From 301f0203e04293c13372c032198665bd75adf81b Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 29 Dec 2020 15:41:19 -0300 Subject: perf bpf examples: Fix bpf.h header include directive in 5sec.c example It was looking at bpf/bpf.h, which caused this problem: # perf trace -e tools/perf/examples/bpf/5sec.c /home/acme/git/perf/tools/perf/examples/bpf/5sec.c:42:10: fatal error: 'bpf/bpf.h' file not found #include ^~~~~~~~~~~ 1 error generated. ERROR: unable to compile tools/perf/examples/bpf/5sec.c Hint: Check error message shown above. Hint: You can also pre-compile it into .o using: clang -target bpf -O2 -c tools/perf/examples/bpf/5sec.c with proper -I and -D options. event syntax error: 'tools/perf/examples/bpf/5sec.c' \___ Failed to load tools/perf/examples/bpf/5sec.c from source: Error when compiling BPF scriptlet # Change that to plain bpf.h, to make it work again: # perf trace -e tools/perf/examples/bpf/5sec.c sleep 5s 0.000 perf_bpf_probe:hrtimer_nanosleep(__probe_ip: -1776891872, rqtp: 5000000000) # perf trace -e tools/perf/examples/bpf/5sec.c/max-stack=16/ sleep 5s 0.000 perf_bpf_probe:hrtimer_nanosleep(__probe_ip: -1776891872, rqtp: 5000000000) hrtimer_nanosleep ([kernel.kallsyms]) common_nsleep ([kernel.kallsyms]) __x64_sys_clock_nanosleep ([kernel.kallsyms]) do_syscall_64 ([kernel.kallsyms]) entry_SYSCALL_64_after_hwframe ([kernel.kallsyms]) __clock_nanosleep_2 (/usr/lib64/libc-2.32.so) # perf trace -e tools/perf/examples/bpf/5sec.c sleep 4s # Cc: Adrian Hunter Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/examples/bpf/5sec.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools') diff --git a/tools/perf/examples/bpf/5sec.c b/tools/perf/examples/bpf/5sec.c index 65c4ff6892d9..e6b6181c6dc6 100644 --- a/tools/perf/examples/bpf/5sec.c +++ b/tools/perf/examples/bpf/5sec.c @@ -39,7 +39,7 @@ Copyright (C) 2018 Red Hat, Inc., Arnaldo Carvalho de Melo */ -#include +#include #define NSEC_PER_SEC 1000000000L -- cgit v1.2.3 From 38c53947a7dcb6d295769830c9085b0409921ec9 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 14 Jan 2021 12:26:08 -0300 Subject: tools headers UAPI: Sync kvm.h headers with the kernel sources To pick the changes in: 647daca25d24fb6e ("KVM: SVM: Add support for booting APs in an SEV-ES guest") That don't cause any tooling change, just silences this perf build warning: Warning: Kernel ABI header at 'tools/include/uapi/linux/kvm.h' differs from latest version at 'include/uapi/linux/kvm.h' diff -u tools/include/uapi/linux/kvm.h include/uapi/linux/kvm.h Cc: Adrian Hunter Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Cc: Paolo Bonzini Cc: Tom Lendacky Signed-off-by: Arnaldo Carvalho de Melo --- tools/include/uapi/linux/kvm.h | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools') diff --git a/tools/include/uapi/linux/kvm.h b/tools/include/uapi/linux/kvm.h index 886802b8ffba..374c67875cdb 100644 --- a/tools/include/uapi/linux/kvm.h +++ b/tools/include/uapi/linux/kvm.h @@ -251,6 +251,7 @@ struct kvm_hyperv_exit { #define KVM_EXIT_X86_RDMSR 29 #define KVM_EXIT_X86_WRMSR 30 #define KVM_EXIT_DIRTY_RING_FULL 31 +#define KVM_EXIT_AP_RESET_HOLD 32 /* For KVM_EXIT_INTERNAL_ERROR */ /* Emulate instruction failed. */ @@ -573,6 +574,7 @@ struct kvm_vapic_addr { #define KVM_MP_STATE_CHECK_STOP 6 #define KVM_MP_STATE_OPERATING 7 #define KVM_MP_STATE_LOAD 8 +#define KVM_MP_STATE_AP_RESET_HOLD 9 struct kvm_mp_state { __u32 mp_state; -- cgit v1.2.3 From addbdff24293ef772a1b8e5d127b570e70f08cdc Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Thu, 14 Jan 2021 12:35:27 -0300 Subject: tools headers: Syncronize linux/build_bug.h with the kernel sources To pick up the changes in: 3a176b94609a18f5 ("Revert "kbuild: avoid static_assert for genksyms"") And silence this perf build warning: Warning: Kernel ABI header at 'tools/include/linux/build_bug.h' differs from latest version at 'include/linux/build_bug.h' diff -u tools/include/linux/build_bug.h include/linux/build_bug.h Cc: Adrian Hunter Cc: Ian Rogers Cc: Jiri Olsa Cc: Masahiro Yamada Cc: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/include/linux/build_bug.h | 5 ----- 1 file changed, 5 deletions(-) (limited to 'tools') diff --git a/tools/include/linux/build_bug.h b/tools/include/linux/build_bug.h index ce365d212768..cc7070c7439b 100644 --- a/tools/include/linux/build_bug.h +++ b/tools/include/linux/build_bug.h @@ -79,9 +79,4 @@ #define __static_assert(expr, msg, ...) _Static_assert(expr, msg) #endif // static_assert -#ifdef __GENKSYMS__ -/* genksyms gets confused by _Static_assert */ -#define _Static_assert(expr, ...) -#endif - #endif /* _LINUX_BUILD_BUG_H */ -- cgit v1.2.3 From a042a82ddbb3434f523c0671f5301d1fe796b4eb Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Thu, 14 Jan 2021 14:06:09 +0900 Subject: perf test: Fix shadow stat test for non-bash shells It was using some bash-specific features and failed to parse when running with a different shell like below: root@kbl-ppc:~/kbl-ws/perf-dev/lck-9077/acme.tmp/tools/perf# ./perf test 83 -vv 83: perf stat metrics (shadow stat) test : --- start --- test child forked, pid 3922 ./tests/shell/stat+shadow_stat.sh: 19: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 24: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 30: ./tests/shell/stat+shadow_stat.sh: [[: not found (standard_in) 2: syntax error ./tests/shell/stat+shadow_stat.sh: 36: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 19: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 24: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 30: ./tests/shell/stat+shadow_stat.sh: [[: not found (standard_in) 2: syntax error ./tests/shell/stat+shadow_stat.sh: 36: ./tests/shell/stat+shadow_stat.sh: [[: not found ./tests/shell/stat+shadow_stat.sh: 45: ./tests/shell/stat+shadow_stat.sh: declare: not found test child finished with -1 ---- end ---- perf stat metrics (shadow stat) test: FAILED! Reported-by: Jin Yao Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: David Laight Cc: Ian Rogers Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210114050609.1258820-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/tests/shell/stat+shadow_stat.sh | 30 ++++++++++++++---------------- 1 file changed, 14 insertions(+), 16 deletions(-) (limited to 'tools') diff --git a/tools/perf/tests/shell/stat+shadow_stat.sh b/tools/perf/tests/shell/stat+shadow_stat.sh index 249dfe48cf6a..ebebd3596cf9 100755 --- a/tools/perf/tests/shell/stat+shadow_stat.sh +++ b/tools/perf/tests/shell/stat+shadow_stat.sh @@ -9,31 +9,29 @@ perf stat -a true > /dev/null 2>&1 || exit 2 test_global_aggr() { - local cyc - perf stat -a --no-big-num -e cycles,instructions sleep 1 2>&1 | \ grep -e cycles -e instructions | \ while read num evt hash ipc rest do # skip not counted events - if [[ $num == "&1 | \ grep ^CPU | \ while read cpu num evt hash ipc rest do # skip not counted events - if [[ $num == " Date: Thu, 14 Jan 2021 13:23:04 -0800 Subject: libperf tests: Avoid uninitialized variable warning The variable 'bf' is read (for a write call) without being initialized triggering a memory sanitizer warning. Use 'bf' in the read and switch the write to reading from a string. Signed-off-by: Ian Rogers Acked-by: Jiri Olsa Acked-by: Namhyung Kim Cc: Alexander Shishkin Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210114212304.4018119-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/lib/perf/tests/test-evlist.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools') diff --git a/tools/lib/perf/tests/test-evlist.c b/tools/lib/perf/tests/test-evlist.c index 6d8ebe0c2504..1b225fe34a72 100644 --- a/tools/lib/perf/tests/test-evlist.c +++ b/tools/lib/perf/tests/test-evlist.c @@ -208,7 +208,6 @@ static int test_mmap_thread(void) char path[PATH_MAX]; int id, err, pid, go_pipe[2]; union perf_event *event; - char bf; int count = 0; snprintf(path, PATH_MAX, "%s/kernel/debug/tracing/events/syscalls/sys_enter_prctl/id", @@ -229,6 +228,7 @@ static int test_mmap_thread(void) pid = fork(); if (!pid) { int i; + char bf; read(go_pipe[0], &bf, 1); @@ -266,7 +266,7 @@ static int test_mmap_thread(void) perf_evlist__enable(evlist); /* kick the child and wait for it to finish */ - write(go_pipe[1], &bf, 1); + write(go_pipe[1], "A", 1); waitpid(pid, NULL, 0); /* -- cgit v1.2.3 From bba2ea17ef553aea0df80cb64399fe2f70f225dd Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 14 Jan 2021 10:02:49 -0800 Subject: libperf tests: If a test fails return non-zero If a test fails return -1 rather than 0. This is consistent with the return value in test-cpumap.c Signed-off-by: Ian Rogers Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210114180250.3853825-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/lib/perf/tests/test-cpumap.c | 2 +- tools/lib/perf/tests/test-evlist.c | 2 +- tools/lib/perf/tests/test-evsel.c | 2 +- tools/lib/perf/tests/test-threadmap.c | 2 +- 4 files changed, 4 insertions(+), 4 deletions(-) (limited to 'tools') diff --git a/tools/lib/perf/tests/test-cpumap.c b/tools/lib/perf/tests/test-cpumap.c index c8d45091e7c2..c70e9e03af3e 100644 --- a/tools/lib/perf/tests/test-cpumap.c +++ b/tools/lib/perf/tests/test-cpumap.c @@ -27,5 +27,5 @@ int main(int argc, char **argv) perf_cpu_map__put(cpus); __T_END; - return 0; + return tests_failed == 0 ? 0 : -1; } diff --git a/tools/lib/perf/tests/test-evlist.c b/tools/lib/perf/tests/test-evlist.c index 1b225fe34a72..220a95be47c3 100644 --- a/tools/lib/perf/tests/test-evlist.c +++ b/tools/lib/perf/tests/test-evlist.c @@ -409,5 +409,5 @@ int main(int argc, char **argv) test_mmap_cpus(); __T_END; - return 0; + return tests_failed == 0 ? 0 : -1; } diff --git a/tools/lib/perf/tests/test-evsel.c b/tools/lib/perf/tests/test-evsel.c index 135722ac965b..0ad82d7a2a51 100644 --- a/tools/lib/perf/tests/test-evsel.c +++ b/tools/lib/perf/tests/test-evsel.c @@ -131,5 +131,5 @@ int main(int argc, char **argv) test_stat_thread_enable(); __T_END; - return 0; + return tests_failed == 0 ? 0 : -1; } diff --git a/tools/lib/perf/tests/test-threadmap.c b/tools/lib/perf/tests/test-threadmap.c index 7dc4d6fbedde..384471441b48 100644 --- a/tools/lib/perf/tests/test-threadmap.c +++ b/tools/lib/perf/tests/test-threadmap.c @@ -27,5 +27,5 @@ int main(int argc, char **argv) perf_thread_map__put(threads); __T_END; - return 0; + return tests_failed == 0 ? 0 : -1; } -- cgit v1.2.3 From 66dd86b2a2bee129c70f7ff054d3a6a2e5f8eb20 Mon Sep 17 00:00:00 2001 From: Ian Rogers Date: Thu, 14 Jan 2021 10:02:50 -0800 Subject: libperf tests: Fail when failing to get a tracepoint id Permissions are necessary to get a tracepoint id. Fail the test when the read fails. Signed-off-by: Ian Rogers Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210114180250.3853825-2-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/lib/perf/tests/test-evlist.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools') diff --git a/tools/lib/perf/tests/test-evlist.c b/tools/lib/perf/tests/test-evlist.c index 220a95be47c3..e2ac0b7f432e 100644 --- a/tools/lib/perf/tests/test-evlist.c +++ b/tools/lib/perf/tests/test-evlist.c @@ -214,6 +214,7 @@ static int test_mmap_thread(void) sysfs__mountpoint()); if (filename__read_int(path, &id)) { + tests_failed++; fprintf(stderr, "error: failed to get tracepoint id: %s\n", path); return -1; } -- cgit v1.2.3 From 3ff1e7180abc7f6db413933c110df69157216715 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 15 Jan 2021 16:11:38 +0900 Subject: perf stat: Introduce struct runtime_stat_data To pass more info to the saved_value in the runtime_stat, add a new struct runtime_stat_data. Currently it only has 'ctx' field but later patch will add more. Note that we intentionally pass 0 as ctx to clock-related events for compatibility. It was already there in a few places. So move the code into the saved_value_lookup() explicitly and add a comment. Suggested-by: Andi Kleen Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Jin Yao Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210115071139.257042-1-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/stat-shadow.c | 346 +++++++++++++++++++++--------------------- 1 file changed, 173 insertions(+), 173 deletions(-) (limited to 'tools') diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index 901265127e36..a1565b6e38f2 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -114,6 +114,10 @@ static struct saved_value *saved_value_lookup(struct evsel *evsel, rblist = &st->value_list; + /* don't use context info for clock events */ + if (type == STAT_NSECS) + dm.ctx = 0; + nd = rblist__find(rblist, &dm); if (nd) return container_of(nd, struct saved_value, rb_node); @@ -191,12 +195,17 @@ void perf_stat__reset_shadow_per_stat(struct runtime_stat *st) reset_stat(st); } +struct runtime_stat_data { + int ctx; +}; + static void update_runtime_stat(struct runtime_stat *st, enum stat_type type, - int ctx, int cpu, u64 count) + int cpu, u64 count, + struct runtime_stat_data *rsd) { - struct saved_value *v = saved_value_lookup(NULL, cpu, true, - type, ctx, st); + struct saved_value *v = saved_value_lookup(NULL, cpu, true, type, + rsd->ctx, st); if (v) update_stats(&v->stats, count); @@ -210,73 +219,75 @@ static void update_runtime_stat(struct runtime_stat *st, void perf_stat__update_shadow_stats(struct evsel *counter, u64 count, int cpu, struct runtime_stat *st) { - int ctx = evsel_context(counter); u64 count_ns = count; struct saved_value *v; + struct runtime_stat_data rsd = { + .ctx = evsel_context(counter), + }; count *= counter->scale; if (evsel__is_clock(counter)) - update_runtime_stat(st, STAT_NSECS, 0, cpu, count_ns); + update_runtime_stat(st, STAT_NSECS, cpu, count_ns, &rsd); else if (evsel__match(counter, HARDWARE, HW_CPU_CYCLES)) - update_runtime_stat(st, STAT_CYCLES, ctx, cpu, count); + update_runtime_stat(st, STAT_CYCLES, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, CYCLES_IN_TX)) - update_runtime_stat(st, STAT_CYCLES_IN_TX, ctx, cpu, count); + update_runtime_stat(st, STAT_CYCLES_IN_TX, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TRANSACTION_START)) - update_runtime_stat(st, STAT_TRANSACTION, ctx, cpu, count); + update_runtime_stat(st, STAT_TRANSACTION, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, ELISION_START)) - update_runtime_stat(st, STAT_ELISION, ctx, cpu, count); + update_runtime_stat(st, STAT_ELISION, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_TOTAL_SLOTS)) update_runtime_stat(st, STAT_TOPDOWN_TOTAL_SLOTS, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_SLOTS_ISSUED)) update_runtime_stat(st, STAT_TOPDOWN_SLOTS_ISSUED, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_SLOTS_RETIRED)) update_runtime_stat(st, STAT_TOPDOWN_SLOTS_RETIRED, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_FETCH_BUBBLES)) update_runtime_stat(st, STAT_TOPDOWN_FETCH_BUBBLES, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_RECOVERY_BUBBLES)) update_runtime_stat(st, STAT_TOPDOWN_RECOVERY_BUBBLES, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_RETIRING)) update_runtime_stat(st, STAT_TOPDOWN_RETIRING, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_BAD_SPEC)) update_runtime_stat(st, STAT_TOPDOWN_BAD_SPEC, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_FE_BOUND)) update_runtime_stat(st, STAT_TOPDOWN_FE_BOUND, - ctx, cpu, count); + cpu, count, &rsd); else if (perf_stat_evsel__is(counter, TOPDOWN_BE_BOUND)) update_runtime_stat(st, STAT_TOPDOWN_BE_BOUND, - ctx, cpu, count); + cpu, count, &rsd); else if (evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) update_runtime_stat(st, STAT_STALLED_CYCLES_FRONT, - ctx, cpu, count); + cpu, count, &rsd); else if (evsel__match(counter, HARDWARE, HW_STALLED_CYCLES_BACKEND)) update_runtime_stat(st, STAT_STALLED_CYCLES_BACK, - ctx, cpu, count); + cpu, count, &rsd); else if (evsel__match(counter, HARDWARE, HW_BRANCH_INSTRUCTIONS)) - update_runtime_stat(st, STAT_BRANCHES, ctx, cpu, count); + update_runtime_stat(st, STAT_BRANCHES, cpu, count, &rsd); else if (evsel__match(counter, HARDWARE, HW_CACHE_REFERENCES)) - update_runtime_stat(st, STAT_CACHEREFS, ctx, cpu, count); + update_runtime_stat(st, STAT_CACHEREFS, cpu, count, &rsd); else if (evsel__match(counter, HW_CACHE, HW_CACHE_L1D)) - update_runtime_stat(st, STAT_L1_DCACHE, ctx, cpu, count); + update_runtime_stat(st, STAT_L1_DCACHE, cpu, count, &rsd); else if (evsel__match(counter, HW_CACHE, HW_CACHE_L1I)) - update_runtime_stat(st, STAT_L1_ICACHE, ctx, cpu, count); + update_runtime_stat(st, STAT_L1_ICACHE, cpu, count, &rsd); else if (evsel__match(counter, HW_CACHE, HW_CACHE_LL)) - update_runtime_stat(st, STAT_LL_CACHE, ctx, cpu, count); + update_runtime_stat(st, STAT_LL_CACHE, cpu, count, &rsd); else if (evsel__match(counter, HW_CACHE, HW_CACHE_DTLB)) - update_runtime_stat(st, STAT_DTLB_CACHE, ctx, cpu, count); + update_runtime_stat(st, STAT_DTLB_CACHE, cpu, count, &rsd); else if (evsel__match(counter, HW_CACHE, HW_CACHE_ITLB)) - update_runtime_stat(st, STAT_ITLB_CACHE, ctx, cpu, count); + update_runtime_stat(st, STAT_ITLB_CACHE, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, SMI_NUM)) - update_runtime_stat(st, STAT_SMI_NUM, ctx, cpu, count); + update_runtime_stat(st, STAT_SMI_NUM, cpu, count, &rsd); else if (perf_stat_evsel__is(counter, APERF)) - update_runtime_stat(st, STAT_APERF, ctx, cpu, count); + update_runtime_stat(st, STAT_APERF, cpu, count, &rsd); if (counter->collect_stat) { v = saved_value_lookup(counter, cpu, true, STAT_NONE, 0, st); @@ -422,11 +433,12 @@ void perf_stat__collect_metric_expr(struct evlist *evsel_list) } static double runtime_stat_avg(struct runtime_stat *st, - enum stat_type type, int ctx, int cpu) + enum stat_type type, int cpu, + struct runtime_stat_data *rsd) { struct saved_value *v; - v = saved_value_lookup(NULL, cpu, false, type, ctx, st); + v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st); if (!v) return 0.0; @@ -434,11 +446,12 @@ static double runtime_stat_avg(struct runtime_stat *st, } static double runtime_stat_n(struct runtime_stat *st, - enum stat_type type, int ctx, int cpu) + enum stat_type type, int cpu, + struct runtime_stat_data *rsd) { struct saved_value *v; - v = saved_value_lookup(NULL, cpu, false, type, ctx, st); + v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st); if (!v) return 0.0; @@ -446,16 +459,15 @@ static double runtime_stat_n(struct runtime_stat *st, } static void print_stalled_cycles_frontend(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -470,16 +482,15 @@ static void print_stalled_cycles_frontend(struct perf_stat_config *config, } static void print_stalled_cycles_backend(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -490,17 +501,15 @@ static void print_stalled_cycles_backend(struct perf_stat_config *config, } static void print_branch_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_BRANCHES, ctx, cpu); + total = runtime_stat_avg(st, STAT_BRANCHES, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -511,18 +520,15 @@ static void print_branch_misses(struct perf_stat_config *config, } static void print_l1_dcache_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) - + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_L1_DCACHE, ctx, cpu); + total = runtime_stat_avg(st, STAT_L1_DCACHE, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -533,18 +539,15 @@ static void print_l1_dcache_misses(struct perf_stat_config *config, } static void print_l1_icache_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) - + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_L1_ICACHE, ctx, cpu); + total = runtime_stat_avg(st, STAT_L1_ICACHE, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -554,17 +557,15 @@ static void print_l1_icache_misses(struct perf_stat_config *config, } static void print_dtlb_cache_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_DTLB_CACHE, ctx, cpu); + total = runtime_stat_avg(st, STAT_DTLB_CACHE, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -574,17 +575,15 @@ static void print_dtlb_cache_misses(struct perf_stat_config *config, } static void print_itlb_cache_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_ITLB_CACHE, ctx, cpu); + total = runtime_stat_avg(st, STAT_ITLB_CACHE, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -594,17 +593,15 @@ static void print_itlb_cache_misses(struct perf_stat_config *config, } static void print_ll_cache_misses(struct perf_stat_config *config, - int cpu, - struct evsel *evsel, - double avg, + int cpu, double avg, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double total, ratio = 0.0; const char *color; - int ctx = evsel_context(evsel); - total = runtime_stat_avg(st, STAT_LL_CACHE, ctx, cpu); + total = runtime_stat_avg(st, STAT_LL_CACHE, cpu, rsd); if (total) ratio = avg / total * 100.0; @@ -662,56 +659,61 @@ static double sanitize_val(double x) return x; } -static double td_total_slots(int ctx, int cpu, struct runtime_stat *st) +static double td_total_slots(int cpu, struct runtime_stat *st, + struct runtime_stat_data *rsd) { - return runtime_stat_avg(st, STAT_TOPDOWN_TOTAL_SLOTS, ctx, cpu); + return runtime_stat_avg(st, STAT_TOPDOWN_TOTAL_SLOTS, cpu, rsd); } -static double td_bad_spec(int ctx, int cpu, struct runtime_stat *st) +static double td_bad_spec(int cpu, struct runtime_stat *st, + struct runtime_stat_data *rsd) { double bad_spec = 0; double total_slots; double total; - total = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_ISSUED, ctx, cpu) - - runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED, ctx, cpu) + - runtime_stat_avg(st, STAT_TOPDOWN_RECOVERY_BUBBLES, ctx, cpu); + total = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_ISSUED, cpu, rsd) - + runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED, cpu, rsd) + + runtime_stat_avg(st, STAT_TOPDOWN_RECOVERY_BUBBLES, cpu, rsd); - total_slots = td_total_slots(ctx, cpu, st); + total_slots = td_total_slots(cpu, st, rsd); if (total_slots) bad_spec = total / total_slots; return sanitize_val(bad_spec); } -static double td_retiring(int ctx, int cpu, struct runtime_stat *st) +static double td_retiring(int cpu, struct runtime_stat *st, + struct runtime_stat_data *rsd) { double retiring = 0; - double total_slots = td_total_slots(ctx, cpu, st); + double total_slots = td_total_slots(cpu, st, rsd); double ret_slots = runtime_stat_avg(st, STAT_TOPDOWN_SLOTS_RETIRED, - ctx, cpu); + cpu, rsd); if (total_slots) retiring = ret_slots / total_slots; return retiring; } -static double td_fe_bound(int ctx, int cpu, struct runtime_stat *st) +static double td_fe_bound(int cpu, struct runtime_stat *st, + struct runtime_stat_data *rsd) { double fe_bound = 0; - double total_slots = td_total_slots(ctx, cpu, st); + double total_slots = td_total_slots(cpu, st, rsd); double fetch_bub = runtime_stat_avg(st, STAT_TOPDOWN_FETCH_BUBBLES, - ctx, cpu); + cpu, rsd); if (total_slots) fe_bound = fetch_bub / total_slots; return fe_bound; } -static double td_be_bound(int ctx, int cpu, struct runtime_stat *st) +static double td_be_bound(int cpu, struct runtime_stat *st, + struct runtime_stat_data *rsd) { - double sum = (td_fe_bound(ctx, cpu, st) + - td_bad_spec(ctx, cpu, st) + - td_retiring(ctx, cpu, st)); + double sum = (td_fe_bound(cpu, st, rsd) + + td_bad_spec(cpu, st, rsd) + + td_retiring(cpu, st, rsd)); if (sum == 0) return 0; return sanitize_val(1.0 - sum); @@ -722,15 +724,15 @@ static double td_be_bound(int ctx, int cpu, struct runtime_stat *st) * the ratios we need to recreate the sum. */ -static double td_metric_ratio(int ctx, int cpu, - enum stat_type type, - struct runtime_stat *stat) +static double td_metric_ratio(int cpu, enum stat_type type, + struct runtime_stat *stat, + struct runtime_stat_data *rsd) { - double sum = runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, ctx, cpu) + - runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, ctx, cpu) + - runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, ctx, cpu) + - runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, ctx, cpu); - double d = runtime_stat_avg(stat, type, ctx, cpu); + double sum = runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, cpu, rsd) + + runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, cpu, rsd) + + runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, cpu, rsd) + + runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, cpu, rsd); + double d = runtime_stat_avg(stat, type, cpu, rsd); if (sum) return d / sum; @@ -742,34 +744,33 @@ static double td_metric_ratio(int ctx, int cpu, * We allow two missing. */ -static bool full_td(int ctx, int cpu, - struct runtime_stat *stat) +static bool full_td(int cpu, struct runtime_stat *stat, + struct runtime_stat_data *rsd) { int c = 0; - if (runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, ctx, cpu) > 0) + if (runtime_stat_avg(stat, STAT_TOPDOWN_RETIRING, cpu, rsd) > 0) c++; - if (runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, ctx, cpu) > 0) + if (runtime_stat_avg(stat, STAT_TOPDOWN_BE_BOUND, cpu, rsd) > 0) c++; - if (runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, ctx, cpu) > 0) + if (runtime_stat_avg(stat, STAT_TOPDOWN_FE_BOUND, cpu, rsd) > 0) c++; - if (runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, ctx, cpu) > 0) + if (runtime_stat_avg(stat, STAT_TOPDOWN_BAD_SPEC, cpu, rsd) > 0) c++; return c >= 2; } -static void print_smi_cost(struct perf_stat_config *config, - int cpu, struct evsel *evsel, +static void print_smi_cost(struct perf_stat_config *config, int cpu, struct perf_stat_output_ctx *out, - struct runtime_stat *st) + struct runtime_stat *st, + struct runtime_stat_data *rsd) { double smi_num, aperf, cycles, cost = 0.0; - int ctx = evsel_context(evsel); const char *color = NULL; - smi_num = runtime_stat_avg(st, STAT_SMI_NUM, ctx, cpu); - aperf = runtime_stat_avg(st, STAT_APERF, ctx, cpu); - cycles = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); + smi_num = runtime_stat_avg(st, STAT_SMI_NUM, cpu, rsd); + aperf = runtime_stat_avg(st, STAT_APERF, cpu, rsd); + cycles = runtime_stat_avg(st, STAT_CYCLES, cpu, rsd); if ((cycles == 0) || (aperf == 0)) return; @@ -930,12 +931,14 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, print_metric_t print_metric = out->print_metric; double total, ratio = 0.0, total2; const char *color = NULL; - int ctx = evsel_context(evsel); + struct runtime_stat_data rsd = { + .ctx = evsel_context(evsel), + }; struct metric_event *me; int num = 1; if (evsel__match(evsel, HARDWARE, HW_INSTRUCTIONS)) { - total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd); if (total) { ratio = avg / total; @@ -945,12 +948,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, print_metric(config, ctxp, NULL, NULL, "insn per cycle", 0); } - total = runtime_stat_avg(st, STAT_STALLED_CYCLES_FRONT, - ctx, cpu); + total = runtime_stat_avg(st, STAT_STALLED_CYCLES_FRONT, cpu, &rsd); total = max(total, runtime_stat_avg(st, STAT_STALLED_CYCLES_BACK, - ctx, cpu)); + cpu, &rsd)); if (total && avg) { out->new_line(config, ctxp); @@ -960,8 +962,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ratio); } } else if (evsel__match(evsel, HARDWARE, HW_BRANCH_MISSES)) { - if (runtime_stat_n(st, STAT_BRANCHES, ctx, cpu) != 0) - print_branch_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_BRANCHES, cpu, &rsd) != 0) + print_branch_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all branches", 0); } else if ( @@ -970,8 +972,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { - if (runtime_stat_n(st, STAT_L1_DCACHE, ctx, cpu) != 0) - print_l1_dcache_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_L1_DCACHE, cpu, &rsd) != 0) + print_l1_dcache_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all L1-dcache accesses", 0); } else if ( @@ -980,8 +982,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { - if (runtime_stat_n(st, STAT_L1_ICACHE, ctx, cpu) != 0) - print_l1_icache_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_L1_ICACHE, cpu, &rsd) != 0) + print_l1_icache_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all L1-icache accesses", 0); } else if ( @@ -990,8 +992,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { - if (runtime_stat_n(st, STAT_DTLB_CACHE, ctx, cpu) != 0) - print_dtlb_cache_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_DTLB_CACHE, cpu, &rsd) != 0) + print_dtlb_cache_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all dTLB cache accesses", 0); } else if ( @@ -1000,8 +1002,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { - if (runtime_stat_n(st, STAT_ITLB_CACHE, ctx, cpu) != 0) - print_itlb_cache_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_ITLB_CACHE, cpu, &rsd) != 0) + print_itlb_cache_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all iTLB cache accesses", 0); } else if ( @@ -1010,27 +1012,27 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, ((PERF_COUNT_HW_CACHE_OP_READ) << 8) | ((PERF_COUNT_HW_CACHE_RESULT_MISS) << 16))) { - if (runtime_stat_n(st, STAT_LL_CACHE, ctx, cpu) != 0) - print_ll_cache_misses(config, cpu, evsel, avg, out, st); + if (runtime_stat_n(st, STAT_LL_CACHE, cpu, &rsd) != 0) + print_ll_cache_misses(config, cpu, avg, out, st, &rsd); else print_metric(config, ctxp, NULL, NULL, "of all LL-cache accesses", 0); } else if (evsel__match(evsel, HARDWARE, HW_CACHE_MISSES)) { - total = runtime_stat_avg(st, STAT_CACHEREFS, ctx, cpu); + total = runtime_stat_avg(st, STAT_CACHEREFS, cpu, &rsd); if (total) ratio = avg * 100 / total; - if (runtime_stat_n(st, STAT_CACHEREFS, ctx, cpu) != 0) + if (runtime_stat_n(st, STAT_CACHEREFS, cpu, &rsd) != 0) print_metric(config, ctxp, NULL, "%8.3f %%", "of all cache refs", ratio); else print_metric(config, ctxp, NULL, NULL, "of all cache refs", 0); } else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_FRONTEND)) { - print_stalled_cycles_frontend(config, cpu, evsel, avg, out, st); + print_stalled_cycles_frontend(config, cpu, avg, out, st, &rsd); } else if (evsel__match(evsel, HARDWARE, HW_STALLED_CYCLES_BACKEND)) { - print_stalled_cycles_backend(config, cpu, evsel, avg, out, st); + print_stalled_cycles_backend(config, cpu, avg, out, st, &rsd); } else if (evsel__match(evsel, HARDWARE, HW_CPU_CYCLES)) { - total = runtime_stat_avg(st, STAT_NSECS, 0, cpu); + total = runtime_stat_avg(st, STAT_NSECS, cpu, &rsd); if (total) { ratio = avg / total; @@ -1039,7 +1041,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, print_metric(config, ctxp, NULL, NULL, "Ghz", 0); } } else if (perf_stat_evsel__is(evsel, CYCLES_IN_TX)) { - total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd); if (total) print_metric(config, ctxp, NULL, @@ -1049,8 +1051,8 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, print_metric(config, ctxp, NULL, NULL, "transactional cycles", 0); } else if (perf_stat_evsel__is(evsel, CYCLES_IN_TX_CP)) { - total = runtime_stat_avg(st, STAT_CYCLES, ctx, cpu); - total2 = runtime_stat_avg(st, STAT_CYCLES_IN_TX, ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES, cpu, &rsd); + total2 = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd); if (total2 < avg) total2 = avg; @@ -1060,21 +1062,19 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, else print_metric(config, ctxp, NULL, NULL, "aborted cycles", 0); } else if (perf_stat_evsel__is(evsel, TRANSACTION_START)) { - total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, - ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd); if (avg) ratio = total / avg; - if (runtime_stat_n(st, STAT_CYCLES_IN_TX, ctx, cpu) != 0) + if (runtime_stat_n(st, STAT_CYCLES_IN_TX, cpu, &rsd) != 0) print_metric(config, ctxp, NULL, "%8.0f", "cycles / transaction", ratio); else print_metric(config, ctxp, NULL, NULL, "cycles / transaction", 0); } else if (perf_stat_evsel__is(evsel, ELISION_START)) { - total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, - ctx, cpu); + total = runtime_stat_avg(st, STAT_CYCLES_IN_TX, cpu, &rsd); if (avg) ratio = total / avg; @@ -1087,28 +1087,28 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, else print_metric(config, ctxp, NULL, NULL, "CPUs utilized", 0); } else if (perf_stat_evsel__is(evsel, TOPDOWN_FETCH_BUBBLES)) { - double fe_bound = td_fe_bound(ctx, cpu, st); + double fe_bound = td_fe_bound(cpu, st, &rsd); if (fe_bound > 0.2) color = PERF_COLOR_RED; print_metric(config, ctxp, color, "%8.1f%%", "frontend bound", fe_bound * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_SLOTS_RETIRED)) { - double retiring = td_retiring(ctx, cpu, st); + double retiring = td_retiring(cpu, st, &rsd); if (retiring > 0.7) color = PERF_COLOR_GREEN; print_metric(config, ctxp, color, "%8.1f%%", "retiring", retiring * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_RECOVERY_BUBBLES)) { - double bad_spec = td_bad_spec(ctx, cpu, st); + double bad_spec = td_bad_spec(cpu, st, &rsd); if (bad_spec > 0.1) color = PERF_COLOR_RED; print_metric(config, ctxp, color, "%8.1f%%", "bad speculation", bad_spec * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_SLOTS_ISSUED)) { - double be_bound = td_be_bound(ctx, cpu, st); + double be_bound = td_be_bound(cpu, st, &rsd); const char *name = "backend bound"; static int have_recovery_bubbles = -1; @@ -1121,43 +1121,43 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, if (be_bound > 0.2) color = PERF_COLOR_RED; - if (td_total_slots(ctx, cpu, st) > 0) + if (td_total_slots(cpu, st, &rsd) > 0) print_metric(config, ctxp, color, "%8.1f%%", name, be_bound * 100.); else print_metric(config, ctxp, NULL, NULL, name, 0); } else if (perf_stat_evsel__is(evsel, TOPDOWN_RETIRING) && - full_td(ctx, cpu, st)) { - double retiring = td_metric_ratio(ctx, cpu, - STAT_TOPDOWN_RETIRING, st); - + full_td(cpu, st, &rsd)) { + double retiring = td_metric_ratio(cpu, + STAT_TOPDOWN_RETIRING, st, + &rsd); if (retiring > 0.7) color = PERF_COLOR_GREEN; print_metric(config, ctxp, color, "%8.1f%%", "retiring", retiring * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_FE_BOUND) && - full_td(ctx, cpu, st)) { - double fe_bound = td_metric_ratio(ctx, cpu, - STAT_TOPDOWN_FE_BOUND, st); - + full_td(cpu, st, &rsd)) { + double fe_bound = td_metric_ratio(cpu, + STAT_TOPDOWN_FE_BOUND, st, + &rsd); if (fe_bound > 0.2) color = PERF_COLOR_RED; print_metric(config, ctxp, color, "%8.1f%%", "frontend bound", fe_bound * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_BE_BOUND) && - full_td(ctx, cpu, st)) { - double be_bound = td_metric_ratio(ctx, cpu, - STAT_TOPDOWN_BE_BOUND, st); - + full_td(cpu, st, &rsd)) { + double be_bound = td_metric_ratio(cpu, + STAT_TOPDOWN_BE_BOUND, st, + &rsd); if (be_bound > 0.2) color = PERF_COLOR_RED; print_metric(config, ctxp, color, "%8.1f%%", "backend bound", be_bound * 100.); } else if (perf_stat_evsel__is(evsel, TOPDOWN_BAD_SPEC) && - full_td(ctx, cpu, st)) { - double bad_spec = td_metric_ratio(ctx, cpu, - STAT_TOPDOWN_BAD_SPEC, st); - + full_td(cpu, st, &rsd)) { + double bad_spec = td_metric_ratio(cpu, + STAT_TOPDOWN_BAD_SPEC, st, + &rsd); if (bad_spec > 0.1) color = PERF_COLOR_RED; print_metric(config, ctxp, color, "%8.1f%%", "bad speculation", @@ -1165,11 +1165,11 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, } else if (evsel->metric_expr) { generic_metric(config, evsel->metric_expr, evsel->metric_events, NULL, evsel->name, evsel->metric_name, NULL, 1, cpu, out, st); - } else if (runtime_stat_n(st, STAT_NSECS, 0, cpu) != 0) { + } else if (runtime_stat_n(st, STAT_NSECS, cpu, &rsd) != 0) { char unit = 'M'; char unit_buf[10]; - total = runtime_stat_avg(st, STAT_NSECS, 0, cpu); + total = runtime_stat_avg(st, STAT_NSECS, cpu, &rsd); if (total) ratio = 1000.0 * avg / total; @@ -1180,7 +1180,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, snprintf(unit_buf, sizeof(unit_buf), "%c/sec", unit); print_metric(config, ctxp, NULL, "%8.3f", unit_buf, ratio); } else if (perf_stat_evsel__is(evsel, SMI_NUM)) { - print_smi_cost(config, cpu, evsel, out, st); + print_smi_cost(config, cpu, out, st, &rsd); } else { num = 0; } -- cgit v1.2.3 From a1bf23052bdfe30ec3c693cf32feb2d79114ac16 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Fri, 15 Jan 2021 16:11:39 +0900 Subject: perf stat: Take cgroups into account for shadow stats As of now it doesn't consider cgroups when collecting shadow stats and metrics so counter values from different cgroups will be saved in a same slot. This resulted in incorrect numbers when those cgroups have different workloads. For example, let's look at the scenario below: cgroups A and C runs same workload which burns a cpu while cgroup B runs a light workload. $ perf stat -a -e cycles,instructions --for-each-cgroup A,B,C sleep 1 Performance counter stats for 'system wide': 3,958,116,522 cycles A 6,722,650,929 instructions A # 2.53 insn per cycle 1,132,741 cycles B 571,743 instructions B # 0.00 insn per cycle 4,007,799,935 cycles C 6,793,181,523 instructions C # 2.56 insn per cycle 1.001050869 seconds time elapsed When I run 'perf stat' with single workload, it usually shows IPC around 1.7. We can verify it (6,722,650,929.0 / 3,958,116,522 = 1.698) for cgroup A. But in this case, since cgroups are ignored, cycles are averaged so it used the lower value for IPC calculation and resulted in around 2.5. avg cycle: (3958116522 + 1132741 + 4007799935) / 3 = 2655683066 IPC (A) : 6722650929 / 2655683066 = 2.531 IPC (B) : 571743 / 2655683066 = 0.0002 IPC (C) : 6793181523 / 2655683066 = 2.557 We can simply compare cgroup pointers in the evsel and it'll be NULL when cgroups are not specified. With this patch, I can see correct numbers like below: $ perf stat -a -e cycles,instructions --for-each-cgroup A,B,C sleep 1 Performance counter stats for 'system wide': 4,171,051,687 cycles A 7,219,793,922 instructions A # 1.73 insn per cycle 1,051,189 cycles B 583,102 instructions B # 0.55 insn per cycle 4,171,124,710 cycles C 7,192,944,580 instructions C # 1.72 insn per cycle 1.007909814 seconds time elapsed Signed-off-by: Namhyung Kim Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ian Rogers Cc: Jin Yao Cc: Mark Rutland Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lore.kernel.org/lkml/20210115071139.257042-2-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/stat-shadow.c | 26 +++++++++++++++++++------- 1 file changed, 19 insertions(+), 7 deletions(-) (limited to 'tools') diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c index a1565b6e38f2..12eafd12a693 100644 --- a/tools/perf/util/stat-shadow.c +++ b/tools/perf/util/stat-shadow.c @@ -8,6 +8,7 @@ #include "evlist.h" #include "expr.h" #include "metricgroup.h" +#include "cgroup.h" #include /* @@ -28,6 +29,7 @@ struct saved_value { enum stat_type type; int ctx; int cpu; + struct cgroup *cgrp; struct runtime_stat *stat; struct stats stats; u64 metric_total; @@ -57,6 +59,9 @@ static int saved_value_cmp(struct rb_node *rb_node, const void *entry) if (a->ctx != b->ctx) return a->ctx - b->ctx; + if (a->cgrp != b->cgrp) + return (char *)a->cgrp < (char *)b->cgrp ? -1 : +1; + if (a->evsel == NULL && b->evsel == NULL) { if (a->stat == b->stat) return 0; @@ -100,7 +105,8 @@ static struct saved_value *saved_value_lookup(struct evsel *evsel, bool create, enum stat_type type, int ctx, - struct runtime_stat *st) + struct runtime_stat *st, + struct cgroup *cgrp) { struct rblist *rblist; struct rb_node *nd; @@ -110,6 +116,7 @@ static struct saved_value *saved_value_lookup(struct evsel *evsel, .type = type, .ctx = ctx, .stat = st, + .cgrp = cgrp, }; rblist = &st->value_list; @@ -197,6 +204,7 @@ void perf_stat__reset_shadow_per_stat(struct runtime_stat *st) struct runtime_stat_data { int ctx; + struct cgroup *cgrp; }; static void update_runtime_stat(struct runtime_stat *st, @@ -205,7 +213,7 @@ static void update_runtime_stat(struct runtime_stat *st, struct runtime_stat_data *rsd) { struct saved_value *v = saved_value_lookup(NULL, cpu, true, type, - rsd->ctx, st); + rsd->ctx, st, rsd->cgrp); if (v) update_stats(&v->stats, count); @@ -223,6 +231,7 @@ void perf_stat__update_shadow_stats(struct evsel *counter, u64 count, struct saved_value *v; struct runtime_stat_data rsd = { .ctx = evsel_context(counter), + .cgrp = counter->cgrp, }; count *= counter->scale; @@ -290,13 +299,14 @@ void perf_stat__update_shadow_stats(struct evsel *counter, u64 count, update_runtime_stat(st, STAT_APERF, cpu, count, &rsd); if (counter->collect_stat) { - v = saved_value_lookup(counter, cpu, true, STAT_NONE, 0, st); + v = saved_value_lookup(counter, cpu, true, STAT_NONE, 0, st, + rsd.cgrp); update_stats(&v->stats, count); if (counter->metric_leader) v->metric_total += count; } else if (counter->metric_leader) { v = saved_value_lookup(counter->metric_leader, - cpu, true, STAT_NONE, 0, st); + cpu, true, STAT_NONE, 0, st, rsd.cgrp); v->metric_total += count; v->metric_other++; } @@ -438,7 +448,7 @@ static double runtime_stat_avg(struct runtime_stat *st, { struct saved_value *v; - v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st); + v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st, rsd->cgrp); if (!v) return 0.0; @@ -451,7 +461,7 @@ static double runtime_stat_n(struct runtime_stat *st, { struct saved_value *v; - v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st); + v = saved_value_lookup(NULL, cpu, false, type, rsd->ctx, st, rsd->cgrp); if (!v) return 0.0; @@ -805,7 +815,8 @@ static int prepare_metric(struct evsel **metric_events, scale = 1e-9; } else { v = saved_value_lookup(metric_events[i], cpu, false, - STAT_NONE, 0, st); + STAT_NONE, 0, st, + metric_events[i]->cgrp); if (!v) break; stats = &v->stats; @@ -933,6 +944,7 @@ void perf_stat__print_shadow_stats(struct perf_stat_config *config, const char *color = NULL; struct runtime_stat_data rsd = { .ctx = evsel_context(evsel), + .cgrp = evsel->cgrp, }; struct metric_event *me; int num = 1; -- cgit v1.2.3 From 5501e9229a80d95a1ea68609f44c447a75d23ed5 Mon Sep 17 00:00:00 2001 From: Adrian Hunter Date: Thu, 7 Jan 2021 19:41:59 +0200 Subject: perf intel-pt: Fix 'CPU too large' error In some cases, the number of cpus (nr_cpus_online) is confused with the maximum cpu number (nr_cpus_avail), which results in the error in the example below: Example on system with 8 cpus: Before: # echo 0 > /sys/devices/system/cpu/cpu2/online # ./perf record --kcore -e intel_pt// taskset --cpu-list 7 uname Linux [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.147 MB perf.data ] # ./perf script --itrace=e Requested CPU 7 too large. Consider raising MAX_NR_CPUS 0x25908 [0x8]: failed to process type: 68 [Invalid argument] After: # ./perf script --itrace=e # Fixes: 8c7274691f0d ("perf machine: Replace MAX_NR_CPUS with perf_env::nr_cpus_online") Fixes: 7df4e36a4785 ("perf session: Replace MAX_NR_CPUS with perf_env::nr_cpus_online") Signed-off-by: Adrian Hunter Tested-by: Kan Liang Cc: Jiri Olsa Cc: stable@vger.kernel.org Link: http://lore.kernel.org/lkml/20210107174159.24897-1-adrian.hunter@intel.com Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/machine.c | 4 ++-- tools/perf/util/session.c | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'tools') diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index f841f3503cae..1e9d3f982b47 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2980,7 +2980,7 @@ int machines__for_each_thread(struct machines *machines, pid_t machine__get_current_tid(struct machine *machine, int cpu) { - int nr_cpus = min(machine->env->nr_cpus_online, MAX_NR_CPUS); + int nr_cpus = min(machine->env->nr_cpus_avail, MAX_NR_CPUS); if (cpu < 0 || cpu >= nr_cpus || !machine->current_tid) return -1; @@ -2992,7 +2992,7 @@ int machine__set_current_tid(struct machine *machine, int cpu, pid_t pid, pid_t tid) { struct thread *thread; - int nr_cpus = min(machine->env->nr_cpus_online, MAX_NR_CPUS); + int nr_cpus = min(machine->env->nr_cpus_avail, MAX_NR_CPUS); if (cpu < 0) return -EINVAL; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 50ff9795a4f1..25adbcce0281 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -2404,7 +2404,7 @@ int perf_session__cpu_bitmap(struct perf_session *session, { int i, err = -1; struct perf_cpu_map *map; - int nr_cpus = min(session->header.env.nr_cpus_online, MAX_NR_CPUS); + int nr_cpus = min(session->header.env.nr_cpus_avail, MAX_NR_CPUS); for (i = 0; i < PERF_TYPE_MAX; ++i) { struct evsel *evsel; -- cgit v1.2.3 From 648b054a4647cd62e13ba79f398b8b97a7c82b19 Mon Sep 17 00:00:00 2001 From: Al Grant Date: Tue, 24 Nov 2020 19:58:17 +0000 Subject: perf inject: Correct event attribute sizes When 'perf inject' reads a perf.data file from an older version of perf, it writes event attributes into the output with the original size field, but lays them out as if they had the size currently used. Readers see a corrupt file. Update the size field to match the layout. Signed-off-by: Al Grant Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Link: http://lore.kernel.org/lkml/20201124195818.30603-1-al.grant@arm.com Signed-off-by: Denis Nikitin Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'tools') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 062383e225a3..c4ed3dc2c8f4 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -3323,6 +3323,14 @@ int perf_session__write_header(struct perf_session *session, attr_offset = lseek(ff.fd, 0, SEEK_CUR); evlist__for_each_entry(evlist, evsel) { + if (evsel->core.attr.size < sizeof(evsel->core.attr)) { + /* + * We are likely in "perf inject" and have read + * from an older file. Update attr size so that + * reader gets the right offset to the ids. + */ + evsel->core.attr.size = sizeof(evsel->core.attr); + } f_attr = (struct perf_file_attr){ .attr = evsel->core.attr, .ids = { -- cgit v1.2.3