Re: [PATCH v3 8/8] perf test: Update sample filtering test

From: Namhyung Kim
Date: Wed Jul 31 2024 - 20:23:56 EST

Next message: Sasha Levin: "[PATCH AUTOSEL 6.10 041/121] drm/amdgpu/pm: Check input value for CUSTOM profile mode setting on legacy SOCs"
Previous message: Sasha Levin: "[PATCH AUTOSEL 6.10 040/121] wifi: ath12k: initialize 'ret' in ath12k_dp_rxdma_ring_sel_config_wcn7850()"
In reply to: Arnaldo Carvalho de Melo: "Re: [PATCH v3 8/8] perf test: Update sample filtering test"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Wed, Jul 31, 2024 at 11:10:07AM -0300, Arnaldo Carvalho de Melo wrote:
> On Wed, Jul 03, 2024 at 03:30:35PM -0700, Namhyung Kim wrote:
> > Now it can run the BPF filtering test with normal user if the BPF
> > objects are pinned by 'sudo perf record --setup-filter pin'. Let's
> > update the test case to verify the behavior. It'll skip the test if the
> > filter check is failed from a normal user, but it shows a message how to
> > set up the filters.
> >
> > First, run the test as a normal user and it fails.
> >
> > $ perf test -vv filtering
> > 95: perf record sample filtering (by BPF) tests:
> > --- start ---
> > test child forked, pid 425677
> > Checking BPF-filter privilege
> > try 'sudo perf record --setup-filter pin' first. <<<--- here
> > bpf-filter test [Skipped permission]
> > ---- end(-2) ----
> > 95: perf record sample filtering (by BPF) tests : Skip
> >
> > According to the message, run the perf record command to pin the BPF
> > objects.
> >
> > $ sudo perf record --setup-filter pin
> >
> > And re-run the test as a normal user.
> >
> > $ perf test -vv filtering
> > 95: perf record sample filtering (by BPF) tests:
> > --- start ---
> > test child forked, pid 424486
> > Checking BPF-filter privilege
> > Basic bpf-filter test
> > Basic bpf-filter test [Success]
> > Failing bpf-filter test
> > Error: task-clock event does not have PERF_SAMPLE_CPU
> > Failing bpf-filter test [Success]
> > Group bpf-filter test
> > Error: task-clock event does not have PERF_SAMPLE_CPU
> > Error: task-clock event does not have PERF_SAMPLE_CODE_PAGE_SIZE
> > Group bpf-filter test [Success]
> > ---- end(0) ----
> > 95: perf record sample filtering (by BPF) tests : Ok
>
> Ok, so I tested one of the examples you provide as a root user:
>
> root@number:~# perf record -o- -e cycles:u --filter 'period < 10' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.024 MB - ]
> perf-exec 228020 53029.825757: 1 cpu_core/cycles/u: 7fe361d1cc11 [unknown] ([unknown])
> perf-exec 228020 53029.825760: 1 cpu_core/cycles/u: 7fe361d1cc11 [unknown] ([unknown])
> perf 228020 53029.826313: 1 cpu_atom/cycles/u: 7fd80d7ba040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228020 53029.826316: 1 cpu_atom/cycles/u: 7fd80d7ba040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228020 53029.838051: 1 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228020 53029.838054: 1 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228020 53029.838055: 9 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228020 53029.844137: 1 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228020 53029.844139: 1 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> root@number:~# perf record -o- -e cycles:u --filter 'period < 100000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.025 MB - ]
> perf-exec 228084 53076.760776: 1 cpu_core/cycles/u: 7f7e7691cc11 [unknown] ([unknown])
> perf-exec 228084 53076.760779: 1 cpu_core/cycles/u: 7f7e7691cc11 [unknown] ([unknown])
> perf-exec 228084 53076.760779: 10 cpu_core/cycles/u: 7f7e7691cc11 [unknown] ([unknown])
> perf-exec 228084 53076.760780: 497 cpu_core/cycles/u: 7f7e7691cc11 [unknown] ([unknown])
> perf-exec 228084 53076.760781: 27924 cpu_core/cycles/u: 7f7e7691cc11 [unknown] ([unknown])
> perf 228084 53076.761318: 1 cpu_atom/cycles/u: 7f317057d040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.761320: 1 cpu_atom/cycles/u: 7f317057d040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.761321: 14 cpu_atom/cycles/u: 7f317057d040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.761322: 518 cpu_atom/cycles/u: 7f317057d040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.761322: 20638 cpu_atom/cycles/u: 7f317057d040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.768070: 1 cpu_core/cycles/u: 7f317056e898 _dl_relocate_object+0x1d8 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.768072: 1 cpu_core/cycles/u: 7f317056e898 _dl_relocate_object+0x1d8 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.768073: 17 cpu_core/cycles/u: 7f317056e898 _dl_relocate_object+0x1d8 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.768073: 836 cpu_core/cycles/u: 7f317056e898 _dl_relocate_object+0x1d8 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.768074: 44346 cpu_core/cycles/u: 7f317056e89b _dl_relocate_object+0x1db (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228084 53076.843976: 1 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228084 53076.843978: 1 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228084 53076.843979: 13 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228084 53076.843979: 563 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228084 53076.843980: 26519 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> perf 228084 53077.482090: 1 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228084 53077.482092: 1 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228084 53077.482093: 15 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228084 53077.482093: 746 cpu_core/cycles/u: 53b062 noploop+0x62 (/home/acme/bin/perf)
> perf 228084 53077.482094: 38315 cpu_core/cycles/u: 53b05c noploop+0x5c (/home/acme/bin/perf)
> root@number:~#
>
> Filtering by period works as advertised, now I have done as root;
>
> root@number:~# perf record --setup-filter pin
> root@number:~# ls -la /sys/fs/bpf/perf_filter/
> total 0
> drwxr-xr-x. 2 root root 0 Jul 31 10:43 .
> drwxr-xr-t. 3 root root 0 Jul 31 10:43 ..
> -rw-rw-rw-. 1 root root 0 Jul 31 10:43 dropped
> -rw-rw-rw-. 1 root root 0 Jul 31 10:43 filters
> -rwxrwxrwx. 1 root root 0 Jul 31 10:43 perf_sample_filter
> -rw-rw-rw-. 1 root root 0 Jul 31 10:43 pid_hash
> -rw-------. 1 root root 0 Jul 31 10:43 sample_f_rodata
> root@number:~# ls -la /sys/fs/bpf/perf_filter/perf_sample_filter
> -rwxrwxrwx. 1 root root 0 Jul 31 10:43 /sys/fs/bpf/perf_filter/perf_sample_filter
> root@number:~#
>
> And as a normal user I try:
>
> acme@number:~$ perf record -o- -e cycles:u perf test -w noploop | perf script -i- | head
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.204 MB - ]
> perf 228218 53158.670585: 1 cpu_atom/cycles/u: 7f2fb1b6e040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.670590: 1 cpu_atom/cycles/u: 7f2fb1b6e040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.670592: 7 cpu_atom/cycles/u: 7f2fb1b6e040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.670593: 117 cpu_atom/cycles/u: 7f2fb1b6e040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.670595: 2152 cpu_atom/cycles/u: 7f2fb1b6e040 _start+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.670604: 38977 cpu_atom/cycles/u: ffffffff99201280 [unknown] ([unknown])
> perf 228218 53158.670650: 167064 cpu_atom/cycles/u: 7f2fb1b67d7c intel_check_word.constprop.0+0x16c (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.671472: 232830 cpu_atom/cycles/u: 7f2fb1b75d98 strcmp+0x78 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.672710: 191183 cpu_atom/cycles/u: 7f2fb1b59311 _dl_map_object_from_fd+0xea1 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 228218 53158.673461: 158125 cpu_atom/cycles/u: 7f2fb1b77148 strcmp+0x1428 (/usr/lib64/ld-linux-x86-64.so.2)
> acme@number:~$
>
> Ok, no filtering, bot samples, lets try to use filtering as with root:
>
> acme@number:~$ perf record -o- -e cycles:u --filter 'period < 10000000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.019 MB - ]
> acme@number:~$ perf record -o- -e cycles:u --filter 'period < 10000000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.019 MB - ]
> acme@number:~$ perf record -o- -e cycles:u --filter 'period < 10000000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.019 MB - ]
> acme@number:~$ perf record -o- -e cycles:u --filter 'period < 10000000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.019 MB - ]
> acme@number:~$

Hmm.. strange. The above command works well for me.

>
> acme@number:~$ perf record -v -e cycles:u --filter 'period < 10000000' perf test -w noploop
> Using CPUID GenuineIntel-6-B7-1
> DEBUGINFOD_URLS=
> nr_cblocks: 0
> affinity: SYS
> mmap flush: 1
> comp level: 0
> Problems creating module maps, continuing anyway...
> pid hash: 228434 -> 13
> pid hash: 228434 -> 14

This part is a little strange as it's using two entries. Hmm, are you
using a hybrid machine? Anyway I think it should work there too..

Also the number is too high.. I expect 1 or 2. Maybe it didn't release
all the entries. Let me think about the case.

Thanks,
Namhyung

> mmap size 528384B
> Control descriptor is not initialized
> Couldn't start the BPF side band thread:
> BPF programs starting from now on won't be annotatable
> [ perf record: Woken up 1 times to write data ]
> failed to write feature CPU_PMU_CAPS
> [ perf record: Captured and wrote 0.009 MB perf.data ]
> acme@number:~$
>
> I also tried with task-clock:
>
> acme@number:~$ perf record -o- -e task-clock -c 10000 perf test -w noploop | perf script -i- | head
> perf 229784 54146.473644: 10000 task-clock:u: 7faf38f1c622 get_common_indices.constprop.0+0xa2 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473654: 10000 task-clock:u: 7faf38f1d323 update_active.constprop.0+0x383 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473664: 10000 task-clock:u: 7faf38f1cd32 intel_check_word.constprop.0+0x122 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473674: 10000 task-clock:u: 7faf38f1cd7c intel_check_word.constprop.0+0x16c (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473684: 10000 task-clock:u: 7faf38f19de5 __tunable_get_val+0x75 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473704: 10000 task-clock:u: 7faf38f190d0 rtld_mutex_dummy+0x0 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473754: 10000 task-clock:u: 7faf38f1a80e _dl_cache_libcmp+0xe (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473864: 10000 task-clock:u: 7faf38f2adb9 strcmp+0x99 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.473954: 10000 task-clock:u: 7faf38f1aa02 search_cache+0x112 (/usr/lib64/ld-linux-x86-64.so.2)
> perf 229784 54146.474024: 10000 task-clock:u: 7faf38f0de38 _dl_map_object_from_fd+0x9c8 (/usr/lib64/ld-linux-x86-64.so.2)
> acme@number:~$
> acme@number:~$ perf record -o- -e task-clock -c 10000 --filter 'ip < 0xffffffff00000000' perf test -w noploop | perf script -i-
> [ perf record: Woken up 1 times to write data ]
> [ perf record: Captured and wrote 0.127 MB - ]
> acme@number:~$
>
> Ideas?
>
> I'm keeping it in my local tree so that I run it through the container
> build tests meanwhile we try to understand this, what am I missing?
>
> - Arnaldo
>
> ⬢[acme@toolbox perf-tools-next]$ uname -a
> Linux toolbox 6.9.10-200.fc40.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Jul 18 21:39:30 UTC 2024 x86_64 GNU/Linux
> ⬢[acme@toolbox perf-tools-next]$ perf -vv
> perf version 6.11.rc1.g77a71e434cf4
> dwarf: [ on ] # HAVE_DWARF_SUPPORT
> dwarf_getlocations: [ on ] # HAVE_DWARF_GETLOCATIONS_SUPPORT
> syscall_table: [ on ] # HAVE_SYSCALL_TABLE_SUPPORT
> libbfd: [ OFF ] # HAVE_LIBBFD_SUPPORT
> debuginfod: [ on ] # HAVE_DEBUGINFOD_SUPPORT
> libelf: [ on ] # HAVE_LIBELF_SUPPORT
> libnuma: [ on ] # HAVE_LIBNUMA_SUPPORT
> numa_num_possible_cpus: [ on ] # HAVE_LIBNUMA_SUPPORT
> libperl: [ on ] # HAVE_LIBPERL_SUPPORT
> libpython: [ on ] # HAVE_LIBPYTHON_SUPPORT
> libslang: [ on ] # HAVE_SLANG_SUPPORT
> libcrypto: [ on ] # HAVE_LIBCRYPTO_SUPPORT
> libunwind: [ on ] # HAVE_LIBUNWIND_SUPPORT
> libdw-dwarf-unwind: [ on ] # HAVE_DWARF_SUPPORT
> libcapstone: [ on ] # HAVE_LIBCAPSTONE_SUPPORT
> zlib: [ on ] # HAVE_ZLIB_SUPPORT
> lzma: [ on ] # HAVE_LZMA_SUPPORT
> get_cpuid: [ on ] # HAVE_AUXTRACE_SUPPORT
> bpf: [ on ] # HAVE_LIBBPF_SUPPORT
> aio: [ on ] # HAVE_AIO_SUPPORT
> zstd: [ on ] # HAVE_ZSTD_SUPPORT
> libpfm4: [ on ] # HAVE_LIBPFM
> libtraceevent: [ on ] # HAVE_LIBTRACEEVENT
> bpf_skeletons: [ on ] # HAVE_BPF_SKEL
> dwarf-unwind-support: [ on ] # HAVE_DWARF_UNWIND_SUPPORT
> libopencsd: [ on ] # HAVE_CSTRACE_SUPPORT
> ⬢[acme@toolbox perf-tools-next]$ git log --oneline -10
> 2a24133dc55000b3 (HEAD -> perf-tools-next) perf test: Update sample filtering test
> d6fed13469889202 perf record: Add --setup-filter option
> d8a2ec627150b7a4 perf record: Fix a potential error handling issue
> b0313e52f43035b5 perf bpf-filter: Support separate lost counts for each filter
> eb29dacbaf215fda perf bpf-filter: Support pin/unpin BPF object
> 086e7d06af7ce4eb perf bpf-filter: Split per-task filter use case
> d3453d1bb80cdbb2 perf bpf-filter: Pass 'target' to perf_bpf_filter__prepare()
> 736cd1c7a7105e1d perf bpf-filter: Make filters map a single entry hashmap
> 96ff640908b9808e perf jevents: Don't stop at the first matched pmu when searching a events table
> 379fe1f78ed5ceaf perf jevents: Use name for special find value (PMU_EVENTS__NOT_FOUND)
> ⬢[acme@toolbox perf-tools-next]$
>
> > Signed-off-by: Namhyung Kim <namhyung@xxxxxxxxxx>
> > ---
> > tools/perf/tests/shell/record_bpf_filter.sh | 13 +++++++------
> > 1 file changed, 7 insertions(+), 6 deletions(-)
> >
> > diff --git a/tools/perf/tests/shell/record_bpf_filter.sh b/tools/perf/tests/shell/record_bpf_filter.sh
> > index 31c593966e8c..c5882d620db7 100755
> > --- a/tools/perf/tests/shell/record_bpf_filter.sh
> > +++ b/tools/perf/tests/shell/record_bpf_filter.sh
> > @@ -22,15 +22,16 @@ trap trap_cleanup EXIT TERM INT
> > test_bpf_filter_priv() {
> > echo "Checking BPF-filter privilege"
> >
> > - if [ "$(id -u)" != 0 ]
> > - then
> > - echo "bpf-filter test [Skipped permission]"
> > - err=2
> > - return
> > - fi
> > if ! perf record -e task-clock --filter 'period > 1' \
> > -o /dev/null --quiet true 2>&1
> > then
> > + if [ "$(id -u)" != 0 ]
> > + then
> > + echo "try 'sudo perf record --setup-filter pin' first."
> > + echo "bpf-filter test [Skipped permission]"
> > + err=2
> > + return
> > + fi
> > echo "bpf-filter test [Skipped missing BPF support]"
> > err=2
> > return
> > --
> > 2.45.2.803.g4e1b14247a-goog
> >

Next message: Sasha Levin: "[PATCH AUTOSEL 6.10 041/121] drm/amdgpu/pm: Check input value for CUSTOM profile mode setting on legacy SOCs"
Previous message: Sasha Levin: "[PATCH AUTOSEL 6.10 040/121] wifi: ath12k: initialize 'ret' in ath12k_dp_rxdma_ring_sel_config_wcn7850()"
In reply to: Arnaldo Carvalho de Melo: "Re: [PATCH v3 8/8] perf test: Update sample filtering test"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]