IDE hard drives & DMA

[dogbert at clue4all.net (Brian Conway)]

I have a multitude of IDE and SCSI disks on one box of mine, and recent ran
into a problem after upgrading to the 2.2.11 kernel.  One of my 3 IDE drives
will no longer initialize with DMA enabled, and this causes a performance
issue.  This problem never showed up in the 2.2.5 or 2.2.10 releases.  Here
are the partial contents of /var/log/dmesg on my rh6 box:

<snip>
hda: Maxtor 72004 AP, ATA DISK drive
hdb: SAMSUNG WN321620A (2.16 GB), ATA DISK drive
hdc: CRD-8160B, ATAPI CDROM drive
hdd: WDC AC22500L, ATA DISK drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: Maxtor 72004 AP, 1916MB w/128kB Cache, CHS=3893/16/63, DMA
hdb: SAMSUNG WN321620A (2.16 GB), 2060MB w/109kB Cache, CHS=4186/16/63, DMA
hdd: WDC AC22500L, 2441MB w/256kB Cache, CHS=4960/16/63, (U)DMA
hdc: ATAPI 16X CD-ROM drive, 128kB Cache
...
Partition check:
 sda: sda1 sda2
 sdb: sdb1 sdb2
 hda: hda1 hda2 hda3
 hdb: [PTBL] [523/128/63] hdb1 hdb2
 hdd:hdd: timeout waiting for DMA
hdd: irq timeout: status=0x58 { DriveReady SeekComplete DataRequest }
hdd: DMA disabled
ide1: reset: success
 hdd1 hdd2
<snip>

I have never run into this before.  I'm curious where I can tell the hdd to
use dma and any other configuration options I've deemed necessary.  I know
hdparm can do some interesting stuff, but I'm curious as to why this is only
showing up now (assuming it's not a big kernel bug).  Ideas?

Brian Conway
dogbert at clue4all.net

Those who do not understand Unix are condemned to reinvent it, poorly.
                -- Henry Spencer




-
Subcription/unsubscription/info requests: send e-mail with
"subscribe", "unsubscribe", or "info" on the first line of the
message body to discuss-request at blu.org (Subject line is ignored).