Most recent kernel where this bug did *NOT* occur: 2.6.20-git11 Distribution: Slackware 11 Hardware Environment: Duron 1300 MHz, VIA KT333 based ECS K7VTA3 v3, 384 MB DDR266, 1x Seagate ST3160812A 160GB HDD, 1x MAXTOR 4K080H4 80GB HDD, 1x HL-DT-ST CD-RW GCE-8523B, 1x HL-DT-ST DVDRAM GSA-4163B, hard drives on the primary channel, optical drives on the secondary Software Environment: standart for Slackware 11; kernel configured with built-in libata pata-via module; Problem Description: all kernels after 2.6.20-git11 set the ATA-7 Seagate at UDMA33, and the ATA-5 Maxtor at UDMA100
Created attachment 10618 [details] kernel config used for 2.6.21-rc2-git3
Created attachment 10619 [details] dmesg output with 2.6.21-rc2-git3
Created attachment 10620 [details] hdparm output for the drive being set at UDMA33
Created attachment 10621 [details] lspci output
Created attachment 10622 [details] rotated messages file from /var/log in the logfile you can see the output of 2.6.20-git11, and later, the newer kernels I compiled, that actually misconfigure the primary master Seagate HDD at UDMA33
the bug was still present in linux-2.6.21-rc5, just downloaded patch-2.6.21-rc5-git6 and will test it as soon as possible...
linux-2.6.21-rc5-git6 solved the issue for me...