The events defined in pmu-events JSON are parsed and added into perf
tool. For fixed counters, we handle the encodings between JSON and perf
by using a static array fixed[].
But the fixed[] has missed an important event "cpu_clk_unhalted.core".
For example, on the Tremont platform,
[root@localhost ~]# perf stat -e cpu_clk_unhalted.core -a
event syntax error: 'cpu_clk_unhalted.core'
\___ parser error
With this patch, the event cpu_clk_unhalted.core can be parsed.
[root@localhost perf]# ./perf stat -e cpu_clk_unhalted.core -a -vvv
------------------------------------------------------------
perf_event_attr:
type 4
size 112
config 0x3c
sample_type IDENTIFIER
read_format TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
disabled 1
inherit 1
exclude_guest 1
------------------------------------------------------------
...
Signed-off-by: Jin Yao <yao.jin@linux.intel.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://lkml.kernel.org/r/20190729072755.2166-1-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
During execution of command 'perf top' the error message:
Not enough memory for annotating '__irf_end' symbol!)
is emitted from this call sequence:
__cmd_top
perf_top__mmap_read
perf_top__mmap_read_idx
perf_event__process_sample
hist_entry_iter__add
hist_iter__top_callback
perf_top__record_precise_ip
hist_entry__inc_addr_samples
symbol__inc_addr_samples
symbol__get_annotation
symbol__alloc_hist
In this function the size of symbol __irf_end is calculated. The size of
a symbol is the difference between its start and end address.
When the symbol was read the first time, its start and end was set to:
symbol__new: __irf_end 0xe954d0-0xe954d0
which is correct and maps with /proc/kallsyms:
root@s8360046:~/linux-4.15.0/tools/perf# fgrep _irf_end /proc/kallsyms
0000000000e954d0 t __irf_end
root@s8360046:~/linux-4.15.0/tools/perf#
In function symbol__alloc_hist() the end of symbol __irf_end is
symbol__alloc_hist sym:__irf_end start:0xe954d0 end:0x3ff80045a8
which is identical with the first module entry in /proc/kallsyms
This results in a symbol size of __irf_req for histogram analyses of
70334140059072 bytes and a malloc() for this requested size fails.
The root cause of this is function
__dso__load_kallsyms()
+-> symbols__fixup_end()
Function symbols__fixup_end() enlarges the last symbol in the kallsyms
map:
# fgrep __irf_end /proc/kallsyms
0000000000e954d0 t __irf_end
#
to the start address of the first module:
# cat /proc/kallsyms | sort | egrep ' [tT] '
....
0000000000e952d0 T __security_initcall_end
0000000000e954d0 T __initramfs_size
0000000000e954d0 t __irf_end
000003ff800045a8 T fc_get_event_number [scsi_transport_fc]
000003ff800045d0 t store_fc_vport_disable [scsi_transport_fc]
000003ff800046a8 T scsi_is_fc_rport [scsi_transport_fc]
000003ff800046d0 t fc_target_setup [scsi_transport_fc]
On s390 the kernel is located around memory address 0x200, 0x10000 or
0x100000, depending on linux version. Modules however start some- where
around 0x3ff xxxx xxxx.
This is different than x86 and produces a large gap for which histogram
allocation fails.
Fix this by detecting the kernel's last symbol and do no adjustment for
it. Introduce a weak function and handle s390 specifics.
Reported-by: Klaus Theurich <klaus.theurich@de.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/20190724122703.3996-2-tmricht@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
On s390 the modules loaded in memory have the text segment located after
the GOT and Relocation table. This can be seen with this output:
[root@m35lp76 perf]# fgrep qeth /proc/modules
qeth 151552 1 qeth_l2, Live 0x000003ff800b2000
...
[root@m35lp76 perf]# cat /sys/module/qeth/sections/.text
0x000003ff800b3990
[root@m35lp76 perf]#
There is an offset of 0x1990 bytes. The size of the qeth module is
151552 bytes (0x25000 in hex).
The location of the GOT/relocation table at the beginning of a module is
unique to s390.
commit 203d8a4aa6 ("perf s390: Fix 'start' address of module's map")
adjusts the start address of a module in the map structures, but does
not adjust the size of the modules. This leads to overlapping of module
maps as this example shows:
[root@m35lp76 perf] # ./perf report -D
0 0 0xfb0 [0xa0]: PERF_RECORD_MMAP -1/0: [0x3ff800b3990(0x25000)
@ 0]: x /lib/modules/.../qeth.ko.xz
0 0 0x1050 [0xb0]: PERF_RECORD_MMAP -1/0: [0x3ff800d85a0(0x8000)
@ 0]: x /lib/modules/.../ip6_tables.ko.xz
The module qeth.ko has an adjusted start address modified to b3990, but
its size is unchanged and the module ends at 0x3ff800d8990. This end
address overlaps with the next modules start address of 0x3ff800d85a0.
When the size of the leading GOT/Relocation table stored in the
beginning of the text segment (0x1990 bytes) is subtracted from module
qeth end address, there are no overlaps anymore:
0x3ff800d8990 - 0x1990 = 0x0x3ff800d7000
which is the same as
0x3ff800b2000 + 0x25000 = 0x0x3ff800d7000.
To fix this issue, also adjust the modules size in function
arch__fix_module_text_start(). Add another function parameter named size
and reduce the size of the module when the text segment start address is
changed.
Output after:
0 0 0xfb0 [0xa0]: PERF_RECORD_MMAP -1/0: [0x3ff800b3990(0x23670)
@ 0]: x /lib/modules/.../qeth.ko.xz
0 0 0x1050 [0xb0]: PERF_RECORD_MMAP -1/0: [0x3ff800d85a0(0x7a60)
@ 0]: x /lib/modules/.../ip6_tables.ko.xz
Reported-by: Stefan Liebler <stli@linux.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Hendrik Brueckner <brueckner@linux.ibm.com>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: stable@vger.kernel.org
Fixes: 203d8a4aa6 ("perf s390: Fix 'start' address of module's map")
Link: http://lkml.kernel.org/r/20190724122703.3996-1-tmricht@linux.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The code to disassemble BPF programs uses binutil's disassembling
routines, and those use in turn fprintf to print to a memstream FILE,
adding a newline at the end of each line, which ends up confusing the
TUI routines called from:
annotate_browser__write()
annotate_line__write()
annotate_browser__printf()
ui_browser__vprintf()
SLsmg_vprintf()
The SLsmg_vprintf() function in the slang library gets confused with the
terminating newline, so make the disasm_line__parse() function that
parses the lines produced by the BPF specific disassembler (that uses
binutil's libopcodes) and the lines produced by the objdump based
disassembler used for everything else (and that doesn't adds this
terminating newline) trim the end of the line in addition of the
beginning.
This way when disasm_line->ops.raw, i.e. for instructions without a
special scnprintf() method, we'll not have that \n getting in the way of
filling the screen right after the instruction with spaces to avoid
leaving what was on the screen before and thus garbling the annotation
screen, breaking scrolling, etc.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Song Liu <songliubraving@fb.com>
Fixes: 6987561c9e ("perf annotate: Enable annotation of BPF programs")
Link: https://lkml.kernel.org/n/tip-unbr5a5efakobfr6rhxq99ta@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Michael reported an issue with perf bench numa failing with binding to
cpu0 with '-0' option.
# perf bench numa mem -p 3 -t 1 -P 512 -s 100 -zZcm0 --thp 1 -M 1 -ddd
# Running 'numa/mem' benchmark:
# Running main, "perf bench numa numa-mem -p 3 -t 1 -P 512 -s 100 -zZcm0 --thp 1 -M 1 -ddd"
binding to node 0, mask: 0000000000000001 => -1
perf: bench/numa.c:356: bind_to_memnode: Assertion `!(ret)' failed.
Aborted (core dumped)
This happens when the cpu0 is not part of node0, which is the benchmark
assumption and we can see that's not the case for some powerpc servers.
Using correct node for cpu0 binding.
Reported-by: Michael Petlan <mpetlan@redhat.com>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Satheesh Rajendran <sathnaga@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/20190801142642.28004-1-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When building our local version of perf with MSAN (Memory Sanitizer) and
running the perf record command, MSAN throws a use of uninitialized
value warning in "tools/perf/util/util.c:333:6".
This warning stems from the "buf" variable being passed into "write".
It originated as the variable "ev" with the type union perf_event*
defined in the "perf_event__synthesize_attr" function in
"tools/perf/util/header.c".
In the "perf_event__synthesize_attr" function they allocate space with a malloc
call using ev, then go on to only assign some of the member variables before
passing "ev" on as a parameter to the "process" function therefore "ev"
contains uninitialized memory. Changing the malloc call to zalloc to initialize
all the members of "ev" which gets rid of the warning.
To reproduce this warning, build perf by running:
make -C tools/perf CLANG=1 CC=clang EXTRA_CFLAGS="-fsanitize=memory\
-fsanitize-memory-track-origins"
(Additionally, llvm might have to be installed and clang might have to
be specified as the compiler - export CC=/usr/bin/clang)
then running:
tools/perf/perf record -o - ls / | tools/perf/perf --no-pager annotate\
-i - --stdio
Please see the cover letter for why false positive warnings may be
generated.
Signed-off-by: Numfor Mbiziwo-Tiapo <nums@google.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mark Drayton <mbd@fb.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20190724234500.253358-2-nums@google.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Pull perf/urgent fixes from Arnaldo Carvalho de Melo:
perf.data:
Alexey Budankov:
- Fix loading of compressed data split across adjacent records
Jiri Olsa:
- Fix buffer size setting for processing CPU topology perf.data header.
perf stat:
Jiri Olsa:
- Fix segfault for event group in repeat mode
Cong Wang:
- Always separate "stalled cycles per insn" line, it was being appended to
the "instructions" line.
perf script:
Andi Kleen:
- Fix --max-blocks man page description.
- Improve man page description of metrics.
- Fix off by one in brstackinsn IPC computation.
perf probe:
Arnaldo Carvalho de Melo:
- Avoid calling freeing routine multiple times for same pointer.
perf build:
- Do not use -Wshadow on gcc < 4.8, avoiding too strict warnings
treated as errors, breaking the build.
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
When perf_add_probe_events() we call cleanup_perf_probe_events() for the
pev pointer it receives, then, as part of handling this failure the main
'perf probe' goes on and calls cleanup_params() and that will again call
cleanup_perf_probe_events()for the same pointer, so just set nevents to
zero when handling the failure of perf_add_probe_events() to avoid the
double free.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lkml.kernel.org/n/tip-x8qgma4g813z96dvtw9w219q@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
So that, when perf_add_probe_events() fails, like in:
# perf probe icmp_rcv:64 "type=icmph->type"
Failed to find 'icmph' in this function.
Error: Failed to add events.
Segmentation fault (core dumped)
#
We don't segfault.
clear_perf_probe_event() was zeroing the whole pev, and since the switch
to zfree() for the members in the pev, that memset() was removed, which
left nargs with its original value, in the above case 1.
With the memset the same pev could be passed to clear_perf_probe_event()
multiple times, since all it would have would be zeroes, and free()
accepts zero, the loop would not happen and we would just memset it
again to zeroes.
Without it we got that segfault, so zero nargs to keep it like it was,
next cset will avoid calling clear_perf_probe_event() for the same pevs
in case of failure.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Fixes: d8f9da2404 ("perf tools: Use zfree() where applicable")
Link: https://lkml.kernel.org/n/tip-802f2jypnwqsvyavvivs8464@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Fix decompression failure found during the loading of compressed trace
collected on larger scale systems (>48 cores).
The error happened due to lack of decompression space for a mmaped
buffer data chunk split across adjacent PERF_RECORD_COMPRESSED records.
$ perf report -i bt.16384.data --stats
failed to decompress (B): 63869 -> 0 : Destination buffer is too small
user stack dump failure
Can't parse sample, err = -14
0x2637e436 [0x4080]: failed to process type: 9
Error:
failed to process sample
$ perf test 71
71: Zstd perf.data compression/decompression : Ok
Signed-off-by: Alexey Budankov <alexey.budankov@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: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/4d839e1b-9c48-89c4-9702-a12217420611@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
The "stalled cycles per insn" is appended to "instructions" when the CPU
has this hardware counter directly. We should always make it a separate
line, which also aligns to the output when we hit the "if (total &&
avg)" branch.
Before:
$ sudo perf stat --all-cpus --field-separator , --log-fd 1 -einstructions,cycles -- sleep 1
4565048704,,instructions,64114578096,100.00,1.34,insn per cycle,,
3396325133,,cycles,64146628546,100.00,,
After:
$ sudo ./tools/perf/perf stat --all-cpus --field-separator , --log-fd 1 -einstructions,cycles -- sleep 1
6721924,,instructions,24026790339,100.00,0.22,insn per cycle
,,,,,0.00,stalled cycles per insn
30939953,,cycles,24025512526,100.00,,
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Andi Kleen <ak@linux.intel.com>
Link: http://lkml.kernel.org/r/20190517221039.8975-1-xiyou.wangcong@gmail.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Numfor Mbiziwo-Tiapo reported segfault on stat of event group in repeat
mode:
# perf stat -e '{cycles,instructions}' -r 10 ls
It's caused by memory corruption due to not cleaned evsel's id array and
index, which needs to be rebuilt in every stat iteration. Currently the
ids index grows, while the array (which is also not freed) has the same
size.
Fixing this by releasing id array and zeroing ids index in
perf_evsel__close function.
We also need to keep the evsel_list alive for stat record (which is
disabled in repeat mode).
Reported-by: Numfor Mbiziwo-Tiapo <nums@google.com>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ian Rogers <irogers@google.com>
Cc: Mark Drayton <mbd@fb.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Song Liu <songliubraving@fb.com>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20190715142121.GC6032@krava
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Jin Yao
Thomas Richter
Ian Rogers
Masanari Iida
He Zhe
Adrian Hunter
Arnaldo Carvalho de Melo
Jiri Olsa
Vince Weaver
Numfor Mbiziwo-Tiapo
Ingo Molnar
Alexey Budankov
Cong Wang
Jiri Olsa
Andi Kleen