Add a test for the perf daemon 'signal' command. The test sends a signal
to configured sessions and verifies the perf data files were generated
accordingly.
Committer testing:
[root@five ~]# perf test daemon
76: daemon operations : Ok
[root@five ~]# perf test -v daemon
76: daemon operations :
--- start ---
test child forked, pid 790017
test daemon list
test daemon reconfig
test daemon stop
test daemon signal
signal 12 sent to session 'test [790268]'
signal 12 sent to session 'test [790268]'
test child finished with 0
---- end ----
daemon operations: Ok
[root@five ~]#
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: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-23-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a test for daemon reconfiguration. The test changes the
configuration file and checks that the session is changed properly.
Committer testing:
[root@five ~]# perf test daemon
76: daemon operations : Ok
[root@five ~]# time perf test daemon
76: daemon operations : Ok
real 0m6.055s
user 0m0.174s
sys 0m0.147s
[root@five ~]# time perf test -v daemon
76: daemon operations :
--- start ---
test child forked, pid 786863
test daemon list
test daemon reconfig
test child finished with 0
---- end ----
daemon operations: Ok
real 0m6.127s
user 0m0.222s
sys 0m0.165s
[root@five ~]#
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: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-21-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a 'ping' command to verify that the 'perf record' session is up and
operational.
It's used in the following patches via test code to make sure 'perf
record' is ready to receive signals.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
[session-sched]
run = -m 20M -e sched:* --overwrite --switch-output -a
Start the daemon:
# perf daemon start
Ping all sessions:
# perf daemon ping
OK cycles
OK sched
Ping specific session:
# perf daemon ping --session sched
OK sched
Committer notes:
Fixed up bug pointed by clang:
Buggy:
if (!pollfd.revents & POLLIN)
Correct code:
if (!(pollfd.revents & POLLIN))
clang warning:
builtin-daemon.c:560:6: error: logical not is only applied to the left hand side of this bitwise operator [-Werror,-Wlogical-not-parentheses]
if (!pollfd.revents & POLLIN) {
^ ~
builtin-daemon.c:560:6: note: add parentheses after the '!' to evaluate the bitwise operator first
Also use designated initialized with pollfd, i.e.:
struct pollfd pollfd = { .events = POLLIN, };
Instead of:
struct pollfd pollfd = { 0, };
To get past:
builtin-daemon.c:510:30: error: missing field 'events' initializer [-Werror,-Wmissing-field-initializers]
struct pollfd pollfd = { 0, };
^
1 error generated.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-16-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add 'lock' file under daemon base and flock it, so only one perf daemon
can run on top of it.
Each daemon tries to create and lock BASE/lock file, if it's successful
we are sure we're the only daemon running over the BASE.
Once daemon is finished, file descriptor to lock file is closed and lock
is released.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
[session-sched]
run = -m 20M -e sched:* --overwrite --switch-output -a
Starting the daemon:
# perf daemon start
And try once more:
# perf daemon start
failed: another perf daemon (pid 775594) owns /opt/perfdata
will end up with an error, because there's already one running
on top of /opt/perfdata.
Committer notes:
Provide lockf(F_TLOCK) when not available, i.e. transform:
lockf(fd, F_TLOCK, 0);
into:
flock(fd, LOCK_EX | LOCK_NB);
Which should be equivalent.
Noticed when cross building to some odd Android NDK.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-14-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Allow the 'perf daemon' to send SIGUSR2 to all running sessions or just
to a specific session.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
[session-sched]
run = -m 20M -e sched:* --overwrite --switch-output -a
Start the daemon:
# perf daemon start
Send signal to all running sessions:
# perf daemon signal
signal 12 sent to session 'cycles [773738]'
signal 12 sent to session 'sched [773739]'
Or to specific one:
# perf daemon signal --session sched
signal 12 sent to session 'sched [773739]'
And verify signals were delivered and perf.data dumped:
# cat /opt/perfdata/session-cycles/output
rounding mmap pages size to 32M (8192 pages)
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2021010220382490 ]
# car /opt/perfdata/session-sched/output
rounding mmap pages size to 32M (8192 pages)
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2021010220382489 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2021010220393745 ]
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-12-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add a 'list' command to display all running sessions. It's the default
command if no other command is specified.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
[session-sched]
run = -m 20M -e sched:* --overwrite --switch-output -a
Start the daemon:
# perf daemon start
List sessions:
# perf daemon
[771394:daemon] base: /opt/perfdata
[771395:cycles] perf record -m 10M -e cycles --overwrite --switch-output -a
[771396:sched] perf record -m 20M -e sched:* --overwrite --switch-output -a
List sessions with more info:
# perf daemon -v
[771394:daemon] base: /opt/perfdata
output: /opt/perfdata/output
[771395:cycles] perf record -m 10M -e cycles --overwrite --switch-output -a
base: /opt/perfdata/session-cycles
output: /opt/perfdata/session-cycles/output
[771396:sched] perf record -m 20M -e sched:* --overwrite --switch-output -a
base: /opt/perfdata/session-sched
output: /opt/perfdata/session-sched/output
The 'output' file is perf record output for specific session.
Note you have to stop all running perf processes manually at this point,
stop command is coming in following patches.
Committer notes:
Fixup union initialization to overcome this in multiple older systems:
22 15.74 debian:8 : FAIL gcc version 4.9.2 (Debian 4.9.2-10+deb8u2)
builtin-daemon.c: In function 'send_cmd_list':
builtin-daemon.c:1386:2: error: missing initializer for field 'csv_sep' of 'struct <anonymous>' [-Werror=missing-field-initializers]
};
^
builtin-daemon.c:641:8: note: 'csv_sep' declared here
char csv_sep;
^
cc1: all warnings being treated as errors
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-11-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add support to detect changes to the daemon's config file triggering a
re-read of the configuration when that happens.
Use a inotify file descriptor plugged into the main fdarray object for
polling.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
Starting the daemon:
# perf daemon start
Check sessions:
# perf daemon
[772262:daemon] base: /opt/perfdata
[772263:cycles] perf record -m 10M -e cycles --overwrite --switch-output -a
Change '-m 10M' to '-m 20M', and check daemon log:
# tail -f /opt/perfdata/output
[2021-01-02 20:31:41.234045] daemon started (pid 772262)
[2021-01-02 20:31:41.235072] reconfig: ruining session [cycles:772263]: -m 10M -e cycles --overwrite --switch-output -a
[2021-01-02 20:32:08.310137] reconfig: session 'cycles' killed
[2021-01-02 20:32:08.310847] reconfig: ruining session [cycles:772338]: -m 20M -e cycles --overwrite --switch-output -a
And the session list:
# perf daemon
[772262:daemon] base: /opt/perfdata
[772338:cycles] perf record -m 20M -e cycles --overwrite --switch-output -a
Note the changed '-m 20M' option is in place.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-8-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Adding support to configure daemon with config file.
Each client or server invocation of perf daemon needs to know the
base directory, where all sessions data is stored.
The base is defined with:
daemon.base
Base path for daemon data. All sessions data are stored under
this path.
The daemon allows to create record sessions. Each session is a
record command spawned and monitored by perf daemon.
The session is defined with:
session-<NAME>.run
Defines new record session for daemon. The value is record's
command line without the 'record' keyword.
Example:
# cat ~/.perfconfig
[daemon]
base=/opt/perfdata
[session-cycles]
run = -m 10M -e cycles --overwrite --switch-output -a
[session-sched]
run = -m 20M -e sched:* --overwrite --switch-output -a
The example above defines '/opt/perfdata' as the base directory and 2
record sessions.
# perf daemon start
[2021-01-28 19:47:33.454413] daemon started (pid 16015)
[2021-01-28 19:47:33.455910] reconfig: ruining session [cycles:16016]: -m 10M -e cycles --overwrite --switch-output -a
[2021-01-28 19:47:33.456599] reconfig: ruining session [sched:16017]: -m 20M -e sched:* --overwrite --switch-output -a
# ps -ef | grep perf
... perf daemon start
... /home/jolsa/.../perf record -m 20M -e cycles --overwrite --switch-output -a
... /home/jolsa/.../perf record -m 20M -e sched:* --overwrite --switch-output -a
The base directory is populated with:
# find /opt/perfdata/
/opt/perfdata/
/opt/perfdata/control <- control socket
/opt/perfdata/session-cycles <- data for session 'cycles':
/opt/perfdata/session-cycles/output <- perf record output
/opt/perfdata/session-cycles/perf.data <- perf data
/opt/perfdata/session-sched <- ditto for session 'sched'
/opt/perfdata/session-sched/output
/opt/perfdata/session-sched/perf.data
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-7-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add support for client socket side that will be used to send commands to
the daemon server socket.
This patch adds only the core support, all commands using this
functionality are coming in the following patches.
Committer notes:
Hat to patch patch it to deal with this in some systems:
cc1: warnings being treated as errors
builtin-daemon.c: In function 'send_cmd': MKDIR /tmp/build/perf/bench/
builtin-daemon.c:1368: error: ignoring return value of 'fwrite', declared with attribute warn_unused_result
MKDIR /tmp/build/perf/tests/
make[3]: *** [/tmp/build/perf/builtin-daemon.o] Error 1
And also to not leak the 'line' buffer allocated by getline(), since you
initialized line to NULL and len to zero, man page says:
If *lineptr is set to NULL and *n is set 0 before the call,
then getline() will allocate a buffer for storing the line.
This buffer should be freed by the user program even if
getline() failed.
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Budankov <abudankov@huawei.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Petlan <mpetlan@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: https://lore.kernel.org/r/20210208200908.1019149-6-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
We need to use "%#" PRIx64 for u64 values, not "%lx", fixing this build
problem on powerpc 32-bit:
72 13.69 ubuntu:18.04-x-powerpc : FAIL powerpc-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
arch/powerpc/util/machine.c: In function 'arch__symbols__fixup_end':
arch/powerpc/util/machine.c:23:12: error: format '%lx' expects argument of type 'long unsigned int', but argument 6 has type 'u64 {aka long long unsigned int}' [-Werror=format=]
pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
^
/git/linux/tools/perf/util/debug.h:18:21: note: in definition of macro 'pr_fmt'
#define pr_fmt(fmt) fmt
^~~
/git/linux/tools/perf/util/debug.h:33:29: note: in expansion of macro 'pr_debugN'
#define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
^~~~~~~~~
/git/linux/tools/perf/util/debug.h:33:42: note: in expansion of macro 'pr_fmt'
#define pr_debug4(fmt, ...) pr_debugN(4, pr_fmt(fmt), ##__VA_ARGS__)
^~~~~~
arch/powerpc/util/machine.c:23:2: note: in expansion of macro 'pr_debug4'
pr_debug4("%s sym:%s end:%#lx\n", __func__, p->name, p->end);
^~~~~~~~~
cc1: all warnings being treated as errors
/git/linux/tools/build/Makefile.build:139: recipe for target 'util' failed
make[5]: *** [util] Error 2
/git/linux/tools/build/Makefile.build:139: recipe for target 'powerpc' failed
make[4]: *** [powerpc] Error 2
/git/linux/tools/build/Makefile.build:139: recipe for target 'arch' failed
make[3]: *** [arch] Error 2
73 30.47 ubuntu:18.04-x-powerpc64 : Ok powerpc64-linux-gnu-gcc (Ubuntu 7.5.0-3ubuntu1~18.04) 7.5.0
Fixes: 557c3eadb7 ("perf powerpc: Fix gap between kernel end and module start")
Cc: Athira Rajeev <atrajeev@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
'perf script' supports '-S' or '--symbol' options to only list the
records for these symbols. A symbol is typically a name or hex address.
If it's hex address, it is the start address of one symbol.
While it would be useful if we can filter trace records by any hex
address (not only the start address of symbol). So now we support
filtering trace records by more conditions, such as:
- symbol name
- start address of symbol
- any hexadecimal address
- address range
The comparison order is defined as:
1. symbol name comparison
2. symbol start address comparison.
3. any hexadecimal address comparison.
4. address range comparison.
The idea is if we can get a valid address from -S list, we add the
address to addr_list for address comparison otherwise we still leave
it to sym_list for symbol comparison.
Some examples:
root@kbl-ppc:~# ./perf script -S ffffffff9a477308
perf 8562 [000] 347303.578858: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [000] 347303.578860: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [000] 347303.578861: 11 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [001] 347303.578903: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [001] 347303.578905: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [001] 347303.578906: 15 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [002] 347303.578952: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
perf 8562 [002] 347303.578953: 1 cycles: ffffffff9a477308 native_write_msr+0x8 ([kernel.kallsyms])
Filter the traced records by hex address ffffffff9a477308.
root@kbl-ppc:~# ./perf script -S ffffffff9a4dd4ce,ffffffff9a4d2de9,ffffffff9a6bf9f4
perf 8562 [001] 347303.578911: 311706 cycles: ffffffff9a6bf9f4 __kmalloc_node+0x204 ([kernel.kallsyms])
perf 8562 [002] 347303.578960: 354477 cycles: ffffffff9a4d2de9 sched_setaffinity+0x49 ([kernel.kallsyms])
perf 8562 [003] 347303.579015: 450958 cycles: ffffffff9a4dd4ce dequeue_task_fair+0x1ae ([kernel.kallsyms])
Filter the traced records by hex address ffffffff9a4dd4ce, ffffffff9a4d2de9, ffffffff9a6bf9f4.
root@kbl-ppc:~# ./perf script -S ffffffff9a477309 --addr-range 16
perf 8562 [000] 347303.578863: 291 cycles: ffffffff9a47730a native_write_msr+0xa ([kernel.kallsyms])
perf 8562 [001] 347303.578907: 411 cycles: ffffffff9a47730a native_write_msr+0xa ([kernel.kallsyms])
perf 8562 [002] 347303.578956: 462 cycles: ffffffff9a47730f native_write_msr+0xf ([kernel.kallsyms])
perf 8562 [003] 347303.579010: 497 cycles: ffffffff9a47730f native_write_msr+0xf ([kernel.kallsyms])
perf 8562 [004] 347303.579059: 429 cycles: ffffffff9a47730f native_write_msr+0xf ([kernel.kallsyms])
perf 8562 [005] 347303.579109: 408 cycles: ffffffff9a47730a native_write_msr+0xa ([kernel.kallsyms])
perf 8562 [006] 347303.579159: 460 cycles: ffffffff9a47730f native_write_msr+0xf ([kernel.kallsyms])
perf 8562 [007] 347303.579213: 436 cycles: ffffffff9a47730f native_write_msr+0xf ([kernel.kallsyms])
Filter the traced records from address range [ffffffff9a477309, ffffffff9a477309 + 15].
root@kbl-ppc:~# ./perf script -S "ffffffff9b163046,rcu_nmi_exit"
perf 8562 [004] 347303.579060: 12013 cycles: ffffffff9b163046 exc_nmi+0x166 ([kernel.kallsyms])
perf 8562 [007] 347303.579214: 12138 cycles: ffffffff9b165944 rcu_nmi_exit+0x34 ([kernel.kallsyms])
Filter by address + symbol
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: Jin Yao <yao.jin@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20210207080935.31784-2-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The TMA method level 2 metrics is supported from the Intel Sapphire
Rapids server, which expose four L2 Topdown metrics events to user
space. There are eight L2 events in total. The other four L2 Topdown
metrics events are calculated from the corresponding L1 and the exposed
L2 events.
Now, the --topdown prints the complete top-down metrics that supported
by the CPU. For the Intel Sapphire Rapids server, there are 4 L1 events
and 8 L2 events displyed in one line.
Add a new option, --td-level, to display the top-down statistics that
equal to or lower than the input level.
The L2 event is marked only when both its L1 parent event and itself
crosse the threshold.
Here is an example:
$ perf stat --topdown --td-level=2 --no-metric-only sleep 1
Topdown accuracy may decrease when measuring long periods.
Please print the result regularly, e.g. -I1000
Performance counter stats for 'sleep 1':
16,734,390 slots
2,100,001 topdown-retiring # 12.6% retiring
2,034,376 topdown-bad-spec # 12.3% bad speculation
4,003,128 topdown-fe-bound # 24.1% frontend bound
328,125 topdown-heavy-ops # 2.0% heavy operations # 10.6% light operations
1,968,751 topdown-br-mispredict # 11.9% branch mispredict # 0.4% machine clears
2,953,127 topdown-fetch-lat # 17.8% fetch latency # 6.3% fetch bandwidth
5,906,255 topdown-mem-bound # 35.6% memory bound # 15.4% core bound
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/1612296553-21962-9-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The instruction latency information can be recorded on some platforms,
e.g., the Intel Sapphire Rapids server. With both memory latency
(weight) and the new instruction latency information, users can easily
locate the expensive load instructions, and also understand the time
spent in different stages. The users can optimize their applications in
different pipeline stages.
The 'weight' field is shared among different architectures. Reusing the
'weight' field may impacts other architectures. Add a new field to store
the instruction latency.
Like the 'weight' support, introduce a 'ins_lat' for the global
instruction latency, and a 'local_ins_lat' for the local instruction
latency version.
Add new sort functions, INSTR Latency and Local INSTR Latency,
accordingly.
Add local_ins_lat to the default_mem_sort_order[].
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/1612296553-21962-7-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The new sample type, PERF_SAMPLE_WEIGHT_STRUCT, is an alternative of the
PERF_SAMPLE_WEIGHT sample type. Users can apply either the
PERF_SAMPLE_WEIGHT sample type or the PERF_SAMPLE_WEIGHT_STRUCT sample
type to retrieve the sample weight, but they cannot apply both sample
types simultaneously.
The new sample type shares the same space as the PERF_SAMPLE_WEIGHT
sample type. The lower 32 bits are exactly the same for both sample
type. The higher 32 bits may be different for different architecture.
Add arch specific arch_evsel__set_sample_weight() to set the new sample
type for X86. Only store the lower 32 bits for the sample->weight if the
new sample type is applied. In practice, no memory access could last
than 4G cycles. No data will be lost.
If the kernel doesn't support the new sample type. Fall back to the
PERF_SAMPLE_WEIGHT sample type.
There is no impact for other architectures.
Committer notes:
Fixup related to PERF_SAMPLE_CODE_PAGE_SIZE, present in acme/perf/core
but not upstream yet.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/1612296553-21962-6-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Two new data source fields, to indicate the block reasons of a load
instruction, are introduced on the Intel Sapphire Rapids server. The
fields can be used by the memory profiling.
Add a new sort function, SORT_MEM_BLOCKED, for the two fields.
For the previous platforms or the block reason is unknown, print "N/A"
for the block reason.
Add blocked as a default mem sort key for perf report and perf mem
report.
Committer testing:
So in machines without this capability we get a "N/A" filling the new "Blocked"
column:
$ perf mem record ls
arch certs CREDITS Documentation include ipc Kconfig lib MAINTAINERS mm samples security usr block
COPYING crypto drivers fs init Kbuild kernel LICENSES Makefile net README scripts sound tools
virt
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.008 MB perf.data (17 samples) ]
$
$ perf mem report --stdio
# To display the perf.data header info, please use --header/--header-only options.
#
# Total Lost Samples: 0
#
# Samples: 6 of event 'cpu/mem-loads,ldlat=30/Pu'
# Total weight : 1381
# Sort order : local_weight,mem,sym,dso,symbol_daddr,dso_daddr,snoop,tlb,locked,blocked
#
# Overhead Samples Local Weight Memory access Symbol Shared Object Data Symbol Data Object Snoop TLB access Locked Blocked
# ........ ....... ............ .................... ....................... ............. ...................... ............ ..... ............ ...... .......
#
32.87% 1 454 Local RAM or RAM hit [.] _dl_relocate_object ld-2.31.so [.] 0x00007fe91cef3078 libc-2.31.so Hit L1 or L2 hit No N/A
25.56% 1 353 LFB or LFB hit [.] strcmp ld-2.31.so [.] 0x00005586973855ca ls None L1 or L2 hit No N/A
22.59% 1 312 LFB or LFB hit [.] _dl_cache_libcmp ld-2.31.so [.] 0x00007fe91d0e3b18 ld.so.cache None L1 or L2 hit No N/A
8.47% 1 117 LFB or LFB hit [.] _dl_relocate_object ld-2.31.so [.] 0x00007fe91ceee570 libc-2.31.so None L1 or L2 hit No N/A
6.88% 1 95 LFB or LFB hit [.] _dl_relocate_object ld-2.31.so [.] 0x00007fe91ceed490 libc-2.31.so None L1 or L2 hit No N/A
3.62% 1 50 LFB or LFB hit [.] _dl_cache_libcmp ld-2.31.so [.] 0x00007fe91d0ebe60 ld.so.cache None L1 or L2 hit No N/A
# Samples: 11 of event 'cpu/mem-stores/Pu'
# Total weight : 11
# Sort order : local_weight,mem,sym,dso,symbol_daddr,dso_daddr,snoop,tlb,locked,blocked
#
# Overhead Samples Local Weight Memory access Symbol Shared Object Data Symbol Data Object Snoop TLB access Locked Blocked
# ........ ....... ............ ............. ....................... ............. ...................... ........... ..... .......... ...... .......
#
9.09% 1 0 L1 hit [.] __strcoll_l libc-2.31.so [.] 0x00007fffe5648fc8 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] _dl_lookup_symbol_x ld-2.31.so [.] 0x00007fffe56490b8 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] _dl_name_match_p ld-2.31.so [.] 0x00007fffe56487d8 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] _dl_start ld-2.31.so [.] start_time+0x0 ld-2.31.so N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] _dl_sysdep_start ld-2.31.so [.] 0x00007fffe56494b8 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] do_lookup_x ld-2.31.so [.] 0x00007fffe5648ff8 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] do_lookup_x ld-2.31.so [.] 0x00007fffe5649064 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 hit [.] do_lookup_x ld-2.31.so [.] 0x00007fffe5649130 [stack] N/A N/A N/A N/A
9.09% 1 0 L1 miss [.] _dl_start ld-2.31.so [.] _rtld_global+0xaf8 ld-2.31.so N/A N/A N/A N/A
9.09% 1 0 L1 miss [.] _dl_start ld-2.31.so [.] _rtld_global+0xc28 ld-2.31.so N/A N/A N/A N/A
9.09% 1 0 L1 miss [.] _dl_start ld-2.31.so [.] 0x00007fffe56495b8 [stack] N/A N/A N/A N/A
# (Tip: Show user configuration overrides: perf config --user --list)
$
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/1612296553-21962-4-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
On the Intel Sapphire Rapids server, an auxiliary event has to be
enabled simultaneously with the load latency event to retrieve complete
Memory Info.
Add X86 specific perf_mem_events__name() to handle the auxiliary event.
- Users are only interested in the samples of the mem-loads event.
Sample read the auxiliary event.
- The auxiliary event must be in front of the load latency event in a
group. Assume the second event to sample if the auxiliary event is the
leader.
- Add a weak is_mem_loads_aux_event() to check the auxiliary event for
X86. For other ARCHs, it always return false.
Parse the unique event name, mem-loads-aux, for the auxiliary event.
Committer notes:
According to 61b985e3e7 ("perf/x86/intel: Add perf core PMU
support for Sapphire Rapids"), ENODATA is only returned by
sys_perf_event_open() when used with these auxiliary events, with this
in evsel__open_strerror():
case ENODATA:
return scnprintf(msg, size, "Cannot collect data source with the load latency event alone. "
"Please add an auxiliary event in front of the load latency event.");
This is Ok at this point in time, but fragile long term, I pointed this
out in the e-mail thread, requesting a follow up patch to check if
ENODATA is really for this specific case.
Fixed up sizeof(MEM_LOADS_AUX_NAME) bug pointed out by Namhyung.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Madhavan Srinivasan <maddy@linux.vnet.ibm.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lore.kernel.org/lkml/20210205152648.GC920417@kernel.org
Link: http://lore.kernel.org/lkml/1612296553-21962-3-git-send-email-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
To enable presenting of Performance Monitor Counter Registers (PMC1 to
PMC6) as part of extended regsiters, this patch adds these to
sample_reg_mask in the tool side (to use with -I? option).
Simplified the PERF_REG_PMU_MASK_300/31 definition. Excluded the
unsupported SPRs (MMCR3, SIER2, SIER3) from extended mask value for
CPU_FTR_ARCH_300.
Signed-off-by: Athira Jajeev <atrajeev@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kajol Jain <kjain@linux.ibm.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: linuxppc-dev@lists.ozlabs.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
