[plug] Random cc segfaults?
Trevor Phillips
T.Phillips at murdoch.edu.au
Wed Feb 4 15:13:28 WST 2004
On Tuesday 03 February 2004 11:36, James Devenish wrote:
> One thought is to examine crash dumps and see the
> compiler is always failing in the same area. Not sure why the problem
> would be intermittent, though. By the way: are you sure it's actually
> segmentation faults that are the cause of failure? Other problems could
> include "ulimits" that restrict the time or memory that a process uses.
Trying again now. It's currently sitting on:
CC arch/i386/kernel/sys_i386.o
CPU usage: 98%.
It's been sitting there about, oh, 20 mins?
I kill it, and restart the build process - this time it compiles sys_i386.o in
under a second?
Next problem:
CC arch/i386/kernel/cpu/mtrr/generic.o
arch/i386/kernel/cpu/mtrr/generic.c: In function `post_set':
arch/i386/kernel/cpu/mtrr/generic.c:283: internal compiler error: Segmentation
fault
Restart... Compiles generic.o fine.
Next problem:
CC arch/i386/kernel/mpparse.o
arch/i386/kernel/mpparse.c: In function `MP_bus_info':
arch/i386/kernel/mpparse.c:224: internal compiler error: Segmentation fault
Restart... Compiles mpparse.o fine
etc...
Doing a clean & restarting from scratch does NOT give errors at the same
points.
gcc version is 3.3.2 (Debian Testing package)
Hmmm. This error was a little different:
CC mm/slab.o
mm/slab.c: In function `kmem_cache_free':
mm/slab.c:2177: internal compiler error: in verify_local_live_at_start, at
flow.c:575
Um, how do enable creation of core dump files? I've tried "limit coredumpsize
xxx" for various values of xxx (under tcsh).
--
. Trevor Phillips - http://jurai.murdoch.edu.au/ .
: Web Technical Administrator - T.Phillips at murdoch.edu.au :
| IT Services - Murdoch University |
>--------------------------------------------------------------------<
| On nights such as this, evil deeds are done. And good deeds, of /
| course. But mostly evil, on the whole. /
\ -- (Terry Pratchett, Wyrd Sisters) /
More information about the plug
mailing list