Subject: Re: latest pthread changes win
To: Daniel Carosone <email@example.com>
From: Harry Waddell <firstname.lastname@example.org>
Date: 05/31/2003 11:19:53
Galeon seems a much happier as well, although it may just be that I'm
loading different pages. It used to die frequently with a pthread error
message and often leave an unkillable process in state closef. Yesterday it
ran for 8 hours until killed. I haven't experienced any other pthread
errors over the last couple days either. My limited experience is anecdotal
at best, but it seems to me that Nathan, et. al. are doing a good job of
shaking out the inevitable bugs.
Looking at what's in current now, the next major release (2.0?) should have
a lot of nice stuff added to it.
Caravan Electronic Publishing
On Sat, 31 May 2003 12:57:55 +1000
Daniel Carosone <email@example.com> wrote:
> Nathan recently made a couple of changes to pthread, particularly
> to do with recycling upcall stacks ASAP, and interrupted calls.
> I'm not sure which of these or related changes makes the difference,
> but with a fresh kernel and userland, xmms has been playing
> continuously for over 5 hours. The same test last week would have
> hung xmms within 10 mins (often less).
> Also interesting to note is that ps -s now shows xmms using 4 lwp's;
> previously it would have had 6.
> I worry I'm speaking too soon, but even this much is a vast
> improvement. I haven't been having many problems with other
> thread-using apps, but for those who have, I think it would be well
> worthwhile testing -current sources and reporting results.
> Thanks Nathan!