Re: DMA in HVM guest on x86_64

[Michael Best <mbest@pendragon.org>]

Michael Best wrote:
> I've been following unstable day to day with mercurial but I'm still 
> having a problem with my HVM testing.
>
> I using the i686 Centos + Bluecurve isntaller and I get the following 
> error in the guest during disk formatting:
>
> <4>hda: dma_timer_expiry: dma status == 0x21
> <4>hda: DMA timeout error
> <4>hda: dma timeout error: status=0x58 { DriveReady SeekComplete 
> DataRequest }
> <4>
> <4>ide: failed opcode was: unknown
>
> Not sure if there is a solution to this at present, or if some patches 
> are still coming down the road.
>
> -Mike 

My dom0 is throwing errors such:

ata2: command 0xea timeout, stat 0x51 host_stat 0x0
ata2: translated ATA stat/err 0x51/04 to SCSI SK/ASC/ASCQ 0xb/00/00
ata2: status=0x51 { DriveReady Seek Complete Error }
ata2: error=0x04 { DriveStatusError }
ata2: command 0xea timeout, stat 0x50 host_stat 0x0
ata2: command 0x35 timeout, stat 0x50 host_stat 0x44
ata2: command 0xea timeout, stat 0x50 host_stat 0x0
ata2: command 0xea timeout, stat 0x50 host_stat 0x0
ata2: command 0x35 timeout, stat 0x50 host_stat 0x44
ata2: command 0xea timeout, stat 0x50 host_stat 0x0


I've been trying a number of different builds of Xen, and the unstable 
branch is causing me the most trouble in this regard (I've seen other 
really odd disk problems as well in non HVM domains).

My build of stable/testing seems to work just fine on the other hand.

Let me know if this sort of bug report is useful.

I'm allocating an LVM partition to be used at the HVM disk, and I'm 
running on an Nvidia 6150 chipset board (sata_nv + forcedeth) with an  - 
AM2 AMD64x2 3800

-Mike