kernel.git/tools/perf/util/evlist.c, branch linux-5.17.y

perf stat: Always keep perf metrics topdown events in a group

2022-06-09T08:26:09Z

[ Upstream commit e8f4f794d7047dd36f090f44f12cd645fba204d2 ] If any member in a group has a different cpu mask than the other members, the current perf stat disables group. when the perf metrics topdown events are part of the group, the below error will be triggered. $ perf stat -e "{slots,topdown-retiring,uncore_imc_free_running_0/dclk/}" -a sleep 1 WARNING: grouped events cpus do not match, disabling group: anon group { slots, topdown-retiring, uncore_imc_free_running_0/dclk/ } Performance counter stats for 'system wide': 141,465,174 slots topdown-retiring 1,605,330,334 uncore_imc_free_running_0/dclk/ The perf metrics topdown events must always be grouped with a slots event as leader. Factor out evsel__remove_from_group() to only remove the regular events from the group. Remove evsel__must_be_in_group(), since no one use it anymore. With the patch, the topdown events aren't broken from the group for the splitting. $ perf stat -e "{slots,topdown-retiring,uncore_imc_free_running_0/dclk/}" -a sleep 1 WARNING: grouped events cpus do not match, disabling group: anon group { slots, topdown-retiring, uncore_imc_free_running_0/dclk/ } Performance counter stats for 'system wide': 346,110,588 slots 124,608,256 topdown-retiring 1,606,869,976 uncore_imc_free_running_0/dclk/ 1.003877592 seconds time elapsed Fixes: a9a1790247bdcf3b ("perf stat: Ensure group is defined on top of the same cpu mask") Signed-off-by: Kan Liang Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Andi Kleen Cc: Ingo Molnar Cc: Jiri Olsa Cc: Jiri Olsa Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Xing Zhengjun Link: https://lore.kernel.org/r/20220518143900.1493980-3-kan.liang@linux.intel.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf evlist: Keep topdown counters in weak group

2022-06-09T08:26:09Z

[ Upstream commit d98079c05b5a5411c6030c47b6256cbeeeff77d0 ] On Intel Icelake, topdown events must always be grouped with a slots event as leader. When a metric is parsed a weak group is formed and retried if perf_event_open fails. The retried events aren't grouped breaking the slots leader requirement. This change modifies the weak group "reset" behavior so that topdown events aren't broken from the group for the retry. $ perf stat -e '{slots,topdown-bad-spec,topdown-be-bound,topdown-fe-bound,topdown-retiring,branch-instructions,branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,mem-loads,mem-stores,ref-cycles,baclears.any,ARITH.DIVIDER_ACTIVE}:W' -a sleep 1 Performance counter stats for 'system wide': 47,867,188,483 slots (92.27%) topdown-bad-spec topdown-be-bound topdown-fe-bound topdown-retiring 2,173,346,937 branch-instructions (92.27%) 10,540,253 branch-misses # 0.48% of all branches (92.29%) 96,291,140 bus-cycles (92.29%) 6,214,202 cache-misses # 20.120 % of all cache refs (92.29%) 30,886,082 cache-references (76.91%) 11,773,726,641 cpu-cycles (84.62%) 11,807,585,307 instructions # 1.00 insn per cycle (92.31%) 0 mem-loads (92.32%) 2,212,928,573 mem-stores (84.69%) 10,024,403,118 ref-cycles (92.35%) 16,232,978 baclears.any (92.35%) 23,832,633 ARITH.DIVIDER_ACTIVE (84.59%) 0.981070734 seconds time elapsed After: $ perf stat -e '{slots,topdown-bad-spec,topdown-be-bound,topdown-fe-bound,topdown-retiring,branch-instructions,branch-misses,bus-cycles,cache-misses,cache-references,cpu-cycles,instructions,mem-loads,mem-stores,ref-cycles,baclears.any,ARITH.DIVIDER_ACTIVE}:W' -a sleep 1 Performance counter stats for 'system wide': 31040189283 slots (92.27%) 8997514811 topdown-bad-spec # 28.2% bad speculation (92.27%) 10997536028 topdown-be-bound # 34.5% backend bound (92.27%) 4778060526 topdown-fe-bound # 15.0% frontend bound (92.27%) 7086628768 topdown-retiring # 22.2% retiring (92.27%) 1417611942 branch-instructions (92.26%) 5285529 branch-misses # 0.37% of all branches (92.28%) 62922469 bus-cycles (92.29%) 1440708 cache-misses # 8.292 % of all cache refs (92.30%) 17374098 cache-references (76.94%) 8040889520 cpu-cycles (84.63%) 7709992319 instructions # 0.96 insn per cycle (92.32%) 0 mem-loads (92.32%) 1515669558 mem-stores (84.68%) 6542411177 ref-cycles (92.35%) 4154149 baclears.any (92.35%) 20556152 ARITH.DIVIDER_ACTIVE (84.59%) 1.010799593 seconds time elapsed Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Andi Kleen Cc: Florian Fischer Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: John Garry Cc: Kim Phillips Cc: Madhavan Srinivasan Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Riccardo Mancini Cc: Shunsuke Nakamura Cc: Stephane Eranian Cc: Xing Zhengjun Link: https://lore.kernel.org/r/20220517052724.283874-2-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo Signed-off-by: Sasha Levin

perf evlist: Avoid iteration for empty evlist.

2022-03-18T21:39:09Z

As seen with 'perf stat --null ..' and reported in: https://lore.kernel.org/lkml/YjCLcpcX2peeQVCH@kernel.org/ v2. Avoids setting evsel in the empty list case as suggested by Jiri Olsa. Committer testing: Before: $ perf stat --null sleep 1 Segmentation fault (core dumped) $ After: $ perf stat --null sleep 1 Performance counter stats for 'sleep 1': 1.010340646 seconds time elapsed 0.001420000 seconds user 0.000000000 seconds sys $ Fixes: 472832d2c000b961 ("perf evlist: Refactor evlist__for_each_cpu()") Reported-by: Arnaldo Carvalho de Melo Signed-off-by: Ian Rogers Tested-by: Arnaldo Carvalho de Melo Acked-by: Jiri Olsa Cc: Adrian Hunter Cc: Ian Rogers Cc: Namhyung Kim Link: https://lore.kernel.org/r/20220317231643.550902-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf evlist: No need to setup affinities when disabling events for pid targets

2022-01-18T12:24:58Z

When the target is a pid, not started by 'perf stat' we need to disable the events, and in that case there is no need to setup affinities as we use a dummy CPU map, with just one entry set to -1. So stop doing it to avoid this needless call to sched_getaffinity(): # strace -ke sched_getaffinity perf stat -e cycles -p 241957 sleep 1 sched_getaffinity(0, 512, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]) = 8 > /usr/lib64/libc-2.33.so(sched_getaffinity@@GLIBC_2.3.4+0x1a) [0xe6eea] > /var/home/acme/bin/perf(affinity__setup+0x6a) [0x532a2a] > /var/home/acme/bin/perf(__evlist__disable.constprop.0+0x27) [0x4b9827] > /var/home/acme/bin/perf(cmd_stat+0x29b5) [0x431725] > /var/home/acme/bin/perf(run_builtin+0x6a) [0x4a2cfa] > /var/home/acme/bin/perf(main+0x612) [0x40f8c2] > /usr/lib64/libc-2.33.so(__libc_start_main+0xd4) [0x27b74] > /var/home/acme/bin/perf(_start+0x2d) [0x40fadd] Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Andi Kleen Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Link: https://lore.kernel.org/r/20220117160931.1191712-5-acme@kernel.org Signed-off-by: Arnaldo Carvalho de Melo

perf evlist: No need to setup affinities when enabling events for pid targets

2022-01-18T12:24:58Z

When the target is a pid, not started by 'perf stat' we need to enable the events, and in that case there is no need to setup affinities as we use a dummy CPU map, with just one entry set to -1. So stop doing it to avoid this needless call to sched_getaffinity(): # strace -ke sched_getaffinity perf stat -e cycles -p 241957 sleep 1 sched_getaffinity(0, 512, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]) = 8 > /usr/lib64/libc-2.33.so(sched_getaffinity@@GLIBC_2.3.4+0x1a) [0xe6eea] > /var/home/acme/bin/perf(affinity__setup+0x6a) [0x5329ca] > /var/home/acme/bin/perf(__evlist__enable.constprop.0+0x23) [0x4b9693] > /var/home/acme/bin/perf(enable_counters+0x14d) [0x42de5d] > /var/home/acme/bin/perf(cmd_stat+0x2358) [0x4310c8] > /var/home/acme/bin/perf(run_builtin+0x6a) [0x4a2cfa] > /var/home/acme/bin/perf(main+0x612) [0x40f8c2] > /usr/lib64/libc-2.33.so(__libc_start_main+0xd4) [0x27b74] > /var/home/acme/bin/perf(_start+0x2d) [0x40fadd] Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Andi Kleen Cc: Ian Rogers Cc: Jiri Olsa Cc: Namhyung Kim Link: https://lore.kernel.org/r/20220117160931.1191712-4-acme@kernel.org Signed-off-by: Arnaldo Carvalho de Melo

perf evlist: No need to do any affinity setup when profiling pids

2022-01-15T20:41:25Z

The cpumap is dummy, so no need to go on figuring out affinity.o This way we reduce the setup time for simple scenarios like: $ perf stat sleep 1 Acked-by: Andi Kleen Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Jiri Olsa Cc: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo

perf cpumap: Give CPUs their own type

2022-01-12T17:28:23Z

A common problem is confusing CPU map indices with the CPU, by wrapping the CPU with a struct then this is avoided. This approach is similar to atomic_t. Committer notes: To make it build with BUILD_BPF_SKEL=1 these files needed the conversions to 'struct perf_cpu' usage: tools/perf/util/bpf_counter.c tools/perf/util/bpf_counter_cgroup.c tools/perf/util/bpf_ftrace.c Also perf_env__get_cpu() was removed back in "perf cpumap: Switch cpu_map__build_map to cpu function". Additionally these needed to be fixed for the ARM builds to complete: tools/perf/arch/arm/util/cs-etm.c tools/perf/arch/arm64/util/pmu.c Suggested-by: John Garry Signed-off-by: Ian Rogers Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kajol Jain Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Mathieu Poirier Cc: Mike Leach Cc: Namhyung Kim Cc: Paul Clarke Cc: Peter Zijlstra Cc: Riccardo Mancini Cc: Stephane Eranian Cc: Suzuki Poulouse Cc: Vineet Singh Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: zhengjun.xing@intel.com Link: https://lore.kernel.org/r/20220105061351.120843-49-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf evlist: Refactor evlist__for_each_cpu()

2022-01-12T17:28:22Z

Previously evlist__for_each_cpu() needed to iterate over the evlist in an inner loop and call "skip" routines. Refactor this so that the iteratr is smarter and the next function can update both the current CPU and evsel. By using a cpu map index, fix apparent off-by-1 in __run_perf_stat's call to perf_evsel__close_cpu(). Signed-off-by: Ian Rogers Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: John Garry Cc: Kajol Jain Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Mathieu Poirier Cc: Mike Leach Cc: Namhyung Kim Cc: Paul Clarke Cc: Peter Zijlstra Cc: Riccardo Mancini Cc: Stephane Eranian Cc: Suzuki Poulouse Cc: Vineet Singh Cc: coresight@lists.linaro.org Cc: linux-arm-kernel@lists.infradead.org Cc: zhengjun.xing@intel.com Link: https://lore.kernel.org/r/20220105061351.120843-35-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf tools: Enable on a list of CPUs for hybrid

2021-08-11T19:07:32Z

The 'perf record' and 'perf stat' commands have supported the option '-C/--cpus' to count or collect only on the list of CPUs provided. This option needs to be supported for hybrid as well. For hybrid support, it needs to check that the cpu list are available on hybrid PMU. One example for AlderLake, cpu0-7 is 'cpu_core', cpu8-11 is 'cpu_atom'. Before: # perf stat -e cpu_core/cycles/ -C11 -- sleep 1 Performance counter stats for 'CPU(s) 11': cpu_core/cycles/ 1.006179431 seconds time elapsed The 'perf stat' command silently returned "" without any helpful information. It should error out pointing out that that cpu11 was not 'cpu_core'. After: # perf stat -e cpu_core/cycles/ -C11 -- sleep 1 WARNING: 11 isn't a 'cpu_core', please use a CPU list in the 'cpu_core' range (0-7) failed to use cpu list 11 We also need to support the events without pmu prefix specified. # perf stat -e cycles -C11 -- sleep 1 WARNING: 11 isn't a 'cpu_core', please use a CPU list in the 'cpu_core' range (0-7) Performance counter stats for 'CPU(s) 11': 1,067,373 cpu_atom/cycles/ 1.005544738 seconds time elapsed The perf tool creates two cycles events automatically, cpu_core/cycles/ and cpu_atom/cycles/. It checks that cpu11 is not 'cpu_core', then shows a warning for cpu_core/cycles/ and only count the cpu_atom/cycles/. If part of cpus are 'cpu_core' and part of cpus are 'cpu_atom', for example, # perf stat -e cycles -C0,11 -- sleep 1 WARNING: use 0 in 'cpu_core' for 'cycles', skip other cpus in list. WARNING: use 11 in 'cpu_atom' for 'cycles', skip other cpus in list. Performance counter stats for 'CPU(s) 0,11': 1,914,704 cpu_core/cycles/ 2,036,983 cpu_atom/cycles/ 1.005815641 seconds time elapsed It now automatically selects cpu0 for cpu_core/cycles/, selects cpu11 for cpu_atom/cycles/, and output with some warnings. Some more complex examples, # perf stat -e cycles,instructions -C0,11 -- sleep 1 WARNING: use 0 in 'cpu_core' for 'cycles', skip other cpus in list. WARNING: use 11 in 'cpu_atom' for 'cycles', skip other cpus in list. WARNING: use 0 in 'cpu_core' for 'instructions', skip other cpus in list. WARNING: use 11 in 'cpu_atom' for 'instructions', skip other cpus in list. Performance counter stats for 'CPU(s) 0,11': 2,780,387 cpu_core/cycles/ 1,583,432 cpu_atom/cycles/ 3,957,277 cpu_core/instructions/ 1,167,089 cpu_atom/instructions/ 1.006005124 seconds time elapsed # perf stat -e cycles,cpu_atom/instructions/ -C0,11 -- sleep 1 WARNING: use 0 in 'cpu_core' for 'cycles', skip other cpus in list. WARNING: use 11 in 'cpu_atom' for 'cycles', skip other cpus in list. WARNING: use 11 in 'cpu_atom' for 'cpu_atom/instructions/', skip other cpus in list. Performance counter stats for 'CPU(s) 0,11': 3,290,301 cpu_core/cycles/ 1,953,073 cpu_atom/cycles/ 1,407,869 cpu_atom/instructions/ 1.006260912 seconds time elapsed Signed-off-by: Jin Yao Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ingo Molnar Cc: Jin Yao Cc: Kan Liang Cc: Peter Zijlstra Link: https //lore.kernel.org/r/20210723063433.7318-4-yao.jin@linux.intel.com Signed-off-by: Arnaldo Carvalho de Melo

perf tools: Create hybrid flag in target

2021-08-11T19:04:33Z

The user may count or collect only on a cpu list via '-C/--cpus' option. Previously cpus for an evsel were retrieved from PMU's sysfs. But if the target cpu list is defined, the retrieved cpus are not kept and the target cpu list is used instead. But for hybrid system, we can't directly use target cpu list. The cpu list may not be available on hybrid pmu (e.g. cpu_core or cpu_atom). So we should not set the 'has_user_cpus' flag for hybrid system. The difficulity is that we can't call perf_pmu__has_hybrid() in evlist.c to check hybrid system otherwise 'perf test python' would be failed (undefined symbol for perf_pmu__has_hybrid). If we add pmu.c to python-ext-sources, too many symbol dependencies are hard to resolve. We use an alternative method by using a new 'hybrid' flag in target for hybrid system checking. Signed-off-by: Jin Yao Acked-by: Jiri Olsa Cc: Alexander Shishkin Cc: Andi Kleen Cc: Ingo Molnar Cc: Jin Yao Cc: Kan Liang Cc: Peter Zijlstra Link: https //lore.kernel.org/r/20210723063433.7318-3-yao.jin@linux.intel.com Signed-off-by: Arnaldo Carvalho de Melo