| Message ID | 20240202234057.2085863-1-irogers@google.com (mailing list archive) |
|---|---|
| Headers | show |
| Series | Clean up libperf cpumap's empty function | expand |
On Fri, Feb 2, 2024 at 3:41 PM Ian Rogers <irogers@google.com> wrote: > > Rename and clean up the use of libperf CPU map functions particularly > focussing on perf_cpu_map__empty that may return true for maps > containing CPUs but also with an "any CPU"/dummy value. > > perf_cpu_map__nr is also troubling in that iterating an empty CPU map > will yield the "any CPU"/dummy value. Reduce the appearance of some > calls to this by using the perf_cpu_map__for_each_cpu macro. > > v3: Address handling of "any" is arm-spe/cs-etm patch. > v2: 6 patches were merged by Arnaldo. New patch added ensure empty > maps are allocated as NULL (suggested by James Clark). Hopefully a > fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". > > Ian Rogers (8): > libperf cpumap: Add any, empty and min helpers > libperf cpumap: Ensure empty cpumap is NULL from alloc > perf arm-spe/cs-etm: Directly iterate CPU maps > perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty > use > perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty > perf arm64 header: Remove unnecessary CPU map get and put > perf stat: Remove duplicate cpus_map_matched function > perf cpumap: Use perf_cpu_map__for_each_cpu when possible Ping. Thanks, Ian > tools/lib/perf/cpumap.c | 33 ++++- > tools/lib/perf/include/perf/cpumap.h | 16 +++ > tools/lib/perf/libperf.map | 4 + > tools/perf/arch/arm/util/cs-etm.c | 114 ++++++++---------- > tools/perf/arch/arm64/util/arm-spe.c | 4 +- > tools/perf/arch/arm64/util/header.c | 13 +- > tools/perf/arch/x86/util/intel-bts.c | 4 +- > tools/perf/arch/x86/util/intel-pt.c | 10 +- > tools/perf/builtin-c2c.c | 6 +- > tools/perf/builtin-stat.c | 31 +---- > tools/perf/tests/bitmap.c | 13 +- > tools/perf/tests/topology.c | 46 +++---- > tools/perf/util/auxtrace.c | 4 +- > tools/perf/util/bpf_kwork.c | 16 +-- > tools/perf/util/bpf_kwork_top.c | 12 +- > tools/perf/util/cpumap.c | 12 +- > tools/perf/util/record.c | 2 +- > .../scripting-engines/trace-event-python.c | 12 +- > tools/perf/util/session.c | 5 +- > tools/perf/util/stat.c | 2 +- > tools/perf/util/svghelper.c | 20 ++- > 21 files changed, 192 insertions(+), 187 deletions(-) > > -- > 2.43.0.594.gd9cf4e227d-goog >
On Wed, Feb 14, 2024 at 2:03 PM Ian Rogers <irogers@google.com> wrote: > > On Fri, Feb 2, 2024 at 3:41 PM Ian Rogers <irogers@google.com> wrote: > > > > Rename and clean up the use of libperf CPU map functions particularly > > focussing on perf_cpu_map__empty that may return true for maps > > containing CPUs but also with an "any CPU"/dummy value. > > > > perf_cpu_map__nr is also troubling in that iterating an empty CPU map > > will yield the "any CPU"/dummy value. Reduce the appearance of some > > calls to this by using the perf_cpu_map__for_each_cpu macro. > > > > v3: Address handling of "any" is arm-spe/cs-etm patch. > > v2: 6 patches were merged by Arnaldo. New patch added ensure empty > > maps are allocated as NULL (suggested by James Clark). Hopefully a > > fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". > > > > Ian Rogers (8): > > libperf cpumap: Add any, empty and min helpers > > libperf cpumap: Ensure empty cpumap is NULL from alloc > > perf arm-spe/cs-etm: Directly iterate CPU maps > > perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty > > use > > perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty > > perf arm64 header: Remove unnecessary CPU map get and put > > perf stat: Remove duplicate cpus_map_matched function > > perf cpumap: Use perf_cpu_map__for_each_cpu when possible > > Ping. Thanks, > Ian Adrian and James, are you ok with this now? Thanks, Namhyung
Hi Ian, Sorry for the late reply. On Fri, Feb 16, 2024 at 5:04 PM Namhyung Kim <namhyung@kernel.org> wrote: > > On Wed, Feb 14, 2024 at 2:03 PM Ian Rogers <irogers@google.com> wrote: > > > > On Fri, Feb 2, 2024 at 3:41 PM Ian Rogers <irogers@google.com> wrote: > > > > > > Rename and clean up the use of libperf CPU map functions particularly > > > focussing on perf_cpu_map__empty that may return true for maps > > > containing CPUs but also with an "any CPU"/dummy value. > > > > > > perf_cpu_map__nr is also troubling in that iterating an empty CPU map > > > will yield the "any CPU"/dummy value. Reduce the appearance of some > > > calls to this by using the perf_cpu_map__for_each_cpu macro. > > > > > > v3: Address handling of "any" is arm-spe/cs-etm patch. > > > v2: 6 patches were merged by Arnaldo. New patch added ensure empty > > > maps are allocated as NULL (suggested by James Clark). Hopefully a > > > fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". > > > > > > Ian Rogers (8): > > > libperf cpumap: Add any, empty and min helpers > > > libperf cpumap: Ensure empty cpumap is NULL from alloc > > > perf arm-spe/cs-etm: Directly iterate CPU maps > > > perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty > > > use > > > perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty > > > perf arm64 header: Remove unnecessary CPU map get and put > > > perf stat: Remove duplicate cpus_map_matched function > > > perf cpumap: Use perf_cpu_map__for_each_cpu when possible > > > > Ping. Thanks, > > Ian > > Adrian and James, are you ok with this now? I think James is fine now and the Intel-pt part seems straight-forward so I'd like to merge this change. Please tell me if you have any concerns. Thanks, Namhyung
On Thu, Mar 07, 2024 at 03:47:00PM -0800, Namhyung Kim wrote: > Hi Ian, > > Sorry for the late reply. > > On Fri, Feb 16, 2024 at 5:04 PM Namhyung Kim <namhyung@kernel.org> wrote: > > > > On Wed, Feb 14, 2024 at 2:03 PM Ian Rogers <irogers@google.com> wrote: > > > > > > On Fri, Feb 2, 2024 at 3:41 PM Ian Rogers <irogers@google.com> wrote: > > > > > > > > Rename and clean up the use of libperf CPU map functions particularly > > > > focussing on perf_cpu_map__empty that may return true for maps > > > > containing CPUs but also with an "any CPU"/dummy value. > > > > > > > > perf_cpu_map__nr is also troubling in that iterating an empty CPU map > > > > will yield the "any CPU"/dummy value. Reduce the appearance of some > > > > calls to this by using the perf_cpu_map__for_each_cpu macro. > > > > > > > > v3: Address handling of "any" is arm-spe/cs-etm patch. > > > > v2: 6 patches were merged by Arnaldo. New patch added ensure empty > > > > maps are allocated as NULL (suggested by James Clark). Hopefully a > > > > fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". > > > > > > > > Ian Rogers (8): > > > > libperf cpumap: Add any, empty and min helpers > > > > libperf cpumap: Ensure empty cpumap is NULL from alloc > > > > perf arm-spe/cs-etm: Directly iterate CPU maps > > > > perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty > > > > use > > > > perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty > > > > perf arm64 header: Remove unnecessary CPU map get and put > > > > perf stat: Remove duplicate cpus_map_matched function > > > > perf cpumap: Use perf_cpu_map__for_each_cpu when possible > > > > > > Ping. Thanks, > > > Ian > > Adrian and James, are you ok with this now? > I think James is fine now and the Intel-pt part seems straight-forward > so I'd like to merge this change. Please tell me if you have any concerns. Namhyung, I noticed this hasn't been merged and applies cleanly, so I'm adding it to perf-tools-next, from your comment above can I take it as an Acked-by or even Reviewed-by? - Arnaldo
On Mon, Mar 18, 2024 at 2:37 PM Arnaldo Carvalho de Melo <acme@kernel.org> wrote: > > On Thu, Mar 07, 2024 at 03:47:00PM -0800, Namhyung Kim wrote: > > Hi Ian, > > > > Sorry for the late reply. > > > > On Fri, Feb 16, 2024 at 5:04 PM Namhyung Kim <namhyung@kernel.org> wrote: > > > > > > On Wed, Feb 14, 2024 at 2:03 PM Ian Rogers <irogers@google.com> wrote: > > > > > > > > On Fri, Feb 2, 2024 at 3:41 PM Ian Rogers <irogers@google.com> wrote: > > > > > > > > > > Rename and clean up the use of libperf CPU map functions particularly > > > > > focussing on perf_cpu_map__empty that may return true for maps > > > > > containing CPUs but also with an "any CPU"/dummy value. > > > > > > > > > > perf_cpu_map__nr is also troubling in that iterating an empty CPU map > > > > > will yield the "any CPU"/dummy value. Reduce the appearance of some > > > > > calls to this by using the perf_cpu_map__for_each_cpu macro. > > > > > > > > > > v3: Address handling of "any" is arm-spe/cs-etm patch. > > > > > v2: 6 patches were merged by Arnaldo. New patch added ensure empty > > > > > maps are allocated as NULL (suggested by James Clark). Hopefully a > > > > > fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". > > > > > > > > > > Ian Rogers (8): > > > > > libperf cpumap: Add any, empty and min helpers > > > > > libperf cpumap: Ensure empty cpumap is NULL from alloc > > > > > perf arm-spe/cs-etm: Directly iterate CPU maps > > > > > perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty > > > > > use > > > > > perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty > > > > > perf arm64 header: Remove unnecessary CPU map get and put > > > > > perf stat: Remove duplicate cpus_map_matched function > > > > > perf cpumap: Use perf_cpu_map__for_each_cpu when possible > > > > > > > > Ping. Thanks, > > > > Ian > > > > Adrian and James, are you ok with this now? > > > I think James is fine now and the Intel-pt part seems straight-forward > > so I'd like to merge this change. Please tell me if you have any concerns. > > Namhyung, > > I noticed this hasn't been merged and applies cleanly, so I'm > adding it to perf-tools-next, from your comment above can I take it as > an Acked-by or even Reviewed-by? Oh, I thought I did it already, but I probably missed pushing it. :( Sure you can add it, I'll do that for the sake of b4. Acked-by: Namhyung Kim <namhyung@kernel.org> Thanks, Namhyung
Rename and clean up the use of libperf CPU map functions particularly focussing on perf_cpu_map__empty that may return true for maps containing CPUs but also with an "any CPU"/dummy value. perf_cpu_map__nr is also troubling in that iterating an empty CPU map will yield the "any CPU"/dummy value. Reduce the appearance of some calls to this by using the perf_cpu_map__for_each_cpu macro. v3: Address handling of "any" is arm-spe/cs-etm patch. v2: 6 patches were merged by Arnaldo. New patch added ensure empty maps are allocated as NULL (suggested by James Clark). Hopefully a fix to "perf arm-spe/cs-etm: Directly iterate CPU maps". Ian Rogers (8): libperf cpumap: Add any, empty and min helpers libperf cpumap: Ensure empty cpumap is NULL from alloc perf arm-spe/cs-etm: Directly iterate CPU maps perf intel-pt/intel-bts: Switch perf_cpu_map__has_any_cpu_or_is_empty use perf cpumap: Clean up use of perf_cpu_map__has_any_cpu_or_is_empty perf arm64 header: Remove unnecessary CPU map get and put perf stat: Remove duplicate cpus_map_matched function perf cpumap: Use perf_cpu_map__for_each_cpu when possible tools/lib/perf/cpumap.c | 33 ++++- tools/lib/perf/include/perf/cpumap.h | 16 +++ tools/lib/perf/libperf.map | 4 + tools/perf/arch/arm/util/cs-etm.c | 114 ++++++++---------- tools/perf/arch/arm64/util/arm-spe.c | 4 +- tools/perf/arch/arm64/util/header.c | 13 +- tools/perf/arch/x86/util/intel-bts.c | 4 +- tools/perf/arch/x86/util/intel-pt.c | 10 +- tools/perf/builtin-c2c.c | 6 +- tools/perf/builtin-stat.c | 31 +---- tools/perf/tests/bitmap.c | 13 +- tools/perf/tests/topology.c | 46 +++---- tools/perf/util/auxtrace.c | 4 +- tools/perf/util/bpf_kwork.c | 16 +-- tools/perf/util/bpf_kwork_top.c | 12 +- tools/perf/util/cpumap.c | 12 +- tools/perf/util/record.c | 2 +- .../scripting-engines/trace-event-python.c | 12 +- tools/perf/util/session.c | 5 +- tools/perf/util/stat.c | 2 +- tools/perf/util/svghelper.c | 20 ++- 21 files changed, 192 insertions(+), 187 deletions(-)