Re: [RFC/PATCH] Deferred disk spinup during system resume

[Jeff Garzik <jgarzik@pobox.com>]

On 01/12/2011 06:21 AM, Tejun Heo wrote:
> Hello, Maksim, Jeff.
>
> On Tue, Jan 11, 2011 at 05:24:17PM -0800, maksim.rayskiy@gmail.com wrote:
>> @@ -4978,6 +4978,13 @@ void ata_qc_issue(struct ata_queued_cmd *qc)
>>   	struct ata_link *link = qc->dev->link;
>>   	u8 prot = qc->tf.protocol;
>>
>> +	if (unlikely(qc->flags&  ATA_QCFLAG_VERIFY)) {
>> +		ata_port_schedule_eh(ap);
>> +		qc->scsidone(qc->scsicmd);
>> +		ata_qc_free(qc);
>> +		return;
>
> I still prefer the original patch where EH is scheduled from the
> translation layer.  This seems unnecessarily intrusive to me.  As I
> wrote before, it's not like we have a clean translation anyway and I
> think this better fits as impedance matching code in the translation
> layer anyway.  But, that said, this is a rather ugly piece of code
> which is necessary just to work around the fact that we live under
> scsi, so as long as it works, one way or the other probably doesn't
> matter all that much (the reason why I prefer the previous one, as it
> doens't try to be pretty and just gets it done), so it's Jeff's call.

The previous patch breaks READ VERIFY translation for all cases -except- 
this one.

The bottom line is that this patch simply wants to trigger an ATA 
command, and return immediately, discarding the command results.  I'm 
not even sure a "run this command in background, and discard results" 
facility requires the EH.

	Jeff