The merge on 08/05/2004 broke the driver for our 5212 mini pci cards.
CVS from 08/04/2004 works fine.=20
Two machines with atheros Mini PCI cards are set up next to each other,
one master mode and one managed. The AP crashes after two or three
pings across the link. Sometimes the AP does not crash, sometimes it
only becomes very slow and not responsive, but either way you can no
longer ping across the link until you reset the card in the AP. When
the machine manages not to crash the driver spits out:=20
"beacon queue 9 did not stop?"
It prints this repeatedly and very fast. 10 times for every status line
"00:02:6F:20:F5:CF: ok 1 err 0 retr 0...." that is printed.
We have reproduced this problem with two different brands of MiniPci
card (Both 5212) and with 2 different models of soekris board (1 amd and
1 geode) and 1 model of the WRAP board (geode).
A search of the message lists reveal at least one other person with this
He describes his platform as:
CARDBUS-controller: TI-1420 (texas instruments)
It appears that the driver is calling ath_hal_stoptxdma from the
ath_beacon_tasklet function, and that ath_hal_stoptxdma is continuously
returning 0. Unfortunately I could not pursue the problem further
because I don't have access to the hal code :-(.
I can however reproduce the problem easily and repeatedly and would be
happy to host any developer that would be interested In examining and
debugging the problem. I can provide ssh access to test machines and
hang out on the telephone to reboot them for you when you crash them.
1800-243-3729 ext 204
Get latest updates about Open Source Projects, Conferences and News.