Could you add one message that consistenly crashes Pan here as an
attachment ? http://bugzilla.gnome.org/createattachment.cgi?id=91536

pretty tricky to do that as as soon as I select it Pan crashes, gonna
try to save it with mozilla

Created attachment 10691 [details]
one of the articles that crashes Pan

Fejj: I'm reassigning to GMime tickets that look like GMime errors,
and CC'ing you, but leaving in Pan, tickets where I'm not sure.
Let me know if you want me to stop doing either of these things...

charles: no prob. I'll take a look at this tonight when I get home.

both gmime's test-parser and test-mbox work fine on this without a
crash. I presume the problem is probably memory corruption in some
unrelated section of code?

how can I import this into Pan?

Can I help by doing some test on the Alpha ?
From what I've seen on the ddd debuging session (I'm not an expert)
The hangig occurs at a message-part object creation, the parameters
becoming random string instead of a valid mime type. 

Bumped remaining bugs to 0.13.2.

jcnbulk@free.fr: does this problem persist in 0.13.1?

Yes I'm afraid so, I've just tried with the current CVS version

I'm following where it segfaults with ddd

I think the problem is in this function

GMimeObject *
g_mime_object_new_type (const char *type, const char *subtype)
{
        struct _type_bucket *bucket;
        GMimeObject *object;
        GType obj_type;
        
        g_return_val_if_fail (type != NULL, NULL);
        
        type_registry_init ();
        
        bucket = g_hash_table_lookup (type_hash, type);
        if (!bucket) {
                bucket = g_hash_table_lookup (type_hash, "*");
                obj_type = bucket ? bucket->object_type : 0;
        } else {
                obj_type = GPOINTER_TO_INT (g_hash_table_lookup
(bucket->subtype_hash, subtype));


HERE
GPointer is 64 bits on Alpha
Is the hint useful ?


                if (!obj_type)
                        obj_type = GPOINTER_TO_INT
(g_hash_table_lookup (bucket->subtype_hash, "*"));
        }
        
        if (!obj_type) {
                /* use the default mime object */
                bucket = g_hash_table_lookup (type_hash, "*");
                if (bucket)
                        obj_type = GPOINTER_TO_INT
(g_hash_table_lookup (bucket->subtype_hash, "*"));
                
                if (!obj_type)
                        return NULL;
        }
        
        object = g_object_new (obj_type, NULL, NULL);
        
        GMIME_OBJECT_GET_CLASS (object)->init (object);
        
        return object;
}
    
    

Bumped remaining bugs to 0.13.3.

fejj: did your 64bit cleanliness patches fix this?

I think so but I'm not sure

jcnbulk@free.fr: Could you give 0.13.2.90 a spin when it comes out
later this week and see if the problem is resolved?

jcnbulk@free.fr: ping

hi, sorry for the late answer
Last time I check (last week) it was broken but I can't say pam is
still broken coz it segfaulted in libglib (libglib is currently broken
in Alpha Debian sid. I gonna try to fis the glib pb and try the last
CVS pam

It's still the same : hangs in libglib
I've appended a ldd output


Program received signal SIGSEGV, Segmentation fault.

+ Trace 31448

ldd /usr/local/src/pan/pan/pan
        libgtk-x11-2.0.so.0 => /usr/lib/libgtk-x11-2.0.so.0
(0x000002000002a000)
        libgdk-x11-2.0.so.0 => /usr/lib/libgdk-x11-2.0.so.0
(0x0000020000358000)
        libatk-1.0.so.0 => /usr/lib/libatk-1.0.so.0 (0x00000200003f2000)
        libgdk_pixbuf-2.0.so.0 => /usr/lib/libgdk_pixbuf-2.0.so.0
(0x0000020000416000)
        libm.so.6.1 => /lib/libm.so.6.1 (0x0000020000442000)
        libpangoxft-1.0.so.0 => /usr/lib/libpangoxft-1.0.so.0
(0x00000200004cc000)
        libpangox-1.0.so.0 => /usr/lib/libpangox-1.0.so.0
(0x000002000050a000)
        libpango-1.0.so.0 => /usr/lib/libpango-1.0.so.0
(0x000002000052e000)
        libgobject-2.0.so.0 => /usr/lib/libgobject-2.0.so.0
(0x0000020000586000)
        libgmodule-2.0.so.0 => /usr/lib/libgmodule-2.0.so.0
(0x00000200005ec000)
        libdl.so.2.1 => /lib/libdl.so.2.1 (0x0000020000600000)
        libglib-2.0.so.0 => /usr/lib/libglib-2.0.so.0 (0x0000020000614000)
        libxml2.so.2 => /usr/lib/libxml2.so.2 (0x00000200006b4000)
        libz.so.1 => /usr/lib/libz.so.1 (0x00000200007b8000)
        libgthread-2.0.so.0 => /usr/lib/libgthread-2.0.so.0
(0x00000200007dc000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x00000200007f2000)
        libc.so.6.1 => /lib/libc.so.6.1 (0x000002000089a000)
        libXi.so.6 => /usr/X11R6/lib/libXi.so.6 (0x0000020000a14000)
        libXft.so.1 => /usr/X11R6/lib/libXft.so.1 (0x0000020000a30000)
        libXrender.so.1 => /usr/X11R6/lib/libXrender.so.1
(0x0000020000a76000)
        libXext.so.6 => /usr/X11R6/lib/libXext.so.6 (0x0000020000a8e000)
        libX11.so.6 => /usr/X11R6/lib/libX11.so.6 (0x0000020000ab4000)
        libfreetype.so.6 => /usr/lib/libfreetype.so.6 (0x0000020000bcc000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x0000020000000000)

Ok I fixed the pango/glib thing by upgrading. But I'm sorry to say
that there is the same (gmime) bug at the same place.
here is the backtrace



Program received signal SIGSEGV, Segmentation fault.

+ Trace 31449

as long as gpointer >= to the size of a gint, it shouldn't matter :-(

I don't really understand the problem here.

okay, I was wrong. GINT_TO_POINTER() and GPOINTER_TO_INT() on 64bit
archs cast to/from 32bit ints and that's why this doesn't work.

Fixed in gmime CVS (should be as easy as copying the new
gmime-object.c file over to pan)