[Samba] More to pdbedit -L segfault

[Karel Kulhavý]

I have recompiled pdbedit 3.0.2 with -g flag to be able to list sources from gdb when
debugging pdbedit and discovered some more info but don't know how to interpret
it:

(gdb) frame 1
#1  0x0806c899 in init_sam_from_buffer (sampass=0x81b8c28, buf=0x81b8d78 "", 
    buflen=208) at passdb/passdb.c:1462
1462            SAFE_FREE(domain);
(gdb) print domain
$8 = 0x10000 <Address 0x10000 out of bounds>

It looks like domain is invalid address. It seems some integer has been sucked
in into the pointer. This architecture is i686-pc-linux-gnu.

Or maybe the whole memory is fatally corrupt and just generic strange things
happen. I suspect the databases got broken by reinstalling one version of
samba to another.

Question: can the databases be broken by reinstalling one 3.0.* samba with
3.0.2? Or can the database be broken by reinstalling one 3.0.* samba with
another 3.0.* samba (possible higher or lower version)?

I can try removing the databases and building them again.

However, I think, no matter how corrupt data files on the samba systems are,
the executables should not crash on segfault. Or is there a case where crashing
on segfault due to invalid input is a correct behaviour for an executable?

Cl<