Re: [PATCH 2/7] ARM: shmobile: r8a7740: add basic pm domain support

[Paul Mundt <lethal@linux-sh.org>]

On Wed, Jul 04, 2012 at 07:48:41PM +0900, Magnus Damm wrote:
> On Wed, Jul 4, 2012 at 7:45 PM, Paul Mundt <lethal@linux-sh.org> wrote:
> > On Tue, Jul 03, 2012 at 08:55:25PM -0700, Kuninori Morimoto wrote:
> >> +void r8a7740_init_pm_domain(struct r8a7740_pm_domain *r8a7740_pd)
> >> +{
> >> +     struct generic_pm_domain *genpd = &r8a7740_pd->genpd;
> >> +     struct dev_power_governor *gov = r8a7740_pd->gov;
> >> +
> >> +     pm_genpd_init(genpd, gov ? : &simple_qos_governor, false);
> >> +     genpd->dev_ops.stop             = r8a7740_stop_dev;
> >> +     genpd->dev_ops.start            = r8a7740_start_dev;
> >> +     genpd->dev_ops.active_wakeup    = pd_active_wakeup;
> >> +     genpd->dev_irq_safe             = true;
> >> +     genpd->power_off                = pd_power_down;
> >> +     genpd->power_on                 = pd_power_up;
> >> +     __pd_power_up(r8a7740_pd, false);
> >> +}
> >> +
> > This pd_ naming really needs to go, it clashes too much with the generic
> > code and it is far more confusing than it needs to be. If this is an
> > r8a7740 specific routine then prefix with r8a7740 accordingly.
> 
> I agree that we want to use soc-specific prefixes for soc-specific code.
> 
> In this case it's not really soc-specific though, so code here needs
> to be reworked for sharing between sh7372 and r8a7740 (and sh73a0 too
> perhaps).
> 
How exactly is the fact it's used on more than one SoC less of a
SoC-specific property? This is not generic infrastructure, and it is way
too close to genpd naming conventions to follow clearly. The discrepancy
between __ prefixing between this and the existing in-tree code already
shows what a joke this is.