oss-sec mailing list archives

Re: Cgit XSS "vulnerability" has no CVE?


From: Peter Bex <peter () more-magic net>
Date: Mon, 7 Mar 2016 19:46:11 +0100

On Mon, Mar 07, 2016 at 06:53:33PM +0100, Jason A. Donenfeld wrote:
On Sat, Mar 5, 2016 at 6:41 PM, Peter Bex <peter () more-magic net> wrote:
This allows for an XSS attack by anyone with write access: If you can
push to a git repository for which the "txt2html" converter is activate,
you can create a README or README.txt and insert arbitrary HTML.

The XSS situation in those release notes does not cover what you've
described here. You're conflating two separate things.

Thanks for clarifying this.  For some reason I expected this to be about
the same issue.

On Mon, Mar 07, 2016 at 06:52:04PM +0100, Jason A. Donenfeld wrote:
At the moment, none of those example filters are XSS-safe. I think
I'll likely rewrite them for the next version to use a framework for
that.

Good to hear that!

But there's never been any guarantee for those filters, and
they've never been provided as anything but potential example filters
for people to tweak and change.

Even so, I think a warning would have been appropriate.  The text issue
is a bit more surprising, as one might certainly expect that to be safe,
though a quick inspection of the code should be enough to know what's
going on.

Considering that it's been "fixed", I thought a CVE might be useful to
trigger distros to include the patch.  Without a CVE, distros like
Debian and RedHat will keep using the unpatched version, which is a
shame if such an easy fix is available.

Cheers,
Peter

Attachment: signature.asc
Description: Digital signature


Current thread: