This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: [libc-alpha] linuxthreads bug in 2.2.4 under ppc linux
- From: Kevin B. Hendricks <kevin dot hendricks at sympatico dot ca>
- To: Franz Sirl <Franz dot Sirl-kernel at lauterbach dot com>,Kaz Kylheku <kaz at ashi dot footprints dot net>
- Cc: <libc-alpha at sources dot redhat dot com>
- Date: Sat, 8 Dec 2001 20:26:10 -0500
- Subject: Re: [libc-alpha] linuxthreads bug in 2.2.4 under ppc linux
- References: <Pine.LNX.4.33.0112072232510.2640-100000@ashi.FootPrints.net> <20011208200105.LMXK20714.tomts9-srv.bellnexxia.net@there> <200112090125.09758@enzo.bigblue.local>
Hi Franz,
I had no idea so much changed between -O2 and -O3. One significant
difference in that all of the wait_node_dequeue and wait_node_free are
actually inlined into the -O3 version and pretty much get integrated in
while they stay as separate routines in -O2.
This is going to be hard to compare.
Kevin
On December 8, 2001 07:46, Franz Sirl wrote:
> On Saturday 08 December 2001 21:04, Kevin B. Hendricks wrote:
> > Hi,
> >
> > As the fates always seem to determine.
> >
> > If I build glic-2.2.4 with -g -O1 -DDEBUG=1 it now won't segfault
> > anymore.
> >
> :-(. Though this would possibly point to some missing barrier in the
> : threads
>
> code if less scheduling fixes it. Well, try with -O2 -g, a bit harder to
> debug though (you can play with -fno-schedule-insns and
> -fno-schedule-insns-2 too). BTW, leaving out -DNDEBUG=1 already switchea
> on the assertions, whereas -DDEBUG=1 enables extra debugging stuff
> (maybe the extra debugging stuff lets the bug go away?).
>
> Franz.