Skip to content

K
Kernel

openeuler / Kernel
1 年多 前同步成功

通知 8
Star 0
Fork 0
K
Kernel
切换分支/标签
  1. 31 10月, 2017 1 次提交
  3. 27 10月, 2017 17 次提交
  5. 26 10月, 2017 1 次提交
    • I
      Merge tag 'perf-core-for-mingo-4.15-20171025' of... · 57646b6f
      由 Ingo Molnar 提交于 
      Merge tag 'perf-core-for-mingo-4.15-20171025' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core

Pull perf/core inline improvements from Arnaldo Carvalho de Melo:

From Milian's cover letter: (Milian Wolff)

"This series of patches completely reworks the way inline frames are
 handled.  Instead of querying for the inline nodes on-demand in the
 individual tools, we now create proper callchain nodes for inlined
 frames. The advantages this approach brings are numerous:

 - Less duplicated code in the individual browser

 - Aggregated cost for inlined frames for the --children top-down list

 - Various bug fixes that arose from querying for a srcline/symbol based on
   the IP of a sample, which will always point to the last inlined frame
   instead of the corresponding non-inlined frame

 - Overall much better support for visualizing cost for heavily-inlined C++
   code, which simply was confusing and unreliably before

 - srcline honors the global setting as to whether full paths or basenames
   should be shown

 - Caches for inlined frames and srcline information, which allow us to
   enable inline frame handling by default"
Signed-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: NIngo Molnar <mingo@kernel.org>
      57646b6f
  7. 25 10月, 2017 5 次提交
    • M
      perf util: Enable handling of inlined frames by default · d8a88dd2
      由 Milian Wolff 提交于 
      Now that we have caches in place to speed up the process of finding
inlined frames and srcline information repeatedly, we can enable this
useful option by default.
Suggested-by: NIngo Molnar <mingo@kernel.org>
Signed-off-by: NMilian Wolff <milian.wolff@kdab.com>
Reviewed-by: NAndi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20171019113836.5548-6-milian.wolff@kdab.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
      d8a88dd2
    • M
      perf report: Use srcline from callchain for hist entries · 1fb7d06a
      由 Milian Wolff 提交于 
      This also removes the symbol name from the srcline column, more on this
below.

This ensures we use the correct srcline, which could originate from a
potentially inlined function. The hist entries used to query for the
srcline based purely on the IP, which leads to wrong results for inlined
entries.

Before:

~~~~~
  perf report --inline -s srcline -g none --stdio
  ...
  # Children      Self  Source:Line
  # ........  ........  ..................................................................................................................................
  #
      94.23%     0.00%  __libc_start_main+18446603487898210537
      94.23%     0.00%  _start+41
      44.58%     0.00%  main+100
      44.58%     0.00%  std::_Norm_helper<true>::_S_do_it<double>+100
      44.58%     0.00%  std::__complex_abs+100
      44.58%     0.00%  std::abs<double>+100
      44.58%     0.00%  std::norm<double>+100
      36.01%     0.00%  hypot+18446603487892193300
      25.81%     0.00%  main+41
      25.81%     0.00%  std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator()+41
      25.81%     0.00%  std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >+41
      25.75%    25.75%  random.h:143
      18.39%     0.00%  main+57
      18.39%     0.00%  std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator()+57
      18.39%     0.00%  std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >+57
      13.80%    13.80%  random.tcc:3330
       5.64%     0.00%  ??:0
       4.13%     4.13%  __hypot_finite+163
       4.13%     0.00%  __hypot_finite+18446603487892193443
...
~~~~~

After:

~~~~~
  perf report --inline -s srcline -g none --stdio
  ...
  # Children      Self  Source:Line
  # ........  ........  ...........................................
  #
      94.30%     1.19%  main.cpp:39
      94.23%     0.00%  __libc_start_main+18446603487898210537
      94.23%     0.00%  _start+41
      48.44%     1.70%  random.h:1823
      48.44%     0.00%  random.h:1814
      46.74%     2.53%  random.h:185
      44.68%     0.10%  complex:589
      44.68%     0.00%  complex:597
      44.68%     0.00%  complex:654
      44.68%     0.00%  complex:664
      40.61%    13.80%  random.tcc:3330
      36.01%     0.00%  hypot+18446603487892193300
      26.81%     0.00%  random.h:151
      26.81%     0.00%  random.h:332
      25.75%    25.75%  random.h:143
       5.64%     0.00%  ??:0
       4.13%     4.13%  __hypot_finite+163
       4.13%     0.00%  __hypot_finite+18446603487892193443
...
~~~~~

Note that this change removes the symbol from the source:line hist
column. If this information is desired, users should explicitly query
for it if needed. I.e. run this command instead:

~~~~~
  perf report --inline -s sym,srcline -g none --stdio
  ...
  # To display the perf.data header info, please use --header/--header-only options.
  #
  #
  # Total Lost Samples: 0
  #
  # Samples: 1K of event 'cycles:uppp'
  # Event count (approx.): 1381229476
  #
  # Children      Self  Symbol                                                                                                                               Source:Line
  # ........  ........  ...................................................................................................................................  ...........................................
  #
      94.30%     1.19%  [.] main                                                                                                                             main.cpp:39
      94.23%     0.00%  [.] __libc_start_main                                                                                                                __libc_start_main+18446603487898210537
      94.23%     0.00%  [.] _start                                                                                                                           _start+41
      48.44%     0.00%  [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)  random.h:1814
      48.44%     0.00%  [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)  random.h:1823
      46.74%     0.00%  [.] std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined)  random.h:185
      44.68%     0.00%  [.] std::_Norm_helper<true>::_S_do_it<double> (inlined)                                                                              complex:654
      44.68%     0.00%  [.] std::__complex_abs (inlined)                                                                                                     complex:589
      44.68%     0.00%  [.] std::abs<double> (inlined)                                                                                                       complex:597
      44.68%     0.00%  [.] std::norm<double> (inlined)                                                                                                      complex:664
      39.80%    13.59%  [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >               random.tcc:3330
      36.01%     0.00%  [.] hypot                                                                                                                            hypot+18446603487892193300
      26.81%     0.00%  [.] std::__detail::__mod<unsigned long, 2147483647ul, 16807ul, 0ul> (inlined)                                                        random.h:151
      26.81%     0.00%  [.] std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>::operator() (inlined)                                 random.h:332
      25.75%     0.00%  [.] std::__detail::_Mod<unsigned long, 2147483647ul, 16807ul, 0ul, true, true>::__calc (inlined)                                     random.h:143
      25.19%    25.19%  [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >               random.h:143
       4.13%     4.13%  [.] __hypot_finite                                                                                                                   __hypot_finite+163
       4.13%     0.00%  [.] __hypot_finite                                                                                                                   __hypot_finite+18446603487892193443
...
~~~~~

Compared to the old behavior, this reduces duplication in the output.
Before we used to print the symbol name in the srcline column even
when the sym column was explicitly requested. I.e. the output was:

~~~~~
  perf report --inline -s sym,srcline -g none --stdio
  ...
  # To display the perf.data header info, please use --header/--header-only options.
  #
  #
  # Total Lost Samples: 0
  #
  # Samples: 1K of event 'cycles:uppp'
  # Event count (approx.): 1381229476
  #
  # Children      Self  Symbol                                                                                                                               Source:Line
  # ........  ........  ...................................................................................................................................  ..................................................................................................................................
  #
      94.23%     0.00%  [.] __libc_start_main                                                                                                                __libc_start_main+18446603487898210537
      94.23%     0.00%  [.] _start                                                                                                                           _start+41
      44.58%     0.00%  [.] main                                                                                                                             main+100
      44.58%     0.00%  [.] std::_Norm_helper<true>::_S_do_it<double> (inlined)                                                                              std::_Norm_helper<true>::_S_do_it<double>+100
      44.58%     0.00%  [.] std::__complex_abs (inlined)                                                                                                     std::__complex_abs+100
      44.58%     0.00%  [.] std::abs<double> (inlined)                                                                                                       std::abs<double>+100
      44.58%     0.00%  [.] std::norm<double> (inlined)                                                                                                      std::norm<double>+100
      36.01%     0.00%  [.] hypot                                                                                                                            hypot+18446603487892193300
      25.81%     0.00%  [.] main                                                                                                                             main+41
      25.81%     0.00%  [.] std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined)  std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator()+41
      25.81%     0.00%  [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)  std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >+41
      25.69%    25.69%  [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >               random.h:143
      18.39%     0.00%  [.] main                                                                                                                             main+57
      18.39%     0.00%  [.] std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined)  std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator()+57
      18.39%     0.00%  [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)  std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >+57
      13.80%    13.80%  [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> >               random.tcc:3330
       4.13%     4.13%  [.] __hypot_finite                                                                                                                   __hypot_finite+163
       4.13%     0.00%  [.] __hypot_finite                                                                                                                   __hypot_finite+18446603487892193443
...
~~~~~
Signed-off-by: NMilian Wolff <milian.wolff@kdab.com>
Reviewed-by: NAndi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20171019113836.5548-5-milian.wolff@kdab.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
      1fb7d06a
    • M
      perf report: Cache srclines for callchain nodes · 21ac9d54
      由 Milian Wolff 提交于 
      On one hand this ensures that the memory is properly freed when the DSO
gets freed. On the other hand this significantly speeds up the
processing of the callchain nodes when lots of srclines are requested.
For one of my data files e.g.:

Before:

 Performance counter stats for 'perf report -s srcline -g srcline --stdio':

      52496.495043      task-clock (msec)         #    0.999 CPUs utilized
               634      context-switches          #    0.012 K/sec
                 2      cpu-migrations            #    0.000 K/sec
           191,561      page-faults               #    0.004 M/sec
   165,074,498,235      cycles                    #    3.144 GHz
   334,170,832,408      instructions              #    2.02  insn per cycle
    90,220,029,745      branches                  # 1718.591 M/sec
       654,525,177      branch-misses             #    0.73% of all branches

      52.533273822 seconds time elapsedProcessed 236605 events and lost 40 chunks!

After:

 Performance counter stats for 'perf report -s srcline -g srcline --stdio':

      22606.323706      task-clock (msec)         #    1.000 CPUs utilized
                31      context-switches          #    0.001 K/sec
                 0      cpu-migrations            #    0.000 K/sec
           185,471      page-faults               #    0.008 M/sec
    71,188,113,681      cycles                    #    3.149 GHz
   133,204,943,083      instructions              #    1.87  insn per cycle
    34,886,384,979      branches                  # 1543.214 M/sec
       278,214,495      branch-misses             #    0.80% of all branches

      22.609857253 seconds time elapsed

Note that the difference is only this large when `--inline` is not
passed. In such situations, we would use the inliner cache and thus do
not run this code path that often.

I think that this cache should actually be used in other places, too.
When looking at the valgrind leak report for perf report, we see tons of
srclines being leaked, most notably from calls to
hist_entry__get_srcline. The problem is that get_srcline has many
different formatting options (show_sym, show_addr, potentially even
unwind_inlines when calling __get_srcline directly). As such, the
srcline cannot easily be cached for all calls, or we'd have to add
caches for all formatting combinations (6 so far). An alternative would
be to remove the formatting options and handle that on a different level
- i.e. print the sym/addr on demand wherever we actually output
something. And the unwind_inlines could be moved into a separate
function that does not return the srcline.
Signed-off-by: NMilian Wolff <milian.wolff@kdab.com>
Reviewed-by: NAndi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20171019113836.5548-4-milian.wolff@kdab.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
      21ac9d54
    • M
      perf report: Cache failed lookups of inlined frames · b38775cf
      由 Milian Wolff 提交于 
      When no inlined frames could be found for a given address, we did not
store this information anywhere. That means we potentially do the costly
inliner lookup repeatedly for cases where we know it can never succeed.

This patch makes dso__parse_addr_inlines always return a valid
inline_node. It will be empty when no inliners are found. This enables
us to cache the empty list in the DSO, thereby improving the performance
when many addresses fail to find the inliners.

For my trivial example, the performance impact is already quite
significant:

Before:

~~~~~
 Performance counter stats for 'perf report --stdio --inline -g srcline -s srcline' (5 runs):

        594.804032      task-clock (msec)         #    0.998 CPUs utilized            ( +-  0.07% )
                53      context-switches          #    0.089 K/sec                    ( +-  4.09% )
                 0      cpu-migrations            #    0.000 K/sec                    ( +-100.00% )
             5,687      page-faults               #    0.010 M/sec                    ( +-  0.02% )
     2,300,918,213      cycles                    #    3.868 GHz                      ( +-  0.09% )
     4,395,839,080      instructions              #    1.91  insn per cycle           ( +-  0.00% )
       939,177,205      branches                  # 1578.969 M/sec                    ( +-  0.00% )
        11,824,633      branch-misses             #    1.26% of all branches          ( +-  0.10% )

       0.596246531 seconds time elapsed                                          ( +-  0.07% )
~~~~~

After:

~~~~~
 Performance counter stats for 'perf report --stdio --inline -g srcline -s srcline' (5 runs):

        113.111405      task-clock (msec)         #    0.990 CPUs utilized            ( +-  0.89% )
                29      context-switches          #    0.255 K/sec                    ( +- 54.25% )
                 0      cpu-migrations            #    0.000 K/sec
             5,380      page-faults               #    0.048 M/sec                    ( +-  0.01% )
       432,378,779      cycles                    #    3.823 GHz                      ( +-  0.75% )
       670,057,633      instructions              #    1.55  insn per cycle           ( +-  0.01% )
       141,001,247      branches                  # 1246.570 M/sec                    ( +-  0.01% )
         2,346,845      branch-misses             #    1.66% of all branches          ( +-  0.19% )

       0.114222393 seconds time elapsed                                          ( +-  1.19% )
~~~~~
Signed-off-by: NMilian Wolff <milian.wolff@kdab.com>
Reviewed-by: NAndi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20171019113836.5548-3-milian.wolff@kdab.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
      b38775cf
    • M
      perf report: Properly handle branch count in match_chain() · bf36eb5c
      由 Milian Wolff 提交于 
      Some of the code paths I introduced before returned too early without
running the code to handle a node's branch count.  By refactoring
match_chain to only have one exit point, this can be remedied.
Signed-off-by: NMilian Wolff <milian.wolff@kdab.com>
Acked-by: NNamhyung Kim <namhyung@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/1707691.qaJ269GSZW@agathebauer
Link: http://lkml.kernel.org/r/20171018185350.14893-2-milian.wolff@kdab.comSigned-off-by: NArnaldo Carvalho de Melo <acme@redhat.com>
      bf36eb5c
  9. 24 10月, 2017 16 次提交