netatm: plan for removal unless an active maintainer is found

[Harti Brandt]

On Wed, 15 Mar 2006, Robert Watson wrote:

RW>
RW>For some time now, the FreeBSD Project has been carrying around at least
RW>three ATM stacks:
RW>
RW>- netatm - HARP, the Host ATM Research Platform
RW>
RW>- netnatm - A fairly minimal socket<->virtual circuit ATM wrapper for ATM
RW>  device drivers
RW>
RW>- ngatm - NetGraph ATM framework
RW>
RW>For a variety of reasons, it would be desirable to do a bit of pruning of ATM
RW>stacks.  In my previous conversations with Harti and Bruce, the conclusion
RW>has generally been that ngatm and netnatm provide almost all functionality
RW>provided by netatm and more, but in modern, and actively maintained,
RW>frameworks.
RW>
RW>The main motivator for pruning has to do with the SMP network stack work:
RW>we're reaching the point, discussed on a number of occasions previously on
RW>this mailing list, where jettisoning unmaintained network stack components
RW>that are unable to run MPSAFE, is highly desirable.  This would allow us to
RW>remove compatibility shims that are increasingly burdonsome in terms of
RW>performance and complexity.  Another aspect of this has to do with upcoming
RW>changes to the socket and pcb code, which will require maintainers of
RW>protocols to do a small but non-trivial amount of work to update protocols to
RW>fit the new socket behavior.  Right now, almost all other sections of the
RW>network stack have active maintainers who are able to do this work, both
RW>development and testing, except for the netatm code.
RW>
RW>We've reached the point where continuing to carry around a third ATM stack is
RW>a significant impediment to continued work on network stack performance and
RW>reliability work.  This should not be a surprise to anyone who reads this
RW>mailing list regularly, since I sent out e-mail on this very topic on July
RW>18, 2005, warning that netatm (and several other components) were at risk as
RW>unmaintained but substantial network stack components.
RW>
RW>The proposal is as follows:
RW>
RW>In order to begin to merge revised socket/pcb code, required to fix a number
RW>of current races manifesting in the TCP code under load, and required for
RW>breaking out the tcbinfo lock which is a significant bottleneck in high
RW>performance TCP and multi-processor TCP scalability, I will disconnect netatm
RW>and dependent components from the build on April 1, 2006.  At that point, I
RW>will merge updated socket and pcb reference counting.
RW>
RW>If a maintainer has not been found who can update and adequately test the
RW>netatm code for the new socket/pcb interfaces by June 30, 2006, the netatm
RW>code will be deleted from CVS HEAD (although remain available in Attic should
RW>someone turn up later).  I'm happy to provide some first cut patches for any
RW>maintainer that arrives that do implement the changes, but I am unable to
RW>test them, and suspect they will be significant deficient for this reason, so
RW>will not commit them myself.
RW>
RW>This will leave us with at least two quite functional ATM implementations.
RW>Please let me know if netatm is critical to your work and you will be able to
RW>work with me to get netatm into shape.  For someone already familiar with the
RW>netatm implementation and set up to perform ATM network testing, this should
RW>be straight forward and I'm happy to help in any way I can.  If you are
RW>available to act in this role, my recommendation would be that we meet at
RW>BSDCan in Ottawa this May to discuss long term maintenance directions, and so
RW>that we can work out a plan for handling SMP behavior for netatm, as well as
RW>its integration with the socket code.

I'm all for this removal. The only two reasons for carrying this around is 
that it supports CLIP over SVCs and that the driver for the IDT77211 chips 
is HARP-only. I have half of an ngatm driver for these chips and sent them 
out two at least two people during the last two years, but never heard 
back. I have also code for CLIP over SVCs but due to ENOTIME was not yet 
able to make this commit ready. If anybody knows somebody who could finish 
the driver it would be great. The CLIP stuff is not so critical - almost 
all people use PVCs only.

harti