Merge tag 'perf-core-for-mingo-5.5-20190925' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux into perf/urgent
Pull perf/core improvements and fixes from Arnaldo Carvalho de Melo:
perf record:
Stephane Eranian:
- Fix priv level with branch sampling for paranoid=2, i.e. the kernel checks
if perf_event_attr_attr.exclude_hv is set in addition to .exclude_kernel,
so reset both to zero.
Arnaldo Carvalho de Melo:
- Don't warn about not being able to read kernel maps (kallsyms, etc) when
kernel samples aren't being collected.
perf list:
Kim Phillips:
- Allow plurals for metric, metricgroup., i.e.:
$ perf list metrics
was showing nothing, which is very confusing, make it work like:
$ perf stat metric
perf stat:
Andi Kleen:
- Free memory access/leaks detected via valgrind, related to metrics.
Libraries:
libperf:
Jiri Olsa:
- Move more stuff from tools/perf, this time a first stab at moving perf_mmap
methods.
libtracevent:
Steven Rostedt (VMware):
- Round up in tep_print_event() time precision.
Tzvetomir Stoyanov (VMware):
- Man pages for event print and related and plugins APIs.
- Move traceevent plugins in its own subdirectory.
Feature detection:
Thomas Richter:
- Add detection of java-11-openjdk-devel package, in addition to the older
versions supported.
Architecture specific:
S/390:
Thomas Richter (2):
- Include JVMTI support for s390
Vendor events:
AMD:
Kim Phillips:
- Add L3 cache events for Family 17h.
- Remove redundant '['.
PowerPC:
Mamatha Inamdar:
- Remove P8 HW events which are not supported.
Cleanups:
Arnaldo Carvalho de Melo:
- Remove needless headers, add needed ones, move things around to reduce the
headers dependency tree, speeding up builds by not doing needless compiles
when unrelated stuff gets changed.
I'm not fully sure if this is the correct fix, but without this I get
crashes on more complex perf stat metric usages. The problem is that
part of the state gets freed when a weak group fails, but then is later
still used. Just don't free the ids, we're going to reuse them anyways
on the weak group retry.
For example:
% perf stat -M IpB,IpCall,IpTB,IPC,Retiring_SMT,Frontend_Bound_SMT,Kernel_Utilization,CPU_Utilization --metric-only -a -I 1000 sleep 2
crashes and gives in valgrind:
=21527== Invalid write of size 8
==21527== at 0x4EE582: hlist_add_head (list.h:644)
==21527== by 0x4EFD3C: perf_evlist__id_hash (evlist.c:477)
==21527== by 0x4EFD99: perf_evlist__id_add (evlist.c:483)
==21527== by 0x4EFF15: perf_evlist__id_add_fd (evlist.c:524)
==21527== by 0x4FC693: store_evsel_ids (evsel.c:2969)
==21527== by 0x4FC76C: perf_evsel__store_ids (evsel.c:2986)
==21527== by 0x450DA7: __run_perf_stat (builtin-stat.c:519)
==21527== by 0x451285: run_perf_stat (builtin-stat.c:636)
==21527== by 0x454619: cmd_stat (builtin-stat.c:1966)
==21527== by 0x4D557D: run_builtin (perf.c:310)
==21527== by 0x4D57EA: handle_internal_command (perf.c:362)
==21527== by 0x4D5931: run_argv (perf.c:406)
==21527== Address 0x12e3f008 is 104 bytes inside a block of size 2,056 free'd
==21527== at 0x4839A0C: free (vg_replace_malloc.c:540)
==21527== by 0x627139: xyarray__delete (xyarray.c:32)
==21527== by 0x4F6BE4: perf_evsel__free_id (evsel.c:1253)
==21527== by 0x4FA11F: evsel__close (evsel.c:1994)
==21527== by 0x4F30A3: perf_evlist__reset_weak_group (evlist.c:1783)
==21527== by 0x450B47: __run_perf_stat (builtin-stat.c:466)
==21527== by 0x451285: run_perf_stat (builtin-stat.c:636)
==21527== by 0x454619: cmd_stat (builtin-stat.c:1966)
==21527== by 0x4D557D: run_builtin (perf.c:310)
==21527== by 0x4D57EA: handle_internal_command (perf.c:362)
==21527== by 0x4D5931: run_argv (perf.c:406)
==21527== by 0x4D5CAE: main (perf.c:531)
==21527== Block was alloc'd at
==21527== at 0x483AB1A: calloc (vg_replace_malloc.c:762)
==21527== by 0x627024: zalloc (zalloc.c:8)
==21527== by 0x627088: xyarray__new (xyarray.c:10)
==21527== by 0x4F6B20: perf_evsel__alloc_id (evsel.c:1237)
==21527== by 0x4FC74E: perf_evsel__store_ids (evsel.c:2983)
==21527== by 0x450DA7: __run_perf_stat (builtin-stat.c:519)
==21527== by 0x451285: run_perf_stat (builtin-stat.c:636)
==21527== by 0x454619: cmd_stat (builtin-stat.c:1966)
==21527== by 0x4D557D: run_builtin (perf.c:310)
==21527== by 0x4D57EA: handle_internal_command (perf.c:362)
==21527== by 0x4D5931: run_argv (perf.c:406)
==21527== by 0x4D5CAE: main (perf.c:531)
Make sure to not free the name passed in by the caller, but free all the
allocated ids when parsing expressions.
The loop at the end knows that the first entry shouldn't be freed, so
make sure the caller name is the first entry.
Fixes
% perf stat -M IpB,IpCall,IpTB,IPC,Retiring_SMT,Frontend_Bound_SMT,Kernel_Utilization,CPU_Utilization --metric-only -a -I 1000 sleep 2
valgrind:
1.009943231 ==21527== Invalid read of size 1
==21527== at 0x483CB74: strcmp (vg_replace_strmem.c:849)
==21527== by 0x582CF8: collect_all_aliases (stat-display.c:554)
==21527== by 0x582EB3: collect_data (stat-display.c:577)
==21527== by 0x583A32: print_counter_aggr (stat-display.c:806)
==21527== by 0x584FAD: perf_evlist__print_counters (stat-display.c:1200)
==21527== by 0x45133A: print_counters (builtin-stat.c:655)
==21527== by 0x450629: process_interval (builtin-stat.c:353)
==21527== by 0x450FBD: __run_perf_stat (builtin-stat.c:564)
==21527== by 0x451285: run_perf_stat (builtin-stat.c:636)
==21527== by 0x454619: cmd_stat (builtin-stat.c:1966)
==21527== by 0x4D557D: run_builtin (perf.c:310)
==21527== by 0x4D57EA: handle_internal_command (perf.c:362)
==21527== Address 0x12826cd0 is 0 bytes inside a block of size 25 free'd
==21527== at 0x4839A0C: free (vg_replace_malloc.c:540)
==21527== by 0x627041: __zfree (zalloc.c:13)
==21527== by 0x57F66A: generic_metric (stat-shadow.c:814)
==21527== by 0x580B21: perf_stat__print_shadow_stats (stat-shadow.c:1057)
==21527== by 0x58418E: print_metric_headers (stat-display.c:943)
==21527== by 0x5844BC: print_interval (stat-display.c:1004)
==21527== by 0x584DEB: perf_evlist__print_counters (stat-display.c:1172)
==21527== by 0x45133A: print_counters (builtin-stat.c:655)
==21527== by 0x450629: process_interval (builtin-stat.c:353)
==21527== by 0x450FBD: __run_perf_stat (builtin-stat.c:564)
==21527== by 0x451285: run_perf_stat (builtin-stat.c:636)
==21527== by 0x454619: cmd_stat (builtin-stat.c:1966)
==21527== Block was alloc'd at
==21527== at 0x483880B: malloc (vg_replace_malloc.c:309)
==21527== by 0x51677DE: strdup (in /usr/lib64/libc-2.29.so)
==21527== by 0x506457: parse_events_name (parse-events.c:1754)
==21527== by 0x5550BB: parse_events_parse (parse-events.y:214)
==21527== by 0x50694D: parse_events__scanner (parse-events.c:1887)
==21527== by 0x506AEF: parse_events (parse-events.c:1927)
==21527== by 0x521D8B: metricgroup__parse_groups (metricgroup.c:527)
==21527== by 0x45156F: parse_metric_groups (builtin-stat.c:721)
==21527== by 0x6228A9: get_value (parse-options.c:243)
==21527== by 0x62363F: parse_short_opt (parse-options.c:348)
==21527== by 0x62363F: parse_options_step (parse-options.c:536)
==21527== by 0x62363F: parse_options_subcommand (parse-options.c:651)
==21527== by 0x453C1D: cmd_stat (builtin-stat.c:1718)
==21527== by 0x4D557D: run_builtin (perf.c:310)
and also a leak report.
Committer testing:
Before:
# perf stat -M IpB,IpCall,IpTB,IPC,Retiring_SMT,Frontend_Bound_SMT,Kernel_Utilization,CPU_Utilization --metric-only -a -I 1000 sleep 2
# time CPU_Utilization
1.000470810 free(): double free detected in tcache 2
Aborted (core dumped)
#
After:
# perf stat -M IpB,IpCall,IpTB,IPC,Retiring_SMT,Frontend_Bound_SMT,Kernel_Utilization,CPU_Utilization --metric-only -a -I 1000 sleep 2
# time CPU_Utilization
1.000494752 0.1
2.001105112 0.1
#
perf tools: Replace needless mmap.h with what is needed, event.h
The perf_sample struct definition and the event_attr_init() are in
util/event.h, but some places were getting it thru an otherwise needless
util/mmap.h header, fix it by including util/event.h directly.
Ditch it, noone is using it, one more stdio.h include in a hot header.
Fix the fallout in parse-events.y, where we end up using a FILE pointer,
I think due to YYDEBUG being set and in some places, like Amazon Linux 1
we don't get stdio.h included by luck, like in most other places, add a
explicit stdio.h include directive.
We already had evsel_fprintf.c, add its counterpart, so that we can
reduce evsel.h a bit more.
We needed a new perf_event_attr_fprintf.c file so as to have a separate
object to link with the python binding in tools/perf/util/python-ext-sources
and not drag symbol_conf, etc into the python binding.
libperf: Use sys/types.h to get ssize_t, not unistd.h
The sys/types.h header looks more sensible, from its name we can gather
it should be there because of some needed typedef, and it is much
smaller than unistd.h, so use it and fix up the fallout in places where
it was being used for something else entirely but being obtained by
sheer luck, indirectly.
perf tools: No need to include internal/lib.h from util/util.h
That was done just to have users of writen() and readn(), that before
had their prototypes in util/util.h to get it without having to add an
include for internal/lib.h, but the right way is to add it and by now
all places already do it.
Fix a fallout were readlink() was used but unistd.h was being obtained
by luck thru util.h -> internal/lib.h, now to check why unistd.h is
being included there...
Add perf_evsel__alloc_id()/perf_evsel__free_id() functions to libperf as
internal functions.
Move 'struct perf_sample_id' to internal/evsel.h header and change
'struct perf_sample_id::evsel' to 'struct perf_evsel' and the related
code that touches it.
perf evlist: Adopt backwards ring buffer state enum
As this isn't used at all in mmap.h but in evlist.h, so to cut down the
header dependency tree, move it to where it is used.
Also add mmap.h to the places using it but previously getting it
indirectly via evlist.h.
Add missing pthread.h to evlist.h, as it has a pthread_t struct member
and was getting the header via mmap.h.
Noticed while processing a Jiri's libperf batch touching mmap.h, where
almost everything gets rebuilt because evlist.h is so popular, so cut
down't this rebuild the world party.
Jiri Olsa [Thu, 12 Sep 2019 08:57:18 +0000 (10:57 +0200)]
tools: Add missing stdio.h include to asm/bug.h header
We have a direct fprintf() call in the header, so we need stdio.h
include, otherwise it could fail compilation if there's no prior stdio.h
include directive.
libtraceevent: Move traceevent plugins in its own subdirectory
All traceevent plugins code is moved to tools/lib/traceevent/plugins
subdirectory. It makes traceevent implementation in trace-cmd and in
kernel tree consistent. There is no changes in the way libtraceevent and
plugins are compiled and installed.
Committer notes:
Applied fixup provided by Steven, fixing the tools/perf/Makefile.perf
target for the plugin dynamic list file. Problem noticed when cross
building to aarch64 from a Ubuntu 19.04 container.
libtraceevent: Add tep_get_event() in event-parse.h
The tep_get_event() function is an official libtracevent API, described
in the library man pages. However, it cannot be used by the library users because
it is not declared in the event-parse.h file, where all libtracevent APIs are.
The function declaration is added in event-parse.h file.
libtraceevent: Man pages fix, changes in event printing APIs
APIs for printing various trace event information were redesigned to be
more simple. However, the main libtraceevent man page was not updated
with those changes. The documentation is updated to describe the new
event print API.
libtraceevent: Man pages fix, rename tep_ref_get() to tep_get_ref()
The tep_ref_get() was renamed to tep_get_ref(), to be more consistent
with the other tep_ref_* APIs. However, in the man pages the API is
still with the old name. The documentation is fixed to reflect the
actual name of the API.
libtraceevent: Round up in tep_print_event() time precision
When testing the output of the old trace-cmd compared to the one that
uses the updated tep_print_event() logic, it was different in that the
time stamp precision in the old format would round up to the nearest
precision, where as the new logic truncates. Bring back the old method
of rounding up.
perf record: Move restricted maps check to after a possible fallback to not collect kernel samples
Before:
[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 ~]$
But we did a fallback and exclude_kernel was set, so no need for
resolving kernel symbols:
perf record: Fix priv level with branch sampling for paranoid=2
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.
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.
Mamatha Inamdar [Thu, 22 Aug 2019 07:20:49 +0000 (12:50 +0530)]
perf session: Return error code for perf_session__new() function on failure
This patch is to return error code of perf_new_session function on
failure instead of NULL.
Test Results:
Before Fix:
$ perf c2c report -input
failed to open nput: No such file or directory
$ echo $?
0
$
After Fix:
$ perf c2c report -input
failed to open nput: No such file or directory
$ echo $?
254
$
Committer notes:
Fix 'perf tests topology' case, where we use that TEST_ASSERT_VAL(...,
session), i.e. we need to pass zero in case of failure, which was the
case before when NULL was returned by perf_session__new() for failure,
but now we need to negate the result of IS_ERR(session) to respect that
TEST_ASSERT_VAL) expectation of zero meaning failure.
perf probe: Fix to clear tev->nargs in clear_probe_trace_event()
Since add_probe_trace_event() can reuse tf->tevs[i] after calling
clear_probe_trace_event(), this can make perf-probe crash if the 1st
attempt of probe event finding fails to find an event argument, and the
2nd attempt fails to find probe point.
E.g.
$ perf probe -D "task_pid_nr tsk"
Failed to find 'tsk' in this function.
Failed to get entry address of warn_bad_vsyscall
Segmentation fault (core dumped)
Committer testing:
After the patch:
$ perf probe -D "task_pid_nr tsk"
Failed to find 'tsk' in this function.
Failed to get entry address of warn_bad_vsyscall
Failed to get entry address of signal_fault
Failed to get entry address of show_signal
Failed to get entry address of umip_printk
Failed to get entry address of __bad_area_nosemaphore
<SNIP>
Failed to get entry address of sock_set_timeout
Failed to get entry address of tcp_recvmsg
Probe point 'task_pid_nr' not found.
Error: Failed to add events.
$
perf probe: Skip same probe address for a given line
Fix to skip making a same probe address on given line.
Since a DWARF line info contains several entries for one line with
different column, perf probe will make a different probe on same address
if user specifies a probe point by "function:line" or "file:line".
e.g.
$ perf probe -D kernel_read:8
p:probe/kernel_read_L8 kernel_read+39
p:probe/kernel_read_L8_1 kernel_read+39
This skips such duplicated probe addresses.
Committer testing:
# uname -a
Linux quaco 5.3.0+ #2 SMP Thu Sep 19 16:13:22 -03 2019 x86_64 x86_64 x86_64 GNU/Linux
#
Jiri Olsa [Thu, 12 Sep 2019 10:52:35 +0000 (12:52 +0200)]
perf tools: Fix segfault in cpu_cache_level__read()
We release wrong pointer on error path in cpu_cache_level__read
function, leading to segfault:
(gdb) r record ls
Starting program: /root/perf/tools/perf/perf record ls
...
[ perf record: Woken up 1 times to write data ]
double free or corruption (out)
Thread 1 "perf" received signal SIGABRT, Aborted.
0x00007ffff7463798 in raise () from /lib64/power9/libc.so.6
(gdb) bt
#0 0x00007ffff7463798 in raise () from /lib64/power9/libc.so.6
#1 0x00007ffff7443bac in abort () from /lib64/power9/libc.so.6
#2 0x00007ffff74af8bc in __libc_message () from /lib64/power9/libc.so.6
#3 0x00007ffff74b92b8 in malloc_printerr () from /lib64/power9/libc.so.6
#4 0x00007ffff74bb874 in _int_free () from /lib64/power9/libc.so.6
#5 0x0000000010271260 in __zfree (ptr=0x7fffffffa0b0) at ../../lib/zalloc..
#6 0x0000000010139340 in cpu_cache_level__read (cache=0x7fffffffa090, cac..
#7 0x0000000010143c90 in build_caches (cntp=0x7fffffffa118, size=<optimiz..
...
tools arch x86: Sync asm/cpufeatures.h with the kernel sources
To pick up the changes from:
b4dd4f6e3648 ("x86/vmware: Add a header file for hypercall definitions") f36cf386e3fe ("x86/speculation/swapgs: Exclude ATOMs from speculation through SWAPGS") be261ffce6f1 ("x86: Remove X86_FEATURE_MFENCE_RDTSC") 018ebca8bd70 ("x86/cpufeatures: Enable a new AVX512 CPU feature")
These don't cause any changes in tooling, just silences this perf build
warning:
Warning: Kernel ABI header at 'tools/arch/x86/include/asm/cpufeatures.h' differs from latest version at 'arch/x86/include/asm/cpufeatures.h'
diff -u tools/arch/x86/include/asm/cpufeatures.h arch/x86/include/asm/cpufeatures.h
To clarify, updating those files cause these bits of tools/perf to rebuild:
CC /tmp/build/perf/bench/mem-memcpy-x86-64-asm.o
CC /tmp/build/perf/bench/mem-memset-x86-64-asm.o
INSTALL GTK UI
LD /tmp/build/perf/bench/perf-in.o
tools arch x86 uapi: Synch asm/unistd.h with the kernel sources
To pick up the change in:
45e29d119e99 ("x86/syscalls: Make __X32_SYSCALL_BIT be unsigned long")
That doesn't trigger any changes in tooling and silences this perf build
warning:
Warning: Kernel ABI header at 'tools/arch/x86/include/uapi/asm/unistd.h' differs from latest version at 'arch/x86/include/uapi/asm/unistd.h'
diff -u tools/arch/x86/include/uapi/asm/unistd.h arch/x86/include/uapi/asm/unistd.h
tools uapi asm-generic: Sync unistd.h with the kernel sources
To pick the change from:
78e05972c5e6 ("ipc: fix semtimedop for generic 32-bit architectures")
Which doesn't trigger any change in tooling and silences this perf build
warning:
Warning: Kernel ABI header at 'tools/include/uapi/asm-generic/unistd.h' differs from latest version at 'include/uapi/asm-generic/unistd.h'
diff -u tools/include/uapi/asm-generic/unistd.h include/uapi/asm-generic/unistd.h
tools headers uapi: Sync prctl.h with the kernel sources
To get the changes in:
63f0c6037965 ("arm64: Introduce prctl() options to control the tagged user addresses ABI")
that introduces prctl options that then automagically gets catched by
the prctl cmd table generator, and thus supported in the 'perf trace'
prctl beautifier for the 'option' argument:
# perf record -e syscalls:sys_enter_prctl --filter="option=*TAGGED_ADDR*"
Should be easy enough, first using tracepoint filters, then via the
augmented_raw_syscalls.c BPF method.
This addresses this perf build warning:
Warning: Kernel ABI header at 'tools/include/uapi/linux/prctl.h' differs from latest version at 'include/uapi/linux/prctl.h'
diff -u tools/include/uapi/linux/prctl.h include/uapi/linux/prctl.h
perf kvm: Add arch neutral function to choose event for perf kvm record
'perf kvm record' uses 'cycles'(if the user did not specify any event)
as the default event to profile the guest.
This will not provide any proper samples from the guest incase of
powerpc architecture, since in powerpc the PMUs are controlled by the
guest rather than the host.
Patch adds a function to pick an arch specific event for 'perf kvm
record', instead of selecting 'cycles' as a default event for all
architectures.
For powerpc this function checks for any user specified event, and if
there isn't any it returns invalid instead of proceeding with 'cycles'
event.
perf kvm: Move kvm-stat header file from conditional inclusion to common include section
Move kvm-stat header file to the common include section, and make the
definitions in the header file under the conditional inclusion `#ifdef
HAVE_KVM_STAT_SUPPORT`.
This helps to define other 'perf kvm' related function prototypes in
kvm-stat header file, which may not need kvm-stat support.
perf stat: Fix a segmentation fault when using repeat forever
Observe a segmentation fault when 'perf stat' is asked to repeat forever
with the interval option.
Without fix:
# perf stat -r 0 -I 5000 -e cycles -a sleep 10
# time counts unit events
5.000211692 3,13,89,82,34,157 cycles
10.000380119 1,53,98,52,22,294 cycles
10.040467280 17,16,79,265 cycles
Segmentation fault
This problem was only observed when we use forever option aka -r 0 and
works with limited repeats. Calling print_counter with ts being set to
NULL, is not a correct option when interval is set. Hence avoid
print_counter(NULL,..) if interval is set.
Did the 'git bisect' to find the cset introducing the problem to add the
Fixes tag below, and at that time the problem reproduced as:
(gdb) run stat -r0 -I500 sleep 1
<SNIP>
Program received signal SIGSEGV, Segmentation fault.
print_interval (prefix=prefix@entry=0x7fffffffc8d0 "", ts=ts@entry=0x0) at builtin-stat.c:866
866 sprintf(prefix, "%6lu.%09lu%s", ts->tv_sec, ts->tv_nsec, csv_sep);
(gdb) bt
#0 print_interval (prefix=prefix@entry=0x7fffffffc8d0 "", ts=ts@entry=0x0) at builtin-stat.c:866
#1 0x000000000041860a in print_counters (ts=ts@entry=0x0, argc=argc@entry=2, argv=argv@entry=0x7fffffffd640) at builtin-stat.c:938
#2 0x0000000000419a7f in cmd_stat (argc=2, argv=0x7fffffffd640, prefix=<optimized out>) at builtin-stat.c:1411
#3 0x000000000045c65a in run_builtin (p=p@entry=0x6291b8 <commands+216>, argc=argc@entry=5, argv=argv@entry=0x7fffffffd640) at perf.c:370
#4 0x000000000045c893 in handle_internal_command (argc=5, argv=0x7fffffffd640) at perf.c:429
#5 0x000000000045c8f1 in run_argv (argcp=argcp@entry=0x7fffffffd4ac, argv=argv@entry=0x7fffffffd4a0) at perf.c:473
#6 0x000000000045cac9 in main (argc=<optimized out>, argv=<optimized out>) at perf.c:588
(gdb)
Mostly the same as just before this patch:
Program received signal SIGSEGV, Segmentation fault.
0x00000000005874a7 in print_interval (config=0xa1f2a0 <stat_config>, evlist=0xbc9b90, prefix=0x7fffffffd1c0 "`", ts=0x0) at util/stat-display.c:964
964 sprintf(prefix, "%6lu.%09lu%s", ts->tv_sec, ts->tv_nsec, config->csv_sep);
(gdb) bt
#0 0x00000000005874a7 in print_interval (config=0xa1f2a0 <stat_config>, evlist=0xbc9b90, prefix=0x7fffffffd1c0 "`", ts=0x0) at util/stat-display.c:964
#1 0x0000000000588047 in perf_evlist__print_counters (evlist=0xbc9b90, config=0xa1f2a0 <stat_config>, _target=0xa1f0c0 <target>, ts=0x0, argc=2, argv=0x7fffffffd670)
at util/stat-display.c:1172
#2 0x000000000045390f in print_counters (ts=0x0, argc=2, argv=0x7fffffffd670) at builtin-stat.c:656
#3 0x0000000000456bb5 in cmd_stat (argc=2, argv=0x7fffffffd670) at builtin-stat.c:1960
#4 0x00000000004dd2e0 in run_builtin (p=0xa30e00 <commands+288>, argc=5, argv=0x7fffffffd670) at perf.c:310
#5 0x00000000004dd54d in handle_internal_command (argc=5, argv=0x7fffffffd670) at perf.c:362
#6 0x00000000004dd694 in run_argv (argcp=0x7fffffffd4cc, argv=0x7fffffffd4c0) at perf.c:406
#7 0x00000000004dda11 in main (argc=5, argv=0x7fffffffd670) at perf.c:531
(gdb)
prev_raw_counts is allocated when using intervals. This is used when
calculating the difference in the counts of events when using interval.
The current counts are stored in prev_raw_counts to calculate the
differences in the next iteration.
On the first interval of the second and subsequent repetitions,
prev_raw_counts would be the values stored in the last interval of the
previous repetitions, while the current counts will only be for the
first interval of the current repetition.
Hence there is a possibility of events showing up as big number.
Fix this by resetting prev_raw_counts whenever perf stat repeats the
command.
With the fix:
# perf stat -r 3 -I 2000 -e faults -e sched:sched_switch -a sleep 5
This was broken since the cset introducing the --interval feature, i.e.
--repeat + --interval wasn't tested at that point, add the Fixes tag so
that automatic scripts can pick this up.
perf tools: Move event synthesizing routines to separate .c file
For better grouping, in time we may end up making most of these static,
i.e. generalizing the 'perf record' synthesizing code so that based on
the target it can do the right thing and call the needed synthesizers.
perf tools: Move event synthesizing routines to separate header
Those are the only routines using the perf_event__handler_t typedef and
are all related, so move to a separate header to reduce the header
dependency tree, lots of places were getting event.h and even stdio.h,
limits.h indirectly, so fix those as well.