oss-sec mailing list archives

Re: Pointer misuse unziping files with busybox


From: cve-assign () mitre org
Date: Tue, 3 Nov 2015 16:00:46 -0500 (EST)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

http://git.busybox.net/busybox/commit/?id=1de25a6e87e0e627aa34298105a3d17c60a1f44e

Unziping a specially crafted zip file results in a computation of an invalid
pointer and a crash reading an invalid address.

BusyBox wouldn't realistically be
used for deployment of a program that remains running to offer an
unzipping service to multiple clients.

There are several distributions including Alpine Linux, widely used in
container environments, which by default use busybox to provide the
unzip utility. Unzipping of any files downloaded by the user, possibly
from untrusted sources, may be affected. I believe CVE is appropriate
for user-facing programs commonly used to open untrusted files even
without an automated process accepting and processing
potentially-malicious files from a client.

We'll try to add some information about what we're looking for.

1. If the product were a library that decompresses untrusted files,
then the existence of a crash would be enough to assign a CVE ID. The
rationale is that a library might have been used to develop a program
that needs to remain running even after one bad file is encountered.

2. Many products that aren't libraries have no need to remain running
after a bad file is encountered. If the only possible problem is "a
crash reading an invalid address" and there is no way to write to an
invalid address or change the flow of control, then there typically
can't be a CVE ID. Typically, a simple and complete workaround for the
crash problem is to not try to unzip the bad file again.

From: Gustavo Grieco <gustavo.grieco () gmail com>
Date: Fri, 30 Oct 2015 09:38:47 -0300

Could you please comment directly about the likelihood of
exploitability for code execution?

To be honest, i don't know. The patched code looks quite complex and i
cannot discard any potential arbitrary write there.

We currently prefer not to assign CVE IDs when the available
information is "a crash reading an invalid address" in combination
with "cannot discard any potential arbitrary write."

- -- 
CVE assignment team, MITRE CVE Numbering Authority
M/S M300
202 Burlington Road, Bedford, MA 01730 USA
[ PGP key available through http://cve.mitre.org/cve/request_id.html ]
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
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=xHYl
-----END PGP SIGNATURE-----


Current thread: