PERF_SAMPLE_CPU contains the cpu number in the first 4 bytes and the
second 4 bytes are reserved. Ensure the reserved bytes are zero in
perf_event__synthesize_sample().
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Link: http://lkml.kernel.org/r/1516108492-21401-3-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Both 'perf inject' and internal tools consume cpu endian samples, so
there is never a need to do any swapping when synthesizing samples.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Link: http://lkml.kernel.org/r/1516108492-21401-2-git-send-email-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
In x86 architecture dependend part function get_cpuid_str() mallocs a
128 byte buffer, but does not check if the memory allocation succeeded
or not.
When the memory allocation fails, function __get_cpuid() is called with
first parameter being a NULL pointer. However this function references
its first parameter and operates on a NULL pointer which might cause
core dumps.
Signed-off-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Link: http://lkml.kernel.org/r/20180117131611.34319-1-tmricht@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previously it was only allowed to use at most 10 time slices in 'perf
script --time'.
This patch removes this limitation.
For example, following command line is OK (12 time slices)
perf script --time 1%/1,1%/2,1%/3,1%/4,1%/5,1%/6,1%/7,1%/8,1%/9,1%/10,1%/11,1%/12
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-9-git-send-email-yao.jin@linux.intel.com
[ No need to check for NULL to call free, use zfree ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previously it was only allowed to use at most 10 time slices in 'perf
report --time'.
This patch removes this limitation.
For example, following command line is OK (12 time slices)
perf report --stdio --time 1%/1,1%/2,1%/3,1%/4,1%/5,1%/6,1%/7,1%/8,1%/9,1%/10,1%/11,1%/12
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-8-git-send-email-yao.jin@linux.intel.com
[ No need to check for NULL to call free, use zfree ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previously we use a magic number 10 to limit the number of time slices.
It's not very good.
This patch creates a new function perf_time__range_alloc() to allocate
time slices buffer. The number of buffer entries is determined by the
number of comma in string but at least it will allocate one entry even
if no comma is found.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-7-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a time slices indication to the perf report header.
For example,
# perf report --stdio --time 10%
# Total Lost Samples: 0
#
# Samples: 9K of event 'cycles:ppp' (time slices: 10%)
# Event count (approx.): 8951288803
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Suggested--by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-6-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previously, the time percent slice needs an index to specify which one
the user wants.
It may be easier to use if the index can be omitted. So with this
patch, for example,
perf report --stdio --time 10%/1 should be equivalent to
perf report --stdio --time 10%
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-5-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The command line like 'perf report --stdio --time 1abc%/1' could be
accepted by perf. It looks not very good.
This patch uses strtod() to replace original atof() and check the entire
string. Now for the same command line, it would return error message
"Invalid time string".
root@skl:/tmp# perf report --stdio --time 1abc%/1
Invalid time string
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-4-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The following message will be returned to user when executing 'perf
script --time' if perf data file doesn't contain the first/last sample
time.
"HINT: no first/last sample time found in perf data.
Please use latest perf binary to execute 'perf record'
(if '--buildid-all' is enabled, needs to set '--timestamp-boundary')."
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-3-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The following message will be returned to user when executing
'perf report --time' if perf data file doesn't contain the
first/last sample time.
"HINT: no first/last sample time found in perf data.
Please use latest perf binary to execute 'perf record'
(if '--buildid-all' is enabled, needs to set '--timestamp-boundary')."
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1515596433-24653-2-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When we use a global DWARF setting as in:
perf record --call-graph dwarf
According to 5c0cf22477 ("perf record: Store data mmaps for dwarf unwind") we need
to set up some extra perf_event_attr bits.
But when we instead do a per event dwarf setting:
perf record -e cycles/call-graph=dwarf/
This was not being done, make them equivalent.
This didn't produce any output changes in my tests while fixing up loose
ends in the per-event settings, I found it just by comparing the
perf_event_attr fields trying to find an explanation for those problems.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Noel Grandin <noelgrandin@gmail.com>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-6476r53h2o38skbs9qa4ust4@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
If we use:
perf trace --max-stack=4
then the syscall events will use DWARF callchains, when available
(libunwind enabled in the build) and the printing will stop at 4 levels.
When we introduced support for tracepoint events this ended up not
applying for them, fix it.
Before:
# perf trace --call-graph=dwarf --no-syscalls -e probe_libc:inet_pton ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.058 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.058/0.058/0.058/0.000 ms
0.000 probe_libc:inet_pton:(7fc6c2a16350))
#
After:
# perf trace --call-graph=dwarf --no-syscalls -e probe_libc:inet_pton ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.087 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.087/0.087/0.087/0.000 ms
0.000 probe_libc:inet_pton:(7fbf9a041350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffaa947cb67f3f] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa947cb68379] (/usr/bin/ping)
#
Reported-by: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-afsu9eegd43ppihiuafhh9qv@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When setting up DWARF callchains on specific events, without using
'record' or 'trace' --call-graph, but instead doing it like:
perf trace -e cycles/call-graph=dwarf/
The unwind__prepare_access() call in thread__insert_map() when we
process PERF_RECORD_MMAP(2) metadata events were not being performed,
precluding us from using per-event DWARF callchains, handling them just
when we asked for all events to be DWARF, using "--call-graph dwarf".
We do it in the PERF_RECORD_MMAP because we have to look at one of the
executable maps to figure out the executable type (64-bit, 32-bit) of
the DSO laid out in that mmap. Also to look at the architecture where
the perf.data file was recorded.
All this probably should be deferred to when we process a sample for
some thread that has callchains, so that we do this processing only for
the threads with samples, not for all of them.
For now, fix using DWARF on specific events.
Before:
# perf trace --no-syscalls -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.048 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.048/0.048/0.048/0.000 ms
0.000 probe_libc:inet_pton:(7fe9597bb350))
Problem processing probe_libc:inet_pton callchain, skipping...
#
After:
# perf trace --no-syscalls -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.060 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.060/0.060/0.060/0.000 ms
0.000 probe_libc:inet_pton:(7fd4aa930350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffaa804e51af3f] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa804e51b379] (/usr/bin/ping)
#
# perf trace --call-graph=dwarf --no-syscalls -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.057 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.057/0.057/0.057/0.000 ms
0.000 probe_libc:inet_pton:(7f9363b9e350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffa9e8a14e0f3f] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffa9e8a14e1379] (/usr/bin/ping)
#
# perf trace --call-graph=fp --no-syscalls -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.077 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.077/0.077/0.077/0.000 ms
0.000 probe_libc:inet_pton:(7f4947e1c350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffaa716d88ef3f] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa716d88f379] (/usr/bin/ping)
#
# perf trace --no-syscalls -e probe_libc:inet_pton/call-graph=fp/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.078 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.078/0.078/0.078/0.000 ms
0.000 probe_libc:inet_pton:(7fa157696350))
__GI___inet_pton (/usr/lib64/libc-2.26.so)
getaddrinfo (/usr/lib64/libc-2.26.so)
[0xffffa9ba39c74f40] (/usr/bin/ping)
#
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/r/20180116182650.GE16107@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When setting the "dwarf" unwinder for a specific event and not
specifying the max-stack, the attr.sample_max_stack ended up using an
uninitialized callchain_param.max_stack, fix it by using designated
initializers for that callchain_param variable, zeroing all non
explicitely initialized struct members.
Here is what happened:
# perf trace -vv --no-syscalls --max-stack 4 -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
callchain: type DWARF
callchain: stack dump size 8192
perf_event_attr:
type 2
size 112
config 0x730
{ sample_period, sample_freq } 1
sample_type IP|TID|TIME|ADDR|CALLCHAIN|CPU|PERIOD|RAW|REGS_USER|STACK_USER|DATA_SRC
exclude_callchain_user 1
{ wakeup_events, wakeup_watermark } 1
sample_regs_user 0xff0fff
sample_stack_user 8192
sample_max_stack 50656
sys_perf_event_open failed, error -75
Value too large for defined data type
# perf trace -vv --no-syscalls --max-stack 4 -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
callchain: type DWARF
callchain: stack dump size 8192
perf_event_attr:
type 2
size 112
config 0x730
sample_type IP|TID|TIME|ADDR|CALLCHAIN|CPU|PERIOD|RAW|REGS_USER|STACK_USER|DATA_SRC
exclude_callchain_user 1
sample_regs_user 0xff0fff
sample_stack_user 8192
sample_max_stack 30448
sys_perf_event_open failed, error -75
Value too large for defined data type
#
Now the attr.sample_max_stack is set to zero and the above works as
expected:
# perf trace --no-syscalls --max-stack 4 -e probe_libc:inet_pton/call-graph=dwarf/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.072 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.072/0.072/0.072/0.000 ms
0.000 probe_libc:inet_pton:(7feb7a998350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffaa39b6108f3f] (/usr/bin/ping)
#
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-is9tramondqa9jlxxsgcm9iz@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
'perf record' and 'perf report --dump-raw-trace' supported in this
release.
Example usage:
# perf record -e arm_spe/ts_enable=1,pa_enable=1/ dd if=/dev/zero of=/dev/null count=10000
# perf report --dump-raw-trace
Note that the perf.data file is portable, so the report can be run on
another architecture host if necessary.
Output will contain raw SPE data and its textual representation, such
as:
0x5c8 [0x30]: PERF_RECORD_AUXTRACE size: 0x200000 offset: 0 ref: 0x1891ad0e idx: 1 tid: 2227 cpu: 1
.
. ... ARM SPE data: size 2097152 bytes
. 00000000: 49 00 LD
. 00000002: b2 c0 3b 29 0f 00 00 ff ff VA 0xffff00000f293bc0
. 0000000b: b3 c0 eb 24 fb 00 00 00 80 PA 0xfb24ebc0 ns=1
. 00000014: 9a 00 00 LAT 0 XLAT
. 00000017: 42 16 EV RETIRED L1D-ACCESS TLB-ACCESS
. 00000019: b0 00 c4 15 08 00 00 ff ff PC 0xff00000815c400 el3 ns=1
. 00000022: 98 00 00 LAT 0 TOT
. 00000025: 71 36 6c 21 2c 09 00 00 00 TS 39395093558
. 0000002e: 49 00 LD
. 00000030: b2 80 3c 29 0f 00 00 ff ff VA 0xffff00000f293c80
. 00000039: b3 80 ec 24 fb 00 00 00 80 PA 0xfb24ec80 ns=1
. 00000042: 9a 00 00 LAT 0 XLAT
. 00000045: 42 16 EV RETIRED L1D-ACCESS TLB-ACCESS
. 00000047: b0 f4 11 16 08 00 00 ff ff PC 0xff0000081611f4 el3 ns=1
. 00000050: 98 00 00 LAT 0 TOT
. 00000053: 71 36 6c 21 2c 09 00 00 00 TS 39395093558
. 0000005c: 48 00 INSN-OTHER
. 0000005e: 42 02 EV RETIRED
. 00000060: b0 2c ef 7f 08 00 00 ff ff PC 0xff0000087fef2c el3 ns=1
. 00000069: 98 00 00 LAT 0 TOT
. 0000006c: 71 d1 6f 21 2c 09 00 00 00 TS 39395094481
...
Other release notes:
- applies to acme's perf/{core,urgent} branches, likely elsewhere
- Report is self-contained within the tool.
Record requires enabling the kernel SPE driver by
setting CONFIG_ARM_SPE_PMU.
- The intel-bts implementation was used as a starting point; its
min/default/max buffer sizes and power of 2 pages granularity need to be
revisited for ARM SPE
- Recording across multiple SPE clusters/domains not supported
- Snapshot support (record -S), and conversion to native perf events
(e.g., via 'perf inject --itrace'), are also not supported
- Technically both cs-etm and spe can be used simultaneously, however
disabled for simplicity in this release
Signed-off-by: Kim Phillips <kim.phillips@arm.com>
Reviewed-by: Dongjiu Geng <gengdongjiu@huawei.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Pawel Moll <pawel.moll@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rob Herring <robh@kernel.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Wang Nan <wangnan0@huawei.com>
Cc: Will Deacon <will.deacon@arm.com>
Link: http://lkml.kernel.org/r/20180114132850.0b127434b704a26bad13268f@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The raw_syscalls:sys_{enter,exit} were first supported in 'perf trace',
together with minor and major page faults, then we supported
--call-graph, then --max-stack, but when the other tracepoints got
supported, and bpf, etc, I forgot to make those global call-graph
settings apply to them.
Fix it by realizing that the global --max-stack and --call-graph
settings are done via:
OPT_CALLBACK(0, "call-graph", &trace.opts,
"record_mode[,record_size]", record_callchain_help,
&record_parse_callchain_opt),
And then, when we go to parse the events in -e via:
OPT_CALLBACK('e', "event", &trace, "event",
"event/syscall selector. use 'perf list' to list available events",
trace__parse_events_option),
And trace__parse_sevents_option() calls:
struct option o = OPT_CALLBACK('e', "event", &trace->evlist, "event",
"event selector. use 'perf list' to list available events",
parse_events_option);
err = parse_events_option(&o, lists[0], 0);
parse_events_option() will override the global --call-graph and
--max-stack if the "call-graph" and/or "max-stack" terms are in the
event definition, such as in the probe_libc:inet_pton event in one of the
examples below (-e probe_libc:inet_pton/max-stack=2).
Before:
# perf trace --mmap 1024 --call-graph dwarf -e sendto,probe_libc:inet_pton ping -6 -c 1 ::1
1.525 ( ): probe_libc:inet_pton:(7f77f3ac9350))
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.071 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.071/0.071/0.071/0.000 ms
1.677 ( 0.081 ms): ping/31296 sendto(fd: 3, buff: 0x55681b652720, len: 64, addr: 0x55681b650640, addr_len: 28) = 64
__libc_sendto (/usr/lib64/libc-2.26.so)
[0xffffaa97e4bc9cef] (/usr/bin/ping)
[0xffffaa97e4bc656d] (/usr/bin/ping)
[0xffffaa97e4bc7d0a] (/usr/bin/ping)
[0xffffaa97e4bca447] (/usr/bin/ping)
[0xffffaa97e4bc2f91] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa97e4bc3379] (/usr/bin/ping)
#
After:
# perf trace --mmap 1024 --call-graph dwarf -e sendto,probe_libc:inet_pton ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.089 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.089/0.089/0.089/0.000 ms
1.955 ( ): probe_libc:inet_pton:(7f383a311350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
[0xffffaa5d91444f3f] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa5d91445379] (/usr/bin/ping)
2.140 ( 0.101 ms): ping/32047 sendto(fd: 3, buff: 0x55a26edd0720, len: 64, addr: 0x55a26edce640, addr_len: 28) = 64
__libc_sendto (/usr/lib64/libc-2.26.so)
[0xffffaa5d9144bcef] (/usr/bin/ping)
[0xffffaa5d9144856d] (/usr/bin/ping)
[0xffffaa5d91449d0a] (/usr/bin/ping)
[0xffffaa5d9144c447] (/usr/bin/ping)
[0xffffaa5d91444f91] (/usr/bin/ping)
__libc_start_main (/usr/lib64/libc-2.26.so)
[0xffffaa5d91445379] (/usr/bin/ping)
#
Same thing for --max-stack, the global one:
# perf trace --max-stack 3 -e sendto,probe_libc:inet_pton ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.097 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.097/0.097/0.097/0.000 ms
1.577 ( ): probe_libc:inet_pton:(7f32f3957350))
__inet_pton (inlined)
gaih_inet.constprop.7 (/usr/lib64/libc-2.26.so)
__GI_getaddrinfo (inlined)
1.738 ( 0.108 ms): ping/32103 sendto(fd: 3, buff: 0x55c3132d7720, len: 64, addr: 0x55c3132d5640, addr_len: 28) = 64
__libc_sendto (/usr/lib64/libc-2.26.so)
[0xffffaa3cecf44cef] (/usr/bin/ping)
[0xffffaa3cecf4156d] (/usr/bin/ping)
#
And then setting up a global setting (dwarf, max-stack=4), that will
affect the raw_syscall:sys_enter for the 'sendto' syscall and that will
be overriden in the probe_libc:inet_pton call to just one entry.
# perf trace --max-stack=4 --call-graph dwarf -e sendto -e probe_libc:inet_pton/max-stack=1/ ping -6 -c 1 ::1
PING ::1(::1) 56 data bytes
64 bytes from ::1: icmp_seq=1 ttl=64 time=0.090 ms
--- ::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.090/0.090/0.090/0.000 ms
2.140 ( ): probe_libc:inet_pton:(7f9fe9337350))
__GI___inet_pton (/usr/lib64/libc-2.26.so)
2.283 ( 0.103 ms): ping/31804 sendto(fd: 3, buff: 0x55c7f3e19720, len: 64, addr: 0x55c7f3e17640, addr_len: 28) = 64
__libc_sendto (/usr/lib64/libc-2.26.so)
[0xffffaa380c402cef] (/usr/bin/ping)
[0xffffaa380c3ff56d] (/usr/bin/ping)
[0xffffaa380c400d0a] (/usr/bin/ping)
#
Install iputils-debuginfo to get those /usr/bin/ping addresses resolved,
those routines are not on its .dymsym nor .symtab :-)
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-qgl2gse8elhh9zztw4ajopg3@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The construct:
if (callchain_param)
perf_evsel__config_callchain(evsel, opts, &callchain_param);
happens in several places, so make perf_evsel__config_callchain() work
just like free(NULL), do nothing if param->enabled is not set.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-ykk0qzxnxwx3o611ctjnmxav@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
We need to increase output offset in each iteration, not decrease it as
we currently do.
I guess we were lucky to finish in most cases in first iteration, so the
bug never showed. However it shows a lot when working with big (~4GB)
size data.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Fixes: 9c9f5a2f19 ("perf tools: Introduce copyfile_offset() function")
Link: http://lkml.kernel.org/r/20180109133923.25406-1-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Since 75562573ba ("perf tools: Add support for
PERF_SAMPLE_IDENTIFIER") we don't need explicitely set
PERF_SAMPLE_IDENTIFIER, as perf_evlist__config() will do this for us,
i.e. when there are more than one evsel in an evlist, it will check if
some evsel has a sample_type different than the one on the first evsel
in the list, setting PERF_SAMPLE_IDENTIFIER in that case.
So, to simplify 'perf trace' codebase, ditch that check.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hendrick Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Thomas Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-12xq6orhwttee2tdtu96ucrp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
There could be different types of memory in the system. E.g normal
System Memory, Persistent Memory. To understand how the workload maps to
those memories, it's important to know the I/O statistics of them. Perf
can collect physical addresses, but those are raw data. It still needs
extra work to resolve the physical addresses. Provide a script to
facilitate the physical addresses resolving and I/O statistics.
Profile with MEM_INST_RETIRED.ALL_LOADS or MEM_UOPS_RETIRED.ALL_LOADS
event if any of them is available.
Look up the /proc/iomem and resolve the physical address. Provide
memory type summary.
Here is an example output:
# perf script report mem-phys-addr
Event: mem_inst_retired.all_loads:P
Memory type count percentage
---------------------------------------- ----------- -----------
System RAM 74 53.2%
Persistent Memory 55 39.6%
N/A
---
Changes since V2:
- Apply the new license rules.
- Add comments for globals
Changes since V1:
- Do not mix DLA and Load Latency. Do not compare the loads and stores.
Only profile the loads.
- Use event name to replace the RAW event
Signed-off-by: Kan Liang <Kan.liang@intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Philippe Ombredanne <pombredanne@nexb.com>
Cc: Stephane Eranian <eranian@google.com>
Link: https://lkml.kernel.org/r/1515099595-34770-1-git-send-email-kan.liang@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The trailing semicolon is an empty statement that does no operation.
Removing it since it doesn't do anything.
Signed-off-by: Luis de Bethencourt <luisbg@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Joe Perches <joe@perches.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20180111155020.9782-1-luisbg@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Commit ("d0565132605f perf evsel: Enable type checking for
perf_evsel_config_term types") assumes PERF_EVSEL__CONFIG_TERM_DRV_CFG
isn't used and as such adds a BUG_ON().
Since the enumeration type is used in macro ADD_CONFIG_TERM() the change
break CoreSight trace acquisition.
This patch restores the original code.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Fixes: d056513260 ("perf evsel: Enable type checking for perf_evsel_config_term types")
Link: http://lkml.kernel.org/r/1515617211-32024-1-git-send-email-mathieu.poirier@linaro.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Similar to --tasks, producing the same output plus /proc/<PID>/maps
similar lines for each mmap record present in a perf.data file.
Please note that not all mmaps are stored, for instance, some of the
non-executable mmaps are only stored when 'perf record --data' is used,
when the user wants to resolve data accesses in addition to asking for
executable mmaps to get the DSO with symtabs.
E.g.:
# perf record sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.018 MB perf.data (7 samples) ]
[root@jouet ~]# perf report --mmaps
# pid tid ppid comm
0 0 -1 |swapper
4137 4137 -1 |sleep
5628a35a1000-5628a37aa000 r-xp 00000000 3147148 /usr/bin/sleep
7fb65ad51000-7fb65b134000 r-xp 00000000 3149795 /usr/lib64/libc-2.26.so
7fb65b134000-7fb65b35e000 r-xp 00000000 3149715 /usr/lib64/ld-2.26.so
7ffd94b9f000-7ffd94ba1000 r-xp 00000000 0 [vdso]
#
# perf record sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.019 MB perf.data (8 samples) ]
# perf report --mmaps
# pid tid ppid comm
0 0 -1 |swapper
4161 4161 -1 |sleep
55afae69a000-55afae8a3000 r-xp 00000000 3147148 /usr/bin/sleep
7f569f00d000-7f569f3f0000 r-xp 00000000 3149795 /usr/lib64/libc-2.26.so
7f569f3f0000-7f569f61a000 r-xp 00000000 3149715 /usr/lib64/ld-2.26.so
7fff6fffe000-7fff70000000 r-xp 00000000 0 [vdso]
#
# perf record time sleep 1
0.00user 0.00system 0:01.00elapsed 0%CPU (0avgtext+0avgdata 2156maxresident)k
0inputs+0outputs (0major+73minor)pagefaults 0swaps
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.019 MB perf.data (14 samples) ]
# perf report --mmaps
# pid tid ppid comm
0 0 -1 |swapper
4281 4281 -1 |time
560560dca000-560560fcf000 r-xp 00000000 3190458 /usr/bin/time
7fc175196000-7fc175579000 r-xp 00000000 3149795 /usr/lib64/libc-2.26.so
7fc175579000-7fc1757a3000 r-xp 00000000 3149715 /usr/lib64/ld-2.26.so
7ffc924f6000-7ffc924f8000 r-xp 00000000 0 [vdso]
4282 4282 4281 | sleep
560560dca000-560560fcf000 r-xp 00000000 3190458 /usr/bin/time
564b4de3c000-564b4e045000 r-xp 00000000 3147148 /usr/bin/sleep
7f6a5a716000-7f6a5aaf9000 r-xp 00000000 3149795 /usr/lib64/libc-2.26.so
7f6a5aaf9000-7f6a5ad23000 r-xp 00000000 3149715 /usr/lib64/ld-2.26.so
7fc175196000-7fc175579000 r-xp 00000000 3149795 /usr/lib64/libc-2.26.so
7fc175579000-7fc1757a3000 r-xp 00000000 3149715 /usr/lib64/ld-2.26.so
7ffc924f6000-7ffc924f8000 r-xp 00000000 0 [vdso]
7ffcec7e6000-7ffcec7e8000 r-xp 00000000 0 [vdso]
#
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-zulwdlg5rfowogr1qznorvvc@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add --stats option to display quick data statistics of event numbers,
without any further processing, like the one at the end of the perf
report -D command.
$ perf report --stat
Aggregated stats:
TOTAL events: 4566
MMAP events: 113
LOST events: 19
COMM events: 3
FORK events: 400
SAMPLE events: 3315
MMAP2 events: 32
FINISHED_ROUND events: 681
THREAD_MAP events: 1
CPU_MAP events: 1
TIME_CONV events: 1
I found this useful when hunting lost events for another change.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20180107160356.28203-12-jolsa@kernel.org
[ Rename it to --stats, plural ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
I want to display the pure events status coming in the next patch and
the tool's warnings are superfluous in the output. Making it optional,
enabled by default.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20180107160356.28203-11-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Adding support to display sample misc field in form
of letter for each bit:
# perf script -F +misc ...
sched-messaging 1414 K 28690.636582: 4590 cycles ...
sched-messaging 1407 U 28690.636600: 325620 cycles ...
sched-messaging 1414 K 28690.636608: 19473 cycles ...
misc field __________/
The misc bits are assigned to following letters:
PERF_RECORD_MISC_KERNEL K
PERF_RECORD_MISC_USER U
PERF_RECORD_MISC_HYPERVISOR H
PERF_RECORD_MISC_GUEST_KERNEL G
PERF_RECORD_MISC_GUEST_USER g
PERF_RECORD_MISC_MMAP_DATA* M
PERF_RECORD_MISC_COMM_EXEC E
PERF_RECORD_MISC_SWITCH_OUT S
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20180107160356.28203-9-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
There's no reason anymore to treat babel trace in a special way, because
a) we no longer display its state b) the needed babeltrace library is
now out and well adopted among distros.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20180107160356.28203-2-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf script has a --time option to limit the time range of output. It
only supports absolute time.
Now this option is extended to support multiple time ranges and support
the percent of time.
For example:
1. Select the first and second 10% time slices:
perf script --time 10%/1,10%/2
2. Select from 0% to 10% and 30% to 40% slices:
perf script --time 0%-10%,30%-40%
Changelog:
v6: Fix the merge issue with latest perf/core branch.
No functional changes.
v5: Add checking of first/last sample time to detect if it's recorded
in perf.data. If it's not recorded, returns error message to user.
v4: Remove perf_time__skip_sample, only uses perf_time__ranges_skip_sample
v3: Since the definitions of first_sample_time/last_sample_time
are moved from perf_session to perf_evlist so change the
related code.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-7-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf report has a --time option to limit the time range of output. It
only supports absolute time.
Now this option is extended to support multiple time ranges and support
the percent of time.
For example:
1. Select the first and second 10% time slices:
perf report --time 10%/1,10%/2
2. Select from 0% to 10% and 30% to 40% slices:
perf report --time 0%-10%,30%-40%
Changelog:
v6: Fix the merge issue with latest perf/core branch.
No functional changes.
v5: Add checking of first/last sample time to detect if it's recorded
in perf.data. If it's not recorded, returns error message to user.
v4: Remove perf_time__skip_sample, only uses perf_time__ranges_skip_sample
v3: Since the definitions of first_sample_time/last_sample_time
are moved from perf_session to perf_evlist so change the
related code.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-6-git-send-email-yao.jin@linux.intel.com
[ Add missing colons at end of examples in the man page ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previous patch supports the multiple time range.
For example, select the first and second 10% time slices.
perf report --time 10%/1,10%/2
We need a function to check if a timestamp is in the ranges of
[0, 10%) and [10%, 20%].
Note that it includes the last element in [10%, 20%] but it doesn't
include the last element in [0, 10%). It's to avoid the overlap.
This patch implments a new function perf_time__ranges_skip_sample
for this checking.
Change log:
v4: Let perf_time__ranges_skip_sample be compatible with
perf_time__skip_sample when only one time range.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-5-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Current perf report/script/... have a --time option to limit the time
range of output. But right now it only supports absolute time, add
support for time percentage.
For example:
1. Select the second 10% time slice
perf report --time 10%/2
2. Select from 0% to 10% time slice
perf report --time 0%-10%
It also support the multiple time ranges.
3. Select the first and second 10% time slices
perf report --time 10%/1,10%/2
4. Select from 0% to 10% and 30% to 40% slices
perf report --time 0%-10%,30%-40%
Changelog:
v4: An issue is found. Following passes.
perf script --time 10%/10x12321xsdfdasfdsafdsafdsa
Now it uses strtol to replace atoi.
Committer notes:
This just puts in place the infrastructure, so the examples in this cset
comment will only work later, after more patches in this series are
applied.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-4-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
In the default 'perf record' configuration, all samples are processed,
to create the HEADER_BUILD_ID table. So it's very easy to get the
first/last samples and save the time to perf file header via the
function write_sample_time().
Later, at post processing time, perf report/script will fetch the time
from perf file header.
Committer testing:
# perf record -a sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 2.099 MB perf.data (1101 samples) ]
[root@jouet home]# perf report --header | grep "time of "
# time of first sample : 22947.909226
# time of last sample : 22948.910704
#
# perf report -D | grep PERF_RECORD_SAMPLE\(
0 22947909226101 0x20bb68 [0x30]: PERF_RECORD_SAMPLE(IP, 0x4001): 0/0: 0xffffffffa21b1af3 period: 1 addr: 0
0 22947909229928 0x20bb98 [0x30]: PERF_RECORD_SAMPLE(IP, 0x4001): 0/0: 0xffffffffa200d204 period: 1 addr: 0
<SNIP>
3 22948910397351 0x219360 [0x30]: PERF_RECORD_SAMPLE(IP, 0x4001): 28251/28251: 0xffffffffa22071d8 period: 169518 addr: 0
0 22948910652380 0x20f120 [0x30]: PERF_RECORD_SAMPLE(IP, 0x4001): 0/0: 0xffffffffa2856816 period: 198807 addr: 0
2 22948910704034 0x2172d0 [0x30]: PERF_RECORD_SAMPLE(IP, 0x4001): 0/0: 0xffffffffa2856816 period: 88111 addr: 0
#
Changelog:
v7: Just update the patch description according to Arnaldo's suggestion.
v6: Currently '--buildid-all' is not enabled at default. So the walking
on all samples is the default operation. There is no big overhead
to calculate the timestamp boundary in process_sample_event handler
once we already go through all samples. So the timestamp boundary
calculation is enabled by default when '--buildid-all' is not enabled.
While if '--buildid-all' is enabled, we creates a new option
"--timestamp-boundary" for user to decide if it enables the
timestamp boundary calculation.
v5: There is an issue that the sample walking can only work when
'--buildid-all' is not enabled. So we need to let the walking
be able to work even if '--buildid-all' is enabled and let the
processing skips the dso hit marking for this case.
At first, I want to provide a new option "--record-time-boundaries".
While after consideration, I think a new option is not very
necessary.
v3: Remove the definitions of first_sample_time and last_sample_time
from struct record and directly save them in perf_evlist.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-3-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf report/script/... have a --time option to limit the time range of
output. That's very useful to slice large traces, e.g. when processing
the output of perf script for some analysis.
But right now --time only supports absolute time. Also there is no fast
way to get the start/end times of a given trace except for looking at
it. This makes it hard to e.g. only decode the first half of the trace,
which is useful for parallelization of scripts
Another problem is that perf records are variable size and there is no
synchronization mechanism. So the only way to find the last sample
reliably would be to walk all samples. But we want to avoid that in perf
report/... because it is already quite expensive. That is why storing
the first sample time and last sample time in perf record is better.
This patch creates a new header feature type HEADER_SAMPLE_TIME and
related ops. Save the first sample time and the last sample time to the
feature section in perf file header. That will be done when, for
instance, processing build-ids, where we already have to process all
samples to create the build-id table, take advantage of that to further
amortize that processing by storing HEADER_SAMPLE_TIME to make 'perf
report/script' faster when using --time.
Committer testing:
After this patch is applied the header is written with zeroes, we need
the next patch, for "perf record" to actually write the timestamps:
# perf report -D | grep PERF_RECORD_SAMPLE\(
22501155244406 0x44f0 [0x28]: PERF_RECORD_SAMPLE(IP, 0x4001): 25016/25016: 0xffffffffa21be8c5 period: 1 addr: 0
<SNIP>
22501155793625 0x4a30 [0x28]: PERF_RECORD_SAMPLE(IP, 0x4001): 25016/25016: 0xffffffffa21ffd50 period: 2828043 addr: 0
# perf report --header | grep "time of "
# time of first sample : 0.000000
# time of last sample : 0.000000
#
Changelog:
v7: 1. Rebase to latest perf/core branch.
2. Add following clarification in patch description according to
Arnaldo's suggestion.
"That will be done when, for instance, processing build-ids,
where we already have to process all samples to create the
build-id table, take advantage of that to further amortize
that processing by storing HEADER_SAMPLE_TIME to make
'perf report/script' faster when using --time."
v4: Use perf script time style for timestamp printing. Also add with
the printing of sample duration.
v3: Remove the definitions of first_sample_time/last_sample_time from
perf_session. Just define them in perf_evlist
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1512738826-2628-2-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When enabling '-b' option in perf record, for example,
perf record -b ...
perf report
and then browsing the annotate browser from perf report (press 'A'), it
would fail (annotate browser can't be displayed).
It's because the '.add_entry_cb' op of struct report is overwritten by
hist_iter__branch_callback() in builtin-report.c. But this function doesn't do
something like mapping symbols and sources. So next, do_annotate() will return
directly.
notes = symbol__annotation(act->ms.sym);
if (!notes->src)
return 0;
This patch adds the lost code to hist_iter__branch_callback (refer to
hist_iter__report_callback).
v2:
Fix a crash bug when perform 'perf report --stdio'.
The reason is that we init the symbol annotation only in browser mode, it
doesn't allocate/init resources for stdio mode.
So now in hist_iter__branch_callback(), it will return directly if it's not in
browser mode.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1514284963-18587-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When a valid vmlinux is not found, 'perf report' falls back to look at
/proc/kcore. In this case, it will report the impossible large offset.
For example:
# perf record -b -e cycles:k find /etc/ > /dev/null
# perf report --stdio --branch-history
22.77% _vm_normal_page+18446603336221188162
|
---page_remove_rmap +18446603336221188324
page_remove_rmap +18446603336221188487 (cycles:5)
unlock_page_memcg +18446603336221188096
page_remove_rmap +18446603336221188327 (cycles:1)
The issue is the value which is passed to parameter 'addr' in
__get_srcline() is the objdump address. It's not correct if we calculate
the offset by using 'addr - sym->start'.
This patch creates a new parameter 'ip' in __get_srcline(). It is not
converted to objdump address.
With this patch, the perf report output is:
22.77% _vm_normal_page+66
|
---page_remove_rmap +228
page_remove_rmap +391 (cycles:5)
unlock_page_memcg +0
page_remove_rmap +231 (cycles:1)
page_remove_rmap +236
Committer testing:
Make sure you get any valid vmlinux out of the way, using '-v' on the
'perf report' case and deleting it from places where perf searches them,
like your kernel build dir and the build-id cache, in ~/.debug/.
Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1514564812-17344-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Fix a compile error:
...
CC util/libunwind/x86_32.o
In file included from util/libunwind/x86_32.c:33:0:
util/libunwind/../../arch/x86/util/unwind-libunwind.c: In function 'libunwind__x86_reg_id':
util/libunwind/../../arch/x86/util/unwind-libunwind.c:110:11: error: 'EINVAL' undeclared (first use in this function)
return -EINVAL;
^
util/libunwind/../../arch/x86/util/unwind-libunwind.c:110:11: note: each undeclared identifier is reported only once for each function it appears in
mv: cannot stat 'util/libunwind/.x86_32.o.tmp': No such file or directory
make[4]: *** [util/libunwind/x86_32.o] Error 1
make[3]: *** [util] Error 2
make[2]: *** [libperf-in.o] Error 2
make[1]: *** [sub-make] Error 2
make: *** [all] Error 2
It happens when libunwind-x86 feature is detected.
Signed-off-by: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20171206015040.114574-1-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The 'perf test bpf' was hooking a eBPF program on the SyS_epoll_wait()
kernel function, that was what the epoll_wait() glibc function ended up
calling, but since at least glibc 2.26, the one that comes with, for
instance, Fedora 27, glibc ends up calling SyS_epoll_pwait() when
epoll_wait() is used, causing this 'perf test' entry to fail.
So switch to using epoll_pwait() and hook the eBPF program to the
SyS_epoll_pwait() kernel function to make it work on a wider range of
glibc and kernel versions.
Tested-by: Wang Nan <wangnan0@huawei.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-zynvquy63er8s5mrgsz65pto@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
To follow standard practice in the kernel sources, documenting the
initialization better and helping quickly finding the value for some
field in a struct with many entries.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-syn3hz9hz7ukxlxbx5x6hv20@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When failing on one of the BPF tests we were just stating:
BPF filter result incorrect
Add some more info to help figuring out the problem:
BPF filter result incorrect, expected 56, got 0 samples
This came out while investigating this failure, first seen after
updating the kernel to the 4.15.0-rc6 tag:
[root@jouet ~]# perf test bpf
39: BPF filter :
39.1: Basic BPF filtering : FAILED!
39.2: BPF pinning : Skip
39.3: BPF prologue generation: Skip
39.4: BPF relocation checker : Skip
[root@jouet ~]#
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: https://lkml.kernel.org/n/tip-403npu7daupv6b2bmxliv5pk@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
It's a follow up patch for a previous patch "perf tool: Return all
events as auto-completions after comma".
With this patch, auto-completion can work well for events with a ':'.
For example:
root@skl:/tmp# perf stat -e block:block_<TAB>
block:block_bio_backmerge block:block_rq_complete
block:block_bio_bounce block:block_rq_insert
block:block_bio_complete block:block_rq_issue
block:block_bio_frontmerge block:block_rq_remap
block:block_bio_queue block:block_rq_requeue
block:block_bio_remap block:block_sleeprq
block:block_dirty_buffer block:block_split
block:block_getrq block:block_touch_buffer
block:block_plug block:block_unplug
root@skl:/tmp# perf stat -e block:block_rq_<TAB>
block:block_rq_complete block:block_rq_issue block:block_rq_requeue
block:block_rq_insert block:block_rq_remap
root@skl:/tmp# perf stat -e block:block_rq_complete<TAB>
block:block_rq_complete
root@skl:/tmp# perf stat -e block:block_rq_complete
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1513973758-19109-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
It's a follow up for one previous patch "perf tool: Improve bash command
line auto-complete for multiple events with comma."
It fixes an issue that no events are displayed when <TAB> is directly
typed after comma.
With this patch, now the result is:
root@skl:/tmp# perf stat -e cpu-cycles,<TAB>
Display all 2389 possibilities? (y or n)
alarmtimer:alarmtimer_cancel
alarmtimer:alarmtimer_fired
alarmtimer:alarmtimer_start
alarmtimer:alarmtimer_suspend
alignment-faults
arith.divider_active
BAClear_Cost
baclears.any
block:block_bio_backmerge
block:block_bio_bounce
block:block_bio_complete
block:block_bio_frontmerge
block:block_bio_queue
block:block_bio_remap
block:block_dirty_buffer
block:block_getrq
block:block_plug
block:block_rq_complete
block:block_rq_insert
block:block_rq_issue
block:block_rq_remap
block:block_rq_requeue
block:block_sleeprq
--More--
One remaining issue is that the auto-completions doesn't work well
for the event with ':'. For example, clk:clk_enable.
Because ':' is set as WORDBREAK by default in bash. Need more work
for this case.
Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1513940255-16528-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf has perf-completion.sh to define command line auto-completion in
bash/zsh.
For record/stat -e it works for single events, but isn't working when
specifying multiple events with comma.
It would be very useful if it could be fixed to make it easier by
supporting multiple events, comma separated.
With this patch, the result can be like this:
1. Support the events returned from 'perf list --raw-dump'
root@skl:/tmp# perf stat -e cpu/cache<TAB>
cpu/cache-misses/ cpu/cache-references/
root@skl:/tmp# perf stat -e cpu/cache-misses/,cpu/branch-<TAB>
cpu/branch-instructions/ cpu/branch-misses/
root@skl:/tmp# perf stat -e cpu/cache-misses/,cpu/branch-i<TAB>
root@skl:/tmp# perf stat -e cpu/cache-misses/,cpu/branch-instructions/
2. Support the events listed in /sys/bus/event_source/devices/cpu/events
root@skl:/tmp# perf stat -e cycle<TAB>
cycle_activity.cycles_l1d_miss cycle_activity.stalls_l3_miss
cycle_activity.cycles_l2_miss cycle_activity.stalls_mem_any
cycle_activity.cycles_l3_miss cycle_activity.stalls_total
cycle_activity.cycles_mem_any cycles-ct
cycle_activity.stalls_l1d_miss cycles-t
cycle_activity.stalls_l2_miss
root@skl:/tmp# perf stat -e cycles-<TAB>
cycles-ct cycles-t
root@skl:/tmp# perf stat -e cycles-t,cpu/c<TAB>
cpu/cache-misses/ cpu/cpu-cycles/ cpu/cycles-t/
cpu/cache-references/ cpu/cycles-ct/
root@skl:/tmp# perf stat -e cycles-t,cpu/cache-<TAB>
cpu/cache-misses/ cpu/cache-references/
root@skl:/tmp# perf stat -e cycles-t,cpu/cache-misses/
3. Support the uppercase event which is with prefix "cpu/"
root@skl:/tmp# perf stat -e cpu/c<TAB>
cpu/cache-misses/ cpu/cpu-cycles/ cpu/cycles-t/
cpu/cache-references/ cpu/cycles-ct/
root@skl:/tmp# perf stat -e cpu/cache-misses/,cpu/C<TAB>
cpu/CACHE-MISSES/ cpu/CPU-CYCLES/ cpu/CYCLES-T/
cpu/CACHE-REFERENCES/ cpu/CYCLES-CT/
root@skl:/tmp# perf stat -e cpu/cache-misses/,cpu/CACHE-REFERENCES/
Note that:
a) This patch only supports bash.
b) It doesn't support the cases like {},{} or {...,...}.
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1513848370-8098-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
On an arm64 machine running a CONFIG_RANDOMIZE_BASE=y kernel, perf
kernel symbol resolution fails. Debugging saw symsrc_init calling the
default elf__needs_adjust_symbols() where checks for an ET_DYN (3)
ehdr.e_type failed when they should have succeeded.
Fix by adopting powerpc version of the weak elf__needs_adjust_symbols()
function, as done in commit d233209833 ("perf probe ppc: Fix symbol
fixup issues due to ELF type").
Prior to this patch, perf test 1 would fail:
$ sudo oldperf test -v 1 |& head
1: vmlinux symtab matches kallsyms :
test child forked, pid 33374
Looking at the vmlinux_path (8 entries long)
Using /usr/lib/debug/boot/vmlinux for symbols
ERR : 0xfffe0000100f1000: do_undefinstr not on kallsyms
ERR : 0xfffe0000100f1320: do_sysinstr not on kallsyms
ERR : 0xfffe0000100f13b0: do_debug_exception not on kallsyms
ERR : 0xfffe0000100f1498: do_mem_abort not on kallsyms
ERR : 0xfffe0000100f1580: do_sp_pc_abort not on kallsyms
...
After applying this patch, perf test 1 now succeeds:
$ sudo ./newperf test -v 1 |& head
1: vmlinux symtab matches kallsyms :
test child forked, pid 33378
Looking at the vmlinux_path (8 entries long)
Using /usr/lib/debug/boot/vmlinux for symbols
WARN: 0xffff000008081000: diff name v: do_undefinstr k: __exception_text_start
WARN: 0xffff0000080819e8: diff name v: __irqentry_text_end k: __softirqentry_text_start
WARN: 0xffff000008081d08: diff name v: __entry_text_start k: __softirqentry_text_end
WARN: 0xffff00000809db5c: diff name v: flush_icache_range k: __flush_cache_user_range
WARN: 0xffff000008101908: diff name v: sys_ni_syscall k: sys_vm86old
...
Signed-off-by: Kim Phillips <kim.phillips@arm.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ganapatrao Kulkarni <ganapatrao.kulkarni@cavium.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lkml.kernel.org/r/20171214175242.e30450f17f93ad675d968fa3@arm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
While monitoring a multithread process with pid option, perf sometimes
may return sys_perf_event_open failure with 3(No such process) if any of
the process's threads die before we open the event. However, we want
perf continue monitoring the remaining threads and do not exit with
error.
Here, the patch enables perf_evsel::ignore_missing_thread for -p option
to ignore complete failure if any of threads die before we open the event.
But it may still return sys_perf_event_open failure with 22(Invalid) if we
monitors several event groups.
sys_perf_event_open: pid 28960 cpu 40 group_fd 118202 flags 0x8
sys_perf_event_open: pid 28961 cpu 40 group_fd 118203 flags 0x8
WARNING: Ignored open failure for pid 28962
sys_perf_event_open: pid 28962 cpu 40 group_fd [118203] flags 0x8
sys_perf_event_open failed, error -22
That is because when we ignore a missing thread, we change the thread_idx
without dealing with its fds, FD(evsel, cpu, thread). Then get_group_fd()
may return a wrong group_fd for the next thread and sys_perf_event_open()
return with 22.
sys_perf_event_open(){
...
if (group_fd != -1)
perf_fget_light()//to get corresponding group_leader by group_fd
...
if (group_leader)
if (group_leader->ctx->task != ctx->task)//should on the same task
goto err_context
...
}
This patch also fixes this bug by introducing perf_evsel__remove_fd() and
update_fds to allow removing fds for the missing thread.
Changes since v1:
- Change group_fd__remove() into a more genetic way without changing code logic
- Remove redundant condition
Changes since v2:
- Use a proper function name and add some comment.
- Multiline comment style fixes.
Committer testing:
Before this patch the recently added 'perf stat --per-thread' for system
wide counting would race while enumerating all threads using /proc:
[root@jouet ~]# perf stat --per-thread
failed to parse CPUs map: No such file or directory
Usage: perf stat [<options>] [<command>]
-C, --cpu <cpu> list of cpus to monitor in system-wide
-a, --all-cpus system-wide collection from all CPUs
[root@jouet ~]# perf stat --per-thread
failed to parse CPUs map: No such file or directory
Usage: perf stat [<options>] [<command>]
-C, --cpu <cpu> list of cpus to monitor in system-wide
-a, --all-cpus system-wide collection from all CPUs
[root@jouet ~]#
When, say, the kernel was being built, so lots of shortlived threads,
after this patch this doesn't happen.
Signed-off-by: Mengting Zhang <zhangmengting@huawei.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Cheng Jian <cj.chengjian@huawei.com>
Cc: Li Bin <huawei.libin@huawei.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1513148513-6974-1-git-send-email-zhangmengting@huawei.com
[ Remove one use 'evlist' alias variable ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
