Re: Customer Reports

[miller@SOL.CS.ROCHESTER.EDU (Brad Miller)]

    Date: Wed, 5 Dec 90 02:32:14 EST
    From: barmar@Think.COM (Barry Margolin)

       Date: Tue, 4 Dec 90 19:12 EST
       From: Brad Miller <miller@SOL.CS.ROCHESTER.EDU>

       Sorry, I meant to say that if you are running under the new flods and create
       a FEP file (e.g. :create fep file, or copying worlds), you will run into
       this problem. The work around is to boot under the old flods (it will handle
       it fine). Don't ask me why, but it does happen.

    I had a feeling this is what you meant, but I simply couldn't believe it.
    The FEP is completely out of the loop when Lisp accesses the disk.  Lisp
    operates the I/O board directly.

    I suppose it's possible that a bug (or new feature, e.g. some form of
    cache?) in the new flods is causing them to write to the disk's directory,
    and this is then confusing Lisp when it tries to update the directory,
    resulting in an inconsistent FEP file system.


I'm not claiming I believe it, only that's what Symbolics told me, and
indeed, it seemed to work that way on our 40. The bomb was a Bus Error
during Clear Machine.

The other intesting tidbit was that we could completely swap the board set
of the 40 that worked with the 40 that didn't, and the result was that the
40 that worked, still worked, and the 40 that didn't, still didn't. It
appears to be a incompatibility with some version of the disk drives...
----
 Brad Miller		U. Rochester Comp Sci Dept.
 miller@cs.rochester.edu {...allegra!rochester!miller}