Message ID | 1460132182-11690-2-git-send-email-Waiman.Long@hpe.com |
---|---|
State | Superseded, archived |
Headers | show |
Hi Waiman, [auto build test ERROR on ext4/dev] [also build test ERROR on v4.6-rc2 next-20160408] [if your patch is applied to the wrong git tree, please drop us a note to help improving the system] url: https://github.com/0day-ci/linux/commits/Waiman-Long/ext4-Improve-parallel-I-O-performance-on-NVDIMM/20160409-002128 base: https://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git dev config: alpha-defconfig (attached as .config) reproduce: wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # save the attached .config to linux build tree make.cross ARCH=alpha Note: the linux-review/Waiman-Long/ext4-Improve-parallel-I-O-performance-on-NVDIMM/20160409-002128 HEAD 712a939b92b9178cb79df4050bba8e6b1d03ca63 builds fine. It only hurts bisectibility. All errors (new ones prefixed by >>): In file included from lib/percpu_stats.c:5:0: include/linux/percpu_stats.h: In function 'percpu_stats_add': >> include/linux/percpu_stats.h:29:2: error: implicit declaration of function 'raw_local_irq_save' [-Werror=implicit-function-declaration] this_cpu_add(pcs->stats[stat], cnt); ^ >> include/linux/percpu_stats.h:29:2: error: implicit declaration of function 'raw_local_irq_restore' [-Werror=implicit-function-declaration] cc1: some warnings being treated as errors vim +/raw_local_irq_save +29 include/linux/percpu_stats.h 23 * @cnt: The value to be added to the statistics count 24 */ 25 static inline void 26 percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt) 27 { 28 BUG_ON((unsigned int)stat >= pcs->nstats); > 29 this_cpu_add(pcs->stats[stat], cnt); 30 } 31 32 static inline void percpu_stats_inc(struct percpu_stats *pcs, int stat) --- 0-DAY kernel test infrastructure Open Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation
Hi Waiman, [auto build test ERROR on ext4/dev] [also build test ERROR on v4.6-rc2 next-20160408] [if your patch is applied to the wrong git tree, please drop us a note to help improving the system] url: https://github.com/0day-ci/linux/commits/Waiman-Long/ext4-Improve-parallel-I-O-performance-on-NVDIMM/20160409-002128 base: https://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4.git dev config: um-x86_64_defconfig (attached as .config) reproduce: # save the attached .config to linux build tree make ARCH=um SUBARCH=x86_64 Note: the linux-review/Waiman-Long/ext4-Improve-parallel-I-O-performance-on-NVDIMM/20160409-002128 HEAD 712a939b92b9178cb79df4050bba8e6b1d03ca63 builds fine. It only hurts bisectibility. All error/warnings (new ones prefixed by >>): In file included from arch/um/include/generated/asm/percpu.h:1:0, from include/linux/percpu.h:12, from include/linux/percpu_stats.h:7, from lib/percpu_stats.c:5: include/linux/percpu_stats.h: In function 'percpu_stats_add': >> include/asm-generic/percpu.h:120:2: error: implicit declaration of function 'raw_local_irq_save' [-Werror=implicit-function-declaration] raw_local_irq_save(__flags); \ ^ >> include/asm-generic/percpu.h:322:34: note: in expansion of macro 'this_cpu_generic_to_op' #define this_cpu_add_1(pcp, val) this_cpu_generic_to_op(pcp, val, +=) ^ >> include/linux/percpu-defs.h:364:11: note: in expansion of macro 'this_cpu_add_1' case 1: stem##1(variable, __VA_ARGS__);break; \ ^ include/linux/percpu-defs.h:496:33: note: in expansion of macro '__pcpu_size_call' #define this_cpu_add(pcp, val) __pcpu_size_call(this_cpu_add_, pcp, val) ^ >> include/linux/percpu_stats.h:29:2: note: in expansion of macro 'this_cpu_add' this_cpu_add(pcs->stats[stat], cnt); ^ >> include/asm-generic/percpu.h:122:2: error: implicit declaration of function 'raw_local_irq_restore' [-Werror=implicit-function-declaration] raw_local_irq_restore(__flags); \ ^ >> include/asm-generic/percpu.h:322:34: note: in expansion of macro 'this_cpu_generic_to_op' #define this_cpu_add_1(pcp, val) this_cpu_generic_to_op(pcp, val, +=) ^ >> include/linux/percpu-defs.h:364:11: note: in expansion of macro 'this_cpu_add_1' case 1: stem##1(variable, __VA_ARGS__);break; \ ^ include/linux/percpu-defs.h:496:33: note: in expansion of macro '__pcpu_size_call' #define this_cpu_add(pcp, val) __pcpu_size_call(this_cpu_add_, pcp, val) ^ >> include/linux/percpu_stats.h:29:2: note: in expansion of macro 'this_cpu_add' this_cpu_add(pcs->stats[stat], cnt); ^ cc1: some warnings being treated as errors vim +/raw_local_irq_save +120 include/asm-generic/percpu.h eba117889a Tejun Heo 2014-06-17 114 __ret; \ 9c28278a24 Tejun Heo 2014-06-17 115 }) 9c28278a24 Tejun Heo 2014-06-17 116 eba117889a Tejun Heo 2014-06-17 117 #define this_cpu_generic_to_op(pcp, val, op) \ 9c28278a24 Tejun Heo 2014-06-17 118 do { \ eba117889a Tejun Heo 2014-06-17 119 unsigned long __flags; \ eba117889a Tejun Heo 2014-06-17 @120 raw_local_irq_save(__flags); \ 9c28278a24 Tejun Heo 2014-06-17 121 *raw_cpu_ptr(&(pcp)) op val; \ eba117889a Tejun Heo 2014-06-17 @122 raw_local_irq_restore(__flags); \ 9c28278a24 Tejun Heo 2014-06-17 123 } while (0) 9c28278a24 Tejun Heo 2014-06-17 124 eba117889a Tejun Heo 2014-06-17 125 #define this_cpu_generic_add_return(pcp, val) \ :::::: The code at line 120 was first introduced by commit :::::: eba117889ac444bea6e8270049cbaeed48169889 percpu: preffity percpu header files :::::: TO: Tejun Heo <tj@kernel.org> :::::: CC: Tejun Heo <tj@kernel.org> --- 0-DAY kernel test infrastructure Open Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation
diff --git a/include/linux/percpu_stats.h b/include/linux/percpu_stats.h new file mode 100644 index 0000000..ed6e8ac --- /dev/null +++ b/include/linux/percpu_stats.h @@ -0,0 +1,42 @@ +#ifndef _LINUX_PERCPU_STATS_H +#define _LINUX_PERCPU_STATS_H +/* + * Simple per-cpu statistics counts that have less overhead than the + * per-cpu counters. + */ +#include <linux/percpu.h> +#include <linux/types.h> + +struct percpu_stats { + unsigned long __percpu *stats; + int nstats; /* Number of statistics counts in stats array */ +}; + +extern void percpu_stats_destroy(struct percpu_stats *pcs); +extern int percpu_stats_init(struct percpu_stats *pcs, int num); +extern uint64_t percpu_stats_sum(struct percpu_stats *pcs, int stat); + +/** + * percpu_stats_add - Add the given value to a statistics count + * @pcs: Pointer to percpu_stats structure + * @stat: The statistics count that needs to be updated + * @cnt: The value to be added to the statistics count + */ +static inline void +percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt) +{ + BUG_ON((unsigned int)stat >= pcs->nstats); + this_cpu_add(pcs->stats[stat], cnt); +} + +static inline void percpu_stats_inc(struct percpu_stats *pcs, int stat) +{ + percpu_stats_add(pcs, stat, 1); +} + +static inline void percpu_stats_dec(struct percpu_stats *pcs, int stat) +{ + percpu_stats_add(pcs, stat, -1); +} + +#endif /* _LINUX_PERCPU_STATS_H */ diff --git a/lib/Makefile b/lib/Makefile index 7bd6fd4..5037c62 100644 --- a/lib/Makefile +++ b/lib/Makefile @@ -40,7 +40,7 @@ obj-y += bcd.o div64.o sort.o parser.o halfmd4.o debug_locks.o random32.o \ gcd.o lcm.o list_sort.o uuid.o flex_array.o iov_iter.o clz_ctz.o \ bsearch.o find_bit.o llist.o memweight.o kfifo.o \ percpu-refcount.o percpu_ida.o rhashtable.o reciprocal_div.o \ - once.o + once.o percpu_stats.o obj-y += string_helpers.o obj-$(CONFIG_TEST_STRING_HELPERS) += test-string_helpers.o obj-y += hexdump.o diff --git a/lib/percpu_stats.c b/lib/percpu_stats.c new file mode 100644 index 0000000..bc9f26d --- /dev/null +++ b/lib/percpu_stats.c @@ -0,0 +1,64 @@ +/* + * Simple per-cpu statistics counts that have less overhead than the + * per-cpu counters. + */ +#include <linux/percpu_stats.h> +#include <linux/bug.h> + +/** + * percpu_stats_init - allocate memory for the percpu statistics counts + * @pcs: Pointer to percpu_stats structure + * @num: Number of statistics counts to be used + * Return: 0 if successful, -ENOMEM if memory allocation fails. + */ +int percpu_stats_init(struct percpu_stats *pcs, int num) +{ + int cpu; + + pcs->nstats = num; + pcs->stats = __alloc_percpu(sizeof(unsigned long) * num, + __alignof__(unsigned long)); + if (!pcs->stats) + return -ENOMEM; + + for_each_possible_cpu(cpu) { + unsigned long *pstats = per_cpu_ptr(pcs->stats, cpu); + int stat; + + for (stat = 0; stat < pcs->nstats; stat++, pstats++) + *pstats = 0; + } + return 0; +} +EXPORT_SYMBOL(percpu_stats_init); + +/** + * percpu_stats_destroy - free the memory used by the statistics counts + * @pcs: Pointer to percpu_stats structure + */ +void percpu_stats_destroy(struct percpu_stats *pcs) +{ + free_percpu(pcs->stats); + pcs->stats = NULL; + pcs->nstats = 0; +} +EXPORT_SYMBOL(percpu_stats_destroy); + +/** + * percpu_stats_sum - compute the percpu sum of the given statistics count + * @pcs : Pointer to percpu_stats structure + * @stat : The statistics count whose sum needs to be computed + * Return: Sum of percpu count values + */ +uint64_t percpu_stats_sum(struct percpu_stats *pcs, int stat) +{ + int cpu; + uint64_t sum = 0; + + BUG_ON((unsigned int)stat >= pcs->nstats); + + for_each_possible_cpu(cpu) + sum += per_cpu(pcs->stats[stat], cpu); + return sum; +} +EXPORT_SYMBOL(percpu_stats_sum);
This patch introduces a set of simple per-cpu statictics count helper functions that can be used by other kernel subsystems for keeping track of the number of events that happens. It is per-cpu based to reduce overhead and improve accuracy of the counter. Using per-cpu counter is usually overkill for such purpose. The following APIs are provided: - int percpu_stats_init(struct percpu_stats *pcs, int num) Initialize the per-cpu statictics counts structure which should have the given number of statistics counts. Return -ENOMEM on error. - void percpu_stats_destroy(struct percpu_stats *pcs) Free the percpu memory allocated. - void percpu_stats_inc(struct percpu_stats *pcs, int stat) void percpu_stats_dec(struct percpu_stats *pcs, int stat) void percpu_stats_add(struct percpu_stats *pcs, int stat, int cnt) Increment, decrement and add to the given per-cpu statistics count. - unsigned long percpu_stats_sum(struct percpu_stats *pcs, int stat) Return the current aggregated sum of the given statistics count. Signed-off-by: Waiman Long <Waiman.Long@hpe.com> --- include/linux/percpu_stats.h | 42 +++++++++++++++++++++++++++ lib/Makefile | 2 +- lib/percpu_stats.c | 64 ++++++++++++++++++++++++++++++++++++++++++ 3 files changed, 107 insertions(+), 1 deletions(-) create mode 100644 include/linux/percpu_stats.h create mode 100644 lib/percpu_stats.c