In message <20050405202654.GA1824@antioche.eu.org>, Manuel Bouyer writes:
>On Tue, Apr 05, 2005 at 03:46:05AM -0500, Peter Seebach wrote:
>> I have an oldish laptop where cardbus doesn't work.  In November 2002,
>> enabling ACPI made it work.  Now it doesn't again.
>> 
>> cbb0 at pci0 dev 10 function 0: Texas Instruments PCI1410 PCI-CardBus Bridge
>> (rev. 0x01)
>> cbb0: NOT USED because of unconfigured interrupt
>> 
>> Under about 1.6J, it worked with ACPI.  Now it doesn't seem to.  Any likely
>> suggestions?
>
>I assume you have
>options   ACPI_PCI_FIXUP
>in your kernel config ?

Yup.  With or without it, no luck.  I have vague memories of being able to
get this one to work as a pcic@isa, once.

Oddly, the 1.6J kernel config file I saved does *not* have the ACPI_PCI_FIXUP
option, or any other ACPI stuff... It's been about two and a half years, but
I thought it was ACPI that fixed it, and I found the thread on port-i386.
The last thing I remember having to try was ACPI, but... Huh.  The config
file doesn't have any ACPI stuff or PCIBIOS stuff, but it apparently worked.
Now I'm mystified; I wonder whether this is the wrong config file.  If it is,
I lose; I already reformatted that.  Oops!

I think this is the general case of the "PCI BIOS didn't program an interrupt
for the chip" problem.  And I know that, on the TS5500, the solution was to
make them fix the BIOS, but... Argh.  I wish I remembered how I did this.

-s