On 11/03/2012 12:48 PM, Alan Cox wrote:

On Sat, 3 Nov 2012 05:26:35 +0100
Borislav Petkov [off-list ref] wrote:

quoted

On Wed, Oct 24, 2012 at 11:57:46AM +0100, Alan Cox wrote:

quoted

Which is an ATA layer bug - adev->dma_mode should never be called
without a DMA mode in normal use.

Ok, it looks like this would take a while to fix.

So a 30 second glance says that the problem is that you seem to have
dma_mode uninitialised as zero which is bogus.

That means either ata_acpi_gtm_xfermask broke (it should have set the
bits to 0xFF if no mode is found), ata_dma_enabled is broken, or
pacpi_qc_issue got called before pacpi_port_start (which seems wildly
unlikely)

Needs someone to go and dump the relevant values in the right places and
see what is breaking in the pata_acpi setup logic.

Agreed -- though the WARN_ONCE() will at least give us trivially better 
poops.

	Jeff