perf/core improvements and fixes
Arnaldo Carvalho de Melo
maintainerHi Ingo, Thomas,
Please consider pulling,
Best regards,
- Arnaldo
Test results at the end of this message, as usual.
The following changes since commit 4511708b9a044f2bc83c7c7f7f8a2c45ec488219:
Merge tag 'perf-core-for-mingo-5.4-20190814' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core (2019-08-15 11:10:38 +0200)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-5.4-20190816
for you to fetch changes up to e2736219e6ca3117e10651e215b96d66775220da:
perf unwind: Remove unnecessary test (2019-08-16 12:30:14 -0300)
----------------------------------------------------------------
perf/core improvements and fixes:
report/script/trace/top:
Arnaldo Carvalho de Melo:
- Allow specifying marker events demarcating when to consider the other events, i.e. one now can state something like:
# perf probe kernel_function
# perf record -e cycles,probe:kernel_function
And then, in 'perf script' or 'perf report' say:
# perf report --switch-on=probe:kernel_function
And then the cycles event samples will be considered only after we find the first probe:kernel_function event.
There is also --switch-off=event, to make it stop considering events out of some window, say to avoid some winding down of a workload.
The same can be done with the "live mode" tools: 'perf top' and 'perf trace'.
There are examples in the cset comments showing how to use it with SDT events in things like 'systemtap', that have those tracepoint-like events for the start/end of passes, etc.
Another example involves selecting scheduler events + entry/exit of a syscall, using the syscalls tracepoints, one can then see the scheduler events that take place while that syscall is being processed.
In the future this should be possible in record/top/trace via eBPF where the perf tools would hook into the marker events and enable events put in place but not enabled when the on/off conditions are the desired ones, reducing the amount of events sampled, but this userspace only solution should be good enough for many scenarios.
perf vendor events intel:
Haiyan Song:
- Add Tremontx event file v1.02.
unwind:
John Keeping:
- Fix callchain unwinding when tid != pid, that was working only for the thread group leader.
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
----------------------------------------------------------------
Arnaldo Carvalho de Melo (13):
perf script: Allow specifying event to switch on processing of other events perf script: Allow showing the --switch-on event
perf script: Allow specifying event to switch off processing of other events perf evswitch: Move struct to a separate header to use in other tools perf evswitch: Move switch logic to use in other tools
perf evswitch: Add the names of on/off events
perf evswitch: Introduce OPTS_EVSWITCH() for cmd line processing
perf evswitch: Introduce init() method to set the on/off evsels from the command line perf evswitch: Move enoent error message printing to separate function perf evswitch: Add hint when not finding specified on/off events
perf trace: Add --switch-on/--switch-off events
perf top: Add --switch-on/--switch-off events
perf report: Add --switch-on/--switch-off events
Haiyan Song (1):
perf vendor events intel: Add Tremontx event file v1.02
John Keeping (3):
perf map: Use zalloc for map_groups
perf unwind: Fix libunwind when tid != pid
perf unwind: Remove unnecessary test
tools/perf/Documentation/perf-report.txt | 17 + tools/perf/Documentation/perf-script.txt | 9 + tools/perf/Documentation/perf-top.txt | 38 ++ tools/perf/Documentation/perf-trace.txt | 9 + tools/perf/builtin-report.c | 10 + tools/perf/builtin-script.c | 10 + tools/perf/builtin-top.c | 10 +- tools/perf/builtin-trace.c | 10 + tools/perf/pmu-events/arch/x86/mapfile.csv | 1 + tools/perf/pmu-events/arch/x86/tremontx/cache.json | 111 ++++++ .../pmu-events/arch/x86/tremontx/frontend.json | 26 ++ .../perf/pmu-events/arch/x86/tremontx/memory.json | 26 ++ tools/perf/pmu-events/arch/x86/tremontx/other.json | 26 ++ .../pmu-events/arch/x86/tremontx/pipeline.json | 111 ++++++ .../arch/x86/tremontx/uncore-memory.json | 73 ++++ .../pmu-events/arch/x86/tremontx/uncore-other.json | 431 +++++++++++++++++++++ .../pmu-events/arch/x86/tremontx/uncore-power.json | 11 + .../arch/x86/tremontx/virtual-memory.json | 86 ++++ tools/perf/util/Build | 1 + tools/perf/util/evswitch.c | 61 +++ tools/perf/util/evswitch.h | 31 ++ tools/perf/util/map.c | 5 +- tools/perf/util/map_groups.h | 4 + tools/perf/util/thread.c | 7 +- tools/perf/util/thread.h | 4 - tools/perf/util/top.h | 2 + tools/perf/util/unwind-libunwind-local.c | 18 +- tools/perf/util/unwind-libunwind.c | 40 +- tools/perf/util/unwind.h | 25 +- 29 files changed, 1158 insertions(+), 55 deletions(-) create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/cache.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/frontend.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/memory.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/other.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/pipeline.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-memory.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-other.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-power.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/virtual-memory.json create mode 100644 tools/perf/util/evswitch.c create mode 100644 tools/perf/util/evswitch.h
Test results:
The first ones are container based builds of tools/perf with and without libelf support. Where clang is available, it is also used to build perf with/without libelf, and building with LIBCLANGLLVM=1 (built-in clang) with gcc and clang when clang and its devel libraries are installed.
The objtool and samples/bpf/ builds are disabled now that I'm switching from using the sources in a local volume to fetching them from a http server to build it inside the container, to make it easier to build in a container cluster. Those will come back later.
Several are cross builds, the ones with -x-ARCH and the android one, and those may not have all the features built, due to lack of multi-arch devel packages, available and being used so far on just a few, like debian:experimental-x-{arm64,mipsel}.
The 'perf test' one will perform a variety of tests exercising tools/perf/util/, tools/lib/{bpf,traceevent,etc}, as well as run perf commands with a variety of command line event specifications to then intercept the sys_perf_event syscall to check that the perf_event_attr fields are set up as expected, among a variety of other unit tests.
Then there is the 'make -C tools/perf build-test' ones, that build tools/perf/ with a variety of feature sets, exercising the build with an incomplete set of features as well as with a complete one. It is planned to have it run on each of the containers mentioned above, using some container orchestration infrastructure. Get in contact if interested in helping having this in place.
Clearlinux is failing when building with libpython, but that is not a perf regression, will try to remove one compiler warning that is causing the problem when building some of the glue code files in the python files, outside perf.
# export PERF_TARBALL=http://192.168.124.1/perf/perf-5.3.0-rc4.tar.xz # dm 1 alpine:3.4 : Ok gcc (Alpine 5.3.0) 5.3.0, clang version 3.8.0 (tags/RELEASE_380/final) 2 alpine:3.5 : Ok gcc (Alpine 6.2.1) 6.2.1 20160822, clang version 3.8.1 (tags/RELEASE_381/final) 3 alpine:3.6 : Ok gcc (Alpine 6.3.0) 6.3.0, clang version 4.0.0 (tags/RELEASE_400/final) 4 alpine:3.7 : Ok gcc (Alpine 6.4.0) 6.4.0, Alpine clang version 5.0.0 (tags/RELEASE_500/final) (based on LLVM 5.0.0) 5 alpine:3.8 : Ok gcc (Alpine 6.4.0) 6.4.0, Alpine clang version 5.0.1 (tags/RELEASE_501/final) (based on LLVM 5.0.1) 6 alpine:3.9 : Ok gcc (Alpine 8.3.0) 8.3.0, Alpine clang version 5.0.1 (tags/RELEASE_502/final) (based on LLVM 5.0.1) 7 alpine:3.10 : Ok gcc (Alpine 8.3.0) 8.3.0, Alpine clang version 8.0.0 (tags/RELEASE_800/final) (based on LLVM 8.0.0) 8 alpine:edge : Ok gcc (Alpine 8.3.0) 8.3.0, Alpine clang version 8.0.1 (tags/RELEASE_801/final) (based on LLVM 8.0.1) 9 amazonlinux:1 : Ok gcc (GCC) 7.2.1 20170915 (Red Hat 7.2.1-2), clang version 3.6.2 (tags/RELEASE_362/final) 10 amazonlinux:2 : Ok gcc (GCC) 7.3.1 20180303 (Red Hat 7.3.1-5), clang version 7.0.1 (Amazon Linux 2 7.0.1-1.amzn2.0.2) 11 android-ndk:r12b-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 12 android-ndk:r15c-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 13 centos:5 : Ok gcc (GCC) 4.1.2 20080704 (Red Hat 4.1.2-55) 14 centos:6 : Ok gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23) 15 centos:7 : Ok gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-36), clang version 3.4.2 (tags/RELEASE_34/dot2-final) 16 clearlinux:latest : Ok gcc (Clear Linux OS for Intel Architecture) 9.1.1 20190808 gcc-9-branch@274204, clang version 8.0.0 (tags/RELEASE_800/final) 17 debian:8 : Ok gcc (Debian 4.9.2-10+deb8u2) 4.9.2, Debian clang version 3.5.0-10 (tags/RELEASE_350/final) (based on LLVM 3.5.0) 18 debian:9 : Ok gcc (Debian 6.3.0-18+deb9u1) 6.3.0 20170516, clang version 3.8.1-24 (tags/RELEASE_381/final) 19 debian:10 : Ok gcc (Debian 8.3.0-6) 8.3.0, clang version 7.0.1-8 (tags/RELEASE_701/final) 20 debian:experimental : Ok gcc (Debian 8.3.0-19) 8.3.0, clang version 7.0.1-9 (tags/RELEASE_701/final) 21 debian:experimental-x-arm64 : Ok aarch64-linux-gnu-gcc (Debian 8.3.0-19) 8.3.0 22 debian:experimental-x-mips : Ok mips-linux-gnu-gcc (Debian 8.3.0-19) 8.3.0 23 debian:experimental-x-mips64 : Ok mips64-linux-gnuabi64-gcc (Debian 8.3.0-7) 8.3.0 24 debian:experimental-x-mipsel : Ok mipsel-linux-gnu-gcc (Debian 8.3.0-19) 8.3.0 25 fedora:20 : Ok gcc (GCC) 4.8.3 20140911 (Red Hat 4.8.3-7), clang version 3.4.2 (tags/RELEASE_34/dot2-final) 26 fedora:22 : Ok gcc (GCC) 5.3.1 20160406 (Red Hat 5.3.1-6), clang version 3.5.0 (tags/RELEASE_350/final) 27 fedora:23 : Ok gcc (GCC) 5.3.1 20160406 (Red Hat 5.3.1-6), clang version 3.7.0 (tags/RELEASE_370/final) 28 fedora:24 : Ok gcc (GCC) 6.3.1 20161221 (Red Hat 6.3.1-1), clang version 3.8.1 (tags/RELEASE_381/final) 29 fedora:24-x-ARC-uClibc : Ok arc-linux-gcc (ARCompact ISA Linux uClibc toolchain 2017.09-rc2) 7.1.1 20170710 30 fedora:25 : Ok gcc (GCC) 6.4.1 20170727 (Red Hat 6.4.1-1), clang version 3.9.1 (tags/RELEASE_391/final) 31 fedora:26 : Ok gcc (GCC) 7.3.1 20180130 (Red Hat 7.3.1-2), clang version 4.0.1 (tags/RELEASE_401/final) 32 fedora:27 : Ok gcc (GCC) 7.3.1 20180712 (Red Hat 7.3.1-6), clang version 5.0.2 (tags/RELEASE_502/final) 33 fedora:28 : Ok gcc (GCC) 8.3.1 20190223 (Red Hat 8.3.1-2), clang version 6.0.1 (tags/RELEASE_601/final) 34 fedora:29 : Ok gcc (GCC) 8.3.1 20190223 (Red Hat 8.3.1-2), clang version 7.0.1 (Fedora 7.0.1-6.fc29) 35 fedora:30 : Ok gcc (GCC) 9.1.1 20190503 (Red Hat 9.1.1-1), clang version 8.0.0 (Fedora 8.0.0-1.fc30) 36 fedora:30-x-ARC-glibc : Ok arc-linux-gcc (ARC HS GNU/Linux glibc toolchain 2019.03-rc1) 8.3.1 20190225 37 fedora:30-x-ARC-uClibc : Ok arc-linux-gcc (ARCv2 ISA Linux uClibc toolchain 2019.03-rc1) 8.3.1 20190225 38 fedora:31 : Ok gcc (GCC) 9.1.1 20190605 (Red Hat 9.1.1-2), clang version 8.0.0 (Fedora 8.0.0-3.fc31.1) 39 fedora:rawhide : Ok gcc (GCC) 9.1.1 20190605 (Red Hat 9.1.1-2), clang version 8.0.0 (Fedora 8.0.0-3.fc31.1) 40 gentoo-stage3-amd64:latest : Ok gcc (Gentoo 8.3.0-r1 p1.1) 8.3.0 41 mageia:5 : Ok gcc (GCC) 4.9.2, clang version 3.5.2 (tags/RELEASE_352/final) 42 mageia:6 : Ok gcc (Mageia 5.5.0-1.mga6) 5.5.0, clang version 3.9.1 (tags/RELEASE_391/final) 43 mageia:7 : Ok gcc (Mageia 8.3.1-0.20190524.1.mga7) 8.3.1 20190524, clang version 8.0.0 (Mageia 8.0.0-1.mga7) 44 manjaro:latest : Ok gcc (GCC) 9.1.0, clang version 8.0.1 (tags/RELEASE_801/final) 45 opensuse:15.0 : Ok gcc (SUSE Linux) 7.4.1 20190424 [gcc-7-branch revision 270538], clang version 5.0.1 (tags/RELEASE_501/final 312548) 46 opensuse:15.1 : Ok gcc (SUSE Linux) 7.4.0, clang version 7.0.1 (tags/RELEASE_701/final 349238) 47 opensuse:42.3 : Ok gcc (SUSE Linux) 4.8.5, clang version 3.8.0 (tags/RELEASE_380/final 262553) 48 opensuse:tumbleweed : Ok gcc (SUSE Linux) 9.1.1 20190723 [gcc-9-branch revision 273734], clang version 8.0.1 (tags/RELEASE_801/final 366581) 49 oraclelinux:6 : Ok gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23.0.1) 50 oraclelinux:7 : Ok gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-39.0.1), clang version 3.4.2 (tags/RELEASE_34/dot2-final) 51 oraclelinux:8 : Ok gcc (GCC) 8.2.1 20180905 (Red Hat 8.2.1-3.0.1), clang version 7.0.1 (tags/RELEASE_701/final) 52 ubuntu:12.04 : Ok gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3, Ubuntu clang version 3.0-6ubuntu3 (tags/RELEASE_30/final) (based on LLVM 3.0) 53 ubuntu:14.04 : Ok gcc (Ubuntu 4.8.4-2ubuntu1~14.04.4) 4.8.4, Ubuntu clang version 3.4-1ubuntu3 (tags/RELEASE_34/final) (based on LLVM 3.4) 54 ubuntu:16.04 : Ok gcc (Ubuntu 5.4.0-6ubuntu1~16.04.11) 5.4.0 20160609, clang version 3.8.0-2ubuntu4 (tags/RELEASE_380/final) 55 ubuntu:16.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 56 ubuntu:16.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 57 ubuntu:16.04-x-powerpc : Ok powerpc-linux-gnu-gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 58 ubuntu:16.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu/IBM 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 59 ubuntu:16.04-x-powerpc64el : Ok powerpc64le-linux-gnu-gcc (Ubuntu/IBM 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 60 ubuntu:16.04-x-s390 : Ok s390x-linux-gnu-gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 61 ubuntu:18.04 : Ok gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0, clang version 6.0.0-1ubuntu2 (tags/RELEASE_600/final) 62 ubuntu:18.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 7.4.0-1ubuntu1~18.04.1) 7.4.0 63 ubuntu:18.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 7.4.0-1ubuntu1~18.04.1) 7.4.0 64 ubuntu:18.04-x-m68k : Ok m68k-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 65 ubuntu:18.04-x-powerpc : Ok powerpc-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 66 ubuntu:18.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 67 ubuntu:18.04-x-powerpc64el : Ok powerpc64le-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 68 ubuntu:18.04-x-riscv64 : Ok riscv64-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 69 ubuntu:18.04-x-s390 : Ok s390x-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 70 ubuntu:18.04-x-sh4 : Ok sh4-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 71 ubuntu:18.04-x-sparc64 : Ok sparc64-linux-gnu-gcc (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0 72 ubuntu:18.10 : Ok gcc (Ubuntu 8.3.0-6ubuntu1~18.10.1) 8.3.0, clang version 7.0.0-3 (tags/RELEASE_700/final) 73 ubuntu:19.04 : Ok gcc (Ubuntu 8.3.0-6ubuntu1) 8.3.0, clang version 8.0.0-3 (tags/RELEASE_800/final) 74 ubuntu:19.04-x-alpha : Ok alpha-linux-gnu-gcc (Ubuntu 8.3.0-6ubuntu1) 8.3.0 75 ubuntu:19.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 8.3.0-6ubuntu1) 8.3.0 76 ubuntu:19.04-x-hppa : Ok hppa-linux-gnu-gcc (Ubuntu 8.3.0-6ubuntu1) 8.3.0 77 ubuntu:19.10 : Ok gcc (Ubuntu 9.1.0-9ubuntu2) 9.1.0, clang version 8.0.1-+rc4-1 (tags/RELEASE_801/rc4)
# uname -a
Linux quaco 5.2.6-200.fc30.x86_64 #1 SMP Mon Aug 5 13:20:47 UTC 2019 x86_64 x86_64 x86_64 GNU/Linux
# git log --oneline -1
e2736219e6ca perf unwind: Remove unnecessary test
# perf version --build-options
perf version 5.3.rc4.ge2736219e6ca
dwarf: [ on ] # HAVE_DWARF_SUPPORT
dwarf_getlocations: [ on ] # HAVE_DWARF_GETLOCATIONS_SUPPORT
glibc: [ on ] # HAVE_GLIBC_SUPPORT
gtk2: [ on ] # HAVE_GTK2_SUPPORT
syscall_table: [ on ] # HAVE_SYSCALL_TABLE_SUPPORT
libbfd: [ on ] # HAVE_LIBBFD_SUPPORT
libelf: [ on ] # HAVE_LIBELF_SUPPORT
libnuma: [ on ] # HAVE_LIBNUMA_SUPPORT
numa_num_possible_cpus: [ on ] # HAVE_LIBNUMA_SUPPORT
libperl: [ on ] # HAVE_LIBPERL_SUPPORT
libpython: [ on ] # HAVE_LIBPYTHON_SUPPORT
libslang: [ on ] # HAVE_SLANG_SUPPORT
libcrypto: [ on ] # HAVE_LIBCRYPTO_SUPPORT
libunwind: [ on ] # HAVE_LIBUNWIND_SUPPORT
libdw-dwarf-unwind: [ on ] # HAVE_DWARF_SUPPORT
zlib: [ on ] # HAVE_ZLIB_SUPPORT
lzma: [ on ] # HAVE_LZMA_SUPPORT
get_cpuid: [ on ] # HAVE_AUXTRACE_SUPPORT
bpf: [ on ] # HAVE_LIBBPF_SUPPORT
aio: [ on ] # HAVE_AIO_SUPPORT
zstd: [ on ] # HAVE_ZSTD_SUPPORT
# perf test
1: vmlinux symtab matches kallsyms : Ok
2: Detect openat syscall event : Ok
3: Detect openat syscall event on all cpus : Ok
4: Read samples using the mmap interface : Ok
5: Test data source output : Ok
6: Parse event definition strings : Ok
7: Simple expression parser : Ok
8: PERF_RECORD_* events & perf_sample fields : Ok
9: Parse perf pmu format : Ok
10: DSO data read : Ok
11: DSO data cache : Ok
12: DSO data reopen : Ok
13: Roundtrip evsel->name : Ok
14: Parse sched tracepoints fields : Ok
15: syscalls:sys_enter_openat event fields : Ok
16: Setup struct perf_event_attr : Ok
17: Match and link multiple hists : Ok
18: 'import perf' in python : Ok
19: Breakpoint overflow signal handler : Ok
20: Breakpoint overflow sampling : Ok
21: Breakpoint accounting : Ok
22: Watchpoint :
22.1: Read Only Watchpoint : Skip
22.2: Write Only Watchpoint : Ok
22.3: Read / Write Watchpoint : Ok
22.4: Modify Watchpoint : Ok
23: Number of exit events of a simple workload : Ok
24: Software clock events period values : Ok
25: Object code reading : Ok
26: Sample parsing : Ok
27: Use a dummy software event to keep tracking : Ok
28: Parse with no sample_id_all bit set : Ok
29: Filter hist entries : Ok
30: Lookup mmap thread : Ok
31: Share thread mg : Ok
32: Sort output of hist entries : Ok
33: Cumulate child hist entries : Ok
34: Track with sched_switch : Ok
35: Filter fds with revents mask in a fdarray : Ok
36: Add fd to a fdarray, making it autogrow : Ok
37: kmod_path__parse : Ok
38: Thread map : Ok
39: LLVM search and compile :
39.1: Basic BPF llvm compile : Ok
39.2: kbuild searching : Ok
39.3: Compile source for BPF prologue generation : Ok
39.4: Compile source for BPF relocation : Ok
40: Session topology : Ok
41: BPF filter :
41.1: Basic BPF filtering : Ok
41.2: BPF pinning : Ok
41.3: BPF prologue generation : Ok
41.4: BPF relocation checker : Ok
42: Synthesize thread map : Ok
43: Remove thread map : Ok
44: Synthesize cpu map : Ok
45: Synthesize stat config : Ok
46: Synthesize stat : Ok
47: Synthesize stat round : Ok
48: Synthesize attr update : Ok
49: Event times : Ok
50: Read backward ring buffer : Ok
51: Print cpu map : Ok
52: Probe SDT events : Ok
53: is_printable_array : Ok
54: Print bitmap : Ok
55: perf hooks : Ok
56: builtin clang support : Skip (not compiled in)
57: unit_number__scnprintf : Ok
58: mem2node : Ok
59: time utils : Ok
60: map_groups__merge_in : Ok
61: x86 rdpmc : Ok
62: Convert perf time to TSC : Ok
63: DWARF unwind : Ok
64: x86 instruction decoder - new instructions : Ok
65: Intel PT packet decoder : Ok
66: x86 bp modify : Ok
67: probe libc's inet_pton & backtrace it with ping : Ok
68: Use vfs_getname probe to get syscall args filenames : Ok
69: Add vfs_getname probe to get syscall args filenames : Ok
70: Check open filename arg using perf trace + vfs_getname: Ok
71: Zstd perf.data compression/decompression : Ok
# $ time make -C tools/perf build-test
make: Entering directory '/home/acme/git/perf/tools/perf'
- tarpkg: ./tests/perf-targz-src-pkg .
make_no_gtk2_O: make NO_GTK2=1
make_no_libaudit_O: make NO_LIBAUDIT=1
make_cscope_O: make cscope
make_debug_O: make DEBUG=1
make_no_backtrace_O: make NO_BACKTRACE=1
make_no_newt_O: make NO_NEWT=1
make_no_libbpf_O: make NO_LIBBPF=1
make_util_map_o_O: make util/map.o
make_install_prefix_O: make install prefix=/tmp/krava
make_with_clangllvm_O: make LIBCLANGLLVM=1
make_no_libbionic_O: make NO_LIBBIONIC=1
make_no_libnuma_O: make NO_LIBNUMA=1
make_clean_all_O: make clean all
make_help_O: make help
make_no_libpython_O: make NO_LIBPYTHON=1
make_util_pmu_bison_o_O: make util/pmu-bison.o
make_no_libdw_dwarf_unwind_O: make NO_LIBDW_DWARF_UNWIND=1
make_install_bin_O: make install-bin
make_no_demangle_O: make NO_DEMANGLE=1
make_no_libperl_O: make NO_LIBPERL=1
make_no_scripts_O: make NO_LIBPYTHON=1 NO_LIBPERL=1
make_minimal_O: make NO_LIBPERL=1 NO_LIBPYTHON=1 NO_NEWT=1 NO_GTK2=1 NO_DEMANGLE=1 NO_LIBELF=1 NO_LIBUNWIND=1 NO_BACKTRACE=1 NO_LIBNUMA=1 NO_LIBAUDIT=1 NO_LIBBIONIC=1 NO_LIBDW_DWARF_UNWIND=1 NO_AUXTRACE=1 NO_LIBBPF=1 NO_LIBCRYPTO=1 NO_SDT=1 NO_JVMTI=1 NO_LIBZSTD=1 NO_LIBCAP=1
make_no_ui_O: make NO_NEWT=1 NO_SLANG=1 NO_GTK2=1
make_no_auxtrace_O: make NO_AUXTRACE=1
make_static_O: make LDFLAGS=-static
make_no_libelf_O: make NO_LIBELF=1
make_no_slang_O: make NO_SLANG=1
make_tags_O: make tags
make_no_libunwind_O: make NO_LIBUNWIND=1
make_doc_O: make doc
make_install_prefix_slash_O: make install prefix=/tmp/krava/
make_with_babeltrace_O: make LIBBABELTRACE=1
make_perf_o_O: make perf.o
make_install_O: make install
make_pure_O: make
OK
make: Leaving directory '/home/acme/git/perf/tools/perf'
$ Arnaldo Carvalho de Melo
maintainerPATCH 1/17 perf vendor events intel: Add Tremontx event file v1.02
From: Haiyan Song <[email protected]>
Add a Intel event file for perf.
Signed-off-by Haiyan Song <[email protected]>
Reviewed-by Kan Liang <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Andi Kleen <[email protected]>
Cc Jin Yao <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Peter Zijlstra <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/pmu-events/arch/x86/mapfile.csv | 1 + .../pmu-events/arch/x86/tremontx/cache.json | 111 +++++ .../arch/x86/tremontx/frontend.json | 26 ++ .../pmu-events/arch/x86/tremontx/memory.json | 26 ++ .../pmu-events/arch/x86/tremontx/other.json | 26 ++ .../arch/x86/tremontx/pipeline.json | 111 +++++ .../arch/x86/tremontx/uncore-memory.json | 73 +++ .../arch/x86/tremontx/uncore-other.json | 431 ++++++++++++++++++ .../arch/x86/tremontx/uncore-power.json | 11 + .../arch/x86/tremontx/virtual-memory.json | 86 ++++ 10 files changed, 902 insertions(+) create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/cache.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/frontend.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/memory.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/other.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/pipeline.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-memory.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-other.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/uncore-power.json create mode 100644 tools/perf/pmu-events/arch/x86/tremontx/virtual-memory.json
tools/perf/pmu-events/arch/x86/mapfile.csv +1 -0
@@ -35,4 +35,5 @@ GenuineIntel-6-55-[01234],v1,skylakex,core GenuineIntel-6-55-[56789ABCDEF],v1,cascadelakex,core GenuineIntel-6-7D,v1,icelake,core GenuineIntel-6-7E,v1,icelake,core +GenuineIntel-6-86,v1,tremontx,core AuthenticAMD-23-[[:xdigit:]]+,v1,amdfam17h,core
tools/perf/pmu-events/arch/x86/tremontx/cache.json new file +111 -0
112 lines not on this page · raw message
tools/perf/pmu-events/arch/x86/tremontx/frontend.json new file +26 -0
@@ -0,0 +1,26 @@ +[ + { + "CollectPEBSRecord": "2", + "PublicDescription": "Counts requests to the Instruction Cache (ICache) for one or more bytes in an ICache Line and that cache line is not in the ICache (miss). The event strives to count on a cache line basis, so that multiple accesses which miss in a single cache line count as one ICACHE.MISS. Specifically, the event counts when straight line code crosses the cache line boundary, or when a branch target is to a new line, and that cache line is not in the ICache.", + "EventCode": "0x80", + "Counter": "0,1,2,3", + "UMask": "0x2", + "PEBScounters": "0,1,2,3", + "EventName": "ICACHE.MISSES", + "PDIR_COUNTER": "na", + "SampleAfterValue": "200003", + "BriefDescription": "Counts requests to the Instruction Cache (ICache) for one or more bytes in a cache line and they do not hit in the ICache (miss)." + }, + { + "CollectPEBSRecord": "2", + "PublicDescription": "Counts requests to the Instruction Cache (ICache) for one or more bytes in an ICache Line. The event strives to count on a cache line basis, so that multiple fetches to a single cache line count as one ICACHE.ACCESS. Specifically, the event counts when accesses from straight line code crosses the cache line boundary, or when a branch target is to a new line.", + "EventCode": "0x80", + "Counter": "0,1,2,3", + "UMask": "0x3", + "PEBScounters": "0,1,2,3", + "EventName": "ICACHE.ACCESSES", + "PDIR_COUNTER": "na", + "SampleAfterValue": "200003", + "BriefDescription": "Counts requests to the Instruction Cache (ICache) for one or more bytes cache Line." + } +]
tools/perf/pmu-events/arch/x86/tremontx/memory.json new file +26 -0
@@ -0,0 +1,26 @@ +[ + { + "PublicDescription": "Offcore response can be programmed only with a specific pair of event select and counter MSR, and with specific event codes and predefine mask bit value in a dedicated MSR to specify attributes of the offcore transaction.", + "EventCode": "0XB7", + "MSRValue": "0x000000003F04000001", + "Counter": "0,1,2,3", + "UMask": "0x1", + "EventName": "OCR.DEMAND_DATA_RD.L3_MISS", + "MSRIndex": "0x1a6,0x1a7", + "SampleAfterValue": "100003", + "BriefDescription": "Counts demand data reads that was not supplied by the L3 cache.", + "Offcore": "1" + }, + { + "PublicDescription": "Offcore response can be programmed only with a specific pair of event select and counter MSR, and with specific event codes and predefine mask bit value in a dedicated MSR to specify attributes of the offcore transaction.", + "EventCode": "0XB7", + "MSRValue": "0x000000003F04000002", + "Counter": "0,1,2,3", + "UMask": "0x1", + "EventName": "OCR.DEMAND_RFO.L3_MISS", + "MSRIndex": "0x1a6,0x1a7", + "SampleAfterValue": "100003", + "BriefDescription": "Counts all demand reads for ownership (RFO) requests and software based prefetches for exclusive ownership (PREFETCHW) that was not supplied by the L3 cache.", + "Offcore": "1" + } +]
tools/perf/pmu-events/arch/x86/tremontx/other.json new file +26 -0
@@ -0,0 +1,26 @@ +[ + { + "PublicDescription": "Offcore response can be programmed only with a specific pair of event select and counter MSR, and with specific event codes and predefine mask bit value in a dedicated MSR to specify attributes of the offcore transaction.", + "EventCode": "0XB7", + "MSRValue": "0x000000000000010001", + "Counter": "0,1,2,3", + "UMask": "0x1", + "EventName": "OCR.DEMAND_DATA_RD.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "SampleAfterValue": "100003", + "BriefDescription": "Counts demand data reads that have any response type.", + "Offcore": "1" + }, + { + "PublicDescription": "Offcore response can be programmed only with a specific pair of event select and counter MSR, and with specific event codes and predefine mask bit value in a dedicated MSR to specify attributes of the offcore transaction.", + "EventCode": "0XB7", + "MSRValue": "0x000000000000010002", + "Counter": "0,1,2,3", + "UMask": "0x1", + "EventName": "OCR.DEMAND_RFO.ANY_RESPONSE", + "MSRIndex": "0x1a6,0x1a7", + "SampleAfterValue": "100003", + "BriefDescription": "Counts all demand reads for ownership (RFO) requests and software based prefetches for exclusive ownership (PREFETCHW) that have any response type.", + "Offcore": "1" + } +]
tools/perf/pmu-events/arch/x86/tremontx/pipeline.json new file +111 -0
112 lines not on this page · raw message
tools/perf/pmu-events/arch/x86/tremontx/uncore-memory.json new file +73 -0
73 lines not on this page · raw message
tools/perf/pmu-events/arch/x86/tremontx/uncore-other.json new file +431 -0
431 lines not on this page · raw message
tools/perf/pmu-events/arch/x86/tremontx/uncore-power.json new file +11 -0
@@ -0,0 +1,11 @@ +[ + { + "BriefDescription": "Clockticks of the power control unit (PCU)", + "Counter": "0,1,2,3", + "CounterType": "PGMABLE", + "EventName": "UNC_P_CLOCKTICKS", + "PerPkg": "1", + "PublicDescription": "Clockticks of the power control unit (PCU)", + "Unit": "PCU" + } +]
tools/perf/pmu-events/arch/x86/tremontx/virtual-memory.json new file +86 -0
87 lines not on this page · raw message
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 2/17 perf script: Allow specifying event to switch on processing of other events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Sometime we want to only consider events after something happens, so allow discarding events till such events is found, e.g.:
Record all scheduler tracepoints and the sys_enter_nanosleep syscall event for the 'sleep 1' workload:
# perf record -e sched:*,syscalls:sys_enter_nanosleep sleep 1 [ perf record: Woken up 31 times to write data ] [ perf record: Captured and wrote 0.032 MB perf.data (10 samples) ] #
So we have these events in the generated perf data file:
# perf evlist sched:sched_kthread_stop sched:sched_kthread_stop_ret sched:sched_waking sched:sched_wakeup sched:sched_wakeup_new sched:sched_switch sched:sched_migrate_task sched:sched_process_free sched:sched_process_exit sched:sched_wait_task sched:sched_process_wait sched:sched_process_fork sched:sched_process_exec sched:sched_stat_wait sched:sched_stat_sleep sched:sched_stat_iowait sched:sched_stat_blocked sched:sched_stat_runtime sched:sched_pi_setprio sched:sched_move_numa sched:sched_stick_numa sched:sched_swap_numa sched:sched_wake_idle_without_ipi syscalls:sys_enter_nanosleep # Tip: use 'perf evlist --trace-fields' to show fields for tracepoint events #
Then show all of the events that actually took place in this 'perf record' session:
# perf script
:13637 13637 [002] 108237.581529: sched:sched_waking: comm=perf pid=13638 prio=120 target_cpu=001
:13637 13637 [002] 108237.581537: sched:sched_wakeup: perf:13638 [120] success=1 CPU:001
sleep 13638 [001] 108237.581992: sched:sched_process_exec: filename=/usr/bin/sleep pid=13638 old_pid=13638
sleep 13638 [001] 108237.582286: syscalls:sys_enter_nanosleep: rqtp: 0x7fff1948ac40, rmtp: 0x00000000
sleep 13638 [001] 108237.582289: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=578104 [ns] vruntime=202889459556 [ns]
sleep 13638 [001] 108237.582291: sched:sched_switch: sleep:13638 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 108238.582428: sched:sched_waking: comm=sleep pid=13638 prio=120 target_cpu=001
swapper 0 [001] 108238.582458: sched:sched_wakeup: sleep:13638 [120] success=1 CPU:001
sleep 13638 [001] 108238.582698: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=173915 [ns] vruntime=202889633471 [ns]
sleep 13638 [001] 108238.582782: sched:sched_process_exit: comm=sleep pid=13638 prio=120
#Now lets see only the ones that took place after a certain "marker":
# perf script --switch-on syscalls:sys_enter_nanosleep
sleep 13638 [001] 108237.582289: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=578104 [ns] vruntime=202889459556 [ns]
sleep 13638 [001] 108237.582291: sched:sched_switch: sleep:13638 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 108238.582428: sched:sched_waking: comm=sleep pid=13638 prio=120 target_cpu=001
swapper 0 [001] 108238.582458: sched:sched_wakeup: sleep:13638 [120] success=1 CPU:001
sleep 13638 [001] 108238.582698: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=173915 [ns] vruntime=202889633471 [ns]
sleep 13638 [001] 108238.582782: sched:sched_process_exit: comm=sleep pid=13638 prio=120
#Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-script.txt | 3 +++ tools/perf/builtin-script.c | 26 ++++++++++++++++++++++++ 2 files changed, 29 insertions(+)
tools/perf/Documentation/perf-script.txt +3 -0
@@ -417,6 +417,9 @@ include::itrace.txt[] For itrace only show specified functions and their callees for itrace. Multiple functions can be separated by comma. +--switch-on EVENT_NAME:: + Only consider events after this event is found. + SEE ALSO -------- linkperf:perf-record[1], linkperf:perf-script-perl[1],
tools/perf/builtin-script.c +26 -0
@@ -1616,6 +1616,11 @@ static int perf_sample__fprintf_synth(struct perf_sample *sample,return 0 ; } +struct evswitch { +struct evsel *on; +bool discarding; +}; +struct perf_script {struct perf_tool tool;struct perf_session *session;
@@ -1628,6 +1633,7 @@ struct perf_script {bool show_bpf_events;bool allocated;bool per_event_dump; +struct evswitch evswitch;struct perf_cpu_map *cpus;struct perf_thread_map *threads;int name_width;
@@ -1805,6 +1811,13 @@ static void process_event(struct perf_script *script,if (!show_event (sample, evsel, thread, al))return ; +if (script->evswitch.on && script->evswitch.discarding) { +if (script->evswitch.on != evsel) +return ; + + script->evswitch.discarding = false; + } + ++es->samples; perf_sample__fprintf_start(sample, thread, evsel,
@@ -3395,6 +3408,7 @@ int cmd_script(int argc, const char **argv)struct utsname uts;char *script_path =NULL ;const char **__argv; +const char *event_switch_on =NULL ;int i, j, err =0 ;struct perf_script script = { .tool = {
@@ -3538,6 +3552,8 @@ int cmd_script(int argc, const char **argv) "file", "file saving guest os /proc/kallsyms"),OPT_STRING (0 ,"guestmodules" , &symbol_conf.default_guest_modules,"file" ,"file saving guest os /proc/modules" ), +OPT_STRING (0 ,"switch-on" , &event_switch_on, +"event" ,"Consider events from the first ocurrence of this event" ), OPT_END() };const char *const script_subcommands[] = {"record" ,"report" ,NULL };
@@ -3862,6 +3878,16 @@ int cmd_script(int argc, const char **argv) script.range_num); } +if (event_switch_on) { + script.evswitch.on =perf_evlist__find_evsel_by_str (session->evlist, event_switch_on); +if (script.evswitch.on ==NULL ) { +fprintf (stderr,"switch-on event not found (%s)\n" , event_switch_on); + err = -ENOENT ; + goto out_delete; + } + script.evswitch.discarding = true; + } + err =__cmd_script (&script);flush_scripting ();
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 3/17 perf script: Allow showing the --switch-on event
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
One may want to see the --switch-on event as well, allow for that, using the previous cset example:
# perf script --switch-on syscalls:sys_enter_nanosleep --show-on-off
sleep 13638 [001] 108237.582286: syscalls:sys_enter_nanosleep: rqtp: 0x7fff1948ac40, rmtp: 0x00000000
sleep 13638 [001] 108237.582289: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=578104 [ns] vruntime=202889459556 [ns]
sleep 13638 [001] 108237.582291: sched:sched_switch: sleep:13638 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 108238.582428: sched:sched_waking: comm=sleep pid=13638 prio=120 target_cpu=001
swapper 0 [001] 108238.582458: sched:sched_wakeup: sleep:13638 [120] success=1 CPU:001
sleep 13638 [001] 108238.582698: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=173915 [ns] vruntime=202889633471 [ns]
sleep 13638 [001] 108238.582782: sched:sched_process_exit: comm=sleep pid=13638 prio=120
#
# perf script --switch-on syscalls:sys_enter_nanosleep
sleep 13638 [001] 108237.582289: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=578104 [ns] vruntime=202889459556 [ns]
sleep 13638 [001] 108237.582291: sched:sched_switch: sleep:13638 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 108238.582428: sched:sched_waking: comm=sleep pid=13638 prio=120 target_cpu=001
swapper 0 [001] 108238.582458: sched:sched_wakeup: sleep:13638 [120] success=1 CPU:001
sleep 13638 [001] 108238.582698: sched:sched_stat_runtime: comm=sleep pid=13638 runtime=173915 [ns] vruntime=202889633471 [ns]
sleep 13638 [001] 108238.582782: sched:sched_process_exit: comm=sleep pid=13638 prio=120
#Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
Cc Adrian Hunter <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Cc Florian Weimer <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-script.txt | 3 +++ tools/perf/builtin-script.c | 6 ++++++ 2 files changed, 9 insertions(+)
tools/perf/Documentation/perf-script.txt +3 -0
@@ -420,6 +420,9 @@ include::itrace.txt[] --switch-on EVENT_NAME:: Only consider events after this event is found. +--show-on-off-events:: + Show the --switch-on event too. + SEE ALSO -------- linkperf:perf-record[1], linkperf:perf-script-perl[1],
tools/perf/builtin-script.c +6 -0
@@ -1619,6 +1619,7 @@ static int perf_sample__fprintf_synth(struct perf_sample *sample,struct evswitch {struct evsel *on;bool discarding; +bool show_on_off_events; };struct perf_script {
@@ -1816,6 +1817,9 @@ static void process_event(struct perf_script *script,return ; script->evswitch.discarding = false; + +if (!script->evswitch.show_on_off_events) +return ; } ++es->samples;
@@ -3554,6 +3558,8 @@ int cmd_script(int argc, const char **argv) "file", "file saving guest os /proc/modules"),OPT_STRING (0 ,"switch-on" , &event_switch_on,"event" ,"Consider events from the first ocurrence of this event" ), +OPT_BOOLEAN (0 ,"show-on-off-events" , &script.evswitch.show_on_off_events, +"Show the on/off switch events, used with --switch-on" ), OPT_END() };const char *const script_subcommands[] = {"record" ,"report" ,NULL };
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 4/17 perf script: Allow specifying event to switch off processing of other events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Counterpart of --switch-on:
# perf record -e sched:*,syscalls:sys_*_nanosleep sleep 1
[ perf record: Woken up 36 times to write data ]
[ perf record: Captured and wrote 0.032 MB perf.data (10 samples) ]
#
# perf script
:20918 20918 [002] 109866.143696: sched:sched_waking: comm=perf pid=20919 prio=120 target_cpu=001
:20918 20918 [002] 109866.143702: sched:sched_wakeup: perf:20919 [120] success=1 CPU:001
sleep 20919 [001] 109866.144081: sched:sched_process_exec: filename=/usr/bin/sleep pid=20919 old_pid=20919
sleep 20919 [001] 109866.144408: syscalls:sys_enter_nanosleep: rqtp: 0x7ffc2384fef0, rmtp: 0x00000000
sleep 20919 [001] 109866.144411: sched:sched_stat_runtime: comm=sleep pid=20919 runtime=521249 [ns] vruntime=202919398131 [n>
sleep 20919 [001] 109866.144412: sched:sched_switch: sleep:20919 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 109867.144568: sched:sched_waking: comm=sleep pid=20919 prio=120 target_cpu=001
swapper 0 [001] 109867.144586: sched:sched_wakeup: sleep:20919 [120] success=1 CPU:001
sleep 20919 [001] 109867.144614: syscalls:sys_exit_nanosleep: 0x0
sleep 20919 [001] 109867.144753: sched:sched_process_exit: comm=sleep pid=20919 prio=120
#
# perf script --switch-off syscalls:sys_exit_nanosleep
:20918 20918 [002] 109866.143696: sched:sched_waking: comm=perf pid=20919 prio=120 target_cpu=001
:20918 20918 [002] 109866.143702: sched:sched_wakeup: perf:20919 [120] success=1 CPU:001
sleep 20919 [001] 109866.144081: sched:sched_process_exec: filename=/usr/bin/sleep pid=20919 old_pid=20919
sleep 20919 [001] 109866.144408: syscalls:sys_enter_nanosleep: rqtp: 0x7ffc2384fef0, rmtp: 0x00000000
sleep 20919 [001] 109866.144411: sched:sched_stat_runtime: comm=sleep pid=20919 runtime=521249 [ns] vruntime=202919398131 [n>
sleep 20919 [001] 109866.144412: sched:sched_switch: sleep:20919 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 109867.144568: sched:sched_waking: comm=sleep pid=20919 prio=120 target_cpu=001
swapper 0 [001] 109867.144586: sched:sched_wakeup: sleep:20919 [120] success=1 CPU:001
sleep 20919 [001] 109867.144753: sched:sched_process_exit: comm=sleep pid=20919 prio=120
#
# perf script --switch-on syscalls:sys_enter_nanosleep --switch-off syscalls:sys_exit_nanosleep
sleep 20919 [001] 109866.144411: sched:sched_stat_runtime: comm=sleep pid=20919 runtime=521249 [ns] vruntime=202919398131 [n>
sleep 20919 [001] 109866.144412: sched:sched_switch: sleep:20919 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 109867.144568: sched:sched_waking: comm=sleep pid=20919 prio=120 target_cpu=001
swapper 0 [001] 109867.144586: sched:sched_wakeup: sleep:20919 [120] success=1 CPU:001
#
# perf script --switch-on syscalls:sys_enter_nanosleep --switch-off syscalls:sys_exit_nanosleep --show-on-off
sleep 20919 [001] 109866.144408: syscalls:sys_enter_nanosleep: rqtp: 0x7ffc2384fef0, rmtp: 0x00000000
sleep 20919 [001] 109866.144411: sched:sched_stat_runtime: comm=sleep pid=20919 runtime=521249 [ns] vruntime=202919398131 [n>
sleep 20919 [001] 109866.144412: sched:sched_switch: sleep:20919 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 109867.144568: sched:sched_waking: comm=sleep pid=20919 prio=120 target_cpu=001
swapper 0 [001] 109867.144586: sched:sched_wakeup: sleep:20919 [120] success=1 CPU:001
sleep 20919 [001] 109867.144614: syscalls:sys_exit_nanosleep: 0x0
#Now think about using this together with 'perf probe' to create custom on/off events in your app :-)
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-script.txt | 5 +++- tools/perf/builtin-script.c | 31 +++++++++++++++++++++--- 2 files changed, 32 insertions(+), 4 deletions(-)
tools/perf/Documentation/perf-script.txt +4 -1
@@ -420,8 +420,11 @@ include::itrace.txt[] --switch-on EVENT_NAME:: Only consider events after this event is found. +--switch-off EVENT_NAME:: + Stop considering events after this event is found. + --show-on-off-events:: - Show the --switch-on event too. + Show the --switch-on/off events too. SEE ALSO --------
tools/perf/builtin-script.c +28 -3
@@ -1617,7 +1617,8 @@ static int perf_sample__fprintf_synth(struct perf_sample *sample, }struct evswitch { -struct evsel *on; +struct evsel *on, + *off;bool discarding;bool show_on_off_events; };
@@ -1820,8 +1821,20 @@ static void process_event(struct perf_script *script,if (!script->evswitch.show_on_off_events)return ; + + goto print_it; } +if (script->evswitch.off && !script->evswitch.discarding) { +if (script->evswitch.off != evsel) + goto print_it; + + script->evswitch.discarding = true; + +if (!script->evswitch.show_on_off_events) +return ; + } +print_it: ++es->samples; perf_sample__fprintf_start(sample, thread, evsel,
@@ -3412,7 +3425,8 @@ int cmd_script(int argc, const char **argv)struct utsname uts;char *script_path =NULL ;const char **__argv; -const char *event_switch_on =NULL ; +const char *event_switch_on =NULL , + *event_switch_off =NULL ;int i, j, err =0 ;struct perf_script script = { .tool = {
@@ -3557,7 +3571,9 @@ int cmd_script(int argc, const char **argv)OPT_STRING (0 ,"guestmodules" , &symbol_conf.default_guest_modules,"file" ,"file saving guest os /proc/modules" ),OPT_STRING (0 ,"switch-on" , &event_switch_on, -"event" ,"Consider events from the first ocurrence of this event" ), +"event" ,"Consider events after the ocurrence of this event" ), +OPT_STRING (0 ,"switch-off" , &event_switch_off, +"event" ,"Stop considering events after the ocurrence of this event" ),OPT_BOOLEAN (0 ,"show-on-off-events" , &script.evswitch.show_on_off_events,"Show the on/off switch events, used with --switch-on" ),OPT_END ()
@@ -3894,6 +3910,15 @@ int cmd_script(int argc, const char **argv) script.evswitch.discarding = true; } +if (event_switch_off) { + script.evswitch.off =perf_evlist__find_evsel_by_str (session->evlist, event_switch_off); +if (script.evswitch.off ==NULL ) { +fprintf (stderr,"switch-off event not found (%s)\n" , event_switch_off); + err = -ENOENT ; + goto out_delete; + } + } + err =__cmd_script (&script);flush_scripting ();
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 5/17 perf evswitch: Move struct to a separate header to use in other tools
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Now that we see that the simple userspace-based "slicing" of events using delimiter events ("markers") works, lets move it to a separate header to make it available to other tools, next step will be having the switch on/off check done at the PERF_RECORD_SAMPLE processing function moved too.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/builtin-script.c | 8 +------- tools/perf/util/evswitch.h | 16 ++++++++++++++++ 2 files changed, 17 insertions(+), 7 deletions(-) create mode 100644 tools/perf/util/evswitch.h
tools/perf/builtin-script.c +1 -7
@@ -16,6 +16,7 @@#include "util/trace-event.h" #include "util/evlist.h" #include "util/evsel.h" +#include "util/evswitch.h" #include "util/sort.h" #include "util/data.h" #include "util/auxtrace.h"
@@ -1616,13 +1617,6 @@ static int perf_sample__fprintf_synth(struct perf_sample *sample,return 0 ; } -struct evswitch { -struct evsel *on, - *off; -bool discarding; -bool show_on_off_events; -}; -struct perf_script {struct perf_tool tool;struct perf_session *session;
tools/perf/util/evswitch.h new file +16 -0
@@ -0,0 +1,16 @@ +// SPDX-License-Identifier: GPL-2.0-only +// Copyright (C) 2019, Red Hat Inc, Arnaldo Carvalho de Melo <[email protected]> +#ifndef __PERF_EVSWITCH_H +#define __PERF_EVSWITCH_H 1 + +#include <stdbool.h> + +struct evsel ; + +struct evswitch { +struct evsel *on, *off; +bool discarding; +bool show_on_off_events; +}; + +#endif /* __PERF_EVSWITCH_H */
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 6/17 perf evswitch: Move switch logic to use in other tools
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Now other tools that want switching can use an evswitch for that, just set it up and add it to the PERF_RECORD_SAMPLE processing function.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/builtin-script.c | 23 ++--------------------- tools/perf/util/Build | 1 + tools/perf/util/evswitch.c | 31 +++++++++++++++++++++++++++++++ tools/perf/util/evswitch.h | 2 ++ 4 files changed, 36 insertions(+), 21 deletions(-) create mode 100644 tools/perf/util/evswitch.c
tools/perf/builtin-script.c +2 -21
@@ -1807,28 +1807,9 @@ static void process_event(struct perf_script *script,if (!show_event (sample, evsel, thread, al))return ; -if (script->evswitch.on && script->evswitch.discarding) { -if (script->evswitch.on != evsel) -return ; - - script->evswitch.discarding = false; - -if (!script->evswitch.show_on_off_events) -return ; - - goto print_it; - } - -if (script->evswitch.off && !script->evswitch.discarding) { -if (script->evswitch.off != evsel) - goto print_it; - - script->evswitch.discarding = true; +if (evswitch__discard (&script->evswitch, evsel)) +return ; -if (!script->evswitch.show_on_off_events) -return ; - } -print_it: ++es->samples; perf_sample__fprintf_start(sample, thread, evsel,
tools/perf/util/Build +1 -0
@@ -9,6 +9,7 @@ perf-y += event.o perf-y += evlist.o perf-y += evsel.o perf-y += evsel_fprintf.o +perf-y += evswitch.o perf-y += find_bit.o perf-y += get_current_dir_name.o perf-y += kallsyms.o
tools/perf/util/evswitch.c new file +31 -0
@@ -0,0 +1,31 @@ +// SPDX-License-Identifier: GPL-2.0-only +// Copyright (C) 2019, Red Hat Inc, Arnaldo Carvalho de Melo <[email protected]> + +#include "evswitch.h" + +bool evswitch__discard (struct evswitch *evswitch,struct evsel *evsel) +{ +if (evswitch->on && evswitch->discarding) { +if (evswitch->on != evsel) +return true; + + evswitch->discarding = false; + +if (!evswitch->show_on_off_events) +return true; + +return false; + } + +if (evswitch->off && !evswitch->discarding) { +if (evswitch->off != evsel) +return false; + + evswitch->discarding = true; + +if (!evswitch->show_on_off_events) +return true; + } + +return false; +}
tools/perf/util/evswitch.h +2 -0
@@ -13,4 +13,6 @@ struct evswitch { bool show_on_off_events; }; +bool evswitch__discard (struct evswitch *evswitch,struct evsel *evsel); +#endif /* __PERF_EVSWITCH_H */
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 7/17 perf evswitch: Add the names of on/off events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
So that we can have macros for the OPT_ entries and also for finding those in an evlist, this way other tools will use this very easily.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/builtin-script.c | 18 ++++++++---------- tools/perf/util/evswitch.h | 1 + 2 files changed, 9 insertions(+), 10 deletions(-)
tools/perf/builtin-script.c +8 -10
@@ -3400,8 +3400,6 @@ int cmd_script(int argc, const char **argv)struct utsname uts;char *script_path =NULL ;const char **__argv; -const char *event_switch_on =NULL , - *event_switch_off =NULL ;int i, j, err =0 ;struct perf_script script = { .tool = {
@@ -3545,9 +3543,9 @@ int cmd_script(int argc, const char **argv) "file", "file saving guest os /proc/kallsyms"),OPT_STRING (0 ,"guestmodules" , &symbol_conf.default_guest_modules,"file" ,"file saving guest os /proc/modules" ), -OPT_STRING (0 ,"switch-on" , &event_switch_on, +OPT_STRING (0 ,"switch-on" , &script.evswitch.on_name,"event" ,"Consider events after the ocurrence of this event" ), -OPT_STRING (0 ,"switch-off" , &event_switch_off, +OPT_STRING (0 ,"switch-off" , &script.evswitch.off_name,"event" ,"Stop considering events after the ocurrence of this event" ),OPT_BOOLEAN (0 ,"show-on-off-events" , &script.evswitch.show_on_off_events, "Show the on/off switch events, used with --switch-on"),
@@ -3875,20 +3873,20 @@ int cmd_script(int argc, const char **argv) script.range_num); } -if (event_switch_on) { - script.evswitch.on =perf_evlist__find_evsel_by_str (session->evlist, event_switch_on); +if (script.evswitch.on_name) { + script.evswitch.on =perf_evlist__find_evsel_by_str (session->evlist, script.evswitch.on_name);if (script.evswitch.on ==NULL ) { -fprintf (stderr,"switch-on event not found (%s)\n" , event_switch_on); +fprintf (stderr,"switch-on event not found (%s)\n" , script.evswitch.on_name); err = -ENOENT ; goto out_delete; } script.evswitch.discarding = true; } -if (event_switch_off) { - script.evswitch.off =perf_evlist__find_evsel_by_str (session->evlist, event_switch_off); +if (script.evswitch.off_name) { + script.evswitch.off =perf_evlist__find_evsel_by_str (session->evlist, script.evswitch.off_name);if (script.evswitch.off ==NULL ) { -fprintf (stderr,"switch-off event not found (%s)\n" , event_switch_off); +fprintf (stderr,"switch-off event not found (%s)\n" , script.evswitch.off_name); err = -ENOENT ; goto out_delete; }
tools/perf/util/evswitch.h +1 -0
@@ -9,6 +9,7 @@ struct evsel;struct evswitch {struct evsel *on, *off; +const char *on_name, *off_name;bool discarding;bool show_on_off_events; };
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 8/17 perf evswitch: Introduce OPTS_EVSWITCH() for cmd line processing
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
All tools will want those, so provide a convenient way to get them.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/builtin-script.c | 7 +------ tools/perf/util/evswitch.h | 8 ++++++++ 2 files changed, 9 insertions(+), 6 deletions(-)
tools/perf/builtin-script.c +1 -6
@@ -3543,12 +3543,7 @@ int cmd_script(int argc, const char **argv) "file", "file saving guest os /proc/kallsyms"),OPT_STRING (0 ,"guestmodules" , &symbol_conf.default_guest_modules,"file" ,"file saving guest os /proc/modules" ), -OPT_STRING (0 ,"switch-on" , &script.evswitch.on_name, -"event" ,"Consider events after the ocurrence of this event" ), -OPT_STRING (0 ,"switch-off" , &script.evswitch.off_name, -"event" ,"Stop considering events after the ocurrence of this event" ), -OPT_BOOLEAN (0 ,"show-on-off-events" , &script.evswitch.show_on_off_events, -"Show the on/off switch events, used with --switch-on" ), +OPTS_EVSWITCH (&script.evswitch), OPT_END() };const char *const script_subcommands[] = {"record" ,"report" ,NULL };
tools/perf/util/evswitch.h +8 -0
@@ -16,4 +16,12 @@ struct evswitch {bool evswitch__discard (struct evswitch *evswitch,struct evsel *evsel); +#define OPTS_EVSWITCH (evswitch) \ + OPT_STRING(0, "switch-on", &(evswitch)->on_name, \ + "event", "Consider events after the ocurrence of this event"), \ + OPT_STRING(0, "switch-off", &(evswitch)->off_name, \ + "event", "Stop considering events after the ocurrence of this event"), \ + OPT_BOOLEAN(0, "show-on-off-events", &(evswitch)->show_on_off_events, \ + "Show the on/off switch events, used with --switch-on and --switch-off") +#endif /* __PERF_EVSWITCH_H */
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 9/17 perf evswitch: Introduce init() method to set the on/off evsels from the command line
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Another step in having all the boilerplate in just one place to then use in the other tools.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/builtin-script.c | 21 +++------------------ tools/perf/util/evswitch.c | 23 +++++++++++++++++++++++ tools/perf/util/evswitch.h | 4 ++++ 3 files changed, 30 insertions(+), 18 deletions(-)
tools/perf/builtin-script.c +3 -18
@@ -3868,24 +3868,9 @@ int cmd_script(int argc, const char **argv) script.range_num); } -if (script.evswitch.on_name) { - script.evswitch.on =perf_evlist__find_evsel_by_str (session->evlist, script.evswitch.on_name); -if (script.evswitch.on ==NULL ) { -fprintf (stderr,"switch-on event not found (%s)\n" , script.evswitch.on_name); - err = -ENOENT ; - goto out_delete; - } - script.evswitch.discarding = true; - } - -if (script.evswitch.off_name) { - script.evswitch.off =perf_evlist__find_evsel_by_str (session->evlist, script.evswitch.off_name); -if (script.evswitch.off ==NULL ) { -fprintf (stderr,"switch-off event not found (%s)\n" , script.evswitch.off_name); - err = -ENOENT ; - goto out_delete; - } - } + err =evswitch__init (&script.evswitch, session->evlist, stderr); +if (err) + goto out_delete; err =__cmd_script (&script);
tools/perf/util/evswitch.c +23 -0
@@ -2,6 +2,7 @@// Copyright (C) 2019, Red Hat Inc, Arnaldo Carvalho de Melo <[email protected]> #include "evswitch.h" +#include "evlist.h" bool evswitch__discard (struct evswitch *evswitch,struct evsel *evsel) {
@@ -29,3 +30,25 @@ bool evswitch__discard(struct evswitch *evswitch, struct evsel *evsel)return false; } + +int evswitch__init (struct evswitch *evswitch,struct evlist *evlist,FILE *fp) +{ +if (evswitch->on_name) { + evswitch->on =perf_evlist__find_evsel_by_str (evlist, evswitch->on_name); +if (evswitch->on ==NULL ) { +fprintf (fp,"switch-on event not found (%s)\n" , evswitch->on_name); +return -ENOENT ; + } + evswitch->discarding = true; + } + +if (evswitch->off_name) { + evswitch->off =perf_evlist__find_evsel_by_str (evlist, evswitch->off_name); +if (evswitch->off ==NULL ) { +fprintf (fp,"switch-off event not found (%s)\n" , evswitch->off_name); +return -ENOENT ; + } + } + +return 0 ; +}
tools/perf/util/evswitch.h +4 -0
@@ -4,8 +4,10 @@#define __PERF_EVSWITCH_H 1#include <stdbool.h> +#include <stdio.h> struct evsel ; +struct evlist ;struct evswitch {struct evsel *on, *off;
@@ -14,6 +16,8 @@ struct evswitch {bool show_on_off_events; }; +int evswitch__init (struct evswitch *evswitch,struct evlist *evlist,FILE *fp); +bool evswitch__discard (struct evswitch *evswitch,struct evsel *evsel);#define OPTS_EVSWITCH (evswitch) \
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 10/17 perf evswitch: Move enoent error message printing to separate function
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Allows adding hints there, will be done in followup patch.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/evswitch.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)
tools/perf/util/evswitch.c +7 -2
@@ -31,12 +31,17 @@ bool evswitch__discard(struct evswitch *evswitch, struct evsel *evsel)return false; } +static int evswitch__fprintf_enoent (FILE *fp,const char *evtype,const char *evname) +{ +return fprintf (fp,"ERROR: switch-%s event not found (%s)\n" , evtype, evname); +} +int evswitch__init (struct evswitch *evswitch,struct evlist *evlist,FILE *fp) {if (evswitch->on_name) { evswitch->on =perf_evlist__find_evsel_by_str (evlist, evswitch->on_name);if (evswitch->on ==NULL ) { -fprintf (fp,"switch-on event not found (%s)\n" , evswitch->on_name); +evswitch__fprintf_enoent (fp,"on" , evswitch->on_name);return -ENOENT ; } evswitch->discarding = true;
@@ -45,7 +50,7 @@ int evswitch__init(struct evswitch *evswitch, struct evlist *evlist, FILE *fp)if (evswitch->off_name) { evswitch->off =perf_evlist__find_evsel_by_str (evlist, evswitch->off_name);if (evswitch->off ==NULL ) { -fprintf (fp,"switch-off event not found (%s)\n" , evswitch->off_name); +evswitch__fprintf_enoent (fp,"off" , evswitch->off_name);return -ENOENT ; } }
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 11/17 perf evswitch: Add hint when not finding specified on/off events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
If the user specifies a on or off switch event and it isn't in the perf.data file, provide a hint about how to see the events in the perf.data evlist:
# perf script --switch-on=syscall:sys_enter_nanosleep --switch-off=syscalls:sys_exit_nanosleep
ERROR: event_on event not found (syscall:sys_enter_nanosleep)
HINT: use 'perf evlist' to see the available event names
#
# perf evlist
sched:sched_kthread_stop
sched:sched_kthread_stop_ret
sched:sched_waking
sched:sched_wakeup
sched:sched_wakeup_new
sched:sched_switch
sched:sched_migrate_task
sched:sched_process_free
sched:sched_process_exit
sched:sched_wait_task
sched:sched_process_wait
sched:sched_process_fork
sched:sched_process_exec
sched:sched_stat_wait
sched:sched_stat_sleep
sched:sched_stat_iowait
sched:sched_stat_blocked
sched:sched_stat_runtime
sched:sched_pi_setprio
sched:sched_move_numa
sched:sched_stick_numa
sched:sched_swap_numa
sched:sched_wake_idle_without_ipi
syscalls:sys_enter_clock_nanosleep
syscalls:sys_exit_clock_nanosleep
syscalls:sys_enter_nanosleep
syscalls:sys_exit_nanosleep
# Tip: use 'perf evlist --trace-fields' to show fields for tracepoint events
#
# perf script --switch-on=syscalls:sys_enter_nanosleep --switch-off=syscalls:sys_exit_nanosleep
sleep 20919 [001] 109866.144411: sched:sched_stat_runtime: comm=sleep pid=20919 runtime=521249 [ns] vruntime=202919398131 [ns]
sleep 20919 [001] 109866.144412: sched:sched_switch: sleep:20919 [120] S ==> swapper/1:0 [120]
swapper 0 [001] 109867.144568: sched:sched_waking: comm=sleep pid=20919 prio=120 target_cpu=001
swapper 0 [001] 109867.144586: sched:sched_wakeup: sleep:20919 [120] success=1 CPU:001
#Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/evswitch.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
tools/perf/util/evswitch.c +3 -1
@@ -33,7 +33,9 @@ bool evswitch__discard(struct evswitch *evswitch, struct evsel *evsel)static int evswitch__fprintf_enoent (FILE *fp,const char *evtype,const char *evname) { -return fprintf (fp,"ERROR: switch-%s event not found (%s)\n" , evtype, evname); +int printed =fprintf (fp,"ERROR: switch-%s event not found (%s)\n" , evtype, evname); + +return printed +=fprintf (fp,"HINT: use 'perf evlist' to see the available event names\n" ); }int evswitch__init (struct evswitch *evswitch,struct evlist *evlist,FILE *fp)
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 12/17 perf trace: Add --switch-on/--switch-off events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Just like with 'perf script':
# perf trace -e sched:*,syscalls:*sleep* sleep 1
0.000 :28345/28345 sched:sched_waking:comm=perf pid=28346 prio=120 target_cpu=005
0.005 :28345/28345 sched:sched_wakeup:perf:28346 [120] success=1 CPU:005
0.383 sleep/28346 sched:sched_process_exec:filename=/usr/bin/sleep pid=28346 old_pid=28346
0.613 sleep/28346 sched:sched_stat_runtime:comm=sleep pid=28346 runtime=607375 [ns] vruntime=23289041218 [ns]
0.689 sleep/28346 syscalls:sys_enter_nanosleep:rqtp: 0x7ffc491789b0
0.693 sleep/28346 sched:sched_stat_runtime:comm=sleep pid=28346 runtime=72021 [ns] vruntime=23289113239 [ns]
0.694 sleep/28346 sched:sched_switch:sleep:28346 [120] S ==> swapper/5:0 [120]
1000.787 :0/0 sched:sched_waking:comm=sleep pid=28346 prio=120 target_cpu=005
1000.824 :0/0 sched:sched_wakeup:sleep:28346 [120] success=1 CPU:005
1000.908 sleep/28346 syscalls:sys_exit_nanosleep:0x0
1001.218 sleep/28346 sched:sched_process_exit:comm=sleep pid=28346 prio=120
# perf trace -e sched:*,syscalls:*sleep* --switch-on=syscalls:sys_enter_nanosleep sleep 1
0.000 sleep/28349 sched:sched_stat_runtime:comm=sleep pid=28349 runtime=603036 [ns] vruntime=23873537697 [ns]
0.001 sleep/28349 sched:sched_switch:sleep:28349 [120] S ==> swapper/4:0 [120]
1000.392 :0/0 sched:sched_waking:comm=sleep pid=28349 prio=120 target_cpu=004
1000.443 :0/0 sched:sched_wakeup:sleep:28349 [120] success=1 CPU:004
1000.540 sleep/28349 syscalls:sys_exit_nanosleep:0x0
1000.852 sleep/28349 sched:sched_process_exit:comm=sleep pid=28349 prio=120
# perf trace -e sched:*,syscalls:*sleep* --switch-on=syscalls:sys_enter_nanosleep --switch-off=syscalls:sys_exit_nanosleep sleep 1
0.000 sleep/28352 sched:sched_stat_runtime:comm=sleep pid=28352 runtime=610543 [ns] vruntime=24811686681 [ns]
0.001 sleep/28352 sched:sched_switch:sleep:28352 [120] S ==> swapper/0:0 [120]
1000.397 :0/0 sched:sched_waking:comm=sleep pid=28352 prio=120 target_cpu=000
1000.440 :0/0 sched:sched_wakeup:sleep:28352 [120] success=1 CPU:000
#
# perf trace -e sched:*,syscalls:*sleep* --switch-on=syscalls:sys_enter_nanosleep --switch-off=syscalls:sys_exit_nanosleep --show-on-off sleep 1
0.000 sleep/28367 syscalls:sys_enter_nanosleep:rqtp: 0x7fffd1a25fc0
0.004 sleep/28367 sched:sched_stat_runtime:comm=sleep pid=28367 runtime=628760 [ns] vruntime=22170052672 [ns]
0.005 sleep/28367 sched:sched_switch:sleep:28367 [120] S ==> swapper/2:0 [120]
1000.367 :0/0 sched:sched_waking:comm=sleep pid=28367 prio=120 target_cpu=002
1000.412 :0/0 sched:sched_wakeup:sleep:28367 [120] success=1 CPU:002
1000.512 sleep/28367 syscalls:sys_exit_nanosleep:0x0
#Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-trace.txt | 9 +++++++++ tools/perf/builtin-trace.c | 10 ++++++++++ 2 files changed, 19 insertions(+)
tools/perf/Documentation/perf-trace.txt +9 -0
@@ -176,6 +176,15 @@ the thread executes on the designated CPUs. Default is to monitor all CPUs. only at exit time or when a syscall is interrupted, i.e. in those cases this option is equivalent to the number of lines printed. +--switch-on EVENT_NAME:: + Only consider events after this event is found. + +--switch-off EVENT_NAME:: + Stop considering events after this event is found. + +--show-on-off-events:: + Show the --switch-on/off events too. + --max-stack:: Set the stack depth limit when parsing the callchain, anything beyond the specified depth will be ignored. Note that at this point
tools/perf/builtin-trace.c +10 -0
@@ -27,6 +27,7 @@#include "util/env.h" #include "util/event.h" #include "util/evlist.h" +#include "util/evswitch.h" #include <subcmd/exec-cmd.h> #include "util/machine.h" #include "util/map.h"
@@ -106,6 +107,7 @@ struct trace {unsigned long nr_events;unsigned long nr_events_printed;unsigned long max_events; +struct evswitch evswitch;struct strlist *ev_qualifier;struct {size_t nr;
@@ -2680,6 +2682,9 @@ static void trace__handle_event(struct trace *trace, union perf_event *event, streturn ; } +if (evswitch__discard (&trace->evswitch, evsel)) +return ; +trace__set_base_time (trace, evsel, sample); if (evsel->core.attr.type == PERF_TYPE_TRACEPOINT &&
@@ -4157,6 +4162,7 @@ int cmd_trace(int argc, const char **argv)OPT_UINTEGER ('D' ,"delay" , &trace.opts.initial_delay,"ms to wait before starting measurement after program " "start" ), +OPTS_EVSWITCH (&trace.evswitch), OPT_END() };bool __maybe_unused max_stack_user_set = true;
@@ -4380,6 +4386,10 @@ int cmd_trace(int argc, const char **argv) } } + err =evswitch__init (&trace.evswitch, trace.evlist, stderr); +if (err) + goto out_close; + err =target__validate (&trace.opts.target);if (err) {target__strerror (&trace.opts.target, err, bf,sizeof (bf));
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 13/17 perf top: Add --switch-on/--switch-off events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Just like 'perf trace' and 'perf script', should be useful for instance to only consider samples after the initialization phase of some workload.
The man page has some examples and considerations about its current interface, that still doesn't handle the on/off events in a special way, behaving just like when multiple events are specified, i.e.:
- In non-group mode (when the event list is not enclosed in {}) show a a menu to allow choosing which event the user wants to see in the histograms browser
- In group mode, be it using {} or asking for --group, show one column per event.
Try for instance:
# perf top -e '{cycles,instructions,probe:icmp_rcv}' --switch-on=probe:icmp_rcvReplace probe:icmp_rcv, that I put in place using:
# perf probe icmp_rcv:59
To hit when broadcast packets arrive, with a probe installed after an initialization phase is over or after some other point of interest, some garbage collection, etc, and also use --switch-off, for instance, on a probe installed after said garbage collection is over.
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-top.txt | 38 +++++++++++++++++++++++++++ tools/perf/builtin-top.c | 10 ++++++- tools/perf/util/top.h | 2 ++ 3 files changed, 49 insertions(+), 1 deletion(-)
tools/perf/Documentation/perf-top.txt +38 -0
@@ -266,6 +266,44 @@ Default is to monitor all CPUS. Record events of type PERF_RECORD_NAMESPACES and display it with the 'cgroup_id' sort key. +--switch-on EVENT_NAME:: + Only consider events after this event is found. + + E.g.: + + Find out where broadcast packets are handled + + perf probe -L icmp_rcv + + Insert a probe there: + + perf probe icmp_rcv:59 + + Start perf top and ask it to only consider the cycles events when a + broadcast packet arrives This will show a menu with two entries and + will start counting when a broadcast packet arrives: + + perf top -e cycles,probe:icmp_rcv --switch-on=probe:icmp_rcv + + Alternatively one can ask for --group and then two overhead columns + will appear, the first for cycles and the second for the switch-on event. + + perf top --group -e cycles,probe:icmp_rcv --switch-on=probe:icmp_rcv + + This may be interesting to measure a workload only after some initialization + phase is over, i.e. insert a perf probe at that point and use the above + examples replacing probe:icmp_rcv with the just-after-init probe. + +--switch-off EVENT_NAME:: + Stop considering events after this event is found. + +--show-on-off-events:: + Show the --switch-on/off events too. This has no effect in 'perf top' now + but probably we'll make the default not to show the switch-on/off events + on the --group mode and if there is only one event besides the off/on ones, + go straight to the histogram browser, just like 'perf top' with no events + explicitely specified does. + INTERACTIVE PROMPTING KEYS --------------------------
tools/perf/builtin-top.c +9 -1
@@ -1148,8 +1148,11 @@ static int deliver_event(struct ordered_events *qe, evsel =perf_evlist__id2evsel (session->evlist, sample.id);assert (evsel !=NULL ); -if (event->header.type ==PERF_RECORD_SAMPLE ) +if (event->header.type ==PERF_RECORD_SAMPLE ) { +if (evswitch__discard (&top->evswitch, evsel)) +return 0 ; ++top->samples; + }switch (sample.cpumode) {case PERF_RECORD_MISC_USER :
@@ -1534,6 +1537,7 @@ int cmd_top(int argc, const char **argv) "number of thread to run event synthesize"),OPT_BOOLEAN (0 ,"namespaces" , &opts->record_namespaces,"Record namespaces events" ), +OPTS_EVSWITCH (&top.evswitch), OPT_END() };struct evlist *sb_evlist =NULL ;
@@ -1567,6 +1571,10 @@ int cmd_top(int argc, const char **argv) goto out_delete_evlist; } + status =evswitch__init (&top.evswitch, top.evlist, stderr); +if (status) + goto out_delete_evlist; +if (symbol_conf.report_hierarchy) {/* disable incompatible options */ symbol_conf.event_group = false;
tools/perf/util/top.h +2 -0
@@ -3,6 +3,7 @@#define __PERF_TOP_H 1#include "tool.h" +#include "evswitch.h" #include "annotate.h" #include <linux/types.h> #include <stddef.h>
@@ -18,6 +19,7 @@ struct perf_top {struct evlist *evlist;struct record_opts record_opts;struct annotation_options annotation_opts; +struct evswitch evswitch;/* * Symbols will be added here in perf_event__process_sample and will * get out after decayed.
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 14/17 perf report: Add --switch-on/--switch-off events
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Since 'perf top' shares the histogram browser with 'perf report', then the same explanation in the previous cset applies.
An additional example uses a pair of SDT events available for systemtap:
# perf probe --exec=/usr/bin/stap '%*:*'
Added new events:
sdt_stap:benchmark__thread__start (on %* in /usr/bin/stap)
sdt_stap:benchmark (on %* in /usr/bin/stap)
sdt_stap:benchmark__thread__end (on %* in /usr/bin/stap)
sdt_stap:pass6__start (on %* in /usr/bin/stap)
sdt_stap:pass6__end (on %* in /usr/bin/stap)
sdt_stap:pass5__start (on %* in /usr/bin/stap)
sdt_stap:pass5__end (on %* in /usr/bin/stap)
sdt_stap:pass0__start (on %* in /usr/bin/stap)
sdt_stap:pass0__end (on %* in /usr/bin/stap)
sdt_stap:pass1a__start (on %* in /usr/bin/stap)
sdt_stap:pass1b__start (on %* in /usr/bin/stap)
sdt_stap:pass1__end (on %* in /usr/bin/stap)
sdt_stap:pass2__start (on %* in /usr/bin/stap)
sdt_stap:pass2__end (on %* in /usr/bin/stap)
sdt_stap:pass3__start (on %* in /usr/bin/stap)
sdt_stap:pass3__end (on %* in /usr/bin/stap)
sdt_stap:pass4__start (on %* in /usr/bin/stap)
sdt_stap:pass4__end (on %* in /usr/bin/stap)
sdt_stap:benchmark__start (on %* in /usr/bin/stap)
sdt_stap:benchmark__end (on %* in /usr/bin/stap)
sdt_stap:cache__get (on %* in /usr/bin/stap)
sdt_stap:cache__clean (on %* in /usr/bin/stap)
sdt_stap:cache__add__module (on %* in /usr/bin/stap)
sdt_stap:cache__add__source (on %* in /usr/bin/stap)
sdt_stap:stap_system__complete (on %* in /usr/bin/stap)
sdt_stap:stap_system__start (on %* in /usr/bin/stap)
sdt_stap:stap_system__spawn (on %* in /usr/bin/stap)
sdt_stap:stap_system__fork (on %* in /usr/bin/stap)
sdt_stap:intern_string (on %* in /usr/bin/stap)
sdt_stap:client__start (on %* in /usr/bin/stap)
sdt_stap:client__end (on %* in /usr/bin/stap)You can now use it in all perf tools, such as:
perf record -e sdt_stap:client__end -aR sleep 1
#
From these we're use the two below to run systemtap's test suite:
# perf record -e sdt_stap:pass2__*,cycles:P make installcheck > /dev/null
^C[ perf record: Woken up 8 times to write data ]
[ perf record: Captured and wrote 2.691 MB perf.data (39638 samples) ]
Terminated
# perf script | grep sdt_stap
stap 28979 [000] 19424.302660: sdt_stap:pass2__start: (561b9a537de3) arg1=140730364262544
stap 28979 [000] 19424.333083: sdt_stap:pass2__end: (561b9a53a9e1) arg1=140730364262544
stap 29045 [006] 19424.933460: sdt_stap:pass2__start: (563edddcede3) arg1=140722674883152
stap 29045 [006] 19424.963794: sdt_stap:pass2__end: (563edddd19e1) arg1=140722674883152
# perf script | grep cycles | wc -l
39634
#Looking at the whole perf.data file:
[root@quaco testsuite]# perf report | grep cycles:P -A25 # Samples: 39K of event 'cycles:P'
# Event count (approx.): 34044267368
#
# Overhead Command Shared Object Symbol
# ........ ....... .................... ................................
#
3.50% cc1 cc1 [.] ht_lookup_with_hash
3.04% cc1 cc1 [.] _cpp_lex_token
2.11% cc1 cc1 [.] ggc_internal_alloc
1.83% cc1 cc1 [.] cpp_get_token_with_location
1.68% cc1 libc-2.29.so [.] _int_malloc
1.41% cc1 cc1 [.] linemap_position_for_column
1.25% cc1 cc1 [.] ggc_internal_cleared_alloc
1.20% cc1 cc1 [.] c_lex_with_flags
1.18% cc1 cc1 [.] get_combined_adhoc_loc
1.05% cc1 libc-2.29.so [.] malloc
1.01% cc1 libc-2.29.so [.] _int_free
0.96% stap stap [.] std::_Hashtable<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, std::__detail::_Identity, std::equal_to<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > >, stringtable_hash, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<true, true, true> >::_M_insert<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__detail::_AllocNode<std::allocator<std::__detail::_Hash_node<std::_
_cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, true> > > >
0.78% stap stap [.] lexer::scan
0.74% cc1 cc1 [.] _cpp_lex_direct
0.70% cc1 cc1 [.] pop_scope
0.70% cc1 cc1 [.] c_parser_declspecs
0.69% stap libc-2.29.so [.] _int_malloc
0.68% cc1 cc1 [.] htab_find_slot
0.68% cc1 [kernel.vmlinux] [k] prepare_exit_to_usermode
0.64% cc1 [kernel.vmlinux] [k] clear_page_erms[root@quaco testsuite]#
And now only what happens in slices demarcated by those start/end SDT events:
[root@quaco testsuite]# perf report --switch-on=sdt_stap:pass2__start --switch-off=sdt_stap:pass2__end | grep cycles:P -A100
# Samples: 240 of event 'cycles:P'
# Event count (approx.): 206491934
#
# Overhead Command Shared Object Symbol
# ........ ....... ................... ................................................
#
38.99% stap stap [.] systemtap_session::register_library_aliases
19.47% stap stap [.] match_key::operator<
15.01% stap libc-2.29.so [.] __memcmp_avx2_movbe
5.19% stap libc-2.29.so [.] _int_malloc
2.50% stap libstdc++.so.6.0.26 [.] std::_Rb_tree_insert_and_rebalance
2.30% stap stap [.] match_node::build_no_more
2.07% stap libc-2.29.so [.] malloc
1.66% stap stap [.] std::_Rb_tree<match_key, std::pair<match_key const, match_node*>, std::_Select1st<std::pair<match_key const, match_node*> >, std::less<match_key>, std::allocator<std::pair<match_key const, match_node*> > >::find
1.66% stap stap [.] match_node::bind
1.58% stap [kernel.vmlinux] [k] prepare_exit_to_usermode
1.17% stap [kernel.vmlinux] [k] native_irq_return_iret
0.87% stap stap [.] 0x0000000000032ec4
0.77% stap libstdc++.so.6.0.26 [.] std::_Rb_tree_increment
0.47% stap stap [.] std::vector<derived_probe_builder*, std::allocator<derived_probe_builder*> >::_M_realloc_insert<derived_probe_builder* const&>
0.47% stap [kernel.vmlinux] [k] get_page_from_freelist
0.47% stap [kernel.vmlinux] [k] swapgs_restore_regs_and_return_to_usermode
0.47% stap [kernel.vmlinux] [k] do_user_addr_fault
0.46% stap [kernel.vmlinux] [k] __pagevec_lru_add_fn
0.46% stap stap [.] std::_Rb_tree<match_key, std::pair<match_key const, match_node*>, std::_Select1st<std::pair<match_key const, match_node*> >, std::less<match_key>, std::allocator<std::pair<match_key const, match_node*> > >::_M_emplace_unique<std::pair<match_key, match_node*> >
0.42% stap libstdc++.so.6.0.26 [.] 0x00000000000c18fa
0.40% stap [kernel.vmlinux] [k] interrupt_entry
0.40% stap [kernel.vmlinux] [k] update_load_avg
0.40% stap [kernel.vmlinux] [k] __intel_pmu_disable_all
0.40% stap [kernel.vmlinux] [k] clear_page_erms
0.39% stap [kernel.vmlinux] [k] __mod_node_page_state
0.39% stap [kernel.vmlinux] [k] error_entry
0.39% stap [kernel.vmlinux] [k] sync_regs
0.38% stap [kernel.vmlinux] [k] __handle_mm_fault
0.38% stap stap [.] derive_probes #
# (Tip: System-wide collection from all CPUs: perf record -a) #
[root@quaco testsuite]#
Cc Adrian Hunter <[email protected]>
Cc Florian Weimer <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc William Cohen <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/Documentation/perf-report.txt | 17 +++++++++++++++++ tools/perf/builtin-report.c | 10 ++++++++++ 2 files changed, 27 insertions(+)
tools/perf/Documentation/perf-report.txt +17 -0
@@ -438,6 +438,23 @@ OPTIONS perf report --time 0%-10%,30%-40% +--switch-on EVENT_NAME:: + Only consider events after this event is found. + + This may be interesting to measure a workload only after some initialization + phase is over, i.e. insert a perf probe at that point and then using this + option with that probe. + +--switch-off EVENT_NAME:: + Stop considering events after this event is found. + +--show-on-off-events:: + Show the --switch-on/off events too. This has no effect in 'perf report' now + but probably we'll make the default not to show the switch-on/off events + on the --group mode and if there is only one event besides the off/on ones, + go straight to the histogram browser, just like 'perf report' with no events + explicitely specified does. + --itrace:: Options for decoding instruction tracing data. The options are:
tools/perf/builtin-report.c +10 -0
@@ -25,6 +25,7 @@#include "util/debug.h" #include "util/evlist.h" #include "util/evsel.h" +#include "util/evswitch.h" #include "util/header.h" #include "util/session.h" #include "util/tool.h"
@@ -60,6 +61,7 @@struct report {struct perf_tool tool;struct perf_session *session; +struct evswitch evswitch;bool use_tui, use_gtk, use_stdio;bool show_full_info;bool show_threads;
@@ -243,6 +245,9 @@ static int process_sample_event(struct perf_tool *tool,return 0 ; } +if (evswitch__discard (&rep->evswitch, evsel)) +return 0 ; +if (machine__resolve (machine, &al, sample) <0 ) {pr_debug ("problem processing %d event, skipping it.\n" , event->header.type);
@@ -1189,6 +1194,7 @@ int cmd_report(int argc, const char **argv)OPT_CALLBACK (0 ,"time-quantum" , &symbol_conf.time_quantum,"time (ms|us|ns|s)" ,"Set time quantum for time sort key (default 100ms)" , parse_time_quantum), +OPTS_EVSWITCH (&report.evswitch),OPT_END () }; struct perf_data data = {
@@ -1257,6 +1263,10 @@ int cmd_report(int argc, const char **argv)if (session ==NULL )return -1 ; + ret =evswitch__init (&report.evswitch, session->evlist, stderr); +if (ret) +return ret; +if (zstd_init (&(session->zstd_data),0 ) <0 )pr_warning ("Decompression initialization failed. Reported data may be incomplete.\n" );
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 15/17 perf map: Use zalloc for map_groups
In reply to Arnaldo Carvalho de MeloFrom: John Keeping <[email protected]>
In the next commit we will add new fields to map_groups and we need these to be null if no value is assigned. The simplest way to achieve this is to request zeroed memory from the allocator.
Signed-off-by John Keeping <[email protected]>
Reviewed-by Jiri Olsa <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Konstantin Khlebnikov <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc john keeping <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/map.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/map.c +1 -1
@@ -636,7 +636,7 @@ bool map_groups__empty(struct map_groups *mg)struct map_groups *map_groups__new (struct machine *machine) { -struct map_groups *mg =malloc (sizeof (*mg)); +struct map_groups *mg =zalloc (sizeof (*mg));if (mg !=NULL )map_groups__init (mg, machine);
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 16/17 perf unwind: Fix libunwind when tid != pid
In reply to Arnaldo Carvalho de MeloFrom: John Keeping <[email protected]>
Commit e5adfc3e7e77 ("perf map: Synthesize maps only for thread group leader") changed the recording side so that we no longer get mmap events for threads other than the thread group leader (when synthesising these events for threads which exist before perf is started).
When a file recorded after this change is loaded, the lack of mmap records mean that unwinding is not set up for any other threads.
This can be seen in a simple record/report scenario:
perf record --call-graph=dwarf -t $TID perf report
If $TID is a process ID then the report will show call graphs, but if $TID is a secondary thread the output is as if --call-graph=none was specified.
Following the rationale in that commit, move the libunwind fields into struct map_groups and update the libunwind functions to take this instead of the struct thread. This is only required for unwind__finish_access which must now be called from map_groups__delete and the others are changed for symmetry.
Note that unwind__get_entries keeps the thread argument since it is required for symbol lookup and the libdw unwind provider uses the thread ID.
Signed-off-by John Keeping <[email protected]>
Reviewed-by Jiri Olsa <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Konstantin Khlebnikov <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Fixes e5adfc3e7e77 ("perf map: Synthesize maps only for thread group leader")
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/map.c | 3 ++- tools/perf/util/map_groups.h | 4 +++ tools/perf/util/thread.c | 7 +++-- tools/perf/util/thread.h | 4 --- tools/perf/util/unwind-libunwind-local.c | 18 ++++++------- tools/perf/util/unwind-libunwind.c | 34 ++++++++++++------------ tools/perf/util/unwind.h | 25 ++++++++--------- 7 files changed, 48 insertions(+), 47 deletions(-)
tools/perf/util/map.c +2 -1
@@ -647,6 +647,7 @@ struct map_groups *map_groups__new(struct machine *machine)void map_groups__delete (struct map_groups *mg) {map_groups__exit (mg); +unwind__finish_access (mg);free (mg); }
@@ -887,7 +888,7 @@ int map_groups__clone(struct thread *thread, struct map_groups *parent)if (new ==NULL ) goto out_unlock; - err =unwind__prepare_access (thread, new,NULL ); + err =unwind__prepare_access (mg, new,NULL );if (err) goto out_unlock;
tools/perf/util/map_groups.h +4 -0
@@ -31,6 +31,10 @@ struct map_groups {struct maps maps;struct machine *machine;refcount_t refcnt; +#ifdef HAVE_LIBUNWIND_SUPPORT +void *addr_space; +struct unwind_libunwind_ops *unwind_libunwind_ops; +#endif };#define KMAP_NAME_LEN 256
tools/perf/util/thread.c +3 -4
@@ -105,7 +105,6 @@ void thread__delete(struct thread *thread) }up_write (&thread->comm_lock); -unwind__finish_access (thread);nsinfo__zput (thread->nsinfo);srccode_state_free (&thread->srccode_state);
@@ -252,7 +251,7 @@ static int ____thread__set_comm(struct thread *thread, const char *str,list_add (&new->list, &thread->comm_list);if (exec) -unwind__flush_access (thread); +unwind__flush_access (thread->mg); } thread->comm_set = true;
@@ -332,7 +331,7 @@ int thread__insert_map(struct thread *thread, struct map *map) {int ret; - ret =unwind__prepare_access (thread, map,NULL ); + ret =unwind__prepare_access (thread->mg, map,NULL );if (ret)return ret;
@@ -352,7 +351,7 @@ static int __thread__prepare_access(struct thread *thread)down_read (&maps->lock);for (map =maps__first (maps); map; map =map__next (map)) { - err =unwind__prepare_access (thread, map, &initialized); + err =unwind__prepare_access (thread->mg, map, &initialized);if (err || initialized)break ; }
tools/perf/util/thread.h +0 -4
@@ -44,10 +44,6 @@ struct thread {struct thread_stack *ts;struct nsinfo *nsinfo;struct srccode_state srccode_state; -#ifdef HAVE_LIBUNWIND_SUPPORT -void *addr_space; -struct unwind_libunwind_ops *unwind_libunwind_ops; -#endif bool filter;int filter_entry_depth; };
tools/perf/util/unwind-libunwind-local.c +9 -9
@@ -616,26 +616,26 @@ static unw_accessors_t accessors = { .get_proc_name = get_proc_name, }; -static int _unwind__prepare_access (struct thread *thread) +static int _unwind__prepare_access (struct map_groups *mg) { - thread->addr_space =unw_create_addr_space (&accessors,0 ); -if (!thread->addr_space) { + mg->addr_space =unw_create_addr_space (&accessors,0 ); +if (!mg->addr_space) {pr_err ("unwind: Can't create unwind address space.\n" );return -ENOMEM ; } -unw_set_caching_policy (thread->addr_space,UNW_CACHE_GLOBAL ); +unw_set_caching_policy (mg->addr_space,UNW_CACHE_GLOBAL );return 0 ; } -static void _unwind__flush_access (struct thread *thread) +static void _unwind__flush_access (struct map_groups *mg) { -unw_flush_cache (thread->addr_space,0 ,0 ); +unw_flush_cache (mg->addr_space,0 ,0 ); } -static void _unwind__finish_access (struct thread *thread) +static void _unwind__finish_access (struct map_groups *mg) { -unw_destroy_addr_space (thread->addr_space); +unw_destroy_addr_space (mg->addr_space); } static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,
@@ -660,7 +660,7 @@ static int get_entries(struct unwind_info *ui, unwind_entry_cb_t cb,*/ if (max_stack -1 >0 ) {WARN_ONCE (!ui->thread,"WARNING: ui->thread is NULL" ); - addr_space = ui->thread->addr_space; + addr_space = ui->thread->mg->addr_space;if (addr_space ==NULL )return -1 ;
tools/perf/util/unwind-libunwind.c +17 -17
@@ -11,13 +11,13 @@ struct unwind_libunwind_ops __weak *local_unwind_libunwind_ops;struct unwind_libunwind_ops __weak *x86_32_unwind_libunwind_ops;struct unwind_libunwind_ops __weak *arm64_unwind_libunwind_ops; -static void unwind__register_ops (struct thread *thread, +static void unwind__register_ops (struct map_groups *mg,struct unwind_libunwind_ops *ops) { - thread->unwind_libunwind_ops = ops; + mg->unwind_libunwind_ops = ops; } -int unwind__prepare_access (struct thread *thread,struct map *map, +int unwind__prepare_access (struct map_groups *mg,struct map *map,bool *initialized) {const char *arch;
@@ -28,7 +28,7 @@ int unwind__prepare_access(struct thread *thread, struct map *map,if (!dwarf_callchain_users)return 0 ; -if (thread->addr_space) { +if (mg->addr_space) {pr_debug ("unwind: thread map already set, dso=%s\n" , map->dso->name);if (initialized)
@@ -37,14 +37,14 @@ int unwind__prepare_access(struct thread *thread, struct map *map, }/* env->arch is NULL for live-mode (i.e. perf top) */ -if (!thread->mg->machine->env || !thread->mg->machine->env->arch) +if (!mg->machine->env || !mg->machine->env->arch) goto out_register; - dso_type =dso__type (map->dso, thread->mg->machine); + dso_type =dso__type (map->dso, mg->machine);if (dso_type ==DSO__TYPE_UNKNOWN )return 0 ; - arch =perf_env__arch (thread->mg->machine->env); + arch =perf_env__arch (mg->machine->env);if (!strcmp (arch,"x86" )) {if (dso_type !=DSO__TYPE_64BIT )
@@ -59,37 +59,37 @@ int unwind__prepare_access(struct thread *thread, struct map *map,return 0 ; } out_register: -unwind__register_ops (thread, ops); +unwind__register_ops (mg, ops); - err = thread->unwind_libunwind_ops->prepare_access (thread); + err = mg->unwind_libunwind_ops->prepare_access (mg);if (initialized) *initialized = err ? false : true;return err; } -void unwind__flush_access (struct thread *thread) +void unwind__flush_access (struct map_groups *mg) {if (!dwarf_callchain_users)return ; -if (thread->unwind_libunwind_ops) - thread->unwind_libunwind_ops->flush_access (thread); +if (mg->unwind_libunwind_ops) + mg->unwind_libunwind_ops->flush_access (mg); } -void unwind__finish_access (struct thread *thread) +void unwind__finish_access (struct map_groups *mg) {if (!dwarf_callchain_users)return ; -if (thread->unwind_libunwind_ops) - thread->unwind_libunwind_ops->finish_access (thread); +if (mg->unwind_libunwind_ops) + mg->unwind_libunwind_ops->finish_access (mg); }int unwind__get_entries (unwind_entry_cb_t cb,void *arg,struct thread *thread,struct perf_sample *data,int max_stack) { -if (thread->unwind_libunwind_ops) -return thread->unwind_libunwind_ops->get_entries (cb, arg, thread, data, max_stack); +if (thread->mg->unwind_libunwind_ops) +return thread->mg->unwind_libunwind_ops->get_entries (cb, arg, thread, data, max_stack);return 0 ; }
tools/perf/util/unwind.h +13 -12
@@ -6,6 +6,7 @@#include <linux/types.h> struct map ; +struct map_groups ;struct perf_sample ;struct symbol ;struct thread ;
@@ -19,9 +20,9 @@ struct unwind_entry {typedef int (*unwind_entry_cb_t )(struct unwind_entry *entry,void *arg);struct unwind_libunwind_ops { -int (*prepare_access)(struct thread *thread); -void (*flush_access)(struct thread *thread); -void (*finish_access)(struct thread *thread); +int (*prepare_access)(struct map_groups *mg); +void (*flush_access)(struct map_groups *mg); +void (*finish_access)(struct map_groups *mg);int (*get_entries)(unwind_entry_cb_t cb,void *arg,struct thread *thread,struct perf_sample *data,int max_stack);
@@ -46,20 +47,20 @@ int unwind__get_entries(unwind_entry_cb_t cb, void *arg,#endif int LIBUNWIND__ARCH_REG_ID (int regnum); -int unwind__prepare_access (struct thread *thread,struct map *map, +int unwind__prepare_access (struct map_groups *mg,struct map *map,bool *initialized); -void unwind__flush_access (struct thread *thread); -void unwind__finish_access (struct thread *thread); +void unwind__flush_access (struct map_groups *mg); +void unwind__finish_access (struct map_groups *mg);#else -static inline int unwind__prepare_access (struct thread *thread__maybe_unused , +static inline int unwind__prepare_access (struct map_groups *mg__maybe_unused ,struct map *map__maybe_unused ,bool *initialized__maybe_unused ) {return 0 ; } -static inline void unwind__flush_access (struct thread *thread__maybe_unused ) {} -static inline void unwind__finish_access (struct thread *thread__maybe_unused ) {} +static inline void unwind__flush_access (struct map_groups *mg__maybe_unused ) {} +static inline void unwind__finish_access (struct map_groups *mg__maybe_unused ) {}#endif #else static inline int
@@ -72,14 +73,14 @@ unwind__get_entries(unwind_entry_cb_t cb __maybe_unused,return 0 ; } -static inline int unwind__prepare_access (struct thread *thread__maybe_unused , +static inline int unwind__prepare_access (struct map_groups *mg__maybe_unused ,struct map *map__maybe_unused ,bool *initialized__maybe_unused ) {return 0 ; } -static inline void unwind__flush_access (struct thread *thread__maybe_unused ) {} -static inline void unwind__finish_access (struct thread *thread__maybe_unused ) {} +static inline void unwind__flush_access (struct map_groups *mg__maybe_unused ) {} +static inline void unwind__finish_access (struct map_groups *mg__maybe_unused ) {}#endif /* HAVE_DWARF_UNWIND_SUPPORT */ #endif /* __UNWIND_H */
Signature · 1 line
2.21.0
Arnaldo Carvalho de Melo
maintainerPATCH 17/17 perf unwind: Remove unnecessary test
In reply to Arnaldo Carvalho de MeloFrom: John Keeping <[email protected]>
If dwarf_callchain_users is false, then unwind__prepare_access() will not set unwind_libunwind_ops so the remaining test here is sufficient.
Signed-off-by John Keeping <[email protected]>
Reviewed-by Jiri Olsa <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Konstantin Khlebnikov <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc john keeping <[email protected]>
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
--- tools/perf/util/unwind-libunwind.c | 6 ------ 1 file changed, 6 deletions(-)
tools/perf/util/unwind-libunwind.c +0 -6
@@ -69,18 +69,12 @@ int unwind__prepare_access(struct map_groups *mg, struct map *map,void unwind__flush_access (struct map_groups *mg) { -if (!dwarf_callchain_users) -return ; -if (mg->unwind_libunwind_ops) mg->unwind_libunwind_ops->flush_access (mg); }void unwind__finish_access (struct map_groups *mg) { -if (!dwarf_callchain_users) -return ; -if (mg->unwind_libunwind_ops) mg->unwind_libunwind_ops->finish_access (mg); }
Signature · 1 line
2.21.0