Commits · 39c2ca43465e0f52ebba3ee96fd03436367c1880 · Kirill Smelkov / linux

29 Aug, 2019 20 commits

Merge tag 'perf-core-for-mingo-5.4-20190829' of... · 39c2ca43

Ingo Molnar authored Aug 29, 2019

Merge tag 'perf-core-for-mingo-5.4-20190829' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core

Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:

perf top:

Namhyung Kim:

- Decay all events in the evlist, we were decaying just the first event
in a group.

- Fix linking of histograms in different evsels in a event group with more
than two events.

With the two fixes above a command line such as:

# perf top -e '{cycles,instructions,cache-misses,cache-references}

Should work as expected, with four columns and with all of them being
decayed over time, i.e. less weight is given for older samples.

perf record:

Arnaldo Carvalho de Melo:

- Fix collection of build-ids when using setns() to get into namespaces,
which had been broken with the introduction of the extra thread to
react to PERF_RECORD_BPF_EVENT, i.e. to collect extra info for BPF
programs. We need to unshare(CLONE_FS) in that thread so that the
main one can do the setns(CLONE_NEWNS) when collectingthe build-ids.
Without that symbol resolution gets more difficult and potentially
misresolves symbols.

core:

Igor Lubashev:

- Further alignment in permission checking via capabilities to how the
kernel checks what tooling tries to do.

PowerPC:

Naveen N. Rao:

- Sync powerpc syscall.tbl, so that 'perf trace' gets the definitions
for recent syscalls.

libperf:

Jiri Olsa:

- Move the rest of the PERF_RECORD_ metadata struct definitions so that
we can use 'union perf_event'.

libtraceevent:

Steven Rostedt (VMware):

- Do not free tep->cmdlines in add_new_comm() on failure.

- Remove unneeded qsort and uses memmove instead
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>

39c2ca43

tools lib traceevent: Remove unneeded qsort and uses memmove instead · 301011ba

Steven Rostedt (VMware) authored Aug 28, 2019

While reading a trace data file that had 100,000s of tasks, the process
took an extremely long time. I profiled it down to add_new_comm(), which
was doing a qsort() call on an array that was pretty much already sorted
(all but the last element. qsort() isn't very efficient when dealing
with mostly sorted arrays, and this definitely showed its issues.

When adding a new task to the task list, instead of using qsort(), do
another bsearch() with a function that will find the element before
where the new task will be inserted in. Then simply shift the rest of
the array, and insert the task where it belongs.

Fixes: f7d82350 ("tools/events: Add files to create libtraceevent.a")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linux-trace-devel@vger.kernel.org
Link: http://lkml.kernel.org/r/20190828191820.127233764@goodmis.orgSigned-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

301011ba

tools lib traceevent: Do not free tep->cmdlines in add_new_comm() on failure · b0215e2d

Steven Rostedt (VMware) authored Aug 28, 2019

If the re-allocation of tep->cmdlines succeeds, then the previous
allocation of tep->cmdlines will be freed. If we later fail in
add_new_comm(), we must not free cmdlines, and also should assign
tep->cmdlines to the new allocation. Otherwise when freeing tep, the
tep->cmdlines will be pointing to garbage.

Fixes: a6d2a61a ("tools lib traceevent: Remove some die() calls")
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: linux-trace-devel@vger.kernel.org
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/20190828191819.970121417@goodmis.orgSigned-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

b0215e2d

perf evlist: Use unshare(CLONE_FS) in sb threads to let setns(CLONE_NEWNS) work · b397f846

Arnaldo Carvalho de Melo authored Aug 28, 2019

When we started using a thread to catch the PERF_RECORD_BPF_EVENT meta
data events to then ask the kernel for further info (BTF, etc) for BPF
programs shortly after they get loaded, we forgot to use
unshare(CLONE_FS) as was done in:

  868a8329 ("perf top: Support lookup of symbols in other mount namespaces.")

Do it so that we can enter the namespaces to read the build-ids at the
end of a 'perf record' session for the DSOs that had hits.

Before:

Starting a 'stress-ng --cpus 8' inside a container and then, outside the
container running:

  # perf record -a --namespaces sleep 5
  # perf buildid-list | grep stress-ng
  #

We would end up with a 'perf.data' file that had no entry in its
build-id table for the /usr/bin/stress-ng binary inside the container
that got tons of PERF_RECORD_SAMPLEs.

After:

  # perf buildid-list | grep stress-ng
  f2ed02c68341183a124b9b0f6e2e6c493c465b29 /usr/bin/stress-ng
  #

Then its just a matter of making sure that that binary debuginfo package
gets available in a place that 'perf report' will look at build-id keyed
ELF files, which, in my case, on a f30 notebook, was a matter of
installing the debuginfo file for the distro used in the container,
fedora 31:

  # rpm -ivh http://fedora.c3sl.ufpr.br/linux/development/31/Everything/x86_64/debug/tree/Packages/s/stress-ng-debuginfo-0.07.29-10.fc31.x86_64.rpm

Then, because perf currently looks for those debuginfo files (richer ELF
symtab) inside that namespace (look at the setns calls):

  openat(AT_FDCWD, "/proc/self/ns/mnt", O_RDONLY) = 137
  openat(AT_FDCWD, "/proc/13169/ns/mnt", O_RDONLY) = 139
  setns(139, CLONE_NEWNS)                 = 0
  stat("/usr/bin/stress-ng", {st_mode=S_IFREG|0755, st_size=3065416, ...}) = 0
  openat(AT_FDCWD, "/usr/bin/stress-ng", O_RDONLY) = 140
  fcntl(140, F_GETFD)                     = 0
  fstat(140, {st_mode=S_IFREG|0755, st_size=3065416, ...}) = 0
  mmap(NULL, 3065416, PROT_READ, MAP_PRIVATE, 140, 0) = 0x7ff2fdc5b000
  munmap(0x7ff2fdc5b000, 3065416)         = 0
  close(140)                              = 0
  stat("stress-ng-0.07.29-10.fc31.x86_64.debug", 0x7fff45d71260) = -1 ENOENT (No such file or directory)
  stat("/usr/bin/stress-ng-0.07.29-10.fc31.x86_64.debug", 0x7fff45d71260) = -1 ENOENT (No such file or directory)
  stat("/usr/bin/.debug/stress-ng-0.07.29-10.fc31.x86_64.debug", 0x7fff45d71260) = -1 ENOENT (No such file or directory)
  stat("/usr/lib/debug/usr/bin/stress-ng-0.07.29-10.fc31.x86_64.debug", 0x7fff45d71260) = -1 ENOENT (No such file or directory)
  stat("/root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29", 0x7fff45d711e0) = -1 ENOENT (No such file or directory)

To only then go back to the "host" namespace to look just in the users's
~/.debug cache:

  setns(137, CLONE_NEWNS)                 = 0
  chdir("/root")                          = 0
  close(137)                              = 0
  close(139)                              = 0
  stat("/root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29/elf", 0x7fff45d732e0) = -1 ENOENT (No such file or directory)

It continues to fail to resolve symbols:

  # perf report | grep stress-ng | head -5
     9.50%  stress-ng-cpu    stress-ng    [.] 0x0000000000021ac1
     8.58%  stress-ng-cpu    stress-ng    [.] 0x0000000000021ab4
     8.51%  stress-ng-cpu    stress-ng    [.] 0x0000000000021489
     7.17%  stress-ng-cpu    stress-ng    [.] 0x00000000000219b6
     3.93%  stress-ng-cpu    stress-ng    [.] 0x0000000000021478
  #

To overcome that we use:

  # perf buildid-cache -v --add /usr/lib/debug/usr/bin/stress-ng-0.07.29-10.fc31.x86_64.debug
  Adding f2ed02c68341183a124b9b0f6e2e6c493c465b29 /usr/lib/debug/usr/bin/stress-ng-0.07.29-10.fc31.x86_64.debug: Ok
  #
  # ls -la /root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29/elf
  -rw-r--r--. 3 root root 2401184 Jul 27 07:03 /root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29/elf
  # file /root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29/elf
  /root/.debug/.build-id/f2/ed02c68341183a124b9b0f6e2e6c493c465b29/elf: ELF 64-bit LSB shared object, x86-64, version 1 (SYSV), dynamically linked, interpreter \004, BuildID[sha1]=f2ed02c68341183a124b9b0f6e2e6c493c465b29, for GNU/Linux 3.2.0, with debug_info, not stripped, too many notes (256)
  #

Now it finally works:

  # perf report | grep stress-ng | head -5
    23.59%  stress-ng-cpu    stress-ng    [.] ackermann
    23.33%  stress-ng-cpu    stress-ng    [.] is_prime
    17.36%  stress-ng-cpu    stress-ng    [.] stress_cpu_sieve
     6.08%  stress-ng-cpu    stress-ng    [.] stress_cpu_correlate
     3.55%  stress-ng-cpu    stress-ng    [.] queens_try
  #

I'll make sure that it looks for the build-id keyed files in both the
"host" namespace (the namespace the user running 'perf record' was a the
time of the recording) and in the container namespace, as it shouldn't
matter where a content based key lookup finds the ELF file to use in
resolving symbols, etc.
Reported-by: Karl Rister <krister@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Krister Johansen <kjlx@templeofstupid.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Stanislav Fomichev <sdf@google.com>
Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
Fixes: 657ee553 ("perf evlist: Introduce side band thread")
Link: https://lkml.kernel.org/n/tip-g79k0jz41adiaeuqud742t2l@git.kernel.orgSigned-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>