In an XDP redirect applications using tracepoint xdp:xdp_redirect to
diagnose TX overrun, I noticed perf_swevent_get_recursion_context()
was consuming 2% CPU. This was reduced to 1.85% with this simple
change.
Looking at the annotated asm code, it was clear that the unlikely case
in_nmi() test was chosen (by the compiler) as the most likely
event/branch. This small adjustment makes the compiler (GCC version
7.1.1 20170622 (Red Hat 7.1.1-3)) put in_nmi() as an unlikely branch.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/150342256382.16595.986861478681783732.stgit@firesoul
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The aux_watermark member of struct ring_buffer represents the period (in
terms of bytes) at which wakeup events should be generated when data is
written to the aux buffer in non-snapshot mode. On hardware that cannot
generate an interrupt when the aux_head reaches an arbitrary wakeup index
(such as ARM SPE), the aux_head sampled from handle->head in
perf_aux_output_{skip,end} may in fact be past the wakeup index. This
can lead to wakeup slowly falling behind the head. For example, consider
the case where hardware can only generate an interrupt on a page-boundary
and the aux buffer is initialised as follows:
// Buffer size is 2 * PAGE_SIZE
rb->aux_head = rb->aux_wakeup = 0
rb->aux_watermark = PAGE_SIZE / 2
following the first perf_aux_output_begin call, the handle is
initialised with:
handle->head = 0
handle->size = 2 * PAGE_SIZE
handle->wakeup = PAGE_SIZE / 2
and the hardware will be programmed to generate an interrupt at
PAGE_SIZE.
When the interrupt is raised, the hardware head will be at PAGE_SIZE,
so calling perf_aux_output_end(handle, PAGE_SIZE) puts the ring buffer
into the following state:
rb->aux_head = PAGE_SIZE
rb->aux_wakeup = PAGE_SIZE / 2
rb->aux_watermark = PAGE_SIZE / 2
and then the next call to perf_aux_output_begin will result in:
handle->head = handle->wakeup = PAGE_SIZE
for which the semantics are unclear and, for a smaller aux_watermark
(e.g. PAGE_SIZE / 4), then the wakeup would in fact be behind head at
this point.
This patch fixes the problem by rounding down the aux_head (as sampled
from the handle) to the nearest aux_watermark boundary when updating
rb->aux_wakeup, therefore taking into account any overruns by the
hardware.
Reported-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/1502900297-21839-2-git-send-email-will.deacon@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch fixes the __output_custom() routine we currently use with
bpf_skb_copy(). I missed that when len is larger than the size of the
current handle, we can issue multiple invocations of copy_func, and
__output_custom() advances destination but also source buffer by the
written amount of bytes. When we have __output_custom(), this is actually
wrong since in that case the source buffer points to a non-linear object,
in our case an skb, which the copy_func helper is supposed to walk.
Therefore, since this is non-linear we thus need to pass the offset into
the helper, so that copy_func can use it for extracting the data from
the source object.
Therefore, adjust the callback signatures properly and pass offset
into the skb_header_pointer() invoked from bpf_skb_copy() callback. The
__DEFINE_OUTPUT_COPY_BODY() is adjusted to accommodate for two things:
i) to pass in whether we should advance source buffer or not; this is
a compile-time constant condition, ii) to pass in the offset for
__output_custom(), which we do with help of __VA_ARGS__, so everything
can stay inlined as is currently. Both changes allow for adapting the
__output_* fast-path helpers w/o extra overhead.
Fixes: 555c8a8623 ("bpf: avoid stack copy and use skb ctx for event output")
Fixes: 7e3f977edd ("perf, events: add non-linear data support for raw records")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support for non-linear data on raw records. It
extends raw records to have one or multiple fragments that will
be written linearly into the ring slot, where each fragment can
optionally have a custom callback handler to walk and extract
complex, possibly non-linear data.
If a callback handler is provided for a fragment, then the new
__output_custom() will be used instead of __output_copy() for
the perf_output_sample() part. perf_prepare_sample() does all
the size calculation only once, so perf_output_sample() doesn't
need to redo the same work anymore, meaning real_size and padding
will be cached in the raw record. The raw record becomes 32 bytes
in size without holes; to not increase it further and to avoid
doing unnecessary recalculations in fast-path, we can reuse
next pointer of the last fragment, idea here is borrowed from
ZERO_OR_NULL_PTR(), which should keep the perf_output_sample()
path for PERF_SAMPLE_RAW minimal.
This facility is needed for BPF's event output helper as a first
user that will, in a follow-up, add an additional perf_raw_frag
to its perf_raw_record in order to be able to more efficiently
dump skb context after a linear head meta data related to it.
skbs can be non-linear and thus need a custom output function to
dump buffers. Currently, the skb data needs to be copied twice;
with the help of __output_custom() this work only needs to be
done once. Future users could be things like XDP/BPF programs
that work on different context though and would thus also have
a different callback function.
The few users of raw records are adapted to initialize their frag
data from the raw record itself, no change in behavior for them.
The code is based upon a PoC diff provided by Peter Zijlstra [1].
[1] http://thread.gmane.org/gmane.linux.network/421294
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
. avoid walking the stack when there is no room left in the buffer
. generalize get_perf_callchain() to be called from bpf helper
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the PEBS interrupt threshold is larger than one record and the
machine supports multiple PEBS events, the records of these events are
mixed up and we need to demultiplex them.
Demuxing the records is hard because the hardware is deficient. The
hardware has two issues that, when combined, create impossible
scenarios to demux.
The first issue is that the 'status' field of the PEBS record is a copy
of the GLOBAL_STATUS MSR at PEBS assist time. To see why this is a
problem let us first describe the regular PEBS cycle:
A) the CTRn value reaches 0:
- the corresponding bit in GLOBAL_STATUS gets set
- we start arming the hardware assist
< some unspecified amount of time later -- this could cover multiple
events of interest >
B) the hardware assist is armed, any next event will trigger it
C) a matching event happens:
- the hardware assist triggers and generates a PEBS record
this includes a copy of GLOBAL_STATUS at this moment
- if we auto-reload we (re)set CTRn
- we clear the relevant bit in GLOBAL_STATUS
Now consider the following chain of events:
A0, B0, A1, C0
The event generated for counter 0 will include a status with counter 1
set, even though its not at all related to the record. A similar thing
can happen with a !PEBS event if it just happens to overflow at the
right moment.
The second issue is that the hardware will only emit one record for two
or more counters if the event that triggers the assist is 'close'. The
'close' can be several cycles. In some cases even the complete assist,
if the event is something that doesn't need retirement.
For instance, consider this chain of events:
A0, B0, A1, B1, C01
Where C01 is an event that triggers both hardware assists, we will
generate but a single record, but again with both counters listed in the
status field.
This time the record pertains to both events.
Note that these two cases are different but undistinguishable with the
data as generated. Therefore demuxing records with multiple PEBS bits
(we can safely ignore status bits for !PEBS counters) is impossible.
Furthermore we cannot emit the record to both events because that might
cause a data leak -- the events might not have the same privileges -- so
what this patch does is discard such events.
The assumption/hope is that such discards will be rare.
Here lists some possible ways you may get high discard rate.
- when you count the same thing multiple times. But it is not a useful
configuration.
- you can be unfortunate if you measure with a userspace only PEBS
event along with either a kernel or unrestricted PEBS event. Imagine
the event triggering and setting the overflow flag right before
entering the kernel. Then all kernel side events will end up with
multiple bits set.
Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
Signed-off-by: Kan Liang <kan.liang@intel.com>
[ Changelog improvements. ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: acme@infradead.org
Cc: eranian@google.com
Link: http://lkml.kernel.org/r/1430940834-8964-4-git-send-email-kan.liang@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This adds support for overwrite mode in the AUX area, which means "keep
collecting data till you're stopped", turning AUX area into a circular
buffer, where new data overwrites old data. It does not depend on data
buffer's overwrite mode, so that it doesn't lose sideband data that is
instrumental for processing AUX data.
Overwrite mode is enabled at mapping AUX area read only. Even though
aux_tail in the buffer's user page might be user writable, it will be
ignored in this mode.
A PERF_RECORD_AUX with PERF_AUX_FLAG_OVERWRITE set is written to the perf
data stream every time an event writes new data to the AUX area. The pmu
driver might not be able to infer the exact beginning of the new data in
each snapshot, some drivers will only provide the tail, which is
aux_offset + aux_size in the AUX record. Consumer has to be able to tell
the new data from the old one, for example, by means of time stamps if
such are provided in the trace.
Consumer is also responsible for disabling any events that might write
to the AUX area (thus potentially racing with the consumer) before
collecting the data.
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Kaixu Xia <kaixu.xia@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Robert Richter <rric@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: acme@infradead.org
Cc: adrian.hunter@intel.com
Cc: kan.liang@intel.com
Cc: markus.t.metzger@intel.com
Cc: mathieu.poirier@linaro.org
Link: http://lkml.kernel.org/r/1421237903-181015-9-git-send-email-alexander.shishkin@linux.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
For pmus that wish to write data to ring buffer's AUX area, provide
perf_aux_output_{begin,end}() calls to initiate/commit data writes,
similarly to perf_output_{begin,end}. These also use the same output
handle structure. Also, similarly to software counterparts, these
will direct inherited events' output to parents' ring buffers.
After the perf_aux_output_begin() returns successfully, handle->size
is set to the maximum amount of data that can be written wrt aux_tail
pointer, so that no data that the user hasn't seen will be overwritten,
therefore this should always be called before hardware writing is
enabled. On success, this will return the pointer to pmu driver's
private structure allocated for this aux area by pmu::setup_aux. Same
pointer can also be retrieved using perf_get_aux() while hardware
writing is enabled.
PMU driver should pass the actual amount of data written as a parameter
to perf_aux_output_end(). All hardware writes should be completed and
visible before this one is called.
Additionally, perf_aux_output_skip() will adjust output handle and
aux_head in case some part of the buffer has to be skipped over to
maintain hardware's alignment constraints.
Nested writers are forbidden and guards are in place to catch such
attempts.
Signed-off-by: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Kaixu Xia <kaixu.xia@linaro.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Robert Richter <rric@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: acme@infradead.org
Cc: adrian.hunter@intel.com
Cc: kan.liang@intel.com
Cc: markus.t.metzger@intel.com
Cc: mathieu.poirier@linaro.org
Link: http://lkml.kernel.org/r/1421237903-181015-8-git-send-email-alexander.shishkin@linux.intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The arch_perf_output_copy_user() default of
__copy_from_user_inatomic() returns bytes not copied, while all other
argument functions given DEFINE_OUTPUT_COPY() return bytes copied.
Since copy_from_user_nmi() is the odd duck out by returning bytes
copied where all other *copy_{to,from}* functions return bytes not
copied, change it over and ammend DEFINE_OUTPUT_COPY() to expect bytes
not copied.
Oddly enough DEFINE_OUTPUT_COPY() already returned bytes not copied
while expecting its worker functions to return bytes copied.
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Acked-by: will.deacon@arm.com
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20131030201622.GR16117@laptop.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Vince's fuzzer once again found holes. This time it spotted a leak in
the locked page accounting.
When an event had redirected output and its close() was the last
reference to the buffer we didn't have a vm context to undo accounting.
Change the code to destroy the buffer on the last munmap() and detach
all redirected events at that time. This provides us the right context
to undo the vm accounting.
Reported-and-tested-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20130604084421.GI8923@twins.programming.kicks-ass.net
Cc: <stable@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Vince reported a problem found by his perf specific trinity
fuzzer.
Al noticed 2 problems with perf's mmap():
- it has issues against fork() since we use vma->vm_mm for accounting.
- it has an rb refcount leak on double mmap().
We fix the issues against fork() by using VM_DONTCOPY; I don't
think there's code out there that uses this; we didn't hear
about weird accounting problems/crashes. If we do need this to
work, the previously proposed VM_PINNED could make this work.
Aside from the rb reference leak spotted by Al, Vince's example
prog was indeed doing a double mmap() through the use of
perf_event_set_output().
This exposes another problem, since we now have 2 events with
one buffer, the accounting gets screwy because we account per
event. Fix this by making the buffer responsible for its own
accounting.
Reported-by: Vince Weaver <vincent.weaver@maine.edu>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Link: http://lkml.kernel.org/r/20130528085548.GA12193@twins.programming.kicks-ass.net
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This patch fixes a flaw in perf_output_space(). In case the size
of the space needed is bigger than the actual buffer size, there
may be situations where the function would return true (i.e.,
there is space) when it should not. head > offset due to
rounding of the masking logic.
The problem can be tested by activating BTS on Intel processors.
A BTS record can be as big as 16 pages. The following command
fails:
$ perf record -m 4 -c 1 -e branches:u my_test_program
You will get a buffer corruption with this. Perf report won't be
able to parse the perf.data.
The fix is to first check that the requested space is smaller
than the buffer size. If so, then the masking logic will work
fine. If not, then there is no chance the record can be saved
and it will be gracefully handled by upper code layers.
[ In v2, we also make the logic for the writable more explicit by
renaming it to rb->overwrite because it tells whether or not the
buffer can overwrite its tail (suggested by PeterZ). ]
Signed-off-by: Stephane Eranian <eranian@google.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: peterz@infradead.org
Cc: jolsa@redhat.com
Cc: fweisbec@gmail.com
Link: http://lkml.kernel.org/r/20130318133327.GA3056@quad
Signed-off-by: Ingo Molnar <mingo@kernel.org>
A few events are interesting not only for a current task.
For example, sched_stat_* events are interesting for a task
which wakes up. For this reason, it will be good if such
events will be delivered to a target task too.
Now a target task can be set by using __perf_task().
The original idea and a draft patch belongs to Peter Zijlstra.
I need these events for profiling sleep times. sched_switch is used for
getting callchains and sched_stat_* is used for getting time periods.
These events are combined in user space, then it can be analyzed by
perf tools.
Inspired-by: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Arun Sharma <asharma@fb.com>
Signed-off-by: Andrew Vagin <avagin@openvz.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1342016098-213063-1-git-send-email-avagin@openvz.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
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>
The nmi parameter indicated if we could do wakeups from the current
context, if not, we would set some state and self-IPI and let the
resulting interrupt do the wakeup.
For the various event classes:
- hardware: nmi=0; PMI is in fact an NMI or we run irq_work_run from
the PMI-tail (ARM etc.)
- tracepoint: nmi=0; since tracepoint could be from NMI context.
- software: nmi=[0,1]; some, like the schedule thing cannot
perform wakeups, and hence need 0.
As one can see, there is very little nmi=1 usage, and the down-side of
not using it is that on some platforms some software events can have a
jiffy delay in wakeup (when arch_irq_work_raise isn't implemented).
The up-side however is that we can remove the nmi parameter and save a
bunch of conditionals in fast paths.
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: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Don Zickus <dzickus@redhat.com>
Link: http://lkml.kernel.org/n/tip-agjev8eu666tvknpb3iaj0fg@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
