From 6066622c97cc0b25d287dc859dbe5f185740e0b4 Mon Sep 17 00:00:00 2001 From: Jiapeng Chong Date: Thu, 25 Feb 2021 18:04:43 +0800 Subject: perf machine: Use true and false for bool variable Fix the following coccicheck warnings: ./tools/perf/util/machine.c:2000:9-10: WARNING: return of 0/1 in function 'symbol__match_regex' with return type bool. Committer notes: Found this in the pile, it was already returning bool, but this patch simplifies it further, from 3 lines to just 1. Reported-by: Abaci Robot Signed-off-by: Jiapeng Chong Suggested-by: David Laight Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Andrii Nakryiko Cc: Daniel Borkmann Cc: Ingo Molnar 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 Cc: bpf@vger.kernel.org Cc: netdev@vger.kernel.org Link: https://lore.kernel.org/r/1614247483-102665-1-git-send-email-jiapeng.chong@linux.alibaba.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/machine.c | 4 +--- 1 file changed, 1 insertion(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 88f31b3a63ac..addfae2f63ef 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2213,9 +2213,7 @@ int machine__process_event(struct machine *machine, union perf_event *event, static bool symbol__match_regex(struct symbol *sym, regex_t *regex) { - if (!regexec(regex, sym->name, 0, NULL, 0)) - return true; - return false; + return regexec(regex, sym->name, 0, NULL, 0) == 0; } static void ip__resolve_ams(struct thread *thread, -- cgit v1.2.3 From 6fcfe54d2c91925ec3054cf25e68064913ca7948 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 1 Sep 2023 16:39:45 -0700 Subject: perf parse-events: Remove unnecessary __maybe_unused The parameter head_terms is always used in get_config_terms. Reviewed-by: James Clark Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Rob Herring Link: https://lore.kernel.org/r/20230901233949.2930562-2-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 65608a3cba81..e9e3623f3fed 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -34,8 +34,7 @@ #ifdef PARSER_DEBUG extern int parse_events_debug; #endif -static int get_config_terms(struct list_head *head_config, - struct list_head *head_terms __maybe_unused); +static int get_config_terms(struct list_head *head_config, struct list_head *head_terms); struct event_symbol event_symbols_hw[PERF_COUNT_HW_MAX] = { [PERF_COUNT_HW_CPU_CYCLES] = { @@ -1079,8 +1078,7 @@ static int config_attr(struct perf_event_attr *attr, return 0; } -static int get_config_terms(struct list_head *head_config, - struct list_head *head_terms __maybe_unused) +static int get_config_terms(struct list_head *head_config, struct list_head *head_terms) { #define ADD_CONFIG_TERM(__type, __weak) \ struct evsel_config_term *__t; \ -- cgit v1.2.3 From 8f91662ef8be473fb025e170601b0dd75838f7d4 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 1 Sep 2023 16:39:46 -0700 Subject: perf parse-events: Tidy up str parameter Add a const and rename str to event_name. Reviewed-by: James Clark Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Rob Herring Link: https://lore.kernel.org/r/20230901233949.2930562-3-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 13 +++++++------ tools/perf/util/parse-events.h | 2 +- 2 files changed, 8 insertions(+), 7 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index e9e3623f3fed..283c559a35b4 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -1482,7 +1482,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, } int parse_events_multi_pmu_add(struct parse_events_state *parse_state, - char *str, struct list_head *head, + const char *event_name, struct list_head *head, struct list_head **listp, void *loc_) { struct parse_events_term *term; @@ -1502,7 +1502,8 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, INIT_LIST_HEAD(head); } - config = strdup(str); + + config = strdup(event_name); if (!config) goto out_err; @@ -1528,7 +1529,7 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, if (parse_events__filter_pmu(parse_state, pmu)) continue; - if (!perf_pmu__have_event(pmu, str)) + if (!perf_pmu__have_event(pmu, event_name)) continue; auto_merge_stats = perf_pmu__auto_merge_stats(pmu); @@ -1539,7 +1540,7 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, strbuf_init(&sb, /*hint=*/ 0); parse_events_term__to_strbuf(orig_head, &sb); - pr_debug("%s -> %s/%s/\n", str, pmu->name, sb.buf); + pr_debug("%s -> %s/%s/\n", event_name, pmu->name, sb.buf); strbuf_release(&sb); ok++; } @@ -1547,13 +1548,13 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, } if (parse_state->fake_pmu) { - if (!parse_events_add_pmu(parse_state, list, str, head, + if (!parse_events_add_pmu(parse_state, list, event_name, head, /*auto_merge_stats=*/true, loc)) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); parse_events_term__to_strbuf(head, &sb); - pr_debug("%s -> %s/%s/\n", str, "fake_pmu", sb.buf); + pr_debug("%s -> %s/%s/\n", event_name, "fake_pmu", sb.buf); strbuf_release(&sb); ok++; } diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index 594e5d2dc67f..36a67ef7b35a 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -217,7 +217,7 @@ struct evsel *parse_events__add_event(int idx, struct perf_event_attr *attr, struct perf_pmu *pmu); int parse_events_multi_pmu_add(struct parse_events_state *parse_state, - char *str, + const char *event_name, struct list_head *head_config, struct list_head **listp, void *loc); -- cgit v1.2.3 From 4163644818e95ea6b0afb3982b34c4d59ed50bb2 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 1 Sep 2023 16:39:47 -0700 Subject: perf parse-events: Avoid enum casts Add term_type to union of values returned by the lexer to avoid casts to and from an integer. Reviewed-by: James Clark Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Rob Herring Link: https://lore.kernel.org/r/20230901233949.2930562-4-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.l | 2 +- tools/perf/util/parse-events.y | 25 +++++++++++-------------- 2 files changed, 12 insertions(+), 15 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l index 4ef4b6f171a0..7bdf0565a92c 100644 --- a/tools/perf/util/parse-events.l +++ b/tools/perf/util/parse-events.l @@ -120,7 +120,7 @@ static int term(yyscan_t scanner, enum parse_events__term_type type) { YYSTYPE *yylval = parse_events_get_lval(scanner); - yylval->num = type; + yylval->term_type = type; return PE_TERM; } diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index 21bfe7e0d944..286ead08065e 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -70,7 +70,7 @@ static void free_list_evsel(struct list_head* list_evsel) %type PE_VALUE_SYM_HW %type PE_VALUE_SYM_SW %type PE_VALUE_SYM_TOOL -%type PE_TERM +%type PE_TERM %type value_sym %type PE_RAW %type PE_NAME @@ -111,6 +111,7 @@ static void free_list_evsel(struct list_head* list_evsel) { char *str; u64 num; + enum parse_events__term_type term_type; struct list_head *list_evsel; struct list_head *list_terms; struct parse_events_term *term; @@ -778,8 +779,7 @@ PE_TERM_HW PE_TERM '=' name_or_legacy { struct parse_events_term *term; - int err = parse_events_term__str(&term, (enum parse_events__term_type)$1, - /*config=*/NULL, $3, &@1, &@3); + int err = parse_events_term__str(&term, $1, /*config=*/NULL, $3, &@1, &@3); if (err) { free($3); @@ -791,8 +791,7 @@ PE_TERM '=' name_or_legacy PE_TERM '=' PE_TERM_HW { struct parse_events_term *term; - int err = parse_events_term__str(&term, (enum parse_events__term_type)$1, - /*config=*/NULL, $3.str, &@1, &@3); + int err = parse_events_term__str(&term, $1, /*config=*/NULL, $3.str, &@1, &@3); if (err) { free($3.str); @@ -804,10 +803,7 @@ PE_TERM '=' PE_TERM_HW PE_TERM '=' PE_TERM { struct parse_events_term *term; - int err = parse_events_term__term(&term, - (enum parse_events__term_type)$1, - (enum parse_events__term_type)$3, - &@1, &@3); + int err = parse_events_term__term(&term, $1, $3, &@1, &@3); if (err) PE_ABORT(err); @@ -818,8 +814,9 @@ PE_TERM '=' PE_TERM PE_TERM '=' PE_VALUE { struct parse_events_term *term; - int err = parse_events_term__num(&term, (enum parse_events__term_type)$1, - /*config=*/NULL, $3, /*novalue=*/false, &@1, &@3); + int err = parse_events_term__num(&term, $1, + /*config=*/NULL, $3, /*novalue=*/false, + &@1, &@3); if (err) PE_ABORT(err); @@ -830,9 +827,9 @@ PE_TERM '=' PE_VALUE PE_TERM { struct parse_events_term *term; - int err = parse_events_term__num(&term, (enum parse_events__term_type)$1, - /*config=*/NULL, /*num=*/1, /*novalue=*/true, - &@1, /*loc_val=*/NULL); + int err = parse_events_term__num(&term, $1, + /*config=*/NULL, /*num=*/1, /*novalue=*/true, + &@1, /*loc_val=*/NULL); if (err) PE_ABORT(err); -- cgit v1.2.3 From 727adeed06e82915841e121762eb329881ae0107 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 1 Sep 2023 16:39:48 -0700 Subject: perf parse-events: Copy fewer term lists When trying to add events to multiple PMUs the term list is copied first as adding the event will rewrite the event's name term into the sysfs and/or json encoding terms (see perf_pmu__check_alias). Change the parse events add API so the passed in term list is const, then copy the list when modification is necessary. Reviewed-by: James Clark Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Rob Herring Link: https://lore.kernel.org/r/20230901233949.2930562-5-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 108 +++++++++++++++++++++++------------------ tools/perf/util/parse-events.h | 7 +-- tools/perf/util/parse-events.y | 17 ++----- 3 files changed, 65 insertions(+), 67 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 283c559a35b4..06a844bcce4a 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -35,6 +35,7 @@ extern int parse_events_debug; #endif static int get_config_terms(struct list_head *head_config, struct list_head *head_terms); +static int parse_events_terms__copy(const struct list_head *src, struct list_head *dest); struct event_symbol event_symbols_hw[PERF_COUNT_HW_MAX] = { [PERF_COUNT_HW_CPU_CYCLES] = { @@ -1367,7 +1368,7 @@ static bool config_term_percore(struct list_head *config_terms) int parse_events_add_pmu(struct parse_events_state *parse_state, struct list_head *list, const char *name, - struct list_head *head_config, + const struct list_head *const_head_terms, bool auto_merge_stats, void *loc_) { struct perf_event_attr attr; @@ -1377,6 +1378,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, struct parse_events_error *err = parse_state->error; YYLTYPE *loc = loc_; LIST_HEAD(config_terms); + LIST_HEAD(head_terms); pmu = parse_state->fake_pmu ?: perf_pmus__find(name); @@ -1390,32 +1392,37 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, return -EINVAL; } + if (const_head_terms) { + int ret = parse_events_terms__copy(const_head_terms, &head_terms); + + if (ret) + return ret; + } + if (verbose > 1) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - if (pmu->selectable && !head_config) { + if (pmu->selectable && list_empty(&head_terms)) { strbuf_addf(&sb, "%s//", name); } else { strbuf_addf(&sb, "%s/", name); - parse_events_term__to_strbuf(head_config, &sb); + parse_events_term__to_strbuf(&head_terms, &sb); strbuf_addch(&sb, '/'); } fprintf(stderr, "Attempt to add: %s\n", sb.buf); strbuf_release(&sb); } - if (head_config) - fix_raw(head_config, pmu); + fix_raw(&head_terms, pmu); if (pmu->default_config) { - memcpy(&attr, pmu->default_config, - sizeof(struct perf_event_attr)); + memcpy(&attr, pmu->default_config, sizeof(struct perf_event_attr)); } else { memset(&attr, 0, sizeof(attr)); } attr.type = pmu->type; - if (!head_config) { + if (list_empty(&head_terms)) { evsel = __add_event(list, &parse_state->idx, &attr, /*init_attr=*/true, /*name=*/NULL, /*metric_id=*/NULL, pmu, @@ -1424,14 +1431,16 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, return evsel ? 0 : -ENOMEM; } - if (!parse_state->fake_pmu && perf_pmu__check_alias(pmu, head_config, &info, err)) + if (!parse_state->fake_pmu && perf_pmu__check_alias(pmu, &head_terms, &info, err)) { + parse_events_terms__purge(&head_terms); return -EINVAL; + } if (verbose > 1) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(head_config, &sb); + parse_events_term__to_strbuf(&head_terms, &sb); fprintf(stderr, "..after resolving event: %s/%s/\n", name, sb.buf); strbuf_release(&sb); } @@ -1440,39 +1449,52 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, * Configure hardcoded terms first, no need to check * return value when called with fail == 0 ;) */ - if (config_attr(&attr, head_config, parse_state->error, config_term_pmu)) + if (config_attr(&attr, &head_terms, parse_state->error, config_term_pmu)) { + parse_events_terms__purge(&head_terms); return -EINVAL; + } - if (get_config_terms(head_config, &config_terms)) + if (get_config_terms(&head_terms, &config_terms)) { + parse_events_terms__purge(&head_terms); return -ENOMEM; + } /* * When using default config, record which bits of attr->config were * changed by the user. */ - if (pmu->default_config && get_config_chgs(pmu, head_config, &config_terms)) + if (pmu->default_config && get_config_chgs(pmu, &head_terms, &config_terms)) { + parse_events_terms__purge(&head_terms); return -ENOMEM; + } - if (!parse_state->fake_pmu && perf_pmu__config(pmu, &attr, head_config, parse_state->error)) { + if (!parse_state->fake_pmu && + perf_pmu__config(pmu, &attr, &head_terms, parse_state->error)) { free_config_terms(&config_terms); + parse_events_terms__purge(&head_terms); return -EINVAL; } evsel = __add_event(list, &parse_state->idx, &attr, /*init_attr=*/true, - get_config_name(head_config), - get_config_metric_id(head_config), pmu, + get_config_name(&head_terms), + get_config_metric_id(&head_terms), pmu, &config_terms, auto_merge_stats, /*cpu_list=*/NULL); - if (!evsel) + if (!evsel) { + parse_events_terms__purge(&head_terms); return -ENOMEM; + } if (evsel->name) evsel->use_config_name = true; evsel->percore = config_term_percore(&evsel->config_terms); - if (parse_state->fake_pmu) + if (parse_state->fake_pmu) { + parse_events_terms__purge(&head_terms); return 0; + } + parse_events_terms__purge(&head_terms); free((char *)evsel->unit); evsel->unit = strdup(info.unit); evsel->scale = info.scale; @@ -1482,25 +1504,25 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, } int parse_events_multi_pmu_add(struct parse_events_state *parse_state, - const char *event_name, struct list_head *head, + const char *event_name, + const struct list_head *const_head_terms, struct list_head **listp, void *loc_) { struct parse_events_term *term; struct list_head *list = NULL; - struct list_head *orig_head = NULL; struct perf_pmu *pmu = NULL; YYLTYPE *loc = loc_; int ok = 0; const char *config; + LIST_HEAD(head_terms); *listp = NULL; - if (!head) { - head = malloc(sizeof(struct list_head)); - if (!head) - goto out_err; + if (const_head_terms) { + int ret = parse_events_terms__copy(const_head_terms, &head_terms); - INIT_LIST_HEAD(head); + if (ret) + return ret; } config = strdup(event_name); @@ -1514,7 +1536,7 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, zfree(&config); goto out_err; } - list_add_tail(&term->list, head); + list_add_tail(&term->list, &head_terms); /* Add it for all PMUs that support the alias */ list = malloc(sizeof(struct list_head)); @@ -1533,27 +1555,25 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, continue; auto_merge_stats = perf_pmu__auto_merge_stats(pmu); - parse_events_copy_term_list(head, &orig_head); if (!parse_events_add_pmu(parse_state, list, pmu->name, - orig_head, auto_merge_stats, loc)) { + &head_terms, auto_merge_stats, loc)) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(orig_head, &sb); + parse_events_term__to_strbuf(&head_terms, &sb); pr_debug("%s -> %s/%s/\n", event_name, pmu->name, sb.buf); strbuf_release(&sb); ok++; } - parse_events_terms__delete(orig_head); } if (parse_state->fake_pmu) { - if (!parse_events_add_pmu(parse_state, list, event_name, head, + if (!parse_events_add_pmu(parse_state, list, event_name, &head_terms, /*auto_merge_stats=*/true, loc)) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(head, &sb); + parse_events_term__to_strbuf(&head_terms, &sb); pr_debug("%s -> %s/%s/\n", event_name, "fake_pmu", sb.buf); strbuf_release(&sb); ok++; @@ -1561,12 +1581,12 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, } out_err: + parse_events_terms__purge(&head_terms); if (ok) *listp = list; else free(list); - parse_events_terms__delete(head); return ok ? 0 : -1; } @@ -2543,27 +2563,19 @@ void parse_events_term__delete(struct parse_events_term *term) free(term); } -int parse_events_copy_term_list(struct list_head *old, - struct list_head **new) +static int parse_events_terms__copy(const struct list_head *src, struct list_head *dest) { - struct parse_events_term *term, *n; - int ret; - - if (!old) { - *new = NULL; - return 0; - } + struct parse_events_term *term; - *new = malloc(sizeof(struct list_head)); - if (!*new) - return -ENOMEM; - INIT_LIST_HEAD(*new); + list_for_each_entry (term, src, list) { + struct parse_events_term *n; + int ret; - list_for_each_entry (term, old, list) { ret = parse_events_term__clone(&n, term); if (ret) return ret; - list_add_tail(&n->list, *new); + + list_add_tail(&n->list, dest); } return 0; } diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index 36a67ef7b35a..e6612856e881 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -209,7 +209,7 @@ int parse_events_add_breakpoint(struct parse_events_state *parse_state, struct list_head *head_config); int parse_events_add_pmu(struct parse_events_state *parse_state, struct list_head *list, const char *name, - struct list_head *head_config, + const struct list_head *const_head_terms, bool auto_merge_stats, void *loc); struct evsel *parse_events__add_event(int idx, struct perf_event_attr *attr, @@ -218,12 +218,9 @@ struct evsel *parse_events__add_event(int idx, struct perf_event_attr *attr, int parse_events_multi_pmu_add(struct parse_events_state *parse_state, const char *event_name, - struct list_head *head_config, + const struct list_head *head_terms, struct list_head **listp, void *loc); -int parse_events_copy_term_list(struct list_head *old, - struct list_head **new); - void parse_events__set_leader(char *name, struct list_head *list); void parse_events_update_lists(struct list_head *list_event, struct list_head *list_all); diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index 286ead08065e..4d7947fb4a98 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -274,23 +274,18 @@ event_pmu: PE_NAME opt_pmu_config { struct parse_events_state *parse_state = _parse_state; - struct list_head *list = NULL, *orig_terms = NULL, *terms= NULL; + /* List of created evsels. */ + struct list_head *list = NULL; char *pattern = NULL; #define CLEANUP \ do { \ parse_events_terms__delete($2); \ - parse_events_terms__delete(orig_terms); \ free(list); \ free($1); \ free(pattern); \ } while(0) - if (parse_events_copy_term_list($2, &orig_terms)) { - CLEANUP; - YYNOMEM; - } - list = alloc_list(); if (!list) { CLEANUP; @@ -320,16 +315,11 @@ PE_NAME opt_pmu_config !perf_pmu__match(pattern, pmu->alias_name, $1)) { bool auto_merge_stats = perf_pmu__auto_merge_stats(pmu); - if (parse_events_copy_term_list(orig_terms, &terms)) { - CLEANUP; - YYNOMEM; - } - if (!parse_events_add_pmu(parse_state, list, pmu->name, terms, + if (!parse_events_add_pmu(parse_state, list, pmu->name, $2, auto_merge_stats, &@1)) { ok++; parse_state->wild_card_pmus = true; } - parse_events_terms__delete(terms); } } @@ -337,7 +327,6 @@ PE_NAME opt_pmu_config /* Failure to add, assume $1 is an event name. */ zfree(&list); ok = !parse_events_multi_pmu_add(parse_state, $1, $2, &list, &@1); - $2 = NULL; } if (!ok) { struct parse_events_error *error = parse_state->error; -- cgit v1.2.3 From 0d3f0e6f94ef58d5532e23b6d153b0890cf0014c Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 1 Sep 2023 16:39:49 -0700 Subject: perf parse-events: Introduce 'struct parse_events_terms' parse_events_terms() existed in function names but was passed a 'struct list_head'. As many parse_events functions take an evsel_config list as well as a parse_event_term list, and the naming head_terms and head_config is inconsistent, there's a potential to switch the lists and get errors. Introduce a 'struct parse_events_terms', that just wraps a list_head, to avoid this. Add the regular init/exit functions and transition the code to use them. Reviewed-by: James Clark Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Rob Herring Link: https://lore.kernel.org/r/20230901233949.2930562-6-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.c | 158 +++++++++++++++++++++-------------------- tools/perf/util/parse-events.h | 29 +++++--- tools/perf/util/parse-events.y | 12 ++-- tools/perf/util/pmu.c | 49 ++++++------- tools/perf/util/pmu.h | 6 +- 5 files changed, 134 insertions(+), 120 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 06a844bcce4a..c56e07bd7dd6 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -34,8 +34,9 @@ #ifdef PARSER_DEBUG extern int parse_events_debug; #endif -static int get_config_terms(struct list_head *head_config, struct list_head *head_terms); -static int parse_events_terms__copy(const struct list_head *src, struct list_head *dest); +static int get_config_terms(struct parse_events_terms *head_config, struct list_head *head_terms); +static int parse_events_terms__copy(const struct parse_events_terms *src, + struct parse_events_terms *dest); struct event_symbol event_symbols_hw[PERF_COUNT_HW_MAX] = { [PERF_COUNT_HW_CPU_CYCLES] = { @@ -153,26 +154,27 @@ const char *event_type(int type) return "unknown"; } -static char *get_config_str(struct list_head *head_terms, enum parse_events__term_type type_term) +static char *get_config_str(struct parse_events_terms *head_terms, + enum parse_events__term_type type_term) { struct parse_events_term *term; if (!head_terms) return NULL; - list_for_each_entry(term, head_terms, list) + list_for_each_entry(term, &head_terms->terms, list) if (term->type_term == type_term) return term->val.str; return NULL; } -static char *get_config_metric_id(struct list_head *head_terms) +static char *get_config_metric_id(struct parse_events_terms *head_terms) { return get_config_str(head_terms, PARSE_EVENTS__TERM_TYPE_METRIC_ID); } -static char *get_config_name(struct list_head *head_terms) +static char *get_config_name(struct parse_events_terms *head_terms) { return get_config_str(head_terms, PARSE_EVENTS__TERM_TYPE_NAME); } @@ -188,11 +190,11 @@ static char *get_config_name(struct list_head *head_terms) * @config_terms: the list of terms that may contain a raw term. * @pmu: the PMU to scan for events from. */ -static void fix_raw(struct list_head *config_terms, struct perf_pmu *pmu) +static void fix_raw(struct parse_events_terms *config_terms, struct perf_pmu *pmu) { struct parse_events_term *term; - list_for_each_entry(term, config_terms, list) { + list_for_each_entry(term, &config_terms->terms, list) { u64 num; if (term->type_term != PARSE_EVENTS__TERM_TYPE_RAW) @@ -356,7 +358,7 @@ static int config_term_common(struct perf_event_attr *attr, struct parse_events_term *term, struct parse_events_error *err); static int config_attr(struct perf_event_attr *attr, - struct list_head *head, + struct parse_events_terms *head, struct parse_events_error *err, config_term_func_t config_term); @@ -442,7 +444,7 @@ bool parse_events__filter_pmu(const struct parse_events_state *parse_state, int parse_events_add_cache(struct list_head *list, int *idx, const char *name, struct parse_events_state *parse_state, - struct list_head *head_config) + struct parse_events_terms *head_config) { struct perf_pmu *pmu = NULL; bool found_supported = false; @@ -520,7 +522,7 @@ static void tracepoint_error(struct parse_events_error *e, int err, static int add_tracepoint(struct list_head *list, int *idx, const char *sys_name, const char *evt_name, struct parse_events_error *err, - struct list_head *head_config, void *loc_) + struct parse_events_terms *head_config, void *loc_) { YYLTYPE *loc = loc_; struct evsel *evsel = evsel__newtp_idx(sys_name, evt_name, (*idx)++); @@ -545,7 +547,7 @@ static int add_tracepoint(struct list_head *list, int *idx, static int add_tracepoint_multi_event(struct list_head *list, int *idx, const char *sys_name, const char *evt_name, struct parse_events_error *err, - struct list_head *head_config, YYLTYPE *loc) + struct parse_events_terms *head_config, YYLTYPE *loc) { char *evt_path; struct dirent *evt_ent; @@ -593,7 +595,7 @@ static int add_tracepoint_multi_event(struct list_head *list, int *idx, static int add_tracepoint_event(struct list_head *list, int *idx, const char *sys_name, const char *evt_name, struct parse_events_error *err, - struct list_head *head_config, YYLTYPE *loc) + struct parse_events_terms *head_config, YYLTYPE *loc) { return strpbrk(evt_name, "*?") ? add_tracepoint_multi_event(list, idx, sys_name, evt_name, @@ -605,7 +607,7 @@ static int add_tracepoint_event(struct list_head *list, int *idx, static int add_tracepoint_multi_sys(struct list_head *list, int *idx, const char *sys_name, const char *evt_name, struct parse_events_error *err, - struct list_head *head_config, YYLTYPE *loc) + struct parse_events_terms *head_config, YYLTYPE *loc) { struct dirent *events_ent; DIR *events_dir; @@ -680,7 +682,7 @@ do { \ int parse_events_add_breakpoint(struct parse_events_state *parse_state, struct list_head *list, u64 addr, char *type, u64 len, - struct list_head *head_config __maybe_unused) + struct parse_events_terms *head_config) { struct perf_event_attr attr; LIST_HEAD(config_terms); @@ -1066,20 +1068,20 @@ static int config_term_tracepoint(struct perf_event_attr *attr, #endif static int config_attr(struct perf_event_attr *attr, - struct list_head *head, + struct parse_events_terms *head, struct parse_events_error *err, config_term_func_t config_term) { struct parse_events_term *term; - list_for_each_entry(term, head, list) + list_for_each_entry(term, &head->terms, list) if (config_term(attr, term, err)) return -EINVAL; return 0; } -static int get_config_terms(struct list_head *head_config, struct list_head *head_terms) +static int get_config_terms(struct parse_events_terms *head_config, struct list_head *head_terms) { #define ADD_CONFIG_TERM(__type, __weak) \ struct evsel_config_term *__t; \ @@ -1112,7 +1114,7 @@ do { \ struct parse_events_term *term; - list_for_each_entry(term, head_config, list) { + list_for_each_entry(term, &head_config->terms, list) { switch (term->type_term) { case PARSE_EVENTS__TERM_TYPE_SAMPLE_PERIOD: ADD_CONFIG_TERM_VAL(PERIOD, period, term->val.num, term->weak); @@ -1193,14 +1195,14 @@ do { \ * Add EVSEL__CONFIG_TERM_CFG_CHG where cfg_chg will have a bit set for * each bit of attr->config that the user has changed. */ -static int get_config_chgs(struct perf_pmu *pmu, struct list_head *head_config, +static int get_config_chgs(struct perf_pmu *pmu, struct parse_events_terms *head_config, struct list_head *head_terms) { struct parse_events_term *term; u64 bits = 0; int type; - list_for_each_entry(term, head_config, list) { + list_for_each_entry(term, &head_config->terms, list) { switch (term->type_term) { case PARSE_EVENTS__TERM_TYPE_USER: type = perf_pmu__format_type(pmu, term->config); @@ -1250,7 +1252,7 @@ static int get_config_chgs(struct perf_pmu *pmu, struct list_head *head_config, int parse_events_add_tracepoint(struct list_head *list, int *idx, const char *sys, const char *event, struct parse_events_error *err, - struct list_head *head_config, void *loc_) + struct parse_events_terms *head_config, void *loc_) { YYLTYPE *loc = loc_; #ifdef HAVE_LIBTRACEEVENT @@ -1283,7 +1285,7 @@ int parse_events_add_tracepoint(struct list_head *list, int *idx, static int __parse_events_add_numeric(struct parse_events_state *parse_state, struct list_head *list, struct perf_pmu *pmu, u32 type, u32 extended_type, - u64 config, struct list_head *head_config) + u64 config, struct parse_events_terms *head_config) { struct perf_event_attr attr; LIST_HEAD(config_terms); @@ -1319,7 +1321,7 @@ static int __parse_events_add_numeric(struct parse_events_state *parse_state, int parse_events_add_numeric(struct parse_events_state *parse_state, struct list_head *list, u32 type, u64 config, - struct list_head *head_config, + struct parse_events_terms *head_config, bool wildcard) { struct perf_pmu *pmu = NULL; @@ -1368,7 +1370,7 @@ static bool config_term_percore(struct list_head *config_terms) int parse_events_add_pmu(struct parse_events_state *parse_state, struct list_head *list, const char *name, - const struct list_head *const_head_terms, + const struct parse_events_terms *const_parsed_terms, bool auto_merge_stats, void *loc_) { struct perf_event_attr attr; @@ -1378,7 +1380,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, struct parse_events_error *err = parse_state->error; YYLTYPE *loc = loc_; LIST_HEAD(config_terms); - LIST_HEAD(head_terms); + struct parse_events_terms parsed_terms; pmu = parse_state->fake_pmu ?: perf_pmus__find(name); @@ -1392,8 +1394,9 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, return -EINVAL; } - if (const_head_terms) { - int ret = parse_events_terms__copy(const_head_terms, &head_terms); + parse_events_terms__init(&parsed_terms); + if (const_parsed_terms) { + int ret = parse_events_terms__copy(const_parsed_terms, &parsed_terms); if (ret) return ret; @@ -1403,17 +1406,17 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - if (pmu->selectable && list_empty(&head_terms)) { + if (pmu->selectable && list_empty(&parsed_terms.terms)) { strbuf_addf(&sb, "%s//", name); } else { strbuf_addf(&sb, "%s/", name); - parse_events_term__to_strbuf(&head_terms, &sb); + parse_events_terms__to_strbuf(&parsed_terms, &sb); strbuf_addch(&sb, '/'); } fprintf(stderr, "Attempt to add: %s\n", sb.buf); strbuf_release(&sb); } - fix_raw(&head_terms, pmu); + fix_raw(&parsed_terms, pmu); if (pmu->default_config) { memcpy(&attr, pmu->default_config, sizeof(struct perf_event_attr)); @@ -1422,7 +1425,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, } attr.type = pmu->type; - if (list_empty(&head_terms)) { + if (list_empty(&parsed_terms.terms)) { evsel = __add_event(list, &parse_state->idx, &attr, /*init_attr=*/true, /*name=*/NULL, /*metric_id=*/NULL, pmu, @@ -1431,8 +1434,8 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, return evsel ? 0 : -ENOMEM; } - if (!parse_state->fake_pmu && perf_pmu__check_alias(pmu, &head_terms, &info, err)) { - parse_events_terms__purge(&head_terms); + if (!parse_state->fake_pmu && perf_pmu__check_alias(pmu, &parsed_terms, &info, err)) { + parse_events_terms__exit(&parsed_terms); return -EINVAL; } @@ -1440,7 +1443,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(&head_terms, &sb); + parse_events_terms__to_strbuf(&parsed_terms, &sb); fprintf(stderr, "..after resolving event: %s/%s/\n", name, sb.buf); strbuf_release(&sb); } @@ -1449,13 +1452,13 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, * Configure hardcoded terms first, no need to check * return value when called with fail == 0 ;) */ - if (config_attr(&attr, &head_terms, parse_state->error, config_term_pmu)) { - parse_events_terms__purge(&head_terms); + if (config_attr(&attr, &parsed_terms, parse_state->error, config_term_pmu)) { + parse_events_terms__exit(&parsed_terms); return -EINVAL; } - if (get_config_terms(&head_terms, &config_terms)) { - parse_events_terms__purge(&head_terms); + if (get_config_terms(&parsed_terms, &config_terms)) { + parse_events_terms__exit(&parsed_terms); return -ENOMEM; } @@ -1463,24 +1466,24 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, * When using default config, record which bits of attr->config were * changed by the user. */ - if (pmu->default_config && get_config_chgs(pmu, &head_terms, &config_terms)) { - parse_events_terms__purge(&head_terms); + if (pmu->default_config && get_config_chgs(pmu, &parsed_terms, &config_terms)) { + parse_events_terms__exit(&parsed_terms); return -ENOMEM; } if (!parse_state->fake_pmu && - perf_pmu__config(pmu, &attr, &head_terms, parse_state->error)) { + perf_pmu__config(pmu, &attr, &parsed_terms, parse_state->error)) { free_config_terms(&config_terms); - parse_events_terms__purge(&head_terms); + parse_events_terms__exit(&parsed_terms); return -EINVAL; } evsel = __add_event(list, &parse_state->idx, &attr, /*init_attr=*/true, - get_config_name(&head_terms), - get_config_metric_id(&head_terms), pmu, + get_config_name(&parsed_terms), + get_config_metric_id(&parsed_terms), pmu, &config_terms, auto_merge_stats, /*cpu_list=*/NULL); if (!evsel) { - parse_events_terms__purge(&head_terms); + parse_events_terms__exit(&parsed_terms); return -ENOMEM; } @@ -1490,11 +1493,11 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, evsel->percore = config_term_percore(&evsel->config_terms); if (parse_state->fake_pmu) { - parse_events_terms__purge(&head_terms); + parse_events_terms__exit(&parsed_terms); return 0; } - parse_events_terms__purge(&head_terms); + parse_events_terms__exit(&parsed_terms); free((char *)evsel->unit); evsel->unit = strdup(info.unit); evsel->scale = info.scale; @@ -1505,7 +1508,7 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, int parse_events_multi_pmu_add(struct parse_events_state *parse_state, const char *event_name, - const struct list_head *const_head_terms, + const struct parse_events_terms *const_parsed_terms, struct list_head **listp, void *loc_) { struct parse_events_term *term; @@ -1514,12 +1517,13 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, YYLTYPE *loc = loc_; int ok = 0; const char *config; - LIST_HEAD(head_terms); + struct parse_events_terms parsed_terms; *listp = NULL; - if (const_head_terms) { - int ret = parse_events_terms__copy(const_head_terms, &head_terms); + parse_events_terms__init(&parsed_terms); + if (const_parsed_terms) { + int ret = parse_events_terms__copy(const_parsed_terms, &parsed_terms); if (ret) return ret; @@ -1536,7 +1540,7 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, zfree(&config); goto out_err; } - list_add_tail(&term->list, &head_terms); + list_add_tail(&term->list, &parsed_terms.terms); /* Add it for all PMUs that support the alias */ list = malloc(sizeof(struct list_head)); @@ -1556,11 +1560,11 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, auto_merge_stats = perf_pmu__auto_merge_stats(pmu); if (!parse_events_add_pmu(parse_state, list, pmu->name, - &head_terms, auto_merge_stats, loc)) { + &parsed_terms, auto_merge_stats, loc)) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(&head_terms, &sb); + parse_events_terms__to_strbuf(&parsed_terms, &sb); pr_debug("%s -> %s/%s/\n", event_name, pmu->name, sb.buf); strbuf_release(&sb); ok++; @@ -1568,12 +1572,12 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, } if (parse_state->fake_pmu) { - if (!parse_events_add_pmu(parse_state, list, event_name, &head_terms, + if (!parse_events_add_pmu(parse_state, list, event_name, &parsed_terms, /*auto_merge_stats=*/true, loc)) { struct strbuf sb; strbuf_init(&sb, /*hint=*/ 0); - parse_events_term__to_strbuf(&head_terms, &sb); + parse_events_terms__to_strbuf(&parsed_terms, &sb); pr_debug("%s -> %s/%s/\n", event_name, "fake_pmu", sb.buf); strbuf_release(&sb); ok++; @@ -1581,7 +1585,7 @@ int parse_events_multi_pmu_add(struct parse_events_state *parse_state, } out_err: - parse_events_terms__purge(&head_terms); + parse_events_terms__exit(&parsed_terms); if (ok) *listp = list; else @@ -1851,7 +1855,7 @@ static int parse_events__scanner(const char *str, /* * parse event config string, return a list of event terms. */ -int parse_events_terms(struct list_head *terms, const char *str, FILE *input) +int parse_events_terms(struct parse_events_terms *terms, const char *str, FILE *input) { struct parse_events_state parse_state = { .terms = NULL, @@ -1860,14 +1864,10 @@ int parse_events_terms(struct list_head *terms, const char *str, FILE *input) int ret; ret = parse_events__scanner(str, input, &parse_state); + if (!ret) + list_splice(&parse_state.terms->terms, &terms->terms); - if (!ret) { - list_splice(parse_state.terms, terms); - zfree(&parse_state.terms); - return 0; - } - - parse_events_terms__delete(parse_state.terms); + zfree(&parse_state.terms); return ret; } @@ -2563,11 +2563,12 @@ void parse_events_term__delete(struct parse_events_term *term) free(term); } -static int parse_events_terms__copy(const struct list_head *src, struct list_head *dest) +static int parse_events_terms__copy(const struct parse_events_terms *src, + struct parse_events_terms *dest) { struct parse_events_term *term; - list_for_each_entry (term, src, list) { + list_for_each_entry (term, &src->terms, list) { struct parse_events_term *n; int ret; @@ -2575,38 +2576,43 @@ static int parse_events_terms__copy(const struct list_head *src, struct list_hea if (ret) return ret; - list_add_tail(&n->list, dest); + list_add_tail(&n->list, &dest->terms); } return 0; } -void parse_events_terms__purge(struct list_head *terms) +void parse_events_terms__init(struct parse_events_terms *terms) +{ + INIT_LIST_HEAD(&terms->terms); +} + +void parse_events_terms__exit(struct parse_events_terms *terms) { struct parse_events_term *term, *h; - list_for_each_entry_safe(term, h, terms, list) { + list_for_each_entry_safe(term, h, &terms->terms, list) { list_del_init(&term->list); parse_events_term__delete(term); } } -void parse_events_terms__delete(struct list_head *terms) +void parse_events_terms__delete(struct parse_events_terms *terms) { if (!terms) return; - parse_events_terms__purge(terms); + parse_events_terms__exit(terms); free(terms); } -int parse_events_term__to_strbuf(struct list_head *term_list, struct strbuf *sb) +int parse_events_terms__to_strbuf(const struct parse_events_terms *terms, struct strbuf *sb) { struct parse_events_term *term; bool first = true; - if (!term_list) + if (!terms) return 0; - list_for_each_entry(term, term_list, list) { + list_for_each_entry(term, &terms->terms, list) { int ret; if (!first) { diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index e6612856e881..63c0a36a4bf1 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -44,7 +44,6 @@ static inline int parse_events(struct evlist *evlist, const char *str, int parse_event(struct evlist *evlist, const char *str); -int parse_events_terms(struct list_head *terms, const char *str, FILE *input); int parse_filter(const struct option *opt, const char *str, int unset); int exclude_perf(const struct option *opt, const char *arg, int unset); @@ -140,6 +139,11 @@ struct parse_events_error { char *first_help; }; +/* A wrapper around a list of terms for the sake of better type safety. */ +struct parse_events_terms { + struct list_head terms; +}; + struct parse_events_state { /* The list parsed events are placed on. */ struct list_head list; @@ -148,7 +152,7 @@ struct parse_events_state { /* Error information. */ struct parse_events_error *error; /* Holds returned terms for term parsing. */ - struct list_head *terms; + struct parse_events_terms *terms; /* Start token. */ int stoken; /* Special fake PMU marker for testing. */ @@ -181,35 +185,38 @@ int parse_events_term__term(struct parse_events_term **term, int parse_events_term__clone(struct parse_events_term **new, struct parse_events_term *term); void parse_events_term__delete(struct parse_events_term *term); -void parse_events_terms__delete(struct list_head *terms); -void parse_events_terms__purge(struct list_head *terms); -int parse_events_term__to_strbuf(struct list_head *term_list, struct strbuf *sb); + +void parse_events_terms__delete(struct parse_events_terms *terms); +void parse_events_terms__init(struct parse_events_terms *terms); +void parse_events_terms__exit(struct parse_events_terms *terms); +int parse_events_terms(struct parse_events_terms *terms, const char *str, FILE *input); +int parse_events_terms__to_strbuf(const struct parse_events_terms *terms, struct strbuf *sb); int parse_events__modifier_event(struct list_head *list, char *str, bool add); int parse_events__modifier_group(struct list_head *list, char *event_mod); int parse_events_name(struct list_head *list, const char *name); int parse_events_add_tracepoint(struct list_head *list, int *idx, const char *sys, const char *event, struct parse_events_error *error, - struct list_head *head_config, void *loc); + struct parse_events_terms *head_config, void *loc); int parse_events_add_numeric(struct parse_events_state *parse_state, struct list_head *list, u32 type, u64 config, - struct list_head *head_config, + struct parse_events_terms *head_config, bool wildcard); int parse_events_add_tool(struct parse_events_state *parse_state, struct list_head *list, int tool_event); int parse_events_add_cache(struct list_head *list, int *idx, const char *name, struct parse_events_state *parse_state, - struct list_head *head_config); + struct parse_events_terms *head_config); int parse_events__decode_legacy_cache(const char *name, int pmu_type, __u64 *config); int parse_events_add_breakpoint(struct parse_events_state *parse_state, struct list_head *list, u64 addr, char *type, u64 len, - struct list_head *head_config); + struct parse_events_terms *head_config); int parse_events_add_pmu(struct parse_events_state *parse_state, struct list_head *list, const char *name, - const struct list_head *const_head_terms, + const struct parse_events_terms *const_parsed_terms, bool auto_merge_stats, void *loc); struct evsel *parse_events__add_event(int idx, struct perf_event_attr *attr, @@ -218,7 +225,7 @@ struct evsel *parse_events__add_event(int idx, struct perf_event_attr *attr, int parse_events_multi_pmu_add(struct parse_events_state *parse_state, const char *event_name, - const struct list_head *head_terms, + const struct parse_events_terms *const_parsed_terms, struct list_head **listp, void *loc); void parse_events__set_leader(char *name, struct list_head *list); diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index 4d7947fb4a98..bf74d6564eba 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -113,7 +113,7 @@ static void free_list_evsel(struct list_head* list_evsel) u64 num; enum parse_events__term_type term_type; struct list_head *list_evsel; - struct list_head *list_terms; + struct parse_events_terms *list_terms; struct parse_events_term *term; struct tracepoint_name { char *sys; @@ -644,26 +644,26 @@ start_terms: event_config event_config: event_config ',' event_term { - struct list_head *head = $1; + struct parse_events_terms *head = $1; struct parse_events_term *term = $3; if (!head) { parse_events_term__delete(term); YYABORT; } - list_add_tail(&term->list, head); + list_add_tail(&term->list, &head->terms); $$ = $1; } | event_term { - struct list_head *head = malloc(sizeof(*head)); + struct parse_events_terms *head = malloc(sizeof(*head)); struct parse_events_term *term = $1; if (!head) YYNOMEM; - INIT_LIST_HEAD(head); - list_add_tail(&term->list, head); + parse_events_terms__init(head); + list_add_tail(&term->list, &head->terms); $$ = head; } diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index d85602aa4b9f..e2159854ab26 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -52,7 +52,7 @@ struct perf_pmu_alias { */ char *topic; /** @terms: Owned list of the original parsed parameters. */ - struct list_head terms; + struct parse_events_terms terms; /** @list: List element of struct perf_pmu aliases. */ struct list_head list; /** @@ -404,7 +404,7 @@ static void perf_pmu_free_alias(struct perf_pmu_alias *newalias) zfree(&newalias->long_desc); zfree(&newalias->topic); zfree(&newalias->pmu_name); - parse_events_terms__purge(&newalias->terms); + parse_events_terms__exit(&newalias->terms); free(newalias); } @@ -484,7 +484,7 @@ static int update_alias(const struct pmu_event *pe, assign_str(pe->name, "topic", &data->alias->topic, pe->topic); data->alias->per_pkg = pe->perpkg; if (pe->event) { - parse_events_terms__purge(&data->alias->terms); + parse_events_terms__exit(&data->alias->terms); ret = parse_events_terms(&data->alias->terms, pe->event, /*input=*/NULL); } if (!ret && pe->unit) { @@ -524,7 +524,7 @@ static int perf_pmu__new_alias(struct perf_pmu *pmu, const char *name, if (!alias) return -ENOMEM; - INIT_LIST_HEAD(&alias->terms); + parse_events_terms__init(&alias->terms); alias->scale = 1.0; alias->unit[0] = '\0'; alias->per_pkg = perpkg; @@ -656,17 +656,17 @@ static int pmu_aliases_parse(struct perf_pmu *pmu) return 0; } -static int pmu_alias_terms(struct perf_pmu_alias *alias, - struct list_head *terms) +static int pmu_alias_terms(struct perf_pmu_alias *alias, struct list_head *terms) { struct parse_events_term *term, *cloned; - LIST_HEAD(list); - int ret; + struct parse_events_terms clone_terms; + + parse_events_terms__init(&clone_terms); + list_for_each_entry(term, &alias->terms.terms, list) { + int ret = parse_events_term__clone(&cloned, term); - list_for_each_entry(term, &alias->terms, list) { - ret = parse_events_term__clone(&cloned, term); if (ret) { - parse_events_terms__purge(&list); + parse_events_terms__exit(&clone_terms); return ret; } /* @@ -674,9 +674,10 @@ static int pmu_alias_terms(struct perf_pmu_alias *alias, * which we don't want for implicit terms in aliases. */ cloned->weak = true; - list_add_tail(&cloned->list, &list); + list_add_tail(&cloned->list, &clone_terms.terms); } - list_splice(&list, terms); + list_splice_init(&clone_terms.terms, terms); + parse_events_terms__exit(&clone_terms); return 0; } @@ -1188,12 +1189,12 @@ static __u64 pmu_format_max_value(const unsigned long *format) * in a config string) later on in the term list. */ static int pmu_resolve_param_term(struct parse_events_term *term, - struct list_head *head_terms, + struct parse_events_terms *head_terms, __u64 *value) { struct parse_events_term *t; - list_for_each_entry(t, head_terms, list) { + list_for_each_entry(t, &head_terms->terms, list) { if (t->type_val == PARSE_EVENTS__TERM_TYPE_NUM && t->config && !strcmp(t->config, term->config)) { t->used = true; @@ -1237,7 +1238,7 @@ error: static int pmu_config_term(struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_term *term, - struct list_head *head_terms, + struct parse_events_terms *head_terms, bool zero, struct parse_events_error *err) { struct perf_pmu_format *format; @@ -1359,13 +1360,13 @@ static int pmu_config_term(struct perf_pmu *pmu, int perf_pmu__config_terms(struct perf_pmu *pmu, struct perf_event_attr *attr, - struct list_head *head_terms, + struct parse_events_terms *terms, bool zero, struct parse_events_error *err) { struct parse_events_term *term; - list_for_each_entry(term, head_terms, list) { - if (pmu_config_term(pmu, attr, term, head_terms, zero, err)) + list_for_each_entry(term, &terms->terms, list) { + if (pmu_config_term(pmu, attr, term, terms, zero, err)) return -EINVAL; } @@ -1378,7 +1379,7 @@ int perf_pmu__config_terms(struct perf_pmu *pmu, * 2) pmu format definitions - specified by pmu parameter */ int perf_pmu__config(struct perf_pmu *pmu, struct perf_event_attr *attr, - struct list_head *head_terms, + struct parse_events_terms *head_terms, struct parse_events_error *err) { bool zero = !!pmu->default_config; @@ -1472,7 +1473,7 @@ static int check_info_data(struct perf_pmu *pmu, * Find alias in the terms list and replace it with the terms * defined for the alias */ -int perf_pmu__check_alias(struct perf_pmu *pmu, struct list_head *head_terms, +int perf_pmu__check_alias(struct perf_pmu *pmu, struct parse_events_terms *head_terms, struct perf_pmu_info *info, struct parse_events_error *err) { struct parse_events_term *term, *h; @@ -1489,7 +1490,7 @@ int perf_pmu__check_alias(struct perf_pmu *pmu, struct list_head *head_terms, info->scale = 0.0; info->snapshot = false; - list_for_each_entry_safe(term, h, head_terms, list) { + list_for_each_entry_safe(term, h, &head_terms->terms, list) { alias = pmu_find_alias(pmu, term); if (!alias) continue; @@ -1634,7 +1635,7 @@ static char *format_alias(char *buf, int len, const struct perf_pmu *pmu, : (int)strlen(pmu->name); int used = snprintf(buf, len, "%.*s/%s", pmu_name_len, pmu->name, alias->name); - list_for_each_entry(term, &alias->terms, list) { + list_for_each_entry(term, &alias->terms.terms, list) { if (term->type_val == PARSE_EVENTS__TERM_TYPE_STR) used += snprintf(buf + used, sub_non_neg(len, used), ",%s=%s", term->config, @@ -1693,7 +1694,7 @@ int perf_pmu__for_each_event(struct perf_pmu *pmu, bool skip_duplicate_pmus, info.desc = event->desc; info.long_desc = event->long_desc; info.encoding_desc = buf + buf_used; - parse_events_term__to_strbuf(&event->terms, &sb); + parse_events_terms__to_strbuf(&event->terms, &sb); buf_used += snprintf(buf + buf_used, sizeof(buf) - buf_used, "%s/%s/", info.pmu_name, sb.buf) + 1; info.topic = event->topic; diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 6a4e170c61d6..bd5d804a6736 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -191,15 +191,15 @@ typedef int (*pmu_event_callback)(void *state, struct pmu_event_info *info); void pmu_add_sys_aliases(struct perf_pmu *pmu); int perf_pmu__config(struct perf_pmu *pmu, struct perf_event_attr *attr, - struct list_head *head_terms, + struct parse_events_terms *head_terms, struct parse_events_error *error); int perf_pmu__config_terms(struct perf_pmu *pmu, struct perf_event_attr *attr, - struct list_head *head_terms, + struct parse_events_terms *terms, bool zero, struct parse_events_error *error); __u64 perf_pmu__format_bits(struct perf_pmu *pmu, const char *name); int perf_pmu__format_type(struct perf_pmu *pmu, const char *name); -int perf_pmu__check_alias(struct perf_pmu *pmu, struct list_head *head_terms, +int perf_pmu__check_alias(struct perf_pmu *pmu, struct parse_events_terms *head_terms, struct perf_pmu_info *info, struct parse_events_error *err); int perf_pmu__find_event(struct perf_pmu *pmu, const char *event, void *state, pmu_event_callback cb); -- cgit v1.2.3 From 9c95e4ef065723496442898614d09a9a916eab81 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Mon, 4 Sep 2023 02:33:36 +0000 Subject: perf evlist: Add evlist__findnew_tracking_event() helper Currently, intel-bts, intel-pt, and arm-spe may add tracking event to the evlist. We may need to search for the tracking event for some settings. Therefore, add evlist__findnew_tracking_event() helper. If system_wide is true, evlist__findnew_tracking_event() set the cpu map of the evsel to all online CPUs. Signed-off-by: Yang Jihong Acked-by: Adrian Hunter Tested-by: Ravi Bangoria Cc: Alexander Shishkin Cc: Andi Kleen Cc: Anshuman Khandual Cc: Ian Rogers Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Thomas Richter Link: https://lore.kernel.org/r/20230904023340.12707-3-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 18 ++++++++++++++++++ tools/perf/util/evlist.h | 1 + 2 files changed, 19 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 7ef43f72098e..25c3ebe2c2f5 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -1694,6 +1694,24 @@ void evlist__set_tracking_event(struct evlist *evlist, struct evsel *tracking_ev tracking_evsel->tracking = true; } +struct evsel *evlist__findnew_tracking_event(struct evlist *evlist, bool system_wide) +{ + struct evsel *evsel; + + evsel = evlist__get_tracking_event(evlist); + if (!evsel__is_dummy_event(evsel)) { + evsel = evlist__add_aux_dummy(evlist, system_wide); + if (!evsel) + return NULL; + + evlist__set_tracking_event(evlist, evsel); + } else if (system_wide) { + perf_evlist__go_system_wide(&evlist->core, &evsel->core); + } + + return evsel; +} + struct evsel *evlist__find_evsel_by_str(struct evlist *evlist, const char *str) { struct evsel *evsel; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 664c6bf7b3e0..98e7ddb2bd30 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -387,6 +387,7 @@ bool evlist_cpu_iterator__end(const struct evlist_cpu_iterator *evlist_cpu_itr); struct evsel *evlist__get_tracking_event(struct evlist *evlist); void evlist__set_tracking_event(struct evlist *evlist, struct evsel *tracking_evsel); +struct evsel *evlist__findnew_tracking_event(struct evlist *evlist, bool system_wide); struct evsel *evlist__find_evsel_by_str(struct evlist *evlist, const char *str); -- cgit v1.2.3 From 95064b3352250a199323b8161eb08aee428a8286 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:05 +0000 Subject: perf kwork: Add `kwork` and `src_type` to work_init() for 'struct kwork_class' To support different types of reports, two parameters `struct perf_kwork * kwork` and `enum kwork_trace_type src_type` are added to work_init() of struct kwork_class for initialization in different scenarios. No functional change intended. Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-5-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/kwork.h | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index 53b7327550b8..736c7a08fb19 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -91,6 +91,7 @@ struct kwork_atom_page { DECLARE_BITMAP(bitmap, NR_ATOM_PER_PAGE); }; +struct perf_kwork; struct kwork_class; struct kwork_work { /* @@ -142,8 +143,10 @@ struct kwork_class { int (*class_init)(struct kwork_class *class, struct perf_session *session); - void (*work_init)(struct kwork_class *class, + void (*work_init)(struct perf_kwork *kwork, + struct kwork_class *class, struct kwork_work *work, + enum kwork_trace_type src_type, struct evsel *evsel, struct perf_sample *sample, struct machine *machine); @@ -152,7 +155,6 @@ struct kwork_class { char *buf, int len); }; -struct perf_kwork; struct trace_kwork_handler { int (*raise_event)(struct perf_kwork *kwork, struct kwork_class *class, struct evsel *evsel, -- cgit v1.2.3 From 38d8d013a525daf4f98a2b53065d8e89d5f810dd Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:08 +0000 Subject: perf kwork: Add sched record support The kwork_class type of sched is added to support recording and parsing of sched_switch events. As follows: # perf kwork -h Usage: perf kwork [] {record|report|latency|timehist} -D, --dump-raw-trace dump raw trace in ASCII -f, --force don't complain, do it -k, --kwork list of kwork to profile (irq, softirq, workqueue, sched, etc) -v, --verbose be more verbose (show symbol address, etc) # perf kwork -k sched record true [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.083 MB perf.data (47 samples) ] # perf evlist sched:sched_switch dummy:HG # Tip: use 'perf evlist --trace-fields' to show fields for tracepoint events Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-8-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/kwork.h | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index 736c7a08fb19..f8e9cdd1371a 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -16,6 +16,7 @@ enum kwork_class_type { KWORK_CLASS_IRQ, KWORK_CLASS_SOFTIRQ, KWORK_CLASS_WORKQUEUE, + KWORK_CLASS_SCHED, KWORK_CLASS_MAX, }; @@ -167,6 +168,10 @@ struct trace_kwork_handler { int (*exit_event)(struct perf_kwork *kwork, struct kwork_class *class, struct evsel *evsel, struct perf_sample *sample, struct machine *machine); + + int (*sched_switch_event)(struct perf_kwork *kwork, + struct kwork_class *class, struct evsel *evsel, + struct perf_sample *sample, struct machine *machine); }; struct perf_kwork { -- cgit v1.2.3 From 55c40e505234588d049384aa45a6931438c3028d Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:10 +0000 Subject: perf kwork top: Introduce new top utility Some common tools for collecting statistics on CPU usage, such as top, obtain statistics from timer interrupt sampling, and then periodically read statistics from /proc/stat. This method has some deviations: 1. In the tick interrupt, the time between the last tick and the current tick is counted in the current task. However, the task may be running only part of the time. 2. For each task, the top tool periodically reads the /proc/{PID}/status information. For tasks with a short life cycle, it may be missed. In conclusion, the top tool cannot accurately collect statistics on the CPU usage and running time of tasks. The statistical method based on sched_switch tracepoint can accurately calculate the CPU usage of all tasks. This method is applicable to scenarios where performance comparison data is of high precision. Example usage: # perf kwork Usage: perf kwork [] {record|report|latency|timehist|top} -D, --dump-raw-trace dump raw trace in ASCII -f, --force don't complain, do it -k, --kwork list of kwork to profile (irq, softirq, workqueue, sched, etc) -v, --verbose be more verbose (show symbol address, etc) # perf kwork -k sched record -- perf bench sched messaging -g 1 -l 10000 # Running 'sched/messaging' benchmark: # 20 sender and receiver processes per group # 1 groups == 40 processes run Total time: 14.074 [sec] [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 15.886 MB perf.data (129472 samples) ] # perf kwork top Total : 115708.178 ms, 8 cpus %Cpu(s): 9.78% id %Cpu0 [||||||||||||||||||||||||||| 90.55%] %Cpu1 [||||||||||||||||||||||||||| 90.51%] %Cpu2 [|||||||||||||||||||||||||| 88.57%] %Cpu3 [||||||||||||||||||||||||||| 91.18%] %Cpu4 [||||||||||||||||||||||||||| 91.09%] %Cpu5 [||||||||||||||||||||||||||| 90.88%] %Cpu6 [|||||||||||||||||||||||||| 88.64%] %Cpu7 [||||||||||||||||||||||||||| 90.28%] PID %CPU RUNTIME COMMMAND ---------------------------------------------------- 4113 22.23 3221.547 ms sched-messaging 4105 21.61 3131.495 ms sched-messaging 4119 21.53 3120.937 ms sched-messaging 4103 21.39 3101.614 ms sched-messaging 4106 21.37 3095.209 ms sched-messaging 4104 21.25 3077.269 ms sched-messaging 4115 21.21 3073.188 ms sched-messaging 4109 21.18 3069.022 ms sched-messaging 4111 20.78 3010.033 ms sched-messaging 4114 20.74 3007.073 ms sched-messaging 4108 20.73 3002.137 ms sched-messaging 4107 20.47 2967.292 ms sched-messaging 4117 20.39 2955.335 ms sched-messaging 4112 20.34 2947.080 ms sched-messaging 4118 20.32 2942.519 ms sched-messaging 4121 20.23 2929.865 ms sched-messaging 4110 20.22 2930.078 ms sched-messaging 4122 20.15 2919.542 ms sched-messaging 4120 19.77 2866.032 ms sched-messaging 4116 19.72 2857.660 ms sched-messaging 4127 16.19 2346.334 ms sched-messaging 4142 15.86 2297.600 ms sched-messaging 4141 15.62 2262.646 ms sched-messaging 4136 15.41 2231.408 ms sched-messaging 4130 15.38 2227.008 ms sched-messaging 4129 15.31 2217.692 ms sched-messaging 4126 15.21 2201.711 ms sched-messaging 4139 15.19 2200.722 ms sched-messaging 4137 15.10 2188.633 ms sched-messaging 4134 15.06 2182.082 ms sched-messaging 4132 15.02 2177.530 ms sched-messaging 4131 14.73 2131.973 ms sched-messaging 4125 14.68 2125.439 ms sched-messaging 4128 14.66 2122.255 ms sched-messaging 4123 14.65 2122.113 ms sched-messaging 4135 14.56 2107.144 ms sched-messaging 4133 14.51 2103.549 ms sched-messaging 4124 14.27 2066.671 ms sched-messaging 4140 14.17 2052.251 ms sched-messaging 4138 13.81 2000.361 ms sched-messaging 0 11.42 1652.009 ms swapper/2 0 11.35 1641.694 ms swapper/6 0 9.71 1405.108 ms swapper/7 0 9.48 1372.338 ms swapper/1 0 9.44 1366.013 ms swapper/0 0 9.11 1318.382 ms swapper/5 0 8.90 1287.582 ms swapper/4 0 8.81 1274.356 ms swapper/3 4100 2.61 379.328 ms perf 4101 1.16 169.487 ms perf-exec 151 0.65 94.741 ms systemd-resolve 249 0.36 53.030 ms sd-resolve 153 0.14 21.405 ms systemd-timesyn 1 0.10 16.200 ms systemd 16 0.09 15.785 ms rcu_preempt 4102 0.06 9.727 ms perf 4095 0.03 5.464 ms kworker/7:1 98 0.02 3.231 ms jbd2/sda-8 353 0.02 4.115 ms sshd 75 0.02 3.889 ms kworker/2:1 73 0.01 1.552 ms kworker/5:1 64 0.01 1.591 ms kworker/4:1 74 0.01 1.952 ms kworker/3:1 61 0.01 2.608 ms kcompactd0 397 0.01 1.602 ms kworker/1:1 69 0.01 1.817 ms kworker/1:1H 10 0.01 2.553 ms kworker/u16:0 2909 0.01 2.684 ms kworker/0:2 1211 0.00 0.426 ms kworker/7:0 97 0.00 0.153 ms kworker/7:1H 51 0.00 0.100 ms ksoftirqd/7 120 0.00 0.856 ms systemd-journal 76 0.00 1.414 ms kworker/6:1 46 0.00 0.246 ms ksoftirqd/6 45 0.00 0.164 ms migration/6 41 0.00 0.098 ms ksoftirqd/5 40 0.00 0.207 ms migration/5 86 0.00 1.339 ms kworker/4:1H 36 0.00 0.252 ms ksoftirqd/4 35 0.00 0.090 ms migration/4 31 0.00 0.156 ms ksoftirqd/3 30 0.00 0.073 ms migration/3 26 0.00 0.180 ms ksoftirqd/2 25 0.00 0.085 ms migration/2 21 0.00 0.106 ms ksoftirqd/1 20 0.00 0.118 ms migration/1 302 0.00 1.440 ms systemd-logind 17 0.00 0.132 ms migration/0 15 0.00 0.255 ms ksoftirqd/0 Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-10-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/kwork.h | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index f8e9cdd1371a..41ed193d5d8b 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -24,6 +24,7 @@ enum kwork_report_type { KWORK_REPORT_RUNTIME, KWORK_REPORT_LATENCY, KWORK_REPORT_TIMEHIST, + KWORK_REPORT_TOP, }; enum kwork_trace_type { @@ -129,6 +130,11 @@ struct kwork_work { u64 max_latency_start; u64 max_latency_end; u64 total_latency; + + /* + * top report + */ + u32 cpu_usage; }; struct kwork_class { @@ -174,6 +180,17 @@ struct trace_kwork_handler { struct perf_sample *sample, struct machine *machine); }; +struct __top_cpus_runtime { + u64 load; + u64 idle; + u64 total; +}; + +struct kwork_top_stat { + DECLARE_BITMAP(all_cpus_bitmap, MAX_NR_CPUS); + struct __top_cpus_runtime *cpus_runtime; +}; + struct perf_kwork { /* * metadata @@ -225,6 +242,11 @@ struct perf_kwork { u64 all_runtime; u64 all_count; u64 nr_skipped_events[KWORK_TRACE_MAX + 1]; + + /* + * perf kwork top data + */ + struct kwork_top_stat top_stat; }; struct kwork_work *perf_kwork_add_work(struct perf_kwork *kwork, -- cgit v1.2.3 From a8792242e4f2915a85bfa2d21a0ee2d9043de996 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:11 +0000 Subject: perf evsel: Add evsel__intval_common() helper Add evsel__intval_common() helper to search for common_field in tracepoint format. Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-11-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 13 +++++++++++++ tools/perf/util/evsel.h | 2 ++ 2 files changed, 15 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index a8a5ff87cc1f..d5363d23f5d3 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -2766,6 +2766,11 @@ struct tep_format_field *evsel__field(struct evsel *evsel, const char *name) return tep_find_field(evsel->tp_format, name); } +struct tep_format_field *evsel__common_field(struct evsel *evsel, const char *name) +{ + return tep_find_common_field(evsel->tp_format, name); +} + void *evsel__rawptr(struct evsel *evsel, struct perf_sample *sample, const char *name) { struct tep_format_field *field = evsel__field(evsel, name); @@ -2831,6 +2836,14 @@ u64 evsel__intval(struct evsel *evsel, struct perf_sample *sample, const char *n return field ? format_field__intval(field, sample, evsel->needs_swap) : 0; } + +u64 evsel__intval_common(struct evsel *evsel, struct perf_sample *sample, const char *name) +{ + struct tep_format_field *field = evsel__common_field(evsel, name); + + return field ? format_field__intval(field, sample, evsel->needs_swap) : 0; +} + #endif bool evsel__fallback(struct evsel *evsel, int err, char *msg, size_t msgsize) diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 848534ec74fa..815be2491938 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -340,6 +340,7 @@ struct perf_sample; #ifdef HAVE_LIBTRACEEVENT void *evsel__rawptr(struct evsel *evsel, struct perf_sample *sample, const char *name); u64 evsel__intval(struct evsel *evsel, struct perf_sample *sample, const char *name); +u64 evsel__intval_common(struct evsel *evsel, struct perf_sample *sample, const char *name); static inline char *evsel__strval(struct evsel *evsel, struct perf_sample *sample, const char *name) { @@ -352,6 +353,7 @@ struct tep_format_field; u64 format_field__intval(struct tep_format_field *field, struct perf_sample *sample, bool needs_swap); struct tep_format_field *evsel__field(struct evsel *evsel, const char *name); +struct tep_format_field *evsel__common_field(struct evsel *evsel, const char *name); static inline bool __evsel__match(const struct evsel *evsel, u32 type, u64 config) { -- cgit v1.2.3 From 2f21f5e4b48d910cbbe234264527b23b7cfc49f2 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:12 +0000 Subject: perf kwork top: Add statistics on hardirq event support Calculate the runtime of the hardirq events and subtract it from the corresponding task runtime to improve the precision. Example usage: # perf kwork -k sched,irq record -- perf record -o perf_record.data -a sleep 10 [ perf record: Woken up 2 times to write data ] [ perf record: Captured and wrote 1.054 MB perf_record.data (18019 samples) ] [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 1.798 MB perf.data (16334 samples) ] # # perf kwork top Total : 139240.869 ms, 8 cpus %Cpu(s): 94.91% id, 0.05% hi %Cpu0 [ 0.05%] %Cpu1 [| 5.00%] %Cpu2 [ 0.43%] %Cpu3 [ 0.57%] %Cpu4 [ 1.19%] %Cpu5 [|||||| 20.46%] %Cpu6 [ 0.48%] %Cpu7 [||| 12.10%] PID %CPU RUNTIME COMMMAND ---------------------------------------------------- 0 99.54 17325.622 ms swapper/2 0 99.54 17327.527 ms swapper/0 0 99.51 17319.909 ms swapper/6 0 99.42 17304.934 ms swapper/3 0 98.80 17197.385 ms swapper/4 0 94.99 16534.991 ms swapper/1 0 87.89 15295.264 ms swapper/7 0 79.53 13843.182 ms swapper/5 4252 36.50 6361.768 ms perf 4256 1.17 205.215 ms bash 151 0.53 93.298 ms systemd-resolve 4254 0.39 69.468 ms perf 423 0.34 59.368 ms bash 412 0.29 51.204 ms sshd 249 0.20 35.288 ms sd-resolve 16 0.17 30.287 ms rcu_preempt 153 0.09 17.266 ms systemd-timesyn 1 0.09 17.078 ms systemd 4253 0.07 12.457 ms perf 4255 0.06 11.559 ms perf 4234 0.03 6.105 ms kworker/u16:1 69 0.03 6.259 ms kworker/1:1H 4251 0.02 4.615 ms perf 4095 0.02 4.890 ms kworker/7:1 61 0.02 4.005 ms kcompactd0 75 0.02 3.546 ms kworker/2:1 97 0.01 3.106 ms kworker/7:1H 98 0.01 1.995 ms jbd2/sda-8 4088 0.01 1.779 ms kworker/u16:3 2909 0.01 1.795 ms kworker/0:2 4246 0.00 1.117 ms kworker/7:2 51 0.00 0.327 ms ksoftirqd/7 50 0.00 0.369 ms migration/7 102 0.00 0.160 ms kworker/6:1H 76 0.00 0.609 ms kworker/6:1 45 0.00 0.779 ms migration/6 87 0.00 0.504 ms kworker/5:1H 73 0.00 1.130 ms kworker/5:1 41 0.00 0.152 ms ksoftirqd/5 40 0.00 0.702 ms migration/5 64 0.00 0.316 ms kworker/4:1 35 0.00 0.791 ms migration/4 353 0.00 2.211 ms sshd 74 0.00 0.272 ms kworker/3:1 30 0.00 0.819 ms migration/3 25 0.00 0.784 ms migration/2 397 0.00 0.539 ms kworker/1:1 21 0.00 1.600 ms ksoftirqd/1 20 0.00 0.773 ms migration/1 17 0.00 1.682 ms migration/0 15 0.00 0.076 ms ksoftirqd/0 Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-12-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/kwork.h | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index 41ed193d5d8b..b3e410d02043 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -183,6 +183,7 @@ struct trace_kwork_handler { struct __top_cpus_runtime { u64 load; u64 idle; + u64 irq; u64 total; }; -- cgit v1.2.3 From e29090d28cc1ac1479a062e07260fdd7aa56f057 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:13 +0000 Subject: perf kwork top: Add statistics on softirq event support Calculate the runtime of the softirq events and subtract it from the corresponding task runtime to improve the precision. Example usage: # perf kwork -k sched,irq,softirq record -- perf record -e cpu-clock -o perf_record.data -a sleep 10 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.467 MB perf_record.data (7154 samples) ] [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 2.152 MB perf.data (22846 samples) ] # perf kwork top Total : 136601.588 ms, 8 cpus %Cpu(s): 95.66% id, 0.04% hi, 0.05% si %Cpu0 [ 0.02%] %Cpu1 [ 0.01%] %Cpu2 [| 4.61%] %Cpu3 [ 0.04%] %Cpu4 [ 0.01%] %Cpu5 [||||| 17.31%] %Cpu6 [ 0.51%] %Cpu7 [||| 11.42%] PID %CPU RUNTIME COMMMAND ---------------------------------------------------- 0 99.98 17073.515 ms swapper/4 0 99.98 17072.173 ms swapper/1 0 99.93 17064.229 ms swapper/3 0 99.62 17011.013 ms swapper/0 0 99.47 16985.180 ms swapper/6 0 95.17 16250.874 ms swapper/2 0 88.51 15111.684 ms swapper/7 0 82.62 14108.577 ms swapper/5 4342 33.00 5644.045 ms perf 4344 0.43 74.351 ms perf 16 0.13 22.296 ms rcu_preempt 4345 0.05 10.093 ms perf 4343 0.05 8.769 ms perf 4341 0.02 4.882 ms perf 4095 0.02 4.605 ms kworker/7:1 75 0.02 4.261 ms kworker/2:1 120 0.01 1.909 ms systemd-journal 98 0.01 2.540 ms jbd2/sda-8 61 0.01 3.404 ms kcompactd0 667 0.01 2.542 ms kworker/u16:2 4340 0.00 1.052 ms kworker/7:2 97 0.00 0.489 ms kworker/7:1H 51 0.00 0.209 ms ksoftirqd/7 50 0.00 0.646 ms migration/7 76 0.00 0.753 ms kworker/6:1 45 0.00 0.572 ms migration/6 87 0.00 0.145 ms kworker/5:1H 73 0.00 0.596 ms kworker/5:1 41 0.00 0.041 ms ksoftirqd/5 40 0.00 0.718 ms migration/5 64 0.00 0.115 ms kworker/4:1 35 0.00 0.556 ms migration/4 353 0.00 2.600 ms sshd 74 0.00 0.205 ms kworker/3:1 33 0.00 1.576 ms kworker/3:0H 30 0.00 0.996 ms migration/3 26 0.00 1.665 ms ksoftirqd/2 25 0.00 0.662 ms migration/2 397 0.00 0.057 ms kworker/1:1 20 0.00 1.005 ms migration/1 2909 0.00 1.053 ms kworker/0:2 17 0.00 0.720 ms migration/0 15 0.00 0.039 ms ksoftirqd/0 Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-13-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/kwork.h | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index b3e410d02043..723b34385df6 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -184,6 +184,7 @@ struct __top_cpus_runtime { u64 load; u64 idle; u64 irq; + u64 softirq; u64 total; }; -- cgit v1.2.3 From 8c98420987cda501924b7a34f2413b982517d244 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:15 +0000 Subject: perf kwork top: Implements BPF-based cpu usage statistics Use BPF to collect statistics on the CPU usage based on perf BPF skeletons. Example usage: # perf kwork top -h Usage: perf kwork top [] -b, --use-bpf Use BPF to measure task cpu usage -C, --cpu list of cpus to profile -i, --input input file name -n, --name event name to profile -s, --sort sort by key(s): rate, runtime, tid --time Time span for analysis (start,stop) # # perf kwork -k sched top -b Starting trace, Hit to stop and report ^C Total : 160702.425 ms, 8 cpus %Cpu(s): 36.00% id, 0.00% hi, 0.00% si %Cpu0 [|||||||||||||||||| 61.66%] %Cpu1 [|||||||||||||||||| 61.27%] %Cpu2 [||||||||||||||||||| 66.40%] %Cpu3 [|||||||||||||||||| 61.28%] %Cpu4 [|||||||||||||||||| 61.82%] %Cpu5 [||||||||||||||||||||||| 77.41%] %Cpu6 [|||||||||||||||||| 61.73%] %Cpu7 [|||||||||||||||||| 63.25%] PID SPID %CPU RUNTIME COMMMAND ------------------------------------------------------------- 0 0 38.72 8089.463 ms [swapper/1] 0 0 38.71 8084.547 ms [swapper/3] 0 0 38.33 8007.532 ms [swapper/0] 0 0 38.26 7992.985 ms [swapper/6] 0 0 38.17 7971.865 ms [swapper/4] 0 0 36.74 7447.765 ms [swapper/7] 0 0 33.59 6486.942 ms [swapper/2] 0 0 22.58 3771.268 ms [swapper/5] 9545 9351 2.48 447.136 ms sched-messaging 9574 9351 2.09 418.583 ms sched-messaging 9724 9351 2.05 372.407 ms sched-messaging 9531 9351 2.01 368.804 ms sched-messaging 9512 9351 2.00 362.250 ms sched-messaging 9514 9351 1.95 357.767 ms sched-messaging 9538 9351 1.86 384.476 ms sched-messaging 9712 9351 1.84 386.490 ms sched-messaging 9723 9351 1.83 380.021 ms sched-messaging 9722 9351 1.82 382.738 ms sched-messaging 9517 9351 1.81 354.794 ms sched-messaging 9559 9351 1.79 344.305 ms sched-messaging 9725 9351 1.77 365.315 ms sched-messaging # perf kwork -k sched top -b -n perf Starting trace, Hit to stop and report ^C Total : 151563.332 ms, 8 cpus %Cpu(s): 26.49% id, 0.00% hi, 0.00% si %Cpu0 [ 0.01%] %Cpu1 [ 0.00%] %Cpu2 [ 0.00%] %Cpu3 [ 0.00%] %Cpu4 [ 0.00%] %Cpu5 [ 0.00%] %Cpu6 [ 0.00%] %Cpu7 [ 0.00%] PID SPID %CPU RUNTIME COMMMAND ------------------------------------------------------------- 9754 9754 0.01 2.303 ms perf # # perf kwork -k sched top -b -C 2,3,4 Starting trace, Hit to stop and report ^C Total : 48016.721 ms, 3 cpus %Cpu(s): 27.82% id, 0.00% hi, 0.00% si %Cpu2 [|||||||||||||||||||||| 74.68%] %Cpu3 [||||||||||||||||||||| 71.06%] %Cpu4 [||||||||||||||||||||| 70.91%] PID SPID %CPU RUNTIME COMMMAND ------------------------------------------------------------- 0 0 29.08 4734.998 ms [swapper/4] 0 0 28.93 4710.029 ms [swapper/3] 0 0 25.31 3912.363 ms [swapper/2] 10248 10158 1.62 264.931 ms sched-messaging 10253 10158 1.62 265.136 ms sched-messaging 10158 10158 1.60 263.013 ms bash 10360 10158 1.49 243.639 ms sched-messaging 10413 10158 1.48 238.604 ms sched-messaging 10531 10158 1.47 234.067 ms sched-messaging 10400 10158 1.47 240.631 ms sched-messaging 10355 10158 1.47 230.586 ms sched-messaging 10377 10158 1.43 234.835 ms sched-messaging 10526 10158 1.42 232.045 ms sched-messaging 10298 10158 1.41 222.396 ms sched-messaging 10410 10158 1.38 221.853 ms sched-messaging 10364 10158 1.38 226.042 ms sched-messaging 10480 10158 1.36 213.633 ms sched-messaging 10370 10158 1.36 223.620 ms sched-messaging 10553 10158 1.34 217.169 ms sched-messaging 10291 10158 1.34 211.516 ms sched-messaging 10251 10158 1.34 218.813 ms sched-messaging 10522 10158 1.33 218.498 ms sched-messaging 10288 10158 1.33 216.787 ms sched-messaging Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-15-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/Build | 1 + tools/perf/util/bpf_kwork_top.c | 286 +++++++++++++++++++++++++++++++ tools/perf/util/bpf_skel/kwork_top.bpf.c | 187 ++++++++++++++++++++ tools/perf/util/kwork.h | 26 +++ 4 files changed, 500 insertions(+) create mode 100644 tools/perf/util/bpf_kwork_top.c create mode 100644 tools/perf/util/bpf_skel/kwork_top.bpf.c (limited to 'tools/perf/util') diff --git a/tools/perf/util/Build b/tools/perf/util/Build index 6d657c9927f7..0ea5a9d368d4 100644 --- a/tools/perf/util/Build +++ b/tools/perf/util/Build @@ -165,6 +165,7 @@ endif ifeq ($(CONFIG_LIBTRACEEVENT),y) perf-$(CONFIG_PERF_BPF_SKEL) += bpf_kwork.o + perf-$(CONFIG_PERF_BPF_SKEL) += bpf_kwork_top.o endif perf-$(CONFIG_LIBELF) += symbol-elf.o diff --git a/tools/perf/util/bpf_kwork_top.c b/tools/perf/util/bpf_kwork_top.c new file mode 100644 index 000000000000..42897ea22c61 --- /dev/null +++ b/tools/perf/util/bpf_kwork_top.c @@ -0,0 +1,286 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * bpf_kwork_top.c + * + * Copyright (c) 2022 Huawei Inc, Yang Jihong + */ + +#include +#include +#include +#include +#include + +#include + +#include "util/debug.h" +#include "util/evsel.h" +#include "util/kwork.h" + +#include +#include + +#include "util/bpf_skel/kwork_top.skel.h" + +/* + * This should be in sync with "util/kwork_top.bpf.c" + */ +#define MAX_COMMAND_LEN 16 + +struct time_data { + __u64 timestamp; +}; + +struct work_data { + __u64 runtime; +}; + +struct task_data { + __u32 tgid; + __u32 is_kthread; + char comm[MAX_COMMAND_LEN]; +}; + +struct work_key { + __u32 type; + __u32 pid; + __u64 task_p; +}; + +struct task_key { + __u32 pid; + __u32 cpu; +}; + +struct kwork_class_bpf { + struct kwork_class *class; + void (*load_prepare)(void); +}; + +static struct kwork_top_bpf *skel; + +void perf_kwork__top_start(void) +{ + struct timespec ts; + + clock_gettime(CLOCK_MONOTONIC, &ts); + skel->bss->from_timestamp = (u64)ts.tv_sec * NSEC_PER_SEC + ts.tv_nsec; + skel->bss->enabled = 1; + pr_debug("perf kwork top start at: %lld\n", skel->bss->from_timestamp); +} + +void perf_kwork__top_finish(void) +{ + struct timespec ts; + + skel->bss->enabled = 0; + clock_gettime(CLOCK_MONOTONIC, &ts); + skel->bss->to_timestamp = (u64)ts.tv_sec * NSEC_PER_SEC + ts.tv_nsec; + pr_debug("perf kwork top finish at: %lld\n", skel->bss->to_timestamp); +} + +static void sched_load_prepare(void) +{ + bpf_program__set_autoload(skel->progs.on_switch, true); +} + +static struct kwork_class_bpf kwork_sched_bpf = { + .load_prepare = sched_load_prepare, +}; + +static struct kwork_class_bpf * +kwork_class_bpf_supported_list[KWORK_CLASS_MAX] = { + [KWORK_CLASS_SCHED] = &kwork_sched_bpf, +}; + +static bool valid_kwork_class_type(enum kwork_class_type type) +{ + return type >= 0 && type < KWORK_CLASS_MAX ? true : false; +} + +static int setup_filters(struct perf_kwork *kwork) +{ + u8 val = 1; + int i, nr_cpus, fd; + struct perf_cpu_map *map; + + if (kwork->cpu_list) { + fd = bpf_map__fd(skel->maps.kwork_top_cpu_filter); + if (fd < 0) { + pr_debug("Invalid cpu filter fd\n"); + return -1; + } + + map = perf_cpu_map__new(kwork->cpu_list); + if (!map) { + pr_debug("Invalid cpu_list\n"); + return -1; + } + + nr_cpus = libbpf_num_possible_cpus(); + for (i = 0; i < perf_cpu_map__nr(map); i++) { + struct perf_cpu cpu = perf_cpu_map__cpu(map, i); + + if (cpu.cpu >= nr_cpus) { + perf_cpu_map__put(map); + pr_err("Requested cpu %d too large\n", cpu.cpu); + return -1; + } + bpf_map_update_elem(fd, &cpu.cpu, &val, BPF_ANY); + } + perf_cpu_map__put(map); + + skel->bss->has_cpu_filter = 1; + } + + return 0; +} + +int perf_kwork__top_prepare_bpf(struct perf_kwork *kwork __maybe_unused) +{ + struct bpf_program *prog; + struct kwork_class *class; + struct kwork_class_bpf *class_bpf; + enum kwork_class_type type; + + skel = kwork_top_bpf__open(); + if (!skel) { + pr_debug("Failed to open kwork top skeleton\n"); + return -1; + } + + /* + * set all progs to non-autoload, + * then set corresponding progs according to config + */ + bpf_object__for_each_program(prog, skel->obj) + bpf_program__set_autoload(prog, false); + + list_for_each_entry(class, &kwork->class_list, list) { + type = class->type; + if (!valid_kwork_class_type(type) || + !kwork_class_bpf_supported_list[type]) { + pr_err("Unsupported bpf trace class %s\n", class->name); + goto out; + } + + class_bpf = kwork_class_bpf_supported_list[type]; + class_bpf->class = class; + + if (class_bpf->load_prepare) + class_bpf->load_prepare(); + } + + if (kwork_top_bpf__load(skel)) { + pr_debug("Failed to load kwork top skeleton\n"); + goto out; + } + + if (setup_filters(kwork)) + goto out; + + if (kwork_top_bpf__attach(skel)) { + pr_debug("Failed to attach kwork top skeleton\n"); + goto out; + } + + return 0; + +out: + kwork_top_bpf__destroy(skel); + return -1; +} + +static void read_task_info(struct kwork_work *work) +{ + int fd; + struct task_data data; + struct task_key key = { + .pid = work->id, + .cpu = work->cpu, + }; + + fd = bpf_map__fd(skel->maps.kwork_top_tasks); + if (fd < 0) { + pr_debug("Invalid top tasks map fd\n"); + return; + } + + if (!bpf_map_lookup_elem(fd, &key, &data)) { + work->tgid = data.tgid; + work->is_kthread = data.is_kthread; + work->name = strdup(data.comm); + } +} +static int add_work(struct perf_kwork *kwork, struct work_key *key, + struct work_data *data, int cpu) +{ + struct kwork_class_bpf *bpf_trace; + struct kwork_work *work; + struct kwork_work tmp = { + .id = key->pid, + .cpu = cpu, + .name = NULL, + }; + enum kwork_class_type type = key->type; + + if (!valid_kwork_class_type(type)) { + pr_debug("Invalid class type %d to add work\n", type); + return -1; + } + + bpf_trace = kwork_class_bpf_supported_list[type]; + tmp.class = bpf_trace->class; + + work = perf_kwork_add_work(kwork, tmp.class, &tmp); + if (!work) + return -1; + + work->total_runtime = data->runtime; + read_task_info(work); + + return 0; +} + +int perf_kwork__top_read_bpf(struct perf_kwork *kwork) +{ + int i, fd, nr_cpus; + struct work_data *data; + struct work_key key, prev; + + fd = bpf_map__fd(skel->maps.kwork_top_works); + if (fd < 0) { + pr_debug("Invalid top runtime fd\n"); + return -1; + } + + nr_cpus = libbpf_num_possible_cpus(); + data = calloc(nr_cpus, sizeof(struct work_data)); + if (!data) + return -1; + + memset(&prev, 0, sizeof(prev)); + while (!bpf_map_get_next_key(fd, &prev, &key)) { + if ((bpf_map_lookup_elem(fd, &key, data)) != 0) { + pr_debug("Failed to lookup top elem\n"); + return -1; + } + + for (i = 0; i < nr_cpus; i++) { + if (data[i].runtime == 0) + continue; + + if (add_work(kwork, &key, &data[i], i)) + return -1; + } + prev = key; + } + free(data); + + return 0; +} + +void perf_kwork__top_cleanup_bpf(void) +{ + kwork_top_bpf__destroy(skel); +} diff --git a/tools/perf/util/bpf_skel/kwork_top.bpf.c b/tools/perf/util/bpf_skel/kwork_top.bpf.c new file mode 100644 index 000000000000..47ad61608ec7 --- /dev/null +++ b/tools/perf/util/bpf_skel/kwork_top.bpf.c @@ -0,0 +1,187 @@ +// SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause) +// Copyright (c) 2022, Huawei + +#include "vmlinux.h" +#include +#include +#include + +/* + * This should be in sync with "util/kwork.h" + */ +enum kwork_class_type { + KWORK_CLASS_IRQ, + KWORK_CLASS_SOFTIRQ, + KWORK_CLASS_WORKQUEUE, + KWORK_CLASS_SCHED, + KWORK_CLASS_MAX, +}; + +#define MAX_ENTRIES 102400 +#define MAX_NR_CPUS 2048 +#define PF_KTHREAD 0x00200000 +#define MAX_COMMAND_LEN 16 + +struct time_data { + __u64 timestamp; +}; + +struct work_data { + __u64 runtime; +}; + +struct task_data { + __u32 tgid; + __u32 is_kthread; + char comm[MAX_COMMAND_LEN]; +}; + +struct work_key { + __u32 type; + __u32 pid; + __u64 task_p; +}; + +struct task_key { + __u32 pid; + __u32 cpu; +}; + +struct { + __uint(type, BPF_MAP_TYPE_TASK_STORAGE); + __uint(map_flags, BPF_F_NO_PREALLOC); + __type(key, int); + __type(value, struct time_data); +} kwork_top_task_time SEC(".maps"); + +struct { + __uint(type, BPF_MAP_TYPE_HASH); + __uint(key_size, sizeof(struct task_key)); + __uint(value_size, sizeof(struct task_data)); + __uint(max_entries, MAX_ENTRIES); +} kwork_top_tasks SEC(".maps"); + +struct { + __uint(type, BPF_MAP_TYPE_PERCPU_HASH); + __uint(key_size, sizeof(struct work_key)); + __uint(value_size, sizeof(struct work_data)); + __uint(max_entries, MAX_ENTRIES); +} kwork_top_works SEC(".maps"); + +struct { + __uint(type, BPF_MAP_TYPE_HASH); + __uint(key_size, sizeof(u32)); + __uint(value_size, sizeof(u8)); + __uint(max_entries, MAX_NR_CPUS); +} kwork_top_cpu_filter SEC(".maps"); + +int enabled = 0; + +int has_cpu_filter = 0; + +__u64 from_timestamp = 0; +__u64 to_timestamp = 0; + +static __always_inline int cpu_is_filtered(__u32 cpu) +{ + __u8 *cpu_val; + + if (has_cpu_filter) { + cpu_val = bpf_map_lookup_elem(&kwork_top_cpu_filter, &cpu); + if (!cpu_val) + return 1; + } + + return 0; +} + +static __always_inline void update_task_info(struct task_struct *task, __u32 cpu) +{ + struct task_key key = { + .pid = task->pid, + .cpu = cpu, + }; + + if (!bpf_map_lookup_elem(&kwork_top_tasks, &key)) { + struct task_data data = { + .tgid = task->tgid, + .is_kthread = task->flags & PF_KTHREAD ? 1 : 0, + }; + BPF_CORE_READ_STR_INTO(&data.comm, task, comm); + + bpf_map_update_elem(&kwork_top_tasks, &key, &data, BPF_ANY); + } +} + +static __always_inline void update_work(struct work_key *key, __u64 delta) +{ + struct work_data *data; + + data = bpf_map_lookup_elem(&kwork_top_works, key); + if (data) { + data->runtime += delta; + } else { + struct work_data new_data = { + .runtime = delta, + }; + + bpf_map_update_elem(&kwork_top_works, key, &new_data, BPF_ANY); + } +} + +static void on_sched_out(struct task_struct *task, __u64 ts, __u32 cpu) +{ + __u64 delta; + struct time_data *pelem; + + pelem = bpf_task_storage_get(&kwork_top_task_time, task, NULL, 0); + if (pelem) + delta = ts - pelem->timestamp; + else + delta = ts - from_timestamp; + + struct work_key key = { + .type = KWORK_CLASS_SCHED, + .pid = task->pid, + .task_p = (__u64)task, + }; + + update_work(&key, delta); + update_task_info(task, cpu); +} + +static void on_sched_in(struct task_struct *task, __u64 ts) +{ + struct time_data *pelem; + + pelem = bpf_task_storage_get(&kwork_top_task_time, task, NULL, + BPF_LOCAL_STORAGE_GET_F_CREATE); + if (pelem) + pelem->timestamp = ts; +} + +SEC("tp_btf/sched_switch") +int on_switch(u64 *ctx) +{ + struct task_struct *prev, *next; + + prev = (struct task_struct *)ctx[1]; + next = (struct task_struct *)ctx[2]; + + if (!enabled) + return 0; + + __u32 cpu = bpf_get_smp_processor_id(); + + if (cpu_is_filtered(cpu)) + return 0; + + __u64 ts = bpf_ktime_get_ns(); + + on_sched_out(prev, ts, cpu); + on_sched_in(next, ts); + + return 0; +} + +char LICENSE[] SEC("license") = "Dual BSD/GPL"; diff --git a/tools/perf/util/kwork.h b/tools/perf/util/kwork.h index 723b34385df6..76fe2a821bcf 100644 --- a/tools/perf/util/kwork.h +++ b/tools/perf/util/kwork.h @@ -135,6 +135,8 @@ struct kwork_work { * top report */ u32 cpu_usage; + u32 tgid; + bool is_kthread; }; struct kwork_class { @@ -264,6 +266,13 @@ void perf_kwork__report_cleanup_bpf(void); void perf_kwork__trace_start(void); void perf_kwork__trace_finish(void); +int perf_kwork__top_prepare_bpf(struct perf_kwork *kwork); +int perf_kwork__top_read_bpf(struct perf_kwork *kwork); +void perf_kwork__top_cleanup_bpf(void); + +void perf_kwork__top_start(void); +void perf_kwork__top_finish(void); + #else /* !HAVE_BPF_SKEL */ static inline int @@ -283,6 +292,23 @@ static inline void perf_kwork__report_cleanup_bpf(void) {} static inline void perf_kwork__trace_start(void) {} static inline void perf_kwork__trace_finish(void) {} +static inline int +perf_kwork__top_prepare_bpf(struct perf_kwork *kwork __maybe_unused) +{ + return -1; +} + +static inline int +perf_kwork__top_read_bpf(struct perf_kwork *kwork __maybe_unused) +{ + return -1; +} + +static inline void perf_kwork__top_cleanup_bpf(void) {} + +static inline void perf_kwork__top_start(void) {} +static inline void perf_kwork__top_finish(void) {} + #endif /* HAVE_BPF_SKEL */ #endif /* PERF_UTIL_KWORK_H */ -- cgit v1.2.3 From d2956b3acf869336e30d903dcaf23645c82d06f6 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:16 +0000 Subject: perf kwork top: Add BPF-based statistics on hardirq event support Use BPF to collect statistics on hardirq events based on perf BPF skeletons. Example usage: # perf kwork top -k sched,irq -b Starting trace, Hit to stop and report ^C Total : 136717.945 ms, 8 cpus %Cpu(s): 17.10% id, 0.01% hi, 0.00% si %Cpu0 [||||||||||||||||||||||||| 84.26%] %Cpu1 [||||||||||||||||||||||||| 84.77%] %Cpu2 [|||||||||||||||||||||||| 83.22%] %Cpu3 [|||||||||||||||||||||||| 80.37%] %Cpu4 [|||||||||||||||||||||||| 81.49%] %Cpu5 [||||||||||||||||||||||||| 84.68%] %Cpu6 [||||||||||||||||||||||||| 84.48%] %Cpu7 [|||||||||||||||||||||||| 80.21%] PID SPID %CPU RUNTIME COMMMAND ------------------------------------------------------------- 0 0 19.78 3482.833 ms [swapper/7] 0 0 19.62 3454.219 ms [swapper/3] 0 0 18.50 3258.339 ms [swapper/4] 0 0 16.76 2842.749 ms [swapper/2] 0 0 15.71 2627.905 ms [swapper/0] 0 0 15.51 2598.206 ms [swapper/6] 0 0 15.31 2561.820 ms [swapper/5] 0 0 15.22 2548.708 ms [swapper/1] 13253 13018 2.95 513.108 ms sched-messaging 13092 13018 2.67 454.167 ms sched-messaging 13401 13018 2.66 454.790 ms sched-messaging 13240 13018 2.64 454.587 ms sched-messaging 13251 13018 2.61 442.273 ms sched-messaging 13075 13018 2.61 438.932 ms sched-messaging 13220 13018 2.60 443.245 ms sched-messaging 13235 13018 2.59 443.268 ms sched-messaging 13222 13018 2.50 426.344 ms sched-messaging 13410 13018 2.49 426.191 ms sched-messaging 13228 13018 2.46 425.121 ms sched-messaging 13379 13018 2.38 409.950 ms sched-messaging 13236 13018 2.37 413.159 ms sched-messaging 13095 13018 2.36 396.572 ms sched-messaging 13325 13018 2.35 408.089 ms sched-messaging 13242 13018 2.32 394.750 ms sched-messaging 13386 13018 2.31 396.997 ms sched-messaging 13046 13018 2.29 383.833 ms sched-messaging 13109 13018 2.28 388.482 ms sched-messaging 13388 13018 2.28 393.576 ms sched-messaging 13238 13018 2.26 388.487 ms sched-messaging Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-16-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_kwork_top.c | 11 +++++ tools/perf/util/bpf_skel/kwork_top.bpf.c | 79 ++++++++++++++++++++++++++++++++ 2 files changed, 90 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_kwork_top.c b/tools/perf/util/bpf_kwork_top.c index 42897ea22c61..3998bd2a938f 100644 --- a/tools/perf/util/bpf_kwork_top.c +++ b/tools/perf/util/bpf_kwork_top.c @@ -79,6 +79,16 @@ void perf_kwork__top_finish(void) pr_debug("perf kwork top finish at: %lld\n", skel->bss->to_timestamp); } +static void irq_load_prepare(void) +{ + bpf_program__set_autoload(skel->progs.on_irq_handler_entry, true); + bpf_program__set_autoload(skel->progs.on_irq_handler_exit, true); +} + +static struct kwork_class_bpf kwork_irq_bpf = { + .load_prepare = irq_load_prepare, +}; + static void sched_load_prepare(void) { bpf_program__set_autoload(skel->progs.on_switch, true); @@ -90,6 +100,7 @@ static struct kwork_class_bpf kwork_sched_bpf = { static struct kwork_class_bpf * kwork_class_bpf_supported_list[KWORK_CLASS_MAX] = { + [KWORK_CLASS_IRQ] = &kwork_irq_bpf, [KWORK_CLASS_SCHED] = &kwork_sched_bpf, }; diff --git a/tools/perf/util/bpf_skel/kwork_top.bpf.c b/tools/perf/util/bpf_skel/kwork_top.bpf.c index 47ad61608ec7..9c7dc62386c7 100644 --- a/tools/perf/util/bpf_skel/kwork_top.bpf.c +++ b/tools/perf/util/bpf_skel/kwork_top.bpf.c @@ -54,6 +54,13 @@ struct { __type(value, struct time_data); } kwork_top_task_time SEC(".maps"); +struct { + __uint(type, BPF_MAP_TYPE_PERCPU_HASH); + __uint(key_size, sizeof(struct work_key)); + __uint(value_size, sizeof(struct time_data)); + __uint(max_entries, MAX_ENTRIES); +} kwork_top_irq_time SEC(".maps"); + struct { __uint(type, BPF_MAP_TYPE_HASH); __uint(key_size, sizeof(struct task_key)); @@ -184,4 +191,76 @@ int on_switch(u64 *ctx) return 0; } +SEC("tp_btf/irq_handler_entry") +int on_irq_handler_entry(u64 *cxt) +{ + struct task_struct *task; + + if (!enabled) + return 0; + + __u32 cpu = bpf_get_smp_processor_id(); + + if (cpu_is_filtered(cpu)) + return 0; + + __u64 ts = bpf_ktime_get_ns(); + + task = (struct task_struct *)bpf_get_current_task(); + if (!task) + return 0; + + struct work_key key = { + .type = KWORK_CLASS_IRQ, + .pid = BPF_CORE_READ(task, pid), + .task_p = (__u64)task, + }; + + struct time_data data = { + .timestamp = ts, + }; + + bpf_map_update_elem(&kwork_top_irq_time, &key, &data, BPF_ANY); + + return 0; +} + +SEC("tp_btf/irq_handler_exit") +int on_irq_handler_exit(u64 *cxt) +{ + __u64 delta; + struct task_struct *task; + struct time_data *pelem; + + if (!enabled) + return 0; + + __u32 cpu = bpf_get_smp_processor_id(); + + if (cpu_is_filtered(cpu)) + return 0; + + __u64 ts = bpf_ktime_get_ns(); + + task = (struct task_struct *)bpf_get_current_task(); + if (!task) + return 0; + + struct work_key key = { + .type = KWORK_CLASS_IRQ, + .pid = BPF_CORE_READ(task, pid), + .task_p = (__u64)task, + }; + + pelem = bpf_map_lookup_elem(&kwork_top_irq_time, &key); + if (pelem && pelem->timestamp != 0) + delta = ts - pelem->timestamp; + else + delta = ts - from_timestamp; + + update_work(&key, delta); + + return 0; +} + char LICENSE[] SEC("license") = "Dual BSD/GPL"; -- cgit v1.2.3 From 36019dff30f7aacb8a26b2b9564fe04b097d502e Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Sat, 12 Aug 2023 08:49:17 +0000 Subject: perf kwork top: Add BPF-based statistics on softirq event support Use BPF to collect statistics on softirq events based on perf BPF skeletons. Example usage: # perf kwork top -b Starting trace, Hit to stop and report ^C Total : 135445.704 ms, 8 cpus %Cpu(s): 28.35% id, 0.00% hi, 0.25% si %Cpu0 [|||||||||||||||||||| 69.85%] %Cpu1 [|||||||||||||||||||||| 74.10%] %Cpu2 [||||||||||||||||||||| 71.18%] %Cpu3 [|||||||||||||||||||| 69.61%] %Cpu4 [|||||||||||||||||||||| 74.05%] %Cpu5 [|||||||||||||||||||| 69.33%] %Cpu6 [|||||||||||||||||||| 69.71%] %Cpu7 [|||||||||||||||||||||| 73.77%] PID SPID %CPU RUNTIME COMMMAND ------------------------------------------------------------- 0 0 30.43 5271.005 ms [swapper/5] 0 0 30.17 5226.644 ms [swapper/3] 0 0 30.08 5210.257 ms [swapper/6] 0 0 29.89 5177.177 ms [swapper/0] 0 0 28.51 4938.672 ms [swapper/2] 0 0 25.93 4223.464 ms [swapper/7] 0 0 25.69 4181.411 ms [swapper/4] 0 0 25.63 4173.804 ms [swapper/1] 16665 16265 2.16 360.600 ms sched-messaging 16537 16265 2.05 356.275 ms sched-messaging 16503 16265 2.01 343.063 ms sched-messaging 16424 16265 1.97 336.876 ms sched-messaging 16580 16265 1.94 323.658 ms sched-messaging 16515 16265 1.92 321.616 ms sched-messaging 16659 16265 1.91 325.538 ms sched-messaging 16634 16265 1.88 327.766 ms sched-messaging 16454 16265 1.87 326.843 ms sched-messaging 16382 16265 1.87 322.591 ms sched-messaging 16642 16265 1.86 320.506 ms sched-messaging 16582 16265 1.86 320.164 ms sched-messaging 16315 16265 1.86 326.872 ms sched-messaging 16637 16265 1.85 323.766 ms sched-messaging 16506 16265 1.82 311.688 ms sched-messaging 16512 16265 1.81 304.643 ms sched-messaging 16560 16265 1.80 314.751 ms sched-messaging 16320 16265 1.80 313.405 ms sched-messaging 16442 16265 1.80 314.403 ms sched-messaging 16626 16265 1.78 295.380 ms sched-messaging 16600 16265 1.77 309.444 ms sched-messaging 16550 16265 1.76 301.161 ms sched-messaging 16525 16265 1.75 296.560 ms sched-messaging 16314 16265 1.75 298.338 ms sched-messaging 16595 16265 1.74 304.390 ms sched-messaging 16555 16265 1.74 287.564 ms sched-messaging 16520 16265 1.74 295.734 ms sched-messaging 16507 16265 1.73 293.956 ms sched-messaging 16593 16265 1.72 296.443 ms sched-messaging 16531 16265 1.72 299.950 ms sched-messaging 16281 16265 1.72 301.339 ms sched-messaging Reviewed-by: Ian Rogers Signed-off-by: Yang Jihong Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Link: https://lore.kernel.org/r/20230812084917.169338-17-yangjihong1@huawei.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_kwork_top.c | 11 +++++ tools/perf/util/bpf_skel/kwork_top.bpf.c | 72 ++++++++++++++++++++++++++++++++ 2 files changed, 83 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_kwork_top.c b/tools/perf/util/bpf_kwork_top.c index 3998bd2a938f..1a607b94f44d 100644 --- a/tools/perf/util/bpf_kwork_top.c +++ b/tools/perf/util/bpf_kwork_top.c @@ -89,6 +89,16 @@ static struct kwork_class_bpf kwork_irq_bpf = { .load_prepare = irq_load_prepare, }; +static void softirq_load_prepare(void) +{ + bpf_program__set_autoload(skel->progs.on_softirq_entry, true); + bpf_program__set_autoload(skel->progs.on_softirq_exit, true); +} + +static struct kwork_class_bpf kwork_softirq_bpf = { + .load_prepare = softirq_load_prepare, +}; + static void sched_load_prepare(void) { bpf_program__set_autoload(skel->progs.on_switch, true); @@ -101,6 +111,7 @@ static struct kwork_class_bpf kwork_sched_bpf = { static struct kwork_class_bpf * kwork_class_bpf_supported_list[KWORK_CLASS_MAX] = { [KWORK_CLASS_IRQ] = &kwork_irq_bpf, + [KWORK_CLASS_SOFTIRQ] = &kwork_softirq_bpf, [KWORK_CLASS_SCHED] = &kwork_sched_bpf, }; diff --git a/tools/perf/util/bpf_skel/kwork_top.bpf.c b/tools/perf/util/bpf_skel/kwork_top.bpf.c index 9c7dc62386c7..84c15ccbab44 100644 --- a/tools/perf/util/bpf_skel/kwork_top.bpf.c +++ b/tools/perf/util/bpf_skel/kwork_top.bpf.c @@ -263,4 +263,76 @@ int on_irq_handler_exit(u64 *cxt) return 0; } +SEC("tp_btf/softirq_entry") +int on_softirq_entry(u64 *cxt) +{ + struct task_struct *task; + + if (!enabled) + return 0; + + __u32 cpu = bpf_get_smp_processor_id(); + + if (cpu_is_filtered(cpu)) + return 0; + + __u64 ts = bpf_ktime_get_ns(); + + task = (struct task_struct *)bpf_get_current_task(); + if (!task) + return 0; + + struct work_key key = { + .type = KWORK_CLASS_SOFTIRQ, + .pid = BPF_CORE_READ(task, pid), + .task_p = (__u64)task, + }; + + struct time_data data = { + .timestamp = ts, + }; + + bpf_map_update_elem(&kwork_top_irq_time, &key, &data, BPF_ANY); + + return 0; +} + +SEC("tp_btf/softirq_exit") +int on_softirq_exit(u64 *cxt) +{ + __u64 delta; + struct task_struct *task; + struct time_data *pelem; + + if (!enabled) + return 0; + + __u32 cpu = bpf_get_smp_processor_id(); + + if (cpu_is_filtered(cpu)) + return 0; + + __u64 ts = bpf_ktime_get_ns(); + + task = (struct task_struct *)bpf_get_current_task(); + if (!task) + return 0; + + struct work_key key = { + .type = KWORK_CLASS_SOFTIRQ, + .pid = BPF_CORE_READ(task, pid), + .task_p = (__u64)task, + }; + + pelem = bpf_map_lookup_elem(&kwork_top_irq_time, &key); + if (pelem) + delta = ts - pelem->timestamp; + else + delta = ts - from_timestamp; + + update_work(&key, delta); + + return 0; +} + char LICENSE[] SEC("license") = "Dual BSD/GPL"; -- cgit v1.2.3 From 2bc12abce8efb9f755bdbc14d426a0fc80ce74f4 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 6 Sep 2023 10:48:59 -0700 Subject: perf tools: Add read_all_cgroups() and __cgroup_find() The read_all_cgroups() is to build a tree of cgroups in the system and users can look up a cgroup using __cgroup_find(). Committer notes: Had to do this to cover that #else block: -static inline u64 __read_cgroup_id(const char *path) { return -1ULL; } +static inline u64 __read_cgroup_id(const char *path __maybe_unused) { return -1ULL; } Reviewed-by: Ian Rogers Signed-off-by: Namhyung Kim Cc: Adrian Hunter Cc: Hao Luo Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230906174903.346486-2-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/cgroup.c | 61 +++++++++++++++++++++++++++++++++++++++++------- tools/perf/util/cgroup.h | 4 ++++ 2 files changed, 57 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c index bfb13306d82c..d03651098dde 100644 --- a/tools/perf/util/cgroup.c +++ b/tools/perf/util/cgroup.c @@ -48,28 +48,36 @@ static int open_cgroup(const char *name) } #ifdef HAVE_FILE_HANDLE -int read_cgroup_id(struct cgroup *cgrp) +static u64 __read_cgroup_id(const char *path) { - char path[PATH_MAX + 1]; - char mnt[PATH_MAX + 1]; struct { struct file_handle fh; uint64_t cgroup_id; } handle; int mount_id; + handle.fh.handle_bytes = sizeof(handle.cgroup_id); + if (name_to_handle_at(AT_FDCWD, path, &handle.fh, &mount_id, 0) < 0) + return -1ULL; + + return handle.cgroup_id; +} + +int read_cgroup_id(struct cgroup *cgrp) +{ + char path[PATH_MAX + 1]; + char mnt[PATH_MAX + 1]; + if (cgroupfs_find_mountpoint(mnt, PATH_MAX + 1, "perf_event")) return -1; scnprintf(path, PATH_MAX, "%s/%s", mnt, cgrp->name); - handle.fh.handle_bytes = sizeof(handle.cgroup_id); - if (name_to_handle_at(AT_FDCWD, path, &handle.fh, &mount_id, 0) < 0) - return -1; - - cgrp->id = handle.cgroup_id; + cgrp->id = __read_cgroup_id(path); return 0; } +#else +static inline u64 __read_cgroup_id(const char *path __maybe_unused) { return -1ULL; } #endif /* HAVE_FILE_HANDLE */ #ifndef CGROUP2_SUPER_MAGIC @@ -562,6 +570,11 @@ struct cgroup *cgroup__findnew(struct perf_env *env, uint64_t id, return cgrp; } +struct cgroup *__cgroup__find(struct rb_root *root, uint64_t id) +{ + return __cgroup__findnew(root, id, /*create=*/false, /*path=*/NULL); +} + struct cgroup *cgroup__find(struct perf_env *env, uint64_t id) { struct cgroup *cgrp; @@ -587,3 +600,35 @@ void perf_env__purge_cgroups(struct perf_env *env) } up_write(&env->cgroups.lock); } + +void read_all_cgroups(struct rb_root *root) +{ + char mnt[PATH_MAX]; + struct cgroup_name *cn; + int prefix_len; + + if (cgroupfs_find_mountpoint(mnt, sizeof(mnt), "perf_event")) + return; + + /* cgroup_name will have a full path, skip the root directory */ + prefix_len = strlen(mnt); + + /* collect all cgroups in the cgroup_list */ + if (nftw(mnt, add_cgroup_name, 20, 0) < 0) + return; + + list_for_each_entry(cn, &cgroup_list, list) { + const char *name; + u64 cgrp_id; + + /* cgroup_name might have a full path, skip the prefix */ + name = cn->name + prefix_len; + if (name[0] == '\0') + name = "/"; + + cgrp_id = __read_cgroup_id(cn->name); + __cgroup__findnew(root, cgrp_id, /*create=*/true, name); + } + + release_cgroup_list(); +} diff --git a/tools/perf/util/cgroup.h b/tools/perf/util/cgroup.h index 12256b78608c..beb6fe1012ed 100644 --- a/tools/perf/util/cgroup.h +++ b/tools/perf/util/cgroup.h @@ -37,6 +37,7 @@ int parse_cgroups(const struct option *opt, const char *str, int unset); struct cgroup *cgroup__findnew(struct perf_env *env, uint64_t id, const char *path); struct cgroup *cgroup__find(struct perf_env *env, uint64_t id); +struct cgroup *__cgroup__find(struct rb_root *root, uint64_t id); void perf_env__purge_cgroups(struct perf_env *env); @@ -49,6 +50,9 @@ static inline int read_cgroup_id(struct cgroup *cgrp __maybe_unused) } #endif /* HAVE_FILE_HANDLE */ +/* read all cgroups in the system and save them in the rbtree */ +void read_all_cgroups(struct rb_root *root); + int cgroup_is_v2(const char *subsys); #endif /* __CGROUP_H__ */ -- cgit v1.2.3 From d0c502e46e9796b19e3b0fd28750cb5b679f7c4b Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 6 Sep 2023 10:49:00 -0700 Subject: perf lock contention: Prepare to handle cgroups Save cgroup info and display cgroup names if requested. This is a preparation for the next patch. Reviewed-by: Ian Rogers Signed-off-by: Namhyung Kim Cc: Adrian Hunter Cc: Hao Luo Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230906174903.346486-3-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_lock_contention.c | 26 +++++++++++++++++++++++++- tools/perf/util/lock-contention.h | 9 +++++++-- 2 files changed, 32 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_lock_contention.c b/tools/perf/util/bpf_lock_contention.c index e7dddf0127bc..c6bd7c9b2d57 100644 --- a/tools/perf/util/bpf_lock_contention.c +++ b/tools/perf/util/bpf_lock_contention.c @@ -1,4 +1,5 @@ // SPDX-License-Identifier: GPL-2.0 +#include "util/cgroup.h" #include "util/debug.h" #include "util/evlist.h" #include "util/machine.h" @@ -151,6 +152,10 @@ int lock_contention_prepare(struct lock_contention *con) skel->bss->needs_callstack = con->save_callstack; skel->bss->lock_owner = con->owner; + if (con->use_cgroup) { + read_all_cgroups(&con->cgroups); + } + bpf_program__set_autoload(skel->progs.collect_lock_syms, false); lock_contention_bpf__attach(skel); @@ -222,6 +227,17 @@ static const char *lock_contention_get_name(struct lock_contention *con, return ""; } + if (con->use_cgroup) { + u64 cgrp_id = key->lock_addr; + struct cgroup *cgrp = __cgroup__find(&con->cgroups, cgrp_id); + + if (cgrp) + return cgrp->name; + + snprintf(name_buf, sizeof(name_buf), "cgroup:%lu", cgrp_id); + return name_buf; + } + /* LOCK_AGGR_CALLER: skip lock internal functions */ while (machine__is_lock_function(machine, stack_trace[idx]) && idx < con->max_stack - 1) @@ -364,12 +380,20 @@ next: return err; } -int lock_contention_finish(void) +int lock_contention_finish(struct lock_contention *con) { if (skel) { skel->bss->enabled = 0; lock_contention_bpf__destroy(skel); } + while (!RB_EMPTY_ROOT(&con->cgroups)) { + struct rb_node *node = rb_first(&con->cgroups); + struct cgroup *cgrp = rb_entry(node, struct cgroup, node); + + rb_erase(node, &con->cgroups); + cgroup__put(cgrp); + } + return 0; } diff --git a/tools/perf/util/lock-contention.h b/tools/perf/util/lock-contention.h index fa16532c971c..70423966d778 100644 --- a/tools/perf/util/lock-contention.h +++ b/tools/perf/util/lock-contention.h @@ -136,6 +136,7 @@ struct lock_contention { struct hlist_head *result; struct lock_filter *filters; struct lock_contention_fails fails; + struct rb_root cgroups; unsigned long map_nr_entries; int max_stack; int stack_skip; @@ -143,6 +144,7 @@ struct lock_contention { int owner; int nr_filtered; bool save_callstack; + bool use_cgroup; }; #ifdef HAVE_BPF_SKEL @@ -151,7 +153,7 @@ int lock_contention_prepare(struct lock_contention *con); int lock_contention_start(void); int lock_contention_stop(void); int lock_contention_read(struct lock_contention *con); -int lock_contention_finish(void); +int lock_contention_finish(struct lock_contention *con); #else /* !HAVE_BPF_SKEL */ @@ -162,7 +164,10 @@ static inline int lock_contention_prepare(struct lock_contention *con __maybe_un static inline int lock_contention_start(void) { return 0; } static inline int lock_contention_stop(void) { return 0; } -static inline int lock_contention_finish(void) { return 0; } +static inline int lock_contention_finish(struct lock_contention *con __maybe_unused) +{ + return 0; +} static inline int lock_contention_read(struct lock_contention *con __maybe_unused) { -- cgit v1.2.3 From 4d1792d0a2564caf1620a2cc9aa6f12284b3a56f Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 6 Sep 2023 10:49:01 -0700 Subject: perf lock contention: Add --lock-cgroup option The --lock-cgroup option shows lock contention stats break down by cgroups. Add LOCK_AGGR_CGROUP mode and use it instead of use_cgroup field. $ sudo ./perf lock con -ab --lock-cgroup sleep 1 contended total wait max wait avg wait cgroup 8 15.70 us 6.34 us 1.96 us / 2 1.48 us 747 ns 738 ns /user.slice/.../app.slice/app-gnome-google\x2dchrome-6442.scope 1 848 ns 848 ns 848 ns /user.slice/.../session.slice/org.gnome.Shell@x11.service 1 220 ns 220 ns 220 ns /user.slice/.../session.slice/pipewire-pulse.service For now, the cgroup mode only works with BPF (-b). Committer notes: Remove -g as it is used in the other tools with a clear meaning of collect/show callchains. As agreed with Namhyung off list. Reviewed-by: Ian Rogers Signed-off-by: Namhyung Kim Tested-by: Arnaldo Carvalho de Melo Cc: Adrian Hunter Cc: Hao Luo Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230906174903.346486-4-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_lock_contention.c | 16 ++++++++----- tools/perf/util/bpf_skel/lock_contention.bpf.c | 31 +++++++++++++++++++++++++- tools/perf/util/bpf_skel/lock_data.h | 3 ++- tools/perf/util/lock-contention.h | 1 - 4 files changed, 42 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_lock_contention.c b/tools/perf/util/bpf_lock_contention.c index c6bd7c9b2d57..42753a0dfdc5 100644 --- a/tools/perf/util/bpf_lock_contention.c +++ b/tools/perf/util/bpf_lock_contention.c @@ -152,7 +152,10 @@ int lock_contention_prepare(struct lock_contention *con) skel->bss->needs_callstack = con->save_callstack; skel->bss->lock_owner = con->owner; - if (con->use_cgroup) { + if (con->aggr_mode == LOCK_AGGR_CGROUP) { + if (cgroup_is_v2("perf_event")) + skel->bss->use_cgroup_v2 = 1; + read_all_cgroups(&con->cgroups); } @@ -214,12 +217,12 @@ static const char *lock_contention_get_name(struct lock_contention *con, return "siglock"; /* global locks with symbols */ - sym = machine__find_kernel_symbol(machine, key->lock_addr, &kmap); + sym = machine__find_kernel_symbol(machine, key->lock_addr_or_cgroup, &kmap); if (sym) return sym->name; /* try semi-global locks collected separately */ - if (!bpf_map_lookup_elem(lock_fd, &key->lock_addr, &flags)) { + if (!bpf_map_lookup_elem(lock_fd, &key->lock_addr_or_cgroup, &flags)) { if (flags == LOCK_CLASS_RQLOCK) return "rq_lock"; } @@ -227,8 +230,8 @@ static const char *lock_contention_get_name(struct lock_contention *con, return ""; } - if (con->use_cgroup) { - u64 cgrp_id = key->lock_addr; + if (con->aggr_mode == LOCK_AGGR_CGROUP) { + u64 cgrp_id = key->lock_addr_or_cgroup; struct cgroup *cgrp = __cgroup__find(&con->cgroups, cgrp_id); if (cgrp) @@ -329,7 +332,8 @@ int lock_contention_read(struct lock_contention *con) ls_key = key.pid; break; case LOCK_AGGR_ADDR: - ls_key = key.lock_addr; + case LOCK_AGGR_CGROUP: + ls_key = key.lock_addr_or_cgroup; break; default: goto next; diff --git a/tools/perf/util/bpf_skel/lock_contention.bpf.c b/tools/perf/util/bpf_skel/lock_contention.bpf.c index 8d3cfbb3cc65..823354999022 100644 --- a/tools/perf/util/bpf_skel/lock_contention.bpf.c +++ b/tools/perf/util/bpf_skel/lock_contention.bpf.c @@ -118,6 +118,9 @@ int needs_callstack; int stack_skip; int lock_owner; +int use_cgroup_v2; +int perf_subsys_id = -1; + /* determine the key of lock stat */ int aggr_mode; @@ -130,6 +133,29 @@ int data_fail; int task_map_full; int data_map_full; +static inline __u64 get_current_cgroup_id(void) +{ + struct task_struct *task; + struct cgroup *cgrp; + + if (use_cgroup_v2) + return bpf_get_current_cgroup_id(); + + task = bpf_get_current_task_btf(); + + if (perf_subsys_id == -1) { +#if __has_builtin(__builtin_preserve_enum_value) + perf_subsys_id = bpf_core_enum_value(enum cgroup_subsys_id, + perf_event_cgrp_id); +#else + perf_subsys_id = perf_event_cgrp_id; +#endif + } + + cgrp = BPF_CORE_READ(task, cgroups, subsys[perf_subsys_id], cgroup); + return BPF_CORE_READ(cgrp, kn, id); +} + static inline int can_record(u64 *ctx) { if (has_cpu) { @@ -364,10 +390,13 @@ int contention_end(u64 *ctx) key.stack_id = pelem->stack_id; break; case LOCK_AGGR_ADDR: - key.lock_addr = pelem->lock; + key.lock_addr_or_cgroup = pelem->lock; if (needs_callstack) key.stack_id = pelem->stack_id; break; + case LOCK_AGGR_CGROUP: + key.lock_addr_or_cgroup = get_current_cgroup_id(); + break; default: /* should not happen */ return 0; diff --git a/tools/perf/util/bpf_skel/lock_data.h b/tools/perf/util/bpf_skel/lock_data.h index 260062a9f2ab..08482daf61be 100644 --- a/tools/perf/util/bpf_skel/lock_data.h +++ b/tools/perf/util/bpf_skel/lock_data.h @@ -6,7 +6,7 @@ struct contention_key { u32 stack_id; u32 pid; - u64 lock_addr; + u64 lock_addr_or_cgroup; }; #define TASK_COMM_LEN 16 @@ -39,6 +39,7 @@ enum lock_aggr_mode { LOCK_AGGR_ADDR = 0, LOCK_AGGR_TASK, LOCK_AGGR_CALLER, + LOCK_AGGR_CGROUP, }; enum lock_class_sym { diff --git a/tools/perf/util/lock-contention.h b/tools/perf/util/lock-contention.h index 70423966d778..a073cc6a82d2 100644 --- a/tools/perf/util/lock-contention.h +++ b/tools/perf/util/lock-contention.h @@ -144,7 +144,6 @@ struct lock_contention { int owner; int nr_filtered; bool save_callstack; - bool use_cgroup; }; #ifdef HAVE_BPF_SKEL -- cgit v1.2.3 From 4fd06bd2dcc893d922985cc4e9082477f6d115e6 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 6 Sep 2023 10:49:02 -0700 Subject: perf lock contention: Add -G/--cgroup-filter option The -G/--cgroup-filter is to limit lock contention collection on the tasks in the specific cgroups only. $ sudo ./perf lock con -abt -G /user.slice/.../vte-spawn-52221fb8-b33f-4a52-b5c3-e35d1e6fc0e0.scope \ ./perf bench sched messaging # Running 'sched/messaging' benchmark: # 20 sender and receiver processes per group # 10 groups == 400 processes run Total time: 0.174 [sec] contended total wait max wait avg wait pid comm 4 114.45 us 60.06 us 28.61 us 214847 sched-messaging 2 111.40 us 60.84 us 55.70 us 214848 sched-messaging 2 106.09 us 59.42 us 53.04 us 214837 sched-messaging 1 81.70 us 81.70 us 81.70 us 214709 sched-messaging 68 78.44 us 6.83 us 1.15 us 214633 sched-messaging 69 73.71 us 2.69 us 1.07 us 214632 sched-messaging 4 72.62 us 60.83 us 18.15 us 214850 sched-messaging 2 71.75 us 67.60 us 35.88 us 214840 sched-messaging 2 69.29 us 67.53 us 34.65 us 214804 sched-messaging 2 69.00 us 68.23 us 34.50 us 214826 sched-messaging ... Export cgroup__new() function as it's needed from outside. Reviewed-by: Ian Rogers Signed-off-by: Namhyung Kim Tested-by: Arnaldo Carvalho de Melo Cc: Adrian Hunter Cc: Hao Luo Cc: Ingo Molnar Cc: Jiri Olsa Cc: Peter Zijlstra Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230906174903.346486-5-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf_lock_contention.c | 15 ++++++++++++++- tools/perf/util/bpf_skel/lock_contention.bpf.c | 17 +++++++++++++++++ tools/perf/util/cgroup.c | 2 +- tools/perf/util/cgroup.h | 1 + tools/perf/util/lock-contention.h | 2 ++ 5 files changed, 35 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_lock_contention.c b/tools/perf/util/bpf_lock_contention.c index 42753a0dfdc5..e105245eb905 100644 --- a/tools/perf/util/bpf_lock_contention.c +++ b/tools/perf/util/bpf_lock_contention.c @@ -21,7 +21,7 @@ static struct lock_contention_bpf *skel; int lock_contention_prepare(struct lock_contention *con) { int i, fd; - int ncpus = 1, ntasks = 1, ntypes = 1, naddrs = 1; + int ncpus = 1, ntasks = 1, ntypes = 1, naddrs = 1, ncgrps = 1; struct evlist *evlist = con->evlist; struct target *target = con->target; @@ -51,6 +51,8 @@ int lock_contention_prepare(struct lock_contention *con) ntasks = perf_thread_map__nr(evlist->core.threads); if (con->filters->nr_types) ntypes = con->filters->nr_types; + if (con->filters->nr_cgrps) + ncgrps = con->filters->nr_cgrps; /* resolve lock name filters to addr */ if (con->filters->nr_syms) { @@ -85,6 +87,7 @@ int lock_contention_prepare(struct lock_contention *con) bpf_map__set_max_entries(skel->maps.task_filter, ntasks); bpf_map__set_max_entries(skel->maps.type_filter, ntypes); bpf_map__set_max_entries(skel->maps.addr_filter, naddrs); + bpf_map__set_max_entries(skel->maps.cgroup_filter, ncgrps); if (lock_contention_bpf__load(skel) < 0) { pr_err("Failed to load lock-contention BPF skeleton\n"); @@ -146,6 +149,16 @@ int lock_contention_prepare(struct lock_contention *con) bpf_map_update_elem(fd, &con->filters->addrs[i], &val, BPF_ANY); } + if (con->filters->nr_cgrps) { + u8 val = 1; + + skel->bss->has_cgroup = 1; + fd = bpf_map__fd(skel->maps.cgroup_filter); + + for (i = 0; i < con->filters->nr_cgrps; i++) + bpf_map_update_elem(fd, &con->filters->cgrps[i], &val, BPF_ANY); + } + /* these don't work well if in the rodata section */ skel->bss->stack_skip = con->stack_skip; skel->bss->aggr_mode = con->aggr_mode; diff --git a/tools/perf/util/bpf_skel/lock_contention.bpf.c b/tools/perf/util/bpf_skel/lock_contention.bpf.c index 823354999022..4900a5dfb4a4 100644 --- a/tools/perf/util/bpf_skel/lock_contention.bpf.c +++ b/tools/perf/util/bpf_skel/lock_contention.bpf.c @@ -92,6 +92,13 @@ struct { __uint(max_entries, 1); } addr_filter SEC(".maps"); +struct { + __uint(type, BPF_MAP_TYPE_HASH); + __uint(key_size, sizeof(__u64)); + __uint(value_size, sizeof(__u8)); + __uint(max_entries, 1); +} cgroup_filter SEC(".maps"); + struct rw_semaphore___old { struct task_struct *owner; } __attribute__((preserve_access_index)); @@ -114,6 +121,7 @@ int has_cpu; int has_task; int has_type; int has_addr; +int has_cgroup; int needs_callstack; int stack_skip; int lock_owner; @@ -194,6 +202,15 @@ static inline int can_record(u64 *ctx) return 0; } + if (has_cgroup) { + __u8 *ok; + __u64 cgrp = get_current_cgroup_id(); + + ok = bpf_map_lookup_elem(&cgroup_filter, &cgrp); + if (!ok) + return 0; + } + return 1; } diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c index d03651098dde..fcb509058499 100644 --- a/tools/perf/util/cgroup.c +++ b/tools/perf/util/cgroup.c @@ -114,7 +114,7 @@ static struct cgroup *evlist__find_cgroup(struct evlist *evlist, const char *str return NULL; } -static struct cgroup *cgroup__new(const char *name, bool do_open) +struct cgroup *cgroup__new(const char *name, bool do_open) { struct cgroup *cgroup = zalloc(sizeof(*cgroup)); diff --git a/tools/perf/util/cgroup.h b/tools/perf/util/cgroup.h index beb6fe1012ed..de8882d6e8d3 100644 --- a/tools/perf/util/cgroup.h +++ b/tools/perf/util/cgroup.h @@ -26,6 +26,7 @@ void cgroup__put(struct cgroup *cgroup); struct evlist; struct rblist; +struct cgroup *cgroup__new(const char *name, bool do_open); struct cgroup *evlist__findnew_cgroup(struct evlist *evlist, const char *name); int evlist__expand_cgroup(struct evlist *evlist, const char *cgroups, struct rblist *metric_events, bool open_cgroup); diff --git a/tools/perf/util/lock-contention.h b/tools/perf/util/lock-contention.h index a073cc6a82d2..1a7248ff3889 100644 --- a/tools/perf/util/lock-contention.h +++ b/tools/perf/util/lock-contention.h @@ -9,9 +9,11 @@ struct lock_filter { int nr_types; int nr_addrs; int nr_syms; + int nr_cgrps; unsigned int *types; unsigned long *addrs; char **syms; + u64 *cgrps; }; struct lock_stat { -- cgit v1.2.3 From 4f19fc1839e54d2c0d1e449a29a7578d029846d1 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Wed, 6 Sep 2023 16:44:15 -0700 Subject: perf list: Avoid a hardcoded cpu PMU name Use the first core PMU instead. On a Raspberry Pi, before: $ perf list ... cpu/t1=v1[,t2=v2,t3 ...]/modifier [Raw hardware event descriptor] [(see 'man perf-list' on how to encode it)] ... After: $ perf list ... armv8_cortex_a72/t1=v1[,t2=v2,t3 ...]/modifier [Raw hardware event descriptor] [(see 'man perf-list' on how to encode it)] ... ``` Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Kang Minchul Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Thomas Richter Cc: Xing Zhengjun Link: https://lore.kernel.org/r/20230906234416.3472339-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/print-events.c | 28 +++++++++++++++++----------- 1 file changed, 17 insertions(+), 11 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/print-events.c b/tools/perf/util/print-events.c index a7566edc86a3..b0fc48be623f 100644 --- a/tools/perf/util/print-events.c +++ b/tools/perf/util/print-events.c @@ -395,6 +395,8 @@ void print_symbol_events(const struct print_callbacks *print_cb, void *print_sta */ void print_events(const struct print_callbacks *print_cb, void *print_state) { + char *tmp; + print_symbol_events(print_cb, print_state, PERF_TYPE_HARDWARE, event_symbols_hw, PERF_COUNT_HW_MAX); print_symbol_events(print_cb, print_state, PERF_TYPE_SOFTWARE, @@ -418,17 +420,21 @@ void print_events(const struct print_callbacks *print_cb, void *print_state) /*long_desc=*/NULL, /*encoding_desc=*/NULL); - print_cb->print_event(print_state, - /*topic=*/NULL, - /*pmu_name=*/NULL, - "cpu/t1=v1[,t2=v2,t3 ...]/modifier", - /*event_alias=*/NULL, - /*scale_unit=*/NULL, - /*deprecated=*/false, - event_type_descriptors[PERF_TYPE_RAW], - "(see 'man perf-list' on how to encode it)", - /*long_desc=*/NULL, - /*encoding_desc=*/NULL); + if (asprintf(&tmp, "%s/t1=v1[,t2=v2,t3 ...]/modifier", + perf_pmus__scan_core(/*pmu=*/NULL)->name) > 0) { + print_cb->print_event(print_state, + /*topic=*/NULL, + /*pmu_name=*/NULL, + tmp, + /*event_alias=*/NULL, + /*scale_unit=*/NULL, + /*deprecated=*/false, + event_type_descriptors[PERF_TYPE_RAW], + "(see 'man perf-list' on how to encode it)", + /*long_desc=*/NULL, + /*encoding_desc=*/NULL); + free(tmp); + } print_cb->print_event(print_state, /*topic=*/NULL, -- cgit v1.2.3 From 6bd8c2ea6b93aabcb4a046ca7a6ad353b574d0a3 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Wed, 6 Sep 2023 16:44:16 -0700 Subject: perf list pfm: Retry supported test with exclude_kernel With paranoia set at 2 evsel__open will fail with EACCES for non-root users. To avoid this stopping libpfm4 events from being printed, retry with exclude_kernel enabled - copying the regular is_event_supported test. Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: Jiri Olsa Cc: Kan Liang Cc: Kang Minchul Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Thomas Richter Cc: Xing Zhengjun Link: https://lore.kernel.org/r/20230906234416.3472339-2-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/pfm.c | 15 ++++++++++++++- 1 file changed, 14 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pfm.c b/tools/perf/util/pfm.c index 862e4a689868..5ccfe4b64cdf 100644 --- a/tools/perf/util/pfm.c +++ b/tools/perf/util/pfm.c @@ -145,7 +145,20 @@ static bool is_libpfm_event_supported(const char *name, struct perf_cpu_map *cpu evsel->is_libpfm_event = true; - if (evsel__open(evsel, cpus, threads) < 0) + ret = evsel__open(evsel, cpus, threads); + if (ret == -EACCES) { + /* + * This happens if the paranoid value + * /proc/sys/kernel/perf_event_paranoid is set to 2 + * Re-run with exclude_kernel set; we don't do that + * by default as some ARM machines do not support it. + * + */ + evsel->core.attr.exclude_kernel = 1; + ret = evsel__open(evsel, cpus, threads); + + } + if (ret < 0) result = false; evsel__close(evsel); -- cgit v1.2.3 From 8a55c1e2c9e123b399b272a7db23f09dbb74af21 Mon Sep 17 00:00:00 2001 From: James Clark Date: Mon, 4 Sep 2023 10:50:45 +0100 Subject: perf util: Add a function for replacing characters in a string It finds all occurrences of a single character and replaces them with a multi character string. This will be used in a test in a following commit. Reviewed-by: Ian Rogers Signed-off-by: James Clark Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Chen Zhongjin Cc: Eduard Zingerman Cc: Haixin Yu Cc: Ingo Molnar Cc: Jing Zhang Cc: Jiri Olsa Cc: John Garry Cc: Kajol Jain Cc: Kan Liang Cc: Leo Yan Cc: Liam Howlett Cc: Madhavan Srinivasan Cc: Mark Rutland Cc: Mike Leach Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Will Deacon Cc: Yang Jihong Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20230904095104.1162928-4-james.clark@arm.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/string.c | 48 +++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/string2.h | 1 + 2 files changed, 49 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/string.c b/tools/perf/util/string.c index cf05b0b56c57..116a642ad99d 100644 --- a/tools/perf/util/string.c +++ b/tools/perf/util/string.c @@ -301,3 +301,51 @@ unsigned int hex(char c) return c - 'a' + 10; return c - 'A' + 10; } + +/* + * Replace all occurrences of character 'needle' in string 'haystack' with + * string 'replace' + * + * The new string could be longer so a new string is returned which must be + * freed. + */ +char *strreplace_chars(char needle, const char *haystack, const char *replace) +{ + int replace_len = strlen(replace); + char *new_s, *to; + const char *loc = strchr(haystack, needle); + const char *from = haystack; + int num = 0; + + /* Count occurrences */ + while (loc) { + loc = strchr(loc + 1, needle); + num++; + } + + /* Allocate enough space for replacements and reset first location */ + new_s = malloc(strlen(haystack) + (num * (replace_len - 1) + 1)); + if (!new_s) + return NULL; + loc = strchr(haystack, needle); + to = new_s; + + while (loc) { + /* Copy original string up to found char and update positions */ + memcpy(to, from, 1 + loc - from); + to += loc - from; + from = loc + 1; + + /* Copy replacement string and update positions */ + memcpy(to, replace, replace_len); + to += replace_len; + + /* needle next occurrence or end of string */ + loc = strchr(from, needle); + } + + /* Copy any remaining chars + null */ + strcpy(to, from); + + return new_s; +} diff --git a/tools/perf/util/string2.h b/tools/perf/util/string2.h index 56c30fef9682..52cb8ba057c7 100644 --- a/tools/perf/util/string2.h +++ b/tools/perf/util/string2.h @@ -39,5 +39,6 @@ char *strpbrk_esc(char *str, const char *stopset); char *strdup_esc(const char *str); unsigned int hex(char c); +char *strreplace_chars(char needle, const char *haystack, const char *replace); #endif /* PERF_STRING_H */ -- cgit v1.2.3 From dc2cfef9a9afe04a1b7683a999065b544d7b54f7 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 11 Sep 2023 10:05:55 -0700 Subject: perf parse-events: Remove unused header files The fnmatch header is now used in the PMU matching logic in pmu.c. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Gaosheng Cui Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Cc: Rob Herring Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230911170559.4037734-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.y | 3 --- 1 file changed, 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index bf74d6564eba..dcadf08f840a 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -9,11 +9,8 @@ #define YYDEBUG 1 #include -#include -#include #include #include -#include #include "pmu.h" #include "pmus.h" #include "evsel.h" -- cgit v1.2.3 From d4ce60190e08d84f88937019defa5e3d23409ac1 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 11 Sep 2023 10:05:56 -0700 Subject: perf parse-events: Make YYDEBUG dependent on doing a debug build YYDEBUG enables line numbers and other error helpers in the generated parse-events-bison.c. These shouldn't be generated when debugging isn't enabled. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Gaosheng Cui Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Cc: Rob Herring Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230911170559.4037734-2-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/parse-events.y | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index dcadf08f840a..c4ce888a2550 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -6,7 +6,9 @@ %{ +#ifndef NDEBUG #define YYDEBUG 1 +#endif #include #include -- cgit v1.2.3 From 1344a7077d78bb2ba72e539fc9119321d98fd06e Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 11 Sep 2023 10:05:57 -0700 Subject: perf expr: Make YYDEBUG dependent on doing a debug build YYDEBUG enables line numbers and other error helpers in the generated expr-bison.c. These shouldn't be generated when debugging isn't enabled. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Gaosheng Cui Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Cc: Rob Herring Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230911170559.4037734-3-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/expr.y | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/expr.y b/tools/perf/util/expr.y index 6c93b358cc2d..e364790babb5 100644 --- a/tools/perf/util/expr.y +++ b/tools/perf/util/expr.y @@ -1,6 +1,8 @@ /* Simple expression parser */ %{ +#ifndef NDEBUG #define YYDEBUG 1 +#endif #include #include #include -- cgit v1.2.3 From f0f4cd1003a03b0ae86be8c9ec8e88b0df384001 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 11 Sep 2023 10:05:58 -0700 Subject: perf pmu: Add YYDEBUG YYDEBUG enables line numbers and other error helpers in the generated pmu-bison.c. Conditionally enabled only for debug builds. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Gaosheng Cui Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Cc: Rob Herring Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230911170559.4037734-4-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/pmu.y | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.y b/tools/perf/util/pmu.y index 600c8c158c8e..198907a8a48a 100644 --- a/tools/perf/util/pmu.y +++ b/tools/perf/util/pmu.y @@ -5,6 +5,10 @@ %{ +#ifndef NDEBUG +#define YYDEBUG 1 +#endif + #include #include #include -- cgit v1.2.3 From 999b81b907ea92fa759e426591068244d9635496 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 11 Sep 2023 10:05:59 -0700 Subject: perf bpf-filter: Add YYDEBUG YYDEBUG enables line numbers and other error helpers in the generated bpf-filter-bison.c. Conditionally enabled only for debug builds. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Gaosheng Cui Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Peter Zijlstra Cc: Rob Herring Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230911170559.4037734-5-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/bpf-filter.y | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf-filter.y b/tools/perf/util/bpf-filter.y index 5dfa948fc986..0e4d6de3c2ad 100644 --- a/tools/perf/util/bpf-filter.y +++ b/tools/perf/util/bpf-filter.y @@ -3,6 +3,10 @@ %{ +#ifndef NDEBUG +#define YYDEBUG 1 +#endif + #include #include #include -- cgit v1.2.3 From 21ce931e55c19c1f74378b4836d9dae631da0e62 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Wed, 13 Sep 2023 21:42:32 -0700 Subject: perf symbol: Avoid an undefined behavior warning The node (nd) may be NULL and pointer arithmetic on NULL is undefined behavior. Move the computation of next below the NULL check on the node. Signed-off-by: Ian Rogers Cc: James Clark Link: https://lore.kernel.org/r/20230914044233.1550195-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/symbol.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 3f36675b7c8f..5b54d2639df4 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -202,11 +202,10 @@ void symbols__fixup_duplicate(struct rb_root_cached *symbols) curr = rb_entry(nd, struct symbol, rb_node); again: nd = rb_next(&curr->rb_node); - next = rb_entry(nd, struct symbol, rb_node); - if (!nd) break; + next = rb_entry(nd, struct symbol, rb_node); if (curr->start != next->start) continue; -- cgit v1.2.3 From 3d0f5f456a5786573ba6a3358178c8db580e4b85 Mon Sep 17 00:00:00 2001 From: James Clark Date: Wed, 13 Sep 2023 16:33:48 +0100 Subject: perf pmu: Move pmu__find_core_pmu() to pmus.c pmu__find_core_pmu() more logically belongs in pmus.c because it iterates over all PMUs, so move it to pmus.c At the same time rename it to perf_pmus__find_core_pmu() to match the naming convention in this file. list_prepare_entry() can't be used in perf_pmus__scan_core() anymore now that it's called from the same compilation unit. This is with -O2 (specifically -O1 -ftree-vrp -finline-functions -finline-small-functions) which allow the bounds of the array access to be determined at compile time. list_prepare_entry() subtracts the offset of the 'list' member in struct perf_pmu from &core_pmus, which isn't a struct perf_pmu. The compiler sees that pmu results in &core_pmus - 8 and refuses to compile. At runtime this works because list_for_each_entry_continue() always adds the offset back again before dereferencing ->next, but it's technically undefined behavior. With -fsanitize=undefined an additional warning is generated. Using list_first_entry_or_null() to get the first entry here avoids doing &core_pmus - 8 but has the same result and fixes both the compile warning and the undefined behavior warning. There are other uses of list_prepare_entry() in pmus.c, but the compiler doesn't seem to be able to see that they can also be called with &core_pmus, so I won't change any at this time. Signed-off-by: James Clark Reviewed-by: Ian Rogers Reviewed-by: John Garry Cc: Ravi Bangoria Cc: Eduard Zingerman Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Haixin Yu Cc: Kan Liang Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20230913153355.138331-2-james.clark@arm.com Signed-off-by: Namhyung Kim --- tools/perf/util/expr.c | 2 +- tools/perf/util/pmu.c | 17 ----------------- tools/perf/util/pmu.h | 2 +- tools/perf/util/pmus.c | 20 +++++++++++++++++++- 4 files changed, 21 insertions(+), 20 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/expr.c b/tools/perf/util/expr.c index 4488f306de78..7be23b3ac082 100644 --- a/tools/perf/util/expr.c +++ b/tools/perf/util/expr.c @@ -509,7 +509,7 @@ double expr__strcmp_cpuid_str(const struct expr_parse_ctx *ctx __maybe_unused, bool compute_ids __maybe_unused, const char *test_id) { double ret; - struct perf_pmu *pmu = pmu__find_core_pmu(); + struct perf_pmu *pmu = perf_pmus__find_core_pmu(); char *cpuid = perf_pmu__getcpuid(pmu); if (!cpuid) diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index e2159854ab26..f50a5636633f 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -2050,20 +2050,3 @@ void perf_pmu__delete(struct perf_pmu *pmu) zfree(&pmu->id); free(pmu); } - -struct perf_pmu *pmu__find_core_pmu(void) -{ - struct perf_pmu *pmu = NULL; - - while ((pmu = perf_pmus__scan_core(pmu))) { - /* - * The cpumap should cover all CPUs. Otherwise, some CPUs may - * not support some events or have different event IDs. - */ - if (RC_CHK_ACCESS(pmu->cpus)->nr != cpu__max_cpu().cpu) - return NULL; - - return pmu; - } - return NULL; -} diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index bd5d804a6736..d7b46085642d 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -264,6 +264,6 @@ int perf_pmu__pathname_fd(int dirfd, const char *pmu_name, const char *filename, struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char *lookup_name); struct perf_pmu *perf_pmu__create_placeholder_core_pmu(struct list_head *core_pmus); void perf_pmu__delete(struct perf_pmu *pmu); -struct perf_pmu *pmu__find_core_pmu(void); +struct perf_pmu *perf_pmus__find_core_pmu(void); #endif /* __PMU_H */ diff --git a/tools/perf/util/pmus.c b/tools/perf/util/pmus.c index 6631367c756f..cec869cbe163 100644 --- a/tools/perf/util/pmus.c +++ b/tools/perf/util/pmus.c @@ -10,6 +10,7 @@ #include #include #include +#include "cpumap.h" #include "debug.h" #include "evsel.h" #include "pmus.h" @@ -268,7 +269,7 @@ struct perf_pmu *perf_pmus__scan_core(struct perf_pmu *pmu) { if (!pmu) { pmu_read_sysfs(/*core_only=*/true); - pmu = list_prepare_entry(pmu, &core_pmus, list); + return list_first_entry_or_null(&core_pmus, typeof(*pmu), list); } list_for_each_entry_continue(pmu, &core_pmus, list) return pmu; @@ -592,3 +593,20 @@ struct perf_pmu *evsel__find_pmu(const struct evsel *evsel) } return pmu; } + +struct perf_pmu *perf_pmus__find_core_pmu(void) +{ + struct perf_pmu *pmu = NULL; + + while ((pmu = perf_pmus__scan_core(pmu))) { + /* + * The cpumap should cover all CPUs. Otherwise, some CPUs may + * not support some events or have different event IDs. + */ + if (RC_CHK_ACCESS(pmu->cpus)->nr != cpu__max_cpu().cpu) + return NULL; + + return pmu; + } + return NULL; +} -- cgit v1.2.3 From 105e5b433e5c743bcdb4956b00b772dfddbee455 Mon Sep 17 00:00:00 2001 From: James Clark Date: Wed, 13 Sep 2023 16:33:49 +0100 Subject: perf pmus: Simplify perf_pmus__find_core_pmu() Currently the while loop always either exits on the first iteration with a core PMU, or exits with NULL on heterogeneous systems or when not all CPUs are online. Both of the latter behaviors are undesirable for platforms other than Arm so simplify it to always return the first core PMU, or NULL if none exist. This behavior was depended on by the Arm version of pmu_metrics_table__find(), so the logic has been moved there instead. Signed-off-by: James Clark Suggested-by: Ian Rogers Reviewed-by: Ian Rogers Reviewed-by: John Garry Cc: Ravi Bangoria Cc: Eduard Zingerman Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Haixin Yu Cc: Kan Liang Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20230913153355.138331-3-james.clark@arm.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmus.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmus.c b/tools/perf/util/pmus.c index cec869cbe163..64e798e68a2d 100644 --- a/tools/perf/util/pmus.c +++ b/tools/perf/util/pmus.c @@ -596,17 +596,5 @@ struct perf_pmu *evsel__find_pmu(const struct evsel *evsel) struct perf_pmu *perf_pmus__find_core_pmu(void) { - struct perf_pmu *pmu = NULL; - - while ((pmu = perf_pmus__scan_core(pmu))) { - /* - * The cpumap should cover all CPUs. Otherwise, some CPUs may - * not support some events or have different event IDs. - */ - if (RC_CHK_ACCESS(pmu->cpus)->nr != cpu__max_cpu().cpu) - return NULL; - - return pmu; - } - return NULL; + return perf_pmus__scan_core(NULL); } -- cgit v1.2.3 From 70360fad919b06100748764eb82fa399a9f37e1c Mon Sep 17 00:00:00 2001 From: James Clark Date: Wed, 13 Sep 2023 16:33:50 +0100 Subject: perf pmu: Remove unused function pmu_events_table__find() is no longer used so remove it and its Arm specific version. Signed-off-by: James Clark Reviewed-by: Ian Rogers Reviewed-by: John Garry Cc: Ravi Bangoria Cc: Eduard Zingerman Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Haixin Yu Cc: Kan Liang Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20230913153355.138331-4-james.clark@arm.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 5 ----- tools/perf/util/pmu.h | 1 - 2 files changed, 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index f50a5636633f..0d81c059c91c 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -776,11 +776,6 @@ char *perf_pmu__getcpuid(struct perf_pmu *pmu) return cpuid; } -__weak const struct pmu_events_table *pmu_events_table__find(void) -{ - return perf_pmu__find_events_table(NULL); -} - __weak const struct pmu_metrics_table *pmu_metrics_table__find(void) { return perf_pmu__find_metrics_table(NULL); diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index d7b46085642d..04b317b17d66 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -238,7 +238,6 @@ void pmu_add_cpu_aliases_table(struct perf_pmu *pmu, const struct pmu_events_table *table); char *perf_pmu__getcpuid(struct perf_pmu *pmu); -const struct pmu_events_table *pmu_events_table__find(void); const struct pmu_metrics_table *pmu_metrics_table__find(void); int perf_pmu__convert_scale(const char *scale, char **end, double *sval); -- cgit v1.2.3 From 3ecf87b2d8ffabb88d1e3901d7a429fd1ecbfc6c Mon Sep 17 00:00:00 2001 From: Yang Li Date: Fri, 15 Sep 2023 14:38:32 +0800 Subject: perf kwork top: Simplify bool conversion ./tools/perf/util/bpf_kwork_top.c:120:53-58: WARNING: conversion to bool not needed here Signed-off-by: Yang Li Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230915063832.120274-1-yang.lee@linux.alibaba.com Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_kwork_top.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_kwork_top.c b/tools/perf/util/bpf_kwork_top.c index 1a607b94f44d..035e02272790 100644 --- a/tools/perf/util/bpf_kwork_top.c +++ b/tools/perf/util/bpf_kwork_top.c @@ -117,7 +117,7 @@ kwork_class_bpf_supported_list[KWORK_CLASS_MAX] = { static bool valid_kwork_class_type(enum kwork_class_type type) { - return type >= 0 && type < KWORK_CLASS_MAX ? true : false; + return type >= 0 && type < KWORK_CLASS_MAX; } static int setup_filters(struct perf_kwork *kwork) -- cgit v1.2.3 From ede72dca45b1893f3e9236b6ad6c4e790db232f6 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 14 Sep 2023 09:40:28 -0700 Subject: perf parse-events: Fix tracepoint name memory leak Fuzzing found that an invalid tracepoint name would create a memory leak with an address sanitizer build: ``` $ perf stat -e '*:o/' true event syntax error: '*:o/' \___ parser error Run 'perf list' for a list of valid events Usage: perf stat [] [] -e, --event event selector. use 'perf list' to list available events ================================================================= ==59380==ERROR: LeakSanitizer: detected memory leaks Direct leak of 4 byte(s) in 2 object(s) allocated from: #0 0x7f38ac07077b in __interceptor_strdup ../../../../src/libsanitizer/asan/asan_interceptors.cpp:439 #1 0x55f2f41be73b in str util/parse-events.l:49 #2 0x55f2f41d08e8 in parse_events_lex util/parse-events.l:338 #3 0x55f2f41dc3b1 in parse_events_parse util/parse-events-bison.c:1464 #4 0x55f2f410b8b3 in parse_events__scanner util/parse-events.c:1822 #5 0x55f2f410d1b9 in __parse_events util/parse-events.c:2094 #6 0x55f2f410e57f in parse_events_option util/parse-events.c:2279 #7 0x55f2f4427b56 in get_value tools/lib/subcmd/parse-options.c:251 #8 0x55f2f4428d98 in parse_short_opt tools/lib/subcmd/parse-options.c:351 #9 0x55f2f4429d80 in parse_options_step tools/lib/subcmd/parse-options.c:539 #10 0x55f2f442acb9 in parse_options_subcommand tools/lib/subcmd/parse-options.c:654 #11 0x55f2f3ec99fc in cmd_stat tools/perf/builtin-stat.c:2501 #12 0x55f2f4093289 in run_builtin tools/perf/perf.c:322 #13 0x55f2f40937f5 in handle_internal_command tools/perf/perf.c:375 #14 0x55f2f4093bbd in run_argv tools/perf/perf.c:419 #15 0x55f2f409412b in main tools/perf/perf.c:535 SUMMARY: AddressSanitizer: 4 byte(s) leaked in 2 allocation(s). ``` Fix by adding the missing destructor. Fixes: 865582c3f48e ("perf tools: Adds the tracepoint name parsing support") Signed-off-by: Ian Rogers Cc: He Kuang Link: https://lore.kernel.org/r/20230914164028.363220-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/parse-events.y | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index c4ce888a2550..4d9f3922a33d 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -103,6 +103,7 @@ static void free_list_evsel(struct list_head* list_evsel) %type groups %destructor { free_list_evsel ($$); } %type tracepoint_name +%destructor { free ($$.sys); free ($$.event); } %type PE_TERM_HW %destructor { free ($$.str); } -- cgit v1.2.3 From f9cdeb58a9cf46c09b56f5f661ea8da24b6458c3 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Fri, 15 Sep 2023 20:56:40 -0700 Subject: perf evlist: Avoid frequency mode for the dummy event Dummy events are created with an attribute where the period and freq are zero. evsel__config will then see the uninitialized values and initialize them in evsel__default_freq_period. As fequency mode is used by default the dummy event would be set to use frequency mode. However, this has no effect on the dummy event but does cause unnecessary timers/interrupts. Avoid this overhead by setting the period to 1 for dummy events. evlist__add_aux_dummy calls evlist__add_dummy then sets freq=0 and period=1. This isn't necessary after this change and so the setting is removed. From Stephane: The dummy event is not counting anything. It is used to collect mmap records and avoid a race condition during the synthesize mmap phase of perf record. As such, it should not cause any overhead during active profiling. Yet, it did. Because of a bug the dummy event was programmed as a sampling event in frequency mode. Events in that mode incur more kernel overheads because on timer tick, the kernel has to look at the number of samples for each event and potentially adjust the sampling period to achieve the desired frequency. The dummy event was therefore adding a frequency event to task and ctx contexts we may otherwise not have any, e.g., perf record -a -e cpu/event=0x3c,period=10000000/. On each timer tick the perf_adjust_freq_unthr_context() is invoked and if ctx->nr_freq is non-zero, then the kernel will loop over ALL the events of the context looking for frequency mode ones. In doing, so it locks the context, and enable/disable the PMU of each hw event. If all the events of the context are in period mode, the kernel will have to traverse the list for nothing incurring overhead. The overhead is multiplied by a very large factor when this happens in a guest kernel. There is no need for the dummy event to be in frequency mode, it does not count anything and therefore should not cause extra overhead for no reason. Fixes: 5bae0250237f ("perf evlist: Introduce perf_evlist__new_dummy constructor") Reported-by: Stephane Eranian Signed-off-by: Ian Rogers Acked-by: Adrian Hunter Cc: Yang Jihong Cc: Kan Liang Link: https://lore.kernel.org/r/20230916035640.1074422-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/evlist.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 25c3ebe2c2f5..e36da58522ef 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -251,6 +251,9 @@ static struct evsel *evlist__dummy_event(struct evlist *evlist) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_DUMMY, .size = sizeof(attr), /* to capture ABI version */ + /* Avoid frequency mode for dummy events to avoid associated timers. */ + .freq = 0, + .sample_period = 1, }; return evsel__new_idx(&attr, evlist->core.nr_entries); @@ -277,8 +280,6 @@ struct evsel *evlist__add_aux_dummy(struct evlist *evlist, bool system_wide) evsel->core.attr.exclude_kernel = 1; evsel->core.attr.exclude_guest = 1; evsel->core.attr.exclude_hv = 1; - evsel->core.attr.freq = 0; - evsel->core.attr.sample_period = 1; evsel->core.system_wide = system_wide; evsel->no_aux_samples = true; evsel->name = strdup("dummy:u"); -- cgit v1.2.3 From 0e501a65d35bf72414379fed0e31a0b6b81ab57d Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 22 Sep 2023 16:44:44 -0700 Subject: perf record: Fix BTF type checks in the off-cpu profiling The BTF func proto for a tracepoint has one more argument than the actual tracepoint function since it has a context argument at the begining. So it should compare to 5 when the tracepoint has 4 arguments. typedef void (*btf_trace_sched_switch)(void *, bool, struct task_struct *, struct task_struct *, unsigned int); Also, recent change in the perf tool would use a hand-written minimal vmlinux.h to generate BTF in the skeleton. So it won't have the info of the tracepoint. Anyway it should use the kernel's vmlinux BTF to check the type in the kernel. Fixes: b36888f71c85 ("perf record: Handle argument change in sched_switch") Reviewed-by: Ian Rogers Acked-by: Song Liu Cc: Hao Luo CC: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20230922234444.3115821-1-namhyung@kernel.org Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_off_cpu.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c index 01f70b8e705a..21f4d9ba023d 100644 --- a/tools/perf/util/bpf_off_cpu.c +++ b/tools/perf/util/bpf_off_cpu.c @@ -98,7 +98,7 @@ static void off_cpu_finish(void *arg __maybe_unused) /* v5.18 kernel added prev_state arg, so it needs to check the signature */ static void check_sched_switch_args(void) { - const struct btf *btf = bpf_object__btf(skel->obj); + const struct btf *btf = btf__load_vmlinux_btf(); const struct btf_type *t1, *t2, *t3; u32 type_id; @@ -116,7 +116,8 @@ static void check_sched_switch_args(void) return; t3 = btf__type_by_id(btf, t2->type); - if (t3 && btf_is_func_proto(t3) && btf_vlen(t3) == 4) { + /* btf_trace func proto has one more argument for the context */ + if (t3 && btf_is_func_proto(t3) && btf_vlen(t3) == 5) { /* new format: pass prev_state as 4th arg */ skel->rodata->has_prev_state = true; } -- cgit v1.2.3 From 2879ff36f5ed80deec5f9d82a7a4107f2347630e Mon Sep 17 00:00:00 2001 From: Jing Zhang Date: Wed, 27 Sep 2023 13:59:45 +0800 Subject: perf pmu: "Compat" supports regular expression matching identifiers The jevent "Compat" is used for uncore PMU alias or metric definitions. The same PMU driver has different PMU identifiers due to different hardware versions and types, but they may have some common PMU event. Since a Compat value can only match one identifier, when adding the same event alias to PMUs with different identifiers, each identifier needs to be defined once, which is not streamlined enough. So let "Compat" support using regular expression to match identifiers for uncore PMU alias. For example, if the "Compat" value is set to "43401|43c01", it would be able to match PMU identifiers such as "43401" or "43c01", which correspond to CMN600_r0p0 or CMN700_r0p0. Signed-off-by: Jing Zhang Reviewed-by: Ian Rogers Tested-by: Ian Rogers Cc: James Clark Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Shuai Xue Cc: Zhuo Song Cc: John Garry Cc: linux-arm-kernel@lists.infradead.org Cc: linux-doc@vger.kernel.org Link: https://lore.kernel.org/r/1695794391-34817-2-git-send-email-renyu.zj@linux.alibaba.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 27 +++++++++++++++++++++++++-- tools/perf/util/pmu.h | 1 + 2 files changed, 26 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 0d81c059c91c..5b417509b8a9 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -28,6 +28,7 @@ #include "strbuf.h" #include "fncache.h" #include "util/evsel_config.h" +#include struct perf_pmu perf_pmu__fake = { .name = "fake", @@ -870,6 +871,28 @@ out: return res; } +bool pmu_uncore_identifier_match(const char *compat, const char *id) +{ + regex_t re; + regmatch_t pmatch[1]; + int match; + + if (regcomp(&re, compat, REG_EXTENDED) != 0) { + /* Warn unable to generate match particular string. */ + pr_info("Invalid regular expression %s\n", compat); + return false; + } + + match = !regexec(&re, id, 1, pmatch, 0); + if (match) { + /* Ensure a full match. */ + match = pmatch[0].rm_so == 0 && (size_t)pmatch[0].rm_eo == strlen(id); + } + regfree(&re); + + return match; +} + static int pmu_add_cpu_aliases_map_callback(const struct pmu_event *pe, const struct pmu_events_table *table __maybe_unused, void *vdata) @@ -910,8 +933,8 @@ static int pmu_add_sys_aliases_iter_fn(const struct pmu_event *pe, if (!pe->compat || !pe->pmu) return 0; - if (!strcmp(pmu->id, pe->compat) && - pmu_uncore_alias_match(pe->pmu, pmu->name)) { + if (pmu_uncore_alias_match(pe->pmu, pmu->name) && + pmu_uncore_identifier_match(pe->compat, pmu->id)) { perf_pmu__new_alias(pmu, pe->name, pe->desc, diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 04b317b17d66..a682c6bc0fce 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -239,6 +239,7 @@ void pmu_add_cpu_aliases_table(struct perf_pmu *pmu, char *perf_pmu__getcpuid(struct perf_pmu *pmu); const struct pmu_metrics_table *pmu_metrics_table__find(void); +bool pmu_uncore_identifier_match(const char *compat, const char *id); int perf_pmu__convert_scale(const char *scale, char **end, double *sval); -- cgit v1.2.3 From 54409997d4b99ab63616bd431cf6244d58f8a597 Mon Sep 17 00:00:00 2001 From: Jing Zhang Date: Wed, 27 Sep 2023 13:59:46 +0800 Subject: perf metric: "Compat" supports regular expression matching identifiers The jevent "Compat" is used for uncore PMU alias or metric definitions. The same PMU driver has different PMU identifiers due to different hardware versions and types, but they may have some common PMU metric. Since a Compat value can only match one identifier, when adding the same metric to PMUs with different identifiers, each identifier needs to be defined once, which is not streamlined enough. So let "Compat" support using regular expression to match multiple identifiers for uncore PMU metric. Signed-off-by: Jing Zhang Reviewed-by: John Garry Reviewed-by: Ian Rogers Tested-by: Ian Rogers Cc: James Clark Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Shuai Xue Cc: Zhuo Song Cc: linux-arm-kernel@lists.infradead.org Cc: linux-doc@vger.kernel.org Link: https://lore.kernel.org/r/1695794391-34817-3-git-send-email-renyu.zj@linux.alibaba.com Signed-off-by: Namhyung Kim --- tools/perf/util/metricgroup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c index 6231044a491e..0484736d9fe4 100644 --- a/tools/perf/util/metricgroup.c +++ b/tools/perf/util/metricgroup.c @@ -498,7 +498,7 @@ static int metricgroup__sys_event_iter(const struct pmu_metric *pm, while ((pmu = perf_pmus__scan(pmu))) { - if (!pmu->id || strcmp(pmu->id, pm->compat)) + if (!pmu->id || !pmu_uncore_identifier_match(pm->compat, pmu->id)) continue; return d->fn(pm, table, d->data); -- cgit v1.2.3 From b1f05622fef39dded385f9e360e859846c1ddaf1 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Sun, 24 Sep 2023 23:23:23 -0700 Subject: perf pmus: Make PMU alias name loading lazy PMU alias names were computed when the first perf_pmu is created, scanning all PMUs in event sources for a file called alias that generally doesn't exist. Switch to trying to load the file when all PMU related files are loaded in lookup. This would cause a PMU name lookup of an alias name to fail if no PMUs were loaded, so in that case all PMUs are loaded and the find repeated. The overhead is similar but in the (very) general case not all PMUs are scanned for the alias file. As the overhead occurs once per invocation it doesn't show in perf bench internals pmu-scan. On a tigerlake machine, the number of openat system calls for an event of cpu/cycles/ with perf stat reduces from 94 to 69 (ie 25 fewer openat calls). Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: James Clark Cc: Leo Yan Cc: Kan Liang Link: https://lore.kernel.org/r/20230925062323.840799-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 39 +++++++++++++++++++++------------------ tools/perf/util/pmu.h | 2 -- tools/perf/util/pmus.c | 10 ++++++++++ 3 files changed, 31 insertions(+), 20 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 5b417509b8a9..6b1b7f8f00fa 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -960,16 +960,27 @@ perf_pmu__get_default_config(struct perf_pmu *pmu __maybe_unused) return NULL; } -const char * __weak -pmu_find_real_name(const char *name) +static char *pmu_find_alias_name(struct perf_pmu *pmu, int dirfd) { - return name; -} + FILE *file = perf_pmu__open_file_at(pmu, dirfd, "alias"); + char *line = NULL; + size_t line_len = 0; + ssize_t ret; -const char * __weak -pmu_find_alias_name(const char *name __maybe_unused) -{ - return NULL; + if (!file) + return NULL; + + ret = getline(&line, &line_len, file); + if (ret < 0) { + fclose(file); + return NULL; + } + /* Remove trailing newline. */ + if (ret > 0 && line[ret - 1] == '\n') + line[--ret] = '\0'; + + fclose(file); + return line; } static int pmu_max_precise(int dirfd, struct perf_pmu *pmu) @@ -980,12 +991,10 @@ static int pmu_max_precise(int dirfd, struct perf_pmu *pmu) return max_precise; } -struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char *lookup_name) +struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char *name) { struct perf_pmu *pmu; __u32 type; - const char *name = pmu_find_real_name(lookup_name); - const char *alias_name; pmu = zalloc(sizeof(*pmu)); if (!pmu) @@ -1018,18 +1027,12 @@ struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char pmu->is_core = is_pmu_core(name); pmu->cpus = pmu_cpumask(dirfd, name, pmu->is_core); - alias_name = pmu_find_alias_name(name); - if (alias_name) { - pmu->alias_name = strdup(alias_name); - if (!pmu->alias_name) - goto err; - } - pmu->type = type; pmu->is_uncore = pmu_is_uncore(dirfd, name); if (pmu->is_uncore) pmu->id = pmu_id(name); pmu->max_precise = pmu_max_precise(dirfd, pmu); + pmu->alias_name = pmu_find_alias_name(pmu, dirfd); pmu->events_table = perf_pmu__find_events_table(pmu); pmu_add_sys_aliases(pmu); list_add_tail(&pmu->list, pmus); diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index a682c6bc0fce..85190d058852 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -252,8 +252,6 @@ void perf_pmu__warn_invalid_formats(struct perf_pmu *pmu); int perf_pmu__match(const char *pattern, const char *name, const char *tok); -const char *pmu_find_real_name(const char *name); -const char *pmu_find_alias_name(const char *name); double perf_pmu__cpu_slots_per_cycle(void); int perf_pmu__event_source_devices_scnprintf(char *pathname, size_t size); int perf_pmu__pathname_scnprintf(char *buf, size_t size, diff --git a/tools/perf/util/pmus.c b/tools/perf/util/pmus.c index 64e798e68a2d..ce4931461741 100644 --- a/tools/perf/util/pmus.c +++ b/tools/perf/util/pmus.c @@ -37,6 +37,8 @@ static LIST_HEAD(other_pmus); static bool read_sysfs_core_pmus; static bool read_sysfs_all_pmus; +static void pmu_read_sysfs(bool core_only); + int pmu_name_len_no_suffix(const char *str, unsigned long *num) { int orig_len, len; @@ -124,6 +126,14 @@ struct perf_pmu *perf_pmus__find(const char *name) pmu = perf_pmu__lookup(core_pmu ? &core_pmus : &other_pmus, dirfd, name); close(dirfd); + if (!pmu) { + /* + * Looking up an inidividual PMU failed. This may mean name is + * an alias, so read the PMUs from sysfs and try to find again. + */ + pmu_read_sysfs(core_pmu); + pmu = pmu_find(name); + } return pmu; } -- cgit v1.2.3 From f2d87895cbc4af80649850dcf5da36de6b2ed3dd Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 28 Sep 2023 10:29:53 +0300 Subject: perf intel-pt: Fix async branch flags Ensure PERF_IP_FLAG_ASYNC is set always for asynchronous branches (i.e. interrupts etc). Fixes: 90e457f7be08 ("perf tools: Add Intel PT support") Cc: stable@vger.kernel.org Signed-off-by: Adrian Hunter Acked-by: Namhyung Kim Link: https://lore.kernel.org/r/20230928072953.19369-1-adrian.hunter@intel.com Signed-off-by: Namhyung Kim --- tools/perf/util/intel-pt.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c index dbf0bc71a63b..f38893e0b036 100644 --- a/tools/perf/util/intel-pt.c +++ b/tools/perf/util/intel-pt.c @@ -1512,9 +1512,11 @@ static void intel_pt_sample_flags(struct intel_pt_queue *ptq) } else if (ptq->state->flags & INTEL_PT_ASYNC) { if (!ptq->state->to_ip) ptq->flags = PERF_IP_FLAG_BRANCH | + PERF_IP_FLAG_ASYNC | PERF_IP_FLAG_TRACE_END; else if (ptq->state->from_nr && !ptq->state->to_nr) ptq->flags = PERF_IP_FLAG_BRANCH | PERF_IP_FLAG_CALL | + PERF_IP_FLAG_ASYNC | PERF_IP_FLAG_VMEXIT; else ptq->flags = PERF_IP_FLAG_BRANCH | PERF_IP_FLAG_CALL | -- cgit v1.2.3 From be7a4caa7c45bd4b0a39cdb260905b52a87c8688 Mon Sep 17 00:00:00 2001 From: Kuan-Wei Chiu Date: Sat, 30 Sep 2023 15:27:19 +0800 Subject: perf hisi-ptt: Fix memory leak in lseek failure handling In the previous code, there was a memory leak issue where the previously allocated memory was not freed upon a failed lseek operation. This patch addresses the problem by releasing the old memory before returning -errno in case of a lseek failure. This ensures that memory is properly managed and avoids potential memory leaks. Signed-off-by: Kuan-Wei Chiu Acked-by: Namhyung Kim Cc: yangyicong@hisilicon.com Cc: jonathan.cameron@huawei.com Link: https://lore.kernel.org/r/20230930072719.1267784-1-visitorckw@gmail.com Signed-off-by: Namhyung Kim --- tools/perf/util/hisi-ptt.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hisi-ptt.c b/tools/perf/util/hisi-ptt.c index 45b614bb73bf..43bd1ca62d58 100644 --- a/tools/perf/util/hisi-ptt.c +++ b/tools/perf/util/hisi-ptt.c @@ -108,8 +108,10 @@ static int hisi_ptt_process_auxtrace_event(struct perf_session *session, data_offset = 0; } else { data_offset = lseek(fd, 0, SEEK_CUR); - if (data_offset == -1) + if (data_offset == -1) { + free(data); return -errno; + } } err = readn(fd, data, size); -- cgit v1.2.3 From 26a5262d30e1452071bcfe86725f2fe4e164db04 Mon Sep 17 00:00:00 2001 From: Athira Rajeev Date: Thu, 28 Sep 2023 13:22:11 +0530 Subject: tools/perf: Add text_end to "struct dso" to save .text section size Update "struct dso" to include new member "text_end". This new field will represent the offset for end of text section for a dso. For elf, this value is derived as: sh_size (Size of section in byes) + sh_offset (Section file offst) of the elf header for text. For bfd, this value is derived as: 1. For PE file, section->size + ( section->vma - dso->text_offset) 2. Other cases: section->filepos (file position) + section->size (size of section) To resolve the address from a sample, perf looks at the DSO maps. In case of address from a kernel module, there were some address found to be not resolved. This was observed while running perf test for "Object code reading". Though the ip falls beteen the start address of the loaded module (perf map->start ) and end address ( perf map->end), it was unresolved. Example: Reading object code for memory address: 0xc008000007f0142c File is: /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko On file address is: 0x1114cc Objdump command is: objdump -z -d --start-address=0x11142c --stop-address=0x1114ac /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko objdump read too few bytes: 128 test child finished with -1 Here, module is loaded at: # cat /proc/modules | grep xfs xfs 2228224 3 - Live 0xc008000007d00000 From objdump for xfs module, text section is: text 0010f7bc 0000000000000000 0000000000000000 000000a0 2**4 Here the offset for 0xc008000007f0142c ie 0x112074 falls out .text section which is up to 0x10f7bc. In this case for module, the address 0xc008000007e11fd4 is pointing to stub instructions. This address range represents the module stubs which is allocated on module load and hence is not part of DSO offset. To identify such address, which falls out of text section and within module end, added the new field "text_end" to "struct dso". Reported-by: Disha Goel Signed-off-by: Athira Rajeev Reviewed-by: Adrian Hunter Reviewed-by: Kajol Jain Acked-by: Namhyung Kim Cc: maddy@linux.ibm.com Cc: disgoel@linux.vnet.ibm.com Cc: linuxppc-dev@lists.ozlabs.org Link: https://lore.kernel.org/r/20230928075213.84392-1-atrajeev@linux.vnet.ibm.com Signed-off-by: Namhyung Kim --- tools/perf/util/dso.h | 1 + tools/perf/util/symbol-elf.c | 4 +++- tools/perf/util/symbol.c | 2 ++ 3 files changed, 6 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/dso.h b/tools/perf/util/dso.h index b41c9782c754..70fe0fe69bef 100644 --- a/tools/perf/util/dso.h +++ b/tools/perf/util/dso.h @@ -181,6 +181,7 @@ struct dso { u8 rel; struct build_id bid; u64 text_offset; + u64 text_end; const char *short_name; const char *long_name; u16 long_name_len; diff --git a/tools/perf/util/symbol-elf.c b/tools/perf/util/symbol-elf.c index 95e99c332d7e..9e7eeaf616b8 100644 --- a/tools/perf/util/symbol-elf.c +++ b/tools/perf/util/symbol-elf.c @@ -1514,8 +1514,10 @@ dso__load_sym_internal(struct dso *dso, struct map *map, struct symsrc *syms_ss, } if (elf_section_by_name(runtime_ss->elf, &runtime_ss->ehdr, &tshdr, - ".text", NULL)) + ".text", NULL)) { dso->text_offset = tshdr.sh_addr - tshdr.sh_offset; + dso->text_end = tshdr.sh_offset + tshdr.sh_size; + } if (runtime_ss->opdsec) opddata = elf_rawdata(runtime_ss->opdsec, NULL); diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 5b54d2639df4..2740d4457c13 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -1732,8 +1732,10 @@ int dso__load_bfd_symbols(struct dso *dso, const char *debugfile) /* PE symbols can only have 4 bytes, so use .text high bits */ dso->text_offset = section->vma - (u32)section->vma; dso->text_offset += (u32)bfd_asymbol_value(symbols[i]); + dso->text_end = (section->vma - dso->text_offset) + section->size; } else { dso->text_offset = section->vma - section->filepos; + dso->text_end = section->filepos + section->size; } } -- cgit v1.2.3 From 6be5d82862ee9b4e517dcfdfb064c2d671dbd45a Mon Sep 17 00:00:00 2001 From: Athira Rajeev Date: Thu, 28 Sep 2023 13:22:12 +0530 Subject: tools/perf: Add "is_kmod" to struct dso to check if it is kernel module Update "struct dso" to include new member "is_kmod". This new field will determine if the file is a kernel module or not. To resolve the address from a sample, perf looks at the DSO maps. In case of address from a kernel module, there were some address found to be not resolved. This was observed while running perf test for "Object code reading". Though the ip falls beteen the start address of the loaded module (perf map->start ) and end address ( perf map->end), it was unresolved. This was happening because in some cases for kernel modules, address from sample points to stub instructions. To identify if the DSO is a kernel module, the new field "is_kmod" is added to "struct dso". Reported-by: Disha Goel Signed-off-by: Athira Rajeev Acked-by: Namhyung Kim Cc: kjain@linux.ibm.com Cc: maddy@linux.ibm.com Cc: disgoel@linux.vnet.ibm.com Cc: linuxppc-dev@lists.ozlabs.org Link: https://lore.kernel.org/r/20230928075213.84392-2-atrajeev@linux.vnet.ibm.com Signed-off-by: Namhyung Kim --- tools/perf/util/dso.c | 2 ++ tools/perf/util/dso.h | 1 + 2 files changed, 3 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/dso.c b/tools/perf/util/dso.c index bdfead36b83a..1f629b6fb7cf 100644 --- a/tools/perf/util/dso.c +++ b/tools/perf/util/dso.c @@ -477,6 +477,7 @@ void dso__set_module_info(struct dso *dso, struct kmod_path *m, dso->comp = m->comp; } + dso->is_kmod = 1; dso__set_short_name(dso, strdup(m->name), true); } @@ -1338,6 +1339,7 @@ struct dso *dso__new_id(const char *name, struct dso_id *id) dso->has_srcline = 1; dso->a2l_fails = 1; dso->kernel = DSO_SPACE__USER; + dso->is_kmod = 0; dso->needs_swap = DSO_SWAP__UNSET; dso->comp = COMP_ID__NONE; RB_CLEAR_NODE(&dso->rb_node); diff --git a/tools/perf/util/dso.h b/tools/perf/util/dso.h index 70fe0fe69bef..3759de8c2267 100644 --- a/tools/perf/util/dso.h +++ b/tools/perf/util/dso.h @@ -162,6 +162,7 @@ struct dso { char *symsrc_filename; unsigned int a2l_fails; enum dso_space_type kernel; + bool is_kmod; enum dso_swap_type needs_swap; enum dso_binary_type symtab_type; enum dso_binary_type binary_type; -- cgit v1.2.3 From 03ff4c6b3e41b3530c6c88cc91a7af599322eb89 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 7 Sep 2023 14:05:33 -0700 Subject: perf parse-events: Avoid erange from hex numbers We specify that a "num_hex" comprises 1 or more digits, however, that allows strtoull to fail with ERANGE. Limit the number of hex digits to being between 1 and 16. Before: ``` $ perf stat -e 'cpu/rE7574c47490475745/' true perf: util/parse-events.c:215: fix_raw: Assertion `errno == 0' failed. Aborted (core dumped) ``` After: ``` $ perf stat -e 'cpu/rE7574c47490475745/' true event syntax error: 'cpu/rE7574c47490475745/' \___ Bad event or PMU Unable to find PMU or event on a PMU of 'cpu' Initial error: event syntax error: 'cpu/rE7574c47490475745/' \___ unknown term 'rE7574c47490475745' for pmu 'cpu' valid terms: event,pc,edge,offcore_rsp,ldlat,inv,umask,frontend,cmask,config,config1,config2,config3,name,period,percore,metric-id Run 'perf list' for a list of valid events Usage: perf stat [] [] -e, --event event selector. use 'perf list' to list available events ``` Issue found through fuzz testing. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Link: https://lore.kernel.org/r/20230907210533.3712979-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/parse-events.l | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.l b/tools/perf/util/parse-events.l index 7bdf0565a92c..e86c45675e1d 100644 --- a/tools/perf/util/parse-events.l +++ b/tools/perf/util/parse-events.l @@ -156,8 +156,8 @@ event_pmu [^,{}/]+[/][^/]*[/][^,{}/]* event [^,{}/]+ num_dec [0-9]+ -num_hex 0x[a-fA-F0-9]+ -num_raw_hex [a-fA-F0-9]+ +num_hex 0x[a-fA-F0-9]{1,16} +num_raw_hex [a-fA-F0-9]{1,16} name [a-zA-Z_*?\[\]][a-zA-Z0-9_*?.\[\]!\-]* name_tag [\'][a-zA-Z_*?\[\]][a-zA-Z0-9_*?\-,\.\[\]:=]*[\'] name_minus [a-zA-Z_*?][a-zA-Z0-9\-_*?.:]* -- cgit v1.2.3 From 29a2fd7c72b3bd8b67bcee363f8068275a4ef351 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 5 Oct 2023 10:29:38 -0300 Subject: perf symbols: Add 'intel_idle_ibrs' to the list of idle symbols This is a longstanding to do list entry: we need a way to see that a sample took place while in idle state, as the current way to do it is to infer that by the name of the functions that in such state have more samples, IOW: a hack. Maybe we can do flip a bit in samples that take place inside the enter/exit idle section in do_idle()? But till then, add one more :-\ Signed-off-by: Arnaldo Carvalho de Melo Acked-by: Namhyung Kim Cc: Frederic Weisbecker Link: https://lore.kernel.org/r/ZR66Qgbcltt+zG7F@kernel.org Signed-off-by: Namhyung Kim --- tools/perf/util/symbol.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 2740d4457c13..96587fd7a5a2 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -718,6 +718,7 @@ static bool symbol__is_idle(const char *name) "cpu_startup_entry", "idle_cpu", "intel_idle", + "intel_idle_ibrs", "default_idle", "native_safe_halt", "enter_idle", -- cgit v1.2.3 From b20576fd7fe39554b212095c3c0d7a3dff512515 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Wed, 27 Sep 2023 17:44:31 -0700 Subject: perf parse-events: Fix for term values that are raw events Raw events can be strings like 'r0xead' but the 0x is optional so they can also be 'read'. On IcelakeX uncore_imc_free_running has an event called 'read' which may be programmed as: ``` $ perf stat -e 'uncore_imc_free_running/event=read/' -a sleep 1 ``` However, the PE_RAW type isn't allowed on the right of a term, even though in this case we just want to interpret it as a string. This leads to the following error on IcelakeX: ``` $ perf stat -e 'uncore_imc_free_running/event=read/' -a sleep 1 event syntax error: '..nning/event=read/' \___ parser error Run 'perf list' for a list of valid events Usage: perf stat [] [] -e, --event event selector. use 'perf list' to list available events ``` Fix this by allowing raw types on the right of terms and treat them as strings, just as is already done for PE_LEGACY_CACHE. Make this consistent by just entirely removing name_or_legacy and always using name_or_raw that covers all three cases. Fixes: 6fd1e5191591 ("perf parse-events: Support PMUs for legacy cache events") Signed-off-by: Ian Rogers Cc: James Clark Cc: Kan Liang Link: https://lore.kernel.org/r/20230928004431.1926969-1-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/parse-events.y | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index 4d9f3922a33d..de098caf0c1c 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -78,7 +78,7 @@ static void free_list_evsel(struct list_head* list_evsel) %type PE_MODIFIER_BP %type PE_EVENT_NAME %type PE_DRV_CFG_TERM -%type name_or_raw name_or_legacy +%type name_or_raw %destructor { free ($$); } %type event_term %destructor { parse_events_term__delete ($$); } @@ -669,8 +669,6 @@ event_term name_or_raw: PE_RAW | PE_NAME | PE_LEGACY_CACHE -name_or_legacy: PE_NAME | PE_LEGACY_CACHE - event_term: PE_RAW { @@ -685,7 +683,7 @@ PE_RAW $$ = term; } | -name_or_raw '=' name_or_legacy +name_or_raw '=' name_or_raw { struct parse_events_term *term; int err = parse_events_term__str(&term, PARSE_EVENTS__TERM_TYPE_USER, $1, $3, &@1, &@3); @@ -765,7 +763,7 @@ PE_TERM_HW $$ = term; } | -PE_TERM '=' name_or_legacy +PE_TERM '=' name_or_raw { struct parse_events_term *term; int err = parse_events_term__str(&term, $1, /*config=*/NULL, $3, &@1, &@3); -- cgit v1.2.3 From e2372136700d460276ca7ff07da523e8f61b69c7 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:08 -0700 Subject: perf env: Remove unnecessary NULL tests clang-tidy was warning: ``` util/env.c:334:23: warning: Access to field 'nr_pmu_mappings' results in a dereference of a null pointer (loaded from variable 'env') [clang-analyzer-core.NullDereference] env->nr_pmu_mappings = pmu_num; ``` As functions are called potentially when !env was true. This condition could never be true as it would produce a segv, so remove the unnecessary NULL tests and silence clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-8-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/env.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/env.c b/tools/perf/util/env.c index a164164001fb..44140b7f596a 100644 --- a/tools/perf/util/env.c +++ b/tools/perf/util/env.c @@ -457,7 +457,7 @@ const char *perf_env__cpuid(struct perf_env *env) { int status; - if (!env || !env->cpuid) { /* Assume local operation */ + if (!env->cpuid) { /* Assume local operation */ status = perf_env__read_cpuid(env); if (status) return NULL; @@ -470,7 +470,7 @@ int perf_env__nr_pmu_mappings(struct perf_env *env) { int status; - if (!env || !env->nr_pmu_mappings) { /* Assume local operation */ + if (!env->nr_pmu_mappings) { /* Assume local operation */ status = perf_env__read_pmu_mappings(env); if (status) return 0; @@ -483,7 +483,7 @@ const char *perf_env__pmu_mappings(struct perf_env *env) { int status; - if (!env || !env->pmu_mappings) { /* Assume local operation */ + if (!env->pmu_mappings) { /* Assume local operation */ status = perf_env__read_pmu_mappings(env); if (status) return NULL; -- cgit v1.2.3 From b3aa09ee78defd3d2e5f7debb5279f8a92b69749 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:09 -0700 Subject: perf jitdump: Avoid memory leak jit_repipe_unwinding_info is called in a loop by jit_process_dump, avoid leaking unwinding_data by free-ing before overwriting. Error detected by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-9-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/jitdump.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/jitdump.c b/tools/perf/util/jitdump.c index 6b2b96c16ccd..1f657ef8975f 100644 --- a/tools/perf/util/jitdump.c +++ b/tools/perf/util/jitdump.c @@ -675,6 +675,7 @@ jit_repipe_unwinding_info(struct jit_buf_desc *jd, union jr_entry *jr) jd->eh_frame_hdr_size = jr->unwinding.eh_frame_hdr_size; jd->unwinding_size = jr->unwinding.unwinding_size; jd->unwinding_mapped_size = jr->unwinding.mapped_size; + free(jd->unwinding_data); jd->unwinding_data = unwinding_data; return 0; -- cgit v1.2.3 From 85f73c377b2ac9988a204b119aebb33ca5c60083 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:10 -0700 Subject: perf mem-events: Avoid uninitialized read pmu should be initialized to NULL before perf_pmus__scan loop. Fix and shrink the scope of pmu at the same time. Issue detected by clang-tidy. Fixes: 5752c20f3787 ("perf mem: Scan all PMUs instead of just core ones") Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-10-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/mem-events.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/mem-events.c b/tools/perf/util/mem-events.c index 39ffe8ceb380..954b235e12e5 100644 --- a/tools/perf/util/mem-events.c +++ b/tools/perf/util/mem-events.c @@ -185,7 +185,6 @@ int perf_mem_events__record_args(const char **rec_argv, int *argv_nr, { int i = *argv_nr, k = 0; struct perf_mem_event *e; - struct perf_pmu *pmu; for (int j = 0; j < PERF_MEM_EVENTS__MAX; j++) { e = perf_mem_events__ptr(j); @@ -202,6 +201,8 @@ int perf_mem_events__record_args(const char **rec_argv, int *argv_nr, rec_argv[i++] = "-e"; rec_argv[i++] = perf_mem_events__name(j, NULL); } else { + struct perf_pmu *pmu = NULL; + if (!e->supported) { perf_mem_events__print_unsupport_hybrid(e, j); return -1; -- cgit v1.2.3 From 52a5ad12f2147506899ee83e680ea2a1d763adeb Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:11 -0700 Subject: perf dlfilter: Be defensive against potential NULL dereference In the unlikely case of having a symbol without a mapping, avoid a NULL dereference that clang-tidy warns about. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-11-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/dlfilter.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/dlfilter.c b/tools/perf/util/dlfilter.c index 1dbf27822ee2..5e54832137a9 100644 --- a/tools/perf/util/dlfilter.c +++ b/tools/perf/util/dlfilter.c @@ -52,8 +52,10 @@ static void al_to_d_al(struct addr_location *al, struct perf_dlfilter_al *d_al) d_al->sym_end = sym->end; if (al->addr < sym->end) d_al->symoff = al->addr - sym->start; - else + else if (al->map) d_al->symoff = al->addr - map__start(al->map) - sym->start; + else + d_al->symoff = 0; d_al->sym_binding = sym->binding; } else { d_al->sym = NULL; -- cgit v1.2.3 From 63d471979e49148e59eae0b33a57c12d535e20c6 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:14 -0700 Subject: perf svghelper: Avoid memory leak On success path the sib_core and sib_thr values weren't being freed. Detected by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-14-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/svghelper.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/svghelper.c b/tools/perf/util/svghelper.c index 0e4dc31c6c9c..1892e9b6aa7f 100644 --- a/tools/perf/util/svghelper.c +++ b/tools/perf/util/svghelper.c @@ -754,6 +754,7 @@ int svg_build_topology_map(struct perf_env *env) int i, nr_cpus; struct topology t; char *sib_core, *sib_thr; + int ret = -1; nr_cpus = min(env->nr_cpus_online, MAX_NR_CPUS); @@ -799,11 +800,11 @@ int svg_build_topology_map(struct perf_env *env) scan_core_topology(topology_map, &t, nr_cpus); - return 0; + ret = 0; exit: zfree(&t.sib_core); zfree(&t.sib_thr); - return -1; + return ret; } -- cgit v1.2.3 From 7875c72c8b0566590c888a2420d7e8fc12f67154 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:16 -0700 Subject: perf parse-events: Fix unlikely memory leak when cloning terms Add missing free on an error path as detected by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-16-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/parse-events.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index c56e07bd7dd6..23c027cf20ae 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -2549,8 +2549,10 @@ int parse_events_term__clone(struct parse_events_term **new, return new_term(new, &temp, /*str=*/NULL, term->val.num); str = strdup(term->val.str); - if (!str) + if (!str) { + zfree(&temp.config); return -ENOMEM; + } return new_term(new, &temp, str, /*num=*/0); } -- cgit v1.2.3 From 97fe038374bdf43fd025ac0e7aebf8bfbdd6d54f Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:18 -0700 Subject: perf trace-event-info: Avoid passing NULL value to closedir If opendir failed then closedir was passed NULL which is erroneous. Caught by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-18-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/trace-event-info.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/trace-event-info.c b/tools/perf/util/trace-event-info.c index 319ccf09a435..c8755679281e 100644 --- a/tools/perf/util/trace-event-info.c +++ b/tools/perf/util/trace-event-info.c @@ -313,7 +313,8 @@ static int record_event_files(struct tracepoint_path *tps) } err = 0; out: - closedir(dir); + if (dir) + closedir(dir); put_tracing_file(path); return err; -- cgit v1.2.3 From 105254501770c8952e50c71618fca6a8b63890f1 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:19 -0700 Subject: perf header: Fix various error path memory leaks Memory leaks were detected by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-19-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/header.c | 60 +++++++++++++++++++++++++++--------------------- 1 file changed, 34 insertions(+), 26 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index d812e1e371a7..e86b9439ffee 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2573,7 +2573,7 @@ error: static int process_cpu_topology(struct feat_fd *ff, void *data __maybe_unused) { u32 nr, i; - char *str; + char *str = NULL; struct strbuf sb; int cpu_nr = ff->ph->env.nr_cpus_avail; u64 size = 0; @@ -2601,7 +2601,7 @@ static int process_cpu_topology(struct feat_fd *ff, void *data __maybe_unused) if (strbuf_add(&sb, str, strlen(str) + 1) < 0) goto error; size += string_size(str); - free(str); + zfree(&str); } ph->env.sibling_cores = strbuf_detach(&sb, NULL); @@ -2620,7 +2620,7 @@ static int process_cpu_topology(struct feat_fd *ff, void *data __maybe_unused) if (strbuf_add(&sb, str, strlen(str) + 1) < 0) goto error; size += string_size(str); - free(str); + zfree(&str); } ph->env.sibling_threads = strbuf_detach(&sb, NULL); @@ -2684,7 +2684,7 @@ static int process_cpu_topology(struct feat_fd *ff, void *data __maybe_unused) if (strbuf_add(&sb, str, strlen(str) + 1) < 0) goto error; size += string_size(str); - free(str); + zfree(&str); } ph->env.sibling_dies = strbuf_detach(&sb, NULL); @@ -2699,6 +2699,7 @@ static int process_cpu_topology(struct feat_fd *ff, void *data __maybe_unused) error: strbuf_release(&sb); + zfree(&str); free_cpu: zfree(&ph->env.cpu); return -1; @@ -2736,10 +2737,9 @@ static int process_numa_topology(struct feat_fd *ff, void *data __maybe_unused) goto error; n->map = perf_cpu_map__new(str); + free(str); if (!n->map) goto error; - - free(str); } ff->ph->env.nr_numa_nodes = nr; ff->ph->env.numa_nodes = nodes; @@ -2913,10 +2913,10 @@ static int process_cache(struct feat_fd *ff, void *data __maybe_unused) return -1; for (i = 0; i < cnt; i++) { - struct cpu_cache_level c; + struct cpu_cache_level *c = &caches[i]; #define _R(v) \ - if (do_read_u32(ff, &c.v))\ + if (do_read_u32(ff, &c->v)) \ goto out_free_caches; \ _R(level) @@ -2926,22 +2926,25 @@ static int process_cache(struct feat_fd *ff, void *data __maybe_unused) #undef _R #define _R(v) \ - c.v = do_read_string(ff); \ - if (!c.v) \ - goto out_free_caches; + c->v = do_read_string(ff); \ + if (!c->v) \ + goto out_free_caches; \ _R(type) _R(size) _R(map) #undef _R - - caches[i] = c; } ff->ph->env.caches = caches; ff->ph->env.caches_cnt = cnt; return 0; out_free_caches: + for (i = 0; i < cnt; i++) { + free(caches[i].type); + free(caches[i].size); + free(caches[i].map); + } free(caches); return -1; } @@ -3585,18 +3588,16 @@ static int perf_header__adds_write(struct perf_header *header, struct feat_copier *fc) { int nr_sections; - struct feat_fd ff; + struct feat_fd ff = { + .fd = fd, + .ph = header, + }; struct perf_file_section *feat_sec, *p; int sec_size; u64 sec_start; int feat; int err; - ff = (struct feat_fd){ - .fd = fd, - .ph = header, - }; - nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; @@ -3623,6 +3624,7 @@ static int perf_header__adds_write(struct perf_header *header, err = do_write(&ff, feat_sec, sec_size); if (err < 0) pr_debug("failed to write feature section\n"); + free(ff.buf); /* TODO: added to silence clang-tidy. */ free(feat_sec); return err; } @@ -3630,11 +3632,11 @@ static int perf_header__adds_write(struct perf_header *header, int perf_header__write_pipe(int fd) { struct perf_pipe_file_header f_header; - struct feat_fd ff; + struct feat_fd ff = { + .fd = fd, + }; int err; - ff = (struct feat_fd){ .fd = fd }; - f_header = (struct perf_pipe_file_header){ .magic = PERF_MAGIC, .size = sizeof(f_header), @@ -3645,7 +3647,7 @@ int perf_header__write_pipe(int fd) pr_debug("failed to write perf pipe header\n"); return err; } - + free(ff.buf); return 0; } @@ -3658,11 +3660,12 @@ static int perf_session__do_write_header(struct perf_session *session, struct perf_file_attr f_attr; struct perf_header *header = &session->header; struct evsel *evsel; - struct feat_fd ff; + struct feat_fd ff = { + .fd = fd, + }; u64 attr_offset; int err; - ff = (struct feat_fd){ .fd = fd}; lseek(fd, sizeof(f_header), SEEK_SET); evlist__for_each_entry(session->evlist, evsel) { @@ -3670,6 +3673,7 @@ static int perf_session__do_write_header(struct perf_session *session, err = do_write(&ff, evsel->core.id, evsel->core.ids * sizeof(u64)); if (err < 0) { pr_debug("failed to write perf header\n"); + free(ff.buf); return err; } } @@ -3695,6 +3699,7 @@ static int perf_session__do_write_header(struct perf_session *session, err = do_write(&ff, &f_attr, sizeof(f_attr)); if (err < 0) { pr_debug("failed to write perf header attribute\n"); + free(ff.buf); return err; } } @@ -3705,8 +3710,10 @@ static int perf_session__do_write_header(struct perf_session *session, if (at_exit) { err = perf_header__adds_write(header, evlist, fd, fc); - if (err < 0) + if (err < 0) { + free(ff.buf); return err; + } } f_header = (struct perf_file_header){ @@ -3728,6 +3735,7 @@ static int perf_session__do_write_header(struct perf_session *session, lseek(fd, 0, SEEK_SET); err = do_write(&ff, &f_header, sizeof(f_header)); + free(ff.buf); if (err < 0) { pr_debug("failed to write perf header\n"); return err; -- cgit v1.2.3 From b84b3f47921568a8172bec77d0370268e9fc62a2 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Mon, 9 Oct 2023 11:39:20 -0700 Subject: perf bpf_counter: Fix a few memory leaks Memory leaks were detected by clang-tidy. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Cc: Ravi Bangoria Cc: Nick Desaulniers Cc: Yang Jihong Cc: Huacai Chen Cc: Nathan Chancellor Cc: Kan Liang Cc: llvm@lists.linux.dev Cc: Ming Wang Cc: Tom Rix Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231009183920.200859-20-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_counter.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_counter.c b/tools/perf/util/bpf_counter.c index 6732cbbcf9b3..7f9b0e46e008 100644 --- a/tools/perf/util/bpf_counter.c +++ b/tools/perf/util/bpf_counter.c @@ -104,7 +104,7 @@ static int bpf_program_profiler_load_one(struct evsel *evsel, u32 prog_id) struct bpf_prog_profiler_bpf *skel; struct bpf_counter *counter; struct bpf_program *prog; - char *prog_name; + char *prog_name = NULL; int prog_fd; int err; @@ -155,10 +155,12 @@ static int bpf_program_profiler_load_one(struct evsel *evsel, u32 prog_id) assert(skel != NULL); counter->skel = skel; list_add(&counter->list, &evsel->bpf_counter_list); + free(prog_name); close(prog_fd); return 0; err_out: bpf_prog_profiler_bpf__destroy(skel); + free(prog_name); free(counter); close(prog_fd); return -1; @@ -180,6 +182,7 @@ static int bpf_program_profiler__load(struct evsel *evsel, struct target *target (*p != '\0' && *p != ',')) { pr_err("Failed to parse bpf prog ids %s\n", target->bpf_str); + free(bpf_str_); return -1; } -- cgit v1.2.3 From a16afcc58a8c5ebc65c852faf001f8f61f05e4ef Mon Sep 17 00:00:00 2001 From: Besar Wicaksono Date: Tue, 10 Oct 2023 18:48:03 -0500 Subject: perf cs-etm: Fix incorrect or missing decoder for raw trace The decoder creation for raw trace uses metadata from the first CPU. On per-cpu mode, this metadata is incorrectly used for every decoder. On per-process/per-thread traces, the first CPU is CPU0. If CPU0 trace is not enabled, its metadata will be marked unused and the decoder is not created. Perf report dump skips the decoding part because the decoder is missing. To fix this, use metadata of the CPU associated with sample object. Signed-off-by: Besar Wicaksono Reviewed-by: James Clark Cc: suzuki.poulose@arm.com Cc: mike.leach@linaro.org Cc: jonathanh@nvidia.com Cc: rwiley@nvidia.com Cc: treding@nvidia.com Cc: vsethi@nvidia.com Cc: ywan@nvidia.com Cc: linux-arm-kernel@lists.infradead.org Cc: coresight@lists.linaro.org Cc: linux-tegra@vger.kernel.org Link: https://lore.kernel.org/r/20231010234803.5419-1-bwicaksono@nvidia.com Signed-off-by: Namhyung Kim --- tools/perf/util/cs-etm.c | 106 +++++++++++++++++++++++++++++------------------ 1 file changed, 65 insertions(+), 41 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cs-etm.c b/tools/perf/util/cs-etm.c index 9729d006550d..a9873d14c632 100644 --- a/tools/perf/util/cs-etm.c +++ b/tools/perf/util/cs-etm.c @@ -283,22 +283,31 @@ static int cs_etm__metadata_set_trace_id(u8 trace_chan_id, u64 *cpu_metadata) } /* - * Get a metadata for a specific cpu from an array. + * Get a metadata index for a specific cpu from an array. * */ -static u64 *get_cpu_data(struct cs_etm_auxtrace *etm, int cpu) +static int get_cpu_data_idx(struct cs_etm_auxtrace *etm, int cpu) { int i; - u64 *metadata = NULL; for (i = 0; i < etm->num_cpu; i++) { if (etm->metadata[i][CS_ETM_CPU] == (u64)cpu) { - metadata = etm->metadata[i]; - break; + return i; } } - return metadata; + return -1; +} + +/* + * Get a metadata for a specific cpu from an array. + * + */ +static u64 *get_cpu_data(struct cs_etm_auxtrace *etm, int cpu) +{ + int idx = get_cpu_data_idx(etm, cpu); + + return (idx != -1) ? etm->metadata[idx] : NULL; } /* @@ -641,66 +650,80 @@ static void cs_etm__packet_dump(const char *pkt_string) } static void cs_etm__set_trace_param_etmv3(struct cs_etm_trace_params *t_params, - struct cs_etm_auxtrace *etm, int idx, - u32 etmidr) + struct cs_etm_auxtrace *etm, int t_idx, + int m_idx, u32 etmidr) { u64 **metadata = etm->metadata; - t_params[idx].protocol = cs_etm__get_v7_protocol_version(etmidr); - t_params[idx].etmv3.reg_ctrl = metadata[idx][CS_ETM_ETMCR]; - t_params[idx].etmv3.reg_trc_id = metadata[idx][CS_ETM_ETMTRACEIDR]; + t_params[t_idx].protocol = cs_etm__get_v7_protocol_version(etmidr); + t_params[t_idx].etmv3.reg_ctrl = metadata[m_idx][CS_ETM_ETMCR]; + t_params[t_idx].etmv3.reg_trc_id = metadata[m_idx][CS_ETM_ETMTRACEIDR]; } static void cs_etm__set_trace_param_etmv4(struct cs_etm_trace_params *t_params, - struct cs_etm_auxtrace *etm, int idx) + struct cs_etm_auxtrace *etm, int t_idx, + int m_idx) { u64 **metadata = etm->metadata; - t_params[idx].protocol = CS_ETM_PROTO_ETMV4i; - t_params[idx].etmv4.reg_idr0 = metadata[idx][CS_ETMV4_TRCIDR0]; - t_params[idx].etmv4.reg_idr1 = metadata[idx][CS_ETMV4_TRCIDR1]; - t_params[idx].etmv4.reg_idr2 = metadata[idx][CS_ETMV4_TRCIDR2]; - t_params[idx].etmv4.reg_idr8 = metadata[idx][CS_ETMV4_TRCIDR8]; - t_params[idx].etmv4.reg_configr = metadata[idx][CS_ETMV4_TRCCONFIGR]; - t_params[idx].etmv4.reg_traceidr = metadata[idx][CS_ETMV4_TRCTRACEIDR]; + t_params[t_idx].protocol = CS_ETM_PROTO_ETMV4i; + t_params[t_idx].etmv4.reg_idr0 = metadata[m_idx][CS_ETMV4_TRCIDR0]; + t_params[t_idx].etmv4.reg_idr1 = metadata[m_idx][CS_ETMV4_TRCIDR1]; + t_params[t_idx].etmv4.reg_idr2 = metadata[m_idx][CS_ETMV4_TRCIDR2]; + t_params[t_idx].etmv4.reg_idr8 = metadata[m_idx][CS_ETMV4_TRCIDR8]; + t_params[t_idx].etmv4.reg_configr = metadata[m_idx][CS_ETMV4_TRCCONFIGR]; + t_params[t_idx].etmv4.reg_traceidr = metadata[m_idx][CS_ETMV4_TRCTRACEIDR]; } static void cs_etm__set_trace_param_ete(struct cs_etm_trace_params *t_params, - struct cs_etm_auxtrace *etm, int idx) + struct cs_etm_auxtrace *etm, int t_idx, + int m_idx) { u64 **metadata = etm->metadata; - t_params[idx].protocol = CS_ETM_PROTO_ETE; - t_params[idx].ete.reg_idr0 = metadata[idx][CS_ETE_TRCIDR0]; - t_params[idx].ete.reg_idr1 = metadata[idx][CS_ETE_TRCIDR1]; - t_params[idx].ete.reg_idr2 = metadata[idx][CS_ETE_TRCIDR2]; - t_params[idx].ete.reg_idr8 = metadata[idx][CS_ETE_TRCIDR8]; - t_params[idx].ete.reg_configr = metadata[idx][CS_ETE_TRCCONFIGR]; - t_params[idx].ete.reg_traceidr = metadata[idx][CS_ETE_TRCTRACEIDR]; - t_params[idx].ete.reg_devarch = metadata[idx][CS_ETE_TRCDEVARCH]; + t_params[t_idx].protocol = CS_ETM_PROTO_ETE; + t_params[t_idx].ete.reg_idr0 = metadata[m_idx][CS_ETE_TRCIDR0]; + t_params[t_idx].ete.reg_idr1 = metadata[m_idx][CS_ETE_TRCIDR1]; + t_params[t_idx].ete.reg_idr2 = metadata[m_idx][CS_ETE_TRCIDR2]; + t_params[t_idx].ete.reg_idr8 = metadata[m_idx][CS_ETE_TRCIDR8]; + t_params[t_idx].ete.reg_configr = metadata[m_idx][CS_ETE_TRCCONFIGR]; + t_params[t_idx].ete.reg_traceidr = metadata[m_idx][CS_ETE_TRCTRACEIDR]; + t_params[t_idx].ete.reg_devarch = metadata[m_idx][CS_ETE_TRCDEVARCH]; } static int cs_etm__init_trace_params(struct cs_etm_trace_params *t_params, struct cs_etm_auxtrace *etm, + bool formatted, + int sample_cpu, int decoders) { - int i; + int t_idx, m_idx; u32 etmidr; u64 architecture; - for (i = 0; i < decoders; i++) { - architecture = etm->metadata[i][CS_ETM_MAGIC]; + for (t_idx = 0; t_idx < decoders; t_idx++) { + if (formatted) + m_idx = t_idx; + else { + m_idx = get_cpu_data_idx(etm, sample_cpu); + if (m_idx == -1) { + pr_warning("CS_ETM: unknown CPU, falling back to first metadata\n"); + m_idx = 0; + } + } + + architecture = etm->metadata[m_idx][CS_ETM_MAGIC]; switch (architecture) { case __perf_cs_etmv3_magic: - etmidr = etm->metadata[i][CS_ETM_ETMIDR]; - cs_etm__set_trace_param_etmv3(t_params, etm, i, etmidr); + etmidr = etm->metadata[m_idx][CS_ETM_ETMIDR]; + cs_etm__set_trace_param_etmv3(t_params, etm, t_idx, m_idx, etmidr); break; case __perf_cs_etmv4_magic: - cs_etm__set_trace_param_etmv4(t_params, etm, i); + cs_etm__set_trace_param_etmv4(t_params, etm, t_idx, m_idx); break; case __perf_cs_ete_magic: - cs_etm__set_trace_param_ete(t_params, etm, i); + cs_etm__set_trace_param_ete(t_params, etm, t_idx, m_idx); break; default: return -EINVAL; @@ -1016,7 +1039,7 @@ out: } static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm, - bool formatted) + bool formatted, int sample_cpu) { struct cs_etm_decoder_params d_params; struct cs_etm_trace_params *t_params = NULL; @@ -1041,7 +1064,7 @@ static struct cs_etm_queue *cs_etm__alloc_queue(struct cs_etm_auxtrace *etm, if (!t_params) goto out_free; - if (cs_etm__init_trace_params(t_params, etm, decoders)) + if (cs_etm__init_trace_params(t_params, etm, formatted, sample_cpu, decoders)) goto out_free; /* Set decoder parameters to decode trace packets */ @@ -1081,14 +1104,15 @@ out_free: static int cs_etm__setup_queue(struct cs_etm_auxtrace *etm, struct auxtrace_queue *queue, unsigned int queue_nr, - bool formatted) + bool formatted, + int sample_cpu) { struct cs_etm_queue *etmq = queue->priv; if (list_empty(&queue->head) || etmq) return 0; - etmq = cs_etm__alloc_queue(etm, formatted); + etmq = cs_etm__alloc_queue(etm, formatted, sample_cpu); if (!etmq) return -ENOMEM; @@ -2816,7 +2840,7 @@ static int cs_etm__process_auxtrace_event(struct perf_session *session, * formatted in piped mode (true). */ err = cs_etm__setup_queue(etm, &etm->queues.queue_array[idx], - idx, true); + idx, true, -1); if (err) return err; @@ -3022,7 +3046,7 @@ static int cs_etm__queue_aux_fragment(struct perf_session *session, off_t file_o idx = auxtrace_event->idx; formatted = !(aux_event->flags & PERF_AUX_FLAG_CORESIGHT_FORMAT_RAW); return cs_etm__setup_queue(etm, &etm->queues.queue_array[idx], - idx, formatted); + idx, formatted, sample->cpu); } /* Wasn't inside this buffer, but there were no parse errors. 1 == 'not found' */ -- cgit v1.2.3 From 1d2dbce9bb9243ca18cacff5ada9dc00fc1e74b7 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 5 Oct 2023 22:04:48 +0300 Subject: perf intel-pt: Simplify intel_pt_get_vmcs() Simplify and remove unnecessary constant expressions. Signed-off-by: Adrian Hunter Acked-by: Ian Rogers Link: https://lore.kernel.org/r/20231005190451.175568-3-adrian.hunter@intel.com Signed-off-by: Namhyung Kim --- tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c | 13 ++++--------- 1 file changed, 4 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c index af9710622a1f..249610cc9284 100644 --- a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c +++ b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c @@ -131,19 +131,14 @@ static int intel_pt_get_cbr(const unsigned char *buf, size_t len, static int intel_pt_get_vmcs(const unsigned char *buf, size_t len, struct intel_pt_pkt *packet) { - unsigned int count = (52 - 5) >> 3; - - if (count < 1 || count > 7) - return INTEL_PT_BAD_PACKET; - - if (len < count + 2) + if (len < 7) return INTEL_PT_NEED_MORE_BYTES; packet->type = INTEL_PT_VMCS; - packet->count = count; - memcpy_le64(&packet->payload, buf + 2, count); + packet->count = 5; + memcpy_le64(&packet->payload, buf + 2, 5); - return count + 2; + return 7; } static int intel_pt_get_ovf(struct intel_pt_pkt *packet) -- cgit v1.2.3 From f058fa5b07556abed26e5ae9d8e8ad13f79e1fa2 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 5 Oct 2023 22:04:49 +0300 Subject: perf intel-pt: Use existing definitions of le16_to_cpu() etc Use definitions from tools/include/linux/kernel.h Signed-off-by: Adrian Hunter Acked-by: Ian Rogers Link: https://lore.kernel.org/r/20231005190451.175568-4-adrian.hunter@intel.com Signed-off-by: Namhyung Kim --- tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c index 249610cc9284..ffd0d647473c 100644 --- a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c +++ b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c @@ -8,6 +8,7 @@ #include #include #include +#include #include #include "intel-pt-pkt-decoder.h" @@ -17,17 +18,11 @@ #define BIT63 ((uint64_t)1 << 63) #if __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__ -#define le16_to_cpu bswap_16 -#define le32_to_cpu bswap_32 -#define le64_to_cpu bswap_64 #define memcpy_le64(d, s, n) do { \ memcpy((d), (s), (n)); \ *(d) = le64_to_cpu(*(d)); \ } while (0) #else -#define le16_to_cpu -#define le32_to_cpu -#define le64_to_cpu #define memcpy_le64 memcpy #endif -- cgit v1.2.3 From 3b4fa67fc666eb88ae4c0ffdfbe338bec989fbee Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 5 Oct 2023 22:04:50 +0300 Subject: perf intel-pt: Use get_unaligned_le16() etc Avoid unaligned access by using get_unaligned_le16(), get_unaligned_le32() and get_unaligned_le64(). Signed-off-by: Adrian Hunter Acked-by: Ian Rogers Link: https://lore.kernel.org/r/20231005190451.175568-5-adrian.hunter@intel.com Signed-off-by: Namhyung Kim --- tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c index ffd0d647473c..7a90218aecb1 100644 --- a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c +++ b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c @@ -10,6 +10,7 @@ #include #include #include +#include #include "intel-pt-pkt-decoder.h" @@ -78,7 +79,7 @@ static int intel_pt_get_long_tnt(const unsigned char *buf, size_t len, if (len < 8) return INTEL_PT_NEED_MORE_BYTES; - payload = le64_to_cpu(*(uint64_t *)buf); + payload = get_unaligned_le64(buf); for (count = 47; count; count--) { if (payload & BIT63) @@ -119,7 +120,7 @@ static int intel_pt_get_cbr(const unsigned char *buf, size_t len, if (len < 4) return INTEL_PT_NEED_MORE_BYTES; packet->type = INTEL_PT_CBR; - packet->payload = le16_to_cpu(*(uint16_t *)(buf + 2)); + packet->payload = get_unaligned_le16(buf + 2); return 4; } @@ -218,12 +219,12 @@ static int intel_pt_get_ptwrite(const unsigned char *buf, size_t len, case 0: if (len < 6) return INTEL_PT_NEED_MORE_BYTES; - packet->payload = le32_to_cpu(*(uint32_t *)(buf + 2)); + packet->payload = get_unaligned_le32(buf + 2); return 6; case 1: if (len < 10) return INTEL_PT_NEED_MORE_BYTES; - packet->payload = le64_to_cpu(*(uint64_t *)(buf + 2)); + packet->payload = get_unaligned_le64(buf + 2); return 10; default: return INTEL_PT_BAD_PACKET; @@ -248,7 +249,7 @@ static int intel_pt_get_mwait(const unsigned char *buf, size_t len, if (len < 10) return INTEL_PT_NEED_MORE_BYTES; packet->type = INTEL_PT_MWAIT; - packet->payload = le64_to_cpu(*(uint64_t *)(buf + 2)); + packet->payload = get_unaligned_le64(buf + 2); return 10; } @@ -455,13 +456,13 @@ static int intel_pt_get_ip(enum intel_pt_pkt_type type, unsigned int byte, if (len < 3) return INTEL_PT_NEED_MORE_BYTES; ip_len = 2; - packet->payload = le16_to_cpu(*(uint16_t *)(buf + 1)); + packet->payload = get_unaligned_le16(buf + 1); break; case 2: if (len < 5) return INTEL_PT_NEED_MORE_BYTES; ip_len = 4; - packet->payload = le32_to_cpu(*(uint32_t *)(buf + 1)); + packet->payload = get_unaligned_le32(buf + 1); break; case 3: case 4: @@ -474,7 +475,7 @@ static int intel_pt_get_ip(enum intel_pt_pkt_type type, unsigned int byte, if (len < 9) return INTEL_PT_NEED_MORE_BYTES; ip_len = 8; - packet->payload = le64_to_cpu(*(uint64_t *)(buf + 1)); + packet->payload = get_unaligned_le64(buf + 1); break; default: return INTEL_PT_BAD_PACKET; -- cgit v1.2.3 From 661ce78105d739f3297de8a2384f4e48a2a16043 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 5 Oct 2023 22:04:51 +0300 Subject: perf intel-pt: Prefer get_unaligned_le64 to memcpy_le64 Use get_unaligned_le64() instead of memcpy_le64(..., 8) because it produces simpler code. Signed-off-by: Adrian Hunter Acked-by: Ian Rogers Link: https://lore.kernel.org/r/20231005190451.175568-6-adrian.hunter@intel.com Signed-off-by: Namhyung Kim --- tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c index 7a90218aecb1..bccb988a7a44 100644 --- a/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c +++ b/tools/perf/util/intel-pt-decoder/intel-pt-pkt-decoder.c @@ -190,7 +190,7 @@ static int intel_pt_get_mnt(const unsigned char *buf, size_t len, if (len < 11) return INTEL_PT_NEED_MORE_BYTES; packet->type = INTEL_PT_MNT; - memcpy_le64(&packet->payload, buf + 3, 8); + packet->payload = get_unaligned_le64(buf + 3); return 11; } @@ -302,7 +302,7 @@ static int intel_pt_get_bip_8(const unsigned char *buf, size_t len, return INTEL_PT_NEED_MORE_BYTES; packet->type = INTEL_PT_BIP; packet->count = buf[0] >> 3; - memcpy_le64(&packet->payload, buf + 1, 8); + packet->payload = get_unaligned_le64(buf + 1); return 9; } @@ -341,7 +341,7 @@ static int intel_pt_get_evd(const unsigned char *buf, size_t len, packet->type = INTEL_PT_EVD; packet->count = buf[2] & 0x3f; packet->payload = buf[3]; - memcpy_le64(&packet->payload, buf + 3, 8); + packet->payload = get_unaligned_le64(buf + 3); return 11; } -- cgit v1.2.3 From aa61360155ac69003dc5c177f03a7f177a435286 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 12 Oct 2023 10:56:39 -0700 Subject: perf pmu: Rename perf_pmu__get_default_config to perf_pmu__arch_init Assign default_config as part of the init. perf_pmu__get_default_config was doing more than just getting the default config and so this is intended to better align with the code. Signed-off-by: Ian Rogers Cc: Ravi Bangoria Cc: James Clark Cc: Suzuki K Poulose Cc: Yang Jihong Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Kajol Jain Cc: Thomas Richter Cc: Alexander Shishkin Cc: Kan Liang Cc: John Garry Cc: linux-arm-kernel@lists.infradead.org Cc: coresight@lists.linaro.org Link: https://lore.kernel.org/r/20231012175645.1849503-2-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 13 ++++++------- tools/perf/util/pmu.h | 2 +- 2 files changed, 7 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 6428e2648289..d075da0eecc0 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -954,12 +954,6 @@ void pmu_add_sys_aliases(struct perf_pmu *pmu) pmu_for_each_sys_event(pmu_add_sys_aliases_iter_fn, pmu); } -struct perf_event_attr * __weak -perf_pmu__get_default_config(struct perf_pmu *pmu __maybe_unused) -{ - return NULL; -} - static char *pmu_find_alias_name(struct perf_pmu *pmu, int dirfd) { FILE *file = perf_pmu__open_file_at(pmu, dirfd, "alias"); @@ -991,6 +985,11 @@ static int pmu_max_precise(int dirfd, struct perf_pmu *pmu) return max_precise; } +void __weak +perf_pmu__arch_init(struct perf_pmu *pmu __maybe_unused) +{ +} + struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char *name) { struct perf_pmu *pmu; @@ -1037,7 +1036,7 @@ struct perf_pmu *perf_pmu__lookup(struct list_head *pmus, int dirfd, const char pmu_add_sys_aliases(pmu); list_add_tail(&pmu->list, pmus); - pmu->default_config = perf_pmu__get_default_config(pmu); + perf_pmu__arch_init(pmu); return pmu; err: diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 85190d058852..588c64e38d6b 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -233,7 +233,7 @@ bool perf_pmu__file_exists(struct perf_pmu *pmu, const char *name); int perf_pmu__test(void); -struct perf_event_attr *perf_pmu__get_default_config(struct perf_pmu *pmu); +void perf_pmu__arch_init(struct perf_pmu *pmu); void pmu_add_cpu_aliases_table(struct perf_pmu *pmu, const struct pmu_events_table *table); -- cgit v1.2.3 From 3a42f4c796ce45a6b1d14975401deb2c457ae79d Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 12 Oct 2023 10:56:42 -0700 Subject: perf pmu: Const-ify file APIs File APIs don't alter the struct pmu so allow const ones to be passed. Signed-off-by: Ian Rogers Reviewed-by: Adrian Hunter Cc: Ravi Bangoria Cc: James Clark Cc: Suzuki K Poulose Cc: Yang Jihong Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Kajol Jain Cc: Thomas Richter Cc: Kan Liang Cc: John Garry Cc: linux-arm-kernel@lists.infradead.org Cc: coresight@lists.linaro.org Link: https://lore.kernel.org/r/20231012175645.1849503-5-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 12 ++++++------ tools/perf/util/pmu.h | 11 ++++++----- tools/perf/util/python.c | 2 +- 3 files changed, 13 insertions(+), 12 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index d075da0eecc0..861e485e0a70 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -575,7 +575,7 @@ static int perf_pmu__new_alias(struct perf_pmu *pmu, const char *name, return 0; } -static inline bool pmu_alias_info_file(char *name) +static inline bool pmu_alias_info_file(const char *name) { size_t len; @@ -1770,7 +1770,7 @@ bool perf_pmu__is_software(const struct perf_pmu *pmu) return !strcmp(pmu->name, "kprobe") || !strcmp(pmu->name, "uprobe"); } -FILE *perf_pmu__open_file(struct perf_pmu *pmu, const char *name) +FILE *perf_pmu__open_file(const struct perf_pmu *pmu, const char *name) { char path[PATH_MAX]; @@ -1781,7 +1781,7 @@ FILE *perf_pmu__open_file(struct perf_pmu *pmu, const char *name) return fopen(path, "r"); } -FILE *perf_pmu__open_file_at(struct perf_pmu *pmu, int dirfd, const char *name) +FILE *perf_pmu__open_file_at(const struct perf_pmu *pmu, int dirfd, const char *name) { int fd; @@ -1792,7 +1792,7 @@ FILE *perf_pmu__open_file_at(struct perf_pmu *pmu, int dirfd, const char *name) return fdopen(fd, "r"); } -int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt, +int perf_pmu__scan_file(const struct perf_pmu *pmu, const char *name, const char *fmt, ...) { va_list args; @@ -1809,7 +1809,7 @@ int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt, return ret; } -int perf_pmu__scan_file_at(struct perf_pmu *pmu, int dirfd, const char *name, +int perf_pmu__scan_file_at(const struct perf_pmu *pmu, int dirfd, const char *name, const char *fmt, ...) { va_list args; @@ -1826,7 +1826,7 @@ int perf_pmu__scan_file_at(struct perf_pmu *pmu, int dirfd, const char *name, return ret; } -bool perf_pmu__file_exists(struct perf_pmu *pmu, const char *name) +bool perf_pmu__file_exists(const struct perf_pmu *pmu, const char *name) { char path[PATH_MAX]; diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 588c64e38d6b..24af7297b522 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -222,14 +222,15 @@ bool pmu__name_match(const struct perf_pmu *pmu, const char *pmu_name); */ bool perf_pmu__is_software(const struct perf_pmu *pmu); -FILE *perf_pmu__open_file(struct perf_pmu *pmu, const char *name); -FILE *perf_pmu__open_file_at(struct perf_pmu *pmu, int dirfd, const char *name); +FILE *perf_pmu__open_file(const struct perf_pmu *pmu, const char *name); +FILE *perf_pmu__open_file_at(const struct perf_pmu *pmu, int dirfd, const char *name); -int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt, ...) __scanf(3, 4); -int perf_pmu__scan_file_at(struct perf_pmu *pmu, int dirfd, const char *name, +int perf_pmu__scan_file(const struct perf_pmu *pmu, const char *name, const char *fmt, ...) + __scanf(3, 4); +int perf_pmu__scan_file_at(const struct perf_pmu *pmu, int dirfd, const char *name, const char *fmt, ...) __scanf(4, 5); -bool perf_pmu__file_exists(struct perf_pmu *pmu, const char *name); +bool perf_pmu__file_exists(const struct perf_pmu *pmu, const char *name); int perf_pmu__test(void); diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c index c29f5f0bb552..8761f51b5c7c 100644 --- a/tools/perf/util/python.c +++ b/tools/perf/util/python.c @@ -98,7 +98,7 @@ struct perf_pmu *evsel__find_pmu(const struct evsel *evsel __maybe_unused) return NULL; } -int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt, ...) +int perf_pmu__scan_file(const struct perf_pmu *pmu, const char *name, const char *fmt, ...) { return EOF; } -- cgit v1.2.3 From 63883cb063846e9d0574038bb702a3a62a1ec046 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 12 Oct 2023 10:56:43 -0700 Subject: perf pmu: Const-ify perf_pmu__config_terms Add const to related APIs, this is so they can be used to default initialize a perf_event_attr from a const pmu. Signed-off-by: Ian Rogers Reviewed-by: Adrian Hunter Cc: Ravi Bangoria Cc: James Clark Cc: Suzuki K Poulose Cc: Yang Jihong Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Kajol Jain Cc: Thomas Richter Cc: Kan Liang Cc: John Garry Cc: linux-arm-kernel@lists.infradead.org Cc: coresight@lists.linaro.org Link: https://lore.kernel.org/r/20231012175645.1849503-6-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/pmu.c | 10 +++++----- tools/perf/util/pmu.h | 2 +- 2 files changed, 6 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 861e485e0a70..8ef675ea7bdd 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -156,7 +156,7 @@ static void __perf_pmu_format__load(struct perf_pmu_format *format, FILE *file) format->loaded = true; } -static void perf_pmu_format__load(struct perf_pmu *pmu, struct perf_pmu_format *format) +static void perf_pmu_format__load(const struct perf_pmu *pmu, struct perf_pmu_format *format) { char path[PATH_MAX]; FILE *file = NULL; @@ -1131,7 +1131,7 @@ void evsel__set_config_if_unset(struct perf_pmu *pmu, struct evsel *evsel, } static struct perf_pmu_format * -pmu_find_format(struct list_head *formats, const char *name) +pmu_find_format(const struct list_head *formats, const char *name) { struct perf_pmu_format *format; @@ -1229,7 +1229,7 @@ static int pmu_resolve_param_term(struct parse_events_term *term, return -1; } -static char *pmu_formats_string(struct list_head *formats) +static char *pmu_formats_string(const struct list_head *formats) { struct perf_pmu_format *format; char *str = NULL; @@ -1255,7 +1255,7 @@ error: * Setup one of config[12] attr members based on the * user input data - term parameter. */ -static int pmu_config_term(struct perf_pmu *pmu, +static int pmu_config_term(const struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_term *term, struct parse_events_terms *head_terms, @@ -1378,7 +1378,7 @@ static int pmu_config_term(struct perf_pmu *pmu, return 0; } -int perf_pmu__config_terms(struct perf_pmu *pmu, +int perf_pmu__config_terms(const struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_terms *terms, bool zero, struct parse_events_error *err) diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 24af7297b522..5a05131aa4ce 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -193,7 +193,7 @@ void pmu_add_sys_aliases(struct perf_pmu *pmu); int perf_pmu__config(struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_terms *head_terms, struct parse_events_error *error); -int perf_pmu__config_terms(struct perf_pmu *pmu, +int perf_pmu__config_terms(const struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_terms *terms, bool zero, struct parse_events_error *error); -- cgit v1.2.3 From 0197da7affab502cd6e25da616ad038b169a7a77 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 12 Oct 2023 10:56:45 -0700 Subject: perf pmu: Lazily compute default config The default config is computed during creation of the PMU and may do things like scanning sysfs, when the PMU may just be used as part of scanning. Change default_config to perf_event_attr_init_default, a callback that is used when a default config needs initializing. This avoids holding onto the memory for a perf_event_attr and copying. On a tigerlake laptop running the pmu-scan benchmark: Before: Running 'internals/pmu-scan' benchmark: Computing performance of sysfs PMU event scan for 100 times Average core PMU scanning took: 28.780 usec (+- 0.503 usec) Average PMU scanning took: 283.480 usec (+- 18.471 usec) Number of openat syscalls: 30,227 After: Running 'internals/pmu-scan' benchmark: Computing performance of sysfs PMU event scan for 100 times Average core PMU scanning took: 27.880 usec (+- 0.169 usec) Average PMU scanning took: 245.260 usec (+- 15.758 usec) Number of openat syscalls: 28,914 Over 3 runs it is a nearly 12% reduction in execution time and a 4.3% of openat calls. Signed-off-by: Ian Rogers Reviewed-by: Adrian Hunter Cc: Ravi Bangoria Cc: James Clark Cc: Suzuki K Poulose Cc: Yang Jihong Cc: Will Deacon Cc: Leo Yan Cc: Mike Leach Cc: Jing Zhang Cc: Kajol Jain Cc: Thomas Richter Cc: Kan Liang Cc: John Garry Cc: linux-arm-kernel@lists.infradead.org Cc: coresight@lists.linaro.org Link: https://lore.kernel.org/r/20231012175645.1849503-8-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/arm-spe.h | 4 +++- tools/perf/util/cs-etm.h | 2 +- tools/perf/util/intel-pt.h | 3 ++- tools/perf/util/parse-events.c | 12 ++++++------ tools/perf/util/pmu.c | 3 +-- tools/perf/util/pmu.h | 7 ++++--- 6 files changed, 17 insertions(+), 14 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/arm-spe.h b/tools/perf/util/arm-spe.h index 98d3235781c3..4f4900c18f3e 100644 --- a/tools/perf/util/arm-spe.h +++ b/tools/perf/util/arm-spe.h @@ -27,5 +27,7 @@ struct auxtrace_record *arm_spe_recording_init(int *err, int arm_spe_process_auxtrace_info(union perf_event *event, struct perf_session *session); -struct perf_event_attr *arm_spe_pmu_default_config(struct perf_pmu *arm_spe_pmu); +void arm_spe_pmu_default_config(const struct perf_pmu *arm_spe_pmu, + struct perf_event_attr *attr); + #endif diff --git a/tools/perf/util/cs-etm.h b/tools/perf/util/cs-etm.h index 7cca37887917..4696267a32f0 100644 --- a/tools/perf/util/cs-etm.h +++ b/tools/perf/util/cs-etm.h @@ -242,7 +242,7 @@ struct cs_etm_packet_queue { int cs_etm__process_auxtrace_info(union perf_event *event, struct perf_session *session); -struct perf_event_attr *cs_etm_get_default_config(struct perf_pmu *pmu); +void cs_etm_get_default_config(const struct perf_pmu *pmu, struct perf_event_attr *attr); enum cs_etm_pid_fmt { CS_ETM_PIDFMT_NONE, diff --git a/tools/perf/util/intel-pt.h b/tools/perf/util/intel-pt.h index c7d6068e3a6b..18fd0be52e6c 100644 --- a/tools/perf/util/intel-pt.h +++ b/tools/perf/util/intel-pt.h @@ -42,6 +42,7 @@ struct auxtrace_record *intel_pt_recording_init(int *err); int intel_pt_process_auxtrace_info(union perf_event *event, struct perf_session *session); -struct perf_event_attr *intel_pt_pmu_default_config(struct perf_pmu *pmu); +void intel_pt_pmu_default_config(const struct perf_pmu *intel_pt_pmu, + struct perf_event_attr *attr); #endif diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 23c027cf20ae..aa2f5c6fc7fc 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -1418,11 +1418,10 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, } fix_raw(&parsed_terms, pmu); - if (pmu->default_config) { - memcpy(&attr, pmu->default_config, sizeof(struct perf_event_attr)); - } else { - memset(&attr, 0, sizeof(attr)); - } + memset(&attr, 0, sizeof(attr)); + if (pmu->perf_event_attr_init_default) + pmu->perf_event_attr_init_default(pmu, &attr); + attr.type = pmu->type; if (list_empty(&parsed_terms.terms)) { @@ -1466,7 +1465,8 @@ int parse_events_add_pmu(struct parse_events_state *parse_state, * When using default config, record which bits of attr->config were * changed by the user. */ - if (pmu->default_config && get_config_chgs(pmu, &parsed_terms, &config_terms)) { + if (pmu->perf_event_attr_init_default && + get_config_chgs(pmu, &parsed_terms, &config_terms)) { parse_events_terms__exit(&parsed_terms); return -ENOMEM; } diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 8ef675ea7bdd..a967d25e899b 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -1402,7 +1402,7 @@ int perf_pmu__config(struct perf_pmu *pmu, struct perf_event_attr *attr, struct parse_events_terms *head_terms, struct parse_events_error *err) { - bool zero = !!pmu->default_config; + bool zero = !!pmu->perf_event_attr_init_default; return perf_pmu__config_terms(pmu, attr, head_terms, zero, err); } @@ -2064,7 +2064,6 @@ void perf_pmu__delete(struct perf_pmu *pmu) perf_cpu_map__put(pmu->cpus); - zfree(&pmu->default_config); zfree(&pmu->name); zfree(&pmu->alias_name); zfree(&pmu->id); diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h index 5a05131aa4ce..d2895d415f08 100644 --- a/tools/perf/util/pmu.h +++ b/tools/perf/util/pmu.h @@ -92,10 +92,11 @@ struct perf_pmu { */ int max_precise; /** - * @default_config: Optional default perf_event_attr determined in - * architecture specific code. + * @perf_event_attr_init_default: Optional function to default + * initialize PMU specific parts of the perf_event_attr. */ - struct perf_event_attr *default_config; + void (*perf_event_attr_init_default)(const struct perf_pmu *pmu, + struct perf_event_attr *attr); /** * @cpus: Empty or the contents of either of: * /bus/event_source/devices//cpumask. -- cgit v1.2.3 From 1f36b190ad2dea68e3a7e84b7b2f24ce8c4063ea Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 10 Oct 2023 16:42:47 -0700 Subject: perf tools: Do not ignore the default vmlinux.h The recent change made it possible to generate vmlinux.h from BTF and to ignore the file. But we also have a minimal vmlinux.h that will be used by default. It should not be ignored by GIT. Fixes: b7a2d774c9c5 ("perf build: Add ability to build with a generated vmlinux.h") Reported-by: kernel test robot Reviewed-by: Ian Rogers Closes: https://lore.kernel.org/oe-kbuild-all/202310110451.rvdUZJEY-lkp@intel.com/ Cc: oe-kbuild-all@lists.linux.dev Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_skel/vmlinux/.gitignore | 1 + 1 file changed, 1 insertion(+) create mode 100644 tools/perf/util/bpf_skel/vmlinux/.gitignore (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_skel/vmlinux/.gitignore b/tools/perf/util/bpf_skel/vmlinux/.gitignore new file mode 100644 index 000000000000..49502c04183a --- /dev/null +++ b/tools/perf/util/bpf_skel/vmlinux/.gitignore @@ -0,0 +1 @@ +!vmlinux.h -- cgit v1.2.3 From 5069211e2f0b47e75119805e23ae6352d871e263 Mon Sep 17 00:00:00 2001 From: Thomas Richter Date: Thu, 19 Oct 2023 10:26:42 +0200 Subject: perf trace: Use the right bpf_probe_read(_str) variant for reading user data Perf test case 111 Check open filename arg using perf trace + vfs_getname fails on s390. This is caused by a failing function bpf_probe_read() in file util/bpf_skel/augmented_raw_syscalls.bpf.c. The root cause is the lookup by address. Function bpf_probe_read() is used. This function works only for architectures with ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE. On s390 is not possible to determine from the address to which address space the address belongs to (user or kernel space). Replace bpf_probe_read() by bpf_probe_read_kernel() and bpf_probe_read_str() by bpf_probe_read_user_str() to explicity specify the address space the address refers to. Output before: # ./perf trace -eopen,openat -- touch /tmp/111 libbpf: prog 'sys_enter': BPF program load failed: Invalid argument libbpf: prog 'sys_enter': -- BEGIN PROG LOAD LOG -- reg type unsupported for arg#0 function sys_enter#75 0: R1=ctx(off=0,imm=0) R10=fp0 ; int sys_enter(struct syscall_enter_args *args) 0: (bf) r6 = r1 ; R1=ctx(off=0,imm=0) R6_w=ctx(off=0,imm=0) ; return bpf_get_current_pid_tgid(); 1: (85) call bpf_get_current_pid_tgid#14 ; R0_w=scalar() 2: (63) *(u32 *)(r10 -8) = r0 ; R0_w=scalar() R10=fp0 fp-8=????mmmm 3: (bf) r2 = r10 ; R2_w=fp0 R10=fp0 ; ..... lines deleted here ..... 23: (bf) r3 = r6 ; R3_w=ctx(off=0,imm=0) R6=ctx(off=0,imm=0) 24: (85) call bpf_probe_read#4 unknown func bpf_probe_read#4 processed 23 insns (limit 1000000) max_states_per_insn 0 \ total_states 2 peak_states 2 mark_read 2 -- END PROG LOAD LOG -- libbpf: prog 'sys_enter': failed to load: -22 libbpf: failed to load object 'augmented_raw_syscalls_bpf' libbpf: failed to load BPF skeleton 'augmented_raw_syscalls_bpf': -22 .... Output after: # ./perf test -Fv 111 111: Check open filename arg using perf trace + vfs_getname : --- start --- 1.085 ( 0.011 ms): touch/320753 openat(dfd: CWD, filename: \ "/tmp/temporary_file.SWH85", \ flags: CREAT|NOCTTY|NONBLOCK|WRONLY, mode: IRUGO|IWUGO) = 3 ---- end ---- Check open filename arg using perf trace + vfs_getname: Ok # Test with the sleep command shows: Output before: # ./perf trace -e *sleep sleep 1.234567890 0.000 (1234.681 ms): sleep/63114 clock_nanosleep(rqtp: \ { .tv_sec: 0, .tv_nsec: 0 }, rmtp: 0x3ffe0979720) = 0 # Output after: # ./perf trace -e *sleep sleep 1.234567890 0.000 (1234.686 ms): sleep/64277 clock_nanosleep(rqtp: \ { .tv_sec: 1, .tv_nsec: 234567890 }, rmtp: 0x3fff3df9ea0) = 0 # Fixes: 14e4b9f4289a ("perf trace: Raw augmented syscalls fix libbpf 1.0+ compatibility") Signed-off-by: Thomas Richter Co-developed-by: Arnaldo Carvalho de Melo Acked-by: Ilya Leoshkevich Tested-by: Arnaldo Carvalho de Melo Cc: Ian Rogers Cc: gor@linux.ibm.com Cc: hca@linux.ibm.com Cc: sumanthk@linux.ibm.com Cc: svens@linux.ibm.com Link: https://lore.kernel.org/r/20231019082642.3286650-1-tmricht@linux.ibm.com Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c b/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c index 939ec769bf4a..52c270330ae0 100644 --- a/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c +++ b/tools/perf/util/bpf_skel/augmented_raw_syscalls.bpf.c @@ -153,7 +153,7 @@ static inline unsigned int augmented_arg__read_str(struct augmented_arg *augmented_arg, const void *arg, unsigned int arg_len) { unsigned int augmented_len = sizeof(*augmented_arg); - int string_len = bpf_probe_read_str(&augmented_arg->value, arg_len, arg); + int string_len = bpf_probe_read_user_str(&augmented_arg->value, arg_len, arg); augmented_arg->size = augmented_arg->err = 0; /* @@ -203,7 +203,7 @@ int sys_enter_connect(struct syscall_enter_args *args) _Static_assert(is_power_of_2(sizeof(augmented_args->saddr)), "sizeof(augmented_args->saddr) needs to be a power of two"); socklen &= sizeof(augmented_args->saddr) - 1; - bpf_probe_read(&augmented_args->saddr, socklen, sockaddr_arg); + bpf_probe_read_user(&augmented_args->saddr, socklen, sockaddr_arg); return augmented__output(args, augmented_args, len + socklen); } @@ -221,7 +221,7 @@ int sys_enter_sendto(struct syscall_enter_args *args) socklen &= sizeof(augmented_args->saddr) - 1; - bpf_probe_read(&augmented_args->saddr, socklen, sockaddr_arg); + bpf_probe_read_user(&augmented_args->saddr, socklen, sockaddr_arg); return augmented__output(args, augmented_args, len + socklen); } @@ -311,7 +311,7 @@ int sys_enter_perf_event_open(struct syscall_enter_args *args) if (augmented_args == NULL) goto failure; - if (bpf_probe_read(&augmented_args->__data, sizeof(*attr), attr) < 0) + if (bpf_probe_read_user(&augmented_args->__data, sizeof(*attr), attr) < 0) goto failure; attr_read = (const struct perf_event_attr_size *)augmented_args->__data; @@ -325,7 +325,7 @@ int sys_enter_perf_event_open(struct syscall_enter_args *args) goto failure; // Now that we read attr->size and tested it against the size limits, read it completely - if (bpf_probe_read(&augmented_args->__data, size, attr) < 0) + if (bpf_probe_read_user(&augmented_args->__data, size, attr) < 0) goto failure; return augmented__output(args, augmented_args, len + size); @@ -347,7 +347,7 @@ int sys_enter_clock_nanosleep(struct syscall_enter_args *args) if (size > sizeof(augmented_args->__data)) goto failure; - bpf_probe_read(&augmented_args->__data, size, rqtp_arg); + bpf_probe_read_user(&augmented_args->__data, size, rqtp_arg); return augmented__output(args, augmented_args, len + size); failure: @@ -385,7 +385,7 @@ int sys_enter(struct syscall_enter_args *args) if (augmented_args == NULL) return 1; - bpf_probe_read(&augmented_args->args, sizeof(augmented_args->args), args); + bpf_probe_read_kernel(&augmented_args->args, sizeof(augmented_args->args), args); /* * Jump to syscall specific augmenter, even if the default one, @@ -406,7 +406,7 @@ int sys_exit(struct syscall_exit_args *args) if (pid_filter__has(&pids_filtered, getpid())) return 0; - bpf_probe_read(&exit_args, sizeof(exit_args), args); + bpf_probe_read_kernel(&exit_args, sizeof(exit_args), args); /* * Jump to syscall specific return augmenter, even if the default one, * "!raw_syscalls:unaugmented" that will just return 1 to return the -- cgit v1.2.3 From c4a852635eddcf4130a93e687bc7b120b19fd1b7 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Fri, 13 Oct 2023 07:59:45 +0000 Subject: perf data: Increase RLIMIT_NOFILE limit when open too many files in perf_data__create_dir() If using parallel threads to collect data, perf record needs at least 6 fds per CPU. (one for sys_perf_event_open, four for pipe msg and ack of the pipe, see record__thread_data_open_pipes(), and one for open perf.data.XXX) For an environment with more than 100 cores, if perf record uses both `-a` and `--threads` options, it is easy to exceed the upper limit of the file descriptor number, when we run out of them try to increase the limits. Before: $ ulimit -n 1024 $ lscpu | grep 'On-line CPU(s)' On-line CPU(s) list: 0-159 $ perf record --threads -a sleep 1 Failed to create data directory: Too many open files After: $ ulimit -n 1024 $ lscpu | grep 'On-line CPU(s)' On-line CPU(s) list: 0-159 $ perf record --threads -a sleep 1 [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.394 MB perf.data (1576 samples) ] Signed-off-by: Yang Jihong Acked-by: Namhyung Kim Link: https://lore.kernel.org/r/20231013075945.698874-1-yangjihong1@huawei.com Signed-off-by: Namhyung Kim --- tools/perf/util/data.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/data.c b/tools/perf/util/data.c index fc16299c915f..098f9e3bb2e7 100644 --- a/tools/perf/util/data.c +++ b/tools/perf/util/data.c @@ -17,6 +17,7 @@ #include "util.h" // rm_rf_perf_data() #include "debug.h" #include "header.h" +#include "evsel.h" #include static void close_dir(struct perf_data_file *files, int nr) @@ -35,6 +36,7 @@ void perf_data__close_dir(struct perf_data *data) int perf_data__create_dir(struct perf_data *data, int nr) { + enum rlimit_action set_rlimit = NO_CHANGE; struct perf_data_file *files = NULL; int i, ret; @@ -54,11 +56,21 @@ int perf_data__create_dir(struct perf_data *data, int nr) goto out_err; } +retry_open: ret = open(file->path, O_RDWR|O_CREAT|O_TRUNC, S_IRUSR|S_IWUSR); if (ret < 0) { + /* + * If using parallel threads to collect data, + * perf record needs at least 6 fds per CPU. + * When we run out of them try to increase the limits. + */ + if (errno == EMFILE && evsel__increase_rlimit(&set_rlimit)) + goto retry_open; + ret = -errno; goto out_err; } + set_rlimit = NO_CHANGE; file->fd = ret; } -- cgit v1.2.3 From e093a222d7cba1eb6c36887e58ce8a4ff249f1c6 Mon Sep 17 00:00:00 2001 From: Yang Jihong Date: Mon, 23 Oct 2023 03:31:44 +0000 Subject: perf evsel: Rename evsel__increase_rlimit to rlimit__increase_nofile evsel__increase_rlimit() helper does nothing with evsel, and description of the functionality is inaccurate, rename it and move to util/rlimit.c. By the way, fix a checkppatch warning about misplaced license tag: WARNING: Misplaced SPDX-License-Identifier tag - use line 1 instead #160: FILE: tools/perf/util/rlimit.h:3: /* SPDX-License-Identifier: LGPL-2.1 */ No functional change. Signed-off-by: Yang Jihong Link: https://lore.kernel.org/r/20231023033144.1011896-1-yangjihong1@huawei.com Signed-off-by: Namhyung Kim --- tools/perf/util/data.c | 4 ++-- tools/perf/util/evsel.c | 30 ++---------------------------- tools/perf/util/evsel.h | 3 --- tools/perf/util/rlimit.c | 28 ++++++++++++++++++++++++++++ tools/perf/util/rlimit.h | 11 ++++++++++- 5 files changed, 42 insertions(+), 34 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/data.c b/tools/perf/util/data.c index 098f9e3bb2e7..c29d8a382b19 100644 --- a/tools/perf/util/data.c +++ b/tools/perf/util/data.c @@ -17,7 +17,7 @@ #include "util.h" // rm_rf_perf_data() #include "debug.h" #include "header.h" -#include "evsel.h" +#include "rlimit.h" #include static void close_dir(struct perf_data_file *files, int nr) @@ -64,7 +64,7 @@ retry_open: * perf record needs at least 6 fds per CPU. * When we run out of them try to increase the limits. */ - if (errno == EMFILE && evsel__increase_rlimit(&set_rlimit)) + if (errno == EMFILE && rlimit__increase_nofile(&set_rlimit)) goto retry_open; ret = -errno; diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index d5363d23f5d3..72a5dfc38d38 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -49,6 +49,7 @@ #include "off_cpu.h" #include "pmu.h" #include "pmus.h" +#include "rlimit.h" #include "../perf-sys.h" #include "util/parse-branch-options.h" #include "util/bpf-filter.h" @@ -1989,33 +1990,6 @@ bool evsel__detect_missing_features(struct evsel *evsel) } } -bool evsel__increase_rlimit(enum rlimit_action *set_rlimit) -{ - int old_errno; - struct rlimit l; - - if (*set_rlimit < INCREASED_MAX) { - old_errno = errno; - - if (getrlimit(RLIMIT_NOFILE, &l) == 0) { - if (*set_rlimit == NO_CHANGE) { - l.rlim_cur = l.rlim_max; - } else { - l.rlim_cur = l.rlim_max + 1000; - l.rlim_max = l.rlim_cur; - } - if (setrlimit(RLIMIT_NOFILE, &l) == 0) { - (*set_rlimit) += 1; - errno = old_errno; - return true; - } - } - errno = old_errno; - } - - return false; -} - static int evsel__open_cpu(struct evsel *evsel, struct perf_cpu_map *cpus, struct perf_thread_map *threads, int start_cpu_map_idx, int end_cpu_map_idx) @@ -2143,7 +2117,7 @@ try_fallback: * perf stat needs between 5 and 22 fds per CPU. When we run out * of them try to increase the limits. */ - if (err == -EMFILE && evsel__increase_rlimit(&set_rlimit)) + if (err == -EMFILE && rlimit__increase_nofile(&set_rlimit)) goto retry_open; if (err != -EINVAL || idx > 0 || thread > 0) diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 815be2491938..d791316a1792 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -330,9 +330,6 @@ int evsel__prepare_open(struct evsel *evsel, struct perf_cpu_map *cpus, struct perf_thread_map *threads); bool evsel__detect_missing_features(struct evsel *evsel); -enum rlimit_action { NO_CHANGE, SET_TO_MAX, INCREASED_MAX }; -bool evsel__increase_rlimit(enum rlimit_action *set_rlimit); - bool evsel__precise_ip_fallback(struct evsel *evsel); struct perf_sample; diff --git a/tools/perf/util/rlimit.c b/tools/perf/util/rlimit.c index 13521d392a22..f857405fe1aa 100644 --- a/tools/perf/util/rlimit.c +++ b/tools/perf/util/rlimit.c @@ -1,5 +1,6 @@ /* SPDX-License-Identifier: LGPL-2.1 */ +#include #include "util/debug.h" #include "util/rlimit.h" #include @@ -27,3 +28,30 @@ void rlimit__bump_memlock(void) } } } + +bool rlimit__increase_nofile(enum rlimit_action *set_rlimit) +{ + int old_errno; + struct rlimit l; + + if (*set_rlimit < INCREASED_MAX) { + old_errno = errno; + + if (getrlimit(RLIMIT_NOFILE, &l) == 0) { + if (*set_rlimit == NO_CHANGE) { + l.rlim_cur = l.rlim_max; + } else { + l.rlim_cur = l.rlim_max + 1000; + l.rlim_max = l.rlim_cur; + } + if (setrlimit(RLIMIT_NOFILE, &l) == 0) { + (*set_rlimit) += 1; + errno = old_errno; + return true; + } + } + errno = old_errno; + } + + return false; +} diff --git a/tools/perf/util/rlimit.h b/tools/perf/util/rlimit.h index 9f59d8e710a3..19050d7fb9d7 100644 --- a/tools/perf/util/rlimit.h +++ b/tools/perf/util/rlimit.h @@ -1,6 +1,15 @@ +/* SPDX-License-Identifier: LGPL-2.1 */ #ifndef __PERF_RLIMIT_H_ #define __PERF_RLIMIT_H_ -/* SPDX-License-Identifier: LGPL-2.1 */ + +enum rlimit_action { + NO_CHANGE, + SET_TO_MAX, + INCREASED_MAX +}; void rlimit__bump_memlock(void); + +bool rlimit__increase_nofile(enum rlimit_action *set_rlimit); + #endif // __PERF_RLIMIT_H_ -- cgit v1.2.3 From d99317f214ca3d381f7b17a75c41263885664e06 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 20 Oct 2023 13:47:39 -0700 Subject: perf lock contention: Clear lock addr after use It checks the current lock to calculated the delta of contention time. The address is saved in the tstamp map which is allocated at begining of contention and released at end of contention. But it's possible for bpf_map_delete_elem() to fail. In that case, the element in the tstamp map kept for the current lock and it makes the next contention for the same lock tracked incorrectly. Specificially the next contention begin will see the existing element for the task and it'd just return. Then the next contention end will see the element and calculate the time using the timestamp for the previous begin. This can result in a large value for two small contentions happened from time to time. Let's clear the lock address so that it can be updated next time even if the bpf_map_delete_elem() failed. Signed-off-by: Namhyung Kim Acked-by: Ian Rogers Cc: Hao Luo Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231020204741.1869520-1-namhyung@kernel.org --- tools/perf/util/bpf_skel/lock_contention.bpf.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_skel/lock_contention.bpf.c b/tools/perf/util/bpf_skel/lock_contention.bpf.c index 4900a5dfb4a4..b11179452e19 100644 --- a/tools/perf/util/bpf_skel/lock_contention.bpf.c +++ b/tools/perf/util/bpf_skel/lock_contention.bpf.c @@ -389,6 +389,7 @@ int contention_end(u64 *ctx) duration = bpf_ktime_get_ns() - pelem->timestamp; if ((__s64)duration < 0) { + pelem->lock = 0; bpf_map_delete_elem(&tstamp, &pid); __sync_fetch_and_add(&time_fail, 1); return 0; @@ -422,6 +423,7 @@ int contention_end(u64 *ctx) data = bpf_map_lookup_elem(&lock_stat, &key); if (!data) { if (data_map_full) { + pelem->lock = 0; bpf_map_delete_elem(&tstamp, &pid); __sync_fetch_and_add(&data_fail, 1); return 0; @@ -445,6 +447,7 @@ int contention_end(u64 *ctx) data_map_full = 1; __sync_fetch_and_add(&data_fail, 1); } + pelem->lock = 0; bpf_map_delete_elem(&tstamp, &pid); return 0; } @@ -458,6 +461,7 @@ int contention_end(u64 *ctx) if (data->min_time > duration) data->min_time = duration; + pelem->lock = 0; bpf_map_delete_elem(&tstamp, &pid); return 0; } -- cgit v1.2.3 From 6a070573f290f99a6129ac3e13b9df521a1a65de Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 20 Oct 2023 13:47:40 -0700 Subject: perf lock contention: Check race in tstamp elem creation When pelem is NULL, it'd create a new entry with zero data. But it might be preempted by IRQ/NMI just before calling bpf_map_update_elem() then there's a chance to call it twice for the same pid. So it'd be better to use BPF_NOEXIST flag and check the return value to prevent the race. Signed-off-by: Namhyung Kim Acked-by: Ian Rogers Cc: Hao Luo Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231020204741.1869520-2-namhyung@kernel.org --- tools/perf/util/bpf_skel/lock_contention.bpf.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_skel/lock_contention.bpf.c b/tools/perf/util/bpf_skel/lock_contention.bpf.c index b11179452e19..69d31fd77cd0 100644 --- a/tools/perf/util/bpf_skel/lock_contention.bpf.c +++ b/tools/perf/util/bpf_skel/lock_contention.bpf.c @@ -328,7 +328,11 @@ int contention_begin(u64 *ctx) if (pelem == NULL) { struct tstamp_data zero = {}; - bpf_map_update_elem(&tstamp, &pid, &zero, BPF_ANY); + if (bpf_map_update_elem(&tstamp, &pid, &zero, BPF_NOEXIST) < 0) { + __sync_fetch_and_add(&task_fail, 1); + return 0; + } + pelem = bpf_map_lookup_elem(&tstamp, &pid); if (pelem == NULL) { __sync_fetch_and_add(&task_fail, 1); -- cgit v1.2.3 From b5711042a1c8cc88ed40a5ebf612b36e83a4e2e4 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 20 Oct 2023 13:47:41 -0700 Subject: perf lock contention: Use per-cpu array map for spinlocks Currently lock contention timestamp is maintained in a hash map keyed by pid. That means it needs to get and release a map element (which is proctected by spinlock!) on each contention begin and end pair. This can impact on performance if there are a lot of contention (usually from spinlocks). It used to go with task local storage but it had an issue on memory allocation in some critical paths. Although it's addressed in recent kernels IIUC, the tool should support old kernels too. So it cannot simply switch to the task local storage at least for now. As spinlocks create lots of contention and they disabled preemption during the spinning, it can use per-cpu array to keep the timestamp to avoid overhead in hashmap update and delete. In contention_begin, it's easy to check the lock types since it can see the flags. But contention_end cannot see it. So let's try to per-cpu array first (unconditionally) if it has an active element (lock != 0). Then it should be used and per-task tstamp map should not be used until the per-cpu array element is cleared which means nested spinlock contention (if any) was finished and it nows see (the outer) lock. Signed-off-by: Namhyung Kim Acked-by: Ian Rogers Cc: Hao Luo Cc: Song Liu Cc: bpf@vger.kernel.org Link: https://lore.kernel.org/r/20231020204741.1869520-3-namhyung@kernel.org --- tools/perf/util/bpf_skel/lock_contention.bpf.c | 89 +++++++++++++++++++++----- 1 file changed, 72 insertions(+), 17 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_skel/lock_contention.bpf.c b/tools/perf/util/bpf_skel/lock_contention.bpf.c index 69d31fd77cd0..95cd8414f6ef 100644 --- a/tools/perf/util/bpf_skel/lock_contention.bpf.c +++ b/tools/perf/util/bpf_skel/lock_contention.bpf.c @@ -42,6 +42,14 @@ struct { __uint(max_entries, MAX_ENTRIES); } tstamp SEC(".maps"); +/* maintain per-CPU timestamp at the beginning of contention */ +struct { + __uint(type, BPF_MAP_TYPE_PERCPU_ARRAY); + __uint(key_size, sizeof(__u32)); + __uint(value_size, sizeof(struct tstamp_data)); + __uint(max_entries, 1); +} tstamp_cpu SEC(".maps"); + /* actual lock contention statistics */ struct { __uint(type, BPF_MAP_TYPE_HASH); @@ -311,34 +319,57 @@ static inline __u32 check_lock_type(__u64 lock, __u32 flags) return 0; } -SEC("tp_btf/contention_begin") -int contention_begin(u64 *ctx) +static inline struct tstamp_data *get_tstamp_elem(__u32 flags) { __u32 pid; struct tstamp_data *pelem; - if (!enabled || !can_record(ctx)) - return 0; + /* Use per-cpu array map for spinlock and rwlock */ + if (flags == (LCB_F_SPIN | LCB_F_READ) || flags == LCB_F_SPIN || + flags == (LCB_F_SPIN | LCB_F_WRITE)) { + __u32 idx = 0; + + pelem = bpf_map_lookup_elem(&tstamp_cpu, &idx); + /* Do not update the element for nested locks */ + if (pelem && pelem->lock) + pelem = NULL; + return pelem; + } pid = bpf_get_current_pid_tgid(); pelem = bpf_map_lookup_elem(&tstamp, &pid); + /* Do not update the element for nested locks */ if (pelem && pelem->lock) - return 0; + return NULL; if (pelem == NULL) { struct tstamp_data zero = {}; if (bpf_map_update_elem(&tstamp, &pid, &zero, BPF_NOEXIST) < 0) { __sync_fetch_and_add(&task_fail, 1); - return 0; + return NULL; } pelem = bpf_map_lookup_elem(&tstamp, &pid); if (pelem == NULL) { __sync_fetch_and_add(&task_fail, 1); - return 0; + return NULL; } } + return pelem; +} + +SEC("tp_btf/contention_begin") +int contention_begin(u64 *ctx) +{ + struct tstamp_data *pelem; + + if (!enabled || !can_record(ctx)) + return 0; + + pelem = get_tstamp_elem(ctx[1]); + if (pelem == NULL) + return 0; pelem->timestamp = bpf_ktime_get_ns(); pelem->lock = (__u64)ctx[0]; @@ -377,24 +408,42 @@ int contention_begin(u64 *ctx) SEC("tp_btf/contention_end") int contention_end(u64 *ctx) { - __u32 pid; + __u32 pid = 0, idx = 0; struct tstamp_data *pelem; struct contention_key key = {}; struct contention_data *data; __u64 duration; + bool need_delete = false; if (!enabled) return 0; - pid = bpf_get_current_pid_tgid(); - pelem = bpf_map_lookup_elem(&tstamp, &pid); - if (!pelem || pelem->lock != ctx[0]) - return 0; + /* + * For spinlock and rwlock, it needs to get the timestamp for the + * per-cpu map. However, contention_end does not have the flags + * so it cannot know whether it reads percpu or hash map. + * + * Try per-cpu map first and check if there's active contention. + * If it is, do not read hash map because it cannot go to sleeping + * locks before releasing the spinning locks. + */ + pelem = bpf_map_lookup_elem(&tstamp_cpu, &idx); + if (pelem && pelem->lock) { + if (pelem->lock != ctx[0]) + return 0; + } else { + pid = bpf_get_current_pid_tgid(); + pelem = bpf_map_lookup_elem(&tstamp, &pid); + if (!pelem || pelem->lock != ctx[0]) + return 0; + need_delete = true; + } duration = bpf_ktime_get_ns() - pelem->timestamp; if ((__s64)duration < 0) { pelem->lock = 0; - bpf_map_delete_elem(&tstamp, &pid); + if (need_delete) + bpf_map_delete_elem(&tstamp, &pid); __sync_fetch_and_add(&time_fail, 1); return 0; } @@ -406,8 +455,11 @@ int contention_end(u64 *ctx) case LOCK_AGGR_TASK: if (lock_owner) key.pid = pelem->flags; - else + else { + if (!need_delete) + pid = bpf_get_current_pid_tgid(); key.pid = pid; + } if (needs_callstack) key.stack_id = pelem->stack_id; break; @@ -428,7 +480,8 @@ int contention_end(u64 *ctx) if (!data) { if (data_map_full) { pelem->lock = 0; - bpf_map_delete_elem(&tstamp, &pid); + if (need_delete) + bpf_map_delete_elem(&tstamp, &pid); __sync_fetch_and_add(&data_fail, 1); return 0; } @@ -452,7 +505,8 @@ int contention_end(u64 *ctx) __sync_fetch_and_add(&data_fail, 1); } pelem->lock = 0; - bpf_map_delete_elem(&tstamp, &pid); + if (need_delete) + bpf_map_delete_elem(&tstamp, &pid); return 0; } @@ -466,7 +520,8 @@ int contention_end(u64 *ctx) data->min_time = duration; pelem->lock = 0; - bpf_map_delete_elem(&tstamp, &pid); + if (need_delete) + bpf_map_delete_elem(&tstamp, &pid); return 0; } -- cgit v1.2.3 From 7a8f349e9d14da1633e021c2fe87234b21ab181d Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:04 -0700 Subject: perf rwsem: Add debug mode that uses a mutex Mutex error check will capture trying to take the lock recursively and other problems that rwlock won't. At the expense of concurrency, adda debug mode that uses a mutex in place of a rwsem. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-2-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/rwsem.c | 34 ++++++++++++++++++++++++++++++++++ tools/perf/util/rwsem.h | 11 +++++++++++ 2 files changed, 45 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/rwsem.c b/tools/perf/util/rwsem.c index f3d29d8ddc99..5109167f27f7 100644 --- a/tools/perf/util/rwsem.c +++ b/tools/perf/util/rwsem.c @@ -2,32 +2,66 @@ #include "util.h" #include "rwsem.h" +#if RWS_ERRORCHECK +#include "mutex.h" +#endif + int init_rwsem(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_init(&sem->mtx); + return 0; +#else return pthread_rwlock_init(&sem->lock, NULL); +#endif } int exit_rwsem(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_destroy(&sem->mtx); + return 0; +#else return pthread_rwlock_destroy(&sem->lock); +#endif } int down_read(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_lock(&sem->mtx); + return 0; +#else return perf_singlethreaded ? 0 : pthread_rwlock_rdlock(&sem->lock); +#endif } int up_read(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_unlock(&sem->mtx); + return 0; +#else return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock); +#endif } int down_write(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_lock(&sem->mtx); + return 0; +#else return perf_singlethreaded ? 0 : pthread_rwlock_wrlock(&sem->lock); +#endif } int up_write(struct rw_semaphore *sem) { +#if RWS_ERRORCHECK + mutex_unlock(&sem->mtx); + return 0; +#else return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock); +#endif } diff --git a/tools/perf/util/rwsem.h b/tools/perf/util/rwsem.h index 94565ad4d494..ef5cbc31d967 100644 --- a/tools/perf/util/rwsem.h +++ b/tools/perf/util/rwsem.h @@ -2,9 +2,20 @@ #define _PERF_RWSEM_H #include +#include "mutex.h" + +/* + * Mutexes have additional error checking. Enable to use a mutex rather than a + * rwlock for debugging. + */ +#define RWS_ERRORCHECK 0 struct rw_semaphore { +#if RWS_ERRORCHECK + struct mutex mtx; +#else pthread_rwlock_t lock; +#endif }; int init_rwsem(struct rw_semaphore *sem); -- cgit v1.2.3 From ab8ce150781d326c6bfbe1e09f175ffde1186f80 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:05 -0700 Subject: perf machine: Avoid out of bounds LBR memory read Running perf top with address sanitizer and "--call-graph=lbr" fails due to reading sample 0 when no samples exist. Add a guard to prevent this. Fixes: e2b23483eb1d ("perf machine: Factor out lbr_callchain_add_lbr_ip()") Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-3-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/machine.c | 22 ++++++++++++---------- 1 file changed, 12 insertions(+), 10 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index addfae2f63ef..e0e2c4a943e4 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2622,16 +2622,18 @@ static int lbr_callchain_add_lbr_ip(struct thread *thread, save_lbr_cursor_node(thread, cursor, i); } - /* Add LBR ip from first entries.to */ - ip = entries[0].to; - flags = &entries[0].flags; - *branch_from = entries[0].from; - err = add_callchain_ip(thread, cursor, parent, - root_al, &cpumode, ip, - true, flags, NULL, - *branch_from); - if (err) - return err; + if (lbr_nr > 0) { + /* Add LBR ip from first entries.to */ + ip = entries[0].to; + flags = &entries[0].flags; + *branch_from = entries[0].from; + err = add_callchain_ip(thread, cursor, parent, + root_al, &cpumode, ip, + true, flags, NULL, + *branch_from); + if (err) + return err; + } return 0; } -- cgit v1.2.3 From 78c32f4cb12f9430e29402118635dcf972b7d325 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:07 -0700 Subject: libperf rc_check: Add RC_CHK_EQUAL Comparing pointers with reference count checking is tricky to avoid a SEGV. Add a convenience macro to simplify and use. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-5-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/callchain.c | 2 +- tools/perf/util/hist.c | 2 +- tools/perf/util/machine.c | 4 ++-- tools/perf/util/sort.c | 2 +- tools/perf/util/symbol.c | 4 ++-- 5 files changed, 7 insertions(+), 7 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index aee937d14fbb..18d545c0629e 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -1142,7 +1142,7 @@ int fill_callchain_info(struct addr_location *al, struct callchain_cursor_node * if (al->map == NULL) goto out; } - if (RC_CHK_ACCESS(al->maps) == RC_CHK_ACCESS(machine__kernel_maps(machine))) { + if (RC_CHK_EQUAL(al->maps, machine__kernel_maps(machine))) { if (machine__is_host(machine)) { al->cpumode = PERF_RECORD_MISC_KERNEL; al->level = 'k'; diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 3dc8a4968beb..cde0078e6c90 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -2142,7 +2142,7 @@ static bool hists__filter_entry_by_thread(struct hists *hists, struct hist_entry *he) { if (hists->thread_filter != NULL && - RC_CHK_ACCESS(he->thread) != RC_CHK_ACCESS(hists->thread_filter)) { + !RC_CHK_EQUAL(he->thread, hists->thread_filter)) { he->filtered |= (1 << HIST_FILTER__THREAD); return true; } diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index e0e2c4a943e4..098600d983c5 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -969,7 +969,7 @@ static int machine__process_ksymbol_unregister(struct machine *machine, if (!map) return 0; - if (RC_CHK_ACCESS(map) != RC_CHK_ACCESS(machine->vmlinux_map)) + if (!RC_CHK_EQUAL(map, machine->vmlinux_map)) maps__remove(machine__kernel_maps(machine), map); else { struct dso *dso = map__dso(map); @@ -2058,7 +2058,7 @@ static void __machine__remove_thread(struct machine *machine, struct thread_rb_n if (!nd) nd = thread_rb_node__find(th, &threads->entries.rb_root); - if (threads->last_match && RC_CHK_ACCESS(threads->last_match) == RC_CHK_ACCESS(th)) + if (threads->last_match && RC_CHK_EQUAL(threads->last_match, th)) threads__set_last_match(threads, NULL); if (lock) diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c index 6aa1c7f2b444..80e4f6132740 100644 --- a/tools/perf/util/sort.c +++ b/tools/perf/util/sort.c @@ -128,7 +128,7 @@ static int hist_entry__thread_filter(struct hist_entry *he, int type, const void if (type != HIST_FILTER__THREAD) return -1; - return th && RC_CHK_ACCESS(he->thread) != RC_CHK_ACCESS(th); + return th && !RC_CHK_EQUAL(he->thread, th); } struct sort_entry sort_thread = { diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 96587fd7a5a2..822f4dcebfe6 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -877,7 +877,7 @@ static int maps__split_kallsyms(struct maps *kmaps, struct dso *dso, u64 delta, *module++ = '\0'; curr_map_dso = map__dso(curr_map); if (strcmp(curr_map_dso->short_name, module)) { - if (RC_CHK_ACCESS(curr_map) != RC_CHK_ACCESS(initial_map) && + if (!RC_CHK_EQUAL(curr_map, initial_map) && dso->kernel == DSO_SPACE__KERNEL_GUEST && machine__is_default_guest(machine)) { /* @@ -1469,7 +1469,7 @@ static int dso__load_kcore(struct dso *dso, struct map *map, list_del_init(&new_node->node); - if (RC_CHK_ACCESS(new_map) == RC_CHK_ACCESS(replacement_map)) { + if (RC_CHK_EQUAL(new_map, replacement_map)) { struct map *map_ref; map__set_start(map, map__start(new_map)); -- cgit v1.2.3 From c1149037f65bcf0334886180ebe3d5efcf214912 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:08 -0700 Subject: perf hist: Add missing puts to hist__account_cycles Caught using reference count checking on perf top with "--call-graph=lbr". After this no memory leaks were detected. Fixes: 57849998e2cd ("perf report: Add processing for cycle histograms") Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-6-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/hist.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index cde0078e6c90..0aa7e231172a 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -2676,8 +2676,6 @@ void hist__account_cycles(struct branch_stack *bs, struct addr_location *al, /* If we have branch cycles always annotate them. */ if (bs && bs->nr && entries[0].flags.cycles) { - int i; - bi = sample__resolve_bstack(sample, al); if (bi) { struct addr_map_symbol *prev = NULL; @@ -2692,7 +2690,7 @@ void hist__account_cycles(struct branch_stack *bs, struct addr_location *al, * Note that perf stores branches reversed from * program order! */ - for (i = bs->nr - 1; i >= 0; i--) { + for (int i = bs->nr - 1; i >= 0; i--) { addr_map_symbol__account_cycles(&bi[i].from, nonany_branch_mode ? NULL : prev, bi[i].flags.cycles); @@ -2701,6 +2699,12 @@ void hist__account_cycles(struct branch_stack *bs, struct addr_location *al, if (total_cycles) *total_cycles += bi[i].flags.cycles; } + for (unsigned int i = 0; i < bs->nr; i++) { + map__put(bi[i].to.ms.map); + maps__put(bi[i].to.ms.maps); + map__put(bi[i].from.ms.map); + maps__put(bi[i].from.ms.maps); + } free(bi); } } -- cgit v1.2.3 From 7b2e444b76ceff71dad53fb8705780741421b570 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:09 -0700 Subject: perf threads: Remove unused dead thread list Commit 40826c45eb0b ("perf thread: Remove notion of dead threads") removed dead threads but the list head wasn't removed. Remove it here. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-7-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/machine.c | 1 - tools/perf/util/machine.h | 1 - 2 files changed, 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 098600d983c5..31fdbb0f27af 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -67,7 +67,6 @@ static void machine__threads_init(struct machine *machine) threads->entries = RB_ROOT_CACHED; init_rwsem(&threads->lock); threads->nr = 0; - INIT_LIST_HEAD(&threads->dead); threads->last_match = NULL; } } diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h index d034ecaf89c1..1279acda6a8a 100644 --- a/tools/perf/util/machine.h +++ b/tools/perf/util/machine.h @@ -35,7 +35,6 @@ struct threads { struct rb_root_cached entries; struct rw_semaphore lock; unsigned int nr; - struct list_head dead; struct thread *last_match; }; -- cgit v1.2.3 From 67a3ebf1c3588ee2ab994a4c96ef19179209c132 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:10 -0700 Subject: perf offcpu: Add missed btf_free Caught by address/leak sanitizer. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-8-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/bpf_off_cpu.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/bpf_off_cpu.c b/tools/perf/util/bpf_off_cpu.c index 21f4d9ba023d..6af36142dc5a 100644 --- a/tools/perf/util/bpf_off_cpu.c +++ b/tools/perf/util/bpf_off_cpu.c @@ -98,22 +98,22 @@ static void off_cpu_finish(void *arg __maybe_unused) /* v5.18 kernel added prev_state arg, so it needs to check the signature */ static void check_sched_switch_args(void) { - const struct btf *btf = btf__load_vmlinux_btf(); + struct btf *btf = btf__load_vmlinux_btf(); const struct btf_type *t1, *t2, *t3; u32 type_id; type_id = btf__find_by_name_kind(btf, "btf_trace_sched_switch", BTF_KIND_TYPEDEF); if ((s32)type_id < 0) - return; + goto cleanup; t1 = btf__type_by_id(btf, type_id); if (t1 == NULL) - return; + goto cleanup; t2 = btf__type_by_id(btf, t1->type); if (t2 == NULL || !btf_is_ptr(t2)) - return; + goto cleanup; t3 = btf__type_by_id(btf, t2->type); /* btf_trace func proto has one more argument for the context */ @@ -121,6 +121,8 @@ static void check_sched_switch_args(void) /* new format: pass prev_state as 4th arg */ skel->rodata->has_prev_state = true; } +cleanup: + btf__free(btf); } int off_cpu_prepare(struct evlist *evlist, struct target *target, -- cgit v1.2.3 From d47d876d72624ee1ef31f2e491be4393d4eacaf0 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:11 -0700 Subject: perf callchain: Make display use of branch_type_stat const Display code doesn't modify the branch_type_stat so switch uses to const. This is done to aid refactoring struct callchain_list where current the branch_type_stat is embedded even if not used. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-9-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/branch.c | 4 ++-- tools/perf/util/branch.h | 4 ++-- tools/perf/util/callchain.c | 6 +++--- 3 files changed, 7 insertions(+), 7 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/branch.c b/tools/perf/util/branch.c index 378f16a24751..ab760e267d41 100644 --- a/tools/perf/util/branch.c +++ b/tools/perf/util/branch.c @@ -109,7 +109,7 @@ const char *get_branch_type(struct branch_entry *e) return branch_type_name(e->flags.type); } -void branch_type_stat_display(FILE *fp, struct branch_type_stat *st) +void branch_type_stat_display(FILE *fp, const struct branch_type_stat *st) { u64 total = 0; int i; @@ -171,7 +171,7 @@ static int count_str_scnprintf(int idx, const char *str, char *bf, int size) return scnprintf(bf, size, "%s%s", (idx) ? " " : " (", str); } -int branch_type_str(struct branch_type_stat *st, char *bf, int size) +int branch_type_str(const struct branch_type_stat *st, char *bf, int size) { int i, j = 0, printed = 0; u64 total = 0; diff --git a/tools/perf/util/branch.h b/tools/perf/util/branch.h index e41bfffe2217..87704d713ff6 100644 --- a/tools/perf/util/branch.h +++ b/tools/perf/util/branch.h @@ -86,8 +86,8 @@ void branch_type_count(struct branch_type_stat *st, struct branch_flags *flags, const char *branch_type_name(int type); const char *branch_new_type_name(int new_type); const char *get_branch_type(struct branch_entry *e); -void branch_type_stat_display(FILE *fp, struct branch_type_stat *st); -int branch_type_str(struct branch_type_stat *st, char *bf, int bfsize); +void branch_type_stat_display(FILE *fp, const struct branch_type_stat *st); +int branch_type_str(const struct branch_type_stat *st, char *bf, int bfsize); const char *branch_spec_desc(int spec); diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 18d545c0629e..cde4860e6f28 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -1339,7 +1339,7 @@ static int count_float_printf(int idx, const char *str, float value, static int branch_to_str(char *bf, int bfsize, u64 branch_count, u64 predicted_count, u64 abort_count, - struct branch_type_stat *brtype_stat) + const struct branch_type_stat *brtype_stat) { int printed, i = 0; @@ -1403,7 +1403,7 @@ static int counts_str_build(char *bf, int bfsize, u64 abort_count, u64 cycles_count, u64 iter_count, u64 iter_cycles, u64 from_count, - struct branch_type_stat *brtype_stat) + const struct branch_type_stat *brtype_stat) { int printed; @@ -1430,7 +1430,7 @@ static int callchain_counts_printf(FILE *fp, char *bf, int bfsize, u64 abort_count, u64 cycles_count, u64 iter_count, u64 iter_cycles, u64 from_count, - struct branch_type_stat *brtype_stat) + const struct branch_type_stat *brtype_stat) { char str[256]; -- cgit v1.2.3 From 6ba29fbb0b3863fa50c809d098680a0f3e0bd625 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:12 -0700 Subject: perf callchain: Make brtype_stat in callchain_list optional struct callchain_list is 352bytes in size, 232 of which are brtype_stat. brtype_stat is only used for certain callchain_list items so make it optional, allocating when necessary. So that printing doesn't need to deal with an optional brtype_stat, pass an empty/zero version. Before: ``` struct callchain_list { u64 ip; /* 0 8 */ struct map_symbol ms; /* 8 24 */ struct { _Bool unfolded; /* 32 1 */ _Bool has_children; /* 33 1 */ }; /* 32 2 */ /* XXX 6 bytes hole, try to pack */ u64 branch_count; /* 40 8 */ u64 from_count; /* 48 8 */ u64 predicted_count; /* 56 8 */ /* --- cacheline 1 boundary (64 bytes) --- */ u64 abort_count; /* 64 8 */ u64 cycles_count; /* 72 8 */ u64 iter_count; /* 80 8 */ u64 iter_cycles; /* 88 8 */ struct branch_type_stat brtype_stat; /* 96 232 */ /* --- cacheline 5 boundary (320 bytes) was 8 bytes ago --- */ const char * srcline; /* 328 8 */ struct list_head list; /* 336 16 */ /* size: 352, cachelines: 6, members: 13 */ /* sum members: 346, holes: 1, sum holes: 6 */ /* last cacheline: 32 bytes */ }; ``` After: ``` struct callchain_list { u64 ip; /* 0 8 */ struct map_symbol ms; /* 8 24 */ struct { _Bool unfolded; /* 32 1 */ _Bool has_children; /* 33 1 */ }; /* 32 2 */ /* XXX 6 bytes hole, try to pack */ u64 branch_count; /* 40 8 */ u64 from_count; /* 48 8 */ u64 predicted_count; /* 56 8 */ /* --- cacheline 1 boundary (64 bytes) --- */ u64 abort_count; /* 64 8 */ u64 cycles_count; /* 72 8 */ u64 iter_count; /* 80 8 */ u64 iter_cycles; /* 88 8 */ struct branch_type_stat * brtype_stat; /* 96 8 */ const char * srcline; /* 104 8 */ struct list_head list; /* 112 16 */ /* size: 128, cachelines: 2, members: 13 */ /* sum members: 122, holes: 1, sum holes: 6 */ }; ``` Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-10-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/callchain.c | 41 +++++++++++++++++++++++++++++++++-------- tools/perf/util/callchain.h | 2 +- 2 files changed, 34 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index cde4860e6f28..5349c6a21849 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -586,7 +586,7 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor) call = zalloc(sizeof(*call)); if (!call) { perror("not enough memory for the code path tree"); - return -1; + return -ENOMEM; } call->ip = cursor_node->ip; call->ms = cursor_node->ms; @@ -602,7 +602,15 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor) * branch_from is set with value somewhere else * to imply it's "to" of a branch. */ - call->brtype_stat.branch_to = true; + if (!call->brtype_stat) { + call->brtype_stat = zalloc(sizeof(*call->brtype_stat)); + if (!call->brtype_stat) { + perror("not enough memory for the code path branch statisitcs"); + free(call->brtype_stat); + return -ENOMEM; + } + } + call->brtype_stat->branch_to = true; if (cursor_node->branch_flags.predicted) call->predicted_count = 1; @@ -610,7 +618,7 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor) if (cursor_node->branch_flags.abort) call->abort_count = 1; - branch_type_count(&call->brtype_stat, + branch_type_count(call->brtype_stat, &cursor_node->branch_flags, cursor_node->branch_from, cursor_node->ip); @@ -618,7 +626,8 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor) /* * It's "from" of a branch */ - call->brtype_stat.branch_to = false; + if (call->brtype_stat && call->brtype_stat->branch_to) + call->brtype_stat->branch_to = false; call->cycles_count = cursor_node->branch_flags.cycles; call->iter_count = cursor_node->nr_loop_iter; @@ -652,6 +661,7 @@ add_child(struct callchain_node *parent, list_del_init(&call->list); map__zput(call->ms.map); maps__zput(call->ms.maps); + zfree(&call->brtype_stat); free(call); } free(new); @@ -762,7 +772,14 @@ static enum match_result match_chain(struct callchain_cursor_node *node, /* * It's "to" of a branch */ - cnode->brtype_stat.branch_to = true; + if (!cnode->brtype_stat) { + cnode->brtype_stat = zalloc(sizeof(*cnode->brtype_stat)); + if (!cnode->brtype_stat) { + perror("not enough memory for the code path branch statisitcs"); + return MATCH_ERROR; + } + } + cnode->brtype_stat->branch_to = true; if (node->branch_flags.predicted) cnode->predicted_count++; @@ -770,7 +787,7 @@ static enum match_result match_chain(struct callchain_cursor_node *node, if (node->branch_flags.abort) cnode->abort_count++; - branch_type_count(&cnode->brtype_stat, + branch_type_count(cnode->brtype_stat, &node->branch_flags, node->branch_from, node->ip); @@ -778,7 +795,8 @@ static enum match_result match_chain(struct callchain_cursor_node *node, /* * It's "from" of a branch */ - cnode->brtype_stat.branch_to = false; + if (cnode->brtype_stat && cnode->brtype_stat->branch_to) + cnode->brtype_stat->branch_to = false; cnode->cycles_count += node->branch_flags.cycles; cnode->iter_count += node->nr_loop_iter; cnode->iter_cycles += node->iter_cycles; @@ -1026,6 +1044,7 @@ merge_chain_branch(struct callchain_cursor *cursor, maps__zput(ms.maps); map__zput(list->ms.map); maps__zput(list->ms.maps); + zfree(&list->brtype_stat); free(list); } @@ -1447,11 +1466,14 @@ static int callchain_counts_printf(FILE *fp, char *bf, int bfsize, int callchain_list_counts__printf_value(struct callchain_list *clist, FILE *fp, char *bf, int bfsize) { + static const struct branch_type_stat empty_brtype_stat = {}; + const struct branch_type_stat *brtype_stat; u64 branch_count, predicted_count; u64 abort_count, cycles_count; u64 iter_count, iter_cycles; u64 from_count; + brtype_stat = clist->brtype_stat ?: &empty_brtype_stat; branch_count = clist->branch_count; predicted_count = clist->predicted_count; abort_count = clist->abort_count; @@ -1463,7 +1485,7 @@ int callchain_list_counts__printf_value(struct callchain_list *clist, return callchain_counts_printf(fp, bf, bfsize, branch_count, predicted_count, abort_count, cycles_count, iter_count, iter_cycles, - from_count, &clist->brtype_stat); + from_count, brtype_stat); } static void free_callchain_node(struct callchain_node *node) @@ -1476,6 +1498,7 @@ static void free_callchain_node(struct callchain_node *node) list_del_init(&list->list); map__zput(list->ms.map); maps__zput(list->ms.maps); + zfree(&list->brtype_stat); free(list); } @@ -1483,6 +1506,7 @@ static void free_callchain_node(struct callchain_node *node) list_del_init(&list->list); map__zput(list->ms.map); maps__zput(list->ms.maps); + zfree(&list->brtype_stat); free(list); } @@ -1569,6 +1593,7 @@ out: list_del_init(&chain->list); map__zput(chain->ms.map); maps__zput(chain->ms.maps); + zfree(&chain->brtype_stat); free(chain); } return -ENOMEM; diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index d2618a47deca..86e8a9e81456 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -129,7 +129,7 @@ struct callchain_list { u64 cycles_count; u64 iter_count; u64 iter_cycles; - struct branch_type_stat brtype_stat; + struct branch_type_stat *brtype_stat; const char *srcline; struct list_head list; }; -- cgit v1.2.3 From dec07fe5d4fd29aed2faf17f56140cd402175d72 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:13 -0700 Subject: perf callchain: Minor layout changes to callchain_list Avoid 6 byte hole for padding. Place more frequently used fields first in an attempt to use just 1 cacheline in the common case. Before: ``` struct callchain_list { u64 ip; /* 0 8 */ struct map_symbol ms; /* 8 24 */ struct { _Bool unfolded; /* 32 1 */ _Bool has_children; /* 33 1 */ }; /* 32 2 */ /* XXX 6 bytes hole, try to pack */ u64 branch_count; /* 40 8 */ u64 from_count; /* 48 8 */ u64 predicted_count; /* 56 8 */ /* --- cacheline 1 boundary (64 bytes) --- */ u64 abort_count; /* 64 8 */ u64 cycles_count; /* 72 8 */ u64 iter_count; /* 80 8 */ u64 iter_cycles; /* 88 8 */ struct branch_type_stat * brtype_stat; /* 96 8 */ const char * srcline; /* 104 8 */ struct list_head list; /* 112 16 */ /* size: 128, cachelines: 2, members: 13 */ /* sum members: 122, holes: 1, sum holes: 6 */ }; ``` After: ``` struct callchain_list { struct list_head list; /* 0 16 */ u64 ip; /* 16 8 */ struct map_symbol ms; /* 24 24 */ const char * srcline; /* 48 8 */ u64 branch_count; /* 56 8 */ /* --- cacheline 1 boundary (64 bytes) --- */ u64 from_count; /* 64 8 */ u64 cycles_count; /* 72 8 */ u64 iter_count; /* 80 8 */ u64 iter_cycles; /* 88 8 */ struct branch_type_stat * brtype_stat; /* 96 8 */ u64 predicted_count; /* 104 8 */ u64 abort_count; /* 112 8 */ struct { _Bool unfolded; /* 120 1 */ _Bool has_children; /* 121 1 */ }; /* 120 2 */ /* size: 128, cachelines: 2, members: 13 */ /* padding: 6 */ }; ``` Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-11-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/callchain.h | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 86e8a9e81456..d5c66345ae31 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -116,22 +116,22 @@ extern struct callchain_param callchain_param; extern struct callchain_param callchain_param_default; struct callchain_list { + struct list_head list; u64 ip; struct map_symbol ms; - struct /* for TUI */ { - bool unfolded; - bool has_children; - }; + const char *srcline; u64 branch_count; u64 from_count; - u64 predicted_count; - u64 abort_count; u64 cycles_count; u64 iter_count; u64 iter_cycles; struct branch_type_stat *brtype_stat; - const char *srcline; - struct list_head list; + u64 predicted_count; + u64 abort_count; + struct /* for TUI */ { + bool unfolded; + bool has_children; + }; }; /* -- cgit v1.2.3 From 56e144fe98260a0f8a17326993ceb576ef859ed5 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Tue, 24 Oct 2023 15:23:14 -0700 Subject: perf mem_info: Add and use map_symbol__exit and addr_map_symbol__exit Fix leak where mem_info__put wouldn't release the maps/map as used by perf mem. Add exit functions and use elsewhere that the maps and map are released. Signed-off-by: Ian Rogers Cc: K Prateek Nayak Cc: Ravi Bangoria Cc: Sandipan Das Cc: Anshuman Khandual Cc: German Gomez Cc: James Clark Cc: Nick Terrell Cc: Sean Christopherson Cc: Changbin Du Cc: liuwenyu Cc: Yang Jihong Cc: Masami Hiramatsu Cc: Miguel Ojeda Cc: Song Liu Cc: Leo Yan Cc: Kajol Jain Cc: Andi Kleen Cc: Kan Liang Cc: Athira Rajeev Cc: Yanteng Si Cc: Liam Howlett Cc: Paolo Bonzini Link: https://lore.kernel.org/r/20231024222353.3024098-12-irogers@google.com Signed-off-by: Namhyung Kim --- tools/perf/util/Build | 1 + tools/perf/util/callchain.c | 27 +++++++++------------------ tools/perf/util/hist.c | 28 ++++++++++++---------------- tools/perf/util/machine.c | 6 ++---- tools/perf/util/map_symbol.c | 15 +++++++++++++++ tools/perf/util/map_symbol.h | 4 ++++ tools/perf/util/symbol.c | 5 ++++- 7 files changed, 47 insertions(+), 39 deletions(-) create mode 100644 tools/perf/util/map_symbol.c (limited to 'tools/perf/util') diff --git a/tools/perf/util/Build b/tools/perf/util/Build index 0ea5a9d368d4..96058f949ec9 100644 --- a/tools/perf/util/Build +++ b/tools/perf/util/Build @@ -49,6 +49,7 @@ perf-y += dso.o perf-y += dsos.o perf-y += symbol.o perf-y += symbol_fprintf.o +perf-y += map_symbol.o perf-y += color.o perf-y += color_config.o perf-y += metricgroup.o diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 5349c6a21849..99cce43ba152 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -659,8 +659,7 @@ add_child(struct callchain_node *parent, list_for_each_entry_safe(call, tmp, &new->val, list) { list_del_init(&call->list); - map__zput(call->ms.map); - maps__zput(call->ms.maps); + map_symbol__exit(&call->ms); zfree(&call->brtype_stat); free(call); } @@ -1040,10 +1039,8 @@ merge_chain_branch(struct callchain_cursor *cursor, }; callchain_cursor_append(cursor, list->ip, &ms, false, NULL, 0, 0, 0, list->srcline); list_del_init(&list->list); - map__zput(ms.map); - maps__zput(ms.maps); - map__zput(list->ms.map); - maps__zput(list->ms.maps); + map_symbol__exit(&ms); + map_symbol__exit(&list->ms); zfree(&list->brtype_stat); free(list); } @@ -1096,8 +1093,7 @@ int callchain_cursor_append(struct callchain_cursor *cursor, } node->ip = ip; - maps__zput(node->ms.maps); - map__zput(node->ms.map); + map_symbol__exit(&node->ms); node->ms = *ms; node->ms.maps = maps__get(ms->maps); node->ms.map = map__get(ms->map); @@ -1496,16 +1492,14 @@ static void free_callchain_node(struct callchain_node *node) list_for_each_entry_safe(list, tmp, &node->parent_val, list) { list_del_init(&list->list); - map__zput(list->ms.map); - maps__zput(list->ms.maps); + map_symbol__exit(&list->ms); zfree(&list->brtype_stat); free(list); } list_for_each_entry_safe(list, tmp, &node->val, list) { list_del_init(&list->list); - map__zput(list->ms.map); - maps__zput(list->ms.maps); + map_symbol__exit(&list->ms); zfree(&list->brtype_stat); free(list); } @@ -1591,8 +1585,7 @@ int callchain_node__make_parent_list(struct callchain_node *node) out: list_for_each_entry_safe(chain, new, &head, list) { list_del_init(&chain->list); - map__zput(chain->ms.map); - maps__zput(chain->ms.maps); + map_symbol__exit(&chain->ms); zfree(&chain->brtype_stat); free(chain); } @@ -1676,10 +1669,8 @@ void callchain_cursor_reset(struct callchain_cursor *cursor) cursor->nr = 0; cursor->last = &cursor->first; - for (node = cursor->first; node != NULL; node = node->next) { - map__zput(node->ms.map); - maps__zput(node->ms.maps); - } + for (node = cursor->first; node != NULL; node = node->next) + map_symbol__exit(&node->ms); } void callchain_param_setup(u64 sample_type, const char *arch) diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 0aa7e231172a..0888b7163b7c 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -515,17 +515,16 @@ err_rawdata: err_infos: if (he->branch_info) { - map__put(he->branch_info->from.ms.map); - map__put(he->branch_info->to.ms.map); + map_symbol__exit(&he->branch_info->from.ms); + map_symbol__exit(&he->branch_info->to.ms); zfree(&he->branch_info); } if (he->mem_info) { - map__put(he->mem_info->iaddr.ms.map); - map__put(he->mem_info->daddr.ms.map); + map_symbol__exit(&he->mem_info->iaddr.ms); + map_symbol__exit(&he->mem_info->daddr.ms); } err: - maps__zput(he->ms.maps); - map__zput(he->ms.map); + map_symbol__exit(&he->ms); zfree(&he->stat_acc); return -ENOMEM; } @@ -1317,20 +1316,19 @@ void hist_entry__delete(struct hist_entry *he) struct hist_entry_ops *ops = he->ops; thread__zput(he->thread); - maps__zput(he->ms.maps); - map__zput(he->ms.map); + map_symbol__exit(&he->ms); if (he->branch_info) { - map__zput(he->branch_info->from.ms.map); - map__zput(he->branch_info->to.ms.map); + map_symbol__exit(&he->branch_info->from.ms); + map_symbol__exit(&he->branch_info->to.ms); zfree_srcline(&he->branch_info->srcline_from); zfree_srcline(&he->branch_info->srcline_to); zfree(&he->branch_info); } if (he->mem_info) { - map__zput(he->mem_info->iaddr.ms.map); - map__zput(he->mem_info->daddr.ms.map); + map_symbol__exit(&he->mem_info->iaddr.ms); + map_symbol__exit(&he->mem_info->daddr.ms); mem_info__zput(he->mem_info); } @@ -2700,10 +2698,8 @@ void hist__account_cycles(struct branch_stack *bs, struct addr_location *al, *total_cycles += bi[i].flags.cycles; } for (unsigned int i = 0; i < bs->nr; i++) { - map__put(bi[i].to.ms.map); - maps__put(bi[i].to.ms.maps); - map__put(bi[i].from.ms.map); - maps__put(bi[i].from.ms.maps); + map_symbol__exit(&bi[i].to.ms); + map_symbol__exit(&bi[i].from.ms); } free(bi); } diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 31fdbb0f27af..90c750150b19 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2389,8 +2389,7 @@ static int add_callchain_ip(struct thread *thread, iter_cycles, branch_from, srcline); out: addr_location__exit(&al); - maps__put(ms.maps); - map__put(ms.map); + map_symbol__exit(&ms); return err; } @@ -3116,8 +3115,7 @@ static int append_inlines(struct callchain_cursor *cursor, struct map_symbol *ms if (ret != 0) return ret; } - map__put(ilist_ms.map); - maps__put(ilist_ms.maps); + map_symbol__exit(&ilist_ms); return ret; } diff --git a/tools/perf/util/map_symbol.c b/tools/perf/util/map_symbol.c new file mode 100644 index 000000000000..bef5079f2403 --- /dev/null +++ b/tools/perf/util/map_symbol.c @@ -0,0 +1,15 @@ +// SPDX-License-Identifier: GPL-2.0 +#include "map_symbol.h" +#include "maps.h" +#include "map.h" + +void map_symbol__exit(struct map_symbol *ms) +{ + maps__zput(ms->maps); + map__zput(ms->map); +} + +void addr_map_symbol__exit(struct addr_map_symbol *ams) +{ + map_symbol__exit(&ams->ms); +} diff --git a/tools/perf/util/map_symbol.h b/tools/perf/util/map_symbol.h index e08817b0c30f..72d5ed938ed6 100644 --- a/tools/perf/util/map_symbol.h +++ b/tools/perf/util/map_symbol.h @@ -22,4 +22,8 @@ struct addr_map_symbol { u64 phys_addr; u64 data_page_size; }; + +void map_symbol__exit(struct map_symbol *ms); +void addr_map_symbol__exit(struct addr_map_symbol *ams); + #endif // __PERF_MAP_SYMBOL diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 822f4dcebfe6..82cc74b9358e 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -2791,8 +2791,11 @@ struct mem_info *mem_info__get(struct mem_info *mi) void mem_info__put(struct mem_info *mi) { - if (mi && refcount_dec_and_test(&mi->refcnt)) + if (mi && refcount_dec_and_test(&mi->refcnt)) { + addr_map_symbol__exit(&mi->iaddr); + addr_map_symbol__exit(&mi->daddr); free(mi); + } } struct mem_info *mem_info__new(void) -- cgit v1.2.3 From 93c65d61433bcec4070fc61b7c737c79fc25f09a Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 27 Oct 2023 10:33:30 -0300 Subject: perf python: Fix binding linkage due to rename and move of evsel__increase_rlimit() The changes in ("perf evsel: Rename evsel__increase_rlimit to rlimit__increase_nofile") ended up breaking the python binding that now references the rlimit__increase_nofile function, add the util/rlimit.o to the tools/perf/util/python-ext-sources to cure that. This was detected by the 'perf test python' regression test: $ perf test python 14: 'import perf' in python : FAILED! $ perf test -v python Couldn't bump rlimit(MEMLOCK), failures may take place when creating BPF maps, etc 14: 'import perf' in python : --- start --- test child forked, pid 2912462 python usage test: "echo "import sys ; sys.path.insert(0, '/tmp/build/perf-tools-next/python'); import perf" | '/usr/bin/python3' " Traceback (most recent call last): File "", line 1, in ImportError: /tmp/build/perf-tools-next/python/perf.cpython-311-x86_64-linux-gnu.so: undefined symbol: rlimit__increase_nofile test child finished with -1 ---- end ---- 'import perf' in python: FAILED! $ Fixes: e093a222d7cba1eb ("perf evsel: Rename evsel__increase_rlimit to rlimit__increase_nofile") Acked-by: Namhyung Kim Acked-by: Yang Jihong Link: https://lore.kernel.org/lkml/ZTrCS5Z3PZAmfPdV@kernel.org Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Namhyung Kim --- tools/perf/util/python-ext-sources | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/python-ext-sources b/tools/perf/util/python-ext-sources index 26e1c8d973ea..593b660ec75e 100644 --- a/tools/perf/util/python-ext-sources +++ b/tools/perf/util/python-ext-sources @@ -40,6 +40,7 @@ util/rwsem.c util/hashmap.c util/perf_regs.c util/fncache.c +util/rlimit.c util/perf-regs-arch/perf_regs_aarch64.c util/perf-regs-arch/perf_regs_arm.c util/perf-regs-arch/perf_regs_csky.c -- cgit v1.2.3 From ee40490dd7cdcda38ece6d081f63ecddd3fdbe25 Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Fri, 27 Oct 2023 09:46:33 +0100 Subject: perf callchain: Fix spelling mistake "statisitcs" -> "statistics" There are a couple of spelling mistakes in perror messages. Fix them. Signed-off-by: Colin Ian King Cc: kernel-janitors@vger.kernel.org Link: https://lore.kernel.org/r/20231027084633.1167530-1-colin.i.king@gmail.com Signed-off-by: Namhyung Kim --- tools/perf/util/callchain.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 99cce43ba152..8262f69118db 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -605,7 +605,7 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor) if (!call->brtype_stat) { call->brtype_stat = zalloc(sizeof(*call->brtype_stat)); if (!call->brtype_stat) { - perror("not enough memory for the code path branch statisitcs"); + perror("not enough memory for the code path branch statistics"); free(call->brtype_stat); return -ENOMEM; } @@ -774,7 +774,7 @@ static enum match_result match_chain(struct callchain_cursor_node *node, if (!cnode->brtype_stat) { cnode->brtype_stat = zalloc(sizeof(*cnode->brtype_stat)); if (!cnode->brtype_stat) { - perror("not enough memory for the code path branch statisitcs"); + perror("not enough memory for the code path branch statistics"); return MATCH_ERROR; } } -- cgit v1.2.3