search for: arch_spin_unlock

...; +{ > + smp_mb(); > + return atomic_read(&lock->val); > +} > + > +#include <asm-generic/qspinlock.h> > + > +/* we need override it as ppc has io_sync stuff */ > +#undef arch_spin_trylock > +#undef arch_spin_lock > +#undef arch_spin_lock_flags > +#undef arch_spin_unlock > +#define arch_spin_trylock arch_spin_trylock > +#define arch_spin_lock arch_spin_lock > +#define arch_spin_lock_flags arch_spin_lock_flags > +#define arch_spin_unlock arch_spin_unlock > + > +static inline int arch_spin_trylock(arch_spinlock_t *lock) > +{ > + CLEAR_IO_SYNC;...

...; +{ > + smp_mb(); > + return atomic_read(&lock->val); > +} > + > +#include <asm-generic/qspinlock.h> > + > +/* we need override it as ppc has io_sync stuff */ > +#undef arch_spin_trylock > +#undef arch_spin_lock > +#undef arch_spin_lock_flags > +#undef arch_spin_unlock > +#define arch_spin_trylock arch_spin_trylock > +#define arch_spin_lock arch_spin_lock > +#define arch_spin_lock_flags arch_spin_lock_flags > +#define arch_spin_unlock arch_spin_unlock > + > +static inline int arch_spin_trylock(arch_spinlock_t *lock) > +{ > + CLEAR_IO_SYNC;...

...from head into tail which would be wrong). > > Unfortunately xadd could result in head overflow as tail is high. > > The other option was repeated cmpxchg which is bad I believe. > Any suggestions? Stupid question... what if we simply move SLOWPATH from .tail to .head? In this case arch_spin_unlock() could do xadd(tickets.head) and check the result In this case __ticket_check_and_clear_slowpath() really needs to cmpxchg the whole .head_tail. Plus obviously more boring changes. This needs a separate patch even _if_ this can work. BTW. If we move "clear slowpath" into "lock&q...

...from head into tail which would be wrong). > > Unfortunately xadd could result in head overflow as tail is high. > > The other option was repeated cmpxchg which is bad I believe. > Any suggestions? Stupid question... what if we simply move SLOWPATH from .tail to .head? In this case arch_spin_unlock() could do xadd(tickets.head) and check the result In this case __ticket_check_and_clear_slowpath() really needs to cmpxchg the whole .head_tail. Plus obviously more boring changes. This needs a separate patch even _if_ this can work. BTW. If we move "clear slowpath" into "lock&q...

...inline int queued_spin_is_locked(struct qspinlock *lock) +{ + smp_mb(); + return atomic_read(&lock->val); +} + +#include <asm-generic/qspinlock.h> + +/* we need override it as ppc has io_sync stuff */ +#undef arch_spin_trylock +#undef arch_spin_lock +#undef arch_spin_lock_flags +#undef arch_spin_unlock +#define arch_spin_trylock arch_spin_trylock +#define arch_spin_lock arch_spin_lock +#define arch_spin_lock_flags arch_spin_lock_flags +#define arch_spin_unlock arch_spin_unlock + +static inline int arch_spin_trylock(arch_spinlock_t *lock) +{ + CLEAR_IO_SYNC; + return queued_spin_trylock(lock); +}...

On 02/06/2015 09:49 AM, Raghavendra K T wrote: > static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) > { > if (TICKET_SLOWPATH_FLAG && > - static_key_false(&paravirt_ticketlocks_enabled)) { > - arch_spinlock_t prev; > + static_key_false(&paravirt_ticketlocks_enabled)) { > + __ticket_t prev_head; > > - prev = *lock; > +...

...ing xadd in unlock, > we would have to make sure lsb bit is cleared so that we can live with 1 > bit overflow to tail which is unused. now either or both of head,tail > lsb bit may be set after unlock. Sorry, can't understand... could you spell? If TICKET_SLOWPATH_FLAG lives in .head arch_spin_unlock() could simply do head = xadd(&lock->tickets.head, TICKET_LOCK_INC); if (head & TICKET_SLOWPATH_FLAG) __ticket_unlock_kick(head); so it can't overflow to .tail? But probably I missed your concern. And we we do this, probably it makes sense to add something like bool tick...

On 02/06/2015 09:49 AM, Raghavendra K T wrote: > static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) > { > if (TICKET_SLOWPATH_FLAG && > - static_key_false(&paravirt_ticketlocks_enabled)) { > - arch_spinlock_t prev; > + static_key_false(&paravirt_ticketlocks_enabled)) { > + __ticket_t prev_head; > > - prev = *lock; > +...

...ing xadd in unlock, > we would have to make sure lsb bit is cleared so that we can live with 1 > bit overflow to tail which is unused. now either or both of head,tail > lsb bit may be set after unlock. Sorry, can't understand... could you spell? If TICKET_SLOWPATH_FLAG lives in .head arch_spin_unlock() could simply do head = xadd(&lock->tickets.head, TICKET_LOCK_INC); if (head & TICKET_SLOWPATH_FLAG) __ticket_unlock_kick(head); so it can't overflow to .tail? But probably I missed your concern. And we we do this, probably it makes sense to add something like bool tick...

...(flags_dis); + local_irq_restore(flags); + do { + HMT_low(); + if (is_shared_processor()) + splpar_spin_yield(lock); + } while (unlikely(lock->slock != 0)); + HMT_medium(); + local_irq_restore(flags_dis); + } +} +#define arch_spin_lock_flags arch_spin_lock_flags + +static inline void arch_spin_unlock(arch_spinlock_t *lock) +{ + __asm__ __volatile__("# arch_spin_unlock\n\t" + PPC_RELEASE_BARRIER: : :"memory"); + lock->slock = 0; +} + +/* + * Read-write spinlocks, allowing multiple readers + * but only one writer. + * + * NOTE! it is quite common to have readers in inter...

...new.tickets.tail || - cmpxchg(&lock->head_tail, old.head_tail, - new.head_tail) != old.head_tail) { - /* - * Lock still has someone queued for it, so wake up an - * appropriate waiter. - */ - __ticket_unlock_kick(lock, old.tickets.head); - } -} - static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) { if (TICKET_SLOWPATH_FLAG && - static_key_false(&paravirt_ticketlocks_enabled)) { - arch_spinlock_t prev; + static_key_false(&paravirt_ticketlocks_enabled)) { + __ticket_t prev_head; - prev = *lock; + prev_head = lock->tickets.head; add_sm...

...new.tickets.tail || - cmpxchg(&lock->head_tail, old.head_tail, - new.head_tail) != old.head_tail) { - /* - * Lock still has someone queued for it, so wake up an - * appropriate waiter. - */ - __ticket_unlock_kick(lock, old.tickets.head); - } -} - static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) { if (TICKET_SLOWPATH_FLAG && - static_key_false(&paravirt_ticketlocks_enabled)) { - arch_spinlock_t prev; + static_key_false(&paravirt_ticketlocks_enabled)) { + __ticket_t prev_head; - prev = *lock; + prev_head = lock->tickets.head; add_sm...

Hi All, this is the fairlock patchset. You can apply them and build successfully. patches are based on linux-next qspinlock can avoid waiter starved issue. It has about the same speed in single-thread and it can be much faster in high contention situations especially when the spinlock is embedded within the data structure to be protected. v7 -> v8: add one patch to drop a function call

Hi All, this is the fairlock patchset. You can apply them and build successfully. patches are based on linux-next qspinlock can avoid waiter starved issue. It has about the same speed in single-thread and it can be much faster in high contention situations especially when the spinlock is embedded within the data structure to be protected. v7 -> v8: add one patch to drop a function call

Hi, Here are some numbers for my version -- also attached is the test code. I found that booting big machines is tediously slow so I lifted the whole lot to userspace. I measure the cycles spend in arch_spin_lock() + arch_spin_unlock(). The machines used are a 4 node (2 socket) AMD Interlagos, and a 2 node (2 socket) Intel Westmere-EP. AMD (ticket) AMD (qspinlock + pending + opt) Local: Local: 1: 324.425530 1: 324.102142 2: 17141.324050 2: 620.185930 3: 52212.232343 3: 2524...

Hi, Here are some numbers for my version -- also attached is the test code. I found that booting big machines is tediously slow so I lifted the whole lot to userspace. I measure the cycles spend in arch_spin_lock() + arch_spin_unlock(). The machines used are a 4 node (2 socket) AMD Interlagos, and a 2 node (2 socket) Intel Westmere-EP. AMD (ticket) AMD (qspinlock + pending + opt) Local: Local: 1: 324.425530 1: 324.102142 2: 17141.324050 2: 620.185930 3: 52212.232343 3: 2524...

...new.tickets.tail || - cmpxchg(&lock->head_tail, old.head_tail, - new.head_tail) != old.head_tail) { - /* - * Lock still has someone queued for it, so wake up an - * appropriate waiter. - */ - __ticket_unlock_kick(lock, old.tickets.head); - } -} - static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) { if (TICKET_SLOWPATH_FLAG && - static_key_false(&paravirt_ticketlocks_enabled)) { - arch_spinlock_t prev; + static_key_false(&paravirt_ticketlocks_enabled)) { + __ticket_t next_head; - prev = *lock; + next_head = lock->tickets.head + TICKET_L...

...new.tickets.tail || - cmpxchg(&lock->head_tail, old.head_tail, - new.head_tail) != old.head_tail) { - /* - * Lock still has someone queued for it, so wake up an - * appropriate waiter. - */ - __ticket_unlock_kick(lock, old.tickets.head); - } -} - static __always_inline void arch_spin_unlock(arch_spinlock_t *lock) { if (TICKET_SLOWPATH_FLAG && - static_key_false(&paravirt_ticketlocks_enabled)) { - arch_spinlock_t prev; + static_key_false(&paravirt_ticketlocks_enabled)) { + __ticket_t next_head; - prev = *lock; + next_head = lock->tickets.head + TICKET_L...

Hi All, this is the fairlock patchset. You can apply them and build successfully. patches are based on linux-next qspinlock can avoid waiter starved issue. It has about the same speed in single-thread and it can be much faster in high contention situations especially when the spinlock is embedded within the data structure to be protected. v8 -> v9: mv qspinlocm config entry to

Hi All, this is the fairlock patchset. You can apply them and build successfully. patches are based on linux-next qspinlock can avoid waiter starved issue. It has about the same speed in single-thread and it can be much faster in high contention situations especially when the spinlock is embedded within the data structure to be protected. v8 -> v9: mv qspinlocm config entry to