Re: [PATCH] libata: skip EH autopsy and recovery during suspend

["Rafael J. Wysocki" <rjw@sisk.pl>]

On Thursday, September 02, 2010, Tejun Heo wrote:
> Hello,
> 
> On 09/02/2010 10:16 PM, Rafael J. Wysocki wrote:
> > Putting the issue at hand aside, I'm not really sure if using SCSI EH for
> > suspending the controller is a good idea.  It seems overly complicated and
> > it doesn't match the new PCI suspend model with separate ->suspend(),
> > ->freeze() and ->poweroff() callbacks.  Moreover, the passing of pm_message_t
> > back and forth doesn't make things clear either.
> > 
> > Would it be possible to rework this thing entirely at one point?
> 
> Well, I think I would need more than that to rework the whole thing.
> There are a lot of benefits in sharing the same path between probing /
> error handling and suspend/resuming.  ATA has a lot of quirks which
> have to be dealt with and it will be very fragile to scatter handling
> logics over multiple separate paths.  We definitely can try to make
> the plumbing from power management easier to follow.

That would be very nice.  In particular, I'd like to get rid of the
pm_message_t thing if possible.  And I'd like to avoid putting the
controller into D3 before creating hibernation image. :-)

Thanks,
Rafael