Re: [PATCH/RFC 2/2] perf tools: Check callchain is corrupted

[Arnaldo Carvalho de Melo <acme@ghostprotocols.net>]

Em Thu, May 31, 2012 at 11:04:37PM +0900, Namhyung Kim escreveu:
> 2012-05-31 (목), 10:45 -0300, Arnaldo Carvalho de Melo:
> > Em Thu, May 31, 2012 at 02:43:27PM +0900, Namhyung Kim escreveu:
> > > We faced segmentation fault on perf top -G at very high sampling
> > > rate due to a corrupted callchain. While the root cause was not
> > > revealed (I failed to figure it out), this patch tries to protect
> > > us from the segfault on such cases.

> > Capping the callchain lenght may even be a feature, not a consistency
> > check, but I don't know how long, for consistency checking, we should
> > consider too much, as there are cases of loooong callchains.

> Maybe we should consider applying (a part of, at least) Arun's multiple
> stack patchset. Also note that Linus complained about the current stack
> depth (255) is too long.

>   https://lkml.org/lkml/2012/4/20/497

Ok, I applied it, 'perf top -G -F 100000' now is holding up, no
segfaults, yay!

Thanks,

- Arnaldo