- 13 Aug, 2024 6 commits
-
-
Ian Rogers authored
Add a for_each callback style API to callchain with sample__for_each_callchain_node(). Possibly in the future such an API can avoid the overhead of constructing the call chain list. Signed-off-by:
Ian Rogers <irogers@google.com> Acked-by:
Namhyung Kim <namhyung@kernel.org> Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Casey Chen <cachen@purestorage.com> Cc: Colin Ian King <colin.i.king@gmail.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: Peter Zijlstra <peterz@infradead.org> Cc: Tom Zanussi <tzanussi@gmail.com> Link: https://lore.kernel.org/r/20240812224119.744968-1-irogers@google.com [ Split from a larger patch that introduced the API and use it ] Signed-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Weilin Wang authored
Intel TPEBS sampling mode is supported through perf record. The counting mode code uses perf record to capture retire_latency value and use it in metric calculation. This test checks the counting mode code on Intel platforms. Committer testing: root@x1:~# perf test tpebs 123: test Intel TPEBS counting mode : Ok root@x1:~# set -o vi root@x1:~# perf test tpebs 123: test Intel TPEBS counting mode : Ok root@x1:~# perf test -v tpebs 123: test Intel TPEBS counting mode : Ok root@x1:~# perf test -vvv tpebs 123: test Intel TPEBS counting mode: --- start --- test child forked, pid 16603 Testing without --record-tpebs Testing with --record-tpebs ---- end(0) ---- 123: test Intel TPEBS counting mode : Ok root@x1:~# Reviewed-by:
Weilin Wang <weilin.wang@intel.com> Acked-by:
-
Weilin Wang authored
TPEBS (Timed PEBS(Precise Event-Based Sampling)) is a new feature Intel PMU from Granite Rapids microarchitecture. It will be used in new TMA (Top-Down Microarchitecture Analysis) releases. Add related introduction to documents while adding new code to support it in 'perf stat'. Reviewed-by:
-
Weilin Wang authored
With this command line option, TPEBS recording is turned off in 'perf stat' on default. It will only be turned on when this option is given in 'perf stat' command. Example with --record-tpebs: perf stat -M tma_split_loads -C1-4 --record-tpebs sleep 1 [ perf record: Woken up 2 times to write data ] [ perf record: Captured and wrote 0.044 MB - ] Performance counter stats for 'CPU(s) 1-4': 53,259,156,071 cpu_core/TOPDOWN.SLOTS/ # 1.6 % tma_split_loads (50.00%) 15,867,565,250 cpu_core/topdown-retiring/ (50.00%) 15,655,580,731 cpu_core/topdown-mem-bound/ (50.00%) 11,738,022,218 cpu_core/topdown-bad-spec/ (50.00%) 6,151,265,424 cpu_core/topdown-fe-bound/ (50.00%) 20,445,917,581 cpu_core/topdown-be-bound/ (50.00%) 6,925,098,013 cpu_core/L1D_PEND_MISS.PENDING/ (50.00%) 3,838,653,421 cpu_core/MEMORY_ACTIVITY.STALLS_L1D_MISS/ (50.00%) 4,797,059,783 cpu_core/EXE_ACTIVITY.BOUND_ON_LOADS/ (50.00%) 11,931,916,714 cpu_core/CPU_CLK_UNHALTED.THREAD/ (50.00%) 102,576,164 cpu_core/MEM_LOAD_COMPLETED.L1_MISS_ANY/ (50.00%) 64,071,854 cpu_core/MEM_INST_RETIRED.SPLIT_LOADS/ (50.00%) 3 cpu_core/MEM_INST_RETIRED.SPLIT_LOADS/R 1.003049679 seconds time elapsed Example without --record-tpebs: perf stat -M tma_contested_accesses -C1 sleep 1 Performance counter stats for 'CPU(s) 1': 50,203,891 cpu_core/TOPDOWN.SLOTS/ # 0.0 % tma_contested_accesses (63.60%) 10,040,777 cpu_core/topdown-retiring/ (63.60%) 6,890,729 cpu_core/topdown-mem-bound/ (63.60%) 2,756,463 cpu_core/topdown-bad-spec/ (63.60%) 10,828,288 cpu_core/topdown-fe-bound/ (63.60%) 28,350,432 cpu_core/topdown-be-bound/ (63.60%) 98 cpu_core/OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HITM/ (63.70%) 577,520 cpu_core/MEMORY_ACTIVITY.STALLS_L2_MISS/ (54.62%) 313,339 cpu_core/MEMORY_ACTIVITY.STALLS_L3_MISS/ (54.62%) 14,155 cpu_core/MEM_LOAD_RETIRED.L1_MISS/ (45.54%) 0 cpu_core/OCR.DEMAND_DATA_RD.L3_HIT.SNOOP_HIT_WITH_FWD/ (36.30%) 8,468,077 cpu_core/CPU_CLK_UNHALTED.THREAD/ (45.38%) 198 cpu_core/MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS/ (45.38%) 8,324 cpu_core/MEM_LOAD_RETIRED.FB_HIT/ (45.38%) 3,388,031,520 TSC 23,226,785 cpu_core/CPU_CLK_UNHALTED.REF_TSC/ (54.46%) 80 cpu_core/MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD/ (54.46%) 0 cpu_core/MEM_LOAD_L3_HIT_RETIRED.XSNP_FWD/R 0 cpu_core/MEM_LOAD_L3_HIT_RETIRED.XSNP_MISS/R 1,006,816,667 ns duration_time 1.002537737 seconds time elapsed Reviewed-by: -
Weilin Wang authored
Add MTL metric JSON file for TMA4.8. Some of the metrics' formulas use TPEBS retire_latency in MTL. This also includes lated E-Core TMA3.6 changes. Reviewed-by:
-
Weilin Wang authored
perf stat: Fork and launch 'perf record' when 'perf stat' needs to get retire latency value for a metric. When retire_latency value is used in a metric formula, evsel would fork a 'perf record' process with "-e" and "-W" options. 'perf record' will collect required retire_latency values in parallel while 'perf stat' is collecting counting values. At the point of time that 'perf stat' stops counting, evsel would stop 'perf record' by sending sigterm signal to 'perf record' process. Sampled data will be processed to get retire latency value. Another thread is required to synchronize between 'perf stat' and 'perf record' when we pass data through pipe. Retire_latency evsel is not opened for 'perf stat' so that there is no counter wasted on it. This commit includes code suggested by Namhyung to adjust reading size for groups that include retire_latency evsels. In current :R parsing implementation, the parser would recognize events with retire_latency modifier and insert them into the evlist like a normal event. Ideally, we need to avoid counting these events. In this commit, at the time when a retire_latency evsel is read, set the retire latency value processed from the sampled data to count value. This sampled retire latency value will be used for metric calculation and final event count print out. No special metric calculation and event print out code required for retire_latency events. Reviewed-by:
-
- 12 Aug, 2024 34 commits
-
-
Weilin Wang authored
When in PIPE mode, allow to use fd dynamically opened and asigned to data->file.fd instead of STDIN_FILENO or STDOUT_FILENO. Reviewed-by:
-
Ian Rogers authored
Retirement latency is a separate sampled count used on newer Intel CPUs. Reviewed-by:
-
Ian Rogers authored
Make tool const now that all uses are const and perf_tool__fill_defaults() won't be used. The aim is to better capture that sessions don't mutate tools. Signed-off-by:
-
Ian Rogers authored
Now all tools are fully initialized prior to use it has no use so remove. Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Ensure tool is initialized to avoid lazy initialization pattern so that more uses of struct perf_tool can be made const. Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Use perf_tool__init() so that more uses of 'struct perf_tool' can be const and not relying on perf_tool__fill_defaults(). Signed-off-by:
-
Ian Rogers authored
Reduce scope of build_id__mark_dso_hit_ops() to the scope of function perf_session__list_build_ids, its only use, and use perf_tool__init() for the default values. Move perf_event__exit_del_thread() to event.[ch] so it can be used in builtin-buildid-list.c. Signed-off-by:
-
Ian Rogers authored
Reduce the scope of the tool from global/static to just that of the cmd_kmem function where the session is scoped. Use the perf_tool__init() to initialize default values. Signed-off-by:
-
Ian Rogers authored
Add init function that behaves like perf_tool__fill_defaults() but assumes all values haven't been initialized. Signed-off-by:
-
Ian Rogers authored
The aim here is to eventually make perf_tool__fill_defaults() an init function so that the tools struct is more const. Create a tool.c to go along with tool.h. Move perf_tool__fill_defaults() out of session.c into tool.c along with the default stub values. Add perf_tool__compressed_is_stub() for a test in perf_session__process_user_event(). perf_session__process_compressed_event() is only used from being default initialized so migrate into tool.c. Signed-off-by:
-
Ian Rogers authored
The tool pointer (to a struct largely of function pointers) is passed around but is unchanged except at initialization. Change parameter and variable types to be const to lower the possibilities of what could happen with a tool. Reviewed-by:
Adrian Hunter <adrian.hunter@intel.com> Signed-off-by:
Leo Yan <leo.yan@arm.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Anshuman Khandual <anshuman.khandual@arm.com> Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com> Cc: Huacai Chen <chenhuacai@kernel.org> Cc: Ilkka Koskinen <ilkka@os.amperecomputing.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Clark <james.clark@arm.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: John Garry <john.g.garry@oracle.com> Cc: Jonathan Cameron <jonathan.cameron@huawei.com> Cc: Kan Liang <kan.liang@linux.intel.com> Cc: Leo Yan <leo.yan@linux.dev> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Mike Leach <mike.leach@linaro.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Nick Desaulniers <ndesaulniers@google.com> Cc: Nick Terrell <terrelln@fb.com> Cc: Oliver Upton <oliver.upton@linux.dev> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Song Liu <song@kernel.org> Cc: Sun Haiyong <sunhaiyong@loongson.cn> Cc: Suzuki Poulouse <suzuki.poulose@arm.com> Cc: Will Deacon <will@kernel.org> Cc: Yanteng Si <siyanteng@loongson.cn> Cc: Yicong Yang <yangyicong@hisilicon.com> Cc: linux-arm-kernel@lists.infradead.org Link: https://lore.kernel.org/r/20240812204720.631678-4-irogers@google.comSigned-off-by:
-
Ian Rogers authored
struct s390_cpumsf_synth was likely cargo culted from other auxtrace examples. It has no users, so remove. Reviewed-by:
-
Ian Rogers authored
Add perf_session__deliver_synth_attr_event that synthesizes a perf_record_header_attr event with one id. Remove use of perf_event__synthesize_attr that necessitates the use of the dummy tool in order to pass the session. Reviewed-by:
-
Ian Rogers authored
The processing of leader samples would turn an individual sample with a group of read values into multiple samples. 'perf inject' would pass through the additional samples increasing the output data file size: $ perf record -g -e "{instructions,cycles}:S" -o perf.orig.data true $ perf script -D -i perf.orig.data | sed -e 's/perf.orig.data/perf.data/g' > orig.txt $ perf inject -i perf.orig.data -o perf.new.data $ perf script -D -i perf.new.data | sed -e 's/perf.new.data/perf.data/g' > new.txt $ diff -u orig.txt new.txt --- orig.txt 2024-07-29 14:29:40.606576769 -0700 +++ new.txt 2024-07-29 14:30:04.142737434 -0700 ... -0xc550@perf.data [0x30]: event: 3 +0xc550@perf.data [0xd0]: event: 9 +. +. ... raw event: size 208 bytes +. 0000: 09 00 00 00 01 00 d0 00 fc 72 01 86 ff ff ff ff .........r...... +. 0010: 74 7d 2c 00 74 7d 2c 00 fb c3 79 f9 ba d5 05 00 t},.t},...y..... +. 0020: e6 cb 1a 00 00 00 00 00 01 00 00 00 00 00 00 00 ................ +. 0030: 02 00 00 00 00 00 00 00 76 01 00 00 00 00 00 00 ........v....... +. 0040: e6 cb 1a 00 00 00 00 00 00 00 00 00 00 00 00 00 ................ +. 0050: 62 18 00 00 00 00 00 00 f6 cb 1a 00 00 00 00 00 b............... +. 0060: 00 00 00 00 00 00 00 00 0c 00 00 00 00 00 00 00 ................ +. 0070: 80 ff ff ff ff ff ff ff fc 72 01 86 ff ff ff ff .........r...... +. 0080: f3 0e 6e 85 ff ff ff ff 0c cb 7f 85 ff ff ff ff ..n............. +. 0090: bc f2 87 85 ff ff ff ff 44 af 7f 85 ff ff ff ff ........D....... +. 00a0: bd be 7f 85 ff ff ff ff 26 d0 7f 85 ff ff ff ff ........&....... +. 00b0: 6d a4 ff 85 ff ff ff ff ea 00 20 86 ff ff ff ff m......... ..... +. 00c0: 00 fe ff ff ff ff ff ff 57 14 4f 43 fc 7e 00 00 ........W.OC.~.. + +1642373909693435 0xc550 [0xd0]: PERF_RECORD_SAMPLE(IP, 0x1): 2915700/2915700: 0xffffffff860172fc period: 1 addr: 0 +... FP chain: nr:12 +..... 0: ffffffffffffff80 +..... 1: ffffffff860172fc +..... 2: ffffffff856e0ef3 +..... 3: ffffffff857fcb0c +..... 4: ffffffff8587f2bc +..... 5: ffffffff857faf44 +..... 6: ffffffff857fbebd +..... 7: ffffffff857fd026 +..... 8: ffffffff85ffa46d +..... 9: ffffffff862000ea +..... 10: fffffffffffffe00 +..... 11: 00007efc434f1457 +... sample_read: +.... group nr 2 +..... id 00000000001acbe6, value 0000000000000176, lost 0 +..... id 00000000001acbf6, value 0000000000001862, lost 0 + +0xc620@perf.data [0x30]: event: 3 ... This behavior is incorrect as in the case above 'perf inject' should have done nothing. Fix this behavior by disabling separating samples for a tool that requests it. Only request this for `perf inject` so as to not affect other perf tools. With the patch and the test above there are no differences between the orig.txt and new.txt. Fixes: e4caec0d ("perf evsel: Add PERF_SAMPLE_READ sample related processing") Signed-off-by: -
Namhyung Kim authored
Now default is to fold everything but it only shows the name of the top-level data type which is not very useful. Instead just expand the top level entry so that it can show the layout at a higher level. Annotate type: 'struct task_struct' (4 samples) Percent Offset Size Field - 100.00 0 9792 struct task_struct { ◆ + 0.50 0 24 struct thread_info thread_info; ▒ 0.00 24 4 unsigned int __state; ▒ 0.00 32 8 void* stack; ▒ + 0.00 40 4 refcount_t usage; ▒ 0.00 44 4 unsigned int flags; ▒ 0.00 48 4 unsigned int ptrace; ▒ 0.00 52 4 int on_cpu; ▒ + 0.00 56 16 struct __call_single_node wake_entry; ▒ 0.00 72 4 unsigned int wakee_flips; ▒ 0.00 80 8 long unsigned int wakee_flip_decay_ts;▒ 0.00 88 8 struct task_struct* last_wakee; ▒ 0.00 96 4 int recent_used_cpu; ▒ 0.00 100 4 int wake_cpu; ▒ 0.00 104 4 int on_rq; ▒ 0.00 108 4 int prio; ▒ 0.00 112 4 int static_prio; ▒ 0.00 116 4 int normal_prio; ▒ 0.00 120 4 unsigned int rt_priority; ▒ + 0.00 128 256 struct sched_entity se; ▒ + 0.00 384 48 struct sched_rt_entity rt; ▒ + 0.00 432 224 struct sched_dl_entity dl; ▒ 0.00 656 8 struct sched_class* sched_class; ▒ ... Committer testing: # perf mem record -a sleep 5s # perf annotate --group --data-type=pthread_mutex_t Annotate type: 'pthread_mutex_t' (13 samples) Percent Offset Size Field - 100.00 0 40 pthread_mutex_t { ▒ - 100.00 0 40 struct __pthread_mutex_s __data { ▒ 39.45 0 4 int __lock; ▒ 0.00 4 4 unsigned int __count; ▒ 7.80 8 4 int __owner; ▒ 6.88 12 4 unsigned int __nusers; ▒ 45.87 16 4 int __kind; ▒ 0.00 20 2 short int __spins; ▒ 0.00 22 2 short int __elision; ▒ + 0.00 24 16 __pthread_list_t __list; ▒ }; ▒ 0.00 0 0 char[] __size; ▒ 39.45 0 8 long int __align; Signed-off-by: -
Namhyung Kim authored
Like 'perf report', use 'e' or 'E' key to toggle folding the current entry so that it can control displaying child entries. Note I didn't add the 'c' and 'C' key to collapse the entry because it's also handled with the 'e'/'E' since it toggles the state. Committer testing: Do some 'perf mem record' for some workload of the whole system, using the target options, as usual (--pid/-p, -C/--cpu, -a for the system wide profiling, etc) and then: # perf annotate --skip-empty --data-type=pthread_mutex_t That, by default, will start as --tui, then press 'E' to see the whole struct unfolded, etc. Signed-off-by:
-
Namhyung Kim authored
Like in the hists browser, it should support folding current entry so that it can hide unwanted details in some data structures. The folded entries will be displayed with the '+' sign, while unfolded entries will have the '-' sign. Entries that have no children will not show any signs. Annotate type: 'struct socket' (1 samples) Percent Offset Size Field - 100.00 0 128 struct socket { ◆ 0.00 0 4 socket_state state; ▒ 0.00 4 2 short int type; ▒ 0.00 8 8 long unsigned int flags; ▒ 0.00 16 8 struct file* file; ▒ 100.00 24 8 struct sock* sk; ▒ 0.00 32 8 struct proto_ops* ops; ▒ - 0.00 64 64 struct socket_wq wq { ▒ - 0.00 64 24 wait_queue_head_t wait { ▒ + 0.00 64 4 spinlock_t lock; ▒ - 0.00 72 16 struct list_head head { ▒ 0.00 72 8 struct list_head* next; ▒ 0.00 80 8 struct list_head* prev; ▒ }; ▒ }; ▒ 0.00 88 8 struct fasync_struct* fasync_list; ▒ 0.00 96 8 long unsigned int flags; ▒ + 0.00 104 16 struct callback_head rcu; ▒ }; ▒ }; ▒ This just adds the display logic for folding, actually folding action will be implemented in the next patch. Signed-off-by: -
Ian Rogers authored
Cache home agent (CHA) events were setting the low rather than high config1 bits. SNR was using CLX CHA events, however its CHA is similar to ICX so remove the events. Incorporate the updates in: https://github.com/intel/perfmon/pull/215 https://github.com/intel/perfmon/pull/216 Fixes: 4cc49942 ("perf vendor events: Update cascadelakex events/metrics") Closes: https://lore.kernel.org/linux-perf-users/CAPhsuW4nem9XZP+b=sJJ7kqXG-cafz0djZf51HsgjCiwkGBA+A@mail.gmail.com/Reported-by:
Song Liu <song@kernel.org> Reviewed-by:
Kan Liang <kan.liang@linux.intel.com> Co-authored-by:
-
