pm24.git/tools/perf/pmu-events, branch master

perf pmu: Move pmu_metrics_table__find and remove ARM override

2024-11-16T19:42:36Z

Move pmu_metrics_table__find() to the jevents.py generated pmu-events.c and remove indirection override for ARM. The movement removes perf_pmu__find_metrics_table that exists to enable the ARM override. The ARM override isn't necessary as just the CPUID, not PMU, is used in the metric table lookup. On non-ARM the CPU argument is just ignored for the CPUID, for ARM -1 is passed so that the CPUID for the first logical CPU is read. Reviewed-by: James Clark Signed-off-by: Ian Rogers Tested-by: Xu Yang Cc: Adrian Hunter Cc: Albert Ou Cc: Alexander Shishkin Cc: Alexandre Ghiti Cc: Athira Rajeev Cc: Ben Zong-You Xie Cc: Benjamin Gray Cc: Bibo Mao Cc: Clément Le Goffic Cc: Dima Kogan Cc: Dr. David Alan Gilbert Cc: Huacai Chen Cc: Ingo Molnar Cc: Jiri Olsa Cc: John Garry Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Masami Hiramatsu Cc: Mike Leach Cc: Namhyung Kim Cc: Palmer Dabbelt Cc: Paul Walmsley Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Cc: Will Deacon Cc: Yicong Yang Cc: linux-arm-kernel@lists.infradead.org Cc: linux-riscv@lists.infradead.org Link: https://lore.kernel.org/r/20241107162035.52206-9-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf jevents: Add map_for_cpu()

2024-11-16T19:41:42Z

The PMU is no longer part of the map finding process and for metrics doesn't make sense as they lack a PMU. Reviewed-by: James Clark Signed-off-by: Ian Rogers Tested-by: Xu Yang Cc: Adrian Hunter Cc: Albert Ou Cc: Alexander Shishkin Cc: Alexandre Ghiti Cc: Athira Rajeev Cc: Ben Zong-You Xie Cc: Benjamin Gray Cc: Bibo Mao Cc: Clément Le Goffic Cc: Dima Kogan Cc: Dr. David Alan Gilbert Cc: Huacai Chen Cc: Ingo Molnar Cc: Jiri Olsa Cc: John Garry Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Masami Hiramatsu Cc: Mike Leach Cc: Namhyung Kim Cc: Palmer Dabbelt Cc: Paul Walmsley Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Cc: Will Deacon Cc: Yicong Yang Cc: linux-arm-kernel@lists.infradead.org Cc: linux-riscv@lists.infradead.org Link: https://lore.kernel.org/r/20241107162035.52206-8-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf header: Pass a perf_cpu rather than a PMU to get_cpuid_str

2024-11-16T19:40:30Z

On ARM the cpuid is dependent on the core type of the CPU in question. The PMU was passed for the sake of the CPU map but this means in places a temporary PMU is created just to pass a CPU value. Just pass the CPU and fix up the callers. As there are no longer PMU users in header.h, shuffle forward declarations earlier to work around build failures. Reviewed-by: James Clark Signed-off-by: Ian Rogers Tested-by: Xu Yang Cc: Adrian Hunter Cc: Albert Ou Cc: Alexander Shishkin Cc: Alexandre Ghiti Cc: Athira Rajeev Cc: Ben Zong-You Xie Cc: Benjamin Gray Cc: Bibo Mao Cc: Clément Le Goffic Cc: Dima Kogan Cc: Dr. David Alan Gilbert Cc: Huacai Chen Cc: Ingo Molnar Cc: Jiri Olsa Cc: John Garry Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Masami Hiramatsu Cc: Mike Leach Cc: Namhyung Kim Cc: Palmer Dabbelt Cc: Paul Walmsley Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Cc: Will Deacon Cc: Yicong Yang Cc: linux-arm-kernel@lists.infradead.org Cc: linux-riscv@lists.infradead.org Link: https://lore.kernel.org/r/20241107162035.52206-7-irogers@google.com Signed-off-by: Arnaldo Carvalho de Melo

perf jevents: fix breakage when do perf stat on system metric

2024-11-16T19:34:15Z

When do perf stat on sys metric, perf tool output nothing now: $ perf stat -a -M imx95_ddr_read.all -I 1000 $ This command runs on an arm64 machine and the Soc has one DDR hw pmu except one armv8_cortex_a55 pmu. Their maps show as follows: const struct pmu_events_map pmu_events_map[] = { { .arch = "arm64", .cpuid = "0x00000000410fd050", .event_table = { .pmus = pmu_events__arm_cortex_a55, .num_pmus = ARRAY_SIZE(pmu_events__arm_cortex_a55) }, .metric_table = { .pmus = NULL, .num_pmus = 0 } }, static const struct pmu_sys_events pmu_sys_event_tables[] = { { .event_table = { .pmus = pmu_events__freescale_imx95_sys, .num_pmus = ARRAY_SIZE(pmu_events__freescale_imx95_sys) }, .metric_table = { .pmus = pmu_metrics__freescale_imx95_sys, .num_pmus = ARRAY_SIZE(pmu_metrics__freescale_imx95_sys) }, .name = "pmu_events__freescale_imx95_sys", }, Currently, pmu_metrics_table__find() will return NULL when only do perf stat on sys metric. Then parse_groups() will never be called to parse sys metric_name, finally perf tool will exit directly. This should be a common problem. To fix the issue, this will keep the logic before commit f20c15d13f01 ("perf pmu-events: Remember the perf_events_map for a PMU") to return a empty metric table rather than a NULL pointer. This should be fine since the removed part just check if the table match provided metric_name. Without these code, the code in parse_groups() will also check the validity of metrci_name too. Fixes: f20c15d13f017d4b ("perf pmu-events: Remember the perf_events_map for a PMU") Reviewed-by: James Clark Signed-off-by: Xu Yang Tested-by: Xu Yang Acked-by: Ian Rogers Cc: Adrian Hunter Cc: Albert Ou Cc: Alexander Shishkin Cc: Alexandre Ghiti Cc: Athira Rajeev Cc: Benjamin Gray Cc: Ben Zong-You Xie Cc: Bibo Mao Cc: Clément Le Goffic Cc: Dima Kogan Cc: Dr. David Alan Gilbert Cc: Huacai Chen Cc: Ingo Molnar Cc: Jiri Olsa Cc: John Garry Cc: Kan Liang Cc: Leo Yan Cc: Mark Rutland Cc: Masami Hiramatsu Cc: Mike Leach Cc: Namhyung Kim Cc: Palmer Dabbelt Cc: Paul Walmsley Cc: Peter Zijlstra Cc: Ravi Bangoria Cc: Sandipan Das Cc: Will Deacon Cc: Yicong Yang Cc: linux-arm-kernel@lists.infradead.org Cc: linux-riscv@lists.infradead.org Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20241107162035.52206-2-irogers@google.com Signed-off-by: Ian Rogers Signed-off-by: Arnaldo Carvalho de Melo

perf vendor events arm64: Add i.MX91 DDR Performance Monitor metrics

2024-10-28T16:37:02Z

Add JSON metrics for i.MX91 DDR Performance Monitor. Signed-off-by: Xu Yang Acked-by: Namhyung Kim Cc: festevam@gmail.com Cc: conor+dt@kernel.org Cc: krzk+dt@kernel.org Cc: robh@kernel.org Cc: shawnguo@kernel.org Cc: will@kernel.org Cc: james.clark@linaro.org Cc: mike.leach@linaro.org Cc: leo.yan@linux.dev Cc: linux-arm-kernel@lists.infradead.org Cc: imx@lists.linux.dev Cc: Frank.li@nxp.com Cc: john.g.garry@oracle.com Cc: kernel@pengutronix.de Cc: s.hauer@pengutronix.de Cc: devicetree@vger.kernel.org Link: https://lore.kernel.org/r/20240924061251.3387850-3-xu.yang_2@nxp.com Signed-off-by: Namhyung Kim

perf vendor events amd: Update Zen 5 data cache fill events

2024-10-19T16:41:51Z

For events that count data cache fills, some combinations of the unit mask bits are useful for counting fills from local caches, DRAM or any far sources. However, named events currently exist for PMCx044 (Any Data Cache Fills) only. Add similar events for the following base events. * PMCx043 (Demand Data Cache Fills) * PMCx059 (Software Prefetch Data Cache Fills) * PMCx05A (Hardware Prefetch Data Cache Fills) While at it, remove "ls_any_fills_from_sys.all_dram_io" since it is a duplicate of "ls_any_fills_from_sys.dram_io_all". Event descriptions can be found in Section 2.1.16.5.2 "Load/Store (LS) Events" of the Processor Programming Reference (PPR) for AMD Family 1Ah Model 02h Revision C1 Processors document available at the link below. Link: https://bugzilla.kernel.org/attachment.cgi?id=307010 Signed-off-by: Sandipan Das Reviewed-by: Ian Rogers Cc: ananth.narayan@amd.com Cc: ravi.bangoria@amd.com Cc: eranian@google.com Link: https://lore.kernel.org/r/e036e3c9fb962c939fa06c855b68e532ee609e01.1729242778.git.sandipan.das@amd.com Signed-off-by: Namhyung Kim

perf vendor events amd: Add Zen 5 data fabric metrics

2024-10-19T16:41:51Z

Add data fabric metrics taken from Section 2.1.16.2 "Performance Measurement" in the Processor Programming Reference (PPR) for AMD Family 1Ah Model 02h Revision C1 Processors document available at the link below. The recommended metrics are sourced from Table 28 "Guidance for Common Performance Statistics with Complex Event Selects". They capture data bandwidth for various links and interfaces in the data fabric. Link: https://bugzilla.kernel.org/attachment.cgi?id=307010 Signed-off-by: Sandipan Das Reviewed-by: Ian Rogers Cc: ananth.narayan@amd.com Cc: ravi.bangoria@amd.com Cc: eranian@google.com Link: https://lore.kernel.org/r/e8757bb9f511907a52bc182de9395c5edec2fccf.1729242778.git.sandipan.das@amd.com Signed-off-by: Namhyung Kim

perf vendor events amd: Add Zen 5 data fabric events

2024-10-19T16:41:51Z

Add data fabric events taken from Section 2.1.16.2 "Performance Measurement" in the Processor Programming Reference (PPR) for AMD Family 1Ah Model 02h Revision C1 Processors document available at the link below. This constitutes events which capture the flow of data beats at various links and interfaces in the data fabric. Link: https://bugzilla.kernel.org/attachment.cgi?id=307010 Signed-off-by: Sandipan Das Reviewed-by: Ian Rogers Cc: ananth.narayan@amd.com Cc: ravi.bangoria@amd.com Cc: eranian@google.com Link: https://lore.kernel.org/r/198049e27366f3980e4991b95cec5eaac6d31d75.1729242778.git.sandipan.das@amd.com Signed-off-by: Namhyung Kim

tools/perf/pmu-events/powerpc: Add support for compat events in json

2024-10-17T18:24:52Z

perf list picks the events supported for specific platform from pmu-events/arch/powerpc/. Example power10 events are in pmu-events/arch/powerpc/power10, power9 events are part of pmu-events/arch/powerpc/power9. The decision of which platform to pick is determined based on PVR value in powerpc. The PVR value is matched from pmu-events/arch/powerpc/mapfile.csv Example: Format: PVR,Version,JSON/file/pathname,Type 0x004[bcd][[:xdigit:]]{4},1,power8,core 0x0066[[:xdigit:]]{4},1,power8,core 0x004e[[:xdigit:]]{4},1,power9,core 0x0080[[:xdigit:]]{4},1,power10,core 0x0082[[:xdigit:]]{4},1,power10,core The code gets the PVR from system using get_cpuid_str function in arch/powerpc/util/headers.c ( from SPRN_PVR ) and compares with value from mapfile.csv In case of compat mode, say when partition is booted in a power9 mode when the system is a power10, add an entry to pick the ISA architected events from "pmu-events/arch/powerpc/compat". Add json file generic-events.json which will contain these events which is supported in compat mode. Suggested-by: Madhavan Srinivasan Signed-off-by: Athira Rajeev Tested-by: Disha Goel Cc: akanksha@linux.ibm.com Cc: hbathini@linux.ibm.com Cc: kjain@linux.ibm.com Cc: linuxppc-dev@lists.ozlabs.org Link: https://lore.kernel.org/r/20241010145107.51211-1-atrajeev@linux.vnet.ibm.com Signed-off-by: Namhyung Kim

perf jevents: Add tool event json under a common architecture

2024-10-11T06:40:33Z

Introduce the notion of a common architecture/model that can be used to find event tables for common PMUs like the tool PMU. By having tool events be json standard PMU attribute configuration, descriptions, etc. can be used and these routines are already optimized for things like binary searching. Signed-off-by: Ian Rogers Acked-by: Namhyung Kim Link: https://lore.kernel.org/r/20241002032016.333748-9-irogers@google.com Signed-off-by: Namhyung Kim