RAID6 deg on 2x300GB IDE process hang during read/bad md5sum

This is a 600GB RAID6 array in degraded mode across 2x300GB PATA IDE drives, one partition as /dev/md0, ext3 file system. Formats ok. Files wrote to it ok (about 75% full). A few dozen out of a few thousand of files have a bad md5sum when compared to original on an 400GB SATA disc with NTFS or FAT32 (formatted in Linux). So far when running md5sum on the files, 2 files (maybe more as I proceed) enter a D state (uninterruptable), but there's no disc I/O, no activity, and the process hangs indefinitely, even if the parent is killed. If I try to md5sum the file again, hang. If I reboot and try, hang. It's consistent which files it hangs on, so far only 2. Ideas? When I finish moving data from SATA to PATA then I will bring the SATA drives in to the array.

I run a command sequence such as this to generate md5sums, then diff source and target MD5SUM files:

cd /mnt/md0/path1
find path2/dir -print0 | xargs -0 -n 1 md5sum >> /mnt/md0/MD5SUM.md0_path2_dir.md5 2> /dev/null &
tail -f /mnt/md0/MD5SUM.md0_path2_dir.md5 &

Any ideas?