Re: PCI bug in 2.6.13

[Jeff Garzik <jgarzik@pobox.com>]

Linus Torvalds wrote:
> 
> On Sun, 11 Sep 2005, Miguel wrote:
> 
>>After applying this patch I don't see anything new so I have added the
>>same WARN_ON in pci_write_config_byte and pci_write_config_word and now
>>dmesg shows this:
> 
> 
> Thanks. Nobody should ever do a byte write to that offset, but clearly 
> something does.
> 
> And yes, that's what I missed even though I quoted it from the hpt366
> driver (heh, and nobody else noticed either):
> 
>         /* FIXME: Not portable */
>         if (dev->resource[PCI_ROM_RESOURCE].start)
>                 pci_write_config_byte(dev, PCI_ROM_ADDRESS,
>                         dev->resource[PCI_ROM_RESOURCE].start | PCI_ROM_ADDRESS_ENABLE);
> 
> I wonder how long that has been like that.
> 
> Change the pci_write_config_byte() into a pci_write_config_dword(), and I 
> bet it works. 
> 
> However, I _also_ suspect it works if you remove those lines entirely. I 
> don't see why it tries to enable the ROM in the first place - it doesn't 
> seem to be _using_ it.

I can't figure out what's going on in hpt driver, either.  Just checked 
in 2.2 (not present) and 2.4 (same write-byte code).

Maybe Alan or Bart has details?

	Jeff