linux_dsm_epyc7002/tools/perf/util/parse-events.c

683 lines
15 KiB
C
Raw Normal View History

#include "../perf.h"
#include "util.h"
#include "parse-options.h"
#include "parse-events.h"
#include "exec_cmd.h"
#include "string.h"
#include "cache.h"
extern char *strcasestr(const char *haystack, const char *needle);
int nr_counters;
struct perf_counter_attr attrs[MAX_COUNTERS];
struct event_symbol {
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
u8 type;
u64 config;
const char *symbol;
const char *alias;
};
char debugfs_path[MAXPATHLEN];
#define CHW(x) .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_##x
#define CSW(x) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_##x
static struct event_symbol event_symbols[] = {
{ CHW(CPU_CYCLES), "cpu-cycles", "cycles" },
{ CHW(INSTRUCTIONS), "instructions", "" },
{ CHW(CACHE_REFERENCES), "cache-references", "" },
{ CHW(CACHE_MISSES), "cache-misses", "" },
{ CHW(BRANCH_INSTRUCTIONS), "branch-instructions", "branches" },
{ CHW(BRANCH_MISSES), "branch-misses", "" },
{ CHW(BUS_CYCLES), "bus-cycles", "" },
{ CSW(CPU_CLOCK), "cpu-clock", "" },
{ CSW(TASK_CLOCK), "task-clock", "" },
{ CSW(PAGE_FAULTS), "page-faults", "faults" },
{ CSW(PAGE_FAULTS_MIN), "minor-faults", "" },
{ CSW(PAGE_FAULTS_MAJ), "major-faults", "" },
{ CSW(CONTEXT_SWITCHES), "context-switches", "cs" },
{ CSW(CPU_MIGRATIONS), "cpu-migrations", "migrations" },
};
#define __PERF_COUNTER_FIELD(config, name) \
((config & PERF_COUNTER_##name##_MASK) >> PERF_COUNTER_##name##_SHIFT)
#define PERF_COUNTER_RAW(config) __PERF_COUNTER_FIELD(config, RAW)
#define PERF_COUNTER_CONFIG(config) __PERF_COUNTER_FIELD(config, CONFIG)
#define PERF_COUNTER_TYPE(config) __PERF_COUNTER_FIELD(config, TYPE)
#define PERF_COUNTER_ID(config) __PERF_COUNTER_FIELD(config, EVENT)
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *hw_event_names[] = {
"cycles",
"instructions",
"cache-references",
"cache-misses",
"branches",
"branch-misses",
"bus-cycles",
};
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *sw_event_names[] = {
"cpu-clock-msecs",
"task-clock-msecs",
"page-faults",
"context-switches",
"CPU-migrations",
"minor-faults",
"major-faults",
};
#define MAX_ALIASES 8
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *hw_cache[][MAX_ALIASES] = {
{ "L1-dcache", "l1-d", "l1d", "L1-data", },
{ "L1-icache", "l1-i", "l1i", "L1-instruction", },
perf_counter tools: Shorten names for events Added new alias for events. On AMD box: $ ./perf stat -e l1d -e l1d-misses -e l1d-write -e l1d-prefetch -e l1d-prefetch-miss -e l1i -e l1i-misses -e l1i-prefetch -e l2 -e l2-misses -e l2-write -e dtlb -e dtlb-misses -e itlb -e itlb-misses -e bpu -e bpu-misses -- ls -lR /usr/include/ > /dev/null Before : Performance counter stats for 'ls -lR /usr/include/': 248064467 L1-data-Cache-Load-Referencees (scaled from 23.27%) 1001433 L1-data-Cache-Load-Misses (scaled from 23.34%) 153691 L1-data-Cache-Store-Referencees (scaled from 23.34%) 423248 L1-data-Cache-Prefetch-Referencees (scaled from 23.33%) 302138 L1-data-Cache-Prefetch-Misses (scaled from 23.25%) 251217546 L1-instruction-Cache-Load-Referencees (scaled from 23.25%) 5757005 L1-instruction-Cache-Load-Misses (scaled from 23.23%) 93435 L1-instruction-Cache-Prefetch-Referencees (scaled from 23.24%) 6496073 L2-Cache-Load-Referencees (scaled from 23.32%) 609485 L2-Cache-Load-Misses (scaled from 23.45%) 6876991 L2-Cache-Store-Referencees (scaled from 23.71%) 248922840 Data-TLB-Cache-Load-Referencees (scaled from 23.94%) 5828386 Data-TLB-Cache-Load-Misses (scaled from 24.17%) 257613506 Instruction-TLB-Cache-Load-Referencees (scaled from 24.20%) 6833 Instruction-TLB-Cache-Load-Misses (scaled from 23.88%) 109043606 Branch-Cache-Load-Referencees (scaled from 23.64%) 5552296 Branch-Cache-Load-Misses (scaled from 23.42%) 0.413702461 seconds time elapsed. After : Peformance counter stats for 'ls -lR /usr/include/': 266590464 L1-d$-loads (scaled from 23.03%) 1222273 L1-d$-load-misses (scaled from 23.58%) 146204 L1-d$-stores (scaled from 23.83%) 406344 L1-d$-prefetches (scaled from 24.09%) 283748 L1-d$-prefetch-misses (scaled from 24.10%) 249650965 L1-i$-loads (scaled from 23.80%) 3353961 L1-i$-load-misses (scaled from 23.82%) 104599 L1-i$-prefetches (scaled from 23.68%) 4836405 LLC-loads (scaled from 23.67%) 498214 LLC-load-misses (scaled from 23.66%) 4953994 LLC-stores (scaled from 23.64%) 243354097 dTLB-loads (scaled from 23.77%) 6468584 dTLB-load-misses (scaled from 23.74%) 249719549 iTLB-loads (scaled from 23.25%) 5060 iTLB-load-misses (scaled from 23.00%) 112343016 branch-loads (scaled from 22.76%) 5528876 branch-load-misses (scaled from 22.54%) 0.427154051 seconds time elapsed. Reported-by : Ingo Molnar <mingo@elte.hu> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> LKML-Reference: <1245934522.5308.39.camel@hpdv5.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-25 19:55:22 +07:00
{ "LLC", "L2" },
{ "dTLB", "d-tlb", "Data-TLB", },
{ "iTLB", "i-tlb", "Instruction-TLB", },
{ "branch", "branches", "bpu", "btb", "bpc", },
};
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *hw_cache_op[][MAX_ALIASES] = {
perf_counter tools: Shorten names for events Added new alias for events. On AMD box: $ ./perf stat -e l1d -e l1d-misses -e l1d-write -e l1d-prefetch -e l1d-prefetch-miss -e l1i -e l1i-misses -e l1i-prefetch -e l2 -e l2-misses -e l2-write -e dtlb -e dtlb-misses -e itlb -e itlb-misses -e bpu -e bpu-misses -- ls -lR /usr/include/ > /dev/null Before : Performance counter stats for 'ls -lR /usr/include/': 248064467 L1-data-Cache-Load-Referencees (scaled from 23.27%) 1001433 L1-data-Cache-Load-Misses (scaled from 23.34%) 153691 L1-data-Cache-Store-Referencees (scaled from 23.34%) 423248 L1-data-Cache-Prefetch-Referencees (scaled from 23.33%) 302138 L1-data-Cache-Prefetch-Misses (scaled from 23.25%) 251217546 L1-instruction-Cache-Load-Referencees (scaled from 23.25%) 5757005 L1-instruction-Cache-Load-Misses (scaled from 23.23%) 93435 L1-instruction-Cache-Prefetch-Referencees (scaled from 23.24%) 6496073 L2-Cache-Load-Referencees (scaled from 23.32%) 609485 L2-Cache-Load-Misses (scaled from 23.45%) 6876991 L2-Cache-Store-Referencees (scaled from 23.71%) 248922840 Data-TLB-Cache-Load-Referencees (scaled from 23.94%) 5828386 Data-TLB-Cache-Load-Misses (scaled from 24.17%) 257613506 Instruction-TLB-Cache-Load-Referencees (scaled from 24.20%) 6833 Instruction-TLB-Cache-Load-Misses (scaled from 23.88%) 109043606 Branch-Cache-Load-Referencees (scaled from 23.64%) 5552296 Branch-Cache-Load-Misses (scaled from 23.42%) 0.413702461 seconds time elapsed. After : Peformance counter stats for 'ls -lR /usr/include/': 266590464 L1-d$-loads (scaled from 23.03%) 1222273 L1-d$-load-misses (scaled from 23.58%) 146204 L1-d$-stores (scaled from 23.83%) 406344 L1-d$-prefetches (scaled from 24.09%) 283748 L1-d$-prefetch-misses (scaled from 24.10%) 249650965 L1-i$-loads (scaled from 23.80%) 3353961 L1-i$-load-misses (scaled from 23.82%) 104599 L1-i$-prefetches (scaled from 23.68%) 4836405 LLC-loads (scaled from 23.67%) 498214 LLC-load-misses (scaled from 23.66%) 4953994 LLC-stores (scaled from 23.64%) 243354097 dTLB-loads (scaled from 23.77%) 6468584 dTLB-load-misses (scaled from 23.74%) 249719549 iTLB-loads (scaled from 23.25%) 5060 iTLB-load-misses (scaled from 23.00%) 112343016 branch-loads (scaled from 22.76%) 5528876 branch-load-misses (scaled from 22.54%) 0.427154051 seconds time elapsed. Reported-by : Ingo Molnar <mingo@elte.hu> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> LKML-Reference: <1245934522.5308.39.camel@hpdv5.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-25 19:55:22 +07:00
{ "load", "loads", "read", },
{ "store", "stores", "write", },
{ "prefetch", "prefetches", "speculative-read", "speculative-load", },
};
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *hw_cache_result[][MAX_ALIASES] = {
perf_counter tools: Shorten names for events Added new alias for events. On AMD box: $ ./perf stat -e l1d -e l1d-misses -e l1d-write -e l1d-prefetch -e l1d-prefetch-miss -e l1i -e l1i-misses -e l1i-prefetch -e l2 -e l2-misses -e l2-write -e dtlb -e dtlb-misses -e itlb -e itlb-misses -e bpu -e bpu-misses -- ls -lR /usr/include/ > /dev/null Before : Performance counter stats for 'ls -lR /usr/include/': 248064467 L1-data-Cache-Load-Referencees (scaled from 23.27%) 1001433 L1-data-Cache-Load-Misses (scaled from 23.34%) 153691 L1-data-Cache-Store-Referencees (scaled from 23.34%) 423248 L1-data-Cache-Prefetch-Referencees (scaled from 23.33%) 302138 L1-data-Cache-Prefetch-Misses (scaled from 23.25%) 251217546 L1-instruction-Cache-Load-Referencees (scaled from 23.25%) 5757005 L1-instruction-Cache-Load-Misses (scaled from 23.23%) 93435 L1-instruction-Cache-Prefetch-Referencees (scaled from 23.24%) 6496073 L2-Cache-Load-Referencees (scaled from 23.32%) 609485 L2-Cache-Load-Misses (scaled from 23.45%) 6876991 L2-Cache-Store-Referencees (scaled from 23.71%) 248922840 Data-TLB-Cache-Load-Referencees (scaled from 23.94%) 5828386 Data-TLB-Cache-Load-Misses (scaled from 24.17%) 257613506 Instruction-TLB-Cache-Load-Referencees (scaled from 24.20%) 6833 Instruction-TLB-Cache-Load-Misses (scaled from 23.88%) 109043606 Branch-Cache-Load-Referencees (scaled from 23.64%) 5552296 Branch-Cache-Load-Misses (scaled from 23.42%) 0.413702461 seconds time elapsed. After : Peformance counter stats for 'ls -lR /usr/include/': 266590464 L1-d$-loads (scaled from 23.03%) 1222273 L1-d$-load-misses (scaled from 23.58%) 146204 L1-d$-stores (scaled from 23.83%) 406344 L1-d$-prefetches (scaled from 24.09%) 283748 L1-d$-prefetch-misses (scaled from 24.10%) 249650965 L1-i$-loads (scaled from 23.80%) 3353961 L1-i$-load-misses (scaled from 23.82%) 104599 L1-i$-prefetches (scaled from 23.68%) 4836405 LLC-loads (scaled from 23.67%) 498214 LLC-load-misses (scaled from 23.66%) 4953994 LLC-stores (scaled from 23.64%) 243354097 dTLB-loads (scaled from 23.77%) 6468584 dTLB-load-misses (scaled from 23.74%) 249719549 iTLB-loads (scaled from 23.25%) 5060 iTLB-load-misses (scaled from 23.00%) 112343016 branch-loads (scaled from 22.76%) 5528876 branch-load-misses (scaled from 22.54%) 0.427154051 seconds time elapsed. Reported-by : Ingo Molnar <mingo@elte.hu> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> LKML-Reference: <1245934522.5308.39.camel@hpdv5.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-25 19:55:22 +07:00
{ "refs", "Reference", "ops", "access", },
{ "misses", "miss", },
};
#define C(x) PERF_COUNT_HW_CACHE_##x
#define CACHE_READ (1 << C(OP_READ))
#define CACHE_WRITE (1 << C(OP_WRITE))
#define CACHE_PREFETCH (1 << C(OP_PREFETCH))
#define COP(x) (1 << x)
/*
* cache operartion stat
* L1I : Read and prefetch only
* ITLB and BPU : Read-only
*/
static unsigned long hw_cache_stat[C(MAX)] = {
[C(L1D)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(L1I)] = (CACHE_READ | CACHE_PREFETCH),
[C(LL)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(DTLB)] = (CACHE_READ | CACHE_WRITE | CACHE_PREFETCH),
[C(ITLB)] = (CACHE_READ),
[C(BPU)] = (CACHE_READ),
};
#define for_each_subsystem(sys_dir, sys_dirent, sys_next, file, st) \
while (!readdir_r(sys_dir, &sys_dirent, &sys_next) && sys_next) \
if (snprintf(file, MAXPATHLEN, "%s/%s", debugfs_path, \
sys_dirent.d_name) && \
(!stat(file, &st)) && (S_ISDIR(st.st_mode)) && \
(strcmp(sys_dirent.d_name, ".")) && \
(strcmp(sys_dirent.d_name, "..")))
static int tp_event_has_id(struct dirent *sys_dir, struct dirent *evt_dir)
{
char evt_path[MAXPATHLEN];
int fd;
snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path,
sys_dir->d_name, evt_dir->d_name);
fd = open(evt_path, O_RDONLY);
if (fd < 0)
return -EINVAL;
close(fd);
return 0;
}
#define for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next, file, st) \
while (!readdir_r(evt_dir, &evt_dirent, &evt_next) && evt_next) \
if (snprintf(file, MAXPATHLEN, "%s/%s/%s", debugfs_path, \
sys_dirent.d_name, evt_dirent.d_name) && \
(!stat(file, &st)) && (S_ISDIR(st.st_mode)) && \
(strcmp(evt_dirent.d_name, ".")) && \
(strcmp(evt_dirent.d_name, "..")) && \
(!tp_event_has_id(&sys_dirent, &evt_dirent)))
#define MAX_EVENT_LENGTH 30
int valid_debugfs_mount(const char *debugfs)
{
struct statfs st_fs;
if (statfs(debugfs, &st_fs) < 0)
return -ENOENT;
else if (st_fs.f_type != (long) DEBUGFS_MAGIC)
return -ENOENT;
return 0;
}
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static const char *tracepoint_id_to_name(u64 config)
{
static char tracepoint_name[2 * MAX_EVENT_LENGTH];
DIR *sys_dir, *evt_dir;
struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent;
struct stat st;
char id_buf[4];
int fd;
u64 id;
char evt_path[MAXPATHLEN];
if (valid_debugfs_mount(debugfs_path))
return "unkown";
sys_dir = opendir(debugfs_path);
if (!sys_dir)
goto cleanup;
for_each_subsystem(sys_dir, sys_dirent, sys_next, evt_path, st) {
evt_dir = opendir(evt_path);
if (!evt_dir)
goto cleanup;
for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next,
evt_path, st) {
snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id",
debugfs_path, sys_dirent.d_name,
evt_dirent.d_name);
fd = open(evt_path, O_RDONLY);
if (fd < 0)
continue;
if (read(fd, id_buf, sizeof(id_buf)) < 0) {
close(fd);
continue;
}
close(fd);
id = atoll(id_buf);
if (id == config) {
closedir(evt_dir);
closedir(sys_dir);
snprintf(tracepoint_name, 2 * MAX_EVENT_LENGTH,
"%s:%s", sys_dirent.d_name,
evt_dirent.d_name);
return tracepoint_name;
}
}
closedir(evt_dir);
}
cleanup:
closedir(sys_dir);
return "unkown";
}
static int is_cache_op_valid(u8 cache_type, u8 cache_op)
{
if (hw_cache_stat[cache_type] & COP(cache_op))
return 1; /* valid */
else
return 0; /* invalid */
}
perf_counter tools: Shorten names for events Added new alias for events. On AMD box: $ ./perf stat -e l1d -e l1d-misses -e l1d-write -e l1d-prefetch -e l1d-prefetch-miss -e l1i -e l1i-misses -e l1i-prefetch -e l2 -e l2-misses -e l2-write -e dtlb -e dtlb-misses -e itlb -e itlb-misses -e bpu -e bpu-misses -- ls -lR /usr/include/ > /dev/null Before : Performance counter stats for 'ls -lR /usr/include/': 248064467 L1-data-Cache-Load-Referencees (scaled from 23.27%) 1001433 L1-data-Cache-Load-Misses (scaled from 23.34%) 153691 L1-data-Cache-Store-Referencees (scaled from 23.34%) 423248 L1-data-Cache-Prefetch-Referencees (scaled from 23.33%) 302138 L1-data-Cache-Prefetch-Misses (scaled from 23.25%) 251217546 L1-instruction-Cache-Load-Referencees (scaled from 23.25%) 5757005 L1-instruction-Cache-Load-Misses (scaled from 23.23%) 93435 L1-instruction-Cache-Prefetch-Referencees (scaled from 23.24%) 6496073 L2-Cache-Load-Referencees (scaled from 23.32%) 609485 L2-Cache-Load-Misses (scaled from 23.45%) 6876991 L2-Cache-Store-Referencees (scaled from 23.71%) 248922840 Data-TLB-Cache-Load-Referencees (scaled from 23.94%) 5828386 Data-TLB-Cache-Load-Misses (scaled from 24.17%) 257613506 Instruction-TLB-Cache-Load-Referencees (scaled from 24.20%) 6833 Instruction-TLB-Cache-Load-Misses (scaled from 23.88%) 109043606 Branch-Cache-Load-Referencees (scaled from 23.64%) 5552296 Branch-Cache-Load-Misses (scaled from 23.42%) 0.413702461 seconds time elapsed. After : Peformance counter stats for 'ls -lR /usr/include/': 266590464 L1-d$-loads (scaled from 23.03%) 1222273 L1-d$-load-misses (scaled from 23.58%) 146204 L1-d$-stores (scaled from 23.83%) 406344 L1-d$-prefetches (scaled from 24.09%) 283748 L1-d$-prefetch-misses (scaled from 24.10%) 249650965 L1-i$-loads (scaled from 23.80%) 3353961 L1-i$-load-misses (scaled from 23.82%) 104599 L1-i$-prefetches (scaled from 23.68%) 4836405 LLC-loads (scaled from 23.67%) 498214 LLC-load-misses (scaled from 23.66%) 4953994 LLC-stores (scaled from 23.64%) 243354097 dTLB-loads (scaled from 23.77%) 6468584 dTLB-load-misses (scaled from 23.74%) 249719549 iTLB-loads (scaled from 23.25%) 5060 iTLB-load-misses (scaled from 23.00%) 112343016 branch-loads (scaled from 22.76%) 5528876 branch-load-misses (scaled from 22.54%) 0.427154051 seconds time elapsed. Reported-by : Ingo Molnar <mingo@elte.hu> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> LKML-Reference: <1245934522.5308.39.camel@hpdv5.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-25 19:55:22 +07:00
static char *event_cache_name(u8 cache_type, u8 cache_op, u8 cache_result)
{
static char name[50];
if (cache_result) {
sprintf(name, "%s-%s-%s", hw_cache[cache_type][0],
hw_cache_op[cache_op][0],
hw_cache_result[cache_result][0]);
} else {
sprintf(name, "%s-%s", hw_cache[cache_type][0],
hw_cache_op[cache_op][1]);
}
return name;
}
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
const char *event_name(int counter)
{
u64 config = attrs[counter].config;
int type = attrs[counter].type;
return __event_name(type, config);
}
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
const char *__event_name(int type, u64 config)
{
static char buf[32];
if (type == PERF_TYPE_RAW) {
sprintf(buf, "raw 0x%llx", config);
return buf;
}
switch (type) {
case PERF_TYPE_HARDWARE:
if (config < PERF_COUNT_HW_MAX)
return hw_event_names[config];
return "unknown-hardware";
case PERF_TYPE_HW_CACHE: {
u8 cache_type, cache_op, cache_result;
cache_type = (config >> 0) & 0xff;
if (cache_type > PERF_COUNT_HW_CACHE_MAX)
return "unknown-ext-hardware-cache-type";
cache_op = (config >> 8) & 0xff;
if (cache_op > PERF_COUNT_HW_CACHE_OP_MAX)
return "unknown-ext-hardware-cache-op";
cache_result = (config >> 16) & 0xff;
if (cache_result > PERF_COUNT_HW_CACHE_RESULT_MAX)
return "unknown-ext-hardware-cache-result";
if (!is_cache_op_valid(cache_type, cache_op))
return "invalid-cache";
perf_counter tools: Shorten names for events Added new alias for events. On AMD box: $ ./perf stat -e l1d -e l1d-misses -e l1d-write -e l1d-prefetch -e l1d-prefetch-miss -e l1i -e l1i-misses -e l1i-prefetch -e l2 -e l2-misses -e l2-write -e dtlb -e dtlb-misses -e itlb -e itlb-misses -e bpu -e bpu-misses -- ls -lR /usr/include/ > /dev/null Before : Performance counter stats for 'ls -lR /usr/include/': 248064467 L1-data-Cache-Load-Referencees (scaled from 23.27%) 1001433 L1-data-Cache-Load-Misses (scaled from 23.34%) 153691 L1-data-Cache-Store-Referencees (scaled from 23.34%) 423248 L1-data-Cache-Prefetch-Referencees (scaled from 23.33%) 302138 L1-data-Cache-Prefetch-Misses (scaled from 23.25%) 251217546 L1-instruction-Cache-Load-Referencees (scaled from 23.25%) 5757005 L1-instruction-Cache-Load-Misses (scaled from 23.23%) 93435 L1-instruction-Cache-Prefetch-Referencees (scaled from 23.24%) 6496073 L2-Cache-Load-Referencees (scaled from 23.32%) 609485 L2-Cache-Load-Misses (scaled from 23.45%) 6876991 L2-Cache-Store-Referencees (scaled from 23.71%) 248922840 Data-TLB-Cache-Load-Referencees (scaled from 23.94%) 5828386 Data-TLB-Cache-Load-Misses (scaled from 24.17%) 257613506 Instruction-TLB-Cache-Load-Referencees (scaled from 24.20%) 6833 Instruction-TLB-Cache-Load-Misses (scaled from 23.88%) 109043606 Branch-Cache-Load-Referencees (scaled from 23.64%) 5552296 Branch-Cache-Load-Misses (scaled from 23.42%) 0.413702461 seconds time elapsed. After : Peformance counter stats for 'ls -lR /usr/include/': 266590464 L1-d$-loads (scaled from 23.03%) 1222273 L1-d$-load-misses (scaled from 23.58%) 146204 L1-d$-stores (scaled from 23.83%) 406344 L1-d$-prefetches (scaled from 24.09%) 283748 L1-d$-prefetch-misses (scaled from 24.10%) 249650965 L1-i$-loads (scaled from 23.80%) 3353961 L1-i$-load-misses (scaled from 23.82%) 104599 L1-i$-prefetches (scaled from 23.68%) 4836405 LLC-loads (scaled from 23.67%) 498214 LLC-load-misses (scaled from 23.66%) 4953994 LLC-stores (scaled from 23.64%) 243354097 dTLB-loads (scaled from 23.77%) 6468584 dTLB-load-misses (scaled from 23.74%) 249719549 iTLB-loads (scaled from 23.25%) 5060 iTLB-load-misses (scaled from 23.00%) 112343016 branch-loads (scaled from 22.76%) 5528876 branch-load-misses (scaled from 22.54%) 0.427154051 seconds time elapsed. Reported-by : Ingo Molnar <mingo@elte.hu> Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> LKML-Reference: <1245934522.5308.39.camel@hpdv5.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-06-25 19:55:22 +07:00
return event_cache_name(cache_type, cache_op, cache_result);
}
case PERF_TYPE_SOFTWARE:
if (config < PERF_COUNT_SW_MAX)
return sw_event_names[config];
return "unknown-software";
case PERF_TYPE_TRACEPOINT:
return tracepoint_id_to_name(config);
default:
break;
}
return "unknown";
}
perf: Enable more compiler warnings Related to a shadowed variable bug fix Valdis Kletnieks noticed that perf does not get built with -Wshadow, which could have helped us avoid the bug. So enable -Wshadow and also enable the following warnings on perf builds, in addition to the already enabled -Wall -Wextra -std=gnu99 warnings: -Wcast-align -Wformat=2 -Wshadow -Winit-self -Wpacked -Wredundant-decls -Wstack-protector -Wstrict-aliasing=3 -Wswitch-default -Wswitch-enum -Wno-system-headers -Wundef -Wvolatile-register-var -Wwrite-strings -Wbad-function-cast -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wold-style-definition -Wstrict-prototypes -Wdeclaration-after-statement And change/fix the perf code to build cleanly under GCC 4.3.2. The list of warnings enablement is rather arbitrary: it's based on my (quick) reading of the GCC manpages and trying them on perf. I categorized the warnings based on individually enabling them and looking whether they trigger something in the perf build. If i liked those warnings (i.e. if they trigger for something that arguably could be improved) i enabled the warning. If the warnings seemed to come from language laywers spamming the build with tons of nuisance warnings i generally kept them off. Most of the sign conversion related warnings were in this category. (A second patch enabling some of the sign warnings might be welcome - sign bugs can be nasty.) I also kept warnings that seem to make sense from their manpage description and which produced no actual warnings on our code base. These warnings might still be turned off if they end up being a nuisance. I also left out a few warnings that are not supported in older compilers. [ Note that these changes might break the build on older compilers i did not test, or on non-x86 architectures that produce different warnings, so more testing would be welcome. ] Reported-by: Valdis.Kletnieks@vt.edu Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <new-submission> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-08-15 17:26:57 +07:00
static int parse_aliases(const char **str, const char *names[][MAX_ALIASES], int size)
{
int i, j;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
int n, longest = -1;
for (i = 0; i < size; i++) {
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
for (j = 0; j < MAX_ALIASES && names[i][j]; j++) {
n = strlen(names[i][j]);
if (n > longest && !strncasecmp(*str, names[i][j], n))
longest = n;
}
if (longest > 0) {
*str += longest;
return i;
}
}
return -1;
}
static int
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
parse_generic_hw_event(const char **str, struct perf_counter_attr *attr)
{
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
const char *s = *str;
int cache_type = -1, cache_op = -1, cache_result = -1;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
cache_type = parse_aliases(&s, hw_cache, PERF_COUNT_HW_CACHE_MAX);
/*
* No fallback - if we cannot get a clear cache type
* then bail out:
*/
if (cache_type == -1)
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
return 0;
while ((cache_op == -1 || cache_result == -1) && *s == '-') {
++s;
if (cache_op == -1) {
cache_op = parse_aliases(&s, hw_cache_op,
PERF_COUNT_HW_CACHE_OP_MAX);
if (cache_op >= 0) {
if (!is_cache_op_valid(cache_type, cache_op))
return 0;
continue;
}
}
if (cache_result == -1) {
cache_result = parse_aliases(&s, hw_cache_result,
PERF_COUNT_HW_CACHE_RESULT_MAX);
if (cache_result >= 0)
continue;
}
/*
* Can't parse this as a cache op or result, so back up
* to the '-'.
*/
--s;
break;
}
/*
* Fall back to reads:
*/
if (cache_op == -1)
cache_op = PERF_COUNT_HW_CACHE_OP_READ;
/*
* Fall back to accesses:
*/
if (cache_result == -1)
cache_result = PERF_COUNT_HW_CACHE_RESULT_ACCESS;
attr->config = cache_type | (cache_op << 8) | (cache_result << 16);
attr->type = PERF_TYPE_HW_CACHE;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
*str = s;
return 1;
}
static int parse_tracepoint_event(const char **strp,
struct perf_counter_attr *attr)
{
const char *evt_name;
char *flags;
char sys_name[MAX_EVENT_LENGTH];
char id_buf[4];
int fd;
unsigned int sys_length, evt_length;
u64 id;
char evt_path[MAXPATHLEN];
if (valid_debugfs_mount(debugfs_path))
return 0;
evt_name = strchr(*strp, ':');
if (!evt_name)
return 0;
sys_length = evt_name - *strp;
if (sys_length >= MAX_EVENT_LENGTH)
return 0;
strncpy(sys_name, *strp, sys_length);
sys_name[sys_length] = '\0';
evt_name = evt_name + 1;
flags = strchr(evt_name, ':');
if (flags) {
*flags = '\0';
flags++;
if (!strncmp(flags, "record", strlen(flags)))
attr->sample_type |= PERF_SAMPLE_RAW;
}
evt_length = strlen(evt_name);
if (evt_length >= MAX_EVENT_LENGTH)
return 0;
snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path,
sys_name, evt_name);
fd = open(evt_path, O_RDONLY);
if (fd < 0)
return 0;
if (read(fd, id_buf, sizeof(id_buf)) < 0) {
close(fd);
return 0;
}
close(fd);
id = atoll(id_buf);
attr->config = id;
attr->type = PERF_TYPE_TRACEPOINT;
*strp = evt_name + evt_length;
return 1;
}
static int check_events(const char *str, unsigned int i)
{
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
int n;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
n = strlen(event_symbols[i].symbol);
if (!strncmp(str, event_symbols[i].symbol, n))
return n;
n = strlen(event_symbols[i].alias);
if (n)
if (!strncmp(str, event_symbols[i].alias, n))
return n;
return 0;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
static int
parse_symbolic_event(const char **strp, struct perf_counter_attr *attr)
{
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
const char *str = *strp;
unsigned int i;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
int n;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
for (i = 0; i < ARRAY_SIZE(event_symbols); i++) {
n = check_events(str, i);
if (n > 0) {
attr->type = event_symbols[i].type;
attr->config = event_symbols[i].config;
*strp = str + n;
return 1;
}
}
return 0;
}
static int parse_raw_event(const char **strp, struct perf_counter_attr *attr)
{
const char *str = *strp;
u64 config;
int n;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
if (*str != 'r')
return 0;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
n = hex2u64(str + 1, &config);
if (n > 0) {
*strp = str + n + 1;
attr->type = PERF_TYPE_RAW;
attr->config = config;
return 1;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
return 0;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
static int
parse_numeric_event(const char **strp, struct perf_counter_attr *attr)
{
const char *str = *strp;
char *endp;
unsigned long type;
u64 config;
type = strtoul(str, &endp, 0);
if (endp > str && type < PERF_TYPE_MAX && *endp == ':') {
str = endp + 1;
config = strtoul(str, &endp, 0);
if (endp > str) {
attr->type = type;
attr->config = config;
*strp = endp;
return 1;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
}
return 0;
}
static int
parse_event_modifier(const char **strp, struct perf_counter_attr *attr)
{
const char *str = *strp;
int eu = 1, ek = 1, eh = 1;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
if (*str++ != ':')
return 0;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
while (*str) {
if (*str == 'u')
eu = 0;
else if (*str == 'k')
ek = 0;
else if (*str == 'h')
eh = 0;
else
break;
++str;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
if (str >= *strp + 2) {
*strp = str;
attr->exclude_user = eu;
attr->exclude_kernel = ek;
attr->exclude_hv = eh;
return 1;
}
return 0;
}
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
/*
* Each event can have multiple symbolic names.
* Symbolic names are (almost) exactly matched.
*/
static int parse_event_symbols(const char **str, struct perf_counter_attr *attr)
{
if (!(parse_tracepoint_event(str, attr) ||
parse_raw_event(str, attr) ||
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
parse_numeric_event(str, attr) ||
parse_symbolic_event(str, attr) ||
parse_generic_hw_event(str, attr)))
return 0;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
parse_event_modifier(str, attr);
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
return 1;
}
int parse_events(const struct option *opt __used, const char *str, int unset __used)
{
struct perf_counter_attr attr;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
for (;;) {
if (nr_counters == MAX_COUNTERS)
return -1;
memset(&attr, 0, sizeof(attr));
if (!parse_event_symbols(&str, &attr))
return -1;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
if (!(*str == 0 || *str == ',' || isspace(*str)))
return -1;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
attrs[nr_counters] = attr;
nr_counters++;
perf_counter tools: Rework event string parsing/syntax This reworks the parser for event descriptors to make it more consistent in what it accepts. It is now structured as a recursive descent parser for the following grammar: events ::= event ( ("," | space) space* event )* event ::= ( raw_event | numeric_event | symbolic_event | generic_hw_event ) [ event_modifier ] raw_event ::= "r" hex_number numeric_event ::= number ":" number number ::= decimal_number | "0x" hex_number | "0" octal_number symbolic_event ::= string_from_event_symbols_array generic_hw_event::= cache_type ( "-" ( cache_op | cache_result ) )* event_modifier ::= ":" ( "u" | "k" | "h" )+ with the extra restriction that you can have at most one cache_op and at most one cache_result. We pass the current string pointer by reference (i.e. as a const char **) to the various parsing functions so that they can advance the pointer to indicate how much they consumed. They return 0 if they didn't recognize the thing at the pointer or 1 if they did (and advance the pointer past it). This also fixes parse_aliases to take the longest matching alias from the table, not the first one. Otherwise "l1-data" would match the "l1-d" alias and the "ata" would not be consumed. This allows event modifiers indicating what processor modes to count in to be applied to any event, not just numeric events, and adds a ":h" modifier to indicate counting in hypervisor mode. Specifying ":u" now sets both exclude_kernel and exclude_hv, and so on. Multiple modes can be specified, e.g. ":uk" will count in user or hypervisor mode (i.e. only exclude_kernel will be set). Signed-off-by: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <19018.53826.843815.189847@cargo.ozlabs.ibm.com> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 10:04:34 +07:00
if (*str == 0)
break;
if (*str == ',')
++str;
while (isspace(*str))
++str;
}
return 0;
}
static const char * const event_type_descriptors[] = {
"",
"Hardware event",
"Software event",
"Tracepoint event",
"Hardware cache event",
};
/*
* Print the events from <debugfs_mount_point>/tracing/events
*/
static void print_tracepoint_events(void)
{
DIR *sys_dir, *evt_dir;
struct dirent *sys_next, *evt_next, sys_dirent, evt_dirent;
struct stat st;
char evt_path[MAXPATHLEN];
if (valid_debugfs_mount(debugfs_path))
return;
sys_dir = opendir(debugfs_path);
if (!sys_dir)
goto cleanup;
for_each_subsystem(sys_dir, sys_dirent, sys_next, evt_path, st) {
evt_dir = opendir(evt_path);
if (!evt_dir)
goto cleanup;
for_each_event(sys_dirent, evt_dir, evt_dirent, evt_next,
evt_path, st) {
snprintf(evt_path, MAXPATHLEN, "%s:%s",
sys_dirent.d_name, evt_dirent.d_name);
fprintf(stderr, " %-40s [%s]\n", evt_path,
event_type_descriptors[PERF_TYPE_TRACEPOINT+1]);
}
closedir(evt_dir);
}
cleanup:
closedir(sys_dir);
}
/*
* Print the help text for the event symbols:
*/
void print_events(void)
{
struct event_symbol *syms = event_symbols;
perf list: Add cache events After: $ ./perf list List of pre-defined events (to be used in -e): cpu-cycles OR cycles [Hardware event] instructions [Hardware event] cache-references [Hardware event] cache-misses [Hardware event] branch-instructions OR branches [Hardware event] branch-misses [Hardware event] bus-cycles [Hardware event] cpu-clock [Software event] task-clock [Software event] page-faults OR faults [Software event] minor-faults [Software event] major-faults [Software event] context-switches OR cs [Software event] cpu-migrations OR migrations [Software event] L1-d$-loads [Hardware cache event] L1-d$-load-misses [Hardware cache event] L1-d$-stores [Hardware cache event] L1-d$-store-misses [Hardware cache event] L1-d$-prefetches [Hardware cache event] L1-d$-prefetch-misses [Hardware cache event] L1-i$-loads [Hardware cache event] L1-i$-load-misses [Hardware cache event] L1-i$-prefetches [Hardware cache event] L1-i$-prefetch-misses [Hardware cache event] LLC-loads [Hardware cache event] LLC-load-misses [Hardware cache event] LLC-stores [Hardware cache event] LLC-store-misses [Hardware cache event] LLC-prefetches [Hardware cache event] LLC-prefetch-misses [Hardware cache event] dTLB-loads [Hardware cache event] dTLB-load-misses [Hardware cache event] dTLB-stores [Hardware cache event] dTLB-store-misses [Hardware cache event] dTLB-prefetches [Hardware cache event] dTLB-prefetch-misses [Hardware cache event] iTLB-loads [Hardware cache event] iTLB-load-misses [Hardware cache event] branch-loads [Hardware cache event] branch-load-misses [Hardware cache event] rNNN [raw hardware event descriptor] Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <1246453578.3072.1.camel@ht.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 20:06:18 +07:00
unsigned int i, type, op, prev_type = -1;
char name[40];
fprintf(stderr, "\n");
fprintf(stderr, "List of pre-defined events (to be used in -e):\n");
for (i = 0; i < ARRAY_SIZE(event_symbols); i++, syms++) {
type = syms->type + 1;
if (type >= ARRAY_SIZE(event_type_descriptors))
type = 0;
if (type != prev_type)
fprintf(stderr, "\n");
if (strlen(syms->alias))
sprintf(name, "%s OR %s", syms->symbol, syms->alias);
else
strcpy(name, syms->symbol);
fprintf(stderr, " %-40s [%s]\n", name,
event_type_descriptors[type]);
prev_type = type;
}
perf list: Add cache events After: $ ./perf list List of pre-defined events (to be used in -e): cpu-cycles OR cycles [Hardware event] instructions [Hardware event] cache-references [Hardware event] cache-misses [Hardware event] branch-instructions OR branches [Hardware event] branch-misses [Hardware event] bus-cycles [Hardware event] cpu-clock [Software event] task-clock [Software event] page-faults OR faults [Software event] minor-faults [Software event] major-faults [Software event] context-switches OR cs [Software event] cpu-migrations OR migrations [Software event] L1-d$-loads [Hardware cache event] L1-d$-load-misses [Hardware cache event] L1-d$-stores [Hardware cache event] L1-d$-store-misses [Hardware cache event] L1-d$-prefetches [Hardware cache event] L1-d$-prefetch-misses [Hardware cache event] L1-i$-loads [Hardware cache event] L1-i$-load-misses [Hardware cache event] L1-i$-prefetches [Hardware cache event] L1-i$-prefetch-misses [Hardware cache event] LLC-loads [Hardware cache event] LLC-load-misses [Hardware cache event] LLC-stores [Hardware cache event] LLC-store-misses [Hardware cache event] LLC-prefetches [Hardware cache event] LLC-prefetch-misses [Hardware cache event] dTLB-loads [Hardware cache event] dTLB-load-misses [Hardware cache event] dTLB-stores [Hardware cache event] dTLB-store-misses [Hardware cache event] dTLB-prefetches [Hardware cache event] dTLB-prefetch-misses [Hardware cache event] iTLB-loads [Hardware cache event] iTLB-load-misses [Hardware cache event] branch-loads [Hardware cache event] branch-load-misses [Hardware cache event] rNNN [raw hardware event descriptor] Signed-off-by: Jaswinder Singh Rajput <jaswinderrajput@gmail.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@redhat.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> LKML-Reference: <1246453578.3072.1.camel@ht.satnam> Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-07-01 20:06:18 +07:00
fprintf(stderr, "\n");
for (type = 0; type < PERF_COUNT_HW_CACHE_MAX; type++) {
for (op = 0; op < PERF_COUNT_HW_CACHE_OP_MAX; op++) {
/* skip invalid cache type */
if (!is_cache_op_valid(type, op))
continue;
for (i = 0; i < PERF_COUNT_HW_CACHE_RESULT_MAX; i++) {
fprintf(stderr, " %-40s [%s]\n",
event_cache_name(type, op, i),
event_type_descriptors[4]);
}
}
}
fprintf(stderr, "\n");
fprintf(stderr, " %-40s [raw hardware event descriptor]\n",
"rNNN");
fprintf(stderr, "\n");
print_tracepoint_events();
exit(129);
}