From 1ab1fa5dfb429c533fbc791e524788cf0cc43775 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 26 Dec 2013 15:11:52 +0900 Subject: perf hists: Add support for showing relative percentage When filtering by thread, dso or symbol on TUI it also update total period so that the output shows different result than no filter - the percentage changed to relative to filtered entries only. Sometimes this is not desired since users might expect same results with filter. So new filtered_* fields to hists->stats to count them separately. They'll be controlled/used by user later. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1397145720-8063-2-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 17 ++++++++++++++--- tools/perf/util/hist.h | 3 +++ 2 files changed, 17 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index f38590d7561b..1ed3e2b86f0b 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -674,8 +674,8 @@ void hists__output_resort(struct hists *hists) next = rb_first(root); hists->entries = RB_ROOT; - hists->nr_entries = 0; - hists->stats.total_period = 0; + hists->nr_entries = hists->nr_non_filtered_entries = 0; + hists->stats.total_period = hists->stats.total_non_filtered_period = 0; hists__reset_col_len(hists); while (next) { @@ -695,11 +695,16 @@ static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h return; ++hists->nr_entries; - if (h->ms.unfolded) + ++hists->nr_non_filtered_entries; + if (h->ms.unfolded) { hists->nr_entries += h->nr_rows; + hists->nr_non_filtered_entries += h->nr_rows; + } h->row_offset = 0; hists->stats.total_period += h->stat.period; + hists->stats.total_non_filtered_period += h->stat.period; hists->stats.nr_events[PERF_RECORD_SAMPLE] += h->stat.nr_events; + hists->stats.nr_non_filtered_samples += h->stat.nr_events; hists__calc_col_len(hists, h); } @@ -722,7 +727,9 @@ void hists__filter_by_dso(struct hists *hists) struct rb_node *nd; hists->nr_entries = hists->stats.total_period = 0; + hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { @@ -755,7 +762,9 @@ void hists__filter_by_thread(struct hists *hists) struct rb_node *nd; hists->nr_entries = hists->stats.total_period = 0; + hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { @@ -786,7 +795,9 @@ void hists__filter_by_symbol(struct hists *hists) struct rb_node *nd; hists->nr_entries = hists->stats.total_period = 0; + hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 1f1f513dfe7f..213551469f36 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -37,9 +37,11 @@ enum hist_filter { */ struct events_stats { u64 total_period; + u64 total_non_filtered_period; u64 total_lost; u64 total_invalid_chains; u32 nr_events[PERF_RECORD_HEADER_MAX]; + u32 nr_non_filtered_samples; u32 nr_lost_warned; u32 nr_unknown_events; u32 nr_invalid_chains; @@ -83,6 +85,7 @@ struct hists { struct rb_root entries; struct rb_root entries_collapsed; u64 nr_entries; + u64 nr_non_filtered_entries; const struct thread *thread_filter; const struct dso *dso_filter; const char *uid_filter_str; -- cgit v1.2.3 From f2148330544a697481219b5bc34261f6dd049bfb Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Jan 2014 11:52:48 +0900 Subject: perf report: Add --percentage option The --percentage option is for controlling overhead percentage displayed. It can only receive either of "relative" or "absolute". "relative" means it's relative to filtered entries only so that the sum of shown entries will be always 100%. "absolute" means it retains the original value before and after the filter is applied. $ perf report -s comm # Overhead Command # ........ ............ # 74.19% cc1 7.61% gcc 6.11% as 4.35% sh 4.14% make 1.13% fixdep ... $ perf report -s comm -c cc1,gcc --percentage absolute # Overhead Command # ........ ............ # 74.19% cc1 7.61% gcc $ perf report -s comm -c cc1,gcc --percentage relative # Overhead Command # ........ ............ # 90.69% cc1 9.31% gcc Note that it has zero effect if no filter was applied. Suggested-by: Arnaldo Carvalho de Melo Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1397145720-8063-3-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 39 ++++++++++++++++++++------------------- tools/perf/util/hist.h | 1 + tools/perf/util/symbol.c | 1 + tools/perf/util/symbol.h | 3 ++- 4 files changed, 24 insertions(+), 20 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 1ed3e2b86f0b..3ebd89a28257 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -321,9 +321,11 @@ void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h) { if (!h->filtered) { hists__calc_col_len(hists, h); - ++hists->nr_entries; - hists->stats.total_period += h->stat.period; + hists->nr_non_filtered_entries++; + hists->stats.total_non_filtered_period += h->stat.period; } + hists->nr_entries++; + hists->stats.total_period += h->stat.period; } static u8 symbol__parent_filter(const struct symbol *parent) @@ -674,8 +676,9 @@ void hists__output_resort(struct hists *hists) next = rb_first(root); hists->entries = RB_ROOT; - hists->nr_entries = hists->nr_non_filtered_entries = 0; - hists->stats.total_period = hists->stats.total_non_filtered_period = 0; + hists->nr_non_filtered_entries = 0; + hists->stats.total_period = 0; + hists->stats.total_non_filtered_period = 0; hists__reset_col_len(hists); while (next) { @@ -694,16 +697,11 @@ static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h if (h->filtered) return; - ++hists->nr_entries; ++hists->nr_non_filtered_entries; - if (h->ms.unfolded) { - hists->nr_entries += h->nr_rows; + if (h->ms.unfolded) hists->nr_non_filtered_entries += h->nr_rows; - } h->row_offset = 0; - hists->stats.total_period += h->stat.period; hists->stats.total_non_filtered_period += h->stat.period; - hists->stats.nr_events[PERF_RECORD_SAMPLE] += h->stat.nr_events; hists->stats.nr_non_filtered_samples += h->stat.nr_events; hists__calc_col_len(hists, h); @@ -726,9 +724,8 @@ void hists__filter_by_dso(struct hists *hists) { struct rb_node *nd; - hists->nr_entries = hists->stats.total_period = 0; - hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; - hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->nr_non_filtered_entries = 0; + hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); @@ -761,9 +758,8 @@ void hists__filter_by_thread(struct hists *hists) { struct rb_node *nd; - hists->nr_entries = hists->stats.total_period = 0; - hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; - hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->nr_non_filtered_entries = 0; + hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); @@ -794,9 +790,8 @@ void hists__filter_by_symbol(struct hists *hists) { struct rb_node *nd; - hists->nr_entries = hists->stats.total_period = 0; - hists->nr_non_filtered_entries = hists->stats.total_non_filtered_period = 0; - hists->stats.nr_events[PERF_RECORD_SAMPLE] = 0; + hists->nr_non_filtered_entries = 0; + hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; hists__reset_col_len(hists); @@ -942,3 +937,9 @@ int hists__link(struct hists *leader, struct hists *other) return 0; } + +u64 hists__total_period(struct hists *hists) +{ + return symbol_conf.filter_relative ? hists->stats.total_non_filtered_period : + hists->stats.total_period; +} diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 213551469f36..3191496bd3b7 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -115,6 +115,7 @@ void hists__collapse_resort(struct hists *hists, struct ui_progress *prog); void hists__decay_entries(struct hists *hists, bool zap_user, bool zap_kernel); void hists__output_recalc_col_len(struct hists *hists, int max_rows); +u64 hists__total_period(struct hists *hists); void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h); void hists__inc_nr_events(struct hists *hists, u32 type); void events_stats__inc(struct events_stats *stats, u32 type); diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 95e249779931..b2eca6c17a70 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -33,6 +33,7 @@ struct symbol_conf symbol_conf = { .try_vmlinux_path = true, .annotate_src = true, .demangle = true, + .filter_relative = true, .symfs = "", }; diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h index 501e4e722e8e..ae94e006a52d 100644 --- a/tools/perf/util/symbol.h +++ b/tools/perf/util/symbol.h @@ -115,7 +115,8 @@ struct symbol_conf { annotate_asm_raw, annotate_src, event_group, - demangle; + demangle, + filter_relative; const char *vmlinux_name, *kallsyms_name, *source_prefix, -- cgit v1.2.3 From 33db4568e1f41efe6d0e4695483f968fc1135bf3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 7 Feb 2014 12:06:07 +0900 Subject: perf top: Add --percentage option The --percentage option is for controlling overhead percentage displayed. It can only receive either of "relative" or "absolute". Move the parser callback function into a common location since it's used by multiple commands now. For more information, please see previous commit same thing done to "perf report". Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1397145720-8063-4-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 13 +++++++++++++ tools/perf/util/hist.h | 5 +++++ 2 files changed, 18 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 3ebd89a28257..3c2dd233b98e 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -943,3 +943,16 @@ u64 hists__total_period(struct hists *hists) return symbol_conf.filter_relative ? hists->stats.total_non_filtered_period : hists->stats.total_period; } + +int parse_filter_percentage(const struct option *opt __maybe_unused, + const char *arg, int unset __maybe_unused) +{ + if (!strcmp(arg, "relative")) + symbol_conf.filter_relative = true; + else if (!strcmp(arg, "absolute")) + symbol_conf.filter_relative = false; + else + return -1; + + return 0; +} diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 3191496bd3b7..a4ec336ae3fe 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -254,4 +254,9 @@ static inline int script_browse(const char *script_opt __maybe_unused) #endif unsigned int hists__sort_list_width(struct hists *hists); + +struct option; +int parse_filter_percentage(const struct option *opt __maybe_unused, + const char *arg, int unset __maybe_unused); + #endif /* __PERF_HIST_H */ -- cgit v1.2.3 From 0b93da1756df4fe930ee0220a6addce263a6e0ab Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Jan 2014 12:02:15 +0900 Subject: perf tools: Add hist.percentage config option Add hist.percentage option for setting default value of the symbol_conf.filter_relative. It affects the output of various perf commands (like perf report, top and diff) only if filter(s) applied. An user can write .perfconfig file like below to show absolute percentage of filtered entries by default: $ cat ~/.perfconfig [hist] percentage = absolute And it can be changed through command line: $ perf report --percentage relative Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1397145720-8063-6-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/config.c | 4 ++++ tools/perf/util/hist.c | 8 ++++++++ tools/perf/util/hist.h | 1 + 3 files changed, 13 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/config.c b/tools/perf/util/config.c index 3e0fdd369ccb..24519e14ac56 100644 --- a/tools/perf/util/config.c +++ b/tools/perf/util/config.c @@ -11,6 +11,7 @@ #include "util.h" #include "cache.h" #include "exec_cmd.h" +#include "util/hist.h" /* perf_hist_config */ #define MAXNAME (256) @@ -355,6 +356,9 @@ int perf_default_config(const char *var, const char *value, if (!prefixcmp(var, "core.")) return perf_default_core_config(var, value); + if (!prefixcmp(var, "hist.")) + return perf_hist_config(var, value); + /* Add other config variables here. */ return 0; } diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 3c2dd233b98e..5a892477aa50 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -956,3 +956,11 @@ int parse_filter_percentage(const struct option *opt __maybe_unused, return 0; } + +int perf_hist_config(const char *var, const char *value) +{ + if (!strcmp(var, "hist.percentage")) + return parse_filter_percentage(NULL, value, 0); + + return 0; +} diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index a4ec336ae3fe..5a0343eb22e2 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -258,5 +258,6 @@ unsigned int hists__sort_list_width(struct hists *hists); struct option; int parse_filter_percentage(const struct option *opt __maybe_unused, const char *arg, int unset __maybe_unused); +int perf_hist_config(const char *var, const char *value); #endif /* __PERF_HIST_H */ -- cgit v1.2.3 From 95ce0ba17d5a0a04bbad61720512381d8165d157 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 14 Jan 2014 12:05:27 +0900 Subject: perf tools: Show absolute percentage by default Now perf report will show absolute percentage on filter entries by default. Suggested-by: Jiri Olsa Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1397145720-8063-8-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/symbol.c | 1 - 1 file changed, 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index b2eca6c17a70..95e249779931 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -33,7 +33,6 @@ struct symbol_conf symbol_conf = { .try_vmlinux_path = true, .annotate_src = true, .demangle = true, - .filter_relative = true, .symfs = "", }; -- cgit v1.2.3 From 7c2f8164e5415ef48954f6929f4acae5764afdb4 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 16 Apr 2014 20:49:02 +0200 Subject: perf tools: Fix pmu object compilation error MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit After applying some patches got another shadowing error: CC util/pmu.o util/pmu.c: In function ‘pmu_alias_terms’: util/pmu.c:287:35: error: declaration of ‘clone’ shadows a global declaration [-Werror=shadow] Renaming clone to cloned. Acked-by: David Ahern Cc: David Ahern Cc: Namhyung Kim Link: http://lkml.kernel.org/r/1397674818-27054-1-git-send-email-jolsa@redhat.com Signed-off-by: Jiri Olsa --- tools/perf/util/pmu.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c index 00a7dcb2f55c..7a811eb61f75 100644 --- a/tools/perf/util/pmu.c +++ b/tools/perf/util/pmu.c @@ -284,17 +284,17 @@ static int pmu_aliases(const char *name, struct list_head *head) static int pmu_alias_terms(struct perf_pmu_alias *alias, struct list_head *terms) { - struct parse_events_term *term, *clone; + struct parse_events_term *term, *cloned; LIST_HEAD(list); int ret; list_for_each_entry(term, &alias->terms, list) { - ret = parse_events_term__clone(&clone, term); + ret = parse_events_term__clone(&cloned, term); if (ret) { parse_events__free_terms(&list); return ret; } - list_add_tail(&clone->list, &list); + list_add_tail(&cloned->list, &list); } list_splice(&list, terms); return 0; -- cgit v1.2.3 From 7780c25bae59fd042281710f5e1243268e2c18dc Mon Sep 17 00:00:00 2001 From: Don Zickus Date: Mon, 7 Apr 2014 14:55:21 -0400 Subject: perf tools: Allow ability to map cpus to nodes easily This patch figures out the max number of cpus and nodes that are on the system and creates a map of cpu to node. This allows us to provide a cpu and quickly get the node associated with it. It was mostly copied from builtin-kmem.c and tweaked slightly to use less memory (use possible cpus instead of max). It also calculates the max number of nodes. Signed-off-by: Don Zickus Reviewed-by: Namhyung Kim Link: http://lkml.kernel.org/r/1396896924-129847-2-git-send-email-dzickus@redhat.com [ Removing out label code in init_cpunode_map ] [ Adding check for snprintf error ] [ Removing unneeded returns ] Signed-off-by: Jiri Olsa --- tools/perf/util/cpumap.c | 160 +++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/cpumap.h | 35 +++++++++++ 2 files changed, 195 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c index 7fe4994eeb63..526da022b3c4 100644 --- a/tools/perf/util/cpumap.c +++ b/tools/perf/util/cpumap.c @@ -317,3 +317,163 @@ int cpu_map__build_core_map(struct cpu_map *cpus, struct cpu_map **corep) { return cpu_map__build_map(cpus, corep, cpu_map__get_core); } + +/* setup simple routines to easily access node numbers given a cpu number */ +static int get_max_num(char *path, int *max) +{ + size_t num; + char *buf; + int err = 0; + + if (filename__read_str(path, &buf, &num)) + return -1; + + buf[num] = '\0'; + + /* start on the right, to find highest node num */ + while (--num) { + if ((buf[num] == ',') || (buf[num] == '-')) { + num++; + break; + } + } + if (sscanf(&buf[num], "%d", max) < 1) { + err = -1; + goto out; + } + + /* convert from 0-based to 1-based */ + (*max)++; + +out: + free(buf); + return err; +} + +/* Determine highest possible cpu in the system for sparse allocation */ +static void set_max_cpu_num(void) +{ + const char *mnt; + char path[PATH_MAX]; + int ret = -1; + + /* set up default */ + max_cpu_num = 4096; + + mnt = sysfs__mountpoint(); + if (!mnt) + goto out; + + /* get the highest possible cpu number for a sparse allocation */ + ret = snprintf(path, PATH_MAX, "%s/devices/system/cpu/kernel_max", mnt); + if (ret == PATH_MAX) { + pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); + goto out; + } + + ret = get_max_num(path, &max_cpu_num); + +out: + if (ret) + pr_err("Failed to read max cpus, using default of %d\n", max_cpu_num); +} + +/* Determine highest possible node in the system for sparse allocation */ +static void set_max_node_num(void) +{ + const char *mnt; + char path[PATH_MAX]; + int ret = -1; + + /* set up default */ + max_node_num = 8; + + mnt = sysfs__mountpoint(); + if (!mnt) + goto out; + + /* get the highest possible cpu number for a sparse allocation */ + ret = snprintf(path, PATH_MAX, "%s/devices/system/node/possible", mnt); + if (ret == PATH_MAX) { + pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); + goto out; + } + + ret = get_max_num(path, &max_node_num); + +out: + if (ret) + pr_err("Failed to read max nodes, using default of %d\n", max_node_num); +} + +static int init_cpunode_map(void) +{ + int i; + + set_max_cpu_num(); + set_max_node_num(); + + cpunode_map = calloc(max_cpu_num, sizeof(int)); + if (!cpunode_map) { + pr_err("%s: calloc failed\n", __func__); + return -1; + } + + for (i = 0; i < max_cpu_num; i++) + cpunode_map[i] = -1; + + return 0; +} + +int cpu__setup_cpunode_map(void) +{ + struct dirent *dent1, *dent2; + DIR *dir1, *dir2; + unsigned int cpu, mem; + char buf[PATH_MAX]; + char path[PATH_MAX]; + const char *mnt; + int n; + + /* initialize globals */ + if (init_cpunode_map()) + return -1; + + mnt = sysfs__mountpoint(); + if (!mnt) + return 0; + + n = snprintf(path, PATH_MAX, "%s/devices/system/node", mnt); + if (n == PATH_MAX) { + pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); + return -1; + } + + dir1 = opendir(path); + if (!dir1) + return 0; + + /* walk tree and setup map */ + while ((dent1 = readdir(dir1)) != NULL) { + if (dent1->d_type != DT_DIR || sscanf(dent1->d_name, "node%u", &mem) < 1) + continue; + + n = snprintf(buf, PATH_MAX, "%s/%s", path, dent1->d_name); + if (n == PATH_MAX) { + pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); + continue; + } + + dir2 = opendir(buf); + if (!dir2) + continue; + while ((dent2 = readdir(dir2)) != NULL) { + if (dent2->d_type != DT_LNK || sscanf(dent2->d_name, "cpu%u", &cpu) < 1) + continue; + cpunode_map[cpu] = mem; + } + closedir(dir2); + } + closedir(dir1); + return 0; +} diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h index b123bb9d6f55..61a654849002 100644 --- a/tools/perf/util/cpumap.h +++ b/tools/perf/util/cpumap.h @@ -4,6 +4,9 @@ #include #include +#include "perf.h" +#include "util/debug.h" + struct cpu_map { int nr; int map[]; @@ -46,4 +49,36 @@ static inline bool cpu_map__empty(const struct cpu_map *map) return map ? map->map[0] == -1 : true; } +int max_cpu_num; +int max_node_num; +int *cpunode_map; + +int cpu__setup_cpunode_map(void); + +static inline int cpu__max_node(void) +{ + if (unlikely(!max_node_num)) + pr_debug("cpu_map not initialized\n"); + + return max_node_num; +} + +static inline int cpu__max_cpu(void) +{ + if (unlikely(!max_cpu_num)) + pr_debug("cpu_map not initialized\n"); + + return max_cpu_num; +} + +static inline int cpu__get_node(int cpu) +{ + if (unlikely(cpunode_map == NULL)) { + pr_debug("cpu_map not initialized\n"); + return -1; + } + + return cpunode_map[cpu]; +} + #endif /* __PERF_CPUMAP_H */ -- cgit v1.2.3 From f5b1f4e483d55a88ed120e2c62b45ba5b939fa72 Mon Sep 17 00:00:00 2001 From: Don Zickus Date: Mon, 7 Apr 2014 14:55:22 -0400 Subject: perf tools: Use cpu/possible instead of cpu/kernel_max The system's max configuration is represented by cpu/possible and cpu/kernel_max can be huge (4096 vs. 128), so save space by keeping smaller structures. Signed-off-by: Don Zickus Reviewed-by: Namhyung Kim Link: http://lkml.kernel.org/r/1396896924-129847-3-git-send-email-dzickus@redhat.com Signed-off-by: Jiri Olsa --- tools/perf/util/cpumap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c index 526da022b3c4..c4e55b71010c 100644 --- a/tools/perf/util/cpumap.c +++ b/tools/perf/util/cpumap.c @@ -365,7 +365,7 @@ static void set_max_cpu_num(void) goto out; /* get the highest possible cpu number for a sparse allocation */ - ret = snprintf(path, PATH_MAX, "%s/devices/system/cpu/kernel_max", mnt); + ret = snprintf(path, PATH_MAX, "%s/devices/system/cpu/possible", mnt); if (ret == PATH_MAX) { pr_err("sysfs path crossed PATH_MAX(%d) size\n", PATH_MAX); goto out; -- cgit v1.2.3 From cff6bb46d477383092f46682a0d12e323e4b84d2 Mon Sep 17 00:00:00 2001 From: Don Zickus Date: Mon, 7 Apr 2014 14:55:24 -0400 Subject: perf callchain: Add generic report parse callchain callback function This takes the parse_callchain_opt function and copies it into the callchain.c file. Now the c2c tool can use it too without duplicating. Update perf-report to use the new routine too. Signed-off-by: Don Zickus Reviewed-by: Namhyung Kim Link: http://lkml.kernel.org/r/1396896924-129847-5-git-send-email-dzickus@redhat.com [ Adding missing braces to multiline if condition ] Signed-off-by: Jiri Olsa --- tools/perf/util/callchain.c | 78 +++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/callchain.h | 1 + 2 files changed, 79 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index 8d9db454f1a9..9a42382b3921 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -25,6 +25,84 @@ __thread struct callchain_cursor callchain_cursor; +int +parse_callchain_report_opt(const char *arg) +{ + char *tok, *tok2; + char *endptr; + + symbol_conf.use_callchain = true; + + if (!arg) + return 0; + + tok = strtok((char *)arg, ","); + if (!tok) + return -1; + + /* get the output mode */ + if (!strncmp(tok, "graph", strlen(arg))) { + callchain_param.mode = CHAIN_GRAPH_ABS; + + } else if (!strncmp(tok, "flat", strlen(arg))) { + callchain_param.mode = CHAIN_FLAT; + } else if (!strncmp(tok, "fractal", strlen(arg))) { + callchain_param.mode = CHAIN_GRAPH_REL; + } else if (!strncmp(tok, "none", strlen(arg))) { + callchain_param.mode = CHAIN_NONE; + symbol_conf.use_callchain = false; + return 0; + } else { + return -1; + } + + /* get the min percentage */ + tok = strtok(NULL, ","); + if (!tok) + goto setup; + + callchain_param.min_percent = strtod(tok, &endptr); + if (tok == endptr) + return -1; + + /* get the print limit */ + tok2 = strtok(NULL, ","); + if (!tok2) + goto setup; + + if (tok2[0] != 'c') { + callchain_param.print_limit = strtoul(tok2, &endptr, 0); + tok2 = strtok(NULL, ","); + if (!tok2) + goto setup; + } + + /* get the call chain order */ + if (!strncmp(tok2, "caller", strlen("caller"))) + callchain_param.order = ORDER_CALLER; + else if (!strncmp(tok2, "callee", strlen("callee"))) + callchain_param.order = ORDER_CALLEE; + else + return -1; + + /* Get the sort key */ + tok2 = strtok(NULL, ","); + if (!tok2) + goto setup; + if (!strncmp(tok2, "function", strlen("function"))) + callchain_param.key = CCKEY_FUNCTION; + else if (!strncmp(tok2, "address", strlen("address"))) + callchain_param.key = CCKEY_ADDRESS; + else + return -1; +setup: + if (callchain_register_param(&callchain_param) < 0) { + pr_err("Can't register callchain params\n"); + return -1; + } + return 0; +} + static void rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, enum chain_mode mode) diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 8ad97e9b119f..dda4cf8b534c 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -157,4 +157,5 @@ int sample__resolve_callchain(struct perf_sample *sample, struct symbol **parent int hist_entry__append_callchain(struct hist_entry *he, struct perf_sample *sample); extern const char record_callchain_help[]; +int parse_callchain_report_opt(const char *arg); #endif /* __PERF_CALLCHAIN_H */ -- cgit v1.2.3 From 6263835a1b1ad137f3c26a1383c0487a9388d06e Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 24 Apr 2014 16:21:46 +0900 Subject: perf hists: Rename hists__inc_stats() The existing hists__inc_nr_entries() is a misnomer as it's not only increasing ->nr_entries but also other stats. So rename it to more general hists__inc_stats(). Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-3-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 6 +++--- tools/perf/util/hist.h | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 5a892477aa50..12d6c1bd761d 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -317,7 +317,7 @@ static struct hist_entry *hist_entry__new(struct hist_entry *template) return he; } -void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h) +void hists__inc_stats(struct hists *hists, struct hist_entry *h) { if (!h->filtered) { hists__calc_col_len(hists, h); @@ -686,7 +686,7 @@ void hists__output_resort(struct hists *hists) next = rb_next(&n->rb_node_in); __hists__insert_output_entry(&hists->entries, n, min_callchain_hits); - hists__inc_nr_entries(hists, n); + hists__inc_stats(hists, n); } } @@ -853,7 +853,7 @@ static struct hist_entry *hists__add_dummy_entry(struct hists *hists, he->hists = hists; rb_link_node(&he->rb_node_in, parent, p); rb_insert_color(&he->rb_node_in, root); - hists__inc_nr_entries(hists, he); + hists__inc_stats(hists, he); he->dummy = true; } out: diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 5a0343eb22e2..51478c94d976 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -116,7 +116,7 @@ void hists__decay_entries(struct hists *hists, bool zap_user, bool zap_kernel); void hists__output_recalc_col_len(struct hists *hists, int max_rows); u64 hists__total_period(struct hists *hists); -void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h); +void hists__inc_stats(struct hists *hists, struct hist_entry *h); void hists__inc_nr_events(struct hists *hists, u32 type); void events_stats__inc(struct events_stats *stats, u32 type); size_t events_stats__fprintf(struct events_stats *stats, FILE *fp); -- cgit v1.2.3 From ae993efc9c6bd109b027d2799a442892067e9230 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 24 Apr 2014 16:25:19 +0900 Subject: perf hists: Move column length calculation out of hists__inc_stats() It's not the part of logic of hists__inc_stats() so it'd be better to move it out of the function. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-4-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 12d6c1bd761d..f5b388e50265 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -320,7 +320,6 @@ static struct hist_entry *hist_entry__new(struct hist_entry *template) void hists__inc_stats(struct hists *hists, struct hist_entry *h) { if (!h->filtered) { - hists__calc_col_len(hists, h); hists->nr_non_filtered_entries++; hists->stats.total_non_filtered_period += h->stat.period; } @@ -687,6 +686,9 @@ void hists__output_resort(struct hists *hists) __hists__insert_output_entry(&hists->entries, n, min_callchain_hits); hists__inc_stats(hists, n); + + if (!n->filtered) + hists__calc_col_len(hists, n); } } -- cgit v1.2.3 From 9283ba9bd77a6940ecad8721429131d773c704b8 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 24 Apr 2014 16:37:26 +0900 Subject: perf hists: Add a couple of hists stat helper functions Add hists__{reset,inc}_[filter_]stats() functions to cleanup accesses to hist stats (for output). Note that number of samples in the stat is not handled here since it belongs to the input stage. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-5-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 59 ++++++++++++++++++++++++++++++++------------------ tools/perf/util/hist.h | 1 + 2 files changed, 39 insertions(+), 21 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index f5b388e50265..b675857883a2 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -317,16 +317,6 @@ static struct hist_entry *hist_entry__new(struct hist_entry *template) return he; } -void hists__inc_stats(struct hists *hists, struct hist_entry *h) -{ - if (!h->filtered) { - hists->nr_non_filtered_entries++; - hists->stats.total_non_filtered_period += h->stat.period; - } - hists->nr_entries++; - hists->stats.total_period += h->stat.period; -} - static u8 symbol__parent_filter(const struct symbol *parent) { if (symbol_conf.exclude_other && parent == NULL) @@ -632,6 +622,35 @@ out: return ret; } +static void hists__reset_filter_stats(struct hists *hists) +{ + hists->nr_non_filtered_entries = 0; + hists->stats.total_non_filtered_period = 0; +} + +void hists__reset_stats(struct hists *hists) +{ + hists->nr_entries = 0; + hists->stats.total_period = 0; + + hists__reset_filter_stats(hists); +} + +static void hists__inc_filter_stats(struct hists *hists, struct hist_entry *h) +{ + hists->nr_non_filtered_entries++; + hists->stats.total_non_filtered_period += h->stat.period; +} + +void hists__inc_stats(struct hists *hists, struct hist_entry *h) +{ + if (!h->filtered) + hists__inc_filter_stats(hists, h); + + hists->nr_entries++; + hists->stats.total_period += h->stat.period; +} + static void __hists__insert_output_entry(struct rb_root *entries, struct hist_entry *he, u64 min_callchain_hits) @@ -675,9 +694,7 @@ void hists__output_resort(struct hists *hists) next = rb_first(root); hists->entries = RB_ROOT; - hists->nr_non_filtered_entries = 0; - hists->stats.total_period = 0; - hists->stats.total_non_filtered_period = 0; + hists__reset_stats(hists); hists__reset_col_len(hists); while (next) { @@ -699,13 +716,13 @@ static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h if (h->filtered) return; - ++hists->nr_non_filtered_entries; if (h->ms.unfolded) hists->nr_non_filtered_entries += h->nr_rows; h->row_offset = 0; - hists->stats.total_non_filtered_period += h->stat.period; + hists->stats.nr_non_filtered_samples += h->stat.nr_events; + hists__inc_filter_stats(hists, h); hists__calc_col_len(hists, h); } @@ -726,9 +743,9 @@ void hists__filter_by_dso(struct hists *hists) { struct rb_node *nd; - hists->nr_non_filtered_entries = 0; - hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; + + hists__reset_filter_stats(hists); hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { @@ -760,9 +777,9 @@ void hists__filter_by_thread(struct hists *hists) { struct rb_node *nd; - hists->nr_non_filtered_entries = 0; - hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; + + hists__reset_filter_stats(hists); hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { @@ -792,9 +809,9 @@ void hists__filter_by_symbol(struct hists *hists) { struct rb_node *nd; - hists->nr_non_filtered_entries = 0; - hists->stats.total_non_filtered_period = 0; hists->stats.nr_non_filtered_samples = 0; + + hists__reset_filter_stats(hists); hists__reset_col_len(hists); for (nd = rb_first(&hists->entries); nd; nd = rb_next(nd)) { diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index 51478c94d976..ef1ad7a948c0 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -116,6 +116,7 @@ void hists__decay_entries(struct hists *hists, bool zap_user, bool zap_kernel); void hists__output_recalc_col_len(struct hists *hists, int max_rows); u64 hists__total_period(struct hists *hists); +void hists__reset_stats(struct hists *hists); void hists__inc_stats(struct hists *hists, struct hist_entry *h); void hists__inc_nr_events(struct hists *hists, u32 type); void events_stats__inc(struct events_stats *stats, u32 type); -- cgit v1.2.3 From 87e90f43285f4096e9ba5fc18b05c2e04caf3fab Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 24 Apr 2014 16:44:16 +0900 Subject: perf hists: Collapse expanded callchains after filter is applied When a filter is applied a hist entry checks whether its callchain was folded and account it to the output stat. But this is rather hacky and only TUI-specific. Simply fold the callchains for the entry looks like a simpler and more generic solution IMHO. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-6-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index b675857883a2..8d5cfcc3bc63 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -716,8 +716,8 @@ static void hists__remove_entry_filter(struct hists *hists, struct hist_entry *h if (h->filtered) return; - if (h->ms.unfolded) - hists->nr_non_filtered_entries += h->nr_rows; + /* force fold unfiltered entry for simplicity */ + h->ms.unfolded = false; h->row_offset = 0; hists->stats.nr_non_filtered_samples += h->stat.nr_events; -- cgit v1.2.3 From 820bc81f4cdaac09a8f25040d3a20d86f3da292b Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 22 Apr 2014 11:44:21 +0900 Subject: perf tools: Account entry stats when it's added to the output tree Currently, accounting each sample is done in multiple places - once when adding them to the input tree, other when adding them to the output tree. It's not only confusing but also can cause a subtle problem since concurrent processing like in perf top might see the updated stats before adding entries into the output tree - like seeing more (blank) lines at the end and/or slight inaccurate percentage. To fix this, only account the entries when it's moved into the output tree so that they cannot be seen prematurely. There're some exceptional cases here and there - they should be addressed separately with comments. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-7-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 1 - 1 file changed, 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 8d5cfcc3bc63..6d0d2d75db68 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -382,7 +382,6 @@ static struct hist_entry *add_hist_entry(struct hists *hists, if (!he) return NULL; - hists->nr_entries++; rb_link_node(&he->rb_node_in, parent, p); rb_insert_color(&he->rb_node_in, hists->entries_in); out: -- cgit v1.2.3 From 3186b6815d49b5e0defbd884223da3778edb59fc Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 22 Apr 2014 13:44:23 +0900 Subject: perf hists: Add missing update on filtered stats in hists__decay_entries() When a filter is used for perf top, its hists->nr_non_filtered_entries was not updated after it removed an entry in hists__decay_entries(). Also hists->stats.total_non_filtered_period was missed too. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-8-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 6d0d2d75db68..7f0236cea4fe 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -225,14 +225,18 @@ static void he_stat__decay(struct he_stat *he_stat) static bool hists__decay_entry(struct hists *hists, struct hist_entry *he) { u64 prev_period = he->stat.period; + u64 diff; if (prev_period == 0) return true; he_stat__decay(&he->stat); + diff = prev_period - he->stat.period; + + hists->stats.total_period -= diff; if (!he->filtered) - hists->stats.total_period -= prev_period - he->stat.period; + hists->stats.total_non_filtered_period -= diff; return he->stat.period == 0; } @@ -259,8 +263,11 @@ void hists__decay_entries(struct hists *hists, bool zap_user, bool zap_kernel) if (sort__need_collapse) rb_erase(&n->rb_node_in, &hists->entries_collapsed); - hist_entry__free(n); --hists->nr_entries; + if (!n->filtered) + --hists->nr_non_filtered_entries; + + hist_entry__free(n); } } } -- cgit v1.2.3 From 268397cb2a47ce6e1c0298d9de1762143867f9d3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 22 Apr 2014 14:49:31 +0900 Subject: perf top/tui: Update nr_entries properly after a filter is applied The hist_browser__reset() is only called right after a filter is applied so it needs to udpate browser->nr_entries properly. We cannot use hists->nr_non_filtered_entreis directly since it's possible that such entries are also filtered out by minimum percentage limit. In addition when a filter is used for perf top, hist browser's nr_entries field was not updated after applying the filter. But it needs to be updated as new samples are coming. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398327843-31845-11-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/hist.h | 6 ++++++ 1 file changed, 6 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index ef1ad7a948c0..38c3e874c164 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -129,6 +129,12 @@ void hists__filter_by_dso(struct hists *hists); void hists__filter_by_thread(struct hists *hists); void hists__filter_by_symbol(struct hists *hists); +static inline bool hists__has_filter(struct hists *hists) +{ + return hists->thread_filter || hists->dso_filter || + hists->symbol_filter_str; +} + u16 hists__col_len(struct hists *hists, enum hist_column col); void hists__set_col_len(struct hists *hists, enum hist_column col, u16 len); bool hists__new_col_len(struct hists *hists, enum hist_column col, u16 len); -- cgit v1.2.3 From 93d5731dcb5b8cb7fa56ee11a5891f10c96c2a45 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 21 Mar 2014 17:57:01 -0300 Subject: perf tools: Allocate thread map_groups's dynamically Moving towards sharing map groups within a process threads. Because of this we need the map groups to be dynamically allocated. No other functional change is intended in here. Based on a patch by Jiri Olsa, but this time _just_ making the conversion from statically allocating thread->mg to turning it into a pointer and instead of initializing it at thread's constructor, introduce a constructor/destructor for the map_groups class and call at thread creation time. Later we will introduce the get/put methods when we move to sharing those map_groups, when the get/put refcounting semantics will be needed. Signed-off-by: Arnaldo Carvalho de Melo Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Corey Ashford Cc: David Ahern Cc: Don Zickus Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1397490723-1992-3-git-send-email-jolsa@redhat.com Signed-off-by: Jiri Olsa --- tools/perf/util/event.c | 2 +- tools/perf/util/map.c | 16 ++++++++++++++++ tools/perf/util/map.h | 3 +++ tools/perf/util/thread.c | 18 ++++++++++++------ tools/perf/util/thread.h | 2 +- 5 files changed, 33 insertions(+), 8 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 9d12aa6dd485..dbcaea1a8180 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -699,7 +699,7 @@ void thread__find_addr_map(struct thread *thread, enum map_type type, u64 addr, struct addr_location *al) { - struct map_groups *mg = &thread->mg; + struct map_groups *mg = thread->mg; bool load_map = false; al->machine = machine; diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c index 39cd2d0faff6..ae4c5e12debd 100644 --- a/tools/perf/util/map.c +++ b/tools/perf/util/map.c @@ -358,6 +358,22 @@ void map_groups__exit(struct map_groups *mg) } } +struct map_groups *map_groups__new(void) +{ + struct map_groups *mg = malloc(sizeof(*mg)); + + if (mg != NULL) + map_groups__init(mg); + + return mg; +} + +void map_groups__delete(struct map_groups *mg) +{ + map_groups__exit(mg); + free(mg); +} + void map_groups__flush(struct map_groups *mg) { int type; diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h index f00f058afb3b..1073e2d8b797 100644 --- a/tools/perf/util/map.h +++ b/tools/perf/util/map.h @@ -61,6 +61,9 @@ struct map_groups { struct machine *machine; }; +struct map_groups *map_groups__new(void); +void map_groups__delete(struct map_groups *mg); + static inline struct kmap *map__kmap(struct map *map) { return (struct kmap *)(map + 1); diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index 3ce0498bdae6..dc51d1632e92 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -15,7 +15,10 @@ struct thread *thread__new(pid_t pid, pid_t tid) struct thread *thread = zalloc(sizeof(*thread)); if (thread != NULL) { - map_groups__init(&thread->mg); + thread->mg = map_groups__new(); + if (thread->mg == NULL) + goto out_free; + thread->pid_ = pid; thread->tid = tid; thread->ppid = -1; @@ -37,6 +40,8 @@ struct thread *thread__new(pid_t pid, pid_t tid) return thread; err_thread: + map_groups__delete(thread->mg); +out_free: free(thread); return NULL; } @@ -45,7 +50,8 @@ void thread__delete(struct thread *thread) { struct comm *comm, *tmp; - map_groups__exit(&thread->mg); + map_groups__delete(thread->mg); + thread->mg = NULL; list_for_each_entry_safe(comm, tmp, &thread->comm_list, list) { list_del(&comm->list); comm__free(comm); @@ -111,13 +117,13 @@ int thread__comm_len(struct thread *thread) size_t thread__fprintf(struct thread *thread, FILE *fp) { return fprintf(fp, "Thread %d %s\n", thread->tid, thread__comm_str(thread)) + - map_groups__fprintf(&thread->mg, verbose, fp); + map_groups__fprintf(thread->mg, verbose, fp); } void thread__insert_map(struct thread *thread, struct map *map) { - map_groups__fixup_overlappings(&thread->mg, map, verbose, stderr); - map_groups__insert(&thread->mg, map); + map_groups__fixup_overlappings(thread->mg, map, verbose, stderr); + map_groups__insert(thread->mg, map); } int thread__fork(struct thread *thread, struct thread *parent, u64 timestamp) @@ -135,7 +141,7 @@ int thread__fork(struct thread *thread, struct thread *parent, u64 timestamp) } for (i = 0; i < MAP__NR_TYPES; ++i) - if (map_groups__clone(&thread->mg, &parent->mg, i) < 0) + if (map_groups__clone(thread->mg, parent->mg, i) < 0) return -ENOMEM; thread->ppid = parent->tid; diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h index 9b29f085aede..bee1eb0f73bc 100644 --- a/tools/perf/util/thread.h +++ b/tools/perf/util/thread.h @@ -13,7 +13,7 @@ struct thread { struct rb_node rb_node; struct list_head node; }; - struct map_groups mg; + struct map_groups *mg; pid_t pid_; /* Not all tools update this */ pid_t tid; pid_t ppid; -- cgit v1.2.3 From a26ca6716a6c683f40bd676cea7e89704653b98d Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 25 Mar 2014 15:26:44 -0300 Subject: perf tools: Reference count map_groups objects We will share it among threads in the same process. Adding map_groups__get/map_groups__put interface for that. Signed-off-by: Arnaldo Carvalho de Melo Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Corey Ashford Cc: David Ahern Cc: Don Zickus Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1397490723-1992-4-git-send-email-jolsa@redhat.com Signed-off-by: Jiri Olsa --- tools/perf/util/map.c | 7 +++++++ tools/perf/util/map.h | 9 +++++++++ tools/perf/util/thread.c | 2 +- 3 files changed, 17 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c index ae4c5e12debd..ba5f5c0c838b 100644 --- a/tools/perf/util/map.c +++ b/tools/perf/util/map.c @@ -323,6 +323,7 @@ void map_groups__init(struct map_groups *mg) INIT_LIST_HEAD(&mg->removed_maps[i]); } mg->machine = NULL; + mg->refcnt = 1; } static void maps__delete(struct rb_root *maps) @@ -374,6 +375,12 @@ void map_groups__delete(struct map_groups *mg) free(mg); } +void map_groups__put(struct map_groups *mg) +{ + if (--mg->refcnt == 0) + map_groups__delete(mg); +} + void map_groups__flush(struct map_groups *mg) { int type; diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h index 1073e2d8b797..d6445b27d672 100644 --- a/tools/perf/util/map.h +++ b/tools/perf/util/map.h @@ -59,11 +59,20 @@ struct map_groups { struct rb_root maps[MAP__NR_TYPES]; struct list_head removed_maps[MAP__NR_TYPES]; struct machine *machine; + int refcnt; }; struct map_groups *map_groups__new(void); void map_groups__delete(struct map_groups *mg); +static inline struct map_groups *map_groups__get(struct map_groups *mg) +{ + ++mg->refcnt; + return mg; +} + +void map_groups__put(struct map_groups *mg); + static inline struct kmap *map__kmap(struct map *map) { return (struct kmap *)(map + 1); diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index dc51d1632e92..b501848a8424 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -50,7 +50,7 @@ void thread__delete(struct thread *thread) { struct comm *comm, *tmp; - map_groups__delete(thread->mg); + map_groups__put(thread->mg); thread->mg = NULL; list_for_each_entry_safe(comm, tmp, &thread->comm_list, list) { list_del(&comm->list); -- cgit v1.2.3 From cddcef607782966f1601808c17fe9c4c5f79f9f4 Mon Sep 17 00:00:00 2001 From: Jiri Olsa Date: Wed, 9 Apr 2014 20:54:29 +0200 Subject: perf tools: Share map_groups among threads of the same group Sharing map groups within all process threads. This way there's only one copy of mmap info and it's reachable from any thread within the process. Original-patch-by: Arnaldo Carvalho de Melo Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Corey Ashford Cc: David Ahern Cc: Don Zickus Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Mike Galbraith Cc: Namhyung Kim Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1397490723-1992-5-git-send-email-jolsa@redhat.com Signed-off-by: Jiri Olsa --- tools/perf/util/machine.c | 11 +++++++++++ tools/perf/util/thread.c | 48 ++++++++++++++++++++++++++++++++++------------- tools/perf/util/thread.h | 1 + 3 files changed, 47 insertions(+), 13 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index a53cd0b8c151..98ec56dc890b 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -316,6 +316,17 @@ static struct thread *__machine__findnew_thread(struct machine *machine, rb_link_node(&th->rb_node, parent, p); rb_insert_color(&th->rb_node, &machine->threads); machine->last_match = th; + + /* + * We have to initialize map_groups separately + * after rb tree is updated. + * + * The reason is that we call machine__findnew_thread + * within thread__init_map_groups to find the thread + * leader and that would screwed the rb tree. + */ + if (thread__init_map_groups(th, machine)) + return NULL; } return th; diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index b501848a8424..2fde0d5e40b5 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -8,6 +8,22 @@ #include "debug.h" #include "comm.h" +int thread__init_map_groups(struct thread *thread, struct machine *machine) +{ + struct thread *leader; + pid_t pid = thread->pid_; + + if (pid == thread->tid) { + thread->mg = map_groups__new(); + } else { + leader = machine__findnew_thread(machine, pid, pid); + if (leader) + thread->mg = map_groups__get(leader->mg); + } + + return thread->mg ? 0 : -1; +} + struct thread *thread__new(pid_t pid, pid_t tid) { char *comm_str; @@ -15,10 +31,6 @@ struct thread *thread__new(pid_t pid, pid_t tid) struct thread *thread = zalloc(sizeof(*thread)); if (thread != NULL) { - thread->mg = map_groups__new(); - if (thread->mg == NULL) - goto out_free; - thread->pid_ = pid; thread->tid = tid; thread->ppid = -1; @@ -40,8 +52,6 @@ struct thread *thread__new(pid_t pid, pid_t tid) return thread; err_thread: - map_groups__delete(thread->mg); -out_free: free(thread); return NULL; } @@ -126,9 +136,26 @@ void thread__insert_map(struct thread *thread, struct map *map) map_groups__insert(thread->mg, map); } +static int thread__clone_map_groups(struct thread *thread, + struct thread *parent) +{ + int i; + + /* This is new thread, we share map groups for process. */ + if (thread->pid_ == parent->pid_) + return 0; + + /* But this one is new process, copy maps. */ + for (i = 0; i < MAP__NR_TYPES; ++i) + if (map_groups__clone(thread->mg, parent->mg, i) < 0) + return -ENOMEM; + + return 0; +} + int thread__fork(struct thread *thread, struct thread *parent, u64 timestamp) { - int i, err; + int err; if (parent->comm_set) { const char *comm = thread__comm_str(parent); @@ -140,13 +167,8 @@ int thread__fork(struct thread *thread, struct thread *parent, u64 timestamp) thread->comm_set = true; } - for (i = 0; i < MAP__NR_TYPES; ++i) - if (map_groups__clone(thread->mg, parent->mg, i) < 0) - return -ENOMEM; - thread->ppid = parent->tid; - - return 0; + return thread__clone_map_groups(thread, parent); } void thread__find_cpumode_addr_location(struct thread *thread, diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h index bee1eb0f73bc..3c0c2724f82c 100644 --- a/tools/perf/util/thread.h +++ b/tools/perf/util/thread.h @@ -30,6 +30,7 @@ struct machine; struct comm; struct thread *thread__new(pid_t pid, pid_t tid); +int thread__init_map_groups(struct thread *thread, struct machine *machine); void thread__delete(struct thread *thread); static inline void thread__exited(struct thread *thread) { -- cgit v1.2.3 From e148c76083dc06ce618d768c0bee0a0edda96a54 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 24 Apr 2014 22:27:32 +0900 Subject: perf tools: Handle EINTR error for readn/writen Those readn/writen functions are to ensure read/write does I/O for a given size exactly. But ion() - its implementation - does not handle in case it returns prematurely due to a signal. As it's not an error itself so just retry the operation. Signed-off-by: Namhyung Kim Link: http://lkml.kernel.org/r/1398346054-3322-1-git-send-email-namhyung@kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/util.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c index 9f66549562bd..7fff6be07f07 100644 --- a/tools/perf/util/util.c +++ b/tools/perf/util/util.c @@ -166,6 +166,8 @@ static ssize_t ion(bool is_read, int fd, void *buf, size_t n) ssize_t ret = is_read ? read(fd, buf, left) : write(fd, buf, left); + if (ret < 0 && errno == EINTR) + continue; if (ret <= 0) return ret; -- cgit v1.2.3 From 2b9032e0ecb57de819bcf40b440e7cbd2d8f3a8c Mon Sep 17 00:00:00 2001 From: Alexander Yarygin Date: Fri, 25 Apr 2014 17:34:05 +0200 Subject: perf tools: Parse tracepoints with '-' in system name Trace events potentially can have a '-' in their trace system name, e.g. kvm on s390 defines kvm-s390:* tracepoints. We could not parse them, because there was no rule for this: $ sudo ./perf top -e "kvm-s390:*" invalid or unsupported event: 'kvm-s390:*' This patch adds an extra rule to event_legacy_tracepoint which handles those cases. Without the patch, perf will not accept such tracepoints in the -e option. Signed-off-by: Alexander Yarygin Tested-by: Christian Borntraeger Acked-by: Christian Borntraeger Link: http://lkml.kernel.org/r/1398440047-6641-2-git-send-email-yarygin@linux.vnet.ibm.com Signed-off-by: Jiri Olsa --- tools/perf/util/parse-events.y | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.y b/tools/perf/util/parse-events.y index 4eb67ec333f1..ac9db9f699f3 100644 --- a/tools/perf/util/parse-events.y +++ b/tools/perf/util/parse-events.y @@ -299,6 +299,18 @@ PE_PREFIX_MEM PE_VALUE sep_dc } event_legacy_tracepoint: +PE_NAME '-' PE_NAME ':' PE_NAME +{ + struct parse_events_evlist *data = _data; + struct list_head *list; + char sys_name[128]; + snprintf(&sys_name, 128, "%s-%s", $1, $3); + + ALLOC_LIST(list); + ABORT_ON(parse_events_add_tracepoint(list, &data->idx, &sys_name, $5)); + $$ = list; +} +| PE_NAME ':' PE_NAME { struct parse_events_evlist *data = _data; -- cgit v1.2.3 From 88080ce7f6af1ad99ad4b2825938411975910116 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Mon, 14 Apr 2014 17:38:39 +0200 Subject: perf tools: Move u64_swap union ... to its single user's header, evsel.h. Signed-off-by: Borislav Petkov Link: http://lkml.kernel.org/r/tip-9os1chjyz12upubfsjc71d99@git.kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/evsel.h | 5 +++++ tools/perf/util/types.h | 5 ----- 2 files changed, 5 insertions(+), 5 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 0c9926cfb292..0c581d0d5eb6 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -91,6 +91,11 @@ struct perf_evsel { char *group_name; }; +union u64_swap { + u64 val64; + u32 val32[2]; +}; + #define hists_to_evsel(h) container_of(h, struct perf_evsel, hists) struct cpu_map; diff --git a/tools/perf/util/types.h b/tools/perf/util/types.h index c51fa6b70a28..5f3689a3d085 100644 --- a/tools/perf/util/types.h +++ b/tools/perf/util/types.h @@ -16,9 +16,4 @@ typedef signed short s16; typedef unsigned char u8; typedef signed char s8; -union u64_swap { - u64 val64; - u32 val32[2]; -}; - #endif /* __PERF_TYPES_H */ -- cgit v1.2.3 From 5ac3e4b6d1d8fb911bb9c497126c51b02033a412 Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Sun, 23 Feb 2014 12:04:53 +0100 Subject: tools: Unify export.h So tools/ has been growing three, at a different stage of their development export.h headers and so we should unite into one. Add tools/include/ to the include path of virtio and liblockdep to pick the shared header now. Signed-off-by: Borislav Petkov Acked-by: Rusty Russell Cc: Arnaldo Carvalho de Melo Cc: Ingo Molnar Cc: Michael S. Tsirkin Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Sasha Levin Cc: virtio-dev@lists.oasis-open.org Cc: virtualization@lists.linux-foundation.org Link: http://lkml.kernel.org/r/1397493185-19521-2-git-send-email-bp@alien8.de Signed-off-by: Jiri Olsa --- tools/perf/util/include/linux/export.h | 6 ------ 1 file changed, 6 deletions(-) delete mode 100644 tools/perf/util/include/linux/export.h (limited to 'tools/perf/util') diff --git a/tools/perf/util/include/linux/export.h b/tools/perf/util/include/linux/export.h deleted file mode 100644 index b43e2dc21e04..000000000000 --- a/tools/perf/util/include/linux/export.h +++ /dev/null @@ -1,6 +0,0 @@ -#ifndef PERF_LINUX_MODULE_H -#define PERF_LINUX_MODULE_H - -#define EXPORT_SYMBOL(name) - -#endif -- cgit v1.2.3 From d944c4eebcf4c0d5e5d9728fec110cbf0047ad7f Mon Sep 17 00:00:00 2001 From: Borislav Petkov Date: Fri, 25 Apr 2014 21:31:02 +0200 Subject: tools: Consolidate types.h Combine all definitions into a common tools/include/linux/types.h and kill the wild growth elsewhere. Move DECLARE_BITMAP to its proper bitmap.h header. Signed-off-by: Borislav Petkov Acked-by: Rusty Russell Link: http://lkml.kernel.org/n/tip-azczs7qcv6h9xek9od10hiv2@git.kernel.org Signed-off-by: Jiri Olsa --- tools/perf/util/annotate.h | 2 +- tools/perf/util/build-id.h | 2 +- tools/perf/util/dso.h | 2 +- tools/perf/util/evsel.h | 4 ++-- tools/perf/util/header.h | 4 ++-- tools/perf/util/include/linux/bitmap.h | 3 +++ tools/perf/util/include/linux/list.h | 1 + tools/perf/util/include/linux/types.h | 29 ----------------------------- tools/perf/util/map.h | 2 +- tools/perf/util/parse-events.h | 3 +-- tools/perf/util/parse-events.y | 2 +- tools/perf/util/perf_regs.h | 2 +- tools/perf/util/pmu.h | 2 +- tools/perf/util/stat.h | 2 +- tools/perf/util/svghelper.c | 2 +- tools/perf/util/svghelper.h | 2 +- tools/perf/util/top.h | 2 +- tools/perf/util/types.h | 19 ------------------- tools/perf/util/unwind-libdw.c | 2 +- tools/perf/util/unwind.h | 2 +- tools/perf/util/util.h | 2 +- tools/perf/util/values.h | 2 +- 22 files changed, 24 insertions(+), 69 deletions(-) delete mode 100644 tools/perf/util/include/linux/types.h delete mode 100644 tools/perf/util/types.h (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index 56ad4f5287de..112d6e268150 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -3,7 +3,7 @@ #include #include -#include "types.h" +#include #include "symbol.h" #include "hist.h" #include "sort.h" diff --git a/tools/perf/util/build-id.h b/tools/perf/util/build-id.h index 845ef865eced..ae392561470b 100644 --- a/tools/perf/util/build-id.h +++ b/tools/perf/util/build-id.h @@ -4,7 +4,7 @@ #define BUILD_ID_SIZE 20 #include "tool.h" -#include "types.h" +#include extern struct perf_tool build_id__mark_dso_hit_ops; struct dso; diff --git a/tools/perf/util/dso.h b/tools/perf/util/dso.h index ab06f1c03655..38efe95a7fdd 100644 --- a/tools/perf/util/dso.h +++ b/tools/perf/util/dso.h @@ -4,7 +4,7 @@ #include #include #include -#include "types.h" +#include #include "map.h" #include "build-id.h" diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 0c581d0d5eb6..a52e9a5bb2d0 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -5,12 +5,12 @@ #include #include #include