perf tools: Revert enable indices setting syntax for BPF map

This reverts commit e571e029bdbf ("perf tools: Enable indices setting
syntax for BPF map").

The reverted commit added a notion of arrays that could be set as
event terms for BPF events. The parsing hasn't worked over multiple
Linux releases. Given the broken nature of the parsing it appears the
code isn't in use, nor could I find a way for it to be used to add a
test.

The original commit contains a test in the commit message,
however, running it yields:
```
$ perf record -e './test_bpf_map_3.c/map:channel.value[0,1,2,3...5]=101/' usleep 2
event syntax error: '..pf_map_3.c/map:channel.value[0,1,2,3...5]=101/'
                                  \___ parser error
Run 'perf list' for a list of valid events

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

    -e, --event <event>   event selector. use 'perf list' to list available events
```

Given the code can't be used this commit reverts and removes it.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: He Kuang <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Wang Nan <[email protected]>
Cc: Wang ShaoBo <[email protected]>
Cc: YueHaibing <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-event: Avoid BPF test SEGV

loc is passed as NULL in tools/perf/tests/bpf.c do_test, meaning
errors trigger a SEGV when trying to access. Add the missing NULL
check.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: He Kuang <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Wang Nan <[email protected]>
Cc: Wang ShaoBo <[email protected]>
Cc: YueHaibing <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Include generated header files properly

The flex and bison generate header files from the source.  When user
specified a build directory with O= option, it'd generate files under
the directory.  The build command has -I option to specify the header
include directory.

But the -I option only affects the files included like <...>.  Let's
change the flex and bison headers to use it instead of "...".

Fixes: 80eeb67fe577aa76 ("perf jevents: Program to convert JSON file")
Signed-off-by: Namhyung Kim <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Anup Sharma <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Update build rule for generated files

The bison and flex generate C files from the source (.y and .l)
files.  When O= option is used, they are saved in a separate directory
but the default build rule assumes the .C files are in the source
directory.  So it might read invalid file if there are generated files
from an old version.  The same is true for the pmu-events files.

For example, the following command would cause a build failure:

  $ git checkout v6.3
  $ make -C tools/perf  # build in the same directory

  $ git checkout v6.5-rc2
  $ mkdir build  # create a build directory
  $ make -C tools/perf O=build  # build in a different directory but it
                                # refers files in the source directory

Let's update the build rule to specify those cases explicitly to depend
on the files in the output directory.

Note that it's not a complete fix and it needs the next patch for the
include path too.

Fixes: 80eeb67fe577aa76 ("perf jevents: Program to convert JSON file")
Signed-off-by: Namhyung Kim <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Anup Sharma <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Remove -Wno-redundant-decls in 2 cases

Properly fix a warning and remove the -Wno-redundant-decls C flag.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Disable fewer bison warnings

If bison is version 3.8.2, reduce the number of bison C warnings
disabled. Earlier bison versions have all C warnings disabled. Avoid
implicit declarations of yylex by adding the declaration in the C
file. A header can't be included as a circular dependency would occur
due to the lexer using the bison defined tokens.

Committer notes:

Some recent versions of gcc and clang (noticed on Alpine Linux 3.17,
edge, clearlinux, fedora 37, etc.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Disable fewer flex warnings

If flex is version 2.6.4, reduce the number of flex C warnings
disabled. Earlier flex versions have all C warnings disabled.

Committer notes:

Added this to the list of ignored warnings to get it building on
a Fedora 36 machine with flex 2.6.4:

  -Wno-misleading-indentation

Noticed when building with:

  $ make LLVM=1 -C tools/perf NO_BPF_SKEL=1 DEBUG=1

Take two:

We can't just try to canonicalize flex versions by just removing the
dots, as we end up with:

2.6.4 >= 2.5.37

becoming:

264 >= 2537

Failing the build on flex 2.5.37, so instead use the back to the past
added $(call version_ge3,$(FLEX_VERSION),2.6.4) variant to check for
that.

Making sure $(FLEX_VERSION) keeps the dots as we may want to use 'sort
-V' or something nicer when available everywhere.

Some other tweaks for other flex versions and combinations with gcc and
clang versions were added, notes on the patch.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

tools build: Add 3-component logical version comparators

The next cset needs to compare if a flex version is greater or
equal/less than another, but since there is no canonical, generally
available way to compare versions in the command line (sort -V, yeah,
but...), just use awk to canonicalize the versions like is also done in
scripts/rust_is_available.sh.

There was a problem spotted in linux-next where a bashism, here
documents, aka the '<<<' stdin redirector, for strings to be used as the
stdin for awk. Use $(shell echo | awk ...) instead.

Cc: Adrian Hunter <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Add Wextra for C++ compilation

Commit d58ac0bf8d1e ("perf build: Add clang and llvm compile and
linking support") added -Wall and -Wno-strict-aliasing for CXXFLAGS,
but not -Wextra. -Wno-strict-aliasing is no longer necessary, adding
-Wextra for CXXFLAGS requires adding -Wno-unused-parameter clang.cpp
and clang-test.cpp for LIBCLANGLLVM=1 to build.

Signed-off-by: Ian Rogers <[email protected]>
Acked-by: James Clark <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Don't always set -funwind-tables and -ggdb3

Commit 6a40cd90f5deb6de ("perf tools: Add libunwind dependency for DWARF
CFI unwinding") added libunwind support but also -funwind-tables and
-ggdb3 to the standard build. These build flags aren't necessary so
remove, set -g when DEBUG is enabled for the build.

Signed-off-by: Ian Rogers <[email protected]>
Acked-by: James Clark <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bpf-loader: Remove unneeded diagnostic pragma

Added during the progress to libbpf 1.0 the deprecated functions are
no longer used and so the pragma can be removed.

Signed-off-by: Ian Rogers <[email protected]>
Acked-by: James Clark <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andrii Nakryiko <[email protected]>
Cc: Eduard Zingerman <[email protected]>
Cc: Gaosheng Cui <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Rob Herring <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf docs: Update metric usage for Alibaba's T-Head PMU driver (arm64)

Alibaba's T-Head ali_drw PMU supports DDR bandwidth metrics. Update
its usage in the documentation.

Reviewed-by: John Garry <[email protected]>
Signed-off-by: Jing Zhang <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Shuai Xue <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Zhuo Song <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf vendor events arm64: Add JSON metrics for Yitian 710 DDR

Add JSON metrics for T-HEAD Yitian 710 SoC DDR.

Reviewed-by: John Garry <[email protected]>
Signed-off-by: Jing Zhang <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Shuai Xue <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Zhuo Song <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf jevents: Add support for Yitian 710 DDR PMU (arm64) aliasing

Add alias support for T-HEAD Yitian 710 SoC DDR PMU events.

Reviewed-by: John Garry <[email protected]>
Reviewed-by: Shuai Xue <[email protected]>
Signed-off-by: Jing Zhang <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Will Deacon <[email protected]>
Cc: Zhuo Song <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf tools: Add a place to put kernel config fragments for test runs

Defconfig doesn't give full coverage for a perf test run, so these can
be merged with defconfig to do so. It's not complete yet, but is a
starting point as a place to add to when a specific test needs something
extra to run.

Signed-off-by: James Clark <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: [email protected]
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Mark Brown <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Masahiro Yamada <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Add command execution for gecko script

This will enable the execution of gecko.py script using record and
report commands in 'perf script'.  And this will be also reflected at
"perf script -l" command.

For Example:
    perf script record gecko
    perf script report gecko

Committer notes:

As discussed on the perf tools office hours, I made -F 99 the default
for the record script and removed the double -- on the report script so
that the existing 'perf script' protocol for the combined operation:

    # perf script gecko

Works, i.e. the record script pipes its stdout into the stdin of the
report script, basically:

  /bin/sh /usr/libexec/perf-core/scripts/python/bin/gecko-record -F 99 -g -a -q -o - | \
  /bin/sh /usr/libexec/perf-core/scripts/python/bin/gecko-report -i -

Testing it:

The resulting JSON file needs to be uploaded to
https://profiler.firefox.com, Anup already has code to start a local
http server on the trace_begin handler of the gecko python script, start
firefox and feed it the JSON.

The example below only collects sample for the specified workload, so
that we don't produce thousands of lines, to collect system wide
samples, use instead:

  # perf script gecko -a sleep 0.5

  # nohup perf script gecko sleep 0.5
  {
    "meta": {
      "interval": 1,
      "processType": 0,
      "product": "x86_64 GNU/Linux",
      "stackwalk": 1,
      "debug": 0,
      "gcpoison": 0,
      "asyncstack": 1,
      "startTime": 274601692.636,
      "shutdownTime": null,
      "version": 24,
      "presymbolicated": true,
      "categories": [
        {
          "name": "User",
          "color": "yellow",
          "subcategories": [
            "Other"
          ]
        },
        {
          "name": "Kernel",
          "color": "orange",
          "subcategories": [
            "Other"
          ]
        }
      ],
      "markerSchema": []
    },
    "libs": [],
    "threads": [
      {
        "tid": 3344498,
        "pid": 3344498,
        "name": "sleep",
        "markers": {
          "schema": {
            "name": 0,
            "startTime": 1,
            "endTime": 2,
            "phase": 3,
            "category": 4,
            "data": 5
          },
          "data": []
        },
        "samples": {
          "schema": {
            "stack": 0,
            "time": 1,
            "responsiveness": 2
          },
          "data": [
            [
              21,
              274601692.636,
              0
            ],
            [
              23,
              274601692.641,
              0
            ],
            [
              29,
              274601692.643,
              0
            ],
            [
              42,
              274601692.648,
              0
            ]
          ]
        },
        "frameTable": {
          "schema": {
            "location": 0,
            "relevantForJS": 1,
            "innerWindowID": 2,
            "implementation": 3,
            "optimizations": 4,
            "line": 5,
            "column": 6,
            "category": 7,
            "subcategory": 8
          },
          "data": [
            [
              0,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              1,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              2,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              3,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              4,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              5,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              6,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              7,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              8,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              9,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              10,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              11,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              12,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              13,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              14,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              15,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              16,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              17,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              18,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              19,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              20,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              21,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              22,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              23,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              24,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              25,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              26,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              27,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              28,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              29,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              30,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              31,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              32,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              33,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              34,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              35,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              36,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              37,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ],
            [
              38,
              false,
              0,
              null,
              null,
              null,
              null,
              1,
              null
            ]
          ]
        },
        "stackTable": {
          "schema": {
            "prefix": 0,
            "frame": 1
          },
          "data": [
            [
              null,
              0
            ],
            [
              0,
              1
            ],
            [
              1,
              2
            ],
            [
              2,
              3
            ],
            [
              3,
              4
            ],
            [
              4,
              5
            ],
            [
              5,
              6
            ],
            [
              6,
              7
            ],
            [
              7,
              8
            ],
            [
              8,
              9
            ],
            [
              9,
              10
            ],
            [
              10,
              11
            ],
            [
              11,
              12
            ],
            [
              12,
              13
            ],
            [
              13,
              14
            ],
            [
              14,
              15
            ],
            [
              15,
              16
            ],
            [
              16,
              17
            ],
            [
              17,
              18
            ],
            [
              18,
              19
            ],
            [
              19,
              20
            ],
            [
              20,
              21
            ],
            [
              20,
              22
            ],
            [
              22,
              23
            ],
            [
              11,
              24
            ],
            [
              24,
              25
            ],
            [
              25,
              26
            ],
            [
              26,
              27
            ],
            [
              27,
              28
            ],
            [
              28,
              29
            ],
            [
              9,
              11
            ],
            [
              30,
              24
            ],
            [
              31,
              25
            ],
            [
              32,
              30
            ],
            [
              33,
              31
            ],
            [
              34,
              32
            ],
            [
              35,
              29
            ],
            [
              36,
              33
            ],
            [
              37,
              34
            ],
            [
              38,
              35
            ],
            [
              39,
              36
            ],
            [
              40,
              37
            ],
            [
              41,
              38
            ]
          ]
        },
        "stringTable": [
          "__func__.0 (in [kernel.kallsyms].rodata)",
          "perf_trace_ext4_fc_track_inode (in [kernel.kallsyms])",
          "perf_trace_ext4_es_insert_delayed_block (in [kernel.kallsyms])",
          "ext4_es_show_pblock (in [kernel.kallsyms])",
          "perf_trace_ext4_ext_rm_leaf (in [kernel.kallsyms])",
          "devcgroup_access_write (in [kernel.kallsyms])",
          "devcgroup_update_access (in [kernel.kallsyms])",
          "propagate_exception (in [kernel.kallsyms])",
          "revalidate_active_exceptions (in [kernel.kallsyms])",
          "perf_trace_ext4_fc_commit_stop (in [kernel.kallsyms])",
          "perf_fetch_caller_regs (in [kernel.kallsyms])",
          "khugepaged (in [kernel.kallsyms])",
          "khugepaged_wait_work (in [kernel.kallsyms])",
          "freezable_schedule_timeout (in [kernel.kallsyms])",
          "freezer_count (in [kernel.kallsyms])",
          "try_to_freeze (in [kernel.kallsyms])",
          "try_to_freeze_unsafe (in [kernel.kallsyms])",
          "split_huge_pages_write (in [kernel.kallsyms])",
          "migrate_pages (in [kernel.kallsyms])",
          "unmap_and_move (in [kernel.kallsyms])",
          "__unmap_and_move (in [kernel.kallsyms])",
          "collect_events (in [kernel.kallsyms])",
          "uncore_down_prepare (in [kernel.kallsyms])",
          "perf_iommu_read (in [kernel.kallsyms])",
          "khugepaged_do_scan (in [kernel.kallsyms])",
          "khugepaged_scan_mm_slot (in [kernel.kallsyms])",
          "khugepaged_scan_file (in [kernel.kallsyms])",
          "need_resched (in [kernel.kallsyms])",
          "get_current (in [kernel.kallsyms])",
          "move_to_new_page (in [kernel.kallsyms])",
          "khugepaged_scan_pmd (in [kernel.kallsyms])",
          "trace_mm_khugepaged_scan_pmd (in [kernel.kallsyms])",
          "migrate_huge_page_move_mapping (in [kernel.kallsyms])",
          "do_huge_pmd_numa_page (in [kernel.kallsyms])",
          "pmd_pfn (in [kernel.kallsyms])",
          "protnone_mask (in [kernel.kallsyms])",
          "__pte_needs_invert (in [kernel.kallsyms])",
          "reclaim_high (in [kernel.kallsyms])",
          "memcg_memory_event (in [kernel.kallsyms])"
        ],
        "registerTime": 0,
        "unregisterTime": null,
        "processType": "default"
      }
    ],
    "processes": [],
    "pausedRanges": []
  }
  #

Signed-off-by: Anup Sharma <[email protected]>
Tested-by: Arnaldo Carvalho de Melo <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/cbf03cda175ea3dd2c6cd87bd3f12d803446cb95.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Implement add sample function and thread processing

The stack has been created for storing func and dso from the callchain.
The sample has been added to a specific thread. It first checks if the
thread exists in the Thread class. Then it call _add_sample function
which is responsible for appending a new entry to the samples list.

Also callchain parsing and storing part is implemented. Moreover removed
the comment from thread.

Signed-off-by: Anup Sharma <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/5a112be85ccdcdcd611e343f6a7a7482d01f6299.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Implement add sample function and thread processing

The intern_stack function is responsible for retrieving
or creating a stack_id based on the provided frame_id and prefix_id.
It first generates a key using the frame_id and prefix_id values.
If the stack corresponding to the key is found in the stackMap,
it is returned. Otherwise, a new stack is created by appending
the prefix_id and frame_id to the stackTable. The key
and the index of the newly created stack are added to the
stackMap for future reference.

The _intern_frame function is responsible for retrieving or
creating a frame_id based on the provided frame string. If the frame_id
corresponding to the frameString is found in the frameMap, it is
returned. Otherwise, a new frame is created by appending relevant
information to the frameTable and adding the frameString to the string_id
through _intern_string.

The _intern_string function will gets a matching string, or saves the new
string and returns a String ID.

Signed-off-by: Anup Sharma <[email protected]>
Link: https://lore.kernel.org/r/4442f4b1ab4c7317cf940560a3a285fcdfbeeb08.1689961706.git.anupnewsmail@gmail.com
Cc: Mark Rutland <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: [email protected]
Cc: [email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Add trace end processing and PRODUCT and CATEGORIES information

The final output will now be presented in JSON format following the Gecko
profile structure. Additionally, the inclusion of PRODUCT allows easy retrieval
of header information for UI.

Furthermore, CATEGORIES have been introduced to enable customization of
kernel and user colors using input arguments. To facilitate this functionality,
an argparse-based parser has been implemented.

Note: The implementation of threads will be addressed in subsequent commits
for now I have commented it out.

Signed-off-by: Anup Sharma <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/fa6d027e4134c48e8a2ea45dd8f6b21e6a3418e4.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Add classes and conversion functions

This commit introduces new classes and conversion functions to
facilitate the representation of Gecko profile information. The new
classes Frame, Stack, Sample, and Thread are added to handle specific
components of the profile data, also link to the origin docs has been
commented out.

Additionally, Inside the Thread class _to_json_dict() method has been
created that converts the current thread data into the corresponding
format expected by the GeckoThread JSON schema, as per the Gecko
profile format specification.

Signed-off-by: Anup Sharma <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/ab7b40bd32df7101a6f8b4a3aa41570b63b831ac.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Extact necessary information from process event

The script takes in a sample event dictionary(param_dict) and retrieves
relevant data such as time stamp, PID, TID, and comm for each event.
Also start time is defined as a global variable as it need to be passed
to trace_end for gecko meta information field creation.

Signed-off-by: Anup Sharma <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/19910fefcfe4be03cd5c2aa3fec11d3f86c0381b.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf scripts python: Add initial script file with usage information

Added necessary modules, including the Perf-Trace-Util
library, and defines the required functions and variables
for using perf script python. The perf_trace_context and
Core modules for tracing and processing events has been
also imported. Added usage information.

Signed-off-by: Anup Sharma <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/f2f1a62f1cc69f44a5414da46a26a4cf124d2744.1689961706.git.anupnewsmail@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf doc: Fix typo in perf.data-file-format.txt

The 'it' should be 'is' here, fix it.

Signed-off-by: Xiu Jianfeng <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf machine: Include data symbols in the kernel map

When 'perf record -d' is used, it needs data mmaps to symbolize global
data.  But it missed to collect kernel data maps so it cannot symbolize
them.  Instead of having a separate map, just increase the kernel map
size to include the data section.

Probably we can have a separate kernel map for data, but the current
code assumes a single kernel map.  So it'd require more changes in other
places and looks error-prone.  I decided not to go that way for now.

Also it seems the kernel module size already includes the data section.

For example, my system has the following.

  $ grep -e _stext -e _etext -e _edata /proc/kallsyms
  ffffffff99800000 T _stext
  ffffffff9a601ac8 T _etext
  ffffffff9b446a00 D _edata

Size of the text section is (0x9a601ac8 - 0x99800000 = 0xe01ac8) and
size including data section is (0x9b446a00 - 0x99800000 = 0x1c46a00).

Before:
  $ perf record -d true

  $ perf report -D | grep MMAP | head -1
  0 0 0x460 [0x60]: PERF_RECORD_MMAP -1/0: [0xffffffff99800000(0xe01ac8) @ 0xffffffff99800000]: x [kernel.kallsyms]_text
                                                               ^^^^^^^^
                                                                 here
After:
  $ perf report -D | grep MMAP | head -1
  0 0 0x460 [0x60]: PERF_RECORD_MMAP -1/0: [0xffffffff99800000(0x1c46a00) @ 0xffffffff99800000]: x [kernel.kallsyms]_text
                                                               ^^^^^^^^^

Instead of just replacing it to _edata, try _edata first and then fall
back to _etext just in case.

Signed-off-by: Namhyung Kim <[email protected]>
Acked-by: Adrian Hunter <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf symbols: Add kallsyms__get_symbol_start()

The kallsyms__get_symbol_start() to get any symbol address from
kallsyms.  The existing kallsyms__get_function_start() only allows text
symbols so create this to allow data symbols too.

Signed-off-by: Namhyung Kim <[email protected]>
Acked-by: Adrian Hunter <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Remove ABORT_ON

Prefer informative messages rather than none with ABORT_ON. Document
one failure mode and add an error message for another.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Improve location for add pmu

Improve the location for add PMU for cases when PMUs aren't found.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Populate error column for BPF/tracepoint events

Follow convention from parse_events_terms__num/str and pass the
YYLTYPE for the location.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Additional error reporting

When no events or PMUs match report an error for event_pmu:

Before:
```
$ perf stat -e 'asdfasdf' -a sleep 1
Run 'perf list' for a list of valid events

 Usage: perf stat [<options>] [<command>]

    -e, --event <event>   event selector. use 'perf list' to list available events
```

After:
```
$ perf stat -e 'asdfasdf' -a sleep 1
event syntax error: 'asdfasdf'
                     \___ Bad event name

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

 Usage: perf stat [<options>] [<command>]

    -e, --event <event>   event selector. use 'perf list' to list available events
```

Fixes the inadvertent removal when hybrid parsing was modified.

Fixes: 70c90e4a6b2fbe77 ("perf parse-events: Avoid scanning PMUs before parsing")
Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Separate ENOMEM memory handling

Add PE_ABORT that will YYNOMEM or YYABORT accordingly.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Move instances of YYABORT to YYNOMEM

Migration to improve error reporting as YYABORT cases should carry
event parsing errors.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Separate YYABORT and YYNOMEM cases

Split cases in event_pmu for greater accuracy.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-event: Add memory allocation test for name terms

If the name memory allocation fails then propagate to the parser.

Committer notes:

Use $(BISON_FALLBACK_FLAGS) on the bison call so that we continue
building with older bison versions, before 3.81, where YYNOMEM isn't
present.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Define YYNOMEM as YYNOABORT for bison < 3.81

YYNOMEM was introduced in bison 3.81, so define it as YYABORT for older
versions, which should provide the previous perf behaviour.

Cc: Adrian Hunter <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Avoid regrouped warning for wild card events

There is logic to avoid printing the regrouping warning for wild card
PMUs, this logic also needs to apply for wild card events.

Before:
```
$ perf stat -e '{data_read,data_write}' -a sleep 1
WARNING: events were regrouped to match PMUs

 Performance counter stats for 'system wide':

          2,979.16 MiB  data_read
            410.26 MiB  data_write

       1.001541923 seconds time elapsed
```
After:
```
$ perf stat -e '{data_read,data_write}' -a sleep 1

 Performance counter stats for 'system wide':

          2,975.94 MiB  data_read
            432.05 MiB  data_write

       1.001119499 seconds time elapsed
```

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Add more comments to 'struct parse_events_state'

Improve documentation of 'struct parse_events_state'.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Remove two unused tokens

The tokens PE_PREFIX_RAW and PE_PREFIX_GROUP are unused so remove them.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Remove unused PE_KERNEL_PMU_EVENT token

Removed by commit 70c90e4a6b2f ("perf parse-events: Avoid scanning
PMUs before parsing").

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Remove unused PE_PMU_EVENT_FAKE token

Removed by commit 70c90e4a6b2f ("perf parse-events: Avoid scanning
PMUs before parsing").

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf build: Add LTO build option

Add an LTO build option, that sets the appropriate CFLAGS and CXXFLAGS
values.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Carsten Haitzler <[email protected]>
Cc: Fangrui Song <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ravi Bangoria <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: Xing Zhengjun <[email protected]>
Cc: Yang Jihong <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf test: Avoid weak symbol for arch_tests

GCC LTO will complain that the array length varies for the arch_tests
weak symbol. Use extern/static and architecture determining #if to
workaround this problem.

Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Carsten Haitzler <[email protected]>
Cc: Fangrui Song <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Nick Desaulniers <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ravi Bangoria <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: Xing Zhengjun <[email protected]>
Cc: Yang Jihong <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf parse-events: Avoid use uninitialized warning

With GCC LTO a potential use uninitialized is spotted:
```
In function ‘parse_events_config_bpf’,
    inlined from ‘parse_events_load_bpf’ at util/parse-events.c:874:8:
util/parse-events.c:792:37: error: ‘error_pos’ may be used uninitialized [-Werror=maybe-uninitialized]
  792 |                                 idx = term->err_term + error_pos;
      |                                     ^
util/parse-events.c: In function ‘parse_events_load_bpf’:
util/parse-events.c:765:13: note: ‘error_pos’ was declared here
  765 |         int error_pos;
      |             ^
```
So initialize at declaration.

Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Carsten Haitzler <[email protected]>
Cc: Fangrui Song <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ravi Bangoria <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: Xing Zhengjun <[email protected]>
Cc: Yang Jihong <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf stat: Avoid uninitialized use of perf_stat_config

perf_event__read_stat_config will assign values based on number of
tags and tag values. Initialize the structs to zero before they are
assigned so that no uninitialized values can be seen.

This potential error was reported by GCC with LTO enabled.

Reviewed-by: Nick Desaulniers <[email protected]>
Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Carsten Haitzler <[email protected]>
Cc: Fangrui Song <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Nathan Chancellor <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Ravi Bangoria <[email protected]>
Cc: Tom Rix <[email protected]>
Cc: Xing Zhengjun <[email protected]>
Cc: Yang Jihong <[email protected]>
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf diff: Replaces some ',' as separator with the more usual ';'

When wrapping code, use ';' better than using ',' which is more in line
with the coding habits of most engineers.

Signed-off-by: Lu Hongfei <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bench uprobe trace_printk: Add entry attaching an BPF program that does a trace_printk

  [root@five ~]# perf bench uprobe all
  # Running uprobe/baseline benchmark...
  # Executed 1,000 usleep(1000) calls
       Total time: 1,053,963 usecs

   1,053.963 usecs/op

  # Running uprobe/empty benchmark...
  # Executed 1,000 usleep(1000) calls
       Total time: 1,056,293 usecs +2,330 to baseline

   1,056.293 usecs/op 2.330 usecs/op to baseline

  # Running uprobe/trace_printk benchmark...
  # Executed 1,000 usleep(1000) calls
       Total time: 1,056,977 usecs +3,014 to baseline +684 to previous

   1,056.977 usecs/op 3.014 usecs/op to baseline 0.684 usecs/op to previous

  [root@five ~]#

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andre Fredette <[email protected]>
Cc: Clark Williams <[email protected]>
Cc: Dave Tucker <[email protected]>
Cc: Derek Barbosa <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Masami Hiramatsu (Google) <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bench uprobe empty: Add entry attaching an empty BPF program

Using libbpf and a BPF skel:

  # perf bench uprobe all
  # Running uprobe/baseline benchmark...
  # Executed 1,000 usleep(1000) calls
       Total time: 1,055,618 usecs

   1,055.618 usecs/op
  # Running uprobe/empty benchmark...
  # Executed 1,000 usleep(1000) calls
       Total time: 1,057,146 usecs +1,528 to baseline

   1,057.146 usecs/op
  #

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andre Fredette <[email protected]>
Cc: Clark Williams <[email protected]>
Cc: Dave Tucker <[email protected]>
Cc: Derek Barbosa <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Masami Hiramatsu (Google) <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bench uprobe: Show diff to previous

Will be useful to show the incremental overhead as we do more stuff in
the BPF program attached to the uprobes.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andre Fredette <[email protected]>
Cc: Clark Williams <[email protected]>
Cc: Dave Tucker <[email protected]>
Cc: Derek Barbosa <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Masami Hiramatsu (Google) <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bench uprobe: Print diff to baseline

This is just prep work to show the diff to the unmodified workload.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andre Fredette <[email protected]>
Cc: Clark Williams <[email protected]>
Cc: Dave Tucker <[email protected]>
Cc: Derek Barbosa <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Masami Hiramatsu (Google) <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf bench uprobe: Add benchmark to test uprobe overhead

This just adds the initial "workload", a call to libc's usleep(1000us)
function:

  $ perf stat --null perf bench uprobe all
  # Running uprobe/baseline benchmark...
  # Executed 1000 usleep(1000) calls
       Total time: 1053533 usecs

   1053.533 usecs/op

   Performance counter stats for 'perf bench uprobe all':

         1.061042896 seconds time elapsed

         0.001079000 seconds user
         0.006499000 seconds sys

  $

More entries will be added using a BPF skel to add various uprobes to
the usleep() function.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Andre Fredette <[email protected]>
Cc: Clark Williams <[email protected]>
Cc: Dave Tucker <[email protected]>
Cc: Derek Barbosa <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Masami Hiramatsu (Google) <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

MAINTAINERS: Add git information for perf-tools and perf-tools-next trees/branches

Now the perf tools development is done on these trees/branches:

  git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools.git perf-tools
  git git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools-next.git perf-tools-next

For a while I'll continue mirroring what is these to the same branches
in my git tree.

Suggested-by: John Garry <[email protected]>
Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Link: https://lore.kernel.org/lkml/CAP-5=fVGOP6-k=BTRd_bn=N0HVy+1ShpdW5rk5ND0ZGhm_fQkg@mail.gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf trace: Free thread_trace->files table

The fd->pathname table that is kept in 'struct thread_trace' and thus in
thread->priv must be freed when a thread is deleted.

This was also detected using -fsanitize=address.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf trace: Really free the evsel->priv area

In 3cb4d5e00e037c70 ("perf trace: Free syscall tp fields in
evsel->priv") it only was freeing if strcmp(evsel->tp_format->system,
"syscalls") returned zero, while the corresponding initialization of
evsel->priv was being performed if it was _not_ zero, i.e. if the tp
system wasn't 'syscalls'.

Just stop looking for that and free it if evsel->priv was set, which
should be equivalent.

Also use the pre-existing evsel_trace__delete() function.

This resolves these leaks, detected with:

  $ make EXTRA_CFLAGS="-fsanitize=address" BUILD_BPF_SKEL=1 CORESIGHT=1 O=/tmp/build/perf-tools-next -C tools/perf install-bin

  =================================================================
  ==481565==ERROR: LeakSanitizer: detected memory leaks

  Direct leak of 40 byte(s) in 1 object(s) allocated from:
      #0 0x7f7343cba097 in calloc (/lib64/libasan.so.8+0xba097)
      #1 0x987966 in zalloc (/home/acme/bin/perf+0x987966)
      #2 0x52f9b9 in evsel_trace__new /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:307
      #3 0x52f9b9 in evsel__syscall_tp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:333
      #4 0x52f9b9 in evsel__init_raw_syscall_tp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:458
      #5 0x52f9b9 in perf_evsel__raw_syscall_newtp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:480
      #6 0x540e8b in trace__add_syscall_newtp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3212
      #7 0x540e8b in trace__run /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3891
      #8 0x540e8b in cmd_trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:5156
      #9 0x5ef262 in run_builtin /home/acme/git/perf-tools-next/tools/perf/perf.c:323
      #10 0x4196da in handle_internal_command /home/acme/git/perf-tools-next/tools/perf/perf.c:377
      #11 0x4196da in run_argv /home/acme/git/perf-tools-next/tools/perf/perf.c:421
      #12 0x4196da in main /home/acme/git/perf-tools-next/tools/perf/perf.c:537
      #13 0x7f7342c4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

  Direct leak of 40 byte(s) in 1 object(s) allocated from:
      #0 0x7f7343cba097 in calloc (/lib64/libasan.so.8+0xba097)
      #1 0x987966 in zalloc (/home/acme/bin/perf+0x987966)
      #2 0x52f9b9 in evsel_trace__new /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:307
      #3 0x52f9b9 in evsel__syscall_tp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:333
      #4 0x52f9b9 in evsel__init_raw_syscall_tp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:458
      #5 0x52f9b9 in perf_evsel__raw_syscall_newtp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:480
      #6 0x540dd1 in trace__add_syscall_newtp /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3205
      #7 0x540dd1 in trace__run /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3891
      #8 0x540dd1 in cmd_trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:5156
      #9 0x5ef262 in run_builtin /home/acme/git/perf-tools-next/tools/perf/perf.c:323
      #10 0x4196da in handle_internal_command /home/acme/git/perf-tools-next/tools/perf/perf.c:377
      #11 0x4196da in run_argv /home/acme/git/perf-tools-next/tools/perf/perf.c:421
      #12 0x4196da in main /home/acme/git/perf-tools-next/tools/perf/perf.c:537
      #13 0x7f7342c4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

  SUMMARY: AddressSanitizer: 80 byte(s) leaked in 2 allocation(s).
  [root@quaco ~]#

With this we plug all leaks with "perf trace sleep 1".

Fixes: 3cb4d5e00e037c70 ("perf trace: Free syscall tp fields in evsel->priv")
Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Riccardo Mancini <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf trace: Register a thread priv destructor

To plug these leaks detected with:

  $ make EXTRA_CFLAGS="-fsanitize=address" BUILD_BPF_SKEL=1 CORESIGHT=1 O=/tmp/build/perf-tools-next -C tools/perf install-bin

  =================================================================
  ==473890==ERROR: LeakSanitizer: detected memory leaks

  Direct leak of 112 byte(s) in 1 object(s) allocated from:
    #0 0x7fdf19aba097 in calloc (/lib64/libasan.so.8+0xba097)
    #1 0x987836 in zalloc (/home/acme/bin/perf+0x987836)
    #2 0x5367ae in thread_trace__new /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:1289
    #3 0x5367ae in thread__trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:1307
    #4 0x5367ae in trace__sys_exit /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:2468
    #5 0x52bf34 in trace__handle_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3177
    #6 0x52bf34 in __trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3685
    #7 0x542927 in trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3712
    #8 0x542927 in trace__run /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:4055
    #9 0x542927 in cmd_trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:5141
    #10 0x5ef1a2 in run_builtin /home/acme/git/perf-tools-next/tools/perf/perf.c:323
    #11 0x4196da in handle_internal_command /home/acme/git/perf-tools-next/tools/perf/perf.c:377
    #12 0x4196da in run_argv /home/acme/git/perf-tools-next/tools/perf/perf.c:421
    #13 0x4196da in main /home/acme/git/perf-tools-next/tools/perf/perf.c:537
    #14 0x7fdf18a4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

  Direct leak of 2048 byte(s) in 1 object(s) allocated from:
    #0 0x7f788fcba6af in __interceptor_malloc (/lib64/libasan.so.8+0xba6af)
    #1 0x5337c0 in trace__sys_enter /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:2342
    #2 0x52bfb4 in trace__handle_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3191
    #3 0x52bfb4 in __trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3699
    #4 0x542883 in trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3726
    #5 0x542883 in trace__run /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:4069
    #6 0x542883 in cmd_trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:5155
    #7 0x5ef232 in run_builtin /home/acme/git/perf-tools-next/tools/perf/perf.c:323
    #8 0x4196da in handle_internal_command /home/acme/git/perf-tools-next/tools/perf/perf.c:377
    #9 0x4196da in run_argv /home/acme/git/perf-tools-next/tools/perf/perf.c:421
    #10 0x4196da in main /home/acme/git/perf-tools-next/tools/perf/perf.c:537
    #11 0x7f788ec4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

  Indirect leak of 48 byte(s) in 1 object(s) allocated from:
    #0 0x7fdf19aba6af in __interceptor_malloc (/lib64/libasan.so.8+0xba6af)
    #1 0x77b335 in intlist__new util/intlist.c:116
    #2 0x5367fd in thread_trace__new /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:1293
    #3 0x5367fd in thread__trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:1307
    #4 0x5367fd in trace__sys_exit /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:2468
    #5 0x52bf34 in trace__handle_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3177
    #6 0x52bf34 in __trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3685
    #7 0x542927 in trace__deliver_event /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:3712
    #8 0x542927 in trace__run /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:4055
    #9 0x542927 in cmd_trace /home/acme/git/perf-tools-next/tools/perf/builtin-trace.c:5141
    #10 0x5ef1a2 in run_builtin /home/acme/git/perf-tools-next/tools/perf/perf.c:323
    #11 0x4196da in handle_internal_command /home/acme/git/perf-tools-next/tools/perf/perf.c:377
    #12 0x4196da in run_argv /home/acme/git/perf-tools-next/tools/perf/perf.c:421
    #13 0x4196da in main /home/acme/git/perf-tools-next/tools/perf/perf.c:537
    #14 0x7fdf18a4a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf thread: Allow tools to register a thread->priv destructor

So that when thread__delete() runs it can be called and free stuff tools
stashed into thread->priv, like 'perf trace' does and will use this
new facility to plug some leaks.

Added an assert(thread__priv_destructor == NULL) as suggested in Ian's
review.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/CAP-5=fV3Er=Ek8=iE=bSGbEBmM56_PJffMWot1g_5Bh8B5hO7A@mail.gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

perf evsel: Free evsel->filter on the destructor

Noticed with:

  make EXTRA_CFLAGS="-fsanitize=address" BUILD_BPF_SKEL=1 CORESIGHT=1 O=/tmp/build/perf-tools-next -C tools/perf install-bin

Direct leak of 45 byte(s) in 1 object(s) allocated from:
    #0 0x7f213f87243b in strdup (/lib64/libasan.so.8+0x7243b)
    #1 0x63d15f in evsel__set_filter util/evsel.c:1371
    #2 0x63d15f in evsel__append_filter util/evsel.c:1387
    #3 0x63d15f in evsel__append_tp_filter util/evsel.c:1400
    #4 0x62cd52 in evlist__append_tp_filter util/evlist.c:1145
    #5 0x62cd52 in evlist__append_tp_filter_pids util/evlist.c:1196
    #6 0x541e49 in trace__set_filter_loop_pids /home/acme/git/perf-tools/tools/perf/builtin-trace.c:3646
    #7 0x541e49 in trace__set_filter_pids /home/acme/git/perf-tools/tools/perf/builtin-trace.c:3670
    #8 0x541e49 in trace__run /home/acme/git/perf-tools/tools/perf/builtin-trace.c:3970
    #9 0x541e49 in cmd_trace /home/acme/git/perf-tools/tools/perf/builtin-trace.c:5141
    #10 0x5ef1a2 in run_builtin /home/acme/git/perf-tools/tools/perf/perf.c:323
    #11 0x4196da in handle_internal_command /home/acme/git/perf-tools/tools/perf/perf.c:377
    #12 0x4196da in run_argv /home/acme/git/perf-tools/tools/perf/perf.c:421
    #13 0x4196da in main /home/acme/git/perf-tools/tools/perf/perf.c:537
    #14 0x7f213e84a50f in __libc_start_call_main (/lib64/libc.so.6+0x2750f)

Free it on evsel__exit().

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

Merge tag 'perf-tools-fixes-for-v6.5-1-2023-07-18' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools

Pull perf tools fixes from Arnaldo Carvalho de Melo:

 - Don't group events when computing metrics that require more than the
   maximum number of simultaneously enabled events on AMD systems.

 - Fix multi CU handling in 'perf probe', add a 'perf test' entry to
   regress test it.

 - Make the 'perf test task_exit' stop generating samples by using the
   'dummy' event, all it is testing is if a PERF_RECORD_EXIT is
   generated at the end of a perf session. This makes this perf test to
   stop sometimes failing on some systems due to a full ring buffer.

 - Avoid SEGV if PMU lookup fails for legacy cache terms.

 - Fix libsubcmd SEGV/use-after-free when commands aren't excluded.

 - Fix OpenCSD (ARM64's CoreSight hardware tracing) library path
   resolution when specifying CSLIBS= in the make command line.

 - Fix broken feature check for libtracefs due to external lib changes,
   use the provided pkgconfig file instead future proof it.

 - Sync drm, fcntl, kvm, mount, prctl, socket, vhost, asound, arm64's
   cputype headers with the kernel sources, in some cases this made the
   tools become aware of new kernel APIs such as ioctls and the
   cachestat sysctl.

* tag 'perf-tools-fixes-for-v6.5-1-2023-07-18' of git://git.kernel.org/pub/scm/linux/kernel/git/perf/perf-tools:
  perf test task_exit: No need for a cycles event to check if we get an PERF_RECORD_EXIT
  tools headers arm64: Sync arm64's cputype.h with the kernel sources
  tools include UAPI: Sync the sound/asound.h copy with the kernel sources
  tools include UAPI: Sync linux/vhost.h with the kernel sources
  perf beauty: Update copy of linux/socket.h with the kernel sources
  perf parse-events: Avoid SEGV if PMU lookup fails for legacy cache terms
  libsubcmd: Avoid SEGV/use-after-free when commands aren't excluded
  tools headers UAPI: Sync linux/prctl.h with the kernel sources
  perf build: Fix broken feature check for libtracefs due to external lib changes
  tools include UAPI: Sync linux/mount.h copy with the kernel sources
  tools headers UAPI: Sync linux/kvm.h with the kernel sources
  tools headers uapi: Sync linux/fcntl.h with the kernel sources
  perf vendor events amd: Fix large metrics
  perf build: Fix library not found error when using CSLIBS
  tools headers UAPI: Sync files changed by new cachestat syscall with the kernel sources
  tools headers UAPI: Sync drm/i915_drm.h with the kernel sources
  perf probe: Read DWARF files from the correct CU
  perf probe: Add test for regression introduced by switch to die_get_decl_file()

Merge tag 'mm-hotfixes-stable-2023-07-18-12-28' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm

Pull hotfixes from Andrew Morton:
 "Seven hotfixes, six of which are cc:stable and one of which addresses
  a post-6.5 issue"

* tag 'mm-hotfixes-stable-2023-07-18-12-28' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm:
  maple_tree: fix node allocation testing on 32 bit
  maple_tree: fix 32 bit mas_next testing
  selftests/mm: mkdirty: fix incorrect position of #endif
  maple_tree: set the node limit when creating a new root node
  mm/mlock: fix vma iterator conversion of apply_vma_lock_flags()
  prctl: move PR_GET_AUXV out of PR_MCE_KILL
  selftests/mm: give scripts execute permission

Merge tag 'linux-kselftest-fixes-6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest

Pull Kselftest fixes from Shuah Khan:
 "Fixes to bugs that are interfering with arm64 and risc workflows. Also
  two fixes to timer and mincore tests that are causing test failures"

* tag 'linux-kselftest-fixes-6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/shuah/linux-kselftest:
  selftests/arm64: fix build failure during the "emit_tests" step
  selftests/riscv: fix potential build failure during the "emit_tests" step
  tools: timers: fix freq average calculation
  selftests/mincore: fix skip condition for check_huge_pages test

Merge tag 'tpmdd-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/jarkko/linux-tpmdd

Pull tpm fixes from Jarkko Sakkinen.

Mostly interrupt storm fixes, with some other minor changes.

* tag 'tpmdd-v6.5-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/jarkko/linux-tpmdd:
  tpm,tpm_tis: Disable interrupts after 1000 unhandled IRQs
  tpm/tpm_tis: Disable interrupts for Lenovo L590 devices
  tpm: Do not remap from ACPI resources again for Pluton TPM
  tpm/tpm_tis: Disable interrupts for Framework Laptop Intel 13th gen
  tpm/tpm_tis: Disable interrupts for Framework Laptop Intel 12th gen
  security: keys: Modify mismatched function name
  tpm: return false from tpm_amd_is_rng_defective on non-x86 platforms
  keys: Fix linking a duplicate key to a keyring's assoc_array
  tpm: tis_i2c: Limit write bursts to I2C_SMBUS_BLOCK_MAX (32) bytes
  tpm: tis_i2c: Limit read bursts to I2C_SMBUS_BLOCK_MAX (32) bytes
  tpm_tis_spi: Release chip select when flow control fails
  tpm: tpm_tis: Disable interrupts *only* for AEON UPX-i11
  tpm: tpm_vtpm_proxy: fix a race condition in /dev/vtpmx creation

maple_tree: fix node allocation testing on 32 bit

Internal node counting was altered and the 64 bit test was updated,
however the 32bit test was missed.

Restore the 32bit test to a functional state.

Link: https://lore.kernel.org/linux-mm/CAMuHMdV4T53fOw7VPoBgPR7fP6RYqf=CBhD_y_vOg53zZX_DnA@mail.gmail.com/
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 541e06b772c1 ("maple_tree: remove GFP_ZERO from kmem_cache_alloc() and kmem_cache_alloc_bulk()")
Signed-off-by: Liam R. Howlett <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

maple_tree: fix 32 bit mas_next testing

The test setup of mas_next is dependent on node entry size to create a 2
level tree, but the tests did not account for this in the expected value
when shifting beyond the scope of the tree.

Fix this by setting up the test to succeed depending on the node entries
which is dependent on the 32/64 bit setup.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 120b116208a0 ("maple_tree: reorganize testing to restore module testing")
Signed-off-by: Liam R. Howlett <[email protected]>
Reported-by: Geert Uytterhoeven <[email protected]>
Closes: https://lore.kernel.org/linux-mm/CAMuHMdV4T53fOw7VPoBgPR7fP6RYqf=CBhD_y_vOg53zZX_DnA@mail.gmail.com/
Tested-by: Geert Uytterhoeven <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

selftests/mm: mkdirty: fix incorrect position of #endif

The #endif is the wrong side of a } causing a build failure when
__NR_userfaultfd is not defined.  Fix this by moving the #end to enclose
the }

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 9eac40fc0cc7 ("selftests/mm: mkdirty: test behavior of (pte|pmd)_mkdirty on VMAs without write permissions")
Signed-off-by: Colin Ian King <[email protected]>
Reviewed-by: David Hildenbrand <[email protected]>
Cc: Shuah Khan <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

maple_tree: set the node limit when creating a new root node

Set the node limit of the root node so that the last pivot of all nodes is
the node limit (if the node is not full).

This patch also fixes a bug in mas_rev_awalk().  Effectively, always
setting a maximum makes mas_logical_pivot() behave as mas_safe_pivot().
Without this fix, it is possible that very small tasks would fail to find
the correct gap.  Although this has not been observed with real tasks, it
has been reported to happen in m68k nommu running the maple tree tests.

Link: https://lkml.kernel.org/r/[email protected]
Link: https://lore.kernel.org/linux-mm/CAMuHMdV4T53fOw7VPoBgPR7fP6RYqf=CBhD_y_vOg53zZX_DnA@mail.gmail.com/
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 54a611b60590 ("Maple Tree: add new data structure")
Signed-off-by: Peng Zhang <[email protected]>
Reviewed-by: Liam R. Howlett <[email protected]>
Tested-by: Geert Uytterhoeven <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

mm/mlock: fix vma iterator conversion of apply_vma_lock_flags()

apply_vma_lock_flags() calls mlock_fixup(), which could merge the VMA
after where the vma iterator is located.  Although this is not an issue,
the next iteration of the loop will check the start of the vma to be equal
to the locally saved 'tmp' variable and cause an incorrect failure
scenario.  Fix the error by setting tmp to the end of the vma iterator
value before restarting the loop.

There is also a potential of the error code being overwritten when the
loop terminates early.  Fix the return issue by directly returning when an
error is encountered since there is nothing to undo after the loop.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 37598f5a9d8b ("mlock: convert mlock to vma iterator")
Signed-off-by: Liam R. Howlett <[email protected]>
Reported-by: Ryan Roberts <[email protected]>
Link: https://lore.kernel.org/linux-mm/[email protected]/
Tested-by: Ryan Roberts <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

prctl: move PR_GET_AUXV out of PR_MCE_KILL

Somehow PR_GET_AUXV got added into PR_MCE_KILL's switch when the patch was
applied [1].

Thus move it out of the switch, to the place the patch added it.

In the recently released v6.4 kernel some user could, in principle, be
already using this feature by mapping the right page and passing the
PR_GET_AUXV constant as a pointer:

    prctl(PR_MCE_KILL, PR_GET_AUXV, ...)

So this does change the behavior for users.  We could keep the bug since
the other subcases in PR_MCE_KILL (PR_MCE_KILL_CLEAR and PR_MCE_KILL_SET)
do not overlap.

However, v6.4 may be recent enough (2 weeks old) that moving the lines
(rather than just adding a new case) does not break anybody?  Moreover,
the documentation in man-pages was just committed today [2].

Link: https://lkml.kernel.org/r/[email protected]
Fixes: ddc65971bb67 ("prctl: add PR_GET_AUXV to copy auxv to userspace")
Link: https://lore.kernel.org/all/d81864a7f7f43bca6afa2a09fc2e850e4050ab42.1680611394.git.josh@joshtriplett.org/
Link: https://git.kernel.org/pub/scm/docs/man-pages/man-pages.git/commit/?id=8cf0c06bfd3c2b219b044d4151c96f0da50af9ad
Signed-off-by: Miguel Ojeda <[email protected]>
Cc: Josh Triplett <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

tpm,tpm_tis: Disable interrupts after 1000 unhandled IRQs

After activation of interrupts for TPM TIS drivers 0-day reports an
interrupt storm on an Inspur NF5180M6 server.

Fix this by detecting the storm and falling back to polling:
Count the number of unhandled interrupts within a 10 ms time interval. In
case that more than 1000 were unhandled deactivate interrupts entirely,
deregister the handler and use polling instead.

Also print a note to point to the tpm_tis_dmi_table.

Since the interrupt deregistration function devm_free_irq() waits for all
interrupt handlers to finish, only trigger a worker in the interrupt
handler and do the unregistration in the worker to avoid a deadlock.

Note: the storm detection logic equals the implementation in
note_interrupt() which uses timestamps and counters stored in struct
irq_desc. Since this structure is private to the generic interrupt core
the TPM TIS core uses its own timestamps and counters. Furthermore the TPM
interrupt handler always returns IRQ_HANDLED to prevent the generic
interrupt core from processing the interrupt storm.

Cc: [email protected] # v6.4+
Fixes: e644b2f498d2 ("tpm, tpm_tis: Enable interrupt test")
Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-lkp/[email protected]/
Suggested-by: Lukas Wunner <[email protected]>
Signed-off-by: Lino Sanfilippo <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm/tpm_tis: Disable interrupts for Lenovo L590 devices

The Lenovo L590 suffers from an irq storm issue like the T490, T490s
and P360 Tiny, so add an entry for it to tpm_tis_dmi_table and force
polling.

Cc: [email protected] # v6.4+
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2214069#c0
Fixes: e644b2f498d2 ("tpm, tpm_tis: Enable interrupt test")
Signed-off-by: Florian Bezdeka <[email protected]>
Reviewed-by: Jerry Snitselaar <[email protected]>
Reviewed-by: Hans de Goede <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: Do not remap from ACPI resources again for Pluton TPM

For Pluton TPM devices, it was assumed that there was no ACPI memory
regions. This is not true for ASUS ROG Ally. ACPI advertises
0xfd500000-0xfd5fffff.

Since remapping is already done in `crb_map_pluton`, remapping again
in `crb_map_io` causes EBUSY error:

[    3.510453] tpm_crb MSFT0101:00: can't request region for resource [mem 0xfd500000-0xfd5fffff]
[    3.510463] tpm_crb: probe of MSFT0101:00 failed with error -16

Cc: [email protected] # v6.3+
Fixes: 4d2732882703 ("tpm_crb: Add support for CRB devices based on Pluton")
Signed-off-by: Valentin David <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm/tpm_tis: Disable interrupts for Framework Laptop Intel 13th gen

This device suffer an irq storm, so add it in tpm_tis_dmi_table to
force polling.

Cc: [email protected] # v6.4+
Link: https://community.frame.work/t/boot-and-shutdown-hangs-with-arch-linux-kernel-6-4-1-mainline-and-arch/33118
Fixes: e644b2f498d2 ("tpm, tpm_tis: Enable interrupt test")
Reported-by: <[email protected]>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217631
Signed-off-by: Christian Hesse <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm/tpm_tis: Disable interrupts for Framework Laptop Intel 12th gen

This device suffer an irq storm, so add it in tpm_tis_dmi_table to
force polling.

Cc: [email protected] # v6.4+
Link: https://community.frame.work/t/boot-and-shutdown-hangs-with-arch-linux-kernel-6-4-1-mainline-and-arch/33118
Fixes: e644b2f498d2 ("tpm, tpm_tis: Enable interrupt test")
Reported-by: <[email protected]>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217631
Signed-off-by: Christian Hesse <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

security: keys: Modify mismatched function name

No functional modification involved.

security/keys/trusted-keys/trusted_tpm2.c:203: warning: expecting prototype for tpm_buf_append_auth(). Prototype was for tpm2_buf_append_auth() instead.

Fixes: 2e19e10131a0 ("KEYS: trusted: Move TPM2 trusted keys code")
Reported-by: Abaci Robot <[email protected]>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=5524
Signed-off-by: Jiapeng Chong <[email protected]>
Reviewed-by: Paul Moore <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: return false from tpm_amd_is_rng_defective on non-x86 platforms

tpm_amd_is_rng_defective is for dealing with an issue related to the
AMD firmware TPM, so on non-x86 architectures just have it inline and
return false.

Cc: [email protected] # v6.3+
Reported-by: Sachin Sant <[email protected]>
Reported-by: Aneesh Kumar K. V <[email protected]>
Closes: https://lore.kernel.org/lkml/[email protected]/
Fixes: f1324bbc4011 ("tpm: disable hwrng for fTPM on some AMD designs")
Signed-off-by: Jerry Snitselaar <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

keys: Fix linking a duplicate key to a keyring's assoc_array

When making a DNS query inside the kernel using dns_query(), the request
code can in rare cases end up creating a duplicate index key in the
assoc_array of the destination keyring. It is eventually found by
a BUG_ON() check in the assoc_array implementation and results in
a crash.

Example report:
[2158499.700025] kernel BUG at ../lib/assoc_array.c:652!
[2158499.700039] invalid opcode: 0000 [#1] SMP PTI
[2158499.700065] CPU: 3 PID: 31985 Comm: kworker/3:1 Kdump: loaded Not tainted 5.3.18-150300.59.90-default #1 SLE15-SP3
[2158499.700096] Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 11/12/2020
[2158499.700351] Workqueue: cifsiod cifs_resolve_server [cifs]
[2158499.700380] RIP: 0010:assoc_array_insert+0x85f/0xa40
[2158499.700401] Code: ff 74 2b 48 8b 3b 49 8b 45 18 4c 89 e6 48 83 e7 fe e8 95 ec 74 00 3b 45 88 7d db 85 c0 79 d4 0f 0b 0f 0b 0f 0b e8 41 f2 be ff <0f> 0b 0f 0b 81 7d 88 ff ff ff 7f 4c 89 eb 4c 8b ad 58 ff ff ff 0f
[2158499.700448] RSP: 0018:ffffc0bd6187faf0 EFLAGS: 00010282
[2158499.700470] RAX: ffff9f1ea7da2fe8 RBX: ffff9f1ea7da2fc1 RCX: 0000000000000005
[2158499.700492] RDX: 0000000000000000 RSI: 0000000000000005 RDI: 0000000000000000
[2158499.700515] RBP: ffffc0bd6187fbb0 R08: ffff9f185faf1100 R09: 0000000000000000
[2158499.700538] R10: ffff9f1ea7da2cc0 R11: 000000005ed8cec8 R12: ffffc0bd6187fc28
[2158499.700561] R13: ffff9f15feb8d000 R14: ffff9f1ea7da2fc0 R15: ffff9f168dc0d740
[2158499.700585] FS:  0000000000000000(0000) GS:ffff9f185fac0000(0000) knlGS:0000000000000000
[2158499.700610] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[2158499.700630] CR2: 00007fdd94fca238 CR3: 0000000809d8c006 CR4: 00000000003706e0
[2158499.700702] Call Trace:
[2158499.700741]  ? key_alloc+0x447/0x4b0
[2158499.700768]  ? __key_link_begin+0x43/0xa0
[2158499.700790]  __key_link_begin+0x43/0xa0
[2158499.700814]  request_key_and_link+0x2c7/0x730
[2158499.700847]  ? dns_resolver_read+0x20/0x20 [dns_resolver]
[2158499.700873]  ? key_default_cmp+0x20/0x20
[2158499.700898]  request_key_tag+0x43/0xa0
[2158499.700926]  dns_query+0x114/0x2ca [dns_resolver]
[2158499.701127]  dns_resolve_server_name_to_ip+0x194/0x310 [cifs]
[2158499.701164]  ? scnprintf+0x49/0x90
[2158499.701190]  ? __switch_to_asm+0x40/0x70
[2158499.701211]  ? __switch_to_asm+0x34/0x70
[2158499.701405]  reconn_set_ipaddr_from_hostname+0x81/0x2a0 [cifs]
[2158499.701603]  cifs_resolve_server+0x4b/0xd0 [cifs]
[2158499.701632]  process_one_work+0x1f8/0x3e0
[2158499.701658]  worker_thread+0x2d/0x3f0
[2158499.701682]  ? process_one_work+0x3e0/0x3e0
[2158499.701703]  kthread+0x10d/0x130
[2158499.701723]  ? kthread_park+0xb0/0xb0
[2158499.701746]  ret_from_fork+0x1f/0x40

The situation occurs as follows:
* Some kernel facility invokes dns_query() to resolve a hostname, for
  example, "abcdef". The function registers its global DNS resolver
  cache as current->cred.thread_keyring and passes the query to
  request_key_net() -> request_key_tag() -> request_key_and_link().
* Function request_key_and_link() creates a keyring_search_context
  object. Its match_data.cmp method gets set via a call to
  type->match_preparse() (resolves to dns_resolver_match_preparse()) to
  dns_resolver_cmp().
* Function request_key_and_link() continues and invokes
  search_process_keyrings_rcu() which returns that a given key was not
  found. The control is then passed to request_key_and_link() ->
  construct_alloc_key().
* Concurrently to that, a second task similarly makes a DNS query for
  "abcdef." and its result gets inserted into the DNS resolver cache.
* Back on the first task, function construct_alloc_key() first runs
  __key_link_begin() to determine an assoc_array_edit operation to
  insert a new key. Index keys in the array are compared exactly as-is,
  using keyring_compare_object(). The operation finds that "abcdef" is
  not yet present in the destination keyring.
* Function construct_alloc_key() continues and checks if a given key is
  already present on some keyring by again calling
  search_process_keyrings_rcu(). This search is done using
  dns_resolver_cmp() and "abcdef" gets matched with now present key
  "abcdef.".
* The found key is linked on the destination keyring by calling
  __key_link() and using the previously calculated assoc_array_edit
  operation. This inserts the "abcdef." key in the array but creates
  a duplicity because the same index key is already present.

Fix the problem by postponing __key_link_begin() in
construct_alloc_key() until an actual key which should be linked into
the destination keyring is determined.

[[email protected]: added a fixes tag and cc to stable]
Cc: [email protected] # v5.3+
Fixes: df593ee23e05 ("keys: Hoist locking out of __key_link_begin()")
Signed-off-by: Petr Pavlu <[email protected]>
Reviewed-by: Joey Lee <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: tis_i2c: Limit write bursts to I2C_SMBUS_BLOCK_MAX (32) bytes

Underlying I2C bus drivers not always support longer transfers and
imx-lpi2c for instance doesn't. The fix is symmetric to previous patch
which fixed the read direction.

Cc: [email protected] # v5.20+
Fixes: bbc23a07b072 ("tpm: Add tpm_tis_i2c backend for tpm_tis_core")
Tested-by: Michael Haener <[email protected]>
Signed-off-by: Alexander Sverdlin <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Reviewed-by: Jerry Snitselaar <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: tis_i2c: Limit read bursts to I2C_SMBUS_BLOCK_MAX (32) bytes

Underlying I2C bus drivers not always support longer transfers and
imx-lpi2c for instance doesn't. SLB 9673 offers 427-bytes packets.

Visible symptoms are:

tpm tpm0: Error left over data
tpm tpm0: tpm_transmit: tpm_recv: error -5
tpm_tis_i2c: probe of 1-002e failed with error -5

Cc: [email protected] # v5.20+
Fixes: bbc23a07b072 ("tpm: Add tpm_tis_i2c backend for tpm_tis_core")
Tested-by: Michael Haener <[email protected]>
Signed-off-by: Alexander Sverdlin <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Reviewed-by: Jerry Snitselaar <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm_tis_spi: Release chip select when flow control fails

The failure paths in tpm_tis_spi_transfer() do not deactivate
chip select. Send an empty message (cs_select == 0) to overcome
this.

The patch is tested by two ways.
One way needs to touch hardware:
   1. force pull MISO pin down to GND, it emulates a forever
      'WAIT' timing.
   2. probe cs pin by an oscilloscope.
   3. load tpm_tis_spi.ko.
After loading, dmesg prints:
    "probe of spi0.0 failed with error -110"
and oscilloscope shows cs pin goes high(deactivated) after
the failure. Before the patch, cs pin keeps low.

Second way is by writing a fake spi controller.
   1. implement .transfer_one method, fill all rx buf with 0.
   2. implement .set_cs method, print the state of cs pin.
   we can see cs goes high after the failure.

Signed-off-by: Peijie Shao <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: tpm_tis: Disable interrupts *only* for AEON UPX-i11

Further restrict with DMI_PRODUCT_VERSION.

Cc: [email protected] # v6.4+
Link: https://lore.kernel.org/linux-integrity/[email protected]/
Fixes: 95a9359ee22f ("tpm: tpm_tis: Disable interrupts for AEON UPX-i11")
Signed-off-by: Peter Ujfalusi <[email protected]>
Reviewed-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

tpm: tpm_vtpm_proxy: fix a race condition in /dev/vtpmx creation

/dev/vtpmx is made visible before 'workqueue' is initialized, which can
lead to a memory corruption in the worst case scenario.

Address this by initializing 'workqueue' as the very first step of the
driver initialization.

Cc: [email protected]
Fixes: 6f99612e2500 ("tpm: Proxy driver for supporting multiple emulated TPMs")
Reviewed-by: Stefan Berger <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>
Signed-off-by: Jarkko Sakkinen <[email protected]>

perf test task_exit: No need for a cycles event to check if we get an PERF_RECORD_EXIT

The intent of this test is to check we get a PERF_RECORD_EXIT as asked
for by setting perf_event_attr.task=1.

When the test was written we didn't had the "dummy" event so we went
with the default event, "cycles".

There were reports of this test failing sometimes, one of these reports
was with a PREEMPT_RT_FULL, but I noticed it failing sometimes with an
aarch64 Firefly board.

In the kernel the call to perf_event_task_output(), that generates the
PERF_RECORD_EXIT may fail when there is not enough memory in the ring
buffer, if the ring buffer is paused, etc.

So switch to using the "dummy" event to use the ring buffer just for
what the test was designed for, avoiding uneeded PERF_RECORD_SAMPLEs.

Acked-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Juri Lelli <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: https://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>

Linux 6.5-rc2

Merge tag 'xtensa-20230716' of https://github.com/jcmvbkbc/linux-xtensa

Pull xtensa fixes from Max Filippov:

 - fix interaction between unaligned exception handler and load/store
   exception handler

 - fix parsing ISS network interface specification string

 - add comment about etherdev freeing to ISS network driver

* tag 'xtensa-20230716' of https://github.com/jcmvbkbc/linux-xtensa:
  xtensa: fix unaligned and load/store configuration interaction
  xtensa: ISS: fix call to split_if_spec
  xtensa: ISS: add comment about etherdev freeing

Merge tag 'perf_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull perf fix from Borislav Petkov:

 - Fix a lockdep warning when the event given is the first one, no event
   group exists yet but the code still goes and iterates over event
   siblings

* tag 'perf_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/x86: Fix lockdep warning in for_each_sibling_event() on SPR

Merge tag 'objtool_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull objtool fixes from Borislav Petkov:

 - Mark copy_iovec_from_user() __noclone in order to prevent gcc from
   doing an inter-procedural optimization and confuse objtool

 - Initialize struct elf fully to avoid build failures

* tag 'objtool_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  iov_iter: Mark copy_iovec_from_user() noclone
  objtool: initialize all of struct elf

Merge tag 'sched_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull scheduler fixes from Borislav Petkov:

 - Remove a cgroup from under a polling process properly

 - Fix the idle sibling selection

* tag 'sched_urgent_for_v6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/psi: use kernfs polling functions for PSI trigger polling
  sched/fair: Use recent_used_cpu to test p->cpus_ptr

Merge tag 'pinctrl-v6.5-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:
 "I'm mostly on vacation but what would vacation be without a few
  critical fixes so people can use their gaming laptops when hiding away
  from the sun (or rain)?

   - Fix a really annoying interrupt storm in the AMD driver affecting
     Asus TUF gaming notebooks

   - Fix device tree parsing in the Renesas driver"

* tag 'pinctrl-v6.5-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: amd: Unify debounce handling into amd_pinconf_set()
  pinctrl: amd: Drop pull up select configuration
  pinctrl: amd: Use amd_pinconf_set() for all config options
  pinctrl: amd: Only use special debounce behavior for GPIO 0
  pinctrl: renesas: rzg2l: Handle non-unique subnode names
  pinctrl: renesas: rzv2m: Handle non-unique subnode names

Merge tag '6.5-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:

 - Two reconnect fixes: important fix to address inFlight count to leak
   (which can leak credits), and fix for better handling a deleted share

 - DFS fix

 - SMB1 cleanup fix

 - deferred close fix

* tag '6.5-rc1-smb3-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: fix mid leak during reconnection after timeout threshold
  cifs: is_network_name_deleted should return a bool
  smb: client: fix missed ses refcounting
  smb: client: Fix -Wstringop-overflow issues
  cifs: if deferred close is disabled then close files immediately

Merge tag 'powerpc-6.5-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux

Pull powerpc fixes from Michael Ellerman:

 - Fix Speculation_Store_Bypass reporting in /proc/self/status on
   Power10

 - Fix HPT with 4K pages since recent changes by implementing pmd_same()

 - Fix 64-bit native_hpte_remove() to be irq-safe

Thanks to Aneesh Kumar K.V, Nageswara R Sastry, and Russell Currey.

* tag 'powerpc-6.5-3' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/mm/book3s64/hash/4k: Add pmd_same callback for 4K page size
  powerpc/64e: Fix obtool warnings in exceptions-64e.S
  powerpc/security: Fix Speculation_Store_Bypass reporting on Power10
  powerpc/64s: Fix native_hpte_remove() to be irq-safe

Merge tag 'hardening-v6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux

Pull hardening fixes from Kees Cook:

 - Remove LTO-only suffixes from promoted global function symbols
   (Yonghong Song)

 - Remove unused .text..refcount section from vmlinux.lds.h (Petr Pavlu)

 - Add missing __always_inline to sparc __arch_xchg() (Arnd Bergmann)

 - Claim maintainership of string routines

* tag 'hardening-v6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  sparc: mark __arch_xchg() as __always_inline
  MAINTAINERS: Foolishly claim maintainership of string routines
  kallsyms: strip LTO-only suffixes from promoted global functions
  vmlinux.lds.h: Remove a reference to no longer used sections .text..refcount

Merge tag 'probes-fixes-v6.5-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull probe fixes from Masami Hiramatsu:

 - fprobe: Add a comment why fprobe will be skipped if another kprobe is
   running in fprobe_kprobe_handler().

 - probe-events: Fix some issues related to fetch-arguments:

    - Fix double counting of the string length for user-string and
      symstr. This will require longer buffer in the array case.

    - Fix not to count error code (minus value) for the total used
      length in array argument. This makes the total used length
      shorter.

    - Fix to update dynamic used data size counter only if fetcharg uses
      the dynamic size data. This may mis-count the used dynamic data
      size and corrupt data.

    - Revert "tracing: Add "(fault)" name injection to kernel probes"
      because that did not work correctly with a bug, and we agreed the
      current '(fault)' output (instead of '"(fault)"' like a string)
      explains what happened more clearly.

    - Fix to record 0-length (means fault access) data_loc data in fetch
      function itself, instead of store_trace_args(). If we record an
      array of string, this will fix to save fault access data on each
      entry of the array correctly.

* tag 'probes-fixes-v6.5-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing/probes: Fix to record 0-length data_loc in fetch_store_string*() if fails
  Revert "tracing: Add "(fault)" name injection to kernel probes"
  tracing/probes: Fix to update dynamic data counter if fetcharg uses it
  tracing/probes: Fix not to count error code to total length
  tracing/probes: Fix to avoid double count of the string length on the array
  fprobes: Add a comment why fprobe_kprobe_handler exits if kprobe is running

Merge tag 'spi-fix-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi

Pull spi fixes from Mark Brown:
 "A couple of fairly minor driver specific fixes here, plus a bunch of
  maintainership and admin updates. Nothing too remarkable"

* tag 'spi-fix-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  mailmap: add entry for Jonas Gorski
  MAINTAINERS: add myself for spi-bcm63xx
  spi: s3c64xx: clear loopback bit after loopback test
  spi: bcm63xx: fix max prepend length
  MAINTAINERS: Add myself as a maintainer for Microchip SPI

Merge tag 'regmap-fix-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap

Pull regmap fix from Mark Brown:
 "One fix for an out of bounds access in the interupt code here"

* tag 'regmap-fix-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
  regmap-irq: Fix out-of-bounds access when allocating config buffers

Merge tag 'iommu-fixes-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu

Pull iommu fixes from Joerg Roedel:

 - Fix a regression causing a crash on sysfs access of iommu-group
   specific files

 - Fix signedness bug in SVA code

* tag 'iommu-fixes-v6.5-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/sva: Fix signedness bug in iommu_sva_alloc_pasid()
  iommu: Fix crash during syfs iommu_groups/N/type

Merge tag 'x86_urgent_for_6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip

Pull x86 CFI fixes from Peter Zijlstra:
 "Fix kCFI/FineIBT weaknesses

  The primary bug Alyssa noticed was that with FineIBT enabled function
  prologues have a spurious ENDBR instruction:

    __cfi_foo:
endbr64
subl $hash, %r10d
jz 1f
ud2
nop
    1:
    foo:
endbr64 <--- *sadface*

  This means that any indirect call that fails to target the __cfi
  symbol and instead targets (the regular old) foo+0, will succeed due
  to that second ENDBR.

  Fixing this led to the discovery of a single indirect call that was
  still doing this: ret_from_fork(). Since that's an assembly stub the
  compiler would not generate the proper kCFI indirect call magic and it
  would not get patched.

  Brian came up with the most comprehensive fix -- convert the thing to
  C with only a very thin asm wrapper. This ensures the kernel thread
  boostrap is a proper kCFI call.

  While discussing all this, Kees noted that kCFI hashes could/should be
  poisoned to seal all functions whose address is never taken, further
  limiting the valid kCFI targets -- much like we already do for IBT.

  So what was a 'simple' observation and fix cascaded into a bunch of
  inter-related CFI infrastructure fixes"

* tag 'x86_urgent_for_6.5_rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/cfi: Only define poison_cfi() if CONFIG_X86_KERNEL_IBT=y
  x86/fineibt: Poison ENDBR at +0
  x86: Rewrite ret_from_fork() in C
  x86/32: Remove schedule_tail_wrapper()
  x86/cfi: Extend ENDBR sealing to kCFI
  x86/alternative: Rename apply_ibt_endbr()
  x86/cfi: Extend {JMP,CAKK}_NOSPEC comment

Merge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "This is a bunch of small driver fixes and a larger rework of zone disk
  handling (which reaches into blk and nvme).

  The aacraid array-bounds fix is now critical since the security people
  turned on -Werror for some build tests, which now fail without it"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: storvsc: Handle SRB status value 0x30
  scsi: block: Improve checks in blk_revalidate_disk_zones()
  scsi: block: virtio_blk: Set zone limits before revalidating zones
  scsi: block: nullblk: Set zone limits before revalidating zones
  scsi: nvme: zns: Set zone limits before revalidating zones
  scsi: sd_zbc: Set zone limits before revalidating zones
  scsi: ufs: core: Add support for qTimestamp attribute
  scsi: aacraid: Avoid -Warray-bounds warning
  scsi: ufs: ufs-mediatek: Add dependency for RESET_CONTROLLER
  scsi: ufs: core: Update contact email for monitor sysfs nodes
  scsi: scsi_debug: Remove dead code
  scsi: qla2xxx: Use vmalloc_array() and vcalloc()
  scsi: fnic: Use vmalloc_array() and vcalloc()
  scsi: qla2xxx: Fix error code in qla2x00_start_sp()
  scsi: qla2xxx: Silence a static checker warning
  scsi: lpfc: Fix a possible data race in lpfc_unregister_fcf_rescan()

Merge tag 'block-6.5-2023-07-14' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:

 - NVMe pull request via Keith:
      - Don't require quirk to use duplicate namespace identifiers
        (Christoph, Sagi)
      - One more BOGUS_NID quirk (Pankaj)
      - IO timeout and error hanlding fixes for PCI (Keith)
      - Enhanced metadata format mask fix (Ankit)
      - Association race condition fix for fibre channel (Michael)
      - Correct debugfs error checks (Minjie)
      - Use PAGE_SECTORS_SHIFT where needed (Damien)
      - Reduce kernel logs for legacy nguid attribute (Keith)
      - Use correct dma direction when unmapping metadata (Ming)

 - Fix for a flush handling regression in this release (Christoph)

 - Fix for batched request time stamping (Chengming)

 - Fix for a regression in the mq-deadline position calculation (Bart)

 - Lockdep fix for blk-crypto (Eric)

 - Fix for a regression in the Amiga partition handling changes
   (Michael)

* tag 'block-6.5-2023-07-14' of git://git.kernel.dk/linux:
  block: queue data commands from the flush state machine at the head
  blk-mq: fix start_time_ns and alloc_time_ns for pre-allocated rq
  nvme-pci: fix DMA direction of unmapping integrity data
  nvme: don't reject probe due to duplicate IDs for single-ported PCIe devices
  block/mq-deadline: Fix a bug in deadline_from_pos()
  nvme: ensure disabling pairs with unquiesce
  nvme-fc: fix race between error recovery and creating association
  nvme-fc: return non-zero status code when fails to create association
  nvme: fix parameter check in nvme_fault_inject_init()
  nvme: warn only once for legacy uuid attribute
  block: remove dead struc request->completion_data field
  nvme: fix the NVME_ID_NS_NVM_STS_MASK definition
  nvmet: use PAGE_SECTORS_SHIFT
  nvme: add BOGUS_NID quirk for Samsung SM953
  blk-crypto: use dynamic lock class for blk_crypto_profile::lock
  block/partition: fix signedness issue for Amiga partitions

Merge tag 'io_uring-6.5-2023-07-14' of git://git.kernel.dk/linux

Pull io_uring fix from Jens Axboe:
 "Just a single tweak for the wait logic in io_uring"

* tag 'io_uring-6.5-2023-07-14' of git://git.kernel.dk/linux:
  io_uring: Use io_schedule* in cqring wait

selftests/arm64: fix build failure during the "emit_tests" step

The build failure reported in [1] occurred because commit <9fc96c7c19df>
("selftests: error out if kernel header files are not yet built") added
a new "kernel_header_files" dependency to "all", and that triggered
another, pre-existing problem. Specifically, the arm64 selftests
override the emit_tests target, and that override improperly declares
itself to depend upon the "all" target.

This is a problem because the "emit_tests" target in lib.mk was not
intended to be overridden. emit_tests is a very simple, sequential build
target that was originally invoked from the "install" target, which in
turn, depends upon "all".

That approach worked for years. But with 9fc96c7c19df in place,
emit_tests failed, because it does not set up all of the elaborate
things that "install" does. And that caused the new
"kernel_header_files" target (which depends upon $(KBUILD_OUTPUT) being
correct) to fail.

Some detail: The "all" target is .PHONY. Therefore, each target that
depends on "all" will cause it to be invoked again, and because
dependencies are managed quite loosely in the selftests Makefiles, many
things will run, even "all" is invoked several times in immediate
succession. So this is not a "real" failure, as far as build steps go:
everything gets built, but "all" reports a problem when invoked a second
time from a bad environment.

To fix this, simply remove the unnecessary "all" dependency from the
overridden emit_tests target. The dependency is still effectively
honored, because again, invocation is via "install", which also depends
upon "all".

An alternative approach would be to harden the emit_tests target so that
it can depend upon "all", but that's a lot more complicated and hard to
get right, and doesn't seem worth it, especially given that emit_tests
should probably not be overridden at all.

[1] https://lore.kernel.org/20230710-kselftest-fix-arm64-v1-1-48e872844f25@kernel.org

Fixes: 9fc96c7c19df ("selftests: error out if kernel header files are not yet built")
Reported-by: Mark Brown <[email protected]>
Signed-off-by: John Hubbard <[email protected]>
Tested-by: Mark Brown <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>

selftests/riscv: fix potential build failure during the "emit_tests" step

The riscv selftests (which were modeled after the arm64 selftests) are
improperly declaring the "emit_tests" target to depend upon the "all"
target. This approach, when combined with commit 9fc96c7c19df
("selftests: error out if kernel header files are not yet built"), has
caused build failures [1] on arm64, and is likely to cause similar
failures for riscv.

To fix this, simply remove the unnecessary "all" dependency from the
emit_tests target. The dependency is still effectively honored, because
again, invocation is via "install", which also depends upon "all".

An alternative approach would be to harden the emit_tests target so that
it can depend upon "all", but that's a lot more complicated and hard to
get right, and doesn't seem worth it, especially given that emit_tests
should probably not be overridden at all.

[1] https://lore.kernel.org/20230710-kselftest-fix-arm64-v1-1-48e872844f25@kernel.org

Fixes: 9fc96c7c19df ("selftests: error out if kernel header files are not yet built")
Signed-off-by: John Hubbard <[email protected]>
Tested-by: Alexandre Ghiti <[email protected]>
Signed-off-by: Shuah Khan <[email protected]>

Merge tag 'riscv-for-linus-6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux

Pull RISC-V fixes from Palmer Dabbelt:

 - fix a formatting error in the hwprobe documentation

 - fix a spurious warning in the RISC-V PMU driver

 - fix memory detection on rv32 (problem does not manifest on any known
   system)

 - avoid parsing legacy parsing of I in ACPI ISA strings

* tag 'riscv-for-linus-6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/riscv/linux:
  RISC-V: Don't include Zicsr or Zifencei in I from ACPI
  riscv: mm: fix truncation warning on RV32
  perf: RISC-V: Remove PERF_HES_STOPPED flag checking in riscv_pmu_start()
  Documentation: RISC-V: hwprobe: Fix a formatting error

Merge tag 'pm-6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "These fix hibernation (after recent changes), frequency QoS and the
  sparc cpufreq driver.

  Specifics:

   - Unbreak the /sys/power/resume interface after recent changes (Azat
     Khuzhin).

   - Allow PM_QOS_DEFAULT_VALUE to be used with frequency QoS (Chungkai
     Yang).

   - Remove __init from cpufreq callbacks in the sparc driver, because
     they may be called after initialization too (Viresh Kumar)"

* tag 'pm-6.5-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq: sparc: Don't mark cpufreq callbacks with __init
  PM: QoS: Restore support for default value on frequency QoS
  PM: hibernate: Fix writing maj:min to /sys/power/resume