prism 2.5 timeout in wi_cmd 0x010b

[Kevin Oberman]

> Date: Sat, 10 Jan 2004 21:19:08 -0500
> From: Peter Radcliffe <pir at pir.net>
> Sender: owner-freebsd-mobile at freebsd.org
> 
> Kevin Oberman <oberman at es.net> probably said:
> > I'll report what I see. It's pretty easy to test...just copy a
> > multi-megabyte file from my laptop to my workstation over the
> > wireless.
> 
> I've been seeing and reporting that problem too, for a while now.

It worked for me! I just copied 850 MB of data at an average transfer
rate of 4.3 Mb/s with no problems. I never came close to this in the
past. I hope Warner can comment on it (and commit it if he does not see
a problem!)

I will at the driver to my default boot to see if it also fixes the
problem of noise on the 2.4G band causing the card to lock up even when
it's not in use. (And pretty much locking up the system at the same
time.) Since this was always very intermittent, I don't know if I will
be able to ever say that it's fixed, but I may be able to say that it's
not! Hopefully other who have seen this (and thee are a LOT of them)
will also give it a try. (That's why I am being bad and cross-posting to
current at .)

Thanks, Marco! This is a big help for a lot of people who have seen this
problem on both Prism and WaveLAN cards.

By the way, the patch for current is at a different location. Here is
the patch for CURRENT:

--- if_wi.c.orig        Sat Jan 10 16:26:25 2004QQ
+++ if_wi.c     Sat Jan 10 16:28:37 2004
@@ -2403,9 +2403,11 @@
                 * set in the event status register.
                 */
                s = CSR_READ_2(sc, WI_EVENT_STAT);
+               DELAY(1);
                if (s & WI_EV_CMD) {
                        /* Ack the event and read result code. */
                        s = CSR_READ_2(sc, WI_STATUS);
+                       DELAY(1);
                        CSR_WRITE_2(sc, WI_EVENT_ACK, WI_EV_CMD);
                        if (s & WI_STAT_CMD_RESULT) {
                                count--;
-- 
R. Kevin Oberman, Network Engineer
Energy Sciences Network (ESnet)
Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab)
E-mail: oberman at es.net			Phone: +1 510 486-8634