Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Thu, 04 Aug 2011 01:50:33 +0200
From: magnum <>
Subject: Re: issues with 1.7.8-jumbo-5

On 2011-08-04 01:11, magnum wrote:
> I just reviewed Jim's changes to wordlist.c. I can't see that -stdin
> mode is different at all from earlier code (just some if clauses) so
> maybe that was not what introduced it. But I have narrowed it down to
> being introduced by *something* in the 0004 patch. Might be a problem
> with NT_fmt. I'll do some real debugging.

Program received signal SIGSEGV, Segmentation fault.
__strcpy_ssse3 () at ../sysdeps/x86_64/multiarch/strcpy.S:94
	in ../sysdeps/x86_64/multiarch/strcpy.S
(gdb) bt
#0  __strcpy_ssse3 () at ../sysdeps/x86_64/multiarch/strcpy.S:94
#1  0x00000000004a0b96 in do_wordlist_crack (db=0x7be1e0, name=0x0, 
rules=0) at wordlist.c:538
#2  0x000000000049372f in john_run () at john.c:452
#3  0x0000000000493b0b in main (argc=5, argv=0x7fffffffda68) at john.c:592

On a Jumbo-5 with wordlist.c reverted to Jumbo-2, the problem is gone. 
But I just can't see why. I think Jim happened to surface an old bug.

There are some weird #ifdefs in the code because in the MPI version, 
nCurLine is ditched in favor of line_number because they are the same 
thing. The enclosed patch simplifies the code (does the same for 
non-MPI) so it's easier to track, but it does not affect this problem (I 
hoped it would).

I have no more time right now.


View attachment "0001-get-rid-of-dupe-variable-in-wordlist.c.patch" of type "text/x-patch" (3013 bytes)

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.