From 27df89689e257cccb604fdf56c91a75a25aa554a Mon Sep 17 00:00:00 2001 From: Matthew Wilcox Date: Mon, 20 Aug 2018 10:19:14 -0400 Subject: [PATCH] locking/spinlocks: Remove an instruction from spin and write locks Both spin locks and write locks currently do: f0 0f b1 17 lock cmpxchg %edx,(%rdi) 85 c0 test %eax,%eax 75 05 jne [slowpath] This 'test' insn is superfluous; the cmpxchg insn sets the Z flag appropriately. Peter pointed out that using atomic_try_cmpxchg_acquire() will let the compiler know this is true. Comparing before/after disassemblies show the only effect is to remove this insn. Take this opportunity to make the spin & write lock code resemble each other more closely and have similar likely() hints. Suggested-by: Peter Zijlstra Signed-off-by: Matthew Wilcox Signed-off-by: Peter Zijlstra (Intel) Acked-by: Will Deacon Cc: Arnd Bergmann Cc: Linus Torvalds Cc: Thomas Gleixner Cc: Waiman Long Link: http://lkml.kernel.org/r/20180820162639.GC25153@bombadil.infradead.org Signed-off-by: Ingo Molnar --- include/asm-generic/qrwlock.h | 7 ++++--- include/asm-generic/qspinlock.h | 16 +++++++++------- 2 files changed, 13 insertions(+), 10 deletions(-) diff --git a/include/asm-generic/qrwlock.h b/include/asm-generic/qrwlock.h index 0f7062bd55e5..36254d2da8e0 100644 --- a/include/asm-generic/qrwlock.h +++ b/include/asm-generic/qrwlock.h @@ -71,8 +71,8 @@ static inline int queued_write_trylock(struct qrwlock *lock) if (unlikely(cnts)) return 0; - return likely(atomic_cmpxchg_acquire(&lock->cnts, - cnts, cnts | _QW_LOCKED) == cnts); + return likely(atomic_try_cmpxchg_acquire(&lock->cnts, &cnts, + _QW_LOCKED)); } /** * queued_read_lock - acquire read lock of a queue rwlock @@ -96,8 +96,9 @@ static inline void queued_read_lock(struct qrwlock *lock) */ static inline void queued_write_lock(struct qrwlock *lock) { + u32 cnts = 0; /* Optimize for the unfair lock case where the fair flag is 0. */ - if (atomic_cmpxchg_acquire(&lock->cnts, 0, _QW_LOCKED) == 0) + if (likely(atomic_try_cmpxchg_acquire(&lock->cnts, &cnts, _QW_LOCKED))) return; queued_write_lock_slowpath(lock); diff --git a/include/asm-generic/qspinlock.h b/include/asm-generic/qspinlock.h index 9cc457597ddf..7541fa707f5b 100644 --- a/include/asm-generic/qspinlock.h +++ b/include/asm-generic/qspinlock.h @@ -66,10 +66,12 @@ static __always_inline int queued_spin_is_contended(struct qspinlock *lock) */ static __always_inline int queued_spin_trylock(struct qspinlock *lock) { - if (!atomic_read(&lock->val) && - (atomic_cmpxchg_acquire(&lock->val, 0, _Q_LOCKED_VAL) == 0)) - return 1; - return 0; + u32 val = atomic_read(&lock->val); + + if (unlikely(val)) + return 0; + + return likely(atomic_try_cmpxchg_acquire(&lock->val, &val, _Q_LOCKED_VAL)); } extern void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val); @@ -80,11 +82,11 @@ extern void queued_spin_lock_slowpath(struct qspinlock *lock, u32 val); */ static __always_inline void queued_spin_lock(struct qspinlock *lock) { - u32 val; + u32 val = 0; - val = atomic_cmpxchg_acquire(&lock->val, 0, _Q_LOCKED_VAL); - if (likely(val == 0)) + if (likely(atomic_try_cmpxchg_acquire(&lock->val, &val, _Q_LOCKED_VAL))) return; + queued_spin_lock_slowpath(lock, val); }