> When I run the following under 2.0.30, I see no problems.
>
> dd if=/dev/hda bs=512k | dd of=/dev/hdc bs=512k
>
> However, under 2.0.31, 2.0.32-pre1, and 2.0.32-pre6, problems eventually show
> up. "Eventully" means that often the command will complete the first time, but
> one of the subsequent attempts to run it will cause problems.
> 2.0.30 also had those to parameters set, but did not have CONFIG_BLK_DEV_TRITON
> set.
> hda: ST52520A, 2446MB w/112kB Cache, CHS=621/128/63, DMA
> hdb: CREATIVE CD1220E, ATAPI CDROM drive
> hdc: ST52520A, 2446MB w/112kB Cache, CHS=4970/16/63, DMA
Triton DMA chipsets have had some hidden hardware bugs before [i've
reported an ugly sporadic one some time ago], so first try to disable it.
Also, since you are mixing a DMA disk with a non-DMA ATAPI CDROM on
interface 0, you might run into buggy devices as well.
just do 'hdparm -d0 /dev/hda' [hdb, hdc] to check this theory out, then
remove the CDROM completely, and find out which pattern introduces the
bug.
-- mingo