[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date: Fri, 25 Dec 2015 16:03:02 -0500
From: Rich Felker <dalias@...c.org>
To: Markus Wichmann <nullplan@....net>
Cc: musl@...ts.openwall.com
Subject: Re: Patch: Negative stack pointer references
On Fri, Dec 25, 2015 at 09:57:34PM +0100, Markus Wichmann wrote:
> Hi all,
>
> I found a few instances in musl where negative stack pointer offsets
> were used in the handwritten assembly. That is problematic, because if a
> signal arrives and is handled during the time that scratch space is in
> use (unlikely but possible), and sigaltstack() is not used, then that
> scratch space will be overwritten.
>
> This was just something I saw while randomly reading the code. Also, I
> only searched for the error pattern using a regex, so the problem may
> persist with (possibly) negative nonconstant offsets to the stack
> pointer, use of the same space with another register as base, or use of
> a negative constant offset my regex failed to match. I searched for
>
> -\d\+(%[er]sp)
> -0[xX]\x\+(%[er]sp)
>
> in all .s, .c, and .h files.
>
> Also, the problem may exist in architectures other than AMD64 or x32.
>
> I'm not a subscriber, so please CC me in this thread.
This is intentional and is safe. Google "x86_64 red zone".
Rich
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.
