[openib-general] Re: Plans for libibverbs 1.0, 1.1 and beyond

[Roland Dreier]

    Michael> One thing that might be important is to use the upcoming
    Michael> madvise(MADV_DONTFORK) on private CQ/QP work request ring
    Michael> buffers.  This would require allocating these from
    Michael> page-sized pools.

We should be able to do this in a compatible way, so I think this can
go into the 1.0 branch.  I'll add it to the README.

libmthca already uses posix_memalign to make sure that CQ and QP
buffers are page-aligned.

    >> I'm also thinking of moving my libibverbs and libmthca
    >> development trees to git (most likely hosted at kernel.org).
    >> This has the drawback of moving their development repositories
    >> out of the common openib.org svn tree.  However, it will make
    >> handling 1.0, 1.1 and feature development branches much easier.
    >> I'd like to hear opinions on this before I make a decision.

    Michael> Well, since you ask, I am pretty happy with how things
    Michael> work with svn.

    Michael> Do you expect sufficient development on branches that
    Michael> needs advanced merging that git provides? I would think
    Michael> we just need an occasional bugfix there, which should be
    Michael> easy to handle by plain patches.

It's really nice to be able to develop features on independent
branches.  And I really like things like git-bisect for debugging.

The more I use git the more I think it's a lot better for my workflow
on libibverbs.  But maybe the hassle for all the consumers of
libibverbs is enough that it's worth it to keep it in svn.  On the
other hand things should be stabilizing enough that nearly everyone
can work from tarballs or binary packages.

 - R.