This a no-op for most of the perf tool. The reference count is set to
1 at allocation, the put will see the 1, decrement it and perform the
delete.
The purpose for adding the reference count is for the python code. Prior
to this change the python code would clone evlists, but this has issues
if events are opened, etc.
This change adds a reference count for the evlists and a later change
will add it to evsels. The combination is needed for the python code to
operate correctly (not hit asserts in the evsel clone), but the changes
are broken apart for the sake of smaller patches.
Assisted-by: Gemini:gemini-3.1-pro-preview
Signed-off-by: Ian Rogers <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alice Rogers <[email protected]>
Cc: Dapeng Mi <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Thomas Richter <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Signed-off-by: Namhyung Kim <[email protected]>
filename__decompress() has an early return path for files that are not
actually compressed. This path returns the fd from open() directly but
never writes to the pathname output parameter, leaving the caller with
an uninitialized buffer despite a successful return.
Callers like dso__decompress_kmodule_path() pass pathname to
decompress_kmodule() which uses it to set the decompressed file path.
If pathname is uninitialized, subsequent operations on the path produce
undefined behavior.
Fix by setting pathname to an empty string on the uncompressed path.
Callers already check for an empty pathname to distinguish temporary
decompressed files (which need unlink) from the original file.
Reported-by: sashiko-bot <[email protected]>
Fixes: 7ac22b088a ("perf tools: Add filename__decompress function")
Cc: Jiri Olsa <[email protected]>
Assisted-by: Claude:claude-opus-4.6
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Instead of using zalloc(nr_entries * sizeof_entry) that is what calloc()
does.
In some places where linux/zalloc.h isn't needed, remove it, add when
needed and was getting it indirectly.
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Signed-off-by: Namhyung Kim <[email protected]>
Without a PMU perf matches an event against any PMU with the
event. Unfortunately some PMU drivers advertise a "cycles" event which
is typically just a core event. As tests assume a core event, switch
to use "cpu-cycles" that avoids the overloaded "cycles" event on
troublesome PMUs and is so far not overloaded. Note, on x86 this
changes a legacy event into a sysfs one.
Signed-off-by: Ian Rogers <[email protected]>
Tested-by: James Clark <[email protected]>
Signed-off-by: Namhyung Kim <[email protected]>
We already only test each kcore map once, but on slow systems
(particularly with network filesystems) even the non-kcore maps are
slow.
The test can test the same objdump output over and over which only wastes
time. Generalize the skipping mechanism to track all DSOs and addresses
so that each section is only tested once.
On a fully loaded ARM Juno (simulating a parallel 'perf test' run) with
a network filesystem, the original runtime is:
real 1m51.126s
user 0m19.445s
sys 1m15.431s
And the new runtime is:
real 0m48.873s
user 0m8.031s
sys 0m32.353s
Committer testing:
# perf test "code read"
22: Object code reading : Ok
#
Reviewed-by: Ian Rogers <[email protected]>
Signed-off-by: James Clark <[email protected]>
Tested-by: Arnaldo Carvalho de Melo <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
The perf_env global variable holds the host perf_env data but its use
is hit and miss. Switch to using local perf_env variables and ensure
scoped perf_env__init and perf_env__exit. This loses command line
setting of the perf_env, but this doesn't matter for tests. So the
perf_env is fully initialized, clear it with memset in perf_env__init.
Signed-off-by: Ian Rogers <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
The struct dump_regs contains 512 bytes of cache_regs, meaning the two
values in perf_sample contribute 1088 bytes of its total 1384 bytes
size. Initializing this much memory has a cost reported by Tavian
Barnes <[email protected]> as about 2.5% when running `perf
script --itrace=i0`:
https://lore.kernel.org/lkml/d841b97b3ad2ca8bcab07e4293375fb7c32dfce7.1736618095.git.tavianator@tavianator.com/
Adrian Hunter <[email protected]> replied that the zero
initialization was necessary and couldn't simply be removed.
This patch aims to strike a middle ground of still zeroing the
perf_sample, but removing 79% of its size by make user_regs and
intr_regs optional pointers to zalloc-ed memory. To support the
allocation accessors are created for user_regs and intr_regs. To
support correct cleanup perf_sample__init and perf_sample__exit
functions are created and added throughout the code base.
Signed-off-by: Ian Rogers <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
Add reference count checking to struct dso, this can help with
implementing correct reference counting discipline. To avoid
RC_CHK_ACCESS everywhere, add accessor functions for the variables in
struct dso.
The majority of the change is mechanical in nature and not easy to
split up.
Committer testing:
'perf test' up to this patch shows no regressions.
But:
util/symbol.c: In function ‘dso__load_bfd_symbols’:
util/symbol.c:1683:9: error: too few arguments to function ‘dso__set_adjust_symbols’
1683 | dso__set_adjust_symbols(dso);
| ^~~~~~~~~~~~~~~~~~~~~~~
In file included from util/symbol.c:21:
util/dso.h:268:20: note: declared here
268 | static inline void dso__set_adjust_symbols(struct dso *dso, bool val)
| ^~~~~~~~~~~~~~~~~~~~~~~
make[6]: *** [/home/acme/git/perf-tools-next/tools/build/Makefile.build:106: /tmp/tmp.ZWHbQftdN6/util/symbol.o] Error 1
MKDIR /tmp/tmp.ZWHbQftdN6/tests/workloads/
make[6]: *** Waiting for unfinished jobs....
This was updated:
- symbols__fixup_end(&dso->symbols, false);
- symbols__fixup_duplicate(&dso->symbols);
- dso->adjust_symbols = 1;
+ symbols__fixup_end(dso__symbols(dso), false);
+ symbols__fixup_duplicate(dso__symbols(dso));
+ dso__set_adjust_symbols(dso);
But not build tested with BUILD_NONDISTRO and libbfd devel files installed
(binutils-devel on fedora).
Add the missing argument:
symbols__fixup_end(dso__symbols(dso), false);
symbols__fixup_duplicate(dso__symbols(dso));
- dso__set_adjust_symbols(dso);
+ dso__set_adjust_symbols(dso, true);
Signed-off-by: Ian Rogers <[email protected]>
Tested-by: Arnaldo Carvalho de Melo <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Ahelenia Ziemiańska <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ben Gainey <[email protected]>
Cc: Changbin Du <[email protected]>
Cc: Chengen Du <[email protected]>
Cc: Colin Ian King <[email protected]>
Cc: Dima Kogan <[email protected]>
Cc: Ilkka Koskinen <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: James Clark <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: K Prateek Nayak <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Li Dong <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Paran Lee <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Song Liu <[email protected]>
Cc: Sun Haiyong <[email protected]>
Cc: Thomas Richter <[email protected]>
Cc: Tiezhu Yang <[email protected]>
Cc: Yanteng Si <[email protected]>
Cc: zhaimingbing <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
PERF_PMU_CAP_EXTENDED_HW_TYPE results in multiple events being opened on
heterogeneous systems. Currently this test only sets its required
attributes on the first event. Not disabling enable_on_exec on the other
events causes the test to fail because the forked objdump processes are
sampled. No tracking event is opened so Perf only knows about its own
mappings causing the objdump samples to give the following error:
$ perf test -vvv "object code reading"
Reading object code for memory address: 0xffff9aaa55ec
thread__find_map failed
---- end(-1) ----
24: Object code reading : FAILED!
Fixes: 251aa04024 ("perf parse-events: Wildcard most "numeric" events")
Reviewed-by: Ian Rogers <[email protected]>
Signed-off-by: James Clark <[email protected]>
Acked-by: Namhyung Kim <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Athira Rajeev <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Kan Liang <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Spoorthy S <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
It tries cycles (or cpu-clock on s390) event with exclude_kernel bit to
open. But other arch on a VM can fail with the hardware event and need
to fallback to the software event in the same way.
So let's get rid of the cpuid check and use generic fallback mechanism
using an array of event candidates. Now event in the odd index excludes
the kernel so use that for the return value.
Reviewed-by: Adrian Hunter <[email protected]>
Signed-off-by: Namhyung Kim <[email protected]>
Tested-by: James Clark <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Thomas Richter <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
The testcase "Object code reading" fails in somecases
for "fs_something" sub test as below:
Reading object code for memory address: 0xc008000007f0142c
File is: /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko
On file address is: 0x1114cc
Objdump command is: objdump -z -d --start-address=0x11142c --stop-address=0x1114ac /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko
objdump read too few bytes: 128
test child finished with -1
This can alo be reproduced when running perf record with
workload that exercises fs_something() code. In the test
setup, this is exercising xfs code since root is xfs.
# perf record ./a.out
# perf report -v |grep "xfs.ko"
0.76% a.out /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko 0xc008000007de5efc B [k] xlog_cil_commit
0.74% a.out /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko 0xc008000007d5ae18 B [k] xfs_btree_key_offset
0.74% a.out /lib/modules/6.5.0-rc3+/kernel/fs/xfs/xfs.ko 0xc008000007e11fd4 B [k] 0x0000000000112074
Here addr "0xc008000007e11fd4" is not resolved. since this is a
kernel module, its offset is from the DSO. Xfs module is loaded
at 0xc008000007d00000
# cat /proc/modules | grep xfs
xfs 2228224 3 - Live 0xc008000007d00000
And size is 0x220000. So its loaded between 0xc008000007d00000
and 0xc008000007f20000. From objdump, text section is:
text 0010f7bc 0000000000000000 0000000000000000 000000a0 2**4
Hence perf captured ip maps to 0x112074 which is:
( ip - start of module ) + a0
This offset 0x112074 falls out .text section which is up to 0x10f7bc
In this case for module, the address 0xc008000007e11fd4 is pointing
to stub instructions. This address range represents the module stubs
which is allocated on module load and hence is not part of DSO offset.
To address this issue in "object code reading", skip the sample if
address falls out of text section and is within the module end.
Use the "text_end" member of "struct dso" to do this check.
To address this issue in "perf report", exploring an option of
having stubs range as part of the /proc/kallsyms, so that perf
report can resolve addresses in stubs range
However this patch uses text_end to skip the stub range for
Object code reading testcase.
Reported-by: Disha Goel <[email protected]>
Signed-off-by: Athira Rajeev <[email protected]>
Tested-by: Disha Goel<[email protected]>
Reviewed-by: Adrian Hunter <[email protected]>
Reviewed-by: Kajol Jain <[email protected]>
Cc: [email protected]
Cc: [email protected]
Cc: [email protected]
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Namhyung Kim <[email protected]>
This test occasionally fails on aarch64 when a sample is taken in
free@plt and it fails with "Bytes read differ from those read by
objdump".
This is because that symbol is near a section boundary in the elf file.
Despite the -z option to always output zeros, objdump uses
bfd_map_over_sections() to iterate through the elf file so it doesn't
see outside of the sections where these zeros are and can't print them.
For example this boundary proceeds free@plt in libc with a gap of 48
bytes between .plt and .text:
objdump -d -z --start-address=0x23cc8 --stop-address=0x23d08 libc-2.30.so
libc-2.30.so: file format elf64-littleaarch64
Disassembly of section .plt:
0000000000023cc8 <*ABS*+0x7fd00@plt+0x8>:
23cc8: 91018210 add x16, x16, #0x60
23ccc: d61f0220 br x17
Disassembly of section .text:
0000000000023d00 <abort@@GLIBC_2.17-0x98>:
23d00: a9bf7bfd stp x29, x30, [sp, #-16]!
23d04: 910003fd mov x29, sp
Taking a sample in free@plt is very rare because it is so small, but the
test can be forced to fail almost every time on any platform by linking
the test with a shared library that has a single empty function and
calling it in a loop.
The fix is to zero the buffers so that when there is a jump in the
addresses output by objdump, zeros are already filled in between.
Signed-off-by: James Clark <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Namhyung Kim <[email protected]>
Link: http://lore.kernel.org/lkml/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
The evlist and the cpu/thread maps should be released together.
Otherwise following error was reported by Asan.
Note that this test still has memory leaks in DSOs so it still fails
even after this change. I'll take a look at that too.
# perf test -v 26
26: Object code reading :
--- start ---
test child forked, pid 154184
Looking at the vmlinux_path (8 entries long)
symsrc__init: build id mismatch for vmlinux.
symsrc__init: cannot get elf header.
Using /proc/kcore for kernel data
Using /proc/kallsyms for symbols
Parsing event 'cycles'
mmap size 528384B
...
=================================================================
==154184==ERROR: LeakSanitizer: detected memory leaks
Direct leak of 439 byte(s) in 1 object(s) allocated from:
#0 0x7fcb66e77037 in __interceptor_calloc ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:154
#1 0x55ad9b7e821e in dso__new_id util/dso.c:1256
#2 0x55ad9b8cfd4a in __machine__addnew_vdso util/vdso.c:132
#3 0x55ad9b8cfd4a in machine__findnew_vdso util/vdso.c:347
#4 0x55ad9b845b7e in map__new util/map.c:176
#5 0x55ad9b8415a2 in machine__process_mmap2_event util/machine.c:1787
#6 0x55ad9b8fab16 in perf_tool__process_synth_event util/synthetic-events.c:64
#7 0x55ad9b8fab16 in perf_event__synthesize_mmap_events util/synthetic-events.c:499
#8 0x55ad9b8fbfdf in __event__synthesize_thread util/synthetic-events.c:741
#9 0x55ad9b8ff3e3 in perf_event__synthesize_thread_map util/synthetic-events.c:833
#10 0x55ad9b738585 in do_test_code_reading tests/code-reading.c:608
#11 0x55ad9b73b25d in test__code_reading tests/code-reading.c:722
#12 0x55ad9b6f28fb in run_test tests/builtin-test.c:428
#13 0x55ad9b6f28fb in test_and_print tests/builtin-test.c:458
#14 0x55ad9b6f4a53 in __cmd_test tests/builtin-test.c:679
#15 0x55ad9b6f4a53 in cmd_test tests/builtin-test.c:825
#16 0x55ad9b760cc4 in run_builtin /home/namhyung/project/linux/tools/perf/perf.c:313
#17 0x55ad9b5eaa88 in handle_internal_command /home/namhyung/project/linux/tools/perf/perf.c:365
#18 0x55ad9b5eaa88 in run_argv /home/namhyung/project/linux/tools/perf/perf.c:409
#19 0x55ad9b5eaa88 in main /home/namhyung/project/linux/tools/perf/perf.c:539
#20 0x7fcb669acd09 in __libc_start_main ../csu/libc-start.c:308
...
SUMMARY: AddressSanitizer: 471 byte(s) leaked in 2 allocation(s).
test child finished with 1
---- end ----
Object code reading: FAILED!
Signed-off-by: Namhyung Kim <[email protected]>
Acked-by: Jiri Olsa <[email protected]>
Cc: Adrian Hunter <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Andi Kleen <[email protected]>
Cc: Ian Rogers <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Leo Yan <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Stephane Eranian <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Detect symbols generated by the OCaml compiler based on their prefix.
Demangle OCaml symbols, returning a newly allocated string (like the
existing Java demangling functionality).
Move a helper function (hex) from tests/code-reading.c to util/string.c
To test:
echo 'Printf.printf "%d\n" (Random.int 42)' > test.ml
perf record ocamlopt.opt test.ml
perf report -d ocamlopt.opt
Signed-off-by: Fabian Hemmer <[email protected]>
Acked-by: Namhyung Kim <[email protected]>
Cc: Alexander Shishkin <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Mark Rutland <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LPU-Reference: 20210203211537.b25ytjb6dq5jfbwx@nyu
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
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.
Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Song Liu <[email protected]>
Link: https://lkml.kernel.org/n/[email protected]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
Ian Rogers