Bill Pringlemeir wrote:
> On 16 Feb 2005, [EMAIL PROTECTED] wrote:
 
> >> #9  0x0816011d in unref_record (model=0x10435055, path=0x2015241,
> >> iter=0x45424543, data=0x5602c300) at search.c:135
> >> rc = (record_t *) 0x18
> >> dups = (GHashTable *) 0x58

> Sorry, the pointer address was on the stack.  First I did "frame 9",
> then I did "print &iter".  Then I did "x /32 &iter" and then I did
> 'printf "%s", &iter'.  The stack address is not a malloc/heap
> location.

In this case, the ASCII might just be trash of a previous function
call to GGEP related code but don't ask me how it got there.

> So, it wasn't what was pointed at that was foobar, it was the actual
> pointer.  Small probability that static CGEP data has smashed the
> stack.  Seems more likely the values aren't correct.

Yes, a smashed stack doesn't look likely at this place.
 
> I guess you will try a passive with GTK1?

I've tried to reproduce the crash with passive searches and GTK2 but
had no success at all. GTKG ran perfectly fine for 9 hrs in UP mode.
However, after opening 20 passive searches the GUI became so horribly
unresponsive that I shut it down forcefully.

In a short session I had a like-wise weird crash (hadn't any crashes
for quite a while now) where the passed-through pointer suddenly
changed from one frame to another. I'm not sure whether the output of
gdb wasn't just junk. So in both cases the symptoms indicate a smashed
stack but that seems rather unlikely here, IMO.

Thus, Purify would come in quite handy as I don't really have an idea
where to start looking at all.

-- 
Christian

Attachment: pgpyHusStv2t3.pgp
Description: PGP signature

Reply via email to