severity 303135 important thanks On Tuesday 05 April 2005 01:11, Sven Mueller wrote: > the ata_piix kernel module blocks the PATA part of the chipset > but doesn't support the PATA devices.
Some additional info on this issue. There is a lot of info in the Ubuntu bug report referenced in this discussion on #d-boot. <Kamion> SATA blocking IDE-CD? <Kamion> is that the same as https://bugzilla.ubuntu.com/show_bug.cgi?id=1440 ? <Kamion> sadly the Ubuntu fix is rather dependent on using hotplug in the installer <fjp> Kamion: Yes, looks like same issue. <Kamion> that bug was, er, a fucking nightmare <Kamion> and that's putting it politely <fjp> I've seen at lease 10 reports on that issue. We've been kind of ignoring it thus far. <Kamion> the fix was basically to make sure the SCSI bus is scanned after PCI and USB <fjp> Kamion: #303135 has a nice summary. <Kamion> but I don't know if it depended on kernel fixes as well <Kamion> right, same thing by the looks of it <Kamion> oh, we turned on LIBATA_ENABLE_ATAPI in the kernel as well <Kamion> that was required <Kamion> but whether that works in 2.6.8 ... who knows <Kamion> I suspect it may not :( <fjp> Yes, that is what the submitter of that mail mentioned. <fjp> Guess it's mainly SATA driver immaturity, will solve itself given time. If only there were less systems using it :-P <Kamion> the submitter mentioned LIBATA_ENABLE_PATA, which is different; and defining that breaks stuff <fjp> Ah. <Kamion> because on some chipsets both piix and ata_piix get the same PCI id in the hotplug pcimap <Kamion> and they both try to claim the disk, which goes totally haywire: you get silent data corruption reading from CD-ROM <Kamion> LIBATA_ENABLE_ATAPI seemed sufficient <Kamion> if the ide_generic workaround works, that would probably be a lot easier for Debian; it would "only" be some hairy hw-detect code <fjp> Kamion: Works for installation, but will probably still cause problems after reboot because the SATA driver will be loaded by the initrd before the generic driver. <Kamion> right -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]