ATAng still causing a lot of pain.

I recently re-installed a SCSI tape drive on my -CURRENT server, and began
loading the IDE bus more than I head in a few months.  I immediate started
getting console logs like:

    Oct 10 14:39:26 kanga kernel: ata0: resetting devices ..
    Oct 10 14:39:26 kanga kernel: done
    Oct 10 14:39:26 kanga kernel: ad0: TIMEOUT - READ_DMA retrying (1 retry left)

immediately followed by panics.  I'd been using a build from 2003-09-19, but
rebuilt world from source from 2003-10-10.  The problem still persists
unless I boot with 'hw.ata.ata_dma="0"' in /boot/loader.conf.local.

The IDE bus in question is:

    atapci0: <VIA 82C596B UDMA66 controller> port 0xd800-0xd80f at device 4.1 on pci0

I understand that there may be a few bugs left in ATAng, and as a user of
-CURRENT, I'm OK with that.  Is there anything I can do that would probably
let me run my system a bit faster than PIO4 will allow and still limit my
likelihood of getting bitten by the READ_DMA timeout panics?
-- 
Kirk Strauser

"94 outdated ports on the box,
 94 outdated ports.
 Portupgrade one, an hour 'til done,
 82 outdated ports on the box."