pata_atiixp takes a very long time to load because the Asus Pundit-R has only
1 ide controller, and pata_atiixp keeps probing for a second controller until
it finally times out and drops a shell.
[ 3.675500] scsi0 : pata_atiixp
[ 3.707256] scsi1 : pata_atiixp
[ 3.709582] ata1: PATA max UDMA/100 cmd 0x1f0 ctl 0x3f6 bmdma 0xff00 irq 14
[ 3.709588] ata2: PATA max UDMA/100 cmd 0x170 ctl 0x376 bmdma 0xff08 irq 15
[ 3.720167] ohci_hcd: 2006 August 04 USB 1.1 'Open' Host Controller (OHCI) Driver
[ 4.051499] ata1.00: ATA-6: ST3160023A, 8.01, max UDMA/100
[ 4.051507] ata1.00: 312581808 sectors, multi 16: LBA48
[ 4.051541] ata1.01: ATAPI: BENQ DVD DD DW1640, BSGB, max UDMA/33
[ 4.064765] ata1.00: configured for UDMA/100
[ 4.081262] ata1.01: configured for UDMA/33
[ 4.318754] ata2.01: failed to IDENTIFY (I/O error, err_mask=0x3)
[ 9.479632] ata2.01: failed to IDENTIFY (I/O error, err_mask=0x3)
[ 14.638291] ata2.01: failed to IDENTIFY (I/O error, err_mask=0x3)
[ 19.801926] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x3)
[ 24.963598] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x3)
[ 30.125286] ata2.00: failed to IDENTIFY (I/O error, err_mask=0x3)
I can confirm this bug on the same machine.
The problem has been reported two years ago here: https:/ /lists. linux-foundatio n.org/pipermail /bugme- new/2006- October/ 015008. html
BUT I didn't experience it with Debian Etch (Kernel 2.6.18), maybe because of fewer probing retries or shorter timeouts.
I haven't found a workaround. The kernel option ata2=noprobe doesn't work (as several other people pointed out with similar problems and different chips)