VKernel progress update - 11 Jan 2007

[Matthew Dillon]

    Hey.  Nobody noticed that all my previous progress updates said
    '2006' instead of '2007' !

    The virtual kernel is a lot more stable now.  My vkernel paniced 
    once while I was doing an installworld... it ran out of room in
    its kernel_map.  I haven't tracked down the reason for that.
    Otherwise though I have not gotten any crashes.  And the real kernel
    has never crashed.

    I have four medium-sized items left to do and because of that I am 
    not going to branch the release until this coming Tuesday (the Branching
    was originally slated for this weekend).  The release date is still
    slated for a little over two weeks from now.

    The four items left on my personal list are:

    * Fix the previously described pmap issue

    * Implement a fine-grained setitimer instead of the callout based itimer
      we have now.  This will fix the timing issues in the vkernel that
      are particular felt when you do things like 'sleep 1'.

    * And try to come up with an async I/O solution for VKE, VKD, and the
      console.

    * Make gdb work properly with the new register frame.

    There are also a few recent bug reports that haven't been acted upon.
    e.g. the NFS mount panic and a few others.  I promise to get to those
    as soon as I finish the VKernel work (but it would be great if others
    could look into them and maybe save me the time :-)).  Hence why I moved
    the branch date to Tuesday.

						-Matt