search for: yield_count

...h> +#include <asm/hvcall.h> +#endif + +#ifdef CONFIG_PPC_SPLPAR +DECLARE_STATIC_KEY_FALSE(shared_processor); + +static inline bool is_shared_processor(void) +{ + return static_branch_unlikely(&shared_processor); +} + +/* If bit 0 is set, the cpu has been preempted */ +static inline u32 yield_count_of(int cpu) +{ + __be32 yield_count = READ_ONCE(lppaca_of(cpu).yield_count); + return be32_to_cpu(yield_count); +} + +static inline void yield_to_preempted(int cpu, u32 yield_count) +{ + plpar_hcall_norets(H_CONFER, get_hard_smp_processor_id(cpu), yield_count); +} +#else +static inline bool is_shar...

...ices to a specified cpu and return. If it is in running state > + * or cpu is -1, then we will check confer. If confer is NULL, we will return > + * otherwise we confer our slices to lpar. > + */ > +void __spin_yield_cpu(int cpu, int confer) > +{ > + unsigned int holder_cpu = cpu, yield_count; As I said at: https://marc.info/?l=linux-kernel&m=147455748619343&w=2 @holder_cpu is not necessary and doesn't help anything. > + > + if (cpu == -1) > + goto yield_to_lpar; > + > + BUG_ON(holder_cpu >= nr_cpu_ids); > + yield_count = be32_to_cpu(lppaca_of(holde...

...ices to a specified cpu and return. If it is in running state > + * or cpu is -1, then we will check confer. If confer is NULL, we will return > + * otherwise we confer our slices to lpar. > + */ > +void __spin_yield_cpu(int cpu, int confer) > +{ > + unsigned int holder_cpu = cpu, yield_count; As I said at: https://marc.info/?l=linux-kernel&m=147455748619343&w=2 @holder_cpu is not necessary and doesn't help anything. > + > + if (cpu == -1) > + goto yield_to_lpar; > + > + BUG_ON(holder_cpu >= nr_cpu_ids); > + yield_count = be32_to_cpu(lppaca_of(holde...

....h> +/* + * confer our slices to a specified cpu and return. If it is in running state + * or cpu is -1, then we will check confer. If confer is NULL, we will return + * otherwise we confer our slices to lpar. + */ +void __spin_yield_cpu(int cpu, int confer) +{ + unsigned int holder_cpu = cpu, yield_count; + + if (cpu == -1) + goto yield_to_lpar; + + BUG_ON(holder_cpu >= nr_cpu_ids); + yield_count = be32_to_cpu(lppaca_of(holder_cpu).yield_count); + + /* if cpu is running, confer slices to lpar conditionally*/ + if ((yield_count & 1) == 0) + goto yield_to_lpar; + + plpar_hcall_norets(H_CONFE...

...-git a/arch/powerpc/lib/locks.c b/arch/powerpc/lib/locks.c index a9ebd71..4109679a 100644 --- a/arch/powerpc/lib/locks.c +++ b/arch/powerpc/lib/locks.c @@ -23,6 +23,38 @@ #include <asm/hvcall.h> #include <asm/smp.h> +void __spin_yield_cpu(int cpu) +{ + unsigned int holder_cpu = cpu, yield_count; + + if (cpu == -1) { + plpar_hcall_norets(H_CEDE); + return; + } + BUG_ON(holder_cpu >= nr_cpu_ids); + yield_count = be32_to_cpu(lppaca_of(holder_cpu).yield_count); + if ((yield_count & 1) == 0) + return; /* virtual cpu is currently running */ + rmb(); + plpar_hcall_norets(H_CONFER, +...

This series adds an option to use queued spinlocks for powerpc, and makes it the default for the Book3S-64 subarch. This effort starts with the generic code so it's very simple but still very performant. There are optimisations that can be made to slowpaths, but I think it's better to attack those incrementally if/when we find things, and try to add the improvements to generic code as

...locks.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arch/powerpc/lib/locks.c b/arch/powerpc/lib/locks.c index 6440d5943c00..47a530de733e 100644 --- a/arch/powerpc/lib/locks.c +++ b/arch/powerpc/lib/locks.c @@ -30,7 +30,7 @@ void splpar_spin_yield(arch_spinlock_t *lock) yield_count = be32_to_cpu(lppaca_of(holder_cpu).yield_count); if ((yield_count & 1) == 0) return; /* virtual cpu is currently running */ - rmb(); + smp_rmb(); if (lock->slock != lock_value) return; /* something has changed */ plpar_hcall_norets(H_CONFER, @@ -56,7 +56,7 @@ void splpar_rw_yi...

v2 is updated to account for feedback from Will, Peter, and Waiman (thank you), and trims off a couple of RFC and unrelated patches. Thanks, Nick Nicholas Piggin (6): powerpc/powernv: must include hvcall.h to get PAPR defines powerpc/pseries: move some PAPR paravirt functions to their own file powerpc: move spinlock implementation to simple_spinlock powerpc/64s: implement queued

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

Updated with everybody's feedback (thanks all), and more performance results. What I've found is I might have been measuring the worst load point for the paravirt case, and by looking at a range of loads it's clear that queued spinlocks are overall better even on PV, doubly so when you look at the generally much improved worst case latencies. I have defaulted it to N even though

...--git a/arch/powerpc/lib/locks.c b/arch/powerpc/lib/locks.c index f7deebd..6e9d3bb 100644 --- a/arch/powerpc/lib/locks.c +++ b/arch/powerpc/lib/locks.c @@ -23,6 +23,35 @@ #include <asm/hvcall.h> #include <asm/smp.h> +void __spin_yield_cpu(int cpu) +{ + unsigned int holder_cpu = cpu, yield_count; + + if (cpu == -1) { + plpar_hcall_norets(H_CEDE); + return; + } + BUG_ON(holder_cpu >= nr_cpu_ids); + yield_count = be32_to_cpu(lppaca_of(holder_cpu).yield_count); + if ((yield_count & 1) == 0) + return; /* virtual cpu is currently running */ + rmb(); + plpar_hcall_norets(H_CONFER, +...

...--git a/arch/powerpc/lib/locks.c b/arch/powerpc/lib/locks.c index f7deebd..6e9d3bb 100644 --- a/arch/powerpc/lib/locks.c +++ b/arch/powerpc/lib/locks.c @@ -23,6 +23,35 @@ #include <asm/hvcall.h> #include <asm/smp.h> +void __spin_yield_cpu(int cpu) +{ + unsigned int holder_cpu = cpu, yield_count; + + if (cpu == -1) { + plpar_hcall_norets(H_CEDE); + return; + } + BUG_ON(holder_cpu >= nr_cpu_ids); + yield_count = be32_to_cpu(lppaca_of(holder_cpu).yield_count); + if ((yield_count & 1) == 0) + return; /* virtual cpu is currently running */ + rmb(); + plpar_hcall_norets(H_CONFER, +...

change fome v1: separate into 6 pathes from one patch some minor code changes. benchmark test results are below. run 3 tests on pseries IBM,8408-E8E with 32cpus, 64GB memory perf bench futex hash perf bench futex lock-pi perf record -advRT || perf bench sched messaging -g 1000 || perf report summary: _____test________________spinlcok______________pv-qspinlcok_____ |futex hash | 556370 ops |

...iff --git a/arch/powerpc/include/asm/paravirt.h b/arch/powerpc/include/asm/paravirt.h > index 7a8546660a63..f2d51f929cf5 100644 > --- a/arch/powerpc/include/asm/paravirt.h > +++ b/arch/powerpc/include/asm/paravirt.h > @@ -29,6 +29,16 @@ static inline void yield_to_preempted(int cpu, u32 yield_count) > { > plpar_hcall_norets(H_CONFER, get_hard_smp_processor_id(cpu), yield_count); > } > + > +static inline void prod_cpu(int cpu) > +{ > + plpar_hcall_norets(H_PROD, get_hard_smp_processor_id(cpu)); > +} > + > +static inline void yield_to_any(void) > +{ >...

change fome v1: separate into 6 pathes from one patch some minor code changes. benchmark test results are below. run 3 tests on pseries IBM,8408-E8E with 32cpus, 64GB memory perf bench futex hash perf bench futex lock-pi perf record -advRT || perf bench sched messaging -g 1000 || perf report summary: _____test________________spinlcok______________pv-qspinlcok_____ |futex hash | 556370 ops |

change from v2: __spin_yeild_cpu() will yield slices to lpar if target cpu is running. remove unnecessary rmb() in __spin_yield/wake_cpu. __pv_wait() will check the *ptr == val. some commit message change change fome v1: separate into 6 pathes from one patch some minor code changes. I do several tests on pseries IBM,8408-E8E with 32cpus, 64GB memory. benchmark test results are below. 2

change from v2: __spin_yeild_cpu() will yield slices to lpar if target cpu is running. remove unnecessary rmb() in __spin_yield/wake_cpu. __pv_wait() will check the *ptr == val. some commit message change change fome v1: separate into 6 pathes from one patch some minor code changes. I do several tests on pseries IBM,8408-E8E with 32cpus, 64GB memory. benchmark test results are below. 2

v3 is updated to use __pv_queued_spin_unlock, noticed by Waiman (thank you). Thanks, Nick Nicholas Piggin (6): powerpc/powernv: must include hvcall.h to get PAPR defines powerpc/pseries: move some PAPR paravirt functions to their own file powerpc: move spinlock implementation to simple_spinlock powerpc/64s: implement queued spinlocks and rwlocks powerpc/pseries: implement paravirt

v3 is updated to use __pv_queued_spin_unlock, noticed by Waiman (thank you). Thanks, Nick Nicholas Piggin (6): powerpc/powernv: must include hvcall.h to get PAPR defines powerpc/pseries: move some PAPR paravirt functions to their own file powerpc: move spinlock implementation to simple_spinlock powerpc/64s: implement queued spinlocks and rwlocks powerpc/pseries: implement paravirt