On Sunday 15 April 2007 21:50, Joey Hess wrote:
> I'm concerned that this same problem might affect _all_ pata
> controllers, since the pata modules have been added, which also claim
> the same pci ids as other modules. I looked for such conflicting claims
> using this rune:
Note that the case for the new pata_* controllers is a bit different from
the piix / ata_piix conflict as ata_piix was an already existing driver.
This command shows which PCI IDs covered by piix are now also covered by
another driver:
/lib/modules/2.6.20-1-686$ grep "^piix[[:space:]]" modules.pcimap | \
awk '{print $2" "$3}' | sort | while read ID; do echo ">> $ID"; \
grep "$ID" modules.pcimap; done
4 not yet covered by a new driver
18 now also covered by ata_piix
1 now also covered by pata_mpiix
1 now also covered by pata_oldpiix
For 2.6.18 all 24 devices were only covered by the piix module.
As disabling ata_piix is not an option and as there is no option with
ata_piix to disable support for devices already covered by piix,
preventing "double" support would require a kernel patch to exclude those
18 PCI IDs for ata_piix.
P.S. 8086:24db and 8086:25a2 are both listed _twice_ for ata_piix!
Attachment:
pgpwCvEliSQo6.pgp
Description: PGP signature