Complete the syscall-less self-profiling feature and address
all complaints, namely:
- capabilities, so we can detect what is actually available at runtime
Add a capabilities field to perf_event_mmap_page to indicate
what is actually available for use.
- on x86: RDPMC weirdness due to being 40/48 bits and not sign-extending
properly.
- ABI documentation as to how all this stuff works.
Also improve the documentation for the new features.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vince Weaver <vweaver1@eecs.utk.edu>
Cc: Arnaldo Carvalho de Melo <acme@infradead.org>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lkml.kernel.org/r/1332433596.2487.33.camel@twins
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Pull cgroup changes from Tejun Heo:
"Out of the 8 commits, one fixes a long-standing locking issue around
tasklist walking and others are cleanups."
* 'for-3.4' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup: Walk task list under tasklist_lock in cgroup_enable_task_cg_list
cgroup: Remove wrong comment on cgroup_enable_task_cg_list()
cgroup: remove cgroup_subsys argument from callbacks
cgroup: remove extra calls to find_existing_css_set
cgroup: replace tasklist_lock with rcu_read_lock
cgroup: simplify double-check locking in cgroup_attach_proc
cgroup: move struct cgroup_pidlist out from the header file
cgroup: remove cgroup_attach_task_current_cg()
With branch stack sampling, it is possible to filter by priv levels.
In system-wide mode, that means it is possible to capture only user
level branches. The builtin SW LBR filter needs to disassemble code
based on LBR captured addresses. For that, it needs to know the task
the addresses are associated with. Because of context switches, the
content of the branch stack buffer may contain addresses from
different tasks.
We need a callback on context switch to either flush the branch stack
or save it. This patch adds a new callback in struct pmu which is called
during context switches. The callback is called only when necessary.
That is when a system-wide context has, at least, one event which
uses PERF_SAMPLE_BRANCH_STACK. The callback is never called for
per-thread context.
In this version, the Intel x86 code simply flushes (resets) the LBR
on context switches (fills it with zeroes). Those zeroed branches are
then filtered out by the SW filter.
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1328826068-11713-11-git-send-email-eranian@google.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch adds the ability to sample taken branches to the
perf_event interface.
The ability to capture taken branches is very useful for all
sorts of analysis. For instance, basic block profiling, call
counts, statistical call graph.
This new capability requires hardware assist and as such may
not be available on all HW platforms. On Intel x86 it is
implemented on top of the Last Branch Record (LBR) facility.
To enable taken branches sampling, the PERF_SAMPLE_BRANCH_STACK
bit must be set in attr->sample_type.
Sampled taken branches may be filtered by type and/or priv
levels.
The patch adds a new field, called branch_sample_type, to the
perf_event_attr structure. It contains a bitmask of filters
to apply to the sampled taken branches.
Filters may be implemented in HW. If the HW filter does not exist
or is not good enough, some arch may also implement a SW filter.
The following generic filters are currently defined:
- PERF_SAMPLE_USER
only branches whose targets are at the user level
- PERF_SAMPLE_KERNEL
only branches whose targets are at the kernel level
- PERF_SAMPLE_HV
only branches whose targets are at the hypervisor level
- PERF_SAMPLE_ANY
any type of branches (subject to priv levels filters)
- PERF_SAMPLE_ANY_CALL
any call branches (may incl. syscall on some arch)
- PERF_SAMPLE_ANY_RET
any return branches (may incl. syscall returns on some arch)
- PERF_SAMPLE_IND_CALL
indirect call branches
Obviously filter may be combined. The priv level bits are optional.
If not provided, the priv level of the associated event are used. It
is possible to collect branches at a priv level different from the
associated event. Use of kernel, hv priv levels is subject to permissions
and availability (hv).
The number of taken branch records present in each sample may vary based
on HW, the type of sampled branches, the executed code. Therefore
each sample contains the number of taken branches it contains.
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1328826068-11713-2-git-send-email-eranian@google.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If kzalloc() for TYPE_DATA failed on a given cpu, previous chunk
of TYPE_INST will be leaked. Fix it.
Thanks to Peter Zijlstra for suggesting this better solution. It
should work as long as the initial value of the region is all
0's and that's the case of static (per-cpu) memory allocation.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: 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/1330391978-28070-1-git-send-email-namhyung.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
So here's a boot tested patch on top of Jason's series that does
all the cleanups I talked about and turns jump labels into a
more intuitive to use facility. It should also address the
various misconceptions and confusions that surround jump labels.
Typical usage scenarios:
#include <linux/static_key.h>
struct static_key key = STATIC_KEY_INIT_TRUE;
if (static_key_false(&key))
do unlikely code
else
do likely code
Or:
if (static_key_true(&key))
do likely code
else
do unlikely code
The static key is modified via:
static_key_slow_inc(&key);
...
static_key_slow_dec(&key);
The 'slow' prefix makes it abundantly clear that this is an
expensive operation.
I've updated all in-kernel code to use this everywhere. Note
that I (intentionally) have not pushed through the rename
blindly through to the lowest levels: the actual jump-label
patching arch facility should be named like that, so we want to
decouple jump labels from the static-key facility a bit.
On non-jump-label enabled architectures static keys default to
likely()/unlikely() branches.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Jason Baron <jbaron@redhat.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: a.p.zijlstra@chello.nl
Cc: mathieu.desnoyers@efficios.com
Cc: davem@davemloft.net
Cc: ddaney.cavm@gmail.com
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/20120222085809.GA26397@elte.hu
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The following patch fixes a bug introduced by the following
commit:
e050e3f0a7 ("perf: Fix broken interrupt rate throttling")
The patch caused the following warning to pop up depending on
the sampling frequency adjustments:
------------[ cut here ]------------
WARNING: at arch/x86/kernel/cpu/perf_event.c:995 x86_pmu_start+0x79/0xd4()
It was caused by the following call sequence:
perf_adjust_freq_unthr_context.part() {
stop()
if (delta > 0) {
perf_adjust_period() {
if (period > 8*...) {
stop()
...
start()
}
}
}
start()
}
Which caused a double start and a double stop, thus triggering
the assert in x86_pmu_start().
The patch fixes the problem by avoiding the double calls. We
pass a new argument to perf_adjust_period() to indicate whether
or not the event is already stopped. We can't just remove the
start/stop from that function because it's called from
__perf_event_overflow where the event needs to be reloaded via a
stop/start back-toback call.
The patch reintroduces the assertion in x86_pmu_start() which
was removed by commit:
84f2b9b ("perf: Remove deprecated WARN_ON_ONCE()")
In this second version, we've added calls to disable/enable PMU
during unthrottling or frequency adjustment based on bug report
of spurious NMI interrupts from Eric Dumazet.
Reported-and-tested-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Stephane Eranian <eranian@google.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: markus@trippelsdorf.de
Cc: paulus@samba.org
Link: http://lkml.kernel.org/r/20120207133956.GA4932@quad
[ Minor edits to the changelog and to the code ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The argument is not used at all, and it's not necessary, because
a specific callback handler of course knows which subsys it
belongs to.
Now only ->pupulate() takes this argument, because the handlers of
this callback always call cgroup_add_file()/cgroup_add_files().
So we reduce a few lines of code, though the shrinking of object size
is minimal.
16 files changed, 113 insertions(+), 162 deletions(-)
text data bss dec hex filename
5486240 656987 7039960 13183187 c928d3 vmlinux.o.orig
5486170 656987 7039960 13183117 c9288d vmlinux.o
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This patch fixes the sampling interrupt throttling mechanism.
It was broken in v3.2. Events were not being unthrottled. The
unthrottling mechanism required that events be checked at each
timer tick.
This patch solves this problem and also separates:
- unthrottling
- multiplexing
- frequency-mode period adjustments
Not all of them need to be executed at each timer tick.
This third version of the patch is based on my original patch +
PeterZ proposal (https://lkml.org/lkml/2012/1/7/87).
At each timer tick, for each context:
- if the current CPU has throttled events, we unthrottle events
- if context has frequency-based events, we adjust sampling periods
- if we have reached the jiffies interval, we multiplex (rotate)
We decoupled rotation (multiplexing) from frequency-mode sampling
period adjustments. They should not necessarily happen at the same
rate. Multiplexing is subject to jiffies_interval (currently at 1
but could be higher once the tunable is exposed via sysfs).
We have grouped frequency-mode adjustment and unthrottling into the
same routine to minimize code duplication. When throttled while in
frequency mode, we scan the events only once.
We have fixed the threshold enforcement code in __perf_event_overflow().
There was a bug whereby it would allow more than the authorized rate
because an increment of hwc->interrupts was not executed at the right
place.
The patch was tested with low sampling limit (2000) and fixed periods,
frequency mode, overcommitted PMU.
On a 2.1GHz AMD CPU:
$ cat /proc/sys/kernel/perf_event_max_sample_rate
2000
We set a rate of 3000 samples/sec (2.1GHz/3000 = 700000):
$ perf record -e cycles,cycles -c 700000 noploop 10
$ perf report -D | tail -21
Aggregated stats:
TOTAL events: 80086
MMAP events: 88
COMM events: 2
EXIT events: 4
THROTTLE events: 19996
UNTHROTTLE events: 19996
SAMPLE events: 40000
cycles stats:
TOTAL events: 40006
MMAP events: 5
COMM events: 1
EXIT events: 4
THROTTLE events: 9998
UNTHROTTLE events: 9998
SAMPLE events: 20000
cycles stats:
TOTAL events: 39996
THROTTLE events: 9998
UNTHROTTLE events: 9998
SAMPLE events: 20000
For 10s, the cap is 2x2000x10 = 40000 samples.
We get exactly that: 20000 samples/event.
Signed-off-by: Stephane Eranian <eranian@google.com>
Cc: <stable@kernel.org> # v3.2+
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20120126160319.GA5655@quad
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The perf_event_time() will call perf_cgroup_event_time()
if @event is a cgroup event. Just do it directly and avoid
the extra check..
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com>
Cc: Namhyung Kim <namhyung@gmail.com>
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/1327021966-27688-2-git-send-email-namhyung.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When alloc_callchain_buffers() fails, it frees all of
entries before return. In addition, calling the
release_callchain_buffers() will cause a NULL pointer
dereference since callchain_cpu_entries is not set.
Signed-off-by: Namhyung Kim <namhyung.kim@lge.com>
Acked-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Namhyung Kim <namhyung@gmail.com>
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/1327021966-27688-1-git-send-email-namhyung.kim@lge.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'for-3.3' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup: (21 commits)
cgroup: fix to allow mounting a hierarchy by name
cgroup: move assignement out of condition in cgroup_attach_proc()
cgroup: Remove task_lock() from cgroup_post_fork()
cgroup: add sparse annotation to cgroup_iter_start() and cgroup_iter_end()
cgroup: mark cgroup_rmdir_waitq and cgroup_attach_proc() as static
cgroup: only need to check oldcgrp==newgrp once
cgroup: remove redundant get/put of task struct
cgroup: remove redundant get/put of old css_set from migrate
cgroup: Remove unnecessary task_lock before fetching css_set on migration
cgroup: Drop task_lock(parent) on cgroup_fork()
cgroups: remove redundant get/put of css_set from css_set_check_fetched()
resource cgroups: remove bogus cast
cgroup: kill subsys->can_attach_task(), pre_attach() and attach_task()
cgroup, cpuset: don't use ss->pre_attach()
cgroup: don't use subsys->can_attach_task() or ->attach_task()
cgroup: introduce cgroup_taskset and use it in subsys->can_attach(), cancel_attach() and attach()
cgroup: improve old cgroup handling in cgroup_attach_proc()
cgroup: always lock threadgroup during migration
threadgroup: extend threadgroup_lock() to cover exit and exec
threadgroup: rename signal->threadgroup_fork_lock to ->group_rwsem
...
Fix up conflict in kernel/cgroup.c due to commit e0197aae59: "cgroups:
fix a css_set not found bug in cgroup_attach_proc" that already
mentioned that the bug is fixed (differently) in Tejun's cgroup
patchset. This one, in other words.
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/trivial: (53 commits)
Kconfig: acpi: Fix typo in comment.
misc latin1 to utf8 conversions
devres: Fix a typo in devm_kfree comment
btrfs: free-space-cache.c: remove extra semicolon.
fat: Spelling s/obsolate/obsolete/g
SCSI, pmcraid: Fix spelling error in a pmcraid_err() call
tools/power turbostat: update fields in manpage
mac80211: drop spelling fix
types.h: fix comment spelling for 'architectures'
typo fixes: aera -> area, exntension -> extension
devices.txt: Fix typo of 'VMware'.
sis900: Fix enum typo 'sis900_rx_bufer_status'
decompress_bunzip2: remove invalid vi modeline
treewide: Fix comment and string typo 'bufer'
hyper-v: Update MAINTAINERS
treewide: Fix typos in various parts of the kernel, and fix some comments.
clockevents: drop unknown Kconfig symbol GENERIC_CLOCKEVENTS_MIGR
gpio: Kconfig: drop unknown symbol 'CS5535_GPIO'
leds: Kconfig: Fix typo 'D2NET_V2'
sound: Kconfig: drop unknown symbol ARCH_CLPS7500
...
Fix up trivial conflicts in arch/powerpc/platforms/40x/Kconfig (some new
kconfig additions, close to removed commented-out old ones)
* 'perf-core-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (106 commits)
perf kvm: Fix copy & paste error in description
perf script: Kill script_spec__delete
perf top: Fix a memory leak
perf stat: Introduce get_ratio_color() helper
perf session: Remove impossible condition check
perf tools: Fix feature-bits rework fallout, remove unused variable
perf script: Add generic perl handler to process events
perf tools: Use for_each_set_bit() to iterate over feature flags
perf tools: Unify handling of features when writing feature section
perf report: Accept fifos as input file
perf tools: Moving code in some files
perf tools: Fix out-of-bound access to struct perf_session
perf tools: Continue processing header on unknown features
perf tools: Improve macros for struct feature_ops
perf: builtin-record: Document and check that mmap_pages must be a power of two.
perf: builtin-record: Provide advice if mmap'ing fails with EPERM.
perf tools: Fix truncated annotation
perf script: look up thread using tid instead of pid
perf tools: Look up thread names for system wide profiling
perf tools: Fix comm for processes with named threads
...
* 'core-rcu-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (64 commits)
cpu: Export cpu_up()
rcu: Apply ACCESS_ONCE() to rcu_boost() return value
Revert "rcu: Permit rt_mutex_unlock() with irqs disabled"
docs: Additional LWN links to RCU API
rcu: Augment rcu_batch_end tracing for idle and callback state
rcu: Add rcutorture tests for srcu_read_lock_raw()
rcu: Make rcutorture test for hotpluggability before offlining CPUs
driver-core/cpu: Expose hotpluggability to the rest of the kernel
rcu: Remove redundant rcu_cpu_stall_suppress declaration
rcu: Adaptive dyntick-idle preparation
rcu: Keep invoking callbacks if CPU otherwise idle
rcu: Irq nesting is always 0 on rcu_enter_idle_common
rcu: Don't check irq nesting from rcu idle entry/exit
rcu: Permit dyntick-idle with callbacks pending
rcu: Document same-context read-side constraints
rcu: Identify dyntick-idle CPUs on first force_quiescent_state() pass
rcu: Remove dynticks false positives and RCU failures
rcu: Reduce latency of rcu_prepare_for_idle()
rcu: Eliminate RCU_FAST_NO_HZ grace-period hang
rcu: Avoid needlessly IPIing CPUs at GP end
...
Extend the mmap control page with fields so that userspace can compute
time deltas relative to the provided time fields.
Currently only implemented for x86 with constant and nonstop TSC.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Arun Sharma <asharma@fb.com>
Link: http://lkml.kernel.org/n/tip-3u1jucza77j3wuvs0x2bic0f@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
There's multiple reason the counter might be unavailable, change the
condition to !->index since perf_event_index() should return 0 for all
those cases.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-1ixr3olci40w8rgv2evv2ldh@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Put the logic to compute the event index into a per pmu method. This
is required because the x86 rules are weird and wonderful and don't
match the capabilities of the current scheme.
AFAIK only powerpc actually has a usable userspace read of the PMCs
but I'm not at all sure anybody actually used that.
ARM is restored to the default since it currently does not support
userspace access at all. And all software events are provided with a
method that reports their index as 0 (disabled).
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Michael Cree <mcree@orcon.net.nz>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Deng-Cheng Zhu <dengcheng.zhu@gmail.com>
Cc: Anton Blanchard <anton@samba.org>
Cc: Eric B Munson <emunson@mgebm.net>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Richard Kuo <rkuo@codeaurora.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Arun Sharma <asharma@fb.com>
Link: http://lkml.kernel.org/n/tip-dfydxodki16lylkt3gl2j7cw@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Apparently we didn't update the mmap control page right after mmap(),
which leads to surprises when userspace wants to use it.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Arun Sharma <asharma@fb.com>
Link: http://lkml.kernel.org/n/tip-dcpi7164djsexmx6ya7lilrc@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Commit 10c6db11 ("perf: Fix loss of notification with multi-event")
seems to unconditionally dereference event->rb in the wakeup handler,
this is wrong, there might not be a buffer attached.
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111213152651.GP20297@mudshark.cambridge.arm.com
[ minor edits ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Now that subsys->can_attach() and attach() take @tset instead of
@task, they can handle per-task operations. Convert
->can_attach_task() and ->attach_task() users to use ->can_attach()
and attach() instead. Most converions are straight-forward.
Noteworthy changes are,
* In cgroup_freezer, remove unnecessary NULL assignments to unused
methods. It's useless and very prone to get out of sync, which
already happened.
* In cpuset, PF_THREAD_BOUND test is checked for each task. This
doesn't make any practical difference but is conceptually cleaner.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Paul Menage <paul@paulmenage.org>
Cc: Balbir Singh <bsingharora@gmail.com>
Cc: Daisuke Nishimura <nishimura@mxp.nes.nec.co.jp>
Cc: James Morris <jmorris@namei.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Change from direct comparison of ->pid with zero to is_idle_task().
Signed-off-by: Paul E. McKenney <paul.mckenney@linaro.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
perf_event_sched_in() shouldn't try to schedule task events if there
are none otherwise task's ctx->is_active will be set and will not be
cleared during sched_out. This will prevent newly added events from
being scheduled into the task context.
Fixes a boo-boo in commit 1d5f003f5a ("perf: Do not set task_ctx
pointer in cpuctx if there are no events in the context").
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111122140821.GF2557@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
jump_lable patching is very expensive operation that involves pausing all
cpus. The patching of perf_sched_events jump_label is easily controllable
from userspace by unprivileged user.
When te user runs a loop like this:
"while true; do perf stat -e cycles true; done"
... the performance of my test application that just increments a counter
for one second drops by 4%.
This is on a 16 cpu box with my test application using only one of
them. An impact on a real server doing real work will be worse.
Performance of KVM PMU drops nearly 50% due to jump_lable for "perf
record" since KVM PMU implementation creates and destroys perf event
frequently.
This patch introduces a way to rate limit jump_label patching and uses
it to fix the above problem.
I believe that as jump_label use will spread the problem will become more
common and thus solving it in a generic code is appropriate. Also fixing
it in the perf code would result in moving jump_label accounting logic to
perf code with all the ifdefs in case of JUMP_LABEL=n kernel. With this
patch all details are nicely hidden inside jump_label code.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Acked-by: Jason Baron <jbaron@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111127155909.GO2557@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Deng-Cheng Zhu reported that sibling events that were created disabled
with enable_on_exec would never get enabled. Iterate all events
instead of the group lists.
Reported-by: Deng-Cheng Zhu <dczhu@mips.com>
Tested-by: Deng-Cheng Zhu <dczhu@mips.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1322048382.14799.41.camel@twins
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Gleb writes:
> Currently pmu is disabled and re-enabled on each timer interrupt even
> when no rotation or frequency adjustment is needed. On Intel CPU this
> results in two writes into PERF_GLOBAL_CTRL MSR per tick. On bare metal
> it does not cause significant slowdown, but when running perf in a virtual
> machine it leads to 20% slowdown on my machine.
Cure this by keeping a perf_event_context::nr_freq counter that counts the
number of active events that require frequency adjustments and use this in a
similar fashion to the already existing nr_events != nr_active test in
perf_rotate_context().
By being able to exclude both rotation and frequency adjustments a-priory for
the common case we can avoid the otherwise superfluous PMU disable.
Suggested-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-515yhoatehd3gza7we9fapaa@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When you do:
$ perf record -e cycles,cycles,cycles noploop 10
You expect about 10,000 samples for each event, i.e., 10s at
1000samples/sec. However, this is not what's happening. You
get much fewer samples, maybe 3700 samples/event:
$ perf report -D | tail -15
Aggregated stats:
TOTAL events: 10998
MMAP events: 66
COMM events: 2
SAMPLE events: 10930
cycles stats:
TOTAL events: 3644
SAMPLE events: 3644
cycles stats:
TOTAL events: 3642
SAMPLE events: 3642
cycles stats:
TOTAL events: 3644
SAMPLE events: 3644
On a Intel Nehalem or even AMD64, there are 4 counters capable
of measuring cycles, so there is plenty of space to measure those
events without multiplexing (even with the NMI watchdog active).
And even with multiplexing, we'd expect roughly the same number
of samples per event.
The root of the problem was that when the event that caused the buffer
to become full was not the first event passed on the cmdline, the user
notification would get lost. The notification was sent to the file
descriptor of the overflowed event but the perf tool was not polling
on it. The perf tool aggregates all samples into a single buffer,
i.e., the buffer of the first event. Consequently, it assumes
notifications for any event will come via that descriptor.
The seemingly straight forward solution of moving the waitq into the
ringbuffer object doesn't work because of life-time issues. One could
perf_event_set_output() on a fd that you're also blocking on and cause
the old rb object to be freed while its waitq would still be
referenced by the blocked thread -> FAIL.
Therefore link all events to the ringbuffer and broadcast the wakeup
from the ringbuffer object to all possible events that could be waited
upon. This is rather ugly, and we're open to better solutions but it
works for now.
Reported-by: Stephane Eranian <eranian@google.com>
Finished-by: Stephane Eranian <eranian@google.com>
Reviewed-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111126014731.GA7030@quad
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch solves the following problem:
Now some samples may be lost due to throttling. The number of samples is
restricted by sysctl_perf_event_sample_rate/HZ. A trace event is
divided on some samples according to event's period. I don't sure, that
we should generate more than one sample on each trace event. I think the
better way to use SAMPLE_PERIOD.
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. It's
bug.
With this patch kernel generates one event on each "sleep" and the time
slice is saved in the field "period". Perf knows how handle it.
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1320670457-2633428-3-git-send-email-avagin@openvz.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Split the callchain code from the perf events core into
a new kernel/events/callchain.c file.
This simplifies a bit the big core.c
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Stephane Eranian <eranian@google.com>
[keep ctx recursion handling inline and use internal headers]
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1318778104-17152-1-git-send-email-fweisbec@gmail.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Do not set task_ctx pointer during sched_in if there are no
events associated with the context. Otherwise if during task
execution total number of events in the system will become zero
perf_event_context_sched_out() will not be called and cpuctx->task_ctx
will be left with a stale value.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20111023171033.GI17571@redhat.com
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'modsplit-Oct31_2011' of git://git.kernel.org/pub/scm/linux/kernel/git/paulg/linux: (230 commits)
Revert "tracing: Include module.h in define_trace.h"
irq: don't put module.h into irq.h for tracking irqgen modules.
bluetooth: macroize two small inlines to avoid module.h
ip_vs.h: fix implicit use of module_get/module_put from module.h
nf_conntrack.h: fix up fallout from implicit moduleparam.h presence
include: replace linux/module.h with "struct module" wherever possible
include: convert various register fcns to macros to avoid include chaining
crypto.h: remove unused crypto_tfm_alg_modname() inline
uwb.h: fix implicit use of asm/page.h for PAGE_SIZE
pm_runtime.h: explicitly requires notifier.h
linux/dmaengine.h: fix implicit use of bitmap.h and asm/page.h
miscdevice.h: fix up implicit use of lists and types
stop_machine.h: fix implicit use of smp.h for smp_processor_id
of: fix implicit use of errno.h in include/linux/of.h
of_platform.h: delete needless include <linux/module.h>
acpi: remove module.h include from platform/aclinux.h
miscdevice.h: delete unnecessary inclusion of module.h
device_cgroup.h: delete needless include <linux/module.h>
net: sch_generic remove redundant use of <linux/module.h>
net: inet_timewait_sock doesnt need <linux/module.h>
...
Fix up trivial conflicts (other header files, and removal of the ab3550 mfd driver) in
- drivers/media/dvb/frontends/dibx000_common.c
- drivers/media/video/{mt9m111.c,ov6650.c}
- drivers/mfd/ab3550-core.c
- include/linux/dmaengine.h
The legacy x86 nmi watchdog code was removed with the implementation
of the perf based nmi watchdog. This broke Oprofile's nmi timer
mode. To run nmi timer mode we relied on a continuous ticking nmi
source which the nmi watchdog provided. The nmi tick was no longer
available and current watchdog can not be used anymore since it runs
with very long periods in the range of seconds. This patch
reimplements the nmi timer mode using a perf counter nmi source.
V2:
* removing pr_info()
* fix undefined reference to `__udivdi3' for 32 bit build
* fix section mismatch of .cpuinit.data:nmi_timer_cpu_nb
* removed nmi timer setup in arch/x86
* implemented function stubs for op_nmi_init/exit()
* made code more readable in oprofile_init()
V3:
* fix architectural initialization in oprofile_init()
* fix CONFIG_OPROFILE_NMI_TIMER dependencies
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Robert Richter <robert.richter@amd.com>
This reverts commit 144060fee0.
It causes a resume regression for Andi on his Acer Aspire 1830T post
3.1. The screen just stays black after wakeup.
Also, it really looks like the wrong way to suspend and resume perf
events: I think they should be done as part of the CPU suspend and
resume, rather than as a notifier that does smp_call_function().
Reported-by: Andi Kleen <andi@firstfloor.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Some kernel components pin user space memory (infiniband and perf) (by
increasing the page count) and account that memory as "mlocked".
The difference between mlocking and pinning is:
A. mlocked pages are marked with PG_mlocked and are exempt from
swapping. Page migration may move them around though.
They are kept on a special LRU list.
B. Pinned pages cannot be moved because something needs to
directly access physical memory. They may not be on any
LRU list.
I recently saw an mlockalled process where mm->locked_vm became
bigger than the virtual size of the process (!) because some
memory was accounted for twice:
Once when the page was mlocked and once when the Infiniband
layer increased the refcount because it needt to pin the RDMA
memory.
This patch introduces a separate counter for pinned pages and
accounts them seperately.
Signed-off-by: Christoph Lameter <cl@linux.com>
Cc: Mike Marciniszyn <infinipath@qlogic.com>
Cc: Roland Dreier <roland@kernel.org>
Cc: Sean Hefty <sean.hefty@intel.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
These files were getting <linux/module.h> via an implicit non-obvious
path, but we want to crush those out of existence since they cost
time during compiles of processing thousands of lines of headers
for no reason. Give them the lightweight header that just contains
the EXPORT_SYMBOL infrastructure.
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
We detected a serious issue with PERF_SAMPLE_READ and
timing information when events were being multiplexing.
Samples would have time_running > time_enabled. That
was easy to reproduce with a libpfm4 example (ran 3
times to cause multiplexing on Core 2):
$ syst_smpl -e uops_retired:freq=1 &
$ syst_smpl -e uops_retired:freq=1 &
$ syst_smpl -e uops_retired:freq=1 &
IIP:0x0000000040062d ... PERIOD:2355332948 ENA=40144625315 RUN=60014875184
syst_smpl: WARNING: time_running > time_enabled
63277537998 uops_retired:freq=1 , scaled
The bug was not present in kernel up to (and including) 3.0. It turns
out the bug was introduced by the following commit:
commit c479429591
events: Move lockless timer calculation into helper function
The parameters of the function got reversed yet the call sites
were not updated to reflect the change. That lead to time_running
and time_enabled being swapped. That had no effect when there was
no multiplexing because in that case time_running = time_enabled
but it would show up in any other scenario.
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20110829124112.GA4828@quad
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The current cgroup context switch code was incorrect leading
to bogus counts. Furthermore, as soon as there was an active
cgroup event on a CPU, the context switch cost on that CPU
would increase by a significant amount as demonstrated by a
simple ping/pong example:
$ ./pong
Both processes pinned to CPU1, running for 10s
10684.51 ctxsw/s
Now start a cgroup perf stat:
$ perf stat -e cycles,cycles -A -a -G test -C 1 -- sleep 100
$ ./pong
Both processes pinned to CPU1, running for 10s
6674.61 ctxsw/s
That's a 37% penalty.
Note that pong is not even in the monitored cgroup.
The results shown by perf stat are bogus:
$ perf stat -e cycles,cycles -A -a -G test -C 1 -- sleep 100
Performance counter stats for 'sleep 100':
CPU1 <not counted> cycles test
CPU1 16,984,189,138 cycles # 0.000 GHz
The second 'cycles' event should report a count @ CPU clock
(here 2.4GHz) as it is counting across all cgroups.
The patch below fixes the bogus accounting and bypasses any
cgroup switches in case the outgoing and incoming tasks are
in the same cgroup.
With this patch the same test now yields:
$ ./pong
Both processes pinned to CPU1, running for 10s
10775.30 ctxsw/s
Start perf stat with cgroup:
$ perf stat -e cycles,cycles -A -a -G test -C 1 -- sleep 10
Run pong outside the cgroup:
$ /pong
Both processes pinned to CPU1, running for 10s
10687.80 ctxsw/s
The penalty is now less than 2%.
And the results for perf stat are correct:
$ perf stat -e cycles,cycles -A -a -G test -C 1 -- sleep 10
Performance counter stats for 'sleep 10':
CPU1 <not counted> cycles test # 0.000 GHz
CPU1 23,933,981,448 cycles # 0.000 GHz
Now perf stat reports the correct counts for
for the non cgroup event.
If we run pong inside the cgroup, then we also get the
correct counts:
$ perf stat -e cycles,cycles -A -a -G test -C 1 -- sleep 10
Performance counter stats for 'sleep 10':
CPU1 22,297,726,205 cycles test # 0.000 GHz
CPU1 23,933,981,448 cycles # 0.000 GHz
10.001457237 seconds time elapsed
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20110825135803.GA4697@quad
Signed-off-by: Ingo Molnar <mingo@elte.hu>
