5.2.1 p9 --> 5.2 current 20040708 upgrade ATA_IDENTIFY no interrupt

[jhandvil]

On Sunday 11 July 2004 10:43 am, jhandvil wrote:
> On Sunday 11 July 2004 07:34 am, Ivan Voras wrote:
> > jhandvil wrote:
> > > Just upgraded from 5.2.1 p9 to 5.2-current.  Received the following
> > > error during boot (copied by hand, please excuse typo's):
> > >
> > > ata0-master: FAILURE - ATA_IDENTIFY no interrupt
> > > ata0-master: FAILURE - ATA_IDENTIFY no interrupt
> > > Mounting root from ufs:/dev/ad0s1a
> > > setrootbyname failed
> > > ffs_mountroot: can't find rootvp
> > > Root mount failed: 6
> >
> > I have a very similar situation at my machine. Could you try toggling
> > WITNESS in the kernel config and report to the list if there are any
> > changes?
>
> Thanks for the fast response.  I toggled turned off WITNESS (and turned on
> DDB), and I got the same response.
>
> I tried digging some more.  I know that there was an issue earlier with the
> via 8237 and dma, so I tried disabling dma (hw.ata.ata_dma=0).  This had no
> effect.  I'll attempt to dig around in the PCI driver and the ata driver a
> little more and see if I can add some traces around the interrupt
> assignment. Any additional suggestions would be much appreciated.

Additional information:  I added a printf at line 236 of ata-queue.c (in 
function ata_completed).  The output shows the values in the ata_request 
structure:

...
command=0xec  (ATA_ATA_IDENTIFY)
flags=0x282 (ATA_R_IMMEDIATE | ATA_R_READ | ATA_R_TIMEOUT)
retries=-1
...

This command is called from ata_identify_devices.  I will trace out further to 
see what is being passed in the ata_channel struct.  Please let me know any 
additional traces that will help to find the root cause.

Thanks,
Justin