Re: STANDBY IMMEDIATE failed on NVIDIA MCP5x controllers when system suspend

[Aaron Lu <aaron.lu@intel.com>]

On 03/13/2013 01:16 PM, Simeon Bird wrote:
>> And also full dmesg for the failed case please, thanks.
> 
> Done (but it is the same boot, so includes the successful case as a subset)
> 
OK, thanks.

So here is the relationship:
  0000:00:05.0			0000:00:05.1		0000:00:05.2
ata1       ata2		    ata3	ata4		ata5    ata6
sd 0:0:0:0 sd 1:0:0:0	    sd 2:0:0:0  sd 3:0:0:0	no disks attched

and ata3's suspend callback is done before sd 3:0:0:0's suspend callback
is finished, and as Robert analyzed, the port suspend callback masked
all isr for the controller, which makes the STANDBY IMMEDIATE command
for sd 3:0:0:0 timedout. The same is true for sd 0:0:0:0, where ata2 is
done before its suspend callback.

So I think Robert's analysis is right, there is a bug in
nv_swncq_port_suspend.

Thanks,
Aaron