Re: REGRESSION: Performance regressions from switchinganon_vma->lock to mutex

[Ingo Molnar]

* Andi Kleen <ak@xxxxxxxxxxxxxxx> wrote:

> > I tried to send uli a patch to just add caching. No go. I sent 
> > *another* patch to at least make glibc use a sane interface (and 
> > the cache if it needs to fall back on /proc/stat for some legacy 
> > reason). We'll see what happens.
> 
> FWIW a rerun with this modified LD_PRELOAD that does caching seems 
> to have the same performance as the version that does 
> sched_getaffinity.
> 
> So you're right. Caching indeed helps and my assumption that the 
> child would only do it once was incorrect.

You should have known that your assumption was wrong not just from a 
quick look at the strace output or a quick look at the glibc sources, 
but also because i pointed out the caching angle to you in the 
sysconf() discussion:

  http://lkml.org/lkml/2011/5/14/9

repeatedly:

  http://lkml.org/lkml/2011/5/17/149

and Denys Vlasenko pointed out the caching angle as well:

  http://lkml.org/lkml/2011/5/17/183

But you kept pushing for your new syscall for upstream integration, 
ignoring all contrary evidence and ignoring all contrary feedback, 
without even *once* checking where and how it would integrate into 
glibc ...

Thanks,

	Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/