Re: [PATCH linux dev-4.17 1/7] clk: Aspeed: Modify clk-aspeed.c driver probe sequence

[Ryan Chen <ryanchen.aspeed@gmail.com>]

On Wed, Jul 18, 2018 at 04:07:14PM +0930, Joel Stanley wrote:
> On 17 July 2018 at 19:15, Benjamin Herrenschmidt
> <benh@kernel.crashing.org> wrote:
> > On Tue, 2018-07-17 at 14:04 +0800, Ryan Chen wrote:
> >> On Wed, Jul 11, 2018 at 03:47:56PM +1000, Benjamin Herrenschmidt wrote:
> >> > On Wed, 2018-07-11 at 13:17 +0800, Ryan Chen wrote:
> >> > > In Aspeed's SoC, all IP clk gating and pll parameter is in scu
> >> > > controller, before IP driver probe, scu driver need prepare for it.
> >> > > So buildin_platform_driver to core_initcall.
> >> > >
> >> > > Signed-off-by: Ryan Chen <ryanchen.aspeed@gmail.com>
> >> > > ---
> >> > >  drivers/clk/clk-aspeed.c | 7 ++++++-
> >> > >  1 file changed, 6 insertions(+), 1 deletion(-)
> >> > >
> >> > > diff --git a/drivers/clk/clk-aspeed.c b/drivers/clk/clk-aspeed.c
> >> > > index 8796b8a..9e55743 100644
> >> > > --- a/drivers/clk/clk-aspeed.c
> >> > > +++ b/drivers/clk/clk-aspeed.c
> >> > > @@ -573,7 +573,12 @@ static struct platform_driver aspeed_clk_driver = {
> >> > >           .suppress_bind_attrs = true,
> >> > >   },
> >> > >  };
> >> > > -builtin_platform_driver(aspeed_clk_driver);
> >> > > +
> >> > > +static int __init aspeed_clk_init(void)
> >> > > +{
> >> > > + return platform_driver_register(&aspeed_clk_driver);
> >> > > +}
> >> > > +core_initcall(aspeed_clk_init);
> >> > >
> >> > >  static void __init aspeed_ast2400_cc(struct regmap *map)
> >> > >  {
> >> >
> >> > It's generally considered dangerous to register drivers at core
> >> > initcall time.
> >>
> >> Understand.
> >>
> >> But if interrupt controller have clk gating.
> >> the scu driver should be eraly than irq chip driver probe.
> >> Is this point is reasonable?
> >
> > I'm not sure I understand what you are trying to solve other than
> > making sure the clock driver is loaded before everything else. Joel, do
> > we have a way to ensure that ? I noticed all other clock drivers use
> > that macro to be initialized at of_clk_init, any reason we don't ?
> >>
> >> >
> >> > Any reason we don't use the generic clock driver registration mechanism
> >> > that runs at of_clk_init() time ?
> >>
> >> I will use "if (gate->reset_idx == aspeed_resets[ASPEED_RESET_SDHCI])",
> >> is it suitable ?
> >
> > I'm on holiday, I don't have the code at hand to check. Joel ? What do
> > you reckon ?
> 
> We don't need to do this. I mentioned this in my review; we have the
> reset added to the clk driver already. With this, reset can be
> released via the normal reset controller call in the sdhci driver.

In my new patch will use following in aspeed_clk_enable function
.....
	- Put in reset
	- enable sd clk
	/* sd ext clk */
	if (gate->reset_idx == aspeed_resets[ASPEED_RESET_SDHCI]) {
		regmap_update_bits(gate->map, ASPEED_CLK_SELECTION, 
				ASPEED_SDIO_CLK_EN, ASPEED_SDIO_CLK_EN);
	}
	- Out of reset
.....
Do you think is it suitable, Joel?