driver loading troubles
wim at djo.tudelft.nl
wim
Sun Oct 12 13:48:56 PDT 2003
Hello list,
just a few phenomena to mention:
although I could flash a prism2.5 card (BenQ AWL100) with FLASH.EXE,
it seems that more of these experiments don't come true that easily.
(That card however still works like a charm.)
Then my eye fell on a harmless looking option in prism2dl (wlan_ng):
-S <str> Sernum) Overwrite the serial number PDR with the given
string. String must be <= 12 characters, any
extra will be truncated.
The serial number doesn't serve anything critical.
I decided to try to change that first.
The program starts, the LED flashes once, then nothing happens for a
long timeout, one error gets printed and after another long timeout the
prompt gets back. Nothing has changed on the card.
Is this a specific prism2.5 property?
Since I couldn't write this way to the prism2.5 card, I then played with a
prism2 card (Belkin F5D6020) that used to easily work with hostap and
wlan_ng and windows.
The serial number change seemed to have taken place (according to windows
flashupdate program). The card still works in windows and with wlan-ng, but
hostap has problems now. IF the card is recognised, it seems to work fine.
The problem is to get it so far. Initially inserting the card will not
always make ik work. Subsequent 'cardctl eject' and 'cardctl insert' will
usually not get it do anything better. (Sometimes it does.)
If not, I must shut down and restart PCMCIA services a number of times,
or even power down the machine.
I cannot find a sequence to smoothly activate the card, except for a power
down.
typical logs are then:
Oct 12 19:11:58 tp600 cardmgr[7165]: initializing socket 0
Oct 12 19:11:58 tp600 cardmgr[7165]: socket 0: Senao SL-2011CD/SL-2011CDPLUS
Oct 12 19:11:58 tp600 cardmgr[7165]: executing: 'modprobe hostap'
Oct 12 19:11:58 tp600 cardmgr[7165]: executing: 'modprobe hostap_cs'
Oct 12 19:11:58 tp600 kernel: hostap_cs: CVS (Jouni Malinen <jkmaline at cc.hut.fi>)
Oct 12 19:11:58 tp600 kernel: hostap_cs: index 0x01: Vcc 5.0, irq 3, io 0x0100-0x013f
Oct 12 19:11:58 tp600 kernel: hostap_cs: Registered netdevice wlan0
hostap_cs: Initialization failed
Oct 12 19:12:02 tp600 cardmgr[7165]: executing: './network start wlan0'
Oct 12 19:12:02 tp600 kernel: hostap_cs: Initialization failed
Oct 12 19:12:24 tp600 /etc/hotplug/net.agent: invoke ifup wlan0
Oct 12 19:12:49 tp600 kernel: wlan0: MAC port 0 enabling failed
Oct 12 19:12:49 tp600 kernel: wlan0: could not enable MAC port
And sometimes messages like this:
cardctl eject
hostap_cs: detach postponed, 'wlan0' still locked
hostap_cs: Shutdown failed
wlan0: interrupt delivery does not seem to work
At shutting down the pcmcia services, the modules (yenta_socket and
pcmcia_cs) in that case cannot be removed (busy) until I physically
remove the card and run pcmcia stop a few times. Or it won't even do
so at all, like now.
Shutting down PCMCIA services: cardmgr modules: Device or resource busy
yenta_socket: Device or resource busy
pcmcia_core: Device or resource busy
Whatever I try, it remains busy.
One could suggest that the (IBM TP600) PCMCIA subsystem is flakey, but
there haven't been equal problems in the past and there are no(t yet)
problems with other cards, or with the same card with other drivers.
So I tend to think that it is hostap that somehow manages to trick the
PCMCIA hardware into a lock-up situation.
(Yesterdays CVS version didn't change anything in that matter.)
Any ideas?
Regards,
Wim Osterholt
More information about the Hostap
mailing list