On Mon, 23 Nov 2009, Tim Retout wrote:
Hi Stuart,
I don't think there's anything complicated going on here. The 'priv'
pointer is not initialized to NULL in 'cb_tree_cursor_changed', so
contains random memory values.
Hi Tim,
This fits with the 32 vs 64 issue I wanted to look for. What I'm
wondering now, is should priv ever get a junk value? It looks like there
are other places that use priv->page with out checking it, so if we get
a bad value here, might it also show up elsewhere?
It looks like the value of priv comes from g_new0() in
prefs_win_create_navtree() where is gets stored with gtk_tree_store_set(). How
could a good value ...
.. never mind, I just convinced myself that it really is a simple case of
potentially using an uninitialized value. I just needed a break from work,
and more than 2 minutes at a time to sit down and look at it.
If you'd like to give 0.2pre1-10 a test drive, I'll take one more look over
it, and then upload it.
http://www.netsweng.com/~anderson/pornview_0.2pre1-10_amd64.deb
Thanks for taking the time to explain your analysis.
On i386 this happens to work a lot of the time, but produces garbage:
$3 = (const gchar * const) 0xbfffcde8 "\b\317\377\277\017o\215\267\270>\036\b"
On amd64 it doesn't:
(gdb) print priv->page->path
Cannot access memory at address 0xd06800278d1a25ff
Note that priv->page->path is meant to be a human-readable string like
"/General".
I can confirm that the other bug (#527714) happens on my i386 sid
system, and is also happening on the buildds for kfreebsd-i386.
Yeah, no contest on this one 8-).
Stuart
Stuart R. Anderson ander...@netsweng.com
Network & Software Engineering http://www.netsweng.com/
1024D/37A79149: 0791 D3B8 9A4C 2CDC A31F
BD03 0A62 E534 37A7 9149
--
To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
