Bug/panic: ATAPI fails with 2.4.20-pre10-ac2

System:
MSI KT3 Ultra2 (KT333/VT8235 bridges)
XP1700+
1G DDR RAM
Seagate ST380021A
Plexwriter PX-W2410A
Radeon 32M SDR
Ensoniq 5880 AudioPCI
SMC EtherPower II
RedHat 8.0
GCC 3.2

Linux 2.4.20-pre10:

Setup A: IDE0 = HD, IDE1 = HD
and
Setup B: IDE0 = HD, IDE1 = ATAPI CD-Writer w/ide-scsi driver

Everything works, but only in PIO mode.

Linux 2.4.20-pre10-ac2:

Setup A: IDE0 = HD, IDE1 = HD

Everything works in UDMA mode 5.

Setup B: IDE0 = HD, IDE1 = ATAPI CD-Writer w/ide-scsi driver

The HD works in UDMA mode 5, but the ATAPI CD-Writer is unusable. When I try to mount a CD-ROM, the kernel spews lots of messages like these:
hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest }
hdc: drive not ready for command
hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest }
hdc: drive not ready for command
hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest }
hdc: drive not ready for command
hdc: status error: status=0x58 { DriveReady SeekComplete DataRequest }
hdc: drive not ready for command
hdc: status error: status=0x49 { DriveReady DataRequest Error }
hdc: status error: error=0x04
hdc: drive not ready for command
hdc: ATAPI reset complete
Usually the CD-ROM can be mounted and read, although it is extremely slow. Once, however, the kernel paniced before it got as far as mounting it. Unfortunately the oops was not captured in the logs, but I think its last words were something like "AIEEE, killed interrupt handler".

If anyone is willing to look into this problem I'd be happy to send you whatever logs, config, and /proc information you need. I can also try out new patches.

Thor