[irrtoolset] Segmentation fault on peval
spz at serpens.de
Wed Nov 2 16:18:34 UTC 2005
Thus wrote Jan Boogman (boogman at ip-plus.net):
> Since a few weeks I am experiencing a similar problem with peval 4.8.2 on linux (gentoo 2.6.11 and 2.4.26):
> jan at gatekeeper ~ $ peval as-swcmlocal
> *** glibc detected *** malloc(): memory corruption: 0x0812eee8 ***
> for really small as-sets it does work though, as does recursion of as-sets without prefix expansion (i.e peval -no-as <as-set>)
> Has anybody been able to fix that??
If it doesn't spit the data out first, it may be a different issue
For the old "peval segfaults when just about done while trying
to clean up its dictionary", this does not happen on sparc64 or m68k, nor
with gcc 3.4.2 on i386 arch, but does happen on i386 with gcc 3.3, so I
strongly suspect a compiler issue instead of the actual code. If you
should have a different gcc handy, could you please verify that?
IRRToolSet 4.8.2 won't compile with gcc 3.4 'as is', but e.g. the patches
from NetBSD pkgsrc net/irrtoolset-nox11 will help you with that.
"peval as-swcmlocal" doesn't segfault on NetBSD/sparc64 -current with
gcc 3.3 nor FreeBSD 5.4-RELEASE on i386 with gcc 3.4 .. not testing with
the m68k right now though. :)
spz at serpens.de (S.P.Zeidler)
More information about the irrtoolset