perf/core improvements and fixes
Subsystems (5)
Arnaldo Carvalho de Melo
maintainerHi Ingo,
Please consider pulling,
- Arnaldo
Test results at the end of this message, as usual.
The following changes since commit 883f4def8b77e6870ce42be279564cca0256c611:
Merge tag 'perf-core-for-mingo-4.21-20181218' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core (2018-12-20 18:51:47 +0100)
are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-4.21-20190103
for you to fetch changes up to b25756df5b28cd7b6e91200fc5012e7c76e8ec69:
perf session: Add comment for perf_session__register_idle_thread() (2019-01-02 11:05:06 -0300)
----------------------------------------------------------------
perf/core improvements and fixes:
perf c2c:
Jiri Olsa:
- Change the default coalesce setup to from '--coalesce pid,iaddr' to just '--coalesce iaddr'.
- Increase the HITM ratio limit for displayed cachelines.
perf script:
Andi Kleen:
- Fix LBR skid dump problems in brstackinsn.
perf trace:
Arnaldo Carvalho de Melo:
- Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter.
- Do not hardcode the size of the tracepoint common_ fields.
- Beautify USBDEFFS_ ioctl commands.
Colin Ian King:
- Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*".
perf python:
Jiri Olsa:
- Do not force closing original perf descriptor in evlist.get_pollfd().
tools misc:
Jiri Olsa:
- Allow overriding CFLAGS and LDFLAGS.
perf build:
Stanislav Fomichev:
- Don't unconditionally link the libbfd feature test to -liberty and -lz
thread-stack:
Adrian Hunter:
- Fix processing for the idle task, having a stack per cpu.
----------------------------------------------------------------
Adrian Hunter (8):
perf thread-stack: Simplify some code in thread_stack__process() perf thread-stack: Tidy thread_stack__bottom() usage perf thread-stack: Avoid direct reference to the thread's stack perf thread-stack: Allow for a thread stack array
perf thread-stack: Factor out thread_stack__init() perf thread-stack: Allocate an array of thread stacks perf thread-stack: Fix thread stack processing for the idle task perf session: Add comment for perf_session__register_idle_thread()
Andi Kleen (1):
perf script: Fix LBR skid dump problems in brstackinsn
Arnaldo Carvalho de Melo (11):
perf trace: Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter perf beauty mmap: PROT_WRITE should come before PROT_EXEC
perf trace: Do not hardcode the size of the tracepoint common_ fields
perf trace: Rename thread_thread->paths to thread_trace->files
perf trace: Move the files table resizing to outside set_pathname()
perf trace: Store the major number for a file when storing its pathname tools headers uapi: Grab a copy of usbdevice_fs.h
perf beauty ioctl: Add generator for USBDEVFS_ ioctl commands
perf trace: Wire up ioctl's USBDEBFS_ cmd table generator
perf trace beauty: Export function to get the files for a thread
perf trace beauty ioctl: Beautify USBDEVFS_ commands
Colin Ian King (1):
perf trace: Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*"
Jiri Olsa (7):
perf python: Do not force closing original perf descriptor in evlist.get_pollfd() perf c2c: Change the default coalesce setup
perf c2c: Increase the HITM ratio limit for displayed cachelines
tools power x86_energy_perf_policy: Override CFLAGS assignments and add LDFLAGS to build command tools thermal tmon: Allow overriding CFLAGS assignments
tools power turbostat: Override CFLAGS assignments and add LDFLAGS to build command tools gpio: Allow overriding CFLAGS
Stanislav Fomichev (1):
perf build: Don't unconditionally link the libbfd feature test to -liberty and -lz
tools/build/Makefile.feature | 4 +- tools/build/feature/Makefile | 10 +- tools/gpio/Makefile | 2 +- tools/include/uapi/linux/usbdevice_fs.h | 201 ++++++++++++++++++ tools/perf/Makefile.config | 44 ++-- tools/perf/Makefile.perf | 8 + tools/perf/builtin-c2c.c | 4 +- tools/perf/builtin-script.c | 21 +- tools/perf/builtin-trace.c | 142 +++++++++---- tools/perf/check-headers.sh | 1 + tools/perf/trace/beauty/beauty.h | 7 + tools/perf/trace/beauty/ioctl.c | 22 ++ tools/perf/trace/beauty/mmap.c | 2 +- tools/perf/trace/beauty/seccomp.c | 4 +- tools/perf/trace/beauty/usbdevfs_ioctl.sh | 19 ++ tools/perf/util/dump-insn.c | 8 + tools/perf/util/dump-insn.h | 2 + tools/perf/util/intel-bts.c | 4 +- .../util/intel-pt-decoder/intel-pt-insn-decoder.c | 8 + tools/perf/util/intel-pt.c | 6 +- tools/perf/util/python.c | 3 +- tools/perf/util/session.c | 7 + tools/perf/util/thread-stack.c | 227 +++++++++++++++------ tools/perf/util/thread-stack.h | 8 +- tools/power/x86/turbostat/Makefile | 8 +- tools/power/x86/x86_energy_perf_policy/Makefile | 6 +- tools/thermal/tmon/Makefile | 8 +- 27 files changed, 620 insertions(+), 166 deletions(-) create mode 100644 tools/include/uapi/linux/usbdevice_fs.h create mode 100755 tools/perf/trace/beauty/usbdevfs_ioctl.sh
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.
$ dm 1 alpine:3.4 : Ok gcc (Alpine 5.3.0) 5.3.0 2 alpine:3.5 : Ok gcc (Alpine 6.2.1) 6.2.1 20160822 3 alpine:3.6 : Ok gcc (Alpine 6.3.0) 6.3.0 4 alpine:3.7 : Ok gcc (Alpine 6.4.0) 6.4.0 5 alpine:3.8 : Ok gcc (Alpine 6.4.0) 6.4.0 6 alpine:edge : Ok gcc (Alpine 8.2.0) 8.2.0 7 amazonlinux:1 : Ok gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-28) 8 amazonlinux:2 : Ok gcc (GCC) 7.3.1 20180303 (Red Hat 7.3.1-5) 9 android-ndk:r12b-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 10 android-ndk:r15c-arm : Ok arm-linux-androideabi-gcc (GCC) 4.9.x 20150123 (prerelease) 11 centos:5 : Ok gcc (GCC) 4.1.2 20080704 (Red Hat 4.1.2-55) 12 centos:6 : Ok gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23) 13 centos:7 : Ok gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-36) 14 clearlinux:latest : Ok gcc (Clear Linux OS for Intel Architecture) 8.2.1 20180502 15 debian:7 : Ok gcc (Debian 4.7.2-5) 4.7.2 16 debian:8 : Ok gcc (Debian 4.9.2-10+deb8u1) 4.9.2 17 debian:9 : Ok gcc (Debian 6.3.0-18+deb9u1) 6.3.0 20170516 18 debian:experimental : Ok gcc (Debian 8.2.0-10) 8.2.0 19 debian:experimental-x-arm64 : Ok aarch64-linux-gnu-gcc (Debian 8.2.0-10) 8.2.0 20 debian:experimental-x-mips64 : Ok mips64-linux-gnuabi64-gcc (Debian 8.2.0-10) 8.2.0 21 debian:experimental : Ok gcc (Debian 8.2.0-13) 8.2.0 22 debian:experimental-x-arm64 : Ok aarch64-linux-gnu-gcc (Debian 8.2.0-11) 8.2.0 23 debian:experimental-x-mips : Ok mips-linux-gnu-gcc (Debian 8.2.0-11) 8.2.0 24 debian:experimental-x-mips64 : Ok mips64-linux-gnuabi64-gcc (Debian 8.2.0-11) 8.2.0 25 debian:experimental-x-mipsel : Ok mipsel-linux-gnu-gcc (Debian 8.2.0-11) 8.2.0 26 fedora:20 : Ok gcc (GCC) 4.8.3 20140911 (Red Hat 4.8.3-7) 27 fedora:22 : Ok gcc (GCC) 5.3.1 20160406 (Red Hat 5.3.1-6) 28 fedora:23 : Ok gcc (GCC) 5.3.1 20160406 (Red Hat 5.3.1-6) 29 fedora:24 : Ok gcc (GCC) 6.3.1 20161221 (Red Hat 6.3.1-1) 30 fedora:24-x-ARC-uClibc : Ok arc-linux-gcc (ARCompact ISA Linux uClibc toolchain 2017.09-rc2) 7.1.1 20170710 31 fedora:25 : Ok gcc (GCC) 6.4.1 20170727 (Red Hat 6.4.1-1) 32 fedora:26 : Ok gcc (GCC) 7.3.1 20180130 (Red Hat 7.3.1-2) 33 fedora:27 : Ok gcc (GCC) 7.3.1 20180712 (Red Hat 7.3.1-6) 34 fedora:28 : Ok gcc (GCC) 8.2.1 20181105 (Red Hat 8.2.1-5) 35 fedora:29 : Ok gcc (GCC) 8.2.1 20181105 (Red Hat 8.2.1-5) 36 fedora:rawhide : Ok gcc (GCC) 8.2.1 20181105 (Red Hat 8.2.1-5) 37 gentoo-stage3-amd64:latest : Ok gcc (Gentoo 7.3.0-r3 p1.4) 7.3.0 38 mageia:5 : Ok gcc (GCC) 4.9.2 39 mageia:6 : Ok gcc (Mageia 5.5.0-1.mga6) 5.5.0 40 opensuse:13.2 : Ok gcc (SUSE Linux) 4.8.3 20140627 [gcc-4_8-branch revision 212064] 41 opensuse:15.0 : Ok gcc (SUSE Linux) 7.3.1 20180323 [gcc-7-branch revision 258812] 42 opensuse:42.1 : Ok gcc (SUSE Linux) 4.8.5 43 opensuse:42.2 : Ok gcc (SUSE Linux) 4.8.5 44 opensuse:42.3 : Ok gcc (SUSE Linux) 4.8.5 45 opensuse:tumbleweed : Ok gcc (SUSE Linux) 8.2.1 20181108 [gcc-8-branch revision 265914] 46 oraclelinux:6 : Ok gcc (GCC) 4.4.7 20120313 (Red Hat 4.4.7-23.0.1) 47 oraclelinux:7 : Ok gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-36.0.1) 48 ubuntu:12.04.5 : Ok gcc (Ubuntu/Linaro 4.6.3-1ubuntu5) 4.6.3 49 ubuntu:14.04.4 : Ok gcc (Ubuntu 4.8.4-2ubuntu1~14.04.4) 4.8.4 50 ubuntu:14.04.4-x-linaro-arm64 : Ok aarch64-linux-gnu-gcc (Linaro GCC 5.5-2017.10) 5.5.0 51 ubuntu:16.04 : Ok gcc (Ubuntu 5.4.0-6ubuntu1~16.04.10) 5.4.0 20160609 52 ubuntu:16.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 53 ubuntu:16.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 54 ubuntu:16.04-x-powerpc : Ok powerpc-linux-gnu-gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 55 ubuntu:16.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu/IBM 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 56 ubuntu:16.04-x-powerpc64el : Ok powerpc64le-linux-gnu-gcc (Ubuntu/IBM 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 57 ubuntu:16.04-x-s390 : Ok s390x-linux-gnu-gcc (Ubuntu 5.4.0-6ubuntu1~16.04.9) 5.4.0 20160609 58 ubuntu:17.10 : Ok gcc (Ubuntu 7.2.0-8ubuntu3.2) 7.2.0 59 ubuntu:18.04 : Ok gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 60 ubuntu:18.04-x-arm : Ok arm-linux-gnueabihf-gcc (Ubuntu/Linaro 7.3.0-27ubuntu1~18.04) 7.3.0 61 ubuntu:18.04-x-arm64 : Ok aarch64-linux-gnu-gcc (Ubuntu/Linaro 7.3.0-27ubuntu1~18.04) 7.3.0 62 ubuntu:18.04-x-m68k : Ok m68k-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 63 ubuntu:18.04-x-powerpc : Ok powerpc-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 64 ubuntu:18.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 65 ubuntu:18.04-x-powerpc64el : Ok powerpc64le-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 66 ubuntu:18.04-x-riscv64 : Ok riscv64-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 67 ubuntu:18.04-x-s390 : Ok s390x-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 68 ubuntu:18.04-x-sh4 : Ok sh4-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 69 ubuntu:18.04-x-sparc64 : Ok sparc64-linux-gnu-gcc (Ubuntu 7.3.0-27ubuntu1~18.04) 7.3.0 70 ubuntu:18.10 : Ok gcc (Ubuntu 8.2.0-7ubuntu1) 8.2.0 71 ubuntu:19.04 : Ok gcc (Ubuntu 8.2.0-12ubuntu1) 8.2.0 72 ubuntu:19.04-x-alpha : Ok alpha-linux-gnu-gcc (Ubuntu 8.2.0-11ubuntu1) 8.2.0 73 ubuntu:19.04-x-hppa : Ok hppa-linux-gnu-gcc (Ubuntu 8.2.0-11ubuntu1) 8.2.0 $
# uname -a
Linux quaco 4.19.8-rt6+ #2 SMP PREEMPT RT Wed Dec 19 12:58:37 -03 2018 x86_64 x86_64 x86_64 GNU/Linux
# git log --oneline -1
b25756df5b28 perf session: Add comment for perf_session__register_idle_thread()
# perf version --build-options
perf version 4.20.rc7.gb25756
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
# 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: x86 rdpmc : Ok
60: Convert perf time to TSC : Ok
61: DWARF unwind : Ok
62: x86 instruction decoder - new instructions : Ok
63: x86 bp modify : Ok
64: probe libc's inet_pton & backtrace it with ping : Ok
65: Use vfs_getname probe to get syscall args filenames : Ok
66: Add vfs_getname probe to get syscall args filenames : Ok
67: Check open filename arg using perf trace + vfs_getname: Ok $ make -C tools/perf build-test
make: Entering directory '/home/acme/git/perf/tools/perf'
- tarpkg: ./tests/perf-targz-src-pkg .
make_no_libdw_dwarf_unwind_O: make NO_LIBDW_DWARF_UNWIND=1
make_no_libpython_O: make NO_LIBPYTHON=1
make_clean_all_O: make clean all
make_perf_o_O: make perf.o
make_no_newt_O: make NO_NEWT=1
make_no_libunwind_O: make NO_LIBUNWIND=1
make_no_libelf_O: make NO_LIBELF=1
make_no_libnuma_O: make NO_LIBNUMA=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
make_no_slang_O: make NO_SLANG=1
make_pure_O: make
make_help_O: make help
make_cscope_O: make cscope
make_with_clangllvm_O: make LIBCLANGLLVM=1
make_util_map_o_O: make util/map.o
make_no_demangle_O: make NO_DEMANGLE=1
make_install_prefix_O: make install prefix=/tmp/krava
make_no_backtrace_O: make NO_BACKTRACE=1
make_doc_O: make doc
make_install_O: make install
make_util_pmu_bison_o_O: make util/pmu-bison.o
make_no_auxtrace_O: make NO_AUXTRACE=1
make_debug_O: make DEBUG=1
make_install_prefix_slash_O: make install prefix=/tmp/krava/
make_no_libaudit_O: make NO_LIBAUDIT=1
make_no_libbionic_O: make NO_LIBBIONIC=1
make_no_libperl_O: make NO_LIBPERL=1
make_with_babeltrace_O: make LIBBABELTRACE=1
make_no_ui_O: make NO_NEWT=1 NO_SLANG=1 NO_GTK2=1
make_install_bin_O: make install-bin
make_static_O: make LDFLAGS=-static
make_no_scripts_O: make NO_LIBPYTHON=1 NO_LIBPERL=1
make_tags_O: make tags
make_no_gtk2_O: make NO_GTK2=1
make_no_libbpf_O: make NO_LIBBPF=1
OK
make: Leaving directory '/home/acme/git/perf/tools/perf'
$Arnaldo Carvalho de Melo
maintainerPATCH 1/29 perf trace: Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter
From: Arnaldo Carvalho de Melo <[email protected]>
While updating 'perf trace' on an machine with an old precompiled augmented_raw_syscalls.o that didn't setup the syscall map the new 'perf trace' codebase notices the augmented_raw_syscalls.o eBPF event, decides to use it instead of the old raw_syscalls:sys_{enter,exit} method, but then because we don't have the syscall map tries to set the tracepoint filter on the sys_{enter,exit} evsels, that are NULL, segfaulting.
Make the code more robust by checking it those tracepoints have their respective evsels in place before trying to set the tp filter.
With this we still get everything to work, just not setting up the syscall filters, which is better than a segfault. Now to update the precompiled augmented_raw_syscalls.o and continue development :-)
--- tools/perf/builtin-trace.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)
tools/perf/builtin-trace.c +3 -1
@@ -2686,7 +2686,9 @@ static int trace__set_ev_qualifier_filter(struct trace *trace) {if (trace->syscalls.map)return trace__set_ev_qualifier_bpf_filter (trace); -return trace__set_ev_qualifier_tp_filter (trace); +if (trace->syscalls.events.sys_enter) +return trace__set_ev_qualifier_tp_filter (trace); +return 0 ; } static int bpf_map__set_filter_pids(struct bpf_map *map __maybe_unused,
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 2/29 perf beauty mmap: PROT_WRITE should come before PROT_EXEC
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
To match strace output:
# cat mmap.c #include <sys/mman.h>
int main(void)
{
mmap(0, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
return 0;
}
# strace -e mmap ./mmap |& grep -v ^+++
mmap(NULL, 103484, PROT_READ, MAP_PRIVATE, 3, 0) = 0x7f5bae400000
mmap(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bae3fe000
mmap(NULL, 3889792, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x7f5bade40000
mmap(0x7f5bae1ec000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x1ac000) = 0x7f5bae1ec000
mmap(0x7f5bae1f2000, 14976, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0) = 0x7f5bae1f2000
mmap(NULL, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x7f5bae419000
# trace -e mmap ./mmap |& grep -v ^+++
mmap(NULL, 103484, PROT_READ, MAP_PRIVATE, 3, 0) = 0x7f6646c25000
mmap(NULL, 8192, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS) = 0x7f6646c23000
mmap(NULL, 3889792, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_DENYWRITE, 3, 0) = 0x7f6646665000
mmap(0x7f6646a11000, 24576, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_DENYWRITE, 3, 0x1ac000) = 0x7f6646a11000
mmap(0x7f6646a17000, 14976, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS) = 0x7f6646a17000
mmap(NULL, 4096, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS) = 0x7f6646c3e000
#--- tools/perf/trace/beauty/mmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/trace/beauty/mmap.c +1 -1
@@ -18,8 +18,8 @@ static size_t syscall_arg__scnprintf_mmap_prot(char *bf, size_t size, }P_MMAP_PROT (READ ); -P_MMAP_PROT (EXEC );P_MMAP_PROT (WRITE ); +P_MMAP_PROT (EXEC );P_MMAP_PROT (SEM );P_MMAP_PROT (GROWSDOWN );P_MMAP_PROT (GROWSUP );
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 3/29 perf build: Don't unconditionally link the libbfd feature test to -liberty and -lz
In reply to Arnaldo Carvalho de MeloFrom: Stanislav Fomichev <[email protected]>
Current libbfd feature test unconditionally links against -liberty and -lz. While it's required on some systems (e.g. opensuse), it's completely unnecessary on the others, where only -lbdf is sufficient (debian). This patch streamlines (and renames) the following feature checks:
feature-libbfd - only link against -lbfd (debian),
see commit 2cf9040714f3 ("perf tools: Fix bfd dependency libraries detection")
feature-libbfd-liberty - link against -lbfd and -liberty
feature-libbfd-liberty-z - link against -lbfd, -liberty and -lz (opensuse),
see commit 280e7c48c3b8 ("perf tools: fix BFD detection on opensuse")
(feature-liberty{,-z} were renamed to feature-libbfd-liberty{,z} for clarity)
The main motivation is to fix this feature test for bpftool which is currently broken on debian (libbfd feature shows OFF, but we still unconditionally link against -lbfd and it works).
Tested on debian with only -lbfd installed (without -liberty); I'd appreciate if somebody on the other systems can test this new detection method.
--- tools/build/Makefile.feature | 4 ++-- tools/build/feature/Makefile | 10 ++++---- tools/perf/Makefile.config | 44 +++++++++++++++++++----------------- 3 files changed, 30 insertions(+), 28 deletions(-)
tools/build/Makefile.feature +2 -2
@@ -82,8 +82,8 @@ FEATURE_TESTS_EXTRA := \ cplus-demangle \ hello \ libbabeltrace \ - liberty \ - liberty-z \ + libbfd-liberty \ + libbfd-liberty-z \ libunwind-debug-frame \ libunwind-debug-frame-arm \ libunwind-debug-frame-aarch64 \
tools/build/feature/Makefile +5 -5
@@ -17,8 +17,8 @@ FILES= \ test-libbfd.bin \ test-disassembler-four-args.bin \ test-reallocarray.bin \ - test-liberty.bin \ - test-liberty-z.bin \ + test-libbfd-liberty.bin \ + test-libbfd-liberty-z.bin \ test-cplus-demangle.bin \ test-libelf.bin \ test-libelf-getphdrnum.bin \
@@ -210,7 +210,7 @@ $(OUTPUT)test-libpython-version.bin: $(BUILD) $(OUTPUT)test-libbfd.bin: - $(BUILD) -DPACKAGE='"perf"' -lbfd -lz -liberty -ldl + $(BUILD) -DPACKAGE='"perf"' -lbfd -ldl $(OUTPUT)test-disassembler-four-args.bin: $(BUILD) -DPACKAGE='"perf"' -lbfd -lopcodes
@@ -218,10 +218,10 @@ $(OUTPUT)test-disassembler-four-args.bin: $(OUTPUT)test-reallocarray.bin: $(BUILD) -$(OUTPUT)test-liberty.bin: +$(OUTPUT)test-libbfd-liberty.bin: $(CC) $(CFLAGS) -Wall -Werror -o $@ test-libbfd.c -DPACKAGE='"perf"' $(LDFLAGS) -lbfd -ldl -liberty -$(OUTPUT)test-liberty-z.bin: +$(OUTPUT)test-libbfd-liberty-z.bin: $(CC) $(CFLAGS) -Wall -Werror -o $@ test-libbfd.c -DPACKAGE='"perf"' $(LDFLAGS) -lbfd -ldl -liberty -lz $(OUTPUT)test-cplus-demangle.bin:
tools/perf/Makefile.config +23 -21
65 lines not on this page · raw message
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 4/29 perf trace: Do not hardcode the size of the tracepoint common_ fields
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
We shouldn't hardcode the size of the tracepoint common_ fields, use the offset of the 'id'/'__syscallnr' field in the sys_enter event instead.
This caused the augmented syscalls code to fail on a particular build of a PREEMPT_RT_FULL kernel where these extra 'common_migrate_disable' and 'common_padding' fields were before the syscall id one:
# cat /sys/kernel/debug/tracing/events/raw_syscalls/sys_enter/format name: sys_enter ID: 22 format: field:unsigned short common_type; offset:0; size:2; signed:0; field:unsigned char common_flags; offset:2; size:1; signed:0; field:unsigned char common_preempt_count; offset:3; size:1; signed:0; field:int common_pid; offset:4; size:4; signed:1; field:unsigned short common_migrate_disable; offset:8; size:2; signed:0; field:unsigned short common_padding; offset:10; size:2; signed:0;
field:long id; offset:16; size:8; signed:1; field:unsigned long args[6]; offset:24; size:48; signed:0;
print fmt: "NR %ld (%lx, %lx, %lx, %lx, %lx, %lx)", REC->id, REC->args[0], REC->args[1], REC->args[2], REC->args[3], REC->args[4], REC->args[5] #
All those 'common_' prefixed fields are zeroed when they hit a BPF tracepoint hook, we better just discard those, i.e. somehow pass an offset to the BPF program from the start of the ctx and make adjustments in the 'perf trace' handlers to adjust the offset of the syscall arg offsets obtained from tracefs.
Till then, fix it the quick way and add this to the augmented_raw_syscalls.c to bet it to work in such kernels:
diff --git a/tools/perf/examples/bpf/augmented_raw_syscalls.c b/tools/perf/examples/bpf/augmented_raw_syscalls.c
index 53c233370fae..1f746f931e13 100644
--- a/tools/perf/examples/bpf/augmented_raw_syscalls.c
+++ b/tools/perf/examples/bpf/augmented_raw_syscalls.c
@@ -38,12 +38,14 @@ struct bpf_map SEC("maps") syscalls = { struct syscall_enter_args {
unsigned long long common_tp_fields;
+ long rt_common_tp_fields;
long syscall_nr;
unsigned long args[6];
};
struct syscall_exit_args {
unsigned long long common_tp_fields;
+ long rt_common_tp_fields;
long syscall_nr;
long ret;
};Just to check that this was the case. Fix it properly later, for now remove the hardcoding of the offset in the 'perf trace' side and document the situation with this patch.
--- tools/perf/builtin-trace.c | 73 +++++++++++++++++++++++++++----------- 1 file changed, 52 insertions(+), 21 deletions(-)
tools/perf/builtin-trace.c +52 -21
@@ -112,8 +112,9 @@ struct trace { } stats;unsigned int max_stack;unsigned int min_stack; -bool sort_events; +int raw_augmented_syscalls_args_size;bool raw_augmented_syscalls; +bool sort_events;bool not_ev_qualifier;bool live;bool full_time;
@@ -283,12 +284,17 @@ static int perf_evsel__init_syscall_tp(struct perf_evsel *evsel)return -ENOENT ; } -static int perf_evsel__init_augmented_syscall_tp (struct perf_evsel *evsel) +static int perf_evsel__init_augmented_syscall_tp (struct perf_evsel *evsel,struct perf_evsel *tp) {struct syscall_tp *sc = evsel->priv =malloc (sizeof (struct syscall_tp )); -if (evsel->priv !=NULL ) {/* field, sizeof_field, offsetof_field */ -if (__tp_field__init_uint (&sc->id,sizeof (long ),sizeof (long long ), evsel->needs_swap)) +if (evsel->priv !=NULL ) { +struct tep_format_field *syscall_id =perf_evsel__field (tp,"id" ); +if (syscall_id ==NULL ) + syscall_id =perf_evsel__field (tp,"__syscall_nr" ); +if (syscall_id ==NULL ) + goto out_delete; +if (__tp_field__init_uint (&sc->id, syscall_id->size, syscall_id->offset, evsel->needs_swap)) goto out_delete;return 0 ;
@@ -1768,16 +1774,16 @@ static int trace__fprintf_sample(struct trace *trace, struct perf_evsel *evsel,return printed; } -static void *syscall__augmented_args (struct syscall *sc,struct perf_sample *sample,int *augmented_args_size,bool raw_augmented) +static void *syscall__augmented_args (struct syscall *sc,struct perf_sample *sample,int *augmented_args_size,int raw_augmented_args_size) {void *augmented_args =NULL ;/* * For now with BPF raw_augmented we hook into raw_syscalls:sys_enter -* and there we get all 6 syscall args plus the tracepoint common -* fields (sizeof(long)) and the syscall_nr (another long). So we check -* if that is the case and if so don't look after the sc->args_size, -* but always after the full raw_syscalls:sys_enter payload, which is -* fixed. +* and there we get all 6 syscall args plus the tracepoint common fields +* that gets calculated at the start and the syscall_nr (another long). +* So we check if that is the case and if so don't look after the +* sc->args_size but always after the full raw_syscalls:sys_enter payload, +* which is fixed. * * We'll revisit this later to pass s->args_size to the BPF augmenter * (now tools/perf/examples/bpf/augmented_raw_syscalls.c, so that it
@@ -1785,7 +1791,7 @@ static void *syscall__augmented_args(struct syscall *sc, struct perf_sample *sam* use syscalls:sys_enter_NAME, so that we reduce the kernel/userspace * traffic to just what is needed for each syscall. */ -int args_size = raw_augmented ? (8 * (int )sizeof (long )) : sc->args_size; +int args_size = raw_augmented_args_size ?: sc->args_size; *augmented_args_size = sample->raw_size - args_size;if (*augmented_args_size >0 )
@@ -1839,7 +1845,7 @@ static int trace__sys_enter(struct trace *trace, struct perf_evsel *evsel,* here and avoid using augmented syscalls when the evsel is the raw_syscalls one. */ if (evsel != trace->syscalls.events.sys_enter) - augmented_args =syscall__augmented_args (sc, sample, &augmented_args_size, trace->raw_augmented_syscalls); + augmented_args =syscall__augmented_args (sc, sample, &augmented_args_size, trace->raw_augmented_syscalls_args_size); ttrace->entry_time = sample->time; msg = ttrace->entry_str; printed +=scnprintf (msg + printed, trace__entry_str_size - printed,"%s(" , sc->name);
@@ -1897,7 +1903,7 @@ static int trace__fprintf_sys_enter(struct trace *trace, struct perf_evsel *evse goto out_put; args =perf_evsel__sc_tp_ptr (evsel, args, sample); - augmented_args =syscall__augmented_args (sc, sample, &augmented_args_size, trace->raw_augmented_syscalls); + augmented_args =syscall__augmented_args (sc, sample, &augmented_args_size, trace->raw_augmented_syscalls_args_size);syscall__scnprintf_args (sc, msg,sizeof (msg), args, augmented_args, augmented_args_size, trace, thread);fprintf (trace->output,"%s" , msg); err =0 ;
@@ -3814,13 +3820,6 @@ int cmd_trace(int argc, const char **argv)* syscall. */ if (trace.syscalls.events.augmented) { - evsel = trace.syscalls.events.augmented; - -if (perf_evsel__init_augmented_syscall_tp (evsel) || -perf_evsel__init_augmented_syscall_tp_args (evsel)) - goto out; - evsel->handler = trace__sys_enter; -evlist__for_each_entry (trace.evlist, evsel) {bool raw_syscalls_sys_exit =strcmp (perf_evsel__name (evsel),"raw_syscalls:sys_exit" ) ==0 ;
@@ -3829,9 +3828,41 @@ int cmd_trace(int argc, const char **argv) goto init_augmented_syscall_tp; } +if (strcmp (perf_evsel__name (evsel),"raw_syscalls:sys_enter" ) ==0 ) { +struct perf_evsel *augmented = trace.syscalls.events.augmented; +if (perf_evsel__init_augmented_syscall_tp (augmented, evsel) || +perf_evsel__init_augmented_syscall_tp_args (augmented)) + goto out; + augmented->handler = trace__sys_enter; + } +if (strstarts (perf_evsel__name (evsel),"syscalls:sys_exit_" )) { +struct syscall_tp *sc; init_augmented_syscall_tp: -perf_evsel__init_augmented_syscall_tp (evsel); +if (perf_evsel__init_augmented_syscall_tp (evsel, evsel)) + goto out; + sc = evsel->priv; +/* +* For now with BPF raw_augmented we hook into +* raw_syscalls:sys_enter and there we get all +* 6 syscall args plus the tracepoint common +* fields and the syscall_nr (another long). +* So we check if that is the case and if so +* don't look after the sc->args_size but +* always after the full raw_syscalls:sys_enter +* payload, which is fixed. +* +* We'll revisit this later to pass +* s->args_size to the BPF augmenter (now +* tools/perf/examples/bpf/augmented_raw_syscalls.c, +* so that it copies only what we need for each +* syscall, like what happens when we use +* syscalls:sys_enter_NAME, so that we reduce +* the kernel/userspace traffic to just what is +* needed for each syscall. +*/ +if (trace.raw_augmented_syscalls) + trace.raw_augmented_syscalls_args_size = (6 +1 ) *sizeof (long ) + sc->id.offset;perf_evsel__init_augmented_syscall_tp_ret (evsel); evsel->handler = trace__sys_exit; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 5/29 perf trace: Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*"
In reply to Arnaldo Carvalho de MeloFrom: Colin Ian King <[email protected]>
The spelling of the SECCOMP is incorrect, fix these.
--- tools/perf/trace/beauty/seccomp.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/trace/beauty/seccomp.c +2 -2
@@ -9,7 +9,7 @@static size_t syscall_arg__scnprintf_seccomp_op (char *bf,size_t size,struct syscall_arg *arg) {bool show_prefix = arg->show_string_prefix; -const char *prefix ="SECOMP_SET_MODE_" ; +const char *prefix ="SECCOMP_SET_MODE_" ;int op = arg->val;size_t printed =0 ;
@@ -34,7 +34,7 @@ static size_t syscall_arg__scnprintf_seccomp_flags(char *bf, size_t size, struct syscall_arg *arg) {bool show_prefix = arg->show_string_prefix; -const char *prefix ="SECOMP_FILTER_FLAG_" ; +const char *prefix ="SECCOMP_FILTER_FLAG_" ;int printed =0 , flags = arg->val;#define P_FLAG (n) \
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 6/29 perf python: Do not force closing original perf descriptor in evlist.get_pollfd()
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
Ondřej reported that when compiled with python3, the python extension regresses in evlist.get_pollfd function behaviour.
The evlist.get_pollfd function creates file objects from evlist's fds and returns them in a list. The python3 version also sets them to 'close the original descriptor' when the object dies (is closed), by passing True via the 'closefd' arg in the PyFile_FromFd call.
The python's closefd doc says:
If closefd is False, the underlying file descriptor will be kept open when the file is closed.
That's why the following line in python3 closes all evlist fds:
evlist.get_pollfd()
the returned list is immediately destroyed and that takes down the original events fds.
Passing closefd as False to PyFile_FromFd to fix this.
--- tools/perf/util/python.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-)
tools/perf/util/python.c +2 -1
@@ -939,7 +939,8 @@ static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist, file =PyFile_FromFile (fp,"perf" ,"r" ,NULL );#else - file =PyFile_FromFd (evlist->pollfd.entries[i].fd,"perf" ,"r" ,-1 ,NULL ,NULL ,NULL ,1 ); + file =PyFile_FromFd (evlist->pollfd.entries[i].fd,"perf" ,"r" ,-1 , +NULL ,NULL ,NULL ,0 );#endif if (file ==NULL ) goto free_list;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 7/29 perf script: Fix LBR skid dump problems in brstackinsn
In reply to Arnaldo Carvalho de MeloFrom: Andi Kleen <[email protected]>
This is a fix for another instance of the skid problem Milian recently found [1]
The LBRs don't freeze at the exact same time as the PMI is triggered. The perf script brstackinsn code that dumps LBR assembler assumes that the last branch in the LBR leads to the sample point. But with skid it's possible that the CPU executes one or more branches before the sample, but which do not appear in the LBR.
What happens then is either that the sample point is before the last LBR branch. In this case the dumper sees a negative length and ignores it. Or it the sample point is long after the last branch. Then the dumper sees a very long block and dumps it upto its block limit (16k bytes), which is noise in the output.
On typical sample session this can happen regularly.
This patch tries to detect and handle the situation. On the last block that is dumped by the LBR dumper we always stop on the first branch. If the block length is negative just scan forward to the first branch. Otherwise scan until a branch is found.
The PT decoder already has a function that uses the instruction decoder to detect branches, so we can just reuse it here.
Then when a terminating branch is found print an indication and stop dumping. This might miss a few instructions, but at least shows no runaway blocks.
--- tools/perf/builtin-script.c | 17 ++++++++++++++++- tools/perf/util/dump-insn.c | 8 ++++++++ tools/perf/util/dump-insn.h | 2 ++ .../intel-pt-decoder/intel-pt-insn-decoder.c | 8 ++++++++ 4 files changed, 34 insertions(+), 1 deletion(-)
tools/perf/builtin-script.c +16 -1
@@ -1073,9 +1073,18 @@ static int perf_sample__fprintf_brstackinsn(struct perf_sample *sample,/* * Print final block upto sample +* +* Due to pipeline delays the LBRs might be missing a branch +* or two, which can result in very large or negative blocks +* between final branch and sample. When this happens just +* continue walking after the last TO until we hit a branch. */ start = br->entries[0 ].to; end = sample->ip; +if (end < start) { +/* Missing jump. Scan 128 bytes for the next branch */ + end = start +128 ; + } len =grab_bb (buffer, start, end, machine, thread, &x.is64bit, &x.cpumode, true); printed +=ip__fprintf_sym (start, thread, x.cpumode, x.cpu, &lastsym, attr, fp);if (len <=0 ) {
@@ -1084,7 +1093,6 @@ static int perf_sample__fprintf_brstackinsn(struct perf_sample *sample, machine, thread, &x.is64bit, &x.cpumode, false);if (len <=0 ) goto out; - printed +=fprintf (fp,"\t%016" PRIx64"\t%s\n" , sample->ip,dump_insn (&x, sample->ip, buffer, len,NULL ));if (PRINT_FIELD (SRCCODE ))
@@ -1096,6 +1104,13 @@ static int perf_sample__fprintf_brstackinsn(struct perf_sample *sample, dump_insn(&x, start + off, buffer + off, len - off, &ilen));if (ilen ==0 )break ; +if (arch_is_branch (buffer + off, len - off, x.is64bit) && start + off != sample->ip) { +/* +* Hit a missing branch. Just stop. +*/ + printed +=fprintf (fp,"\t... not reaching sample ...\n" ); +break ; + }if (PRINT_FIELD (SRCCODE ))print_srccode (thread, x.cpumode, start + off); }
tools/perf/util/dump-insn.c +8 -0
@@ -13,3 +13,11 @@ const char *dump_insn(struct perf_insn *x __maybe_unused, *lenp =0 ;return "?" ; } + +__weak +int arch_is_branch (const unsigned char *buf__maybe_unused , +size_t len__maybe_unused , +int x86_64__maybe_unused ) +{ +return 0 ; +}
tools/perf/util/dump-insn.h +2 -0
@@ -20,4 +20,6 @@ struct perf_insn {const char *dump_insn (struct perf_insn *x,u64 ip,u8 *inbuf,int inlen,int *lenp); +int arch_is_branch (const unsigned char *buf,size_t len,int x86_64); +#endif
tools/perf/util/intel-pt-decoder/intel-pt-insn-decoder.c +8 -0
@@ -180,6 +180,14 @@ int intel_pt_get_insn(const unsigned char *buf, size_t len, int x86_64,return 0 ; } +int arch_is_branch (const unsigned char *buf,size_t len,int x86_64) +{ +struct intel_pt_insn in; +if (intel_pt_get_insn (buf, len, x86_64, &in) <0 ) +return -1 ; +return in.branch !=INTEL_PT_BR_NO_BRANCH ; +} +const char *dump_insn (struct perf_insn *x,uint64_t ip__maybe_unused ,u8 *inbuf,int inlen,int *lenp) {
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 8/29 perf trace: Rename thread_thread->paths to thread_trace->files
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
So that we can add more per file attributes besides the pathname, such as which ioctl beautifier to use, for cases such as the sound and usbdeffs ioctls, that both use the 'U' command, so we have to differentiate at the major number for the device file.
--- tools/perf/builtin-trace.c | 42 +++++++++++++++++++++----------------- 1 file changed, 23 insertions(+), 19 deletions(-)
tools/perf/builtin-trace.c +23 -19
@@ -957,6 +957,10 @@ static size_t fprintf_duration(unsigned long t, bool calculated, FILE *fp)return printed +fprintf (fp,"): " ); } +struct file { +char *pathname; +}; +/** * filename.ptr: The filename char pointer that will be vfs_getname'd * filename.entry_str_pos: Where to insert the string translated from
@@ -980,9 +984,9 @@ struct thread_trace {char *name; } filename;struct { -int max; -char **table; - } paths; +int max; +struct file *table; + } files;struct intlist *syscall_stats; };
@@ -992,7 +996,7 @@ static struct thread_trace *thread_trace__new(void)struct thread_trace *ttrace =zalloc (sizeof (struct thread_trace ));if (ttrace) - ttrace->paths.max =-1 ; + ttrace->files.max =-1 ; ttrace->syscall_stats =intlist__new (NULL );
@@ -1040,26 +1044,26 @@ static int trace__set_fd_pathname(struct thread *thread, int fd, const char *pat {struct thread_trace *ttrace =thread__priv (thread); -if (fd > ttrace->paths.max) { -char **npath =realloc (ttrace->paths.table, (fd +1 ) *sizeof (char *)); +if (fd > ttrace->files.max) { +struct file *nfiles =realloc (ttrace->files.table, (fd +1 ) *sizeof (struct file )); -if (npath ==NULL ) +if (nfiles ==NULL )return -1 ; -if (ttrace->paths.max !=-1 ) { -memset (npath + ttrace->paths.max +1 ,0 , - (fd - ttrace->paths.max) *sizeof (char *)); +if (ttrace->files.max !=-1 ) { +memset (nfiles + ttrace->files.max +1 ,0 , + (fd - ttrace->files.max) *sizeof (struct file )); }else { -memset (npath,0 , (fd +1 ) *sizeof (char *)); +memset (nfiles,0 , (fd +1 ) *sizeof (struct file )); } - ttrace->paths.table = npath; - ttrace->paths.max = fd; + ttrace->files.table = nfiles; + ttrace->files.max = fd; } - ttrace->paths.table[fd] =strdup (pathname); + ttrace->files.table[fd].pathname =strdup (pathname); -return ttrace->paths.table[fd] !=NULL ?0 :-1 ; +return ttrace->files.table[fd].pathname !=NULL ?0 :-1 ; }static int thread__read_fd_path (struct thread *thread,int fd)
@@ -1099,7 +1103,7 @@ static const char *thread__fd_path(struct thread *thread, int fd,if (fd <0 )return NULL ; -if ((fd > ttrace->paths.max || ttrace->paths.table[fd] ==NULL )) { +if ((fd > ttrace->files.max || ttrace->files.table[fd].pathname ==NULL )) {if (!trace->live)return NULL ; ++trace->stats.proc_getname;
@@ -1107,7 +1111,7 @@ static const char *thread__fd_path(struct thread *thread, int fd,return NULL ; } -return ttrace->paths.table[fd]; +return ttrace->files.table[fd].pathname; }size_t syscall_arg__scnprintf_fd (char *bf,size_t size,struct syscall_arg *arg)
@@ -1146,8 +1150,8 @@ static size_t syscall_arg__scnprintf_close_fd(char *bf, size_t size,size_t printed =syscall_arg__scnprintf_fd (bf, size, arg);struct thread_trace *ttrace =thread__priv (arg->thread); -if (ttrace && fd >=0 && fd <= ttrace->paths.max) -zfree (&ttrace->paths.table[fd]); +if (ttrace && fd >=0 && fd <= ttrace->files.max) +zfree (&ttrace->files.table[fd].pathname);return printed; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 9/29 perf trace: Move the files table resizing to outside set_pathname()
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
So that we can have that table expanded when setting other attributes.
--- tools/perf/builtin-trace.c | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-)
tools/perf/builtin-trace.c +16 -6
@@ -1040,15 +1040,13 @@ void syscall_arg__set_ret_scnprintf(struct syscall_arg *arg,static const size_t trace__entry_str_size =2048 ; -static int trace__set_fd_pathname (struct thread *thread,int fd,const char *pathname) +static struct file *thread_trace__files_entry (struct thread_trace *ttrace,int fd) { -struct thread_trace *ttrace =thread__priv (thread); -if (fd > ttrace->files.max) {struct file *nfiles =realloc (ttrace->files.table, (fd +1 ) *sizeof (struct file ));if (nfiles ==NULL ) -return -1 ; +return NULL ;if (ttrace->files.max !=-1 ) { memset(nfiles + ttrace->files.max + 1, 0,
@@ -1061,9 +1059,21 @@ static int trace__set_fd_pathname(struct thread *thread, int fd, const char *pat ttrace->files.max = fd; } - ttrace->files.table[fd].pathname =strdup (pathname); +return ttrace->files.table + fd; +} + +static int trace__set_fd_pathname (struct thread *thread,int fd,const char *pathname) +{ +struct thread_trace *ttrace =thread__priv (thread); +struct file *file =thread_trace__files_entry (ttrace, fd); + +if (file !=NULL ) { + file->pathname =strdup (pathname); +if (file->pathname) +return 0 ; + } -return ttrace->files.table[fd].pathname !=NULL ?0 :-1 ; +return -1 ; }static int thread__read_fd_path (struct thread *thread,int fd)
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 10/29 perf trace: Store the major number for a file when storing its pathname
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
We keep a table for the fds to map them back to pathnames when showing 'fd' based APIs such as write(), store as well the major number for the device the path is in, to use in things like choosing the right ioctl 'cmd' beautifier.
--- tools/perf/builtin-trace.c | 5 +++++ 1 file changed, 5 insertions(+)
tools/perf/builtin-trace.c +5 -0
@@ -60,6 +60,7 @@#include <linux/stringify.h> #include <linux/time64.h> #include <fcntl.h> +#include <sys/sysmacros.h> #include "sane_ctype.h"
@@ -959,6 +960,7 @@ static size_t fprintf_duration(unsigned long t, bool calculated, FILE *fp)struct file {char *pathname; +int dev_maj; };/**
@@ -1068,6 +1070,9 @@ static int trace__set_fd_pathname(struct thread *thread, int fd, const char *patstruct file *file =thread_trace__files_entry (ttrace, fd);if (file !=NULL ) { +struct stat st; +if (stat (pathname, &st) ==0 ) + file->dev_maj =major (st.st_rdev); file->pathname =strdup (pathname);if (file->pathname)return 0 ;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 11/29 tools headers uapi: Grab a copy of usbdevice_fs.h
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Will be used to generate the string table for the USBDEVFS_ prefixed ioctl commands.
--- tools/include/uapi/linux/usbdevice_fs.h | 201 ++++++++++++++++++++++++ tools/perf/check-headers.sh | 1 + 2 files changed, 202 insertions(+) create mode 100644 tools/include/uapi/linux/usbdevice_fs.h
tools/include/uapi/linux/usbdevice_fs.h new file +201 -0
@@ -0,0 +1,201 @@ +/* SPDX-License-Identifier: GPL-2.0+ WITH Linux-syscall-note */ +/*****************************************************************************/ + +/* +* usbdevice_fs.h -- USB device file system. +* +* Copyright (C) 2000 +* Thomas Sailer ([email protected]) +* +* This program is free software; you can redistribute it and/or modify +* it under the terms of the GNU General Public License as published by +* the Free Software Foundation; either version 2 of the License, or +* (at your option) any later version. +* +* This program is distributed in the hope that it will be useful, +* but WITHOUT ANY WARRANTY; without even the implied warranty of +* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +* GNU General Public License for more details. +* +* You should have received a copy of the GNU General Public License +* along with this program; if not, write to the Free Software +* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. +* +* History: +* 0.1 04.01.2000 Created +*/ + +/*****************************************************************************/ + +#ifndef _UAPI_LINUX_USBDEVICE_FS_H +#define _UAPI_LINUX_USBDEVICE_FS_H + +#include <linux/types.h> +#include <linux/magic.h> + +/* --------------------------------------------------------------------- */ + +/* usbdevfs ioctl codes */ + +struct usbdevfs_ctrltransfer { +__u8 bRequestType; +__u8 bRequest; +__u16 wValue; +__u16 wIndex; +__u16 wLength; +__u32 timeout;/* in milliseconds */ +void __user *data; +}; + +struct usbdevfs_bulktransfer { +unsigned int ep; +unsigned int len; +unsigned int timeout;/* in milliseconds */ +void __user *data; +}; + +struct usbdevfs_setinterface { +unsigned int interface; +unsigned int altsetting; +}; + +struct usbdevfs_disconnectsignal { +unsigned int signr; +void __user *context; +}; + +#define USBDEVFS_MAXDRIVERNAME 255 + +struct usbdevfs_getdriver { +unsigned int interface; +char driver[USBDEVFS_MAXDRIVERNAME +1 ]; +}; + +struct usbdevfs_connectinfo { +unsigned int devnum; +unsigned char slow; +}; + +#define USBDEVFS_URB_SHORT_NOT_OK 0x01 +#define USBDEVFS_URB_ISO_ASAP 0x02 +#define USBDEVFS_URB_BULK_CONTINUATION 0x04 +#define USBDEVFS_URB_NO_FSBR 0x20/* Not used */ +#define USBDEVFS_URB_ZERO_PACKET 0x40 +#define USBDEVFS_URB_NO_INTERRUPT 0x80 + +#define USBDEVFS_URB_TYPE_ISO 0 +#define USBDEVFS_URB_TYPE_INTERRUPT 1 +#define USBDEVFS_URB_TYPE_CONTROL 2 +#define USBDEVFS_URB_TYPE_BULK 3 + +struct usbdevfs_iso_packet_desc { +unsigned int length; +unsigned int actual_length; +unsigned int status; +}; + +struct usbdevfs_urb { +unsigned char type; +unsigned char endpoint; +int status; +unsigned int flags; +void __user *buffer; +int buffer_length; +int actual_length; +int start_frame; +union { +int number_of_packets;/* Only used for isoc urbs */ +unsigned int stream_id;/* Only used with bulk streams */ + }; +int error_count; +unsigned int signr;/* signal to be sent on completion, +or 0 if none should be sent. */ +void __user *usercontext; +struct usbdevfs_iso_packet_desc iso_frame_desc[0 ]; +}; + +/* ioctls for talking directly to drivers */ +struct usbdevfs_ioctl { +int ifno;/* interface 0..N ; negative numbers reserved */ +int ioctl_code;/* MUST encode size + direction of data so the +* macros in <asm/ioctl.h> give correct values */ +void __user *data;/* param buffer (in, or out) */ +}; + +/* You can do most things with hubs just through control messages, +* except find out what device connects to what port. */ +struct usbdevfs_hub_portinfo { +char nports;/* number of downstream ports in this hub */ +char port [127 ];/* e.g. port 3 connects to device 27 */ +}; + +/* System and bus capability flags */ +#define USBDEVFS_CAP_ZERO_PACKET 0x01 +#define USBDEVFS_CAP_BULK_CONTINUATION 0x02 +#define USBDEVFS_CAP_NO_PACKET_SIZE_LIM 0x04 +#define USBDEVFS_CAP_BULK_SCATTER_GATHER 0x08 +#define USBDEVFS_CAP_REAP_AFTER_DISCONNECT 0x10 +#define USBDEVFS_CAP_MMAP 0x20 +#define USBDEVFS_CAP_DROP_PRIVILEGES 0x40 + +/* USBDEVFS_DISCONNECT_CLAIM flags & struct */ + +/* disconnect-and-claim if the driver matches the driver field */ +#define USBDEVFS_DISCONNECT_CLAIM_IF_DRIVER 0x01 +/* disconnect-and-claim except when the driver matches the driver field */ +#define USBDEVFS_DISCONNECT_CLAIM_EXCEPT_DRIVER 0x02 + +struct usbdevfs_disconnect_claim { +unsigned int interface; +unsigned int flags; +char driver[USBDEVFS_MAXDRIVERNAME +1 ]; +}; + +struct usbdevfs_streams { +unsigned int num_streams;/* Not used by USBDEVFS_FREE_STREAMS */ +unsigned int num_eps; +unsigned char eps[0 ]; +}; + +/* +* USB_SPEED_* values returned by USBDEVFS_GET_SPEED are defined in +* linux/usb/ch9.h +*/ + +#define USBDEVFS_CONTROL _IOWR('U', 0, struct usbdevfs_ctrltransfer) +#define USBDEVFS_CONTROL32 _IOWR('U', 0, struct usbdevfs_ctrltransfer32) +#define USBDEVFS_BULK _IOWR('U', 2, struct usbdevfs_bulktransfer) +#define USBDEVFS_BULK32 _IOWR('U', 2, struct usbdevfs_bulktransfer32) +#define USBDEVFS_RESETEP _IOR('U', 3, unsigned int) +#define USBDEVFS_SETINTERFACE _IOR('U', 4, struct usbdevfs_setinterface) +#define USBDEVFS_SETCONFIGURATION _IOR('U', 5, unsigned int) +#define USBDEVFS_GETDRIVER _IOW('U', 8, struct usbdevfs_getdriver) +#define USBDEVFS_SUBMITURB _IOR('U', 10, struct usbdevfs_urb) +#define USBDEVFS_SUBMITURB32 _IOR('U', 10, struct usbdevfs_urb32) +#define USBDEVFS_DISCARDURB _IO('U', 11) +#define USBDEVFS_REAPURB _IOW('U', 12, void *) +#define USBDEVFS_REAPURB32 _IOW('U', 12, __u32) +#define USBDEVFS_REAPURBNDELAY _IOW('U', 13, void *) +#define USBDEVFS_REAPURBNDELAY32 _IOW('U', 13, __u32) +#define USBDEVFS_DISCSIGNAL _IOR('U', 14, struct usbdevfs_disconnectsignal) +#define USBDEVFS_DISCSIGNAL32 _IOR('U', 14, struct usbdevfs_disconnectsignal32) +#define USBDEVFS_CLAIMINTERFACE _IOR('U', 15, unsigned int) +#define USBDEVFS_RELEASEINTERFACE _IOR('U', 16, unsigned int) +#define USBDEVFS_CONNECTINFO _IOW('U', 17, struct usbdevfs_connectinfo) +#define USBDEVFS_IOCTL _IOWR('U', 18, struct usbdevfs_ioctl) +#define USBDEVFS_IOCTL32 _IOWR('U', 18, struct usbdevfs_ioctl32) +#define USBDEVFS_HUB_PORTINFO _IOR('U', 19, struct usbdevfs_hub_portinfo) +#define USBDEVFS_RESET _IO('U', 20) +#define USBDEVFS_CLEAR_HALT _IOR('U', 21, unsigned int) +#define USBDEVFS_DISCONNECT _IO('U', 22) +#define USBDEVFS_CONNECT _IO('U', 23) +#define USBDEVFS_CLAIM_PORT _IOR('U', 24, unsigned int) +#define USBDEVFS_RELEASE_PORT _IOR('U', 25, unsigned int) +#define USBDEVFS_GET_CAPABILITIES _IOR('U', 26, __u32) +#define USBDEVFS_DISCONNECT_CLAIM _IOR('U', 27, struct usbdevfs_disconnect_claim) +#define USBDEVFS_ALLOC_STREAMS _IOR('U', 28, struct usbdevfs_streams) +#define USBDEVFS_FREE_STREAMS _IOR('U', 29, struct usbdevfs_streams) +#define USBDEVFS_DROP_PRIVILEGES _IOW('U', 30, __u32) +#define USBDEVFS_GET_SPEED _IO('U', 31) + +#endif /* _UAPI_LINUX_USBDEVICE_FS_H */
tools/perf/check-headers.sh +1 -0
@@ -14,6 +14,7 @@ include/uapi/linux/perf_event.h include/uapi/linux/prctl.h include/uapi/linux/sched.h include/uapi/linux/stat.h +include/uapi/linux/usbdevice_fs.h include/uapi/linux/vhost.h include/uapi/sound/asound.h include/linux/bits.h
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 12/29 perf beauty ioctl: Add generator for USBDEVFS_ ioctl commands
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
Will be associated with fds with the right device major.
$ tools/perf/trace/beauty/usbdevfs_ioctl.sh
static const char *usbdevfs_ioctl_cmds[] = {
[0] = "CONTROL",
[10] = "SUBMITURB",
[11] = "DISCARDURB",
[12] = "REAPURB",
[13] = "REAPURBNDELAY",
[14] = "DISCSIGNAL",
[15] = "CLAIMINTERFACE",
[16] = "RELEASEINTERFACE",
[17] = "CONNECTINFO",
[18] = "IOCTL",
[19] = "HUB_PORTINFO",
[20] = "RESET",
[21] = "CLEAR_HALT",
[22] = "DISCONNECT",
[23] = "CONNECT",
[24] = "CLAIM_PORT",
[25] = "RELEASE_PORT",
[26] = "GET_CAPABILITIES",
[27] = "DISCONNECT_CLAIM",
[28] = "ALLOC_STREAMS",
[29] = "FREE_STREAMS",
[2] = "BULK",
[30] = "DROP_PRIVILEGES",
[31] = "GET_SPEED",
[3] = "RESETEP",
[4] = "SETINTERFACE",
[5] = "SETCONFIGURATION",
[8] = "GETDRIVER",
};#if 0
static const char *usbdevfs_ioctl_32_cmds[] = {
[0] = "CONTROL32",
[10] = "SUBMITURB32",
[12] = "REAPURB32",
[13] = "REAPURBNDELAY32",
[14] = "DISCSIGNAL32",
[18] = "IOCTL32",
[2] = "BULK32", };
#endif
$
Leaving the '32' variants commented, later we can try to support those as well, from some other hint (maybe something about the thread issuing the ioctls) and from the _IOC_SIZE(cmd).
--- tools/perf/trace/beauty/usbdevfs_ioctl.sh | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) create mode 100755 tools/perf/trace/beauty/usbdevfs_ioctl.sh
tools/perf/trace/beauty/usbdevfs_ioctl.sh new file +19 -0
@@ -0,0 +1,19 @@ +#!/bin/sh +# SPDX-License-Identifier: LGPL-2.1 + +[ $# -eq 1 ] && header_dir=$1 || header_dir=tools/include/uapi/linux/ + +printf "static const char *usbdevfs_ioctl_cmds[] = {\n" +regex="^#[[:space:]]*define[[:space:]]+USBDEVFS_(\w+)[[:space:]]+_IO[WR]{0,2}\([[:space:]]*'U'[[:space:]]*,[[:space:]]*([[:digit:]]+).*" +egrep $regex ${header_dir}/usbdevice_fs.h | egrep -v 'USBDEVFS_\w+32[[:space:]]' | \ + sed -r "s/$regex/\2 \1/g" | \ + sort | xargs printf "\t[%s] = \"%s\",\n" +printf "};\n\n" +printf "#if 0\n" +printf "static const char *usbdevfs_ioctl_32_cmds[] = {\n" +regex="^#[[:space:]]*define[[:space:]]+USBDEVFS_(\w+)[[:space:]]+_IO[WR]{0,2}\([[:space:]]*'U'[[:space:]]*,[[:space:]]*([[:digit:]]+).*" +egrep $regex ${header_dir}/usbdevice_fs.h | egrep 'USBDEVFS_\w+32[[:space:]]' | \ + sed -r "s/$regex/\2 \1/g" | \ + sort | xargs printf "\t[%s] = \"%s\",\n" +printf "};\n" +printf "#endif\n"
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 13/29 perf trace: Wire up ioctl's USBDEBFS_ cmd table generator
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
That ends up generating this:
[acme@quaco perf]$ cat /tmp/build/perf/trace/beauty/generated/ioctl/usbdevfs_ioctl_array.c
static const char *usbdevfs_ioctl_cmds[] = {
[0] = "CONTROL",
[10] = "SUBMITURB",
[11] = "DISCARDURB",
[12] = "REAPURB",
[13] = "REAPURBNDELAY",
[14] = "DISCSIGNAL",
[15] = "CLAIMINTERFACE",
[16] = "RELEASEINTERFACE",
[17] = "CONNECTINFO",
[18] = "IOCTL",
[19] = "HUB_PORTINFO",
[2] = "BULK",
[20] = "RESET",
[21] = "CLEAR_HALT",
[22] = "DISCONNECT",
[23] = "CONNECT",
[24] = "CLAIM_PORT",
[25] = "RELEASE_PORT",
[26] = "GET_CAPABILITIES",
[27] = "DISCONNECT_CLAIM",
[28] = "ALLOC_STREAMS",
[29] = "FREE_STREAMS",
[3] = "RESETEP",
[30] = "DROP_PRIVILEGES",
[31] = "GET_SPEED",
[4] = "SETINTERFACE",
[5] = "SETCONFIGURATION",
[8] = "GETDRIVER",};
#if 0
static const char *usbdevfs_ioctl_32_cmds[] = {
[0] = "CONTROL32",
[10] = "SUBMITURB32",
[12] = "REAPURB32",
[13] = "REAPURBNDELAY32",
[14] = "DISCSIGNAL32",
[18] = "IOCTL32",
[2] = "BULK32", };
#endif
$
--- tools/perf/Makefile.perf | 8 ++++++++ 1 file changed, 8 insertions(+)
tools/perf/Makefile.perf +8 -0
@@ -497,6 +497,12 @@ prctl_option_tbl := $(srctree)/tools/perf/trace/beauty/prctl_option.sh $(prctl_option_array): $(prctl_hdr_dir)/prctl.h $(prctl_option_tbl) $(Q)$(SHELL) '$(prctl_option_tbl)' $(prctl_hdr_dir) > $@ +usbdevfs_ioctl_array := $(beauty_ioctl_outdir)/usbdevfs_ioctl_array.c +usbdevfs_ioctl_tbl := $(srctree)/tools/perf/trace/beauty/usbdevfs_ioctl.sh + +$(usbdevfs_ioctl_array): $(linux_uapi_dir)/usbdevice_fs.h $(usbdevfs_ioctl_tbl) + $(Q)$(SHELL) '$(usbdevfs_ioctl_tbl)' $(linux_uapi_dir) > $@ + x86_arch_prctl_code_array := $(beauty_outdir)/x86_arch_prctl_code_array.c x86_arch_prctl_code_tbl := $(srctree)/tools/perf/trace/beauty/x86_arch_prctl.sh
@@ -624,6 +630,7 @@ prepare: $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h archheaders $(drm_ioc $(mount_flags_array) \ $(perf_ioctl_array) \ $(prctl_option_array) \ + $(usbdevfs_ioctl_array) \ $(x86_arch_prctl_code_array) \ $(rename_flags_array) \ $(arch_errno_name_array)
@@ -923,6 +930,7 @@ clean:: $(LIBTRACEEVENT)-clean $(LIBAPI)-clean $(LIBBPF)-clean $(LIBSUBCMD)-clea $(OUTPUT)$(vhost_virtio_ioctl_array) \ $(OUTPUT)$(perf_ioctl_array) \ $(OUTPUT)$(prctl_option_array) \ + $(OUTPUT)$(usbdevfs_ioctl_array) \ $(OUTPUT)$(x86_arch_prctl_code_array) \ $(OUTPUT)$(rename_flags_array) \ $(OUTPUT)$(arch_errno_name_array)
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 14/29 perf trace beauty: Export function to get the files for a thread
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
So that beautifiers can access things like dev_maj.
--- tools/perf/builtin-trace.c | 10 +++++----- tools/perf/trace/beauty/beauty.h | 7 +++++++ 2 files changed, 12 insertions(+), 5 deletions(-)
tools/perf/builtin-trace.c +5 -5
@@ -958,11 +958,6 @@ static size_t fprintf_duration(unsigned long t, bool calculated, FILE *fp)return printed +fprintf (fp,"): " ); } -struct file { -char *pathname; -int dev_maj; -}; -/** * filename.ptr: The filename char pointer that will be vfs_getname'd * filename.entry_str_pos: Where to insert the string translated from
@@ -1064,6 +1059,11 @@ static struct file *thread_trace__files_entry(struct thread_trace *ttrace, int freturn ttrace->files.table + fd; } +struct file *thread__files_entry (struct thread *thread,int fd) +{ +return thread_trace__files_entry (thread__priv (thread), fd); +} +static int trace__set_fd_pathname (struct thread *thread,int fd,const char *pathname) {struct thread_trace *ttrace =thread__priv (thread);
tools/perf/trace/beauty/beauty.h +7 -0
@@ -32,6 +32,13 @@ size_t strarray__scnprintf_flags(struct strarray *sa, char *bf, size_t size, boostruct trace ;struct thread ; +struct file { +char *pathname; +int dev_maj; +}; + +struct file *thread__files_entry (struct thread *thread,int fd); +struct strarrays {int nr_entries;struct strarray **entries;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 15/29 perf trace beauty ioctl: Beautify USBDEVFS_ commands
In reply to Arnaldo Carvalho de MeloFrom: Arnaldo Carvalho de Melo <[email protected]>
For instance, while debugging the 'galileo' python utility to synchronize fitbit trackers:
# perf trace -e ioctl ./run --force ioctl(0</dev/pts/8>, TCSETS, 0x7ffe28666420) = 0 ioctl(0</dev/pts/8>, TCSETS, 0x7ffe28666290) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe28666290) = 0 ioctl(2</dev/pts/8>, TCSETS, 0x7ffe28666290) = 0 ioctl(3</home/acme/hg/galileo/run>, TCSETS, 0x7ffe286663f0) = -1 ENOTTY (Inappropriate ioctl for device) ioctl(1</dev/pts/8>, TCSETS, 0x7ffe286655a0) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe28665470) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe28665470) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe286654a0) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe286654a0) = 0 ioctl(1</dev/pts/8>, TCSETS, 0x7ffe28665400) = 0 ioctl(1</dev/pts/8>, TIOCSWINSZ, 0x7ffe286654c0) = 0 ioctl(0</dev/pts/8>, TIOCSWINSZ, 0x7ffe28665560) = 0 ioctl(0</dev/pts/8>, TIOCSWINSZ, 0x7ffe28665560) = 0 ioctl(0</dev/pts/8>, TIOCMGET, 0x7ffe28665560) = 0 ioctl(0</dev/pts/8>, TCSETS, 0x7ffe28665530) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_GET_CAPABILITIES, 0x561468dad048) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_GETDRIVER, 0x7ffe28665500) = -1 ENODATA (No data available) ioctl(10</dev/bus/usb/001/011>, USBDEVFS_GETDRIVER, 0x7ffe28665500) = -1 ENODATA (No data available) ioctl(10</dev/bus/usb/001/011>, USBDEVFS_SETCONFIGURATION, 0x7ffe2866513c) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_CLAIMINTERFACE, 0x7ffe286647bc) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_SUBMITURB, 0x561468dace40) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664c10) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664c10) = -1 EAGAIN (Resource temporarily unavailable) ioctl(10</dev/bus/usb/001/011>, USBDEVFS_SUBMITURB, 0x561468dace40) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664dd0) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664dd0) = -1 EAGAIN (Resource temporarily unavailable) <SNIP> ioctl(10</dev/bus/usb/001/011>, USBDEVFS_SUBMITURB, 0x561468e72ec0) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664cc0) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_REAPURBNDELAY, 0x7ffe28664cc0) = -1 EAGAIN (Resource temporarily unavailable) ioctl(10</dev/bus/usb/001/011>, USBDEVFS_RELEASEINTERFACE, 0x7ffe2866463c) = 0 ioctl(10</dev/bus/usb/001/011>, USBDEVFS_RELEASEINTERFACE, 0x7ffe2866463c) = 0 Tracker: 813F4690C3D1: Synchronisation successful #
--- tools/perf/trace/beauty/ioctl.c | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+)
tools/perf/trace/beauty/ioctl.c +22 -0
@@ -112,6 +112,17 @@ static size_t ioctl__scnprintf_perf_cmd(int nr, int dir, char *bf, size_t size)return scnprintf (bf, size,"(%#x, %#x, %#x)" ,0xAE , nr, dir); } +static size_t ioctl__scnprintf_usbdevfs_cmd (int nr,int dir,char *bf,size_t size) +{ +#include "trace/beauty/generated/ioctl/usbdevfs_ioctl_array.c" + static DEFINE_STRARRAY(usbdevfs_ioctl_cmds, ""); + +if (nr < strarray__usbdevfs_ioctl_cmds.nr_entries && strarray__usbdevfs_ioctl_cmds.entries[nr] !=NULL ) +return scnprintf (bf, size,"USBDEVFS_%s" , strarray__usbdevfs_ioctl_cmds.entries[nr]); + +return scnprintf (bf, size,"(%c, %#x, %#x)" ,'U' , nr, dir); +} +static size_t ioctl__scnprintf_cmd (unsigned long cmd,char *bf,size_t size,bool show_prefix) {const char *prefix ="_IOC_" ;
@@ -157,9 +168,20 @@ static size_t ioctl__scnprintf_cmd(unsigned long cmd, char *bf, size_t size, booreturn printed +scnprintf (bf + printed, size - printed,", %#x, %#x, %#x)" , type, nr, sz); } +#ifndef USB_DEVICE_MAJOR +#define USB_DEVICE_MAJOR 189 +#endif // USB_DEVICE_MAJOR +size_t syscall_arg__scnprintf_ioctl_cmd (char *bf,size_t size,struct syscall_arg *arg) {unsigned long cmd = arg->val; +unsigned int fd =syscall_arg__val (arg,0 ); +struct file *file =thread__files_entry (arg->thread, fd); + +if (file !=NULL ) { +if (file->dev_maj ==USB_DEVICE_MAJOR ) +return ioctl__scnprintf_usbdevfs_cmd (_IOC_NR (cmd),_IOC_DIR (cmd), bf, size); + }return ioctl__scnprintf_cmd (cmd, bf, size, arg->show_string_prefix); }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 16/29 perf c2c: Change the default coalesce setup
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
Joe suggested to have the coalesce default set just to 'iaddr', because it's easier to read on the default 'perf c2c report' output.
By removing the "pid" field from the default -c/--coalesce option, the 'perf c2c' report will group all the relevant PIDs under the instruction address ('iaddr') bucket. User can always run "-c pid,iaddr" for a more fine grained output on particular PIDs.
--- tools/perf/builtin-c2c.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/builtin-c2c.c +1 -1
@@ -68,7 +68,7 @@ struct c2c_hist_entry {struct hist_entry he; }; -static char const *coalesce_default ="pid,iaddr" ; +static char const *coalesce_default ="iaddr" ;struct perf_c2c {struct perf_tool tool;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 17/29 perf c2c: Increase the HITM ratio limit for displayed cachelines
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
The cachelines being reported are the ones with percentages all the way down to 0.05%. That makes for very long output files. Raising that to 0.1%. The user can always specify --show-all if they want all the cachelines with hits.
--- tools/perf/builtin-c2c.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/builtin-c2c.c +1 -1
@@ -1878,7 +1878,7 @@ static int c2c_hists__reinit(struct c2c_hists *c2c_hists,return hpp_list__parse (&c2c_hists->list, output, sort); } -#define DISPLAY_LINE_LIMIT 0.0005 +#define DISPLAY_LINE_LIMIT 0.001static bool he__display (struct hist_entry *he,struct c2c_stats *stats) {
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 18/29 tools power x86_energy_perf_policy: Override CFLAGS assignments and add LDFLAGS to build command
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
So user could specify outside CFLAGS/LDFLAGS values.
--- tools/power/x86/x86_energy_perf_policy/Makefile | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
tools/power/x86/x86_energy_perf_policy/Makefile +3 -3
@@ -9,12 +9,12 @@ ifeq ("$(origin O)", "command line") endif x86_energy_perf_policy : x86_energy_perf_policy.c -CFLAGS += -Wall -CFLAGS += -DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"' +override CFLAGS += -Wall +override CFLAGS += -DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"' %: %.c @mkdir -p $(BUILD_OUTPUT) - $(CC) $(CFLAGS) $< -o $(BUILD_OUTPUT)/$@ + $(CC) $(CFLAGS) $< -o $(BUILD_OUTPUT)/$@ $(LDFLAGS) .PHONY : clean clean :
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 19/29 tools thermal tmon: Allow overriding CFLAGS assignments
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
So that the user can provide, e.g. distro package alternative values.
--- tools/thermal/tmon/Makefile | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
tools/thermal/tmon/Makefile +4 -4
@@ -6,13 +6,13 @@ VERSION = 1.0 BINDIR=usr/bin WARNFLAGS=-Wall -Wshadow -W -Wformat -Wimplicit-function-declaration -Wimplicit-int -CFLAGS+= -O1 ${WARNFLAGS} +override CFLAGS+= -O1 ${WARNFLAGS} # Add "-fstack-protector" only if toolchain supports it. -CFLAGS+= $(call cc-option,-fstack-protector) +override CFLAGS+= $(call cc-option,-fstack-protector-strong) CC?= $(CROSS_COMPILE)gcc PKG_CONFIG?= pkg-config -CFLAGS+=-D VERSION=\"$(VERSION)\" +override CFLAGS+=-D VERSION=\"$(VERSION)\" LDFLAGS+= TARGET=tmon
@@ -29,7 +29,7 @@ TMON_LIBS += $(shell $(PKG_CONFIG) --libs $(STATIC) panelw ncursesw 2> /dev/null $(PKG_CONFIG) --libs $(STATIC) panel ncurses 2> /dev/null || \ echo -lpanel -lncurses) -CFLAGS += $(shell $(PKG_CONFIG) --cflags $(STATIC) panelw ncursesw 2> /dev/null || \ +override CFLAGS += $(shell $(PKG_CONFIG) --cflags $(STATIC) panelw ncursesw 2> /dev/null || \ $(PKG_CONFIG) --cflags $(STATIC) panel ncurses 2> /dev/null) OBJS = tmon.o tui.o sysfs.o pid.o
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 20/29 tools power turbostat: Override CFLAGS assignments and add LDFLAGS to build command
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
So that the user can specify outside CFLAGS/LDFLAGS values.
--- tools/power/x86/turbostat/Makefile | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)
tools/power/x86/turbostat/Makefile +4 -4
@@ -9,13 +9,13 @@ ifeq ("$(origin O)", "command line") endif turbostat : turbostat.c -CFLAGS += -Wall -CFLAGS += -DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"' -CFLAGS += -DINTEL_FAMILY_HEADER='"../../../../arch/x86/include/asm/intel-family.h"' +override CFLAGS += -Wall +override CFLAGS += -DMSRHEADER='"../../../../arch/x86/include/asm/msr-index.h"' +override CFLAGS += -DINTEL_FAMILY_HEADER='"../../../../arch/x86/include/asm/intel-family.h"' %: %.c @mkdir -p $(BUILD_OUTPUT) - $(CC) $(CFLAGS) $< -o $(BUILD_OUTPUT)/$@ + $(CC) $(CFLAGS) $< -o $(BUILD_OUTPUT)/$@ $(LDFLAGS) .PHONY : clean clean :
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 21/29 tools gpio: Allow overriding CFLAGS
In reply to Arnaldo Carvalho de MeloFrom: Jiri Olsa <[email protected]>
So that the user can specify outside CFLAGS values.
--- tools/gpio/Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
tools/gpio/Makefile +1 -1
@@ -12,7 +12,7 @@ endif # (this improves performance and avoids hard-to-debug behaviour); MAKEFLAGS += -r -CFLAGS += -O2 -Wall -g -D_GNU_SOURCE -I$(OUTPUT)include +override CFLAGS += -O2 -Wall -g -D_GNU_SOURCE -I$(OUTPUT)include ALL_TARGETS := lsgpio gpio-hammer gpio-event-mon ALL_PROGRAMS := $(patsubst %,$(OUTPUT)%,$(ALL_TARGETS))
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 22/29 perf thread-stack: Simplify some code in thread_stack__process()
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, simplify some code in thread_stack__process().
--- tools/perf/util/thread-stack.c | 18 +++++++----------- 1 file changed, 7 insertions(+), 11 deletions(-)
tools/perf/util/thread-stack.c +7 -11
@@ -593,17 +593,13 @@ int thread_stack__process(struct thread *thread, struct comm *comm,struct thread_stack *ts = thread->ts;int err =0 ; -if (ts) { -if (!ts->crp) { -/* Supersede thread_stack__event() */ -thread_stack__free (thread); - thread->ts =thread_stack__new (thread, crp); -if (!thread->ts) -return -ENOMEM ; - ts = thread->ts; - ts->comm = comm; - } - }else { +if (ts && !ts->crp) { +/* Supersede thread_stack__event() */ +thread_stack__free (thread); + ts =NULL ; + } + +if (!ts) { thread->ts =thread_stack__new (thread, crp);if (!thread->ts)return -ENOMEM ;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 23/29 perf thread-stack: Tidy thread_stack__bottom() usage
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, tidy thread_stack__bottom() usage. Specifically, the parameter 'thread' is not needed.
--- tools/perf/util/thread-stack.c | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)
tools/perf/util/thread-stack.c +3 -4
@@ -449,7 +449,7 @@ static int thread_stack__pop_cp(struct thread *thread, struct thread_stack *ts,return 1 ; } -static int thread_stack__bottom (struct thread *thread,struct thread_stack *ts, +static int thread_stack__bottom (struct thread_stack *ts,struct perf_sample *sample,struct addr_location *from_al,struct addr_location *to_al,u64 ref)
@@ -474,7 +474,7 @@ static int thread_stack__bottom(struct thread *thread, struct thread_stack *ts,if (!cp)return -ENOMEM ; -return thread_stack__push_cp (thread->ts, ip, sample->time, ref, cp, +return thread_stack__push_cp (ts, ip, sample->time, ref, cp, true, false); }
@@ -617,8 +617,7 @@ int thread_stack__process(struct thread *thread, struct comm *comm,/* If the stack is empty, put the current symbol on the stack */ if (!ts->cnt) { - err =thread_stack__bottom (thread, ts, sample, from_al, to_al, - ref); + err =thread_stack__bottom (ts, sample, from_al, to_al, ref);if (err)return err; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 24/29 perf thread-stack: Avoid direct reference to the thread's stack
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, avoid direct reference to the thread's stack. The thread stack will change to an array of thread stacks, at which point the meaning of the direct reference will change.
--- tools/perf/util/thread-stack.c | 81 ++++++++++++++++++++-------------- 1 file changed, 49 insertions(+), 32 deletions(-)
tools/perf/util/thread-stack.c +49 -32
@@ -111,9 +111,16 @@ static struct thread_stack *thread_stack__new(struct thread *thread, ts->kernel_start =1ULL <<63 ; ts->crp = crp; + thread->ts = ts; +return ts; } +static inline struct thread_stack *thread__stack (struct thread *thread) +{ +return thread ? thread->ts :NULL ; +} +static int thread_stack__push (struct thread_stack *ts,u64 ret_addr,bool trace_end) {
@@ -226,8 +233,10 @@ static int __thread_stack__flush(struct thread *thread, struct thread_stack *ts)int thread_stack__flush (struct thread *thread) { -if (thread->ts) -return __thread_stack__flush (thread, thread->ts); +struct thread_stack *ts = thread->ts; + +if (ts) +return __thread_stack__flush (thread, ts);return 0 ; }
@@ -235,16 +244,18 @@ int thread_stack__flush(struct thread *thread)int thread_stack__event (struct thread *thread,u32 flags,u64 from_ip,u64 to_ip,u16 insn_len,u64 trace_nr) { +struct thread_stack *ts =thread__stack (thread); +if (!thread)return -EINVAL ; -if (!thread->ts) { - thread->ts =thread_stack__new (thread,NULL ); -if (!thread->ts) { +if (!ts) { + ts =thread_stack__new (thread,NULL ); +if (!ts) {pr_warning ("Out of memory: no thread stack\n" );return -ENOMEM ; } - thread->ts->trace_nr = trace_nr; + ts->trace_nr = trace_nr; }/*
@@ -252,14 +263,14 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,* the stack might be completely invalid. Better to report nothing than * to report something misleading, so flush the stack. */ -if (trace_nr != thread->ts->trace_nr) { -if (thread->ts->trace_nr) -__thread_stack__flush (thread, thread->ts); - thread->ts->trace_nr = trace_nr; +if (trace_nr != ts->trace_nr) { +if (ts->trace_nr) +__thread_stack__flush (thread, ts); + ts->trace_nr = trace_nr; }/* Stop here if thread_stack__process() is in use */ -if (thread->ts->crp) +if (ts->crp)return 0 ;if (flags &PERF_IP_FLAG_CALL ) {
@@ -270,7 +281,7 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip, ret_addr = from_ip + insn_len;if (ret_addr == to_ip)return 0 ;/* Zero-length calls are excluded */ -return thread_stack__push (thread->ts, ret_addr, +return thread_stack__push (ts, ret_addr, flags &PERF_IP_FLAG_TRACE_END ); }else if (flags &PERF_IP_FLAG_TRACE_BEGIN ) {/*
@@ -280,10 +291,10 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,* address, so try to pop that. Also, do not expect a call made * when the trace ended, to return, so pop that. */ -thread_stack__pop (thread->ts, to_ip); -thread_stack__pop_trace_end (thread->ts); +thread_stack__pop (ts, to_ip); +thread_stack__pop_trace_end (ts); }else if ((flags &PERF_IP_FLAG_RETURN ) && from_ip) { -thread_stack__pop (thread->ts, to_ip); +thread_stack__pop (ts, to_ip); }return 0 ;
@@ -291,21 +302,25 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,void thread_stack__set_trace_nr (struct thread *thread,u64 trace_nr) { -if (!thread || !thread->ts) +struct thread_stack *ts =thread__stack (thread); + +if (!ts)return ; -if (trace_nr != thread->ts->trace_nr) { -if (thread->ts->trace_nr) -__thread_stack__flush (thread, thread->ts); - thread->ts->trace_nr = trace_nr; +if (trace_nr != ts->trace_nr) { +if (ts->trace_nr) +__thread_stack__flush (thread, ts); + ts->trace_nr = trace_nr; } }void thread_stack__free (struct thread *thread) { -if (thread->ts) { -__thread_stack__flush (thread, thread->ts); -zfree (&thread->ts->stack); +struct thread_stack *ts = thread->ts; + +if (ts) { +__thread_stack__flush (thread, ts); +zfree (&ts->stack);zfree (&thread->ts); } }
@@ -318,6 +333,7 @@ static inline u64 callchain_context(u64 ip, u64 kernel_start)void thread_stack__sample (struct thread *thread,struct ip_callchain *chain,size_t sz,u64 ip,u64 kernel_start) { +struct thread_stack *ts =thread__stack (thread);u64 context =callchain_context (ip, kernel_start);u64 last_context;size_t i, j;
@@ -330,15 +346,15 @@ void thread_stack__sample(struct thread *thread, struct ip_callchain *chain, chain->ips[0 ] = context; chain->ips[1 ] = ip; -if (!thread || !thread->ts) { +if (!ts) { chain->nr =2 ;return ; } last_context = context; -for (i =2 , j =1 ; i < sz && j <= thread->ts->cnt; i++, j++) { - ip = thread->ts->stack[thread->ts->cnt - j].ret_addr; +for (i =2 , j =1 ; i < sz && j <= ts->cnt; i++, j++) { + ip = ts->stack[ts->cnt - j].ret_addr; context =callchain_context (ip, kernel_start);if (context != last_context) {if (i >= sz -1 )
@@ -590,7 +606,7 @@ int thread_stack__process(struct thread *thread, struct comm *comm, struct addr_location *to_al, u64 ref, struct call_return_processor *crp) { - struct thread_stack *ts = thread->ts; + struct thread_stack *ts = thread__stack(thread);int err =0 ;if (ts && !ts->crp) {
@@ -600,10 +616,9 @@ int thread_stack__process(struct thread *thread, struct comm *comm, }if (!ts) { - thread->ts =thread_stack__new (thread, crp); -if (!thread->ts) + ts =thread_stack__new (thread, crp); +if (!ts)return -ENOMEM ; - ts = thread->ts; ts->comm = comm; }
@@ -668,7 +683,9 @@ int thread_stack__process(struct thread *thread, struct comm *comm,size_t thread_stack__depth (struct thread *thread) { -if (!thread->ts) +struct thread_stack *ts =thread__stack (thread); + +if (!ts)return 0 ; -return thread->ts->cnt; +return ts->cnt; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 25/29 perf thread-stack: Allow for a thread stack array
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, allow for a thread stack array.
--- tools/perf/util/thread-stack.c | 40 +++++++++++++++++++++++++++++----- 1 file changed, 34 insertions(+), 6 deletions(-)
tools/perf/util/thread-stack.c +34 -6
@@ -60,6 +60,7 @@ struct thread_stack_entry {* @last_time: last timestamp * @crp: call/return processor * @comm: current comm +* @arr_sz: size of array if this is the first element of an array */ struct thread_stack {struct thread_stack_entry *stack;
@@ -71,6 +72,7 @@ struct thread_stack {u64 last_time;struct call_return_processor *crp;struct comm *comm; +unsigned int arr_sz; };static int thread_stack__grow (struct thread_stack *ts)
@@ -100,6 +102,8 @@ static struct thread_stack *thread_stack__new(struct thread *thread,if (!ts)return NULL ; + ts->arr_sz =1 ; +if (thread_stack__grow (ts)) {free (ts);return NULL ;
@@ -234,11 +238,19 @@ static int __thread_stack__flush(struct thread *thread, struct thread_stack *ts)int thread_stack__flush (struct thread *thread) {struct thread_stack *ts = thread->ts; +unsigned int pos; +int err =0 ; -if (ts) -return __thread_stack__flush (thread, ts); +if (ts) { +for (pos =0 ; pos < ts->arr_sz; pos++) { +int ret =__thread_stack__flush (thread, ts + pos); -return 0 ; +if (ret) + err = ret; + } + } + +return err; } int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip,
@@ -314,13 +326,29 @@ void thread_stack__set_trace_nr(struct thread *thread, u64 trace_nr) } } +static void __thread_stack__free (struct thread *thread,struct thread_stack *ts) +{ +__thread_stack__flush (thread, ts); +zfree (&ts->stack); +} + +static void thread_stack__reset (struct thread *thread,struct thread_stack *ts) +{ +unsigned int arr_sz = ts->arr_sz; + +__thread_stack__free (thread, ts); +memset (ts,0 ,sizeof (*ts)); + ts->arr_sz = arr_sz; +} +void thread_stack__free (struct thread *thread) {struct thread_stack *ts = thread->ts; +unsigned int pos;if (ts) { -__thread_stack__flush (thread, ts); -zfree (&ts->stack); +for (pos =0 ; pos < ts->arr_sz; pos++) +__thread_stack__free (thread, ts + pos);zfree (&thread->ts); } }
@@ -611,7 +639,7 @@ int thread_stack__process(struct thread *thread, struct comm *comm,if (ts && !ts->crp) {/* Supersede thread_stack__event() */ -thread_stack__free (thread); +thread_stack__reset (thread, ts); ts =NULL ; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 26/29 perf thread-stack: Factor out thread_stack__init()
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, factor out thread_stack__init().
--- tools/perf/util/thread-stack.c | 26 +++++++++++++++++++------- 1 file changed, 19 insertions(+), 7 deletions(-)
tools/perf/util/thread-stack.c +19 -7
@@ -93,6 +93,24 @@ static int thread_stack__grow(struct thread_stack *ts)return 0 ; } +static int thread_stack__init (struct thread_stack *ts,struct thread *thread, +struct call_return_processor *crp) +{ +int err; + + err =thread_stack__grow (ts); +if (err) +return err; + +if (thread->mg && thread->mg->machine) + ts->kernel_start =machine__kernel_start (thread->mg->machine); +else + ts->kernel_start =1ULL <<63 ; + ts->crp = crp; + +return 0 ; +} +static struct thread_stack *thread_stack__new (struct thread *thread,struct call_return_processor *crp) {
@@ -104,17 +122,11 @@ static struct thread_stack *thread_stack__new(struct thread *thread, ts->arr_sz =1 ; -if (thread_stack__grow (ts)) { +if (thread_stack__init (ts, thread, crp)) {free (ts);return NULL ; } -if (thread->mg && thread->mg->machine) - ts->kernel_start =machine__kernel_start (thread->mg->machine); -else - ts->kernel_start =1ULL <<63 ; - ts->crp = crp; - thread->ts = ts;return ts;
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 27/29 perf thread-stack: Allocate an array of thread stacks
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
In preparation for fixing thread stack processing for the idle task, allocate an array of thread stacks.
--- tools/perf/util/thread-stack.c | 29 +++++++++++++++++------------ 1 file changed, 17 insertions(+), 12 deletions(-)
tools/perf/util/thread-stack.c +17 -12
@@ -114,20 +114,25 @@ static int thread_stack__init(struct thread_stack *ts, struct thread *thread,static struct thread_stack *thread_stack__new (struct thread *thread,struct call_return_processor *crp) { -struct thread_stack *ts; - - ts =zalloc (sizeof (struct thread_stack )); -if (!ts) -return NULL ; - - ts->arr_sz =1 ; - -if (thread_stack__init (ts, thread, crp)) { -free (ts); -return NULL ; +struct thread_stack *ts = thread->ts, *new_ts; +unsigned int old_sz = ts ? ts->arr_sz :0 ; +unsigned int new_sz =1 ; + +if (!ts || new_sz > old_sz) { + new_ts =calloc (new_sz,sizeof (*ts)); +if (!new_ts) +return NULL ; +if (ts) +memcpy (new_ts, ts, old_sz *sizeof (*ts)); + new_ts->arr_sz = new_sz; +zfree (&thread->ts); + thread->ts = new_ts; + ts = new_ts; } - thread->ts = ts; +if (!ts->stack && +thread_stack__init (ts, thread, crp)) +return NULL ;return ts; }
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 28/29 perf thread-stack: Fix thread stack processing for the idle task
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
perf creates a single 'struct thread' to represent the idle task. That is because threads are identified by PID and TID, and the idle task always has PID == TID == 0.
However, there are actually separate idle tasks for each CPU. That creates a problem for thread stack processing which assumes that each thread has a single stack, not one stack per CPU.
Fix that by passing through the CPU number, and in the case of the idle "thread", pick the thread stack from an array based on the CPU number.
--- tools/perf/builtin-script.c | 4 +- tools/perf/util/intel-bts.c | 4 +- tools/perf/util/intel-pt.c | 6 +-- tools/perf/util/thread-stack.c | 72 +++++++++++++++++++++++++++------- tools/perf/util/thread-stack.h | 8 ++-- 5 files changed, 69 insertions(+), 25 deletions(-)
tools/perf/builtin-script.c +2 -2
@@ -1182,7 +1182,7 @@ static int perf_sample__fprintf_callindent(struct perf_sample *sample, struct addr_location *al, FILE *fp) { struct perf_event_attr *attr = &evsel->attr; -size_t depth =thread_stack__depth (thread); +size_t depth =thread_stack__depth (thread, sample->cpu);const char *name =NULL ;static int spacing;int len =0 ;
@@ -1716,7 +1716,7 @@ static bool show_event(struct perf_sample *sample,struct thread *thread, struct addr_location *al) { - int depth = thread_stack__depth(thread); + int depth = thread_stack__depth(thread, sample->cpu);if (!symbol_conf.graph_function)return true;
tools/perf/util/intel-bts.c +2 -2
@@ -451,7 +451,7 @@ static int intel_bts_process_buffer(struct intel_bts_queue *btsq,continue ;intel_bts_get_branch_type (btsq, branch);if (btsq->bts->synth_opts.thread_stack) - thread_stack__event(thread, btsq->sample_flags, + thread_stack__event(thread, btsq->cpu, btsq->sample_flags, le64_to_cpu(branch->from), le64_to_cpu(branch->to), btsq->intel_pt_insn.length,
@@ -523,7 +523,7 @@ static int intel_bts_process_queue(struct intel_bts_queue *btsq, u64 *timestamp) !btsq->bts->synth_opts.thread_stack && thread && (!old_buffer || btsq->bts->sampling_mode || (btsq->bts->snapshot_mode && !buffer->consecutive))) -thread_stack__set_trace_nr (thread, buffer->buffer_nr +1 ); +thread_stack__set_trace_nr (thread, btsq->cpu, buffer->buffer_nr +1 ); err =intel_bts_process_buffer (btsq, buffer, thread);
tools/perf/util/intel-pt.c +3 -3
@@ -1174,7 +1174,7 @@ static void intel_pt_prep_sample(struct intel_pt *pt,intel_pt_prep_b_sample (pt, ptq, event, sample);if (pt->synth_opts.callchain) { -thread_stack__sample (ptq->thread, ptq->chain, +thread_stack__sample (ptq->thread, ptq->cpu, ptq->chain, pt->synth_opts.callchain_sz +1 , sample->ip, pt->kernel_start); sample->callchain = ptq->chain;
@@ -1526,11 +1526,11 @@ static int intel_pt_sample(struct intel_pt_queue *ptq)return 0 ;if (pt->synth_opts.callchain || pt->synth_opts.thread_stack) -thread_stack__event (ptq->thread, ptq->flags, state->from_ip, +thread_stack__event (ptq->thread, ptq->cpu, ptq->flags, state->from_ip, state->to_ip, ptq->insn_len, state->trace_nr);else -thread_stack__set_trace_nr (ptq->thread, state->trace_nr); +thread_stack__set_trace_nr (ptq->thread, ptq->cpu, state->trace_nr);if (pt->sample_branches) { err =intel_pt_synth_branch_sample (ptq);
tools/perf/util/thread-stack.c +58 -14
156 lines not on this page · raw message
tools/perf/util/thread-stack.h +4 -4
@@ -80,14 +80,14 @@ struct call_return_processor {void *data; }; -int thread_stack__event (struct thread *thread,u32 flags,u64 from_ip, +int thread_stack__event (struct thread *thread,int cpu,u32 flags,u64 from_ip,u64 to_ip,u16 insn_len,u64 trace_nr); -void thread_stack__set_trace_nr (struct thread *thread,u64 trace_nr); -void thread_stack__sample (struct thread *thread,struct ip_callchain *chain, +void thread_stack__set_trace_nr (struct thread *thread,int cpu,u64 trace_nr); +void thread_stack__sample (struct thread *thread,int cpu,struct ip_callchain *chain,size_t sz,u64 ip,u64 kernel_start);int thread_stack__flush (struct thread *thread);void thread_stack__free (struct thread *thread); -size_t thread_stack__depth (struct thread *thread); +size_t thread_stack__depth (struct thread *thread,int cpu);struct call_return_processor * call_return_processor__new(int (*process)(struct call_return *cr, void *data),
Signature · 1 line
2.19.2
Arnaldo Carvalho de Melo
maintainerPATCH 29/29 perf session: Add comment for perf_session__register_idle_thread()
In reply to Arnaldo Carvalho de MeloFrom: Adrian Hunter <[email protected]>
Add a comment to perf_session__register_idle_thread() to bring attention to a pitfall with the idle task thread structure. The pitfall is that there should really be a 'struct thread' for the idle task of each cpu, but there is only one that can have pid == tid == 0.
--- tools/perf/util/session.c | 7 +++++++ 1 file changed, 7 insertions(+)
tools/perf/util/session.c +7 -0
@@ -1527,6 +1527,13 @@ struct thread *perf_session__findnew(struct perf_session *session, pid_t pid)return machine__findnew_thread (&session->machines.host,-1 , pid); } +/* +* Threads are identified by pid and tid, and the idle task has pid == tid == 0. +* So here a single thread is created for that, but actually there is a separate +* idle task per cpu, so there should be one 'struct thread' per cpu, but there +* is only 1. That causes problems for some tools, requiring workarounds. For +* example get_idle_thread() in builtin-sched.c, or thread_stack__per_cpu(). +*/ int perf_session__register_idle_thread (struct perf_session *session) {struct thread *thread;
Signature · 1 line
2.19.2
Quoted 141 lines · Arnaldo Carvalho de Melo
Hi Ingo,
Please consider pulling,
- Arnaldo
Test results at the end of this message, as usual.
The following changes since commit
883f4def8b77e6870ce42be279564cca0256c611:Merge tag 'perf-core-for-mingo-4.21-20181218' of
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linuxinto perf/core (2018-12-20 18:51:47 +0100)are available in the Git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.gittags/perf-core-for-mingo-4.21-20190103for you to fetch changes up to
b25756df5b28cd7b6e91200fc5012e7c76e8ec69:perf session: Add comment for
perf_session__register_idle_thread()(2019-01-02 11:05:06 -0300)----------------------------------------------------------------
perf/core improvements and fixes:perf c2c:
Jiri Olsa:
- Change the default coalesce setup to from '--coalesce pid,iaddr' to just '--coalesce iaddr'.
- Increase the HITM ratio limit for displayed cachelines.
perf script:
Andi Kleen:
- Fix LBR skid dump problems in brstackinsn.
perf trace:
Arnaldo Carvalho de Melo:
- Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter.
- Do not hardcode the size of the tracepoint
common_fields.- Beautify
USBDEFFS_ioctl commands.Colin Ian King:
- Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*".
perf python:
Jiri Olsa:
- Do not force closing original perf descriptor in
evlist.get_pollfd().tools misc:
Jiri Olsa:
- Allow overriding CFLAGS and LDFLAGS.
perf build:
Stanislav Fomichev:
- Don't unconditionally link the libbfd feature test to -liberty and -lz
thread-stack:
Adrian Hunter:
- Fix processing for the idle task, having a stack per cpu.
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>----------------------------------------------------------------
Adrian Hunter (8):
perf thread-stack: Simplify some code inthread_stack__process()perf thread-stack: Tidythread_stack__bottom()usage perf thread-stack: Avoid direct reference to the thread's stack perf thread-stack: Allow for a thread stack array
perf thread-stack: Factor outthread_stack__init()perf thread-stack: Allocate an array of thread stacks perf thread-stack: Fix thread stack processing for the idle task perf session: Add comment forperf_session__register_idle_thread()Andi Kleen (1):
perf script: Fix LBR skid dump problems in brstackinsnArnaldo Carvalho de Melo (11):
perf trace: Check if the raw_syscalls:sys_{enter,exit} are setup before setting tp filter perf beauty mmap:PROT_WRITEshould come beforePROT_EXEC
perf trace: Do not hardcode the size of the tracepointcommon_fields
perf trace: Renamethread_thread->pathstothread_trace->files
perf trace: Move the files table resizing to outsideset_pathname()
perf trace: Store the major number for a file when storing its pathname tools headers uapi: Grab a copy ofusbdevice_fs.h
perf beauty ioctl: Add generator forUSBDEVFS_ioctl commands
perf trace: Wire up ioctl'sUSBDEBFS_cmd table generator
perf trace beauty: Export function to get the files for a thread
perf trace beauty ioctl: BeautifyUSBDEVFS_commandsColin Ian King (1):
perf trace: Use correct SECCOMP prefix spelling, "SECOMP_*" -> "SECCOMP_*"Jiri Olsa (7):
perf python: Do not force closing original perf descriptor inevlist.get_pollfd()perf c2c: Change the default coalesce setup
perf c2c: Increase the HITM ratio limit for displayed cachelines
tools powerx86_energy_perf_policy: Override CFLAGS assignments and add LDFLAGS to build command tools thermal tmon: Allow overriding CFLAGS assignments
tools power turbostat: Override CFLAGS assignments and add LDFLAGS to build command tools gpio: Allow overriding CFLAGSStanislav Fomichev (1):
perf build: Don't unconditionally link the libbfd feature test to -liberty and -lztools/build/Makefile.feature | 4 +- tools/build/feature/Makefile | 10 +- tools/gpio/Makefile | 2 +- tools/include/uapi/linux/usbdevice_fs.h | 201 ++++++++++++++++++ tools/perf/Makefile.config | 44 ++-- tools/perf/Makefile.perf | 8 + tools/perf/builtin-c2c.c | 4 +- tools/perf/builtin-script.c | 21 +- tools/perf/builtin-trace.c | 142 +++++++++---- tools/perf/check-headers.sh | 1 + tools/perf/trace/beauty/beauty.h | 7 + tools/perf/trace/beauty/ioctl.c | 22 ++ tools/perf/trace/beauty/mmap.c | 2 +- tools/perf/trace/beauty/seccomp.c | 4 +- tools/perf/trace/beauty/usbdevfs_ioctl.sh | 19 ++ tools/perf/util/dump-insn.c | 8 + tools/perf/util/dump-insn.h | 2 + tools/perf/util/intel-bts.c | 4 +- .../util/intel-pt-decoder/intel-pt-insn-decoder.c | 8 + tools/perf/util/intel-pt.c | 6 +- tools/perf/util/python.c | 3 +- tools/perf/util/session.c | 7 +
Quoted 8 lines
tools/perf/util/thread-stack.c | 227 +++++++++++++++------ tools/perf/util/thread-stack.h | 8 +- tools/power/x86/turbostat/Makefile | 8 +- tools/power/x86/x86_energy_perf_policy/Makefile | 6 +- tools/thermal/tmon/Makefile | 8 +- 27 files changed, 620 insertions(+), 166 deletions(-) create mode 100644 tools/include/uapi/linux/usbdevice_fs.h create mode 100755 tools/perf/trace/beauty/usbdevfs_ioctl.sh
Pulled, thanks a lot Arnaldo!
Ingo