enabled = 0; @@ -628,18 +654,23 @@ static int __ftrace_update_code(void *ignore) stop = ftrace_now(raw_smp_processor_id()); ftrace_update_time = stop - start; ftrace_update_tot_cnt += ftrace_update_cnt; + ftraced_trigger = 0; ftrace_enabled = save_ftrace_enabled; + ftrace_record_suspend--; return 0; } -static void ftrace_update_code(void) +static int ftrace_update_code(void) { - if (unlikely(ftrace_disabled)) - return; + if (unlikely(ftrace_disabled) || + !ftrace_enabled || !ftraced_trigger) + return 0; stop_machine_run(__ftrace_update_code, NULL, NR_CPUS); + + return 1; } static int ftraced(void *ignore) @@ -658,14 +689,13 @@ static int ftraced(void *ignore) mutex_lock(&ftrace_sysctl_lock); mutex_lock(&ftraced_lock); - if (ftrace_enabled && ftraced_trigger && !ftraced_suspend) { - ftrace_record_suspend++; - ftrace_update_code(); + if (!ftraced_suspend && !ftraced_stop && + ftrace_update_code()) { usecs = nsecs_to_usecs(ftrace_update_time); if (ftrace_update_tot_cnt > 100000) { ftrace_update_tot_cnt = 0; pr_info("hm, dftrace overflow: %lu change%s" - " (%lu total) in %lu usec%s\n", + " (%lu total) in %lu usec%s\n", ftrace_update_cnt, ftrace_update_cnt != 1 ? "s" : "", ftrace_update_tot_cnt, @@ -673,15 +703,10 @@ static int ftraced(void *ignore) ftrace_disabled = 1; WARN_ON_ONCE(1); } - ftraced_trigger = 0; - ftrace_record_suspend--; } - ftraced_iteration_counter++; mutex_unlock(&ftraced_lock); mutex_unlock(&ftrace_sysctl_lock); - wake_up_interruptible(&ftraced_waiters); - ftrace_shutdown_replenish(); } __set_current_state(TASK_RUNNING); @@ -1219,6 +1244,55 @@ ftrace_notrace_release(struct inode *inode, struct file *file) return ftrace_regex_release(inode, file, 0); } +static ssize_t +ftraced_read(struct file *filp, char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + /* don't worry about races */ + char *buf = ftraced_stop ? "disabled\n" : "enabled\n"; + int r = strlen(buf); + + return simple_read_from_buffer(ubuf, cnt, ppos, buf, r); +} + +static ssize_t +ftraced_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + char buf[64]; + long val; + int ret; + + if (cnt >= sizeof(buf)) + return -EINVAL; + + if (copy_from_user(&buf, ubuf, cnt)) + return -EFAULT; + + if (strncmp(buf, "enable", 6) == 0) + val = 1; + else if (strncmp(buf, "disable", 7) == 0) + val = 0; + else { + buf[cnt] = 0; + + ret = strict_strtoul(buf, 10, &val); + if (ret < 0) + return ret; + + val = !!val; + } + + if (val) + ftrace_enable_daemon(); + else + ftrace_disable_daemon(); + + filp->f_pos += cnt; + + return cnt; +} + static struct file_operations ftrace_avail_fops = { .open = ftrace_avail_open, .read = seq_read, @@ -1242,51 +1316,34 @@ static struct file_operations ftrace_notrace_fops = { .release = ftrace_notrace_release, }; +static struct file_operations ftraced_fops = { + .open = tracing_open_generic, + .read = ftraced_read, + .write = ftraced_write, +}; + /** * ftrace_force_update - force an update to all recording ftrace functions - * - * The ftrace dynamic update daemon only wakes up once a second. - * There may be cases where an update needs to be done immediately - * for tests or internal kernel tracing to begin. This function - * wakes the daemon to do an update and will not return until the - * update is complete. */ int ftrace_force_update(void) { - unsigned long last_counter; - DECLARE_WAITQUEUE(wait, current); int ret = 0; if (unlikely(ftrace_disabled)) return -ENODEV; + mutex_lock(&ftrace_sysctl_lock); mutex_lock(&ftraced_lock); - last_counter = ftraced_iteration_counter; - - set_current_state(TASK_INTERRUPTIBLE); - add_wait_queue(&ftraced_waiters, &wait); - if (unlikely(!ftraced_task)) { - ret = -ENODEV; - goto out; - } - - do { - mutex_unlock(&ftraced_lock); - wake_up_process(ftraced_task); - schedule(); - mutex_lock(&ftraced_lock); - if (signal_pending(current)) { - ret = -EINTR; - break; - } - set_current_state(TASK_INTERRUPTIBLE); - } while (last_counter == ftraced_iteration_counter); + /* + * If ftraced_trigger is not set, then there is nothing + * to update. + */ + if (ftraced_trigger && !ftrace_update_code()) + ret = -EBUSY; - out: mutex_unlock(&ftraced_lock); - remove_wait_queue(&ftraced_waiters, &wait); - set_current_state(TASK_RUNNING); + mutex_unlock(&ftrace_sysctl_lock); return ret; } @@ -1331,6 +1388,12 @@ static __init int ftrace_init_debugfs(void) if (!entry) pr_warning("Could not create debugfs " "'set_ftrace_notrace' entry\n"); + + entry = debugfs_create_file("ftraced_enabled", 0644, d_tracer, + NULL, &ftraced_fops); + if (!entry) + pr_warning("Could not create debugfs " + "'ftraced_enabled' entry\n"); return 0; } -- cgit v1.2.3 From 0eb967012ea15e6e8cfab483d9fa37bc602d400c Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sun, 1 Jun 2008 21:47:30 +0530 Subject: ftrace: prevent freeing of all failed updates Prevent freeing of records which cause problems and correspond to function from core kernel text. A new flag, FTRACE_FL_CONVERTED is used to mark a record as "converted". All other records are patched lazily to NOPs. Failed records now also remain on frace_hash table. Each invocation of ftrace_record_ip now checks whether the traced function has ever been recorded (including past failures) and doesn't re-record it again. Signed-off-by: Abhishek Sagar Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 76 +++++++++++++++++++++++++++++++-------------------- 1 file changed, 46 insertions(+), 30 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index f762f5a2d331..ec54cb7d69d6 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -216,6 +216,12 @@ ftrace_add_hash(struct dyn_ftrace *node, unsigned long key) hlist_add_head_rcu(&node->node, &ftrace_hash[key]); } +/* called from kstop_machine */ +static inline void ftrace_del_hash(struct dyn_ftrace *node) +{ + hlist_del(&node->node); +} + static void ftrace_free_rec(struct dyn_ftrace *rec) { /* no locking, only called from kstop_machine */ @@ -332,12 +338,11 @@ ftrace_record_ip(unsigned long ip) #define FTRACE_ADDR ((long)(ftrace_caller)) #define MCOUNT_ADDR ((long)(mcount)) -static void +static int __ftrace_replace_code(struct dyn_ftrace *rec, unsigned char *old, unsigned char *new, int enable) { unsigned long ip, fl; - int failed; ip = rec->ip; @@ -364,7 +369,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, if ((fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) || (fl == 0) || (rec->flags & FTRACE_FL_NOTRACE)) - return; + return 0; /* * If it is enabled disable it, @@ -388,7 +393,7 @@ __ftrace_replace_code(struct dyn_ftrace *rec, */ fl = rec->flags & (FTRACE_FL_NOTRACE | FTRACE_FL_ENABLED); if (fl == FTRACE_FL_NOTRACE) - return; + return 0; new = ftrace_call_replace(ip, FTRACE_ADDR); } else @@ -396,34 +401,24 @@ __ftrace_replace_code(struct dyn_ftrace *rec, if (enable) { if (rec->flags & FTRACE_FL_ENABLED) - return; + return 0; rec->flags |= FTRACE_FL_ENABLED; } else { if (!(rec->flags & FTRACE_FL_ENABLED)) - return; + return 0; rec->flags &= ~FTRACE_FL_ENABLED; } } - failed = ftrace_modify_code(ip, old, new); - if (failed) { - unsigned long key; - /* It is possible that the function hasn't been converted yet */ - key = hash_long(ip, FTRACE_HASHBITS); - if (!ftrace_ip_in_hash(ip, key)) { - rec->flags |= FTRACE_FL_FAILED; - ftrace_free_rec(rec); - } - - } + return ftrace_modify_code(ip, old, new); } static void ftrace_replace_code(int enable) { + int i, failed; unsigned char *new = NULL, *old = NULL; struct dyn_ftrace *rec; struct ftrace_page *pg; - int i; if (enable) old = ftrace_nop_replace(); @@ -438,7 +433,15 @@ static void ftrace_replace_code(int enable) if (rec->flags & FTRACE_FL_FAILED) continue; - __ftrace_replace_code(rec, old, new, enable); + failed = __ftrace_replace_code(rec, old, new, enable); + if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { + rec->flags |= FTRACE_FL_FAILED; + if ((system_state == SYSTEM_BOOTING) || + !kernel_text_address(rec->ip)) { + ftrace_del_hash(rec); + ftrace_free_rec(rec); + } + } } } } @@ -467,7 +470,6 @@ ftrace_code_disable(struct dyn_ftrace *rec) failed = ftrace_modify_code(ip, call, nop); if (failed) { rec->flags |= FTRACE_FL_FAILED; - ftrace_free_rec(rec); return 0; } return 1; @@ -621,8 +623,7 @@ unsigned long ftrace_update_tot_cnt; static int __ftrace_update_code(void *ignore) { struct dyn_ftrace *p; - struct hlist_head head; - struct hlist_node *t; + struct hlist_node *t, *n; int save_ftrace_enabled; cycle_t start, stop; int i; @@ -637,18 +638,33 @@ static int __ftrace_update_code(void *ignore) /* No locks needed, the machine is stopped! */ for (i = 0; i < FTRACE_HASHSIZE; i++) { - if (hlist_empty(&ftrace_hash[i])) - continue; + /* all CPUS are stopped, we are safe to modify code */ + hlist_for_each_entry_safe(p, t, n, &ftrace_hash[i], node) { + /* Skip over failed records which have not been + * freed. */ + if (p->flags & FTRACE_FL_FAILED) + continue; - head = ftrace_hash[i]; - INIT_HLIST_HEAD(&ftrace_hash[i]); + /* Unconverted records are always at the head of the + * hash bucket. Once we encounter a converted record, + * simply skip over to the next bucket. Saves ftraced + * some processor cycles (ftrace does its bid for + * global warming :-p ). */ + if (p->flags & (FTRACE_FL_CONVERTED)) + break; - /* all CPUS are stopped, we are safe to modify code */ - hlist_for_each_entry(p, t, &head, node) { - if (ftrace_code_disable(p)) + if (ftrace_code_disable(p)) { + p->flags |= FTRACE_FL_CONVERTED; ftrace_update_cnt++; - } + } else { + if ((system_state == SYSTEM_BOOTING) || + !kernel_text_address(p->ip)) { + ftrace_del_hash(p); + ftrace_free_rec(p); + } + } + } } stop = ftrace_now(raw_smp_processor_id()); -- cgit v1.2.3 From 1d74f2a0f64b4091e5e91b55ac1b17dff93f4b59 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sun, 1 Jun 2008 21:47:42 +0530 Subject: ftrace: remove ftrace_ip_converted() Remove the unneeded function ftrace_ip_converted(). Signed-off-by: Abhishek Sagar Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 7 ------- 1 file changed, 7 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index ec54cb7d69d6..a8929e4c77c1 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -306,13 +306,6 @@ ftrace_record_ip(unsigned long ip) if (ftrace_ip_in_hash(ip, key)) goto out_unlock; - /* - * There's a slight race that the ftraced will update the - * hash and reset here. If it is already converted, skip it. - */ - if (ftrace_ip_converted(ip)) - goto out_unlock; - node = ftrace_alloc_dyn_node(ip); if (!node) goto out_unlock; -- cgit v1.2.3 From eb9a7bf09172f409c10ec9560adeea95bb4045f5 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sun, 1 Jun 2008 21:47:54 +0530 Subject: ftrace: add debugfs entry 'failures' Identify functions which had their mcount call-site updates failed. This can help us track functions which ftrace shouldn't fiddle with, and are thus not being traced. If there is no race with any external agent which is modifying the mcount call-site, then this file displays no entries (normal case). Signed-off-by: Abhishek Sagar Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 40 +++++++++++++++++++++++++++++++++++++++- 1 file changed, 39 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index a8929e4c77c1..ad568c742bfc 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -768,6 +768,7 @@ enum { FTRACE_ITER_FILTER = (1 << 0), FTRACE_ITER_CONT = (1 << 1), FTRACE_ITER_NOTRACE = (1 << 2), + FTRACE_ITER_FAILURES = (1 << 3), }; #define FTRACE_BUFF_MAX (KSYM_SYMBOL_LEN+4) /* room for wildcards */ @@ -799,9 +800,16 @@ t_next(struct seq_file *m, void *v, loff_t *pos) } } else { rec = &iter->pg->records[iter->idx++]; - if ((rec->flags & FTRACE_FL_FAILED) || + if ((!(iter->flags & FTRACE_ITER_FAILURES) && + (rec->flags & FTRACE_FL_FAILED)) || + + ((iter->flags & FTRACE_ITER_FAILURES) && + (!(rec->flags & FTRACE_FL_FAILED) || + (rec->flags & FTRACE_FL_FREE))) || + ((iter->flags & FTRACE_ITER_FILTER) && !(rec->flags & FTRACE_FL_FILTER)) || + ((iter->flags & FTRACE_ITER_NOTRACE) && !(rec->flags & FTRACE_FL_NOTRACE))) { rec = NULL; @@ -896,6 +904,24 @@ int ftrace_avail_release(struct inode *inode, struct file *file) return 0; } +static int +ftrace_failures_open(struct inode *inode, struct file *file) +{ + int ret; + struct seq_file *m; + struct ftrace_iterator *iter; + + ret = ftrace_avail_open(inode, file); + if (!ret) { + m = (struct seq_file *)file->private_data; + iter = (struct ftrace_iterator *)m->private; + iter->flags = FTRACE_ITER_FAILURES; + } + + return ret; +} + + static void ftrace_filter_reset(int enable) { struct ftrace_page *pg; @@ -1309,6 +1335,13 @@ static struct file_operations ftrace_avail_fops = { .release = ftrace_avail_release, }; +static struct file_operations ftrace_failures_fops = { + .open = ftrace_failures_open, + .read = seq_read, + .llseek = seq_lseek, + .release = ftrace_avail_release, +}; + static struct file_operations ftrace_filter_fops = { .open = ftrace_filter_open, .read = ftrace_regex_read, @@ -1386,6 +1419,11 @@ static __init int ftrace_init_debugfs(void) pr_warning("Could not create debugfs " "'available_filter_functions' entry\n"); + entry = debugfs_create_file("failures", 0444, + d_tracer, NULL, &ftrace_failures_fops); + if (!entry) + pr_warning("Could not create debugfs 'failures' entry\n"); + entry = debugfs_create_file("set_ftrace_filter", 0644, d_tracer, NULL, &ftrace_filter_fops); if (!entry) -- cgit v1.2.3 From 34078a5e44db3cbed2e0ed580c29a39d94e0cd97 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Tue, 3 Jun 2008 08:33:41 +0530 Subject: ftrace: prevent freeing of all failed updates Steven Rostedt wrote: > If we unload a module and reload it, will it ever get converted again? The intent was always to filter core kernel functions to prevent their freeing. Here's a fix which should allow re-recording of module call-sites. Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index ad568c742bfc..0118979e211f 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -430,7 +430,7 @@ static void ftrace_replace_code(int enable) if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { rec->flags |= FTRACE_FL_FAILED; if ((system_state == SYSTEM_BOOTING) || - !kernel_text_address(rec->ip)) { + !core_kernel_text(rec->ip)) { ftrace_del_hash(rec); ftrace_free_rec(rec); } @@ -651,10 +651,9 @@ static int __ftrace_update_code(void *ignore) ftrace_update_cnt++; } else { if ((system_state == SYSTEM_BOOTING) || - !kernel_text_address(p->ip)) { + !core_kernel_text(p->ip)) { ftrace_del_hash(p); ftrace_free_rec(p); - } } } -- cgit v1.2.3 From 040ec23d07f95285e9777a85cda29cb339a3065b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Mon, 9 Jun 2008 01:45:29 -0700 Subject: sched: sched_clock() lockdep fix Sitsofe Wheeler bisected the following commit to cause a lockdep to warn about itself and turn itself off: > commit c6531cce6e6e4b99bcda46b6268d6f2d9e30aea4 > Author: Ingo Molnar > Date: Mon May 12 21:21:14 2008 +0200 > > sched: do not trace sched_clock do not use raw irq flags in cpu_clock() as it causes lockdep to lose track of the true state of the IRQ flag. Reported-and-bisected-by: Sitsofe Wheeler Signed-off-by: Ingo Molnar Signed-off-by: Andrew Morton --- kernel/sched.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched.c b/kernel/sched.c index 6590a828138f..b8c9fe676221 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -889,7 +889,7 @@ unsigned long long notrace cpu_clock(int cpu) unsigned long long prev_cpu_time, time, delta_time; unsigned long flags; - raw_local_irq_save(flags); + local_irq_save(flags); prev_cpu_time = per_cpu(prev_cpu_time, cpu); time = __cpu_clock(cpu) + per_cpu(time_offset, cpu); delta_time = time-prev_cpu_time; @@ -898,7 +898,7 @@ unsigned long long notrace cpu_clock(int cpu) time = __sync_cpu_clock(time, cpu); per_cpu(prev_cpu_time, cpu) = time; } - raw_local_irq_restore(flags); + local_irq_restore(flags); return time; } -- cgit v1.2.3 From 2b1bce1787700768cbc87c8509851c6f49d252dc Mon Sep 17 00:00:00 2001 From: Ankita Garg Date: Mon, 9 Jun 2008 14:10:25 +0530 Subject: ftrace: disable tracing when current_tracer is set to "none" Found that inspite of setting the current_tracer to "none", trace from the previous trace type continued to be collected. The patch below fixes this and causes the trace to be disabled when the "none" type is selected. Compile and boot tested the patch for functionality. Signed-off-by: Ankita Garg Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 22 +++++++++++++++++----- 1 file changed, 17 insertions(+), 5 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 12f5e817380e..dde6f0ace6dc 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -43,11 +43,6 @@ static cpumask_t __read_mostly tracing_buffer_mask; #define for_each_tracing_cpu(cpu) \ for_each_cpu_mask(cpu, tracing_buffer_mask) -/* dummy trace to disable tracing */ -static struct tracer no_tracer __read_mostly = { - .name = "none", -}; - static int trace_alloc_page(void); static int trace_free_page(void); @@ -135,6 +130,23 @@ static DECLARE_WAIT_QUEUE_HEAD(trace_wait); /* trace_flags holds iter_ctrl options */ unsigned long trace_flags = TRACE_ITER_PRINT_PARENT; +static notrace void no_trace_init(struct trace_array *tr) +{ + int cpu; + + if(tr->ctrl) + for_each_online_cpu(cpu) + tracing_reset(tr->data[cpu]); + tracer_enabled = 0; +} + +/* dummy trace to disable tracing */ +static struct tracer no_tracer __read_mostly = { + .name = "none", + .init = no_trace_init +}; + + /** * trace_wake_up - wake up tasks waiting for trace input * -- cgit v1.2.3 From 20764ff1efb440640353053ec83263e69e1259e0 Mon Sep 17 00:00:00 2001 From: Jiri Slaby Date: Thu, 12 Jun 2008 11:27:03 +0200 Subject: ftrace: fix printout Do not print loglevel before "entries of %ld bytes". Move it to the previous pr_info. Signed-off-by: Jiri Slaby Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index dde6f0ace6dc..6e9dae7eb418 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -3051,9 +3051,8 @@ __init static int tracer_alloc_buffers(void) } max_tr.entries = global_trace.entries; - pr_info("tracer: %d pages allocated for %ld", - pages, trace_nr_entries); - pr_info(" entries of %ld bytes\n", (long)TRACE_ENTRY_SIZE); + pr_info("tracer: %d pages allocated for %ld entries of %ld bytes\n", + pages, trace_nr_entries, (long)TRACE_ENTRY_SIZE); pr_info(" actual entries %ld\n", global_trace.entries); tracer_init_debugfs(); -- cgit v1.2.3 From a4500b84c51645bbc86be3ca84f2252b7ada060f Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sat, 14 Jun 2008 11:59:39 +0530 Subject: ftrace: fix "notrace" filtering priority This is a fix to give notrace filter rules priority over "set_ftrace_filter" rules. This fix ensures that functions which are set to be filtered and are concurrently marked as "notrace" don't get recorded. As of now, if a record is marked as FTRACE_FL_FILTER and is enabled, then the notrace flag is not checked. Tested on x86-32. Signed-off-by: Abhishek Sagar Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 0118979e211f..b532e4a68c74 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -355,20 +355,26 @@ __ftrace_replace_code(struct dyn_ftrace *rec, * If this record is set not to trace then * do nothing. * + * If this record is set not to trace and + * it is enabled then disable it. + * * If this record is not set to be filtered and * it is enabled, disable it. */ - fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_ENABLED); + + fl = rec->flags & (FTRACE_FL_FILTER | FTRACE_FL_NOTRACE | + FTRACE_FL_ENABLED); if ((fl == (FTRACE_FL_FILTER | FTRACE_FL_ENABLED)) || - (fl == 0) || (rec->flags & FTRACE_FL_NOTRACE)) + (fl == (FTRACE_FL_FILTER | FTRACE_FL_NOTRACE)) || + !fl || (fl == FTRACE_FL_NOTRACE)) return 0; /* * If it is enabled disable it, * otherwise enable it! */ - if (fl == FTRACE_FL_ENABLED) { + if (fl & FTRACE_FL_ENABLED) { /* swap new and old */ new = old; old = ftrace_call_replace(ip, FTRACE_ADDR); -- cgit v1.2.3 From f22529351f7060d61eff3b76d7c9706f90aaedf3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 22 May 2008 10:37:48 +0200 Subject: namespacecheck: fixes Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 2 +- kernel/trace/trace_sched_switch.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index b532e4a68c74..0d5bcf69952d 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -50,7 +50,7 @@ static struct ftrace_ops ftrace_list_end __read_mostly = static struct ftrace_ops *ftrace_list __read_mostly = &ftrace_list_end; ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub; -void ftrace_list_func(unsigned long ip, unsigned long parent_ip) +static void ftrace_list_func(unsigned long ip, unsigned long parent_ip) { struct ftrace_ops *op = ftrace_list; diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index c16935d3bc5c..93a662009151 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -196,7 +196,7 @@ static void tracing_sched_unregister(void) &ctx_trace); } -void tracing_start_sched_switch(void) +static void tracing_start_sched_switch(void) { long ref; @@ -205,7 +205,7 @@ void tracing_start_sched_switch(void) tracing_sched_register(); } -void tracing_stop_sched_switch(void) +static void tracing_stop_sched_switch(void) { long ref; -- cgit v1.2.3 From 395a59d0f8e86bb39cd700c3d185d30c670bb958 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sat, 21 Jun 2008 23:47:27 +0530 Subject: ftrace: store mcount address in rec->ip Record the address of the mcount call-site. Currently all archs except sparc64 record the address of the instruction following the mcount call-site. Some general cleanups are entailed. Storing mcount addresses in rec->ip enables looking them up in the kprobe hash table later on to check if they're kprobe'd. Signed-off-by: Abhishek Sagar Cc: davem@davemloft.net Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 0d5bcf69952d..f1e9e5c74e64 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -27,6 +27,8 @@ #include #include +#include + #include "trace.h" /* ftrace_enabled is a method to turn ftrace on or off */ @@ -329,7 +331,6 @@ ftrace_record_ip(unsigned long ip) } #define FTRACE_ADDR ((long)(ftrace_caller)) -#define MCOUNT_ADDR ((long)(mcount)) static int __ftrace_replace_code(struct dyn_ftrace *rec, -- cgit v1.2.3 From ecea656d1d5e912d2f3d332657ea4a6d8380f891 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sat, 21 Jun 2008 23:47:53 +0530 Subject: ftrace: freeze kprobe'd records Let records identified as being kprobe'd be marked as "frozen". The trouble with records which have a kprobe installed on their mcount call-site is that they don't get updated. So if such a function which is currently being traced gets its tracing disabled due to a new filter rule (or because it was added to the notrace list) then it won't be updated and continue being traced. This patch allows scanning of all frozen records during tracing to check if they should be traced. Signed-off-by: Abhishek Sagar Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 72 ++++++++++++++++++++++++++++++++++++++++++++++++++- kernel/trace/trace.c | 3 +++ 2 files changed, 74 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index f1e9e5c74e64..d1238163155f 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -163,6 +163,8 @@ enum { }; static int ftrace_filtered; +static int tracing_on; +static int frozen_record_count; static struct hlist_head ftrace_hash[FTRACE_HASHSIZE]; @@ -195,6 +197,71 @@ static int ftrace_record_suspend; static struct dyn_ftrace *ftrace_free_records; + +#ifdef CONFIG_KPROBES +static inline void freeze_record(struct dyn_ftrace *rec) +{ + if (!(rec->flags & FTRACE_FL_FROZEN)) { + rec->flags |= FTRACE_FL_FROZEN; + frozen_record_count++; + } +} + +static inline void unfreeze_record(struct dyn_ftrace *rec) +{ + if (rec->flags & FTRACE_FL_FROZEN) { + rec->flags &= ~FTRACE_FL_FROZEN; + frozen_record_count--; + } +} + +static inline int record_frozen(struct dyn_ftrace *rec) +{ + return rec->flags & FTRACE_FL_FROZEN; +} +#else +# define freeze_record(rec) ({ 0; }) +# define unfreeze_record(rec) ({ 0; }) +# define record_frozen(rec) ({ 0; }) +#endif /* CONFIG_KPROBES */ + +int skip_trace(unsigned long ip) +{ + unsigned long fl; + struct dyn_ftrace *rec; + struct hlist_node *t; + struct hlist_head *head; + + if (frozen_record_count == 0) + return 0; + + head = &ftrace_hash[hash_long(ip, FTRACE_HASHBITS)]; + hlist_for_each_entry_rcu(rec, t, head, node) { + if (rec->ip == ip) { + if (record_frozen(rec)) { + if (rec->flags & FTRACE_FL_FAILED) + return 1; + + if (!(rec->flags & FTRACE_FL_CONVERTED)) + return 1; + + if (!tracing_on || !ftrace_enabled) + return 1; + + if (ftrace_filtered) { + fl = rec->flags & (FTRACE_FL_FILTER | + FTRACE_FL_NOTRACE); + if (!fl || (fl & FTRACE_FL_NOTRACE)) + return 1; + } + } + break; + } + } + + return 0; +} + static inline int ftrace_ip_in_hash(unsigned long ip, unsigned long key) { @@ -489,8 +556,11 @@ static int __ftrace_modify_code(void *data) */ __ftrace_update_code(NULL); ftrace_replace_code(1); - } else if (*command & FTRACE_DISABLE_CALLS) + tracing_on = 1; + } else if (*command & FTRACE_DISABLE_CALLS) { ftrace_replace_code(0); + tracing_on = 0; + } if (*command & FTRACE_UPDATE_TRACE_FUNC) ftrace_update_ftrace_func(ftrace_trace_function); diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 6e9dae7eb418..9ade79369bfb 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -988,6 +988,9 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) if (unlikely(!tracer_enabled)) return; + if (skip_trace(ip)) + return; + local_irq_save(flags); cpu = raw_smp_processor_id(); data = tr->data[cpu]; -- cgit v1.2.3 From f22f9a89ce6857d377bf22dba4c1a8cd256c5136 Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Sat, 21 Jun 2008 23:50:29 +0530 Subject: ftrace: avoid modifying kprobe'd records Avoid modifying the mcount call-site if there is a kprobe installed on it. These records are not marked as failed however. This allowed the filter rules on them to remain up-to-date. Whenever the kprobe on the corresponding record is removed, the record gets updated as normal. Signed-off-by: Abhishek Sagar Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 35 +++++++++++++++++++++++++++++++---- 1 file changed, 31 insertions(+), 4 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index d1238163155f..85e841335417 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -21,6 +21,7 @@ #include #include #include +#include #include #include #include @@ -500,6 +501,10 @@ static void ftrace_replace_code(int enable) if (rec->flags & FTRACE_FL_FAILED) continue; + /* ignore updates to this record's mcount site */ + if (get_kprobe((void *)rec->ip)) + continue; + failed = __ftrace_replace_code(rec, old, new, enable); if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { rec->flags |= FTRACE_FL_FAILED; @@ -692,11 +697,11 @@ unsigned long ftrace_update_tot_cnt; static int __ftrace_update_code(void *ignore) { + int i, save_ftrace_enabled; + cycle_t start, stop; struct dyn_ftrace *p; struct hlist_node *t, *n; - int save_ftrace_enabled; - cycle_t start, stop; - int i; + struct hlist_head *head, temp_list; /* Don't be recording funcs now */ ftrace_record_suspend++; @@ -708,8 +713,11 @@ static int __ftrace_update_code(void *ignore) /* No locks needed, the machine is stopped! */ for (i = 0; i < FTRACE_HASHSIZE; i++) { + INIT_HLIST_HEAD(&temp_list); + head = &ftrace_hash[i]; + /* all CPUS are stopped, we are safe to modify code */ - hlist_for_each_entry_safe(p, t, n, &ftrace_hash[i], node) { + hlist_for_each_entry_safe(p, t, n, head, node) { /* Skip over failed records which have not been * freed. */ if (p->flags & FTRACE_FL_FAILED) @@ -723,6 +731,19 @@ static int __ftrace_update_code(void *ignore) if (p->flags & (FTRACE_FL_CONVERTED)) break; + /* Ignore updates to this record's mcount site. + * Reintroduce this record at the head of this + * bucket to attempt to "convert" it again if + * the kprobe on it is unregistered before the + * next run. */ + if (get_kprobe((void *)p->ip)) { + ftrace_del_hash(p); + INIT_HLIST_NODE(&p->node); + hlist_add_head(&p->node, &temp_list); + continue; + } + + /* convert record (i.e, patch mcount-call with NOP) */ if (ftrace_code_disable(p)) { p->flags |= FTRACE_FL_CONVERTED; ftrace_update_cnt++; @@ -734,6 +755,12 @@ static int __ftrace_update_code(void *ignore) } } } + + hlist_for_each_entry_safe(p, t, n, &temp_list, node) { + hlist_del(&p->node); + INIT_HLIST_NODE(&p->node); + hlist_add_head(&p->node, head); + } } stop = ftrace_now(raw_smp_processor_id()); -- cgit v1.2.3 From 3e61e0c976532a542b23bbb74c8f631815171078 Mon Sep 17 00:00:00 2001 From: Pekka Paalanen Date: Mon, 30 Jun 2008 23:48:37 +0300 Subject: mmiotrace broken in linux-next (8-bit writes only) The moment mmiotrace is enabled, I hit a NULL deref in: IP: [] __trace_special+0x17c/0x23a Call Trace: [] ftrace_special+0x6f/0x9a [] down+0x19/0x4a [] acquire_console_sem+0x42/0x58 [] con_flush_chars+0x28/0x43 [] write_chan+0x22e/0x334 [] ? default_wake_function+0x0/0xf [] tty_write+0x195/0x228 [] ? write_chan+0x0/0x334 [] vfs_write+0xae/0x137 [] sys_write+0x47/0x70 [] system_call_after_swapgs+0x7b/0x80 which means 'entry' in __trace_special() is NULL. [ mingo@elte.hu: that ftrace_special() was a leftover. ] Signed-off-by: Pekka Paalanen Cc: Steven Rostedt Cc: proski@gnu.org Cc: "Vegard Nossum" Signed-off-by: Ingo Molnar --- kernel/semaphore.c | 1 - 1 file changed, 1 deletion(-) (limited to 'kernel') diff --git a/kernel/semaphore.c b/kernel/semaphore.c index 1a064adab658..aaaeae8244e7 100644 --- a/kernel/semaphore.c +++ b/kernel/semaphore.c @@ -54,7 +54,6 @@ void down(struct semaphore *sem) { unsigned long flags; - ftrace_special(sem->count, 0, __LINE__); spin_lock_irqsave(&sem->lock, flags); if (likely(sem->count > 0)) sem->count--; -- cgit v1.2.3 From 98a05ed4bd7774f533ab185fe0bf2fdc58292d7c Mon Sep 17 00:00:00 2001 From: Abhishek Sagar Date: Thu, 26 Jun 2008 22:51:51 +0530 Subject: ftrace: prevent ftrace modifications while being kprobe'd, v2 add two missing chunks for ftrace+kprobe. Signed-off-by: Abhishek Sagar Cc: Steven Rostedt Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 9 ++++++++- 1 file changed, 8 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 85e841335417..0f271c45cd02 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -502,8 +502,12 @@ static void ftrace_replace_code(int enable) continue; /* ignore updates to this record's mcount site */ - if (get_kprobe((void *)rec->ip)) + if (get_kprobe((void *)rec->ip)) { + freeze_record(rec); continue; + } else { + unfreeze_record(rec); + } failed = __ftrace_replace_code(rec, old, new, enable); if (failed && (rec->flags & FTRACE_FL_CONVERTED)) { @@ -740,7 +744,10 @@ static int __ftrace_update_code(void *ignore) ftrace_del_hash(p); INIT_HLIST_NODE(&p->node); hlist_add_head(&p->node, &temp_list); + freeze_record(p); continue; + } else { + unfreeze_record(p); } /* convert record (i.e, patch mcount-call with NOP) */ -- cgit v1.2.3 From 007c05d4d2ce42fabd58cb54ed98e0a1714d9d86 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:09 -0400 Subject: ftrace: move sched_switch enable after markers We have two markers now that are enabled on sched_switch. One that records the context switching and the other that records task wake ups. Currently we enable the tracing first and then set the markers. This causes some confusing traces: # tracer: sched_switch # # TASK-PID CPU# TIMESTAMP FUNCTION # | | | | | trace-cmd-3973 [00] 115.834817: 3973:120:R + 3: 0:S trace-cmd-3973 [01] 115.834910: 3973:120:R + 6: 0:S trace-cmd-3973 [02] 115.834910: 3973:120:R + 9: 0:S trace-cmd-3973 [03] 115.834910: 3973:120:R + 12: 0:S trace-cmd-3973 [02] 115.834910: 3973:120:R + 9: 0:S -0 [02] 115.834910: 0:140:R ==> 3973:120:R Here we see that trace-cmd with PID 3973 wakes up task 9 but the next line shows the idle task doing a context switch to task 3973. Enabling the tracing to _after_ the markers are set creates a much saner output: # tracer: sched_switch # # TASK-PID CPU# TIMESTAMP FUNCTION # | | | | | -0 [02] 7922.634225: 0:140:R ==> 4790:120:R trace-cmd-4789 [03] 7922.634225: 0:140:R + 4790:120:R Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace_sched_switch.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_switch.c b/kernel/trace/trace_sched_switch.c index 93a662009151..cb817a209aa0 100644 --- a/kernel/trace/trace_sched_switch.c +++ b/kernel/trace/trace_sched_switch.c @@ -227,14 +227,14 @@ void tracing_stop_cmdline_record(void) static void start_sched_trace(struct trace_array *tr) { sched_switch_reset(tr); - tracer_enabled = 1; tracing_start_cmdline_record(); + tracer_enabled = 1; } static void stop_sched_trace(struct trace_array *tr) { - tracing_stop_cmdline_record(); tracer_enabled = 0; + tracing_stop_cmdline_record(); } static void sched_switch_trace_init(struct trace_array *tr) -- cgit v1.2.3 From 001b6767b1d0c89e458e5ddb039245b268f569fb Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:10 -0400 Subject: ftrace: define function trace nop When CONFIG_FTRACE is not enabled, the tracing_start_functon_trace and tracing_stop_function_trace should be nops. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace.h | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 6b8bd8800d04..df840f42be39 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -216,8 +216,6 @@ void trace_function(struct trace_array *tr, unsigned long parent_ip, unsigned long flags); -void tracing_start_function_trace(void); -void tracing_stop_function_trace(void); void tracing_start_cmdline_record(void); void tracing_stop_cmdline_record(void); int register_tracer(struct tracer *type); @@ -234,6 +232,14 @@ void update_max_tr_single(struct trace_array *tr, extern cycle_t ftrace_now(int cpu); +#ifdef CONFIG_FTRACE +void tracing_start_function_trace(void); +void tracing_stop_function_trace(void); +#else +# define tracing_start_function_trace() do { } while (0) +# define tracing_stop_function_trace() do { } while (0) +#endif + #ifdef CONFIG_CONTEXT_SWITCH_TRACER typedef void (*tracer_switch_func_t)(void *private, -- cgit v1.2.3 From 1e16c0a081f6c93f04c6af784d6a160955269f91 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:11 -0400 Subject: ftrace: trace schedule After the sched_clock code has been removed from sched.c we can now trace the scheduler. The scheduler has a lot of functions that would be worth tracing. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/Makefile b/kernel/Makefile index ca2433e84873..480976275d98 100644 --- a/kernel/Makefile +++ b/kernel/Makefile @@ -11,7 +11,7 @@ obj-y = sched.o fork.o exec_domain.o panic.o printk.o profile.o \ hrtimer.o rwsem.o nsproxy.o srcu.o semaphore.o \ notifier.o ksysfs.o pm_qos_params.o sched_clock.o -CFLAGS_REMOVE_sched.o = -pg -mno-spe +CFLAGS_REMOVE_sched.o = -mno-spe ifdef CONFIG_FTRACE # Do not trace debug files and internal ftrace files -- cgit v1.2.3 From b5c21b4514b38f450848feb432f7120376d01ffe Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:12 -0400 Subject: ftrace: check proper config for preempt type There is no CONFIG_PREEMPT_DESKTOP. Use the proper entry CONFIG_PREEMPT. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index 9ade79369bfb..f8fdb9cedc24 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -1341,7 +1341,7 @@ print_trace_header(struct seq_file *m, struct trace_iterator *iter) "server", #elif defined(CONFIG_PREEMPT_VOLUNTARY) "desktop", -#elif defined(CONFIG_PREEMPT_DESKTOP) +#elif defined(CONFIG_PREEMPT) "preempt", #else "unknown", -- cgit v1.2.3 From ad591240ceadcaf41b2a88855ca5f1c77c5a0298 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:13 -0400 Subject: ftrace: start wakeup tracing after setting function tracer Enabling the wakeup tracer before enabling the function tracing causes some strange results due to the dynamic enabling of the functions. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace_sched_wakeup.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index bf7e91caef57..3c8d61df4474 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -352,9 +352,10 @@ static void start_wakeup_tracer(struct trace_array *tr) */ smp_wmb(); - tracer_enabled = 1; register_ftrace_function(&trace_ops); + tracer_enabled = 1; + return; fail_deprobe_wake_new: marker_probe_unregister("kernel_sched_wakeup_new", -- cgit v1.2.3 From 26bc83f4cb911a0b4dabfe23b700aaf3235f2955 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:14 -0400 Subject: ftrace: use current CPU for function startup This is more of a clean up. Currently the function tracer initializes the tracer with which ever CPU was last used for tracing. This value isn't realy useful for function tracing, but at least it should be something other than a random number. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace_functions.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 7ee7dcd76b7d..312144897970 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -28,7 +28,10 @@ static void function_reset(struct trace_array *tr) static void start_function_trace(struct trace_array *tr) { + tr->cpu = get_cpu(); function_reset(tr); + put_cpu(); + tracing_start_cmdline_record(); tracing_start_function_trace(); } -- cgit v1.2.3 From a2bb6a3d85ef3124cd336403a95abc0540d3fbe2 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:15 -0400 Subject: ftrace: add ftrace_kill_atomic It has been suggested that I add a way to disable the function tracer on an oops. This code adds a ftrace_kill_atomic. It is not meant to be used in normal situations. It will disable the ftrace tracer, but will not perform the nice shutdown that requires scheduling. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 0f271c45cd02..1359632668a4 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1601,6 +1601,21 @@ core_initcall(ftrace_dynamic_init); # define ftrace_force_shutdown() do { } while (0) #endif /* CONFIG_DYNAMIC_FTRACE */ +/** + * ftrace_kill_atomic - kill ftrace from critical sections + * + * This function should be used by panic code. It stops ftrace + * but in a not so nice way. If you need to simply kill ftrace + * from a non-atomic section, use ftrace_kill. + */ +void ftrace_kill_atomic(void) +{ + ftrace_disabled = 1; + ftrace_enabled = 0; + ftraced_suspend = -1; + clear_ftrace_function(); +} + /** * ftrace_kill - totally shutdown ftrace * -- cgit v1.2.3 From 60bc080090e3bf6afa29c62cb25f913706551010 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Thu, 10 Jul 2008 20:58:16 -0400 Subject: ftrace: separate out the function enabled variable Currently the function tracer uses the global tracer_enabled variable that is used to keep track if the tracer is enabled or not. The function tracing startup needs to be separated out, otherwise the internal happenings of the tracer startup is also recorded. This patch creates a ftrace_function_enabled variable to all the starting of the function traces to happen after everything has been started. Signed-off-by: Steven Rostedt Cc: Steven Rostedt Cc: Peter Zijlstra Cc: Andrew Morton Signed-off-by: Ingo Molnar --- kernel/trace/trace.c | 22 +++++++++++++++++++--- 1 file changed, 19 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index f8fdb9cedc24..2e37857f7dfe 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -96,6 +96,9 @@ static DEFINE_PER_CPU(struct trace_array_cpu, max_data); /* tracer_enabled is used to toggle activation of a tracer */ static int tracer_enabled = 1; +/* function tracing enabled */ +int ftrace_function_enabled; + /* * trace_nr_entries is the number of entries that is allocated * for a buffer. Note, the number of entries is always rounded @@ -134,6 +137,7 @@ static notrace void no_trace_init(struct trace_array *tr) { int cpu; + ftrace_function_enabled = 0; if(tr->ctrl) for_each_online_cpu(cpu) tracing_reset(tr->data[cpu]); @@ -985,7 +989,7 @@ function_trace_call(unsigned long ip, unsigned long parent_ip) long disabled; int cpu; - if (unlikely(!tracer_enabled)) + if (unlikely(!ftrace_function_enabled)) return; if (skip_trace(ip)) @@ -1010,11 +1014,15 @@ static struct ftrace_ops trace_ops __read_mostly = void tracing_start_function_trace(void) { + ftrace_function_enabled = 0; register_ftrace_function(&trace_ops); + if (tracer_enabled) + ftrace_function_enabled = 1; } void tracing_stop_function_trace(void) { + ftrace_function_enabled = 0; unregister_ftrace_function(&trace_ops); } #endif @@ -1850,8 +1858,10 @@ __tracing_open(struct inode *inode, struct file *file, int *ret) m->private = iter; /* stop the trace while dumping */ - if (iter->tr->ctrl) + if (iter->tr->ctrl) { tracer_enabled = 0; + ftrace_function_enabled = 0; + } if (iter->trace && iter->trace->open) iter->trace->open(iter); @@ -1884,8 +1894,14 @@ int tracing_release(struct inode *inode, struct file *file) iter->trace->close(iter); /* reenable tracing if it was previously enabled */ - if (iter->tr->ctrl) + if (iter->tr->ctrl) { tracer_enabled = 1; + /* + * It is safe to enable function tracing even if it + * isn't used + */ + ftrace_function_enabled = 1; + } mutex_unlock(&trace_types_lock); seq_release(inode, file); -- cgit v1.2.3 From b2613e370dbeb69edbff989382fa54f2395aa471 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 11 Jul 2008 16:44:27 +0200 Subject: ftrace: build fix for ftraced_suspend fix: kernel/trace/ftrace.c:1615: error: 'ftraced_suspend' undeclared (first use in this function) kernel/trace/ftrace.c:1615: error: (Each undeclared identifier is reported only once kernel/trace/ftrace.c:1615: error: for each function it appears in.) Signed-off-by: Ingo Molnar --- kernel/trace/ftrace.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'kernel') diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 1359632668a4..4231a3dc224a 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -1612,7 +1612,9 @@ void ftrace_kill_atomic(void) { ftrace_disabled = 1; ftrace_enabled = 0; +#ifdef CONFIG_DYNAMIC_FTRACE ftraced_suspend = -1; +#endif clear_ftrace_function(); } -- cgit v1.2.3