Re: Exmh releases, and a large bunch of small changes in the exmh tree...

On Mon, 26 Jul 2004 12:09:44 PDT, Kevin Cosgrove <[email protected]>  said:

> Yes, security problems are the best (sadly) reason to roll a
> 2.7.1 version.  I'd like to see the easy to implement security
> fixes happen as fast as reasonable and release before the harder
> to implement fixes, just to give us a bit more safety while
> waiting for the remaining updates.

The fact of the matter is that the fixes are in general *easy* - the file/exec
cleanup snarfed up a bunch of them (although there's a second round of
less-critical auditing needed there still), and the real biggie was a 2-line fix.

The *hard* part is going through the code, looking at each 'exec', 'eval', or 'file'
or similar call, and asking "Is there *any* possible code path where the parameters
weren't totally under our control?". (Perl's 'taint' mode makes this a LOT easier).

I was just writing a message to Brent and discovered I missed a nasty 'exec' call.
I'll probably commit a fix for it in the morning - the general nature of the fix is
an obvious one-liner, but I need to double-check what the problematic variable is
used for....

Somebody needs to go through the lib/html_* files - html_cache.tcl in particular
doesn't give me warm-n-fuzzies, even before I look at the code..;)

Does anybody else have any "I've been meaning to..." patches to toss in, while
we're at it?  I'd hate to finish the code audit, Brent cuts 2.7.1, and then other
stuff shows up.. .;)

_______________________________________________
Exmh-workers mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/exmh-workers

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Exmh version 2.5 07/13/2001

iD8DBQFBBayYcC3lWbTT17ARAgxbAJ41JY5I85QKJzu1s39G+w0h2HPqMQCcCEzE
yOGAGcErgod7ZF6HKfSuEiQ=
=VVsU
-----END PGP SIGNATURE-----