Subject: Re: Problems with 1.5_BETA CardBus
To: None <current-users@netbsd.org>
From: Tero Kivinen <kivinen@ssh.fi>
List: current-users
Date: 11/22/2000 05:10:01
> I've seen a few reports now of PCMCIA cards that work when using a
> PCMCIA controller, but fail on the same laptop when the PCMCIA
> controller is attached as a CardBus controller (see my recent
> messages to port-i386 for an example).

At least that happens both in my laptop (HP OmniBook 6000) and with my
friends laptop (Dell Inspiron 5000e).

If I turn on cardbus support in my kernel, then only the lower PCMCIA
slot is working. If I turn the cardbus support both of them work. The
problem in my machine seems to be in the interrupt allocation system.
The PCIBIOS seems to allocate IRQ 10 for both cardbus slots, the
internal 3com 3c556 MiniPCI 10/100 ethernet card (ex0) and the usb
chip.

Only the pcmcia0 slot is working, all others does not get interrupts
properly (or at least the pcmcia1 and the ex0 cards don't get
interrupts at all, don't know about the usb).

When I turn off the cardbus support both pcmcia slots gets different
irq and then they work.

I have been iterating with the PCIBIOS, PCIBIOS_INTR_FIXUP,
PCIBIOS_IRQS_HINT, PCIBIOS_INTR_GUESS, and PCIBIOS_INTR_FIXUP_FORCE,
but I have not yet found a combination that would give me a working
ex0 driver (I don't care about the second pcmcia slot if I can get the
builtin ethernet work).

On the Dell Inspiron 5000e the problem is that if the cardbus support
is turned on then the WaveLan card does not work at all, and if it is
turned off then his cardbus ethernet card does not work... He is
currently rebooting between two kernels one having cardbus support and
another not having it...

He also have a problem on the Dell Inspiron 5000e that if he tries to
enable the apm on the kernel then the kernel will crash in the boot,
while doing the first apmcall (kernel page fault or something).
This same thing happens on the linux, and somebody from the Dell said
that those machines don't support APM at all only ACPI...

If anybody has any idea what should I try next to get those interrupts
going correctly send me email. I am almost ready to put a timer in
kernel that will call the ex_intr every millisecond to get packets
going :-)
-- 
kivinen@ssh.fi                               Work : +358 303 9870
SSH Communications Security                  http://www.ssh.fi/
SSH IPSEC Toolkit                            http://www.ssh.fi/ipsec/