kernel.git/tools/perf/util/intel-tpebs.c, branch master

perf auxtrace: Remove errno.h from auxtrace.h and fix transitive dependencies

2025-11-14T07:03:11Z

errno.h isn't used in auxtrace.h so remove it and fix build failures caused by transitive dependencies through auxtrace.h on errno.h. Signed-off-by: Ian Rogers Reviewed-by: James Clark Signed-off-by: Namhyung Kim

perf tool: Add the perf_tool argument to all callbacks

2025-11-07T21:25:05Z

Getting context for what a tool is doing, such as the perf_inject instance, using container_of the tool is a common pattern in the code. This isn't possible event_op2, event_op3 and event_op4 callbacks as the tool isn't passed. Add the argument and then fix function signatures to match. As tools maybe reading a tool from somewhere else, change that code to use the passed in tool. Signed-off-by: Ian Rogers Signed-off-by: Namhyung Kim

perf sample: Remove arch notion of sample parsing

2025-07-25T17:37:58Z

By definition arch sample parsing and synthesis will inhibit certain kinds of cross-platform record then analysis (report, script, etc.). Remove arch_perf_parse_sample_weight and arch_perf_synthesize_sample_weight replacing with a common implementation. Combine perf_sample p_stage_cyc and retire_lat as weight3 to capture the differing uses regardless of compiled for architecture. Signed-off-by: Ian Rogers Link: https://lore.kernel.org/r/20250724163302.596743-21-irogers@google.com Signed-off-by: Namhyung Kim

perf intel-tpebs: Avoid race when evlist is being deleted

2025-05-28T13:12:47Z

Reading through the evsel->evlist may seg fault if a sample arrives when the evlist is being deleted. Detect this case and ignore samples arriving when the evlist is being deleted. Fixes: bcfab08db7fb38bf ("perf intel-tpebs: Filter non-workload samples") Signed-off-by: Ian Rogers Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alex Gaynor Cc: Alice Ryhl Cc: Andi Kleen Cc: Andreas Hindborg Cc: Benno Lossin Cc: Björn Roy Baron Cc: Boqun Feng Cc: Danilo Krummrich Cc: Dmitriy Vyukov Cc: Gary Guo Cc: Howard Chu Cc: Ingo Molnar Cc: James Clark Cc: Jiapeng Chong Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Miguel Ojeda Cc: Namhyung Kim Cc: Peter Zijlstra Cc: Stephen Brennan Cc: Trevor Gross Cc: Weilin Wang Link: https://lore.kernel.org/r/20250528032637.198960-5-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf intel-tpebs: Filter non-workload samples

2025-05-16T14:31:10Z

If perf is running with a benchmark then we want the retirement latency samples associated with the benchmark rather than from the system as a whole. Use the workload's PID to filter out samples that aren't from the workload or its children. Signed-off-by: Ian Rogers Tested-by: Weilin Wang Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Ingo Molnar Cc: James Clark Cc: Jiri Olsa Cc: Kan Liang Cc: Mark Rutland Cc: Namhyung Kim Cc: Peter Zijlstra Link: https://lore.kernel.org/r/20250430200108.243234-1-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf pmu-events: Add retirement latency to JSON events inside of perf

2025-04-25T15:31:39Z

The updated Intel vendor events add retirement latency for graniterapids: https://lore.kernel.org/lkml/20250322063403.364981-14-irogers@google.com/ This change makes those values available within an alias/event within a PMU and saves them into the evsel at event parse time. When no TPEBS data is available the default values are substituted in for TMA metrics that are using retirement latency events - currently just those on graniterapids. Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Tested-by: Weilin Wang Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Andreas Färber Cc: Caleb Biggers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Manivannan Sadhasivam Cc: Mark Rutland Cc: Maxime Coquelin Cc: Perry Taylor Cc: Peter Zijlstra Cc: Thomas Falcon Link: https://lore.kernel.org/r/20250414174134.3095492-16-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf stat: Add mean, min, max and last --tpebs-mode options

2025-04-25T15:31:36Z

Add command line configuration option for how retirement latency events are combined. The default "mean" gives the average of retirement latency. "min" or "max" give the smallest or largest retirment latency times respectively. "last" uses the last retirment latency sample's time. Committer notes: Enclose parse_tpebs_mode() under HAVE_ARCH_X86_64_SUPPORT to match the ifdef block where it is used, fixing the build in systems like: 20 5.60 debian:experimental-x-mips : FAIL gcc version 14.2.0 (Debian 14.2.0-1) builtin-stat.c:2330:12: error: 'parse_tpebs_mode' defined but not used [-Werror=unused-function] 2330 | static int parse_tpebs_mode(const struct option *opt, const char *str, | ^~~~~~~~~~~~~~~~ Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Tested-by: Weilin Wang Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Andreas Färber Cc: Caleb Biggers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Manivannan Sadhasivam Cc: Mark Rutland Cc: Maxime Coquelin Cc: Perry Taylor Cc: Peter Zijlstra Cc: Thomas Falcon Link: https://lore.kernel.org/r/20250414174134.3095492-15-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf intel-tpebs: Use stats for retirement latency statistics

2025-04-25T15:31:32Z

struct stats provides access to mean, min and max. It also provides uniformity with statistics code used elsewhere in perf. Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Tested-by: Weilin Wang Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Andreas Färber Cc: Caleb Biggers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Manivannan Sadhasivam Cc: Mark Rutland Cc: Maxime Coquelin Cc: Perry Taylor Cc: Peter Zijlstra Cc: Thomas Falcon Link: https://lore.kernel.org/r/20250414174134.3095492-14-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf intel-tpebs: Don't close record on read

2025-04-25T15:31:28Z

Factor sending record control fd code into its own function. Rather than killing the record process send it a ping when reading. Timeouts were witnessed if done too frequently, so only ping for the first tpebs events. Don't kill the record command send it a stop command. As close isn't reliably called also close on evsel__exit. Add extra checks on the record being terminated to avoid warnings. Adjust the locking as needed and incorporate extra -Wthread-safety checks. Check to do six 500ms poll timeouts when sending commands, rather than the larger 3000ms, to allow the record process terminating to be better witnessed. Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Tested-by: Weilin Wang Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Andreas Färber Cc: Caleb Biggers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Manivannan Sadhasivam Cc: Mark Rutland Cc: Maxime Coquelin Cc: Perry Taylor Cc: Peter Zijlstra Cc: Thomas Falcon Link: https://lore.kernel.org/r/20250414174134.3095492-13-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf intel-tpebs: Add mutex for tpebs_results

2025-04-25T15:31:26Z

Ensure sample reader isn't racing with events being added/removed. Reviewed-by: Kan Liang Signed-off-by: Ian Rogers Tested-by: Weilin Wang Acked-by: Namhyung Kim Cc: Adrian Hunter Cc: Alexander Shishkin Cc: Alexandre Torgue Cc: Andreas Färber Cc: Caleb Biggers Cc: Ingo Molnar Cc: Jiri Olsa Cc: Manivannan Sadhasivam Cc: Mark Rutland Cc: Maxime Coquelin Cc: Perry Taylor Cc: Peter Zijlstra Cc: Thomas Falcon Link: https://lore.kernel.org/r/20250414174134.3095492-12-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo