perf test shell record_bpf_filter: Skip 6.2 kernel

The BPF sample filtering requires two kernel changes below:

 * bpf_cast_to_kernel_ctx() kfunc (added in v6.2)

 * setting perf_sample_data->sample_flags (finished in v6.3)

The perf tools can check bpf_cast_to_kernel_ctx() easily so it can
refuse BPF filters on those old kernels (v6.1 and earlier).

But checking sample_flags appears to be difficult so current code won't
work on v6.2 kernel.  That's unfortunate but I don't know what's the
correct way to handle it.

For now, let's skip v6.2 kernels explicitly (if failed) in the test.

Fixes: 9575ecdd198a50e9 ("perf test: Add perf record sample filtering test")
Reported-by: Arnaldo Carvalho de Melo <acme@kernel.org>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: bpf@vger.kernel.org
Link: https://lore.kernel.org/r/20230825164152.165610-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

libperf: Get rid of attr.id field

Now there's no in-tree user of the field.  To remove the possible bug
later, let's get rid of the 'id' field and add a comment for that.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230825152552.112913-4-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Convert to perf_record_header_attr_id()

Instead of accessing the attr.id directly, use the
perf_record_header_attr_id() helper to handle old versions.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230825152552.112913-3-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

libperf: Add perf_record_header_attr_id()

The HEADER_ATTR record has an event attr followed by the id array.  But
perf data from a different version could have different size of attr.

So it cannot just use event->attr.id to access the array.  Let's add the
perf_record_header_attr_id() macro to calculate the start of the array.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230825152552.112913-2-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Handle old data in PERF_RECORD_ATTR

The PERF_RECORD_ATTR is used for a pipe mode to describe an event with
attribute and IDs.  The ID table comes after the attr and it calculate
size of the table using the total record size and the attr size.

  n_ids = (total_record_size - end_of_the_attr_field) / sizeof(u64)

This is fine for most use cases, but sometimes it saves the pipe output
in a file and then process it later.  And it becomes a problem if there
is a change in attr size between the record and report.

  $ perf record -o- > perf-pipe.data  # old version
  $ perf report -i- < perf-pipe.data  # new version

For example, if the attr size is 128 and it has 4 IDs, then it would
save them in 168 byte like below:

   8 byte: perf event header { .type = PERF_RECORD_ATTR, .size = 168 },
 128 byte: perf event attr { .size = 128, ... },
  32 byte: event IDs [] = { 1234, 1235, 1236, 1237 },

But when report later, it thinks the attr size is 136 then it only read
the last 3 entries as ID.

   8 byte: perf event header { .type = PERF_RECORD_ATTR, .size = 168 },
 136 byte: perf event attr { .size = 136, ... },
  24 byte: event IDs [] = { 1235, 1236, 1237 },  // 1234 is missing

So it should use the recorded version of the attr.  The attr has the
size field already then it should honor the size when reading data.

Fixes: 2c46dbb517a10b18 ("perf: Convert perf header attrs into attr events")
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Tom Zanussi <zanussi@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230825152552.112913-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmus: Skip duplicate PMUs and don't print list suffix by default

Add a PMUs scan that ignores duplicates. When there are multiple PMUs
that differ only by suffix, by default just list the first one and
skip all others. The scan routine checks that the PMU names match but
doesn't enforce that the numbers are consecutive as for some PMUs
there are gaps. If "-v" is passed to "perf list" then list all PMUs.

With the previous change duplicate PMUs are no longer printed but the
suffix of the first is printed. When duplicate PMUs are being skipped
avoid printing the suffix.

Before:

  $ perf list
  ...
    uncore_imc_free_running_0/data_read/               [Kernel PMU event]
    uncore_imc_free_running_0/data_total/              [Kernel PMU event]
    uncore_imc_free_running_0/data_write/              [Kernel PMU event]
    uncore_imc_free_running_1/data_read/               [Kernel PMU event]
    uncore_imc_free_running_1/data_total/              [Kernel PMU event]
    uncore_imc_free_running_1/data_write/              [Kernel PMU event]

After:

  $ perf list
  ...
    uncore_imc_free_running/data_read/                 [Kernel PMU event]
    uncore_imc_free_running/data_total/                [Kernel PMU event]
    uncore_imc_free_running/data_write/                [Kernel PMU event]
  ...
  $ perf list -v
    uncore_imc_free_running_0/data_read/               [Kernel PMU event]
    uncore_imc_free_running_0/data_total/              [Kernel PMU event]
    uncore_imc_free_running_0/data_write/              [Kernel PMU event]
    uncore_imc_free_running_1/data_read/               [Kernel PMU event]
    uncore_imc_free_running_1/data_total/              [Kernel PMU event]
    uncore_imc_free_running_1/data_write/              [Kernel PMU event]
  ...

Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20230825135237.921058-3-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmus: Sort pmus by name then suffix

Sort PMUs by name. If two PMUs have the same name but differ by
suffix, sort the suffixes numerically.

For example, "breakpoint" comes before "cpu",
"uncore_imc_free_running_0" comes before "uncore_imc_free_running_1".

Suffixes need to be treated specially as otherwise they will be ordered
like 0, 1, 10, 11, .., 2, 20, 21, .., etc. Only PMUs starting 'uncore_'
are considered to have a potential suffix.

Sorting of PMUs is done so that later patches can skip duplicate uncore
PMUs that differ only by there suffix.

Committer notes:

Used the more compact, intention revealing strstarts() function we got
from the kernel sources:

-       if (strncmp(str, "uncore_", 7))
+       if (!strstarts(str, "uncore_"))

Also in pmus_cmp() the lhs_num and rhs_num variables may end up not
being set for non "uncore_" prefixed PMUs in pmu_name_len_no_suffix(),
or at least gcc 7.5 in some distros (opensuse 15.5, to be EOLed in
Dec/2024) thins so, so initialize both to zero.

Reviewed-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lore.kernel.org/r/20230825135237.921058-2-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf beauty mmap_flags: Use "test -f" instead of "[-f FILE]"

"[" is part of the shell builtin test (and a synonym for it),
 not a link to the external command /usr/bin/test.

Using the "test" is simpler because it avoids a lot of "[]".

Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Yanteng Si <siyanteng@loongson.cn>
Acked-by: Huacai Chen <chenhuacai@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.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: loongarch@lists.linux.dev
Cc: loongson-kernel@lists.loongnix.cn
Link: https://lore.kernel.org/r/c50bc0a92dce0ff0fa6504c1a52fb53e2ac007bf.1692962043.git.siyanteng@loongson.cn
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf beauty mmap_flags: Fix script for archs that use the generic mman.h

To address this error:

  grep: /root/linux-next/tools/arch/xxxxx/include/uapi/asm//mman.h:
  No such file or directory

Signed-off-by: Yanteng Si <siyanteng@loongson.cn>
Acked-by: Huacai Chen <chenhuacai@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.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: loongarch@lists.linux.dev
Cc: loongson-kernel@lists.loongnix.cn
Link: https://lore.kernel.org/r/42e8e3565d6035302907426c1e65483b2a4007f5.1692962043.git.siyanteng@loongson.cn
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tools: Allow to use cpuinfo on LoongArch

Define these macros so that the CPU name can be displayed when running
'perf report' and 'perf timechart'.

Committer notes:

No need to have:

if (strcasestr(buf, "Model Name")) {
strlcpy(cpu_m, &buf[13], 255);
break;
} else if (strcasestr(buf, "model name")) {
strlcpy(cpu_m, &buf[13], 255);
break;
}

As the point of strcasestr() is to be case insensitive to both the
haystack and the needle, so simplify the above to just:

if (strcasestr(buf, "model name")) {
strlcpy(cpu_m, &buf[13], 255);
break;
}

Signed-off-by: Yanteng Si <siyanteng@loongson.cn>
Acked-by: Huacai Chen <chenhuacai@loongson.cn>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.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: loongarch@lists.linux.dev
Cc: loongson-kernel@lists.loongnix.cn
Link: https://lore.kernel.org/r/db968a186a10e4629fe10c26a1210f7126ad41ec.1692962043.git.siyanteng@loongson.cn
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf lock contention: Fix typo in max-stack option description

Fix typo in max-stack option description by changing lopck contention
to lock contention.

Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20230825104700.440809-1-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tui slang: Tidy casts

Casts were necessary for older versions of libslang, however, these
are now 15 years old and so we no longer need to care about supporting
them. Tidy the casts and remove unnecessary logic.

Move the ENABLE_SLFUTURE_CONST to the libslang.h common include file,
and also enable ENABLE_SLFUTURE_VOID.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
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: K Prateek Nayak <kprateek.nayak@amd.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: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Wei Li <liwei391@huawei.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230825024002.801955-7-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf build-id: Simplify build_id_cache__cachedir()

Initialize realname to NULL, rather than name.

This avoids a cast and as realpath is either NULL or an allocated
string, free can be called unconditionally.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
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: K Prateek Nayak <kprateek.nayak@amd.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: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Wei Li <liwei391@huawei.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230825024002.801955-6-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Make id const and add missing free

The struct pmu id is initialized from pmu_id that is read into allocated
memory from a file, as such it needs free-ing in pmu__delete().

Make the id value const so that we can remove casts in tests.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
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: K Prateek Nayak <kprateek.nayak@amd.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: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Wei Li <liwei391@huawei.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230825024002.801955-5-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Make term's config const

This avoids casts in tests. Use zfree in a few places to avoid
warnings about a freeing a const pointer.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
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: K Prateek Nayak <kprateek.nayak@amd.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: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Wei Li <liwei391@huawei.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230825024002.801955-4-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Remove logic for PMU name being NULL

The PMU name could be NULL in the case of the fake_pmu. Initialize the
name for the fake_pmu to "fake" so that all other logic can assume it
is initialized. Add a const to the type of name so that a literal can
be used to avoid additional initialization code. Propagate the cost
through related routines and remove now unnecessary "(char *)"
casts. Doing this located a bug in builtin-list for the pmu_glob that
was missing a strdup.

Signed-off-by: Ian Rogers <irogers@google.com>
Link: https://lore.kernel.org/r/20230825024002.801955-3-irogers@google.com
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: James Clark <james.clark@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Wei Li <liwei391@huawei.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Will Deacon <will@kernel.org>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Ming Wang <wangming01@loongson.cn>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: linux-kernel@vger.kernel.org
Cc: linux-perf-users@vger.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf header: Fix missing PMU caps

PMU caps are written as HEADER_PMU_CAPS or for the special case of the
PMU "cpu" as HEADER_CPU_PMU_CAPS. As the PMU "cpu" is special, and not
any "core" PMU, the logic had become broken and core PMUs not called
"cpu" were not having their caps written.

This affects ARM and s390 non-hybrid PMUs.

Simplify the PMU caps writing logic to scan one fewer time and to be
more explicit in its behavior.

Fixes: 178ddf3bad981380 ("perf header: Avoid hybrid PMU list in write_pmu_caps")
Reported-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ian Rogers <irogers@google.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: K Prateek Nayak <kprateek.nayak@amd.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: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230825024002.801955-2-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf jevents: Don't append Unit to desc

Unit with the PMU name is appended to desc in jevents.py, but on
hybrid platforms it causes the desc to differ from the regular
non-hybrid system with a PMU of 'cpu'. Having differing descs means
the events don't deduplicate. To make the perf list output not differ,
append the Unit on again in the perf list printing code.

On x86 reduces the binary size by 409,600 bytes or about 4%. Update
pmu-events test expectations to match the differently generated
pmu-events.c code.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824183212.374787-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts python gecko: Launch the profiler UI on the default browser with the appropriate URL

All required libraries have been imported and make sure that none of
them are external dependencies. To achieve this, created a virt env and
verified.

Modified usage information and added combined command.

Modified the main() function to read the --save-only command-line option
and set the output_file variable accordingly.

Modified the trace_end() function to check for the output_file variable.
If it is set, the profiler data is saved to a local file in Gecko
Profile format, or the profiler.firefox.com is opened on the default
browser.

Included trace_begin() to initialize the Firefox Profiler and launch the
default browser to display the profiler.firefox.com.

Added a new function launchFirefox() to start a local server and launch
the profiler UI on the default browser with the appropriate URL.

Created the "CORSRequestHandler" class to enable Cross-Origin Resource
Sharing.

Summary:

This integration now includes a exiting feature to conveniently host the
Gecko Profile data on a local server and open it directly in the default
web browser.

This means that users can now effortlessly visualize and analyze the
profiler results with just a single click.

The addition of the --save-only command-line option allows users to save
the profiler output to a local file in Gecko Profile format, but the
real highlight lies in the capability to seamlessly launch a local
server, making the data accessible to Firefox Profiler via a web
browser.

In addition, it's important to highlight that all data are hosted
locally, eliminating any concerns about data privacy rules and
regulations.

Signed-off-by: Anup Sharma <anupnewsmail@gmail.com>
Tested-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.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>
Link: https://lore.kernel.org/r/ZNOS0vo58DnVLpD8@yoga
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts python: Add support for input args in gecko script

Refines the argument handling mechanism in the "gecko-report" script to
enable better compatibility and improved user experience.

The script now differentiates between scenarios where arguments are
provided for record and report cases where gecko.py arguments are
passed.

Signed-off-by: Anup Sharma <anupnewsmail@gmail.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.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>
Link: https://lore.kernel.org/r/ZNf7W+EIrrCSHZN0@yoga
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf jevents: Sort strings in the big C string to reduce faults

Sort the strings within the big C string based on whether they were
for a metric and then by when they were added. This helps group
related strings and reduce minor faults by approximately 10 in 1740,
about 0.57%.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-18-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Lazily load sysfs aliases

Don't load sysfs aliases for a PMU when the PMU is first created, defer
until an alias needs to be found. For the pmu-scan benchmark, average
core PMU scanning is reduced by 30.8%, and average PMU scanning by
12.6%.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-17-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Be lazy about loading event info files from sysfs

Event info is only needed when an event is parsed or when merging data
from an JSON and sysfs event. Be lazy in its loading to reduce file
accesses.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-16-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Scan type early to fail an invalid PMU quickly

Scan sysfs PMU's type early so that format and aliases aren't
attempted to be loaded if the PMU name is invalid.

This is the case for event_pmu tokens in parse-events.y where a wildcard
name is first assumed to be a PMU name.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-15-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Lazily add JSON events

Rather than scanning all JSON events and adding them when a PMU is
created, add the alias when the JSON event is needed.

Average core PMU scanning run time reduced by 60.2%. Average PMU
scanning run time reduced by 15%. Page faults with no events reduced by
74 page faults, 4% of total.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-14-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Cache JSON events table

Cache the JSON events table so that finding it isn't done per
event/alias.

Change the events table find so that when the PMU is given, if the PMU
has no JSON events return null.

Update usage to always use the PMU variable.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-13-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Merge JSON events with sysfs at load time

Rather than load all sysfs events then parsing all JSON events and
merging with ones that already exist. When a sysfs event is loaded, look
for a corresponding JSON event and merge immediately.

To simplify the logic, early exit the perf_pmu__new_alias function if an
alias is attempted to be added twice - as merging has already been
explicitly handled.

Fix the copying of terms to a merged alias and some ENOMEM paths.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-12-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Prefer passing pmu to aliases list

The aliases list is part of the PMU. Rather than pass the aliases
list, pass the full PMU simplifying some callbacks.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-11-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Parse sysfs events directly from a file

Rather than read a sysfs events file into a 256 byte char buffer, pass
the FILE* directly to the lex/yacc parser.

This avoids there being a maximum events file size.

While changing the API, constify some arguments to remove unnecessary
casts.

Allocating the read buffer decreases the performance of pmu-scan by
around 3%.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-10-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu-events: Add pmu_events_table__find_event()

jevents stores events sorted by name. Add a find function that will
binary search event names avoiding the need to linearly search through
events.

Add a test in tests/pmu-events.c. If the PMU or event aren't found -1000
is returned. If the event is found but no callback function given, 0 is
returned.

This allows the find function also act as a test for existence.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-9-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu-events: Reduce processed events by passing PMU

Pass the PMU to pmu_events_table__for_each_event so that entries that
don't match don't need to be processed by callback.

If a NULL PMU is passed then all PMUs are processed.

'perf bench internals pmu-scan's "Average PMU scanning" performance is
reduced by about 5% on an Intel tigerlake.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-8-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf s390 s390_cpumcfdg_dump: Don't scan all PMUs

Rather than scanning all PMUs for a counter name, scan the PMU
associated with the evsel of the sample. This is done to remove a
dependence on pmu-events.h.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-7-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Improve error message for double setting

Double setting information for an event would produce an error message
associated with the PMU rather than the term that was double setting.
Improve the error message to be on the term.

Before:

  $ perf stat -e 'cpu/inst_retired.any,inst_retired.any/' true
  event syntax error: 'cpu/inst_retired.any,inst_retired.any/'
                       \___ Bad event or PMU

  Unabled to find PMU or event on a PMU of 'cpu'
  Run 'perf list' for a list of valid events
  $

After:

  $ perf stat -e 'cpu/inst_retired.any,inst_retired.any/' true
  event syntax error: '..etired.any,inst_retired.any/'
                                    \___ Bad event or PMU

  Unabled to find PMU or event on a PMU of 'cpu'

  Initial error:

  event syntax error: '..etired.any,inst_retired.any/'
                                    \___ Attempt to set event's scale twice
  Run 'perf list' for a list of valid events

Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-6-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf jevents: Group events by PMU

Prior to this change a cpuid would map to a list of events where the PMU
would be encoded alongside the event information. This change breaks
apart each group of events so that there is a group per PMU. A new table
is added with the PMU's name and the list of events, the original table
now holding an array of these per PMU tables.

These changes are to make it easier to get per PMU information about
events, rather than the current approach of scanning all events. The
perf binary size with BPF skeletons on x86 is reduced by about 1%. The
unidentified PMU is now always expanded to "cpu".

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-5-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu-events: Add extra underscore to function names

Add extra underscore before "for" of pmu_events_table_for_each_event
and pmu_metrics_table_for_each_metric.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-4-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Abstract alias/event struct

In order to be able to lazily compute aliases/events for a PMU, move
the struct perf_pmu_alias into pmu.c.

Add perf_pmu__find_event and perf_pmu__for_each_event that take a
callback that is called for the found event or for each event.

The layout of struct pmu and the event/alias list is unchanged but the
API is altered so that aliases are no longer directly accessed, allowing
for later changes.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-3-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Make the loading of formats lazy

The sysfs format files are loaded eagerly in a PMU. Add a flag so that
we create the format but only load the contents when necessary.

Reduce the size of the value in struct perf_pmu_format and avoid holes
so there is no additional space requirement.

For "perf stat -e cycles true" this reduces the number of openat calls
from 648 to 573 (about 12%). The benchmark pmu scan speed is improved
by roughly 5%.

Before:

  $ perf bench internals pmu-scan
  Computing performance of sysfs PMU event scan for 100 times
    Average core PMU scanning took: 1061.100 usec (+- 9.965 usec)
    Average PMU scanning took: 4725.300 usec (+- 260.599 usec)

After:

  $ perf bench internals pmu-scan
  Computing performance of sysfs PMU event scan for 100 times
    Average core PMU scanning took: 989.170 usec (+- 6.873 usec)
    Average PMU scanning took: 4520.960 usec (+- 251.272 usec)

Committer testing:

On a AMD Ryzen 5950x:

Before:

  $ perf bench internals pmu-scan -i1000
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 563.466 usec (+- 1.008 usec)
    Average PMU scanning took: 1619.174 usec (+- 23.627 usec)
  $ perf stat -r5 perf bench internals pmu-scan -i1000
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 583.401 usec (+- 2.098 usec)
    Average PMU scanning took: 1677.352 usec (+- 24.636 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 553.254 usec (+- 0.825 usec)
    Average PMU scanning took: 1635.655 usec (+- 24.312 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 557.733 usec (+- 0.980 usec)
    Average PMU scanning took: 1600.659 usec (+- 23.344 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 554.906 usec (+- 0.774 usec)
    Average PMU scanning took: 1595.338 usec (+- 23.288 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 551.798 usec (+- 0.967 usec)
    Average PMU scanning took: 1623.213 usec (+- 23.998 usec)

   Performance counter stats for 'perf bench internals pmu-scan -i1000' (5 runs):

             3276.82 msec task-clock:u                     #    0.990 CPUs utilized               ( +-  0.82% )
                   0      context-switches:u               #    0.000 /sec
                   0      cpu-migrations:u                 #    0.000 /sec
                1008      page-faults:u                    #  307.615 /sec                        ( +-  0.04% )
         12049614778      cycles:u                         #    3.677 GHz                         ( +-  0.07% )  (83.34%)
           117507478      stalled-cycles-frontend:u        #    0.98% frontend cycles idle        ( +-  0.33% )  (83.32%)
            27106761      stalled-cycles-backend:u         #    0.22% backend cycles idle         ( +-  9.55% )  (83.36%)
         33294953848      instructions:u                   #    2.76  insn per cycle
                                                           #    0.00  stalled cycles per insn     ( +-  0.03% )  (83.31%)
          6849825049      branches:u                       #    2.090 G/sec                       ( +-  0.03% )  (83.37%)
            71533903      branch-misses:u                  #    1.04% of all branches             ( +-  0.20% )  (83.30%)

              3.3088 +- 0.0302 seconds time elapsed  ( +-  0.91% )

  $

After:

  $ perf stat -r5 perf bench internals pmu-scan -i1000
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 550.702 usec (+- 0.958 usec)
    Average PMU scanning took: 1566.577 usec (+- 22.747 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 548.315 usec (+- 0.555 usec)
    Average PMU scanning took: 1565.499 usec (+- 22.760 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 548.073 usec (+- 0.555 usec)
    Average PMU scanning took: 1586.097 usec (+- 23.299 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 561.184 usec (+- 2.709 usec)
    Average PMU scanning took: 1567.153 usec (+- 22.548 usec)
  # Running 'internals/pmu-scan' benchmark:
  Computing performance of sysfs PMU event scan for 1000 times
    Average core PMU scanning took: 546.987 usec (+- 0.553 usec)
    Average PMU scanning took: 1562.814 usec (+- 22.729 usec)

   Performance counter stats for 'perf bench internals pmu-scan -i1000' (5 runs):

             3170.86 msec task-clock:u                     #    0.992 CPUs utilized               ( +-  0.22% )
                   0      context-switches:u               #    0.000 /sec
                   0      cpu-migrations:u                 #    0.000 /sec
                1010      page-faults:u                    #  318.526 /sec                        ( +-  0.04% )
         11890047674      cycles:u                         #    3.750 GHz                         ( +-  0.14% )  (83.27%)
           119090499      stalled-cycles-frontend:u        #    1.00% frontend cycles idle        ( +-  0.46% )  (83.40%)
            32502449      stalled-cycles-backend:u         #    0.27% backend cycles idle         ( +-  8.32% )  (83.30%)
         33119141261      instructions:u                   #    2.79  insn per cycle
                                                    #    0.00  stalled cycles per insn     ( +-  0.01% )  (83.37%)
          6812816561      branches:u                       #    2.149 G/sec                       ( +-  0.01% )  (83.29%)
            70157855      branch-misses:u                  #    1.03% of all branches             ( +-  0.28% )  (83.38%)

             3.19710 +- 0.00826 seconds time elapsed  ( +-  0.26% )

  $

Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230824041330.266337-2-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf build: Allow customization of clang options for BPF target

This also puts an unconditional -Werror under control of WERROR. The
clang includes added during the build can lead to a warning that may be
turned into an error. In addition, hardened clang produces a warning
about lack of support for -fstack-protector* options for the BPF target:

  clang -g -O2 -target bpf -Wall -Werror -Ilinux/tools/perf/util/bpf_skel/.tmp/.. \
    -I -idirafter /usr/lib/llvm/16/bin/../../../../lib/clang/16/include -idirafter /usr/local/include \
    -idirafter /usr/include  -Ilinux/tools/include/uapi -c util/bpf_skel/bperf_follower.bpf.c \
    -o linux/tools/perf/util/bpf_skel/.tmp/bperf_follower.bpf.o && llvm-strip -g linux/tools/perf/util/bpf_skel/.tmp/bperf_follower.bpf.o
  clang-16: error: /usr/lib/llvm/16/bin/../../../../lib/clang/16/include: 'linker' input unused [-Werror,-Wunused-command-line-argument]
  clang-16: error: ignoring '-fstack-protector-strong' option as it is not currently supported for target 'bpf' [-Werror,-Woption-ignored]
  make[1]: *** [Makefile.perf:1082: linux/tools/perf/util/bpf_skel/.tmp/bpf_prog_profiler.bpf.o] Error 1

Signed-off-by: Guilherme Amadio <amadio@gentoo.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZOZQ2LDA+3Wg8x2T@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Pass PMU rather than aliases and format

Pass the pmu so the aliases and format list can be better abstracted
and later lazily loaded.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-9-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Avoid passing format list to perf_pmu__format_bits()

Pass the PMU so the format list can be better abstracted and later
lazily loaded.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-8-irogers@google.com
[ Did missing conversions in tools/perf/arch/arm*/util/cs-etm.c ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Avoid passing format list to perf_pmu__format_type

Pass the pmu so the format list can be better abstracted and later
lazily loaded.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-7-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Avoid passing format list to perf_pmu__config_terms()

Abstract the format list better, hiding it in the PMU, by changing
perf_pmu__config_terms() the PMU rather than the format list in the PMU.

Change the PMU test to pass a dummy PMU for this purpose. Changing the
test allows perf_pmu__del_formats() to become static.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-6-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Reduce scope of perf_pmu_error()

Move declaration from header file to pmu.y and make static.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-5-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Move perf_pmu__set_format to pmu.y

Avoid having the function in the C and header file, as it is only used
locally by pmu.y.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-4-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf pmu: Avoid a path name copy

Rather than read a base path and append into a 2nd path, read the base
path directly into output buffer and append to that.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-3-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script ibs: Remove unused include

Done to reduce dependencies on pmu-events.h.

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Gaosheng Cui <cuigaosheng1@huawei.com>
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: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20230823080828.1460376-2-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bench breakpoint: Skip run if no breakpoints available

Based on commit 7d54a4acd8c1de3e ("perf test: Skip watchpoint tests if
no watchpoints available"), hardware breakpoints are not available for
power9 platform and because of that 'perf bench breakpoint' run fails on
power9 platform.

Add code to check for the return value of perf_event_open() in the
breakpoint run and skip the 'perf bench breakpoint' run, if hardware
breakpoints are not available.

Result on power9 system before patch changes:

  [command]# perf bench breakpoint thread
  perf_event_open: No such device

Result on power9 system after patch changes:

  [command]# ./perf bench breakpoint thread
  Skipping perf bench breakpoint thread: No hardware support

Reported-by: Disha Goel <disgoel@linux.vnet.ibm.com>
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Acked-by: Naveen N Rao <naveen@kernel.org>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230823075103.190565-1-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf lzma: Convert some pr_err() to pr_debug() as callers already use pr_debug()

I noticed some error with:

  # perf list ex_ret_brn
  lzma: fopen failed on /usr/lib/modules/5.15.14-100.fc34.x86_64/kernel/net/bluetooth/bnep/bnep.ko.xz: 'No such file or directory'
  lzma: fopen failed on /usr/lib/modules/5.16.16-200.fc35.x86_64/kernel/drivers/gpu/drm/drm_kms_helper.ko.xz: 'No such file or directory'
  lzma: fopen failed on /usr/lib/modules/5.18.16-200.fc36.x86_64/kernel/arch/x86/crypto/crct10dif-pclmul.ko.xz: 'No such file or directory'
  lzma: fopen failed on /usr/lib/modules/5.16.16-200.fc35.x86_64/kernel/drivers/i2c/busses/i2c-piix4.ko.xz: 'No such file or directory'
  <BIG SNIP>

Then using 'perf probe' + 'perf trace' to debug 'perf list', it seems
its some inconsistency in the ~/.debug/ cache where broken build id
symlinks that ends up making it try to uncompress some kernel modules
using the lzma routines:

   395.309 perf/3594447 probe_perf:lzma_decompress_to_file(__probe_ip: 6118448, input_string: "/usr/lib/modules/5.18.17-200.fc36.x86_64/kernel/drivers/nvme/host/nvme.ko.xz")
                                       lzma_decompress_to_file (/var/home/acme/bin/perf)
                                       filename__decompress (/var/home/acme/bin/perf)
                                       filename__read_build_id (/var/home/acme/bin/perf)
                                       filename__sprintf_build_id (inlined)
                                       build_id_cache__valid_id (inlined)
                                       build_id_cache__list_all (/var/home/acme/bin/perf)
                                       print_sdt_events (/var/home/acme/bin/perf)
                                       cmd_list (/var/home/acme/bin/perf)
                                       run_builtin (/var/home/acme/bin/perf)
                                       handle_internal_command (inlined)
                                       run_argv (inlined)
                                       main (/var/home/acme/bin/perf)
                                       __libc_start_call_main (/usr/lib64/libc.so.6)
                                       __libc_start_main@@GLIBC_2.34 (/usr/lib64/libc.so.6)
                                       _start (/var/home/acme/bin/perf)

But callers of filename__decompress() already check its return and use
pr_debug(), so be consistent and make functions it calls also use
pr_debug().

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZOUD0+GkuCVkYF7n@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf stat-display: Check if snprintf()'s fmt argument is NULL

It is undefined behavior to pass NULL as snprintf()'s fmt argument.
Here is an example to trigger the problem:

  $ perf stat --metric-only -x, -e instructions -- sleep 1
  insn per cycle,
  Segmentation fault (core dumped)

With this patch:

  $ perf stat --metric-only -x, -e instructions -- sleep 1
  insn per cycle,
  ,

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Kaige Ye <ye@kaige.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.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>
Link: https://lore.kernel.org/r/01CA7674B690CA24+20230804020907.144562-2-ye@kaige.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf augmented_raw_syscalls: Add an assert to make sure sizeof(augmented_arg->value) is a power of two.

Similar to what was done in the previous cset for sizeof(saddr), we need
to make sure sizeof(augmented_arg->value) is a power of two to do bounds
checking using &=:

  augmented_len &= sizeof(augmented_arg->value) - 1;

Suggested-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZONrPo0NSqdbXiGx@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf augmented_raw_syscalls: Add an assert to make sure sizeof(saddr) is a power of two.

We're using the BPF verifier suggestion:

    22: (85) call bpf_probe_read#4
    R2 min value is negative, either use unsigned or 'var &= const'

That works only when const is a (power of two - 1) so add an assert to
make sure that that is the case.

Suggested-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZONrFmJBNlQpSpZj@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: AmpereOne: Remove unsupported events

Some of the events included in the ampereone/core-imp-def are not
supported on AmpereOne, remove them.

Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Dave Kleikamp <dave.kleikamp@oracle.com>
Cc: Ian Rogers <irogers@google.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: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230803211331.140553-5-ilkka@os.amperecomputing.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Add AmpereOne metrics

This patch adds AmpereOne metrics. The metrics also work around
the issue related to some of the events.

Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Dave Kleikamp <dave.kleikamp@oracle.com>
Cc: Ian Rogers <irogers@google.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: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230803211331.140553-4-ilkka@os.amperecomputing.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: AmpereOne: Mark affected STALL_* events impacted by errata

Per errata AC03_CPU_29, STALL_SLOT_FRONTEND, STALL_FRONTEND, and STALL
events are not counting as expected. The follow up metrics patch will
include correct way to calculate the impacted events.

Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Dave Kleikamp <dave.kleikamp@oracle.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230803211331.140553-3-ilkka@os.amperecomputing.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Remove L1D_CACHE_LMISS from AmpereOne list

amperene/cache.json file tried to include L1D_CACHE_LMISS while it
doesn't exist in common-and-microarch.json. While this bug doesn't seem to
cause issue in newer kernels with jevents.py script, it prevents building
older perf tools with the backported patch.

Fixes: a9650b7f6fc09d16 ("perf vendor events arm64: Add AmpereOne core PMU events")
Reported-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
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: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Closes: https://lore.kernel.org/all/76bb2e47-ce44-76ae-838e-53279047084d@oracle.com/
Link: https://lore.kernel.org/r/20230803211331.140553-2-ilkka@os.amperecomputing.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf jevents: Raise exception for no definition of a arch std event

Recently Ilkka reported that the JSONs for the AmpereOne arm64-based
platform included a dud event which referenced a non-existent arch std
event [0].

Previously in the times of jevents.c, we would raise an exception for this.

This is still invalid, even though the current code just ignores such an
event.

Re-introduce code to raise an exception for when no definition exists to
help catch as many invalid JSONs as possible.

[0] https://lore.kernel.org/linux-perf-users/9e851e2a-26c7-ba78-cb20-be4337b2916a@oracle.com/

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: John Garry <john.g.garry@oracle.com>
Tested-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Link: https://lore.kernel.org/r/20230807111631.3033102-1-john.g.garry@oracle.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Use heuristic when deciding if a syscall tracepoint "const char *" field is really a string

'perf trace' tries to find BPF progs associated with a syscall that have
a signature that is similar to syscalls without one to try and reuse,
so, for instance, the 'open' signature can be reused with many other
syscalls that have as its first arg a string.

It uses the tracefs events format file for finding a signature that can
be reused, but then comes the "write" syscall with its second argument
as a "const char *":

  # cat /sys/kernel/debug/tracing/events/syscalls/sys_enter_write/format
  name: sys_enter_write
  ID: 746
  format:
   field:unsigned short common_type; offset:0; size:2; signed:0;
   field:unsigned char common_flags; offset:2; size:1; signed:0;
   field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
   field:int common_pid; offset:4; size:4; signed:1;

   field:int __syscall_nr; offset:8; size:4; signed:1;
   field:unsigned int fd; offset:16; size:8; signed:0;
   field:const char * buf; offset:24; size:8; signed:0;
   field:size_t count; offset:32; size:8; signed:0;

  print fmt: "fd: 0x%08lx, buf: 0x%08lx, count: 0x%08lx", ((unsigned long)(REC->fd)), ((unsigned long)(REC->buf)), ((unsigned long)(REC->count))
  #

Which isn't a string (the man page for glibc has buf as "void *"), so we
have to use the name of the argument as an heuristic, to consider a
string just args that are "const char *" and that have in its name  the
"path", "file", etc substrings.

With that now it reuses:

  [root@quaco ~]# perf trace -v --max-events=1 |& grep Reus
  Reusing "open" BPF sys_enter augmenter for "stat"
  Reusing "open" BPF sys_enter augmenter for "lstat"
  Reusing "open" BPF sys_enter augmenter for "access"
  Reusing "connect" BPF sys_enter augmenter for "accept"
  Reusing "sendto" BPF sys_enter augmenter for "recvfrom"
  Reusing "connect" BPF sys_enter augmenter for "bind"
  Reusing "connect" BPF sys_enter augmenter for "getsockname"
  Reusing "connect" BPF sys_enter augmenter for "getpeername"
  Reusing "open" BPF sys_enter augmenter for "execve"
  Reusing "open" BPF sys_enter augmenter for "truncate"
  Reusing "open" BPF sys_enter augmenter for "chdir"
  Reusing "open" BPF sys_enter augmenter for "mkdir"
  Reusing "open" BPF sys_enter augmenter for "rmdir"
  Reusing "open" BPF sys_enter augmenter for "creat"
  Reusing "open" BPF sys_enter augmenter for "link"
  Reusing "open" BPF sys_enter augmenter for "unlink"
  Reusing "open" BPF sys_enter augmenter for "symlink"
  Reusing "open" BPF sys_enter augmenter for "readlink"
  Reusing "open" BPF sys_enter augmenter for "chmod"
  Reusing "open" BPF sys_enter augmenter for "chown"
  Reusing "open" BPF sys_enter augmenter for "lchown"
  Reusing "open" BPF sys_enter augmenter for "mknod"
  Reusing "open" BPF sys_enter augmenter for "statfs"
  Reusing "open" BPF sys_enter augmenter for "pivot_root"
  Reusing "open" BPF sys_enter augmenter for "chroot"
  Reusing "open" BPF sys_enter augmenter for "acct"
  Reusing "open" BPF sys_enter augmenter for "swapon"
  Reusing "open" BPF sys_enter augmenter for "swapoff"
  Reusing "open" BPF sys_enter augmenter for "delete_module"
  Reusing "open" BPF sys_enter augmenter for "setxattr"
  Reusing "open" BPF sys_enter augmenter for "lsetxattr"
  Reusing "openat" BPF sys_enter augmenter for "fsetxattr"
  Reusing "open" BPF sys_enter augmenter for "getxattr"
  Reusing "open" BPF sys_enter augmenter for "lgetxattr"
  Reusing "openat" BPF sys_enter augmenter for "fgetxattr"
  Reusing "open" BPF sys_enter augmenter for "listxattr"
  Reusing "open" BPF sys_enter augmenter for "llistxattr"
  Reusing "open" BPF sys_enter augmenter for "removexattr"
  Reusing "open" BPF sys_enter augmenter for "lremovexattr"
  Reusing "fsetxattr" BPF sys_enter augmenter for "fremovexattr"
  Reusing "open" BPF sys_enter augmenter for "mq_open"
  Reusing "open" BPF sys_enter augmenter for "mq_unlink"
  Reusing "fsetxattr" BPF sys_enter augmenter for "add_key"
  Reusing "fremovexattr" BPF sys_enter augmenter for "request_key"
  Reusing "fremovexattr" BPF sys_enter augmenter for "inotify_add_watch"
  Reusing "fremovexattr" BPF sys_enter augmenter for "mkdirat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "mknodat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "fchownat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "futimesat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "newfstatat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "unlinkat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "linkat"
  Reusing "open" BPF sys_enter augmenter for "symlinkat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "readlinkat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "fchmodat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "faccessat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "utimensat"
  Reusing "connect" BPF sys_enter augmenter for "accept4"
  Reusing "fremovexattr" BPF sys_enter augmenter for "name_to_handle_at"
  Reusing "fremovexattr" BPF sys_enter augmenter for "renameat2"
  Reusing "open" BPF sys_enter augmenter for "memfd_create"
  Reusing "fremovexattr" BPF sys_enter augmenter for "execveat"
  Reusing "fremovexattr" BPF sys_enter augmenter for "statx"
  [root@quaco ~]#

Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alan Maguire <alan.maguire@oracle.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZN5lrdeEdSMCn7hk@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Use the augmented_raw_syscall BPF skel only for tracing syscalls

It is possible to use 'perf trace' with tracepoints and in that case we
can't initialize/use the augmented_raw_syscalls BPF skel.

For instance, this usecase:

  # perf trace -e sched:*exec --max-events=5
         ? (         ): NetworkManager/1183  ... [continued]: poll())                                             = 1
     0.043 ( 0.007 ms): NetworkManager/1183 epoll_wait(epfd: 17<anon_inode:[eventpoll]>, events: 0x55555f90e920, maxevents: 6) = 0
     0.060 ( 0.007 ms): NetworkManager/1183 write(fd: 3<anon_inode:[eventfd]>, buf: 0x7ffc5a27cd30, count: 8)     = 8
     0.073 ( 0.005 ms): NetworkManager/1183 epoll_wait(epfd: 24<anon_inode:[eventpoll]>, events: 0x7ffc5a27cd20, maxevents: 2) = 1
     0.082 ( 0.010 ms): NetworkManager/1183 recvmmsg(fd: 26<socket:[30298]>, mmsg: 0x7ffc5a27caa0, vlen: 8)       = 1
  #

Where we want to trace just some sched tracepoints ending in 'exec' ends
up tracing all syscalls.

Fix it by checking existing trace->trace_syscalls boolean to see if we
need the augmenter.

A followup patch will move those sections of code used only with the
augmenter to separate functions, to get it cleaner and remove the goto,
done just for reviewing purposes.

With this patch in place the previous behaviour is restored: no syscalls
when we have other events and no syscall names:

  [root@quaco ~]# perf probe do_filp_open "filename=pathname->name:string"
  Added new event:
    probe:do_filp_open   (on do_filp_open with filename=pathname->name:string)

  You can now use it in all perf tools, such as:

  perf record -e probe:do_filp_open -aR sleep 1

  [root@quaco ~]# perf trace --max-events=10 -e probe:do_filp_open sleep 1
     0.000 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/etc/ld.so.cache")
     0.056 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/lib64/libc.so.6")
     0.481 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/locale-archive")
     0.501 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/share/locale/locale.alias")
     0.572 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/en_US.UTF-8/LC_IDENTIFICATION")
     0.581 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/en_US.utf8/LC_IDENTIFICATION")
     0.616 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib64/gconv/gconv-modules.cache")
     0.656 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/en_US.UTF-8/LC_MEASUREMENT")
     0.664 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/en_US.utf8/LC_MEASUREMENT")
     0.696 sleep/455122 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/en_US.UTF-8/LC_TELEPHONE")
  [root@quaco ~]#

As well as mixing syscalls with tracepoints, getting the syscall
tracepoints used augmented using the BPF skel:

  [root@quaco ~]# perf trace --max-events=10 -e open*,probe:do_filp_open sleep 1
     0.000 (         ): sleep/455124 openat(dfd: CWD, filename: "/etc/ld.so.cache", flags: RDONLY|CLOEXEC) ...
     0.005 (         ): sleep/455124 probe:do_filp_open(__probe_ip: -1186560412, filename: "/etc/ld.so.cache")
     0.000 ( 0.011 ms): sleep/455124  ... [continued]: openat())                                           = 3
     0.031 (         ): sleep/455124 openat(dfd: CWD, filename: "/lib64/libc.so.6", flags: RDONLY|CLOEXEC) ...
     0.033 (         ): sleep/455124 probe:do_filp_open(__probe_ip: -1186560412, filename: "/lib64/libc.so.6")
     0.031 ( 0.006 ms): sleep/455124  ... [continued]: openat())                                           = 3
     0.258 (         ): sleep/455124 openat(dfd: CWD, filename: "/usr/lib/locale/locale-archive", flags: RDONLY|CLOEXEC) ...
     0.261 (         ): sleep/455124 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/lib/locale/locale-archive")
     0.258 ( 0.006 ms): sleep/455124  ... [continued]: openat())                                           = -1 ENOENT (No such file or directory)
     0.272 (         ): sleep/455124 openat(dfd: CWD, filename: "/usr/share/locale/locale.alias", flags: RDONLY|CLOEXEC) ...
     0.273  (        ): sleep/455124 probe:do_filp_open(__probe_ip: -1186560412, filename: "/usr/share/locale/locale.alias")

A final note: the probe:do_filp_open uses a kprobe (probably optimized
as its in the start of a function) that uses the kprobe_tracer mechanism
in the kernel to collect the pathname->name string and stash it into the
tracepoint created by 'perf probe' for that:

  [root@quaco ~]# cat /sys/kernel/debug/tracing/kprobe_events
  p:probe/do_filp_open _text+4621920 filename=+0(+0(%si)):string
  [root@quaco ~]#

While the syscalls:sys_enter_openat tracepoint gets its string from a
BPF program attached to raw_syscalls:sys_enter that tail calls into
another BPF program that knows the types for the openat syscall args and
thus can bpf_probe_read it right after the normal
sys_enter/sys_enter_openat tracepoint payload that comes prefixed with
whatever perf_event_open asked for (CPU, timestamp, etc):

  [root@quaco ~]# bpftool prog | grep -E "sys_enter |sys_enter_opena" -A3
  3176: tracepoint  name sys_enter  tag 0bc3fc9d11754ba1  gpl
loaded_at 2023-08-17T12:32:20-0300  uid 0
xlated 272B  jited 257B  memlock 4096B  map_ids 2462,2466,2463
btf_id 2976
  --
  3180: tracepoint  name sys_enter_opena  tag 19dd077f00ec2f58  gpl
  loaded_at 2023-08-17T12:32:20-0300  uid 0
  xlated 328B  jited 206B  memlock 4096B  map_ids 2466,2465
  btf_id 2976
  [root@quaco ~]#

Fixes: 5e6da6be3082f77b ("perf trace: Migrate BPF augmentation to use a skeleton")
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/lkml/ZN4+s2Wl+zYmXTDj@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf lock: Don't pass an ERR_PTR() directly to perf_session__delete()

While debugging a segfault on 'perf lock contention' without an
available perf.data file I noticed that it was basically calling:

perf_session__delete(ERR_PTR(-1))

Resulting in:

  (gdb) run lock contention
  Starting program: /root/bin/perf lock contention
  [Thread debugging using libthread_db enabled]
  Using host libthread_db library "/lib64/libthread_db.so.1".
  failed to open perf.data: No such file or directory  (try 'perf record' first)
  Initializing perf session failed

  Program received signal SIGSEGV, Segmentation fault.
  0x00000000005e7515 in auxtrace__free (session=0xffffffffffffffff) at util/auxtrace.c:2858
  2858 if (!session->auxtrace)
  (gdb) p session
  $1 = (struct perf_session *) 0xffffffffffffffff
  (gdb) bt
  #0  0x00000000005e7515 in auxtrace__free (session=0xffffffffffffffff) at util/auxtrace.c:2858
  #1  0x000000000057bb4d in perf_session__delete (session=0xffffffffffffffff) at util/session.c:300
  #2  0x000000000047c421 in __cmd_contention (argc=0, argv=0x7fffffffe200) at builtin-lock.c:2161
  #3  0x000000000047dc95 in cmd_lock (argc=0, argv=0x7fffffffe200) at builtin-lock.c:2604
  #4  0x0000000000501466 in run_builtin (p=0xe597a8 <commands+552>, argc=2, argv=0x7fffffffe200) at perf.c:322
  #5  0x00000000005016d5 in handle_internal_command (argc=2, argv=0x7fffffffe200) at perf.c:375
  #6  0x0000000000501824 in run_argv (argcp=0x7fffffffe02c, argv=0x7fffffffe020) at perf.c:419
  #7  0x0000000000501b11 in main (argc=2, argv=0x7fffffffe200) at perf.c:535
  (gdb)

So just set it to NULL after using PTR_ERR(session) to decode the error
as perf_session__delete(NULL) is supported.

Fixes: eef4fee5e52071d5 ("perf lock: Dynamically allocate lockhash_table")
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: K Prateek Nayak <kprateek.nayak@amd.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mamatha Inamdar <mamatha4@linux.vnet.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Ross Zwisler <zwisler@chromium.org>
Cc: Sean Christopherson <seanjc@google.com>
Cc: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Yang Jihong <yangjihong1@huawei.com>
Link: https://lore.kernel.org/lkml/ZN4R1AYfsD2J8lRs@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf top: Don't pass an ERR_PTR() directly to perf_session__delete()

While debugging a segfault on 'perf lock contention' without an
available perf.data file I noticed that it was basically calling:

perf_session__delete(ERR_PTR(-1))

Resulting in:

  (gdb) run lock contention
  Starting program: /root/bin/perf lock contention
  [Thread debugging using libthread_db enabled]
  Using host libthread_db library "/lib64/libthread_db.so.1".
  failed to open perf.data: No such file or directory  (try 'perf record' first)
  Initializing perf session failed

  Program received signal SIGSEGV, Segmentation fault.
  0x00000000005e7515 in auxtrace__free (session=0xffffffffffffffff) at util/auxtrace.c:2858
  2858 if (!session->auxtrace)
  (gdb) p session
  $1 = (struct perf_session *) 0xffffffffffffffff
  (gdb) bt
  #0  0x00000000005e7515 in auxtrace__free (session=0xffffffffffffffff) at util/auxtrace.c:2858
  #1  0x000000000057bb4d in perf_session__delete (session=0xffffffffffffffff) at util/session.c:300
  #2  0x000000000047c421 in __cmd_contention (argc=0, argv=0x7fffffffe200) at builtin-lock.c:2161
  #3  0x000000000047dc95 in cmd_lock (argc=0, argv=0x7fffffffe200) at builtin-lock.c:2604
  #4  0x0000000000501466 in run_builtin (p=0xe597a8 <commands+552>, argc=2, argv=0x7fffffffe200) at perf.c:322
  #5  0x00000000005016d5 in handle_internal_command (argc=2, argv=0x7fffffffe200) at perf.c:375
  #6  0x0000000000501824 in run_argv (argcp=0x7fffffffe02c, argv=0x7fffffffe020) at perf.c:419
  #7  0x0000000000501b11 in main (argc=2, argv=0x7fffffffe200) at perf.c:535
  (gdb)

So just set it to NULL after using PTR_ERR(session) to decode the error
as perf_session__delete(NULL) is supported.

The same problem was found in 'perf top' after an audit of all
perf_session__new() failure handling.

Fixes: 6ef81c55a2b6584c ("perf session: Return error code for perf_session__new() function on failure")
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexey Budankov <alexey.budankov@linux.intel.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Jeremie Galarneau <jeremie.galarneau@efficios.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kate Stewart <kstewart@linuxfoundation.org>
Cc: Mamatha Inamdar <mamatha4@linux.vnet.ibm.com>
Cc: Mukesh Ojha <mojha@codeaurora.org>
Cc: Nageswara R Sastry <rnsastry@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Shawn Landden <shawn@git.icu>
Cc: Song Liu <songliubraving@fb.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Tzvetomir Stoyanov <tstoyanov@vmware.com>
Link: https://lore.kernel.org/lkml/ZN4Q2rxxsL08A8rd@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Update N2 and V2 metrics and events using Arm telemetry repo

Apart from some slight naming and grouping differences, the new metrics
are functionally the same as the existing ones. Any missing metrics were
manually appended to the end of the auto generated file.

For the events, the new data includes descriptions that may have product
specific details and new groupings that will be consistent with other
products.

After generating the metrics from the telemetry repo [1], the following
manual steps were performed:

 * Change the topdown expressions to compare on CPUID and use
   #slots so that the same data can be shared between N2 and V2. Apart
   from these modifications, the expressions now match more closely with
   the Arm telemetry data which will hopefully make future updates
   easier.

 * Append some metrics from the old N2/V2 data that aren't present in
   the telemetry data. These will possibly be added to the
   telemetry-solution repo at a later time:

    l3d_cache_mpki, l3d_cache_miss_rate, branch_pki, ipc_rate, spec_ipc,
    retired_rate, wasted_rate, branch_immed_spec_rate,
    branch_return_spec_rate, branch_indirect_spec_rate

[1]: https://gitlab.arm.com/telemetry-solution/telemetry-solution/-/blob/main/data/pmu/cpu/neoverse/neoverse-n2.json

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Haixin Yu <yuhaixin.yhx@linux.alibaba.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230816114841.1679234-7-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Update stall_slot workaround for N2 r0p3

N2 r0p3 doesn't require the workaround [1], so gating on (#slots - 5) no
longer works because all N2s have 5 slots. Use the new expression
builtin that allows calling strcmp_cpuid_str() and comparing CPUIDs in
metric formulas.

In this case, the commented formula looks like this:

  strcmp_cpuid_str(0x410fd493)        # greater than or equal to N2 r0p3
  | strcmp_cpuid_str(0x410fd490) ^ 1  # OR NOT any version of N2

[1]: https://gitlab.arm.com/telemetry-solution/telemetry-solution/-/blob/main/data/pmu/cpu/neoverse/neoverse-n2-r0p3.json

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Haixin Yu <yuhaixin.yhx@linux.alibaba.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230816114841.1679234-6-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf jevents: Add a new expression builtin strcmp_cpuid_str()

This will allow writing formulas that are conditional on a specific
CPU type or CPU version. It calls through to the existing
strcmp_cpuid_str() function in Perf which has a default weak version,
and an arch specific version for x86 and arm64.

The function takes an 'ID' type value, which is a string. But in this
case Arm CPU IDs are hex numbers prefixed with '0x'. metric.py
assumes strings are only used by event names, and that they can't start
with a number ('0'), so an additional change has to be made to the
regex to convert hex numbers back to 'ID' types.

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Haixin Yu <yuhaixin.yhx@linux.alibaba.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230816114841.1679234-5-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Add a test for the new Arm CPU ID comparison behavior

Now that variant and revision fields are taken into account the behavior
is slightly more complicated so add a test to ensure that this behaves
as expected.

Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: James Clark <james.clark@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Haixin Yu <yuhaixin.yhx@linux.alibaba.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230816114841.1679234-3-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf arm64: Allow version comparisons of CPU IDs

Currently variant and revision fields are masked out of the MIDR so
it's not possible to compare different versions of the same CPU.
In a later commit a workaround will be removed just for N2 r0p3, so
enable comparisons on version.

This has the side effect of changing the MIDR stored in the header of
the perf.data file to no longer have masked version fields. It also
affects the lookups in mapfile.csv, but as that currently only has
zeroed version fields, it has no actual effect. The mapfile.csv
documentation also states to zero the version fields, so unless this
isn't done it will continue to have no effect.

There is an existing weak default strcmp_cpuid_str() function, and an
x86 version. This adds another version for arm64.

Signed-off-by: James Clark <james.clark@arm.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Haixin Yu <yuhaixin.yhx@linux.alibaba.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230816114841.1679234-2-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf_skel augmented_raw_syscalls: Cap the socklen parameter using &= sizeof(saddr)

This works with:

  $ clang -v
  clang version 14.0.5 (Fedora 14.0.5-2.fc36)
  $

But not with:

  $ clang -v
  clang version 16.0.6 (Fedora 16.0.6-2.fc38)
  $

  [root@quaco ~]# perf trace -e connect*,sendto* ping -c 10 localhost
  libbpf: prog 'sys_enter_sendto': BPF program load failed: Permission denied
  libbpf: prog 'sys_enter_sendto': -- BEGIN PROG LOAD LOG --
  reg type unsupported for arg#0 function sys_enter_sendto#59
  0: R1=ctx(off=0,imm=0) R10=fp0
  ; int sys_enter_sendto(struct syscall_enter_args *args)
  0: (bf) r6 = r1                       ; R1=ctx(off=0,imm=0) R6_w=ctx(off=0,imm=0)
  1: (b7) r1 = 0                        ; R1_w=0
  ; int key = 0;
  2: (63) *(u32 *)(r10 -4) = r1         ; R1_w=0 R10=fp0 fp-8=0000????
  3: (bf) r2 = r10                      ; R2_w=fp0 R10=fp0
  ;
  4: (07) r2 += -4                      ; R2_w=fp-4
  ; return bpf_map_lookup_elem(&augmented_args_tmp, &key);
  5: (18) r1 = 0xffff8de5a5b8bc00       ; R1_w=map_ptr(off=0,ks=4,vs=8272,imm=0)
  7: (85) call bpf_map_lookup_elem#1    ; R0_w=map_value_or_null(id=1,off=0,ks=4,vs=8272,imm=0)
  8: (bf) r7 = r0                       ; R0_w=map_value_or_null(id=1,off=0,ks=4,vs=8272,imm=0) R7_w=map_value_or_null(id=1,off=0,ks=4,vs=8272,imm=0)
  9: (b7) r0 = 1                        ; R0_w=1
  ; if (augmented_args == NULL)
  10: (15) if r7 == 0x0 goto pc+25      ; R7_w=map_value(off=0,ks=4,vs=8272,imm=0)
  ; unsigned int socklen = args->args[5];
  11: (79) r1 = *(u64 *)(r6 +56)        ; R1_w=scalar() R6_w=ctx(off=0,imm=0)
  ;
  12: (bf) r2 = r1                      ; R1_w=scalar(id=2) R2_w=scalar(id=2)
  13: (67) r2 <<= 32                    ; R2_w=scalar(smax=9223372032559808512,umax=18446744069414584320,var_off=(0x0; 0xffffffff00000000),s32_min=0,s32_max=0,u32_max=0)
  14: (77) r2 >>= 32                    ; R2_w=scalar(umax=4294967295,var_off=(0x0; 0xffffffff))
  15: (b7) r8 = 128                     ; R8=128
  ; if (socklen > sizeof(augmented_args->saddr))
  16: (25) if r2 > 0x80 goto pc+1       ; R2=scalar(umax=128,var_off=(0x0; 0xff))
  17: (bf) r8 = r1                      ; R1=scalar(id=2) R8_w=scalar(id=2)
  ; const void *sockaddr_arg = (const void *)args->args[4];
  18: (79) r3 = *(u64 *)(r6 +48)        ; R3_w=scalar() R6=ctx(off=0,imm=0)
  ; bpf_probe_read(&augmented_args->saddr, socklen, sockaddr_arg);
  19: (bf) r1 = r7                      ; R1_w=map_value(off=0,ks=4,vs=8272,imm=0) R7=map_value(off=0,ks=4,vs=8272,imm=0)
  20: (07) r1 += 64                     ; R1_w=map_value(off=64,ks=4,vs=8272,imm=0)
  ; bpf_probe_read(&augmented_args->saddr, socklen, sockaddr_arg);
  21: (bf) r2 = r8                      ; R2_w=scalar(id=2) R8_w=scalar(id=2)
  22: (85) call bpf_probe_read#4
  R2 min value is negative, either use unsigned or 'var &= const'
  processed 22 insns (limit 1000000) max_states_per_insn 0 total_states 1 peak_states 1 mark_read 1
  -- END PROG LOAD LOG --
  libbpf: prog 'sys_enter_sendto': failed to load: -13
  libbpf: failed to load object 'augmented_raw_syscalls_bpf'
  libbpf: failed to load BPF skeleton 'augmented_raw_syscalls_bpf': -13

So use the suggested &= variant since sizeof(saddr) == 128 bytes.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Update metric events for power10 platform

Update JSON/events for power10 platform with additional metrics.

Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-7-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Update metric event names for power10 platform

Update metric event name for some of the JSON/metric events for
power10 platform.

Fixes: 3ca3af7d1f230d1f ("perf vendor events power10: Add metric events JSON file for power10 platform")
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-6-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Update JSON/events for power10 platform

Update JSON/events for power10 platform with additional events.

Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-5-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Move JSON/events to appropriate files for power10 platform

Move some of the power10 JSON/events to appropriate files.

Fixes: 32daa5d7899e0343 ("perf vendor events: Initial JSON/events list for power10 platform")
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-4-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Drop STORES_PER_INST metric event for power10 platform

Drop STORES_PER_INST metric event for the power10 platform, as the
metric expression of STORES_PER_INST metric event using dropped event
PM_ST_FIN.

Fixes: 3ca3af7d1f230d1f ("perf vendor events power10: Add metric events JSON file for power10 platform")
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-3-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Drop some of the JSON/events for power10 platform

Drop some of the JSON/events for power10 platform due to counter
data mismatch.

Fixes: 32daa5d7899e0343 ("perf vendor events: Initial JSON/events list for power10 platform")
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-2-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events: Update the JSON/events descriptions for power10 platform

Update the description for some of the JSON/events for power10 platform.

Fixes: 32daa5d7899e0343 ("perf vendor events: Initial JSON/events list for power10 platform")
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.ibm.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linuxppc-dev@lists.ozlabs.org
Link: https://lore.kernel.org/r/20230814112803.1508296-1-kjain@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tests mmap-basic: Adapt for riscv

riscv now supports mmaping hardware counters to userspace so adapt the
test to run on this architecture.

Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Anup Patel <anup@brainfault.org>
Cc: Atish Patra <atishp@atishpatra.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Rémi Denis-Courmont <remi@remlab.net>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-doc@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230802080328.1213905-11-alexghiti@rivosinc.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

libperf: Implement riscv mmap support

riscv now supports mmaping hardware counters so add what's needed to
take advantage of that in libperf.

Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Atish Patra <atishp@rivosinc.com>
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Anup Patel <anup@brainfault.org>
Cc: Atish Patra <atishp@atishpatra.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Jonathan Corbet <corbet@lwn.net>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Rémi Denis-Courmont <remi@remlab.net>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-doc@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230802080328.1213905-10-alexghiti@rivosinc.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-regs: Move out arch specific header from util/perf_regs.h

util/perf_regs.h includes another perf_regs.h:

  #include <perf_regs.h>

Here it includes architecture specific header, for example, if we build
arm64 target, the header tools/perf/arch/arm64/include/perf_regs.h is
included.

We use this implicit way to include architecture specific header, which
is not directive; furthermore, util/perf_regs.c is coupled with the
architecture specific definitions.

This patch moves out arch specific header from util/perf_regs.h for
generalizing the 'util' folder, as a result, the source files in 'arch'
folder explicitly include architecture's perf_regs.h.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-7-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-regs: Remove PERF_REGS_{MAX|MASK} from common code

The macros PERF_REGS_MAX and PERF_REGS_MASK are architecture specific,
let's remove them from the common file util/perf_regs.c.

As a side effect, the weak functions arch__intr_reg_mask() and
arch__user_reg_mask() just return zeros, every arch defines its own
functions in the 'arch' folder for returning right values.

Note, we don't need to return intr/user register masks dynamically, this
is because these two functions are invoked during recording phase but
not decoding phase, they are always invoked on the native environment,
thus we don't need to parse them dynamically.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-6-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-regs: Remove unused macros PERF_REG_{IP|SP}

The macros PERF_REG_{IP|SP} have been replaced by using functions
perf_arch_reg_{ip|sp}(), remove them!

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-5-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf unwind: Use perf_arch_reg_{ip|sp}() to substitute macros

We use perf_arch_reg_ip() and perf_arch_reg_sp() to substitute macros
for obtaining the register numbers of SP and IP.  This modification
enables cross analysis in the unwinding, therefore, the unwinding is
not restricted to the predefined values by the macros.

Consequently, the macros LIBUNWIND__ARCH_REG_{IP|SP} are removed since
they are no longer used.

Committer notes:

Add missing "util/env.h" header to make sure we have the definition for
perf_env__arch(), that when built with NO_LIBUNWIND=1 isn't available,
i.e. it was being included by sheer luck.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-4-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-regs: Introduce functions perf_arch_reg_{ip|sp}()

The current code uses macros PERF_REG_IP and PERF_REG_SP for parsing
registers and we build perf with these macros statically, which means it
only can correctly analyze CPU registers for the native architecture and
fails to support cross analysis (e.g. we build perf on x86 and cannot
analyze Arm64's registers).

We need to generalize util/perf_regs.c for support multi architectures,
as a first step, this commit introduces new functions perf_arch_reg_ip()
and perf_arch_reg_sp(), these two functions dynamically return IP and SP
register index respectively according to the parameter "arch".

Every architecture has its own functions (like __perf_reg_ip_arm64 and
__perf_reg_sp_arm64), these architecture specific functions are defined
in each arch source file under folder util/perf-regs-arch; at the end
all of them are built into the tool for cross analysis.

Committer notes:

Make DWARF_MINIMAL_REGS() an inline function, so that we can use the
__maybe_unused attribute for the 'arch' parameter, as this will avoid a
build failure when that variable is unused in the callers. That happens
when building on unsupported architectures, the ones without
HAVE_PERF_REGS_SUPPORT defined.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-3-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-regs: Refactor arch register parsing functions

Every architecture has a specific register parsing function for
returning register name based on register index, to support cross
analysis (e.g. we use perf x86 binary to parse Arm64's perf data), we
build all these register parsing functions into the tool, this is why
we place all related functions into util/perf_regs.c.

Unfortunately, since util/perf_regs.c needs to include every arch's
perf_regs.h, this easily introduces duplicated definitions coming from
multiple headers, finally it's fragile for building and difficult for
maintenance.

We cannot simply move these register parsing functions into the
corresponding 'arch' folder, the folder is only conditionally built
based on the target architecture.

Therefore, this commit creates a new folder util/perf-regs-arch/ and
uses a dedicated source file to keep every architecture's register
parsing function to avoid definition conflicts.

This is only a refactoring, no functionality change is expected.

Committer notes:

Had to add util/perf-regs-arch/*.c to tools/perf/util/python-ext-sources
to keep 'perf test python' passing.

Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Eric Lin <eric.lin@sifive.com>
Cc: Fangrui Song <maskray@google.com>
Cc: Guo Ren <guoren@kernel.org>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Ivan Babrou <ivan@cloudflare.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: John Garry <john.g.garry@oracle.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Ming Wang <wangming01@loongson.cn>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Paul Walmsley <paul.walmsley@sifive.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sandipan Das <sandipan.das@amd.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: linux-csky@vger.kernel.org
Cc: linux-riscv@lists.infradead.org
Link: https://lore.kernel.org/r/20230606014559.21783-2-leo.yan@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf docs: Fix format of unordered lists

Fix the format of unordered lists so the can wrap properly.

Signed-off-by: Changbin Du <changbin.du@huawei.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.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>
Link: https://lore.kernel.org/r/20230718085242.3090797-1-changbin.du@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf vendor events arm64: Update scale units and descriptions of common topdown metrics

Metrics will be published here [1] going forwards, but they have
slightly different scale units. To allow autogenerated metrics to be
added more easily, update the scale units to match.

The more detailed descriptions have also been taken and added to the
common file.

[1]: https://gitlab.arm.com/telemetry-solution/telemetry-solution/-/tree/main/data/pmu/cpu/

Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: James Clark <james.clark@arm.com>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: coresight@lists.linaro.org
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230811144017.491628-5-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf cs-etm: Don't duplicate FIELD_GET()

linux/bitfield.h can be included as long as linux/kernel.h is included
first, so change the order of the includes and drop the duplicate macro.

Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: James Clark <james.clark@arm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jing Zhang <renyu.zj@linux.alibaba.com>
Cc: Jiri Olsa <jolsa@kernel.org>
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: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nick Forrington <nick.forrington@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Sohom Datta <sohomdatta1@gmail.com>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: coresight@lists.linaro.org
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230811144017.491628-2-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dlfilter: Add al_cleanup()

Add perf_dlfilter_fns.al_cleanup() to do addr_location__exit() on data
passed via perf_dlfilter_fns.resolve_address().

Add dlfilter-test-api-v2 to the "dlfilter C API" test to test it.

Update documentation, clarifying that data returned by APIs should not
be dereferenced after filter_event() and filter_event_early() return.

Fixes: 0dd5041c9a0eaf8c ("perf addr_location: Add init/exit/copy functions")
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20230731091857.10681-3-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dlfilter: Initialize addr_location before passing it to thread__find_symbol_fb()

As thread__find_symbol_fb() will end up calling thread__find_map() and
it in turn will call these on uninitialized memory:

        maps__zput(al->maps);
        map__zput(al->map);
        thread__zput(al->thread);

Fixes: 0dd5041c9a0eaf8c ("perf addr_location: Add init/exit/copy functions")
Reviewed-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Disha Goel <disgoel@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20230731091857.10681-2-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf dlfilter: Add a test for resolve_address()

Extend the "dlfilter C API" test to test
perf_dlfilter_fns.resolve_address(). The test currently fails, but passes
after a subsequent patch.

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20230731091857.10681-1-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts python: Update audit-libs package name for python3

'audit-libs-python' is the package for python2, update it for python3.
On Ubuntu and Fedora, the new package is 'python3-audit'.

Signed-off-by: Wei Li <liwei391@huawei.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Li Bin <huawei.libin@huawei.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230815131805.1237491-1-liwei391@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf scripts python: Support syscall name parsing on arm64

In the result of "perf script syscall-counts" on arm64, the syscall
events are not resolved currently. Add "aarch64" to audit uname list to
support name parsing.

* After the patch:

  [root@localhost ~]# perf script syscall-counts sleep 1
  Press control+C to stop and show the summary

  syscall events:

  event                                          count
  ----------------------------------------  -----------
  mmap                                               6
  close                                              5
  mprotect                                           4
  brk                                                3
  newfstatat                                         3
  openat                                             3
  getrandom                                          1
  prlimit64                                          1
  munmap                                             1
  clock_nanosleep                                    1
  set_robust_list                                    1
  set_tid_address                                    1
  exit_group                                         1
  read                                               1
  faccessat                                          1

Signed-off-by: Wei Li <liwei391@huawei.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Li Bin <huawei.libin@huawei.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230815131735.1237221-1-liwei391@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf evsel: Remove duplicate check for `field` in evsel__intval()

The `file` parameter in evsel__intval() is checked repeatedly, fix it.

No functional change.

Signed-off-by: Yang Jihong <yangjihong1@huawei.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.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: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Sandipan Das <sandipan.das@amd.com>
Link: https://lore.kernel.org/r/20230815221009.3641751-1-yangjihong1@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test: Add perf record sample filtering test

  $ sudo ./perf test 'sample filter' -v
   94: perf record sample filtering (by BPF) tests                     :
  --- start ---
  test child forked, pid 3817527
  Checking BPF-filter privilege
  Basic bpf-filter test
  Basic bpf-filter test [Success]
  Failing bpf-filter test
  Error: task-clock event does not have PERF_SAMPLE_CPU
  Failing bpf-filter test [Success]
  Group bpf-filter test
  Error: task-clock event does not have PERF_SAMPLE_CPU
  Error: task-clock event does not have PERF_SAMPLE_CODE_PAGE_SIZE
  Group bpf-filter test [Success]
  test child finished with 0
  ---- end ----
  perf record sample filtering (by BPF) tests: Ok

Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230811025822.3859771-2-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf-filter: Fix sample flag check with ||

For logical OR operator, the actual sample_flags are in the 'groups'
list so it needs to check entries in the list instead.  Otherwise it
would show the following error message.

  $ sudo perf record -a -e cycles:p --filter 'period > 100 || weight > 0' sleep 1
  Error: cycles:p event does not have sample flags 0
  failed to set filter "BPF" on event cycles:p with 2 (No such file or directory)

Actually it should warn on 'weight' is used without WEIGHT flag.

  Error: cycles:p event does not have PERF_SAMPLE_WEIGHT
   Hint: please add -W option to perf record
  failed to set filter "BPF" on event cycles:p with 2 (No such file or directory)

Fixes: 4310551b76e0d676 ("perf bpf filter: Show warning for missing sample flags")
Reviewed-by: Ian Rogers <irogers@google.com>
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230811025822.3859771-1-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Tidy comments related to BPF + syscall augmentation

Now tools/perf/examples/bpf/augmented_syscalls.c is
tools/perf/util/bpf_skel/augmented_syscalls.bpf.c and not enabled as a
BPF event, tidy the comments to reflect this.

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: bpf@vger.kernel.org
Cc: llvm@lists.linux.dev
Link: https://lore.kernel.org/r/20230810184853.2860737-5-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf examples: With no BPF events remove examples

The examples were used to give demonstrations of BPF events but such
functionality is now subsumed by using --filter with 'perf record' or
the direct use of BPF skeletons.

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: bpf@vger.kernel.org
Cc: llvm@lists.linux.dev
Link: https://lore.kernel.org/r/20230810184853.2860737-4-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf trace: Migrate BPF augmentation to use a skeleton

Previously a BPF event of augmented_raw_syscalls.c could be used to
enable augmentation of syscalls by perf trace. As BPF events are no
longer supported, switch to using a BPF skeleton which when attached
explicitly opens the sysenter and sysexit tracepoints.

The dump map is removed as debugging wasn't supported by the
augmentation and bpf_printk can be used when necessary.

Remove tools/perf/examples/bpf/augmented_raw_syscalls.c so that the
rename/migration to a BPF skeleton captures that this was the source.

Committer notes:

Some minor stylistic changes to help visualizing the diff.

Use libbpf_strerror when failing to load the augmented raw syscalls BPF.

Use  bpf_object__for_each_program(prog, trace.skel->obj) to disable auto
attachment for all but the sys_enter, sys_exit tracepoints, to avoid
having to add extra lines as we go adding support for more pointer
receiving syscalls.

Committer testing:

  # perf trace -e open*  --max-events=10
     0.000 ( 0.022 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC)    = 11
   208.833 (         ): gnome-terminal/3223 openat(dfd: CWD, filename: "/proc/51250/cmdline")                  ...
   249.993 ( 0.024 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC)    = 11
   250.118 ( 0.030 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.pressure", flags: RDONLY|CLOEXEC) = 11
   250.205 ( 0.016 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.current", flags: RDONLY|CLOEXEC) = 11
   250.244 ( 0.014 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.min", flags: RDONLY|CLOEXEC) = 11
   250.282 ( 0.014 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.low", flags: RDONLY|CLOEXEC) = 11
   250.320 ( 0.014 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.swap.current", flags: RDONLY|CLOEXEC) = 11
   250.355 ( 0.014 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/memory.stat", flags: RDONLY|CLOEXEC) = 11
   250.717 ( 0.016 ms): systemd-oomd/1151 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1001.slice/user@1001.service/memory.pressure", flags: RDONLY|CLOEXEC) = 11
  #
  # perf trace -e *nanosleep*  --max-events=10
         ? (         ): SCTP timer/28304  ... [continued]: clock_nanosleep())                                  = 0
     0.007 (10.058 ms): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) = 0
    10.069 (         ): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) ...
    10.069 (10.056 ms): SCTP timer/28304  ... [continued]: clock_nanosleep())                                  = 0
    17.059 (         ): podman/3572 nanosleep(rqtp: 0x7fc4f4d75be0)                                    ...
    17.059 (10.061 ms): podman/3572  ... [continued]: nanosleep())                                        = 0
    20.131 (10.059 ms): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) = 0
    30.195 (10.038 ms): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) = 0
    40.238 (10.057 ms): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) = 0
    50.301 (         ): SCTP timer/28304 clock_nanosleep(rqtp: { .tv_sec: 0, .tv_nsec: 10000000 }, rmtp: 0x7f0466b78de0) ...
  #

  # perf trace -e perf_event*  -- perf stat -e instructions,cycles,cache-misses sleep 0.1
     0.000 ( 0.011 ms): perf/51331 perf_event_open(attr_uptr: { type: 0 (PERF_TYPE_HARDWARE), size: 136, config: 0x1 (PERF_COUNT_HW_INSTRUCTIONS), sample_type: IDENTIFIER, read_format: TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING, disabled: 1, inherit: 1, enable_on_exec: 1, exclude_guest: 1 }, pid: 51332 (perf), cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 3
     0.013 ( 0.003 ms): perf/51331 perf_event_open(attr_uptr: { type: 0 (PERF_TYPE_HARDWARE), size: 136, config: 0 (PERF_COUNT_HW_CPU_CYCLES), sample_type: IDENTIFIER, read_format: TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING, disabled: 1, inherit: 1, enable_on_exec: 1, exclude_guest: 1 }, pid: 51332 (perf), cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 4
     0.017 ( 0.002 ms): perf/51331 perf_event_open(attr_uptr: { type: 0 (PERF_TYPE_HARDWARE), size: 136, config: 0x3 (PERF_COUNT_HW_CACHE_MISSES), sample_type: IDENTIFIER, read_format: TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING, disabled: 1, inherit: 1, enable_on_exec: 1, exclude_guest: 1 }, pid: 51332 (perf), cpu: -1, group_fd: -1, flags: FD_CLOEXEC) = 5

 Performance counter stats for 'sleep 0.1':

         1,495,051      instructions                     #    1.11  insn per cycle
         1,347,641      cycles
            35,424      cache-misses

       0.100935279 seconds time elapsed

       0.000924000 seconds user
       0.000000000 seconds sys

  #

  # perf trace -e connect*  ssh localhost
       0.000 ( 0.012 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.118 ( 0.004 ms): ssh/51346 connect(fd: 6, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.399 ( 0.007 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.426 ( 0.003 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.754 ( 0.009 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: INET, port: 22, addr: 127.0.0.1 }, addrlen: 16) = 0
       0.771 ( 0.010 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: INET6, port: 22, addr: ::1 }, addrlen: 28) = 0
       0.798 ( 0.053 ms): ssh/51346 connect(fd: 4, uservaddr: { .family: INET6, port: 22, addr: ::1 }, addrlen: 28) = 0
       0.870 ( 0.004 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.904 ( 0.003 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.930 ( 0.003 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.957 ( 0.003 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       0.981 ( 0.003 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       1.006 ( 0.004 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
       1.036 ( 0.005 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/lib/sss/pipes/nss }, addrlen: 110) = -1 ECONNREFUSED (Connection refused)
      65.077 ( 0.022 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/run/.heim_org.h5l.kcm-socket }, addrlen: 110) = 0
      66.608 ( 0.014 ms): ssh/51346 connect(fd: 5, uservaddr: { .family: LOCAL, path: /var/run/.heim_org.h5l.kcm-socket }, addrlen: 110) = 0
  root@localhost's password:
  #

  # perf trace -e sendto*  ping -c 2 localhost
  PING localhost(localhost (::1)) 56 data bytes
  64 bytes from localhost (::1): icmp_seq=1 ttl=64 time=0.024 ms
       0.000 ( 0.011 ms): ping/51357 sendto(fd: 5, buff: 0x7ffcca35e620, len: 20, addr: { .family: NETLINK }, addr_len: 0xc) = 20
       0.135 ( 0.026 ms): ping/51357 sendto(fd: 4, buff: 0x5601398f7b20, len: 64, addr: { .family: INET6, port: 58, addr: ::1 }, addr_len: 0x1c) = 64
    1014.929 ( 0.050 ms): ping/51357 sendto(fd: 4, buff: 0x5601398f7b20, len: 64, flags: CONFIRM, addr: { .family: INET6, port: 58, addr: ::1 }, addr_len: 0x1c) = 64
  64 bytes from localhost (::1): icmp_seq=2 ttl=64 time=0.046 ms

  --- localhost ping statistics ---
  2 packets transmitted, 2 received, 0% packet loss, time 1015ms
  rtt min/avg/max/mdev = 0.024/0.035/0.046/0.011 ms
  #

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: bpf@vger.kernel.org
Cc: llvm@lists.linux.dev
Link: https://lore.kernel.org/r/20230810184853.2860737-3-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf parse-events: Remove BPF event support

New features like the BPF --filter support in perf record have made the
BPF event functionality somewhat redundant. As shown by commit
fcb027c1a4f6 ("perf tools: Revert enable indices setting syntax for BPF
map") and commit 14e4b9f4289a ("perf trace: Raw augmented syscalls fix
libbpf 1.0+ compatibility") the BPF event support hasn't been well
maintained and it adds considerable complexity in areas like event
parsing, not least as '/' is a separator for event modifiers as well as
in paths.

This patch removes support in the event parser for BPF events and then
the associated functions are removed. This leads to the removal of whole
source files like bpf-loader.c.  Removing support means that augmented
syscalls in perf trace is broken, this will be fixed in a later commit
adding support using BPF skeletons.

The removal of BPF events causes an unused label warning from flex
generated code, so update build to ignore it:

  ```
  util/parse-events-flex.c:2704:1: error: label ‘find_rule’ defined but not used [-Werror=unused-label]
  2704 | find_rule: /* we branch to this label when backing up */
  ```

Committer notes:

Extracted from a larger patch that was also removing the support for
linking with libllvm and libclang, that were an alternative to using an
external clang execution to compile the .c event source code into BPF
bytecode.

Testing it:

  # perf trace -e /home/acme/git/perf/tools/perf/examples/bpf/augmented_raw_syscalls.c
  event syntax error: '/home/acme/git/perf/tools/perf/examples/bpf/augmented_raw_syscalls.c'
                        \___ Bad event or PMU

  Unabled to find PMU or event on a PMU of 'home'

  Initial error:
  event syntax error: '/home/acme/git/perf/tools/perf/examples/bpf/augmented_raw_syscalls.c'
                        \___ Cannot find PMU `home'. Missing kernel support?
  Run 'perf list' for a list of valid events

   Usage: perf trace [<options>] [<command>]
      or: perf trace [<options>] -- <command> [<options>]
      or: perf trace record [<options>] [<command>]
      or: perf trace record [<options>] -- <command> [<options>]

      -e, --event <event>   event/syscall selector. use 'perf list' to list available events
  #

Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Cc: bpf@vger.kernel.org
Cc: llvm@lists.linux.dev
Link: https://lore.kernel.org/r/20230810184853.2860737-2-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf bpf: Remove support for embedding clang for compiling BPF events (-e foo.c)

This never was in the default build for perf, is difficult to maintain
as it uses clang/llvm internals so ditch it, keeping, for now, the
external compilation of .c BPF into .o bytecode and its subsequent
loading, that is also going to be removed, do it separately to help
bisection and to properly document what is being removed and why.

Committer notes:

Extracted from a larger patch and removed some leftovers, namely
deleting these now unused feature tests:

    tools/build/feature/test-clang.cpp
    tools/build/feature/test-cxx.cpp
    tools/build/feature/test-llvm-version.cpp
    tools/build/feature/test-llvm.cpp

Testing the use of BPF events after applying this patch:

To use the external clang/llvm toolchain to compile a .c event and then
use libbpf to load it, to get the syscalls:sys_enter_open* tracepoints
and read the filename pointer, putting it into the ring buffer right
after the usual tracepoint payload for 'perf trace' to then print it:

  [root@quaco ~]# perf trace -e /home/acme/git/perf-tools-next/tools/perf/examples/bpf/augmented_raw_syscalls.c,open* --max-events=10
     0.000 systemd-oomd/959 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC) = 12
     0.083 abrt-dump-jour/1453 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
     0.063 abrt-dump-jour/1454 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
     0.082 abrt-dump-jour/1455 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
   250.124 systemd-oomd/959 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC) = 12
   250.521 systemd-oomd/959 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/app.slice/memory.pressure", flags: RDONLY|CLOEXEC) = 12
   251.047 systemd-oomd/959 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/app.slice/memory.current", flags: RDONLY|CLOEXEC) = 12
   251.162 systemd-oomd/959 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/app.slice/memory.min", flags: RDONLY|CLOEXEC) = 12
   251.242 systemd-oomd/959 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/app.slice/memory.low", flags: RDONLY|CLOEXEC) = 12
   251.353 systemd-oomd/959 openat(dfd: CWD, filename: "/sys/fs/cgroup/user.slice/user-1000.slice/user@1000.service/app.slice/memory.swap.current", flags: RDONLY|CLOEXEC) = 12
  [root@quaco ~]#

Same thing, but with a prebuilt .o BPF bytecode:

  [root@quaco ~]# perf trace -e /home/acme/git/perf-tools-next/tools/perf/examples/bpf/augmented_raw_syscalls.o,open* --max-events=10
     0.000 systemd-oomd/959 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC) = 12
     0.083 abrt-dump-jour/1453 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
     0.083 abrt-dump-jour/1455 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
     0.062 abrt-dump-jour/1454 openat(dfd: CWD, filename: "/var/log/journal/d6a97235307247e09f13f326fb607e3c/system.journal", flags: RDONLY|CLOEXEC|NONBLOCK) = 4
   249.985 systemd-oomd/959 openat(dfd: CWD, filename: "/proc/meminfo", flags: RDONLY|CLOEXEC) = 12
   466.763 thermald/1234 openat(dfd: CWD, filename: "/sys/class/powercap/intel-rapl/intel-rapl:0/intel-rapl:0:2/energy_uj") = 13
   467.145 thermald/1234 openat(dfd: CWD, filename: "/sys/class/powercap/intel-rapl/intel-rapl:0/energy_uj") = 13
   467.311 thermald/1234 openat(dfd: CWD, filename: "/sys/class/thermal/thermal_zone2/temp") = 13
   500.040 cgroupify/24006 openat(dfd: 4, filename: ".", flags: RDONLY|CLOEXEC|DIRECTORY|NONBLOCK) = 5
   500.295 cgroupify/24006 openat(dfd: 4, filename: "24616/cgroup.procs") = 5
  [root@quaco ~]#

Signed-off-by: Ian Rogers <irogers@google.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Carsten Haitzler <carsten.haitzler@arm.com>
Cc: Eduard Zingerman <eddyz87@gmail.com>
Cc: Fangrui Song <maskray@google.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: "Naveen N. Rao" <naveen.n.rao@linux.vnet.ibm.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@amd.com>
Cc: Rob Herring <robh@kernel.org>
Cc: Tiezhu Yang <yangtiezhu@loongson.cn>
Cc: Tom Rix <trix@redhat.com>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Wang ShaoBo <bobo.shaobowang@huawei.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Yonghong Song <yhs@fb.com>
Cc: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/lkml/ZNZWsAXg2px1sm2h@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf tests trace+probe_vfs_getname.sh: Accept quotes surrounding the filename

With augmented_raw_syscalls transformed into a BPF skel made the output have a "
around the filenames, which is not what the old perf probe vfs_getname
method of obtaining filenames did, so accept the augmented way, with the
quotes.

At this point probably removing all the logic for the vfs_getname method
is in order, will do it at some point.

For now lets accept with/without quotes and make that test pass.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf test trace+probe_vfs_getname.sh: Remove stray \ before /

Running on fedora:38 in verbose mode I noticed:

  # perf test -v 117
  grep: warning: stray \ before /
  117: Check open filename arg using perf trace + vfs_getname          :

Remove that \ before /.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/lkml/ZNvTDsSMO3nw9Tnp@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

perf script python: Add stub for PMU symbol to the python binding

Fix missing symbol seen in:

  ```
   19: 'import perf' in python                                         :
  --- start ---
  test child forked, pid 2640936
  python usage test: "echo "import sys ; sys.path.insert(0, 'python'); import perf" | '/usr/bin/python3' "
  Traceback (most recent call last):
    File "<stdin>", line 1, in <module>
  ImportError: tools/perf/python/perf.cpython-311-x86_64-linux-gnu.so: undefined symbol: perf_pmus__supports_extended_type
  test child finished with -1
  ---- end ----
  'import perf' in python: FAILED!
  ```

Signed-off-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: James Clark <james.clark@arm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Leo Yan <leo.yan@linaro.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Yang Jihong <yangjihong1@huawei.com>
Link: https://lore.kernel.org/r/20230810180944.2794188-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>