On Fri, 28 Jan 2000 00:42:47 -0800 
 Ethan Solomita <ethan@geocast.com> wrote:

 > 	I don't expect so, at least not with only one fastrm. The "fix" from
 > Kirk handles a single process very effectively but doesn't scale. And
 > yes, the scenario you mentioned for news servers would probably fail
 > spectacularly with soft updates. Either the code needs to be further
 > improved, or such users can't use soft updates.

As I said, Frank van der Linden and I have discussed ways to fix the
softupdates memory management issues, but it involves using kernel
threads for the softupdates code (rather than doing everything from
within an interrupt context) ... probably one thread per file system.

In any case, what needs to be done is fairly straightforward, but it
will be a lot of work.

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>