Re: [PATCH] ahci: platform support for suspend/resume

Re: [PATCH] ahci: platform support for suspend/resume

[Brian Norris]

+linux-pm mailing list

Hi Felipe,

On Wed, Nov 16, 2011 at 12:34 PM, Felipe Balbi <balbi@ti.com> wrote:
> On Wed, Nov 16, 2011 at 12:27:59PM -0800, Brian Norris wrote:
>> +     /* Does platform support suspend/resume? */
>> +     if (!pdata->suspend || !pdata->resume)
>> +             return -EINVAL;
>
> should you really prevent suspend is platform doesn't provide
> ->suspend() ?

No, we shouldn't prevent suspend entirely; that would just be my
accidental side effect.

> I mean, we could some platform where all clocks are autogated and
> there's no need for platform to do anything. Would it be better to only
> call pdata->suspend if it exists but still let the AHCI-specific part go
> through ?

Perhaps. But would this have any unintended effects on systems that
don't implement AHCI power management at all? My intention was just to
provide an 'opt-in' interface for platform developers. I probably got
this wrong.

Brian

Re: [PATCH] ahci: platform support for suspend/resume

[Felipe Balbi]

[-- Attachment #1.1: Type: text/plain, Size: 1218 bytes --]

On Thu, Nov 17, 2011 at 04:31:32PM -0800, Brian Norris wrote:
> +linux-pm mailing list
> 
> Hi Felipe,
> 
> On Wed, Nov 16, 2011 at 12:34 PM, Felipe Balbi <balbi@ti.com> wrote:
> > On Wed, Nov 16, 2011 at 12:27:59PM -0800, Brian Norris wrote:
> >> +     /* Does platform support suspend/resume? */
> >> +     if (!pdata->suspend || !pdata->resume)
> >> +             return -EINVAL;
> >
> > should you really prevent suspend is platform doesn't provide
> > ->suspend() ?
> 
> No, we shouldn't prevent suspend entirely; that would just be my
> accidental side effect.
> 
> > I mean, we could some platform where all clocks are autogated and
> > there's no need for platform to do anything. Would it be better to only
> > call pdata->suspend if it exists but still let the AHCI-specific part go
> > through ?
> 
> Perhaps. But would this have any unintended effects on systems that
> don't implement AHCI power management at all? My intention was just to
> provide an 'opt-in' interface for platform developers. I probably got
> this wrong.

I guess you could add a flag for that, but until we know that the AHCI
powermanagement is broken somewhere should we really care ? :-)

-- 
balbi

[-- Attachment #1.2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

[-- Attachment #2: Type: text/plain, Size: 0 bytes --]