[kaffe] backtrace of KJC on darwin during class library compile

[Dalibor Topic]

Ciao Riccardo,

Riccardo wrote:
> #0  0x900739c0 in bzero ()
> #1  0x00284f04 in gc_heap_malloc (sz=32776) at ../../../kaffe/kaffe/kaffevm/mem/gc-mem.c:388
> #2  0x00257fb4 in gcMalloc (gcif=0x87a008, size=32776, fidx=GC_ALLOC_FIXED) at ../../../kaffe/kaffe/kaffevm/mem/gc-incremental.c:1027

Thanks for the backtrace.

Looks like something breaks in the gc. There was a very hard to track 
down bug in the gc a while ago, and it ended up being in the libtool-gc 
interaction. I'm inclined to blame it on the libtool lt_malloc 
interface, which has been removed in libtool CVS HEAD, so I'll stop 
kaffe from telling libltdl to use our gc allocation functions. Oh, and 
I'll upgrade kaffe to libtool 1.5.6.

On a side note, there is a darwinport of Kaffe 1.1.3 here: 
http://www.opendarwin.org/pipermail/cvs-darwinports-all/2004-February/023763.html 
I'd be interested to see if the CFLAGS used make any difference.

cheers,
dalibor topic