> On Wed, Oct 04, 2006 at 11:42:40AM +0900, YAMAMOTO Takashi wrote:
> > 
> > iirc, what you want to do is something about unreliable backing store, right?
> > i doubt if you really need to call fileassoc with spin locks held
> > because, in general, i/o from/to backing store can't be done with
> > spin locks held.
> > 
> 
> Correct.

then, let's try not to call it with spinlock held.
"nosleep" dances are evil in general.

i guess you are talking about getpages with PGO_LOCKED, right?
i think it's acceptable to add "never allocate new pages"
to the definition of PGO_LOCKED so that you don't need to check
finger prints in that case.  i believe it's already true for all in-tree
filesystems.  (please confirm.)

YAMAMOTO Takashi