linux-stable.git/tools/perf, branch v6.12.91

perf tool_pmu: Fix aggregation on duration_time

2026-05-23T11:04:58+00:00

[ Upstream commit 68cb1567439fa325ba980f3b5b67f95d3953eafd ]

evsel__count_has_error() fails counters when the enabled or running time
are 0. The duration_time event reads 0 when the cpu_map_idx != 0 to
avoid aggregating time over CPUs. Change the enable and running time
to always have a ratio of 100% so that evsel__count_has_error won't
fail.

Before:
```
$ sudo /tmp/perf/perf stat --per-core -a -M UNCORE_FREQ sleep 1

 Performance counter stats for 'system wide':

S0-D0-C0              1      2,615,819,485      UNC_CLOCK.SOCKET                 #     2.61 UNCORE_FREQ
S0-D0-C0              2            duration_time

       1.002111784 seconds time elapsed
```

After:
```
$ perf stat --per-core -a -M UNCORE_FREQ sleep 1

 Performance counter stats for 'system wide':

S0-D0-C0              1        758,160,296      UNC_CLOCK.SOCKET                 #     0.76 UNCORE_FREQ
S0-D0-C0              2      1,003,438,246      duration_time

       1.002486017 seconds time elapsed
```

Note: the metric reads the value a different way and isn't impacted.

Fixes: 240505b2d0adcdc8 ("perf tool_pmu: Factor tool events into their own PMU")
Reported-by: Stephane Eranian 
Reviewed-by: James Clark 
Signed-off-by: Ian Rogers 
Cc: Adrian Hunter 
Cc: Alexander Shishkin 
Cc: Ian Rogers 
Cc: Ingo Molnar 
Cc: Jiri Olsa 
Cc: Kan Liang 
Cc: Mark Rutland 
Cc: Namhyung Kim 
Cc: Peter Zijlstra 
Cc: Thomas Richter 
Link: https://lore.kernel.org/r/20250423050358.94310-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo 
Signed-off-by: Sasha Levin

perf util: Kill die() prototype, dead for a long time

2026-05-23T11:04:43+00:00

[ Upstream commit e5cce1b9c82fbd48e2f1f7a25a9fad8ee228176f ]

In fef2a735167a827a ("perf tools: Kill die()") the die() function was
removed, but not the prototype in util.h, now when building with
LIBPERL=1, during a 'make -C tools/perf build-test' routine test, it is
failing as perl likes die() calls and then this clashes with this
remnant, remove it.

Fixes: fef2a735167a827a ("perf tools: Kill die()")
Reviewed-by: Ian Rogers 
Signed-off-by: Arnaldo Carvalho de Melo 
Signed-off-by: Namhyung Kim 
Signed-off-by: Sasha Levin

perf maps: Fix copy_from that can break sorted by name order

2026-05-23T11:04:43+00:00

[ Upstream commit f552b132e4d5248715828e7e5c2bf7889bf05b2e ]

When an parent is copied into a child the name array is populated in
address not name order. Make sure the name array isn't flagged as sorted.

Fixes: 659ad3492b91 ("perf maps: Switch from rbtree to lazily sorted array for addresses")
Signed-off-by: Ian Rogers 
Signed-off-by: Namhyung Kim 
Signed-off-by: Sasha Levin

perf cgroup: Update metric leader in evlist__expand_cgroup

2026-05-23T11:04:43+00:00

[ Upstream commit c9ef786c0970991578397043f1c819229e2b7197 ]

When the evlist is expanded the metric leader wasn't being updated. As
the original evsel is deleted this creates a use-after-free in
stat-shadow's prepare_metric. This was detected running the "perf stat
--bpf-counters --for-each-cgroup test" with sanitizers.

The change itself puts the copied evsel into the priv field (known
unused because of evsel__clone use) and then in a second pass over the
list updates the copied values using the priv pointer.

Fixes: d1c5a0e86a4e ("perf stat: Add --for-each-cgroup option")
Signed-off-by: Ian Rogers 
Acked-by: Sun Jian 
Signed-off-by: Namhyung Kim 
Signed-off-by: Sasha Levin

perf python: Add parse_events function

2026-05-23T11:04:43+00:00

[ Upstream commit f081defccd934a8db309c90a61178e4f2eef386c ]

Add basic parse_events function that takes a string and returns an
evlist. As the python evlist is embedded in a pyrf_evlist, and the
evsels are embedded in pyrf_evsels, copy the parsed data into those
structs and update evsel__clone to enable this.

Signed-off-by: Ian Rogers 
Cc: Adrian Hunter 
Cc: Alexander Shishkin 
Cc: Andi Kleen 
Cc: Athira Rajeev 
Cc: Colin Ian King 
Cc: Dapeng Mi 
Cc: Howard Chu 
Cc: Ilya Leoshkevich 
Cc: Ingo Molnar 
Cc: James Clark 
Cc: Jiri Olsa 
Cc: Josh Poimboeuf 
Cc: Kan Liang 
Cc: Mark Rutland 
Cc: Michael Petlan 
Cc: Namhyung Kim 
Cc: Peter Zijlstra 
Cc: Thomas Richter 
Cc: Veronika Molnarova 
Cc: Weilin Wang 
Link: https://lore.kernel.org/r/20241119011644.971342-20-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo 
Stable-dep-of: c9ef786c0970 ("perf cgroup: Update metric leader in evlist__expand_cgroup")
Signed-off-by: Sasha Levin

perf tool_pmu: Factor tool events into their own PMU

2026-05-23T11:04:43+00:00

[ Upstream commit 240505b2d0adcdc8fd018117e88dc27b09734735 ]

Rather than treat tool events as a special kind of event, create a
tool only PMU where the events/aliases match the existing
duration_time, user_time and system_time events. Remove special
parsing and printing support for the tool events, but add function
calls for when PMU functions are called on a tool_pmu.

Move the tool PMU code in evsel into tool_pmu.c to better encapsulate
the tool event behavior in that file.

Signed-off-by: Ian Rogers 
Acked-by: Namhyung Kim 
Link: https://lore.kernel.org/r/20241002032016.333748-5-irogers@google.com
Signed-off-by: Namhyung Kim 
Stable-dep-of: c9ef786c0970 ("perf cgroup: Update metric leader in evlist__expand_cgroup")
Signed-off-by: Sasha Levin

perf evsel: Add alternate_hw_config and use in evsel__match

2026-05-23T11:04:43+00:00

[ Upstream commit 22a4db3c36034e2b034c5b88414680857fc59cf4 ]

There are cases where we want to match events like instructions and
cycles with legacy hardware values, in particular in stat-shadow's
hard coded metrics. An evsel's name isn't a good point of reference as
it gets altered, strstr would be too imprecise and re-parsing the
event from its name is silly. Instead, hold the legacy hardware event
name, determined during parsing, in the evsel for this matching case.

Inline evsel__match2 that is only used in builtin-diff.

Acked-by: Namhyung Kim 
Signed-off-by: Ian Rogers 
Acked-by: Kan Liang 
Signed-off-by: James Clark 
Cc: Yang Jihong 
Cc: Dominique Martinet 
Cc: Colin Ian King 
Cc: Howard Chu 
Cc: Yunseong Kim 
Cc: Ze Gao 
Cc: Yicong Yang 
Cc: Weilin Wang 
Cc: Will Deacon 
Cc: Mike Leach 
Cc: Jing Zhang 
Cc: Yang Li 
Cc: Leo Yan 
Cc: ak@linux.intel.com
Cc: Athira Rajeev 
Cc: linux-arm-kernel@lists.infradead.org
Cc: Sun Haiyong 
Cc: John Garry 
Link: https://lore.kernel.org/r/20240926144851.245903-2-james.clark@linaro.org
Signed-off-by: Namhyung Kim 
Stable-dep-of: c9ef786c0970 ("perf cgroup: Update metric leader in evlist__expand_cgroup")
Signed-off-by: Sasha Levin

perf expr: Return -EINVAL for syntax error in expr__find_ids()

2026-05-23T11:04:43+00:00

[ Upstream commit 3a61fd866ef9aaa1d3158b460f852b74a2df07f4 ]

expr__find_ids() propagates the parser return value directly.  For syntax
errors, the parser can return a positive value, but callers treat it as
success, e.g., for below case on Arm64 platform:

  metric expr 100 * (STALL_SLOT_BACKEND / (CPU_CYCLES * #slots) - BR_MIS_PRED * 3 / CPU_CYCLES) for backend_bound
  parsing metric: 100 * (STALL_SLOT_BACKEND / (CPU_CYCLES * #slots) - BR_MIS_PRED * 3 / CPU_CYCLES)
  Failure to read '#slots' literal: #slots = nan
  syntax error

Convert positive parser returns in expr__find_ids() to -EINVAL, as a
result, the error value will be respected by callers.

Before:

  perf stat -C 5
  Failure to read '#slots'Failure to read '#slots'Failure to read '#slots'Failure to read '#slots'Segmentation fault

After:

  perf stat -C 5
  Failure to read '#slots'Cannot find metric or group `Default'

Fixes: ded80bda8bc9 ("perf expr: Migrate expr ids table to a hashmap")
Signed-off-by: Leo Yan 
Reviewed-by: Ian Rogers 
Signed-off-by: Namhyung Kim 
Signed-off-by: Sasha Levin

perf tools: Fix module symbol resolution for non-zero .text sh_addr

2026-05-23T11:04:43+00:00

[ Upstream commit 9a82bfde4775b7a87cd1a7e791f46f83ae442848 ]

When perf resolves symbols from kernel module ELF files (ET_REL),
it converts symbol addresses to file offsets so that sample IPs
can be matched to the correct symbol. The conversion adjusts each
symbol's st_value:

sym->st_value -= shdr->sh_addr - shdr->sh_offset;

For vmlinux (ET_EXEC), st_value is a virtual address and sh_addr
is the section's virtual base, so subtracting sh_addr and adding
sh_offset correctly yields a file offset.

For kernel modules (ET_REL), st_value is a section-relative
offset. The module loader ignores sh_addr entirely and places
symbols at module_base + st_value. Converting to file offset
requires only adding sh_offset; subtracting sh_addr introduces an
error equal to sh_addr bytes.

When .text has sh_addr == 0 -- the historical norm for simple
modules -- both formulas produce the same result and the bug is
latent. As modules gain more metadata sections before .text (.note,
.static_call.text, etc.), the linker assigns .text a non-zero
sh_addr, exposing the defect. For example, nfsd.ko on this kernel
has sh_addr=0xa80, kvm-intel.ko has sh_addr=0x1e90.

The effect is that all .text symbols in affected modules
shift by sh_addr bytes relative to sample IPs, causing perf
report to attribute samples to incorrect, nearby symbols. This
was observed as 13% of LLC-load-miss samples misattributed
to nfsd_file_get_dio_attrs when the actual hot function was
nfsd_cache_lookup, approximately 0xa80 bytes away in the symbol
table.

Use the existing dso__rel() flag (already set for ET_REL modules)
to select the correct adjustment: add sh_offset for ET_REL,
subtract (sh_addr - sh_offset) for ET_EXEC/ET_DYN.

Fixes: 0131c4ec794a ("perf tools: Make it possible to read object code from kernel modules")
Signed-off-by: Chuck Lever
Reviewed-by: Ian Rogers
Tested-by: Thomas Richter
Signed-off-by: Namhyung Kim
Signed-off-by: Sasha Levin

perf stat: Fix opt->value type for parse_cache_level

2026-05-23T11:04:43+00:00

[ Upstream commit 44311ae84ad9177fb311aee856027861c22f17b2 ]

Commit f5803651b4a4 ("perf stat: Choose the most disaggregate command
line option") changed aggregation option handling for `perf stat` but
not `perf stat report` leading to parse_cache_level being passed a
struct in the `perf stat` case but erroneously an aggr_mode enum value
for `perf stat report`. Change the `perf stat report` aggregation
handling to use the same opt_aggr_mode as `perf stat`. Also, just pass
the boolean for consistency with other boolean argument handling.

Fixes: f5803651b4a4 ("perf stat: Choose the most disaggregate command line option")
Signed-off-by: Ian Rogers 
Signed-off-by: Namhyung Kim 
Signed-off-by: Sasha Levin