diff options
Diffstat (limited to 'kernel')
| -rw-r--r-- | kernel/futex.c | 100 | ||||
| -rw-r--r-- | kernel/jump_label.c | 158 | ||||
| -rw-r--r-- | kernel/locking/Makefile | 1 | ||||
| -rw-r--r-- | kernel/locking/qrwlock.c | 47 | ||||
| -rw-r--r-- | kernel/locking/qspinlock.c | 6 | ||||
| -rw-r--r-- | kernel/locking/qspinlock_paravirt.h | 102 | ||||
| -rw-r--r-- | kernel/locking/rtmutex-tester.c | 420 | ||||
| -rw-r--r-- | kernel/locking/rtmutex.c | 2 | ||||
| -rw-r--r-- | kernel/locking/rtmutex_common.h | 22 | ||||
| -rw-r--r-- | kernel/sched/core.c | 6 |
10 files changed, 306 insertions, 558 deletions
diff --git a/kernel/futex.c b/kernel/futex.c index c4a182f5357e..6e443efc65f4 100644 --- a/kernel/futex.c +++ b/kernel/futex.c @@ -64,6 +64,7 @@ #include <linux/hugetlb.h> #include <linux/freezer.h> #include <linux/bootmem.h> +#include <linux/fault-inject.h> #include <asm/futex.h> @@ -258,6 +259,66 @@ static unsigned long __read_mostly futex_hashsize; static struct futex_hash_bucket *futex_queues; +/* + * Fault injections for futexes. + */ +#ifdef CONFIG_FAIL_FUTEX + +static struct { + struct fault_attr attr; + + u32 ignore_private; +} fail_futex = { + .attr = FAULT_ATTR_INITIALIZER, + .ignore_private = 0, +}; + +static int __init setup_fail_futex(char *str) +{ + return setup_fault_attr(&fail_futex.attr, str); +} +__setup("fail_futex=", setup_fail_futex); + +static bool should_fail_futex(bool fshared) +{ + if (fail_futex.ignore_private && !fshared) + return false; + + return should_fail(&fail_futex.attr, 1); +} + +#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS + +static int __init fail_futex_debugfs(void) +{ + umode_t mode = S_IFREG | S_IRUSR | S_IWUSR; + struct dentry *dir; + + dir = fault_create_debugfs_attr("fail_futex", NULL, + &fail_futex.attr); + if (IS_ERR(dir)) + return PTR_ERR(dir); + + if (!debugfs_create_bool("ignore-private", mode, dir, + &fail_futex.ignore_private)) { + debugfs_remove_recursive(dir); + return -ENOMEM; + } + + return 0; +} + +late_initcall(fail_futex_debugfs); + +#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */ + +#else +static inline bool should_fail_futex(bool fshared) +{ + return false; +} +#endif /* CONFIG_FAIL_FUTEX */ + static inline void futex_get_mm(union futex_key *key) { atomic_inc(&key->private.mm->mm_count); @@ -413,6 +474,9 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw) if (unlikely(!access_ok(rw, uaddr, sizeof(u32)))) return -EFAULT; + if (unlikely(should_fail_futex(fshared))) + return -EFAULT; + /* * PROCESS_PRIVATE futexes are fast. * As the mm cannot disappear under us and the 'key' only needs @@ -428,6 +492,10 @@ get_futex_key(u32 __user *uaddr, int fshared, union futex_key *key, int rw) } again: + /* Ignore any VERIFY_READ mapping (futex common case) */ + if (unlikely(should_fail_futex(fshared))) + return -EFAULT; + err = get_user_pages_fast(address, 1, 1, &page); /* * If write access is not required (eg. FUTEX_WAIT), try @@ -516,7 +584,7 @@ again: * A RO anonymous page will never change and thus doesn't make * sense for futex operations. */ - if (ro) { + if (unlikely(should_fail_futex(fshared)) || ro) { err = -EFAULT; goto out; } @@ -974,6 +1042,9 @@ static int lock_pi_update_atomic(u32 __user *uaddr, u32 uval, u32 newval) { u32 uninitialized_var(curval); + if (unlikely(should_fail_futex(true))) + return -EFAULT; + if (unlikely(cmpxchg_futex_value_locked(&curval, uaddr, uval, newval))) return -EFAULT; @@ -1015,12 +1086,18 @@ static int futex_lock_pi_atomic(u32 __user *uaddr, struct futex_hash_bucket *hb, if (get_futex_value_locked(&uval, uaddr)) return -EFAULT; + if (unlikely(should_fail_futex(true))) + return -EFAULT; + /* * Detect deadlocks. */ if ((unlikely((uval & FUTEX_TID_MASK) == vpid))) return -EDEADLK; + if ((unlikely(should_fail_futex(true)))) + return -EDEADLK; + /* * Lookup existing state first. If it exists, try to attach to * its pi_state. @@ -1155,6 +1232,9 @@ static int wake_futex_pi(u32 __user *uaddr, u32 uval, struct futex_q *this, */ newval = FUTEX_WAITERS | task_pid_vnr(new_owner); + if (unlikely(should_fail_futex(true))) + ret = -EFAULT; + if (cmpxchg_futex_value_locked(&curval, uaddr, uval, newval)) ret = -EFAULT; else if (curval != uval) @@ -1457,6 +1537,9 @@ static int futex_proxy_trylock_atomic(u32 __user *pifutex, if (get_futex_value_locked(&curval, pifutex)) return -EFAULT; + if (unlikely(should_fail_futex(true))) + return -EFAULT; + /* * Find the top_waiter and determine if there are additional waiters. * If the caller intends to requeue more than 1 waiter to pifutex, @@ -2268,8 +2351,11 @@ static long futex_wait_restart(struct restart_block *restart) /* * Userspace tried a 0 -> TID atomic transition of the futex value * and failed. The kernel side here does the whole locking operation: - * if there are waiters then it will block, it does PI, etc. (Due to - * races the kernel might see a 0 value of the futex too.) + * if there are waiters then it will block as a consequence of relying + * on rt-mutexes, it does PI, etc. (Due to races the kernel might see + * a 0 value of the futex too.). + * + * Also serves as futex trylock_pi()'ing, and due semantics. */ static int futex_lock_pi(u32 __user *uaddr, unsigned int flags, ktime_t *time, int trylock) @@ -2300,6 +2386,10 @@ retry_private: ret = futex_lock_pi_atomic(uaddr, hb, &q.key, &q.pi_state, current, 0); if (unlikely(ret)) { + /* + * Atomic work succeeded and we got the lock, + * or failed. Either way, we do _not_ block. + */ switch (ret) { case 1: /* We got the lock. */ @@ -2530,7 +2620,7 @@ int handle_early_requeue_pi_wakeup(struct futex_hash_bucket *hb, * futex_wait_requeue_pi() - Wait on uaddr and take uaddr2 * @uaddr: the futex we initially wait on (non-pi) * @flags: futex flags (FLAGS_SHARED, FLAGS_CLOCKRT, etc.), they must be - * the same type, no requeueing from private to shared, etc. + * the same type, no requeueing from private to shared, etc. * @val: the expected value of uaddr * @abs_time: absolute timeout * @bitset: 32 bit wakeup bitset set by userspace, defaults to all @@ -3005,6 +3095,8 @@ SYSCALL_DEFINE6(futex, u32 __user *, uaddr, int, op, u32, val, if (utime && (cmd == FUTEX_WAIT || cmd == FUTEX_LOCK_PI || cmd == FUTEX_WAIT_BITSET || cmd == FUTEX_WAIT_REQUEUE_PI)) { + if (unlikely(should_fail_futex(!(op & FUTEX_PRIVATE_FLAG)))) + return -EFAULT; if (copy_from_user(&ts, utime, sizeof(ts)) != 0) return -EFAULT; if (!timespec_valid(&ts)) diff --git a/kernel/jump_label.c b/kernel/jump_label.c index 52ebaca1b9fc..f7dd15d537f9 100644 --- a/kernel/jump_label.c +++ b/kernel/jump_label.c @@ -54,7 +54,7 @@ jump_label_sort_entries(struct jump_entry *start, struct jump_entry *stop) sort(start, size, sizeof(struct jump_entry), jump_label_cmp, NULL); } -static void jump_label_update(struct static_key *key, int enable); +static void jump_label_update(struct static_key *key); void static_key_slow_inc(struct static_key *key) { @@ -63,13 +63,8 @@ void static_key_slow_inc(struct static_key *key) return; jump_label_lock(); - if (atomic_read(&key->enabled) == 0) { - if (!jump_label_get_branch_default(key)) - jump_label_update(key, JUMP_LABEL_ENABLE); - else - jump_label_update(key, JUMP_LABEL_DISABLE); - } - atomic_inc(&key->enabled); + if (atomic_inc_return(&key->enabled) == 1) + jump_label_update(key); jump_label_unlock(); } EXPORT_SYMBOL_GPL(static_key_slow_inc); @@ -87,10 +82,7 @@ static void __static_key_slow_dec(struct static_key *key, atomic_inc(&key->enabled); schedule_delayed_work(work, rate_limit); } else { - if (!jump_label_get_branch_default(key)) - jump_label_update(key, JUMP_LABEL_DISABLE); - else - jump_label_update(key, JUMP_LABEL_ENABLE); + jump_label_update(key); } jump_label_unlock(); } @@ -149,7 +141,7 @@ static int __jump_label_text_reserved(struct jump_entry *iter_start, return 0; } -/* +/* * Update code which is definitely not currently executing. * Architectures which need heavyweight synchronization to modify * running code can override this to make the non-live update case @@ -158,37 +150,54 @@ static int __jump_label_text_reserved(struct jump_entry *iter_start, void __weak __init_or_module arch_jump_label_transform_static(struct jump_entry *entry, enum jump_label_type type) { - arch_jump_label_transform(entry, type); + arch_jump_label_transform(entry, type); +} + +static inline struct jump_entry *static_key_entries(struct static_key *key) +{ + return (struct jump_entry *)((unsigned long)key->entries & ~JUMP_TYPE_MASK); +} + +static inline bool static_key_type(struct static_key *key) +{ + return (unsigned long)key->entries & JUMP_TYPE_MASK; +} + +static inline struct static_key *jump_entry_key(struct jump_entry *entry) +{ + return (struct static_key *)((unsigned long)entry->key & ~1UL); +} + +static bool jump_entry_branch(struct jump_entry *entry) +{ + return (unsigned long)entry->key & 1UL; +} + +static enum jump_label_type jump_label_type(struct jump_entry *entry) +{ + struct static_key *key = jump_entry_key(entry); + bool enabled = static_key_enabled(key); + bool branch = jump_entry_branch(entry); + + /* See the comment in linux/jump_label.h */ + return enabled ^ branch; } static void __jump_label_update(struct static_key *key, struct jump_entry *entry, - struct jump_entry *stop, int enable) + struct jump_entry *stop) { - for (; (entry < stop) && - (entry->key == (jump_label_t)(unsigned long)key); - entry++) { + for (; (entry < stop) && (jump_entry_key(entry) == key); entry++) { /* * entry->code set to 0 invalidates module init text sections * kernel_text_address() verifies we are not in core kernel * init code, see jump_label_invalidate_module_init(). */ if (entry->code && kernel_text_address(entry->code)) - arch_jump_label_transform(entry, enable); + arch_jump_label_transform(entry, jump_label_type(entry)); } } -static enum jump_label_type jump_label_type(struct static_key *key) -{ - bool true_branch = jump_label_get_branch_default(key); - bool state = static_key_enabled(key); - - if ((!true_branch && state) || (true_branch && !state)) - return JUMP_LABEL_ENABLE; - - return JUMP_LABEL_DISABLE; -} - void __init jump_label_init(void) { struct jump_entry *iter_start = __start___jump_table; @@ -202,8 +211,11 @@ void __init jump_label_init(void) for (iter = iter_start; iter < iter_stop; iter++) { struct static_key *iterk; - iterk = (struct static_key *)(unsigned long)iter->key; - arch_jump_label_transform_static(iter, jump_label_type(iterk)); + /* rewrite NOPs */ + if (jump_label_type(iter) == JUMP_LABEL_NOP) + arch_jump_label_transform_static(iter, JUMP_LABEL_NOP); + + iterk = jump_entry_key(iter); if (iterk == key) continue; @@ -222,6 +234,16 @@ void __init jump_label_init(void) #ifdef CONFIG_MODULES +static enum jump_label_type jump_label_init_type(struct jump_entry *entry) +{ + struct static_key *key = jump_entry_key(entry); + bool type = static_key_type(key); + bool branch = jump_entry_branch(entry); + + /* See the comment in linux/jump_label.h */ + return type ^ branch; +} + struct static_key_mod { struct static_key_mod *next; struct jump_entry *entries; @@ -243,17 +265,15 @@ static int __jump_label_mod_text_reserved(void *start, void *end) start, end); } -static void __jump_label_mod_update(struct static_key *key, int enable) +static void __jump_label_mod_update(struct static_key *key) { - struct static_key_mod *mod = key->next; + struct static_key_mod *mod; - while (mod) { + for (mod = key->next; mod; mod = mod->next) { struct module *m = mod->mod; __jump_label_update(key, mod->entries, - m->jump_entries + m->num_jump_entries, - enable); - mod = mod->next; + m->jump_entries + m->num_jump_entries); } } @@ -276,7 +296,9 @@ void jump_label_apply_nops(struct module *mod) return; for (iter = iter_start; iter < iter_stop; iter++) { - arch_jump_label_transform_static(iter, JUMP_LABEL_DISABLE); + /* Only write NOPs for arch_branch_static(). */ + if (jump_label_init_type(iter) == JUMP_LABEL_NOP) + arch_jump_label_transform_static(iter, JUMP_LABEL_NOP); } } @@ -297,7 +319,7 @@ static int jump_label_add_module(struct module *mod) for (iter = iter_start; iter < iter_stop; iter++) { struct static_key *iterk; - iterk = (struct static_key *)(unsigned long)iter->key; + iterk = jump_entry_key(iter); if (iterk == key) continue; @@ -318,8 +340,9 @@ static int jump_label_add_module(struct module *mod) jlm->next = key->next; key->next = jlm; - if (jump_label_type(key) == JUMP_LABEL_ENABLE) - __jump_label_update(key, iter, iter_stop, JUMP_LABEL_ENABLE); + /* Only update if we've changed from our initial state */ + if (jump_label_type(iter) != jump_label_init_type(iter)) + __jump_label_update(key, iter, iter_stop); } return 0; @@ -334,10 +357,10 @@ static void jump_label_del_module(struct module *mod) struct static_key_mod *jlm, **prev; for (iter = iter_start; iter < iter_stop; iter++) { - if (iter->key == (jump_label_t)(unsigned long)key) + if (jump_entry_key(iter) == key) continue; - key = (struct static_key *)(unsigned long)iter->key; + key = jump_entry_key(iter); if (within_module(iter->key, mod)) continue; @@ -439,14 +462,14 @@ int jump_label_text_reserved(void *start, void *end) return ret; } -static void jump_label_update(struct static_key *key, int enable) +static void jump_label_update(struct static_key *key) { struct jump_entry *stop = __stop___jump_table; - struct jump_entry *entry = jump_label_get_entries(key); + struct jump_entry *entry = static_key_entries(key); #ifdef CONFIG_MODULES struct module *mod; - __jump_label_mod_update(key, enable); + __jump_label_mod_update(key); preempt_disable(); mod = __module_address((unsigned long)key); @@ -456,7 +479,44 @@ static void jump_label_update(struct static_key *key, int enable) #endif /* if there are no users, entry can be NULL */ if (entry) - __jump_label_update(key, entry, stop, enable); + __jump_label_update(key, entry, stop); } -#endif +#ifdef CONFIG_STATIC_KEYS_SELFTEST +static DEFINE_STATIC_KEY_TRUE(sk_true); +static DEFINE_STATIC_KEY_FALSE(sk_false); + +static __init int jump_label_test(void) +{ + int i; + + for (i = 0; i < 2; i++) { + WARN_ON(static_key_enabled(&sk_true.key) != true); + WARN_ON(static_key_enabled(&sk_false.key) != false); + + WARN_ON(!static_branch_likely(&sk_true)); + WARN_ON(!static_branch_unlikely(&sk_true)); + WARN_ON(static_branch_likely(&sk_false)); + WARN_ON(static_branch_unlikely(&sk_false)); + + static_branch_disable(&sk_true); + static_branch_enable(&sk_false); + + WARN_ON(static_key_enabled(&sk_true.key) == true); + WARN_ON(static_key_enabled(&sk_false.key) == false); + + WARN_ON(static_branch_likely(&sk_true)); + WARN_ON(static_branch_unlikely(&sk_true)); + WARN_ON(!static_branch_likely(&sk_false)); + WARN_ON(!static_branch_unlikely(&sk_false)); + + static_branch_enable(&sk_true); + static_branch_disable(&sk_false); + } + + return 0; +} +late_initcall(jump_label_test); +#endif /* STATIC_KEYS_SELFTEST */ + +#endif /* HAVE_JUMP_LABEL */ diff --git a/kernel/locking/Makefile b/kernel/locking/Makefile index 7dd5c9918e4c..36942047ffc0 100644 --- a/kernel/locking/Makefile +++ b/kernel/locking/Makefile @@ -20,7 +20,6 @@ obj-$(CONFIG_PROVE_LOCKING) += spinlock.o obj-$(CONFIG_QUEUED_SPINLOCKS) += qspinlock.o obj-$(CONFIG_RT_MUTEXES) += rtmutex.o obj-$(CONFIG_DEBUG_RT_MUTEXES) += rtmutex-debug.o -obj-$(CONFIG_RT_MUTEX_TESTER) += rtmutex-tester.o obj-$(CONFIG_DEBUG_SPINLOCK) += spinlock.o obj-$(CONFIG_DEBUG_SPINLOCK) += spinlock_debug.o obj-$(CONFIG_RWSEM_GENERIC_SPINLOCK) += rwsem-spinlock.o diff --git a/kernel/locking/qrwlock.c b/kernel/locking/qrwlock.c index 6c5da483966b..f17a3e3b3550 100644 --- a/kernel/locking/qrwlock.c +++ b/kernel/locking/qrwlock.c @@ -55,27 +55,29 @@ rspin_until_writer_unlock(struct qrwlock *lock, u32 cnts) { while ((cnts & _QW_WMASK) == _QW_LOCKED) { cpu_relax_lowlatency(); - cnts = smp_load_acquire((u32 *)&lock->cnts); + cnts = atomic_read_acquire(&lock->cnts); } } /** - * queue_read_lock_slowpath - acquire read lock of a queue rwlock + * queued_read_lock_slowpath - acquire read lock of a queue rwlock * @lock: Pointer to queue rwlock structure + * @cnts: Current qrwlock lock value */ -void queue_read_lock_slowpath(struct qrwlock *lock) +void queued_read_lock_slowpath(struct qrwlock *lock, u32 cnts) { - u32 cnts; - /* * Readers come here when they cannot get the lock without waiting */ if (unlikely(in_interrupt())) { /* - * Readers in interrupt context will spin until the lock is - * available without waiting in the queue. + * Readers in interrupt context will get the lock immediately + * if the writer is just waiting (not holding the lock yet). + * The rspin_until_writer_unlock() function returns immediately + * in this case. Otherwise, they will spin (with ACQUIRE + * semantics) until the lock is available without waiting in + * the queue. */ - cnts = smp_load_acquire((u32 *)&lock->cnts); rspin_until_writer_unlock(lock, cnts); return; } @@ -87,16 +89,11 @@ void queue_read_lock_slowpath(struct qrwlock *lock) arch_spin_lock(&lock->lock); /* - * At the head of the wait queue now, wait until the writer state - * goes to 0 and then try to increment the reader count and get - * the lock. It is possible that an incoming writer may steal the - * lock in the interim, so it is necessary to check the writer byte - * to make sure that the write lock isn't taken. + * The ACQUIRE semantics of the following spinning code ensure + * that accesses can't leak upwards out of our subsequent critical + * section in the case that the lock is currently held for write. */ - while (atomic_read(&lock->cnts) & _QW_WMASK) - cpu_relax_lowlatency(); - - cnts = atomic_add_return(_QR_BIAS, &lock->cnts) - _QR_BIAS; + cnts = atomic_add_return_acquire(_QR_BIAS, &lock->cnts) - _QR_BIAS; rspin_until_writer_unlock(lock, cnts); /* @@ -104,13 +101,13 @@ void queue_read_lock_slowpath(struct qrwlock *lock) */ arch_spin_unlock(&lock->lock); } -EXPORT_SYMBOL(queue_read_lock_slowpath); +EXPORT_SYMBOL(queued_read_lock_slowpath); /** - * queue_write_lock_slowpath - acquire write lock of a queue rwlock + * queued_write_lock_slowpath - acquire write lock of a queue rwlock * @lock : Pointer to queue rwlock structure */ -void queue_write_lock_slowpath(struct qrwlock *lock) +void queued_write_lock_slowpath(struct qrwlock *lock) { u32 cnts; @@ -119,7 +116,7 @@ void queue_write_lock_slowpath(struct qrwlock *lock) /* Try to acquire the lock directly if no reader is present */ if (!atomic_read(&lock->cnts) && - (atomic_cmpxchg(&lock->cnts, 0, _QW_LOCKED) == 0)) + (atomic_cmpxchg_acquire(&lock->cnts, 0, _QW_LOCKED) == 0)) goto unlock; /* @@ -130,7 +127,7 @@ void queue_write_lock_slowpath(struct qrwlock *lock) struct __qrwlock *l = (struct __qrwlock *)lock; if (!READ_ONCE(l->wmode) && - (cmpxchg(&l->wmode, 0, _QW_WAITING) == 0)) + (cmpxchg_relaxed(&l->wmode, 0, _QW_WAITING) == 0)) break; cpu_relax_lowlatency(); @@ -140,8 +137,8 @@ void queue_write_lock_slowpath(struct qrwlock *lock) for (;;) { cnts = atomic_read(&lock->cnts); if ((cnts == _QW_WAITING) && - (atomic_cmpxchg(&lock->cnts, _QW_WAITING, - _QW_LOCKED) == _QW_WAITING)) + (atomic_cmpxchg_acquire(&lock->cnts, _QW_WAITING, + _QW_LOCKED) == _QW_WAITING)) break; cpu_relax_lowlatency(); @@ -149,4 +146,4 @@ void queue_write_lock_slowpath(struct qrwlock *lock) unlock: arch_spin_unlock(&lock->lock); } -EXPORT_SYMBOL(queue_write_lock_slowpath); +EXPORT_SYMBOL(queued_write_lock_slowpath); diff --git a/kernel/locking/qspinlock.c b/kernel/locking/qspinlock.c index 38c49202d532..337c8818541d 100644 --- a/kernel/locking/qspinlock.c +++ b/kernel/locking/qspinlock.c @@ -239,8 +239,8 @@ static __always_inline void set_locked(struct qspinlock *lock) static __always_inline void __pv_init_node(struct mcs_spinlock *node) { } static __always_inline void __pv_wait_node(struct mcs_spinlock *node) { } -static __always_inline void __pv_kick_node(struct mcs_spinlock *node) { } - +static __always_inline void __pv_kick_node(struct qspinlock *lock, + struct mcs_spinlock *node) { } static __always_inline void __pv_wait_head(struct qspinlock *lock, struct mcs_spinlock *node) { } @@ -440,7 +440,7 @@ queue: cpu_relax(); arch_mcs_spin_unlock_contended(&next->locked); - pv_kick_node(next); + pv_kick_node(lock, next); release: /* diff --git a/kernel/locking/qspinlock_paravirt.h b/kernel/locking/qspinlock_paravirt.h index df19ae4debd0..c8e6e9a596f5 100644 --- a/kernel/locking/qspinlock_paravirt.h +++ b/kernel/locking/qspinlock_paravirt.h @@ -22,9 +22,14 @@ #define _Q_SLOW_VAL (3U << _Q_LOCKED_OFFSET) +/* + * Queue node uses: vcpu_running & vcpu_halted. + * Queue head uses: vcpu_running & vcpu_hashed. + */ enum vcpu_state { vcpu_running = 0, - vcpu_halted, + vcpu_halted, /* Used only in pv_wait_node */ + vcpu_hashed, /* = pv_hash'ed + vcpu_halted */ }; struct pv_node { @@ -153,7 +158,8 @@ static void pv_init_node(struct mcs_spinlock *node) /* * Wait for node->locked to become true, halt the vcpu after a short spin. - * pv_kick_node() is used to wake the vcpu again. + * pv_kick_node() is used to set _Q_SLOW_VAL and fill in hash table on its + * behalf. */ static void pv_wait_node(struct mcs_spinlock *node) { @@ -172,9 +178,9 @@ static void pv_wait_node(struct mcs_spinlock *node) * * [S] pn->state = vcpu_halted [S] next->locked = 1 * MB MB - * [L] pn->locked [RmW] pn->state = vcpu_running + * [L] pn->locked [RmW] pn->state = vcpu_hashed * - * Matches the xchg() from pv_kick_node(). + * Matches the cmpxchg() from pv_kick_node(). */ smp_store_mb(pn->state, vcpu_halted); @@ -182,9 +188,10 @@ static void pv_wait_node(struct mcs_spinlock *node) pv_wait(&pn->state, vcpu_halted); /* - * Reset the vCPU state to avoid unncessary CPU kicking + * If pv_kick_node() changed us to vcpu_hashed, retain that value + * so that pv_wait_head() knows to not also try to hash this lock. */ - WRITE_ONCE(pn->state, vcpu_running); + cmpxchg(&pn->state, vcpu_halted, vcpu_running); /* * If the locked flag is still not set after wakeup, it is a @@ -194,6 +201,7 @@ static void pv_wait_node(struct mcs_spinlock *node) * MCS lock will be released soon. */ } + /* * By now our node->locked should be 1 and our caller will not actually * spin-wait for it. We do however rely on our caller to do a @@ -202,24 +210,35 @@ static void pv_wait_node(struct mcs_spinlock *node) } /* - * Called after setting next->locked = 1, used to wake those stuck in - * pv_wait_node(). + * Called after setting next->locked = 1 when we're the lock owner. + * + * Instead of waking the waiters stuck in pv_wait_node() advance their state such + * that they're waiting in pv_wait_head(), this avoids a wake/sleep cycle. */ -static void pv_kick_node(struct mcs_spinlock *node) +static void pv_kick_node(struct qspinlock *lock, struct mcs_spinlock *node) { struct pv_node *pn = (struct pv_node *)node; + struct __qspinlock *l = (void *)lock; /* - * Note that because node->locked is already set, this actual - * mcs_spinlock entry could be re-used already. + * If the vCPU is indeed halted, advance its state to match that of + * pv_wait_node(). If OTOH this fails, the vCPU was running and will + * observe its next->locked value and advance itself. * - * This should be fine however, kicking people for no reason is - * harmless. + * Matches with smp_store_mb() and cmpxchg() in pv_wait_node() + */ + if (cmpxchg(&pn->state, vcpu_halted, vcpu_hashed) != vcpu_halted) + return; + + /* + * Put the lock into the hash table and set the _Q_SLOW_VAL. * - * See the comment in pv_wait_node(). + * As this is the same vCPU that will check the _Q_SLOW_VAL value and + * the hash table later on at unlock time, no atomic instruction is + * needed. */ - if (xchg(&pn->state, vcpu_running) == vcpu_halted) - pv_kick(pn->cpu); + WRITE_ONCE(l->locked, _Q_SLOW_VAL); + (void)pv_hash(lock, pn); } /* @@ -233,6 +252,13 @@ static void pv_wait_head(struct qspinlock *lock, struct mcs_spinlock *node) struct qspinlock **lp = NULL; int loop; + /* + * If pv_kick_node() already advanced our state, we don't need to + * insert ourselves into the hash table anymore. + */ + if (READ_ONCE(pn->state) == vcpu_hashed) + lp = (struct qspinlock **)1; + for (;;) { for (loop = SPIN_THRESHOLD; loop; loop--) { if (!READ_ONCE(l->locked)) @@ -240,17 +266,22 @@ static void pv_wait_head(struct qspinlock *lock, struct mcs_spinlock *node) cpu_relax(); } - WRITE_ONCE(pn->state, vcpu_halted); if (!lp) { /* ONCE */ + WRITE_ONCE(pn->state, vcpu_hashed); lp = pv_hash(lock, pn); + /* - * lp must be set before setting _Q_SLOW_VAL + * We must hash before setting _Q_SLOW_VAL, such that + * when we observe _Q_SLOW_VAL in __pv_queued_spin_unlock() + * we'll be sure to be able to observe our hash entry. * - * [S] lp = lock [RmW] l = l->locked = 0 - * MB MB - * [S] l->locked = _Q_SLOW_VAL [L] lp + * [S] pn->state + * [S] <hash> [Rmw] l->locked == _Q_SLOW_VAL + * MB RMB + * [RmW] l->locked = _Q_SLOW_VAL [L] <unhash> + * [L] pn->state * - * Matches the cmpxchg() in __pv_queued_spin_unlock(). + * Matches the smp_rmb() in __pv_queued_spin_unlock(). */ if (!cmpxchg(&l->locked, _Q_LOCKED_VAL, _Q_SLOW_VAL)) { /* @@ -287,24 +318,34 @@ __visible void __pv_queued_spin_unlock(struct qspinlock *lock) { struct __qspinlock *l = (void *)lock; struct pv_node *node; - u8 lockval = cmpxchg(&l->locked, _Q_LOCKED_VAL, 0); + u8 locked; /* * We must not unlock if SLOW, because in that case we must first * unhash. Otherwise it would be possible to have multiple @lock * entries, which would be BAD. */ - if (likely(lockval == _Q_LOCKED_VAL)) + locked = cmpxchg(&l->locked, _Q_LOCKED_VAL, 0); + if (likely(locked == _Q_LOCKED_VAL)) return; - if (unlikely(lockval != _Q_SLOW_VAL)) { - if (debug_locks_silent) - return; - WARN(1, "pvqspinlock: lock %p has corrupted value 0x%x!\n", lock, atomic_read(&lock->val)); + if (unlikely(locked != _Q_SLOW_VAL)) { + WARN(!debug_locks_silent, + "pvqspinlock: lock 0x%lx has corrupted value 0x%x!\n", + (unsigned long)lock, atomic_read(&lock->val)); return; } /* + * A failed cmpxchg doesn't provide any memory-ordering guarantees, + * so we need a barrier to order the read of the node data in + * pv_unhash *after* we've read the lock being _Q_SLOW_VAL. + * + * Matches the cmpxchg() in pv_wait_head() setting _Q_SLOW_VAL. + */ + smp_rmb(); + + /* * Since the above failed to release, this must be the SLOW path. * Therefore start by looking up the blocked node and unhashing it. */ @@ -319,8 +360,11 @@ __visible void __pv_queued_spin_unlock(struct qspinlock *lock) /* * At this point the memory pointed at by lock can be freed/reused, * however we can still use the pv_node to kick the CPU. + * The other vCPU may not really be halted, but kicking an active + * vCPU is harmless other than the additional latency in completing + * the unlock. */ - if (READ_ONCE(node->state) == vcpu_halted) + if (READ_ONCE(node->state) == vcpu_hashed) pv_kick(node->cpu); } /* diff --git a/kernel/locking/rtmutex-tester.c b/kernel/locking/rtmutex-tester.c deleted file mode 100644 index 1d96dd0d93c1..000000000000 --- a/kernel/locking/rtmutex-tester.c +++ /dev/null @@ -1,420 +0,0 @@ -/* - * RT-Mutex-tester: scriptable tester for rt mutexes - * - * started by Thomas Gleixner: - * - * Copyright (C) 2006, Timesys Corp., Thomas Gleixner <tglx@timesys.com> - * - */ -#include <linux/device.h> -#include <linux/kthread.h> -#include <linux/export.h> -#include <linux/sched.h> -#include <linux/sched/rt.h> -#include <linux/spinlock.h> -#include <linux/timer.h> -#include <linux/freezer.h> -#include <linux/stat.h> - -#include "rtmutex.h" - -#define MAX_RT_TEST_THREADS 8 -#define MAX_RT_TEST_MUTEXES 8 - -static spinlock_t rttest_lock; -static atomic_t rttest_event; - -struct test_thread_data { - int opcode; - int opdata; - int mutexes[MAX_RT_TEST_MUTEXES]; - int event; - struct device dev; -}; - -static struct test_thread_data thread_data[MAX_RT_TEST_THREADS]; -static struct task_struct *threads[MAX_RT_TEST_THREADS]; -static struct rt_mutex mutexes[MAX_RT_TEST_MUTEXES]; - -enum test_opcodes { - RTTEST_NOP = 0, - RTTEST_SCHEDOT, /* 1 Sched other, data = nice */ - RTTEST_SCHEDRT, /* 2 Sched fifo, data = prio */ - RTTEST_LOCK, /* 3 Lock uninterruptible, data = lockindex */ - RTTEST_LOCKNOWAIT, /* 4 Lock uninterruptible no wait in wakeup, data = lockindex */ - RTTEST_LOCKINT, /* 5 Lock interruptible, data = lockindex */ - RTTEST_LOCKINTNOWAIT, /* 6 Lock interruptible no wait in wakeup, data = lockindex */ - RTTEST_LOCKCONT, /* 7 Continue locking after the wakeup delay */ - RTTEST_UNLOCK, /* 8 Unlock, data = lockindex */ - /* 9, 10 - reserved for BKL commemoration */ - RTTEST_SIGNAL = 11, /* 11 Signal other test thread, data = thread id */ - RTTEST_RESETEVENT = 98, /* 98 Reset event counter */ - RTTEST_RESET = 99, /* 99 Reset all pending operations */ -}; - -static int handle_op(struct test_thread_data *td, int lockwakeup) -{ - int i, id, ret = -EINVAL; - - switch(td->opcode) { - - case RTTEST_NOP: - return 0; - - case RTTEST_LOCKCONT: - td->mutexes[td->opdata] = 1; - td->event = atomic_add_return(1, &rttest_event); - return 0; - - case RTTEST_RESET: - for (i = 0; i < MAX_RT_TEST_MUTEXES; i++) { - if (td->mutexes[i] == 4) { - rt_mutex_unlock(&mutexes[i]); - td->mutexes[i] = 0; - } - } - return 0; - - case RTTEST_RESETEVENT: - atomic_set(&rttest_event, 0); - return 0; - - default: - if (lockwakeup) - return ret; - } - - switch(td->opcode) { - - case RTTEST_LOCK: - case RTTEST_LOCKNOWAIT: - id = td->opdata; - if (id < 0 || id >= MAX_RT_TEST_MUTEXES) - return ret; - - td->mutexes[id] = 1; - td->event = atomic_add_return(1, &rttest_event); - rt_mutex_lock(&mutexes[id]); - td->event = atomic_add_return(1, &rttest_event); - td->mutexes[id] = 4; - return 0; - - case RTTEST_LOCKINT: - case RTTEST_LOCKINTNOWAIT: - id = td->opdata; - if (id < 0 || id >= MAX_RT_TEST_MUTEXES) - return ret; - - td->mutexes[id] = 1; - td->event = atomic_add_return(1, &rttest_event); - ret = rt_mutex_lock_interruptible(&mutexes[id], 0); - td->event = atomic_add_return(1, &rttest_event); - td->mutexes[id] = ret ? 0 : 4; - return ret ? -EINTR : 0; - - case RTTEST_UNLOCK: - id = td->opdata; - if (id < 0 || id >= MAX_RT_TEST_MUTEXES || td->mutexes[id] != 4) - return ret; - - td->event = atomic_add_return(1, &rttest_event); - rt_mutex_unlock(&mutexes[id]); - td->event = atomic_add_return(1, &rttest_event); - td->mutexes[id] = 0; - return 0; - - default: - break; - } - return ret; -} - -/* - * Schedule replacement for rtsem_down(). Only called for threads with - * PF_MUTEX_TESTER set. - * - * This allows us to have finegrained control over the event flow. - * - */ -void schedule_rt_mutex_test(struct rt_mutex *mutex) -{ - int tid, op, dat; - struct test_thread_data *td; - - /* We have to lookup the task */ - for (tid = 0; tid < MAX_RT_TEST_THREADS; tid++) { - if (threads[tid] == current) - break; - } - - BUG_ON(tid == MAX_RT_TEST_THREADS); - - td = &thread_data[tid]; - - op = td->opcode; - dat = td->opdata; - - switch (op) { - case RTTEST_LOCK: - case RTTEST_LOCKINT: - case RTTEST_LOCKNOWAIT: - case RTTEST_LOCKINTNOWAIT: - if (mutex != &mutexes[dat]) - break; - - if (td->mutexes[dat] != 1) - break; - - td->mutexes[dat] = 2; - td->event = atomic_add_return(1, &rttest_event); - break; - - default: - break; - } - - schedule(); - - - switch (op) { - case RTTEST_LOCK: - case RTTEST_LOCKINT: - if (mutex != &mutexes[dat]) - return; - - if (td->mutexes[dat] != 2) - return; - - td->mutexes[dat] = 3; - td->event = atomic_add_return(1, &rttest_event); - break; - - case RTTEST_LOCKNOWAIT: - case RTTEST_LOCKINTNOWAIT: - if (mutex != &mutexes[dat]) - return; - - if (td->mutexes[dat] != 2) - return; - - td->mutexes[dat] = 1; - td->event = atomic_add_return(1, &rttest_event); - return; - - default: - return; - } - - td->opcode = 0; - - for (;;) { - set_current_state(TASK_INTERRUPTIBLE); - - if (td->opcode > 0) { - int ret; - - set_current_state(TASK_RUNNING); - ret = handle_op(td, 1); - set_current_state(TASK_INTERRUPTIBLE); - if (td->opcode == RTTEST_LOCKCONT) - break; - td->opcode = ret; - } - - /* Wait for the next command to be executed */ - schedule(); - } - - /* Restore previous command and data */ - td->opcode = op; - td->opdata = dat; -} - -static int test_func(void *data) -{ - struct test_thread_data *td = data; - int ret; - - current->flags |= PF_MUTEX_TESTER; - set_freezable(); - allow_signal(SIGHUP); - - for(;;) { - - set_current_state(TASK_INTERRUPTIBLE); - - if (td->opcode > 0) { - set_current_state(TASK_RUNNING); - ret = handle_op(td, 0); - set_current_state(TASK_INTERRUPTIBLE); - td->opcode = ret; - } - - /* Wait for the next command to be executed */ - schedule(); - try_to_freeze(); - - if (signal_pending(current)) - flush_signals(current); - - if(kthread_should_stop()) - break; - } - return 0; -} - -/** - * sysfs_test_command - interface for test commands - * @dev: thread reference - * @buf: command for actual step - * @count: length of buffer - * - * command syntax: - * - * opcode:data - */ -static ssize_t sysfs_test_command(struct device *dev, struct device_attribute *attr, - const char *buf, size_t count) -{ - struct sched_param schedpar; - struct test_thread_data *td; - char cmdbuf[32]; - int op, dat, tid, ret; - - td = container_of(dev, struct test_thread_data, dev); - tid = td->dev.id; - - /* strings from sysfs write are not 0 terminated! */ - if (count >= sizeof(cmdbuf)) - return -EINVAL; - - /* strip of \n: */ - if (buf[count-1] == '\n') - count--; - if (count < 1) - return -EINVAL; - - memcpy(cmdbuf, buf, count); - cmdbuf[count] = 0; - - if (sscanf(cmdbuf, "%d:%d", &op, &dat) != 2) - return -EINVAL; - - switch (op) { - case RTTEST_SCHEDOT: - schedpar.sched_priority = 0; - ret = sched_setscheduler(threads[tid], SCHED_NORMAL, &schedpar); - if (ret) - return ret; - set_user_nice(current, 0); - break; - - case RTTEST_SCHEDRT: - schedpar.sched_priority = dat; - ret = sched_setscheduler(threads[tid], SCHED_FIFO, &schedpar); - if (ret) - return ret; - break; - - case RTTEST_SIGNAL: - send_sig(SIGHUP, threads[tid], 0); - break; - - default: - if (td->opcode > 0) - return -EBUSY; - td->opdata = dat; - td->opcode = op; - wake_up_process(threads[tid]); - } - - return count; -} - -/** - * sysfs_test_status - sysfs interface for rt tester - * @dev: thread to query - * @buf: char buffer to be filled with thread status info - */ -static ssize_t sysfs_test_status(struct device *dev, struct device_attribute *attr, - char *buf) -{ - struct test_thread_data *td; - struct task_struct *tsk; - char *curr = buf; - int i; - - td = container_of(dev, struct test_thread_data, dev); - tsk = threads[td->dev.id]; - - spin_lock(&rttest_lock); - - curr += sprintf(curr, - "O: %4d, E:%8d, S: 0x%08lx, P: %4d, N: %4d, B: %p, M:", - td->opcode, td->event, tsk->state, - (MAX_RT_PRIO - 1) - tsk->prio, - (MAX_RT_PRIO - 1) - tsk->normal_prio, - tsk->pi_blocked_on); - - for (i = MAX_RT_TEST_MUTEXES - 1; i >=0 ; i--) - curr += sprintf(curr, "%d", td->mutexes[i]); - - spin_unlock(&rttest_lock); - - curr += sprintf(curr, ", T: %p, R: %p\n", tsk, - mutexes[td->dev.id].owner); - - return curr - buf; -} - -static DEVICE_ATTR(status, S_IRUSR, sysfs_test_status, NULL); -static DEVICE_ATTR(command, S_IWUSR, NULL, sysfs_test_command); - -static struct bus_type rttest_subsys = { - .name = "rttest", - .dev_name = "rttest", -}; - -static int init_test_thread(int id) -{ - thread_data[id].dev.bus = &rttest_subsys; - thread_data[id].dev.id = id; - - threads[id] = kthread_run(test_func, &thread_data[id], "rt-test-%d", id); - if (IS_ERR(threads[id])) - return PTR_ERR(threads[id]); - - return device_register(&thread_data[id].dev); -} - -static int init_rttest(void) -{ - int ret, i; - - spin_lock_init(&rttest_lock); - - for (i = 0; i < MAX_RT_TEST_MUTEXES; i++) - rt_mutex_init(&mutexes[i]); - - ret = subsys_system_register(&rttest_subsys, NULL); - if (ret) - return ret; - - for (i = 0; i < MAX_RT_TEST_THREADS; i++) { - ret = init_test_thread(i); - if (ret) - break; - ret = device_create_file(&thread_data[i].dev, &dev_attr_status); - if (ret) - break; - ret = device_create_file(&thread_data[i].dev, &dev_attr_command); - if (ret) - break; - } - - printk("Initializing RT-Tester: %s\n", ret ? "Failed" : "OK" ); - - return ret; -} - -device_initcall(init_rttest); diff --git a/kernel/locking/rtmutex.c b/kernel/locking/rtmutex.c index 5674b073473c..7781d801212f 100644 --- a/kernel/locking/rtmutex.c +++ b/kernel/locking/rtmutex.c @@ -1120,7 +1120,7 @@ __rt_mutex_slowlock(struct rt_mutex *lock, int state, debug_rt_mutex_print_deadlock(waiter); - schedule_rt_mutex(lock); + schedule(); raw_spin_lock(&lock->wait_lock); set_current_state(state); diff --git a/kernel/locking/rtmutex_common.h b/kernel/locking/rtmutex_common.h index 7844f8f0e639..4f5f83c7d2d3 100644 --- a/kernel/locking/rtmutex_common.h +++ b/kernel/locking/rtmutex_common.h @@ -15,28 +15,6 @@ #include <linux/rtmutex.h> /* - * The rtmutex in kernel tester is independent of rtmutex debugging. We - * call schedule_rt_mutex_test() instead of schedule() for the tasks which - * belong to the tester. That way we can delay the wakeup path of those - * threads to provoke lock stealing and testing of complex boosting scenarios. - */ -#ifdef CONFIG_RT_MUTEX_TESTER - -extern void schedule_rt_mutex_test(struct rt_mutex *lock); - -#define schedule_rt_mutex(_lock) \ - do { \ - if (!(current->flags & PF_MUTEX_TESTER)) \ - schedule(); \ - else \ - schedule_rt_mutex_test(_lock); \ - } while (0) - -#else -# define schedule_rt_mutex(_lock) schedule() -#endif - -/* * This is the control structure for tasks blocked on a rt_mutex, * which is allocated on the kernel stack on of the blocked task. * diff --git a/kernel/sched/core.c b/kernel/sched/core.c index d8420c233ff7..3595403921bd 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -164,14 +164,12 @@ struct static_key sched_feat_keys[__SCHED_FEAT_NR] = { static void sched_feat_disable(int i) { - if (static_key_enabled(&sched_feat_keys[i])) - static_key_slow_dec(&sched_feat_keys[i]); + static_key_disable(&sched_feat_keys[i]); } static void sched_feat_enable(int i) { - if (!static_key_enabled(&sched_feat_keys[i])) - static_key_slow_inc(&sched_feat_keys[i]); + static_key_enable(&sched_feat_keys[i]); } #else static void sched_feat_disable(int i) { }; |