From e3b03b6c1a4f3b4564be08809f58584592621a0a Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Thu, 5 May 2016 16:04:03 -0700 Subject: perf stat: Avoid fractional digits for integer scales When the scaling factor is a full integer don't display fractional digits. This avoids unnecessary .00 output for topdown metrics with scale factors. v2: Remove redundant check. Signed-off-by: Andi Kleen Acked-by: Jiri Olsa Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1462489447-31832-7-git-send-email-andi@firstfloor.org [ Rename 'round' to 'stat_round' as 'round' is defined in math.h, included by this patch, and this breaks the build on ubuntu 12.04 ] Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 5645a83..16a923c 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -66,6 +66,7 @@ #include #include #include +#include #define DEFAULT_SEPARATOR " " #define CNTR_NOT_SUPPORTED "" @@ -986,12 +987,12 @@ static void abs_printout(int id, int nr, struct perf_evsel *evsel, double avg) const char *fmt; if (csv_output) { - fmt = sc != 1.0 ? "%.2f%s" : "%.0f%s"; + fmt = floor(sc) != sc ? "%.2f%s" : "%.0f%s"; } else { if (big_num) - fmt = sc != 1.0 ? "%'18.2f%s" : "%'18.0f%s"; + fmt = floor(sc) != sc ? "%'18.2f%s" : "%'18.0f%s"; else - fmt = sc != 1.0 ? "%18.2f%s" : "%18.0f%s"; + fmt = floor(sc) != sc ? "%18.2f%s" : "%18.0f%s"; } aggr_printout(evsel, id, nr); @@ -1995,7 +1996,7 @@ static int process_stat_round_event(struct perf_tool *tool __maybe_unused, union perf_event *event, struct perf_session *session) { - struct stat_round_event *round = &event->stat_round; + struct stat_round_event *stat_round = &event->stat_round; struct perf_evsel *counter; struct timespec tsh, *ts = NULL; const char **argv = session->header.env.cmdline_argv; @@ -2004,12 +2005,12 @@ static int process_stat_round_event(struct perf_tool *tool __maybe_unused, evlist__for_each(evsel_list, counter) perf_stat_process_counter(&stat_config, counter); - if (round->type == PERF_STAT_ROUND_TYPE__FINAL) - update_stats(&walltime_nsecs_stats, round->time); + if (stat_round->type == PERF_STAT_ROUND_TYPE__FINAL) + update_stats(&walltime_nsecs_stats, stat_round->time); - if (stat_config.interval && round->time) { - tsh.tv_sec = round->time / NSECS_PER_SEC; - tsh.tv_nsec = round->time % NSECS_PER_SEC; + if (stat_config.interval && stat_round->time) { + tsh.tv_sec = stat_round->time / NSECS_PER_SEC; + tsh.tv_nsec = stat_round->time % NSECS_PER_SEC; ts = &tsh; } -- cgit v0.10.2 From 6ae98ba611ed1c11ddc5645475bc03b46a3c04e7 Mon Sep 17 00:00:00 2001 From: He Kuang Date: Thu, 12 May 2016 08:43:11 +0000 Subject: perf symbols: Store vdso buildid unconditionally When unwinding callchains on a different machine, vdso info should be available so the unwind process won't be interrupted if address falls into vdso region. But in most cases, the addresses of sample events are not in vdso range, the buildid of a zero hit vdso won't be stored into perf.data. This patch stores vdso buildid regardless of whether the vdso is hit or not. Signed-off-by: He Kuang Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Andi Kleen Cc: David Ahern Cc: Ekaterina Tumanova Cc: Jiri Olsa Cc: Josh Poimboeuf Cc: Kan Liang Cc: Masami Hiramatsu Cc: Namhyung Kim Cc: Pekka Enberg Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Sukadev Bhattiprolu Cc: Wang Nan Link: http://lkml.kernel.org/r/1463042596-61703-3-git-send-email-hekuang@huawei.com Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index bff425e..67e5966 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -256,7 +256,7 @@ static int machine__write_buildid_table(struct machine *machine, int fd) size_t name_len; bool in_kernel = false; - if (!pos->hit) + if (!pos->hit && !dso__is_vdso(pos)) continue; if (dso__is_vdso(pos)) { diff --git a/tools/perf/util/dso.c b/tools/perf/util/dso.c index 3357479..75b7561 100644 --- a/tools/perf/util/dso.c +++ b/tools/perf/util/dso.c @@ -7,6 +7,7 @@ #include "auxtrace.h" #include "util.h" #include "debug.h" +#include "vdso.h" char dso__symtab_origin(const struct dso *dso) { @@ -1169,7 +1170,7 @@ bool __dsos__read_build_ids(struct list_head *head, bool with_hits) struct dso *pos; list_for_each_entry(pos, head, node) { - if (with_hits && !pos->hit) + if (with_hits && !pos->hit && !dso__is_vdso(pos)) continue; if (pos->has_build_id) { have_build_id = true; -- cgit v0.10.2 From b0404be8d6186f9f3c23e2b5ff247e667be90652 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 13 May 2016 15:01:01 +0900 Subject: perf stat: Fix indentation of stalled backend cycle The commit 140aeadc1fb5 ("perf stat: Abstract stat metrics printing") changed how shadow metrics are printed, but it missed to update the width of the stalled backend cycles event to 7.2% like others. This resulted in misaligned output like below: Performance counter stats for 'pwd': 0.638313 task-clock (msec) # 0.567 CPUs utilized 0 context-switches # 0.000 K/sec 0 cpu-migrations # 0.000 K/sec 54 page-faults # 0.085 M/sec 885,600 cycles # 1.387 GHz 558,438 stalled-cycles-frontend # 63.06% frontend cycles idle 431,355 stalled-cycles-backend # 48.71% backend cycles idle 674,956 instructions # 0.76 insn per cycle # 0.83 stalled cycles per insn 130,380 branches # 204.257 M/sec branch-misses 0.001125426 seconds time elapsed Signed-off-by: Namhyung Kim Cc: Andi Kleen Cc: Jiri Olsa Cc: Peter Zijlstra Fixes: 140aeadc1fb5 ("perf stat: Abstract stat metrics printing") Link: http://lkml.kernel.org/r/1463119263-5569-1-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index fdb7196..61200fc 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -188,7 +188,7 @@ static void print_stalled_cycles_backend(int cpu, color = get_ratio_color(GRC_STALLED_CYCLES_BE, ratio); - out->print_metric(out->ctx, color, "%6.2f%%", "backend cycles idle", ratio); + out->print_metric(out->ctx, color, "%7.2f%%", "backend cycles idle", ratio); } static void print_branch_misses(int cpu, -- cgit v0.10.2 From daf4f4786e8af371048e72cb37ac05190e89198a Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 13 May 2016 15:01:02 +0900 Subject: perf stat: Update runtime using cpu-clock event Currently only the task-clock event updates the runtime_nsec so it cannot show the metric when using cpu-clock events. However cpu clock works basically same as task-clock, so no need to not update the runtime IMHO. Before: # perf stat -a -e cpu-clock,context-switches,page-faults,cycles sleep 0.1 Performance counter stats for 'system wide': 1217.759506 cpu-clock (msec) 93 context-switches 61 page-faults 18,958,022 cycles 0.101393794 seconds time elapsed After: Performance counter stats for 'system wide': 1220.471884 cpu-clock (msec) # 12.013 CPUs utilized 118 context-switches # 0.097 K/sec 59 page-faults # 0.048 K/sec 17,941,247 cycles # 0.015 GHz 0.101594777 seconds time elapsed Signed-off-by: Namhyung Kim Tested-by: Arnaldo Carvalho de Melo Cc: Andi Kleen Cc: Jiri Olsa Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1463119263-5569-2-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index 61200fc..aa9efe0 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -94,7 +94,8 @@ void perf_stat__update_shadow_stats(struct perf_evsel *counter, u64 *count, { int ctx = evsel_context(counter); - if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK)) + if (perf_evsel__match(counter, SOFTWARE, SW_TASK_CLOCK) || + perf_evsel__match(counter, SOFTWARE, SW_CPU_CLOCK)) update_stats(&runtime_nsecs_stats[cpu], count[0]); else if (perf_evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) update_stats(&runtime_cycles_stats[ctx][cpu], count[0]); @@ -444,7 +445,8 @@ void perf_stat__print_shadow_stats(struct perf_evsel *evsel, ratio = total / avg; print_metric(ctxp, NULL, "%8.0f", "cycles / elision", ratio); - } else if (perf_evsel__match(evsel, SOFTWARE, SW_TASK_CLOCK)) { + } else if (perf_evsel__match(evsel, SOFTWARE, SW_TASK_CLOCK) || + perf_evsel__match(evsel, SOFTWARE, SW_CPU_CLOCK)) { if ((ratio = avg_stats(&walltime_nsecs_stats)) != 0) print_metric(ctxp, NULL, "%8.3f", "CPUs utilized", avg / ratio); -- cgit v0.10.2 From a1f3d56761df31f0ffeb215b974e26d5613e92a4 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 13 May 2016 15:01:03 +0900 Subject: perf stat: Use cpu-clock event for cpu targets Currently 'perf stat' always counts task-clock event by default. But it's somewhat confusing for system-wide targets (especially with 'sleep N' as the 'sleep' task just sleeps and doesn't use cputime). Changing to cpu-clock event instead for that case makes more sense IMHO. Before: # perf stat -a sleep 0.1 Performance counter stats for 'system wide': 403.038603 task-clock (msec) # 4.001 CPUs utilized 150 context-switches # 0.372 K/sec 7 cpu-migrations # 0.017 K/sec 71 page-faults # 0.176 K/sec 23,705,169 cycles # 0.059 GHz 15,888,166 instructions # 0.67 insn per cycle 3,326,078 branches # 8.253 M/sec 87,643 branch-misses # 2.64% of all branches 0.100737009 seconds time elapsed # After: # perf stat -a sleep 0.1 Performance counter stats for 'system wide': 404.271182 cpu-clock (msec) # 4.000 CPUs utilized 143 context-switches # 0.354 K/sec 13 cpu-migrations # 0.032 K/sec 73 page-faults # 0.181 K/sec 22,119,220 cycles # 0.055 GHz 13,622,065 instructions # 0.62 insn per cycle 2,918,769 branches # 7.220 M/sec 85,033 branch-misses # 2.91% of all branches 0.101073089 seconds time elapsed # Signed-off-by: Namhyung Kim Tested-by: Arnaldo Carvalho de Melo Cc: Andi Kleen Cc: Jiri Olsa Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1463119263-5569-3-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index 16a923c..efdd232 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -1905,6 +1905,9 @@ static int add_default_attributes(void) } if (!evsel_list->nr_entries) { + if (target__has_cpu(&target)) + default_attrs0[0].config = PERF_COUNT_SW_CPU_CLOCK; + if (perf_evlist__add_default_attrs(evsel_list, default_attrs0) < 0) return -1; if (pmu_have_event("cpu", "stalled-cycles-frontend")) { -- cgit v0.10.2 From 0a77582f0407e7f9b5d775bebc31297a1b890be0 Mon Sep 17 00:00:00 2001 From: Masami Hiramatsu Date: Sun, 15 May 2016 12:19:40 +0900 Subject: perf symbols: Introduce DSO__NAME_KALLSYMS and DSO__NAME_KCORE Instead of using a raw string, use DSO__NAME_KALLSYMS and DSO__NAME_KCORE macros for kallsyms and kcore. Signed-off-by: Masami Hiramatsu Cc: Ananth N Mavinakayanahalli Cc: Brendan Gregg Cc: Hemant Kumar Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20160515031935.4017.50971.stgit@devbox Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/builtin-buildid-cache.c b/tools/perf/builtin-buildid-cache.c index 632efc6..d75bded 100644 --- a/tools/perf/builtin-buildid-cache.c +++ b/tools/perf/builtin-buildid-cache.c @@ -119,8 +119,8 @@ static int build_id_cache__add_kcore(const char *filename, bool force) if (build_id_cache__kcore_buildid(from_dir, sbuildid) < 0) return -1; - scnprintf(to_dir, sizeof(to_dir), "%s/[kernel.kcore]/%s", - buildid_dir, sbuildid); + scnprintf(to_dir, sizeof(to_dir), "%s/%s/%s", + buildid_dir, DSO__NAME_KCORE, sbuildid); if (!force && !build_id_cache__kcore_existing(from_dir, to_dir, sizeof(to_dir))) { @@ -131,8 +131,8 @@ static int build_id_cache__add_kcore(const char *filename, bool force) if (build_id_cache__kcore_dir(dir, sizeof(dir))) return -1; - scnprintf(to_dir, sizeof(to_dir), "%s/[kernel.kcore]/%s/%s", - buildid_dir, sbuildid, dir); + scnprintf(to_dir, sizeof(to_dir), "%s/%s/%s/%s", + buildid_dir, DSO__NAME_KCORE, sbuildid, dir); if (mkdir_p(to_dir, 0755)) return -1; diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 4db73d5..b811924 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -1122,7 +1122,7 @@ int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize) } else if (dso__is_kcore(dso)) { goto fallback; } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || - strstr(command, "[kernel.kallsyms]") || + strstr(command, DSO__NAME_KALLSYMS) || access(symfs_filename, R_OK)) { free(filename); fallback: diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 639a290..18dd96b 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -709,7 +709,7 @@ static struct dso *machine__get_kernel(struct machine *machine) if (machine__is_host(machine)) { vmlinux_name = symbol_conf.vmlinux_name; if (!vmlinux_name) - vmlinux_name = "[kernel.kallsyms]"; + vmlinux_name = DSO__NAME_KALLSYMS; kernel = machine__findnew_kernel(machine, vmlinux_name, "[kernel]", DSO_TYPE_KERNEL); diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 7fb3330..2252b54 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -1662,8 +1662,8 @@ static char *dso__find_kallsyms(struct dso *dso, struct map *map) build_id__sprintf(dso->build_id, sizeof(dso->build_id), sbuild_id); - scnprintf(path, sizeof(path), "%s/[kernel.kcore]/%s", buildid_dir, - sbuild_id); + scnprintf(path, sizeof(path), "%s/%s/%s", buildid_dir, + DSO__NAME_KCORE, sbuild_id); /* Use /proc/kallsyms if possible */ if (is_host) { @@ -1699,8 +1699,8 @@ static char *dso__find_kallsyms(struct dso *dso, struct map *map) if (!find_matching_kcore(map, path, sizeof(path))) return strdup(path); - scnprintf(path, sizeof(path), "%s/[kernel.kallsyms]/%s", - buildid_dir, sbuild_id); + scnprintf(path, sizeof(path), "%s/%s/%s", + buildid_dir, DSO__NAME_KALLSYMS, sbuild_id); if (access(path, F_OK)) { pr_err("No kallsyms or vmlinux with build-id %s was found\n", @@ -1769,7 +1769,7 @@ do_kallsyms: if (err > 0 && !dso__is_kcore(dso)) { dso->binary_type = DSO_BINARY_TYPE__KALLSYMS; - dso__set_long_name(dso, "[kernel.kallsyms]", false); + dso__set_long_name(dso, DSO__NAME_KALLSYMS, false); map__fixup_start(map); map__fixup_end(map); } diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h index 2b5e4ed..25f2fd67 100644 --- a/tools/perf/util/symbol.h +++ b/tools/perf/util/symbol.h @@ -44,6 +44,9 @@ Elf_Scn *elf_section_by_name(Elf *elf, GElf_Ehdr *ep, #define DMGL_ANSI (1 << 1) /* Include const, volatile, etc */ #endif +#define DSO__NAME_KALLSYMS "[kernel.kallsyms]" +#define DSO__NAME_KCORE "[kernel.kcore]" + /** struct symbol - symtab entry * * @ignore - resolvable but tools ignore it (e.g. idle routines) -- cgit v0.10.2 From a831100aeefbe6d9f3e47a3e2712f82c042f1f5c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 10 May 2016 16:34:53 -0300 Subject: perf core: Generalize max_stack sysctl handler So that it can be used for other stack related knobs, such as the upcoming one to tweak the max number of of contexts per stack sample. In all those cases we can only change the value if there are no perf sessions collecting stacks, so they need to grab that mutex, etc. Cc: David Ahern Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-8t3fk94wuzp8m2z1n4gc0s17@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c index b9325e7..7fc8993 100644 --- a/kernel/events/callchain.c +++ b/kernel/events/callchain.c @@ -228,7 +228,8 @@ exit_put: int perf_event_max_stack_handler(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos) { - int new_value = sysctl_perf_event_max_stack, ret; + int *value = table->data; + int new_value = *value, ret; struct ctl_table new_table = *table; new_table.data = &new_value; @@ -240,7 +241,7 @@ int perf_event_max_stack_handler(struct ctl_table *table, int write, if (atomic_read(&nr_callchain_events)) ret = -EBUSY; else - sysctl_perf_event_max_stack = new_value; + *value = new_value; mutex_unlock(&callchain_mutex); diff --git a/kernel/sysctl.c b/kernel/sysctl.c index c8b3186..0ec6907 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1149,7 +1149,7 @@ static struct ctl_table kern_table[] = { }, { .procname = "perf_event_max_stack", - .data = NULL, /* filled in by handler */ + .data = &sysctl_perf_event_max_stack, .maxlen = sizeof(sysctl_perf_event_max_stack), .mode = 0644, .proc_handler = perf_event_max_stack_handler, -- cgit v0.10.2 From cfbcf468454ab4b20f0b4b62da51920b99fdb19e Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 28 Apr 2016 12:30:53 -0300 Subject: perf core: Pass max stack as a perf_callchain_entry context This makes perf_callchain_{user,kernel}() receive the max stack as context for the perf_callchain_entry, instead of accessing the global sysctl_perf_event_max_stack. Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Brendan Gregg Cc: David Ahern Cc: Frederic Weisbecker Cc: He Kuang Cc: Jiri Olsa Cc: Linus Torvalds Cc: Masami Hiramatsu Cc: Milian Wolff Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Vince Weaver Cc: Wang Nan Cc: Zefan Li Link: http://lkml.kernel.org/n/tip-kolmn1yo40p7jhswxwrc7rrd@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/arch/arc/kernel/perf_event.c b/arch/arc/kernel/perf_event.c index 8b134cf..6fd4802 100644 --- a/arch/arc/kernel/perf_event.c +++ b/arch/arc/kernel/perf_event.c @@ -48,7 +48,7 @@ struct arc_callchain_trace { static int callchain_trace(unsigned int addr, void *data) { struct arc_callchain_trace *ctrl = data; - struct perf_callchain_entry *entry = ctrl->perf_stuff; + struct perf_callchain_entry_ctx *entry = ctrl->perf_stuff; perf_callchain_store(entry, addr); if (ctrl->depth++ < 3) @@ -58,7 +58,7 @@ static int callchain_trace(unsigned int addr, void *data) } void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct arc_callchain_trace ctrl = { .depth = 0, @@ -69,7 +69,7 @@ perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) } void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { /* * User stack can't be unwound trivially with kernel dwarf unwinder diff --git a/arch/arm/kernel/perf_callchain.c b/arch/arm/kernel/perf_callchain.c index 27563be..bc552e8 100644 --- a/arch/arm/kernel/perf_callchain.c +++ b/arch/arm/kernel/perf_callchain.c @@ -31,7 +31,7 @@ struct frame_tail { */ static struct frame_tail __user * user_backtrace(struct frame_tail __user *tail, - struct perf_callchain_entry *entry) + struct perf_callchain_entry_ctx *entry) { struct frame_tail buftail; unsigned long err; @@ -59,7 +59,7 @@ user_backtrace(struct frame_tail __user *tail, } void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct frame_tail __user *tail; @@ -75,7 +75,7 @@ perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) tail = (struct frame_tail __user *)regs->ARM_fp - 1; - while ((entry->nr < sysctl_perf_event_max_stack) && + while ((entry->entry->nr < entry->max_stack) && tail && !((unsigned long)tail & 0x3)) tail = user_backtrace(tail, entry); } @@ -89,13 +89,13 @@ static int callchain_trace(struct stackframe *fr, void *data) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; perf_callchain_store(entry, fr->pc); return 0; } void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct stackframe fr; diff --git a/arch/arm64/kernel/perf_callchain.c b/arch/arm64/kernel/perf_callchain.c index 32c3c6e..0d60150 100644 --- a/arch/arm64/kernel/perf_callchain.c +++ b/arch/arm64/kernel/perf_callchain.c @@ -31,7 +31,7 @@ struct frame_tail { */ static struct frame_tail __user * user_backtrace(struct frame_tail __user *tail, - struct perf_callchain_entry *entry) + struct perf_callchain_entry_ctx *entry) { struct frame_tail buftail; unsigned long err; @@ -76,7 +76,7 @@ struct compat_frame_tail { static struct compat_frame_tail __user * compat_user_backtrace(struct compat_frame_tail __user *tail, - struct perf_callchain_entry *entry) + struct perf_callchain_entry_ctx *entry) { struct compat_frame_tail buftail; unsigned long err; @@ -106,7 +106,7 @@ compat_user_backtrace(struct compat_frame_tail __user *tail, } #endif /* CONFIG_COMPAT */ -void perf_callchain_user(struct perf_callchain_entry *entry, +void perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) { @@ -122,7 +122,7 @@ void perf_callchain_user(struct perf_callchain_entry *entry, tail = (struct frame_tail __user *)regs->regs[29]; - while (entry->nr < sysctl_perf_event_max_stack && + while (entry->entry->nr < entry->max_stack && tail && !((unsigned long)tail & 0xf)) tail = user_backtrace(tail, entry); } else { @@ -132,7 +132,7 @@ void perf_callchain_user(struct perf_callchain_entry *entry, tail = (struct compat_frame_tail __user *)regs->compat_fp - 1; - while ((entry->nr < sysctl_perf_event_max_stack) && + while ((entry->entry->nr < entry->max_stack) && tail && !((unsigned long)tail & 0x3)) tail = compat_user_backtrace(tail, entry); #endif @@ -146,12 +146,12 @@ void perf_callchain_user(struct perf_callchain_entry *entry, */ static int callchain_trace(struct stackframe *frame, void *data) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; perf_callchain_store(entry, frame->pc); return 0; } -void perf_callchain_kernel(struct perf_callchain_entry *entry, +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct stackframe frame; diff --git a/arch/metag/kernel/perf_callchain.c b/arch/metag/kernel/perf_callchain.c index 252abc1..b3261a9 100644 --- a/arch/metag/kernel/perf_callchain.c +++ b/arch/metag/kernel/perf_callchain.c @@ -29,7 +29,7 @@ static bool is_valid_call(unsigned long calladdr) static struct metag_frame __user * user_backtrace(struct metag_frame __user *user_frame, - struct perf_callchain_entry *entry) + struct perf_callchain_entry_ctx *entry) { struct metag_frame frame; unsigned long calladdr; @@ -56,7 +56,7 @@ user_backtrace(struct metag_frame __user *user_frame, } void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long sp = regs->ctx.AX[0].U0; struct metag_frame __user *frame; @@ -65,7 +65,7 @@ perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) --frame; - while ((entry->nr < sysctl_perf_event_max_stack) && frame) + while ((entry->entry->nr < entry->max_stack) && frame) frame = user_backtrace(frame, entry); } @@ -78,13 +78,13 @@ static int callchain_trace(struct stackframe *fr, void *data) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; perf_callchain_store(entry, fr->pc); return 0; } void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct stackframe fr; diff --git a/arch/mips/kernel/perf_event.c b/arch/mips/kernel/perf_event.c index 5021c54..22395c7 100644 --- a/arch/mips/kernel/perf_event.c +++ b/arch/mips/kernel/perf_event.c @@ -25,8 +25,8 @@ * the user stack callchains, we will add it here. */ -static void save_raw_perf_callchain(struct perf_callchain_entry *entry, - unsigned long reg29) +static void save_raw_perf_callchain(struct perf_callchain_entry_ctx *entry, + unsigned long reg29) { unsigned long *sp = (unsigned long *)reg29; unsigned long addr; @@ -35,14 +35,14 @@ static void save_raw_perf_callchain(struct perf_callchain_entry *entry, addr = *sp++; if (__kernel_text_address(addr)) { perf_callchain_store(entry, addr); - if (entry->nr >= sysctl_perf_event_max_stack) + if (entry->entry->nr >= entry->max_stack) break; } } } -void perf_callchain_kernel(struct perf_callchain_entry *entry, - struct pt_regs *regs) +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, + struct pt_regs *regs) { unsigned long sp = regs->regs[29]; #ifdef CONFIG_KALLSYMS @@ -59,7 +59,7 @@ void perf_callchain_kernel(struct perf_callchain_entry *entry, } do { perf_callchain_store(entry, pc); - if (entry->nr >= sysctl_perf_event_max_stack) + if (entry->entry->nr >= entry->max_stack) break; pc = unwind_stack(current, &sp, pc, &ra); } while (pc); diff --git a/arch/powerpc/perf/callchain.c b/arch/powerpc/perf/callchain.c index 22d9015..c9260c1 100644 --- a/arch/powerpc/perf/callchain.c +++ b/arch/powerpc/perf/callchain.c @@ -47,7 +47,7 @@ static int valid_next_sp(unsigned long sp, unsigned long prev_sp) } void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long sp, next_sp; unsigned long next_ip; @@ -232,7 +232,7 @@ static int sane_signal_64_frame(unsigned long sp) puc == (unsigned long) &sf->uc; } -static void perf_callchain_user_64(struct perf_callchain_entry *entry, +static void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long sp, next_sp; @@ -247,7 +247,7 @@ static void perf_callchain_user_64(struct perf_callchain_entry *entry, sp = regs->gpr[1]; perf_callchain_store(entry, next_ip); - while (entry->nr < sysctl_perf_event_max_stack) { + while (entry->entry->nr < entry->max_stack) { fp = (unsigned long __user *) sp; if (!valid_user_sp(sp, 1) || read_user_stack_64(fp, &next_sp)) return; @@ -319,7 +319,7 @@ static int read_user_stack_32(unsigned int __user *ptr, unsigned int *ret) return rc; } -static inline void perf_callchain_user_64(struct perf_callchain_entry *entry, +static inline void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { } @@ -439,7 +439,7 @@ static unsigned int __user *signal_frame_32_regs(unsigned int sp, return mctx->mc_gregs; } -static void perf_callchain_user_32(struct perf_callchain_entry *entry, +static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned int sp, next_sp; @@ -453,7 +453,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, sp = regs->gpr[1]; perf_callchain_store(entry, next_ip); - while (entry->nr < sysctl_perf_event_max_stack) { + while (entry->entry->nr < entry->max_stack) { fp = (unsigned int __user *) (unsigned long) sp; if (!valid_user_sp(sp, 0) || read_user_stack_32(fp, &next_sp)) return; @@ -487,7 +487,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, } void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { if (current_is_64bit()) perf_callchain_user_64(entry, regs); diff --git a/arch/s390/kernel/perf_event.c b/arch/s390/kernel/perf_event.c index c3e4099..87035fa 100644 --- a/arch/s390/kernel/perf_event.c +++ b/arch/s390/kernel/perf_event.c @@ -224,13 +224,13 @@ arch_initcall(service_level_perf_register); static int __perf_callchain_kernel(void *data, unsigned long address) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; perf_callchain_store(entry, address); return 0; } -void perf_callchain_kernel(struct perf_callchain_entry *entry, +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { if (user_mode(regs)) diff --git a/arch/sh/kernel/perf_callchain.c b/arch/sh/kernel/perf_callchain.c index cc80b61..fa2c0cd 100644 --- a/arch/sh/kernel/perf_callchain.c +++ b/arch/sh/kernel/perf_callchain.c @@ -21,7 +21,7 @@ static int callchain_stack(void *data, char *name) static void callchain_address(void *data, unsigned long addr, int reliable) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; if (reliable) perf_callchain_store(entry, addr); @@ -33,7 +33,7 @@ static const struct stacktrace_ops callchain_ops = { }; void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { perf_callchain_store(entry, regs->pc); diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index a4b8b5a..bcc5376 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -1711,7 +1711,7 @@ static int __init init_hw_perf_events(void) } pure_initcall(init_hw_perf_events); -void perf_callchain_kernel(struct perf_callchain_entry *entry, +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long ksp, fp; @@ -1756,7 +1756,7 @@ void perf_callchain_kernel(struct perf_callchain_entry *entry, } } #endif - } while (entry->nr < sysctl_perf_event_max_stack); + } while (entry->entry->nr < entry->max_stack); } static inline int @@ -1769,7 +1769,7 @@ valid_user_frame(const void __user *fp, unsigned long size) return (__range_not_ok(fp, size, TASK_SIZE) == 0); } -static void perf_callchain_user_64(struct perf_callchain_entry *entry, +static void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long ufp; @@ -1790,10 +1790,10 @@ static void perf_callchain_user_64(struct perf_callchain_entry *entry, pc = sf.callers_pc; ufp = (unsigned long)sf.fp + STACK_BIAS; perf_callchain_store(entry, pc); - } while (entry->nr < sysctl_perf_event_max_stack); + } while (entry->entry->nr < entry->max_stack); } -static void perf_callchain_user_32(struct perf_callchain_entry *entry, +static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { unsigned long ufp; @@ -1822,11 +1822,11 @@ static void perf_callchain_user_32(struct perf_callchain_entry *entry, ufp = (unsigned long)sf.fp; } perf_callchain_store(entry, pc); - } while (entry->nr < sysctl_perf_event_max_stack); + } while (entry->entry->nr < entry->max_stack); } void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { u64 saved_fault_address = current_thread_info()->fault_address; u8 saved_fault_code = get_thread_fault_code(); diff --git a/arch/tile/kernel/perf_event.c b/arch/tile/kernel/perf_event.c index 8767060..6394c1c 100644 --- a/arch/tile/kernel/perf_event.c +++ b/arch/tile/kernel/perf_event.c @@ -941,7 +941,7 @@ arch_initcall(init_hw_perf_events); /* * Tile specific backtracing code for perf_events. */ -static inline void perf_callchain(struct perf_callchain_entry *entry, +static inline void perf_callchain(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct KBacktraceIterator kbt; @@ -992,13 +992,13 @@ static inline void perf_callchain(struct perf_callchain_entry *entry, } } -void perf_callchain_user(struct perf_callchain_entry *entry, +void perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { perf_callchain(entry, regs); } -void perf_callchain_kernel(struct perf_callchain_entry *entry, +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { perf_callchain(entry, regs); diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c index 5e5e76a..07f2b01 100644 --- a/arch/x86/events/core.c +++ b/arch/x86/events/core.c @@ -2202,7 +2202,7 @@ static int backtrace_stack(void *data, char *name) static int backtrace_address(void *data, unsigned long addr, int reliable) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; return perf_callchain_store(entry, addr); } @@ -2214,7 +2214,7 @@ static const struct stacktrace_ops backtrace_ops = { }; void -perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { if (perf_guest_cbs && perf_guest_cbs->is_in_guest()) { /* TODO: We don't support guest os callchain now */ @@ -2268,7 +2268,7 @@ static unsigned long get_segment_base(unsigned int segment) #include static inline int -perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) +perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry_ctx *entry) { /* 32-bit process in 64-bit kernel. */ unsigned long ss_base, cs_base; @@ -2283,7 +2283,7 @@ perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) fp = compat_ptr(ss_base + regs->bp); pagefault_disable(); - while (entry->nr < sysctl_perf_event_max_stack) { + while (entry->entry->nr < entry->max_stack) { unsigned long bytes; frame.next_frame = 0; frame.return_address = 0; @@ -2309,14 +2309,14 @@ perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) } #else static inline int -perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry *entry) +perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry_ctx *entry) { return 0; } #endif void -perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) +perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { struct stack_frame frame; const void __user *fp; @@ -2343,7 +2343,7 @@ perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs) return; pagefault_disable(); - while (entry->nr < sysctl_perf_event_max_stack) { + while (entry->entry->nr < entry->max_stack) { unsigned long bytes; frame.next_frame = NULL; frame.return_address = 0; diff --git a/arch/xtensa/kernel/perf_event.c b/arch/xtensa/kernel/perf_event.c index a6b00b3..ef90479 100644 --- a/arch/xtensa/kernel/perf_event.c +++ b/arch/xtensa/kernel/perf_event.c @@ -323,23 +323,23 @@ static void xtensa_pmu_read(struct perf_event *event) static int callchain_trace(struct stackframe *frame, void *data) { - struct perf_callchain_entry *entry = data; + struct perf_callchain_entry_ctx *entry = data; perf_callchain_store(entry, frame->pc); return 0; } -void perf_callchain_kernel(struct perf_callchain_entry *entry, +void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { - xtensa_backtrace_kernel(regs, sysctl_perf_event_max_stack, + xtensa_backtrace_kernel(regs, entry->max_stack, callchain_trace, NULL, entry); } -void perf_callchain_user(struct perf_callchain_entry *entry, +void perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { - xtensa_backtrace_user(regs, sysctl_perf_event_max_stack, + xtensa_backtrace_user(regs, entry->max_stack, callchain_trace, entry); } diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 9e1c3ad..dbd1824 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -61,6 +61,11 @@ struct perf_callchain_entry { __u64 ip[0]; /* /proc/sys/kernel/perf_event_max_stack */ }; +struct perf_callchain_entry_ctx { + struct perf_callchain_entry *entry; + u32 max_stack; +}; + struct perf_raw_record { u32 size; void *data; @@ -1063,19 +1068,20 @@ extern void perf_event_fork(struct task_struct *tsk); /* Callchains */ DECLARE_PER_CPU(struct perf_callchain_entry, perf_callchain_entry); -extern void perf_callchain_user(struct perf_callchain_entry *entry, struct pt_regs *regs); -extern void perf_callchain_kernel(struct perf_callchain_entry *entry, struct pt_regs *regs); +extern void perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs); +extern void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs); extern struct perf_callchain_entry * get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, - bool crosstask, bool add_mark); + u32 max_stack, bool crosstask, bool add_mark); extern int get_callchain_buffers(void); extern void put_callchain_buffers(void); extern int sysctl_perf_event_max_stack; -static inline int perf_callchain_store(struct perf_callchain_entry *entry, u64 ip) +static inline int perf_callchain_store(struct perf_callchain_entry_ctx *ctx, u64 ip) { - if (entry->nr < sysctl_perf_event_max_stack) { + struct perf_callchain_entry *entry = ctx->entry; + if (entry->nr < ctx->max_stack) { entry->ip[entry->nr++] = ip; return 0; } else { diff --git a/kernel/bpf/stackmap.c b/kernel/bpf/stackmap.c index f5a1954..a82d760 100644 --- a/kernel/bpf/stackmap.c +++ b/kernel/bpf/stackmap.c @@ -136,7 +136,8 @@ static u64 bpf_get_stackid(u64 r1, u64 r2, u64 flags, u64 r4, u64 r5) BPF_F_FAST_STACK_CMP | BPF_F_REUSE_STACKID))) return -EINVAL; - trace = get_perf_callchain(regs, init_nr, kernel, user, false, false); + trace = get_perf_callchain(regs, init_nr, kernel, user, + sysctl_perf_event_max_stack, false, false); if (unlikely(!trace)) /* couldn't fetch the stack trace */ diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c index 7fc8993..af95ad9 100644 --- a/kernel/events/callchain.c +++ b/kernel/events/callchain.c @@ -32,12 +32,12 @@ static DEFINE_MUTEX(callchain_mutex); static struct callchain_cpus_entries *callchain_cpus_entries; -__weak void perf_callchain_kernel(struct perf_callchain_entry *entry, +__weak void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { } -__weak void perf_callchain_user(struct perf_callchain_entry *entry, +__weak void perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs) { } @@ -176,14 +176,15 @@ perf_callchain(struct perf_event *event, struct pt_regs *regs) if (!kernel && !user) return NULL; - return get_perf_callchain(regs, 0, kernel, user, crosstask, true); + return get_perf_callchain(regs, 0, kernel, user, sysctl_perf_event_max_stack, crosstask, true); } struct perf_callchain_entry * get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, - bool crosstask, bool add_mark) + u32 max_stack, bool crosstask, bool add_mark) { struct perf_callchain_entry *entry; + struct perf_callchain_entry_ctx ctx; int rctx; entry = get_callchain_entry(&rctx); @@ -193,12 +194,15 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, if (!entry) goto exit_put; + ctx.entry = entry; + ctx.max_stack = max_stack; + entry->nr = init_nr; if (kernel && !user_mode(regs)) { if (add_mark) - perf_callchain_store(entry, PERF_CONTEXT_KERNEL); - perf_callchain_kernel(entry, regs); + perf_callchain_store(&ctx, PERF_CONTEXT_KERNEL); + perf_callchain_kernel(&ctx, regs); } if (user) { @@ -214,8 +218,8 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, goto exit_put; if (add_mark) - perf_callchain_store(entry, PERF_CONTEXT_USER); - perf_callchain_user(entry, regs); + perf_callchain_store(&ctx, PERF_CONTEXT_USER); + perf_callchain_user(&ctx, regs); } } -- cgit v0.10.2 From 3b1fff08038bd0792b1aa1e9703b2dd0512a3fd0 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 10 May 2016 18:08:32 -0300 Subject: perf core: Add a 'nr' field to perf_event_callchain_context We will use it to count how many addresses are in the entry->ip[] array, excluding PERF_CONTEXT_{KERNEL,USER,etc} entries, so that we can really return the number of entries specified by the user via the relevant sysctl, kernel.perf_event_max_contexts, or via the per event perf_event_attr.sample_max_stack knob. This way we keep the perf_sample->ip_callchain->nr meaning, that is the number of entries, be it real addresses or PERF_CONTEXT_ entries, while honouring the max_stack knobs, i.e. the end result will be max_stack entries if we have at least that many entries in a given stack trace. Cc: David Ahern Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-s8teto51tdqvlfhefndtat9r@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/arch/arm/kernel/perf_callchain.c b/arch/arm/kernel/perf_callchain.c index bc552e8..22bf1f6 100644 --- a/arch/arm/kernel/perf_callchain.c +++ b/arch/arm/kernel/perf_callchain.c @@ -75,7 +75,7 @@ perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs tail = (struct frame_tail __user *)regs->ARM_fp - 1; - while ((entry->entry->nr < entry->max_stack) && + while ((entry->nr < entry->max_stack) && tail && !((unsigned long)tail & 0x3)) tail = user_backtrace(tail, entry); } diff --git a/arch/arm64/kernel/perf_callchain.c b/arch/arm64/kernel/perf_callchain.c index 0d60150..713ca82 100644 --- a/arch/arm64/kernel/perf_callchain.c +++ b/arch/arm64/kernel/perf_callchain.c @@ -122,7 +122,7 @@ void perf_callchain_user(struct perf_callchain_entry_ctx *entry, tail = (struct frame_tail __user *)regs->regs[29]; - while (entry->entry->nr < entry->max_stack && + while (entry->nr < entry->max_stack && tail && !((unsigned long)tail & 0xf)) tail = user_backtrace(tail, entry); } else { @@ -132,7 +132,7 @@ void perf_callchain_user(struct perf_callchain_entry_ctx *entry, tail = (struct compat_frame_tail __user *)regs->compat_fp - 1; - while ((entry->entry->nr < entry->max_stack) && + while ((entry->nr < entry->max_stack) && tail && !((unsigned long)tail & 0x3)) tail = compat_user_backtrace(tail, entry); #endif diff --git a/arch/metag/kernel/perf_callchain.c b/arch/metag/kernel/perf_callchain.c index b3261a9..3e8e048 100644 --- a/arch/metag/kernel/perf_callchain.c +++ b/arch/metag/kernel/perf_callchain.c @@ -65,7 +65,7 @@ perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs --frame; - while ((entry->entry->nr < entry->max_stack) && frame) + while ((entry->nr < entry->max_stack) && frame) frame = user_backtrace(frame, entry); } diff --git a/arch/mips/kernel/perf_event.c b/arch/mips/kernel/perf_event.c index 22395c7..d64056e 100644 --- a/arch/mips/kernel/perf_event.c +++ b/arch/mips/kernel/perf_event.c @@ -35,7 +35,7 @@ static void save_raw_perf_callchain(struct perf_callchain_entry_ctx *entry, addr = *sp++; if (__kernel_text_address(addr)) { perf_callchain_store(entry, addr); - if (entry->entry->nr >= entry->max_stack) + if (entry->nr >= entry->max_stack) break; } } @@ -59,7 +59,7 @@ void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, } do { perf_callchain_store(entry, pc); - if (entry->entry->nr >= entry->max_stack) + if (entry->nr >= entry->max_stack) break; pc = unwind_stack(current, &sp, pc, &ra); } while (pc); diff --git a/arch/powerpc/perf/callchain.c b/arch/powerpc/perf/callchain.c index c9260c1..f68f213 100644 --- a/arch/powerpc/perf/callchain.c +++ b/arch/powerpc/perf/callchain.c @@ -247,7 +247,7 @@ static void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, sp = regs->gpr[1]; perf_callchain_store(entry, next_ip); - while (entry->entry->nr < entry->max_stack) { + while (entry->nr < entry->max_stack) { fp = (unsigned long __user *) sp; if (!valid_user_sp(sp, 1) || read_user_stack_64(fp, &next_sp)) return; @@ -453,7 +453,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, sp = regs->gpr[1]; perf_callchain_store(entry, next_ip); - while (entry->entry->nr < entry->max_stack) { + while (entry->nr < entry->max_stack) { fp = (unsigned int __user *) (unsigned long) sp; if (!valid_user_sp(sp, 0) || read_user_stack_32(fp, &next_sp)) return; diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index bcc5376..710f327 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -1756,7 +1756,7 @@ void perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, } } #endif - } while (entry->entry->nr < entry->max_stack); + } while (entry->nr < entry->max_stack); } static inline int @@ -1790,7 +1790,7 @@ static void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, pc = sf.callers_pc; ufp = (unsigned long)sf.fp + STACK_BIAS; perf_callchain_store(entry, pc); - } while (entry->entry->nr < entry->max_stack); + } while (entry->nr < entry->max_stack); } static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, @@ -1822,7 +1822,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, ufp = (unsigned long)sf.fp; } perf_callchain_store(entry, pc); - } while (entry->entry->nr < entry->max_stack); + } while (entry->nr < entry->max_stack); } void diff --git a/arch/x86/events/core.c b/arch/x86/events/core.c index 07f2b01..5de96a1 100644 --- a/arch/x86/events/core.c +++ b/arch/x86/events/core.c @@ -2283,7 +2283,7 @@ perf_callchain_user32(struct pt_regs *regs, struct perf_callchain_entry_ctx *ent fp = compat_ptr(ss_base + regs->bp); pagefault_disable(); - while (entry->entry->nr < entry->max_stack) { + while (entry->nr < entry->max_stack) { unsigned long bytes; frame.next_frame = 0; frame.return_address = 0; @@ -2343,7 +2343,7 @@ perf_callchain_user(struct perf_callchain_entry_ctx *entry, struct pt_regs *regs return; pagefault_disable(); - while (entry->entry->nr < entry->max_stack) { + while (entry->nr < entry->max_stack) { unsigned long bytes; frame.next_frame = NULL; frame.return_address = 0; diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index dbd1824..3803bb1 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -64,6 +64,7 @@ struct perf_callchain_entry { struct perf_callchain_entry_ctx { struct perf_callchain_entry *entry; u32 max_stack; + u32 nr; }; struct perf_raw_record { @@ -1080,9 +1081,10 @@ extern int sysctl_perf_event_max_stack; static inline int perf_callchain_store(struct perf_callchain_entry_ctx *ctx, u64 ip) { - struct perf_callchain_entry *entry = ctx->entry; - if (entry->nr < ctx->max_stack) { + if (ctx->nr < ctx->max_stack) { + struct perf_callchain_entry *entry = ctx->entry; entry->ip[entry->nr++] = ip; + ++ctx->nr; return 0; } else { return -1; /* no more room, stop walking the stack */ diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c index af95ad9..8774ff8 100644 --- a/kernel/events/callchain.c +++ b/kernel/events/callchain.c @@ -196,8 +196,7 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, ctx.entry = entry; ctx.max_stack = max_stack; - - entry->nr = init_nr; + ctx.nr = entry->nr = init_nr; if (kernel && !user_mode(regs)) { if (add_mark) -- cgit v0.10.2 From 3e4de4ec4cfea40994b47a79767610153edbf45b Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 12 May 2016 13:01:50 -0300 Subject: perf core: Add perf_callchain_store_context() helper We need have different helpers to account how many contexts we have in the sample and for real addresses, so do it now as a prep patch, to ease review. Cc: David Ahern Cc: Frederic Weisbecker Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lkml.kernel.org/n/tip-q964tnyuqrxw5gld18vizs3c@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/arch/powerpc/perf/callchain.c b/arch/powerpc/perf/callchain.c index f68f213..f62597d 100644 --- a/arch/powerpc/perf/callchain.c +++ b/arch/powerpc/perf/callchain.c @@ -76,7 +76,7 @@ perf_callchain_kernel(struct perf_callchain_entry_ctx *entry, struct pt_regs *re next_ip = regs->nip; lr = regs->link; level = 0; - perf_callchain_store(entry, PERF_CONTEXT_KERNEL); + perf_callchain_store_context(entry, PERF_CONTEXT_KERNEL); } else { if (level == 0) @@ -274,7 +274,7 @@ static void perf_callchain_user_64(struct perf_callchain_entry_ctx *entry, read_user_stack_64(&uregs[PT_R1], &sp)) return; level = 0; - perf_callchain_store(entry, PERF_CONTEXT_USER); + perf_callchain_store_context(entry, PERF_CONTEXT_USER); perf_callchain_store(entry, next_ip); continue; } @@ -473,7 +473,7 @@ static void perf_callchain_user_32(struct perf_callchain_entry_ctx *entry, read_user_stack_32(&uregs[PT_R1], &sp)) return; level = 0; - perf_callchain_store(entry, PERF_CONTEXT_USER); + perf_callchain_store_context(entry, PERF_CONTEXT_USER); perf_callchain_store(entry, next_ip); continue; } diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 3803bb1..2024b14 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -1079,6 +1079,8 @@ extern void put_callchain_buffers(void); extern int sysctl_perf_event_max_stack; +#define perf_callchain_store_context(ctx, context) perf_callchain_store(ctx, context) + static inline int perf_callchain_store(struct perf_callchain_entry_ctx *ctx, u64 ip) { if (ctx->nr < ctx->max_stack) { diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c index 8774ff8..ca64573 100644 --- a/kernel/events/callchain.c +++ b/kernel/events/callchain.c @@ -200,7 +200,7 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, if (kernel && !user_mode(regs)) { if (add_mark) - perf_callchain_store(&ctx, PERF_CONTEXT_KERNEL); + perf_callchain_store_context(&ctx, PERF_CONTEXT_KERNEL); perf_callchain_kernel(&ctx, regs); } @@ -217,7 +217,7 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, goto exit_put; if (add_mark) - perf_callchain_store(&ctx, PERF_CONTEXT_USER); + perf_callchain_store_context(&ctx, PERF_CONTEXT_USER); perf_callchain_user(&ctx, regs); } } -- cgit v0.10.2 From c85b03349640b34f3545503c8429fc43005e9a92 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 12 May 2016 13:06:21 -0300 Subject: perf core: Separate accounting of contexts and real addresses in a stack trace The perf_sample->ip_callchain->nr value includes all the entries in the ip_callchain->ip[] array, real addresses and PERF_CONTEXT_{KERNEL,USER,etc}, while what the user expects is that what is in the kernel.perf_event_max_stack sysctl or in the upcoming per event perf_event_attr.sample_max_stack knob be honoured in terms of IP addresses in the stack trace. So allocate a bunch of extra entries for contexts, and do the accounting via perf_callchain_entry_ctx struct members. A new sysctl, kernel.perf_event_max_contexts_per_stack is also introduced for investigating possible bugs in the callchain implementation by some arch. Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Brendan Gregg Cc: David Ahern Cc: Frederic Weisbecker Cc: He Kuang Cc: Jiri Olsa Cc: Masami Hiramatsu Cc: Milian Wolff Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Vince Weaver Cc: Wang Nan Cc: Zefan Li Link: http://lkml.kernel.org/n/tip-3b4wnqk340c4sg4gwkfdi9yk@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/Documentation/sysctl/kernel.txt b/Documentation/sysctl/kernel.txt index daabdd7..a3683ce 100644 --- a/Documentation/sysctl/kernel.txt +++ b/Documentation/sysctl/kernel.txt @@ -61,6 +61,7 @@ show up in /proc/sys/kernel: - perf_cpu_time_max_percent - perf_event_paranoid - perf_event_max_stack +- perf_event_max_contexts_per_stack - pid_max - powersave-nap [ PPC only ] - printk @@ -668,6 +669,19 @@ The default value is 127. ============================================================== +perf_event_max_contexts_per_stack: + +Controls maximum number of stack frame context entries for +(attr.sample_type & PERF_SAMPLE_CALLCHAIN) configured events, for +instance, when using 'perf record -g' or 'perf trace --call-graph fp'. + +This can only be done when no events are in use that have callchains +enabled, otherwise writing to this file will return -EBUSY. + +The default value is 8. + +============================================================== + pid_max: PID allocation wrap value. When the kernel's next PID value diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 2024b14..6b87be9 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -65,6 +65,8 @@ struct perf_callchain_entry_ctx { struct perf_callchain_entry *entry; u32 max_stack; u32 nr; + short contexts; + bool contexts_maxed; }; struct perf_raw_record { @@ -1078,12 +1080,24 @@ extern int get_callchain_buffers(void); extern void put_callchain_buffers(void); extern int sysctl_perf_event_max_stack; +extern int sysctl_perf_event_max_contexts_per_stack; -#define perf_callchain_store_context(ctx, context) perf_callchain_store(ctx, context) +static inline int perf_callchain_store_context(struct perf_callchain_entry_ctx *ctx, u64 ip) +{ + if (ctx->contexts < sysctl_perf_event_max_contexts_per_stack) { + struct perf_callchain_entry *entry = ctx->entry; + entry->ip[entry->nr++] = ip; + ++ctx->contexts; + return 0; + } else { + ctx->contexts_maxed = true; + return -1; /* no more room, stop walking the stack */ + } +} static inline int perf_callchain_store(struct perf_callchain_entry_ctx *ctx, u64 ip) { - if (ctx->nr < ctx->max_stack) { + if (ctx->nr < ctx->max_stack && !ctx->contexts_maxed) { struct perf_callchain_entry *entry = ctx->entry; entry->ip[entry->nr++] = ip; ++ctx->nr; diff --git a/include/uapi/linux/perf_event.h b/include/uapi/linux/perf_event.h index 43fc8d2..36ce552 100644 --- a/include/uapi/linux/perf_event.h +++ b/include/uapi/linux/perf_event.h @@ -862,6 +862,7 @@ enum perf_event_type { }; #define PERF_MAX_STACK_DEPTH 127 +#define PERF_MAX_CONTEXTS_PER_STACK 8 enum perf_callchain_context { PERF_CONTEXT_HV = (__u64)-32, diff --git a/kernel/events/callchain.c b/kernel/events/callchain.c index ca64573..179ef46 100644 --- a/kernel/events/callchain.c +++ b/kernel/events/callchain.c @@ -19,11 +19,13 @@ struct callchain_cpus_entries { }; int sysctl_perf_event_max_stack __read_mostly = PERF_MAX_STACK_DEPTH; +int sysctl_perf_event_max_contexts_per_stack __read_mostly = PERF_MAX_CONTEXTS_PER_STACK; static inline size_t perf_callchain_entry__sizeof(void) { return (sizeof(struct perf_callchain_entry) + - sizeof(__u64) * sysctl_perf_event_max_stack); + sizeof(__u64) * (sysctl_perf_event_max_stack + + sysctl_perf_event_max_contexts_per_stack)); } static DEFINE_PER_CPU(int, callchain_recursion[PERF_NR_CONTEXTS]); @@ -197,6 +199,8 @@ get_perf_callchain(struct pt_regs *regs, u32 init_nr, bool kernel, bool user, ctx.entry = entry; ctx.max_stack = max_stack; ctx.nr = entry->nr = init_nr; + ctx.contexts = 0; + ctx.contexts_maxed = false; if (kernel && !user_mode(regs)) { if (add_mark) @@ -228,6 +232,10 @@ exit_put: return entry; } +/* + * Used for sysctl_perf_event_max_stack and + * sysctl_perf_event_max_contexts_per_stack. + */ int perf_event_max_stack_handler(struct ctl_table *table, int write, void __user *buffer, size_t *lenp, loff_t *ppos) { diff --git a/kernel/sysctl.c b/kernel/sysctl.c index 0ec6907..bec4c11 100644 --- a/kernel/sysctl.c +++ b/kernel/sysctl.c @@ -1156,6 +1156,15 @@ static struct ctl_table kern_table[] = { .extra1 = &zero, .extra2 = &six_hundred_forty_kb, }, + { + .procname = "perf_event_max_contexts_per_stack", + .data = &sysctl_perf_event_max_contexts_per_stack, + .maxlen = sizeof(sysctl_perf_event_max_contexts_per_stack), + .mode = 0644, + .proc_handler = perf_event_max_stack_handler, + .extra1 = &zero, + .extra2 = &one_thousand, + }, #endif #ifdef CONFIG_KMEMCHECK { -- cgit v0.10.2 From a29d5c9b8167dbc21a7ca8c0302e3799f9063b4e Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 16 May 2016 21:16:54 -0300 Subject: perf tools: Separate accounting of contexts and real addresses in a stack trace The perf_sample->ip_callchain->nr value includes all the entries in the ip_callchain->ip[] array, real addresses and PERF_CONTEXT_{KERNEL,USER,etc}, while what the user expects is that what is in the kernel.perf_event_max_stack sysctl or in the upcoming per event perf_event_attr.sample_max_stack knob be honoured in terms of IP addresses in the stack trace. So match the kernel support and validate chain->nr taking into account both kernel.perf_event_max_stack and kernel.perf_event_max_contexts_per_stack. Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexei Starovoitov Cc: Brendan Gregg Cc: David Ahern Cc: Frederic Weisbecker Cc: He Kuang Cc: Jiri Olsa Cc: Masami Hiramatsu Cc: Milian Wolff Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Thomas Gleixner Cc: Vince Weaver Cc: Wang Nan Cc: Zefan Li Link: http://lkml.kernel.org/n/tip-mgx0jpzfdq4uq4abfa40byu0@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo diff --git a/tools/perf/perf.c b/tools/perf/perf.c index 7970008..15982ce 100644 --- a/tools/perf/perf.c +++ b/tools/perf/perf.c @@ -549,6 +549,9 @@ int main(int argc, const char **argv) if (sysctl__read_int("kernel/perf_event_max_stack", &value) == 0) sysctl_perf_event_max_stack = value; + if (sysctl__read_int("kernel/perf_event_max_contexts_per_stack", &value) == 0) + sysctl_perf_event_max_contexts_per_stack = value; + cmd = extract_argv0_path(argv[0]); if (!cmd) cmd = "perf-help"; diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 18dd96b..7ba9fad 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -1811,9 +1811,9 @@ static int thread__resolve_callchain_sample(struct thread *thread, { struct branch_stack *branch = sample->branch_stack; struct ip_callchain *chain = sample->callchain; - int chain_nr = min(max_stack, (int)chain->nr); + int chain_nr = chain->nr; u8 cpumode = PERF_RECORD_MISC_USER; - int i, j, err; + int i, j, err, nr_entries, nr_contexts; int skip_idx = -1; int first_call = 0; @@ -1828,7 +1828,7 @@ static int thread__resolve_callchain_sample(struct thread *thread, * Based on DWARF debug information, some architectures skip * a callchain entry saved by the kernel. */ - if (chain->nr < sysctl_perf_event_max_stack) + if (chain_nr < sysctl_perf_event_max_stack) skip_idx = arch_skip_callchain_idx(thread, chain); /* @@ -1889,12 +1889,8 @@ static int thread__resolve_callchain_sample(struct thread *thread, } check_calls: - if (chain->nr > sysctl_perf_event_max_stack && (int)chain->nr > max_stack) { - pr_warning("corrupted callchain. skipping...\n"); - return 0; - } - - for (i = first_call; i < chain_nr; i++) { + for (i = first_call, nr_entries = 0, nr_contexts = 0; + i < chain_nr && nr_entries < max_stack; i++) { u64 ip; if (callchain_param.order == ORDER_CALLEE) @@ -1908,6 +1904,14 @@ check_calls: #endif ip = chain->ips[j]; + if (ip >= PERF_CONTEXT_MAX) { + if (++nr_contexts > sysctl_perf_event_max_contexts_per_stack) + goto out_corrupted_callchain; + } else { + if (++nr_entries > sysctl_perf_event_max_stack) + goto out_corrupted_callchain; + } + err = add_callchain_ip(thread, cursor, parent, root_al, &cpumode, ip); if (err) @@ -1915,6 +1919,10 @@ check_calls: } return 0; + +out_corrupted_callchain: + pr_warning("corrupted callchain. skipping...\n"); + return 0; } static int unwind_entry(struct unwind_entry *entry, void *arg) diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c index eab077a..23504ad 100644 --- a/tools/perf/util/util.c +++ b/tools/perf/util/util.c @@ -33,7 +33,8 @@ struct callchain_param callchain_param = { unsigned int page_size; int cacheline_size; -unsigned int sysctl_perf_event_max_stack = PERF_MAX_STACK_DEPTH; +int sysctl_perf_event_max_stack = PERF_MAX_STACK_DEPTH; +int sysctl_perf_event_max_contexts_per_stack = PERF_MAX_CONTEXTS_PER_STACK; bool test_attr__enabled; diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h index 7651633..1e8c316 100644 --- a/tools/perf/util/util.h +++ b/tools/perf/util/util.h @@ -261,7 +261,8 @@ void sighandler_dump_stack(int sig); extern unsigned int page_size; extern int cacheline_size; -extern unsigned int sysctl_perf_event_max_stack; +extern int sysctl_perf_event_max_stack; +extern int sysctl_perf_event_max_contexts_per_stack; struct parse_tag { char tag; -- cgit v0.10.2