Next-in-Thread Next Message

Sad Problems with 3.1.15 + wavelan2_cs driver 

Forum: PCMCIA Wireless Network Adapter Issues
Date: 2000, Jun 08
From: Bjorn Knutsson bjorn

Today I upgraded my TP600X to kernel 2.2.16 and while I was at it, I also upgraded från pcmcia-cs 3.1.14 to 3.1.15. Since I'm using a WaveLAN (Silver 11Mbps), I also added the wavelan2_cs driver, like I had in 3.1.14.

When I inserted the WaveLAN card, cardctl promptly died, and checking the logs, this is what I found:

 Jun  8 13:23:17 foobar cardmgr[427]: initializing socket 0
 Jun  8 13:23:17 foobar cardmgr[427]: socket 0: Lucent Technologies WaveLAN/IEEE
 Jun  8 13:23:17 foobar kernel: cs: memory probe 0xa0000000-0xa0ffffff: excluding 0xa0000000-0xa0ffffff 
 Jun  8 13:23:17 foobar kernel: cs: memory probe 0x60000000-0x60ffffff: clean. 
 Jun  8 13:23:17 foobar cardmgr[427]: executing: 'modprobe wavelan2_cs station_name=foobar enable_encryption=y transmit_key_id=1 channel=0 port_type=1 network_name=FOOBAR key_1=\"snafu\"'
 Jun  8 13:23:17 foobar kernel: wvlan_cs: WaveLAN/IEEE PCMCIA driver v1.0.4 
 Jun  8 13:23:17 foobar kernel: wvlan_cs: (c) Andreas Neuhaus <andy@fasta.fh-dortmund.de> 

It seems it tried to load two different drivers. I then configured, compiled and installed from the 3.1.14 sources, and 3.1.14 + wavelan2_cs works fine under 2.2.16. Something must have been introduced in 3.1.15 that either breaks with wavelan2_cs, or with kernel 2.2.16.

Next-in-Thread Next Message

Messages Inline: 0 1

None Fishy modules.conf setup??

Re: Sad Problems with 3.1.15 + wavelan2_cs driver (Bjorn Knutsson)
Date: 2000, Jun 08
From: David Hinds <dhinds@pcmcia.sourceforge.org>

On Thu, Jun 08, 2000 at 04:39:29AM -0700, Bjorn Knutsson wrote:

>  Jun  8 13:23:17 foobar cardmgr[427]: executing: 'modprobe wavelan2_cs station_name=foobar enable_encryption=y transmit_key_id=1 channel=0 port_type=1 network_name=FOOBAR key_1=\"snafu\"'
>  Jun  8 13:23:17 foobar kernel: wvlan_cs: WaveLAN/IEEE PCMCIA driver v1.0.4 

Well it certainly asked for wavelan2_cs to be loaded.

The only thing I can think of is that you've got a line in
/etc/modules.conf that aliases wavelan2_cs to wvlan_cs.  The 3.1.15
drivers use modprobe by default so aliases will be followed... but
3.1.14 uses insmod.  As for where the alias came from, you've got me.

-- Dave

Disagree No fish in my conf.modules.

Re: None Fishy modules.conf setup?? (David Hinds)
Date: 2000, Jun 08
From: Bjorn Knutsson bjorn

I wish it was that simple, but my conf.modules is fishless. It does not deal with anything related to PCMCIA modules, only sound and tpctl (Thinkpad control program) modules.

And don't forget: cardmgr died when the card was inserted. If it was only a broken alias, I would hope that cardmgr would be robust enough to survive that.

If there are any additional things you want me to do, I'll be happy to experiment.

None Re: Disagree: No fish in my conf.modules.

Re: Disagree No fish in my conf.modules. (Bjorn Knutsson)
Date: 2000, Jun 09
From: David Hinds <dhinds@pcmcia.sourceforge.org>

I would try issuing that modprobe command by hand, adding a "-v", and
see what it does.

The log doesn't seem to include any hints about how/when cardmgr might
have died.  I guess you just noticed it wasn't running any more after
you inserted the card?

I guess I'd also look at /lib/modules/.../modules.dep and examine the
dependencies for wavelan2_cs.o.

-- Dave

News Same old problem, new pcmcia_cs-version.

Re: None Re: Disagree: No fish in my conf.modules. (David Hinds)
Date: 2000, Jul 18
From: Bjorn Knutsson bjorn

I'm sorry about not getting back about the problem, but I was swamped with work.

I just installed pcmcia-cs 3.1.18 + wavelan2_cs 6.01BETA from contrib, and my troubles continue. When I insert the Wavelan card, cardmgr dies.

> I would try issuing that modprobe command by hand, adding a "-v", and see what it does.

Loading the module is not a problem. It just echos the insmod-line and nothing more.

> The log doesn't seem to include any hints about how/when cardmgr might have died. I guess you just noticed it wasn't running any more after you inserted the card?

It now definitely dies at the point where it loads the module. It now logs no more messages after the modprobe for wavelan2_cs, and checking the system immediately after, the wavelan2_cs module is loaded, and cardmgr is not around anymore.

> I guess I'd also look at /lib/modules/.../modules.dep and examine the dependencies for wavelan2_cs.o.

No problems there either, from what I can see.

I have experimented some more. This is what happens now:

If I insert a card, say my old 3c562C, it configures OK. So I unplug it, and insert my Wavelan. It comes to the modprobe wavelan2_cs, and then I see no more messages in the logs (I run cardmgr with -v). And looking around, cardmgr is no longer running. Now, if I at this point manually run cardmgr -v -f, it will complain about "could not adjust resource" for memory and IO ports, but will then initialize socket 0 and correctly configure the wavelan2_cs driver and everything is fine.

Remove all cards and restart pcmcia. I now insert my 3c562c in socket 0. It configures OK. I then insert the Wavelan in socket 1. cardmgr dies, as before, after the modprobe. I, again, run 'cardmgr -v -f'. Lots of "could not adjust resource" as before, but the card in socket 0 is now identified as "Anonymous Memory". The wavelan2_cs driver is again configured correctly, and everything is fine with the wavelan card.

So to conclude: cardmgr seems to die when or soon after the modprobe for wavelan2_cs is issued. It seems that it will configure correctly when the module is already loaded (and whatever else cardmgr may have node). I did, however, try to modprobe the module and then insert the card, but then my machine locked up.

Here's the log:

Jul 18 17:07:14 foobar cardmgr[480]: initializing socket 0
Jul 18 17:07:14 foobar cardmgr[480]: socket 0: Lucent Technologies WaveLAN/IEEE
Jul 18 17:07:14 foobar kernel: cs: memory probe 0xa0000000-0xa0ffffff: excluding 0xa0000000-0xa0ffffff 
Jul 18 17:07:14 foobar kernel: cs: memory probe 0x60000000-0x60ffffff: clean. 
Jul 18 17:07:14 foobar cardmgr[480]:   product info: "Lucent Technologies", "WaveLAN/IEEE", "Version 01.01", ""
Jul 18 17:07:14 foobar cardmgr[480]:   manfid: 0x0156, 0x0002  function: 6 (network)
Jul 18 17:07:14 foobar cardmgr[480]: executing: 'modprobe wavelan2_cs station_name=foobar enable_encryption=y transmit_key_id=1 channel=0 port_type=1 network_name=FOOBAR key_1=\"snafu\"'

Restarted cardmgr:

Jul 18 17:07:31 foobar cardmgr[625]: starting, version is 3.1.18
Jul 18 17:07:31 foobar cardmgr[625]: watching 2 sockets
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: IO ports 0xc00-0xcff: Device or resource busy
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: IO ports 0x800-0x8ff: Device or resource busy
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: IO ports 0x100-0x4ff: Device or resource busy
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: memory 0xc0000-0xfffff: Input/output error
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: memory 0x60000000-0x60ffffff: Input/output error
Jul 18 17:07:31 foobar cardmgr[625]: could not adjust resource: IO ports 0xa00-0xaff: Device or resource busy
Jul 18 17:07:31 foobar cardmgr[625]: initializing socket 0
Jul 18 17:07:31 foobar cardmgr[625]: socket 0: Lucent Technologies WaveLAN/IEEE
Jul 18 17:07:31 foobar cardmgr[625]:   product info: "Lucent Technologies", "WaveLAN/IEEE", "Version 01.01", ""
Jul 18 17:07:31 foobar cardmgr[625]:   manfid: 0x0156, 0x0002  function: 6 (network)
Jul 18 17:07:31 foobar kernel: eth0: WaveLAN/IEEE, io_addr 0x100, irq 3, mac_address xx:xx:xx:xx:xx:xx 
Jul 18 17:07:31 foobar cardmgr[625]: executing: './network start eth0'
Jul 18 17:07:32 foobar pumpd[638]: starting at (uptime 0 days, 0:03:55) Tue Jul 18 17:07:32 2000  
Jul 18 17:07:32 foobar pumpd[638]: configured interface eth0
Jul 18 17:07:32 foobar cardmgr[625]: exiting

(I edited the mac_address)

None Fixed: a really dumb bug

Re: News Same old problem, new pcmcia_cs-version. (Bjorn Knutsson)
Date: 2000, Jul 22
From: David Hinds <dhinds@pcmcia.sourceforge.org>

The bug is fixed in the 21-Jul-00 beta on projects.sourceforge.net in
/pub/pcmcia-cs/NEW.  The long list of module options was overrunning a
fixed-length buffer.

-- Dave

Ok Re: Fixed: a really dumb bug

Re: None Fixed: a really dumb bug (David Hinds)
Date: 2000, Jul 22
From: Bjorn Knutsson bjorn

> The bug is fixed in the 21-Jul-00 beta on projects.sourceforge.net in
> /pub/pcmcia-cs/NEW.  The long list of module options was overrunning a
> fixed-length buffer.

I'm out of office again, but I brought the WaveLAN card this time. I compiled and installed the new version, and it now configures the card properly (well, since I don't have a WaveLAN net around, all I really know is that it gets past the point where cardmgr used to die and it starts DHCP:ing for an address).

Thanks! Great work!

Problems with 3.1.15 + wavelan2_cs driver


Members Subscribe No Admin Mode Show Frames Help for HyperNews at pcmcia-cs.sourceforge.net 1.10
[ Edit This Forum ]