On Wed, Sep 04, 2002 at 10:24:37PM +0200, Manuel Bouyer wrote:
> On Wed, Sep 04, 2002 at 01:19:15PM -0700, Jason R Thorpe wrote:
> > We already "sort of" have a way to do this ... the minphys entry point.
> > 
> > The problem is that only physio uses minphys.  The file system does not.
> 
> And the request size is cut after the transfer request has been set up.
> It would be better if the upper level had the info to set up a buf of the
> proper size.

Hum, and if we have a variable, per-device MAXPHYS, the per-device minphys
can probably die, isn't it ?
This would avoid several function pointer dereference in the I/O path,
to only end up calling the global minphys() for most of the drivers.

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
--