[Bug debuginfod/29975] New: Search concurrency doesn't respect CPU affinity

[Bug debuginfod/29975] New: Search concurrency doesn't respect CPU affinity

[ross at burtonini dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29975

            Bug ID: 29975
           Summary: Search concurrency doesn't respect CPU affinity
           Product: elfutils
           Version: unspecified
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: debuginfod
          Assignee: unassigned at sourceware dot org
          Reporter: ross at burtonini dot com
                CC: elfutils-devel at sourceware dot org
  Target Milestone: ---

I've a machine with lots of cores (256) but debuginfod doesn't respect CPU
affinity:

$ taskset -c 2 debuginfod -v
...
[Mon Jan  9 17:33:20 2023] (2351762/2351762): search concurrency 256

Instead of looking at the number of cores present, it should look at the number
of cores available to it (sched_getaffinity()).

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug debuginfod/29975] Search concurrency doesn't respect CPU affinity

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29975

Frank Ch. Eigler <fche at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |fche at redhat dot com

--- Comment #1 from Frank Ch. Eigler <fche at redhat dot com> ---
> number of cores available to it (sched_getaffinity()).

Do you happen to be aware of a c++y front-end to this, which we could
use instead of std::thread::hardware_concurrency() ?

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug debuginfod/29975] Search concurrency doesn't respect CPU affinity

[ross at burtonini dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29975

--- Comment #2 from Ross Burton <ross at burtonini dot com> ---
No, but the C isn't that difficult:

#define _GNU_SOURCE
#include <sched.h>

int ret, count;
cpu_set_t mask;

CPU_ZERO(&mask);
ret = sched_getaffinity(0, sizeof(mask), &mask);
// if non-zero, errno is set
count = CPU_COUNT(&mask);

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug debuginfod/29975] Search concurrency doesn't respect CPU affinity

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29975

--- Comment #3 from Frank Ch. Eigler <fche at redhat dot com> ---
please check out commit 7399e3bd7eb72d045 on elfutils.git for a test patch

-- 
You are receiving this mail because:
You are on the CC list for the bug.

[Bug debuginfod/29975] Search concurrency doesn't respect CPU affinity

[fche at redhat dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=29975

Frank Ch. Eigler <fche at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |FIXED

--- Comment #4 from Frank Ch. Eigler <fche at redhat dot com> ---

Pushed to master as dcb40f9caa7ca30

Author: Frank Ch. Eigler <fche@redhat.com>
Date:   Tue Jan 10 17:59:35 2023 -0500

    debuginfod PR29975 & PR29976: decrease default concurrency

    ... based on rlimit (rlimig -n NUM)
    ... based on cpu-affinity (taskset -c A,B,C,D ...)

    Signed-off-by: Frank Ch. Eigler <fche@redhat.com>

-- 
You are receiving this mail because:
You are on the CC list for the bug.