Thomas Sailer (sailer@ife.ee.ethz.ch)
Mon, 26 Jul 1999 20:07:11 +0200
Tim Waugh wrote:
> How does this look for supporting ECP/EPP address transfers with ppdev?
The interface looks fine that way.
Today, I ported my app to ppdev. The app internally has a parport
abstraction
interface similar to what is in the kernel, the backend (or low level
driver)
may either directly use inb/outb, or use ppdev or ppuser.
Now, my tests with 2.3.11 (with the ppdev patch from pre-2.3.12-5
applied) gave
the following results:
- hardware EPP: works
- software emul EPP: the program hanged forever or until a PTRACE
attach,
in which case it immediately returned (having no bytes transmitted).
- hardware ECP: ecp_write_addr works, ecp_read_data returned
immediately, having
no bytes read
- software ECP: same as hw ECP
will investigate further some time.
Under 2.2.10 with the ppdev1 patch, I somehow managed (using
strace and ctrl-c) to make ppdev or the kernel forget to unregister port
and to decrement ppdev use count, but don't ask me how I did that...
Another issue: the parport probing routines play games with
parport_pc_ops,
replacing method pointers there, and then a reference to parport_pc_ops
is stored in parport->ops. Doesn't that mean that all PC parports share
the same method table? That does only work if all ports have the
same hw capabilities...
Tom
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2.0b3 on Mon 26 Jul 1999 - 14:13:23 EDT