When converting from atomic_t to refcount_t we didn't follow the usual
step of initializing it to one before taking any new reference, which
trips over checking if taking a reference for a freed refcount_t, fix
it.
Brendan's report:
---
It's 4.12-rc7, with node v4.4.1. I'm building 4.13-rc1 now, as I hit
what I think is another unrelated perf bug and I'm starting to wonder
what else is broken on that version:
(root) /mnt/src/linux-4.12-rc7/tools/perf # ./perf record -F 99 -a -e
cpu-clock --cgroup=docker/f9e9d5df065b14646e8a11edc837a13877fd90c171137b2ba3feb67a0201cb65
-g
perf: /mnt/src/linux-4.12-rc7/tools/include/linux/refcount.h:108:
refcount_inc: Assertion `!(!refcount_inc_not_zero(r))' failed.
Aborted
that used to work...
---
Testing it:
Before:
# perf stat -e cycles -C 0 --cgroup /
perf: /home/acme/git/linux/tools/include/linux/refcount.h:108: refcount_inc: Assertion `!(!refcount_inc_not_zero(r))' failed.
Aborted (core dumped)
#
After:
# perf stat -e cycles -C 0 --cgroup /
^C
Performance counter stats for 'CPU(s) 0':
132,081,393 cycles /
2.492942763 seconds time elapsed
#
Reported-by: Brendan Gregg <brendan.d.gregg@gmail.com>
Acked-by: Elena Reshetova <elena.reshetova@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: David Carrillo-Cisneros <davidcc@google.com>
Cc: Kees Kook <keescook@chromium.org>
Cc: Krister Johansen <kjlx@templeofstupid.com>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Sudeep Holla <Sudeep.Holla@arm.com>
Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
Cc: Wang Nan <wangnan0@huawei.com>
Fixes: 79c5fe6db8 ("perf cgroup: Convert cgroup_sel.refcnt from atomic_t to refcount_t")
Link: http://lkml.kernel.org/n/tip-l7ovfblq14ip2i08m1g0fkhv@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
On s390x the kernel text segment starts at address 0x0. When perf
report reads kernel symbols from vmlinux file it adds an offset of
0x1000.
For example see symbol set_reset_devices:
[root@s8360047 linux-devel]# nm -A vmlinux| fgrep set_reset_devices
vmlinux:0000000001379000 t set_reset_devices
[root@s8360047 linux-devel]#
[root@s8360047 linux-devel]# fgrep set_reset_devices /proc/kallsyms
0000000001379000 t set_reset_devices
[root@s8360047 linux-devel]#
The kernel symbol table and the vmlinux file have the same address for
symbol set_reset_devices namely 1379000.
When perf report reads this symbols it displays it with address
symbol__new: set_reset_devices 0x137a000-0x137a018
There is a difference between perf report and vmlinux of 0x1000.
The reason for the difference is at kernel symbol load time in function
dso__load_sym(). The vmlinux file is investigated with its ELF header.
Command readelf shows this:
Section Headers:
[Nr] Name Type Address Offset
Size EntSize Flags Link Info Align
[ 0] NULL 0000000000000000 00000000
0000000000000000 0000000000000000 0 0 0
[ 1] .text PROGBITS 0000000000000000 00001000
0000000000b0e0c2 0000000000000000 AX 0 0 128
This leads to an invalid calculation of the symbol start address, see
file utit/symbol-elf.c line 974:
/* Adjust symbol to map to file offset */
if (adjust_kernel_syms)
sym.st_value -= shdr.sh_addr - shdr.sh_offset;
With shdr.sh_addr set to 0x0 and shdr.sh_offset set to 0x1000 as read
from the ELF .text section 0x1000 is added to the symbol address.
I would like to fix this by introducing an archticture specific function
named elf__needs_adjust_symbols(). This is the same approach as done by
PowerPC. The function currently does not exist for s390x and the
default weak one is used. The s390x specific one returns false when
symsrc_init() is invoked for kernel symbols and results in variable
adjust_kernel_syms being false. This omits the adjustment and the
correct address is displayed (when symbol resolvement does not work).
The s390x specific function returns false for kernel symbol adjustment
and returns true for kernel modules, processes and shared libraries.
Signed-off-by: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Cc: Thomas-Mich Richter <tmricht@linux.vnet.ibm.com>
LPU-Reference: 20170713130252.6167-1-tmricht@linux.vnet.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Pull perf fixes from Ingo Molnar:
"Two hw-enablement patches, two race fixes, three fixes for regressions
of semantics, plus a number of tooling fixes"
* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86/intel: Add proper condition to run sched_task callbacks
perf/core: Fix locking for children siblings group read
perf/core: Fix scheduling regression of pinned groups
perf/x86/intel: Fix debug_store reset field for freq events
perf/x86/intel: Add Goldmont Plus CPU PMU support
perf/x86/intel: Enable C-state residency events for Apollo Lake
perf symbols: Accept zero as the kernel base address
Revert "perf/core: Drop kernel samples even though :u is specified"
perf annotate: Fix broken arrow at row 0 connecting jmp instruction to its target
perf evsel: State in the default event name if attr.exclude_kernel is set
perf evsel: Fix attr.exclude_kernel setting for default cycles:p
To make it more clear that it is the sum of all the nr_samples fields in the
addr[] entries, i.e.:
sym_hist->nr_samples = sum(sym_hist->addr[0 .. symbol__size(sym)]->nr_samples)
Committer notes:
Taeung had renamed it to total_samples, but using nr_samples, as in the
added explanation above, looks clearer and establishes the direct
connection, making clear it is about the _number_ of samples.
Signed-off-by: Taeung Song <treeze.taeung@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/1500500211-16599-1-git-send-email-treeze.taeung@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
struct sym_hist has addr[] but it should have not only number of samples
but also the sample period. So use new struct symhist_entry to pave the
way to have that.
Committer notes:
This initial patch will only introduce the struct sym_hist_entry and use
only the nr_samples member, which makes the code clearer and paves the
way to save the period as well.
Signed-off-by: Taeung Song <treeze.taeung@gmail.com>
Suggested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/r/1500500205-16553-1-git-send-email-treeze.taeung@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Avoiding a loop, so now its quite convenient to ssh to a machine and
then simply do:
# perf trace
To trace all syscalls without causing a loop.
This was possible using --filter-pids, i.e. once you noticed the loop,
get the sshd pid and add it to --filter-pids, restarting the 'perf
trace'.
Now to figure out how to do that in a X terminal, the other common
scenario, which is way more involved, as there are multiple processes
communicating to process terminal activity...
Using --filter-pids + '-e \!syscall,names,you,dont,need' may be a good
approximation when having to do syswide tracing on your workstation.
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: http://lkml.kernel.org/n/tip-68rjeao9wnpylla41htk7xps@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Previously we only used the syscall_fmt when we had sc->tp_format set,
i.e. when we found the (enter, exit) pair in tracefs/events/syscalls/.
But we really only need to use what is in sc->arg_fmt to apply the arg
beautifiers to the syscall argument values, so do it.
With this we will be able to provide formatters to the "clone" syscall,
which doesn't have entries in tracefs/events/syscalls/.
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: http://lkml.kernel.org/n/tip-y41nl41jrayjo5ucnde2peix@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
An earlier kernel patch allowed enabling PT and LBR at the same time on
Goldmont.
commit ccbebba4c6 ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity
if the core supports it")
However, users still cannot use Intel PT and LBRs simultaneously. $
sudo perf record -e cycles,intel_pt//u -b -- sleep 1 Error: PMU
Hardware doesn't support sampling/overflow-interrupts.
PT implicitly adds dummy event in perf tool. dummy event is software
event which doesn't support LBR.
Always setting no branch for dummy event in Intel PT.
Signed-off-by: Kan Liang <kan.liang@intel.com>
Acked-by: Jiri Olsa <jolsa@redhat.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20170630141656.1626-2-kan.liang@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Show branch type in callchain entry. The branch type is printed
with other LBR information (such as cycles/abort/...).
For example:
perf record -g -j any,save_type
perf report --branch-history --stdio --no-children
38.50% div.c:45 [.] main div
|
---main div.c:42 (RET CROSS_2M cycles:2)
compute_flag div.c:28 (cycles:2)
compute_flag div.c:27 (RET CROSS_2M cycles:1)
rand rand.c:28 (cycles:1)
rand rand.c:28 (RET CROSS_2M cycles:1)
__random random.c:298 (cycles:1)
__random random.c:297 (COND_BWD CROSS_2M cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (COND_BWD CROSS_2M cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (RET CROSS_2M cycles:9)
Change log
v6: Remove the branch_type_str() since it's moved to branch.c.
v5: Rewrite the branch info print code in util/callchain.c.
v4: Comparing to previous version, the major changes are:
Signed-off-by: Yao Jin <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: Michael Ellerman <mpe@ellerman.id.au>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1500379995-6449-8-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Show the branch type statistics at the end of perf report --stdio.
For example:
perf report --stdio
COND_FWD: 28.5%
COND_BWD: 9.4%
CROSS_4K: 0.7%
CROSS_2M: 14.1%
COND: 37.9%
UNCOND: 0.2%
IND: 6.7%
CALL: 26.5%
RET: 28.7%
SYSRET: 0.0%
The branch types are:
COND_FWD: conditional forward
COND_BWD: conditional backward
COND: conditional branch
UNCOND: unconditional branch
IND: indirect
CALL: function call
IND_CALL: indirect function call
RET: function return
SYSCALL: syscall
SYSRET: syscall return
COND_CALL: conditional function call
COND_RET: conditional function return
CROSS_4K and CROSS_2M:
They are the metrics checking for branches cross 4K or 2MB pages.
It's an approximate computing. We don't know if the area is 4K or
2MB, so always compute both.
To make the output simple, if a branch crosses 2M area, CROSS_4K
will not be incremented.
Change log
v7: Since the common branch type definitions are changed, some
tags/strings are updated accordingly.
v6: Remove branch_type_stat_display() since it's moved to branch.c.
v5: Remove the unnecessary sort__mode checking in
hist_iter__branch_callback().
v4: Comparing to previous version, the major changes are:
Add the computing of JCC forward/JCC backward and cross page checking
by using the from and to addresses.
Signed-off-by: Yao Jin <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: Michael Ellerman <mpe@ellerman.id.au>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1500379995-6449-7-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Create new util/branch.c and util/branch.h to contain the common branch
functions. Such as:
branch_type_count(): Count the numbers of branch types
branch_type_name() : Return the name of branch type
branch_type_stat_display(): Display branch type statistics info
branch_type_str(): Construct the branch type string.
The branch type is saved in branch_flags.
Change log:
v8: Change PERF_BR_NONE to PERF_BR_UNKNOWN.
v7: Since the common branch type name is changed (e.g. JCC->COND),
this patch is performed the modification accordingly.
v6: Move that multiline conditional code inside {} brackets.
Move branch_type_stat_display() from builtin-report.c to
branch.c.
Move branch_type_str() from callchain.c to branch.c.
v5: It's a new patch in v5 patch series.
Signed-off-by: Yao Jin <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: Michael Ellerman <mpe@ellerman.id.au>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1500379995-6449-6-git-send-email-yao.jin@linux.intel.com
[ Don't use 'index' and 'stat' as names for variables, it shadows global decls in older distros ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The branch info such as predicted/cycles/... are printed at the
callchain entries.
For example: perf report --branch-history --no-children --stdio
--1.07%--main div.c:39 (predicted:52.4% cycles:1 iterations:17)
main div.c:44 (predicted:52.4% cycles:1)
main div.c:42 (cycles:2)
compute_flag div.c:28 (cycles:2)
compute_flag div.c:27 (cycles:1)
rand rand.c:28 (cycles:1)
rand rand.c:28 (cycles:1)
__random random.c:298 (cycles:1)
__random random.c:297 (cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (cycles:1)
But the current code is difficult to maintain and extend. This patch
refactors the code for easy maintenance.
Change log:
v6: 1. Put the multiline condition code into {} brackets in
counts_str_build()
2. Keep the original display order, that is:
predicted, abort, cycles, iterations
v5: It's a new patch in v5 patch series.
Signed-off-by: Yao Jin <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: Michael Ellerman <mpe@ellerman.id.au>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1500379995-6449-5-git-send-email-yao.jin@linux.intel.com
[ Don't use 'index' as a name for a variable, it shadows a globa decl in older distros ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Add header record types to pipe-mode, reusing the functions
used in file-mode and leveraging the new struct feat_fd.
For alignment, check that synthesized events don't exceed
pagesize.
Add the perf_event__synthesize_feature event call back to
process the new header records.
Before this patch:
$ perf record -o - -e cycles sleep 1 | perf report --stdio --header
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
...
After this patch:
$ perf record -o - -e cycles sleep 1 | perf report --stdio --header
# ========
# captured on: Mon May 22 16:33:43 2017
# ========
#
# hostname : my_hostname
# os release : 4.11.0-dbx-up_perf
# perf version : 4.11.rc6.g6277c80
# arch : x86_64
# nrcpus online : 72
# nrcpus avail : 72
# cpudesc : Intel(R) Xeon(R) CPU E5-2696 v3 @ 2.30GHz
# cpuid : GenuineIntel,6,63,2
# total memory : 263457192 kB
# cmdline : /root/perf record -o - -e cycles -c 100000 sleep 1
# HEADER_CPU_TOPOLOGY info available, use -I to display
# HEADER_NUMA_TOPOLOGY info available, use -I to display
# pmu mappings: intel_bts = 6, uncore_imc_4 = 22, uncore_sbox_1 = 47, uncore_cbox_5 = 33, uncore_ha_0 = 16, uncore_cbox
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.000 MB - ]
...
Support added for the subcommands: report, inject, annotate and script.
Signed-off-by: David Carrillo-Cisneros <davidcc@google.com>
Acked-by: David Ahern <dsahern@gmail.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Simon Que <sque@chromium.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20170718042549.145161-16-davidcc@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
There are three FEAT_OP* macros:
- FEAT_OPA: for features without process record.
- FEAT_OPP: for features with process record.
- FEAT_OPF: like FEAT_OPP but to show only if show_full_info flags
is set.
To add pipe-mode headers we need yet another variation of the macros
(one to specify whether a feature generates an auxiliar record).
Instead, we redefine macros so that:
- show_full_info is specified as an argument (to remove the
FEAT_OPF variation) and,
- it always sets "process" handler (to remove the FEAT_OPA variation).
Individual process handlers can be NULLed individually.
This allows to define two variations only:
- FEAT_OPR: synthesizes auxiliar event record.
- FEAT_OPN: doesn't synthesize an auxiliar event record.
Signed-off-by: David Carrillo-Cisneros <davidcc@google.com>
Acked-by: David Ahern <dsahern@gmail.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Turner <pjt@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Simon Que <sque@chromium.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20170718042549.145161-14-davidcc@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
