[augeas-devel] Segfault in 0.7.0

Matthew Palmer matt at anchor.net.au
Thu Jan 28 01:00:38 UTC 2010


I'm getting the following backtrace after a segfault in 0.7.0:

#0  tree_replace (origin=<value optimized out>, path=<value optimized out>,
#sub=0x0) at augeas.c:783
#1  0x00007fc539b6a75d in transform_load (aug=0x20c7020, xfm=0x20cc3a0) at
#transform.c:435
#2  0x00007fc539b51c54 in aug_load (aug=0x20c7020) at augeas.c:458
#3  0x00007fc539b52123 in aug_init (root=0x4b52c40 "", loadpath=0x49aaa20
#"\002", flags=<value optimized out>) at augeas.c:421
#4  0x0000000000403351 in main (argc=1, argv=0x7fff41fb0c88) at
#augtool.c:1137

(I was also getting a segfault in 0.5.0, but I didn't have debugging symbols
available to get a decent backtrace, so I don't *know* it's the same
segfault).  The same (0.5.0) package and lenses are running OK on several hundred
other machines, so this is a real hum-dinger of an problem...

Running with either of --noload or --nostdinc allows startup to occur, so
it's almost certainly a wonky lens, but (a) segfaults are never cool, IMAO,
and (b) which lens?  On the "which lens" question, I managed to narrow it
down to the "aptpreferences" lens by moving everything out of the way and
adding them all back in, but a '-v' type option (prints "Loading
blah.aug..." sort of thing) would have been handy.

I'm going to continue working on this, but I thought it'd be worth dropping
an early warning here in case it looks obvious to anyone -- either "I got
that too, glad I'm not alone" or "There's a patch for that(TM)".

- Matt




More information about the augeas-devel mailing list