Jeff Garzik wrote:

Tejun Heo wrote:

quoted

Alan wrote:

quoted

quoted

Some SATA controllers use 0xff to indicate empty port.  This seldomly
matters as we have the almighty SStatus register to check device
presence (there is a bug regarding this, patch pending).

This GoVault drive fails because ata_piix doesn't have SCR while using
0xff to indicate port not ready (dunno exact which state causes 0xff
status tho) while the GoVault drive fails to clear that state in 150ms
(not 30s).  The libata sees 0xff after SRST if GoVault drive is 
attached

So we can also cut this down by only doing the extra polling on a device
which is SATA and lacks SCR ?

That's true but the offending one is ata_piix, so the cutting down is
not as effective.  If we can live with the extra two secs per empty port

..

While I don't mind making changes for this device, and taking into 
consideration Alan's recent comments that some ATAPI workarounds are 
still yet to appear for libata, I still dislike making changes for one 
specific device with non-standard behavior.

How does drivers/ide manage with this device, or does it?
It would be useful here to patch the PCI ID into drivers/ide
for that PIIX variant, and try it.. just to see what the
behaviour is.

There are other possible ways to avoid a 2-second per-port delay at boot.
Eg. by attempting write+readback on some of the other registers.

It all sounds very messy, but that's the ATA/ATAPI world.

Cheers