Re: On the subject of the VFS layer (was Re: VFS questions)

Alan Cox (
Sun, 4 May 1997 15:24:07 +0100 (BST)

> additionally to that will defintely lead to problems at some point.
> The only place where strategic decisions can be done are in the lowlevel
> driver. You can supply a set of generic routines for that for the current
> set of streamline controllers but anything else is a fatal mistake.

This is akin to the decision that was made for the networking. The upper
layers assume every driver is the ultimate supersmart control unit. If its
not the driver calls back into what is effectively library code to implement
the smarts.

For a slower device you are executing the same amount of code as before, but
something like a very high end smart scsi controller doesnt need to take
the hit of the generic scsi goings on, the disk queue scheduling and other
stuff it would rather do itself.

I am more convinced than ever that this is the right model for all species
of device driver