Re: 2.6.25-rc2 + smartd = hang

[Mark Lord <liml@rtr.ca>]

Anders Eriksson wrote:
> 
> jeff@garzik.org said:
>> The sysrq-e output is probably just standard ext3 journalling unrelated  to
>> the problem...  what does dmesg say?  lspci?  What's your hardware setup? 
> 
> 
> dmesg ; smartd ; dmesg yields no new entries in dmesg. It seems on disk 
> accesses are dead. it still routes packets fine.
> 
> This is an old PII-300 with 2 IDS disks and a DVD R/W. 
...
> Feb 22 17:38:49 tippex Uniform Multi-Platform E-IDE driver
> Feb 22 17:38:49 tippex ide: Assuming 33MHz system bus speed for PIO modes; override with idebus=xx
> Feb 22 17:38:49 tippex PIIX4: IDE controller (0x8086:0x7111 rev 0x01) at  PCI slot 0000:00:07.1
> Feb 22 17:38:49 tippex PIIX4: not 100% native mode: will probe irqs later
> Feb 22 17:38:49 tippex ide0: BM-DMA at 0xffa0-0xffa7, BIOS settings: hda:DMA, hdb:PIO
> Feb 22 17:38:49 tippex ide1: BM-DMA at 0xffa8-0xffaf, BIOS settings: hdc:PIO, hdd:PIO
> Feb 22 17:38:49 tippex Probing IDE interface ide0...
> Feb 22 17:38:49 tippex hdb: IC35L120AVV207-0, ATA DISK drive
> Feb 22 17:38:49 tippex hda: IBM-DTTA-371010, ATA DISK drive
> Feb 22 17:38:49 tippex hda: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> Feb 22 17:38:49 tippex hda: UDMA/33 mode selected
> Feb 22 17:38:49 tippex hdb: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> Feb 22 17:38:49 tippex hdb: UDMA/33 mode selected
> Feb 22 17:38:49 tippex Probing IDE interface ide1...
> Feb 22 17:38:49 tippex hdd: Maxtor 6L250R0, ATA DISK drive
> Feb 22 17:38:49 tippex hdc: AOPEN DUW1608/ARR, ATAPI CD/DVD-ROM drive
> Feb 22 17:38:49 tippex hdc: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> Feb 22 17:38:49 tippex hdc: UDMA/33 mode selected
> Feb 22 17:38:49 tippex hdd: host max PIO4 wanted PIO255(auto-tune) selected PIO4
> Feb 22 17:38:49 tippex hdd: UDMA/33 mode selected
> Feb 22 17:38:49 tippex ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
> Feb 22 17:38:49 tippex ide1 at 0x170-0x177,0x376 on irq 15
..

So that's using the old IDE drivers.
And the network and USB are sharing IRQ#11 with each other.

If you are going to be using newer kernels like this (2.6.23+),
then you might consider shifting those drives over to libata drivers.

This involves a little bit of work -- building a kernel with libata
and "ata_piix" built-in instead of the old IDE drivers,
and then rearranging /etc/fstab to match the new device names
(eg. /dev/sda instead of /dev/hda).

But at this point libata is working much better than the old IDE stuff,
and it really is worth moving things over if you can.

Cheers