[GIT PULL 00/19] perf/core improvements
From: Arnaldo Carvalho de Melo
Date: Mon Apr 11 2016 - 21:56:54 EST
Hi Ingo,
Please consider pulling, tested with 'perf test', 'make -C tools/perf
build-test' and building on these userspaces, using docker:
# dm
alldeps-fedora-rawhide-minus-python-dev: Ok
alldeps-fedora-20: Ok
alldeps-ubuntu-12.04: Ok
minimal-debian-experimental-x-mips64: Ok
minimal-debian-experimental-x-mips64el: Ok
minimal-debian-experimental-x-mipsel: Ok
minimal-ubuntu-x-arm: Ok
minimal-ubuntu-x-arm64: Ok
minimal-ubuntu-x-ppc64: Ok
minimal-ubuntu-x-ppc64el: Ok
alldeps-debian: Ok
alldeps-mageia: Ok
alldeps-rhel7: Ok
alldeps-centos: Ok
alldeps-opensuse: Ok
alldeps-ubuntu: Ok
#
This is on top of my previous pull request, that is not yet
merged: perf-core-for-mingo-20160408.
Best regards,
- Arnaldo
The following changes since commit 99e87f7bb7268cf644add87130590966fd5d0d17:
perf symbols: Adjust symbol for shared objects (2016-04-08 09:58:15 -0300)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-20160411
for you to fetch changes up to 00768a2bd3245eace0690fcf2c02776a256b66d7:
perf trace: Print unresolved symbol names as addresses (2016-04-11 22:18:25 -0300)
----------------------------------------------------------------
perf/core improvements:
- Automagically create a 'bpf-output' event, easing the setup of BPF
C "scripts" that produce output via the perf ring buffer. Now it is
just a matter of calling any perf tool, such as 'trace', with a C
source file that references the __bpf_stdout__ output channel and
that channel will be created and connected to the script:
# trace -e nanosleep --event test_bpf_stdout.c usleep 1
0.013 ( 0.013 ms): usleep/2818 nanosleep(rqtp: 0x7ffcead45f40 ) ...
0.013 ( ): __bpf_stdout__:Raise a BPF event!..)
0.015 ( ): perf_bpf_probe:func_begin:(ffffffff81112460))
0.261 ( ): __bpf_stdout__:Raise a BPF event!..)
0.262 ( ): perf_bpf_probe:func_end:(ffffffff81112460 <- ffffffff81003d92))
0.264 ( 0.264 ms): usleep/2818 ... [continued]: nanosleep()) = 0
#
Further work is needed to reduce the number of lines in a perf bpf C source
file, this being the part where we greatly reduce the command line setup (Wang Nan)
- 'perf trace' now supports callchains, with 'trace --call-graph dwarf' using
libunwind, just like 'perf top', to ask the kernel for stack dumps for CFI
processing. This reduces the overhead by asking just for userspace callchains
and also only for the syscall exit tracepoint (raw_syscalls:sys_exit)
(Milian Wolff, Arnaldo Carvalho de Melo)
Try it with, for instance:
# perf trace --call dwarf ping 127.0.0.1
An excerpt of a system wide 'perf trace --call dwarf" session is at:
https://fedorapeople.org/~acme/perf/perf-trace--call-graph-dwarf--all-cpus.txt
You may need to bump the number of mmap pages, using -m/--mmap-pages,
but on a Broadwell machine the defaults allowed system wide tracing to
work without losing that many records, experiment with just some
syscalls, like:
# perf trace --call dwarf -e nanosleep,futex
All the targets available for 'perf record', 'perf top' (--pid, --tid, --cpu,
etc) should work. Also --duration may be interesting to try.
To get filenames from in various syscalls pointer args (open, ettc), add this
to the mix:
# perf probe 'vfs_getname=getname_flags:72 pathname=filename:string'
Making this work is next in line:
# trace --call dwarf --ev sched:sched_switch/call-graph=fp/ usleep 1
I.e. honouring per-tracepoint callchains in 'perf trace' in addition to
in raw_syscalls:sys_exit.
Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
----------------------------------------------------------------
Arnaldo Carvalho de Melo (15):
perf script: Use readdir() instead of deprecated readdir_r()
perf thread_map: Use readdir() instead of deprecated readdir_r()
perf tools: Use readdir() instead of deprecated readdir_r()
perf tools: Use readdir() instead of deprecated readdir_r()
perf dwarf: Guard !x86_64 definitions under #ifdef else clause
perf evsel: Allow passing a left alignment when printing a symbol
perf evsel: Rename print_ip() to fprintf_sym()
perf evsel: Introduce fprintf_callchain() method out of fprintf_sym()
perf trace: Exclude the kernel part of the callchain leading to a syscall
perf evsel: Do not use globals in config()
perf evlist: Add (reset,set)_sample_bit methods
perf evsel: Rename config_callgraph() to config_callchain() and make it public
perf trace: Make "--call-graph" affect just "raw_syscalls:sys_exit"
perf evsel: Allow unresolved symbol names to be printed as addresses
perf trace: Print unresolved symbol names as addresses
Milian Wolff (2):
perf evsel: Allow specifying a file to output in perf_evsel__print_ip
perf trace: Add support for printing call chains on sys_exit events.
Wang Nan (2):
perf bpf: Clone bpf stdout events in multiple bpf scripts
perf bpf: Automatically create bpf-output event __bpf_stdout__
tools/perf/Documentation/perf-trace.txt | 9 ++
tools/perf/arch/x86/tests/perf-time-to-tsc.c | 2 +-
tools/perf/arch/x86/util/dwarf-regs.c | 8 +-
tools/perf/builtin-kvm.c | 2 +-
tools/perf/builtin-record.c | 10 +-
tools/perf/builtin-script.c | 78 +++++++--------
tools/perf/builtin-top.c | 2 +-
tools/perf/builtin-trace.c | 65 +++++++++++-
tools/perf/tests/bpf.c | 2 +-
tools/perf/tests/code-reading.c | 2 +-
tools/perf/tests/keep-tracking.c | 2 +-
tools/perf/tests/openat-syscall-tp-fields.c | 2 +-
tools/perf/tests/perf-record.c | 2 +-
tools/perf/tests/switch-tracking.c | 2 +-
tools/perf/util/bpf-loader.c | 143 +++++++++++++++++++++++++++
tools/perf/util/bpf-loader.h | 19 ++++
tools/perf/util/event.c | 12 +--
tools/perf/util/evlist.c | 18 ++++
tools/perf/util/evlist.h | 16 ++-
tools/perf/util/evsel.c | 16 +--
tools/perf/util/evsel.h | 14 ++-
tools/perf/util/parse-events.c | 60 +++++------
tools/perf/util/record.c | 5 +-
tools/perf/util/session.c | 95 ++++++++++++------
tools/perf/util/session.h | 8 +-
tools/perf/util/symbol.c | 25 ++++-
tools/perf/util/symbol.h | 6 ++
tools/perf/util/thread_map.c | 8 +-
28 files changed, 487 insertions(+), 146 deletions(-)