glibc-2.36 failure

glibc-2.36 failure

[Paul Zimmermann]

       Hi,

is the following failure known for glibc 2.36 make check?

Summary of test results:
      1 FAIL
   5083 PASS
     78 UNSUPPORTED
     18 XFAIL
      4 XPASS

FAIL: stdlib/tst-arc4random-thread

zimmerma@biscotte:/localdisk/zimmerma/glibc-2.36/build$ cat stdlib/tst-arc4random-thread.out
info: outer_threads=8 inner_threads=4
info: arc4random: minimum of 655360 blob results expected
info: arc4random: 662011 blob results observed
info: arc4random_buf: minimum of 655360 blob results expected
info: arc4random_buf: 655360 blob results observed
info: arc4random_uniform: minimum of 655360 blob results expected
Timed out: killed the child process
Termination time: 2022-08-22T08:57:46.154936205
Last write to standard output: 2022-08-22T08:57:32.198256849

I used ../configure --prefix=/usr, on a AMD EPYC 7282 16-Core Processor,
with gcc version 12.1.0 (Debian 12.1.0-8).

Paul

Re: glibc-2.36 failure

[Adhemerval Zanella Netto]

On 22/08/22 06:29, Paul Zimmermann via Libc-alpha wrote:
>        Hi,
> 
> is the following failure known for glibc 2.36 make check?
> 
> Summary of test results:
>       1 FAIL
>    5083 PASS
>      78 UNSUPPORTED
>      18 XFAIL
>       4 XPASS
> 
> FAIL: stdlib/tst-arc4random-thread
> 
> zimmerma@biscotte:/localdisk/zimmerma/glibc-2.36/build$ cat stdlib/tst-arc4random-thread.out
> info: outer_threads=8 inner_threads=4
> info: arc4random: minimum of 655360 blob results expected
> info: arc4random: 662011 blob results observed
> info: arc4random_buf: minimum of 655360 blob results expected
> info: arc4random_buf: 655360 blob results observed
> info: arc4random_uniform: minimum of 655360 blob results expected
> Timed out: killed the child process
> Termination time: 2022-08-22T08:57:46.154936205
> Last write to standard output: 2022-08-22T08:57:32.198256849
> 
> I used ../configure --prefix=/usr, on a AMD EPYC 7282 16-Core Processor,
> with gcc version 12.1.0 (Debian 12.1.0-8).

I have seen the test timeout on some environments, I think we will need to
tune it down similar to what we have done for tst-arc4random-fork.

Re: glibc-2.36 failure

[Florian Weimer]

* Paul Zimmermann via Libc-alpha:

>        Hi,
>
> is the following failure known for glibc 2.36 make check?
>
> Summary of test results:
>       1 FAIL
>    5083 PASS
>      78 UNSUPPORTED
>      18 XFAIL
>       4 XPASS
>
> FAIL: stdlib/tst-arc4random-thread
>
> zimmerma@biscotte:/localdisk/zimmerma/glibc-2.36/build$ cat stdlib/tst-arc4random-thread.out
> info: outer_threads=8 inner_threads=4
> info: arc4random: minimum of 655360 blob results expected
> info: arc4random: 662011 blob results observed
> info: arc4random_buf: minimum of 655360 blob results expected
> info: arc4random_buf: 655360 blob results observed
> info: arc4random_uniform: minimum of 655360 blob results expected
> Timed out: killed the child process
> Termination time: 2022-08-22T08:57:46.154936205
> Last write to standard output: 2022-08-22T08:57:32.198256849
>
> I used ../configure --prefix=/usr, on a AMD EPYC 7282 16-Core Processor,
> with gcc version 12.1.0 (Debian 12.1.0-8).

What's your kernel version?  Earlier ones have a serialized getrandom
implementation, I think.  Only later versions have per-CPU state and
decent scaling across multiple cores.

Thanks,
Florian

Re: glibc-2.36 failure

[Paul Zimmermann]

       Dear Florian,

I have:

zimmerma@biscotte:~$ uname -a
Linux biscotte 5.10.0-10-amd64 #1 SMP Debian 5.10.84-1 (2021-12-08) x86_64 GNU/Linux

Best regards,
Paul

> From: Florian Weimer <fweimer@redhat.com>
> Cc: Paul Zimmermann <Paul.Zimmermann@inria.fr>
> Date: Mon, 22 Aug 2022 19:30:06 +0200
> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (gnu/linux)
> 
> * Paul Zimmermann via Libc-alpha:
> 
> >        Hi,
> >
> > is the following failure known for glibc 2.36 make check?
> >
> > Summary of test results:
> >       1 FAIL
> >    5083 PASS
> >      78 UNSUPPORTED
> >      18 XFAIL
> >       4 XPASS
> >
> > FAIL: stdlib/tst-arc4random-thread
> >
> > zimmerma@biscotte:/localdisk/zimmerma/glibc-2.36/build$ cat stdlib/tst-arc4random-thread.out
> > info: outer_threads=8 inner_threads=4
> > info: arc4random: minimum of 655360 blob results expected
> > info: arc4random: 662011 blob results observed
> > info: arc4random_buf: minimum of 655360 blob results expected
> > info: arc4random_buf: 655360 blob results observed
> > info: arc4random_uniform: minimum of 655360 blob results expected
> > Timed out: killed the child process
> > Termination time: 2022-08-22T08:57:46.154936205
> > Last write to standard output: 2022-08-22T08:57:32.198256849
> >
> > I used ../configure --prefix=/usr, on a AMD EPYC 7282 16-Core Processor,
> > with gcc version 12.1.0 (Debian 12.1.0-8).
> 
> What's your kernel version?  Earlier ones have a serialized getrandom
> implementation, I think.  Only later versions have per-CPU state and
> decent scaling across multiple cores.
> 
> Thanks,
> Florian

Re: glibc-2.36 failure

[Florian Weimer]

* Paul Zimmermann:

>        Dear Florian,
>
> I have:
>
> zimmerma@biscotte:~$ uname -a
> Linux biscotte 5.10.0-10-amd64 #1 SMP Debian 5.10.84-1 (2021-12-08) x86_64 GNU/Linux

Yeah, 5.10 probably predates the per-CPU kernel optimizations.

Thanks,
Florian