6.0BETA3 - READ_DMA timeouts on UDMA33 drive?

[Bryan Fullerton]

Howdy,

I've been tracking 6.0BETA on a P3/450 machine that's been running
extremely solidly for several years. I've been cvsup'ing and
rebuilding once a week or so. It's not a hugely busy machine, just has
a squid cache and DNS for my home network and storage of remote
backups.

After an update to 6.0BETA3 as of last week I've started seeing
intermittant READ_DMA timeout errors (warnings?) from a standard
UDMA33 drive. I'm building BETA4 now and will see if that still has
problems.

Aug 31 00:55:25 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=33212179
Aug 31 00:58:58 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=18832671
Aug 31 02:21:49 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=12707583
Aug 31 05:55:41 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=30539199
Aug 31 20:24:29 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=2663743
Aug 31 20:55:46 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=17535935
Aug 31 20:58:06 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=1371539
Sep  1 00:48:14 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=17523487
Sep  1 07:26:19 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=75323491
Sep  1 08:19:52 katana kernel: ad0: TIMEOUT - READ_DMA retrying (1
retry left) LBA=9808191

ATA related items from dmesg:

atapci0: <Intel PIIX4 UDMA33 controller> port
0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0xb800-0xb80f at device 4.1 on
pci0
ata0: <ATA channel 0> on atapci0
ata1: <ATA channel 1> on atapci0
ad0: 39266MB <IC35L040AVER07 0 ER4OA44A> at ata0-master UDMA33
ad1: 29314MB <IBM DTLA-307030 TX4OA60A> at ata0-slave UDMA33
acd0: CDRW <YAMAHA CRW2100E/1.0H> at ata1-master PIO4

Any ideas what's up with that? Are these errors or warnings? Is this
something that's always been happening but wasn't being reported
before, or something being erroneously reported?

smartmontools reports that ad0 has zero SMART errors or failures, so
it doesn't appear to be an issue with the drive itself. A short SMART
test also passes with no errors.

Thanks,
Bryan