Re: Current amd64 new error or warning from today's current with ruby r320323

From: Konstantin Belousov <kostikbel_at_gmail.com>
Date: Sun, 25 Jun 2017 05:04:41 +0300
On Sat, Jun 24, 2017 at 06:48:03PM -0700, Manfred Antar wrote:
> 
> > On Jun 24, 2017, at 6:23 PM, Konstantin Belousov <kostikbel_at_gmail.com> wrote:
> > 
> > On Sat, Jun 24, 2017 at 06:08:50PM -0700, Manfred Antar wrote:
> >> New world and kernel  r320323
> >> I get a new error or message when using ruby:
> >> 
> >> 
> >> /usr/local/sbin/portupgrade -av
> >> <main>: warning: pthread_create failed for timer: Resource temporarily unavailable, scheduling broken
> >> 
> >> everything works just this message when using ruby. I recompiled ruby , still same message
> >> 
> >> /usr/local/bin/ruby -v
> >> <main>: warning: pthread_create failed for timer: Resource temporarily unavailable, scheduling broken
> >> ruby 2.3.4p301 (2017-03-30 revision 58214) [amd64-freebsd12]
> >> 
> >> Not sure what???s changed, I noticed some commits to vm stuff, maybe thats it.
> > 
> > ktrace your failing ruby invocation, then post output of kdump -H somewhere.
> > 
> 
> Ok not sure  if this is right , but this is what i did:
> 
> (tmp)4637}ktrace /usr/local/bin/ruby -v
> <main>: warning: pthread_create failed for timer: Resource temporarily unavailable, scheduling broken
> ruby 2.3.4p301 (2017-03-30 revision 58214) [amd64-freebsd12]
> 
> (tmp)4638}kdump -H -f ./ktrace.out >  kdump.txt
> 
> you can get kdump.txt at:
> 
> http://www.pozo.com/kernel/kdump <http://www.pozo.com/kernel/kdump>.txt
> 
> It???s not failing, I don???t think , I can do portupgrade and it works fine.
> I just get this new message

I see what is going on, but it is somewhat strange that it happens.

Do you run ruby in a jail with old (say, stable/10) libthr ?
Or do you have environment variable LIBPTHREAD_SPLITSTACK_MAIN set in
your environment ?

Anyway, the rework of the stack grow indeed have incompatibility with the
old (pre-11) libthr, which tries to split main thread stack into smaller
stacks for the new threads.  New stack grow code was specifically designed
to prevent this.  Some hack would be needed there, to allow reuse of
the main stack gap.
Received on Sun Jun 25 2017 - 00:04:47 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:41:12 UTC