Re: [BUG] perf/inject: crash in pipe mode

From: Arnaldo Carvalho de Melo
Date: Sat May 26 2018 - 08:18:34 EST


Em Fri, May 25, 2018 at 05:19:54PM -0700, Stephane Eranian escreveu:
>
> With the latest tip.git perf, if you run
>
> $ perf record -a -o - sleep 2 | perf inject -b -i - | perf buildid-list -i -
> SEGFAULT in perf inject:

I'm not being able to reproduce this problem here:

[root@seventh ~]# perf record -a -o - sleep 2 | perf inject -b -i - 2>
/dev/null | perf buildid-list -i - | wc -l
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
1038
[root@seventh ~]# perf --version
perf version 4.17.rc5.g615aaf
[root@seventh ~]#

This is with my private perf/core branch, lets see with tip.git/master

[acme@seventh perf]$ git checkout -b tip-master tip/master
Branch tip-master set up to track remote branch master from tip.
Switched to a new branch 'tip-master'
[acme@seventh perf]$ rm -rf /tmp/build/perf/ ; mkdir -p /tmp/build/perf ; make -C tools/perf O=/tmp/build/perf install-bin
[root@seventh ~]# perf --version
perf version 4.17.rc6.g99a7f2
[root@seventh ~]#
[root@seventh ~]# perf record -a -o - sleep 2 | perf inject -b -i - 2> /dev/null | perf buildid-list -i - | wc -l
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
1038
[root@seventh ~]#
[acme@seventh perf]$ git log --oneline -5
99a7f2c81712 (HEAD -> tip-master, tip/master) Merge branch 'perf/core'
861410270ab5 (tip/perf/core, jouet/perf/core, acme/perf/core) Merge tag
'perf-core-for-mingo-4.18-20180523' of
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
dc0a16dd9952 Merge branch 'x86/pti'
88a582678aa9 Merge branch 'linus'
bee797529d7c (torvalds/master) Merge tag 'mfd-fixes-4.17' of
git://git.kernel.org/pub/scm/linux/kernel/git/lee/mfd
[acme@seventh perf]$
[acme@seventh perf]$ perf version --build-options
perf version 4.17.rc6.g99a7f2
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
[acme@seventh perf]$
[acme@seventh perf]$ ldd ~/bin/perf
linux-vdso.so.1 (0x00007fff2095f000)
libunwind-x86_64.so.8 => /lib64/libunwind-x86_64.so.8 (0x00007fd0e7f01000)
libunwind.so.8 => /lib64/libunwind.so.8 (0x00007fd0e7ce7000)
liblzma.so.5 => /lib64/liblzma.so.5 (0x00007fd0e7ac1000)
libpthread.so.0 => /lib64/libpthread.so.0 (0x00007fd0e78a3000)
librt.so.1 => /lib64/librt.so.1 (0x00007fd0e769b000)
libm.so.6 => /lib64/libm.so.6 (0x00007fd0e7350000)
libdl.so.2 => /lib64/libdl.so.2 (0x00007fd0e714c000)
libelf.so.1 => /lib64/libelf.so.1 (0x00007fd0e6f34000)
libdw.so.1 => /lib64/libdw.so.1 (0x00007fd0e6cec000)
libcrypto.so.1.1 => /lib64/libcrypto.so.1.1 (0x00007fd0e6864000)
libslang.so.2 => /lib64/libslang.so.2 (0x00007fd0e639e000)
libperl.so.5.26 => /lib64/libperl.so.5.26 (0x00007fd0e5f97000)
libresolv.so.2 => /lib64/libresolv.so.2 (0x00007fd0e5d80000)
libnsl.so.1 => /lib64/libnsl.so.1 (0x00007fd0e5b68000)
libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007fd0e5932000)
libutil.so.1 => /lib64/libutil.so.1 (0x00007fd0e572f000)
libc.so.6 => /lib64/libc.so.6 (0x00007fd0e5379000) /lib64/ld-linux-x86-64.so.2 (0x00007fd0e8120000)
libpython2.7.so.1.0 => /lib64/libpython2.7.so.1.0 (0x00007fd0e4f3a000)
libz.so.1 => /lib64/libz.so.1 (0x00007fd0e4d23000)
libnuma.so.1 => /lib64/libnuma.so.1 (0x00007fd0e4b18000)
libbabeltrace-ctf.so.1 => /lib64/libbabeltrace-ctf.so.1 (0x00007fd0e48c8000)
libgcc_s.so.1 => /lib64/libgcc_s.so.1 (0x00007fd0e46b1000)
libbz2.so.1 => /lib64/libbz2.so.1 (0x00007fd0e44a0000)
libfreebl3.so => /lib64/libfreebl3.so (0x00007fd0e429d000)
libbabeltrace.so.1 => /lib64/libbabeltrace.so.1 (0x00007fd0e4090000)
libpopt.so.0 => /lib64/libpopt.so.0 (0x00007fd0e3e83000)
libuuid.so.1 => /lib64/libuuid.so.1 (0x00007fd0e3c7e000)
libgmodule-2.0.so.0 => /lib64/libgmodule-2.0.so.0 (0x00007fd0e3a7a000)
libglib-2.0.so.0 => /lib64/libglib-2.0.so.0 (0x00007fd0e3766000)
libpcre.so.1 => /lib64/libpcre.so.1 (0x00007fd0e34f4000)
[acme@seventh perf]$
[root@seventh ~]# cat /etc/fedora-release
Fedora release 27 (Twenty Seven)
[root@seventh ~]#

> free_dup_event (oe=0x555555d25b88, oe=0x555555d25b88,
> event=0x3030310931303031) at util/ordered-events.c:86
> 86 oe->cur_alloc_size -= event->header.size;
> (gdb) bt
> #0 free_dup_event (oe=0x555555d25b88, oe=0x555555d25b88,
> event=0x3030310931303031) at util/ordered-events.c:86
> #1 ordered_events__free (oe=oe@entry=0x555555d25b88) at
> util/ordered-events.c:310
> #2 0x00005555557964f8 in __perf_session__process_pipe_events
> (session=0x555555d1f910) at util/session.c:1778
> #3 perf_session__process_events (session=session@entry=0x555555d1f910) at
> util/session.c:1958
> #4 0x00005555556ef9b2 in __cmd_inject (inject=0x7fffffffda40) at
> builtin-inject.c:697
> #5 cmd_inject (argc=<optimized out>, argv=<optimized out>) at
> builtin-inject.c:871
> #6 0x000055555572e8b1 in run_builtin (p=0x555555be8f98 <commands+600>,
> argc=4, argv=0x7fffffffe460) at perf.c:303
> #7 0x000055555572ebae in handle_internal_command (argc=4,
> argv=0x7fffffffe460) at perf.c:355
> #8 0x00005555556ae1e1 in run_argv (argcp=<synthetic pointer>,
> argv=<synthetic pointer>) at perf.c:399
> #9 main (argc=<optimized out>, argv=0x7fffffffe460) at perf.c:521
>
> In general I think the pipe mode is not very well tested. I think it could
> be made the default file format. I believe perf can
> autodetect file vs. pipe mode perf.data using the header.size field. This
> would simplify a few things inside perf and ensure
> the pipe mode format is well tested.