- 19 Feb, 2023 24 commits
-
-
Ian Rogers authored
Update the jaketown events from 21 to 22. Generation was done using https://github.com/intel/perfmon. Notable changes are improved event descriptions, TMA metrics are updated to version 4.5, TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
Ian Rogers <irogers@google.com> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexandre Torgue <alexandre.torgue@foss.st.com> Cc: Andrii Nakryiko <andrii@kernel.org> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Caleb Biggers <caleb.biggers@intel.com> Cc: Eduard Zingerman <eddyz87@gmail.com> Cc: Florian Fischer <florian.fischer@muhq.space> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jing Zhang <renyu.zj@linux.alibaba.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: John Garry <john.g.garry@oracle.com> Cc: Kajol Jain <kjain@linux.ibm.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Maxime Coquelin <mcoquelin.stm32@gmail.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Perry Taylor <perry.taylor@intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ravi Bangoria <ravi.bangoria@amd.com> Cc: Sandipan Das <sandipan.das@amd.com> Cc: Sean Christopherson <seanjc@google.com> Cc: Stephane Eranian <eranian@google.com> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Xing Zhengjun <zhengjun.xing@linux.intel.com> Cc: linux-arm-kernel@lists.infradead.org Cc: linux-stm32@st-md-mailman.stormreply.com Link: https://lore.kernel.org/r/20230219092848.639226-24-irogers@google.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Ian Rogers authored
Update the ivytown metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the ivybridge metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the icelakex metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, "Sample with" documentation is added to many TMA metrics, smi_cost and transaction metric groups are added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the icelake events from 1.15 to 1.17. Generation was done using https://github.com/intel/perfmon. Notable changes are new events and event descriptions, TMA metrics are updated to version 4.5, TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, smi_cost and transaction metric groups are added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the haswellx metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added and "Sample with" documentation is added to many TMA metrics, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the haswell metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Add version 1.00 of the graniterapids events from https://github.com/intel/perfmon. Signed-off-by:
-
Ian Rogers authored
Update the cascadelakex events from 1.16 to 1.17. Generation was done using https://github.com/intel/perfmon. Notable changes are new events and event descriptions, TMA metrics are updated to version 4.5, TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, "Sample with" documentation is added to many TMA metrics, smi_cost and transaction metric groups are added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the broadwellx metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, "Sample with" documentation is added to many TMA metrics, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the broadwellde metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the broadwell metrics to TMA version 4.5. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the alderlake-n events from 1.16 to 1.18 (no change) and metrics. Generation was done using https://github.com/intel/perfmon. Notable changes are TMA metrics are updated to version 4.5, TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added and the smi_cost metric group is added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Update the alderlake events from 1.16 to 1.18. Generation was done using https://github.com/intel/perfmon. Notable changes are new events and event descriptions, TMA metrics are updated to version 4.5, TMA info metrics are renamed from their node name to be lower case and prefixed by tma_info_, MetricThreshold expressions are added, smi_cost and transaction metric groups are added replicating existing hard coded metrics in stat-shadow. Signed-off-by:
-
Ian Rogers authored
Encoding enums like '1\0' wastes a byte and could be '1' (no '\0' terminator) if the 0 case is '0', it also removes a branch for decompressing. Signed-off-by:
-
Ian Rogers authored
Rename metric_constraint to event_grouping to better explain what the variable is used for. Switch to use an enum for encoding instead of a string. Rather than just no constraint/grouping information or "NO_NMI_WATCHDOG", have 4 enum values. The values encode whether to group or not, and two cases where the behavior is dependent on either the NMI watchdog being enabled or SMT being enabled. Signed-off-by:
-
Ian Rogers authored
Knowing the topology of online CPUs is useful for more than just expr literals. Move to a global function that caches the value. An additional upside is that this may also avoid computing the CPU topology in some situations. Signed-off-by:
-
Ian Rogers authored
Switch to a more natural bool rather than string encoding, where NULL implicitly meant false. The only value of 'PerPkg' in the event json is '1'. Signed-off-by:
-
Ian Rogers authored
Switch to a more natural bool rather than string encoding, where NULL implicitly meant false. Signed-off-by:
-
Ian Rogers authored
Rather than use a string to encode aggr_mode, use an enum value. Signed-off-by:
-
Ian Rogers authored
aggr_mode is used on Power to set a flag for metrics. For pmu_event it is unused. Signed-off-by:
-
Ian Rogers authored
has_constraint implies the NMI_WATCHDOG_CONSTRAINT and if the constraint is detected it causes events not to be grouped. Most of the code cares about whether events are grouped or not, so rename has_constraint to group_events. Also remove group from metricgroup___watchdog_constraint_hint as the warning is specific to a metric. Make the warning message agree with this too. Signed-off-by:
-
Ian Rogers authored
Use the evsel__name accessor as otherwise name may be NULL resulting in a segv. This was observed with the perf stat shell test. Signed-off-by:
-
Namhyung Kim authored
On Fedora 36, the 'perf record' offcpu profiling tests are failing. It was because the BPF checks the prev task's state being S or D but actually it has more bits set. Let's check the LSB 8 bits for the purpose of offcpu profiling. Reported-by:
Arnaldo Carvalho de Melo <acme@kernel.org> Signed-off-by:
Namhyung Kim <namhyung@kernel.org> Acked-by:
-
- 17 Feb, 2023 3 commits
-
-
Kajol Jain authored
Testcase stat_all_metrics.sh fails in powerpc: 98: perf all metrics test : FAILED! Logs with verbose: [command]# ./perf test 98 -vv 98: perf all metrics test : --- start --- test child forked, pid 13262 Testing BRU_STALL_CPI Testing COMPLETION_STALL_CPI ---- Testing TOTAL_LOCAL_NODE_PUMPS_P23 Metric 'TOTAL_LOCAL_NODE_PUMPS_P23' not printed in: Error: Invalid event (hv_24x7/PM_PB_LNS_PUMP23,chip=3/) in per-thread mode, enable system wide with '-a'. Testing TOTAL_LOCAL_NODE_PUMPS_RETRIES_P01 Metric 'TOTAL_LOCAL_NODE_PUMPS_RETRIES_P01' not printed in: Error: Invalid event (hv_24x7/PM_PB_RTY_LNS_PUMP01,chip=3/) in per-thread mode, enable system wide with '-a'. ---- Based on above logs, we could see some of the hv-24x7 metric events fails, and logs suggest to run the metric event with -a option. This change happened after the commit a4b8cfca ("perf stat: Delay metric parsing"), which delayed the metric parsing phase and now before metric parsing phase perf tool identifies, whether target is system-wide or not. With this change, perf_event_open will fails with workload monitoring for uncore events as expected. The perf all metric test case fails as some of the hv-24x7 metric events may need bigger workload with system wide monitoring to get the data. Fix this issue by changing current system wide check from true workload to sleep 0.01 workload. Result with the patch changes in powerpc: 98: perf all metrics test : Ok Fixes: a4b8cfca ("perf stat: Delay metric parsing") Suggested-by:
Athira Rajeev <atrajeev@linux.vnet.ibm.com> Signed-off-by:
Kajol Jain <kjain@linux.ibm.com> Tested-by:
Disha Goel <disgoel@linux.ibm.com> Tested-by:
-
Athira Rajeev authored
Power10 Performance Monitoring Unit (PMU) provides events to understand stall cycles of different pipeline stages. These events along with completed instructions provides useful metrics for application tuning. Patch implements the JSON changes to collect counter statistics to present the high level CPI stall breakdown metrics. New metric group is named as "CPI_STALL_RATIO" and this new metric group presents these stall metrics: - DISPATCHED_CPI ( Dispatch stall cycles per insn ) - ISSUE_STALL_CPI ( Issue stall cycles per insn ) - EXECUTION_STALL_CPI ( Execution stall cycles per insn ) - COMPLETION_STALL_CPI ( Completition stall cycles per insn ) To avoid multipling of events, PM_RUN_INST_CMPL event has been modified to use PMC5(performance monitoring counter5) instead of PMC4. This change is needed, since completion stall event is using PMC4. Usage example: ./perf stat --metric-no-group -M CPI_STALL_RATIO <workload> Performance counter stats for 'workload': 63,056,817,982 PM_CMPL_STALL # 0.28 COMPLETION_STALL_CPI 1,743,988,038,896 PM_ISSUE_STALL # 7.73 ISSUE_STALL_CPI 225,597,495,030 PM_RUN_INST_CMPL # 6.18 DISPATCHED_CPI # 37.48 EXECUTION_STALL_CPI 1,393,916,546,654 PM_DISP_STALL_CYC 8,455,376,836,463 PM_EXEC_STALL "--metric-no-group" is used for forcing PM_RUN_INST_CMPL to be scheduled in all group for more accuracy. Signed-off-by: -
Steinar H. Gunderson authored
There is no good reason why we cannot synthesize "cycle" events from Intel PT just as we can synthesize "instruction" events, in particular when CYC packets are available. This enables using PT to getting much more accurate cycle profiles than regular sampling (record -e cycles) when the work last for very short periods (<10 ms). Thus, add support for this, based off of the existing IPC calculation framework. The new option to --itrace is "y" (for cYcles), as c was taken for calls. Cycle and instruction events can be synthesized together, and are by default. The only real caveat is that CYC packets are only emitted whenever some other packet is, which in practice is when a branch instruction is encountered (and not even all branches). Thus, even at no subsampling (e.g. --itrace=y0ns), it is impossible to get more accuracy than a single basic block, and all cycles spent executing that block will get attributed to the branch instruction that ends the packet. Thus, one cannot know whether the cycles came from e.g. a specific load, a mispredicted branch, or something else. When subsampling (which is the default), the cycle events will get smeared out even more, but will still be generally useful to attribute cycle counts to functions. Reviewed-by:
Adrian Hunter <adrian.hunter@intel.com> Signed-off-by:
Steinar H. Gunderson <sesse@google.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: https://lore.kernel.org/r/20220322082452.1429091-1-sesse@google.comSigned-off-by:
-
- 16 Feb, 2023 1 commit
-
-
Feng Tang authored
Many platforms have feature of adjacent cachelines prefetch, when it is enabled, for data in RAM of 2 cachelines (2N and 2N+1) granularity, if one is fetched to cache, the other one could likely be fetched too, which sort of extends the cacheline size to double, thus the false sharing could happens in adjacent cachelines. 0Day has captured performance changed related with this [1], and some commercial software explicitly makes its hot global variables 128 bytes aligned (2 cache lines) to avoid this kind of extended false sharing. So add an option "--double-cl" for 'perf c2c report' to show false sharing in double cache line granularity, which acts just like the cacheline size is doubled. There is no change to c2c record. The hardware events of shared cacheline are still per cacheline, and this option just changes the granularity of how events are grouped and displayed. In the 'perf c2c report' output below (will-it-scale's 'pagefault2' case on old kernel): ---------------------------------------------------------------------- 26 31 2 0 0 0 0xffff888103ec6000 ---------------------------------------------------------------------- 35.48% 50.00% 0.00% 0.00% 0.00% 0x10 0 1 0xffffffff8133148b 1153 66 971 3748 74 [k] get_mem_cgroup_from_mm 6.45% 0.00% 0.00% 0.00% 0.00% 0x10 0 1 0xffffffff813396e4 570 0 1531 879 75 [k] mem_cgroup_charge 25.81% 50.00% 0.00% 0.00% 0.00% 0x54 0 1 0xffffffff81331472 949 70 593 3359 74 [k] get_mem_cgroup_from_mm 19.35% 0.00% 0.00% 0.00% 0.00% 0x54 0 1 0xffffffff81339686 1352 0 1073 1022 74 [k] mem_cgroup_charge 9.68% 0.00% 0.00% 0.00% 0.00% 0x54 0 1 0xffffffff813396d6 1401 0 863 768 74 [k] mem_cgroup_charge 3.23% 0.00% 0.00% 0.00% 0.00% 0x54 0 1 0xffffffff81333106 618 0 804 11 9 [k] uncharge_batch The offset 0x10 and 0x54 used to displayed in 2 groups, and now they are listed together to give users a hint of extended false sharing. [1]. https://lore.kernel.org/lkml/20201102091543.GM31092@shao2-debian/ Committer notes: Link: https://lore.kernel.org/r/Y+wvVNWqXb70l4uy@feng-clx Removed -a, leaving just as --double-cl, as this probably is not used so frequently and perhaps will be even auto-detected if we manage to record the MSR where this is configured. Reviewed-by:Andi Kleen <ak@linux.intel.com> Reviewed-by:
Leo Yan <leo.yan@linaro.org> Signed-off-by:
Feng Tang <feng.tang@intel.com> Tested-by:
Joe Mario <jmario@redhat.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Tim Chen <tim.c.chen@intel.com> Cc: Xing Zhengjun <zhengjun.xing@linux.intel.com> Link: https://lore.kernel.org/r/20230214075823.246414-1-feng.tang@intel.comSigned-off-by:
-
- 15 Feb, 2023 1 commit
-
-
Yang Jihong authored
When --overwrite and --max-size options of perf record are used together, a segmentation fault occurs. The following is an example: # perf record -e sched:sched* --overwrite --max-size 1K -a -- sleep 1 [ perf record: Woken up 1 times to write data ] perf: Segmentation fault Obtained 12 stack frames. ./perf/perf(+0x197673) [0x55f99710b673] /lib/x86_64-linux-gnu/libc.so.6(+0x3ef0f) [0x7fa45f3cff0f] ./perf/perf(+0x8eb40) [0x55f997002b40] ./perf/perf(+0x1f6882) [0x55f99716a882] ./perf/perf(+0x794c2) [0x55f996fed4c2] ./perf/perf(+0x7b7c7) [0x55f996fef7c7] ./perf/perf(+0x9074b) [0x55f99700474b] ./perf/perf(+0x12e23c) [0x55f9970a223c] ./perf/perf(+0x12e54a) [0x55f9970a254a] ./perf/perf(+0x7db60) [0x55f996ff1b60] /lib/x86_64-linux-gnu/libc.so.6(__libc_start_main+0xe6) [0x7fa45f3b2c86] ./perf/perf(+0x7dfe9) [0x55f996ff1fe9] Segmentation fault (core dumped) backtrace of the core file is as follows: (gdb) bt #0 record__bytes_written (rec=0x55f99755a200 <record>) at builtin-record.c:234 #1 record__output_max_size_exceeded (rec=0x55f99755a200 <record>) at builtin-record.c:242 #2 record__write (map=0x0, size=12816, bf=0x55f9978da2e0, rec=0x55f99755a200 <record>) at builtin-record.c:263 #3 process_synthesized_event (tool=tool@entry=0x55f99755a200 <record>, event=event@entry=0x55f9978da2e0, sample=sample@entry=0x0, machine=machine@entry=0x55f997893658) at builtin-record.c:618 #4 0x000055f99716a883 in __perf_event__synthesize_id_index (tool=tool@entry=0x55f99755a200 <record>, process=process@entry=0x55f997002aa0 <process_synthesized_event>, evlist=0x55f9978928b0, machine=machine@entry=0x55f997893658, from=from@entry=0) at util/synthetic-events.c:1895 #5 0x000055f99716a91f in perf_event__synthesize_id_index (tool=tool@entry=0x55f99755a200 <record>, process=process@entry=0x55f997002aa0 <process_synthesized_event>, evlist=<optimized out>, machine=machine@entry=0x55f997893658) at util/synthetic-events.c:1905 #6 0x000055f996fed4c3 in record__synthesize (tail=tail@entry=true, rec=0x55f99755a200 <record>) at builtin-record.c:1997 #7 0x000055f996fef7c8 in __cmd_record (argc=argc@entry=2, argv=argv@entry=0x7ffc67551260, rec=0x55f99755a200 <record>) at builtin-record.c:2802 #8 0x000055f99700474c in cmd_record (argc=<optimized out>, argv=0x7ffc67551260) at builtin-record.c:4258 #9 0x000055f9970a223d in run_builtin (p=0x55f997564d88 <commands+264>, argc=10, argv=0x7ffc67551260) at perf.c:330 #10 0x000055f9970a254b in handle_internal_command (argc=10, argv=0x7ffc67551260) at perf.c:384 #11 0x000055f996ff1b61 in run_argv (argcp=<synthetic pointer>, argv=<synthetic pointer>) at perf.c:428 #12 main (argc=<optimized out>, argv=0x7ffc67551260) at perf.c:562 The reason is that record__bytes_written accesses the freed memory rec->thread_data, The process is as follows: __cmd_record -> record__free_thread_data -> zfree(&rec->thread_data) // free rec->thread_data -> record__synthesize -> perf_event__synthesize_id_index -> process_synthesized_event -> record__write -> record__bytes_written // access rec->thread_data We add a member variable "thread_bytes_written" in the struct "record" to save the data size written by the threads. Fixes: 6d575816 ("perf record: Add support for limit perf output file size") Signed-off-by:Yang Jihong <yangjihong1@huawei.com> Acked-by:
-
- 09 Feb, 2023 1 commit
-
-
Ian Rogers authored
The added perf_stat_merge_counters combines uncore counters. When metrics are enabled, the counts are merged into a metric_leader via the stat-shadow saved_value logic. As the leader now is passed an aggregated count, it leads to all counters being added together twice and counts appearing approximately doubled in metrics. This change disables the saved_value merging of counts for evsels that are merged. It is recommended that later changes remove the saved_value entirely as the two layers of aggregation in the code is confusing. Fixes: 942c5593 ("perf stat: Add perf_stat_merge_counters()") Reported-by:
Perry Taylor <perry.taylor@intel.com> Signed-off-by:
-
- 08 Feb, 2023 5 commits
-
-
Thomas Richter authored
I have downloaded linux-next and build the perf tool using # make LIBPFM4=1 to have libpfm4 support built into perf. The build fails: # make LIBPFM4=1 .... INSTALL libbpf_headers CC util/pfm.o util/pfm.c: In function ‘print_libpfm_event’: util/pfm.c:189:9: error: too many arguments to function ‘print_cb->print_event’ 189 | print_cb->print_event(print_state, | ^~~~~~~~ util/pfm.c:220:25: error: too many arguments to function ‘print_cb->print_event’ 220 | print_cb->print_event(print_state, The build error is caused by commit d9dc8874 ("perf pmu-events: Remove now unused event and metric variables") which changes the function prototype of struct print_callbacks { ... void (*print_event)(...); --> last two parameters removed. }; but does not adjust the usage of this function prototype in util/pfm.c. In file util/pfm.c function print_event() is still invoked with 13 parameters instead of 11. The compile fails. When I adjust the file util/pfm.c as in this patch, the build works file. Please check this patch for correctness, I have just fixed the compile issue. Fixes: d9dc8874 ("perf pmu-events: Remove now unused event and metric variables") Signed-off-by:Thomas Richter <tmricht@linux.ibm.com> Tested-by:
-
Yicong Yang authored
On aarch64 CPU related events are not under event_source/devices/cpu/events, they're under event_source/devices/armv8_pmuv3_0/events on my machine. Using current auto-complete script will generate below error: [root@localhost bin]# perf stat -e ls: cannot access '/sys/bus/event_source/devices/cpu/events': No such file or directory Fix this by not testing /sys/bus/event_source/devices/cpu/events on aarch64 machine. Fixes: 74cd5815 ("perf tool: Improve bash command line auto-complete for multiple events with comma") Reviewed-by:
James Clark <james.clark@arm.com> Signed-off-by:
Yicong Yang <yangyicong@hisilicon.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Jin Yao <yao.jin@linux.intel.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: linux-arm-kernel@lists.infradead.org Cc: linuxarm@huawei.com Cc: prime.zeng@hisilicon.com Link: https://lore.kernel.org/r/20230207035057.43394-1-yangyicong@huawei.comSigned-off-by:
-
Namhyung Kim authored
The old kernel has a different type of the owner field in rwsem. We can check it using bpf_core_type_matches() builtin in clang but it also needs its own version check since it's available on recent versions. Signed-off-by:
-
Namhyung Kim authored
When there're many lock contentions in the system, people sometimes want to know who caused the contention, IOW who's the owner of the locks. The -o/--lock-owner option tries to follow the lock owners for the contended mutexes and rwsems from BPF, and then attributes the contention time to the owner instead of the waiter. It's a best effort approach to get the owner info at the time of the contention and doesn't guarantee to have the precise tracking of owners if it's changing over time. Currently it only handles mutex and rwsem that have owner field in their struct and it basically points to a task_struct that owns the lock at the moment. Technically its type is atomic_long_t and it comes with some LSB bits used for other meanings. So it needs to clear them when casting it to a pointer to task_struct. Also the atomic_long_t is a typedef of the atomic 32 or 64 bit types depending on arch which is a wrapper struct for the counter value. I'm not aware of proper ways to access those kernel atomic types from BPF so I just read the internal counter value directly. Please let me know if there's a better way. When -o/--lock-owner option is used, it goes to the task aggregation mode like -t/--threads option does. However it cannot get the owner for other lock types like spinlock and sometimes even for mutex. $ sudo ./perf lock con -abo -- ./perf bench sched pipe # Running 'sched/pipe' benchmark: # Executed 1000000 pipe operations between two processes Total time: 4.766 [sec] 4.766540 usecs/op 209795 ops/sec contended total wait max wait avg wait pid owner 403 565.32 us 26.81 us 1.40 us -1 Unknown 4 27.99 us 8.57 us 7.00 us 1583145 sched-pipe 1 8.25 us 8.25 us 8.25 us 1583144 sched-pipe 1 2.03 us 2.03 us 2.03 us 5068 chrome As you can see, the owner is unknown for the most cases. But if we filter only for the mutex locks, it'd more likely get the onwers. $ sudo ./perf lock con -abo -Y mutex -- ./perf bench sched pipe # Running 'sched/pipe' benchmark: # Executed 1000000 pipe operations between two processes Total time: 4.910 [sec] 4.910435 usecs/op 203647 ops/sec contended total wait max wait avg wait pid owner 2 15.50 us 8.29 us 7.75 us 1582852 sched-pipe 7 7.20 us 2.47 us 1.03 us -1 Unknown 1 6.74 us 6.74 us 6.74 us 1582851 sched-pipe Signed-off-by: -
Namhyung Kim authored
The previous change missed to set the con->save_callstack for the LOCK_AGGR_CALLER mode resulting in no caller information. Fixes: ebab2916 ("perf lock contention: Support filters for different aggregation") Signed-off-by:
-
- 06 Feb, 2023 5 commits
-
-
Athira Rajeev authored
Perf BPF filter test fails in environment where "kernel-debuginfo" is not installed. Test failure logs: <<>> 42: BPF filter : 42.1: Basic BPF filtering : Ok 42.2: BPF pinning : Ok 42.3: BPF prologue generation : FAILED! <<>> Enabling verbose option provided debug logs, which says debuginfo needs to be installed. Snippet of verbose logs: <<>> 42.3: BPF prologue generation : --- start --- test child forked, pid 28218 <<>> Rebuild with CONFIG_DEBUG_INFO=y, or install an appropriate debuginfo package. bpf_probe: failed to convert perf probe events Failed to add events selected by BPF test child finished with -1 ---- end ---- BPF filter subtest 3: FAILED! <<>> Here the subtest "BPF prologue generation" failed and logs shows debuginfo is needed. After installing kernel-debuginfo package, testcase passes. The "BPF prologue generation" subtest failed because, the do_test() returns TEST_FAIL without checking the error type returned by parse_events_load_bpf_obj(). parse_events_load_bpf_obj() can also return error of type -ENODATA incase kernel-debuginfo package is not installed. Fix this by adding check for -ENODATA error. Test result after the patch changes: Test failure logs: <<>> 42: BPF filter : 42.1: Basic BPF filtering : Ok 42.2: BPF pinning : Ok 42.3: BPF prologue generation : Skip (clang/debuginfo isn't installed or environment missing BPF support) <<>> Fixes: ba1fae43 ("perf test: Add 'perf test BPF'") Signed-off-by:
-
Athira Rajeev authored
try_to_find_probe_trace_events() uses return error code as ENOENT in two places. First place is after open_debuginfo() when opening debuginfo fails and secondly, after when not finding the probe point. This function is invoked during BPF load and there are other exit points in this code path which returns ENOENT. This makes it difficult to understand the exact reason for exit. Patches changes the exit code from ENOENT to: - ENODATA when it fails to find debuginfo - ENODEV when it fails to find probe point Signed-off-by:
-
Kan Liang authored
The 'perf script' documentation is missing the fields option for Retire Latency. Add it. Signed-off-by:
Kan Liang <kan.liang@linux.intel.com> Cc: Andi Kleen <ak@linux.intel.com> Cc: Ian Rogers <irogers@google.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: https://lore.kernel.org/r/20230206162100.3329395-2-kan.liang@linux.intel.comSigned-off-by:
-
Kan Liang authored
In arch_perf_synthesize_sample_weight(), the retire_lat was mistakenly missed, add it. perf test -v "x86 sample parsing" 74: x86 Sample parsing : --- start --- test child forked, pid 72526 Samples differ at 'retire_lat' parsing failed for sample_type 0x1000000 test child finished with -1 ---- end ---- x86 Sample parsing: FAILED! Reported-by:
-
Kan Liang authored
Add test for the new field for Retire Latency in the X86 specific test. Signed-off-by:
-
