linux.git/tools/perf/tests, branch v6.4-rc2

Revert "perf build: Make BUILD_BPF_SKEL default, rename to NO_BPF_SKEL"

2023-05-06T21:07:37+00:00

This reverts commit a980755beb5aca9002e1c95ba519b83a44242b5b.

We need to better polish building with BPF skels, so revert back to
making it an experimental feature that has to be explicitely enabled
using BUILD_BPF_SKEL=1.

Signed-off-by: Arnaldo Carvalho de Melo

perf test record+probe_libc_inet_pton: Fix call chain match on x86_64

2023-05-03T14:02:21+00:00

The test case probe libc's inet_pton & backtrace it with ping fails with
Fedora 38 on x86_64.

Function getaddrinfo() does not show up in the call chain anymore:

  # ./perf script
  ping  1803 [000] 728.567146: probe_libc:inet_pton: (7f5275afc840)
            133840 __GI___inet_pton+0x0 (/usr/lib64/libc.so.6)
             27b4a __libc_start_call_main+0x7a (/usr/lib64/libc.so.6)
             27c0b __libc_start_main@@GLIBC_2.34+0x8b (/usr/lib64/libc.so.6)

  ping  1803 [000]   728.567184: probe_libc:inet_pton: (7f5275afc840)
            133840 __GI___inet_pton+0x0 (/usr/lib64/libc.so.6)
              493e main+0xcde (/usr/bin/ping)
             27b4a __libc_start_call_main+0x7a (/usr/lib64/libc.so.6)

  #

which causes the test case to fail. Remove function getaddrinfo()
from list of expected functions.

Output before:

  # ./perf test 'libc'
  91: probe libc's inet_pton & backtrace it with ping    : FAILED!
  #

Output after:

  # ./perf test 'libc'
  91: probe libc's inet_pton & backtrace it with ping    : Ok
  #

Signed-off-by: Thomas Richter 
Tested-by: Arnaldo Carvalho de Melo 
Cc: Heiko Carstens 
Cc: Sumanth Korikkar 
Cc: Sven Schnelle 
Cc: Vasily Gorbik 
Link: https://lore.kernel.org/r/20230503081255.3372986-1-tmricht@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo

perf test record+probe_libc_inet_pton: Fix call chain match on s390

2023-05-03T14:00:44+00:00

With Fedora 38 the perf test 86 probe libc's inet_pton fails on s390.
The call chain of the ping command changed.  The functions
text_to_binary_address() and gaih_inet() do not show up in the call
chain anymore.

Output before:

  # ./perf test -v 86
  86: probe libc's inet_pton & backtrace it with ping                 :
  --- start ---
  test child forked, pid 541050
  fgrep: warning: fgrep is obsolescent; using grep -F
  fgrep: warning: fgrep is obsolescent; using grep -F
  BFD: DWARF error: could not find variable specification at offset 0x22011
  ...

  ping 541078 [002] 348826.679581: probe_libc:inet_pton_1: (3ffad84b940)
  14b940 __GI___inet_pton+0x0 (/usr/lib64/libc.so.6)
  10e9c3 __GI_getaddrinfo+0xeb3 (inlined)
  4397 main+0x737 (/usr/bin/ping)
   FAIL: expected backtrace entry "gaih_inet.*\+0x[[:xdigit:]]\
          +[[:space:]]\(/usr/lib64/libc.so.6|inlined\)$"
          got "4397 main+0x737 (/usr/bin/ping)"
  test child finished with -1
  ---- end ----
  probe libc's inet_pton & backtrace it with ping: FAILED!
  #

Output after:

  # ./perf test -v 86
  86: probe libc's inet_pton & backtrace it with ping                 :
  --- start ---
  test child forked, pid 541098
  fgrep: warning: fgrep is obsolescent; using grep -F
  fgrep: warning: fgrep is obsolescent; using grep -F
  BFD: DWARF error: could not find variable specification at offset 0x309d1
  ...

  ping 541126 [006] 349309.099067: probe_libc:inet_pton_1: (3ffb7f4b940)
  14b940 __GI___inet_pton+0x0 (/usr/lib64/libc.so.6)
  10e9c3 __GI_getaddrinfo+0xeb3 (inlined)
  4397 main+0x737 (/usr/bin/ping)
  test child finished with 0
  ---- end ----
  probe libc's inet_pton & backtrace it with ping: Ok
  #

Signed-off-by: Thomas Richter 
Cc: Heiko Carstens 
Cc: Sumanth Korikkar 
Cc: Sven Schnelle 
Cc: Vasily Gorbik 
Link: https://lore.kernel.org/r/20230503081134.3372415-1-tmricht@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo

perf build: Fix unescaped # in perf build-test

2023-05-02T11:36:14+00:00

With the following bash and make versions:

  $ make --version
  GNU Make 4.2.1
  Built for aarch64-unknown-linux-gnu

  $ bash --version
  GNU bash, version 5.0.17(1)-release (aarch64-unknown-linux-gnu)

This error is encountered when running the build-test target:

  $ make -C tools/perf build-test
  tests/make:181: *** unterminated call to function 'shell': missing ')'.  Stop.
  make: *** [Makefile:103: build-test] Error 2

Fix it by escaping the # which was causing make to interpret the rest of
the line as a comment leaving the unclosed opening bracket.

Fixes: 56d5229471ee1634 ("tools build: Pass libbpf feature only if libbpf 1.0+")
Signed-off-by: James Clark 
Cc: Adrian Hunter 
Cc: Alexander Shishkin 
Cc: Ian Rogers 
Cc: Ingo Molnar 
Cc: Jiri Olsa 
Cc: Mark Rutland 
Cc: Namhyung Kim 
Cc: Peter Zijlstra 
Cc: bpf@vger.kernel.org
Link: https://lore.kernel.org/r/20230425104414.1723571-1-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo

perf cs-etm: Add separate decode paths for timeless and per-thread modes

2023-04-24T17:42:20+00:00

Timeless and per-thread are orthogonal concepts that are currently
treated as if they are the same (per-thread == timeless). This breaks
when you modify the command line or itrace options to something that the
current logic doesn't expect.

For example:

  # Force timeless with Z
  --itrace=Zi10i

  # Or inconsistent record options
  -e cs_etm/timestamp=1/ --per-thread

Adding Z for decoding in per-cpu mode is particularly bad because in
per-thread mode trace channel IDs are discarded and all assumed to be 0,
which would mix trace from different CPUs in per-cpu mode.

Although the results might not be perfect in all scenarios, if the user
requests no timestamps, it should still be possible to decode in either
mode. Especially if the relative times of samples in different processes
aren't interesting, quite a bit of space can be saved by turning off
timestamps in per-cpu mode.

Signed-off-by: James Clark 
Cc: Adrian Hunter 
Cc: Alexander Shishkin 
Cc: Denis Nikitin 
Cc: Ian Rogers 
Cc: Ingo Molnar 
Cc: Jiri Olsa 
Cc: John Garry 
Cc: Leo Yan 
Cc: Mark Rutland 
Cc: Mathieu Poirier 
Cc: Mike Leach 
Cc: Namhyung Kim 
Cc: Peter Zijlstra 
Cc: Suzuki Poulouse 
Cc: Will Deacon 
Cc: Yang Shi 
Cc: coresight@lists.linaro.org
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230424134748.228137-8-james.clark@arm.com
Signed-off-by: Arnaldo Carvalho de Melo

perf evsel: Introduce evsel__name_is() method to check if the evsel name is equal to a given string

2023-04-24T17:28:11+00:00

This makes the logic a bit clear by avoiding the !strcmp() pattern and
also a way to intercept the pointer if we need to do extra validation on
it or to do lazy setting of evsel->name via evsel__name(evsel).

Reviewed-by: "Liang, Kan" 
Cc: Adrian Hunter 
Cc: Ian Rogers 
Cc: Jiri Olsa 
Cc: Namhyung Kim 
Link: https://lore.kernel.org/lkml/ZEGLM8VehJbS0gP2@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo

perf test: Fix maps use after put

2023-04-20T11:23:52+00:00

Fix a use after put reference count issue. maps is copied from leader,
but the leader is put on line 79 and then maps is used to read the
reference count below - so a use after put, with the put of maps
happening within thread__put. Fix by reversing the order of puts so
that the leader is put last.

To explain the reference count checker, I wrote this up as a little
example here:
https://perf.wiki.kernel.org/index.php/Reference_Count_Checking

Note, the bug was introduced by the committer and wasn't present in
the original reference count patch set.

Committer notes:

Yes, the bug predated your patch and is detected by the reference count
checking you contributed.

This was just part of splitting up your series into smaller chunks, in
this case either we fix the problem detected while developing this
reference counting infrastructure before the patch introducing REFCNT_CHECKING
or fix it later after the merged infrastructure, when built with
EXTRA_CFLAGS="-DREFCNT_CHECKING=1" detects it when running 'perf test', which
is what this patch does.

Signed-off-by: Ian Rogers 
Cc: Adrian Hunter 
Cc: Alexander Shishkin 
Cc: Jiri Olsa 
Cc: Mark Rutland 
Cc: Namhyung Kim 
Cc: Peter Zijlstra 
Link: http://lore.kernel.org/lkml/20230420030430.489243-1-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo

perf build: Test the refcnt check build

2023-04-19T16:47:03+00:00

Make sure we test build the currently added REFCNT_CHECKING
infrastructure.

Signed-off-by: Arnaldo Carvalho de Melo

perf map: Add reference count checking

2023-04-19T15:57:53+00:00

There's no strict get/put policy with map that leads to leaks or use
after free. Reference count checking identifies correct pairing of gets
and puts.

Committer notes:

Extracted from a larger patch removing bits that were covered by the use
of pre-existing map__ accessors (e.g. maps__nr_maps()) and new ones
added (map__refcnt() and the maps__set_ ones) to reduce
RC_CHK_ACCESS(maps)-> source code pollution.

Signed-off-by: Ian Rogers 
Cc: Adrian Hunter 
Cc: Alexey Bayduraev 
Cc: Dmitriy Vyukov 
Cc: Jiri Olsa 
Cc: Namhyung Kim 
Cc: Riccardo Mancini 
Cc: Stephane Eranian 
Cc: Stephen Brennan 
Link: https://lore.kernel.org/lkml/20230407230405.2931830-6-irogers@google.com
Signed-off-by: Arnaldo Carvalho de Melo

perf map: Add set_ methods for map->{start,end,pgoff,pgoff,reloc,erange_warned,dso,map_ip,unmap_ip,priv}

2023-04-19T15:54:41+00:00

To have a way to intercept usage of the reference counted struct map.

Signed-off-by: Arnaldo Carvalho de Melo