RFC: PPP MPPE merge into kernel 2.6
I'd like to open a discussion on the merge of the PPP MPPE support with kernel 2.6. What can be done to achieve it? What is hindering it? What can we do to remove the roadblocks?
MPPE is Microsoft's encryption scheme for PPP links.
The Linux MPPE support consists of five new source files, and patches to ppp.c, ppp_generic.c and ppp-comp.h, and can be found in the CVS repository for the PPP project in linux/mppe. The patches are relative to 2.2 and 2.4. See http://samba.org/ftp/unpacked/ppp/linux/mppe/
The author is Frank Cusack <[email protected]>, and the code is covered by the "BSD without advertisement clause" license. Frank released this around March 2002. See http://samba.org/ftp/unpacked/ppp/README.MPPE
MPPE support is used by PPTP. People who wish to use PPTP on their system often have to modify their kernel to add MPPE support. PPTP is not a particularly good tunnelling protocol, but many people are required to use it as a condition of employment. In particular, MPPE is not applied to non-data packets in the PPP stream, apparently by design.
Many distributions do not ship MPPE support (e.g. Red Hat). Some distributions have shipped an older version of MPPE support covered by the OpenSSL license, which is incompatible with the kernel license (e.g. SuSE, Mandrake). Some provide a patch (e.g. Debian).
The PPTP projects have adapted to the lack of MPPE support in the upstream kernel by providing custom kernels, version specific patches, module build scripts, and even binary modules.
Since the release of 2.6.0, two or three people have provided me with patches relative to 2.6.
The OpenSSL licensed MPPE support was a separate implementation, and was provided as a patch to 2.4.1. This was adopted by SuSE and Mandrake for their kernels, but is incompatible with the MPPE support in the newly released PPP 2.4.2.
It would be good to unify this situation and remove some of the pain.
Some effort has also been put into MPPC support, but as MPPC is covered by a patent I'm not proposing that.
I'm the PPTP client project leader.
--
James Cameron http://quozl.netrek.org/
HP Open Source, Volunteer http://opensource.hp.com/
PPTP Client Project, Release Engineer http://pptpclient.sourceforge.net/
Hi James,
Quoted 3 lines
I'd like to open a discussion on the merge of the PPP MPPE support with kernel 2.6. What can be done to achieve it? What is hindering it? What can we do to remove the roadblocks?
Sounds like an excellent idea, but it may be too late to have anything other than bug fixes merged into 2.6. 2.7/2.8 should be possible, though.
I think there is a strong argument for the support being in the kernel, given the number of other devices that have to support it, the obvious end-user demand (witness pptpclient and poptop :-) and the hassle for users of finding the right patch and patching their kernels (given the patch confusion which exists). Some users are having problems with patching due to failed hunks, and Joe Bloggs shouldn't have to worry about that.
Let's go for it :-) What do we have to lose?
Cheers, Chris.
Signature · 4 lines
_ __ __ _ / __/ / ,__(_)_ | Chris Wilson -- UNIX Firewall Lead Developer | / (_ ,\/ _/ /_ \ | NetServers.co.uk http://www.netservers.co.uk | \__/_/_/_//_/___/ | 21 Signet Court, Cambridge, UK. 01223 576516 |
Quoted 17 lines · Chris Wilson
Hi James,
Quoted 3 lines
I'd like to open a discussion on the merge of the PPP MPPE support with kernel 2.6. What can be done to achieve it? What is hindering it? What can we do to remove the roadblocks?
Sounds like an excellent idea, but it may be too late to have anything other than bug fixes merged into 2.6. 2.7/2.8 should be possible, though.
I think there is a strong argument for the support being in the kernel, given the number of other devices that have to support it, the obvious end-user demand (witness pptpclient and poptop :-) and the hassle for users of finding the right patch and patching their kernels (given the patch confusion which exists). Some users are having problems with patching due to failed hunks, and Joe Bloggs shouldn't have to worry about that.
Quoted 2 lines
Let's go for it :-) What do we have to lose?
Should the MPPC support (by Jan Dubiec) be included as well?
http://www.polbox.com/h/hs001/
It also contains some additions to the MPPE support.
-- Pasi Kärkkäinen
^
. .
Linux
/ - \
Choice.of.the
.Next.Generation.Quoted 2 lines · Pasi K?rkk?inen
Should the MPPC support (by Jan Dubiec) be included as well?
http://www.polbox.com/h/hs001/
I don't think so, due to the software patent issue. While my personal belief differs, while working for HP I am constrained to follow patents.
Quoted 1 line
It also contains some additions to the MPPE support.
Yes, 56-bit keys. I haven't found any demand for that. The options are also slightly different, though that would be easy to ignore.
Jan's work though looks robust and thorough. Having the work hosted like that, with current patches for all current kernels, is more than Frank is doing. ;-)
Signature · 3 lines
James Cameron http://quozl.netrek.org/ HP Open Source, Volunteer http://opensource.hp.com/ PPTP Client Project, Release Engineer http://pptpclient.sourceforge.net/