perf tools: add support for libpfm4
Subsystems (4)
Ian Rogers
reviewerPATCH v7 1/1 perf tools: add support for libpfm4
From: Stephane Eranian <[email protected]>
This patch links perf with the libpfm4 library if it is available and NO_LIBPFM4 isn't passed to the build. The libpfm4 library contains hardware event tables for all processors supported by perf_events. It is a helper library that helps convert from a symbolic event name to the event encoding required by the underlying kernel interface. This library is open-source and available from: http://perfmon2.sf.net.
With this patch, it is possible to specify full hardware events by name. Hardware filters are also supported. Events must be specified via the --pfm-events and not -e option. Both options are active at the same time and it is possible to mix and match:
$ perf stat --pfm-events inst_retired:any_p:c=1:i -e cycles ....
v7 rebases and adds fallback code for libpfm4 events.
The fallback code is to force user only priv level in case the perf_event_open() syscall failed for permissions reason.
the fallback forces a user privilege level restriction on the event string, so depending on the syntax either u or :u is needed.
But libpfm4 can use a : or . as the separator, so simply searching for ':' vs. '/' is not good enough to determine the syntax needed. Therefore, this patch introduces a new evsel boolean field to mark events coming from libpfm4. The field is then used to adjust the fallback string.
v6 is a rebase.
v5 is a rebase.
v4 is a rebase on git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git
branch perf/core and re-adds the tools/build/feature/test-libpfm4.c missed in v3.
v3 is against acme/perf/core and removes a diagnostic warning.
v2 of this patch makes the --pfm-events man page documentation conditional on libpfm4 behing configured. It tidies some of the documentation and adds the feature test missed in the v1 patch.
--- tools/build/Makefile.feature | 6 +- tools/build/feature/Makefile | 6 +- tools/build/feature/test-libpfm4.c | 8 + tools/perf/Documentation/Makefile | 4 +- tools/perf/Documentation/perf-record.txt | 11 + tools/perf/Documentation/perf-stat.txt | 10 + tools/perf/Documentation/perf-top.txt | 11 + tools/perf/Makefile.config | 12 ++ tools/perf/Makefile.perf | 6 +- tools/perf/builtin-list.c | 16 ++ tools/perf/builtin-record.c | 20 ++ tools/perf/builtin-stat.c | 21 ++ tools/perf/builtin-top.c | 20 ++ tools/perf/util/evsel.c | 6 + tools/perf/util/evsel.h | 3 + tools/perf/util/parse-events.c | 247 +++++++++++++++++++++++ tools/perf/util/parse-events.h | 5 + tools/perf/util/pmu.c | 11 + tools/perf/util/pmu.h | 1 + 19 files changed, 417 insertions(+), 7 deletions(-) create mode 100644 tools/build/feature/test-libpfm4.c
tools/build/Makefile.feature +4 -2
@@ -73,7 +73,8 @@ FEATURE_TESTS_BASIC := \ libaio \ libzstd \ disassembler-four-args \ - file-handle + file-handle \ + libpfm4 # FEATURE_TESTS_BASIC + FEATURE_TESTS_EXTRA is the complete list # of all feature tests
@@ -128,7 +129,8 @@ FEATURE_DISPLAY ?= \ bpf \ libaio \ libzstd \ - disassembler-four-args + disassembler-four-args \ + libpfm4 # Set FEATURE_CHECK_(C|LD)FLAGS-all for all FEATURE_TESTS features. # If in the future we need per-feature checks/flags for features not
tools/build/feature/Makefile +5 -1
@@ -68,7 +68,8 @@ FILES= \ test-llvm-version.bin \ test-libaio.bin \ test-libzstd.bin \ - test-file-handle.bin + test-file-handle.bin \ + test-libpfm4.bin FILES := $(addprefix $(OUTPUT),$(FILES))
@@ -325,6 +326,9 @@ $(OUTPUT)test-libzstd.bin: $(OUTPUT)test-file-handle.bin: $(BUILD) +$(OUTPUT)test-libpfm4.bin: + $(BUILD) -lpfm + ############################### clean:
tools/build/feature/test-libpfm4.c new file +8 -0
@@ -0,0 +1,8 @@ +#include <sys/types.h> +#include <perfmon/pfmlib.h> + +int main (void ) +{ +pfm_initialize (); +return 0 ; +}
tools/perf/Documentation/Makefile +2 -2
@@ -48,7 +48,7 @@ man5dir=$(mandir)/man5 man7dir=$(mandir)/man7 ASCIIDOC=asciidoc -ASCIIDOC_EXTRA = --unsafe -f asciidoc.conf +ASCIIDOC_EXTRA += --unsafe -f asciidoc.conf ASCIIDOC_HTML = xhtml11 MANPAGE_XSL = manpage-normal.xsl XMLTO_EXTRA =
@@ -59,7 +59,7 @@ HTML_REF = origin/html ifdef USE_ASCIIDOCTOR ASCIIDOC = asciidoctor -ASCIIDOC_EXTRA = -a compat-mode +ASCIIDOC_EXTRA += -a compat-mode ASCIIDOC_EXTRA += -I. -rasciidoctor-extensions ASCIIDOC_EXTRA += -a mansource="perf" -a manmanual="perf Manual" ASCIIDOC_HTML = xhtml5
tools/perf/Documentation/perf-record.txt +11 -0
@@ -596,6 +596,17 @@ Make a copy of /proc/kcore and place it into a directory with the perf data file Limit the sample data max size, <size> is expected to be a number with appended unit character - B/K/M/G +ifdef::HAVE_LIBPFM[] +--pfm-events events:: +Select a PMU event using libpfm4 syntax (see http://perfmon2.sf.net) +including support for event filters. For example '--pfm-events +inst_retired:any_p:u:c=1:i'. More than one event can be passed to the +option using the comma separator. Hardware events and generic hardware +events cannot be mixed together. The latter must be used with the -e +option. The -e option and this one can be mixed and matched. Events +can be grouped using the {} notation. +endif::HAVE_LIBPFM[] + SEE ALSO -------- linkperf:perf-stat[1], linkperf:perf-list[1], linkperf:perf-intel-pt[1]
tools/perf/Documentation/perf-stat.txt +10 -0
@@ -71,6 +71,16 @@ report:: --tid=<tid>:: stat events on existing thread id (comma separated list) +ifdef::HAVE_LIBPFM[] +--pfm-events events:: +Select a PMU event using libpfm4 syntax (see http://perfmon2.sf.net) +including support for event filters. For example '--pfm-events +inst_retired:any_p:u:c=1:i'. More than one event can be passed to the +option using the comma separator. Hardware events and generic hardware +events cannot be mixed together. The latter must be used with the -e +option. The -e option and this one can be mixed and matched. Events +can be grouped using the {} notation. +endif::HAVE_LIBPFM[] -a:: --all-cpus::
tools/perf/Documentation/perf-top.txt +11 -0
@@ -319,6 +319,17 @@ Default is to monitor all CPUS. go straight to the histogram browser, just like 'perf top' with no events explicitely specified does. +ifdef::HAVE_LIBPFM[] +--pfm-events events:: +Select a PMU event using libpfm4 syntax (see http://perfmon2.sf.net) +including support for event filters. For example '--pfm-events +inst_retired:any_p:u:c=1:i'. More than one event can be passed to the +option using the comma separator. Hardware events and generic hardware +events cannot be mixed together. The latter must be used with the -e +option. The -e option and this one can be mixed and matched. Events +can be grouped using the {} notation. +endif::HAVE_LIBPFM[] + INTERACTIVE PROMPTING KEYS --------------------------
tools/perf/Makefile.config +12 -0
@@ -1012,6 +1012,18 @@ ifdef LIBCLANGLLVM endif endif +ifndef NO_LIBPFM4 + ifeq ($(feature-libpfm4), 1) + CFLAGS += -DHAVE_LIBPFM + EXTLIBS += -lpfm + ASCIIDOC_EXTRA = -aHAVE_LIBPFM=1 + $(call detected,CONFIG_LIBPFM4) + else + msg := $(warning libpfm4 not found, disables libpfm4 support. Please install libpfm4-dev); + NO_LIBPFM4 := 1 + endif +endif + # Among the variables below, these: # perfexecdir # perf_include_dir
tools/perf/Makefile.perf +4 -2
@@ -118,6 +118,8 @@ include ../scripts/utilities.mak # # Define LIBBPF_DYNAMIC to enable libbpf dynamic linking. # +# Define NO_LIBPFM4 to disable libpfm4 extension. +# # As per kernel Makefile, avoid funny character set dependencies unexport LC_ALL
@@ -188,7 +190,7 @@ AWK = awk # non-config cases config := 1 -NON_CONFIG_TARGETS := clean python-clean TAGS tags cscope help install-doc install-man install-html install-info install-pdf doc man html info pdf +NON_CONFIG_TARGETS := clean python-clean TAGS tags cscope help ifdef MAKECMDGOALS ifeq ($(filter-out $(NON_CONFIG_TARGETS),$(MAKECMDGOALS)),)
@@ -832,7 +834,7 @@ INSTALL_DOC_TARGETS += quick-install-doc quick-install-man quick-install-html # 'make doc' should call 'make -C Documentation all' $(DOC_TARGETS): - $(Q)$(MAKE) -C $(DOC_DIR) O=$(OUTPUT) $(@:doc=all) + $(Q)$(MAKE) -C $(DOC_DIR) O=$(OUTPUT) $(@:doc=all) ASCIIDOC_EXTRA=$(ASCIIDOC_EXTRA) TAG_FOLDERS= . ../lib ../include TAG_FILES= ../../include/uapi/linux/perf_event.h
tools/perf/builtin-list.c +16 -0
@@ -18,6 +18,10 @@#include <subcmd/parse-options.h> #include <stdio.h> +#ifdef HAVE_LIBPFM +#include <perfmon/pfmlib.h> +#endif +static bool desc_flag = true;static bool details_flag;
@@ -56,6 +60,18 @@ int cmd_list(int argc, const char **argv)if (!raw_dump &&pager_in_use ())printf ("\nList of pre-defined events (to be used in -e):\n\n" ); +#ifdef HAVE_LIBPFM + { +int ret; + ret =pfm_initialize (); +if (ret !=PFM_SUCCESS ) { +fprintf (stderr, +"warning libpfm failed to initialize: %s\n" , +pfm_strerror (ret)); + } + } +#endif +if (argc ==0 ) {print_events (NULL , raw_dump, !desc_flag, long_desc_flag, details_flag, deprecated);
tools/perf/builtin-record.c +20 -0
@@ -64,6 +64,10 @@#include <linux/zalloc.h> #include <linux/bitmap.h> +#ifdef HAVE_LIBPFM +#include <perfmon/pfmlib.h> +#endif +struct switch_output {bool enabled;bool signal;
@@ -2421,6 +2425,11 @@ static struct option __record_options[] = {#endif OPT_CALLBACK(0, "max-size", &record.output_max_size, "size", "Limit the maximum size of the output file", parse_output_max_size), +#ifdef HAVE_LIBPFM +OPT_CALLBACK (0 ,"pfm-events" , &record.evlist,"event" , +"libpfm4 event selector. use 'perf list' to list available events" , + parse_libpfm_events_option), +#endifOPT_END () };
@@ -2461,6 +2470,17 @@ int cmd_record(int argc, const char **argv)if (rec->evlist ==NULL )return -ENOMEM ; +#ifdef HAVE_LIBPFM + { +int ret; + ret =pfm_initialize (); +if (ret !=PFM_SUCCESS ) { +ui__warning ("warning libpfm failed to initialize: %s\n" , +pfm_strerror (ret)); + } + } +#endif + err =perf_config (perf_record_config, rec);if (err)return err;
tools/perf/builtin-stat.c +21 -0
@@ -89,6 +89,10 @@#include <linux/ctype.h> #include <perf/evlist.h> +#ifdef HAVE_LIBPFM +#include <perfmon/pfmlib.h> +#endif +#define DEFAULT_SEPARATOR " "#define FREEZE_ON_SMI_PATH "devices/cpu/freeze_on_smi"
@@ -933,6 +937,11 @@ static struct option stat_options[] = { "Use with 'percore' event qualifier to show the event " "counts of one hardware thread by sum up total hardware " "threads of same physical core"), +#ifdef HAVE_LIBPFM +OPT_CALLBACK (0 ,"pfm-events" , &evsel_list,"event" , +"libpfm4 event selector. use 'perf list' to list available events" , + parse_libpfm_events_option), +#endifOPT_END () };
@@ -1871,6 +1880,18 @@ int cmd_stat(int argc, const char **argv)unsigned int interval, timeout;const char *const stat_subcommands[] = {"record" ,"report" }; +#ifdef HAVE_LIBPFM + { +int ret; + ret =pfm_initialize (); +if (ret !=PFM_SUCCESS ) { +fprintf (stderr, +"warning libpfm failed to initialize: %s\n" , +pfm_strerror (ret)); + } + } +#endif +setlocale (LC_ALL ,"" ); evsel_list =evlist__new ();
tools/perf/builtin-top.c +20 -0
@@ -84,6 +84,10 @@#include <linux/ctype.h> #include <perf/mmap.h> +#ifdef HAVE_LIBPFM +#include <perfmon/pfmlib.h> +#endif +static volatile int done;static volatile int resize;
@@ -1571,6 +1575,11 @@ int cmd_top(int argc, const char **argv) "Sort the output by the event at the index n in group. " "If n is invalid, sort by the first event. " "WARNING: should be used on grouped events."), +#ifdef HAVE_LIBPFM +OPT_CALLBACK (0 ,"pfm-events" , &top.evlist,"event" , +"libpfm4 event selector. use 'perf list' to list available events" , + parse_libpfm_events_option), +#endifOPTS_EVSWITCH (&top.evswitch),OPT_END () };
@@ -1584,6 +1593,17 @@ int cmd_top(int argc, const char **argv)if (status <0 )return status; +#ifdef HAVE_LIBPFM + { +int ret; + ret =pfm_initialize (); +if (ret !=PFM_SUCCESS ) { +ui__warning ("warning libpfm failed to initialize: %s\n" , +pfm_strerror (ret)); + } + } +#endif + top.annotation_opts.min_pcnt =5 ; top.annotation_opts.context =4 ;
tools/perf/util/evsel.c +6 -0
@@ -2447,9 +2447,15 @@ bool perf_evsel__fallback(struct evsel *evsel, int err,const char *sep =":" ;/* Is there already the separator in the name. */ +#ifndef HAVE_LIBPFM if (strchr (name,'/' ) ||strchr (name,':' )) sep ="" ; +#else +if (strchr (name,'/' ) || + (strchr (name,':' ) && !evsel->is_libpfm_event)) + sep ="" ; +#endif if (asprintf (&new_name,"%s%su" , name, sep) <0 )return false;
tools/perf/util/evsel.h +3 -0
@@ -76,6 +76,9 @@ struct evsel {bool ignore_missing_thread;bool forced_leader;bool use_uncore_alias; +#ifdef HAVE_LIBPFM +bool is_libpfm_event; +#endif /* parse modifier helper */ int exclude_GH;int sample_read;
tools/perf/util/parse-events.c +247 -0
262 lines not on this page · raw message
tools/perf/util/parse-events.h +5 -0
@@ -37,6 +37,11 @@ int parse_events_terms(struct list_head *terms, const char *str);int parse_filter (const struct option *opt,const char *str,int unset);int exclude_perf (const struct option *opt,const char *arg,int unset); +#ifdef HAVE_LIBPFM +extern int parse_libpfm_events_option (const struct option *opt,const char *str, +int unset); +#endif +#define EVENTS_HELP_MAX (128*1024)enum perf_pmu_event_symbol_type {
tools/perf/util/pmu.c +11 -0
@@ -869,6 +869,17 @@ static struct perf_pmu *pmu_find(const char *name)return NULL ; } +struct perf_pmu *perf_pmu__find_by_type (unsigned int type) +{ +struct perf_pmu *pmu; + +list_for_each_entry (pmu, &pmus, list) +if (pmu->type == type) +return pmu; + +return NULL ; +} +struct perf_pmu *perf_pmu__scan (struct perf_pmu *pmu) {/*
tools/perf/util/pmu.h +1 -0
@@ -65,6 +65,7 @@ struct perf_pmu_alias { };struct perf_pmu *perf_pmu__find (const char *name); +struct perf_pmu *perf_pmu__find_by_type (unsigned int type);int perf_pmu__config (struct perf_pmu *pmu,struct perf_event_attr *attr,struct list_head *head_terms,struct parse_events_error *error);
Signature · 1 line
2.26.0.292.g33ef6b2f38-goog
Quoted 22 lines · Ian Rogers
From: Stephane Eranian <[email protected]>
This patch links perf with the libpfm4 library if it is available and
NO_LIBPFM4isn't passed to the build. The libpfm4 library contains hardware event tables for all processors supported byperf_events. It is a helper library that helps convert from a symbolic event name to the event encoding required by the underlying kernel interface. This library is open-source and available from: http://perfmon2.sf.net.With this patch, it is possible to specify full hardware events by name. Hardware filters are also supported. Events must be specified via the --pfm-events and not -e option. Both options are active at the same time and it is possible to mix and match:
$ perf stat --pfm-events inst_retired:any_p:c=1:i -e cycles ....v7 rebases and adds fallback code for libpfm4 events.
The fallback code is to force user only priv level in case theperf_event_open()syscall failed for permissions reason.
the fallback forces a user privilege level restriction on the event string, so depending on the syntax either u or :u is needed.
Quoted 4 lines
But libpfm4 can use a : or . as the separator, so simply searching for ':' vs. '/' is not good enough to determine the syntax needed. Therefore, this patch introduces a new evsel boolean field to mark events coming from libpfm4. The field is then used to adjust the fallback string.
heya,
I made bunch of comments for v5, not sure you saw them:
https://lore.kernel.org/lkml/[email protected]/
jirka
Quoted 10 lines
v6 is a rebase.
v5 is a rebase.
v4 is a rebase ongit://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git
branch perf/core and re-adds thetools/build/feature/test-libpfm4.cmissed in v3.
v3 is against acme/perf/core and removes a diagnostic warning.
v2 of this patch makes the --pfm-events man page documentation conditional on libpfm4 behing configured. It tidies some of the documentation and adds the feature test missed in the v1 patch.
SNIP
Ian Rogers
reviewerQuoted 29 lines · Jiri Olsa
Quoted 22 lines · Ian Rogers
From: Stephane Eranian <[email protected]>
This patch links perf with the libpfm4 library if it is available and
NO_LIBPFM4isn't passed to the build. The libpfm4 library contains hardware event tables for all processors supported byperf_events. It is a helper library that helps convert from a symbolic event name to the event encoding required by the underlying kernel interface. This library is open-source and available from: http://perfmon2.sf.net.With this patch, it is possible to specify full hardware events by name. Hardware filters are also supported. Events must be specified via the --pfm-events and not -e option. Both options are active at the same time and it is possible to mix and match:
$ perf stat --pfm-events inst_retired:any_p:c=1:i -e cycles ....v7 rebases and adds fallback code for libpfm4 events.
The fallback code is to force user only priv level in case theperf_event_open()syscall failed for permissions reason.
the fallback forces a user privilege level restriction on the event string, so depending on the syntax either u or :u is needed.Quoted 4 lines
But libpfm4 can use a : or . as the separator, so simply searching for ':' vs. '/' is not good enough to determine the syntax needed. Therefore, this patch introduces a new evsel boolean field to mark events coming from libpfm4. The field is then used to adjust the fallback string.
Quoted 5 lines
heya,
I made bunch of comments for v5, not sure you saw them:
https://lore.kernel.org/lkml/[email protected]/jirka
Sorry for missing this, I will work on fixing these and thanks!
Ian
Quoted 13 lines
Quoted 10 lines
v6 is a rebase.
v5 is a rebase.
v4 is a rebase ongit://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git
branch perf/core and re-adds thetools/build/feature/test-libpfm4.cmissed in v3.
v3 is against acme/perf/core and removes a diagnostic warning.
v2 of this patch makes the --pfm-events man page documentation conditional on libpfm4 behing configured. It tidies some of the documentation and adds the feature test missed in the v1 patch.SNIP
Ian Rogers
reviewerQuoted 35 lines · Ian Rogers
Quoted 33 lines · Jiri Olsa
Quoted 22 lines · Ian Rogers
From: Stephane Eranian <[email protected]>
This patch links perf with the libpfm4 library if it is available and
NO_LIBPFM4isn't passed to the build. The libpfm4 library contains hardware event tables for all processors supported byperf_events. It is a helper library that helps convert from a symbolic event name to the event encoding required by the underlying kernel interface. This library is open-source and available from: http://perfmon2.sf.net.With this patch, it is possible to specify full hardware events by name. Hardware filters are also supported. Events must be specified via the --pfm-events and not -e option. Both options are active at the same time and it is possible to mix and match:
$ perf stat --pfm-events inst_retired:any_p:c=1:i -e cycles ....v7 rebases and adds fallback code for libpfm4 events.
The fallback code is to force user only priv level in case theperf_event_open()syscall failed for permissions reason.
the fallback forces a user privilege level restriction on the event string, so depending on the syntax either u or :u is needed.Quoted 4 lines
But libpfm4 can use a : or . as the separator, so simply searching for ':' vs. '/' is not good enough to determine the syntax needed. Therefore, this patch introduces a new evsel boolean field to mark events coming from libpfm4. The field is then used to adjust the fallback string.
heya,
I made bunch of comments for v5, not sure you saw them:
https://lore.kernel.org/lkml/[email protected]/
Quoted 5 lines
Quoted 1 line
jirka
Sorry for missing this, I will work on fixing these and thanks!
Ian
v8 is available now:
https://lore.kernel.org/lkml/[email protected]/T/#t
Thanks,
Ian
Quoted 13 lines
Quoted 13 lines
Quoted 10 lines
v6 is a rebase.
v5 is a rebase.
v4 is a rebase ongit://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git
branch perf/core and re-adds thetools/build/feature/test-libpfm4.cmissed in v3.
v3 is against acme/perf/core and removes a diagnostic warning.
v2 of this patch makes the --pfm-events man page documentation conditional on libpfm4 behing configured. It tidies some of the documentation and adds the feature test missed in the v1 patch.SNIP