Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 17 Apr 2013 01:50:59 +0200
From: magnum <john.magnum@...hmail.com>
To: john-dev@...ts.openwall.com
Subject: Re: Segfaults probably caused by DEBUG code in memory.c

On 17 Apr, 2013, at 1:33 , "jfoug" <jfoug@....net> wrote:
> I just tested unstable, and MEM_ALLOC_SIZE==0 worked fine.

Yup, committed now.

> I would rather not have dyna use that define, but it certainly hurts
> nothing.

I applied your patch too but kept the memory stuff in memory.c only. This make it safer against someone reusing MEM_ALLOC_SIZE like you did, in the future.

Hopefully this is settled now. The root issue was this: without this, Valgrind may tell you that a buffer allocated in this or that line, in this or that file[1], was busted in some other line and file[2]. But [1] was totally misleading, because that was just a much *earlier* call to mem_alloc_tiny() where there was an actual malloc() allocation going on behind the scenes. Extremely confusing until I understood what was going on and added this stuff.

So any time you are tracking a buffer problem down, be sure to use -DDEBUG.

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.