perf record: Fix sample cgroup & namespace tracking
Namhyung Kim
maintainerPATCH 1/2 perf record: Fix sample cgroup & namespace tracking
The recent change in perf_tool constification broke the cgroup and/or namespace tracking by resetting tool fields. It should set the values after perf_tool__init().
--- tools/perf/builtin-record.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-)
tools/perf/builtin-record.c +3 -6
@@ -2374,13 +2374,8 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)signal (SIGTERM , sig_handler);signal (SIGSEGV , sigsegv_handler); -if (rec->opts.record_namespaces) - tool->namespace_events = true; -if (rec->opts.record_cgroup) { -#ifdef HAVE_FILE_HANDLE - tool->cgroup_events = true; -#else +#ifndef HAVE_FILE_HANDLE pr_err ("cgroup tracking is not supported\n" );return -1 ;#endif
@@ -2406,6 +2401,8 @@ static int __cmd_record(struct record *rec, int argc, const char **argv) tool->mmap2 = build_id__process_mmap2; tool->itrace_start = process_timestamp_boundary; tool->aux = process_timestamp_boundary; + tool->namespace_events = rec->opts.record_namespaces; + tool->cgroup_events = rec->opts.record_cgroup; session =perf_session__new (data, tool);if (IS_ERR (session)) {pr_err ("Perf session creation failed.\n" );
Signature · 1 line
2.46.0.184.g6999bdac58-goog
Namhyung Kim
maintainerPATCH 2/2 perf test: Add cgroup sampling test
Add it to the record.sh shell test to verify if it tracks cgroup information correctly. It records with --all-cgroups option can check if it has PERF_RECORD_CGROUP and the names are not "unknown".
$ sudo ./perf test -vv 95 95: perf record tests: --- start --- test child forked, pid 2871922 169c90-169cd0 g test_loop perf does have symbol 'test_loop' Basic --per-thread mode test Basic --per-thread mode test [Success] Register capture test Register capture test [Success] Basic --system-wide mode test Basic --system-wide mode test [Success] Basic target workload test Basic target workload test [Success] Branch counter test branch counter feature not supported on all core PMUs (/sys/bus/event_source/devices/cpu) [Skipped] Cgroup sampling test Cgroup sampling test [Success] ---- end(0) ---- 95: perf record tests : Ok
--- tools/perf/tests/shell/record.sh | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+)
tools/perf/tests/shell/record.sh +23 -0
@@ -206,6 +206,28 @@ test_branch_counter() { echo "Branch counter test [Success]" } +test_cgroup() { + echo "Cgroup sampling test" + if ! perf record -aB --synth=cgroup --all-cgroups -o "${perfdata}" ${testprog} 2> /dev/null + then + echo "Cgroup sampling [Skipped not supported]" + return + fi + if ! perf report -i "${perfdata}" -D | grep -q "CGROUP" + then + echo "Cgroup sampling [Failed missing output]" + err=1 + return + fi + if ! perf script -i "${perfdata}" -F cgroup | grep -q -v "unknown" + then + echo "Cgroup sampling [Failed cannot resolve cgroup names]" + err=1 + return + fi + echo "Cgroup sampling test [Success]" +} + # raise the limit of file descriptors to minimum if [[ $default_fd_limit -lt $min_fd_limit ]]; then ulimit -Sn $min_fd_limit
@@ -216,6 +238,7 @@ test_register_capture test_system_wide test_workload test_branch_counter +test_cgroup # restore the default value ulimit -Sn $default_fd_limit
Signature · 1 line
2.46.0.184.g6999bdac58-goog
Re: [PATCH 2/2] perf test: Add cgroup sampling test
Quoted 26 lines · Namhyung Kim
Add it to the
record.shshell test to verify if it tracks cgroup information correctly. It records with --all-cgroups option can check if it hasPERF_RECORD_CGROUPand the names are not "unknown".$ sudo ./perf test -vv 95 95: perf record tests: --- start --- test child forked, pid 2871922 169c90-169cd0 g test_loop perf does have symbol 'test_loop' Basic --per-thread mode test Basic --per-thread mode test [Success] Register capture test Register capture test [Success] Basic --system-wide mode test Basic --system-wide mode test [Success] Basic target workload test Basic target workload test [Success] Branch counter test branch counter feature not supported on all core PMUs (/sys/bus/event_source/devices/cpu) [Skipped] Cgroup sampling test Cgroup sampling test [Success] ---- end(0) ---- 95: perf record tests : Ok
Quoted 1 line
Signed-off-by Namhyung Kim <[email protected]>
Thanks,
Ian
Quoted 48 lines
--- tools/perf/tests/shell/record.sh | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+)tools/perf/tests/shell/record.sh +23
-1@@ -206,6 +206,28 @@ test_branch_counter() { echo "Branch counter test [Success]" } +test_cgroup() { + echo "Cgroup sampling test" + if ! perf record -aB --synth=cgroup --all-cgroups -o "${perfdata}" ${testprog} 2> /dev/null + then + echo "Cgroup sampling [Skipped not supported]" + return + fi + if ! perf report -i "${perfdata}" -D | grep -q "CGROUP" + then + echo "Cgroup sampling [Failed missing output]" + err=1 + return + fi + if ! perf script -i "${perfdata}" -F cgroup | grep -q -v "unknown" + then + echo "Cgroup sampling [Failed cannot resolve cgroup names]" + err=1 + return + fi + echo "Cgroup sampling test [Success]" +} + # raise the limit of file descriptors to minimum if [[ $default_fd_limit -lt $min_fd_limit ]]; then ulimit -Sn $min_fd_limit@@ -216,6 +238,7 @@ test_register_capture test_system_wide test_workload test_branch_counter +test_cgroup # restore the default value ulimit -Sn $default_fd_limit --2.46.0.184.g6999bdac58-goog
Quoted 7 lines · Namhyung Kim
The recent change in
perf_toolconstification broke the cgroup and/or namespace tracking by resetting tool fields. It should set the values afterperf_tool__init().Fixes cecb1cf154b30 ("perf record: Use perf_tool__init()")Signed-off-by Namhyung Kim <[email protected]>
Thanks,
Ian
Quoted 35 lines
--- tools/perf/builtin-record.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-)tools/perf/builtin-record.c +3
-7@@ -2374,13 +2374,8 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)signal (SIGTERM , sig_handler);signal (SIGSEGV , sigsegv_handler); -if (rec->opts.record_namespaces) - tool->namespace_events = true; -if (rec->opts.record_cgroup) { -#ifdef HAVE_FILE_HANDLE - tool->cgroup_events = true; -#else +#ifndef HAVE_FILE_HANDLE pr_err ("cgroup tracking is not supported\n" );return -1 ;#endif @@ -2406,6 +2401,8 @@ static int __cmd_record(struct record *rec, int argc, const char **argv) tool->mmap2 = build_id__process_mmap2; tool->itrace_start = process_timestamp_boundary; tool->aux = process_timestamp_boundary; + tool->namespace_events = rec->opts.record_namespaces; + tool->cgroup_events = rec->opts.record_cgroup; session =perf_session__new (data, tool);if (IS_ERR (session)) {pr_err ("Perf session creation failed.\n" ); --2.46.0.184.g6999bdac58-goog
Quoted 10 lines · Ian Rogers
Quoted 7 lines · Namhyung Kim
The recent change in
perf_toolconstification broke the cgroup and/or namespace tracking by resetting tool fields. It should set the values afterperf_tool__init().Fixes cecb1cf154b30 ("perf record: Use perf_tool__init()")Signed-off-by Namhyung Kim <[email protected]>Reviewed-by Ian Rogers <[email protected]>
Thanks, applied to perf-tools-next,
- Arnaldo