PCMCIA issues

[Giuseppe Argentieri]

On Wed, Jul 13, 2005 at 01:10:29PM +0200, Stefan Bauer wrote:

> as hint:
 
> if its a 16bit pcmcia card u have to map the space to _right_ position.
> a good value would be at least
 
> sysctl -w hw.cbb.start_memory="0x21000000"
 
This is a good point but... how can I find the _right_ position?
There must be a better method than trying all possible offsets rebooting
hundreds of times.


> next step i would recomend is to set the debug level higher and then
> remove & plugin the wi card again:

> do a

> sysctl hw.wi.debug=1

> and have a look if there is any output.

No output. I built a kernel.debug, too. Booting in verbose mode, with

hw.cbb.debug=1
hw.pccard.debug=1
hw.cardbus.debug=1
hw.pccard.cis_debug=1
hw.cardbus.cis_debug=1
hw.wi.debug=1

I have the same messages. Well, not exactly the same: this time there is
not the "Bad Vcc request" error but the pcmcia slot still not works.

I think there is no way to increase the debug level. I peeked at the
sources in dev/pccard/pccard.c dev/cardbus/cardbus.c dev/pccbb/pccbb.c
but those TUNABLES seem to be only flags. I mean: putting
hw.pccard.debug=1 or hw.pccard.debug=3 doesn't change anything.

In pccbb.c there are two o2micro_power_hack() functions. Maybe O2Micro
Controllers are not well supported at the present time. I'm going to put
some printf around in order to have more information about those
variables and structs, but honestly I have no idea of what they mean.


> in case nothing happend, reboot the machine without acpi and see if you
> get the same behaviors.

Disabling ACPI the machine goes in kernel panic while booting.

According to acpi(4) it's possible to disable only some acpi sub-drivers
via "debug.acpi.disabled" environment variable. I'll try these tricks.



-- 
http://swpat.ffii.org/

Giuseppe