提交 · 4598a0a6d22fadfb7b37f2b44ee7fdcb24632fcf · gsplhtlxg / clone-Linux

02 10月, 2014 1 次提交

perf symbols: Improve DSO long names lookup speed with rbtree · 4598a0a6

由 Waiman Long 提交于 9月 30, 2014

With workload that spawns and destroys many threads and processes, it
was found that perf-mem could took a long time to post-process the perf
data after the target workload had completed its operation.

The performance bottleneck was found to be the lookup and insertion of
the new DSO structures (thousands of them in this case).

In a dual-socket Ivy-Bridge E7-4890 v2 machine (30-core, 60-thread), the
perf profile below shows what perf was doing after the profiled AIM7
shared workload completed:

-     83.94%  perf  libc-2.11.3.so     [.] __strcmp_sse42
   - __strcmp_sse42
      - 99.82% map__new
           machine__process_mmap_event
           perf_session_deliver_event
           perf_session__process_event
           __perf_session__process_events
           cmd_record
           cmd_mem
           run_builtin
           main
           __libc_start_main
-     13.17%  perf  perf               [.] __dsos__findnew
     __dsos__findnew
     map__new
     machine__process_mmap_event
     perf_session_deliver_event
     perf_session__process_event
     __perf_session__process_events
     cmd_record
     cmd_mem
     run_builtin
     main
     __libc_start_main

So about 97% of CPU times were spent in the map__new() function trying
to insert new DSO entry into the DSO linked list. The whole
post-processing step took about 9 minutes.

The DSO structures are currently searched linearly. So the total
processing time will be proportional to n^2.

To overcome this performance problem, the DSO code is modified to also
put the DSO structures in a RB tree sorted by its long name in
additional to being in a simple linked list. With this change, the
processing time will become proportional to n*log(n) which will be much
quicker for large n. However, the short name will still be searched
using the old linear searching method.  With that patch in place, the
same perf-mem post-processing step took less than 30 seconds to
complete.
Signed-off-by: NWaiman Long <Waiman.Long@hp.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Douglas Hatch <doug.hatch@hp.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Scott J Norton <scott.norton@hp.com>
Link: http://lkml.kernel.org/r/1412098575-27863-3-git-send-email-Waiman.Long@hp.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

4598a0a6

30 9月, 2014 5 次提交

perf symbols: Encapsulate dsos list head into struct dsos · 8fa7d87f

由 Waiman Long 提交于 9月 29, 2014

This is a precursor patch to enable long name searching of DSOs using
a rbtree.

In this patch, a new dsos structure is created which contains only a
list head structure for the moment.

The new dsos structure is used, in turn, in the machine structure for
the user_dsos and kernel_dsos fields.

Only the following 3 dsos functions are modified to accept the new dsos
structure parameter instead of list_head:

 - dsos__add()
 - dsos__find()
 - __dsos__findnew()
Signed-off-by: NWaiman Long <Waiman.Long@hp.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Don Zickus <dzickus@redhat.com>
Cc: Douglas Hatch <doug.hatch@hp.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Scott J Norton <scott.norton@hp.com>
Link: http://lkml.kernel.org/r/1412021249-19201-2-git-send-email-Waiman.Long@hp.com
[ Move struct dsos to dso.h to reduce the dso methods depends on machine.h ]
Signed-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

8fa7d87f

perf bench futex: Sanitize -q option in requeue · e19685ed

由 Davidlohr Bueso 提交于 9月 29, 2014

When given the number of threads to requeue at once by user input,
there's always the risk of this value being larger than the total number
of threads.  This doesn't make any sense, and the kernel can easily deal
with such sort of situations, hence no big deal. We should however
prevent bogus output such as:

./perf bench --repeat 2 futex requeue -q 10
Run summary [PID 22210]: Requeuing 4 threads (from [private] 0x99ef3c to 0x99ef38), 10 at a time.

[Run 1]: Requeued 10 of 4 threads in 0.0040 ms
[Run 2]: Requeued 10 of 4 threads in 0.0030 ms
Requeued 10 of 4 threads in 0.0035 ms (+-14.29%)
Signed-off-by: NDavidlohr Bueso <dbueso@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Link: http://lkml.kernel.org/r/1412008868-22328-2-git-send-email-dave@stgolabs.netSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

e19685ed

perf bench futex: Support operations for shared futexes · 86c87e13

由 Davidlohr Bueso 提交于 9月 29, 2014

Unlike futex-hash, requeuing and wakeup benchmarks do not support shared
futexes, limiting the usefulness of the programs. Correct this, and
allow using the local -S parameter. The default remains using private
futexes.
Signed-off-by: NDavidlohr Bueso <dbueso@suse.de>
Cc: Davidlohr Bueso <dbueso@suse.de>
Link: http://lkml.kernel.org/r/1412008868-22328-1-git-send-email-dave@stgolabs.netSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

86c87e13

perf trace: Fix mmap return address truncation to 32-bit · 2c82c3ad

由 Chang Hyun Park 提交于 9月 26, 2014

Using 'perf trace' for mmap is truncating return values by stripping the
top 32 bits, actually printing only the lower 32 bits.

This was because the ret value was of an 'int' type and not a 'long'
type.

  The Problem:

  991258501.244 ( 0.004 ms): mmap(len: 40001536, prot: READ|WRITE, flags: PRIVATE|ANONYMOUS, fd: -1) = 0x56691000
  991258501.257 ( 0.000 ms): minfault [_int_malloc+0x1038] => //anon@0x7fa056691008 //(d.)

The first line shows an mmap, which succeeds and returns 0x56691000.

However the next line shows a memory access to that virtual memory area,
specifically to 0x7fa056691008. The upper 32 bit is lost due to the
problem mentioned above, and thus mmap's return value didn't have the
upper 0x7fa0.

Tested on 3.17-rc5 from the linus's tree, and the HEAD of tip/master
Signed-off-by: NChang Hyun Park <heartinpiece@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1411736041-8017-1-git-send-email-heartinpiece@gmail.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

2c82c3ad

perf tools: Refactor unit and scale function parameters · 46441bdc

由 Matt Fleming 提交于 9月 24, 2014

Passing pointers to alias modifiers 'unit' and 'scale' isn't very
future-proof since if we add more modifiers to the list we'll end up
passing more arguments.

Instead wrap everything up in a struct perf_pmu_info, which can easily
be expanded when additional alias modifiers are necessary in the future.
Signed-off-by: NMatt Fleming <matt.fleming@intel.com>
Acked-by: NJiri Olsa <jolsa@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1411567455-31264-3-git-send-email-matt@console-pimps.orgSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

46441bdc

27 9月, 2014 1 次提交

Merge tag 'perf-core-for-mingo' of... · 07394b5f

由 Ingo Molnar 提交于 9月 27, 2014

Merge tag 'perf-core-for-mingo' 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:

User visible changes:

  o Restore "--callchain graph" output, broken in recent cset to end
    up being the same as "fractal" (Namhyung Kim)

  o Allow profiling when kptr_restrict == 1 for non root users,
    kernel samples will just remain unresolved (Andi Kleen)

  o Allow configuring default options for callchains in config file (Namhyung Kim)

  o Fix line number in the config file error message (Jiri Olsa)

  o Fix --per-core on multi socket systems (Andi Kleen)

Cleanups:

  o Use ACCESS_ONCE() instead of volatile cast. (Pranith Kumar)

  o Modify error code for when perf_session__new() fails (Taeung Song)
Signed-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: NIngo Molnar <mingo@kernel.org>

07394b5f

26 9月, 2014 25 次提交

perf tools: Fix line number in the config file error message · 49757c9c

由 Jiri Olsa 提交于 9月 23, 2014

If we fail to parse the config file within the callback function,
the line number counter 'could be' already on the next line.

This results in wrong line number report like:

  $ cat ~/.perfconfig
  [call-graph]
          sort-key = krava
  $ perf record ls
  Fatal: bad config file line 3 in /home/jolsa/.perfconfig

Fixing this by saving the current line number for this case.
Signed-off-by: NJiri Olsa <jolsa@kernel.org>
Acked-by: NNamhyung Kim <namhyung@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Milian Wolff <mail@milianw.de>
Cc: Namhyung Kim <namhyung.kim@lge.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20140923115656.GC2979@krava.brq.redhat.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>

49757c9c

perf tools: Convert {record,top}.call-graph option to call-graph.record-mode · 5a2e5e85