Re: [BProc] Re: pthreads+bproc& SIGSTOP -- possible fix

[<er...@he...>]

On Thu, Apr 10, 2003 at 01:23:02PM -0400, Nicholas Henke wrote:
> Hey Erik -- Ok it looks like there may be >=2 problems combining here to
> make life difficult. 
> 
> 1) Odd ass-ed SIGSTOP problem. I think this may have been solved. We
> talked to a few of the other guys here, and apparently the latest
> versions of glibc 2.2.4-{31,32} from redhat for 7.2 are SCREWED for
> pthreads. Something to do with each thread getting 64k for their stack
> when there is a total of 64K of total stack for their threads.  We
> reverted to glibc-2.2.4-30, and I have not seen the SIGSTOP problem
> again.

That's interesting.  Still, it shouldn't be possible to get a SIGSTOP
to the slave daemon no matter how screwed up a thing you do.

> 2) kill -9 being ignored. Dunno here -- I am going to try the patch
> suggested, and see what happens there. Hopefully the trace information
> can shed some light on this.

I believe the patch that J.A. Magallon suggested is aimed at problem
1.  I'm pretty sure #2 is purely a BProc problem.

- Erik