Prev: mdadm software raid + ext4, capped at ~350MiB/s limitation/bug?
Next: [PATCH 2/2] power_supply: bq27x00: fix voltage and current units
From: Ingo Molnar on 27 Feb 2010 09:30
Linus,

Please pull the latest perf-core-for-linus git tree from:

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git perf-core-for-linus

It was a very busy cycle for perf, with lots of improvements:

131 files changed, 8461 insertions(+), 3335 deletions(-)

A few highlights:

- New 'session' support, with cross-arch features: it's possible to record a
session on 32-bit PA/RISC and then display a correct 'perf report' on
64-bit x86. (this specific combination was tested in practice)

- Rewritten auto-freq sampling code, there should be no sampling artifacts in
'perf report' output anymore:

45.08% multi-fault-all [kernel] [k] smp_invalidate_interrupt
19.45% multi-fault-all [kernel] [k] intel_pmu_enable_all
14.17% multi-fault-all [kernel] [k] flush_tlb_others_ipi

That 'intel_pmu_enable_all()' entry is now gone.

- New x86 event constraint handling code, to correctly measure certain event
combinations.

- Improved script engine for 'perf trace' and Python support.

- New 'perf lock' tool to profile locking patterns, which
plugs into the lockstat callbacks.

- Expanded x86 CPU support: Westmere support and support for intel core solo/duo.

- Various 'perf probe' enhancements and refinements.

- Improved symbol handling, buildid-cache in the .debug directory hierarchy
and improved debuginfo processing support.

- Improved 'perf annotate'.

- New 'perf archive' tool: create archive with object files with build-ids
found in the perf.data file, into perf.data.tar.bz2, to transport
self-contained profiling info and profiled images across machines.

( There's two more perf trees pending: perf/nmi and perf/probes, with two
other features that we kept separate. )

Thanks,

Ingo

------------------>
Alexander Beregalov (1):
perf: Fix memory leak: counterwidth

Amerigo Wang (1):
perf: Use format string of printf to align strings

Anton Blanchard (1):
perf: Fix inconsistency between IP and callchain sampling

Arnaldo Carvalho de Melo (63):
perf session: Move the event processing routines to session.c
perf tools: Move the map class definition to a separate header
perf session: Share the common trace sample_check routine as perf_session__has_traces
perf session: Remove sample_type_check from event_ops
perf session: Move total_unknown to perf_session->unknown events
perf session: Move full_paths config to symbol_conf
perf session: Remove redundant prefix & suffix from perf_event_ops
perf record: Introduce a symtab cache
perf header: Do_read shouldn't die
perf header: perf_header__push_event() shouldn't die
perf report: Add --hide-unresolved/-U command line option
perf diff: Don't add the period for unresolved symbols
perf diff: Fix support for all --sort combinations
perf symbols: Generalise the kallsyms parsing routine
perf symbols: Export symbol_type__is_a
perf tools: Create write_padded routine out of __dsos__write_buildid_table
perf session: Keep pointers to the vmlinux maps
perf report: Fix --no-call-chain option handling
perf tools: Handle relocatable kernels
perf buildid-list: No need to process the header sections again
perf symbols: Record the domain of DSOs in HEADER_BUILD_ID header table
perf tools: Create typedef for common event synthesizing callback
perf tools: Encode kernel module mappings in perf.data
perf tools: Don't cast RIP to pointers
perf tools: Cross platform perf.data analysis support
perf tools: Convert getpagesize() uses to sysconf(_SC_GETPAGESIZE)
perf symbols: Don't try to load kallsyms if doesn't match the record build-id
perf symbols: Cache /proc/kallsyms files by build-id
perf symbols: Use dso->long_name in dsos__find()
perf record: Encode the domain while synthesizing MMAP events
perf symbols: Create thread__find_addr_map from thread__find_addr_location
perf buildid-list: Introduce --with-hits option
perf symbols: The synthesized kernel modules MMAP must use the pathnames
perf archive: Add helper script to package files needed to do analysis
perf record: Intercept all events
perf symbols: Accept an alias when looking for "_text"
perf archive: Add documentation
perf archive: Fix installation steps in the Makefile
perf symbols: Set dso->kernel when handling the fake vmlinux MMAP event
perf symbols: Use only --vmlinux if specified
perf buildid-cache: Add new command to manage build-id cache
perf top: Fix sample counting
perf top: Handle PERF_RECORD_{FORK,EXIT} events
perf symbols: Fix inverted logic for showing kallsyms as the source of symbols
perf symbols: Use the right variable to check for kallsyms in the cache
perf top: Exit if specified --vmlinux can't be used
perf symbols: Factor out dso__load_vmlinux_path()
perf symbols: Split helpers used when creating kernel dso object
perf session: Create kernel maps in the constructor
perf symbols: Remove perf_session usage in symbols layer
perf symbols: Fixup vsyscall maps
perf symbols: Ditch vdso global variable
perf probe: Don't use a perf_session instance just to resolve symbols
perf build-id: Move the routine to find DSOs with hits to the lib
perf record: Stop intercepting events, use postprocessing to get build-ids
perf tools: Adjust some verbosity levels
perf top: Use address pattern in lookup_sym_source
perf symbols: Fix up map end too on modular kernels with no modules installed
perf tools: Don't use parent comm if not set at fork time
perf top: Use a macro instead of a constant variable
perf symbols: Improve debugging information about symtab origins
perf annotate: Defer allocating sym_priv->hist array
perf annotate: Handle samples not at objdump output addr boundaries

David S. Miller (1):
perf tools: Flush maps on COMM events

Frederic Weisbecker (11):
perf: Drop useless check for ignored frame
perf/core: Split context's event group list into pinned and non-pinned lists
list: Introduce list_rotate_left()
perf: Round robin flexible groups of events using list_rotate_left()
perf: Export software-only event group characteristic as a flag
perf: Fix implicit declaration of getline in util.c
perf: Make __perf_event_sched_out static
perf: Allow pinned and flexible groups to be scheduled separately
perf: Don't schedule out/in pinned events on task tick
perf: Better order flexible and pinned scheduling
perf/scripts: Tag syscall_name helper as not yet available

Hitoshi Mitake (7):
perf probe: Fix build error of builtin-probe.c
perf trace: Add -i option for choosing input file
Revert "perf record: Intercept all events"
perf tools: Add __data_loc support
perf: Add util/include/linuxhash.h to include hash.h of kernel
perf lock: Enhance information of lock trace events
perf lock: Introduce new tool "perf lock", for analyzing lock statistics

Ingo Molnar (3):
perf, x86: Clean up event constraints code a bit
perf lock: Clean up various details
perf tools: Fix session init on non-modular kernels

Jamie Iles (1):
sched/perf: Make sure irqs are disabled for perf_event_task_sched_in()

John Kacur (1):
perf: Ignore perf-archive temp file

K.Prasad (2):
x86/debug: Clear reserved bits of DR6 in do_debug()
x86/hw-breakpoints: Optimize return code from notifier chain in hw_breakpoint_handler

Kirill Smelkov (5):
perf: Fix few typos + cosmetics
perf top: Fix code typo in prompt_symbol()
perf annotate: Fix it for non-prelinked *.so
perf top: Teach it to autolocate vmlinux
perf top: Fix annotate for userspace

Li Zefan (1):
perf events: Remove CONFIG_EVENT_PROFILE

Liming Wang (3):
perf tools: Unify event type description
perf probe: Change CONFIG_KPROBE_TRACER to CONFIG_KPROBE_EVENT
perf tools: Fix --pid option for stat

Masami Hiramatsu (15):
tracing/kprobe: Update example output in documentation
tracing/kprobe: Drop function argument access syntax
x86/ptrace: Remove unused regs_get_argument_nth API
perf probe: Remove newline from die()
perf probe: Show probe list in pager
perf tools: Support tracepoint glob matching
perf tools: Enhance glob string matching
perf probe: Support --line option to show probable source-code lines
tracing/kprobe: Update kprobe tracing self test for new syntax
kprobes: Disable booster when CONFIG_PREEMPT=y
ftrace/alternatives: Introducing *_text_reserved functions
kprobes: Check probe address is reserved
ftrace: Remove record freezing
kprobes: Add mcount to the kprobes blacklist
x86/alternatives: Fix build warning

Mike Galbraith (1):
perf annotate: Fix perf top module symbol annotation

Motohiro KOSAKI (2):
kprobetrace, doc: Shell needs single quote to use $ character
kprobetrace, doc: Add the explanation to remove probe points

Naga Chumbalkar (1):
x86, perfctr: Remove unused func avail_to_resrv_perfctr_nmi()

Peter Zijlstra (26):
perf events: Remove arg from perf sched hooks
perf: Reimplement frequency driven sampling
perf_event: x86: Allocate the fake_cpuc
perf_event: x86: Fixup constraints typing issue
perf_event: x86: Clean up some of the u64/long bitmask casting
perf_event: x86: Reduce some overly long lines with some MACROs
bitops: Provide compile time HWEIGHT{8,16,32,64}
perf_event: x86: Optimize the constraint searching bits
perf_event: x86: Optimize constraint weight computation
perf_event: x86: Optimize the fast path a little more
perf_event: x86: Optimize x86_pmu_disable()
perf_event: x86: Deduplicate the disable code
perf_events, x86: Fix event constraint masks
perf_events, x86: Clean up hw_perf_*_all() implementation
perf_events, x86: Implement Intel Westmere support
perf_events, x86: Remove spurious counter reset from x86_pmu_enable()
perf_events: Fix sample_period transfer on inherit
perf_events: Optimize perf_event_task_tick()
perf_events, x86: Implement intel core solo/duo support
bitops: Ensure the compile time HWEIGHT is only used for such
perf_events: Report the MMAP pgoff value in bytes
perf_events: Simplify code by removing cpu argument to hw_perf_group_sched_in()
perf_events, x86: Remove superflous MSR writes
perf_events, x86: Split PMU definitions into separate files
perf_event: Fix preempt warning in perf_clock()
perf_event, amd: Fix spinlock initialization

Stephane Eranian (5):
perf_events, x86: Improve x86 event scheduling
perf_events: Add fast-path to the rescheduling code
perf_events, x86: Fix bug in hw_perf_enable()
perf_events: Add new start/stop PMU callbacks
perf_events, x86: AMD event scheduling

Thiago Farina (1):
tools/perf/perf.c: Clean up trivial style issues

Tom Zanussi (9):
perf/scripts: Fix supported language listing option
perf/scripts: Fix bug in Util.pm
perf/scripts: Move common code out of Perl-specific files
perf/scripts: Move Perl scripting files to scripting-engines dir
perf/scripts: Remove check-perf-trace from listed scripts
perf/scripts: Add Python scripting engine
perf/scripts: Add syscall tracing scripts
perf/scripts: Remove unnecessary PyTuple resizes
perf/scripts: Add perf-trace-python Documentation

Ulrich Drepper (1):
perf tools: Do a few more directory handling optimizations

Wenji Huang (1):
perf: Make cmd_to_page() function more compact

Xiao Guangrong (7):
perf tools: Mount debugfs automatically
perf trace: Clean up find_debugfs()
perf trace: Fix forgotten close of file/dir
perf: Factorize trace events raw sample buffer operations
tracing/kprobe: Cleanup unused return value of tracing functions
perf tools: Use O_LARGEFILE to open perf data file
perf tools: Clean up O_LARGEFILE et al usage

Zhang, Yanmin (1):
perf symbols: Check the right return variable

austin_zhang(a)linux.intel.com (1):
perf record: Fix existing process callgraph symbol


Documentation/trace/kprobetrace.txt | 57 +-
arch/ia64/kernel/kprobes.c | 2 +-
arch/powerpc/kernel/perf_callchain.c | 3 -
arch/powerpc/kernel/perf_event.c | 10 +-
arch/sh/kernel/perf_callchain.c | 3 -
arch/sparc/kernel/perf_event.c | 10 +-
arch/x86/include/asm/alternative.h | 5 +
arch/x86/include/asm/debugreg.h | 3 +
arch/x86/include/asm/nmi.h | 1 -
arch/x86/include/asm/perf_event.h | 16 +-
arch/x86/include/asm/ptrace.h | 4 -
arch/x86/include/asm/stacktrace.h | 2 -
arch/x86/kernel/alternative.c | 18 +
arch/x86/kernel/cpu/perf_event.c | 1854 ++++++--------------
arch/x86/kernel/cpu/perf_event_amd.c | 416 +++++
arch/x86/kernel/cpu/perf_event_intel.c | 971 ++++++++++
arch/x86/kernel/cpu/perf_event_p6.c | 157 ++
arch/x86/kernel/cpu/perfctr-watchdog.c | 11 -
arch/x86/kernel/dumpstack_32.c | 5 -
arch/x86/kernel/dumpstack_64.c | 5 -
arch/x86/kernel/hw_breakpoint.c | 10 +-
arch/x86/kernel/kprobes.c | 5 +-
arch/x86/kernel/ptrace.c | 24 -
arch/x86/kernel/traps.c | 3 +
include/linux/bitops.h | 29 +-
include/linux/ftrace.h | 7 +-
include/linux/ftrace_event.h | 20 +-
include/linux/list.h | 14 +
include/linux/perf_event.h | 55 +-
include/linux/syscalls.h | 4 +-
include/trace/events/lock.h | 29 +-
include/trace/ftrace.h | 60 +-
include/trace/syscall.h | 4 +-
init/Kconfig | 13 -
kernel/kprobes.c | 34 +-
kernel/perf_event.c | 627 +++++---
kernel/sched.c | 12 +-
kernel/trace/Makefile | 4 +-
kernel/trace/ftrace.c | 54 +-
kernel/trace/trace_event_profile.c | 52 +-
kernel/trace/trace_events_filter.c | 4 +-
kernel/trace/trace_kprobe.c | 196 +--
kernel/trace/trace_syscalls.c | 76 +-
tools/perf/.gitignore | 1 +
tools/perf/Documentation/perf-archive.txt | 22 +
tools/perf/Documentation/perf-buildid-cache.txt | 33 +
tools/perf/Documentation/perf-probe.txt | 20 +
tools/perf/Documentation/perf-top.txt | 2 +-
tools/perf/Documentation/perf-trace-perl.txt | 2 +-
tools/perf/Documentation/perf-trace-python.txt | 625 +++++++
tools/perf/Documentation/perf-trace.txt | 15 +-
tools/perf/Documentation/perf.txt | 2 +-
tools/perf/Makefile | 51 +-
tools/perf/builtin-annotate.c | 240 ++-
tools/perf/builtin-buildid-cache.c | 133 ++
tools/perf/builtin-buildid-list.c | 31 +-
tools/perf/builtin-diff.c | 74 +-
tools/perf/builtin-help.c | 5 +-
tools/perf/builtin-kmem.c | 48 +-
tools/perf/builtin-lock.c | 678 +++++++
tools/perf/builtin-probe.c | 94 +-
tools/perf/builtin-record.c | 50 +-
tools/perf/builtin-report.c | 58 +-
tools/perf/builtin-sched.c | 32 +-
tools/perf/builtin-stat.c | 106 +-
tools/perf/builtin-timechart.c | 25 +-
tools/perf/builtin-top.c | 109 +-
tools/perf/builtin-trace.c | 34 +-
tools/perf/builtin.h | 2 +
tools/perf/command-list.txt | 2 +
tools/perf/design.txt | 8 +-
tools/perf/perf-archive.sh | 32 +
tools/perf/perf.c | 25 +-
tools/perf/scripts/perl/Perf-Trace-Util/Context.c | 5 +-
tools/perf/scripts/perl/Perf-Trace-Util/Context.xs | 3 +-
.../perl/Perf-Trace-Util/lib/Perf/Trace/Util.pm | 2 +-
.../perf/scripts/perl/bin/check-perf-trace-record | 7 +-
.../perf/scripts/perl/bin/check-perf-trace-report | 6 -
tools/perf/scripts/perl/bin/failed-syscalls-record | 2 +
tools/perf/scripts/perl/bin/failed-syscalls-report | 4 +
tools/perf/scripts/perl/failed-syscalls.pl | 38 +
.../perf/scripts/python/Perf-Trace-Util/Context.c | 88 +
.../python/Perf-Trace-Util/lib/Perf/Trace/Core.py | 91 +
.../python/Perf-Trace-Util/lib/Perf/Trace/Util.py | 25 +
.../python/bin/failed-syscalls-by-pid-record | 2 +
.../python/bin/failed-syscalls-by-pid-report | 4 +
.../python/bin/syscall-counts-by-pid-record | 2 +
.../python/bin/syscall-counts-by-pid-report | 4 +
.../perf/scripts/python/bin/syscall-counts-record | 2 +
.../perf/scripts/python/bin/syscall-counts-report | 4 +
tools/perf/scripts/python/check-perf-trace.py | 83 +
.../perf/scripts/python/failed-syscalls-by-pid.py | 68 +
tools/perf/scripts/python/syscall-counts-by-pid.py | 64 +
tools/perf/scripts/python/syscall-counts.py | 58 +
tools/perf/util/build-id.c | 39 +
tools/perf/util/build-id.h | 8 +
tools/perf/util/data_map.c | 252 ---
tools/perf/util/debug.c | 1 +
tools/perf/util/debugfs.c | 17 +-
tools/perf/util/debugfs.h | 2 +-
tools/perf/util/event.c | 220 ++-
tools/perf/util/event.h | 79 +-
tools/perf/util/header.c | 284 +++-
tools/perf/util/header.h | 9 +-
tools/perf/util/include/linux/hash.h | 5 +
tools/perf/util/include/linux/kernel.h | 1 +
tools/perf/util/map.c | 52 +-
tools/perf/util/map.h | 94 +
tools/perf/util/parse-events.c | 48 +-
tools/perf/util/probe-event.c | 105 ++-
tools/perf/util/probe-event.h | 2 +
tools/perf/util/probe-finder.c | 203 ++-
tools/perf/util/probe-finder.h | 33 +
.../{ => scripting-engines}/trace-event-perl.c | 115 +--
.../util/scripting-engines/trace-event-python.c | 573 ++++++
tools/perf/util/session.c | 431 +++++-
tools/perf/util/session.h | 55 +-
tools/perf/util/string.c | 65 +-
tools/perf/util/symbol.c | 529 ++++--
tools/perf/util/symbol.h | 52 +-
tools/perf/util/thread.c | 52 +-
tools/perf/util/thread.h | 24 +-
tools/perf/util/trace-event-info.c | 64 +-
tools/perf/util/trace-event-parse.c | 24 +
tools/perf/util/trace-event-perl.h | 55 -
tools/perf/util/trace-event-read.c | 18 +-
tools/perf/util/trace-event-scripting.c | 167 ++
tools/perf/util/trace-event.h | 10 +-
tools/perf/util/util.c | 94 +
tools/perf/util/util.h | 3 +
tools/perf/util/values.c | 1 +
131 files changed, 8461 insertions(+), 3335 deletions(-)
create mode 100644 arch/x86/kernel/cpu/perf_event_amd.c
create mode 100644 arch/x86/kernel/cpu/perf_event_intel.c
create mode 100644 arch/x86/kernel/cpu/perf_event_p6.c
create mode 100644 tools/perf/Documentation/perf-archive.txt
create mode 100644 tools/perf/Documentation/perf-buildid-cache.txt
create mode 100644 tools/perf/Documentation/perf-trace-python.txt
create mode 100644 tools/perf/builtin-buildid-cache.c
create mode 100644 tools/perf/builtin-lock.c
create mode 100644 tools/perf/perf-archive.sh
delete mode 100644 tools/perf/scripts/perl/bin/check-perf-trace-report
create mode 100644 tools/perf/scripts/perl/bin/failed-syscalls-record
create mode 100644 tools/perf/scripts/perl/bin/failed-syscalls-report
create mode 100644 tools/perf/scripts/perl/failed-syscalls.pl
create mode 100644 tools/perf/scripts/python/Perf-Trace-Util/Context.c
create mode 100644 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Core.py
create mode 100644 tools/perf/scripts/python/Perf-Trace-Util/lib/Perf/Trace/Util.py
create mode 100644 tools/perf/scripts/python/bin/failed-syscalls-by-pid-record
create mode 100644 tools/perf/scripts/python/bin/failed-syscalls-by-pid-report
create mode 100644 tools/perf/scripts/python/bin/syscall-counts-by-pid-record
create mode 100644 tools/perf/scripts/python/bin/syscall-counts-by-pid-report
create mode 100644 tools/perf/scripts/python/bin/syscall-counts-record
create mode 100644 tools/perf/scripts/python/bin/syscall-counts-report
create mode 100644 tools/perf/scripts/python/check-perf-trace.py
create mode 100644 tools/perf/scripts/python/failed-syscalls-by-pid.py
create mode 100644 tools/perf/scripts/python/syscall-counts-by-pid.py
create mode 100644 tools/perf/scripts/python/syscall-counts.py
create mode 100644 tools/perf/util/build-id.c
create mode 100644 tools/perf/util/build-id.h
delete mode 100644 tools/perf/util/data_map.c
create mode 100644 tools/perf/util/include/linux/hash.h
create mode 100644 tools/perf/util/map.h
rename tools/perf/util/{ => scripting-engines}/trace-event-perl.c (85%)
create mode 100644 tools/perf/util/scripting-engines/trace-event-python.c
delete mode 100644 tools/perf/util/trace-event-perl.h
create mode 100644 tools/perf/util/trace-event-scripting.c
create mode 100644 tools/perf/util/util.c

[ 500K+ patch left out due to lkml size limits ]
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: mdadm software raid + ext4, capped at ~350MiB/s limitation/bug?
Next: [PATCH 2/2] power_supply: bq27x00: fix voltage and current units