On Wed, 30 Jun 1999, Jason Thorpe wrote:

:  > The main applications I see are in new filesystems where userland daemons
:  > help the kernel out. Like for the data migration system I'm working on,
:  > where a daemon needs to load a file in from tape. By the time the fs gets
:  > all the calls, it has no clue of the path to the file. But with these
:  > calls, userland can be told how to find an daccess the file.
: 
: This sort of thing could be used for e.g. ARLA or CODA, too (where the file
: system itself is in userland).

Isn't that application the direct converse of the fh*() call set?

-- 
-- Todd Vierling (tv@pobox.com)