Skip to contentSkip to conversation

perf/core improvements and fixes

GIT PULL New Arnaldo Carvalho de Melo 1 participant 10 messages Jul 19, 2017 perf events

lore
Share & tools
lore

Arnaldo Carvalho de Melo

author maintainer
#0 lore raw

Hi Ingo,

Unusually big one, please conside pulling, details on the signed tag,

- Arnaldo

Test results at the end of this message, as usual.

The following changes since commit 4b1303d0b01440f224cf81493b7e8e43d9b4965e:

perf symbols: Accept zero as the kernel base address (2017-07-12 11:47:05 -0300)

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo-4.13-20170718

for you to fetch changes up to b851dd49868e295e18c5d72fc3bad85ff1c444b1:

perf report: Show branch type in callchain entry (2017-07-18 23:14:42 -0300)

----------------------------------------------------------------
perf/core improvements and fixes:

User visible:

. Initial support for namespaces, using setns to access files in
namespaces, grabbing their build-ids, etc. We still need to work more to deal with namespaces that vanish before we can get the needed data to do analysis, but this should be as good as what is in bcc now (Krister Johansen)

. Add header record types to pipe-mode, now this command:

  $ perf record -o - -e cycles sleep 1 | perf report --stdio --header

Will show the same as in non-pipe mode, i.e. involving a perf.data file (David Carrillo-Cisneros)

. Implement a visual marker for fused x86 instructions in the annotate
TUI browser, available now in 'perf report', more work needed to have it available as well in 'perf top' (Jin Yao)

Further explanation from one of Jin's patches:

       │   ┌──cmpl   $0x0,argp_program_version_hook
 81.93 │   ├──je     20
       │   │  lock   cmpxchg %esi,0x38a9a4(%rip)
       │   │↓ jne    29
       │   │↓ jmp    43
 11.47 │20:└─→cmpxch %esi,0x38a999(%rip)

That means the cmpl+je is a fused instruction pair and they should be considered together.

. Record the branch type and then show statistics and info about
in callchain entries (Jin Yao)

Example from one of Jin's patches:

  # perf record -g -j any,save_type
  # perf report --branch-history --stdio --no-children

38.50% div.c:45 [.] main div

          |
          ---main div.c:42 (RET CROSS_2M cycles:2)
             compute_flag div.c:28 (cycles:2)
             compute_flag div.c:27 (RET CROSS_2M cycles:1)
             rand rand.c:28 (cycles:1)
             rand rand.c:28 (RET CROSS_2M cycles:1)
             __random random.c:298 (cycles:1)
             __random random.c:297 (COND_BWD CROSS_2M cycles:1)
             __random random.c:295 (cycles:1)
             __random random.c:295 (COND_BWD CROSS_2M cycles:1)
             __random random.c:295 (cycles:1)
             __random random.c:295 (RET CROSS_2M cycles:9)

. Beautify the fcntl syscall, which is an interesting one in the sense
that infrastructure had to be put in place to change the formatters of some arguments according to the value in a previous one, i.e. cmd dictates how arg and the syscall return will be formatted. (Arnaldo Carvalho de Melo

Infrastructure:

. 'perf test attr' fixes (Jiri Olsa)

Vendor events:

- Add POWER9 PMU events Sukadev (Bhattiprolu)

- Support additional POWER8+ PVR in PMU mapfile (Shriya)

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

----------------------------------------------------------------
Arnaldo Carvalho de Melo (39):
perf trace: Remove F_ from some of the fcntl command strings
perf trace: Beautify linux specific fcntl commands
tools: Update include/uapi/linux/fcntl.h copy from the kernel
perf trace beauty: Export the strarrays scnprintf method
perf trace: Only build tools/perf/trace/beauty/ when building 'perf trace' perf trace beauty: Mask ignored fcntl 'arg' parameter
perf trace beauty: Allow accessing syscall args values in a syscall arg formatter perf trace beauty: Export the "int" and "hex" syscall arg formatters perf trace beauty: Introduce syscall arg beautifier for long integers tools include uapi asm-generic: Grab a copy of fcntl.h
perf trace beauty fcntl: Basic 'arg' beautifier
perf trace: Beautify new write hint fcntl commands
perf beauty open: Detach the syscall_arg agnostic bits from the flags formatter perf trace: Allow syscall_arg beautifiers to set a different return formatter perf trace beauty open flags: Support O_TMPFILE and O_NOFOLLOW
perf trace beauty open flags: Do not depend on the system's O_LARGEFILE define perf trace beauty fcntl: Beautify F_GETFL return value
perf trace beauty open flags: Move RDRW to the start of the output perf trace beauty fcntl flags: Beautify F_SETFL arg
perf trace beauty fcntl: Beautify F_[GS]ETFD arg/return value
perf trace beauty: Give syscall return beautifier more context
perf trace beauty: Export the fd beautifier for use in more places perf trace beauty fcntl: Augment the return of F_DUPFD(_CLOEXEC) perf trace beauty: Export the pid beautifier for use in more places perf trace beauty fcntl: Beautify F_GETOWN and F_SETOWN
tools include uapi x86: Grab a copy of unistd.h
tools include uapi x86: Add __NR_setns, if missing
tools build: Add test for setns()
perf evsel: Allow asking for max precise_ip in new_cycles()
perf evlist: Allow asking for max precise_ip in add_default()
perf record: Do not ask for precise_ip with --no-samples
perf test sdt: Handle realpath() failure
perf trace beauty: Export strarray for use in per-object beautifiers perf trace beauty fcntl: Beautify F_GETLEASE and F_SETLEASE arg/return perf trace: Group per syscall arg formatter info into one struct perf trace: Allow syscall arg formatters to request non suppression of zeros perf trace beauty fcntl: Do not suppress 'cmd' when zero, should be DUPFD perf trace beauty fcntl: Beautify the 'arg' for DUPFD
perf trace beauty: Simplify syscall return formatting

David Carrillo-Cisneros (16):
perf header: Encapsulate read and swap
perf header: Add PROCESS_STR_FUN macro
perf header: Fail on write_padded error
perf util: Add const modifier to buf in "writen" function perf header: Revamp do_write()
perf header: Add struct feat_fd for write
perf header: Use struct feat_fd for print
perf header: Use struct feat_fd to process header records perf header: Don't pass struct perf_file_section to process_##_feat perf header: Use struct feat_fd in read header records perf header: Make write_pmu_mappings pipe-mode friendly perf header: Add a buffer to struct feat_fd
perf header: Change FEAT_OP* macros
perf tool: Add show_feature_header to perf_tool
perf tools: Add feature header record to pipe-mode
perf header: Add event desc to pipe-mode header

Jin Yao (10):
perf annotate: Check for fused instructions
perf annotate: Implement visual marker for macro fusion perf report: Enable finding kernel inline functions perf/core: Define the common branch type classification perf/x86/intel: Record branch type
perf record: Create a new option save_type in --branch-filter perf report: Refactor the branch info printing code perf util: Create branch.c/.h for common branch functions perf report: Show branch type statistics for stdio mode perf report: Show branch type in callchain entry

Jiri Olsa (13):
perf tests attr: Do not store failed events
perf tests attr: Add test_attr__ready function perf tests attr: Make compare_data global
perf tests attr: Rename compare_data to data_equal perf tests attr: Add 1s for exclude_kernel and task base bits perf tests attr: Fix record dwarf test
perf tests attr: Fix no-delay test
perf tests attr: Add proper return values
perf tests attr: Fix cpu test disabled term setup perf tests attr: Fix sample_period setup
perf tests attr: Fix precise_ip setup
perf tests attr: Fix stat sample_type setup
perf tests attr: Add optional term

Krister Johansen (5):
perf symbols: Find symbols in different mount namespace perf maps: Lookup maps in both intitial mountns and inner mountns. perf probe: Allow placing uprobes in alternate namespaces. perf buildid-cache: Support binary objects from other namespaces perf buildid-cache: Cache debuginfo

Shriya (1):
perf pmu-events: Support additional POWER8+ PVR in mapfile

Sukadev Bhattiprolu (2):
perf vendor events: Add POWER9 PMU events perf vendor events: Add POWER9 PVRs to mapfile

 arch/x86/events/intel/lbr.c                        |   52 +-
 include/uapi/linux/perf_event.h                    |   27 +-
 tools/arch/x86/include/asm/unistd_32.h             |    3 +
 tools/arch/x86/include/asm/unistd_64.h             |    3 +
 tools/arch/x86/include/uapi/asm/unistd.h           |   17 +
 tools/build/Makefile.feature                       |    3 +-
 tools/build/feature/Makefile                       |    6 +-
 tools/build/feature/test-all.c                     |    5 +
 tools/build/feature/test-setns.c                   |    7 +
 tools/include/uapi/asm-generic/fcntl.h             |  220 +++++
 tools/include/uapi/linux/fcntl.h                   |   21 +
 tools/include/uapi/linux/perf_event.h              |   27 +-
 tools/perf/Build                                   |    2 +-
 tools/perf/Documentation/perf-buildid-cache.txt    |    5 +
 tools/perf/Documentation/perf-probe.txt            |   14 +
 tools/perf/Documentation/perf-record.txt           |    1 +
 tools/perf/Documentation/perf.data-file-format.txt |   10 +-
 tools/perf/Makefile.config                         |    5 +
 tools/perf/arch/powerpc/util/sym-handling.c        |    2 +-
 tools/perf/arch/x86/annotate/instructions.c        |   46 +
 tools/perf/builtin-annotate.c                      |    1 +
 tools/perf/builtin-buildid-cache.c                 |   54 +-
 tools/perf/builtin-inject.c                        |    1 +
 tools/perf/builtin-probe.c                         |   45 +-
 tools/perf/builtin-record.c                        |    9 +-
 tools/perf/builtin-report.c                        |   30 +
 tools/perf/builtin-script.c                        |    4 +
 tools/perf/builtin-top.c                           |    2 +-
 tools/perf/builtin-trace.c                         |  602 ++++++------
 tools/perf/check-headers.sh                        |    1 +
 tools/perf/perf.h                                  |    1 +
 tools/perf/pmu-events/arch/powerpc/mapfile.csv     |    4 +
 .../perf/pmu-events/arch/powerpc/power9/cache.json |  176 ++++
 .../arch/powerpc/power9/floating-point.json        |   44 +
 .../pmu-events/arch/powerpc/power9/frontend.json   |  446 +++++++++
 .../pmu-events/arch/powerpc/power9/marked.json     |  782 +++++++++++++++
 .../pmu-events/arch/powerpc/power9/memory.json     |  158 +++
 .../perf/pmu-events/arch/powerpc/power9/other.json |  836 ++++++++++++++++
 .../pmu-events/arch/powerpc/power9/pipeline.json   |  680 +++++++++++++
 tools/perf/pmu-events/arch/powerpc/power9/pmc.json |  146 +++
 .../arch/powerpc/power9/translation.json           |  272 ++++++
 tools/perf/tests/attr.c                            |   12 +-
 tools/perf/tests/attr.py                           |   50 +-
 tools/perf/tests/attr/base-record                  |    6 +-
 tools/perf/tests/attr/base-stat                    |    4 +-
 tools/perf/tests/attr/test-record-C0               |    1 +
 tools/perf/tests/attr/test-record-basic            |    1 +
 tools/perf/tests/attr/test-record-branch-any       |    2 +-
 .../perf/tests/attr/test-record-branch-filter-any  |    2 +-
 .../tests/attr/test-record-branch-filter-any_call  |    2 +-
 .../tests/attr/test-record-branch-filter-any_ret   |    2 +-
 tools/perf/tests/attr/test-record-branch-filter-hv |    2 +-
 .../tests/attr/test-record-branch-filter-ind_call  |    2 +-
 tools/perf/tests/attr/test-record-branch-filter-k  |    2 +-
 tools/perf/tests/attr/test-record-branch-filter-u  |    2 +-
 tools/perf/tests/attr/test-record-count            |    1 +
 tools/perf/tests/attr/test-record-data             |    3 +-
 tools/perf/tests/attr/test-record-freq             |    1 +
 tools/perf/tests/attr/test-record-graph-default    |    1 +
 tools/perf/tests/attr/test-record-graph-dwarf      |    4 +-
 tools/perf/tests/attr/test-record-graph-fp         |    1 +
 tools/perf/tests/attr/test-record-group            |    1 +
 tools/perf/tests/attr/test-record-group-sampling   |    1 +
 tools/perf/tests/attr/test-record-group1           |    1 +
 ...st-record-no-delay => test-record-no-buffering} |    4 +-
 tools/perf/tests/attr/test-record-no-inherit       |    1 +
 tools/perf/tests/attr/test-record-no-samples       |    1 +
 tools/perf/tests/attr/test-record-period           |    1 +
 tools/perf/tests/attr/test-record-raw              |    2 +-
 tools/perf/tests/attr/test-stat-C0                 |    4 +-
 tools/perf/tests/attr/test-stat-default            |    2 +
 tools/perf/tests/attr/test-stat-detailed-1         |    2 +
 tools/perf/tests/attr/test-stat-detailed-2         |    3 +
 tools/perf/tests/attr/test-stat-detailed-3         |    5 +
 tools/perf/tests/sdt.c                             |    8 +-
 tools/perf/trace/beauty/Build                      |    1 +
 tools/perf/trace/beauty/beauty.h                   |   65 ++
 tools/perf/trace/beauty/fcntl.c                    |  100 ++
 tools/perf/trace/beauty/open_flags.c               |   29 +-
 tools/perf/trace/beauty/pid.c                      |    4 +-
 tools/perf/ui/browser.c                            |   29 +
 tools/perf/ui/browser.h                            |    2 +
 tools/perf/ui/browsers/annotate.c                  |   30 +-
 tools/perf/ui/browsers/hists.c                     |    3 -
 tools/perf/ui/gtk/annotate.c                       |    2 +-
 tools/perf/ui/stdio/hist.c                         |    3 -
 tools/perf/util/Build                              |    5 +
 tools/perf/util/annotate.c                         |   29 +-
 tools/perf/util/annotate.h                         |    4 +-
 tools/perf/util/branch.c                           |  147 +++
 tools/perf/util/branch.h                           |   24 +
 tools/perf/util/build-id.c                         |  129 ++-
 tools/perf/util/build-id.h                         |   16 +-
 tools/perf/util/callchain.c                        |  134 +--
 tools/perf/util/callchain.h                        |    5 +-
 tools/perf/util/dso.c                              |   21 +-
 tools/perf/util/dso.h                              |    3 +
 tools/perf/util/event.c                            |    1 +
 tools/perf/util/event.h                            |   11 +-
 tools/perf/util/evlist.c                           |    4 +-
 tools/perf/util/evlist.h                           |    9 +-
 tools/perf/util/evsel.c                            |   18 +-
 tools/perf/util/evsel.h                            |    3 +-
 tools/perf/util/header.c                           | 1015 +++++++++++---------
 tools/perf/util/header.h                           |   16 +-
 tools/perf/util/hist.c                             |    5 +-
 tools/perf/util/machine.c                          |   33 +-
 tools/perf/util/map.c                              |   23 +-
 tools/perf/util/map.h                              |    2 +-
 tools/perf/util/namespaces.c                       |  211 ++++
 tools/perf/util/namespaces.h                       |   38 +
 tools/perf/util/parse-branch-options.c             |    1 +
 tools/perf/util/parse-events.c                     |    2 +-
 tools/perf/util/probe-event.c                      |   86 +-
 tools/perf/util/probe-event.h                      |   10 +-
 tools/perf/util/probe-file.c                       |   19 +-
 tools/perf/util/probe-file.h                       |    4 +-
 tools/perf/util/python-ext-sources                 |    1 +
 tools/perf/util/session.c                          |    4 +
 tools/perf/util/setns.c                            |    8 +
 tools/perf/util/symbol.c                           |   92 +-
 tools/perf/util/thread.c                           |    3 +
 tools/perf/util/thread.h                           |    1 +
 tools/perf/util/tool.h                             |   10 +-
 tools/perf/util/util.c                             |   40 +-
 tools/perf/util/util.h                             |    8 +-
 126 files changed, 6339 insertions(+), 1031 deletions(-)
 create mode 100644 tools/arch/x86/include/uapi/asm/unistd.h
 create mode 100644 tools/build/feature/test-setns.c
 create mode 100644 tools/include/uapi/asm-generic/fcntl.h
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/cache.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/floating-point.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/frontend.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/marked.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/memory.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/other.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/pipeline.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/pmc.json
 create mode 100644 tools/perf/pmu-events/arch/powerpc/power9/translation.json
 rename tools/perf/tests/attr/{test-record-no-delay => test-record-no-buffering} (61%)
 create mode 100644 tools/perf/trace/beauty/fcntl.c
 create mode 100644 tools/perf/util/branch.c
 create mode 100644 tools/perf/util/branch.h
 create mode 100644 tools/perf/util/setns.c

Test results at the end of this message, as usual.

Test results:

The first ones are container (docker) based builds of tools/perf with and without libelf support, objtool where it is supported and samples/bpf/, ditto. Where clang is available, it is also used to build perf with/without libelf.

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.

The fedora:rawhide case is being investigated, doesn't seem to have been introduced by this batch:

LINK /tmp/build/perf/perf
LINK /tmp/build/perf/libperf-gtk.so
/usr/bin/ld: /tmp/build/perf/perf-in.o: relocation R_X86_64_32 against .rodata.str1.1' can not be used when making a shared object; recompile with -fPIC /usr/bin/ld: /tmp/build/perf/libperf.a(libperf-in.o): relocation R_X86_64_32S against .text' can not be used when making a shared object; recompile with -fPIC /usr/bin/ld: final link failed: Nonrepresentable section on output
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile.perf:420: /tmp/build/perf/perf] Error 1

  # dm
   1 alpine:3.4: Ok
   2 alpine:3.5: Ok
   3 alpine:3.6: Ok
   4 alpine:edge: Ok
   5 android-ndk:r12b-arm: Ok
   6 archlinux:latest: Ok
   7 centos:5: Ok
   8 centos:6: Ok
   9 centos:7: Ok
  10 debian:7: Ok
  11 debian:8: Ok
  12 debian:9: Ok
  13 debian:experimental: Ok
  14 debian:experimental-x-arm64: Ok
  15 debian:experimental-x-mips: Ok
  16 debian:experimental-x-mips64: Ok
  17 debian:experimental-x-mipsel: Ok
  18 fedora:20: Ok
  19 fedora:21: Ok
  20 fedora:22: Ok
  21 fedora:23: Ok
  22 fedora:24: Ok
  23 fedora:24-x-ARC-uClibc: Ok
  24 fedora:25: Ok
  25 fedora:26: Ok
  26 fedora:rawhide: FAIL
  27 mageia:5: Ok
  28 opensuse:13.2: Ok
  29 opensuse:42.1: Ok
  30 opensuse:42.2: Ok
  31 opensuse:tumbleweed: Ok
  32 oraclelinux:6: Ok
  33 oraclelinux:7: Ok
  34 ubuntu:12.04.5: Ok
  35 ubuntu:14.04.4: Ok
  36 ubuntu:14.04.4-x-linaro-arm64: Ok
  37 ubuntu:15.10: Ok
  38 ubuntu:16.04: Ok
  39 ubuntu:16.04-x-arm: Ok
  40 ubuntu:16.04-x-arm64: Ok
  41 ubuntu:16.04-x-powerpc: Ok
  42 ubuntu:16.04-x-powerpc64: Ok
  43 ubuntu:16.04-x-powerpc64el: Ok
  44 ubuntu:16.04-x-s390: Ok
  45 ubuntu:16.10: Ok
  46 ubuntu:17.04: Ok
  47 ubuntu:17.10: Ok
  #
  # uname -a
  Linux jouet 4.12.0-rc6+ #3 SMP Tue Jun 27 15:12:38 -03 2017 x86_64 x86_64 x86_64 GNU/Linux
  # 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: Parse event definition strings             : Ok
   6: Simple expression parser                   : Ok
   7: PERF_RECORD_* events & perf_sample fields  : Ok
   8: Parse perf pmu format                      : Ok
   9: DSO data read                              : Ok
  10: DSO data cache                             : Ok
  11: DSO data reopen                            : Ok
  12: Roundtrip evsel->name                      : Ok
  13: Parse sched tracepoints fields             : Ok
  14: syscalls:sys_enter_openat event fields     : Ok
  15: Setup struct perf_event_attr               : Ok
  16: Match and link multiple hists              : Ok
  17: 'import perf' in python                    : Ok
  18: Breakpoint overflow signal handler         : Ok
  19: Breakpoint overflow sampling               : Ok
  20: Number of exit events of a simple workload : Ok
  21: Software clock events period values        : Ok
  22: Object code reading                        : Ok
  23: Sample parsing                             : Ok
  24: Use a dummy software event to keep tracking: Ok
  25: Parse with no sample_id_all bit set        : Ok
  26: Filter hist entries                        : Ok
  27: Lookup mmap thread                         : Ok
  28: Share thread mg                            : Ok
  29: Sort output of hist entries                : Ok
  30: Cumulate child hist entries                : Ok
  31: Track with sched_switch                    : Ok
  32: Filter fds with revents mask in a fdarray  : Ok
  33: Add fd to a fdarray, making it autogrow    : Ok
  34: kmod_path__parse                           : Ok
  35: Thread map                                 : Ok
  36: LLVM search and compile                    :
  36.1: Basic BPF llvm compile                    : Ok
  36.2: kbuild searching                          : Ok
  36.3: Compile source for BPF prologue generation: Ok
  36.4: Compile source for BPF relocation         : Ok
  37: Session topology                           : Ok
  38: BPF filter                                 :
  38.1: Basic BPF filtering                      : Ok
  38.2: BPF pinning                              : Ok
  38.3: BPF prologue generation                  : Ok
  38.4: BPF relocation checker                   : Ok
  39: Synthesize thread map                      : Ok
  40: Remove thread map                          : Ok
  41: Synthesize cpu map                         : Ok
  42: Synthesize stat config                     : Ok
  43: Synthesize stat                            : Ok
  44: Synthesize stat round                      : Ok
  45: Synthesize attr update                     : Ok
  46: Event times                                : Ok
  47: Read backward ring buffer                  : Ok
  48: Print cpu map                              : Ok
  49: Probe SDT events                           : Ok
  50: is_printable_array                         : Ok
  51: Print bitmap                               : Ok
  52: perf hooks                                 : Ok
  53: builtin clang support                      : Skip (not compiled in)
  54: unit_number__scnprintf                     : Ok
  55: x86 rdpmc                                  : Ok
  56: Convert perf time to TSC                   : Ok
  57: DWARF unwind                               : Ok
  58: x86 instruction decoder - new instructions : Ok
  59: Intel cqm nmi context read                 : Skip
  #
author maintainer
#1 lore raw

PATCH 1/86 perf annotate: Check for fused instructions

From: Jin Yao <[email protected]>

Macro fusion merges two instructions to a single micro-op. Intel core platform performs this hardware optimization under limited circumstances.

For example, CMP + JCC can be "fused" and executed /retired together. While with sampling this can result in the sample sometimes being on the JCC and sometimes on the CMP. So for the fused instruction pair, they could be considered together.

On Nehalem, fused instruction pairs:

cmp/test + jcc.

On other new CPU:

cmp/test/add/sub/and/inc/dec + jcc.

This patch adds an x86-specific function which checks if 2 instructions are in a "fused" pair. For non-x86 arch, the function is just NULL.

Changelog:

v4: Move the CPU model checking to symbol__disassemble and save the CPU
family/model in arch structure.

It avoids checking every time when jump arrow printed.

v3: Add checking for Nehalem (CMP, TEST). For other newer Intel CPUs
just check it by default (CMP, TEST, ADD, SUB, AND, INC, DEC).

v2: Remove the original weak function. Arnaldo points out that doing it
as a weak function that will be overridden by the host arch doesn't work. So now it's implemented as an arch-specific function.

Committer fix:

Do not access evsel->evlist->env->cpuid, ->env can be null, introduce perf_evsel__env_cpuid(), just like perf_evsel__env_arch(), also used in this function call.

The original patch was segfaulting 'perf top' + annotation.

But this essentially disables this fused instructions augmentation in 'perf top', the right thing is to get the cpuid from the running kernel, left for a later patch tho.

Signed-off-by Yao Jin <[email protected]>
Tested-by Arnaldo Carvalho de Melo <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Andi Kleen <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Kan Liang <[email protected]>
Cc Peter Zijlstra <[email protected]>
Link http://lkml.kernel.org/r/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/arch/x86/annotate/instructions.c | 46 +++++++++++++++++++++++++++++
 tools/perf/builtin-top.c                    |  2 +-
 tools/perf/ui/browsers/annotate.c           |  4 ++-
 tools/perf/ui/gtk/annotate.c                |  2 +-
 tools/perf/util/annotate.c                  | 22 ++++++++++++--
 tools/perf/util/annotate.h                  |  3 +-
 tools/perf/util/evsel.c                     |  7 +++++
 tools/perf/util/evsel.h                     |  1 +
 8 files changed, 81 insertions(+), 6 deletions(-)
tools/perf/arch/x86/annotate/instructions.c +46 -0
index c1625f256df3..d84b72063a30 100644
@@ -76,3 +76,49 @@ static struct ins x86__instructions[] = {
 	{ .name = "xbeginq",	.ops = &jump_ops, },
 	{ .name = "retq",	.ops = &ret_ops,  },
 };
+
+static bool x86__ins_is_fused(struct arch *arch, const char *ins1,
+			      const char *ins2)
+{
+	if (arch->family != 6 || arch->model < 0x1e || strstr(ins2, "jmp"))
+		return false;
+
+	if (arch->model == 0x1e) {
+		/* Nehalem */
+		if ((strstr(ins1, "cmp") && !strstr(ins1, "xchg")) ||
+		     strstr(ins1, "test")) {
+			return true;
+		}
+	} else {
+		/* Newer platform */
+		if ((strstr(ins1, "cmp") && !strstr(ins1, "xchg")) ||
+		     strstr(ins1, "test") ||
+		     strstr(ins1, "add") ||
+		     strstr(ins1, "sub") ||
+		     strstr(ins1, "and") ||
+		     strstr(ins1, "inc") ||
+		     strstr(ins1, "dec")) {
+			return true;
+		}
+	}
+
+	return false;
+}
+
+static int x86__cpuid_parse(struct arch *arch, char *cpuid)
+{
+	unsigned int family, model, stepping;
+	int ret;
+
+	/*
+	 * cpuid = "GenuineIntel,family,model,stepping"
+	 */
+	ret = sscanf(cpuid, "%*[^,],%u,%u,%u", &family, &model, &stepping);
+	if (ret == 3) {
+		arch->family = family;
+		arch->model = model;
+		return 0;
+	}
+
+	return -1;
+}
tools/perf/builtin-top.c +1 -1
index 6052376634c0..022486dc67f5 100644
@@ -134,7 +134,7 @@ static int perf_top__parse_source(struct perf_top *top, struct hist_entry *he)
 		return err;
 	}
 
-	err = symbol__disassemble(sym, map, NULL, 0, NULL);
+	err = symbol__disassemble(sym, map, NULL, 0, NULL, NULL);
 	if (err == 0) {
 out_assign:
 		top->sym_filter_entry = he;
tools/perf/ui/browsers/annotate.c +3 -1
index 27f41f28dcb4..c4336138b673 100644
@@ -9,6 +9,7 @@
 #include "../../util/symbol.h"
 #include "../../util/evsel.h"
 #include "../../util/config.h"
+#include "../../util/evlist.h"
 #include <inttypes.h>
 #include <pthread.h>
 #include <linux/kernel.h>

@@ -1074,7 +1075,8 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map,
 	}
 
 	err = symbol__disassemble(sym, map, perf_evsel__env_arch(evsel),
-				  sizeof_bdl, &browser.arch);
+				  sizeof_bdl, &browser.arch,
+				  perf_evsel__env_cpuid(evsel));
 	if (err) {
 		char msg[BUFSIZ];
 		symbol__strerror_disassemble(sym, map, err, msg, sizeof(msg));
tools/perf/ui/gtk/annotate.c +1 -1
index d903fd493416..87e3760624f2 100644
@@ -169,7 +169,7 @@ static int symbol__gtk_annotate(struct symbol *sym, struct map *map,
 		return -1;
 
 	err = symbol__disassemble(sym, map, perf_evsel__env_arch(evsel),
-				  0, NULL);
+				  0, NULL, NULL);
 	if (err) {
 		char msg[BUFSIZ];
 		symbol__strerror_disassemble(sym, map, err, msg, sizeof(msg));
tools/perf/util/annotate.c +20 -2
index be1caabb9290..8748ebb3f932 100644
@@ -47,7 +47,12 @@ struct arch {
 	bool		sorted_instructions;
 	bool		initialized;
 	void		*priv;
+	unsigned int	model;
+	unsigned int	family;
 	int		(*init)(struct arch *arch);
+	bool		(*ins_is_fused)(struct arch *arch, const char *ins1,
+					const char *ins2);
+	int		(*cpuid_parse)(struct arch *arch, char *cpuid);
 	struct		{
 		char comment_char;
 		char skip_functions_char;

@@ -129,6 +134,8 @@ static struct arch architectures[] = {
 		.name = "x86",
 		.instructions = x86__instructions,
 		.nr_instructions = ARRAY_SIZE(x86__instructions),
+		.ins_is_fused = x86__ins_is_fused,
+		.cpuid_parse = x86__cpuid_parse,
 		.objdump =  {
 			.comment_char = '#',
 		},

@@ -171,6 +178,14 @@ int ins__scnprintf(struct ins *ins, char *bf, size_t size,
 	return ins__raw_scnprintf(ins, bf, size, ops);
 }
 
+bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2)
+{
+	if (!arch || !arch->ins_is_fused)
+		return false;
+
+	return arch->ins_is_fused(arch, ins1, ins2);
+}
+
 static int call__parse(struct arch *arch, struct ins_operands *ops, struct map *map)
 {
 	char *endptr, *tok, *name;

@@ -1381,7 +1396,7 @@ static const char *annotate__norm_arch(const char *arch_name)
 
 int symbol__disassemble(struct symbol *sym, struct map *map,
 			const char *arch_name, size_t privsize,
-			struct arch **parch)
+			struct arch **parch, char *cpuid)
 {
 	struct dso *dso = map->dso;
 	char command[PATH_MAX * 2];

@@ -1418,6 +1433,9 @@ int symbol__disassemble(struct symbol *sym, struct map *map,
 		}
 	}
 
+	if (arch->cpuid_parse && cpuid)
+		arch->cpuid_parse(arch, cpuid);
+
 	pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__,
 		 symfs_filename, sym->name, map->unmap_ip(map, sym->start),
 		 map->unmap_ip(map, sym->end));

@@ -1907,7 +1925,7 @@ int symbol__tty_annotate(struct symbol *sym, struct map *map,
 	u64 len;
 
 	if (symbol__disassemble(sym, map, perf_evsel__env_arch(evsel),
-				0, NULL) < 0)
+				0, NULL, NULL) < 0)
 		return -1;
 
 	len = symbol__size(sym);
tools/perf/util/annotate.h +2 -1
index 21055034aedd..72d72728a0fc 100644
@@ -53,6 +53,7 @@ bool ins__is_jump(const struct ins *ins);
 bool ins__is_call(const struct ins *ins);
 bool ins__is_ret(const struct ins *ins);
 int ins__scnprintf(struct ins *ins, char *bf, size_t size, struct ins_operands *ops);
+bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2);
 
 struct annotation;
 

@@ -160,7 +161,7 @@ void symbol__annotate_zero_histograms(struct symbol *sym);
 
 int symbol__disassemble(struct symbol *sym, struct map *map,
 			const char *arch_name, size_t privsize,
-			struct arch **parch);
+			struct arch **parch, char *cpuid);
 
 enum symbol_disassemble_errno {
 	SYMBOL_ANNOTATE_ERRNO__SUCCESS		= 0,
tools/perf/util/evsel.c +7 -0
index 413f74df08de..0e4cd6092564 100644
@@ -2610,3 +2610,10 @@ char *perf_evsel__env_arch(struct perf_evsel *evsel)
 		return evsel->evlist->env->arch;
 	return NULL;
 }
+
+char *perf_evsel__env_cpuid(struct perf_evsel *evsel)
+{
+	if (evsel && evsel->evlist && evsel->evlist->env)
+		return evsel->evlist->env->cpuid;
+	return NULL;
+}
tools/perf/util/evsel.h +1 -0
index d101695c482c..219ad0cdb9f4 100644
@@ -436,5 +436,6 @@ int perf_event_attr__fprintf(FILE *fp, struct perf_event_attr *attr,
 			     attr__fprintf_f attr__fprintf, void *priv);
 
 char *perf_evsel__env_arch(struct perf_evsel *evsel);
+char *perf_evsel__env_cpuid(struct perf_evsel *evsel);
 
 #endif /* __PERF_EVSEL_H */
Signature · 1 line
2.9.4
author maintainer
#2 lore raw

PATCH 2/86 perf annotate: Implement visual marker for macro fusion

In reply to Arnaldo Carvalho de Melo

From: Jin Yao <[email protected]>

For marking fused instructions clearly this patch adds a line before the first instruction of pair and joins it with the arrow of the jump to its target.

For example, when "je" is selected in annotate view, the line before cmpl is displayed and joins the arrow of "je".

       │   ┌──cmpl   $0x0,argp_program_version_hook
 81.93 │   ├──je     20
       │   │  lock   cmpxchg %esi,0x38a9a4(%rip)
       │   │↓ jne    29
       │   │↓ jmp    43
 11.47 │20:└─→cmpxch %esi,0x38a999(%rip)

That means the cmpl+je is a fused instruction pair and they should be considered together.

Changelog:

v3: Use Arnaldo's fix to improve the arrow origin rendering. To get the

    evsel->evlist->env->cpuid, save the evsel in annotate_browser.

v2: new function "ins__is_fused" to check if the instructions are fused.

Signed-off-by Yao Jin <[email protected]>
Tested-by Arnaldo Carvalho de Melo <[email protected]>
Cc Alexander Shishkin <[email protected]>
Cc Andi Kleen <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Kan Liang <[email protected]>
Cc Peter Zijlstra <[email protected]>
Link http://lkml.kernel.org/r/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/ui/browser.c           | 29 +++++++++++++++++++++++++++++
 tools/perf/ui/browser.h           |  2 ++
 tools/perf/ui/browsers/annotate.c | 26 ++++++++++++++++++++++++++
 tools/perf/util/annotate.c        |  5 +++++
 tools/perf/util/annotate.h        |  1 +
 5 files changed, 63 insertions(+)
tools/perf/ui/browser.c +29 -0
index 83874b0e266c..f73f3f13e01d 100644
@@ -738,6 +738,35 @@ void __ui_browser__line_arrow(struct ui_browser *browser, unsigned int column,
 		__ui_browser__line_arrow_down(browser, column, start, end);
 }
 
+void ui_browser__mark_fused(struct ui_browser *browser, unsigned int column,
+			    unsigned int row, bool arrow_down)
+{
+	unsigned int end_row;
+
+	if (row >= browser->top_idx)
+		end_row = row - browser->top_idx;
+	else
+		return;
+
+	SLsmg_set_char_set(1);
+
+	if (arrow_down) {
+		ui_browser__gotorc(browser, end_row, column - 1);
+		SLsmg_write_char(SLSMG_ULCORN_CHAR);
+		ui_browser__gotorc(browser, end_row, column);
+		SLsmg_draw_hline(2);
+		ui_browser__gotorc(browser, end_row + 1, column - 1);
+		SLsmg_write_char(SLSMG_LTEE_CHAR);
+	} else {
+		ui_browser__gotorc(browser, end_row, column - 1);
+		SLsmg_write_char(SLSMG_LTEE_CHAR);
+		ui_browser__gotorc(browser, end_row, column);
+		SLsmg_draw_hline(2);
+	}
+
+	SLsmg_set_char_set(0);
+}
+
 void ui_browser__init(void)
 {
 	int i = 0;
tools/perf/ui/browser.h +2 -0
index be3b70eb5fca..a12eff75638b 100644
@@ -43,6 +43,8 @@ void ui_browser__printf(struct ui_browser *browser, const char *fmt, ...);
 void ui_browser__write_graph(struct ui_browser *browser, int graph);
 void __ui_browser__line_arrow(struct ui_browser *browser, unsigned int column,
 			      u64 start, u64 end);
+void ui_browser__mark_fused(struct ui_browser *browser, unsigned int column,
+			    unsigned int row, bool arrow_down);
 void __ui_browser__show_title(struct ui_browser *browser, const char *title);
 void ui_browser__show_title(struct ui_browser *browser, const char *title);
 int ui_browser__show(struct ui_browser *browser, const char *title,
tools/perf/ui/browsers/annotate.c +26 -0
index c4336138b673..8d3f6f53c122 100644
@@ -273,6 +273,25 @@ static bool disasm_line__is_valid_jump(struct disasm_line *dl, struct symbol *sy
 	return true;
 }
 
+static bool is_fused(struct annotate_browser *ab, struct disasm_line *cursor)
+{
+	struct disasm_line *pos = list_prev_entry(cursor, node);
+	const char *name;
+
+	if (!pos)
+		return false;
+
+	if (ins__is_lock(&pos->ins))
+		name = pos->ops.locked.ins.name;
+	else
+		name = pos->ins.name;
+
+	if (!name || !cursor->ins.name)
+		return false;
+
+	return ins__is_fused(ab->arch, name, cursor->ins.name);
+}
+
 static void annotate_browser__draw_current_jump(struct ui_browser *browser)
 {
 	struct annotate_browser *ab = container_of(browser, struct annotate_browser, b);

@@ -308,6 +327,13 @@ static void annotate_browser__draw_current_jump(struct ui_browser *browser)
 	ui_browser__set_color(browser, HE_COLORSET_JUMP_ARROWS);
 	__ui_browser__line_arrow(browser, pcnt_width + 2 + ab->addr_width,
 				 from, to);
+
+	if (is_fused(ab, cursor)) {
+		ui_browser__mark_fused(browser,
+				       pcnt_width + 3 + ab->addr_width,
+				       from - 1,
+				       to > from ? true : false);
+	}
 }
 
 static unsigned int annotate_browser__refresh(struct ui_browser *browser)
tools/perf/util/annotate.c +5 -0
index 8748ebb3f932..ef434b53d849 100644
@@ -517,6 +517,11 @@ bool ins__is_ret(const struct ins *ins)
 	return ins->ops == &ret_ops;
 }
 
+bool ins__is_lock(const struct ins *ins)
+{
+	return ins->ops == &lock_ops;
+}
+
 static int ins__key_cmp(const void *name, const void *insp)
 {
 	const struct ins *ins = insp;
tools/perf/util/annotate.h +1 -0
index 72d72728a0fc..bac698d7cc6a 100644
@@ -52,6 +52,7 @@ struct ins_ops {
 bool ins__is_jump(const struct ins *ins);
 bool ins__is_call(const struct ins *ins);
 bool ins__is_ret(const struct ins *ins);
+bool ins__is_lock(const struct ins *ins);
 int ins__scnprintf(struct ins *ins, char *bf, size_t size, struct ins_operands *ops);
 bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2);
Signature · 1 line
2.9.4
author maintainer
#3 lore raw

PATCH 3/86 perf trace: Remove F_ from some of the fcntl command strings

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

The initial ones already had that "F_" prefix stripped to make things shorter, some hadn't, do it now.

We do this to make the 'perf trace' output more compact. At some point perhaps the best thing to do is to have the tool do this stripping automatically, letting the user also decide if this is to be done or not. For now, be consistent.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-trace.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
tools/perf/builtin-trace.c +3 -3
index 4b2a5d298197..cfe1858ed074 100644
@@ -407,9 +407,9 @@ static DEFINE_STRARRAY(whences);
 
 static const char *fcntl_cmds[] = {
 	"DUPFD", "GETFD", "SETFD", "GETFL", "SETFL", "GETLK", "SETLK",
-	"SETLKW", "SETOWN", "GETOWN", "SETSIG", "GETSIG", "F_GETLK64",
-	"F_SETLK64", "F_SETLKW64", "F_SETOWN_EX", "F_GETOWN_EX",
-	"F_GETOWNER_UIDS",
+	"SETLKW", "SETOWN", "GETOWN", "SETSIG", "GETSIG", "GETLK64",
+	"SETLK64", "SETLKW64", "SETOWN_EX", "GETOWN_EX",
+	"GETOWNER_UIDS",
 };
 static DEFINE_STRARRAY(fcntl_cmds);
Signature · 1 line
2.9.4
author maintainer
#4 lore raw

PATCH 4/86 perf trace: Beautify linux specific fcntl commands

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

We were only beautifying (transforming from an integer to its name) the non-linux specific fcntl syscall cmd args, fix it:

Before:

  # perf trace -e fcntl -p 2472
     0.000 ( 0.017 ms): gnome-terminal/2472 fcntl(fd: 55, cmd: 1030) = 56
  ^C#

After:

  # trace -e fcntl -p 2472
     0.000 ( 0.015 ms): gnome-terminal/2472 fcntl(fd: 55, cmd: DUPFD_CLOEXEC) = 56
  ^C#
Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-trace.c | 57 +++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 54 insertions(+), 3 deletions(-)
tools/perf/builtin-trace.c +54 -3
index cfe1858ed074..431ef70067ed 100644
@@ -64,6 +64,10 @@
 # define O_CLOEXEC		02000000
 #endif
 
+#ifndef F_LINUX_SPECIFIC_BASE
+# define F_LINUX_SPECIFIC_BASE	1024
+#endif
+
 struct trace {
 	struct perf_tool	tool;
 	struct syscalltbl	*sctbl;

@@ -317,6 +321,38 @@ static size_t syscall_arg__scnprintf_strarray(char *bf, size_t size,
 
 #define SCA_STRARRAY syscall_arg__scnprintf_strarray
 
+struct strarrays {
+	int		nr_entries;
+	struct strarray **entries;
+};
+
+#define DEFINE_STRARRAYS(array) struct strarrays strarrays__##array = { \
+	.nr_entries = ARRAY_SIZE(array), \
+	.entries = array, \
+}
+
+static size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size,
+					       struct syscall_arg *arg)
+{
+	struct strarrays *sas = arg->parm;
+	int i;
+
+	for (i = 0; i < sas->nr_entries; ++i) {
+		struct strarray *sa = sas->entries[i];
+		int idx = arg->val - sa->offset;
+
+		if (idx >= 0 && idx < sa->nr_entries) {
+			if (sa->entries[idx] == NULL)
+				break;
+			return scnprintf(bf, size, "%s", sa->entries[idx]);
+		}
+	}
+
+	return scnprintf(bf, size, "%d", arg->val);
+}
+
+#define SCA_STRARRAYS syscall_arg__scnprintf_strarrays
+
 #if defined(__i386__) || defined(__x86_64__)
 /*
  * FIXME: Make this available to all arches as soon as the ioctl beautifier

@@ -413,6 +449,20 @@ static const char *fcntl_cmds[] = {
 };
 static DEFINE_STRARRAY(fcntl_cmds);
 
+static const char *fcntl_linux_specific_cmds[] = {
+	"SETLEASE", "GETLEASE", "NOTIFY", [5] =	"CANCELLK", "DUPFD_CLOEXEC",
+	"SETPIPE_SZ", "GETPIPE_SZ", "ADD_SEALS", "GET_SEALS",
+};
+
+static DEFINE_STRARRAY_OFFSET(fcntl_linux_specific_cmds, F_LINUX_SPECIFIC_BASE);
+
+static struct strarray *fcntl_cmds_arrays[] = {
+	&strarray__fcntl_cmds,
+	&strarray__fcntl_linux_specific_cmds,
+};
+
+static DEFINE_STRARRAYS(fcntl_cmds_arrays);
+
 static const char *rlimit_resources[] = {
 	"CPU", "FSIZE", "DATA", "STACK", "CORE", "RSS", "NPROC", "NOFILE",
 	"MEMLOCK", "AS", "LOCKS", "SIGPENDING", "MSGQUEUE", "NICE", "RTPRIO",

@@ -613,8 +663,8 @@ static struct syscall_fmt {
 	{ .name	    = "fchownat",   .errmsg = true,
 	  .arg_scnprintf = { [0] = SCA_FDAT, /* fd */ }, },
 	{ .name	    = "fcntl",	    .errmsg = true,
-	  .arg_scnprintf = { [1] = SCA_STRARRAY, /* cmd */ },
-	  .arg_parm	 = { [1] = &strarray__fcntl_cmds, /* cmd */ }, },
+	  .arg_scnprintf = { [1] = SCA_STRARRAYS, /* cmd */ },
+	  .arg_parm	 = { [1] = &strarrays__fcntl_cmds_arrays, /* cmd */ }, },
 	{ .name	    = "fdatasync",  .errmsg = true, },
 	{ .name	    = "flock",	    .errmsg = true,
 	  .arg_scnprintf = { [1] = SCA_FLOCK, /* cmd */ }, },

@@ -1356,7 +1406,8 @@ static size_t syscall__scnprintf_args(struct syscall *sc, char *bf, size_t size,
  			 */
 			if (val == 0 &&
 			    !(sc->arg_scnprintf &&
-			      sc->arg_scnprintf[arg.idx] == SCA_STRARRAY &&
+			      (sc->arg_scnprintf[arg.idx] == SCA_STRARRAY ||
+			       sc->arg_scnprintf[arg.idx] == SCA_STRARRAYS) &&
 			      sc->arg_parm[arg.idx]))
 				continue;
Signature · 1 line
2.9.4
author maintainer
#5 lore raw

PATCH 5/86 tools: Update include/uapi/linux/fcntl.h copy from the kernel

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

To get the changes in the commit c75b1d9421f8 ("fs: add fcntl() interface for setting/getting write life time hints").

Silencing this perf build warning:

Warning: include/uapi/linux/fcntl.h differs from kernel

We already beautify the fcntl cmd argument, so an upcoming cset will update the 'cmd' strarray to cover these new commands.

The hints are in the 3rd arg, a pointer, so not yet supported in 'perf trace', for that we need to copy it somehow, probably using eBPF, a new attempt at doing that is planned.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jens Axboe <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Martin K. Petersen <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/include/uapi/linux/fcntl.h | 21 +++++++++++++++++++++
 1 file changed, 21 insertions(+)
tools/include/uapi/linux/fcntl.h +21 -0
index 813afd6eee71..ec69d55bcec7 100644
@@ -43,6 +43,27 @@
 /* (1U << 31) is reserved for signed error codes */
 
 /*
+ * Set/Get write life time hints. {GET,SET}_RW_HINT operate on the
+ * underlying inode, while {GET,SET}_FILE_RW_HINT operate only on
+ * the specific file.
+ */
+#define F_GET_RW_HINT		(F_LINUX_SPECIFIC_BASE + 11)
+#define F_SET_RW_HINT		(F_LINUX_SPECIFIC_BASE + 12)
+#define F_GET_FILE_RW_HINT	(F_LINUX_SPECIFIC_BASE + 13)
+#define F_SET_FILE_RW_HINT	(F_LINUX_SPECIFIC_BASE + 14)
+
+/*
+ * Valid hint values for F_{GET,SET}_RW_HINT. 0 is "not set", or can be
+ * used to clear any hints previously set.
+ */
+#define RWF_WRITE_LIFE_NOT_SET	0
+#define RWH_WRITE_LIFE_NONE	1
+#define RWH_WRITE_LIFE_SHORT	2
+#define RWH_WRITE_LIFE_MEDIUM	3
+#define RWH_WRITE_LIFE_LONG	4
+#define RWH_WRITE_LIFE_EXTREME	5
+
+/*
  * Types of directory notifications that may be requested.
  */
 #define DN_ACCESS	0x00000001	/* File accessed */
Signature · 1 line
2.9.4
author maintainer
#6 lore raw

PATCH 6/86 perf trace beauty: Export the strarrays scnprintf method

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

As we'll call it from the fcntl cmd scnprintf method, that needs to look at the cmd to mask the next fcntl argument when it is ignored.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-trace.c       | 6 ++----
 tools/perf/trace/beauty/beauty.h | 3 +++
 2 files changed, 5 insertions(+), 4 deletions(-)
tools/perf/builtin-trace.c +2 -4
index 431ef70067ed..ef1b1d4ea007 100644
@@ -331,8 +331,8 @@ struct strarrays {
 	.entries = array, \
 }
 
-static size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size,
-					       struct syscall_arg *arg)
+size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size,
+					struct syscall_arg *arg)
 {
 	struct strarrays *sas = arg->parm;
 	int i;

@@ -351,8 +351,6 @@ static size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size,
 	return scnprintf(bf, size, "%d", arg->val);
 }
 
-#define SCA_STRARRAYS syscall_arg__scnprintf_strarrays
-
 #if defined(__i386__) || defined(__x86_64__)
 /*
  * FIXME: Make this available to all arches as soon as the ioctl beautifier
tools/perf/trace/beauty/beauty.h +3 -0
index cf50be3f17a4..a6348073a6e9 100644
@@ -15,6 +15,9 @@ struct syscall_arg {
 	u8	      mask;
 };
 
+size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size, struct syscall_arg *arg);
+#define SCA_STRARRAYS syscall_arg__scnprintf_strarrays
+
 size_t syscall_arg__scnprintf_statx_flags(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_STATX_FLAGS syscall_arg__scnprintf_statx_flags
Signature · 1 line
2.9.4
author maintainer
#7 lore raw

PATCH 7/86 perf trace: Only build tools/perf/trace/beauty/ when building 'perf trace'

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

As it calls functions in builtin-trace.c.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/Build | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/Build +1 -1
index bd8eeb60533c..b48ca40fccf9 100644
@@ -50,6 +50,6 @@ libperf-y += util/
 libperf-y += arch/
 libperf-y += ui/
 libperf-y += scripts/
-libperf-y += trace/beauty/
+libperf-$(CONFIG_AUDIT) += trace/beauty/
 
 gtk-y += ui/gtk/
Signature · 1 line
2.9.4
author maintainer
#8 lore raw

PATCH 8/86 perf trace beauty: Mask ignored fcntl 'arg' parameter

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

A series of fcntl cmds ignore the third argument, so mask it.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-trace.c       |  2 +-
 tools/perf/trace/beauty/Build    |  1 +
 tools/perf/trace/beauty/beauty.h |  3 +++
 tools/perf/trace/beauty/fcntl.c  | 23 +++++++++++++++++++++++
 4 files changed, 28 insertions(+), 1 deletion(-)
 create mode 100644 tools/perf/trace/beauty/fcntl.c
tools/perf/builtin-trace.c +1 -1
index ef1b1d4ea007..b7f79dea3c44 100644
@@ -661,7 +661,7 @@ static struct syscall_fmt {
 	{ .name	    = "fchownat",   .errmsg = true,
 	  .arg_scnprintf = { [0] = SCA_FDAT, /* fd */ }, },
 	{ .name	    = "fcntl",	    .errmsg = true,
-	  .arg_scnprintf = { [1] = SCA_STRARRAYS, /* cmd */ },
+	  .arg_scnprintf = { [1] = SCA_FCNTL_CMD, /* cmd */ },
 	  .arg_parm	 = { [1] = &strarrays__fcntl_cmds_arrays, /* cmd */ }, },
 	{ .name	    = "fdatasync",  .errmsg = true, },
 	{ .name	    = "flock",	    .errmsg = true,
tools/perf/trace/beauty/Build +1 -0
index be95ac6ce845..c9e215b806f1 100644
@@ -1 +1,2 @@
+libperf-y += fcntl.o
 libperf-y += statx.o
tools/perf/trace/beauty/beauty.h +3 -0
index a6348073a6e9..ce01079d8422 100644
@@ -18,6 +18,9 @@ struct syscall_arg {
 size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_STRARRAYS syscall_arg__scnprintf_strarrays
 
+size_t syscall_arg__scnprintf_fcntl_cmd(char *bf, size_t size, struct syscall_arg *arg);
+#define SCA_FCNTL_CMD syscall_arg__scnprintf_fcntl_cmd
+
 size_t syscall_arg__scnprintf_statx_flags(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_STATX_FLAGS syscall_arg__scnprintf_statx_flags
tools/perf/trace/beauty/fcntl.c new file +23 -0
new file mode 100644
index 000000000000..7e4582c9308e
@@ -0,0 +1,23 @@
+/*
+ * trace/beauty/fcntl.c
+ *
+ *  Copyright (C) 2017, Red Hat Inc, Arnaldo Carvalho de Melo <[email protected]>
+ *
+ * Released under the GPL v2. (and only v2, not any later version)
+ */
+
+#include "trace/beauty/beauty.h"
+#include <uapi/linux/fcntl.h>
+
+size_t syscall_arg__scnprintf_fcntl_cmd(char *bf, size_t size, struct syscall_arg *arg)
+{
+	/*
+	 * Some commands ignore the third fcntl argument, "arg", so mask it
+	 */
+	if (arg->val == F_GETFD	   || arg->val == F_GETFL     ||
+	    arg->val == F_GETOWN   || arg->val == F_GET_SEALS ||
+	    arg->val == F_GETLEASE || arg->val == F_GETSIG)
+		arg->mask |= (1 << 2);
+
+	return syscall_arg__scnprintf_strarrays(bf, size, arg);
+}
Signature · 1 line
2.9.4
author maintainer
#9 lore raw

PATCH 9/86 perf trace beauty: Allow accessing syscall args values in a syscall arg formatter

In reply to Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

For instance, fcntl's upcoming 'arg' formatter needs to look at the 'cmd' value to decide how to format its value, sometimes it is a file flags, sometimes an fd, a pointer to a structure, etc.

Cc Adrian Hunter <[email protected]>
Cc David Ahern <[email protected]>
Cc Jiri Olsa <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Wang Nan <[email protected]>
Link http://lkml.kernel.org/n/[email protected] open here
Signed-off-by Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-trace.c       | 23 ++++++++++++++++-------
 tools/perf/trace/beauty/beauty.h | 13 +++++++++++++
 2 files changed, 29 insertions(+), 7 deletions(-)
tools/perf/builtin-trace.c +16 -7
index b7f79dea3c44..40bc0a326096 100644
@@ -1369,19 +1369,32 @@ static int trace__validate_ev_qualifier(struct trace *trace)
  * variable to read it. Most notably this avoids extended load instructions
  * on unaligned addresses
  */
+static unsigned long __syscall_arg__val(unsigned char *args, u8 idx)
+{
+	unsigned long val;
+	unsigned char *p = args + sizeof(unsigned long) * idx;
+
+	memcpy(&val, p, sizeof(val));
+	return val;
+}
+
+unsigned long syscall_arg__val(struct syscall_arg *arg, u8 idx)
+{
+	return __syscall_arg__val(arg->args, idx);
+}
 
 static size_t syscall__scnprintf_args(struct syscall *sc, char *bf, size_t size,
 				      unsigned char *args, struct trace *trace,
 				      struct thread *thread)
 {
 	size_t printed = 0;
-	unsigned char *p;
 	unsigned long val;
 
 	if (sc->args != NULL) {
 		struct format_field *field;
 		u8 bit = 1;
 		struct syscall_arg arg = {
+			.args	= args,
 			.idx	= 0,
 			.mask	= 0,
 			.trace  = trace,

@@ -1393,9 +1406,7 @@ static size_t syscall__scnprintf_args(struct syscall *sc, char *bf, size_t size,
 			if (arg.mask & bit)
 				continue;
 
-			/* special care for unaligned accesses */
-			p = args + sizeof(unsigned long) * arg.idx;
-			memcpy(&val, p, sizeof(val));
+			val = syscall_arg__val(&arg, arg.idx);
 
 			/*
  			 * Suppress this argument if its value is zero and

@@ -1431,9 +1442,7 @@ static size_t syscall__scnprintf_args(struct syscall *sc, char *bf, size_t size,
 		int i = 0;
 
 		while (i < 6) {
-			/* special care for unaligned accesses */
-			p = args + sizeof(unsigned long) * i;
-			memcpy(&val, p, sizeof(val));
+			val = __syscall_arg__val(args, i);
 			printed += scnprintf(bf + printed, size - printed,
 					     "%sarg%d: %ld",
 					     printed ? ", " : "", i, val);
tools/perf/trace/beauty/beauty.h +13 -0
index ce01079d8422..6fbac0c8120d 100644
@@ -6,8 +6,19 @@
 struct trace;
 struct thread;
 
+/**
+ * @val: value of syscall argument being formatted
+ * @args: All the args, use syscall_args__val(arg, nth) to access one
+ * @thread: tid state (maps, pid, tid, etc)
+ * @trace: 'perf trace' internals: all threads, etc
+ * @parm: private area, may be an strarray, for instance
+ * @idx: syscall arg idx (is this the first?)
+ * @mask: a syscall arg may mask another arg, see syscall_arg__scnprintf_futex_op
+ */
+
 struct syscall_arg {
 	unsigned long val;
+	unsigned char *args;
 	struct thread *thread;
 	struct trace  *trace;
 	void	      *parm;

@@ -15,6 +26,8 @@ struct syscall_arg {
 	u8	      mask;
 };
 
+unsigned long syscall_arg__val(struct syscall_arg *arg, u8 idx);
+
 size_t syscall_arg__scnprintf_strarrays(char *bf, size_t size, struct syscall_arg *arg);
 #define SCA_STRARRAYS syscall_arg__scnprintf_strarrays
Signature · 1 line
2.9.4