I get such truncated annotation results in 'perf top':
: Disassembly of section .text: ▒
: ▒
: ffffffff810966a8 <nr_iowait_cpu>: ▒
4.94 : ffffffff810966a8: movslq %edi,%rdi ▒
3.70 : ffffffff810966ab: mov $0x13700,%rax ▒
0.00 : ffffffff810966b2: add -0x7e32cb00(,%rdi,8),%rax ▒
8.64 : ffffffff810966ba: mov 0x7e0(%rax),%eax ▒
82.72 : ffffffff810966c0: cltq ▒
Note the missing 'retq' which is there in the original function:
ffffffff810966a8 <nr_iowait_cpu>:
ffffffff810966a8: 48 63 ff movslq %edi,%rdi
ffffffff810966ab: 48 c7 c0 00 37 01 00 mov $0x13700,%rax
ffffffff810966b2: 48 03 04 fd 00 35 cd add -0x7e32cb00(,%rdi,8),%rax
ffffffff810966b9: 81
ffffffff810966ba: 8b 80 e0 07 00 00 mov 0x7e0(%rax),%eax
ffffffff810966c0: 48 98 cltq
ffffffff810966c2: c3 retq
ffffffff810966c3 <this_cpu_load>:
I'm using a fairly recent binutils:
GNU objdump version 2.21.51.0.6-2.fc16 20110118
AFAICS the bug is simply that sym->end points to the last byte
of the symbol in question - while objdump's --stop-address
expects the last byte plus 1 to disassemble the full range.
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111223130804.GA24305@elte.hu
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf does not properly handle monitoring of processes with named threads.
For example:
$ ps -C myapp -L
PID LWP TTY TIME CMD
25118 25118 ? 00:00:00 myapp
25118 25119 ? 00:00:00 myapp:worker
perf record -e cs -c 1 -fo /tmp/perf.data -p 25118 -- sleep 10
perf report --stdio -i /tmp/perf.data
100.00% myapp:worker [kernel.kallsyms] [k] perf_event_task_sched_out
The process name is set to the name of the last thread it finds for the
process.
The Problem:
perf-top and perf-record both create a thread_map of threads to be
monitored. That map is used in perf_event__synthesize_thread_map which
loops over the entries in thread_map and calls __event__synthesize_thread
to generate COMM and MMAP events.
__event__synthesize_thread calls perf_event__synthesize_comm which opens
/proc/pid/status, reads the name of the task and its thread group id.
That's all fine. The problem is that it then reads /proc/pid/task and
generates COMM events for each task it finds - but using the name found
in /proc/pid/status where pid is the thread of interest.
The end result (looping over thread_map + synthesizing comm events for
each thread each time) means the name of the last thread processed sets
the name for all threads in the process - which is not good for
multithreaded processes with named threads.
The Fix:
perf_event__synthesize_comm has an input argument (full) that decides
whether to process task entries for each pid it is passed. It currently
never set to 0 (perf_event__synthesize_comm has a single caller and it
always passes the value 1). Let's fix that.
Add the full input argument to __event__synthesize_thread which passes
it to perf_event__synthesize_comm. For thread/process monitoring set full
to 0 which means COMM and MMAP events are only generated for the pid
passed to it. For system wide monitoring set full to 1 so that COMM events
are generated for all threads in a process.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1324578603-12762-2-git-send-email-dsahern@gmail.com
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The compare script compare-ktest-sample.pl checks for options
that are defined in ktest.pl and not documented in samples.conf,
as well as samples in samples.conf that are not used in ktest.pl.
With the switch to the hash format to initialize the ktest variables
the compare script needs to be updated to handle the change.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
It becomes quite annoying when you go to run a test and then
realize that you typed an option name wrong, and the test starts
doing the default action and not what you expected it to do.
It is even more annoying when you wake up the next day after
running the test over night when you discover this.
By testing if all options specified in a config file are
used by either ktest or were used in one of the option's values
we can see if there are any dangling options that were not used.
In such a case, show the user the options that were not used
and ask them if they want to continue or not.
The option IGNORE_UNUSED was also added to allow the user to
override this feature.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Currently the patchcheck, bisect, and config_bisect variables
are only able to be set per test. You can not set a default
value for them.
By letting default values be set, it makes some config files
a bit easier, and also makes it easier to find typos in the
option names.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Initializing each default value by specifying the hash name is
ugly. This is one of the rare cases that the "perl way" is actually
much cleaner and easier to read.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
For machines that do no use grub, it may be needed to update an
external image (tftp) before doing a reboot into either the
test image or the known good image.
The option SWITCH_TO_GOOD is added, where if it is defined, the
command that is specified as its value will be executed before
doing a reboot into a known good image.
The option SWITCH_TO_TEST is added, where if it is defined, the
command that is specified as its value will be executed before
doing a reboot into the test image.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When running the ktest git bisect test, if the BISECT_TYPE is "test",
the bisect is determined to be good or bad based off of the error
code of the test that is run. Currently, if the test returns 0,
it is considered a pass (good), a non-zero is considered a fail (bad).
But it has been requested to add more options, and also change
the meanings of the error codes of the test. For example, one may
want the test to detect if the commit is not good or bad,
(maybe the bisect came to a point where the code in question
does not exist). The test could report an error code that should tell
ktest to skip the commit.
Also, a test could detect that something is horribly wrong and the
biscet should just be aborted.
The new options:
BISECT_RET_GOOD
BISECT_RET_BAD
BISECT_RET_SKIP
BISECT_RET_ABORT
BISECT_RET_DEFAULT
have been added. The first 4 take an integer value that will
represent if the test should be considered a pass, fail, neither
good nor bad, or abort respectively.
The BISECT_RET_DEFAULT will bo whatever is not defined by the
above codes. If only BISECT_RET_DEFAULT is defined, then all tests
will do the default.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
All options can take variables "${var}". Before doing any processing
or decision making on the content of an option, evaluate it incase
there are variables that may change the outcome.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The install process may also need to know what the kernel version
is, to add it to the name. Evaluate it for both install and
post install.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
If all the tests are only for build or install, do not ask
for options not needed to do the install, if the options do
not exist.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When creating a new config, ask for the BUILD_OPTIONS variable
that lets users add things like -j20 to the make.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When creating a ktest config or if te config only has build only
tests, some of the manditory config options are not needed.
Do not ask for them if all tests in the config file are just build
tests.
Suggested-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When no argument is supplied to ktest, or the config applied does
not exist and a new config is being created, instead of just using
the default test type, give the user an option to pick the test type
of either 'build, install, or boot'. Other options may be added later
but then those would require more questions as they require more
fields. But that's for another release of ktest to add that feature.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
If a bisect is killed for some reason, have ktest detect that a bisect
is in progress and if so, allow the user to start the bisect where
it left off.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The problem is that when SAMPLE_PERIOD is not set, the kernel generates
a number of samples in proportion to an event's period. Number of these
samples may be too big and the kernel throttles all samples above a
defined limit.
E.g.: I want to trace when a process sleeps. I created a process which
sleeps for 1ms and for 4ms. perf got 100 events in both cases.
swapper 0 [000] 1141.371830: sched_stat_sleep: comm=foo pid=1801 delay=1386750 [ns]
swapper 0 [000] 1141.369444: sched_stat_sleep: comm=foo pid=1801 delay=4499585 [ns]
In the first case a kernel want to send 4499585 events and in the second
case it wants to send 1386750 events. perf-reports shows that process
sleeps in both places equal time.
Instead of this we can get only one sample with an attribute period. As
result we have less data transferring between kernel and user-space and
we avoid throttling of samples.
The patch "events: Don't divide events if it has field period" added a
kernel part of this functionality.
Acked-by: Arun Sharma <asharma@fb.com>
Cc: Arun Sharma <asharma@fb.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: devel@openvz.org
Link: http://lkml.kernel.org/r/1324391565-1369947-1-git-send-email-avagin@openvz.org
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Field names were shortened: "pkg" is now "pk", "core" is now "cr"
Signed-off-by: Arun Thomas <arun.thomas@gmail.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
The option is documented in man perf-script but was not yet implemented:
-a
Force system-wide collection. Scripts run without a
<command> normally use -a by default, while scripts run
with a <command> normally don't - this option allows the
latter to be run in system-wide mode.
As with perf record you now can profile in system-wide mode for the
runtime of a given command, e.g.:
# perf script -a syscall-counts sleep 2
Cc: Ingo Molnar <mingo@elte.hu>
Link: http://lkml.kernel.org/r/1322229925-10075-1-git-send-email-robert.richter@amd.com
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf: Do no try to schedule task events if there are none
lockdep, kmemcheck: Annotate ->lock in lockdep_init_map()
perf header: Use event_name() to get an event name
perf stat: Failure with "Operation not supported"
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
ftrace: Fix hash record accounting bug
perf: Fix parsing of __print_flags() in TP_printk()
jump_label: jump_label_inc may return before the code is patched
ftrace: Remove force undef config value left for testing
tracing: Restore system filter behavior
tracing: fix event_subsystem ref counting
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf: Fix loss of notification with multi-event
perf, x86: Force IBS LVT offset assignment for family 10h
perf, x86: Disable PEBS on SandyBridge chips
trace_events_filter: Use rcu_assign_pointer() when setting ftrace_event_call->filter
perf session: Fix crash with invalid CPU list
perf python: Fix undefined symbol problem
perf/x86: Enable raw event access to Intel offcore events
perf: Don't use -ENOSPC for out of PMU resources
perf: Do not set task_ctx pointer in cpuctx if there are no events in the context
perf/x86: Fix PEBS instruction unwind
oprofile, x86: Fix crash when unloading module (nmi timer mode)
oprofile: Fix crash when unloading module (hr timer mode)
A update is made to the sched:sched_switch event that adds some
logic to the first parameter of the __print_flags() that shows the
state of tasks. This change cause perf to fail parsing the flags.
A simple fix is needed to have the parser be able to process ops
within the argument.
Cc: stable@vger.kernel.org
Reported-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
perf stat is failing on PowerPC:
Error: open_counter returned with 95 (Operation not supported). /bin/dmesg may provide additional information.
Fatal: Not all events could be opened.
commit 370faf1dd0 (perf stat: Fail softly on unsupported events)
added a check for failure returning ENOENT, but the POWER backend
returns EOPNOTSUPP. It looks like alpha, blackfin and mips do the
same.
With the patch applied, things work as expected:
Performance counter stats for '/bin/true':
0.362176 task-clock # 0.623 CPUs utilized
0 context-switches # 0.000 M/sec
0 CPU-migrations # 0.000 M/sec
28 page-faults # 0.077 M/sec
1,677,020 cycles # 4.630 GHz
<not supported> stalled-cycles-frontend
<not supported> stalled-cycles-backend
431,220 instructions # 0.26 insns per cycle
101,889 branches # 281.325 M/sec
4,145 branch-misses # 4.07% of all branches
0.000581361 seconds time elapsed
Cc: <stable@kernel.org> # 3.0+
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/20111202093833.5fef7226@kryten
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
