A new branch sample type PERF_SAMPLE_BRANCH_HW_INDEX has been introduced
in latest kernel.
Enable HW_INDEX by default in LBR call stack mode.
If kernel doesn't support the sample type, switching it off.
Add HW_INDEX in attr_fprintf as well. User can check whether the branch
sample type is set via debug information or header.
Committer testing:
First collect some samples with LBR callchains, system wide, for a few
seconds:
# perf record --call-graph lbr -a sleep 5
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.625 MB perf.data (224 samples) ]
#
Now lets use 'perf evlist -v' to look at the branch_sample_type:
# perf evlist -v
cycles: size: 120, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: USER|CALL_STACK|NO_FLAGS|NO_CYCLES|HW_INDEX
#
So the machine has the kernel feature, and it was correctly added to
perf_event_attr.branch_sample_type, for the default 'cycles' event.
If we do it in another machine, where the kernel lacks the HW_INDEX
feature, we get:
# perf record --call-graph lbr -a sleep 2s
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 1.690 MB perf.data (499 samples) ]
# perf evlist -v
cycles: size: 120, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|CALLCHAIN|CPU|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, task: 1, precise_ip: 3, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: USER|CALL_STACK|NO_FLAGS|NO_CYCLES
#
No HW_INDEX in attr.branch_sample_type.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexey Budankov <alexey.budankov@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Pavel Gerasimov <pavel.gerasimov@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vitaly Slobodskoy <vitaly.slobodskoy@intel.com>
Link: http://lore.kernel.org/lkml/20200228163011.19358-3-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The low level index of raw branch records for the most recent branch can
be recorded in a sample with PERF_SAMPLE_BRANCH_HW_INDEX
branch_sample_type. Extend struct branch_stack to support it.
However, if the PERF_SAMPLE_BRANCH_HW_INDEX is not applied, only nr and
entries[] will be output by kernel. The pointer of entries[] could be
wrong, since the output format is different with new struct
branch_stack. Add a variable no_hw_idx in struct perf_sample to
indicate whether the hw_idx is output. Add get_branch_entry() to return
corresponding pointer of entries[0].
To make dummy branch sample consistent as new branch sample, add hw_idx
in struct dummy_branch_stack for cs-etm and intel-pt.
Apply the new struct branch_stack for synthetic events as well.
Extend test case sample-parsing to support new struct branch_stack.
Committer notes:
Renamed get_branch_entries() to perf_sample__branch_entries() to have
proper namespacing and pave the way for this to be moved to libperf,
eventually.
Add 'static' to that inline as it is in a header.
Add 'hw_idx' to 'struct dummy_branch_stack' in cs-etm.c to fix the build
on arm64.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexey Budankov <alexey.budankov@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Pavel Gerasimov <pavel.gerasimov@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Stephane Eranian <eranian@google.com>
Cc: Vitaly Slobodskoy <vitaly.slobodskoy@intel.com>
Link: http://lore.kernel.org/lkml/20200228163011.19358-2-kan.liang@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
perf with CoreSight fails to record trace data with command:
perf record -e cs_etm/@tmc_etr0/u --per-thread ls
failed to set sink "" on event cs_etm/@tmc_etr0/u with 21 (Is a
directory)/perf/
This failure is root caused with the commit 1dc925568f ("perf
parse: Add a deep delete for parse event terms").
The log shows, cs_etm fails to parse the sink attribution; cs_etm event
relies on the event configuration to pass sink name, but the event
specific configuration data cannot be passed properly with flow:
get_config_terms()
ADD_CONFIG_TERM(DRV_CFG, term->val.str);
__t->val.str = term->val.str;
`> __t->val.str is assigned to term->val.str;
parse_events_terms__purge()
parse_events_term__delete()
zfree(&term->val.str);
`> term->val.str is freed and assigned to NULL pointer;
cs_etm_set_sink_attr()
sink = __t->val.str;
`> sink string has been freed.
To fix this issue, in the function get_config_terms(), this patch
changes to use strdup() for allocation a new duplicate string rather
than directly assignment string pointer.
This patch addes a new field 'free_str' in the data structure
perf_evsel_config_term; 'free_str' is set to true when the union is used
as a string pointer; thus it can tell perf_evsel__free_config_terms() to
free the string.
Fixes: 1dc925568f ("perf parse: Add a deep delete for parse event terms")
Suggested-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Leo Yan <leo.yan@linaro.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Poirier <mathieu.poirier@linaro.org>
Cc: Mike Leach <mike.leach@linaro.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Suzuki Poulouse <suzuki.poulose@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Link: http://lore.kernel.org/lkml/20200117055251.24058-2-leo.yan@linaro.org
[ Use zfree() in perf_evsel__free_config_terms ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
:# modified: tools/perf/util/evsel_config.h
Restructure the event opening in perf stat to cycle through the events
by CPU after setting affinity to that CPU.
This eliminates IPI overhead in the perf API.
We have to loop through the CPU in the outter builtin-stat code instead
of leaving that to low level functions.
It has to change the weak group fallback strategy slightly. Since we
cannot easily undo the opens for other CPUs move the weak group retry to
a separate loop.
Before with a large test case with 94 CPUs:
% time seconds usecs/call calls errors syscall
------ ----------- ----------- --------- --------- ----------------
42.75 4.050910 67 60046 110 perf_event_open
After:
26.86 0.944396 16 58069 110 perf_event_open
(the number changes slightly because the weak group retries
work differently and the test case relies on weak groups)
Committer notes:
Added one of the hunks in a patch provided by Andi after I noticed that
the "event times" 'perf test' entry was segfaulting.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: http://lore.kernel.org/lkml/20191121001522.180827-10-andi@firstfloor.org
Link: http://lore.kernel.org/lkml/20191127232657.GL84886@tassilo.jf.intel.com # Fix
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Default config for a PMU is defined before selected events are parsed.
That allows the user-entered config to override the default config.
However that does not allow for changing the default config based on
other options.
For example, if the user chooses AUX area sampling mode, in the case of
Intel PT, the psb_period needs to be small for sampling, so there is a
need to set the default psb_period to 0 (2 KiB) in that case. However
that should not override a value set by the user. To allow for that,
when using default config, record which bits of config were changed by
the user.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20191115124225.5247-13-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
To allow individual events to be selected for AUX area sampling, add
aux-sample-size config term. attr.aux_sample_size is updated by
auxtrace_parse_sample_options() so that the existing validation will see
the value. Any event that has a non-zero aux_sample_size will cause AUX
area sampling to be configured, irrespective of the --aux-sample option.
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20191115124225.5247-8-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add kernel AUX area sampling definitions, which brings perf_event.h into
line with the kernel version.
New sample type PERF_SAMPLE_AUX requests a sample of the AUX area
buffer. New perf_event_attr member 'aux_sample_size' specifies the
desired size of the sample.
Also add support for parsing samples containing AUX area data i.e.
PERF_SAMPLE_AUX.
Committer notes:
I squashed the first two patches in this series to avoid breaking
automatic bisection, i.e. after applying only the original first patch
in this series we would have:
# perf test -v parsing
26: Sample parsing :
--- start ---
test child forked, pid 17018
sample format has changed, some new PERF_SAMPLE_ bit was introduced - test needs updating
test child finished with -1
---- end ----
Sample parsing: FAILED!
#
With the two paches combined:
# perf test parsing
26: Sample parsing : Ok
#
Signed-off-by: Adrian Hunter <adrian.hunter@intel.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Link: http://lore.kernel.org/lkml/20191115124225.5247-3-adrian.hunter@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Perf record with verbose=2 already prints this information along with
whole lot of other traces which requires lot of scrolling. Introduce
an option to print only perf_event_open() arguments and return value.
Sample o/p:
$ perf --debug perf-event-open=1 record -- ls > /dev/null
------------------------------------------------------------
perf_event_attr:
size 112
{ sample_period, sample_freq } 4000
sample_type IP|TID|TIME|PERIOD
read_format ID
disabled 1
inherit 1
exclude_kernel 1
mmap 1
comm 1
freq 1
enable_on_exec 1
task 1
precise_ip 3
sample_id_all 1
exclude_guest 1
mmap2 1
comm_exec 1
ksymbol 1
bpf_event 1
------------------------------------------------------------
sys_perf_event_open: pid 4308 cpu 0 group_fd -1 flags 0x8 = 4
sys_perf_event_open: pid 4308 cpu 1 group_fd -1 flags 0x8 = 5
sys_perf_event_open: pid 4308 cpu 2 group_fd -1 flags 0x8 = 6
sys_perf_event_open: pid 4308 cpu 3 group_fd -1 flags 0x8 = 8
sys_perf_event_open: pid 4308 cpu 4 group_fd -1 flags 0x8 = 9
sys_perf_event_open: pid 4308 cpu 5 group_fd -1 flags 0x8 = 10
sys_perf_event_open: pid 4308 cpu 6 group_fd -1 flags 0x8 = 11
sys_perf_event_open: pid 4308 cpu 7 group_fd -1 flags 0x8 = 12
------------------------------------------------------------
perf_event_attr:
type 1
size 112
config 0x9
watermark 1
sample_id_all 1
bpf_event 1
{ wakeup_events, wakeup_watermark } 1
------------------------------------------------------------
sys_perf_event_open: pid -1 cpu 0 group_fd -1 flags 0x8
sys_perf_event_open failed, error -13
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.002 MB perf.data (9 samples) ]
Committer notes:
Just like the 'verbose' variable this new 'debug_peo_args' needs to be
added to util/python.c, since we don't link the debug.o file in the
python binding, which ended up making 'perf test python' fail with:
# perf test -v python
18: 'import perf' in python :
--- start ---
test child forked, pid 19237
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
ImportError: /tmp/build/perf/python/perf.so: undefined symbol: debug_peo_args
test child finished with -1
---- end ----
'import perf' in python: FAILED!
#
After adding that new variable to util/python.c:
# perf test -v python
18: 'import perf' in python :
--- start ---
test child forked, pid 22364
test child finished with 0
---- end ----
'import perf' in python: Ok
#
Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Link: http://lore.kernel.org/lkml/20191108094128.28769-1-ravi.bangoria@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Now that the default perf_events paranoid level is set to 2, a regular
user cannot monitor kernel level activity anymore. As such, with the
following cmdline:
$ perf record -e cycles date
The perf tool first tries cycles:uk but then falls back to cycles:u as
can be seen in the perf report --header-only output:
cmdline : /export/hda3/tmp/perf.tip record -e cycles ls
event : name = cycles:u, , id = { 436186, ... }
This is okay as long as there is way to learn the priv level was changed
internally by the tool.
But consider a similar example:
$ perf record -b -e cycles date
Error:
You may not have permission to collect stats.
Consider tweaking /proc/sys/kernel/perf_event_paranoid,
which controls use of the performance events system by
unprivileged users (without CAP_SYS_ADMIN).
...
Why is that treated differently given that the branch sampling inherits the
priv level of the first event in this case, i.e., cycles:u? It turns out
that the branch sampling code is more picky and also checks exclude_hv.
In the fallback path, perf record is setting exclude_kernel = 1, but it
does not change exclude_hv. This does not seem to match the restriction
imposed by paranoid = 2.
This patch fixes the problem by forcing exclude_hv = 1 in the fallback
for paranoid=2. With this in place:
$ perf record -b -e cycles date
cmdline : /export/hda3/tmp/perf.tip record -b -e cycles ls
event : name = cycles:u, , id = { 436847, ... }
And the command succeeds as expected.
V2 fix a white space.
Committer testing:
After aplying the patch we get:
[acme@quaco ~]$ perf record -b -e cycles date
WARNING: Kernel address maps (/proc/{kallsyms,modules}) are restricted,
check /proc/sys/kernel/kptr_restrict and /proc/sys/kernel/perf_event_paranoid.
Samples in kernel functions may not be resolved if a suitable vmlinux
file is not found in the buildid cache or in the vmlinux path.
Samples in kernel modules won't be resolved at all.
If some relocation was applied (e.g. kexec) symbols may be misresolved
even with a suitable vmlinux or kallsyms file.
Mon 23 Sep 2019 11:00:59 AM -03
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.005 MB perf.data (14 samples) ]
[acme@quaco ~]$ perf evlist -v
cycles:u: size: 112, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|PERIOD|BRANCH_STACK, read_format: ID, disabled: 1, inherit: 1, exclude_kernel: 1, exclude_hv: 1, mmap: 1, comm: 1, freq: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1, ksymbol: 1, bpf_event: 1, branch_sample_type: ANY
[acme@quaco ~]$
That warning about restricted kernel maps will be suppressed in a follow
up patch, as perf_event_attr.exclude_kernel is set, i.e. no samples for
the kernel will be taken and thus no need for those maps.
Signed-off-by: Stephane Eranian <eranian@google.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20190920230356.41420-1-eranian@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
