On Mon, May 07, 2012 at 11:54:18PM +0200, Julien Cristau wrote: > On Mon, May 7, 2012 at 21:35:58 +0200, Laurent Fousse wrote: > > > Hello, > > > > * Julien Cristau [Mon, May 07, 2012 at 06:08:40PM +0200]: > > > it looks like the update from gmp 2:5.0.4+dfsg-1 to 2:5.0.5+dfsg-1 > > > causes gcc to ICE on the buildds. > > > See e.g. > > > https://buildd.debian.org/status/fetch.php?pkg=ufraw&arch=amd64&ver=0.18-1.1%2Bb2&stamp=1336402119 > > > > It would be interesting to see if the bug goes away when mpfr is > > rebuilt against the newest gmp package. > > > > I will look into it this evening, but feel free to beat me to it. > > > I haven't tried that, won't get to it today. However as a data point > the crash seemed to (mostly? only?) affect VM guests, rather than > physical hosts.
Matches my experience: I cannot reproduce the crash using reproducer.c
on my physical amd64.
> A local rebuild with --disable-fat seemed not to crash,
> but I haven't tried a clean rebuild without source change so I'm not
> sure if it's the rebuild or the configure option change that fixed
> things for me.
Interesting. It could well be the configure option. For the past 6
years, Debian built using --disable-fat on amd64. I changed it
because there are both bug workarounds and code speedups that are
processor-based. So it makes more sense for the code to detect the
processor at runtime.
On the other hand, I just looked at the source diff and there is a
change in configure.guess that changes the behaviour for a 64-bit
"vmkernel". Could this be affecting your virtual machine?
@@ -1315,6 +1325,9 @@
i*86:AROS:*:*)
echo ${UNAME_MACHINE}-pc-aros
exit ;;
+ x86_64:VMkernel:*:*)
+ echo ${UNAME_MACHINE}-unknown-esx
+ exit ;;
Dear gmp-bugs readers: any ideas?
Thanks,
-Steve
signature.asc
Description: Digital signature
