Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sun, 5 Feb 2023 13:13:50 +0100
From: Helmut Grohne <helmut@...divi.de>
To: oss-security@...ts.openwall.com
Subject: Re: sox: patches for old vulnerabilities

Hi,

On Sat, Feb 04, 2023 at 12:19:14AM +0100, Steffen Nurpmeso wrote:
> But i was only wondering a bit, have you checked against the
> [master] branch?  For example

I did a (too) rough survey of the upstream repository and (too quickly)
concluded that it wouldn't help me with fixing these in Debian, so I
worked from Debian's fork. I should have made this more clear.

>   02-fix-resource-leak-hcom.patch

Still needed in git.

>   03-fix-regression-in-CVE-2017-11358.patch

I'll be replacing the Debian-specific, broken fix of CVE-2017-11358 with
the one committed upstream. Thanks.

>   04-fix-hcom-big-endian.patch#

Indeed, I should have revisited the upstream tree. Upstream also fixes a
double free and I'll be replacing my patch with the upstream one.

>   06-CVE-2021-33844.patch

The code is refactored, but I think the issue persists in wav_read_fmt
where wav->bitsPerSample isn't checked.

> and
>   07-CVE-2021-3643.patch

The hunk context changed and channels are now verified, but the size
validation is still missing. During further analysis I also found that
my patch is insufficient still.

If uc becomes 1, we assign it to v->size, later we pass 6 - v->size as
the second parameter to lsx_adpcm_init, which is used as an index into a
static array of 5 elements. We thus have an out-of-bounds read access
here. I don't yet know where exactly the check belongs as v->size == 1
may be valid in some contexts still.

Updated patch attached.

> The rest just apply fine, and 02- was needed here, 03- seemed an
> unrolled dup, 04- in parts (stdint via sox.h, but overflow, sure),
> it is too late to check the rest, 'will do tomorrow.

Thank you.

> (I an maintaining an official contrib now private sox port for
> CRUX Linux based upon 42b3557e13e0fe0 as of 20211029.)

I think it would be good to have a maintained upstream repository of sox
eventually. It seems like multiple distributions are maintaining
diverging patch piles now.

Helmut

View attachment "CVE-2021-3643.patch" of type "text/x-diff" (672 bytes)

Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

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