diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2011-07-22 16:44:39 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2011-07-22 16:44:39 -0700 |
commit | 4d4abdcb1dee03a4f9d6d2021622ed07e14dfd17 (patch) | |
tree | 4ed4c74b70240451065165fda5fb2059f8c6b1e5 /kernel | |
parent | 0342cbcfced2ee937d7c8e1c63f3d3082da7c7dc (diff) | |
parent | 7fcfd1abd6480d3b9ef17f5759c175e036e835cf (diff) | |
download | linux-4d4abdcb1dee03a4f9d6d2021622ed07e14dfd17.tar.gz linux-4d4abdcb1dee03a4f9d6d2021622ed07e14dfd17.tar.bz2 linux-4d4abdcb1dee03a4f9d6d2021622ed07e14dfd17.zip |
Merge branch 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip
* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (123 commits)
perf: Remove the nmi parameter from the oprofile_perf backend
x86, perf: Make copy_from_user_nmi() a library function
perf: Remove perf_event_attr::type check
x86, perf: P4 PMU - Fix typos in comments and style cleanup
perf tools: Make test use the preset debugfs path
perf tools: Add automated tests for events parsing
perf tools: De-opt the parse_events function
perf script: Fix display of IP address for non-callchain path
perf tools: Fix endian conversion reading event attr from file header
perf tools: Add missing 'node' alias to the hw_cache[] array
perf probe: Support adding probes on offline kernel modules
perf probe: Add probed module in front of function
perf probe: Introduce debuginfo to encapsulate dwarf information
perf-probe: Move dwarf library routines to dwarf-aux.{c, h}
perf probe: Remove redundant dwarf functions
perf probe: Move strtailcmp to string.c
perf probe: Rename DIE_FIND_CB_FOUND to DIE_FIND_CB_END
tracing/kprobe: Update symbol reference when loading module
tracing/kprobes: Support module init function probing
kprobes: Return -ENOENT if probe point doesn't exist
...
Diffstat (limited to 'kernel')
-rw-r--r-- | kernel/async.c | 12 | ||||
-rw-r--r-- | kernel/events/Makefile | 2 | ||||
-rw-r--r-- | kernel/events/core.c | 938 | ||||
-rw-r--r-- | kernel/events/hw_breakpoint.c | 10 | ||||
-rw-r--r-- | kernel/events/internal.h | 96 | ||||
-rw-r--r-- | kernel/events/ring_buffer.c | 380 | ||||
-rw-r--r-- | kernel/kprobes.c | 33 | ||||
-rw-r--r-- | kernel/sched.c | 2 | ||||
-rw-r--r-- | kernel/stacktrace.c | 12 | ||||
-rw-r--r-- | kernel/trace/ftrace.c | 157 | ||||
-rw-r--r-- | kernel/trace/ring_buffer.c | 66 | ||||
-rw-r--r-- | kernel/trace/ring_buffer_benchmark.c | 2 | ||||
-rw-r--r-- | kernel/trace/trace.c | 326 | ||||
-rw-r--r-- | kernel/trace/trace.h | 61 | ||||
-rw-r--r-- | kernel/trace/trace_entries.h | 3 | ||||
-rw-r--r-- | kernel/trace/trace_events.c | 139 | ||||
-rw-r--r-- | kernel/trace/trace_events_filter.c | 6 | ||||
-rw-r--r-- | kernel/trace/trace_functions.c | 3 | ||||
-rw-r--r-- | kernel/trace/trace_functions_graph.c | 225 | ||||
-rw-r--r-- | kernel/trace/trace_irqsoff.c | 4 | ||||
-rw-r--r-- | kernel/trace/trace_kprobe.c | 324 | ||||
-rw-r--r-- | kernel/trace/trace_output.c | 11 | ||||
-rw-r--r-- | kernel/trace/trace_sched_wakeup.c | 4 | ||||
-rw-r--r-- | kernel/trace/trace_stack.c | 13 | ||||
-rw-r--r-- | kernel/watchdog.c | 8 |
25 files changed, 1655 insertions, 1182 deletions
diff --git a/kernel/async.c b/kernel/async.c index cd9dbb913c77..d5fe7af0de2e 100644 --- a/kernel/async.c +++ b/kernel/async.c @@ -49,12 +49,13 @@ asynchronous and synchronous parts of the kernel. */ #include <linux/async.h> +#include <linux/atomic.h> +#include <linux/ktime.h> #include <linux/module.h> #include <linux/wait.h> #include <linux/sched.h> #include <linux/slab.h> #include <linux/workqueue.h> -#include <asm/atomic.h> static async_cookie_t next_cookie = 1; @@ -128,7 +129,8 @@ static void async_run_entry_fn(struct work_struct *work) /* 2) run (and print duration) */ if (initcall_debug && system_state == SYSTEM_BOOTING) { - printk("calling %lli_%pF @ %i\n", (long long)entry->cookie, + printk(KERN_DEBUG "calling %lli_%pF @ %i\n", + (long long)entry->cookie, entry->func, task_pid_nr(current)); calltime = ktime_get(); } @@ -136,7 +138,7 @@ static void async_run_entry_fn(struct work_struct *work) if (initcall_debug && system_state == SYSTEM_BOOTING) { rettime = ktime_get(); delta = ktime_sub(rettime, calltime); - printk("initcall %lli_%pF returned 0 after %lld usecs\n", + printk(KERN_DEBUG "initcall %lli_%pF returned 0 after %lld usecs\n", (long long)entry->cookie, entry->func, (long long)ktime_to_ns(delta) >> 10); @@ -270,7 +272,7 @@ void async_synchronize_cookie_domain(async_cookie_t cookie, ktime_t starttime, delta, endtime; if (initcall_debug && system_state == SYSTEM_BOOTING) { - printk("async_waiting @ %i\n", task_pid_nr(current)); + printk(KERN_DEBUG "async_waiting @ %i\n", task_pid_nr(current)); starttime = ktime_get(); } @@ -280,7 +282,7 @@ void async_synchronize_cookie_domain(async_cookie_t cookie, endtime = ktime_get(); delta = ktime_sub(endtime, starttime); - printk("async_continuing @ %i after %lli usec\n", + printk(KERN_DEBUG "async_continuing @ %i after %lli usec\n", task_pid_nr(current), (long long)ktime_to_ns(delta) >> 10); } diff --git a/kernel/events/Makefile b/kernel/events/Makefile index 1ce23d3d8394..89e5e8aa4c36 100644 --- a/kernel/events/Makefile +++ b/kernel/events/Makefile @@ -2,5 +2,5 @@ ifdef CONFIG_FUNCTION_TRACER CFLAGS_REMOVE_core.o = -pg endif -obj-y := core.o +obj-y := core.o ring_buffer.o obj-$(CONFIG_HAVE_HW_BREAKPOINT) += hw_breakpoint.o diff --git a/kernel/events/core.c b/kernel/events/core.c index 9efe7108ccaf..b8785e26ee1c 100644 --- a/kernel/events/core.c +++ b/kernel/events/core.c @@ -36,6 +36,8 @@ #include <linux/ftrace_event.h> #include <linux/hw_breakpoint.h> +#include "internal.h" + #include <asm/irq_regs.h> struct remote_function_call { @@ -200,6 +202,22 @@ __get_cpu_context(struct perf_event_context *ctx) return this_cpu_ptr(ctx->pmu->pmu_cpu_context); } +static void perf_ctx_lock(struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx) +{ + raw_spin_lock(&cpuctx->ctx.lock); + if (ctx) + raw_spin_lock(&ctx->lock); +} + +static void perf_ctx_unlock(struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx) +{ + if (ctx) + raw_spin_unlock(&ctx->lock); + raw_spin_unlock(&cpuctx->ctx.lock); +} + #ifdef CONFIG_CGROUP_PERF /* @@ -340,11 +358,8 @@ void perf_cgroup_switch(struct task_struct *task, int mode) rcu_read_lock(); list_for_each_entry_rcu(pmu, &pmus, entry) { - cpuctx = this_cpu_ptr(pmu->pmu_cpu_context); - perf_pmu_disable(cpuctx->ctx.pmu); - /* * perf_cgroup_events says at least one * context on this CPU has cgroup events. @@ -353,6 +368,8 @@ void perf_cgroup_switch(struct task_struct *task, int mode) * events for a context. */ if (cpuctx->ctx.nr_cgroups > 0) { + perf_ctx_lock(cpuctx, cpuctx->task_ctx); + perf_pmu_disable(cpuctx->ctx.pmu); if (mode & PERF_CGROUP_SWOUT) { cpu_ctx_sched_out(cpuctx, EVENT_ALL); @@ -372,9 +389,9 @@ void perf_cgroup_switch(struct task_struct *task, int mode) cpuctx->cgrp = perf_cgroup_from_task(task); cpu_ctx_sched_in(cpuctx, EVENT_ALL, task); } + perf_pmu_enable(cpuctx->ctx.pmu); + perf_ctx_unlock(cpuctx, cpuctx->task_ctx); } - - perf_pmu_enable(cpuctx->ctx.pmu); } rcu_read_unlock(); @@ -731,6 +748,7 @@ static u64 perf_event_time(struct perf_event *event) /* * Update the total_time_enabled and total_time_running fields for a event. + * The caller of this function needs to hold the ctx->lock. */ static void update_event_times(struct perf_event *event) { @@ -1105,6 +1123,10 @@ static int __perf_remove_from_context(void *info) raw_spin_lock(&ctx->lock); event_sched_out(event, cpuctx, ctx); list_del_event(event, ctx); + if (!ctx->nr_events && cpuctx->task_ctx == ctx) { + ctx->is_active = 0; + cpuctx->task_ctx = NULL; + } raw_spin_unlock(&ctx->lock); return 0; @@ -1454,8 +1476,24 @@ static void add_event_to_ctx(struct perf_event *event, event->tstamp_stopped = tstamp; } -static void perf_event_context_sched_in(struct perf_event_context *ctx, - struct task_struct *tsk); +static void task_ctx_sched_out(struct perf_event_context *ctx); +static void +ctx_sched_in(struct perf_event_context *ctx, + struct perf_cpu_context *cpuctx, + enum event_type_t event_type, + struct task_struct *task); + +static void perf_event_sched_in(struct perf_cpu_context *cpuctx, + struct perf_event_context *ctx, + struct task_struct *task) +{ + cpu_ctx_sched_in(cpuctx, EVENT_PINNED, task); + if (ctx) + ctx_sched_in(ctx, cpuctx, EVENT_PINNED, task); + cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE, task); + if (ctx) + ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE, task); +} /* * Cross CPU call to install and enable a performance event @@ -1466,20 +1504,37 @@ static int __perf_install_in_context(void *info) { struct perf_event *event = info; struct perf_event_context *ctx = event->ctx; - struct perf_event *leader = event->group_leader; struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); - int err; + struct perf_event_context *task_ctx = cpuctx->task_ctx; + struct task_struct *task = current; + + perf_ctx_lock(cpuctx, task_ctx); + perf_pmu_disable(cpuctx->ctx.pmu); /* - * In case we're installing a new context to an already running task, - * could also happen before perf_event_task_sched_in() on architectures - * which do context switches with IRQs enabled. + * If there was an active task_ctx schedule it out. */ - if (ctx->task && !cpuctx->task_ctx) - perf_event_context_sched_in(ctx, ctx->task); + if (task_ctx) + task_ctx_sched_out(task_ctx); + + /* + * If the context we're installing events in is not the + * active task_ctx, flip them. + */ + if (ctx->task && task_ctx != ctx) { + if (task_ctx) + raw_spin_unlock(&task_ctx->lock); + raw_spin_lock(&ctx->lock); + task_ctx = ctx; + } + + if (task_ctx) { + cpuctx->task_ctx = task_ctx; + task = task_ctx->task; + } + + cpu_ctx_sched_out(cpuctx, EVENT_ALL); - raw_spin_lock(&ctx->lock); - ctx->is_active = 1; update_context_time(ctx); /* * update cgrp time only if current cgrp @@ -1490,43 +1545,13 @@ static int __perf_install_in_context(void *info) add_event_to_ctx(event, ctx); - if (!event_filter_match(event)) - goto unlock; - - /* - * Don't put the event on if it is disabled or if - * it is in a group and the group isn't on. - */ - if (event->state != PERF_EVENT_STATE_INACTIVE || - (leader != event && leader->state != PERF_EVENT_STATE_ACTIVE)) - goto unlock; - /* - * An exclusive event can't go on if there are already active - * hardware events, and no hardware event can go on if there - * is already an exclusive event on. + * Schedule everything back in */ - if (!group_can_go_on(event, cpuctx, 1)) - err = -EEXIST; - else - err = event_sched_in(event, cpuctx, ctx); - - if (err) { - /* - * This event couldn't go on. If it is in a group - * then we have to pull the whole group off. - * If the event group is pinned then put it in error state. - */ - if (leader != event) - group_sched_out(leader, cpuctx, ctx); - if (leader->attr.pinned) { - update_group_times(leader); - leader->state = PERF_EVENT_STATE_ERROR; - } - } + perf_event_sched_in(cpuctx, task_ctx, task); -unlock: - raw_spin_unlock(&ctx->lock); + perf_pmu_enable(cpuctx->ctx.pmu); + perf_ctx_unlock(cpuctx, task_ctx); return 0; } @@ -1739,7 +1764,7 @@ out: raw_spin_unlock_irq(&ctx->lock); } -static int perf_event_refresh(struct perf_event *event, int refresh) +int perf_event_refresh(struct perf_event *event, int refresh) { /* * not supported on inherited events @@ -1752,36 +1777,35 @@ static int perf_event_refresh(struct perf_event *event, int refresh) return 0; } +EXPORT_SYMBOL_GPL(perf_event_refresh); static void ctx_sched_out(struct perf_event_context *ctx, struct perf_cpu_context *cpuctx, enum event_type_t event_type) { struct perf_event *event; + int is_active = ctx->is_active; - raw_spin_lock(&ctx->lock); - perf_pmu_disable(ctx->pmu); - ctx->is_active = 0; + ctx->is_active &= ~event_type; if (likely(!ctx->nr_events)) - goto out; + return; + update_context_time(ctx); update_cgrp_time_from_cpuctx(cpuctx); - if (!ctx->nr_active) - goto out; + return; - if (event_type & EVENT_PINNED) { + perf_pmu_disable(ctx->pmu); + if ((is_active & EVENT_PINNED) && (event_type & EVENT_PINNED)) { list_for_each_entry(event, &ctx->pinned_groups, group_entry) group_sched_out(event, cpuctx, ctx); } - if (event_type & EVENT_FLEXIBLE) { + if ((is_active & EVENT_FLEXIBLE) && (event_type & EVENT_FLEXIBLE)) { list_for_each_entry(event, &ctx->flexible_groups, group_entry) group_sched_out(event, cpuctx, ctx); } -out: perf_pmu_enable(ctx->pmu); - raw_spin_unlock(&ctx->lock); } /* @@ -1929,8 +1953,10 @@ static void perf_event_context_sched_out(struct task_struct *task, int ctxn, rcu_read_unlock(); if (do_switch) { + raw_spin_lock(&ctx->lock); ctx_sched_out(ctx, cpuctx, EVENT_ALL); cpuctx->task_ctx = NULL; + raw_spin_unlock(&ctx->lock); } } @@ -1965,8 +1991,7 @@ void __perf_event_task_sched_out(struct task_struct *task, perf_cgroup_sched_out(task); } -static void task_ctx_sched_out(struct perf_event_context *ctx, - enum event_type_t event_type) +static void task_ctx_sched_out(struct perf_event_context *ctx) { struct perf_cpu_context *cpuctx = __get_cpu_context(ctx); @@ -1976,7 +2001,7 @@ static void task_ctx_sched_out(struct perf_event_context *ctx, if (WARN_ON_ONCE(ctx != cpuctx->task_ctx)) return; - ctx_sched_out(ctx, cpuctx, event_type); + ctx_sched_out(ctx, cpuctx, EVENT_ALL); cpuctx->task_ctx = NULL; } @@ -2055,11 +2080,11 @@ ctx_sched_in(struct perf_event_context *ctx, struct task_struct *task) { u64 now; + int is_active = ctx->is_active; - raw_spin_lock(&ctx->lock); - ctx->is_active = 1; + ctx->is_active |= event_type; if (likely(!ctx->nr_events)) - goto out; + return; now = perf_clock(); ctx->timestamp = now; @@ -2068,15 +2093,12 @@ ctx_sched_in(struct perf_event_context *ctx, * First go through the list and put on any pinned groups * in order to give them the best chance of going on. */ - if (event_type & EVENT_PINNED) + if (!(is_active & EVENT_PINNED) && (event_type & EVENT_PINNED)) ctx_pinned_sched_in(ctx, cpuctx); /* Then walk through the lower prio flexible groups */ - if (event_type & EVENT_FLEXIBLE) + if (!(is_active & EVENT_FLEXIBLE) && (event_type & EVENT_FLEXIBLE)) ctx_flexible_sched_in(ctx, cpuctx); - -out: - raw_spin_unlock(&ctx->lock); } static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx, @@ -2088,19 +2110,6 @@ static void cpu_ctx_sched_in(struct perf_cpu_context *cpuctx, ctx_sched_in(ctx, cpuctx, event_type, task); } -static void task_ctx_sched_in(struct perf_event_context *ctx, - enum event_type_t event_type) -{ - struct perf_cpu_context *cpuctx; - - cpuctx = __get_cpu_context(ctx); - if (cpuctx->task_ctx == ctx) - return; - - ctx_sched_in(ctx, cpuctx, event_type, NULL); - cpuctx->task_ctx = ctx; -} - static void perf_event_context_sched_in(struct perf_event_context *ctx, struct task_struct *task) { @@ -2110,6 +2119,7 @@ static void perf_event_context_sched_in(struct perf_event_context *ctx, if (cpuctx->task_ctx == ctx) return; + perf_ctx_lock(cpuctx, ctx); perf_pmu_disable(ctx->pmu); /* * We want to keep the following priority order: @@ -2118,18 +2128,18 @@ static void perf_event_context_sched_in(struct perf_event_context *ctx, */ cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE); - ctx_sched_in(ctx, cpuctx, EVENT_PINNED, task); - cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE, task); - ctx_sched_in(ctx, cpuctx, EVENT_FLEXIBLE, task); + perf_event_sched_in(cpuctx, ctx, task); cpuctx->task_ctx = ctx; + perf_pmu_enable(ctx->pmu); + perf_ctx_unlock(cpuctx, ctx); + /* * Since these rotations are per-cpu, we need to ensure the * cpu-context we got scheduled on is actually rotating. */ perf_pmu_rotate_start(ctx->pmu); - perf_pmu_enable(ctx->pmu); } /* @@ -2269,7 +2279,6 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period) u64 interrupts, now; s64 delta; - raw_spin_lock(&ctx->lock); list_for_each_entry_rcu(event, &ctx->event_list, event_entry) { if (event->state != PERF_EVENT_STATE_ACTIVE) continue; @@ -2301,7 +2310,6 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period) if (delta > 0) perf_adjust_period(event, period, delta); } - raw_spin_unlock(&ctx->lock); } /* @@ -2309,16 +2317,12 @@ static void perf_ctx_adjust_freq(struct perf_event_context *ctx, u64 period) */ static void rotate_ctx(struct perf_event_context *ctx) { - raw_spin_lock(&ctx->lock); - /* * Rotate the first entry last of non-pinned groups. Rotation might be * disabled by the inheritance code. */ if (!ctx->rotate_disable) list_rotate_left(&ctx->flexible_groups); - - raw_spin_unlock(&ctx->lock); } /* @@ -2345,6 +2349,7 @@ static void perf_rotate_context(struct perf_cpu_context *cpuctx) rotate = 1; } + perf_ctx_lock(cpuctx, cpuctx->task_ctx); perf_pmu_disable(cpuctx->ctx.pmu); perf_ctx_adjust_freq(&cpuctx->ctx, interval); if (ctx) @@ -2355,21 +2360,20 @@ static void perf_rotate_context(struct perf_cpu_context *cpuctx) cpu_ctx_sched_out(cpuctx, EVENT_FLEXIBLE); if (ctx) - task_ctx_sched_out(ctx, EVENT_FLEXIBLE); + ctx_sched_out(ctx, cpuctx, EVENT_FLEXIBLE); rotate_ctx(&cpuctx->ctx); if (ctx) rotate_ctx(ctx); - cpu_ctx_sched_in(cpuctx, EVENT_FLEXIBLE, current); - if (ctx) - task_ctx_sched_in(ctx, EVENT_FLEXIBLE); + perf_event_sched_in(cpuctx, ctx, current); done: if (remove) list_del_init(&cpuctx->rotation_list); perf_pmu_enable(cpuctx->ctx.pmu); + perf_ctx_unlock(cpuctx, cpuctx->task_ctx); } void perf_event_task_tick(void) @@ -2424,9 +2428,9 @@ static void perf_event_enable_on_exec(struct perf_event_context *ctx) * in. */ perf_cgroup_sched_out(current); - task_ctx_sched_out(ctx, EVENT_ALL); raw_spin_lock(&ctx->lock); + task_ctx_sched_out(ctx); list_for_each_entry(event, &ctx->pinned_groups, group_entry) { ret = event_enable_on_exec(event, ctx); @@ -2835,16 +2839,12 @@ retry: unclone_ctx(ctx); ++ctx->pin_count; raw_spin_unlock_irqrestore(&ctx->lock, flags); - } - - if (!ctx) { + } else { ctx = alloc_perf_context(pmu, task); err = -ENOMEM; if (!ctx) goto errout; - get_ctx(ctx); - err = 0; mutex_lock(&task->perf_event_mutex); /* @@ -2856,14 +2856,14 @@ retry: else if (task->perf_event_ctxp[ctxn]) err = -EAGAIN; else { + get_ctx(ctx); ++ctx->pin_count; rcu_assign_pointer(task->perf_event_ctxp[ctxn], ctx); } mutex_unlock(&task->perf_event_mutex); if (unlikely(err)) { - put_task_struct(task); - kfree(ctx); + put_ctx(ctx); if (err == -EAGAIN) goto retry; @@ -2890,7 +2890,7 @@ static void free_event_rcu(struct rcu_head *head) kfree(event); } -static void perf_buffer_put(struct perf_buffer *buffer); +static void ring_buffer_put(struct ring_buffer *rb); static void free_event(struct perf_event *event) { @@ -2913,9 +2913,9 @@ static void free_event(struct perf_event *event) } } - if (event->buffer) { - perf_buffer_put(event->buffer); - event->buffer = NULL; + if (event->rb) { + ring_buffer_put(event->rb); + event->rb = NULL; } if (is_cgroup_event(event)) @@ -2934,12 +2934,6 @@ int perf_event_release_kernel(struct perf_event *event) { struct perf_event_context *ctx = event->ctx; - /* - * Remove from the PMU, can't get re-enabled since we got - * here because the last ref went. - */ - perf_event_disable(event); - WARN_ON_ONCE(ctx->parent_ctx); /* * There are two ways this annotation is useful: @@ -2956,8 +2950,8 @@ int perf_event_release_kernel(struct perf_event *event) mutex_lock_nested(&ctx->mutex, SINGLE_DEPTH_NESTING); raw_spin_lock_irq(&ctx->lock); perf_group_detach(event); - list_del_event(event, ctx); raw_spin_unlock_irq(&ctx->lock); + perf_remove_from_context(event); mutex_unlock(&ctx->mutex); free_event(event); @@ -3149,13 +3143,13 @@ perf_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) static unsigned int perf_poll(struct file *file, poll_table *wait) { struct perf_event *event = file->private_data; - struct perf_buffer *buffer; + struct ring_buffer *rb; unsigned int events = POLL_HUP; rcu_read_lock(); - buffer = rcu_dereference(event->buffer); - if (buffer) - events = atomic_xchg(&buffer->poll, 0); + rb = rcu_dereference(event->rb); + if (rb) + events = atomic_xchg(&rb->poll, 0); rcu_read_unlock(); poll_wait(file, &event->waitq, wait); @@ -3358,6 +3352,18 @@ static int perf_event_index(struct perf_event *event) return event->hw.idx + 1 - PERF_EVENT_INDEX_OFFSET; } +static void calc_timer_values(struct perf_event *event, + u64 *running, + u64 *enabled) +{ + u64 now, ctx_time; + + now = perf_clock(); + ctx_time = event->shadow_ctx_time + now; + *enabled = ctx_time - event->tstamp_enabled; + *running = ctx_time - event->tstamp_running; +} + /* * Callers need to ensure there can be no nesting of this function, otherwise * the seqlock logic goes bad. We can not serialize this because the arch @@ -3366,14 +3372,25 @@ static int perf_event_index(struct perf_event *event) void perf_event_update_userpage(struct perf_event *event) { struct perf_event_mmap_page *userpg; - struct perf_buffer *buffer; + struct ring_buffer *rb; + u64 enabled, running; rcu_read_lock(); - buffer = rcu_dereference(event->buffer); - if (!buffer) + /* + * compute total_time_enabled, total_time_running + * based on snapshot values taken when the event + * was last scheduled in. + * + * we cannot simply called update_context_time() + * because of locking issue as we can be called in + * NMI context + */ + calc_timer_values(event, &enabled, &running); + rb = rcu_dereference(event->rb); + if (!rb) goto unlock; - userpg = buffer->user_page; + userpg = rb->user_page; /* * Disable preemption so as to not let the corresponding user-space @@ -3387,10 +3404,10 @@ void perf_event_update_userpage(struct perf_event *event) if (event->state == PERF_EVENT_STATE_ACTIVE) userpg->offset -= local64_read(&event->hw.prev_count); - userpg->time_enabled = event->total_time_enabled + + userpg->time_enabled = enabled + atomic64_read(&event->child_total_time_enabled); - userpg->time_running = event->total_time_running + + userpg->time_running = running + atomic64_read(&event->child_total_time_running); barrier(); @@ -3400,220 +3417,10 @@ unlock: rcu_read_unlock(); } -static unsigned long perf_data_size(struct perf_buffer *buffer); - -static void -perf_buffer_init(struct perf_buffer *buffer, long watermark, int flags) -{ - long max_size = perf_data_size(buffer); - - if (watermark) - buffer->watermark = min(max_size, watermark); - - if (!buffer->watermark) - buffer->watermark = max_size / 2; - - if (flags & PERF_BUFFER_WRITABLE) - buffer->writable = 1; - - atomic_set(&buffer->refcount, 1); -} - -#ifndef CONFIG_PERF_USE_VMALLOC - -/* - * Back perf_mmap() with regular GFP_KERNEL-0 pages. - */ - -static struct page * -perf_mmap_to_page(struct perf_buffer *buffer, unsigned long pgoff) -{ - if (pgoff > buffer->nr_pages) - return NULL; - - if (pgoff == 0) - return virt_to_page(buffer->user_page); - - return virt_to_page(buffer->data_pages[pgoff - 1]); -} - -static void *perf_mmap_alloc_page(int cpu) -{ - struct page *page; - int node; - - node = (cpu == -1) ? cpu : cpu_to_node(cpu); - page = alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0); - if (!page) - return NULL; - - return page_address(page); -} - -static struct perf_buffer * -perf_buffer_alloc(int nr_pages, long watermark, int cpu, int flags) -{ - struct perf_buffer *buffer; - unsigned long size; - int i; - - size = sizeof(struct perf_buffer); - size += nr_pages * sizeof(void *); - - buffer = kzalloc(size, GFP_KERNEL); - if (!buffer) - goto fail; - - buffer->user_page = perf_mmap_alloc_page(cpu); - if (!buffer->user_page) - goto fail_user_page; - - for (i = 0; i < nr_pages; i++) { - buffer->data_pages[i] = perf_mmap_alloc_page(cpu); - if (!buffer->data_pages[i]) - goto fail_data_pages; - } - - buffer->nr_pages = nr_pages; - - perf_buffer_init(buffer, watermark, flags); - - return buffer; - -fail_data_pages: - for (i--; i >= 0; i--) - free_page((unsigned long)buffer->data_pages[i]); - - free_page((unsigned long)buffer->user_page); - -fail_user_page: - kfree(buffer); - -fail: - return NULL; -} - -static void perf_mmap_free_page(unsigned long addr) -{ - struct page *page = virt_to_page((void *)addr); - - page->mapping = NULL; - __free_page(page); -} - -static void perf_buffer_free(struct perf_buffer *buffer) -{ - int i; - - perf_mmap_free_page((unsigned long)buffer->user_page); - for (i = 0; i < buffer->nr_pages; i++) - perf_mmap_free_page((unsigned long)buffer->data_pages[i]); - kfree(buffer); -} - -static inline int page_order(struct perf_buffer *buffer) -{ - return 0; -} - -#else - -/* - * Back perf_mmap() with vmalloc memory. - * - * Required for architectures that have d-cache aliasing issues. - */ - -static inline int page_order(struct perf_buffer *buffer) -{ - return buffer->page_order; -} - -static struct page * -perf_mmap_to_page(struct perf_buffer *buffer, unsigned long pgoff) -{ - if (pgoff > (1UL << page_order(buffer))) - return NULL; - - return vmalloc_to_page((void *)buffer->user_page + pgoff * PAGE_SIZE); -} - -static void perf_mmap_unmark_page(void *addr) -{ - struct page *page = vmalloc_to_page(addr); - - page->mapping = NULL; -} - -static void perf_buffer_free_work(struct work_struct *work) -{ - struct perf_buffer *buffer; - void *base; - int i, nr; - - buffer = container_of(work, struct perf_buffer, work); - nr = 1 << page_order(buffer); - - base = buffer->user_page; - for (i = 0; i < nr + 1; i++) - perf_mmap_unmark_page(base + (i * PAGE_SIZE)); - - vfree(base); - kfree(buffer); -} - -static void perf_buffer_free(struct perf_buffer *buffer) -{ - schedule_work(&buffer->work); -} - -static struct perf_buffer * -perf_buffer_alloc(int nr_pages, long watermark, int cpu, int flags) -{ - struct perf_buffer *buffer; - unsigned long size; - void *all_buf; - - size = sizeof(struct perf_buffer); - size += sizeof(void *); - - buffer = kzalloc(size, GFP_KERNEL); - if (!buffer) - goto fail; - - INIT_WORK(&buffer->work, perf_buffer_free_work); - - all_buf = vmalloc_user((nr_pages + 1) * PAGE_SIZE); - if (!all_buf) - goto fail_all_buf; - - buffer->user_page = all_buf; - buffer->data_pages[0] = all_buf + PAGE_SIZE; - buffer->page_order = ilog2(nr_pages); - buffer->nr_pages = 1; - - perf_buffer_init(buffer, watermark, flags); - - return buffer; - -fail_all_buf: - kfree(buffer); - -fail: - return NULL; -} - -#endif - -static unsigned long perf_data_size(struct perf_buffer *buffer) -{ - return buffer->nr_pages << (PAGE_SHIFT + page_order(buffer)); -} - static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf) { struct perf_event *event = vma->vm_file->private_data; - struct perf_buffer *buffer; + struct ring_buffer *rb; int ret = VM_FAULT_SIGBUS; if (vmf->flags & FAULT_FLAG_MKWRITE) { @@ -3623,14 +3430,14 @@ static int perf_mmap_fault(struct vm_area_struct *vma, struct vm_fault *vmf) } rcu_read_lock(); - buffer = rcu_dereference(event->buffer); - if (!buffer) + rb = rcu_dereference(event->rb); + if (!rb) goto unlock; if (vmf->pgoff && (vmf->flags & FAULT_FLAG_WRITE)) goto unlock; - vmf->page = perf_mmap_to_page(buffer, vmf->pgoff); + vmf->page = perf_mmap_to_page(rb, vmf->pgoff); if (!vmf->page) goto unlock; @@ -3645,35 +3452,35 @@ unlock: return ret; } -static void perf_buffer_free_rcu(struct rcu_head *rcu_head) +static void rb_free_rcu(struct rcu_head *rcu_head) { - struct perf_buffer *buffer; + struct ring_buffer *rb; - buffer = container_of(rcu_head, struct perf_buffer, rcu_head); - perf_buffer_free(buffer); + rb = container_of(rcu_head, struct ring_buffer, rcu_head); + rb_free(rb); } -static struct perf_buffer *perf_buffer_get(struct perf_event *event) +static struct ring_buffer *ring_buffer_get(struct perf_event *event) { - struct perf_buffer *buffer; + struct ring_buffer *rb; rcu_read_lock(); - buffer = rcu_dereference(event->buffer); - if (buffer) { - if (!atomic_inc_not_zero(&buffer->refcount)) - buffer = NULL; + rb = rcu_dereference(event->rb); + if (rb) { + if (!atomic_inc_not_zero(&rb->refcount)) + rb = NULL; } rcu_read_unlock(); - return buffer; + return rb; } -static void perf_buffer_put(struct perf_buffer *buffer) +static void ring_buffer_put(struct ring_buffer *rb) { - if (!atomic_dec_and_test(&buffer->refcount)) + if (!atomic_dec_and_test(&rb->refcount)) return; - call_rcu(&buffer->rcu_head, perf_buffer_free_rcu); + call_rcu(&rb->rcu_head, rb_free_rcu); } static void perf_mmap_open(struct vm_area_struct *vma) @@ -3688,16 +3495,16 @@ static void perf_mmap_close(struct vm_area_struct *vma) struct perf_event *event = vma->vm_file->private_data; if (atomic_dec_and_mutex_lock(&event->mmap_count, &event->mmap_mutex)) { - unsigned long size = perf_data_size(event->buffer); + unsigned long size = perf_data_size(event->rb); struct user_struct *user = event->mmap_user; - struct perf_buffer *buffer = event->buffer; + struct ring_buffer *rb = event->rb; atomic_long_sub((size >> PAGE_SHIFT) + 1, &user->locked_vm); vma->vm_mm->locked_vm -= event->mmap_locked; - rcu_assign_pointer(event->buffer, NULL); + rcu_assign_pointer(event->rb, NULL); mutex_unlock(&event->mmap_mutex); - perf_buffer_put(buffer); + ring_buffer_put(rb); free_uid(user); } } @@ -3715,7 +3522,7 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) unsigned long user_locked, user_lock_limit; struct user_struct *user = current_user(); unsigned long locked, lock_limit; - struct perf_buffer *buffer; + struct ring_buffer *rb; unsigned long vma_size; unsigned long nr_pages; long user_extra, extra; @@ -3724,7 +3531,7 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) /* * Don't allow mmap() of inherited per-task counters. This would * create a performance issue due to all children writing to the - * same buffer. + * same rb. */ if (event->cpu == -1 && event->attr.inherit) return -EINVAL; @@ -3736,7 +3543,7 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) nr_pages = (vma_size / PAGE_SIZE) - 1; /* - * If we have buffer pages ensure they're a power-of-two number, so we + * If we have rb pages ensure they're a power-of-two number, so we * can do bitmasks instead of modulo. */ if (nr_pages != 0 && !is_power_of_2(nr_pages)) @@ -3750,9 +3557,9 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) WARN_ON_ONCE(event->ctx->parent_ctx); mutex_lock(&event->mmap_mutex); - if (event->buffer) { - if (event->buffer->nr_pages == nr_pages) - atomic_inc(&event->buffer->refcount); + if (event->rb) { + if (event->rb->nr_pages == nr_pages) + atomic_inc(&event->rb->refcount); else ret = -EINVAL; goto unlock; @@ -3782,18 +3589,20 @@ static int perf_mmap(struct file *file, struct vm_area_struct *vma) goto unlock; } - WARN_ON(event->buffer); + WARN_ON(event->rb); if (vma->vm_flags & VM_WRITE) - flags |= PERF_BUFFER_WRITABLE; + flags |= RING_BUFFER_WRITABLE; - buffer = perf_buffer_alloc(nr_pages, event->attr.wakeup_watermark, - event->cpu, flags); - if (!buffer) { + rb = rb_alloc(nr_pages, + event->attr.watermark ? event->attr.wakeup_watermark : 0, + event->cpu, flags); + + if (!rb) { ret = -ENOMEM; goto unlock; } - rcu_assign_pointer(event->buffer, buffer); + rcu_assign_pointer(event->rb, rb); atomic_long_add(user_extra, &user->locked_vm); event->mmap_locked = extra; @@ -3892,117 +3701,6 @@ int perf_unregister_guest_info_callbacks(struct perf_guest_info_callbacks *cbs) } EXPORT_SYMBOL_GPL(perf_unregister_guest_info_callbacks); -/* - * Output - */ -static bool perf_output_space(struct perf_buffer *buffer, unsigned long tail, - unsigned long offset, unsigned long head) -{ - unsigned long mask; - - if (!buffer->writable) - return true; - - mask = perf_data_size(buffer) - 1; - - offset = (offset - tail) & mask; - head = (head - tail) & mask; - - if ((int)(head - offset) < 0) - return false; - - return true; -} - -static void perf_output_wakeup(struct perf_output_handle *handle) -{ - atomic_set(&handle->buffer->poll, POLL_IN); - - if (handle->nmi) { - handle->event->pending_wakeup = 1; - irq_work_queue(&handle->event->pending); - } else - perf_event_wakeup(handle->event); -} - -/* - * We need to ensure a later event_id doesn't publish a head when a former - * event isn't done writing. However since we need to deal with NMIs we - * cannot fully serialize things. - * - * We only publish the head (and generate a wakeup) when the outer-most - * event completes. - */ -static void perf_output_get_handle(struct perf_output_handle *handle) -{ - struct perf_buffer *buffer = handle->buffer; - - preempt_disable(); - local_inc(&buffer->nest); - handle->wakeup = local_read(&buffer->wakeup); -} - -static void perf_output_put_handle(struct perf_output_handle *handle) -{ - struct perf_buffer *buffer = handle->buffer; - unsigned long head; - -again: - head = local_read(&buffer->head); - - /* - * IRQ/NMI can happen here, which means we can miss a head update. - */ - - if (!local_dec_and_test(&buffer->nest)) - goto out; - - /* - * Publish the known good head. Rely on the full barrier implied - * by atomic_dec_and_test() order the buffer->head read and this - * write. - */ - buffer->user_page->data_head = head; - - /* - * Now check if we missed an update, rely on the (compiler) - * barrier in atomic_dec_and_test() to re-read buffer->head. - */ - if (unlikely(head != local_read(&buffer->head))) { - local_inc(&buffer->nest); - goto again; - } - - if (handle->wakeup != local_read(&buffer->wakeup)) - perf_output_wakeup(handle); - -out: - preempt_enable(); -} - -__always_inline void perf_output_copy(struct perf_output_handle *handle, - const void *buf, unsigned int len) -{ - do { - unsigned long size = min_t(unsigned long, handle->size, len); - - memcpy(handle->addr, buf, size); - - len -= size; - handle->addr += size; - buf += size; - handle->size -= size; - if (!handle->size) { - struct perf_buffer *buffer = handle->buffer; - - handle->page++; - handle->page &= buffer->nr_pages - 1; - handle->addr = buffer->data_pages[handle->page]; - handle->size = PAGE_SIZE << page_order(buffer); - } - } while (len); -} - static void __perf_event_header__init_id(struct perf_event_header *header, struct perf_sample_data *data, struct perf_event *event) @@ -4033,9 +3731,9 @@ static void __perf_event_header__init_id(struct perf_event_header *header, } } -static void perf_event_header__init_id(struct perf_event_header *header, - struct perf_sample_data *data, - struct perf_event *event) +void perf_event_header__init_id(struct perf_event_header *header, + struct perf_sample_data *data, + struct perf_event *event) { if (event->attr.sample_id_all) __perf_event_header__init_id(header, data, event); @@ -4062,121 +3760,14 @@ static void __perf_event__output_id_sample(struct perf_output_handle *handle, perf_output_put(handle, data->cpu_entry); } -static void perf_event__output_id_sample(struct perf_event *event, - struct perf_output_handle *handle, - struct perf_sample_data *sample) +void perf_event__output_id_sample(struct perf_event *event, + struct perf_output_handle *handle, + struct perf_sample_data *sample) { if (event->attr.sample_id_all) __perf_event__output_id_sample(handle, sample); } -int perf_output_begin(struct perf_output_handle *handle, - struct perf_event *event, unsigned int size, - int nmi, int sample) -{ - struct perf_buffer *buffer; - unsigned long tail, offset, head; - int have_lost; - struct perf_sample_data sample_data; - struct { - struct perf_event_header header; - u64 id; - u64 lost; - } lost_event; - - rcu_read_lock(); - /* - * For inherited events we send all the output towards the parent. - */ - if (event->parent) - event = event->parent; - - buffer = rcu_dereference(event->buffer); - if (!buffer) - goto out; - - handle->buffer = buffer; - handle->event = event; - handle->nmi = nmi; - handle->sample = sample; - - if (!buffer->nr_pages) - goto out; - - have_lost = local_read(&buffer->lost); - if (have_lost) { - lost_event.header.size = sizeof(lost_event); - perf_event_header__init_id(&lost_event.header, &sample_data, - event); - size += lost_event.header.size; - } - - perf_output_get_handle(handle); - - do { - /* - * Userspace could choose to issue a mb() before updating the - * tail pointer. So that all reads will be completed before the - * write is issued. - */ - tail = ACCESS_ONCE(buffer->user_page->data_tail); - smp_rmb(); - offset = head = local_read(&buffer->head); - head += size; - if (unlikely(!perf_output_space(buffer, tail, offset, head))) - goto fail; - } while (local_cmpxchg(&buffer->head, offset, head) != offset); - - if (head - local_read(&buffer->wakeup) > buffer->watermark) - local_add(buffer->watermark, &buffer->wakeup); - - handle->page = offset >> (PAGE_SHIFT + page_order(buffer)); - handle->page &= buffer->nr_pages - 1; - handle->size = offset & ((PAGE_SIZE << page_order(buffer)) - 1); - handle->addr = buffer->data_pages[handle->page]; - handle->addr += handle->size; - handle->size = (PAGE_SIZE << page_order(buffer)) - handle->size; - - if (have_lost) { - lost_event.header.type = PERF_RECORD_LOST; - lost_event.header.misc = 0; - lost_event.id = event->id; - lost_event.lost = local_xchg(&buffer->lost, 0); - - perf_output_put(handle, lost_event); - perf_event__output_id_sample(event, handle, &sample_data); - } - - return 0; - -fail: - local_inc(&buffer->lost); - perf_output_put_handle(handle); -out: - rcu_read_unlock(); - - return -ENOSPC; -} - -void perf_output_end(struct perf_output_handle *handle) -{ - struct perf_event *event = handle->event; - struct perf_buffer *buffer = handle->buffer; - - int wakeup_events = event->attr.wakeup_events; - - if (handle->sample && wakeup_events) { - int events = local_inc_return(&buffer->events); - if (events >= wakeup_events) { - local_sub(wakeup_events, &buffer->events); - local_inc(&buffer->wakeup); - } - } - - perf_output_put_handle(handle); - rcu_read_unlock(); -} - static void perf_output_read_one(struct perf_output_handle *handle, struct perf_event *event, u64 enabled, u64 running) @@ -4197,7 +3788,7 @@ static void perf_output_read_one(struct perf_output_handle *handle, if (read_format & PERF_FORMAT_ID) values[n++] = primary_event_id(event); - perf_output_copy(handle, values, n * sizeof(u64)); + __output_copy(handle, values, n * sizeof(u64)); } /* @@ -4227,7 +3818,7 @@ static void perf_output_read_group(struct perf_output_handle *handle, if (read_format & PERF_FORMAT_ID) values[n++] = primary_event_id(leader); - perf_output_copy(handle, values, n * sizeof(u64)); + __output_copy(handle, values, n * sizeof(u64)); list_for_each_entry(sub, &leader->sibling_list, group_entry) { n = 0; @@ -4239,7 +3830,7 @@ static void perf_output_read_group(struct perf_output_handle *handle, if (read_format & PERF_FORMAT_ID) values[n++] = primary_event_id(sub); - perf_output_copy(handle, values, n * sizeof(u64)); + __output_copy(handle, values, n * sizeof(u64)); } } @@ -4249,7 +3840,7 @@ static void perf_output_read_group(struct perf_output_handle *handle, static void perf_output_read(struct perf_output_handle *handle, struct perf_event *event) { - u64 enabled = 0, running = 0, now, ctx_time; + u64 enabled = 0, running = 0; u64 read_format = event->attr.read_format; /* @@ -4261,12 +3852,8 @@ static void perf_output_read(struct perf_output_handle *handle, * because of locking issue as we are called in * NMI context */ - if (read_format & PERF_FORMAT_TOTAL_TIMES) { - now = perf_clock(); - ctx_time = event->shadow_ctx_time + now; - enabled = ctx_time - event->tstamp_enabled; - running = ctx_time - event->tstamp_running; - } + if (read_format & PERF_FORMAT_TOTAL_TIMES) + calc_timer_values(event, &enabled, &running); if (event->attr.read_format & PERF_FORMAT_GROUP) perf_output_read_group(handle, event, enabled, running); @@ -4319,7 +3906,7 @@ void perf_output_sample(struct perf_output_handle *handle, size *= sizeof(u64); - perf_output_copy(handle, data->callchain, size); + __output_copy(handle, data->callchain, size); } else { u64 nr = 0; perf_output_put(handle, nr); @@ -4329,8 +3916,8 @@ void perf_output_sample(struct perf_output_handle *handle, if (sample_type & PERF_SAMPLE_RAW) { if (data->raw) { perf_output_put(handle, data->raw->size); - perf_output_copy(handle, data->raw->data, - data->raw->size); + __output_copy(handle, data->raw->data, + data->raw->size); } else { struct { u32 size; @@ -4342,6 +3929,20 @@ void perf_output_sample(struct perf_output_handle *handle, perf_output_put(handle, raw); } } + + if (!event->attr.watermark) { + int wakeup_events = event->attr.wakeup_events; + + if (wakeup_events) { + struct ring_buffer *rb = handle->rb; + int events = local_inc_return(&rb->events); + + if (events >= wakeup_events) { + local_sub(wakeup_events, &rb->events); + local_inc(&rb->wakeup); + } + } + } } void perf_prepare_sample(struct perf_event_header *header, @@ -4386,7 +3987,7 @@ void perf_prepare_sample(struct perf_event_header *header, } } -static void perf_event_output(struct perf_event *event, int nmi, +static void perf_event_output(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs) { @@ -4398,7 +3999,7 @@ static void perf_event_output(struct perf_event *event, int nmi, perf_prepare_sample(&header, data, event, regs); - if (perf_output_begin(&handle, event, header.size, nmi, 1)) + if (perf_output_begin(&handle, event, header.size)) goto exit; perf_output_sample(&handle, &header, data, event); @@ -4438,7 +4039,7 @@ perf_event_read_event(struct perf_event *event, int ret; perf_event_header__init_id(&read_event.header, &sample, event); - ret = perf_output_begin(&handle, event, read_event.header.size, 0, 0); + ret = perf_output_begin(&handle, event, read_event.header.size); if (ret) return; @@ -4481,7 +4082,7 @@ static void perf_event_task_output(struct perf_event *event, perf_event_header__init_id(&task_event->event_id.header, &sample, event); ret = perf_output_begin(&handle, event, - task_event->event_id.header.size, 0, 0); + task_event->event_id.header.size); if (ret) goto out; @@ -4618,7 +4219,7 @@ static void perf_event_comm_output(struct perf_event *event, perf_event_header__init_id(&comm_event->event_id.header, &sample, event); ret = perf_output_begin(&handle, event, - comm_event->event_id.header.size, 0, 0); + comm_event->event_id.header.size); if (ret) goto out; @@ -4627,7 +4228,7 @@ static void perf_event_comm_output(struct perf_event *event, comm_event->event_id.tid = perf_event_tid(event, comm_event->task); perf_output_put(&handle, comm_event->event_id); - perf_output_copy(&handle, comm_event->comm, + __output_copy(&handle, comm_event->comm, comm_event->comm_size); perf_event__output_id_sample(event, &handle, &sample); @@ -4765,7 +4366,7 @@ static void perf_event_mmap_output(struct perf_event *event, perf_event_header__init_id(&mmap_event->event_id.header, &sample, event); ret = perf_output_begin(&handle, event, - mmap_event->event_id.header.size, 0, 0); + mmap_event->event_id.header.size); if (ret) goto out; @@ -4773,7 +4374,7 @@ static void perf_event_mmap_output(struct perf_event *event, mmap_event->event_id.tid = perf_event_tid(event, current); perf_output_put(&handle, mmap_event->event_id); - perf_output_copy(&handle, mmap_event->file_name, + __output_copy(&handle, mmap_event->file_name, mmap_event->file_size); perf_event__output_id_sample(event, &handle, &sample); @@ -4829,7 +4430,7 @@ static void perf_event_mmap_event(struct perf_mmap_event *mmap_event) if (file) { /* - * d_path works from the end of the buffer backwards, so we + * d_path works from the end of the rb backwards, so we * need to add enough zero bytes after the string to handle * the 64bit alignment we do later. */ @@ -4960,7 +4561,7 @@ static void perf_log_throttle(struct perf_event *event, int enable) perf_event_header__init_id(&throttle_event.header, &sample, event); ret = perf_output_begin(&handle, event, - throttle_event.header.size, 1, 0); + throttle_event.header.size); if (ret) return; @@ -4973,7 +4574,7 @@ static void perf_log_throttle(struct perf_event *event, int enable) * Generic event overflow handling, sampling. */ -static int __perf_event_overflow(struct perf_event *event, int nmi, +static int __perf_event_overflow(struct perf_event *event, int throttle, struct perf_sample_data *data, struct pt_regs *regs) { @@ -5016,34 +4617,28 @@ static int __perf_event_overflow(struct perf_event *event, int nmi, if (events && atomic_dec_and_test(&event->event_limit)) { ret = 1; event->pending_kill = POLL_HUP; - if (nmi) { - event->pending_disable = 1; - irq_work_queue(&event->pending); - } else - perf_event_disable(event); + event->pending_disable = 1; + irq_work_queue(&event->pending); } if (event->overflow_handler) - event->overflow_handler(event, nmi, data, regs); + event->overflow_handler(event, data, regs); else - perf_event_output(event, nmi, data, regs); + perf_event_output(event, data, regs); if (event->fasync && event->pending_kill) { - if (nmi) { - event->pending_wakeup = 1; - irq_work_queue(&event->pending); - } else - perf_event_wakeup(event); + event->pending_wakeup = 1; + irq_work_queue(&event->pending); } return ret; } -int perf_event_overflow(struct perf_event *event, int nmi, +int perf_event_overflow(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs) { - return __perf_event_overflow(event, nmi, 1, data, regs); + return __perf_event_overflow(event, 1, data, regs); } /* @@ -5092,7 +4687,7 @@ again: } static void perf_swevent_overflow(struct perf_event *event, u64 overflow, - int nmi, struct perf_sample_data *data, + struct perf_sample_data *data, struct pt_regs *regs) { struct hw_perf_event *hwc = &event->hw; @@ -5106,7 +4701,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow, return; for (; overflow; overflow--) { - if (__perf_event_overflow(event, nmi, throttle, + if (__perf_event_overflow(event, throttle, data, regs)) { /* * We inhibit the overflow from happening when @@ -5119,7 +4714,7 @@ static void perf_swevent_overflow(struct perf_event *event, u64 overflow, } static void perf_swevent_event(struct perf_event *event, u64 nr, - int nmi, struct perf_sample_data *data, + struct perf_sample_data *data, struct pt_regs *regs) { struct hw_perf_event *hwc = &event->hw; @@ -5133,12 +4728,12 @@ static void perf_swevent_event(struct perf_event *event, u64 nr, return; if (nr == 1 && hwc->sample_period == 1 && !event->attr.freq) - return perf_swevent_overflow(event, 1, nmi, data, regs); + return perf_swevent_overflow(event, 1, data, regs); if (local64_add_negative(nr, &hwc->period_left)) return; - perf_swevent_overflow(event, 0, nmi, data, regs); + perf_swevent_overflow(event, 0, data, regs); } static int perf_exclude_event(struct perf_event *event, @@ -5226,7 +4821,7 @@ find_swevent_head(struct swevent_htable *swhash, struct perf_event *event) } static void do_perf_sw_event(enum perf_type_id type, u32 event_id, - u64 nr, int nmi, + u64 nr, struct perf_sample_data *data, struct pt_regs *regs) { @@ -5242,7 +4837,7 @@ static void do_perf_sw_event(enum perf_type_id type, u32 event_id, hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_swevent_match(event, type, event_id, data, regs)) - perf_swevent_event(event, nr, nmi, data, regs); + perf_swevent_event(event, nr, data, regs); } end: rcu_read_unlock(); @@ -5263,8 +4858,7 @@ inline void perf_swevent_put_recursion_context(int rctx) put_recursion_context(swhash->recursion, rctx); } -void __perf_sw_event(u32 event_id, u64 nr, int nmi, - struct pt_regs *regs, u64 addr) +void __perf_sw_event(u32 event_id, u64 nr, struct pt_regs *regs, u64 addr) { struct perf_sample_data data; int rctx; @@ -5276,7 +4870,7 @@ void __perf_sw_event(u32 event_id, u64 nr, int nmi, perf_sample_data_init(&data, addr); - do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, nmi, &data, regs); + do_perf_sw_event(PERF_TYPE_SOFTWARE, event_id, nr, &data, regs); perf_swevent_put_recursion_context(rctx); preempt_enable_notrace(); @@ -5524,7 +5118,7 @@ void perf_tp_event(u64 addr, u64 count, void *record, int entry_size, hlist_for_each_entry_rcu(event, node, head, hlist_entry) { if (perf_tp_event_match(event, &data, regs)) - perf_swevent_event(event, count, 1, &data, regs); + perf_swevent_event(event, count, &data, regs); } perf_swevent_put_recursion_context(rctx); @@ -5617,7 +5211,7 @@ void perf_bp_event(struct perf_event *bp, void *data) perf_sample_data_init(&sample, bp->attr.bp_addr); if (!bp->hw.state && !perf_exclude_event(bp, regs)) - perf_swevent_event(bp, 1, 1, &sample, regs); + perf_swevent_event(bp, 1, &sample, regs); } #endif @@ -5646,7 +5240,7 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) if (regs && !perf_exclude_event(event, regs)) { if (!(event->attr.exclude_idle && current->pid == 0)) - if (perf_event_overflow(event, 0, &data, regs)) + if (perf_event_overflow(event, &data, regs)) ret = HRTIMER_NORESTART; } @@ -5986,6 +5580,7 @@ free_dev: } static struct lock_class_key cpuctx_mutex; +static struct lock_class_key cpuctx_lock; int perf_pmu_register(struct pmu *pmu, char *name, int type) { @@ -6036,6 +5631,7 @@ skip_type: cpuctx = per_cpu_ptr(pmu->pmu_cpu_context, cpu); __perf_event_init_context(&cpuctx->ctx); lockdep_set_class(&cpuctx->ctx.mutex, &cpuctx_mutex); + lockdep_set_class(&cpuctx->ctx.lock, &cpuctx_lock); cpuctx->ctx.type = cpu_context; cpuctx->ctx.pmu = pmu; cpuctx->jiffies_interval = 1; @@ -6150,7 +5746,8 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, struct task_struct *task, struct perf_event *group_leader, struct perf_event *parent_event, - perf_overflow_handler_t overflow_handler) + perf_overflow_handler_t overflow_handler, + void *context) { struct pmu *pmu; struct perf_event *event; @@ -6208,10 +5805,13 @@ perf_event_alloc(struct perf_event_attr *attr, int cpu, #endif } - if (!overflow_handler && parent_event) + if (!overflow_handler && parent_event) { overflow_handler = parent_event->overflow_handler; + context = parent_event->overflow_handler_context; + } event->overflow_handler = overflow_handler; + event->overflow_handler_context = context; if (attr->disabled) event->state = PERF_EVENT_STATE_OFF; @@ -6326,13 +5926,6 @@ static int perf_copy_attr(struct perf_event_attr __user *uattr, if (ret) return -EFAULT; - /* - * If the type exists, the corresponding creation will verify - * the attr->config. - */ - if (attr->type >= PERF_TYPE_MAX) - return -EINVAL; - if (attr->__reserved_1) return -EINVAL; @@ -6354,7 +5947,7 @@ err_size: static int perf_event_set_output(struct perf_event *event, struct perf_event *output_event) { - struct perf_buffer *buffer = NULL, *old_buffer = NULL; + struct ring_buffer *rb = NULL, *old_rb = NULL; int ret = -EINVAL; if (!output_event) @@ -6371,7 +5964,7 @@ perf_event_set_output(struct perf_event *event, struct perf_event *output_event) goto out; /* - * If its not a per-cpu buffer, it must be the same task. + * If its not a per-cpu rb, it must be the same task. */ if (output_event->cpu == -1 && output_event->ctx != event->ctx) goto out; @@ -6383,20 +5976,20 @@ set: goto unlock; if (output_event) { - /* get the buffer we want to redirect to */ - buffer = perf_buffer_get(output_event); - if (!buffer) + /* get the rb we want to redirect to */ + rb = ring_buffer_get(output_event); + if (!rb) goto unlock; } - old_buffer = event->buffer; - rcu_assign_pointer(event->buffer, buffer); + old_rb = event->rb; + rcu_assign_pointer(event->rb, rb); ret = 0; unlock: mutex_unlock(&event->mmap_mutex); - if (old_buffer) - perf_buffer_put(old_buffer); + if (old_rb) + ring_buffer_put(old_rb); out: return ret; } @@ -6478,7 +6071,8 @@ SYSCALL_DEFINE5(perf_event_open, } } - event = perf_event_alloc(&attr, cpu, task, group_leader, NULL, NULL); + event = perf_event_alloc(&attr, cpu, task, group_leader, NULL, + NULL, NULL); if (IS_ERR(event)) { err = PTR_ERR(event); goto err_task; @@ -6663,7 +6257,8 @@ err_fd: struct perf_event * perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, struct task_struct *task, - perf_overflow_handler_t overflow_handler) + perf_overflow_handler_t overflow_handler, + void *context) { struct perf_event_context *ctx; struct perf_event *event; @@ -6673,7 +6268,8 @@ perf_event_create_kernel_counter(struct perf_event_attr *attr, int cpu, * Get the target context (task or percpu): */ - event = perf_event_alloc(attr, cpu, task, NULL, NULL, overflow_handler); + event = perf_event_alloc(attr, cpu, task, NULL, NULL, + overflow_handler, context); if (IS_ERR(event)) { err = PTR_ERR(event); goto err; @@ -6780,7 +6376,6 @@ static void perf_event_exit_task_context(struct task_struct *child, int ctxn) * our context. */ child_ctx = rcu_dereference_raw(child->perf_event_ctxp[ctxn]); - task_ctx_sched_out(child_ctx, EVENT_ALL); /* * Take the context lock here so that if find_get_context is @@ -6788,6 +6383,7 @@ static void perf_event_exit_task_context(struct task_struct *child, int ctxn) * incremented the context's refcount before we do put_ctx below. */ raw_spin_lock(&child_ctx->lock); + task_ctx_sched_out(child_ctx); child->perf_event_ctxp[ctxn] = NULL; /* * If this context is a clone; unclone it so it can't get @@ -6957,7 +6553,7 @@ inherit_event(struct perf_event *parent_event, parent_event->cpu, child, group_leader, parent_event, - NULL); + NULL, NULL); if (IS_ERR(child_event)) return child_event; get_ctx(child_ctx); @@ -6984,6 +6580,8 @@ inherit_event(struct perf_event *parent_event, child_event->ctx = child_ctx; child_event->overflow_handler = parent_event->overflow_handler; + child_event->overflow_handler_context + = parent_event->overflow_handler_context; /* * Precalculate sample_data sizes diff --git a/kernel/events/hw_breakpoint.c b/kernel/events/hw_breakpoint.c index 086adf25a55e..b7971d6f38bf 100644 --- a/kernel/events/hw_breakpoint.c +++ b/kernel/events/hw_breakpoint.c @@ -431,9 +431,11 @@ int register_perf_hw_breakpoint(struct perf_event *bp) struct perf_event * register_user_hw_breakpoint(struct perf_event_attr *attr, perf_overflow_handler_t triggered, + void *context, struct task_struct *tsk) { - return perf_event_create_kernel_counter(attr, -1, tsk, triggered); + return perf_event_create_kernel_counter(attr, -1, tsk, triggered, + context); } EXPORT_SYMBOL_GPL(register_user_hw_breakpoint); @@ -502,7 +504,8 @@ EXPORT_SYMBOL_GPL(unregister_hw_breakpoint); */ struct perf_event * __percpu * register_wide_hw_breakpoint(struct perf_event_attr *attr, - perf_overflow_handler_t triggered) + perf_overflow_handler_t triggered, + void *context) { struct perf_event * __percpu *cpu_events, **pevent, *bp; long err; @@ -515,7 +518,8 @@ register_wide_hw_breakpoint(struct perf_event_attr *attr, get_online_cpus(); for_each_online_cpu(cpu) { pevent = per_cpu_ptr(cpu_events, cpu); - bp = perf_event_create_kernel_counter(attr, cpu, NULL, triggered); + bp = perf_event_create_kernel_counter(attr, cpu, NULL, + triggered, context); *pevent = bp; diff --git a/kernel/events/internal.h b/kernel/events/internal.h new file mode 100644 index 000000000000..09097dd8116c --- /dev/null +++ b/kernel/events/internal.h @@ -0,0 +1,96 @@ +#ifndef _KERNEL_EVENTS_INTERNAL_H +#define _KERNEL_EVENTS_INTERNAL_H + +#define RING_BUFFER_WRITABLE 0x01 + +struct ring_buffer { + atomic_t refcount; + struct rcu_head rcu_head; +#ifdef CONFIG_PERF_USE_VMALLOC + struct work_struct work; + int page_order; /* allocation order */ +#endif + int nr_pages; /* nr of data pages */ + int writable; /* are we writable */ + + atomic_t poll; /* POLL_ for wakeups */ + + local_t head; /* write position */ + local_t nest; /* nested writers */ + local_t events; /* event limit */ + local_t wakeup; /* wakeup stamp */ + local_t lost; /* nr records lost */ + + long watermark; /* wakeup watermark */ + + struct perf_event_mmap_page *user_page; + void *data_pages[0]; +}; + +extern void rb_free(struct ring_buffer *rb); +extern struct ring_buffer * +rb_alloc(int nr_pages, long watermark, int cpu, int flags); +extern void perf_event_wakeup(struct perf_event *event); + +extern void +perf_event_header__init_id(struct perf_event_header *header, + struct perf_sample_data *data, + struct perf_event *event); +extern void +perf_event__output_id_sample(struct perf_event *event, + struct perf_output_handle *handle, + struct perf_sample_data *sample); + +extern struct page * +perf_mmap_to_page(struct ring_buffer *rb, unsigned long pgoff); + +#ifdef CONFIG_PERF_USE_VMALLOC +/* + * Back perf_mmap() with vmalloc memory. + * + * Required for architectures that have d-cache aliasing issues. + */ + +static inline int page_order(struct ring_buffer *rb) +{ + return rb->page_order; +} + +#else + +static inline int page_order(struct ring_buffer *rb) +{ + return 0; +} +#endif + +static unsigned long perf_data_size(struct ring_buffer *rb) +{ + return rb->nr_pages << (PAGE_SHIFT + page_order(rb)); +} + +static inline void +__output_copy(struct perf_output_handle *handle, + const void *buf, unsigned int len) +{ + do { + unsigned long size = min_t(unsigned long, handle->size, len); + + memcpy(handle->addr, buf, size); + + len -= size; + handle->addr += size; + buf += size; + handle->size -= size; + if (!handle->size) { + struct ring_buffer *rb = handle->rb; + + handle->page++; + handle->page &= rb->nr_pages - 1; + handle->addr = rb->data_pages[handle->page]; + handle->size = PAGE_SIZE << page_order(rb); + } + } while (len); +} + +#endif /* _KERNEL_EVENTS_INTERNAL_H */ diff --git a/kernel/events/ring_buffer.c b/kernel/events/ring_buffer.c new file mode 100644 index 000000000000..a2a29205cc0f --- /dev/null +++ b/kernel/events/ring_buffer.c @@ -0,0 +1,380 @@ +/* + * Performance events ring-buffer code: + * + * Copyright (C) 2008 Thomas Gleixner <tglx@linutronix.de> + * Copyright (C) 2008-2011 Red Hat, Inc., Ingo Molnar + * Copyright (C) 2008-2011 Red Hat, Inc., Peter Zijlstra <pzijlstr@redhat.com> + * Copyright © 2009 Paul Mackerras, IBM Corp. <paulus@au1.ibm.com> + * + * For licensing details see kernel-base/COPYING + */ + +#include <linux/perf_event.h> +#include <linux/vmalloc.h> +#include <linux/slab.h> + +#include "internal.h" + +static bool perf_output_space(struct ring_buffer *rb, unsigned long tail, + unsigned long offset, unsigned long head) +{ + unsigned long mask; + + if (!rb->writable) + return true; + + mask = perf_data_size(rb) - 1; + + offset = (offset - tail) & mask; + head = (head - tail) & mask; + + if ((int)(head - offset) < 0) + return false; + + return true; +} + +static void perf_output_wakeup(struct perf_output_handle *handle) +{ + atomic_set(&handle->rb->poll, POLL_IN); + + handle->event->pending_wakeup = 1; + irq_work_queue(&handle->event->pending); +} + +/* + * We need to ensure a later event_id doesn't publish a head when a former + * event isn't done writing. However since we need to deal with NMIs we + * cannot fully serialize things. + * + * We only publish the head (and generate a wakeup) when the outer-most + * event completes. + */ +static void perf_output_get_handle(struct perf_output_handle *handle) +{ + struct ring_buffer *rb = handle->rb; + + preempt_disable(); + local_inc(&rb->nest); + handle->wakeup = local_read(&rb->wakeup); +} + +static void perf_output_put_handle(struct perf_output_handle *handle) +{ + struct ring_buffer *rb = handle->rb; + unsigned long head; + +again: + head = local_read(&rb->head); + + /* + * IRQ/NMI can happen here, which means we can miss a head update. + */ + + if (!local_dec_and_test(&rb->nest)) + goto out; + + /* + * Publish the known good head. Rely on the full barrier implied + * by atomic_dec_and_test() order the rb->head read and this + * write. + */ + rb->user_page->data_head = head; + + /* + * Now check if we missed an update, rely on the (compiler) + * barrier in atomic_dec_and_test() to re-read rb->head. + */ + if (unlikely(head != local_read(&rb->head))) { + local_inc(&rb->nest); + goto again; + } + + if (handle->wakeup != local_read(&rb->wakeup)) + perf_output_wakeup(handle); + +out: + preempt_enable(); +} + +int perf_output_begin(struct perf_output_handle *handle, + struct perf_event *event, unsigned int size) +{ + struct ring_buffer *rb; + unsigned long tail, offset, head; + int have_lost; + struct perf_sample_data sample_data; + struct { + struct perf_event_header header; + u64 id; + u64 lost; + } lost_event; + + rcu_read_lock(); + /* + * For inherited events we send all the output towards the parent. + */ + if (event->parent) + event = event->parent; + + rb = rcu_dereference(event->rb); + if (!rb) + goto out; + + handle->rb = rb; + handle->event = event; + + if (!rb->nr_pages) + goto out; + + have_lost = local_read(&rb->lost); + if (have_lost) { + lost_event.header.size = sizeof(lost_event); + perf_event_header__init_id(&lost_event.header, &sample_data, + event); + size += lost_event.header.size; + } + + perf_output_get_handle(handle); + + do { + /* + * Userspace could choose to issue a mb() before updating the + * tail pointer. So that all reads will be completed before the + * write is issued. + */ + tail = ACCESS_ONCE(rb->user_page->data_tail); + smp_rmb(); + offset = head = local_read(&rb->head); + head += size; + if (unlikely(!perf_output_space(rb, tail, offset, head))) + goto fail; + } while (local_cmpxchg(&rb->head, offset, head) != offset); + + if (head - local_read(&rb->wakeup) > rb->watermark) + local_add(rb->watermark, &rb->wakeup); + + handle->page = offset >> (PAGE_SHIFT + page_order(rb)); + handle->page &= rb->nr_pages - 1; + handle->size = offset & ((PAGE_SIZE << page_order(rb)) - 1); + handle->addr = rb->data_pages[handle->page]; + handle->addr += handle->size; + handle->size = (PAGE_SIZE << page_order(rb)) - handle->size; + + if (have_lost) { + lost_event.header.type = PERF_RECORD_LOST; + lost_event.header.misc = 0; + lost_event.id = event->id; + lost_event.lost = local_xchg(&rb->lost, 0); + + perf_output_put(handle, lost_event); + perf_event__output_id_sample(event, handle, &sample_data); + } + + return 0; + +fail: + local_inc(&rb->lost); + perf_output_put_handle(handle); +out: + rcu_read_unlock(); + + return -ENOSPC; +} + +void perf_output_copy(struct perf_output_handle *handle, + const void *buf, unsigned int len) +{ + __output_copy(handle, buf, len); +} + +void perf_output_end(struct perf_output_handle *handle) +{ + perf_output_put_handle(handle); + rcu_read_unlock(); +} + +static void +ring_buffer_init(struct ring_buffer *rb, long watermark, int flags) +{ + long max_size = perf_data_size(rb); + + if (watermark) + rb->watermark = min(max_size, watermark); + + if (!rb->watermark) + rb->watermark = max_size / 2; + + if (flags & RING_BUFFER_WRITABLE) + rb->writable = 1; + + atomic_set(&rb->refcount, 1); +} + +#ifndef CONFIG_PERF_USE_VMALLOC + +/* + * Back perf_mmap() with regular GFP_KERNEL-0 pages. + */ + +struct page * +perf_mmap_to_page(struct ring_buffer *rb, unsigned long pgoff) +{ + if (pgoff > rb->nr_pages) + return NULL; + + if (pgoff == 0) + return virt_to_page(rb->user_page); + + return virt_to_page(rb->data_pages[pgoff - 1]); +} + +static void *perf_mmap_alloc_page(int cpu) +{ + struct page *page; + int node; + + node = (cpu == -1) ? cpu : cpu_to_node(cpu); + page = alloc_pages_node(node, GFP_KERNEL | __GFP_ZERO, 0); + if (!page) + return NULL; + + return page_address(page); +} + +struct ring_buffer *rb_alloc(int nr_pages, long watermark, int cpu, int flags) +{ + struct ring_buffer *rb; + unsigned long size; + int i; + + size = sizeof(struct ring_buffer); + size += nr_pages * sizeof(void *); + + rb = kzalloc(size, GFP_KERNEL); + if (!rb) + goto fail; + + rb->user_page = perf_mmap_alloc_page(cpu); + if (!rb->user_page) + goto fail_user_page; + + for (i = 0; i < nr_pages; i++) { + rb->data_pages[i] = perf_mmap_alloc_page(cpu); + if (!rb->data_pages[i]) + goto fail_data_pages; + } + + rb->nr_pages = nr_pages; + + ring_buffer_init(rb, watermark, flags); + + return rb; + +fail_data_pages: + for (i--; i >= 0; i--) + free_page((unsigned long)rb->data_pages[i]); + + free_page((unsigned long)rb->user_page); + +fail_user_page: + kfree(rb); + +fail: + return NULL; +} + +static void perf_mmap_free_page(unsigned long addr) +{ + struct page *page = virt_to_page((void *)addr); + + page->mapping = NULL; + __free_page(page); +} + +void rb_free(struct ring_buffer *rb) +{ + int i; + + perf_mmap_free_page((unsigned long)rb->user_page); + for (i = 0; i < rb->nr_pages; i++) + perf_mmap_free_page((unsigned long)rb->data_pages[i]); + kfree(rb); +} + +#else + +struct page * +perf_mmap_to_page(struct ring_buffer *rb, unsigned long pgoff) +{ + if (pgoff > (1UL << page_order(rb))) + return NULL; + + return vmalloc_to_page((void *)rb->user_page + pgoff * PAGE_SIZE); +} + +static void perf_mmap_unmark_page(void *addr) +{ + struct page *page = vmalloc_to_page(addr); + + page->mapping = NULL; +} + +static void rb_free_work(struct work_struct *work) +{ + struct ring_buffer *rb; + void *base; + int i, nr; + + rb = container_of(work, struct ring_buffer, work); + nr = 1 << page_order(rb); + + base = rb->user_page; + for (i = 0; i < nr + 1; i++) + perf_mmap_unmark_page(base + (i * PAGE_SIZE)); + + vfree(base); + kfree(rb); +} + +void rb_free(struct ring_buffer *rb) +{ + schedule_work(&rb->work); +} + +struct ring_buffer *rb_alloc(int nr_pages, long watermark, int cpu, int flags) +{ + struct ring_buffer *rb; + unsigned long size; + void *all_buf; + + size = sizeof(struct ring_buffer); + size += sizeof(void *); + + rb = kzalloc(size, GFP_KERNEL); + if (!rb) + goto fail; + + INIT_WORK(&rb->work, rb_free_work); + + all_buf = vmalloc_user((nr_pages + 1) * PAGE_SIZE); + if (!all_buf) + goto fail_all_buf; + + rb->user_page = all_buf; + rb->data_pages[0] = all_buf + PAGE_SIZE; + rb->page_order = ilog2(nr_pages); + rb->nr_pages = 1; + + ring_buffer_init(rb, watermark, flags); + + return rb; + +fail_all_buf: + kfree(rb); + +fail: + return NULL; +} + +#endif diff --git a/kernel/kprobes.c b/kernel/kprobes.c index 77981813a1e7..b30fd54eb985 100644 --- a/kernel/kprobes.c +++ b/kernel/kprobes.c @@ -1255,19 +1255,29 @@ static int __kprobes in_kprobes_functions(unsigned long addr) /* * If we have a symbol_name argument, look it up and add the offset field * to it. This way, we can specify a relative address to a symbol. + * This returns encoded errors if it fails to look up symbol or invalid + * combination of parameters. */ static kprobe_opcode_t __kprobes *kprobe_addr(struct kprobe *p) { kprobe_opcode_t *addr = p->addr; + + if ((p->symbol_name && p->addr) || + (!p->symbol_name && !p->addr)) + goto invalid; + if (p->symbol_name) { - if (addr) - return NULL; kprobe_lookup_name(p->symbol_name, addr); + if (!addr) + return ERR_PTR(-ENOENT); } - if (!addr) - return NULL; - return (kprobe_opcode_t *)(((char *)addr) + p->offset); + addr = (kprobe_opcode_t *)(((char *)addr) + p->offset); + if (addr) + return addr; + +invalid: + return ERR_PTR(-EINVAL); } /* Check passed kprobe is valid and return kprobe in kprobe_table. */ @@ -1311,8 +1321,8 @@ int __kprobes register_kprobe(struct kprobe *p) kprobe_opcode_t *addr; addr = kprobe_addr(p); - if (!addr) - return -EINVAL; + if (IS_ERR(addr)) + return PTR_ERR(addr); p->addr = addr; ret = check_kprobe_rereg(p); @@ -1335,6 +1345,8 @@ int __kprobes register_kprobe(struct kprobe *p) */ probed_mod = __module_text_address((unsigned long) p->addr); if (probed_mod) { + /* Return -ENOENT if fail. */ + ret = -ENOENT; /* * We must hold a refcount of the probed module while updating * its code to prohibit unexpected unloading. @@ -1351,6 +1363,7 @@ int __kprobes register_kprobe(struct kprobe *p) module_put(probed_mod); goto fail_with_jump_label; } + /* ret will be updated by following code */ } preempt_enable(); jump_label_unlock(); @@ -1399,7 +1412,7 @@ out: fail_with_jump_label: preempt_enable(); jump_label_unlock(); - return -EINVAL; + return ret; } EXPORT_SYMBOL_GPL(register_kprobe); @@ -1686,8 +1699,8 @@ int __kprobes register_kretprobe(struct kretprobe *rp) if (kretprobe_blacklist_size) { addr = kprobe_addr(&rp->kp); - if (!addr) - return -EINVAL; + if (IS_ERR(addr)) + return PTR_ERR(addr); for (i = 0; kretprobe_blacklist[i].name != NULL; i++) { if (kretprobe_blacklist[i].addr == addr) diff --git a/kernel/sched.c b/kernel/sched.c index c518b05fd062..84b9e076812e 100644 --- a/kernel/sched.c +++ b/kernel/sched.c @@ -2220,7 +2220,7 @@ void set_task_cpu(struct task_struct *p, unsigned int new_cpu) if (task_cpu(p) != new_cpu) { p->se.nr_migrations++; - perf_sw_event(PERF_COUNT_SW_CPU_MIGRATIONS, 1, 1, NULL, 0); + perf_sw_event(PERF_COUNT_SW_CPU_MIGRATIONS, 1, NULL, 0); } __set_task_cpu(p, new_cpu); diff --git a/kernel/stacktrace.c b/kernel/stacktrace.c index eb212f8f8bc8..d20c6983aad9 100644 --- a/kernel/stacktrace.c +++ b/kernel/stacktrace.c @@ -26,12 +26,18 @@ void print_stack_trace(struct stack_trace *trace, int spaces) EXPORT_SYMBOL_GPL(print_stack_trace); /* - * Architectures that do not implement save_stack_trace_tsk get this - * weak alias and a once-per-bootup warning (whenever this facility - * is utilized - for example by procfs): + * Architectures that do not implement save_stack_trace_tsk or + * save_stack_trace_regs get this weak alias and a once-per-bootup warning + * (whenever this facility is utilized - for example by procfs): */ __weak void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace) { WARN_ONCE(1, KERN_INFO "save_stack_trace_tsk() not implemented yet.\n"); } + +__weak void +save_stack_trace_regs(struct pt_regs *regs, struct stack_trace *trace) +{ + WARN_ONCE(1, KERN_INFO "save_stack_trace_regs() not implemented yet.\n"); +} diff --git a/kernel/trace/ftrace.c b/kernel/trace/ftrace.c index 908038f57440..c3e4575e7829 100644 --- a/kernel/trace/ftrace.c +++ b/kernel/trace/ftrace.c @@ -32,7 +32,6 @@ #include <trace/events/sched.h> -#include <asm/ftrace.h> #include <asm/setup.h> #include "trace_output.h" @@ -82,14 +81,14 @@ static int ftrace_disabled __read_mostly; static DEFINE_MUTEX(ftrace_lock); -static struct ftrace_ops ftrace_list_end __read_mostly = -{ +static struct ftrace_ops ftrace_list_end __read_mostly = { .func = ftrace_stub, }; static struct ftrace_ops *ftrace_global_list __read_mostly = &ftrace_list_end; static struct ftrace_ops *ftrace_ops_list __read_mostly = &ftrace_list_end; ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub; +static ftrace_func_t __ftrace_trace_function_delay __read_mostly = ftrace_stub; ftrace_func_t __ftrace_trace_function __read_mostly = ftrace_stub; ftrace_func_t ftrace_pid_function __read_mostly = ftrace_stub; static struct ftrace_ops global_ops; @@ -148,9 +147,11 @@ void clear_ftrace_function(void) { ftrace_trace_function = ftrace_stub; __ftrace_trace_function = ftrace_stub; + __ftrace_trace_function_delay = ftrace_stub; ftrace_pid_function = ftrace_stub; } +#undef CONFIG_HAVE_FUNCTION_TRACE_MCOUNT_TEST #ifndef CONFIG_HAVE_FUNCTION_TRACE_MCOUNT_TEST /* * For those archs that do not test ftrace_trace_stop in their @@ -210,7 +211,12 @@ static void update_ftrace_function(void) #ifdef CONFIG_HAVE_FUNCTION_TRACE_MCOUNT_TEST ftrace_trace_function = func; #else +#ifdef CONFIG_DYNAMIC_FTRACE + /* do not update till all functions have been modified */ + __ftrace_trace_function_delay = func; +#else __ftrace_trace_function = func; +#endif ftrace_trace_function = ftrace_test_stop_func; #endif } @@ -785,8 +791,7 @@ static void unregister_ftrace_profiler(void) unregister_ftrace_graph(); } #else -static struct ftrace_ops ftrace_profile_ops __read_mostly = -{ +static struct ftrace_ops ftrace_profile_ops __read_mostly = { .func = function_profile_call, }; @@ -806,19 +811,10 @@ ftrace_profile_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { unsigned long val; - char buf[64]; /* big enough to hold a number */ int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; val = !!val; @@ -1182,8 +1178,14 @@ alloc_and_copy_ftrace_hash(int size_bits, struct ftrace_hash *hash) return NULL; } +static void +ftrace_hash_rec_disable(struct ftrace_ops *ops, int filter_hash); +static void +ftrace_hash_rec_enable(struct ftrace_ops *ops, int filter_hash); + static int -ftrace_hash_move(struct ftrace_hash **dst, struct ftrace_hash *src) +ftrace_hash_move(struct ftrace_ops *ops, int enable, + struct ftrace_hash **dst, struct ftrace_hash *src) { struct ftrace_func_entry *entry; struct hlist_node *tp, *tn; @@ -1193,9 +1195,16 @@ ftrace_hash_move(struct ftrace_hash **dst, struct ftrace_hash *src) unsigned long key; int size = src->count; int bits = 0; + int ret; int i; /* + * Remove the current set, update the hash and add + * them back. + */ + ftrace_hash_rec_disable(ops, enable); + + /* * If the new source is empty, just free dst and assign it * the empty_hash. */ @@ -1215,9 +1224,10 @@ ftrace_hash_move(struct ftrace_hash **dst, struct ftrace_hash *src) if (bits > FTRACE_HASH_MAX_BITS) bits = FTRACE_HASH_MAX_BITS; + ret = -ENOMEM; new_hash = alloc_ftrace_hash(bits); if (!new_hash) - return -ENOMEM; + goto out; size = 1 << src->size_bits; for (i = 0; i < size; i++) { @@ -1236,7 +1246,16 @@ ftrace_hash_move(struct ftrace_hash **dst, struct ftrace_hash *src) rcu_assign_pointer(*dst, new_hash); free_ftrace_hash_rcu(old_hash); - return 0; + ret = 0; + out: + /* + * Enable regardless of ret: + * On success, we enable the new hash. + * On failure, we re-enable the original hash. + */ + ftrace_hash_rec_enable(ops, enable); + + return ret; } /* @@ -1596,6 +1615,12 @@ static int __ftrace_modify_code(void *data) { int *command = data; + /* + * Do not call function tracer while we update the code. + * We are in stop machine, no worrying about races. + */ + function_trace_stop++; + if (*command & FTRACE_ENABLE_CALLS) ftrace_replace_code(1); else if (*command & FTRACE_DISABLE_CALLS) @@ -1609,6 +1634,18 @@ static int __ftrace_modify_code(void *data) else if (*command & FTRACE_STOP_FUNC_RET) ftrace_disable_ftrace_graph_caller(); +#ifndef CONFIG_HAVE_FUNCTION_TRACE_MCOUNT_TEST + /* + * For archs that call ftrace_test_stop_func(), we must + * wait till after we update all the function callers + * before we update the callback. This keeps different + * ops that record different functions from corrupting + * each other. + */ + __ftrace_trace_function = __ftrace_trace_function_delay; +#endif + function_trace_stop--; + return 0; } @@ -1744,10 +1781,36 @@ static cycle_t ftrace_update_time; static unsigned long ftrace_update_cnt; unsigned long ftrace_update_tot_cnt; +static int ops_traces_mod(struct ftrace_ops *ops) +{ + struct ftrace_hash *hash; + + hash = ops->filter_hash; + return !!(!hash || !hash->count); +} + static int ftrace_update_code(struct module *mod) { struct dyn_ftrace *p; cycle_t start, stop; + unsigned long ref = 0; + + /* + * When adding a module, we need to check if tracers are + * currently enabled and if they are set to trace all functions. + * If they are, we need to enable the module functions as well + * as update the reference counts for those function records. + */ + if (mod) { + struct ftrace_ops *ops; + + for (ops = ftrace_ops_list; + ops != &ftrace_list_end; ops = ops->next) { + if (ops->flags & FTRACE_OPS_FL_ENABLED && + ops_traces_mod(ops)) + ref++; + } + } start = ftrace_now(raw_smp_processor_id()); ftrace_update_cnt = 0; @@ -1760,7 +1823,7 @@ static int ftrace_update_code(struct module *mod) p = ftrace_new_addrs; ftrace_new_addrs = p->newlist; - p->flags = 0L; + p->flags = ref; /* * Do the initial record conversion from mcount jump @@ -1783,7 +1846,7 @@ static int ftrace_update_code(struct module *mod) * conversion puts the module to the correct state, thus * passing the ftrace_make_call check. */ - if (ftrace_start_up) { + if (ftrace_start_up && ref) { int failed = __ftrace_replace_code(p, 1); if (failed) { ftrace_bug(failed, p->ip); @@ -2407,10 +2470,9 @@ ftrace_match_module_records(struct ftrace_hash *hash, char *buff, char *mod) */ static int -ftrace_mod_callback(char *func, char *cmd, char *param, int enable) +ftrace_mod_callback(struct ftrace_hash *hash, + char *func, char *cmd, char *param, int enable) { - struct ftrace_ops *ops = &global_ops; - struct ftrace_hash *hash; char *mod; int ret = -EINVAL; @@ -2430,11 +2492,6 @@ ftrace_mod_callback(char *func, char *cmd, char *param, int enable) if (!strlen(mod)) return ret; - if (enable) - hash = ops->filter_hash; - else - hash = ops->notrace_hash; - ret = ftrace_match_module_records(hash, func, mod); if (!ret) ret = -EINVAL; @@ -2760,7 +2817,7 @@ static int ftrace_process_regex(struct ftrace_hash *hash, mutex_lock(&ftrace_cmd_mutex); list_for_each_entry(p, &ftrace_commands, list) { if (strcmp(p->name, command) == 0) { - ret = p->func(func, command, next, enable); + ret = p->func(hash, func, command, next, enable); goto out_unlock; } } @@ -2857,7 +2914,11 @@ ftrace_set_regex(struct ftrace_ops *ops, unsigned char *buf, int len, ftrace_match_records(hash, buf, len); mutex_lock(&ftrace_lock); - ret = ftrace_hash_move(orig_hash, hash); + ret = ftrace_hash_move(ops, enable, orig_hash, hash); + if (!ret && ops->flags & FTRACE_OPS_FL_ENABLED + && ftrace_enabled) + ftrace_run_update_code(FTRACE_ENABLE_CALLS); + mutex_unlock(&ftrace_lock); mutex_unlock(&ftrace_regex_lock); @@ -3040,18 +3101,12 @@ ftrace_regex_release(struct inode *inode, struct file *file) orig_hash = &iter->ops->notrace_hash; mutex_lock(&ftrace_lock); - /* - * Remove the current set, update the hash and add - * them back. - */ - ftrace_hash_rec_disable(iter->ops, filter_hash); - ret = ftrace_hash_move(orig_hash, iter->hash); - if (!ret) { - ftrace_hash_rec_enable(iter->ops, filter_hash); - if (iter->ops->flags & FTRACE_OPS_FL_ENABLED - && ftrace_enabled) - ftrace_run_update_code(FTRACE_ENABLE_CALLS); - } + ret = ftrace_hash_move(iter->ops, filter_hash, + orig_hash, iter->hash); + if (!ret && (iter->ops->flags & FTRACE_OPS_FL_ENABLED) + && ftrace_enabled) + ftrace_run_update_code(FTRACE_ENABLE_CALLS); + mutex_unlock(&ftrace_lock); } free_ftrace_hash(iter->hash); @@ -3330,7 +3385,7 @@ static int ftrace_process_locs(struct module *mod, { unsigned long *p; unsigned long addr; - unsigned long flags; + unsigned long flags = 0; /* Shut up gcc */ mutex_lock(&ftrace_lock); p = start; @@ -3348,12 +3403,18 @@ static int ftrace_process_locs(struct module *mod, } /* - * Disable interrupts to prevent interrupts from executing - * code that is being modified. + * We only need to disable interrupts on start up + * because we are modifying code that an interrupt + * may execute, and the modification is not atomic. + * But for modules, nothing runs the code we modify + * until we are finished with it, and there's no + * reason to cause large interrupt latencies while we do it. */ - local_irq_save(flags); + if (!mod) + local_irq_save(flags); ftrace_update_code(mod); - local_irq_restore(flags); + if (!mod) + local_irq_restore(flags); mutex_unlock(&ftrace_lock); return 0; diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c index b0c7aa407943..731201bf4acc 100644 --- a/kernel/trace/ring_buffer.c +++ b/kernel/trace/ring_buffer.c @@ -997,15 +997,21 @@ static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer, unsigned nr_pages) { struct buffer_page *bpage, *tmp; - unsigned long addr; LIST_HEAD(pages); unsigned i; WARN_ON(!nr_pages); for (i = 0; i < nr_pages; i++) { + struct page *page; + /* + * __GFP_NORETRY flag makes sure that the allocation fails + * gracefully without invoking oom-killer and the system is + * not destabilized. + */ bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()), - GFP_KERNEL, cpu_to_node(cpu_buffer->cpu)); + GFP_KERNEL | __GFP_NORETRY, + cpu_to_node(cpu_buffer->cpu)); if (!bpage) goto free_pages; @@ -1013,10 +1019,11 @@ static int rb_allocate_pages(struct ring_buffer_per_cpu *cpu_buffer, list_add(&bpage->list, &pages); - addr = __get_free_page(GFP_KERNEL); - if (!addr) + page = alloc_pages_node(cpu_to_node(cpu_buffer->cpu), + GFP_KERNEL | __GFP_NORETRY, 0); + if (!page) goto free_pages; - bpage->page = (void *)addr; + bpage->page = page_address(page); rb_init_page(bpage->page); } @@ -1045,7 +1052,7 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu) { struct ring_buffer_per_cpu *cpu_buffer; struct buffer_page *bpage; - unsigned long addr; + struct page *page; int ret; cpu_buffer = kzalloc_node(ALIGN(sizeof(*cpu_buffer), cache_line_size()), @@ -1067,10 +1074,10 @@ rb_allocate_cpu_buffer(struct ring_buffer *buffer, int cpu) rb_check_bpage(cpu_buffer, bpage); cpu_buffer->reader_page = bpage; - addr = __get_free_page(GFP_KERNEL); - if (!addr) + page = alloc_pages_node(cpu_to_node(cpu), GFP_KERNEL, 0); + if (!page) goto fail_free_reader; - bpage->page = (void *)addr; + bpage->page = page_address(page); rb_init_page(bpage->page); INIT_LIST_HEAD(&cpu_buffer->reader_page->list); @@ -1314,7 +1321,6 @@ int ring_buffer_resize(struct ring_buffer *buffer, unsigned long size) unsigned nr_pages, rm_pages, new_pages; struct buffer_page *bpage, *tmp; unsigned long buffer_size; - unsigned long addr; LIST_HEAD(pages); int i, cpu; @@ -1375,16 +1381,24 @@ int ring_buffer_resize(struct ring_buffer *buffer, unsigned long size) for_each_buffer_cpu(buffer, cpu) { for (i = 0; i < new_pages; i++) { + struct page *page; + /* + * __GFP_NORETRY flag makes sure that the allocation + * fails gracefully without invoking oom-killer and + * the system is not destabilized. + */ bpage = kzalloc_node(ALIGN(sizeof(*bpage), cache_line_size()), - GFP_KERNEL, cpu_to_node(cpu)); + GFP_KERNEL | __GFP_NORETRY, + cpu_to_node(cpu)); if (!bpage) goto free_pages; list_add(&bpage->list, &pages); - addr = __get_free_page(GFP_KERNEL); - if (!addr) + page = alloc_pages_node(cpu_to_node(cpu), + GFP_KERNEL | __GFP_NORETRY, 0); + if (!page) goto free_pages; - bpage->page = (void *)addr; + bpage->page = page_address(page); rb_init_page(bpage->page); } } @@ -3730,16 +3744,17 @@ EXPORT_SYMBOL_GPL(ring_buffer_swap_cpu); * Returns: * The page allocated, or NULL on error. */ -void *ring_buffer_alloc_read_page(struct ring_buffer *buffer) +void *ring_buffer_alloc_read_page(struct ring_buffer *buffer, int cpu) { struct buffer_data_page *bpage; - unsigned long addr; + struct page *page; - addr = __get_free_page(GFP_KERNEL); - if (!addr) + page = alloc_pages_node(cpu_to_node(cpu), + GFP_KERNEL | __GFP_NORETRY, 0); + if (!page) return NULL; - bpage = (void *)addr; + bpage = page_address(page); rb_init_page(bpage); @@ -3978,20 +3993,11 @@ rb_simple_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { unsigned long *p = filp->private_data; - char buf[64]; unsigned long val; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; if (val) diff --git a/kernel/trace/ring_buffer_benchmark.c b/kernel/trace/ring_buffer_benchmark.c index 302f8a614635..a5457d577b98 100644 --- a/kernel/trace/ring_buffer_benchmark.c +++ b/kernel/trace/ring_buffer_benchmark.c @@ -106,7 +106,7 @@ static enum event_status read_page(int cpu) int inc; int i; - bpage = ring_buffer_alloc_read_page(buffer); + bpage = ring_buffer_alloc_read_page(buffer, cpu); if (!bpage) return EVENT_DROPPED; diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index ee9c921d7f21..e5df02c69b1d 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -343,26 +343,27 @@ unsigned long trace_flags = TRACE_ITER_PRINT_PARENT | TRACE_ITER_PRINTK | static int trace_stop_count; static DEFINE_SPINLOCK(tracing_start_lock); +static void wakeup_work_handler(struct work_struct *work) +{ + wake_up(&trace_wait); +} + +static DECLARE_DELAYED_WORK(wakeup_work, wakeup_work_handler); + /** * 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. + * Schedules a delayed work to wake 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) { - int cpu; + const unsigned long delay = msecs_to_jiffies(2); if (trace_flags & TRACE_ITER_BLOCK) return; - /* - * The runqueue_is_locked() can fail, but this is the best we - * have for now: - */ - cpu = get_cpu(); - if (!runqueue_is_locked(cpu)) - wake_up(&trace_wait); - put_cpu(); + schedule_delayed_work(&wakeup_work, delay); } static int __init set_buf_size(char *str) @@ -424,6 +425,7 @@ static const char *trace_options[] = { "graph-time", "record-cmd", "overwrite", + "disable_on_free", NULL }; @@ -1191,6 +1193,18 @@ void trace_nowake_buffer_unlock_commit(struct ring_buffer *buffer, } EXPORT_SYMBOL_GPL(trace_nowake_buffer_unlock_commit); +void trace_nowake_buffer_unlock_commit_regs(struct ring_buffer *buffer, + struct ring_buffer_event *event, + unsigned long flags, int pc, + struct pt_regs *regs) +{ + ring_buffer_unlock_commit(buffer, event); + + ftrace_trace_stack_regs(buffer, flags, 0, pc, regs); + ftrace_trace_userstack(buffer, flags, pc); +} +EXPORT_SYMBOL_GPL(trace_nowake_buffer_unlock_commit_regs); + void trace_current_buffer_discard_commit(struct ring_buffer *buffer, struct ring_buffer_event *event) { @@ -1234,30 +1248,103 @@ ftrace(struct trace_array *tr, struct trace_array_cpu *data, } #ifdef CONFIG_STACKTRACE + +#define FTRACE_STACK_MAX_ENTRIES (PAGE_SIZE / sizeof(unsigned long)) +struct ftrace_stack { + unsigned long calls[FTRACE_STACK_MAX_ENTRIES]; +}; + +static DEFINE_PER_CPU(struct ftrace_stack, ftrace_stack); +static DEFINE_PER_CPU(int, ftrace_stack_reserve); + static void __ftrace_trace_stack(struct ring_buffer *buffer, unsigned long flags, - int skip, int pc) + int skip, int pc, struct pt_regs *regs) { struct ftrace_event_call *call = &event_kernel_stack; struct ring_buffer_event *event; struct stack_entry *entry; struct stack_trace trace; + int use_stack; + int size = FTRACE_STACK_ENTRIES; + + trace.nr_entries = 0; + trace.skip = skip; + + /* + * Since events can happen in NMIs there's no safe way to + * use the per cpu ftrace_stacks. We reserve it and if an interrupt + * or NMI comes in, it will just have to use the default + * FTRACE_STACK_SIZE. + */ + preempt_disable_notrace(); + + use_stack = ++__get_cpu_var(ftrace_stack_reserve); + /* + * We don't need any atomic variables, just a barrier. + * If an interrupt comes in, we don't care, because it would + * have exited and put the counter back to what we want. + * We just need a barrier to keep gcc from moving things + * around. + */ + barrier(); + if (use_stack == 1) { + trace.entries = &__get_cpu_var(ftrace_stack).calls[0]; + trace.max_entries = FTRACE_STACK_MAX_ENTRIES; + + if (regs) + save_stack_trace_regs(regs, &trace); + else + save_stack_trace(&trace); + + if (trace.nr_entries > size) + size = trace.nr_entries; + } else + /* From now on, use_stack is a boolean */ + use_stack = 0; + + size *= sizeof(unsigned long); event = trace_buffer_lock_reserve(buffer, TRACE_STACK, - sizeof(*entry), flags, pc); + sizeof(*entry) + size, flags, pc); if (!event) - return; - entry = ring_buffer_event_data(event); - memset(&entry->caller, 0, sizeof(entry->caller)); + goto out; + entry = ring_buffer_event_data(event); - trace.nr_entries = 0; - trace.max_entries = FTRACE_STACK_ENTRIES; - trace.skip = skip; - trace.entries = entry->caller; + memset(&entry->caller, 0, size); + + if (use_stack) + memcpy(&entry->caller, trace.entries, + trace.nr_entries * sizeof(unsigned long)); + else { + trace.max_entries = FTRACE_STACK_ENTRIES; + trace.entries = entry->caller; + if (regs) + save_stack_trace_regs(regs, &trace); + else + save_stack_trace(&trace); + } + + entry->size = trace.nr_entries; - save_stack_trace(&trace); if (!filter_check_discard(call, entry, buffer, event)) ring_buffer_unlock_commit(buffer, event); + + out: + /* Again, don't let gcc optimize things here */ + barrier(); + __get_cpu_var(ftrace_stack_reserve)--; + preempt_enable_notrace(); + +} + +void ftrace_trace_stack_regs(struct ring_buffer *buffer, unsigned long flags, + int skip, int pc, struct pt_regs *regs) +{ + if (!(trace_flags & TRACE_ITER_STACKTRACE)) + return; + + __ftrace_trace_stack(buffer, flags, skip, pc, regs); } void ftrace_trace_stack(struct ring_buffer *buffer, unsigned long flags, @@ -1266,13 +1353,13 @@ void ftrace_trace_stack(struct ring_buffer *buffer, unsigned long flags, if (!(trace_flags & TRACE_ITER_STACKTRACE)) return; - __ftrace_trace_stack(buffer, flags, skip, pc); + __ftrace_trace_stack(buffer, flags, skip, pc, NULL); } void __trace_stack(struct trace_array *tr, unsigned long flags, int skip, int pc) { - __ftrace_trace_stack(tr->buffer, flags, skip, pc); + __ftrace_trace_stack(tr->buffer, flags, skip, pc, NULL); } /** @@ -1288,7 +1375,7 @@ void trace_dump_stack(void) local_save_flags(flags); /* skipping 3 traces, seems to get us at the caller of this function */ - __ftrace_trace_stack(global_trace.buffer, flags, 3, preempt_count()); + __ftrace_trace_stack(global_trace.buffer, flags, 3, preempt_count(), NULL); } static DEFINE_PER_CPU(int, user_stack_count); @@ -1536,7 +1623,12 @@ peek_next_entry(struct trace_iterator *iter, int cpu, u64 *ts, ftrace_enable_cpu(); - return event ? ring_buffer_event_data(event) : NULL; + if (event) { + iter->ent_size = ring_buffer_event_length(event); + return ring_buffer_event_data(event); + } + iter->ent_size = 0; + return NULL; } static struct trace_entry * @@ -2051,6 +2143,9 @@ void trace_default_header(struct seq_file *m) { struct trace_iterator *iter = m->private; + if (!(trace_flags & TRACE_ITER_CONTEXT_INFO)) + return; + if (iter->iter_flags & TRACE_FILE_LAT_FMT) { /* print nothing if the buffers are empty */ if (trace_empty(iter)) @@ -2701,20 +2796,11 @@ tracing_ctrl_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { struct trace_array *tr = filp->private_data; - char buf[64]; unsigned long val; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; val = !!val; @@ -2767,7 +2853,7 @@ int tracer_init(struct tracer *t, struct trace_array *tr) return t->init(tr); } -static int tracing_resize_ring_buffer(unsigned long size) +static int __tracing_resize_ring_buffer(unsigned long size) { int ret; @@ -2819,6 +2905,41 @@ static int tracing_resize_ring_buffer(unsigned long size) return ret; } +static ssize_t tracing_resize_ring_buffer(unsigned long size) +{ + int cpu, ret = size; + + mutex_lock(&trace_types_lock); + + tracing_stop(); + + /* disable all cpu buffers */ + for_each_tracing_cpu(cpu) { + if (global_trace.data[cpu]) + atomic_inc(&global_trace.data[cpu]->disabled); + if (max_tr.data[cpu]) + atomic_inc(&max_tr.data[cpu]->disabled); + } + + if (size != global_trace.entries) + ret = __tracing_resize_ring_buffer(size); + + if (ret < 0) + ret = -ENOMEM; + + for_each_tracing_cpu(cpu) { + if (global_trace.data[cpu]) + atomic_dec(&global_trace.data[cpu]->disabled); + if (max_tr.data[cpu]) + atomic_dec(&max_tr.data[cpu]->disabled); + } + + tracing_start(); + mutex_unlock(&trace_types_lock); + + return ret; +} + /** * tracing_update_buffers - used by tracing facility to expand ring buffers @@ -2836,7 +2957,7 @@ int tracing_update_buffers(void) mutex_lock(&trace_types_lock); if (!ring_buffer_expanded) - ret = tracing_resize_ring_buffer(trace_buf_size); + ret = __tracing_resize_ring_buffer(trace_buf_size); mutex_unlock(&trace_types_lock); return ret; @@ -2860,7 +2981,7 @@ static int tracing_set_tracer(const char *buf) mutex_lock(&trace_types_lock); if (!ring_buffer_expanded) { - ret = tracing_resize_ring_buffer(trace_buf_size); + ret = __tracing_resize_ring_buffer(trace_buf_size); if (ret < 0) goto out; ret = 0; @@ -2966,20 +3087,11 @@ tracing_max_lat_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { unsigned long *ptr = filp->private_data; - char buf[64]; unsigned long val; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; *ptr = val * 1000; @@ -3434,67 +3546,54 @@ tracing_entries_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { unsigned long val; - char buf[64]; - int ret, cpu; - - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; + int ret; - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; /* must have at least 1 entry */ if (!val) return -EINVAL; - mutex_lock(&trace_types_lock); - - tracing_stop(); - - /* disable all cpu buffers */ - for_each_tracing_cpu(cpu) { - if (global_trace.data[cpu]) - atomic_inc(&global_trace.data[cpu]->disabled); - if (max_tr.data[cpu]) - atomic_inc(&max_tr.data[cpu]->disabled); - } - /* value is in KB */ val <<= 10; - if (val != global_trace.entries) { - ret = tracing_resize_ring_buffer(val); - if (ret < 0) { - cnt = ret; - goto out; - } - } + ret = tracing_resize_ring_buffer(val); + if (ret < 0) + return ret; *ppos += cnt; - /* If check pages failed, return ENOMEM */ - if (tracing_disabled) - cnt = -ENOMEM; - out: - for_each_tracing_cpu(cpu) { - if (global_trace.data[cpu]) - atomic_dec(&global_trace.data[cpu]->disabled); - if (max_tr.data[cpu]) - atomic_dec(&max_tr.data[cpu]->disabled); - } + return cnt; +} - tracing_start(); - mutex_unlock(&trace_types_lock); +static ssize_t +tracing_free_buffer_write(struct file *filp, const char __user *ubuf, + size_t cnt, loff_t *ppos) +{ + /* + * There is no need to read what the user has written, this function + * is just to make sure that there is no error when "echo" is used + */ + + *ppos += cnt; return cnt; } +static int +tracing_free_buffer_release(struct inode *inode, struct file *filp) +{ + /* disable tracing ? */ + if (trace_flags & TRACE_ITER_STOP_ON_FREE) + tracing_off(); + /* resize the ring buffer to 0 */ + tracing_resize_ring_buffer(0); + + return 0; +} + static int mark_printk(const char *fmt, ...) { int ret; @@ -3640,6 +3739,11 @@ static const struct file_operations tracing_entries_fops = { .llseek = generic_file_llseek, }; +static const struct file_operations tracing_free_buffer_fops = { + .write = tracing_free_buffer_write, + .release = tracing_free_buffer_release, +}; + static const struct file_operations tracing_mark_fops = { .open = tracing_open_generic, .write = tracing_mark_write, @@ -3696,7 +3800,7 @@ tracing_buffers_read(struct file *filp, char __user *ubuf, return 0; if (!info->spare) - info->spare = ring_buffer_alloc_read_page(info->tr->buffer); + info->spare = ring_buffer_alloc_read_page(info->tr->buffer, info->cpu); if (!info->spare) return -ENOMEM; @@ -3853,7 +3957,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos, ref->ref = 1; ref->buffer = info->tr->buffer; - ref->page = ring_buffer_alloc_read_page(ref->buffer); + ref->page = ring_buffer_alloc_read_page(ref->buffer, info->cpu); if (!ref->page) { kfree(ref); break; @@ -3862,8 +3966,7 @@ tracing_buffers_splice_read(struct file *file, loff_t *ppos, r = ring_buffer_read_page(ref->buffer, &ref->page, len, info->cpu, 1); if (r < 0) { - ring_buffer_free_read_page(ref->buffer, - ref->page); + ring_buffer_free_read_page(ref->buffer, ref->page); kfree(ref); break; } @@ -4099,19 +4202,10 @@ trace_options_write(struct file *filp, const char __user *ubuf, size_t cnt, { struct trace_option_dentry *topt = filp->private_data; unsigned long val; - char buf[64]; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; if (val != 0 && val != 1) @@ -4159,20 +4253,11 @@ trace_options_core_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { long index = (long)filp->private_data; - char buf[64]; unsigned long val; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; if (val != 0 && val != 1) @@ -4365,6 +4450,9 @@ static __init int tracer_init_debugfs(void) trace_create_file("buffer_size_kb", 0644, d_tracer, &global_trace, &tracing_entries_fops); + trace_create_file("free_buffer", 0644, d_tracer, + &global_trace, &tracing_free_buffer_fops); + trace_create_file("trace_marker", 0220, d_tracer, NULL, &tracing_mark_fops); diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h index 229f8591f61d..3f381d0b20a8 100644 --- a/kernel/trace/trace.h +++ b/kernel/trace/trace.h @@ -278,6 +278,29 @@ struct tracer { }; +/* Only current can touch trace_recursion */ +#define trace_recursion_inc() do { (current)->trace_recursion++; } while (0) +#define trace_recursion_dec() do { (current)->trace_recursion--; } while (0) + +/* Ring buffer has the 10 LSB bits to count */ +#define trace_recursion_buffer() ((current)->trace_recursion & 0x3ff) + +/* for function tracing recursion */ +#define TRACE_INTERNAL_BIT (1<<11) +#define TRACE_GLOBAL_BIT (1<<12) +/* + * Abuse of the trace_recursion. + * As we need a way to maintain state if we are tracing the function + * graph in irq because we want to trace a particular function that + * was called in irq context but we have irq tracing off. Since this + * can only be modified by current, we can reuse trace_recursion. + */ +#define TRACE_IRQ_BIT (1<<13) + +#define trace_recursion_set(bit) do { (current)->trace_recursion |= (bit); } while (0) +#define trace_recursion_clear(bit) do { (current)->trace_recursion &= ~(bit); } while (0) +#define trace_recursion_test(bit) ((current)->trace_recursion & (bit)) + #define TRACE_PIPE_ALL_CPU -1 int tracer_init(struct tracer *t, struct trace_array *tr); @@ -389,6 +412,9 @@ void update_max_tr_single(struct trace_array *tr, void ftrace_trace_stack(struct ring_buffer *buffer, unsigned long flags, int skip, int pc); +void ftrace_trace_stack_regs(struct ring_buffer *buffer, unsigned long flags, + int skip, int pc, struct pt_regs *regs); + void ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc); @@ -400,6 +426,12 @@ static inline void ftrace_trace_stack(struct ring_buffer *buffer, { } +static inline void ftrace_trace_stack_regs(struct ring_buffer *buffer, + unsigned long flags, int skip, + int pc, struct pt_regs *regs) +{ +} + static inline void ftrace_trace_userstack(struct ring_buffer *buffer, unsigned long flags, int pc) { @@ -507,8 +539,18 @@ static inline int ftrace_graph_addr(unsigned long addr) return 1; for (i = 0; i < ftrace_graph_count; i++) { - if (addr == ftrace_graph_funcs[i]) + if (addr == ftrace_graph_funcs[i]) { + /* + * If no irqs are to be traced, but a set_graph_function + * is set, and called by an interrupt handler, we still + * want to trace it. + */ + if (in_irq()) + trace_recursion_set(TRACE_IRQ_BIT); + else + trace_recursion_clear(TRACE_IRQ_BIT); return 1; + } } return 0; @@ -609,6 +651,7 @@ enum trace_iterator_flags { TRACE_ITER_GRAPH_TIME = 0x80000, TRACE_ITER_RECORD_CMD = 0x100000, TRACE_ITER_OVERWRITE = 0x200000, + TRACE_ITER_STOP_ON_FREE = 0x400000, }; /* @@ -677,6 +720,7 @@ struct event_subsystem { struct dentry *entry; struct event_filter *filter; int nr_events; + int ref_count; }; #define FILTER_PRED_INVALID ((unsigned short)-1) @@ -784,19 +828,4 @@ extern const char *__stop___trace_bprintk_fmt[]; FTRACE_ENTRY(call, struct_name, id, PARAMS(tstruct), PARAMS(print)) #include "trace_entries.h" -/* Only current can touch trace_recursion */ -#define trace_recursion_inc() do { (current)->trace_recursion++; } while (0) -#define trace_recursion_dec() do { (current)->trace_recursion--; } while (0) - -/* Ring buffer has the 10 LSB bits to count */ -#define trace_recursion_buffer() ((current)->trace_recursion & 0x3ff) - -/* for function tracing recursion */ -#define TRACE_INTERNAL_BIT (1<<11) -#define TRACE_GLOBAL_BIT (1<<12) - -#define trace_recursion_set(bit) do { (current)->trace_recursion |= (bit); } while (0) -#define trace_recursion_clear(bit) do { (current)->trace_recursion &= ~(bit); } while (0) -#define trace_recursion_test(bit) ((current)->trace_recursion & (bit)) - #endif /* _LINUX_KERNEL_TRACE_H */ diff --git a/kernel/trace/trace_entries.h b/kernel/trace/trace_entries.h index e32744c84d94..93365907f219 100644 --- a/kernel/trace/trace_entries.h +++ b/kernel/trace/trace_entries.h @@ -161,7 +161,8 @@ FTRACE_ENTRY(kernel_stack, stack_entry, TRACE_STACK, F_STRUCT( - __array( unsigned long, caller, FTRACE_STACK_ENTRIES ) + __field( int, size ) + __dynamic_array(unsigned long, caller ) ), F_printk("\t=> (%08lx)\n\t=> (%08lx)\n\t=> (%08lx)\n\t=> (%08lx)\n" diff --git a/kernel/trace/trace_events.c b/kernel/trace/trace_events.c index 686ec399f2a8..581876f9f387 100644 --- a/kernel/trace/trace_events.c +++ b/kernel/trace/trace_events.c @@ -244,6 +244,35 @@ static void ftrace_clear_events(void) mutex_unlock(&event_mutex); } +static void __put_system(struct event_subsystem *system) +{ + struct event_filter *filter = system->filter; + + WARN_ON_ONCE(system->ref_count == 0); + if (--system->ref_count) + return; + + if (filter) { + kfree(filter->filter_string); + kfree(filter); + } + kfree(system->name); + kfree(system); +} + +static void __get_system(struct event_subsystem *system) +{ + WARN_ON_ONCE(system->ref_count == 0); + system->ref_count++; +} + +static void put_system(struct event_subsystem *system) +{ + mutex_lock(&event_mutex); + __put_system(system); + mutex_unlock(&event_mutex); +} + /* * __ftrace_set_clr_event(NULL, NULL, NULL, set) will set/unset all events. */ @@ -486,20 +515,11 @@ event_enable_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { struct ftrace_event_call *call = filp->private_data; - char buf[64]; unsigned long val; int ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; ret = tracing_update_buffers(); @@ -528,7 +548,7 @@ system_enable_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) { const char set_to_char[4] = { '?', '0', '1', 'X' }; - const char *system = filp->private_data; + struct event_subsystem *system = filp->private_data; struct ftrace_event_call *call; char buf[2]; int set = 0; @@ -539,7 +559,7 @@ system_enable_read(struct file *filp, char __user *ubuf, size_t cnt, if (!call->name || !call->class || !call->class->reg) continue; - if (system && strcmp(call->class->system, system) != 0) + if (system && strcmp(call->class->system, system->name) != 0) continue; /* @@ -569,21 +589,13 @@ static ssize_t system_enable_write(struct file *filp, const char __user *ubuf, size_t cnt, loff_t *ppos) { - const char *system = filp->private_data; + struct event_subsystem *system = filp->private_data; + const char *name = NULL; unsigned long val; - char buf[64]; ssize_t ret; - if (cnt >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, cnt)) - return -EFAULT; - - buf[cnt] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, cnt, 10, &val); + if (ret) return ret; ret = tracing_update_buffers(); @@ -593,7 +605,14 @@ system_enable_write(struct file *filp, const char __user *ubuf, size_t cnt, if (val != 0 && val != 1) return -EINVAL; - ret = __ftrace_set_clr_event(NULL, system, NULL, val); + /* + * Opening of "enable" adds a ref count to system, + * so the name is safe to use. + */ + if (system) + name = system->name; + + ret = __ftrace_set_clr_event(NULL, name, NULL, val); if (ret) goto out; @@ -826,6 +845,52 @@ event_filter_write(struct file *filp, const char __user *ubuf, size_t cnt, return cnt; } +static LIST_HEAD(event_subsystems); + +static int subsystem_open(struct inode *inode, struct file *filp) +{ + struct event_subsystem *system = NULL; + int ret; + + if (!inode->i_private) + goto skip_search; + + /* Make sure the system still exists */ + mutex_lock(&event_mutex); + list_for_each_entry(system, &event_subsystems, list) { + if (system == inode->i_private) { + /* Don't open systems with no events */ + if (!system->nr_events) { + system = NULL; + break; + } + __get_system(system); + break; + } + } + mutex_unlock(&event_mutex); + + if (system != inode->i_private) + return -ENODEV; + + skip_search: + ret = tracing_open_generic(inode, filp); + if (ret < 0 && system) + put_system(system); + + return ret; +} + +static int subsystem_release(struct inode *inode, struct file *file) +{ + struct event_subsystem *system = inode->i_private; + + if (system) + put_system(system); + + return 0; +} + static ssize_t subsystem_filter_read(struct file *filp, char __user *ubuf, size_t cnt, loff_t *ppos) @@ -963,17 +1028,19 @@ static const struct file_operations ftrace_event_filter_fops = { }; static const struct file_operations ftrace_subsystem_filter_fops = { - .open = tracing_open_generic, + .open = subsystem_open, .read = subsystem_filter_read, .write = subsystem_filter_write, .llseek = default_llseek, + .release = subsystem_release, }; static const struct file_operations ftrace_system_enable_fops = { - .open = tracing_open_generic, + .open = subsystem_open, .read = system_enable_read, .write = system_enable_write, .llseek = default_llseek, + .release = subsystem_release, }; static const struct file_operations ftrace_show_header_fops = { @@ -1002,8 +1069,6 @@ static struct dentry *event_trace_events_dir(void) return d_events; } -static LIST_HEAD(event_subsystems); - static struct dentry * event_subsystem_dir(const char *name, struct dentry *d_events) { @@ -1013,6 +1078,7 @@ event_subsystem_dir(const char *name, struct dentry *d_events) /* First see if we did not already create this dir */ list_for_each_entry(system, &event_subsystems, list) { if (strcmp(system->name, name) == 0) { + __get_system(system); system->nr_events++; return system->entry; } @@ -1035,6 +1101,7 @@ event_subsystem_dir(const char *name, struct dentry *d_events) } system->nr_events = 1; + system->ref_count = 1; system->name = kstrdup(name, GFP_KERNEL); if (!system->name) { debugfs_remove(system->entry); @@ -1062,8 +1129,7 @@ event_subsystem_dir(const char *name, struct dentry *d_events) "'%s/filter' entry\n", name); } - trace_create_file("enable", 0644, system->entry, - (void *)system->name, + trace_create_file("enable", 0644, system->entry, system, &ftrace_system_enable_fops); return system->entry; @@ -1184,16 +1250,9 @@ static void remove_subsystem_dir(const char *name) list_for_each_entry(system, &event_subsystems, list) { if (strcmp(system->name, name) == 0) { if (!--system->nr_events) { - struct event_filter *filter = system->filter; - debugfs_remove_recursive(system->entry); list_del(&system->list); - if (filter) { - kfree(filter->filter_string); - kfree(filter); - } - kfree(system->name); - kfree(system); + __put_system(system); } break; } diff --git a/kernel/trace/trace_events_filter.c b/kernel/trace/trace_events_filter.c index 8008ddcfbf20..256764ecccd6 100644 --- a/kernel/trace/trace_events_filter.c +++ b/kernel/trace/trace_events_filter.c @@ -1886,6 +1886,12 @@ int apply_subsystem_event_filter(struct event_subsystem *system, mutex_lock(&event_mutex); + /* Make sure the system still has events */ + if (!system->nr_events) { + err = -ENODEV; + goto out_unlock; + } + if (!strcmp(strstrip(filter_string), "0")) { filter_free_subsystem_preds(system); remove_filter_string(system->filter); diff --git a/kernel/trace/trace_functions.c b/kernel/trace/trace_functions.c index 8d0e1cc4e974..c7b0c6a7db09 100644 --- a/kernel/trace/trace_functions.c +++ b/kernel/trace/trace_functions.c @@ -324,7 +324,8 @@ ftrace_trace_onoff_unreg(char *glob, char *cmd, char *param) } static int -ftrace_trace_onoff_callback(char *glob, char *cmd, char *param, int enable) +ftrace_trace_onoff_callback(struct ftrace_hash *hash, + char *glob, char *cmd, char *param, int enable) { struct ftrace_probe_ops *ops; void *count = (void *)-1; diff --git a/kernel/trace/trace_functions_graph.c b/kernel/trace/trace_functions_graph.c index 962cdb24ed81..a7d2a4c653d8 100644 --- a/kernel/trace/trace_functions_graph.c +++ b/kernel/trace/trace_functions_graph.c @@ -74,6 +74,20 @@ static struct tracer_flags tracer_flags = { static struct trace_array *graph_array; +/* + * DURATION column is being also used to display IRQ signs, + * following values are used by print_graph_irq and others + * to fill in space into DURATION column. + */ +enum { + DURATION_FILL_FULL = -1, + DURATION_FILL_START = -2, + DURATION_FILL_END = -3, +}; + +static enum print_line_t +print_graph_duration(unsigned long long duration, struct trace_seq *s, + u32 flags); /* Add a function return address to the trace stack on thread info.*/ int @@ -213,7 +227,7 @@ int __trace_graph_entry(struct trace_array *tr, static inline int ftrace_graph_ignore_irqs(void) { - if (!ftrace_graph_skip_irqs) + if (!ftrace_graph_skip_irqs || trace_recursion_test(TRACE_IRQ_BIT)) return 0; return in_irq(); @@ -577,32 +591,6 @@ get_return_for_leaf(struct trace_iterator *iter, return next; } -/* Signal a overhead of time execution to the output */ -static int -print_graph_overhead(unsigned long long duration, struct trace_seq *s, - u32 flags) -{ - /* If duration disappear, we don't need anything */ - if (!(flags & TRACE_GRAPH_PRINT_DURATION)) - return 1; - - /* Non nested entry or return */ - if (duration == -1) - return trace_seq_printf(s, " "); - - if (flags & TRACE_GRAPH_PRINT_OVERHEAD) { - /* Duration exceeded 100 msecs */ - if (duration > 100000ULL) - return trace_seq_printf(s, "! "); - - /* Duration exceeded 10 msecs */ - if (duration > 10000ULL) - return trace_seq_printf(s, "+ "); - } - - return trace_seq_printf(s, " "); -} - static int print_graph_abs_time(u64 t, struct trace_seq *s) { unsigned long usecs_rem; @@ -625,34 +613,36 @@ print_graph_irq(struct trace_iterator *iter, unsigned long addr, addr >= (unsigned long)__irqentry_text_end) return TRACE_TYPE_UNHANDLED; - /* Absolute time */ - if (flags & TRACE_GRAPH_PRINT_ABS_TIME) { - ret = print_graph_abs_time(iter->ts, s); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - } + if (trace_flags & TRACE_ITER_CONTEXT_INFO) { + /* Absolute time */ + if (flags & TRACE_GRAPH_PRINT_ABS_TIME) { + ret = print_graph_abs_time(iter->ts, s); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + } - /* Cpu */ - if (flags & TRACE_GRAPH_PRINT_CPU) { - ret = print_graph_cpu(s, cpu); - if (ret == TRACE_TYPE_PARTIAL_LINE) - return TRACE_TYPE_PARTIAL_LINE; - } + /* Cpu */ + if (flags & TRACE_GRAPH_PRINT_CPU) { + ret = print_graph_cpu(s, cpu); + if (ret == TRACE_TYPE_PARTIAL_LINE) + return TRACE_TYPE_PARTIAL_LINE; + } - /* Proc */ - if (flags & TRACE_GRAPH_PRINT_PROC) { - ret = print_graph_proc(s, pid); - if (ret == TRACE_TYPE_PARTIAL_LINE) - return TRACE_TYPE_PARTIAL_LINE; - ret = trace_seq_printf(s, " | "); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; + /* Proc */ + if (flags & TRACE_GRAPH_PRINT_PROC) { + ret = print_graph_proc(s, pid); + if (ret == TRACE_TYPE_PARTIAL_LINE) + return TRACE_TYPE_PARTIAL_LINE; + ret = trace_seq_printf(s, " | "); + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; + } } /* No overhead */ - ret = print_graph_overhead(-1, s, flags); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; + ret = print_graph_duration(DURATION_FILL_START, s, flags); + if (ret != TRACE_TYPE_HANDLED) + return ret; if (type == TRACE_GRAPH_ENT) ret = trace_seq_printf(s, "==========>"); @@ -662,9 +652,10 @@ print_graph_irq(struct trace_iterator *iter, unsigned long addr, if (!ret) return TRACE_TYPE_PARTIAL_LINE; - /* Don't close the duration column if haven't one */ - if (flags & TRACE_GRAPH_PRINT_DURATION) - trace_seq_printf(s, " |"); + ret = print_graph_duration(DURATION_FILL_END, s, flags); + if (ret != TRACE_TYPE_HANDLED) + return ret; + ret = trace_seq_printf(s, "\n"); if (!ret) @@ -716,9 +707,49 @@ trace_print_graph_duration(unsigned long long duration, struct trace_seq *s) } static enum print_line_t -print_graph_duration(unsigned long long duration, struct trace_seq *s) +print_graph_duration(unsigned long long duration, struct trace_seq *s, + u32 flags) { - int ret; + int ret = -1; + + if (!(flags & TRACE_GRAPH_PRINT_DURATION) || + !(trace_flags & TRACE_ITER_CONTEXT_INFO)) + return TRACE_TYPE_HANDLED; + + /* No real adata, just filling the column with spaces */ + switch (duration) { + case DURATION_FILL_FULL: + ret = trace_seq_printf(s, " | "); + return ret ? TRACE_TYPE_HANDLED : TRACE_TYPE_PARTIAL_LINE; + case DURATION_FILL_START: + ret = trace_seq_printf(s, " "); + return ret ? TRACE_TYPE_HANDLED : TRACE_TYPE_PARTIAL_LINE; + case DURATION_FILL_END: + ret = trace_seq_printf(s, " |"); + return ret ? TRACE_TYPE_HANDLED : TRACE_TYPE_PARTIAL_LINE; + } + + /* Signal a overhead of time execution to the output */ + if (flags & TRACE_GRAPH_PRINT_OVERHEAD) { + /* Duration exceeded 100 msecs */ + if (duration > 100000ULL) + ret = trace_seq_printf(s, "! "); + /* Duration exceeded 10 msecs */ + else if (duration > 10000ULL) + ret = trace_seq_printf(s, "+ "); + } + + /* + * The -1 means we either did not exceed the duration tresholds + * or we dont want to print out the overhead. Either way we need + * to fill out the space. + */ + if (ret == -1) + ret = trace_seq_printf(s, " "); + + /* Catching here any failure happenned above */ + if (!ret) + return TRACE_TYPE_PARTIAL_LINE; ret = trace_print_graph_duration(duration, s); if (ret != TRACE_TYPE_HANDLED) @@ -767,18 +798,11 @@ print_graph_entry_leaf(struct trace_iterator *iter, cpu_data->enter_funcs[call->depth] = 0; } - /* Overhead */ - ret = print_graph_overhead(duration, s, flags); - if (!ret) + /* Overhead and duration */ + ret = print_graph_duration(duration, s, flags); + if (ret == TRACE_TYPE_PARTIAL_LINE) return TRACE_TYPE_PARTIAL_LINE; - /* Duration */ - if (flags & TRACE_GRAPH_PRINT_DURATION) { - ret = print_graph_duration(duration, s); - if (ret == TRACE_TYPE_PARTIAL_LINE) - return TRACE_TYPE_PARTIAL_LINE; - } - /* Function */ for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) { ret = trace_seq_printf(s, " "); @@ -815,17 +839,10 @@ print_graph_entry_nested(struct trace_iterator *iter, cpu_data->enter_funcs[call->depth] = call->func; } - /* No overhead */ - ret = print_graph_overhead(-1, s, flags); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - /* No time */ - if (flags & TRACE_GRAPH_PRINT_DURATION) { - ret = trace_seq_printf(s, " | "); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - } + ret = print_graph_duration(DURATION_FILL_FULL, s, flags); + if (ret != TRACE_TYPE_HANDLED) + return ret; /* Function */ for (i = 0; i < call->depth * TRACE_GRAPH_INDENT; i++) { @@ -865,6 +882,9 @@ print_graph_prologue(struct trace_iterator *iter, struct trace_seq *s, return TRACE_TYPE_PARTIAL_LINE; } + if (!(trace_flags & TRACE_ITER_CONTEXT_INFO)) + return 0; + /* Absolute time */ if (flags & TRACE_GRAPH_PRINT_ABS_TIME) { ret = print_graph_abs_time(iter->ts, s); @@ -1078,18 +1098,11 @@ print_graph_return(struct ftrace_graph_ret *trace, struct trace_seq *s, if (print_graph_prologue(iter, s, 0, 0, flags)) return TRACE_TYPE_PARTIAL_LINE; - /* Overhead */ - ret = print_graph_overhead(duration, s, flags); - if (!ret) + /* Overhead and duration */ + ret = print_graph_duration(duration, s, flags); + if (ret == TRACE_TYPE_PARTIAL_LINE) return TRACE_TYPE_PARTIAL_LINE; - /* Duration */ - if (flags & TRACE_GRAPH_PRINT_DURATION) { - ret = print_graph_duration(duration, s); - if (ret == TRACE_TYPE_PARTIAL_LINE) - return TRACE_TYPE_PARTIAL_LINE; - } - /* Closing brace */ for (i = 0; i < trace->depth * TRACE_GRAPH_INDENT; i++) { ret = trace_seq_printf(s, " "); @@ -1146,17 +1159,10 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent, if (print_graph_prologue(iter, s, 0, 0, flags)) return TRACE_TYPE_PARTIAL_LINE; - /* No overhead */ - ret = print_graph_overhead(-1, s, flags); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - /* No time */ - if (flags & TRACE_GRAPH_PRINT_DURATION) { - ret = trace_seq_printf(s, " | "); - if (!ret) - return TRACE_TYPE_PARTIAL_LINE; - } + ret = print_graph_duration(DURATION_FILL_FULL, s, flags); + if (ret != TRACE_TYPE_HANDLED) + return ret; /* Indentation */ if (depth > 0) @@ -1207,7 +1213,7 @@ print_graph_comment(struct trace_seq *s, struct trace_entry *ent, enum print_line_t -__print_graph_function_flags(struct trace_iterator *iter, u32 flags) +print_graph_function_flags(struct trace_iterator *iter, u32 flags) { struct ftrace_graph_ent_entry *field; struct fgraph_data *data = iter->private; @@ -1270,18 +1276,7 @@ __print_graph_function_flags(struct trace_iterator *iter, u32 flags) static enum print_line_t print_graph_function(struct trace_iterator *iter) { - return __print_graph_function_flags(iter, tracer_flags.val); -} - -enum print_line_t print_graph_function_flags(struct trace_iterator *iter, - u32 flags) -{ - if (trace_flags & TRACE_ITER_LATENCY_FMT) - flags |= TRACE_GRAPH_PRINT_DURATION; - else - flags |= TRACE_GRAPH_PRINT_ABS_TIME; - - return __print_graph_function_flags(iter, flags); + return print_graph_function_flags(iter, tracer_flags.val); } static enum print_line_t @@ -1309,8 +1304,7 @@ static void print_lat_header(struct seq_file *s, u32 flags) seq_printf(s, "#%.*s / _----=> need-resched \n", size, spaces); seq_printf(s, "#%.*s| / _---=> hardirq/softirq \n", size, spaces); seq_printf(s, "#%.*s|| / _--=> preempt-depth \n", size, spaces); - seq_printf(s, "#%.*s||| / _-=> lock-depth \n", size, spaces); - seq_printf(s, "#%.*s|||| / \n", size, spaces); + seq_printf(s, "#%.*s||| / \n", size, spaces); } static void __print_graph_headers_flags(struct seq_file *s, u32 flags) @@ -1329,7 +1323,7 @@ static void __print_graph_headers_flags(struct seq_file *s, u32 flags) if (flags & TRACE_GRAPH_PRINT_PROC) seq_printf(s, " TASK/PID "); if (lat) - seq_printf(s, "|||||"); + seq_printf(s, "||||"); if (flags & TRACE_GRAPH_PRINT_DURATION) seq_printf(s, " DURATION "); seq_printf(s, " FUNCTION CALLS\n"); @@ -1343,7 +1337,7 @@ static void __print_graph_headers_flags(struct seq_file *s, u32 flags) if (flags & TRACE_GRAPH_PRINT_PROC) seq_printf(s, " | | "); if (lat) - seq_printf(s, "|||||"); + seq_printf(s, "||||"); if (flags & TRACE_GRAPH_PRINT_DURATION) seq_printf(s, " | | "); seq_printf(s, " | | | |\n"); @@ -1358,15 +1352,16 @@ void print_graph_headers_flags(struct seq_file *s, u32 flags) { struct trace_iterator *iter = s->private; + if (!(trace_flags & TRACE_ITER_CONTEXT_INFO)) + return; + if (trace_flags & TRACE_ITER_LATENCY_FMT) { /* print nothing if the buffers are empty */ if (trace_empty(iter)) return; print_trace_header(s, iter); - flags |= TRACE_GRAPH_PRINT_DURATION; - } else - flags |= TRACE_GRAPH_PRINT_ABS_TIME; + } __print_graph_headers_flags(s, flags); } diff --git a/kernel/trace/trace_irqsoff.c b/kernel/trace/trace_irqsoff.c index c77424be284d..667aa8cc0cfc 100644 --- a/kernel/trace/trace_irqsoff.c +++ b/kernel/trace/trace_irqsoff.c @@ -226,7 +226,9 @@ static void irqsoff_trace_close(struct trace_iterator *iter) } #define GRAPH_TRACER_FLAGS (TRACE_GRAPH_PRINT_CPU | \ - TRACE_GRAPH_PRINT_PROC) + TRACE_GRAPH_PRINT_PROC | \ + TRACE_GRAPH_PRINT_ABS_TIME | \ + TRACE_GRAPH_PRINT_DURATION) static enum print_line_t irqsoff_print_line(struct trace_iterator *iter) { diff --git a/kernel/trace/trace_kprobe.c b/kernel/trace/trace_kprobe.c index 27d13b36b8be..5fb3697bf0e5 100644 --- a/kernel/trace/trace_kprobe.c +++ b/kernel/trace/trace_kprobe.c @@ -343,6 +343,14 @@ DEFINE_BASIC_FETCH_FUNCS(deref) DEFINE_FETCH_deref(string) DEFINE_FETCH_deref(string_size) +static __kprobes void update_deref_fetch_param(struct deref_fetch_param *data) +{ + if (CHECK_FETCH_FUNCS(deref, data->orig.fn)) + update_deref_fetch_param(data->orig.data); + else if (CHECK_FETCH_FUNCS(symbol, data->orig.fn)) + update_symbol_cache(data->orig.data); +} + static __kprobes void free_deref_fetch_param(struct deref_fetch_param *data) { if (CHECK_FETCH_FUNCS(deref, data->orig.fn)) @@ -377,6 +385,19 @@ DEFINE_BASIC_FETCH_FUNCS(bitfield) #define fetch_bitfield_string_size NULL static __kprobes void +update_bitfield_fetch_param(struct bitfield_fetch_param *data) +{ + /* + * Don't check the bitfield itself, because this must be the + * last fetch function. + */ + if (CHECK_FETCH_FUNCS(deref, data->orig.fn)) + update_deref_fetch_param(data->orig.data); + else if (CHECK_FETCH_FUNCS(symbol, data->orig.fn)) + update_symbol_cache(data->orig.data); +} + +static __kprobes void free_bitfield_fetch_param(struct bitfield_fetch_param *data) { /* @@ -389,6 +410,7 @@ free_bitfield_fetch_param(struct bitfield_fetch_param *data) free_symbol_cache(data->orig.data); kfree(data); } + /* Default (unsigned long) fetch type */ #define __DEFAULT_FETCH_TYPE(t) u##t #define _DEFAULT_FETCH_TYPE(t) __DEFAULT_FETCH_TYPE(t) @@ -536,6 +558,7 @@ struct probe_arg { /* Flags for trace_probe */ #define TP_FLAG_TRACE 1 #define TP_FLAG_PROFILE 2 +#define TP_FLAG_REGISTERED 4 struct trace_probe { struct list_head list; @@ -555,16 +578,49 @@ struct trace_probe { (sizeof(struct probe_arg) * (n))) -static __kprobes int probe_is_return(struct trace_probe *tp) +static __kprobes int trace_probe_is_return(struct trace_probe *tp) { return tp->rp.handler != NULL; } -static __kprobes const char *probe_symbol(struct trace_probe *tp) +static __kprobes const char *trace_probe_symbol(struct trace_probe *tp) { return tp->symbol ? tp->symbol : "unknown"; } +static __kprobes unsigned long trace_probe_offset(struct trace_probe *tp) +{ + return tp->rp.kp.offset; +} + +static __kprobes bool trace_probe_is_enabled(struct trace_probe *tp) +{ + return !!(tp->flags & (TP_FLAG_TRACE | TP_FLAG_PROFILE)); +} + +static __kprobes bool trace_probe_is_registered(struct trace_probe *tp) +{ + return !!(tp->flags & TP_FLAG_REGISTERED); +} + +static __kprobes bool trace_probe_has_gone(struct trace_probe *tp) +{ + return !!(kprobe_gone(&tp->rp.kp)); +} + +static __kprobes bool trace_probe_within_module(struct trace_probe *tp, + struct module *mod) +{ + int len = strlen(mod->name); + const char *name = trace_probe_symbol(tp); + return strncmp(mod->name, name, len) == 0 && name[len] == ':'; +} + +static __kprobes bool trace_probe_is_on_module(struct trace_probe *tp) +{ + return !!strchr(trace_probe_symbol(tp), ':'); +} + static int register_probe_event(struct trace_probe *tp); static void unregister_probe_event(struct trace_probe *tp); @@ -646,6 +702,16 @@ error: return ERR_PTR(ret); } +static void update_probe_arg(struct probe_arg *arg) +{ + if (CHECK_FETCH_FUNCS(bitfield, arg->fetch.fn)) + update_bitfield_fetch_param(arg->fetch.data); + else if (CHECK_FETCH_FUNCS(deref, arg->fetch.fn)) + update_deref_fetch_param(arg->fetch.data); + else if (CHECK_FETCH_FUNCS(symbol, arg->fetch.fn)) + update_symbol_cache(arg->fetch.data); +} + static void free_probe_arg(struct probe_arg *arg) { if (CHECK_FETCH_FUNCS(bitfield, arg->fetch.fn)) @@ -671,7 +737,7 @@ static void free_trace_probe(struct trace_probe *tp) kfree(tp); } -static struct trace_probe *find_probe_event(const char *event, +static struct trace_probe *find_trace_probe(const char *event, const char *group) { struct trace_probe *tp; @@ -683,13 +749,96 @@ static struct trace_probe *find_probe_event(const char *event, return NULL; } +/* Enable trace_probe - @flag must be TP_FLAG_TRACE or TP_FLAG_PROFILE */ +static int enable_trace_probe(struct trace_probe *tp, int flag) +{ + int ret = 0; + + tp->flags |= flag; + if (trace_probe_is_enabled(tp) && trace_probe_is_registered(tp) && + !trace_probe_has_gone(tp)) { + if (trace_probe_is_return(tp)) + ret = enable_kretprobe(&tp->rp); + else + ret = enable_kprobe(&tp->rp.kp); + } + + return ret; +} + +/* Disable trace_probe - @flag must be TP_FLAG_TRACE or TP_FLAG_PROFILE */ +static void disable_trace_probe(struct trace_probe *tp, int flag) +{ + tp->flags &= ~flag; + if (!trace_probe_is_enabled(tp) && trace_probe_is_registered(tp)) { + if (trace_probe_is_return(tp)) + disable_kretprobe(&tp->rp); + else + disable_kprobe(&tp->rp.kp); + } +} + +/* Internal register function - just handle k*probes and flags */ +static int __register_trace_probe(struct trace_probe *tp) +{ + int i, ret; + + if (trace_probe_is_registered(tp)) + return -EINVAL; + + for (i = 0; i < tp->nr_args; i++) + update_probe_arg(&tp->args[i]); + + /* Set/clear disabled flag according to tp->flag */ + if (trace_probe_is_enabled(tp)) + tp->rp.kp.flags &= ~KPROBE_FLAG_DISABLED; + else + tp->rp.kp.flags |= KPROBE_FLAG_DISABLED; + + if (trace_probe_is_return(tp)) + ret = register_kretprobe(&tp->rp); + else + ret = register_kprobe(&tp->rp.kp); + + if (ret == 0) + tp->flags |= TP_FLAG_REGISTERED; + else { + pr_warning("Could not insert probe at %s+%lu: %d\n", + trace_probe_symbol(tp), trace_probe_offset(tp), ret); + if (ret == -ENOENT && trace_probe_is_on_module(tp)) { + pr_warning("This probe might be able to register after" + "target module is loaded. Continue.\n"); + ret = 0; + } else if (ret == -EILSEQ) { + pr_warning("Probing address(0x%p) is not an " + "instruction boundary.\n", + tp->rp.kp.addr); + ret = -EINVAL; + } + } + + return ret; +} + +/* Internal unregister function - just handle k*probes and flags */ +static void __unregister_trace_probe(struct trace_probe *tp) +{ + if (trace_probe_is_registered(tp)) { + if (trace_probe_is_return(tp)) + unregister_kretprobe(&tp->rp); + else + unregister_kprobe(&tp->rp.kp); + tp->flags &= ~TP_FLAG_REGISTERED; + /* Cleanup kprobe for reuse */ + if (tp->rp.kp.symbol_name) + tp->rp.kp.addr = NULL; + } +} + /* Unregister a trace_probe and probe_event: call with locking probe_lock */ static void unregister_trace_probe(struct trace_probe *tp) { - if (probe_is_return(tp)) - unregister_kretprobe(&tp->rp); - else - unregister_kprobe(&tp->rp.kp); + __unregister_trace_probe(tp); list_del(&tp->list); unregister_probe_event(tp); } @@ -702,41 +851,65 @@ static int register_trace_probe(struct trace_probe *tp) mutex_lock(&probe_lock); - /* register as an event */ - old_tp = find_probe_event(tp->call.name, tp->call.class->system); + /* Delete old (same name) event if exist */ + old_tp = find_trace_probe(tp->call.name, tp->call.class->system); if (old_tp) { - /* delete old event */ unregister_trace_probe(old_tp); free_trace_probe(old_tp); } + + /* Register new event */ ret = register_probe_event(tp); if (ret) { pr_warning("Failed to register probe event(%d)\n", ret); goto end; } - tp->rp.kp.flags |= KPROBE_FLAG_DISABLED; - if (probe_is_return(tp)) - ret = register_kretprobe(&tp->rp); - else - ret = register_kprobe(&tp->rp.kp); - - if (ret) { - pr_warning("Could not insert probe(%d)\n", ret); - if (ret == -EILSEQ) { - pr_warning("Probing address(0x%p) is not an " - "instruction boundary.\n", - tp->rp.kp.addr); - ret = -EINVAL; - } + /* Register k*probe */ + ret = __register_trace_probe(tp); + if (ret < 0) unregister_probe_event(tp); - } else + else list_add_tail(&tp->list, &probe_list); + end: mutex_unlock(&probe_lock); return ret; } +/* Module notifier call back, checking event on the module */ +static int trace_probe_module_callback(struct notifier_block *nb, + unsigned long val, void *data) +{ + struct module *mod = data; + struct trace_probe *tp; + int ret; + + if (val != MODULE_STATE_COMING) + return NOTIFY_DONE; + + /* Update probes on coming module */ + mutex_lock(&probe_lock); + list_for_each_entry(tp, &probe_list, list) { + if (trace_probe_within_module(tp, mod)) { + __unregister_trace_probe(tp); + ret = __register_trace_probe(tp); + if (ret) + pr_warning("Failed to re-register probe %s on" + "%s: %d\n", + tp->call.name, mod->name, ret); + } + } + mutex_unlock(&probe_lock); + + return NOTIFY_DONE; +} + +static struct notifier_block trace_probe_module_nb = { + .notifier_call = trace_probe_module_callback, + .priority = 1 /* Invoked after kprobe module callback */ +}; + /* Split symbol and offset. */ static int split_symbol_offset(char *symbol, unsigned long *offset) { @@ -962,8 +1135,8 @@ static int create_trace_probe(int argc, char **argv) { /* * Argument syntax: - * - Add kprobe: p[:[GRP/]EVENT] KSYM[+OFFS]|KADDR [FETCHARGS] - * - Add kretprobe: r[:[GRP/]EVENT] KSYM[+0] [FETCHARGS] + * - Add kprobe: p[:[GRP/]EVENT] [MOD:]KSYM[+OFFS]|KADDR [FETCHARGS] + * - Add kretprobe: r[:[GRP/]EVENT] [MOD:]KSYM[+0] [FETCHARGS] * Fetch args: * $retval : fetch return value * $stack : fetch stack address @@ -1025,7 +1198,7 @@ static int create_trace_probe(int argc, char **argv) return -EINVAL; } mutex_lock(&probe_lock); - tp = find_probe_event(event, group); + tp = find_trace_probe(event, group); if (!tp) { mutex_unlock(&probe_lock); pr_info("Event %s/%s doesn't exist.\n", group, event); @@ -1144,7 +1317,7 @@ error: return ret; } -static void cleanup_all_probes(void) +static void release_all_trace_probes(void) { struct trace_probe *tp; @@ -1158,7 +1331,6 @@ static void cleanup_all_probes(void) mutex_unlock(&probe_lock); } - /* Probes listing interfaces */ static void *probes_seq_start(struct seq_file *m, loff_t *pos) { @@ -1181,15 +1353,16 @@ static int probes_seq_show(struct seq_file *m, void *v) struct trace_probe *tp = v; int i; - seq_printf(m, "%c", probe_is_return(tp) ? 'r' : 'p'); + seq_printf(m, "%c", trace_probe_is_return(tp) ? 'r' : 'p'); seq_printf(m, ":%s/%s", tp->call.class->system, tp->call.name); if (!tp->symbol) seq_printf(m, " 0x%p", tp->rp.kp.addr); else if (tp->rp.kp.offset) - seq_printf(m, " %s+%u", probe_symbol(tp), tp->rp.kp.offset); + seq_printf(m, " %s+%u", trace_probe_symbol(tp), + tp->rp.kp.offset); else - seq_printf(m, " %s", probe_symbol(tp)); + seq_printf(m, " %s", trace_probe_symbol(tp)); for (i = 0; i < tp->nr_args; i++) seq_printf(m, " %s=%s", tp->args[i].name, tp->args[i].comm); @@ -1209,7 +1382,7 @@ static int probes_open(struct inode *inode, struct file *file) { if ((file->f_mode & FMODE_WRITE) && (file->f_flags & O_TRUNC)) - cleanup_all_probes(); + release_all_trace_probes(); return seq_open(file, &probes_seq_op); } @@ -1397,7 +1570,8 @@ static __kprobes void kprobe_trace_func(struct kprobe *kp, struct pt_regs *regs) store_trace_args(sizeof(*entry), tp, regs, (u8 *)&entry[1], dsize); if (!filter_current_check_discard(buffer, call, entry, event)) - trace_nowake_buffer_unlock_commit(buffer, event, irq_flags, pc); + trace_nowake_buffer_unlock_commit_regs(buffer, event, + irq_flags, pc, regs); } /* Kretprobe handler */ @@ -1429,7 +1603,8 @@ static __kprobes void kretprobe_trace_func(struct kretprobe_instance *ri, store_trace_args(sizeof(*entry), tp, regs, (u8 *)&entry[1], dsize); if (!filter_current_check_discard(buffer, call, entry, event)) - trace_nowake_buffer_unlock_commit(buffer, event, irq_flags, pc); + trace_nowake_buffer_unlock_commit_regs(buffer, event, + irq_flags, pc, regs); } /* Event entry printers */ @@ -1511,30 +1686,6 @@ partial: return TRACE_TYPE_PARTIAL_LINE; } -static int probe_event_enable(struct ftrace_event_call *call) -{ - struct trace_probe *tp = (struct trace_probe *)call->data; - - tp->flags |= TP_FLAG_TRACE; - if (probe_is_return(tp)) - return enable_kretprobe(&tp->rp); - else - return enable_kprobe(&tp->rp.kp); -} - -static void probe_event_disable(struct ftrace_event_call *call) -{ - struct trace_probe *tp = (struct trace_probe *)call->data; - - tp->flags &= ~TP_FLAG_TRACE; - if (!(tp->flags & (TP_FLAG_TRACE | TP_FLAG_PROFILE))) { - if (probe_is_return(tp)) - disable_kretprobe(&tp->rp); - else - disable_kprobe(&tp->rp.kp); - } -} - #undef DEFINE_FIELD #define DEFINE_FIELD(type, item, name, is_signed) \ do { \ @@ -1596,7 +1747,7 @@ static int __set_print_fmt(struct trace_probe *tp, char *buf, int len) const char *fmt, *arg; - if (!probe_is_return(tp)) { + if (!trace_probe_is_return(tp)) { fmt = "(%lx)"; arg = "REC->" FIELD_STRING_IP; } else { @@ -1713,49 +1864,25 @@ static __kprobes void kretprobe_perf_func(struct kretprobe_instance *ri, head = this_cpu_ptr(call->perf_events); perf_trace_buf_submit(entry, size, rctx, entry->ret_ip, 1, regs, head); } - -static int probe_perf_enable(struct ftrace_event_call *call) -{ - struct trace_probe *tp = (struct trace_probe *)call->data; - - tp->flags |= TP_FLAG_PROFILE; - - if (probe_is_return(tp)) - return enable_kretprobe(&tp->rp); - else - return enable_kprobe(&tp->rp.kp); -} - -static void probe_perf_disable(struct ftrace_event_call *call) -{ - struct trace_probe *tp = (struct trace_probe *)call->data; - - tp->flags &= ~TP_FLAG_PROFILE; - - if (!(tp->flags & TP_FLAG_TRACE)) { - if (probe_is_return(tp)) - disable_kretprobe(&tp->rp); - else - disable_kprobe(&tp->rp.kp); - } -} #endif /* CONFIG_PERF_EVENTS */ static __kprobes int kprobe_register(struct ftrace_event_call *event, enum trace_reg type) { + struct trace_probe *tp = (struct trace_probe *)event->data; + switch (type) { case TRACE_REG_REGISTER: - return probe_event_enable(event); + return enable_trace_probe(tp, TP_FLAG_TRACE); case TRACE_REG_UNREGISTER: - probe_event_disable(event); + disable_trace_probe(tp, TP_FLAG_TRACE); return 0; #ifdef CONFIG_PERF_EVENTS case TRACE_REG_PERF_REGISTER: - return probe_perf_enable(event); + return enable_trace_probe(tp, TP_FLAG_PROFILE); case TRACE_REG_PERF_UNREGISTER: - probe_perf_disable(event); + disable_trace_probe(tp, TP_FLAG_PROFILE); return 0; #endif } @@ -1805,7 +1932,7 @@ static int register_probe_event(struct trace_probe *tp) /* Initialize ftrace_event_call */ INIT_LIST_HEAD(&call->class->fields); - if (probe_is_return(tp)) { + if (trace_probe_is_return(tp)) { call->event.funcs = &kretprobe_funcs; call->class->define_fields = kretprobe_event_define_fields; } else { @@ -1844,6 +1971,9 @@ static __init int init_kprobe_trace(void) struct dentry *d_tracer; struct dentry *entry; + if (register_module_notifier(&trace_probe_module_nb)) + return -EINVAL; + d_tracer = tracing_init_dentry(); if (!d_tracer) return 0; @@ -1897,12 +2027,12 @@ static __init int kprobe_trace_self_tests_init(void) warn++; } else { /* Enable trace point */ - tp = find_probe_event("testprobe", KPROBE_EVENT_SYSTEM); + tp = find_trace_probe("testprobe", KPROBE_EVENT_SYSTEM); if (WARN_ON_ONCE(tp == NULL)) { pr_warning("error on getting new probe.\n"); warn++; } else - probe_event_enable(&tp->call); + enable_trace_probe(tp, TP_FLAG_TRACE); } ret = command_trace_probe("r:testprobe2 kprobe_trace_selftest_target " @@ -1912,12 +2042,12 @@ static __init int kprobe_trace_self_tests_init(void) warn++; } else { /* Enable trace point */ - tp = find_probe_event("testprobe2", KPROBE_EVENT_SYSTEM); + tp = find_trace_probe("testprobe2", KPROBE_EVENT_SYSTEM); if (WARN_ON_ONCE(tp == NULL)) { pr_warning("error on getting new probe.\n"); warn++; } else - probe_event_enable(&tp->call); + enable_trace_probe(tp, TP_FLAG_TRACE); } if (warn) @@ -1938,7 +2068,7 @@ static __init int kprobe_trace_self_tests_init(void) } end: - cleanup_all_probes(); + release_all_trace_probes(); if (warn) pr_cont("NG: Some tests are failed. Please check them.\n"); else diff --git a/kernel/trace/trace_output.c b/kernel/trace/trace_output.c index e37de492a9e1..51999309a6cf 100644 --- a/kernel/trace/trace_output.c +++ b/kernel/trace/trace_output.c @@ -1107,19 +1107,20 @@ static enum print_line_t trace_stack_print(struct trace_iterator *iter, { struct stack_entry *field; struct trace_seq *s = &iter->seq; - int i; + unsigned long *p; + unsigned long *end; trace_assign_type(field, iter->ent); + end = (unsigned long *)((long)iter->ent + iter->ent_size); if (!trace_seq_puts(s, "<stack trace>\n")) goto partial; - for (i = 0; i < FTRACE_STACK_ENTRIES; i++) { - if (!field->caller[i] || (field->caller[i] == ULONG_MAX)) - break; + + for (p = field->caller; p && *p != ULONG_MAX && p < end; p++) { if (!trace_seq_puts(s, " => ")) goto partial; - if (!seq_print_ip_sym(s, field->caller[i], flags)) + if (!seq_print_ip_sym(s, *p, flags)) goto partial; if (!trace_seq_puts(s, "\n")) goto partial; diff --git a/kernel/trace/trace_sched_wakeup.c b/kernel/trace/trace_sched_wakeup.c index f029dd4fd2ca..e4a70c0c71b6 100644 --- a/kernel/trace/trace_sched_wakeup.c +++ b/kernel/trace/trace_sched_wakeup.c @@ -227,7 +227,9 @@ static void wakeup_trace_close(struct trace_iterator *iter) graph_trace_close(iter); } -#define GRAPH_TRACER_FLAGS (TRACE_GRAPH_PRINT_PROC) +#define GRAPH_TRACER_FLAGS (TRACE_GRAPH_PRINT_PROC | \ + TRACE_GRAPH_PRINT_ABS_TIME | \ + TRACE_GRAPH_PRINT_DURATION) static enum print_line_t wakeup_print_line(struct trace_iterator *iter) { diff --git a/kernel/trace/trace_stack.c b/kernel/trace/trace_stack.c index b0b53b8e4c25..77575b386d97 100644 --- a/kernel/trace/trace_stack.c +++ b/kernel/trace/trace_stack.c @@ -156,20 +156,11 @@ stack_max_size_write(struct file *filp, const char __user *ubuf, { long *ptr = filp->private_data; unsigned long val, flags; - char buf[64]; int ret; int cpu; - if (count >= sizeof(buf)) - return -EINVAL; - - if (copy_from_user(&buf, ubuf, count)) - return -EFAULT; - - buf[count] = 0; - - ret = strict_strtoul(buf, 10, &val); - if (ret < 0) + ret = kstrtoul_from_user(ubuf, count, 10, &val); + if (ret) return ret; local_irq_save(flags); diff --git a/kernel/watchdog.c b/kernel/watchdog.c index 3d0c56ad4792..36491cd5b7d4 100644 --- a/kernel/watchdog.c +++ b/kernel/watchdog.c @@ -200,6 +200,7 @@ static int is_softlockup(unsigned long touch_ts) } #ifdef CONFIG_HARDLOCKUP_DETECTOR + static struct perf_event_attr wd_hw_attr = { .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_CPU_CYCLES, @@ -209,7 +210,7 @@ static struct perf_event_attr wd_hw_attr = { }; /* Callback function for perf event subsystem */ -static void watchdog_overflow_callback(struct perf_event *event, int nmi, +static void watchdog_overflow_callback(struct perf_event *event, struct perf_sample_data *data, struct pt_regs *regs) { @@ -368,10 +369,11 @@ static int watchdog_nmi_enable(int cpu) if (event != NULL) goto out_enable; - /* Try to register using hardware perf events */ wd_attr = &wd_hw_attr; wd_attr->sample_period = hw_nmi_get_sample_period(watchdog_thresh); - event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback); + + /* Try to register using hardware perf events */ + event = perf_event_create_kernel_counter(wd_attr, cpu, NULL, watchdog_overflow_callback, NULL); if (!IS_ERR(event)) { printk(KERN_INFO "NMI watchdog enabled, takes one hw-pmu counter.\n"); goto out_save; |