Jiri Olsa [Mon, 21 May 2012 07:12:50 +0000 (09:12 +0200)]
perf tools: Add support for displaying event parser debug info
Adding PARSER_DEBUG Makefile variable to enable building event scanner/
parser with debug enabled. This results in verbose output right out of
the scanner/parser.
It's useful for debuging the event parser. Keeping this only for event
parser so far.
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/1337584373-2741-3-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Ingo Molnar [Mon, 21 May 2012 07:17:31 +0000 (09:17 +0200)]
Merge branch 'perf/core' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Fixes for perf/core:
- Rename some perf_target methods to avoid double negation, from Namhyung Kim.
- Revert change to use per task events with inheritance, from Namhyung Kim.
- Events should start disabled till children starts running, from David Ahern.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org>
David Ahern [Mon, 14 May 2012 04:01:28 +0000 (22:01 -0600)]
perf evsel: Create events initially disabled -- again
764e16a changed perf-record to create events disabled by default and
enable them once perf initializations are done. This setting was dropped
by 0f82ebc. Now perf events are once again generated during perf's
initialization phase (e.g., generating maps).
As an example, perf opens a lot of files at startup. Unpatched:
perf record -e syscalls:sys_enter_open -ga -fo /tmp/perf.data -- sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.087 MB /tmp/perf.data (~3798 samples) ]
Using perf-script to look at the samples shows the perf command generating
563 of the 566 total events.
Patched:
perf record -e syscalls:sys_enter_open -ga -fo /tmp/perf.data -- sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.028 MB /tmp/perf.data (~1206 samples) ]
Using perf-script to look at the samples does not show perf command.
Merge remote-tracking branch 'tip/perf/urgent' into perf/core
Merge reason: We are going to queue up a dependent patch:
"perf tools: Move parse event automated tests to separated object"
That depends on:
commit e7c72d8
perf tools: Add 'G' and 'H' modifiers to event parsing
Conflicts:
tools/perf/builtin-stat.c
Conflicted with the recent 'perf_target' patches when checking the
result of perf_evsel open routines to see if a retry is needed to cope
with older kernels where the exclude guest/host perf_event_attr bits
were not used.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jiri Olsa [Wed, 16 May 2012 06:59:08 +0000 (08:59 +0200)]
perf hists: Fix callchain ip printf format
The callchain address is stored as u64. Current code uses following
format string to display callchain address:
"%p\n", (void *)(long)chain->ip
This way we lose upper 32 bits if we report 64 bit addresses in 32 bit
environment. Fixing this to always display whole 64 bits.
Note, running following to test perf endianity handling:
test 1)
- origin system:
# perf record -a -- sleep 10 (any perf record will do)
# perf report > report.origin
# perf archive perf.data
- copy the perf.data, report.origin and perf.data.tar.bz2
to a target system and run:
# tar xjvf perf.data.tar.bz2 -C ~/.debug
# perf report > report.target
# diff -u report.origin report.target
- the diff should produce no output
(besides some white space stuff and possibly different
date/TZ output)
test 2)
- origin system:
# perf record -ag -fo /tmp/perf.data -- sleep 1
- mount origin system root to the target system on /mnt/origin
- target system:
# perf script --symfs /mnt/origin -I -i /mnt/origin/tmp/perf.data \
--kallsyms /mnt/origin/proc/kallsyms
- complete perf.data header is displayed
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com> Cc: David Ahern <dsahern@gmail.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/1337151548-2396-8-git-send-email-jolsa@redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Wed, 16 May 2012 09:45:49 +0000 (18:45 +0900)]
perf target: Add uses_mmap field
If perf doesn't mmap on event (like perf stat), it should not create
per-task-per-cpu events. So just use a dummy cpu map to create a
per-task event for this case.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1337161549-9870-3-git-send-email-namhyung.kim@lge.com
[ committer note: renamed .need_mmap to .uses_mmap ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Steven Rostedt [Mon, 30 Apr 2012 13:17:03 +0000 (09:17 -0400)]
ftrace: Remove selecting FRAME_POINTER with FUNCTION_TRACER
The function tracer will enable the -pg option with gcc, which requires
that frame pointers. When FRAME_POINTER is defined in the kernel config
it adds the gcc option -fno-omit-frame-pointer which causes some problems
on some architectures. For those architectures, the FRAME_POINTER select
was not set.
When FUNCTION_TRACER was selected on these architectures that can not have
-fno-omit-frame-pointer, the -pg option is still set. But when
FRAME_POINTER is not selected, the kernel config would add the gcc option
-fomit-frame-pointer. Adding this option is incompatible with -pg
even on archs that do not need frame pointers with -pg.
The answer to this was to just not add either -fno-omit-frame-pointer
or -fomit-frame-pointer on these archs that want function tracing
but do not set FRAME_POINTER.
As it turns out, for archs that require frame pointers for function
tracing, the same can be used. If gcc requires frame pointers with
-pg, it will simply add it. The best thing to do is not select FRAME_POINTER
when function tracing is selected, and let gcc add it if needed.
Only add the -fno-omit-frame-pointer when something else selects
FRAME_POINTER, but do not add -fomit-frame-pointer if function tracing
is selected.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Fri, 27 Apr 2012 13:13:18 +0000 (09:13 -0400)]
ftrace/x86: Have x86 ftrace use the ftrace_modify_all_code()
To remove duplicate code, have the ftrace arch_ftrace_update_code()
use the generic ftrace_modify_all_code(). This requires that the
default ftrace_replace_code() becomes a weak function so that an
arch may override it.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Thu, 26 Apr 2012 18:59:43 +0000 (14:59 -0400)]
ftrace: Make ftrace_modify_all_code() global for archs to use
Rename __ftrace_modify_code() to ftrace_modify_all_code() and make
it global for all archs to use. This will remove the duplication
of code, as archs that can modify code without stop_machine()
can use it directly outside of the stop_machine() call.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Wed, 25 Apr 2012 18:39:54 +0000 (14:39 -0400)]
ftrace: Return record ip addr for ftrace_location()
ftrace_location() is passed an addr, and returns 1 if the addr is
on a ftrace nop (or caller to ftrace_caller), and 0 otherwise.
To let kprobes know if it should move a breakpoint or not, it
must return the actual addr that is the start of the ftrace nop.
This way a kprobe placed on the location of a ftrace nop, can
instead be placed on the instruction after the nop. Even if the
probe addr is on the second or later byte of the nop, it can
simply be moved forward.
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Wed, 25 Apr 2012 17:48:13 +0000 (13:48 -0400)]
ftrace: Consolidate ftrace_location() and ftrace_text_reserved()
Both ftrace_location() and ftrace_text_reserved() do basically the same thing.
They search to see if an address is in the ftace table (contains an address
that may change from nop to call ftrace_caller). The difference is
that ftrace_location() searches a single address, but ftrace_text_reserved()
searches a range.
This also makes the ftrace_text_reserved() faster as it now uses a bsearch()
instead of linearly searching all the addresses within a page.
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Wed, 25 Apr 2012 14:14:43 +0000 (10:14 -0400)]
ftrace: Speed up search by skipping pages by address
As all records in a page of the ftrace table are sorted, we can
speed up the search algorithm by checking if the address to look for
falls in between the first and last record ip on the page.
This speeds up both the ftrace_location() and ftrace_text_reserved()
algorithms, as it can skip full pages when the search address is
not in them.
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Wed, 25 Apr 2012 03:45:26 +0000 (23:45 -0400)]
ftrace: Remove extra helper functions
The ftrace_record_ip() and ftrace_alloc_dyn_node() were from the
time of the ftrace daemon. Although they were still used, they
still make things a bit more complex than necessary.
Move the code into the one function that uses it, and remove the
helper functions.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Namhyung Kim [Mon, 23 Apr 2012 01:11:57 +0000 (10:11 +0900)]
tracing: Check return value of tracing_dentry_percpu()
If tracing_dentry_percpu() failed, tracing_init_debugfs_percpu()
will try to create each cpu directories on debugfs' root directory
as d_percpu is NULL.
Steven Rostedt [Wed, 16 May 2012 23:46:32 +0000 (19:46 -0400)]
ring-buffer: Reset head page before running self test
When the ring buffer does its consistency test on itself, it
removes the head page, runs the tests, and then adds it back
to what the "head_page" pointer was. But because the head_page
pointer may lack behind the real head page (held by the link
list pointer). The reset may be incorrect.
Instead, if the head_page exists (it does not on first allocation)
reset it back to the real head page before running the consistency
tests. Then it will be put back to its original location after
the tests are complete.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Steven Rostedt [Mon, 14 May 2012 21:02:33 +0000 (17:02 -0400)]
ring-buffer: Add integrity check at end of iter read
There use to be ring buffer integrity checks after updating the
size of the ring buffer. But now that the ring buffer can modify
the size while the system is running, the integrity checks were
removed, as they require the ring buffer to be disabed to perform
the check.
Move the integrity check to the reading of the ring buffer via the
iterator reads (the "trace" file). As reading via an iterator requires
disabling the ring buffer, it is a perfect place to have it.
If the ring buffer happens to be disabled when updating the size,
we still perform the integrity check.
Cc: Vaibhav Nagarnaik <vnagarnaik@google.com> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
ring-buffer: Make addition of pages in ring buffer atomic
This patch adds the capability to add new pages to a ring buffer
atomically while write operations are going on. This makes it possible
to expand the ring buffer size without reinitializing the ring buffer.
The new pages are attached between the head page and its previous page.
ring-buffer: Make removal of ring buffer pages atomic
This patch adds the capability to remove pages from a ring buffer
without destroying any existing data in it.
This is done by removing the pages after the tail page. This makes sure
that first all the empty pages in the ring buffer are removed. If the
head page is one in the list of pages to be removed, then the page after
the removed ones is made the head page. This removes the oldest data
from the ring buffer and keeps the latest data around to be read.
To do this in a non-racey manner, tracing is stopped for a very short
time while the pages to be removed are identified and unlinked from the
ring buffer. The pages are freed after the tracing is restarted to
minimize the time needed to stop tracing.
The context in which the pages from the per-cpu ring buffer are removed
runs on the respective CPU. This minimizes the events not traced to only
NMI trace contexts.
Steven Rostedt [Sat, 12 May 2012 03:28:49 +0000 (23:28 -0400)]
tracing: Clean up tracing_mark_write()
On gcc 4.5 the function tracing_mark_write() would give a warning
of page2 being uninitialized. This is due to a bug in gcc because
the logic prevents page2 from being used uninitialized, and
gcc 4.6+ does not complain (correctly).
Instead of adding a "unitialized" around page2, which could show
a bug later on, I combined page1 and page2 into an array map_pages[].
This binds the two and the two are modified according to nr_pages
(what gcc 4.5 seems to ignore). This no longer gives a warning with
gcc 4.5 nor with gcc 4.6.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Namhyung Kim [Wed, 16 May 2012 09:45:48 +0000 (18:45 +0900)]
Revert 'perf evlist: Fix creation of cpu map'
The commit 55261f46702c ("perf evlist: Fix creation of cpu map") changed
to create a per-task event when no cpu target is specified. However it
caused a problem since perf-task do not allow event inheritance due to
scalability issues so that the result will contain samples only from
parent, not from its children.
So we should use perf-task-per-cpu events anyway to get the right
result. Revert it.
Reported-by: Linus Torvalds <torvalds@linux-foundation.org> Analysed-by: Ingo Molnar <mingo@kernel.org> Acked-and-tested-by: Peter Zijlstra <a.p.zijlstra@chello.nl> Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1337161549-9870-2-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Wed, 16 May 2012 09:45:47 +0000 (18:45 +0900)]
perf target: Rename functions to avoid double negation
Rename perf_target__no_{cpu,task} to perf_target__has_{cpu,task} because
it's more intuitive and easy to parse (for human beings) when used with
negation.
The names are came out from David Ahern. It is intended to be a
mechanical substitution without any functional change.
The perf_target__none remains unchanged since I couldn't find a right
name and it is hardly used with negation.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Suggested-by: David Ahern <dsahern@gmail.com> Suggested-by: Ingo Molnar <mingo@kernel.org> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1337161549-9870-1-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Just press 'J' and see how many places jump to jump targets.
The hottest jump target appears in red, targets with more than one
source have a different color than single source jump targets.
Suggested-by: Arjan van de Ven <arjan@infradead.org> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.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-7452y0dmc02a20ooins7rn79@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf annotate browser: Count the numbers of jump sources to a target
Instead of simply marking an offset as a jump target. So that we can
implement a new feature: showing "jumpy" targets, I.e. addresses that
lots of places jump to.
Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.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-vc7b0u5yxgrubig0q61ayhxf@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Ingo Molnar [Fri, 11 May 2012 06:13:55 +0000 (08:13 +0200)]
Merge tag 'perf-core-for-mingo' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Fixes and improvements for perf/core:
- perf_target: abstraction for --uid, --pid, --tid, --cpu, --all-cpus handling,
eliminating code duplicated in the tools, having constraints that apply to
all of them, from Namhyung Kim
- Fixes for handling fallback to cpu-clock on PPC, from David Ahern
- Fix for processing events with unknown size, from Jiri Olsa
- Compilation fix on 32-bit, from Jiri Olsa
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org>
[console]
event trace: Could not enable event function
This is because when enabling all events via the debugfs system
it ignores events that do not have a ->reg() function assigned.
This was to skip over the ftrace internal events (as they are
not TRACE_EVENTs). But as the ftrace function event now has
a ->reg() function attached to it for use with perf, it is no
longer ignored.
Worse yet, this ->reg() function is being called when it should
not be. It returns an error and causes the above warning to
be printed.
By adding a new event_call flag (TRACE_EVENT_FL_IGNORE_ENABLE)
and have all ftrace internel event structures have it set,
setting the events/enable will no longe try to incorrectly enable
the function event and does not warn.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
David Ahern [Tue, 8 May 2012 15:29:16 +0000 (09:29 -0600)]
perf stat: handle ENXIO error for perf_event_open
perf stat on PPC currently fails to run:
$ perf stat -- sleep 1
Error: open_counter returned with 6 (No such device or address). /bin/dmesg may provide additional information.
Fatal: Not all events could be opened.
The problem is that until 2.6.37 (behavior changed with commit b0a873e)
perf on PPC returns ENXIO when hw_perf_event_init() fails. With this
patch we get the expected behavior:
$ perf stat -v -- sleep 1
cycles event is not supported by the kernel.
stalled-cycles-frontend event is not supported by the kernel.
stalled-cycles-backend event is not supported by the kernel.
instructions event is not supported by the kernel.
branches event is not supported by the kernel.
branch-misses event is not supported by the kernel.
David Ahern [Tue, 8 May 2012 16:49:49 +0000 (10:49 -0600)]
perf top: Update event name when falling back to cpu-clock
The 'perf top' command falls back to cpu-clock if the H/W cycles event
is not supported, but the event name is not updated leading to a
misleading header:
David Ahern [Tue, 8 May 2012 15:29:16 +0000 (09:29 -0600)]
perf stat: handle ENXIO error for perf_event_open
perf stat on PPC currently fails to run:
$ perf stat -- sleep 1
Error: open_counter returned with 6 (No such device or address). /bin/dmesg may provide additional information.
Fatal: Not all events could be opened.
The problem is that until 2.6.37 (behavior changed with commit b0a873e)
perf on PPC returns ENXIO when hw_perf_event_init() fails. With this
patch we get the expected behavior:
$ perf stat -v -- sleep 1
cycles event is not supported by the kernel.
stalled-cycles-frontend event is not supported by the kernel.
stalled-cycles-backend event is not supported by the kernel.
instructions event is not supported by the kernel.
branches event is not supported by the kernel.
branch-misses event is not supported by the kernel.
David Ahern [Tue, 8 May 2012 15:28:57 +0000 (09:28 -0600)]
perf record: Fix fallback to cpu-clock on ppc
perf-record on PPC is not falling back to cpu-clock:
$ perf record -ag -fo /tmp/perf.data -- sleep 1
Error: sys_perf_event_open() syscall returned with 6 (No such device or address). /bin/dmesg may provide additional information.
Fatal: No CONFIG_PERF_EVENTS=y kernel support configured?
The problem is that until 2.6.37 (behavior changed with commit b0a873e)
perf on PPC returns ENXIO when hw_perf_event_init() fails. With this
patch we get the expected behavior:
$ perf record -ag -fo /tmp/perf.data -v -- sleep 1
Old kernel, cannot exclude guest or host samples.
The cycles event is not supported, trying to fall back to cpu-clock-ticks
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.151 MB /tmp/perf.data (~6592 samples) ]
Peter Zijlstra [Tue, 8 May 2012 16:56:04 +0000 (18:56 +0200)]
sched, perf: Use a single callback into the scheduler
We can easily use a single callback for both sched-in and sched-out. This
reduces the code footprint in the scheduler path as well as removes
the PMU black spot otherwise present between the out and in callback.
Robert Richter [Mon, 2 Apr 2012 18:19:18 +0000 (20:19 +0200)]
perf/x86-ibs: Fix usage of IBS op current count
The value of IbsOpCurCnt rolls over when it reaches IbsOpMaxCnt. Thus,
it is reset to zero by hardware. To get the correct count we need to
add the max count to it in case we received an ibs sample (valid bit
set).
Robert Richter [Mon, 2 Apr 2012 18:19:17 +0000 (20:19 +0200)]
perf/x86-ibs: Catch spurious interrupts after stopping IBS
After disabling IBS there could be still incomming NMIs with samples
that even have the valid bit cleared. Mark all this NMIs as handled to
avoid spurious interrupt messages.
Robert Richter [Mon, 2 Apr 2012 18:19:16 +0000 (20:19 +0200)]
perf/x86-ibs: Implement workaround for IBS erratum #420
When disabling ibs there might be the case where hardware continuously
generates interrupts. This is described in erratum #420 (Instruction-
Based Sampling Engine May Generate Interrupt that Cannot Be Cleared).
To avoid this we must clear the counter mask first and then clear the
enable bit. This patch implements this.
See Revision Guide for AMD Family 10h Processors, Publication #41322.
Note: We now keep track of the last read ibs config value which is
then used to disable ibs. To update the config value we pass now a
pointer to the functions reading it.
Robert Richter [Mon, 2 Apr 2012 18:19:15 +0000 (20:19 +0200)]
perf/x86-ibs: Extend hw period that triggers overflow
If the last hw period is too short we might hit the irq handler which
biases the results. Thus try to have a max last period that triggers
the sw overflow.
Robert Richter [Mon, 2 Apr 2012 18:19:14 +0000 (20:19 +0200)]
perf/x86-ibs: Trigger overflow if remaining period is too small
There are cases where the remaining period is smaller than the minimal
possible value. In this case the counter is restarted with the minimal
period. This is of no use as the interrupt handler will trigger
immediately again and most likely hits itself. This biases the
results.
So, if the remaining period is within the min range, we better do not
restart the counter and instead trigger the overflow.
Robert Richter [Mon, 12 Mar 2012 11:54:32 +0000 (12:54 +0100)]
perf/x86-ibs: Precise event sampling with IBS for AMD CPUs
This patch adds support for precise event sampling with IBS. There are
two counting modes to count either cycles or micro-ops. If the
corresponding performance counter events (hw events) are setup with
the precise flag set, the request is redirected to the ibs pmu:
perf record -a -e cpu-cycles:p ... # use ibs op counting cycle count
perf record -a -e r076:p ... # same as -e cpu-cycles:p
perf record -a -e r0C1:p ... # use ibs op counting micro-ops
Each ibs sample contains a linear address that points to the
instruction that was causing the sample to trigger. With ibs we have
skid 0. Thus, ibs supports precise levels 1 and 2. Samples are marked
with the PERF_EFLAGS_EXACT flag set. In rare cases the rip is invalid
when IBS was not able to record the rip correctly. Then the
PERF_EFLAGS_EXACT flag is cleared and the rip is taken from pt_regs.
V2:
* don't drop samples in precise level 2 if rip is invalid, instead
support the PERF_EFLAGS_EXACT flag
Robert Richter [Mon, 2 Apr 2012 18:19:11 +0000 (20:19 +0200)]
perf/x86-ibs: Take instruction pointer from ibs sample
Each IBS sample contains a linear address of the instruction that
caused the sample to trigger. This address is more precise than the
rip that was taken from the interrupt handler's stack. Update the rip
with that address. We use this in the next patch to implement
precise-event sampling on AMD systems using IBS.
Robert Richter [Mon, 2 Apr 2012 18:19:10 +0000 (20:19 +0200)]
perf/x86-ibs: Fix frequency profiling
Fixing profiling at a fixed frequency, in this case the freq value and
sample period was setup incorrectly. Since sampling periods are
adjusted we also allow periods that have lower 4 bits set.
Another fix is the setup of the hw counter: If we modify
hwc->sample_period, we also need to update hwc->last_period and
hwc->period_left.
Robert Richter [Mon, 2 Apr 2012 18:19:08 +0000 (20:19 +0200)]
perf: Pass last sampling period to perf_sample_data_init()
We always need to pass the last sample period to
perf_sample_data_init(), otherwise the event distribution will be
wrong. Thus, modifiyng the function interface with the required period
as argument. So basically a pattern like this:
Robert Richter [Mon, 2 Apr 2012 18:19:07 +0000 (20:19 +0200)]
perf/x86-ibs: Fix update of period
The last sw period was not correctly updated on overflow and thus led
to wrong distribution of events. We always need to properly initialize
data.period in struct perf_sample_data.
Steven Rostedt [Wed, 9 May 2012 00:57:53 +0000 (20:57 -0400)]
tracing: Remove ftrace_disable/enable_cpu()
The ftrace_disable_cpu() and ftrace_enable_cpu() functions were
needed back before the ring buffer was lockless. Now that the
ring buffer is lockless (and has been for some time), these functions
serve no purpose, and unnecessarily slow down operations of the tracer.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Jiri Olsa [Wed, 25 Apr 2012 08:23:39 +0000 (10:23 +0200)]
tracing: Use seq_*_private interface for some seq files
It's appropriate to use __seq_open_private interface to open
some of trace seq files, because it covers all steps we are
duplicating in tracing code - zallocating the iterator and
setting it as seq_file's private.
Using this for following files:
trace
available_filter_functions
enabled_functions
Link: http://lkml.kernel.org/r/1335342219-2782-5-git-send-email-jolsa@redhat.com Signed-off-by: Jiri Olsa <jolsa@redhat.com>
[
Fixed warnings for:
kernel/trace/trace.c: In function '__tracing_open':
kernel/trace/trace.c:2418:11: warning: unused variable 'ret' [-Wunused-variable]
kernel/trace/trace.c:2417:19: warning: unused variable 'm' [-Wunused-variable]
]
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Ingo Molnar [Tue, 8 May 2012 14:55:14 +0000 (16:55 +0200)]
Merge branch 'perf/annotate' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/core
Perf annotate browser improvements:
- Get back the line separating the overheads from the disassembly, requested by
Peter Zijlstra, Linus agreed now that it is a solid line and more column real
state was harvested. Also it has the jump->arrow lines separated from it by
the address/jump target column.
- Don't change asm line color when toggling source code view. Requested by
Peter Zijlstra.
Current snapshot:
avtab_search_node
│ push %rbp
│ mov %rsp,%rbp
│ → callq mcount
│ movzwl 0x6(%rsi),%edx
│ and $0x7fff,%dx
│ test %rdi,%rdi
│ ↓ jne 20
0.42 │17:┌─→xor %eax,%eax
│19:│ leaveq
0.42 │ │← retq
│ │ nopl 0x0(%rax,%rax,1)
│20:│ mov (%rdi),%rax
0.08 │ │ test %rax,%rax
│ └──je 17
│ movzwl (%rsi),%ecx
│ movzwl 0x2(%rsi),%r9d
│ movzwl 0x4(%rsi),%r8d
│ movzwl %cx,%esi
│ movzwl %r9w,%r10d
│ shl $0x9,%esi
│ lea (%rsi,%r10,4),%esi
│ lea (%r8,%rsi,1),%esi
│ and 0x10(%rdi),%si
│ movzwl %si,%esi
│ mov (%rax,%rsi,8),%rax
1.01 │ test %rax,%rax
│ ↑ je 19
│ nopw 0x0(%rax,%rax,1)
3.19 │60: cmp %cx,(%rax)
│ ↓ jne 7e
0.08 │ cmp %r9w,0x2(%rax)
│ ↓ jne 7e
│ cmp %r8w,0x4(%rax)
│ ↓ jne 79
│ test %dx,0x6(%rax)
│ ↑ jne 19
│79: cmp %r8w,0x4(%rax)
83.45 │7e: ↑ ja 17
3.36 │ mov 0x10(%rax),%rax
7.98 │ test %rax,%rax
│ ↑ jne 60
│ leaveq
│ ← retq
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: Ingo Molnar <mingo@kernel.org>
Just suppress the nop operands, future infrastructure that will record
the instruction lenght (and its contents) in struct ins will allow
rendering them as nopN, i.e. nop5 for a 5-byte nop.
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.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-qddbeglfzqdlal8vj2yaj67y@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Mon, 7 May 2012 05:09:03 +0000 (14:09 +0900)]
perf target: Consolidate target task/cpu checking
There are places that check whether target task/cpu is given or not and
some of them didn't check newly introduced uid or cpu list. Add and use
three of helper functions to treat them properly.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1336367344-28071-7-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Mon, 7 May 2012 05:09:02 +0000 (14:09 +0900)]
perf tools: Introduce perf_target__strerror()
The perf_target__strerror() sets @buf to a string that describes the
(perf_target-specific) error condition that is passed via @errnum.
This is similar to strerror_r() and does same thing if @errnum has a
standard errno value.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Suggested-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1336367344-28071-6-git-send-email-namhyung.kim@lge.com
[ committer note: No need to use PERF_ERRNO_TARGET__SUCCESS, use shorter idiom ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Mon, 7 May 2012 05:09:00 +0000 (14:09 +0900)]
perf target: Introduce perf_target_errno
The perf_target_errno enumerations are used to indicate specific error
cases on perf target operations. It'd help libperf being a more generic
library.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Suggested-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1336367344-28071-4-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Check if neither of --pid, --tid or --uid was specified and if so, set
system_wide appropriately.
Namhyung's patch would make using any of the above target specifiers
emit a warning in perf_target__validate, since it would see
target.system_wide set and one of the others as well.
So set system_wide after validation.
Suggested-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Frederic Weisbecker <fweisbec@gmail.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-6e4zrji1uw0rinfyoitl0wi4@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf: Turn off compiler warnings for flex and bison generated files
We don't know what types of warnings different versions of flex
and bison combined with different versions of gcc is going to
generate, so just punt and don't warn about anything.
This fixes the build of perf for me on an openSUSE 12.1 system.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Cc: Paul Mackerras <paulus@samba.org> Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> Link: http://lkml.kernel.org/r/20120504183254.GA11154@kroah.com Signed-off-by: Ingo Molnar <mingo@kernel.org>
Jiri Olsa [Mon, 16 Apr 2012 18:42:51 +0000 (20:42 +0200)]
perf session: Fail on processing event with unknown size
Currently if we cannot decide the size of the event, we guess next
event possition by:
"... check alignment, and increment a single u64 in the hope
to catch on again 'soon'"
This usually ends up with segfault or endless loop. It's better
to admit the failure right away, then pretend nothing happened.
It makes the life easier ;)
Signed-off-by: Jiri Olsa <jolsa@redhat.com> Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl> 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/20120416184251.GA11503@m.brq.redhat.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Thu, 26 Apr 2012 05:15:22 +0000 (14:15 +0900)]
perf target: Split out perf_target handling code
For further work on perf_target, it'd be better off splitting the code
into a separate file.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1335417327-11796-9-git-send-email-namhyung.kim@lge.com
[ committer note: Fixed perl build by using stdbool and types.h in target.h ] Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The perf_target__validate function is used to check given PID/TID/UID/CPU
target options and warn if some combination is impossible. Also this can
make some arguments of parse_target_uid() function useless as it is checked
before the call via our new helper.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1335417327-11796-5-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Namhyung Kim [Thu, 26 Apr 2012 05:15:15 +0000 (14:15 +0900)]
perf tools: Introduce struct perf_target
The perf_target struct will be used for taking care of cpu/thread maps
based on user's input. Since it is used on various subcommands it'd
better factoring it out.
Thanks to Arnaldo for suggesting the better name.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com> Reviewed-by: David Ahern <dsahern@gmail.com> Cc: David Ahern <dsahern@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Namhyung Kim <namhyung@gmail.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl> Link: http://lkml.kernel.org/r/1335417327-11796-2-git-send-email-namhyung.kim@lge.com Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
projects / linux-beck.git / log