Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 08 Sep 2012 16:24:04 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: New formats: KRB5-18 and KRB5-23

On 09/08/2012 04:16 PM, Dhiru Kholia wrote:
> On Sat, Sep 8, 2012 at 6:17 PM, magnum <john.magnum@...hmail.com> wrote:
>> That looks correct to me. Is there something similar in krb5-23?
>>
>> I tried this on OSX and now got this:
>>
>> Benchmarking: KRB5 aes256-cts-hmac-sha1-96 [32/64]...
>> Program received signal EXC_BAD_ACCESS, Could not access memory.
>> Reason: 13 at address: 0x0000000000000000
>> 0x00000001000408ac in crypt_all (count=98745856) at krb5-18_fmt.c:198
>> 198                   crypt_out[index][i] = (key.contents[4 * i]) |
>> (gdb) bt
>> #0  0x00000001000408ac in crypt_all (count=98745856) at krb5-18_fmt.c:198
>> #1  0x0000000100099b46 in fmt_self_test (format=0x1002dff20) at formats.c:104
>> Previous frame inner to this frame (gdb could not unwind past this frame)
>>
>> This is a non-OMP build. Note the count=98745856... I'm puzzled this only happens on OSX.
>
> I built john (on OSX) with debugging on and generated a core dump.
> However, when I try to load the core in gdb the whole VM hangs.
>
> count=98745856, why is the count so high? This seems to be a bug!

Yeah, that has to be a buffer overrun, it should be 1 in non-OMP. I 
would think the bug exist on Linux too but for various reasons it does 
not show up. Valgrind under Linux does not complain over anything. And 
Valgrind does not yet support OSX 10.8.

BTW that 'i' variable declaration must be moved in krb5-18 too.

magnum


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.