Most recent kernel where this bug did not occur: ? (around 2.6.9 was OK)
Hardware Environment: Intel TD945GTP motherboard
Software Environment: Gentoo
The disks connected to the SATA controller are not visible (before they were
visible as /dev/sd[abc]). The kernel log mentions disabling them:
Dec 13 19:51:37 tachyon libata version 1.12 loaded.
Dec 13 19:51:37 tachyon ata_piix version 1.04
Dec 13 19:51:37 tachyon ACPI: PCI Interrupt 0000:00:1f.2[B] -> GSI 19 (level,
low) -> IRQ 18
Dec 13 19:51:37 tachyon PCI: Setting latency timer of device 0000:00:1f.2 to 64
Dec 13 19:51:37 tachyon ata1: SATA max UDMA/133 cmd 0x20C8 ctl 0x20EE bmdma
0x20A0 irq 18
Dec 13 19:51:37 tachyon ata2: SATA max UDMA/133 cmd 0x20C0 ctl 0x20EA bmdma
0x20A8 irq 18
Dec 13 19:51:37 tachyon ATA: abnormal status 0x7F on port 0x20CF
Dec 13 19:51:37 tachyon ata1: disabling port
Dec 13 19:51:37 tachyon scsi0 : ata_piix
Dec 13 19:51:37 tachyon ATA: abnormal status 0x7F on port 0x20C7
Dec 13 19:51:37 tachyon ata2: disabling port
Dec 13 19:51:37 tachyon scsi1 : ata_piix
Bug originally posted as http://bugs.gentoo.org/show_bug.cgi?id=115441
Confirmed with vanilla kernel 2.6.15_rc7.
Steps to reproduce:
1. install kernel 2.6.14 with ata_piix built-in or modular
2. reboot into the new kernel
3. ls /dev
Works fine with 2.6.12 (Gentoo 2005.1 LiveCD).
I made some additional tests with different kernel versions. None of the
"vanilla" kernels from 2.6.12 up to (and including) 2.6.15 works. The
interesting fact is, that up to version 126.96.36.199 (libata 1.12, ata_piix 1.04)
the system log contains:
Jan 13 20:47:59 tachyon ata1: SATA port has no device.
but starting from version 2.6.14 (the same libata 1.12, ata_piix 1.04), it says:
Jan 13 23:01:47 tachyon ata1: disabling port
I don't know what makes it work on the Gentoo 2005.1 LiveCD. It may be either
one of the patches or the configuration.
Created attachment 7021 [details]
my 2.6.15 .config
It may be the case that I am just missing something in my kernel config...
Could anyone verify that?
Have you tried latest kernels, is the problem still there?
This should have been fixed now. Please reopen if the problem persists.