- 28 Mar, 2017 4 commits
-
-
Ravi Bangoria authored
I found couple of events using al, bl, cl and dl registers for argument. These are not directly accepted by uprobe_events and thus needs to be mapped to ax, bx, cx and dx respectively. Few ex, /usr/bin/qemu-system-s390x css_adapter_interrupt: 1@%bl css_chpid_add: 1@%cl 1@%sil 1@%dl dma_bdrv_io: 8@%rbx 8@%rbp -8@%r14 1@%al /usr/bin/postgres buffer__read__done: ... -1@-bash -1@%al buffer__read__start: ... -1@%al I don't find any sdt events using ah, bh,... registers. But I also don't see any reason to not use them, so there might be rare events using these registers, and if so, perf should have a renaming logic for them too. Signed-off-by:Ravi Bangoria <ravi.bangoria@linux.vnet.ibm.com> Acked-by:
Masami Hiramatsu <mhiramat@kernel.org> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Alexis Berlemont <alexis.berlemont@gmail.com> Cc: Hemant Kumar <hemant@linux.vnet.ibm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20170328094754.3156-2-ravi.bangoria@linux.vnet.ibm.comSigned-off-by:
Arnaldo Carvalho de Melo <acme@redhat.com>
-
Arnaldo Carvalho de Melo authored
This came from 'git', but isn't documented anywhere in tools/perf/Documentation/, looks like baggage we can do without, ditch it. 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-e7uwkn60t4hmlnwj99ba4t2s@git.kernel.orgSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-4.12-20170327' 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: New features: - Handle inline functions in callchains (Jin Yao) - Enable sorting by srcline as key (Milian Wolff) Fixes: - Fix no_size logic in addr_filter__resolve_kernel_syms() in the auxtrace code (Adrian Hunter) - Fix some thread refcount leaks in 'perf trace' (Arnaldo Carvalho de Melo) - Fix divide by zero when calculating percent for an event in a group in the annotate by source line code (Taeung Song) - build-id files now aren't anymore symlinks, their parent directories are, so readlink the later (Taeung Song) - Assorted fixes for null termination problems, mostly related to readlink, detected by valgrind (Tommi Rantala) Infrastructure changes: - Make vfs_getname probe point logic in 'perf trace' more robust wrt length of pathname (Arnaldo Carvalho de Melo) - Remove unused 'prefix' parameter from builtins main functions (Arnaldo Carvalho de Melo) - Show 'perf list sdt' option in man page (Ravi Bangoria) Signed-off-by:
Ingo Molnar <mingo@kernel.org>
-
Ingo Molnar authored
Signed-off-by:
-
- 27 Mar, 2017 17 commits
-
-
Tommi Rantala authored
Simplification: it is easier to open /proc/self/exe than /proc/$pid/exe. Signed-off-by:
Tommi Rantala <tommi.t.rantala@nokia.com> Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/20170322130624.21881-7-tommi.t.rantala@nokia.comSigned-off-by:
-
Tommi Rantala authored
Ensure that the string that we read from the data file is null terminated. Valgrind was complaining: ==31357== Invalid read of size 1 ==31357== at 0x4EC8C1: __strtok_r_1c (string2.h:200) ==31357== by 0x4EC8C1: parse_ftrace_printk (trace-event-parse.c:161) ==31357== by 0x4F82A8: read_ftrace_printk (trace-event-read.c:204) ==31357== by 0x4F82A8: trace_report (trace-event-read.c:468) ==31357== by 0x4CD552: process_tracing_data (header.c:1576) ==31357== by 0x4D3397: perf_file_section__process (header.c:2705) ==31357== by 0x4D3397: perf_header__process_sections (header.c:2488) ==31357== by 0x4D3397: perf_session__read_header (header.c:2925) ==31357== by 0x4E71E2: perf_session__open (session.c:32) ==31357== by 0x4E71E2: perf_session__new (session.c:139) ==31357== by 0x429F5D: cmd_annotate (builtin-annotate.c:472) ==31357== by 0x497150: run_builtin (perf.c:359) ==31357== by 0x428CE0: handle_internal_command (perf.c:421) ==31357== by 0x428CE0: run_argv (perf.c:467) ==31357== by 0x428CE0: main (perf.c:614) ==31357== Address 0x8ac0efb is 0 bytes after a block of size 1,963 alloc'd ==31357== at 0x4C2DB9D: malloc (vg_replace_malloc.c:299) ==31357== by 0x4F827B: read_ftrace_printk (trace-event-read.c:195) ==31357== by 0x4F827B: trace_report (trace-event-read.c:468) ==31357== by 0x4CD552: process_tracing_data (header.c:1576) ==31357== by 0x4D3397: perf_file_section__process (header.c:2705) ==31357== by 0x4D3397: perf_header__process_sections (header.c:2488) ==31357== by 0x4D3397: perf_session__read_header (header.c:2925) ==31357== by 0x4E71E2: perf_session__open (session.c:32) ==31357== by 0x4E71E2: perf_session__new (session.c:139) ==31357== by 0x429F5D: cmd_annotate (builtin-annotate.c:472) ==31357== by 0x497150: run_builtin (perf.c:359) ==31357== by 0x428CE0: handle_internal_command (perf.c:421) ==31357== by 0x428CE0: run_argv (perf.c:467) ==31357== by 0x428CE0: main (perf.c:614) Signed-off-by:
-
Tommi Rantala authored
Ensure that we have space for the null byte in buf. Signed-off-by:
-
Tommi Rantala authored
Ensure that the string in buf is null terminated. Signed-off-by:
-
Tommi Rantala authored
Valgrind was complaining: $ valgrind ./perf list >/dev/null ==11643== Memcheck, a memory error detector ==11643== Copyright (C) 2002-2015, and GNU GPL'd, by Julian Seward et al. ==11643== Using Valgrind-3.12.0 and LibVEX; rerun with -h for copyright info ==11643== Command: ./perf list ==11643== ==11643== Conditional jump or move depends on uninitialised value(s) ==11643== at 0x4C30620: rindex (vg_replace_strmem.c:199) ==11643== by 0x49DAA9: build_id_cache__origname (build-id.c:198) ==11643== by 0x49E1C7: build_id_cache__valid_id (build-id.c:222) ==11643== by 0x49E1C7: build_id_cache__list_all (build-id.c:507) ==11643== by 0x4B9C8F: print_sdt_events (parse-events.c:2067) ==11643== by 0x4BB0B3: print_events (parse-events.c:2313) ==11643== by 0x439501: cmd_list (builtin-list.c:53) ==11643== by 0x497150: run_builtin (perf.c:359) ==11643== by 0x428CE0: handle_internal_command (perf.c:421) ==11643== by 0x428CE0: run_argv (perf.c:467) ==11643== by 0x428CE0: main (perf.c:614) [...] Additionally, a zero length result from readlink() is not very interesting. Signed-off-by:
-
Tommi Rantala authored
Valgrind was complaining: ==2633== Syscall param open(filename) points to unaddressable byte(s) ==2633== at 0x5281CC0: __open_nocancel (syscall-template.S:84) ==2633== by 0x537D38: open (fcntl2.h:53) ==2633== by 0x537D38: get_sdt_note_list (symbol-elf.c:2017) ==2633== by 0x5396FD: probe_cache__scan_sdt (probe-file.c:700) ==2633== by 0x49EA2C: build_id_cache__add_sdt_cache (build-id.c:625) ==2633== by 0x49EA2C: build_id_cache__add_s (build-id.c:697) ==2633== by 0x49EE72: build_id_cache__add_b (build-id.c:717) ==2633== by 0x49EE72: dso__cache_build_id (build-id.c:782) ==2633== by 0x49F190: __dsos__cache_build_ids (build-id.c:793) ==2633== by 0x49F190: machine__cache_build_ids (build-id.c:801) ==2633== by 0x49F190: perf_session__cache_build_ids (build-id.c:815) ==2633== by 0x4CD4F2: write_build_id (header.c:165) ==2633== by 0x4D26F7: do_write_feat (header.c:2296) ==2633== by 0x4D26F7: perf_header__adds_write (header.c:2335) ==2633== by 0x4D26F7: perf_session__write_header (header.c:2414) ==2633== by 0x43B324: __cmd_record (builtin-record.c:1154) ==2633== by 0x43B324: cmd_record (builtin-record.c:1839) ==2633== by 0x455A07: __cmd_record (builtin-kmem.c:1868) ==2633== by 0x455A07: cmd_kmem (builtin-kmem.c:1944) ==2633== by 0x497150: run_builtin (perf.c:359) ==2633== by 0x428CE0: handle_internal_command (perf.c:421) ==2633== by 0x428CE0: run_argv (perf.c:467) ==2633== by 0x428CE0: main (perf.c:614) ==2633== Address 0x0 is not stack'd, malloc'd or (recently) free'd Signed-off-by:
-
Taeung Song authored
Currently perf-annotate with --print-line can print -nan(0x8000000000000) because of division by zero when calculating percent. The division by zero happens when a sum of samples is zero in symbol__get_source_line(), so fix it. For example: After running 'perf record' like below, $ perf record -e "{cycles,page-faults,branch-misses}" ./a.out Before: $ perf annotate --stdio -l Sorted summary for file /home/taeung/workspace/a.out ---------------------------------------------- 32.89 -nan 7.04 a.c:38 25.14 -nan 0.00 a.c:34 16.26 -nan 56.34 a.c:31 15.88 -nan 1.41 a.c:37 5.67 -nan 0.00 a.c:39 1.13 -nan 35.21 a.c:26 0.95 -nan 0.00 a.c:44 0.57 -nan 0.00 a.c:32 Percent | Source code & Disassembly of a.out for cycles (529 samples) ----------------------------------------------------------------------------------------- : ... a.c:26 0.57 -nan 4.23 : 40081a: mov %edi,-0x24(%rbp) a.c:26 0.00 -nan 9.86 : 40081d: mov %rsi,-0x30(%rbp) ... However, if a sum of samples is zero (e.g. 'page-faults'), skip calculating percent. After: $ perf annotate --stdio -l Sorted summary for file /home/taeung/workspace/a.out ---------------------------------------------- 32.89 0.00 7.04 a.c:38 25.14 0.00 0.00 a.c:34 16.26 0.00 56.34 a.c:31 15.88 0.00 1.41 a.c:37 5.67 0.00 0.00 a.c:39 1.13 0.00 35.21 a.c:26 0.95 0.00 0.00 a.c:44 0.57 0.00 0.00 a.c:32 Percent | Source code & Disassembly of old for cycles (529 samples) ----------------------------------------------------------------------------------------- : ... a.c:26 0.57 0.00 4.23 : 40081a: mov %edi,-0x24(%rbp) a.c:26 0.00 0.00 9.86 : 40081d: mov %rsi,-0x30(%rbp) ... Signed-off-by:Taeung Song <treeze.taeung@gmail.com> Tested-by:
-
Taeung Song authored
It is wrong way to read link name from a build-id file. Because a build-id file is not anymore a symbolic link but build-id directory of it is symbolic link, so fix it. For example, if build-id file name gotten from dso__build_id_filename() is as below, /root/.debug/.build-id/4f/75c7d197c951659d1c1b8b5fd49bcdf8f3f8b1/elf To correctly read link name of build-id, use the build-id dir path that is a symbolic link, instead of the above build-id file name like below. /root/.debug/.build-id/4f/75c7d197c951659d1c1b8b5fd49bcdf8f3f8b1 Signed-off-by:
-
Milian Wolff authored
Often it is interesting to know how costly a given source line is in total. Previously, one had to build these sums manually based on all addresses that pointed to the same source line. This patch introduces srcline as a sort key, which will do the aggregation for us. Paired with the recent addition of showing inline frames, this makes perf report much more useful for many C++ work loads. The following shows the new feature in action. First, let's show the status quo output when we sort by address. The result contains many hist entries that generate the same output: ~~~~~~~~~~~~~~~~ $ perf report --stdio --inline -g address # Children Self Command Shared Object Symbol # ........ ........ ............ ................... ......................................... # 99.89% 35.34% cpp-inlining cpp-inlining [.] main | |--64.55%--main complex:655 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/complex:664 (inline) | | | |--60.31%--hypot +20 | | | | | |--8.52%--__hypot_finite +273 | | | | | |--7.32%--__hypot_finite +411 ... --35.34%--_start +4194346 __libc_start_main +241 | |--6.65%--main random.tcc:3326 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/bits/random.h:1809 (inline) | /usr/include/c++/6.3.1/bits/random.h:1818 (inline) | /usr/include/c++/6.3.1/bits/random.h:185 (inline) | |--2.70%--main random.tcc:3326 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/bits/random.h:1809 (inline) | /usr/include/c++/6.3.1/bits/random.h:1818 (inline) | /usr/include/c++/6.3.1/bits/random.h:185 (inline) | |--1.69%--main random.tcc:3326 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/bits/random.h:1809 (inline) | /usr/include/c++/6.3.1/bits/random.h:1818 (inline) | /usr/include/c++/6.3.1/bits/random.h:185 (inline) ... ~~~~~~~~~~~~~~~~ With this patch and `-g srcline` we instead get the following output: ~~~~~~~~~~~~~~~~ $ perf report --stdio --inline -g srcline # Children Self Command Shared Object Symbol # ........ ........ ............ ................... ......................................... # 99.89% 35.34% cpp-inlining cpp-inlining [.] main | |--64.55%--main complex:655 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/complex:664 (inline) | | | |--64.02%--hypot | | | | | --59.81%--__hypot_finite | | | --0.53%--cabs | --35.34%--_start __libc_start_main | |--12.48%--main random.tcc:3326 | /home/milian/projects/kdab/rnd/hotspot/tests/test-clients/cpp-inlining/main.cpp:39 (inline) | /usr/include/c++/6.3.1/bits/random.h:1809 (inline) | /usr/include/c++/6.3.1/bits/random.h:1818 (inline) | /usr/include/c++/6.3.1/bits/random.h:185 (inline) ... ~~~~~~~~~~~~~~~~ Signed-off-by:Milian Wolff <milian.wolff@kdab.com> Cc: Jiri Olsa <jolsa@kernel.org> Cc: Yao Jin <yao.jin@linux.intel.com> Link: http://lkml.kernel.org/r/20170318214928.9047-1-milian.wolff@kdab.comSigned-off-by:
-
Jin Yao authored
If the address belongs to an inlined function, the source information back to the first non-inlined function will be printed. For example: 1. Show inlined function name perf report -g function --inline - 0.69% 0.00% inline ld-2.23.so [.] dl_main - dl_main 0.56% _dl_relocate_object _dl_relocate_object (inline) elf_dynamic_do_Rela (inline) 2. Show the file/line information perf report -g address --inline - 0.69% 0.00% inline ld-2.23.so [.] _dl_start _dl_start rtld.c:307 /build/glibc-GKVZIf/glibc-2.23/elf/rtld.c:413 (inline) + _dl_sysdep_start dl-sysdep.c:250 Signed-off-by:Yao Jin <yao.jin@linux.intel.com> Tested-by:
-
Jin Yao authored
If the address belongs to an inlined function, the source information back to the first non-inlined function will be printed. For example: 1. Show inlined function name perf report --stdio -g function --inline 0.69% 0.00% inline ld-2.23.so [.] dl_main | ---dl_main | --0.56%--_dl_relocate_object _dl_relocate_object (inline) elf_dynamic_do_Rela (inline) 2. Show the file/line information perf report --stdio -g address --inline 0.69% 0.00% inline ld-2.23.so [.] _dl_start_user | ---_dl_start_user .:0 _dl_start rtld.c:307 /build/glibc-GKVZIf/glibc-2.23/elf/rtld.c:413 (inline) _dl_sysdep_start dl-sysdep.c:250 | --0.56%--dl_main rtld.c:2076 Committer tests: # perf record --call-graph dwarf ~/bin/perf stat usleep 1 Performance counter stats for 'usleep 1': 0.443020 task-clock (msec) # 0.449 CPUs utilized 1 context-switches # 0.002 M/sec 0 cpu-migrations # 0.000 K/sec 52 page-faults # 0.117 M/sec 1,049,423 cycles # 2.369 GHz 801,456 instructions # 0.76 insn per cycle 155,609 branches # 351.246 M/sec 7,026 branch-misses # 4.52% of all branches 0.000987570 seconds time elapsed [ perf record: Woken up 2 times to write data ] [ perf record: Captured and wrote 0.553 MB perf.data (66 samples) ] # perf report --stdio --inline fs__get_mountpoint <SNIP> 1.73% 0.00% perf perf [.] fs__get_mountpoint | ---fs__get_mountpoint fs__get_mountpoint (inline) fs__check_mounts (inline) __statfs entry_SYSCALL_64 sys_statfs SYSC_statfs user_statfs user_path_at_empty filename_lookup path_lookupat link_path_walk inode_permission __inode_permission kernfs_iop_permission kernfs_refresh_inode security_inode_notifysecctx selinux_inode_notifysecctx selinux_inode_setsecurity security_context_to_sid security_context_to_sid_core string_to_context_struct symcmp Signed-off-by: -
Jin Yao authored
It takes some time to look for inline stack for callgraph addresses. So it provides new option "--inline" to let user decide if enable this feature. --inline: If a callgraph address belongs to an inlined function, the inline stack will be printed. Each entry is the inline function name or file/line. Signed-off-by:
-
Jin Yao authored
It would be useful for perf to support a mode to query the inline stack for a given callgraph address. This would simplify finding the right code in code that does a lot of inlining. The srcline.c has contained the code which supports to translate the address to filename:line_nr. This patch just extends the function to let it support getting the inline stacks. It introduces the inline_list which will store the inline function result (filename:line_nr and funcname). If BFD lib is not supported, the result is only filename:line_nr. Signed-off-by:
-
Jin Yao authored
Introduce dso__name() and filename_split() out of existing code because these codes will be used in several places in next patch. For filename_split(), it may also solve a potential memory leak in existing code. In existing addr2line(), sep = strchr(filename, ':'); if (sep) { *sep++ = '\0'; *file = filename; *line_nr = strtoul(sep, NULL, 0); ret = 1; } out: pclose(fp); return ret; If sep is NULL, filename is not freed or returned via file. Signed-off-by: -
Arnaldo Carvalho de Melo authored
We got it from the git sources but never used it for anything, with the place where this would be somehow used remaining: static int run_builtin(struct cmd_struct *p, int argc, const char **argv) { prefix = NULL; if (p->option & RUN_SETUP) prefix = NULL; /* setup_perf_directory(); */ Ditch it. 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-uw5swz05vol0qpr32c5lpvus@git.kernel.orgSigned-off-by: -
Ravi Bangoria authored
Commit 40218dae ("perf list: Show SDT and pre-cached events") added sdt support in perf list, but it missed to update documentation. Show sdt option in man perf-list. Signed-off-by:
-
Adrian Hunter authored
Address filtering with kernel symbols incorrectly resulted in the error "Cannot determine size of symbol" because the no_size logic was the wrong way around. Signed-off-by:
Adrian Hunter <adrian.hunter@intel.com> Tested-by:
Andi Kleen <ak@linux.intel.com> Cc: stable@vger.kernel.org # v4.9+ Link: http://lkml.kernel.org/r/1490357752-27942-1-git-send-email-adrian.hunter@intel.comSigned-off-by:
-
- 24 Mar, 2017 4 commits
-
-
Arnaldo Carvalho de Melo authored
In trace__vfs_getname() and when checking if a thread is filtered in trace__process_sample() we were not dropping the reference obtained via machine__findnew_thread(), fix it. 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-9gc470phavxwxv5d9w7ck8ev@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
Trivial fix removing a tab in an error path. Link: http://lkml.kernel.org/n/tip-c14mk6cqaiby8gf5rpft3d9r@git.kernel.orgSigned-off-by:
-
Arnaldo Carvalho de Melo authored
It shouldn't be zero, but if the 'perf probe' on getname_flags() (or elsewhere in the future we need to probe to catch the pathname for syscalls like 'open' being copied from userspace to the kernel) is misplaced somehow, then we will end up not allocating space and trying to copy the "" empty string to ttrace->filename.name, causing a segfault, fix it. 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-c4f1t6sx1nczuzop19r5si5s@git.kernel.orgSigned-off-by:
-
Ingo Molnar authored
Merge tag 'perf-core-for-mingo-4.12-20170324' 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: - Allow suppressing 'uncore_' when specifying PMU events (Andi Kleen) - Collapse identically named PMU events in 'perf stat', allow not merging it via --no-merge (Andi Kleen) Fixes: - Use more precise 'grep -v' to suppress unwanted 'objdump -dS' disassembly output to not ditch line:number lines needed by 'perf annotate --print-lines' logic (Taeung Song) Infrastructure changes: - SDT (Statically Defined Tracing)/uprobes_events arguments improvements (Alexis Berlemont, Ravi Bangoria) - Improvements for the handling of JSON described vendor events, including having an expression parser to calculate metrics from multiple vendor events (Andi Kleen) - Update Intel JSON vendor event files (Andi Kleen) - Restore error reporting in 'perf probe -d' when none of the events requested to be deleted exist. (Kefeng Wang) - Bump MAX_CMDLEN in 'perf probe' to match what the kernel accepts (Ravi Bangoria) Signed-off-by:
-
- 23 Mar, 2017 7 commits
-
-
Andi Kleen authored
Move the printing of perf expressions and internal events to a new clearer --details flag, instead of lumping it together with other debug options in --debug. This makes it clearer to use. Before perf list --debug ... unc_m_power_critical_throttle_cycles [Cycles all ranks are in critical thermal throttle. Unit: uncore_imc] uncore_imc_2/event=0x86/ MetricName: power_critical_throttle_cycles % MetricExpr: (unc_m_power_critical_throttle_cycles / unc_m_clockticks) * 100. after perf list --details ... unc_m_power_critical_throttle_cycles [Cycles all ranks are in critical thermal throttle. Unit: uncore_imc] uncore_imc_2/event=0x86/ MetricName: power_critical_throttle_cycles % MetricExpr: (unc_m_power_critical_throttle_cycles / unc_m_clockticks) * 100. Signed-off-by:Jiri Olsa <jolsa@kernel.org> Tested-by:
-
Andi Kleen authored
Add support for a new JSON event attribute to name MetricExpr for better output in perf stat. If the event has no MetricName it uses the normal event name instead to describe the metric. Before % perf stat -a -I 1000 -e '{unc_p_clockticks,unc_p_freq_max_os_cycles}' --metric-only time unc_p_freq_max_os_cycles 1.000149775 15.7 2.000344807 19.3 3.000502544 16.7 4.000640656 6.6 5.000779955 9.9 After % perf stat -a -I 1000 -e '{unc_p_clockticks,unc_p_freq_max_os_cycles}' --metric-only time freq_max_os_cycles % 1.000149775 15.7 2.000344807 19.3 3.000502544 16.7 4.000640656 6.6 5.000779955 9.9 Signed-off-by: -
Andi Kleen authored
Output the metric expr in perf list when --debug is specified, so that the user can check the formula. Before: % perf list ... unc_m_power_channel_ppd [Cycles where DRAM ranks are in power down (CKE) mode. Derived from unc_m_power_channel_ppd. Unit: uncore_imc] uncore_imc_2/event=0x85/ After: % perf list --debug ... unc_m_power_channel_ppd [Cycles where DRAM ranks are in power down (CKE) mode. Derived from unc_m_power_channel_ppd. Unit: uncore_imc] Perf: uncore_imc_2/event=0x85/ MetricExpr: (unc_m_power_channel_ppd / unc_m_clockticks) * 100. Signed-off-by: -
Andi Kleen authored
Add generic infrastructure to perf stat to output ratios for "MetricExpr" entries in the event lists. Many events are more useful as ratios than in raw form, typically some count in relation to total ticks. Transfer the MetricExpr information from the alias to the evsel. We mark the events that need to be collected for MetricExpr, and also link the events using them with a pointer. The code is careful to always prefer the right event in the same group to minimize multiplexing errors. At the moment only a single relation is supported. Then add a rblist to the stat shadow code that remembers stats based on the cpu and context. Then finally update and retrieve and print these values similarly to the existing hardcoded perf metrics. We use the simple expression parser added earlier to evaluate the expression. Normally we just output the result without further commentary, but for --metric-only this would lead to empty columns. So for this case use the original event as description. There is no attempt to automatically add the MetricExpr event, if it is missing, however we suggest it to the user, because the user tool doesn't have enough information to reliably construct a group that is guaranteed to schedule. So we leave that to the user. % perf stat -a -I 1000 -e '{unc_p_clockticks,unc_p_freq_max_os_cycles}' 1.000147889 800,085,181 unc_p_clockticks 1.000147889 93,126,241 unc_p_freq_max_os_cycles # 11.6 2.000448381 800,218,217 unc_p_clockticks 2.000448381 142,516,095 unc_p_freq_max_os_cycles # 17.8 3.000639852 800,243,057 unc_p_clockticks 3.000639852 162,292,689 unc_p_freq_max_os_cycles # 20.3 % perf stat -a -I 1000 -e '{unc_p_clockticks,unc_p_freq_max_os_cycles}' --metric-only # time freq_max_os_cycles % 1.000127077 0.9 2.000301436 0.7 3.000456379 0.0 v2: Change from DivideBy to MetricExpr v3: Use expr__ prefix. Support more than one other event. v4: Update description v5: Only print warning message once for multiple PMUs. Signed-off-by: -
Andi Kleen authored
Add support for parsing the MetricExpr header in the JSON event lists and storing them in the alias structure. Used in the next patch. v2: Change DividedBy to MetricExpr v3: Really catch all uses of DividedBy Signed-off-by:
-
Andi Kleen authored
- Add MetricName to describe Metric - Remove redundant "derived from" in descriptions - Rename UNC_M_CAS_COUNT to LLC_MISSES.READ Signed-off-by:
-
Andi Kleen authored
Add a simple expression parser good enough to parse JSON relation expressions. The parser is implemented using bison. This is just intended as an simple parser for internal usage in the event lists, not the beginning of a "perf scripting language" v2: Use expr__ prefix instead of expr_ Support multiple free variables for parser Committer note: The v2 patch had: %define api.pure full In expr.y, that is a feature introduced in bison 2.7, to have reentrant parsers, not using global variables, which would make tools/perf stop building with the bison version shipped in older distros, so Andi realised that the other parsers (e.g. parse-events.y) were using: %pure-parser Which is present in older versions of bison and fits the bill. I added: CFLAGS_expr-bison.o += -DYYENABLE_NLS=0 -DYYLTYPE_IS_TRIVIAL=0 -w To finally make it build, copying what was there for pmu-bison.o, another parser. Signed-off-by:
-
- 21 Mar, 2017 8 commits
-
-
Andi Kleen authored
Special case uncore_ prefix in PMU match, to allow for shorter event uncore specifications. Before: perf stat -a -e uncore_cbox/event=0x35,umask=0x1,filter_opc=0x19C/ sleep 1 After perf stat -a -e cbox/event=0x35,umask=0x1,filter_opc=0x19C/ sleep 1 Committer tests: # perf list uncore List of pre-defined events (to be used in -e): uncore_cbox_0/clockticks/ [Kernel PMU event] uncore_cbox_1/clockticks/ [Kernel PMU event] uncore_imc/data_reads/ [Kernel PMU event] uncore_imc/data_writes/ [Kernel PMU event] # perf stat -a -e cbox_0/clockticks/ sleep 1 Performance counter stats for 'system wide': 281,474,976,653,084 cbox_0/clockticks/ 1.000870129 seconds time elapsed # Signed-off-by: -
Andi Kleen authored
When the user specifies a pmu directly, expand it automatically with a prefix match for all available PMUs, similar as we do for the normal aliases now. This allows to specify attributes for duplicated boxes quickly. For example uncore_cbox_{0,6}/.../ can be now specified as uncore_cbox/.../ and it gets automatically expanded for all boxes. This generally makes it more concise to write uncore specifications, and also avoids the need to know the exact topology of the system. Before: % perf stat -a -e uncore_cbox_0/event=0x35,umask=0x1,filter_opc=0x19C/,\ uncore_cbox_1/event=0x35,umask=0x1,filter_opc=0x19C/,\ uncore_cbox_2/event=0x35,umask=0x1,filter_opc=0x19C/,\ uncore_cbox_3/event=0x35,umask=0x1,filter_opc=0x19C/,\ uncore_cbox_4/event=0x35,umask=0x1,filter_opc=0x19C/,\ uncore_cbox_5/event=0x35,umask=0x1,filter_opc=0x19C/ sleep 1 After: % perf stat -a -e uncore_cbox/event=0x35,umask=0x1,filter_opc=0x19C/ sleep 1 v2: Handle all bison rules. Move multi add code to separate function. Handle uncore_ prefix correctly. v3: Move parse_events_multi_pmu_add to separate patch. Move uncore prefix check to separate patch. Signed-off-by: -
Andi Kleen authored
Factor out the PMU name matching in the event parser into a separate function, to use the same code for other grammar rules later. Signed-off-by:
-
Andi Kleen authored
When any result that is being merged is bad, mark them all bad to give consistent output in interval mode. No before/after, because the issue was only found in theoretical review and it is hard to reproduce Signed-off-by:
-
Andi Kleen authored
The uncore PMU has a lot of duplicated PMUs for different subsystems. When expanding an uncore alias we usually end up with a large number of identically named aliases, which makes perf stat output difficult to read. Automatically sum them up in perf stat, unless --no-merge is specified. This can be default because only the uncores generally have duplicated aliases. Other PMUs have unique names. Before: % perf stat --no-merge -a -e unc_c_llc_lookup.any sleep 1 Performance counter stats for 'system wide': 694,976 Bytes unc_c_llc_lookup.any 706,304 Bytes unc_c_llc_lookup.any 956,608 Bytes unc_c_llc_lookup.any 782,720 Bytes unc_c_llc_lookup.any 605,696 Bytes unc_c_llc_lookup.any 442,816 Bytes unc_c_llc_lookup.any 659,328 Bytes unc_c_llc_lookup.any 509,312 Bytes unc_c_llc_lookup.any 263,936 Bytes unc_c_llc_lookup.any 592,448 Bytes unc_c_llc_lookup.any 672,448 Bytes unc_c_llc_lookup.any 608,640 Bytes unc_c_llc_lookup.any 641,024 Bytes unc_c_llc_lookup.any 856,896 Bytes unc_c_llc_lookup.any 808,832 Bytes unc_c_llc_lookup.any 684,864 Bytes unc_c_llc_lookup.any 710,464 Bytes unc_c_llc_lookup.any 538,304 Bytes unc_c_llc_lookup.any 1.002577660 seconds time elapsed After: % perf stat -a -e unc_c_llc_lookup.any sleep 1 Performance counter stats for 'system wide': 2,685,120 Bytes unc_c_llc_lookup.any 1.002648032 seconds time elapsed v2: Split collect_aliases. Rename alias flag. v3: Make sure unsupported/not counted is always printed. v4: Factor out callback change into separate patch. v5: Move check for bad results here Move merged check into collect_data Signed-off-by: -
Andi Kleen authored
To be used in next patch to support automatic summing of alias events. v2: Move check for bad results to next patch v3: Remove trivial addition. v4: Use perf_evsel__cpus instead of evsel->cpus Signed-off-by:
-
Arnaldo Carvalho de Melo authored
The source code line number (lineno) needs to be kept in accross calls to symbol__parse_objdump_line() when parsing the output of 'objdump -l -dS', so that it can associate it with the instructions till the next line. See disasm_line__new() and struct disasm_line::line_nr. 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: Taeung Song <treeze.taeung@gmail.com> Cc: Wang Nan <wangnan0@huawei.com> Link: http://lkml.kernel.org/n/tip-7hpx8f8ybdpiujceysaj229w@git.kernel.orgSigned-off-by:
-
Taeung Song authored
The 'grep -v "filename"' applied to the objdump command output cause a side effect eliminating filename:linenr of output of 'objdump -l' if the object file name and source file name are the same, fix it. E.g. the output of the following objdump command in symbol__disassemble(): $ objdump -l -d -S -C /home/taeung/hello --start-address=... /home/taeung/hello: file format elf64-x86-64 Disassembly of section .text: 0000000000400526 <main>: main(): /home/taeung/hello.c:4 void main() { 400526: 55 push %rbp 400527: 48 89 e5 mov %rsp,%rbp /home/taeung/hello.c:5 ... But it uses grep -v "filename" e.g. "/home/taeung/hello" in the objdump command to remove the first line containing file name and file format ("/home/taeung/hello: file format elf64-x86-64"): Before: $ objdump -l -d -S -C /home/taeung/hello | grep /home/taeung/hello But this causes a side effect, removing filename:linenr too, because the object file and source file have the same name e.g. "/home/taueng/hello", "/home/taeung/hello.c" So more do a better match by using grep -v as below to correctly remove that first line: "/home/taeung/hello: file format elf64-x86-64" After: $ objdump -l -d -S -C /home/taeung/hello | grep /home/taeung/hello: Signed-off-by:
-
