Ingo Molnar [Tue, 13 Nov 2012 18:05:09 +0000 (19:05 +0100)]
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:
* Don't show scripts menu for 'perf top', fix from Feng Tang
* Add framework for automated perf_event_attr tests, where tools with
different command line options will be run from a 'perf test', via
python glue, and the perf syscall will be intercepted to verify that
the perf_event_attr fields set by the tool are those expected,
from Jiri Olsa
* Use normalized arch name for searching objdump path. This fixes cases
where the system's objdump (e.g. x86_64) supports the architecture in
the perf.data file (e.g. i686), but is not the same,
fix from Namhyung Kim.
* Postpone objdump check until annotation requested, from Namhyung Kim.
* Add a 'link' method for hists, so that we can have the leader with
buckets for all the entries in all the hists. This new method
is now used in the default 'diff' output, making the sum of the 'baseline'
column be 100%, eliminating blind spots. Now we need to use this
for 'diff' with > 2 perf.data files and for multi event 'report' and
'annotate'.
* libtraceevent fixes for compiler warnings trying to make perf it build
on some distros, like fedora 14, 32-bit, some of the warnings really
pointed to real bugs.
* Remove temp dir on failure in 'perf test', fix from Jiri Olsa.
* Fixes for handling data, stack mmaps, from Namhyung Kim.
* Fix live annotation bug related to recent objdump lookup patches, from
Namhyung Kim
* Don't try to follow jump target on PLT symbols in the annotation browser,
fix from Namhyung Kim.
* Fix leak on hist_entry delete, from Namhyung Kim.
* Fix a CPU_ALLOC related build error on builtin-test, from Zheng Liu.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org>
tools lib traceevent: Use 'const' in variables pointing to const strings
Fixing the build on fedora 14, 32-bit:
tools/lib/traceevent/event-parse.c: In function ‘find_cmdline’:
tools/lib/traceevent/event-parse.c:183:3: error: return discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c:186:3: error: return discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c:195:2: error: return discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c: In function ‘process_func_handler’:
tools/lib/traceevent/event-parse.c:2658:9: error: assignment discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c:2660:9: error: assignment discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c: In function ‘print_mac_arg’:
tools/lib/traceevent/event-parse.c:3892:14: error: initialization discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c:3906:7: error: assignment discards qualifiers from pointer target type
tools/lib/traceevent/event-parse.c: In function ‘pevent_print_event’:
tools/lib/traceevent/event-parse.c:4412:24: error: initialization discards qualifiers from pointer target type
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/n/tip-0k5g8urwu7vwkgbcbt2x05fe@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools lib traceevent: No need to check for < 0 on an unsigned enum
gcc on f14 32-bit complains:
tools/lib/traceevent/event-parse.c: In function ‘pevent_register_print_function’:
tools/lib/traceevent/event-parse.c:5366:3: error: comparison of unsigned expression < 0 is always false
This is because:
enum pevent_func_arg_type type;
this enum doesn't have any negative value, so gcc makes it an 'unsigned
int'. Fix it by removing the < 0 test.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/n/tip-6vnd6ud6fbpn48zax4a5ru01@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools lib traceevent: Avoid comparisions between signed/unsigned
Fixing this warning-as-error on f14 32-bit:
tools/lib/traceevent/event-parse.c:5564:17: error: comparison between signed and unsigned integer expressions
tools/lib/traceevent/event-parse.c:5586:17: error: comparison between signed and unsigned integer expressions
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/n/tip-stmix8hy4nu5ervpynn8yj2z@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
tools lib traceevent: Add __maybe_unused to unused parameters
Fixing the build on 32-bit Fedora 14:
tools/lib/traceevent/event-parse.c: In function ‘print_event_fields’:
tools/lib/traceevent/event-parse.c:3934:69: error: unused parameter ‘size’
tools/lib/traceevent/event-parse.c: In function ‘pevent_strerror’:
tools/lib/traceevent/event-parse.c:5074:36: error: unused parameter ‘pevent’
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Steven Rostedt <rostedt@goodmis.org> Link: http://lkml.kernel.org/n/tip-soe4gqcz8fd4ecik6exvyqox@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Fri, 9 Nov 2012 05:58:49 +0000 (14:58 +0900)]
perf annotate: Merge same lines in summary view
The --print-line option of perf annotate command shows summary for
each source line. But it didn't merge same lines so that it can
appear multiple times.
Namhyung Kim [Fri, 9 Nov 2012 17:27:15 +0000 (02:27 +0900)]
perf annotate: Don't try to follow jump target on PLT symbols
The perf annotate browser on TUI can identify a jump target for a
selected instruction. It assumes that the jump target is within the
function but it's not the case of PLT symbols which have offset out of
the function as a target.
Since it caused a segmentation fault, do not try to follow jump target
on the PLT symbols.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Andi Kleen <ak@linux.intel.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/r/1352482044-3443-5-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Zheng Liu [Fri, 9 Nov 2012 00:58:46 +0000 (16:58 -0800)]
perf test: fix a build error on builtin-test
Recently I build perf and get a build error on builtin-test.c. The error is as
following:
$ make
CC perf.o
CC builtin-test.o
cc1: warnings being treated as errors
builtin-test.c: In function ‘sched__get_first_possible_cpu’:
builtin-test.c:977: warning: implicit declaration of function ‘CPU_ALLOC’
builtin-test.c:977: warning: nested extern declaration of ‘CPU_ALLOC’
builtin-test.c:977: warning: assignment makes pointer from integer without a cast
builtin-test.c:978: warning: implicit declaration of function ‘CPU_ALLOC_SIZE’
builtin-test.c:978: warning: nested extern declaration of ‘CPU_ALLOC_SIZE’
builtin-test.c:979: warning: implicit declaration of function ‘CPU_ZERO_S’
builtin-test.c:979: warning: nested extern declaration of ‘CPU_ZERO_S’
builtin-test.c:982: warning: implicit declaration of function ‘CPU_FREE’
builtin-test.c:982: warning: nested extern declaration of ‘CPU_FREE’
builtin-test.c:992: warning: implicit declaration of function ‘CPU_ISSET_S’
builtin-test.c:992: warning: nested extern declaration of ‘CPU_ISSET_S’
builtin-test.c:998: warning: implicit declaration of function ‘CPU_CLR_S’
builtin-test.c:998: warning: nested extern declaration of ‘CPU_CLR_S’
make: *** [builtin-test.o] Error 1
This problem is introduced in 3e7c439a. CPU_ALLOC and related macros are
missing in sched__get_first_possible_cpu function. In 54489c18, commiter
mentioned that CPU_ALLOC has been removed. So CPU_ALLOC calls in this
function are removed to let perf to be built.
Signed-off-by: Vinson Lee <vlee@twitter.com> Signed-off-by: Zheng Liu <wenqing.lz@taobao.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Cc: Vinson Lee <vlee@twitter.com> Cc: Zheng Liu <wenqing.lz@taobao.com> Cc: stable@vger.kernel.org Link: http://lkml.kernel.org/r/1352422726-31114-1-git-send-email-vlee@twitter.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf diff: Use hists__link when not pairing just with baseline
Previously there were blind spots because we were not looking at symbols
that didn't ocurred in the latest run:
# perf record usleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.018 MB perf.data (~801 samples) ]
# perf record usleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.018 MB perf.data (~801 samples) ]
That given two hists will find the hist_entries (buckets) in the second
hists that are for the same bucket in the first and link them, then it
will look for all buckets in the second that don't have a counterpart in
the first and will create a dummy counterpart that will then be linked
to the entry in the second.
For multiple events this will be done pairing the leader with all the
other events in the group, so that in the end the leader will have all
the buckets in all the hists in a group, dummy or not while the other
hists will be left untouched.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-l9l9ieozqdhn9lieokd95okw@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf diff: Start moving to support matching more than two hists
We want to match more than two hists, so that we can match more than two
perf.data files and moreover, match hist_entries (buckets) in multiple
events in a group.
So the "baseline"/"leader" will instead of a ->pair pointer, use a
list_head, that will link to the pairs and hists__match use it.
Following that perf_evlist__link will link the hists in its evsel
groups.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-2kbmzepoi544ygj9godseqpv@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Wed, 7 Nov 2012 01:30:15 +0000 (10:30 +0900)]
perf tools: Don't try to lookup objdump for live mode
Arnaldo reported that annotation during perf top resulted in a segfault.
It was because the env->arch was NULL and we don't set it for a live
session. In fact, no need to look up objdump in this case since we can
use system's default (native) objdump.
Reported-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Irina Tirdea <irina.tirdea@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1352251815-12615-1-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Olsa [Thu, 8 Nov 2012 16:01:01 +0000 (17:01 +0100)]
perf tests: Move attr.py temp dir cleanup into finally section
Currently if there's 'Unsup' exception raised, we do not clean up the
temp directory. Solving this by adding 'finally' to make the cleanup in
any case.
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1352390461-15404-1-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Wed, 7 Nov 2012 07:27:11 +0000 (16:27 +0900)]
perf tools: Fix detection of stack area
Output of /proc/<pid>/maps contains helpful information to anonymous
mappings like stack, heap, ... For the case of stack, it can show
multiple stack area for each thread in the process:
Namhyung Kim [Fri, 2 Nov 2012 05:50:05 +0000 (14:50 +0900)]
perf tools: Introduce struct hist_browser_timer
Currently various hist browser functions receive 3 arguments for
refreshing histogram but only used from a few places. Also it's only
for perf top command so that it can be NULL for other (and probably
most) cases. Pack them into a struct in order to reduce number of those
unused arguments.
This is a mechanical change and does not intend a functional change.
Signed-off-by: Namhyung Kim <namhyung@kernel.org> Tested-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Irina Tirdea <irina.tirdea@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1351835406-15208-2-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Fri, 2 Nov 2012 05:50:04 +0000 (14:50 +0900)]
perf tools: Use normalized arch name for searching objdump path
David reported that perf report for i686 target data on x86_64 host
failed to work because it tried to find out cross-compiled objdump.
However objdump for x86_64 is compatible to i686 so that it doesn't need
to do it at all. To prevent similar artifacts, normalize arch name when
comparing host and file architectures.
Reported-by: David Ahern <dsahern@gmail.com> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Reviewed-by: David Ahern <dsahern@gmail.com> Tested-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: Irina Tirdea <irina.tirdea@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1351835406-15208-1-git-send-email-namhyung@kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Steven Rostedt [Fri, 2 Nov 2012 02:56:07 +0000 (22:56 -0400)]
tracing: Add trace_options kernel command line parameter
Add trace_options to the kernel command line parameter to be able to
set options at early boot. For example, to enable stack dumps of
events, add the following:
trace_options=stacktrace
This along with the trace_event option, you can get not only
traces of the events but also the stack dumps with them.
Requested-by: Frederic Weisbecker <fweisbec@gmail.com> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Fri, 2 Nov 2012 00:54:21 +0000 (20:54 -0400)]
tracing: Use irq_work for wake ups and remove *_nowake_*() functions
Have the ring buffer commit function use the irq_work infrastructure to
wake up any waiters waiting on the ring buffer for new data. The irq_work
was created for such a purpose, where doing the actual wake up at the
time of adding data is too dangerous, as an event or function trace may
be in the midst of the work queue locks and cause deadlocks. The irq_work
will either delay the action to the next timer interrupt, or trigger an IPI
to itself forcing an interrupt to do the work (in a safe location).
With irq_work, all ring buffer commits can safely do wakeups, removing
the need for the ring buffer commit "nowake" variants, which were used
by events and function tracing. All commits can now safely use the
normal commit, and the "nowake" variants can be removed.
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Thu, 1 Nov 2012 15:51:40 +0000 (11:51 -0400)]
tracing: Remove deprecated tracing_enabled file
The tracing_enabled file was used as a quick way to stop
tracers, and try to bring down overhead for things like
the latency tracers (irqsoff, wakeup, etc). But it didn't
work that well.
The tracing_on file was created as a really fast way to
stop recording into the ftrace ring buffer and can interact
with the kernel. That is a tracing_off() call in the kernel
can disable recording of events, and then from userspace one
could echo 1 into the tracing_on file to continue it. The
tracing_enabled function did too much to allow for this.
The tracing_on has taken over as a way to start and stop tracing
and the tracing_enabled file should not be used. But because of
its existance, it still confuses people. Over a year ago the
following commit was added:
This commit added a WARN_ON() if the tracing_enabled file's variable
was changed. After this was added, only LatencyTop complained, and
they soon fixed their tool as there was no reason that LatencyTop
should touch this file as it was using the perf ring buffers which
this file does not interact with. But since that time no one else
has complained about this WARN_ON(). Thus it is safe to assume that
this file is no longer needed. Time to get rid of it.
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Fri, 11 May 2012 18:25:30 +0000 (14:25 -0400)]
tracing: Make tracing_enabled be equal to tracing_on
The tracing_enabled file has been deprecated as it never was able
to serve its purpose well. The tracing_on file has taken over.
Instead of having code to keep tracing_enabled, have the tracing_enabled
file just set tracing_on, and remove the tracing_enabled variable.
This allows us to remove the tracing_enabled file. The reason that
the remove is in a different change set and not removed here is
in case we find some lonely userspace tool that requires the file
to exist. Then the removal patch will get reverted, but this one
will not.
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Sat, 12 May 2012 00:54:53 +0000 (20:54 -0400)]
tracing: Remove unused function unregister_tracer()
The function register_tracer() is only used by kernel core code,
that never needs to remove the tracer. As trace_events have become
the main way to add new tracing to the kernel, the need to
unregister a tracer has diminished. Remove the unused function
unregister_tracer(). If a need arises where we need it, then we
can always add it back.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Thu, 3 May 2012 18:57:28 +0000 (14:57 -0400)]
tracing: Separate open function from set_event and available_events
The open function used by available_events is the same as set_event even
though it uses different seq functions. This causes a side effect of
writing into available_events clearing all events, even though
available_events is suppose to be read only.
There's no reason to keep a single function for just the open and have
both use different functions for everything else. It is a little
confusing and causes strange behavior. Just have each have their own
function.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
David Sharp [Thu, 11 Oct 2012 23:27:52 +0000 (16:27 -0700)]
tracing: Reset ring buffer when changing trace_clocks
Because the "tsc" clock isn't in nanoseconds, the ring buffer must be
reset when changing clocks so that incomparable timestamps don't end up
in the same trace.
Tested: Confirmed switching clocks resets the trace buffer.
tracing: Cleanup unnecessary function declarations
The functions defined in include/trace/syscalls.h are not used directly
since struct ftrace_event_class was introduced. Remove them from the
header file and rearrange the ftrace_event_class declarations in
trace_syscalls.c.
Steven Rostedt [Thu, 11 Oct 2012 16:14:25 +0000 (12:14 -0400)]
tracing: Cache comms only after an event occurred
Whenever an event is registered, the comm of tasks are saved at
every task switch instead of saving them at every event. But if
an event isn't executed much, the comm cache will be filled up
by tasks that did not record the event and you lose out on the comms
that did.
Here's an example, if you enable the following events:
Note, there's no kvm running on this machine so the first event will
never be triggered, but because it is enabled, the storing of comms
will continue. If we now disable the network event:
Steven Rostedt [Thu, 11 Oct 2012 01:44:34 +0000 (21:44 -0400)]
tracing: Expand ring buffer when trace_printk() is used
Since tracing is not used by 99% of Linux users, even though tracing
may be configured in, it does not make sense to allocate 1.4 Megs
per CPU for the ring buffers if they are not used. Thus, on boot up
the ring buffers are set to a minimal size until something needs the
and they are expanded.
This works well for events and tracers (function, etc), but for the
asynchronous use of trace_printk() which can write to the ring buffer
at any time, does not expand the buffers.
On boot up a check is made to see if any trace_printk() is used to
see if the trace_printk() temp buffer pages should be allocated. This
same code can be used to expand the buffers as well.
Suggested-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The existing 'overrun' counter is incremented when the ring
buffer wraps around, with overflow on (the default). We wanted
a way to count requests lost from the buffer filling up with
overflow off, too. I decided to add a new counter instead
of retro-fitting the existing one because it seems like a
different statistic to count conceptually, and also because
of how the code was structured.
Steven Rostedt [Fri, 5 Oct 2012 16:13:07 +0000 (12:13 -0400)]
tracing: Allow tracers to start at core initcall
There's times during debugging that it is helpful to see traces of early
boot functions. But the tracers are initialized at device_initcall()
which is quite late during the boot process. Setting the kernel command
line parameter ftrace=function will not show anything until the function
tracer is initialized. This prevents being able to trace functions before
device_initcall().
There's no reason that the tracers need to be initialized so late in the
boot process. Move them up to core_initcall() as they still need to come
after early_initcall() which initializes the tracing buffers.
Cc: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Jiri Olsa [Wed, 31 Oct 2012 14:52:47 +0000 (15:52 +0100)]
perf tests: Add framework for automated perf_event_attr tests
The idea is run perf session with kidnapping sys_perf_event_open
function. For each sys_perf_event_open call we store the perf_event_attr
data to the file to be checked later against what we expect.
You can run this by:
$ python ./tests/attr.py -d ./tests/attr/ -p ./perf -v
v2 changes:
- preserve errno value in the hook
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/20121031145247.GB1027@krava.brq.redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Wed, 31 Oct 2012 16:00:55 +0000 (00:00 +0800)]
perf browser: Don't show scripts menu for 'perf top'
As 'perf top' has no data files to run scripts against. Also add a
is_report_browser() helper function to judge whether the running browser
is for 'perf report'.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351699257-5102-1-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf tools: Handle --version string generation on machines without git
If git is installed we'll have a 'perf --version' output of this form:
$ make -j8 -C tools/perf/ O=/home/acme/git/build/perf install
$ perf --version
perf version 3.7.rc3.g3afad6
Now on a machine without git installed:
$ mv /home/acme/bin/git /home/acme/bin/git.OFF
$ make -j8 -C tools/perf/ O=/home/acme/git/build/perf install
$ perf --version
perf version 3.7.0-rc2
That is, no error message due to git not being installed will appear on the
screen and instead the version string in the top level Makefile will be
used.
Requested-by: Ingo Molnar <mingo@kernel.org> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-am6yp6phvxyjmyndxogpunjv@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Ingo Molnar [Tue, 30 Oct 2012 08:54:41 +0000 (09:54 +0100)]
perf tools: Further speed up the perf build
There's another source of overhead in the perf version string generator:
git update-index -q --refresh
... which will iterate the whole checked out tree. This can be pretty
slow on NFS volumes, but takes some time even with local SSD disks and a
fully cached kernel tree:
So remove the .dirty differentiator as well - it adds little information
because locally patched git trees are common, but seldom are the perf
tools modified.
So a lot of version strings are reported as 'dirty' while in fact they
are pristine perf builds. For example 99% of my perf builds are not
patched but the kernel tree is slightly patched, which adds the .dirty
tag.
Eliminating that tag speeds up version generation by another order of
magnitude:
Ingo Molnar [Tue, 30 Oct 2012 08:46:00 +0000 (09:46 +0100)]
perf tools: Speed up the perf build time by simplifying the perf --version string generation
Building perf is pretty slow on trees that have a lot of commits
relative to the nearest Git tag. This slowness manifests itself during
version string generation:
The build can be even slower than that, when one over NFS volumes.
The reason for the slowness is that util/PERF-VERSION-GEN uses "git
describe" to generate the string, which has to count the "number of
commits distance" from the nearest tag - the ".1458." count in the
output above. For that Git had to extract and decompress 1458 Git
objects, which takes time and bandwidth.
But this "number of commits" value is mostly irrelevant in practice. We
either want to know an approximate tag name, or we want to know the
precise sha1.
So this patch simplifies the version string to:
PERF_VERSION = 3.7.rc3.g5399b3b.dirty
which speeds up the version string generation script by an order of
magnitude:
Namhyung Kim [Mon, 29 Oct 2012 15:06:46 +0000 (00:06 +0900)]
perf tools: Warn about missing libelf
When perf detects no libelf during the build, it'll use internal mini
elf parser instead of libelf. But as it only supports minimal
functionalities, it also disables support to 'probe' builtin command.
Reported-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/n/tip-8ww8zc4hhpxabfskxs3u5ede@git.kernel.org
[ committer note: The package needed is elfutils-libelf-devel, not elfutils-devel ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Peter Huewe [Mon, 29 Oct 2012 20:48:17 +0000 (21:48 +0100)]
perf/x86: Fix sparse warnings
FYI, there are new sparse warnings:
arch/x86/kernel/cpu/perf_event.c:1356:18: sparse: symbol 'events_attr' was not declared. Should it be static?
This patch makes it static and also adds the static keyword to
fix arch/x86/kernel/cpu/perf_event.c:1344:9: warning: symbol
'events_sysfs_show' was not declared.
Signed-off-by: Peter Huewe <peterhuewe@gmx.de> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Yuanhan Liu <yuanhan.liu@linux.intel.com> Cc: fengguang.wu@intel.com Cc: Jiri Olsa <jolsa@redhat.com> Link: http://lkml.kernel.org/n/tip-lerdpXlnruh0yvWs2owwuizl@git.kernel.org Signed-off-by: Ingo Molnar <mingo@kernel.org>
Ingo Molnar [Tue, 30 Oct 2012 07:35:33 +0000 (08:35 +0100)]
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Pull perf/core improvements, fixes and code move from Arnaldo Carvalho de Melo:
* Initialize 'page_size' variable in the python binding, this was sent
for perf/urgent by mistake, then when merging Ingo removed it, fixing
the problem for perf/urgent, but when perf/urgent was merged with
perf/core, where that initialization is needed, made the python
binding mmap call to fail, fix it by initializing page_size again.
* Add a browser for 'perf script' and make it available from the report
and annotate browsers. It does filtering to find the scripts that
handle events found in the perf.data file used. From Feng Tang
* Move some functions from symbol.c to more appropriate files, creating
dso.[ch] in the process, no code changes. From Jiri Olsa
* Fix mmap error output message for when perf_mmap fails and returns
!-EPERM, where the default for mmap_pages, INT_MAX, was causing a
!power of 2 error message, fix from Jiri Olsa.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org>
'perf tools: Have the page size value available for all tools'
Broke the python binding because the global variable 'page_size' is
initialized on the main() routine, that is not called when using
just the python binding, causing evlist.mmap() to fail because it
expects that variable to be initialized to the system's page size.
Fix it by initializing it on the binding init routine.
Reported-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Jiri Olsa <jolsa@redhat.com> Cc: Mike Galbraith <efault@gmx.de> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Stephane Eranian <eranian@google.com> Link: http://lkml.kernel.org/n/tip-vrvp3azmbfzexnpmkhmvtzzc@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Olsa [Sat, 20 Oct 2012 14:33:19 +0000 (16:33 +0200)]
perf record: Fix mmap error output condition
The mmap_pages default value is not power of 2 (UINT_MAX).
Together with perf_evlist__mmap function returning error value different
from EPERM, we get misleading error message: "--mmap_pages/-m value must
be a power of two."
Fixing this by adding extra check for UINT_MAX value for this error
condition.
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1350743599-4805-12-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:07 +0000 (11:56 +0800)]
perf header: Add is_perf_magic() func
With this function, other modules can basically check whether a file is
a legal perf data file by checking its first 8 bytes against all
possible perf magic numbers.
Change the function name from check_perf_magic to more meaningful
is_perf_magic as suggested by acme.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-7-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:06 +0000 (11:56 +0800)]
perf hists browser: Integrate script browser into main hists browser
Integrate the script browser into "perf report" framework, users can use
function key 'r' or the drop down menu to list all perf scripts and
select one of them, just like they did for the annotation.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-6-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:05 +0000 (11:56 +0800)]
perf annotate browser: Integrate script browser into annotation browser
Integrate the script browser into annotation, users can press function
key 'r' to list all perf scripts and select one of them to run that
script, the output will be shown in a separate browser.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-5-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:04 +0000 (11:56 +0800)]
perf scripts browser: Add a browser for perf script
Create a script browser, so that user can check all the available
scripts for current perf data file and run them inside the main perf
report or annotation browsers, for all perf samples or for samples
belong to one thread/symbol.
Please be noted: current script browser is only for report use, and
doesn't cover the record phase, IOW it must run against one existing
perf data file.
The work flow is, users can use function key to list all the available
scripts for current perf data file in system and chose one, which will
be executed with popen("perf script -s xxx.xx",) and all the output
lines are put into one ui browser, pressing 'q' or left arrow key will
make it return to previous browser.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-4-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:03 +0000 (11:56 +0800)]
perf script: Add more filter to find_scripts()
As suggested by Arnaldo, many scripts have their own usages and need
capture specific events or tracepoints, so only those scripts whose
target events match the events in current perf data file should be
listed in the script browser menu.
This patch will add the event match checking, by opening "xxx-record"
script to cherry pick out all events name and comparing them with
those inside the perf data file.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-3-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Feng Tang [Tue, 30 Oct 2012 03:56:02 +0000 (11:56 +0800)]
perf tools: Add a global variable "const char *input_name"
Currently many perf commands annotate/evlist/report/script/lock etc all
support "-i" option to chose a specific perf data, and all of them
create a local "input_name" to save the file name for that perf data.
Since most of these commands need it, we can add a global variable for
it, also it can some other benefits:
1. When calling script browser inside hists/annotation browser, it needs
to know the perf data file name to run that script.
2. For further feature like runtime switching to another perf data file,
this variable can also help.
Signed-off-by: Feng Tang <feng.tang@intel.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Link: http://lkml.kernel.org/r/1351569369-26732-2-git-send-email-feng.tang@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Olsa [Sat, 27 Oct 2012 21:18:32 +0000 (23:18 +0200)]
perf tools: Move dso_* related functions into dso object
Moving dso_* related functions into dso object.
Keeping symbol loading related functions still in the symbol object as
it seems more convenient.
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Reviewed-by: Namhyung Kim <namhyung@kernel.org> Tested-by: Namhyung Kim <namhyung@kernel.org> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@elte.hu> Cc: Namhyung Kim <namhyung@kernel.org> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1351372712-21104-6-git-send-email-jolsa@redhat.com
[ committer note: Use "symbol.h" instead of <symbol.h> to make it build with O= ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
projects / karo-tx-linux.git / log