Interrupts lost on IDE disks in RAID1 array

[Simon Vallet <debian-ppc_user@castalie.org>]

Hi,

I have a RAID1 setup on a BeigeG3, with one disk on /dev/hda, and the
second one on /dev/hdc. The machine does not seem to like the setup,
however, because I often get the following error :

ide-pmac lost interrupt, dma status: 8080
hdc: lost interrupt
hdc: dma_intr: status=0x58 { DriveReady SeekComplete DataRequest }

ide: failed opcode was: unknown

This happens at random intervals, and is _very_ annoying. I already
tried to change the disk & the cables, to no avail. In fact, I'm a bit
desperate here, and I'd like to know if this is a kernel:settings
problem (which would be fixable), or a hardware limitation: maybe this
machine does not support DMA2 on two controllers ?

There are other devices on the two IDEs : /dev/hdb is a CDROM drive
(DMA1), and /dev/hdd an IDE ZIP drive. Removing them does not help...

Thanks for any info/pointers on this.
Simon

-- 
Simon Vallet
Due to massive spam, the address shown in the From: header 
only accepts mail from Debian hosts. If you wish to mail me 
privately, just use the 'user' local part.