perf_events: add cgroup support to perf tool (v7)
PATCH 5/5 perf_events: add cgroup support to perf tool (v7)
This patch adds the ability to filter monitoring based on container groups (cgroups) for both perf stat and perf record. It is possible to monitor multiple cgroup in parallel. There is one cgroup per event. The cgroups to monitor are passed via a new -G option followed by a comma separated list of cgroup names.
The cgroup filesystem has to be mounted. Given a cgroup name, the perf tool finds the corresponding directory in the cgroup filesystem and opens it. It then passes that file descriptor to the kernel.
Example:
$ perf stat -B -a -e cycles:u,cycles:u,cycles:u -G test1,,test2 -- sleep 1 Performance counter stats for 'sleep 1':
2,368,667,414 cycles test1
2,369,661,459 cycles
<not counted> cycles test21.001856890 seconds time elapsed
Signed-off-by Stephane Eranian <[email protected]>
---
tools/perf/Documentation/perf-record.txt +8 -0
@@ -134,6 +134,14 @@ Do not update the builid cache. This saves some overhead in situations where the information in the perf.data file (which includes buildids) is sufficient. +-G name,...:: +--cgroup name,...:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. + SEE ALSO -------- linkperf:perf-stat[1], linkperf:perf-list[1]
tools/perf/Documentation/perf-stat.txt +8 -0
@@ -83,6 +83,14 @@ This option is only valid in system-wide mode. print counts using a CSV-style output to make it easy to import directly into spreadsheets. Columns are separated by the string specified in SEP. +-G name:: +--cgroup name:: +monitor only in the container (cgroup) called "name". This option is available only +in per-cpu mode. The cgroup filesystem must be mounted. All threads belonging to +container "name" are monitored when they run on the monitored CPUs. Multiple cgroups +can be provided. Each cgroup is applied to the corresponding event, i.e., first cgroup +to first event, second cgroup to second event and so on. + EXAMPLES --------
tools/perf/Makefile +2 -0
@@ -424,6 +424,7 @@ LIB_H += util/probe-event.h LIB_H += util/pstack.h LIB_H += util/cpumap.h LIB_H += $(ARCH_INCLUDE) +LIB_H += util/cgroup.h LIB_OBJS += $(OUTPUT)util/abspath.o LIB_OBJS += $(OUTPUT)util/alias.o
@@ -471,6 +472,7 @@ LIB_OBJS += $(OUTPUT)util/hist.o LIB_OBJS += $(OUTPUT)util/probe-event.o LIB_OBJS += $(OUTPUT)util/util.o LIB_OBJS += $(OUTPUT)util/cpumap.o +LIB_OBJS += $(OUTPUT)util/cgroup.o BUILTIN_OBJS += $(OUTPUT)builtin-annotate.o
tools/perf/builtin-record.c +26 -1
@@ -22,6 +22,7 @@#include "util/session.h" #include "util/symbol.h" #include "util/cpumap.h" +#include "util/cgroup.h" #include <unistd.h> #include <sched.h>
@@ -234,6 +235,8 @@ static void create_counter(int counter, int cpu)char *filter = filters[counter];struct perf_event_attr *attr = attrs + counter;struct perf_header_attr *h_attr; +unsigned long flags =0 ; +int pid;int track = !counter;/* only the first counter needs these */ int thread_index;int ret;
@@ -308,6 +311,9 @@ static void create_counter(int counter, int cpu) attr->sample_type |=PERF_SAMPLE_CPU ; } +if (cgroups[counter]) + flags =PERF_FLAG_PID_CGROUP ; + attr->mmap = track; attr->comm = track; attr->inherit = !no_inherit;
@@ -320,8 +326,13 @@ retry_sample_id:for (thread_index =0 ; thread_index < thread_num; thread_index++) { try_again: +if (cgroups[counter]) + pid = cgroups_fd[counter]; +else + pid = all_tids[thread_index]; + fd[nr_cpu][counter][thread_index] =sys_perf_event_open (attr, - all_tids[thread_index], cpu, group_fd,0 ); + pid, cpu, group_fd, flags);if (fd[nr_cpu][counter][thread_index] <0 ) {int err = errno;
@@ -882,6 +893,9 @@ const struct option record_options[] = { "do not update the buildid cache"),OPT_BOOLEAN ('B' ,"no-buildid" , &no_buildid,"do not collect buildids in perf.data" ), +OPT_CALLBACK ('G' ,"cgroup" ,NULL ,"name" , +"monitor in cgroup name only" , + parse_cgroups), OPT_END() };
@@ -905,6 +919,12 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) write_mode =WRITE_FORCE ; } +if (nr_cgroups && !system_wide) { +fprintf (stderr,"cgroup monitoring only available in " +" system-wide mode\n" ); +usage_with_options (record_usage, record_options); + } +symbol__init ();if (no_buildid_cache || no_buildid)
@@ -916,6 +936,9 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) attrs[0 ].config =PERF_COUNT_HW_CPU_CYCLES ; } +if (open_cgroups ()) +usage_with_options (record_usage, record_options); +if (target_pid !=-1 ) { target_tid = target_pid; thread_num =find_all_tid (target_pid, &all_tids);
@@ -925,6 +948,7 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)usage_with_options (record_usage, record_options); } }else { + err = -ENOMEM ; all_tids=malloc (sizeof (pid_t));if (!all_tids) goto out_symbol_exit;
@@ -976,5 +1000,6 @@ out_free_fd: all_tids =NULL ; out_symbol_exit:symbol__exit (); +close_cgroups ();return err; }
tools/perf/builtin-stat.c +54 -8
154 lines not on this page · raw message
tools/perf/util/cgroup.c new file +123 -0
123 lines not on this page · raw message
tools/perf/util/cgroup.h new file +14 -0
@@ -0,0 +1,14 @@ +#ifndef __CGROUP_H__ +#define __CGROUP_H__ + +struct option ; + +extern char *cgroups[MAX_COUNTERS ]; +extern int cgroups_fd[MAX_COUNTERS ]; +extern int nr_cgroups;/* number of explicit cgroups defined */ + +extern int open_cgroups (void ); +extern void close_cgroups (void ); +extern int parse_cgroups (const struct option *opt,const char *str,int unset); + +#endif /* __CGROUP_H__ */