linux-kernel - perf top -p broken for multithreaded processes since 5.19

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 02 Sep 2022 16:46:45 +0200
From:   Tomáš Trnka <trnka@....com>
To:     Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>,
        Arnaldo Carvalho de Melo <acme@...nel.org>
Cc:     linux-perf-users@...r.kernel.org, linux-kernel@...r.kernel.org,
        Adrian Hunter <adrian.hunter@...el.com>
Subject: perf top -p broken for multithreaded processes since 5.19

Hello,

A bug in perf v5.19 and newer completely breaks monitoring multithreaded
processes using "perf top -p". The tool fails to start with "Failed to mmap
with 22 (Invalid argument)". It still seems to work fine on single-threaded
processes. "perf record" is also unaffected.

I have bisected the issue to the following commit:

commit ae4f8ae16a07896403c90305d4b9be27f657c1fc
Author: Adrian Hunter <adrian.hunter@...el.com>
Date:   Tue May 24 10:54:31 2022 +0300

    libperf evlist: Allow mixing per-thread and per-cpu mmaps
    
    mmap_per_evsel() will skip events that do not match the CPU, so all CPUs
    can be iterated in any case.

The issue can be easily reproduced using the following test:

$ python - <<EOF
import time
import threading
th = threading.Thread(target=time.sleep, args=(3600,))
th.start()
th.join()
EOF

stracing "perf top -p $(pgrep python)" yields this:

145184 perf_event_open({type=PERF_TYPE_HARDWARE, size=PERF_ATTR_SIZE_VER7, config=PERF_COUNT_HW_CPU_CYCLES, sample_freq=4000, sample_type=PERF_SAMPLE_IP|PERF_SAMPLE_TID|PERF_SAMPLE_TIME|PERF_SAMPLE_PERIOD, read_format=PERF_FORMAT_ID, disabled=1, exclude_kernel=1, mmap=1, comm=1, freq=1, task=1, precise_ip=0 /* arbitrary skid */, sample_id_all=1, mmap2=1, comm_exec=1, ksymbol=1, ...}, 92061, -1, -1, PERF_FLAG_FD_CLOEXEC) = 3
 > tools/perf/perf(evsel__open_cpu+0x287) [0x4c8ad7]
 > tools/perf/perf(cmd_top+0x1996) [0x439b26]
 > tools/perf/perf(run_builtin+0x68) [0x4a91f8]
 > tools/perf/perf(main+0x645) [0x40cad5]
145184 perf_event_open({type=PERF_TYPE_HARDWARE, size=PERF_ATTR_SIZE_VER7, config=PERF_COUNT_HW_CPU_CYCLES, sample_freq=4000, sample_type=PERF_SAMPLE_IP|PERF_SAMPLE_TID|PERF_SAMPLE_TIME|PERF_SAMPLE_PERIOD, read_format=PERF_FORMAT_ID, disabled=1, exclude_kernel=1, mmap=1, comm=1, freq=1, task=1, precise_ip=0 /* arbitrary skid */, sample_id_all=1, mmap2=1, comm_exec=1, ksymbol=1, ...}, 104619, -1, -1, PERF_FLAG_FD_CLOEXEC) = 5
 > tools/perf/perf(evsel__open_cpu+0x287) [0x4c8ad7]
 > tools/perf/perf(cmd_top+0x1996) [0x439b26]
 > tools/perf/perf(run_builtin+0x68) [0x4a91f8]
 > tools/perf/perf(main+0x645) [0x40cad5]
…(snip)…
145184 ioctl(5, PERF_EVENT_IOC_SET_OUTPUT, 3) = -1 EINVAL (Invalid argument)
 > tools/perf/perf(perf_evlist__mmap_ops+0x2cf) [0x5d497f]
 > tools/perf/perf(evlist__mmap+0xa7) [0x4c09b7]
 > perf/perf(cmd_top+0x1ccd) [0x439e5d]
 > tools/perf/perf(run_builtin+0x68) [0x4a91f8]
 > tools/perf/perf(main+0x645) [0x40cad5]

Best regards,

Tomáš
--
Tomáš Trnka
Software for Chemistry & Materials B.V.