Re: [PATCH rt-tests v2 1/3] rt-numa: Use sched_getaffinity() instead of pthread_getaffinity_np()

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Fri, Jul 09, 2021 at 01:58:51PM -0400, John Kacur wrote:
> It looks okay on the surface but I would like to know where you tested.

x86_64 and armv7. FWIW, pthread_getaffinity_np() is a fancy wrapper
around sched_getaffinity():

  https://github.com/bminor/glibc/blob/master/nptl/pthread_getaffinity.c#L34

> Also, in your message you explain that you can't do a static build
> but we never made static builds a requirment for rt-tests.

So you are saying static builds are not supported at all?

> I know from
> your cover letter than you are trying to do something with arm, if this
> is your motivation you should put it in your message. Even if this is your 
> motivation is there a reason we have to have a static build there?

Sure thing, I'll add more to the commit message. I just find it strange
that you are starting to argue static builds are not supported. This is
very handy especially since the linker dependency on libnuma. I can drop
the binary on the target without the need to install libnuma into the
rootfs. It is not available on all distro for armv7.

Another user of static builds is the LAVA test suite. It ships the
rt-tests programs as static builds so that the test can run without the
need to pre populate the rootfs with the correct binary:

  https://github.com/Linaro/test-definitions/tree/master/automated/linux/cyclictest/bin
  https://github.com/Linaro/test-definitions/blob/master/automated/linux/cyclictest/cyclictest.sh#L41

I am not really a big fan of this either and wont defend it. I am just
listening existing users.

> Finally, you put "fixes f240656b056b", it is useful to know where this
> was introduced in trying to figure out whether we can really replace a 
> pthread call with the sched call, but then phrase it that way, because
> as far as I know that commit is not a problem needing fixing, so put
> "pthread_getaffinity_np call introduced in f240656b056b"
> or something like that.

>From my POV it is fixing a bug. If you decide static builds are not
supported, please say so. This allows me to discuss this issue with the
LAVA folks.

Thanks,
Daniel




[Index of Archives]     [RT Stable]     [Kernel Newbies]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Samba]     [Video 4 Linux]     [Device Mapper]

  Powered by Linux