[Bug linuxthreads/25774] Suggestion: count number of retries in __lll_lock_wait

["carlos at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=25774

Carlos O'Donell <carlos at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
                 CC|                            |carlos at redhat dot com
   Last reconfirmed|                            |2020-04-04
     Ever confirmed|0                           |1

--- Comment #1 from Carlos O'Donell <carlos at redhat dot com> ---
(In reply to François Rigault from comment #0)
> ```
> --- glibc-2.30/nptl/lowlevellock.c      2019-08-01 06:29:31.000000000 +0200
> +++ ../glibc/glibc-2.30/nptl/lowlevellock.c     2020-04-04
> 11:24:50.367896172 +0200
> @@ -42,13 +42,16 @@
>  void
>  __lll_lock_wait (int *futex, int private)
>  {
> +  int count = 0;
> +
>    if (atomic_load_relaxed (futex) == 2)
>      goto futex;
> 
>    while (atomic_exchange_acquire (futex, 2) != 0)
>      {
> +      ++count;
>      futex:
> -      LIBC_PROBE (lll_lock_wait, 1, futex);
> +      LIBC_PROBE (lll_lock_wait, 2, futex, count);
>        lll_futex_wait (futex, 2, private); /* Wait if *futex == 2.  */
>      }
>  }
> ```

It adds a small amount of code in the slow path before the wait, which seems
fine to me, we're about to enter the kernel and sleep.

The value is local, and should be 'unsigned' to allow wrapping.

I think your suggestion is a good one.

Would you like to put a patch together for this?

I'd expect:
- The change to the probe.
- Add the probe to manual/probes.texi under a new node "POSIX Thread Probes"
and document the behaviour (previously undocumented) of just this probe.

Have a look at our contribution checklist:
https://sourceware.org/glibc/wiki/Contribution%20checklist

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