From 44472d1b83127e579c798ff92a07ae86d98b61b9 Mon Sep 17 00:00:00 2001 From: Arnd Bergmann Date: Mon, 6 Oct 2025 13:07:32 +0200 Subject: atomic: Skip alignment check for try_cmpxchg() old arg The 'old' argument in atomic_try_cmpxchg() and related functions is a pointer to a normal non-atomic integer number, which does not require to be naturally aligned, unlike the atomic_t/atomic64_t types themselves. In order to add an alignment check with CONFIG_DEBUG_ATOMIC into the normal instrument_atomic_read_write() helper, change this check to use the non-atomic instrument_read_write(), the same way that was done earlier for try_cmpxchg() in commit ec570320b09f ("locking/atomic: Correct (cmp)xchg() instrumentation"). This prevents warnings on m68k calling the 32-bit atomic_try_cmpxchg() with 16-bit aligned arguments as well as several more architectures including x86-32 when calling atomic64_try_cmpxchg() with 32-bit aligned u64 arguments. Reported-by: Finn Thain Signed-off-by: Arnd Bergmann Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/all/cover.1757810729.git.fthain@linux-m68k.org/ --- include/linux/atomic/atomic-instrumented.h | 26 +++++++++++++------------- 1 file changed, 13 insertions(+), 13 deletions(-) (limited to 'include/linux') diff --git a/include/linux/atomic/atomic-instrumented.h b/include/linux/atomic/atomic-instrumented.h index 9409a6ddf3e0..37ab6314a9f7 100644 --- a/include/linux/atomic/atomic-instrumented.h +++ b/include/linux/atomic/atomic-instrumented.h @@ -1276,7 +1276,7 @@ atomic_try_cmpxchg(atomic_t *v, int *old, int new) { kcsan_mb(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_try_cmpxchg(v, old, new); } @@ -1298,7 +1298,7 @@ static __always_inline bool atomic_try_cmpxchg_acquire(atomic_t *v, int *old, int new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_try_cmpxchg_acquire(v, old, new); } @@ -1321,7 +1321,7 @@ atomic_try_cmpxchg_release(atomic_t *v, int *old, int new) { kcsan_release(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_try_cmpxchg_release(v, old, new); } @@ -1343,7 +1343,7 @@ static __always_inline bool atomic_try_cmpxchg_relaxed(atomic_t *v, int *old, int new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_try_cmpxchg_relaxed(v, old, new); } @@ -2854,7 +2854,7 @@ atomic64_try_cmpxchg(atomic64_t *v, s64 *old, s64 new) { kcsan_mb(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic64_try_cmpxchg(v, old, new); } @@ -2876,7 +2876,7 @@ static __always_inline bool atomic64_try_cmpxchg_acquire(atomic64_t *v, s64 *old, s64 new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic64_try_cmpxchg_acquire(v, old, new); } @@ -2899,7 +2899,7 @@ atomic64_try_cmpxchg_release(atomic64_t *v, s64 *old, s64 new) { kcsan_release(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic64_try_cmpxchg_release(v, old, new); } @@ -2921,7 +2921,7 @@ static __always_inline bool atomic64_try_cmpxchg_relaxed(atomic64_t *v, s64 *old, s64 new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic64_try_cmpxchg_relaxed(v, old, new); } @@ -4432,7 +4432,7 @@ atomic_long_try_cmpxchg(atomic_long_t *v, long *old, long new) { kcsan_mb(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_long_try_cmpxchg(v, old, new); } @@ -4454,7 +4454,7 @@ static __always_inline bool atomic_long_try_cmpxchg_acquire(atomic_long_t *v, long *old, long new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_long_try_cmpxchg_acquire(v, old, new); } @@ -4477,7 +4477,7 @@ atomic_long_try_cmpxchg_release(atomic_long_t *v, long *old, long new) { kcsan_release(); instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_long_try_cmpxchg_release(v, old, new); } @@ -4499,7 +4499,7 @@ static __always_inline bool atomic_long_try_cmpxchg_relaxed(atomic_long_t *v, long *old, long new) { instrument_atomic_read_write(v, sizeof(*v)); - instrument_atomic_read_write(old, sizeof(*old)); + instrument_read_write(old, sizeof(*old)); return raw_atomic_long_try_cmpxchg_relaxed(v, old, new); } @@ -5050,4 +5050,4 @@ atomic_long_dec_if_positive(atomic_long_t *v) #endif /* _LINUX_ATOMIC_INSTRUMENTED_H */ -// 8829b337928e9508259079d32581775ececd415b +// f618ac667f868941a84ce0ab2242f1786e049ed4 -- cgit v1.2.3 From cc39f3872c0865bef992b713338df369554fa9e0 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 9 Oct 2025 22:11:54 +0200 Subject: seqlock: Introduce scoped_seqlock_read() The read_seqbegin/need_seqretry/done_seqretry API is cumbersome and error prone. With the new helper the "typical" code like int seq, nextseq; unsigned long flags; nextseq = 0; do { seq = nextseq; flags = read_seqbegin_or_lock_irqsave(&seqlock, &seq); // read-side critical section nextseq = 1; } while (need_seqretry(&seqlock, seq)); done_seqretry_irqrestore(&seqlock, seq, flags); can be rewritten as scoped_seqlock_read (&seqlock, ss_lock_irqsave) { // read-side critical section } Original idea by Oleg Nesterov; with contributions from Linus. Originally-by: Oleg Nesterov Signed-off-by: Peter Zijlstra (Intel) --- include/linux/seqlock.h | 111 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 111 insertions(+) (limited to 'include/linux') diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h index 5ce48eab7a2a..b7bcc4111e90 100644 --- a/include/linux/seqlock.h +++ b/include/linux/seqlock.h @@ -1209,4 +1209,115 @@ done_seqretry_irqrestore(seqlock_t *lock, int seq, unsigned long flags) if (seq & 1) read_sequnlock_excl_irqrestore(lock, flags); } + +enum ss_state { + ss_done = 0, + ss_lock, + ss_lock_irqsave, + ss_lockless, +}; + +struct ss_tmp { + enum ss_state state; + unsigned long data; + spinlock_t *lock; + spinlock_t *lock_irqsave; +}; + +static inline void __scoped_seqlock_cleanup(struct ss_tmp *sst) +{ + if (sst->lock) + spin_unlock(sst->lock); + if (sst->lock_irqsave) + spin_unlock_irqrestore(sst->lock_irqsave, sst->data); +} + +extern void __scoped_seqlock_invalid_target(void); + +#if defined(CONFIG_CC_IS_GCC) && CONFIG_GCC_VERSION < 90000 +/* + * For some reason some GCC-8 architectures (nios2, alpha) have trouble + * determining that the ss_done state is impossible in __scoped_seqlock_next() + * below. + */ +static inline void __scoped_seqlock_bug(void) { } +#else +/* + * Canary for compiler optimization -- if the compiler doesn't realize this is + * an impossible state, it very likely generates sub-optimal code here. + */ +extern void __scoped_seqlock_bug(void); +#endif + +static inline void +__scoped_seqlock_next(struct ss_tmp *sst, seqlock_t *lock, enum ss_state target) +{ + switch (sst->state) { + case ss_done: + __scoped_seqlock_bug(); + return; + + case ss_lock: + case ss_lock_irqsave: + sst->state = ss_done; + return; + + case ss_lockless: + if (!read_seqretry(lock, sst->data)) { + sst->state = ss_done; + return; + } + break; + } + + switch (target) { + case ss_done: + __scoped_seqlock_invalid_target(); + return; + + case ss_lock: + sst->lock = &lock->lock; + spin_lock(sst->lock); + sst->state = ss_lock; + return; + + case ss_lock_irqsave: + sst->lock_irqsave = &lock->lock; + spin_lock_irqsave(sst->lock_irqsave, sst->data); + sst->state = ss_lock_irqsave; + return; + + case ss_lockless: + sst->data = read_seqbegin(lock); + return; + } +} + +#define __scoped_seqlock_read(_seqlock, _target, _s) \ + for (struct ss_tmp _s __cleanup(__scoped_seqlock_cleanup) = \ + { .state = ss_lockless, .data = read_seqbegin(_seqlock) }; \ + _s.state != ss_done; \ + __scoped_seqlock_next(&_s, _seqlock, _target)) + +/** + * scoped_seqlock_read (lock, ss_state) - execute the read side critical + * section without manual sequence + * counter handling or calls to other + * helpers + * @lock: pointer to seqlock_t protecting the data + * @ss_state: one of {ss_lock, ss_lock_irqsave, ss_lockless} indicating + * the type of critical read section + * + * Example: + * + * scoped_seqlock_read (&lock, ss_lock) { + * // read-side critical section + * } + * + * Starts with a lockess pass first. If it fails, restarts the critical + * section with the lock held. + */ +#define scoped_seqlock_read(_seqlock, _target) \ + __scoped_seqlock_read(_seqlock, _target, __UNIQUE_ID(seqlock)) + #endif /* __LINUX_SEQLOCK_H */ -- cgit v1.2.3 From b94d45b6bbb42571ec225d3be0e7457c8765a5b4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 28 Oct 2025 09:56:38 +0100 Subject: seqlock: Allow KASAN to fail optimizing Some KASAN builds are failing to properly optimize this code -- luckily we don't care about core quality for KASAN builds, so just exclude it. Reported-by: kernel test robot Signed-off-by: Peter Zijlstra (Intel) Closes: https://lore.kernel.org/oe-kbuild-all/202510251641.idrNXhv5-lkp@intel.com/ --- include/linux/seqlock.h | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h index b7bcc4111e90..a8a8661839b6 100644 --- a/include/linux/seqlock.h +++ b/include/linux/seqlock.h @@ -1234,11 +1234,14 @@ static inline void __scoped_seqlock_cleanup(struct ss_tmp *sst) extern void __scoped_seqlock_invalid_target(void); -#if defined(CONFIG_CC_IS_GCC) && CONFIG_GCC_VERSION < 90000 +#if (defined(CONFIG_CC_IS_GCC) && CONFIG_GCC_VERSION < 90000) || defined(CONFIG_KASAN) /* * For some reason some GCC-8 architectures (nios2, alpha) have trouble * determining that the ss_done state is impossible in __scoped_seqlock_next() * below. + * + * Similarly KASAN is known to confuse compilers enough to break this. But we + * don't care about code quality for KASAN builds anyway. */ static inline void __scoped_seqlock_bug(void) { } #else -- cgit v1.2.3 From 51d7a054521de7085783a9a1ba15c3530863409a Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Wed, 5 Nov 2025 15:23:50 +0100 Subject: locking/mutex: Redo __mutex_init() to reduce generated code size mutex_init() invokes __mutex_init() providing the name of the lock and a pointer to a the lock class. With LOCKDEP enabled this information is useful but without LOCKDEP it not used at all. Passing the pointer information of the lock class might be considered negligible but the name of the lock is passed as well and the string is stored. This information is wasting storage. Split __mutex_init() into a _genereic() variant doing the initialisation of the lock and a _lockdep() version which does _genereic() plus the lockdep bits. Restrict the lockdep version to lockdep enabled builds allowing the compiler to remove the unused parameter. This results in the following size reduction: text data bss dec filename | 30237599 8161430 1176624 39575653 vmlinux.defconfig | 30233269 8149142 1176560 39558971 vmlinux.defconfig.patched -4.2KiB -12KiB | 32455099 8471098 12934684 53860881 vmlinux.defconfig.lockdep | 32455100 8471098 12934684 53860882 vmlinux.defconfig.patched.lockdep | 27152407 7191822 2068040 36412269 vmlinux.defconfig.preempt_rt | 27145937 7183630 2067976 36397543 vmlinux.defconfig.patched.preempt_rt -6.3KiB -8KiB | 29382020 7505742 13784608 50672370 vmlinux.defconfig.preempt_rt.lockdep | 29376229 7505742 13784544 50666515 vmlinux.defconfig.patched.preempt_rt.lockdep -5.6KiB [peterz: folded fix from boqun] Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Boqun Feng Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Reviewed-by: Waiman Long Link: https://lkml.kernel.org/r/20251125145425.68319-1-boqun.feng@gmail.com Link: https://patch.msgid.link/20251105142350.Tfeevs2N@linutronix.de --- include/linux/mutex.h | 45 +++++++++++++++++++++++++++++++++++---------- 1 file changed, 35 insertions(+), 10 deletions(-) (limited to 'include/linux') diff --git a/include/linux/mutex.h b/include/linux/mutex.h index 847b81ca6436..bf535f0118bb 100644 --- a/include/linux/mutex.h +++ b/include/linux/mutex.h @@ -86,8 +86,23 @@ do { \ #define DEFINE_MUTEX(mutexname) \ struct mutex mutexname = __MUTEX_INITIALIZER(mutexname) -extern void __mutex_init(struct mutex *lock, const char *name, - struct lock_class_key *key); +#ifdef CONFIG_DEBUG_LOCK_ALLOC +void mutex_init_lockep(struct mutex *lock, const char *name, struct lock_class_key *key); + +static inline void __mutex_init(struct mutex *lock, const char *name, + struct lock_class_key *key) +{ + mutex_init_lockep(lock, name, key); +} +#else +extern void mutex_init_generic(struct mutex *lock); + +static inline void __mutex_init(struct mutex *lock, const char *name, + struct lock_class_key *key) +{ + mutex_init_generic(lock); +} +#endif /* !CONFIG_DEBUG_LOCK_ALLOC */ /** * mutex_is_locked - is the mutex locked @@ -111,17 +126,27 @@ extern bool mutex_is_locked(struct mutex *lock); #define DEFINE_MUTEX(mutexname) \ struct mutex mutexname = __MUTEX_INITIALIZER(mutexname) -extern void __mutex_rt_init(struct mutex *lock, const char *name, - struct lock_class_key *key); - #define mutex_is_locked(l) rt_mutex_base_is_locked(&(l)->rtmutex) -#define __mutex_init(mutex, name, key) \ -do { \ - rt_mutex_base_init(&(mutex)->rtmutex); \ - __mutex_rt_init((mutex), name, key); \ -} while (0) +#ifdef CONFIG_DEBUG_LOCK_ALLOC +extern void mutex_rt_init_lockdep(struct mutex *mutex, const char *name, + struct lock_class_key *key); + +static inline void __mutex_init(struct mutex *lock, const char *name, + struct lock_class_key *key) +{ + mutex_rt_init_lockdep(lock, name, key); +} +#else +extern void mutex_rt_init_generic(struct mutex *mutex); + +static inline void __mutex_init(struct mutex *lock, const char *name, + struct lock_class_key *key) +{ + mutex_rt_init_generic(lock); +} +#endif /* !CONFIG_LOCKDEP */ #endif /* CONFIG_PREEMPT_RT */ #ifdef CONFIG_DEBUG_MUTEXES -- cgit v1.2.3 From 719e357fc09c63238956eb7cd546627f9e050640 Mon Sep 17 00:00:00 2001 From: Vincent Mailhol Date: Thu, 27 Nov 2025 15:41:40 +0100 Subject: locking/local_lock: s/l/__l/ and s/tl/__tl/ to reduce the risk of shadowing The Linux kernel coding style advises to avoid common variable names in function-like macros to reduce the risk of namespace collisions. Throughout local_lock_internal.h, several macros use the rather common variable names 'l' and 'tl'. This already resulted in an actual collision: the __local_lock_acquire() function like macro is currently shadowing the parameter 'l' of the: class_##_name##_t class_##_name##_constructor(_type *l) function factory from . Rename the variable 'l' to '__l' and the variable 'tl' to '__tl' throughout the file to fix the current namespace collision and to prevent future ones. [ bigeasy: Rebase, update all l and tl instances in macros ] Signed-off-by: Vincent Mailhol Signed-off-by: Sebastian Andrzej Siewior Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Acked-by: Waiman Long Link: https://patch.msgid.link/20251127144140.215722-3-bigeasy@linutronix.de --- include/linux/local_lock_internal.h | 62 ++++++++++++++++++------------------- 1 file changed, 31 insertions(+), 31 deletions(-) (limited to 'include/linux') diff --git a/include/linux/local_lock_internal.h b/include/linux/local_lock_internal.h index a4dc479157b5..8f82b4eb542f 100644 --- a/include/linux/local_lock_internal.h +++ b/include/linux/local_lock_internal.h @@ -99,18 +99,18 @@ do { \ #define __local_lock_acquire(lock) \ do { \ - local_trylock_t *tl; \ - local_lock_t *l; \ + local_trylock_t *__tl; \ + local_lock_t *__l; \ \ - l = (local_lock_t *)(lock); \ - tl = (local_trylock_t *)l; \ + __l = (local_lock_t *)(lock); \ + __tl = (local_trylock_t *)__l; \ _Generic((lock), \ local_trylock_t *: ({ \ - lockdep_assert(tl->acquired == 0); \ - WRITE_ONCE(tl->acquired, 1); \ + lockdep_assert(__tl->acquired == 0); \ + WRITE_ONCE(__tl->acquired, 1); \ }), \ local_lock_t *: (void)0); \ - local_lock_acquire(l); \ + local_lock_acquire(__l); \ } while (0) #define __local_lock(lock) \ @@ -133,36 +133,36 @@ do { \ #define __local_trylock(lock) \ ({ \ - local_trylock_t *tl; \ + local_trylock_t *__tl; \ \ preempt_disable(); \ - tl = (lock); \ - if (READ_ONCE(tl->acquired)) { \ + __tl = (lock); \ + if (READ_ONCE(__tl->acquired)) { \ preempt_enable(); \ - tl = NULL; \ + __tl = NULL; \ } else { \ - WRITE_ONCE(tl->acquired, 1); \ + WRITE_ONCE(__tl->acquired, 1); \ local_trylock_acquire( \ - (local_lock_t *)tl); \ + (local_lock_t *)__tl); \ } \ - !!tl; \ + !!__tl; \ }) #define __local_trylock_irqsave(lock, flags) \ ({ \ - local_trylock_t *tl; \ + local_trylock_t *__tl; \ \ local_irq_save(flags); \ - tl = (lock); \ - if (READ_ONCE(tl->acquired)) { \ + __tl = (lock); \ + if (READ_ONCE(__tl->acquired)) { \ local_irq_restore(flags); \ - tl = NULL; \ + __tl = NULL; \ } else { \ - WRITE_ONCE(tl->acquired, 1); \ + WRITE_ONCE(__tl->acquired, 1); \ local_trylock_acquire( \ - (local_lock_t *)tl); \ + (local_lock_t *)__tl); \ } \ - !!tl; \ + !!__tl; \ }) /* preemption or migration must be disabled before calling __local_lock_is_locked */ @@ -170,16 +170,16 @@ do { \ #define __local_lock_release(lock) \ do { \ - local_trylock_t *tl; \ - local_lock_t *l; \ + local_trylock_t *__tl; \ + local_lock_t *__l; \ \ - l = (local_lock_t *)(lock); \ - tl = (local_trylock_t *)l; \ - local_lock_release(l); \ + __l = (local_lock_t *)(lock); \ + __tl = (local_trylock_t *)__l; \ + local_lock_release(__l); \ _Generic((lock), \ local_trylock_t *: ({ \ - lockdep_assert(tl->acquired == 1); \ - WRITE_ONCE(tl->acquired, 0); \ + lockdep_assert(__tl->acquired == 1); \ + WRITE_ONCE(__tl->acquired, 0); \ }), \ local_lock_t *: (void)0); \ } while (0) @@ -223,12 +223,12 @@ typedef spinlock_t local_trylock_t; #define INIT_LOCAL_LOCK(lockname) __LOCAL_SPIN_LOCK_UNLOCKED((lockname)) #define INIT_LOCAL_TRYLOCK(lockname) __LOCAL_SPIN_LOCK_UNLOCKED((lockname)) -#define __local_lock_init(l) \ +#define __local_lock_init(__l) \ do { \ - local_spin_lock_init((l)); \ + local_spin_lock_init((__l)); \ } while (0) -#define __local_trylock_init(l) __local_lock_init(l) +#define __local_trylock_init(__l) __local_lock_init(__l) #define __local_lock(__lock) \ do { \ -- cgit v1.2.3 From 43decb6b628eb033a1b6188e5018773c0d38be1d Mon Sep 17 00:00:00 2001 From: Randy Dunlap Date: Thu, 27 Nov 2025 22:59:25 -0800 Subject: locking/local_lock: Fix all kernel-doc warnings Modify kernel-doc comments in local_lock.h to prevent warnings: Warning: include/linux/local_lock.h:9 function parameter 'lock' not described in 'local_lock_init' Warning: include/linux/local_lock.h:56 function parameter 'lock' not described in 'local_trylock_init' Warning: include/linux/local_lock.h:56 expecting prototype for local_lock_init(). Prototype was for local_trylock_init() instead Signed-off-by: Randy Dunlap Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://patch.msgid.link/20251128065925.917917-1-rdunlap@infradead.org --- include/linux/local_lock.h | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/local_lock.h b/include/linux/local_lock.h index 0d91d060e3e9..b0e6ab329b00 100644 --- a/include/linux/local_lock.h +++ b/include/linux/local_lock.h @@ -6,6 +6,7 @@ /** * local_lock_init - Runtime initialize a lock instance + * @lock: The lock variable */ #define local_lock_init(lock) __local_lock_init(lock) @@ -52,7 +53,8 @@ __local_unlock_irqrestore(this_cpu_ptr(lock), flags) /** - * local_lock_init - Runtime initialize a lock instance + * local_trylock_init - Runtime initialize a lock instance + * @lock: The lock variable */ #define local_trylock_init(lock) __local_trylock_init(lock) -- cgit v1.2.3