Tim Waugh (firstname.lastname@example.org)
Fri, 19 Mar 1999 18:48:48 +0000 (GMT)
On Fri, 19 Mar 1999, Philip Blundell wrote:
> If you're dealing with 256k of data the overhead of set_fs() is
> negligible. I agree it's not ideal that you have to do this but the
> alternative would be to force any driver that *does* want to supply
> data from user space (eg, lp) to bounce it through a kernel buffer.
And this has to happen anyway. Phil, I thought we had this discussion?
The new port->ops functions that do block reads/writes can return
immediately, with the actual work being done later. A user-space buffer
isn't much good then.
> It might possibly be worth investigating the idea of providing "fast
> path" IEEE1284 functions for at least the common cases.
If you're talking about IEEE 1284 transfers without the fuss of
negotiation, the functions in port->ops do that. You can start talking in
ECP mode straight away.
-- To unsubscribe, send mail to: email@example.com --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Sun 28 Mar 1999 - 17:02:56 EST