From ce4326d2752fd51726be50e94b750d3f74670ae3 Mon Sep 17 00:00:00 2001 From: David Sharp Date: Fri, 11 Sep 2020 19:56:52 -0700 Subject: perf record: Set PERF_RECORD_PERIOD if attr->freq is set. evsel__config() would only set PERF_RECORD_PERIOD if it set attr->freq from perf record options. When it is set by libpfm events, it would not get set. This changes evsel__config to see if attr->freq is set outside of whether or not it changes attr->freq itself. Signed-off-by: David Sharp Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andi Kleen Cc: Andrii Nakryiko Cc: Athira Jajeev Cc: Daniel Borkmann Cc: Ian Rogers Cc: Jiri Olsa Cc: John Fastabend Cc: KP Singh Cc: Mark Rutland Cc: Martin KaFai Lau Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Song Liu Cc: Stephane Eranian Cc: Yonghong Song Cc: david sharp Link: http://lore.kernel.org/lkml/20200912025655.1337192-2-irogers@google.com Signed-off-by: Ian Rogers Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index fd865002cbbd..3e985016da7e 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -979,13 +979,18 @@ void evsel__config(struct evsel *evsel, struct record_opts *opts, if (!attr->sample_period || (opts->user_freq != UINT_MAX || opts->user_interval != ULLONG_MAX)) { if (opts->freq) { - evsel__set_sample_bit(evsel, PERIOD); attr->freq = 1; attr->sample_freq = opts->freq; } else { attr->sample_period = opts->default_interval; } } + /* + * If attr->freq was set (here or earlier), ask for period + * to be sampled. + */ + if (attr->freq) + evsel__set_sample_bit(evsel, PERIOD); if (opts->no_samples) attr->sample_freq = 0; -- cgit v1.2.3 From ae5dcc8abe311ceafd8753fb488e38c4f5e8e699 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Fri, 11 Sep 2020 19:56:53 -0700 Subject: perf record: Prevent override of attr->sample_period for libpfm4 events Before: $ perf record -c 10000 --pfm-events=cycles:period=77777 Would yield a cycles event with period=10000, instead of 77777. the event string and perf record initializing the event. This was due to an ordering issue between libpfm4 parsing events with attr->sample_period != 0 by the time intent of the author. perf_evsel__config() is invoked. This seems to have been the This patch fixes the problem by preventing override for Signed-off-by: Stephane Eranian Reviewed-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andi Kleen Cc: Andrii Nakryiko Cc: Athira Jajeev Cc: Daniel Borkmann Cc: Jiri Olsa Cc: John Fastabend Cc: KP Singh Cc: Mark Rutland Cc: Martin KaFai Lau Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Song Liu Cc: Yonghong Song Link: http://lore.kernel.org/lkml/20200912025655.1337192-3-irogers@google.com Signed-off-by: Ian Rogers Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 3e985016da7e..459b51e90063 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -976,8 +976,7 @@ void evsel__config(struct evsel *evsel, struct record_opts *opts, * We default some events to have a default interval. But keep * it a weak assumption overridable by the user. */ - if (!attr->sample_period || (opts->user_freq != UINT_MAX || - opts->user_interval != ULLONG_MAX)) { + if (!attr->sample_period) { if (opts->freq) { attr->freq = 1; attr->sample_freq = opts->freq; -- cgit v1.2.3 From 3b0a18c1aa6cbfa7b0dd513b6be9893ef6e6ac30 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 11 Sep 2020 19:56:54 -0700 Subject: perf record: Don't clear event's period if set by a term If events in a group explicitly set a frequency or period with leader sampling, don't disable the samples on those events. Prior to 5.8: perf record -e '{cycles/period=12345000/,instructions/period=6789000/}:S' would clear the attributes then apply the config terms. In commit 5f34278867b7 leader sampling configuration was moved to after applying the config terms, in the example, making the instructions' event have its period cleared. This change makes it so that sampling is only disabled if configuration terms aren't present. Committer testing: Before: # perf record -e '{cycles/period=1/,instructions/period=2/}:S' sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.051 MB perf.data (6 samples) ] # # perf evlist -v cycles/period=1/: size: 120, { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|READ|ID, read_format: ID|GROUP, disabled: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 instructions/period=2/: size: 120, config: 0x1, sample_type: IP|TID|TIME|READ|ID, read_format: ID|GROUP, sample_id_all: 1, exclude_guest: 1 # After: # perf record -e '{cycles/period=1/,instructions/period=2/}:S' sleep 0.0001 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.052 MB perf.data (4 samples) ] # perf evlist -v cycles/period=1/: size: 120, { sample_period, sample_freq }: 1, sample_type: IP|TID|TIME|READ|ID, read_format: ID|GROUP, disabled: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1 instructions/period=2/: size: 120, config: 0x1, { sample_period, sample_freq }: 2, sample_type: IP|TID|TIME|READ|ID, read_format: ID|GROUP, sample_id_all: 1, exclude_guest: 1 # Fixes: 5f34278867b7 ("perf evlist: Move leader-sampling configuration") Signed-off-by: Ian Rogers Acked-by: Adrian Hunter Acked-by: Jiri Olsa Tested-by: Arnaldo Carvalho de Melo Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andi Kleen Cc: Andrii Nakryiko Cc: Athira Jajeev Cc: Daniel Borkmann Cc: John Fastabend Cc: KP Singh Cc: Mark Rutland Cc: Martin KaFai Lau Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Song Liu Cc: Stephane Eranian Cc: Yonghong Song Link: http://lore.kernel.org/lkml/20200912025655.1337192-4-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/record.c | 34 ++++++++++++++++++++++++++-------- 1 file changed, 26 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/record.c b/tools/perf/util/record.c index a4cc11592f6b..ea9aa1d7cf50 100644 --- a/tools/perf/util/record.c +++ b/tools/perf/util/record.c @@ -2,6 +2,7 @@ #include "debug.h" #include "evlist.h" #include "evsel.h" +#include "evsel_config.h" #include "parse-events.h" #include #include @@ -33,11 +34,24 @@ static struct evsel *evsel__read_sampler(struct evsel *evsel, struct evlist *evl return leader; } +static u64 evsel__config_term_mask(struct evsel *evsel) +{ + struct evsel_config_term *term; + struct list_head *config_terms = &evsel->config_terms; + u64 term_types = 0; + + list_for_each_entry(term, config_terms, list) { + term_types |= 1 << term->type; + } + return term_types; +} + static void evsel__config_leader_sampling(struct evsel *evsel, struct evlist *evlist) { struct perf_event_attr *attr = &evsel->core.attr; struct evsel *leader = evsel->leader; struct evsel *read_sampler; + u64 term_types, freq_mask; if (!leader->sample_read) return; @@ -47,16 +61,20 @@ static void evsel__config_leader_sampling(struct evsel *evsel, struct evlist *ev if (evsel == read_sampler) return; + term_types = evsel__config_term_mask(evsel); /* - * Disable sampling for all group members other than the leader in - * case the leader 'leads' the sampling, except when the leader is an - * AUX area event, in which case the 2nd event in the group is the one - * that 'leads' the sampling. + * Disable sampling for all group members except those with explicit + * config terms or the leader. In the case of an AUX area event, the 2nd + * event in the group is the one that 'leads' the sampling. */ - attr->freq = 0; - attr->sample_freq = 0; - attr->sample_period = 0; - attr->write_backward = 0; + freq_mask = (1 << EVSEL__CONFIG_TERM_FREQ) | (1 << EVSEL__CONFIG_TERM_PERIOD); + if ((term_types & freq_mask) == 0) { + attr->freq = 0; + attr->sample_freq = 0; + attr->sample_period = 0; + } + if ((term_types & (1 << EVSEL__CONFIG_TERM_OVERWRITE)) == 0) + attr->write_backward = 0; /* * We don't get a sample for slave events, we make them when delivering -- cgit v1.2.3 From 22fe5a25b5d8c4f8008dc4a8738d6d8a5f5ddbe9 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:18 +0900 Subject: perf test: Free aliases for PMU event map aliases test The aliases were never released causing the following leaks: Indirect leak of 1224 byte(s) in 9 object(s) allocated from: #0 0x7feefb830628 in malloc (/lib/x86_64-linux-gnu/libasan.so.5+0x107628) #1 0x56332c8f1b62 in __perf_pmu__new_alias util/pmu.c:322 #2 0x56332c8f401f in pmu_add_cpu_aliases_map util/pmu.c:778 #3 0x56332c792ce9 in __test__pmu_event_aliases tests/pmu-events.c:295 #4 0x56332c792ce9 in test_aliases tests/pmu-events.c:367 #5 0x56332c76a09b in run_test tests/builtin-test.c:410 #6 0x56332c76a09b in test_and_print tests/builtin-test.c:440 #7 0x56332c76ce69 in __cmd_test tests/builtin-test.c:695 #8 0x56332c76ce69 in cmd_test tests/builtin-test.c:807 #9 0x56332c7d2214 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:312 #10 0x56332c6701a8 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:364 #11 0x56332c6701a8 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:408 #12 0x56332c6701a8 in main /home/namhyung/project/linux/tools/perf/perf.c:538 #13 0x7feefb359cc9 in __libc_start_main ../csu/libc-start.c:308 Fixes: 956a78356c24c ("perf test: Test pmu-events aliases") Signed-off-by: Namhyung Kim Reviewed-by: John Garry Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-11-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/tests/pmu-events.c | 5 +++++ tools/perf/util/pmu.c | 2 +- tools/perf/util/pmu.h | 1 + 3 files changed, 7 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/tests/pmu-events.c b/tools/perf/tests/pmu-events.c index eb19f9a0bc15..d3517a74d95e 100644 --- a/tools/perf/tests/pmu-events.c +++ b/tools/perf/tests/pmu-events.c @@ -274,6 +274,7 @@ static int __test__pmu_event_aliases(char *pmu_name, int *count) int res = 0; bool use_uncore_table; struct pmu_events_map *map = __test_pmu_get_events_map(); + struct perf_pmu_alias *a, *tmp; if (!map) return -1; @@ -347,6 +348,10 @@ static int __test__pmu_event_aliases(char *pmu_name, int *count) pmu_name, alias->name); } + list_for_each_entry_safe(a, tmp, &aliases, list) { + list_del(&a->list); + perf_pmu_free_alias(a); + } free(pmu); return res; } diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index f1688e1f6ed7..555cb3524c25 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -274,7 +274,7 @@ static void perf_pmu_update_alias(struct perf_pmu_alias *old, } /* Delete an alias entry. */ -static void perf_pmu_free_alias(struct perf_pmu_alias *newalias) +void perf_pmu_free_alias(struct perf_pmu_alias *newalias) { zfree(&newalias->name); zfree(&newalias->desc); diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 44ccbdbb1c37..b63c4c5e335e 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -113,6 +113,7 @@ void pmu_add_cpu_aliases_map(struct list_head *head, struct perf_pmu *pmu, struct pmu_events_map *perf_pmu__find_map(struct perf_pmu *pmu); bool pmu_uncore_alias_match(const char *pmu_name, const char *name); +void perf_pmu_free_alias(struct perf_pmu_alias *alias); int perf_pmu__convert_scale(const char *scale, char **end, double *sval); -- cgit v1.2.3 From 4f57a1ed749a81ec553d89233cab53db9365e193 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:09 +0900 Subject: perf metric: Fix some memory leaks I found some memory leaks while reading the metric code. Some are real and others only occur in the error path. When it failed during metric or event parsing, it should release all resources properly. Fixes: b18f3e365019d ("perf stat: Support JSON metrics in perf stat") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: John Garry Cc: Kajol Jain Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-2-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/metricgroup.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index 8831b964288f..af664d6218d6 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -530,6 +530,9 @@ void metricgroup__print(bool metrics, bool metricgroups, char *filter, continue; strlist__add(me->metrics, s); } + + if (!raw) + free(s); } free(omg); } @@ -1040,7 +1043,7 @@ static int parse_groups(struct evlist *perf_evlist, const char *str, ret = metricgroup__add_metric_list(str, metric_no_group, &extra_events, &metric_list, map); if (ret) - return ret; + goto out; pr_debug("adding %s\n", extra_events.buf); bzero(&parse_error, sizeof(parse_error)); ret = __parse_events(perf_evlist, extra_events.buf, &parse_error, fake_pmu); @@ -1048,11 +1051,11 @@ static int parse_groups(struct evlist *perf_evlist, const char *str, parse_events_print_error(&parse_error, extra_events.buf); goto out; } - strbuf_release(&extra_events); ret = metricgroup__setup_events(&metric_list, metric_no_merge, perf_evlist, metric_events); out: metricgroup__free_metrics(&metric_list); + strbuf_release(&extra_events); return ret; } -- cgit v1.2.3 From b033ab11ad0c7f9a9c9fa4a1ac2ffad14454a9f3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:10 +0900 Subject: perf metric: Fix some memory leaks - part 2 The metric_event_delete() missed to free expr->metric_events and it should free an expr when metric_refs allocation failed. Fixes: 4ea2896715e67 ("perf metric: Collect referenced metrics in struct metric_expr") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: John Garry Cc: Kajol Jain Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/metricgroup.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index af664d6218d6..b28c09447c10 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -85,6 +85,7 @@ static void metric_event_delete(struct rblist *rblist __maybe_unused, list_for_each_entry_safe(expr, tmp, &me->head, nd) { free(expr->metric_refs); + free(expr->metric_events); free(expr); } @@ -316,6 +317,7 @@ static int metricgroup__setup_events(struct list_head *groups, if (!metric_refs) { ret = -ENOMEM; free(metric_events); + free(expr); break; } -- cgit v1.2.3 From bfd1b83d75e44a9f65de30accb3dd3b5940bd3ac Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:11 +0900 Subject: perf evlist: Fix cpu/thread map leak Asan reported leak of cpu and thread maps as they have one more refcount than released. I found that after setting evlist maps it should release it's refcount. It seems to be broken from the beginning so I chose the original commit as the culprit. But not sure how it's applied to stable trees since there are many changes in the code after that. Fixes: 7e2ed097538c5 ("perf evlist: Store pointer to the cpu and thread maps") Fixes: 4112eb1899c0e ("perf evlist: Default to syswide target when no thread/cpu maps set") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index e3fa3bf7498a..c0768c61eb43 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -946,6 +946,10 @@ int perf_evlist__create_maps(struct evlist *evlist, struct target *target) perf_evlist__set_maps(&evlist->core, cpus, threads); + /* as evlist now has references, put count here */ + perf_cpu_map__put(cpus); + perf_thread_map__put(threads); + return 0; out_delete_threads: @@ -1273,11 +1277,12 @@ static int perf_evlist__create_syswide_maps(struct evlist *evlist) goto out_put; perf_evlist__set_maps(&evlist->core, cpus, threads); -out: - return err; + + perf_thread_map__put(threads); out_put: perf_cpu_map__put(cpus); - goto out; +out: + return err; } int evlist__open(struct evlist *evlist) -- cgit v1.2.3 From b12eea5ad8e77f8a380a141e3db67c07432dde16 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:13 +0900 Subject: perf parse-event: Fix memory leak in evsel->unit The evsel->unit borrows a pointer of pmu event or alias instead of owns a string. But tool event (duration_time) passes a result of strdup() caused a leak. It was found by ASAN during metric test: Direct leak of 210 byte(s) in 70 object(s) allocated from: #0 0x7fe366fca0b5 in strdup (/lib/x86_64-linux-gnu/libasan.so.5+0x920b5) #1 0x559fbbcc6ea3 in add_event_tool util/parse-events.c:414 #2 0x559fbbcc6ea3 in parse_events_add_tool util/parse-events.c:1414 #3 0x559fbbd8474d in parse_events_parse util/parse-events.y:439 #4 0x559fbbcc95da in parse_events__scanner util/parse-events.c:2096 #5 0x559fbbcc95da in __parse_events util/parse-events.c:2141 #6 0x559fbbc28555 in check_parse_id tests/pmu-events.c:406 #7 0x559fbbc28555 in check_parse_id tests/pmu-events.c:393 #8 0x559fbbc28555 in check_parse_cpu tests/pmu-events.c:415 #9 0x559fbbc28555 in test_parsing tests/pmu-events.c:498 #10 0x559fbbc0109b in run_test tests/builtin-test.c:410 #11 0x559fbbc0109b in test_and_print tests/builtin-test.c:440 #12 0x559fbbc03e69 in __cmd_test tests/builtin-test.c:695 #13 0x559fbbc03e69 in cmd_test tests/builtin-test.c:807 #14 0x559fbbc691f4 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:312 #15 0x559fbbb071a8 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:364 #16 0x559fbbb071a8 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:408 #17 0x559fbbb071a8 in main /home/namhyung/project/linux/tools/perf/perf.c:538 #18 0x7fe366b68cc9 in __libc_start_main ../csu/libc-start.c:308 Fixes: f0fbb114e3025 ("perf stat: Implement duration_time as a proper event") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-6-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index c4d2394e2b2d..667cbca1547a 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -411,7 +411,7 @@ static int add_event_tool(struct list_head *list, int *idx, return -ENOMEM; evsel->tool_event = tool_event; if (tool_event == PERF_TOOL_DURATION_TIME) - evsel->unit = strdup("ns"); + evsel->unit = "ns"; return 0; } -- cgit v1.2.3 From 437822bf38c1dc3bda6ffde52689e4e667046a6a Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:15 +0900 Subject: perf metric: Release expr_parse_ctx after testing The test_generic_metric() missed to release entries in the pctx. Asan reported following leak (and more): Direct leak of 128 byte(s) in 1 object(s) allocated from: #0 0x7f4c9396980e in calloc (/lib/x86_64-linux-gnu/libasan.so.5+0x10780e) #1 0x55f7e748cc14 in hashmap_grow (/home/namhyung/project/linux/tools/perf/perf+0x90cc14) #2 0x55f7e748d497 in hashmap__insert (/home/namhyung/project/linux/tools/perf/perf+0x90d497) #3 0x55f7e7341667 in hashmap__set /home/namhyung/project/linux/tools/perf/util/hashmap.h:111 #4 0x55f7e7341667 in expr__add_ref util/expr.c:120 #5 0x55f7e7292436 in prepare_metric util/stat-shadow.c:783 #6 0x55f7e729556d in test_generic_metric util/stat-shadow.c:858 #7 0x55f7e712390b in compute_single tests/parse-metric.c:128 #8 0x55f7e712390b in __compute_metric tests/parse-metric.c:180 #9 0x55f7e712446d in compute_metric tests/parse-metric.c:196 #10 0x55f7e712446d in test_dcache_l2 tests/parse-metric.c:295 #11 0x55f7e712446d in test__parse_metric tests/parse-metric.c:355 #12 0x55f7e70be09b in run_test tests/builtin-test.c:410 #13 0x55f7e70be09b in test_and_print tests/builtin-test.c:440 #14 0x55f7e70c101a in __cmd_test tests/builtin-test.c:661 #15 0x55f7e70c101a in cmd_test tests/builtin-test.c:807 #16 0x55f7e7126214 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:312 #17 0x55f7e6fc41a8 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:364 #18 0x55f7e6fc41a8 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:408 #19 0x55f7e6fc41a8 in main /home/namhyung/project/linux/tools/perf/perf.c:538 #20 0x7f4c93492cc9 in __libc_start_main ../csu/libc-start.c:308 Fixes: 6d432c4c8aa56 ("perf tools: Add test_generic_metric function") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-8-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/stat-shadow.c | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index e1ba6c1b916a..a5f42c22c484 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -853,14 +853,16 @@ static void generic_metric(struct perf_stat_config *config, double test_generic_metric(struct metric_expr *mexp, int cpu, struct runtime_stat *st) { struct expr_parse_ctx pctx; - double ratio; + double ratio = 0.0; if (prepare_metric(mexp->metric_events, mexp->metric_refs, &pctx, cpu, st) < 0) - return 0.; + goto out; if (expr__parse(&ratio, &pctx, mexp->metric_expr, 1)) - return 0.; + ratio = 0.0; +out: + expr__ctx_clear(&pctx); return ratio; } -- cgit v1.2.3 From 27adafcda3d8b8a818e85d58eea95b85b5f513f9 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:16 +0900 Subject: perf metric: Free metric when it failed to resolve The metricgroup__add_metric() can find multiple match for a metric group and it's possible to fail. Also it can fail in the middle like in resolve_metric() even for single metric. In those cases, the intermediate list and ids will be leaked like: Direct leak of 3 byte(s) in 1 object(s) allocated from: #0 0x7f4c938f40b5 in strdup (/lib/x86_64-linux-gnu/libasan.so.5+0x920b5) #1 0x55f7e71c1bef in __add_metric util/metricgroup.c:683 #2 0x55f7e71c31d0 in add_metric util/metricgroup.c:906 #3 0x55f7e71c3844 in metricgroup__add_metric util/metricgroup.c:940 #4 0x55f7e71c488d in metricgroup__add_metric_list util/metricgroup.c:993 #5 0x55f7e71c488d in parse_groups util/metricgroup.c:1045 #6 0x55f7e71c60a4 in metricgroup__parse_groups_test util/metricgroup.c:1087 #7 0x55f7e71235ae in __compute_metric tests/parse-metric.c:164 #8 0x55f7e7124650 in compute_metric tests/parse-metric.c:196 #9 0x55f7e7124650 in test_recursion_fail tests/parse-metric.c:318 #10 0x55f7e7124650 in test__parse_metric tests/parse-metric.c:356 #11 0x55f7e70be09b in run_test tests/builtin-test.c:410 #12 0x55f7e70be09b in test_and_print tests/builtin-test.c:440 #13 0x55f7e70c101a in __cmd_test tests/builtin-test.c:661 #14 0x55f7e70c101a in cmd_test tests/builtin-test.c:807 #15 0x55f7e7126214 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:312 #16 0x55f7e6fc41a8 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:364 #17 0x55f7e6fc41a8 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:408 #18 0x55f7e6fc41a8 in main /home/namhyung/project/linux/tools/perf/perf.c:538 #19 0x7f4c93492cc9 in __libc_start_main ../csu/libc-start.c:308 Fixes: 83de0b7d535de ("perf metric: Collect referenced metrics in struct metric_ref_node") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-9-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/metricgroup.c | 17 ++++++++++++----- 1 file changed, 12 insertions(+), 5 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index b28c09447c10..c8904e471a71 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -939,7 +939,7 @@ static int metricgroup__add_metric(const char *metric, bool metric_no_group, ret = add_metric(&list, pe, metric_no_group, &m, NULL, &ids); if (ret) - return ret; + goto out; /* * Process any possible referenced metrics @@ -948,12 +948,14 @@ static int metricgroup__add_metric(const char *metric, bool metric_no_group, ret = resolve_metric(metric_no_group, &list, map, &ids); if (ret) - return ret; + goto out; } /* End of pmu events. */ - if (!has_match) - return -EINVAL; + if (!has_match) { + ret = -EINVAL; + goto out; + } list_for_each_entry(m, &list, nd) { if (events->len > 0) @@ -968,9 +970,14 @@ static int metricgroup__add_metric(const char *metric, bool metric_no_group, } } +out: + /* + * add to metric_list so that they can be released + * even if it's failed + */ list_splice(&list, metric_list); expr_ids__exit(&ids); - return 0; + return ret; } static int metricgroup__add_metric_list(const char *list, bool metric_no_group, -- cgit v1.2.3 From 6f47ed6cd12a0ea9b55b19b5d6e4a3b490a97bdf Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:17 +0900 Subject: perf metric: Do not free metric when failed to resolve It's dangerous to free the original metric when it's called from resolve_metric() as it's already in the metric_list and might have other resources too. Instead, it'd better let them bail out and be released properly at the later stage. So add a check when it's called from metricgroup__add_metric() and release it. Also make sure that mp is set properly. Fixes: 83de0b7d535de ("perf metric: Collect referenced metrics in struct metric_ref_node") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-10-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/metricgroup.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index c8904e471a71..ab5030fcfed4 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -672,7 +672,6 @@ static int __add_metric(struct list_head *metric_list, m->has_constraint = metric_no_group || metricgroup__has_constraint(pe); INIT_LIST_HEAD(&m->metric_refs); m->metric_refs_cnt = 0; - *mp = m; parent = expr_ids__alloc(ids); if (!parent) { @@ -685,6 +684,7 @@ static int __add_metric(struct list_head *metric_list, free(m); return -ENOMEM; } + *mp = m; } else { /* * We got here for the referenced metric, via the @@ -719,8 +719,11 @@ static int __add_metric(struct list_head *metric_list, * all the metric's IDs and add it to the parent context. */ if (expr__find_other(pe->metric_expr, NULL, &m->pctx, runtime) < 0) { - expr__ctx_clear(&m->pctx); - free(m); + if (m->metric_refs_cnt == 0) { + expr__ctx_clear(&m->pctx); + free(m); + *mp = NULL; + } return -EINVAL; } -- cgit v1.2.3 From d26383dcb2b4b8629fde05270b4e3633be9e3d4b Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 15 Sep 2020 12:18:19 +0900 Subject: perf test: Free formats for perf pmu parse test The following leaks were detected by ASAN: Indirect leak of 360 byte(s) in 9 object(s) allocated from: #0 0x7fecc305180e in calloc (/lib/x86_64-linux-gnu/libasan.so.5+0x10780e) #1 0x560578f6dce5 in perf_pmu__new_format util/pmu.c:1333 #2 0x560578f752fc in perf_pmu_parse util/pmu.y:59 #3 0x560578f6a8b7 in perf_pmu__format_parse util/pmu.c:73 #4 0x560578e07045 in test__pmu tests/pmu.c:155 #5 0x560578de109b in run_test tests/builtin-test.c:410 #6 0x560578de109b in test_and_print tests/builtin-test.c:440 #7 0x560578de401a in __cmd_test tests/builtin-test.c:661 #8 0x560578de401a in cmd_test tests/builtin-test.c:807 #9 0x560578e49354 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:312 #10 0x560578ce71a8 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:364 #11 0x560578ce71a8 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:408 #12 0x560578ce71a8 in main /home/namhyung/project/linux/tools/perf/perf.c:538 #13 0x7fecc2b7acc9 in __libc_start_main ../csu/libc-start.c:308 Fixes: cff7f956ec4a1 ("perf tests: Move pmu tests into separate object") Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20200915031819.386559-12-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/tests/pmu.c | 1 + tools/perf/util/pmu.c | 11 +++++++++++ tools/perf/util/pmu.h | 1 + 3 files changed, 13 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/tests/pmu.c b/tools/perf/tests/pmu.c index 5c11fe2b3040..714e6830a758 100644 --- a/tools/perf/tests/pmu.c +++ b/tools/perf/tests/pmu.c @@ -173,6 +173,7 @@ int test__pmu(struct test *test __maybe_unused, int subtest __maybe_unused) ret = 0; } while (0); + perf_pmu__del_formats(&formats); test_format_dir_put(format); return ret; } diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 555cb3524c25..d41caeb35cf6 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -1354,6 +1354,17 @@ void perf_pmu__set_format(unsigned long *bits, long from, long to) set_bit(b, bits); } +void perf_pmu__del_formats(struct list_head *formats) +{ + struct perf_pmu_format *fmt, *tmp; + + list_for_each_entry_safe(fmt, tmp, formats, list) { + list_del(&fmt->list); + free(fmt->name); + free(fmt); + } +} + static int sub_non_neg(int a, int b) { if (b > a) diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index b63c4c5e335e..a64e9c9ce731 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -94,6 +94,7 @@ int perf_pmu__new_format(struct list_head *list, char *name, int config, unsigned long *bits); void perf_pmu__set_format(unsigned long *bits, long from, long to); int perf_pmu__format_parse(char *dir, struct list_head *head); +void perf_pmu__del_formats(struct list_head *formats); struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu); -- cgit v1.2.3 From ce9c13f31b10560d4e4d801f9f94895f595756aa Mon Sep 17 00:00:00 2001 From: Qi Liu Date: Wed, 16 Sep 2020 18:48:51 +0800 Subject: perf stat: Fix the ratio comments of miss-events 'perf stat' displays miss ratio of L1-dcache, L1-icache, dTLB cache, iTLB cache and LL-cache. Take L1-dcache for example, miss ratio is caculated as "L1-dcache-load-misses/L1-dcache-loads". So "of all L1-dcache hits" is unsuitable to describe it, and "of all L1-dcache accesses" seems better. The comments of L1-icache, dTLB cache, iTLB cache and LL-cache are fixed in the same way. Signed-off-by: Qi Liu Reviewed-by: Andi Kleen Cc: linuxarm@huawei.com Link: http://lore.kernel.org/lkml/1600253331-10535-1-git-send-email-liuqi115@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/stat-shadow.c | 20 ++++++++++---------- 1 file changed, 10 insertions(+), 10 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index a5f42c22c484..924b54d15d54 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -517,7 +517,7 @@ static void print_l1_dcache_misses(struct perf_stat_config *config, color = get_ratio_color(GRC_CACHE_MISSES, ratio); - out->print_metric(config, out->ctx, color, "%7.2f%%", "of all L1-dcache hits", ratio); + out->print_metric(config, out->ctx, color, "%7.2f%%", "of all L1-dcache accesses", ratio); } static void print_l1_icache_misses(struct perf_stat_config *config, @@ -538,7 +538,7 @@ static void print_l1_icache_misses(struct perf_stat_config *config, ratio = avg / total * 100.0; color = get_ratio_color(GRC_CACHE_MISSES, ratio); - out->print_metric(config, out->ctx, color, "%7.2f%%", "of all L1-icache hits", ratio); + out->print_metric(config, out->ctx, color, "%7.2f%%", "of all L1-icache accesses", ratio); } static void print_dtlb_cache_misses(struct perf_stat_config *config, @@ -558,7 +558,7 @@ static void print_dtlb_cache_misses(struct perf_stat_config *config, ratio = avg / total * 100.0; color = get_ratio_color(GRC_CACHE_MISSES, ratio); - out->print_metric(config, out->ctx, color, "%7.2f%%", "of all dTLB cache hits", ratio); + out->print_metric(config, out->ctx, color, "%7.2f%%", "of all dTLB cache accesses", ratio); } static void print_itlb_cache_misses(struct perf_stat_config *config, @@ -578,7 +578,7 @@ static void print_itlb_cache_misses(struct perf_stat_config *config, ratio = avg / total * 100.0; color = get_ratio_color(GRC_CACHE_MISSES, ratio); - out->print_metric(config, out->ctx, color, "%7.2f%%", "of all iTLB cache hits", ratio); + out->print_metric(config, out->ctx, color, "%7.2f%%", "of all iTLB cache accesses", ratio); } static void print_ll_cache_misses(struct perf_stat_config *config, @@ -598,7 +598,7 @@ static void print_ll_cache_misses(struct perf_stat_config *config, ratio = avg / total * 100.0; color = get_ratio_color(GRC_CACHE_MISSES, ratio); - out->print_metric(config, out->ctx, color, "%7.2f%%", "of all LL-cache hits", ratio); + out->print_metric(config, out->ctx, color, "%7.2f%%", "of all LL-cache accesses", ratio); } /* @@ -920,7 +920,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (runtime_stat_n(st, STAT_L1_DCACHE, ctx, cpu) != 0) print_l1_dcache_misses(config, cpu, evsel, avg, out, st); else - print_metric(config, ctxp, NULL, NULL, "of all L1-dcache hits", 0); + print_metric(config, ctxp, NULL, NULL, "of all L1-dcache accesses", 0); } else if ( evsel->core.attr.type == PERF_TYPE_HW_CACHE && evsel->core.attr.config == ( PERF_COUNT_HW_CACHE_L1I | @@ -930,7 +930,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (runtime_stat_n(st, STAT_L1_ICACHE, ctx, cpu) != 0) print_l1_icache_misses(config, cpu, evsel, avg, out, st); else - print_metric(config, ctxp, NULL, NULL, "of all L1-icache hits", 0); + print_metric(config, ctxp, NULL, NULL, "of all L1-icache accesses", 0); } else if ( evsel->core.attr.type == PERF_TYPE_HW_CACHE && evsel->core.attr.config == ( PERF_COUNT_HW_CACHE_DTLB | @@ -940,7 +940,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (runtime_stat_n(st, STAT_DTLB_CACHE, ctx, cpu) != 0) print_dtlb_cache_misses(config, cpu, evsel, avg, out, st); else - print_metric(config, ctxp, NULL, NULL, "of all dTLB cache hits", 0); + print_metric(config, ctxp, NULL, NULL, "of all dTLB cache accesses", 0); } else if ( evsel->core.attr.type == PERF_TYPE_HW_CACHE && evsel->core.attr.config == ( PERF_COUNT_HW_CACHE_ITLB | @@ -950,7 +950,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (runtime_stat_n(st, STAT_ITLB_CACHE, ctx, cpu) != 0) print_itlb_cache_misses(config, cpu, evsel, avg, out, st); else - print_metric(config, ctxp, NULL, NULL, "of all iTLB cache hits", 0); + print_metric(config, ctxp, NULL, NULL, "of all iTLB cache accesses", 0); } else if ( evsel->core.attr.type == PERF_TYPE_HW_CACHE && evsel->core.attr.config == ( PERF_COUNT_HW_CACHE_LL | @@ -960,7 +960,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (runtime_stat_n(st, STAT_LL_CACHE, ctx, cpu) != 0) print_ll_cache_misses(config, cpu, evsel, avg, out, st); else - print_metric(config, ctxp, NULL, NULL, "of all LL-cache hits", 0); + print_metric(config, ctxp, NULL, NULL, "of all LL-cache accesses", 0); } else if (evsel__match(evsel, HARDWARE, HW_CACHE_MISSES)) { total = runtime_stat_avg(st, STAT_CACHEREFS, ctx, cpu); -- cgit v1.2.3