search for: __add

On 02/10, Jeremy Fitzhardinge wrote: > > On 02/10/2015 05:26 AM, Oleg Nesterov wrote: > > On 02/10, Raghavendra K T wrote: > >> 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

On 02/10, Jeremy Fitzhardinge wrote: > > On 02/10/2015 05:26 AM, Oleg Nesterov wrote: > > On 02/10, Raghavendra K T wrote: > >> 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

...vent read-reordering. x86 memory ordering rules state that all writes are seen in a globally consistent order, and are globally ordered wrt reads *on the same addresses*, but reads to different addresses can be reordered wrt to writes. So, if the unlocking add were not a locked operation: __add(&lock->tickets.head, TICKET_LOCK_INC); /* not locked */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); Then the read of lock->tickets.tail can be reordered before the unlock, which introduces a race: /* read r...

...vent read-reordering. x86 memory ordering rules state that all writes are seen in a globally consistent order, and are globally ordered wrt reads *on the same addresses*, but reads to different addresses can be reordered wrt to writes. So, if the unlocking add were not a locked operation: __add(&lock->tickets.head, TICKET_LOCK_INC); /* not locked */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); Then the read of lock->tickets.tail can be reordered before the unlock, which introduces a race: /* read r...

Hi, I''m trying to stub File.open and whenever I do, rspec blows up with the following error message (undefined method `add'' for nil:NilClass), which seems to happen because method __add in class Proxy is calling #add on global $rspec_mocks, which in turn is nil. Can someone explain what I''m doing wrong, as I can''t seem to stub anything out! Here''s my code: class Foo def Foo.open( name, mode ) return name end end describe Something...

..., line 1066, in batchRun batch.run() File "up2dateBatch.py", line 74, in run File "up2dateBatch.py", line 159, in __dryRun File "up2date.py", line 395, in dryRun File "depSolver.py", line 357, in setup File "depSolver.py", line 390, in __add File "headers.py", line 37, in __getitem__ File "headers.py", line 42, in __retrievePackage File "rpcServer.py", line 304, in doCall File "repoDirector.py", line 31, in getHeader File "rpmSource.py", line 210, in getHeader File "/us...

...091 100644 --- a/arch/x86/include/asm/spinlock.h +++ b/arch/x86/include/asm/spinlock.h @@ -42,6 +42,10 @@ extern struct static_key paravirt_ticketlocks_enabled; static __always_inline bool static_key_false(struct static_key *key); +static inline void ___ticket_unlock(arch_spinlock_t *lock) +{ + __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); +} #ifdef CONFIG_PARAVIRT_SPINLOCKS static inline void __ticket_enter_slowpath(arch_spinlock_t *lock) @@ -64,6 +68,11 @@ static inline void __ticket_clear_slowpath(arch_spinlock_t *lock, __ticket_t head) { } + +static...

...prev = *lock; - add_smp(&lock->tickets.head, TICKET_LOCK_INC); + add_smp(&lock->tickets.head, TICKET_LOCK_UNLOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else - __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); + __add(&lock->tickets.head, TICKET_LOCK_UNLOCK_INC, + UNLOCK_LOCK_PREFIX); } static inline int arch_spin_is_locked(arch_spinlock_t *lock) @@ -171,7 +218,7 @@ static inline int arch_spin_is_contended(arch_spinlock_t...

...prev = *lock; - add_smp(&lock->tickets.head, TICKET_LOCK_INC); + add_smp(&lock->tickets.head, TICKET_LOCK_UNLOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else - __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); + __add(&lock->tickets.head, TICKET_LOCK_UNLOCK_INC, + UNLOCK_LOCK_PREFIX); } static inline int arch_spin_is_locked(arch_spinlock_t *lock) @@ -171,7 +218,7 @@ static inline int arch_spin_is_contended(arch_spinlock_t...

...tic_key_false(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...tic_key_false(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...tic_key_false(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...tic_key_false(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...tic_key_false(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...static_branch(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...static_branch(&paravirt_ticketlocks_enabled))) { arch_spinlock_t prev; prev = *lock; add_smp(&lock->tickets.head, TICKET_LOCK_INC); /* add_smp() is a full mb() */ if (unlikely(lock->tickets.tail & TICKET_SLOWPATH_FLAG)) __ticket_unlock_slowpath(lock, prev); } else __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); which generates: push %rbp mov %rsp,%rbp nop5 # replaced by 5-byte jmp 2f when PV enabled # non-PV unlock addb $0x2,(%rdi) 1: pop %rbp retq ### PV unlock ### 2: movzwl (%rdi),%esi # Fetch prev lock addb $0x2,...

...in practice. > + BUILD_BUG_ON(((__ticket_t)NR_CPUS) != NR_CPUS); > + __ticket_unlock_kick(lock, prev_head); Should be "prev_head + TICKET_LOCK_INC" to match the previous code, otherwise it won't find the CPU waiting for the new head. J > + } > } else > __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); > } > @@ -164,7 +162,7 @@ static inline int arch_spin_is_locked(arch_spinlock_t *lock) > { > struct __raw_tickets tmp = READ_ONCE(lock->tickets); > > - return tmp.tail != tmp.head; > + return (tmp.tail &...

...in practice. > + BUILD_BUG_ON(((__ticket_t)NR_CPUS) != NR_CPUS); > + __ticket_unlock_kick(lock, prev_head); Should be "prev_head + TICKET_LOCK_INC" to match the previous code, otherwise it won't find the CPU waiting for the new head. J > + } > } else > __add(&lock->tickets.head, TICKET_LOCK_INC, UNLOCK_LOCK_PREFIX); > } > @@ -164,7 +162,7 @@ static inline int arch_spin_is_locked(arch_spinlock_t *lock) > { > struct __raw_tickets tmp = READ_ONCE(lock->tickets); > > - return tmp.tail != tmp.head; > + return (tmp.tail &...

Paravirtualized spinlocks produce some overhead even if the kernel is running on bare metal. The main reason are the more complex locking and unlocking functions. Especially unlocking is no longer just one instruction but so complex that it is no longer inlined. This patch series addresses this issue by adding two more pvops functions to reduce the size of the inlined spinlock functions. When

Paravirtualized spinlocks produce some overhead even if the kernel is running on bare metal. The main reason are the more complex locking and unlocking functions. Especially unlocking is no longer just one instruction but so complex that it is no longer inlined. This patch series addresses this issue by adding two more pvops functions to reduce the size of the inlined spinlock functions. When