- 25 Oct, 2017 2 commits
-
-
Milian Wolff authored
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:Milian Wolff <milian.wolff@kdab.com> Reviewed-by:
Andi 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:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Milian Wolff authored
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:
Namhyung 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:
-
- 24 Oct, 2017 11 commits
-
-
Milian Wolff authored
Similar to the callstack frame matching, we also have to compare the symbol name when sorting hist entries. The reason is twofold: On one hand, multiple inlined functions will use the same symbol start/end values of the parent, non-inlined symbol. As such, all of these symbols often end up missing from top-level report, as they get merged with the non-inlined frame. On the other hand, multiple different functions may end up inlining the same function, and we need to aggregate these values properly. Before: ~~~~~ perf report --stdio --inline -g none # Children Self Command Shared Object Symbol # ........ ........ ............ ............. ................................... # 100.00% 39.69% cpp-inlining cpp-inlining [.] main 100.00% 0.00% cpp-inlining cpp-inlining [.] _start 100.00% 0.00% cpp-inlining libc-2.25.so [.] __libc_start_main 97.03% 0.00% cpp-inlining cpp-inlining [.] std::norm<double> (inlined) 59.53% 4.26% cpp-inlining libm-2.25.so [.] hypot 55.21% 55.08% cpp-inlining libm-2.25.so [.] __hypot_finite 0.52% 0.52% cpp-inlining libm-2.25.so [.] cabs ~~~~~ After: ~~~~~ perf report --stdio --inline -g none # Children Self Command Shared Object Symbol # ........ ........ ............ ............. ................................................................................................................................... # 100.00% 39.69% cpp-inlining cpp-inlining [.] main 100.00% 0.00% cpp-inlining cpp-inlining [.] _start 100.00% 0.00% cpp-inlining libc-2.25.so [.] __libc_start_main 62.57% 0.00% cpp-inlining cpp-inlining [.] std::_Norm_helper<true>::_S_do_it<double> (inlined) 62.57% 0.00% cpp-inlining cpp-inlining [.] std::__complex_abs (inlined) 62.57% 0.00% cpp-inlining cpp-inlining [.] std::abs<double> (inlined) 62.57% 0.00% cpp-inlining cpp-inlining [.] std::norm<double> (inlined) 59.53% 4.26% cpp-inlining libm-2.25.so [.] hypot 55.21% 55.08% cpp-inlining libm-2.25.so [.] __hypot_finite 34.46% 0.00% cpp-inlining cpp-inlining [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined) 32.39% 0.00% cpp-inlining cpp-inlining [.] std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined) 32.39% 0.00% cpp-inlining cpp-inlining [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined) 12.29% 0.00% cpp-inlining cpp-inlining [.] std::__detail::_Mod<unsigned long, 2147483647ul, 16807ul, 0ul, true, true>::__calc (inlined) 12.29% 0.00% cpp-inlining cpp-inlining [.] std::__detail::__mod<unsigned long, 2147483647ul, 16807ul, 0ul> (inlined) 12.29% 0.00% cpp-inlining cpp-inlining [.] std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>::operator() (inlined) 0.52% 0.52% cpp-inlining libm-2.25.so [.] cabs ~~~~~ Signed-off-by:Jiri Olsa <jolsa@redhat.com> Reviewed-by:
-
Milian Wolff authored
The fake symbols we create for inlined frames will represent different functions but can use the symbol start address. This leads to issues when different inline branches all lead to the same function. Before: ~~~~~ $ perf report -s sym -i perf.inlining.data --inline --stdio -g function ... --38.86%--_start __libc_start_main main | --37.57%--std::norm<double> (inlined) std::_Norm_helper<true>::_S_do_it<double> (inlined) | --36.36%--std::abs<double> (inlined) std::__complex_abs (inlined) | --12.24%--std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>::operator() (inlined) std::__detail::__mod<unsigned long, 2147483647ul, 16807ul, 0ul> (inlined) std::__detail::_Mod<unsigned long, 2147483647ul, 16807ul, 0ul, true, true>::__calc (inlined) ~~~~~ Note that this backtrace representation is completely bogus. Complex abs does not call the linear congruential engine! It is just a side-effect of a longer inlined stack being appended to a shorter, different inlined stack, both of which originate in the same function (main). This patch fixes the issue: ~~~~~ $ perf report -s sym -i perf.inlining.data --inline --stdio -g function ... --38.86%--_start __libc_start_main main | |--35.59%--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> > (inlined) | | | --34.37%--std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined) | std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined) | | | --12.24%--std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>::operator() (inlined) | std::__detail::__mod<unsigned long, 2147483647ul, 16807ul, 0ul> (inlined) | std::__detail::_Mod<unsigned long, 2147483647ul, 16807ul, 0ul, true, true>::__calc (inlined) | --1.99%--std::norm<double> (inlined) std::_Norm_helper<true>::_S_do_it<double> (inlined) std::abs<double> (inlined) std::__complex_abs (inlined) ~~~~~ Signed-off-by: -
Milian Wolff authored
Instead of showing the (repeated) DSO name of the non-inlined frame, we now show the "(inlined)" suffix instead. Before: 214f7 __hypot_finite (/usr/lib/libm-2.25.so) ace3 hypot (/usr/lib/libm-2.25.so) a4a std::__complex_abs (/home/milian/projects/src/perf-tests/inlining) a4a std::abs<double> (/home/milian/projects/src/perf-tests/inlining) a4a std::_Norm_helper<true>::_S_do_it<double> (/home/milian/projects/src/perf-tests/inlining) a4a std::norm<double> (/home/milian/projects/src/perf-tests/inlining) a4a main (/home/milian/projects/src/perf-tests/inlining) 20510 __libc_start_main (/usr/lib/libc-2.25.so) bd9 _start (/home/milian/projects/src/perf-tests/inlining) After: 214f7 __hypot_finite (/usr/lib/libm-2.25.so) ace3 hypot (/usr/lib/libm-2.25.so) a4a std::__complex_abs (inlined) a4a std::abs<double> (inlined) a4a std::_Norm_helper<true>::_S_do_it<double> (inlined) a4a std::norm<double> (inlined) a4a main (/home/milian/projects/src/perf-tests/inlining) 20510 __libc_start_main (/usr/lib/libc-2.25.so) bd9 _start (/home/milian/projects/src/perf-tests/inlining) Signed-off-by: -
Milian Wolff authored
The original patch that introduced inline frame output in the various browsers used this suffix already. The new centralized approach that uses fake symbols for inlined frames was missing this approach so far. Instead of changing the symbol name itself, we only print the suffix where needed. This allows us to efficiently lookup the symbol for a given name without first having to append the suffix before the lookup. Signed-off-by:
-
Milian Wolff authored
When a callchain entry has no srcline available, we ended up comparing the instruction pointer. I consider this to be not too useful. Rather, I think we should group the entries by function name, which this patch adds. For people who want to split the data on the IP boundary, using `-g address` is the correct choice. Before: ~~~~~ 100.00% 38.86% [.] main | |--61.14%--main inlining.cpp:14 | std::norm<double> complex:664 | std::_Norm_helper<true>::_S_do_it<double> complex:654 | std::abs<double> complex:597 | std::__complex_abs complex:589 | | | |--56.03%--hypot | | | | | |--8.45%--__hypot_finite | | | | | |--7.62%--__hypot_finite | | | | | |--2.29%--__hypot_finite | | | | | |--2.24%--__hypot_finite | | | | | |--2.06%--__hypot_finite | | | | | |--1.81%--__hypot_finite ... ~~~~~ After: ~~~~~ 100.00% 38.86% [.] main | |--61.14%--main inlining.cpp:14 | std::norm<double> complex:664 | std::_Norm_helper<true>::_S_do_it<double> complex:654 | std::abs<double> complex:597 | std::__complex_abs complex:589 | | | |--60.29%--hypot | | | | | --56.03%--__hypot_finite | | | --0.85%--cabs ~~~~~ Signed-off-by: -
Milian Wolff authored
The inline_node structs are maintained by the new dso->inlines tree. This in turn keeps ownership of the fake symbols and srcline string representing an inline frame. This tree is sorted by address to allow quick lookups. All other entries of the symbol beside the function name are unused for inline frames. The advantage of this approach is that all existing users of the callchain API can now transparently display inlined frames without having to patch their code. Signed-off-by:
-
Milian Wolff authored
This is a preparation for the creation of real callchain entries for inlined frames. The rest of the perf code uses the srcline string. As such, using that also for the srcline API allows us to simplify some of the upcoming code. Most notably, it will allow us to cache the srcline for a given inline node and reuse it for different callchain entries. Signed-off-by:
-
Milian Wolff authored
This is a requirement to create real callchain entries for inlined frames. Since the list of inlines usually contains the target symbol too, i.e. the location where the frames get inlined to, we alias that symbol and reuse it as-is is. This ensures that other dependent functionality keeps working, most notably annotation of the target frames. For all other entries in the inline_list, a fake symbol is created. These are marked by new 'inlined' member which is set to true. Only those symbols are managed by the inline_list and get freed when the inline_list is deleted from within inline_node__delete. Signed-off-by:
-
Milian Wolff authored
This is mostly a preparation to enable the creation of full callchain nodes for inline frames. Such frames will reference the IP of the non-inlined frame, but hold the symbol and srcline for an inlined location. As such, we won't be able to query the srcline on-demand based on the IP alone. Instead, we will leverage the functionality provided by this patch here, and store the srcline for the inlined nodes in the new srcline member of callchain_cursor_node. Note that this patch on its own leaks the srcline, as there is no free_callchain_cursor_node or similar. A future patch will add caching of the srcline and handle deletion properly. Signed-off-by:
-
Milian Wolff authored
The follow-up commits will make inline frames first-class citizens in the callchain, thereby obsoleting all of this special code. Signed-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-4.15-20171023' 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: - Update vendor events JSON metrics for Intel's Broadwell, Broadwell Server, Haswell, Haswell Server, IvyBridge, IvyTown, JakeTown, Sandy Bridge, Skylake and SkyLake Server (Andi Kleen) - Add vendor event file for Intel's Goldmont Plus V1 (Kan Liang) - Move perf_mmap methods from 'perf record' and evlist.c to a separate mmap.[ch] pair, to better separate things and pave the way for further work on multithreading tools (Arnaldo Carvalho de Melo) - Do not check ABI headers in a detached tarball build, as it the kernel headers from where we copied tools/include/ are by definition not available (Arnaldo Carvalho de Melo) - Make 'perf script' use fprintf() like printing, i.e. receiving a FILE pointer so that it gets consistent with other tools/ code and allows for printing to per-event files (Arnaldo Carvalho de Melo) - Error handling fixes (resource release on exit) for 'perf script' and 'perf kmem' (Christophe JAILLET) - Make some 'perf event attr' tests optional on virtual machines, where tested counters are not available (Jiri Olsa) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
- 23 Oct, 2017 25 commits
-
-
Kan Liang authored
Add a Intel event file for perf. Signed-off-by:
Kan Liang <Kan.liang@intel.com> Acked-by:
-
Arnaldo Carvalho de Melo authored
We don't need perf.h, that is a kitchen sink, all we need is perf_events.h for perf_ns_link_info, sys/types.h for pid_t and linux/types.h for u64, list_head. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Li Zhijian <lizhijian@cn.fujitsu.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-f2uxyaj4s2hmntkrezpa6dqz@git.kernel.orgSigned-off-by:
-
Christophe JAILLET authored
If the string passed in '--time' is invalid, we must do some cleanup before leaving. As in the other error handling paths of this function. Signed-off-by:
Christophe JAILLET <christophe.jaillet@wanadoo.fr> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: kernel-janitors@vger.kernel.org Fixes: 2a865bd8 ("perf kmem: Add option to specify time window of interest") Link: http://lkml.kernel.org/r/20170916060936.28199-1-christophe.jaillet@wanadoo.frSigned-off-by:
-
Christophe JAILLET authored
If the string passed in '--time' is invalid, or if failed to set libtraceevent function resolver, we must do some cleanup before leaving. As in the other error handling paths of this function. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
We've been mixing print() with fprintf() style printing for a while, but now we need to use fprintf() like syntax uniformly as a preparatory patch for supporting printing to different files, one per event. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Cc: yuzhoujian <yuzhoujian@didichuxing.com> Link: http://lkml.kernel.org/n/tip-kv5z3v8ptfghbarv3a9usvin@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Out of print_binary() but receiving a fp pointer and expecting that the printer be a fprintf like function, i.e. receive a FILE pointer and return the number of characters printed. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Cc: yuzhoujian <yuzhoujian@didichuxing.com> Link: http://lkml.kernel.org/n/tip-6oqnxr6lmgqe6q6p3iugnscx@git.kernel.orgSigned-off-by:
-
Andi Kleen authored
Some of the metrics use an incorrect syntax for specifying the cmask for an event. Convert to perf syntax so that they can be resolved. Fixes metrics on Broadwell, SandyBridge. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
When we use one of: [acme@jouet linux]$ make help | grep perf perf-tar-src-pkg - Build perf-4.14.0-rc3.tar source tarball perf-targz-src-pkg - Build perf-4.14.0-rc3.tar.gz source tarball perf-tarbz2-src-pkg - Build perf-4.14.0-rc3.tar.bz2 source tarball perf-tarxz-src-pkg - Build perf-4.14.0-rc3.tar.xz source tarball [acme@jouet linux]$ I.e. when we create a detached tarball to build perf outside outside the enveloping kernel sources (from a kernel tarball or a checked out linux.git directory) we by definition can't check for differences among the tools/{include,arch}, etc files we originally copied from the kernel, so bail out in that case, to avoid warnings when doing the detached builds. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-vbrga0mhplv7niwxr3ghjyxv@git.kernel.orgSigned-off-by: -
Jiri Olsa authored
There's no need for extra cpuid_parse arch callback, it can be handled directly in init callback. Adding the init function to x86 to cover the cpuid initialization. Signed-off-by:
Jiri Olsa <jolsa@kernel.org> Cc: Andi Kleen <andi@firstfloor.org> Cc: David Ahern <dsahern@gmail.com> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20171011150158.11895-2-jolsa@kernel.orgSigned-off-by:
-
Andi Kleen authored
Fix an incorrect description in the 'perf list' manpage. When a group does not fit into the hardware it is partially scheduled, but does not error out. Signed-off-by:
-
Jiri Olsa authored
Otherwise we fail on virtual machines with no support for specific HW events. Signed-off-by:
-
Arnaldo Carvalho de Melo authored
The previous prep patch was just to show exactly what changed in that function, now its time to move that method and things only it uses to the right place, mmap.[ch] Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-aaxywfgw3d44x6xlu8zm1avu@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
It becomes a perf_mmap method, "push", that build reads from a mmap and "pushes" it to a consumer, that in the initial case, for 'perf record', just writes it to the perf.data file descriptor, but may be used by 'top', etc. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-u4l1qjbi6l76r2k0nv99220n@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
To better organize the sources, and we may end up even using it directly, without evlists and evsels. Cc: Adrian Hunter <adrian.hunter@intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-oiqrm7grflurnnzo2ovfnslg@git.kernel.orgSigned-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Andi Kleen authored
Signed-off-by:
-
Kees Cook authored
The jprobes subsystem is being removed, so convert to using kprobes instead. Signed-off-by:
Kees Cook <keescook@chromium.org> Acked-by:
Masami Hiramatsu <mhiramat@kernel.org> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Masami Hiramatsu <mhiramat@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Thomas Gleixner <tglx@linutronix.de> Link: http://lkml.kernel.org/r/20171020133127.GA18360@beastSigned-off-by:
-
- 20 Oct, 2017 2 commits
-
-
Masami Hiramatsu authored
Remove jprobes related documentation from kprobes.txt. Also add some migration advice for the people who are still using jprobes. Signed-off-by:
-
Masami Hiramatsu authored
Remove the jprobes sample module because jprobes are deprecated. Signed-off-by:
-
