On Tuesday, Feb 10, 2004, at 12:09 Australia/Sydney, David Young wrote:
>
> See if there is any improvement with the patches I made to wi(4) today.

Seems much better! I've banged on the machine a bit doing the things
that usually set it off (ttcp over ethernet, cdda2wav from a SCSI CD,
and just driving the wi0 interface) and haven't had it go OACTIVE.
BTW: with wi_debug set, it only spat out the following while I was
giving it a work out:
Feb 10 21:34:14 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 08
Feb 10 21:42:48 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:44:02 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 07
Feb 10 21:44:12 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 09
Feb 10 21:45:12 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:45:23 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 08
Feb 10 21:45:55 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:48:22 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:54:54 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:58:32 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:58:32 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 07
Feb 10 21:58:42 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e
Feb 10 21:58:42 stix-pc /netbsd: wi0: wi_tx_intr null node, rssdesc 09
Feb 10 21:59:03 stix-pc /netbsd: wi0: wi_tx_intr bad idx 6e

I still get high packet loss, esp when using -paranoia with cdda2wav,
but no longer runs out of descriptors. Basically back to where it was
before I upgraded, except with 4 times the transmit bandwidth - a big
improvement!

Thanks,
--
Paul Ripke
Unix/OpenVMS/TSM/DBA
I love deadlines. I like the whooshing sound they make as they fly by.
-- Douglas Adams