PATCH: uniata: v0.41b2 does not see IDE "slave" drives on some controllers[upstream]

Created an attachment (id=6611)
Debug log of testing ReactOS build with UNIATA 0.41b2 attempting to boot from a CD set as "slave"

I wish I had noticed this back during the call for testing on the new UNIATA, but it seems that the new UNIATA 0.41b2 has problems seeing IDE drives (both CD and hard drives) connected as "slave" to the controller on real hardware.

I have been able to replicate this on two different controllers (SB700/SB800 and a Via VT82C686B), and it also occurs in an old version of QEMU (version 0.6.2)

Trunk builds 51735 and prior will recognize the "slave" drive just fine while build 51738 (after the uniata 0.41b2 update) and later fails to see it.

I have tried this in a number of configurations to verify this behavior.

If a CD is the "master" device, and HD is "slave", I can boot a live CD but the HD is not visible as a drive. An installer CD will say it can't find the drive. Attempting to boot from the HD will crash.

If a HD is the "master" device, and the CD is "slave", I can boot the HD but the CD will not be visible as a drive. Attempting to boot a UNIATA 0.41b2 ReactOS CD will crash.

To simplify things, it is also possible to connect just the CD as "slave" without any other device on the controller. Again, attempting to boot a 0.41b2 UNIATA ReactOS CD will crash. (And earlier versions work fine).

I'm attaching a debug log with ReactOS attempting to boot from the "slave" CD, no HD attached, using the newer 0.41b2 (this is from an older ReactOS revision that was posted with the newer debug UNIATA for testing prior to it being added to the trunk)