Mixed success with new MA301 (orinoco_plx) code

[Martijn van Oosterhout]

Summary for all those people who don't want to read it all:

The MA301 (PLX-based card) seems to work now although it appears that the
testing version of David's driver (Nov 12) may have broken something on one
of my two cards.

-------------------------------------------------------------------------
Well, I finally got around to testing out the patch posted a few days ago
and while it's an improvement, it's not quite there yet.

The code I used was pulled from David's ozlabs page (which seems to be gone
right now) out the of testing directory on Nov 12. The patch didn't apply
cleanly to the 0.08a version available as a tarball.

First thing, interrupts now seem to work (yay!).

> cat /proc/interrupts | grep eth2
 11:        416          XT-PIC  eth2

That's about one for each transmitted packet. I'm running a ping from both
ends.

However, receiving packets still seems to be an issue. Here is a tcpdump
from the end with the PLX card (192.168.15.148) with an Orinoco Silver:

# tcpdump -i eth2 -n
tcpdump: listening on eth2
Nov 17 14:39:07 kleptog kernel: eth2: Multicast list longer than mc_count.
Nov 17 14:39:07 kleptog kernel: device eth2 entered promiscuous mode
14:39:07.725128 arp who-has 192.168.15.17 tell 192.168.15.148
14:39:08.725171 arp who-has 192.168.15.17 tell 192.168.15.148

And this is from my laptop (192.168.15.17) with an MA401:

# tcpdump -i eth1 -n
tcpdump: listening on eth1
Nov 17 14:37:54 vali kernel: eth1: Multicast list longer than mc_count.
Nov 17 14:37:54 vali kernel: device eth1 entered promiscuous mode
14:37:54.686084 arp who-has 192.168.15.17 tell 192.168.15.148
14:37:54.686115 arp reply 192.168.15.17 is-at 0:30:ab:c:9:16
14:37:54.906034 192.168.15.17 > 192.168.15.148: icmp: echo request (DF)
14:37:55.686096 arp who-has 192.168.15.17 tell 192.168.15.148
14:37:55.686125 arp reply 192.168.15.17 is-at 0:30:ab:c:9:16
14:37:55.906068 192.168.15.17 > 192.168.15.148: icmp: echo request (DF)

So my laptop is receiving the packets sent, but the replies are not being
received by the other end.

Also, this error pops up on the PLX end every now and then:

Nov 17 14:31:01 kleptog kernel: eth2: Unable to determine current bitrate (0x0002)

OK, now I'm repeating the tests but with the cards swapped, so now the MA401
is in the PLX card and the Orinoco Silver is in the laptop.

Guess what. It may not be the PLX card at all. Now the desktop with PLX card
is showing:

# tcpdump -i eth2 -n
tcpdump: listening on eth2
14:53:26.125100 192.168.15.148 > 192.168.15.17: icmp: echo request (DF)
14:53:26.220596 arp who-has 192.168.15.148 tell 192.168.15.17
14:53:26.220642 arp reply 192.168.15.148 is-at 0:30:ab:c:9:16

And the laptop is showing:

# tcpdump -i eth1 -n
tcpdump: listening on eth1
14:52:02.125895 arp who-has 192.168.15.148 tell 192.168.15.17
14:52:03.126100 arp who-has 192.168.15.148 tell 192.168.15.17

The PLX card machine still shows:

Nov 17 14:52:27 kleptog kernel: eth2: Unable to determine current bitrate (0x0002)

So they may still be a problem with the PLX card still but it doesn't seem
to hurt anything.

I'm now going to trying some other stuff, maybe backporting the patch a few
versions to see if I can isolate the fault.

HTH,
-- 
Martijn van Oosterhout <kleptog at svana.org>
http://svana.org/kleptog/
> Magnetism, electricity and motion are like a three-for-two special offer:
> if you have two of them, the third one comes free.