[PATCH] clk: rockchip: convert rk3036 pll type to use internal lock status

[PATCH] clk: rockchip: convert rk3036 pll type to use internal lock status

[Heiko Stuebner]

From: Heiko Stuebner <heiko.stuebner@theobroma-systems.com>

The rk3036 pll type exposes its lock status in both its pllcon registers
as well as the General Register Files. To remove one dependency convert
it to the "internal" lock status, similar to how rk3399 handles it.

Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com>
---
 drivers/clk/rockchip/clk-pll.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/rockchip/clk-pll.c b/drivers/clk/rockchip/clk-pll.c
index 198417d56300..37378ded0993 100644
--- a/drivers/clk/rockchip/clk-pll.c
+++ b/drivers/clk/rockchip/clk-pll.c
@@ -118,12 +118,30 @@ static int rockchip_pll_wait_lock(struct rockchip_clk_pll *pll)
 #define RK3036_PLLCON1_REFDIV_SHIFT		0
 #define RK3036_PLLCON1_POSTDIV2_MASK		0x7
 #define RK3036_PLLCON1_POSTDIV2_SHIFT		6
+#define RK3036_PLLCON1_LOCK_STATUS		BIT(10)
 #define RK3036_PLLCON1_DSMPD_MASK		0x1
 #define RK3036_PLLCON1_DSMPD_SHIFT		12
+#define RK3036_PLLCON1_PWRDOWN			BIT(13)
 #define RK3036_PLLCON2_FRAC_MASK		0xffffff
 #define RK3036_PLLCON2_FRAC_SHIFT		0
 
-#define RK3036_PLLCON1_PWRDOWN			(1 << 13)
+static int rockchip_rk3036_pll_wait_lock(struct rockchip_clk_pll *pll)
+{
+	u32 pllcon;
+	int delay = 24000000;
+
+	/* poll check the lock status in rk3399 xPLLCON2 */
+	while (delay > 0) {
+		pllcon = readl_relaxed(pll->reg_base + RK3036_PLLCON(1));
+		if (pllcon & RK3036_PLLCON1_LOCK_STATUS)
+			return 0;
+
+		delay--;
+	}
+
+	pr_err("%s: timeout waiting for pll to lock\n", __func__);
+	return -ETIMEDOUT;
+}
 
 static void rockchip_rk3036_pll_get_params(struct rockchip_clk_pll *pll,
 					struct rockchip_pll_rate_table *rate)
@@ -221,7 +239,7 @@ static int rockchip_rk3036_pll_set_params(struct rockchip_clk_pll *pll,
 	writel_relaxed(pllcon, pll->reg_base + RK3036_PLLCON(2));
 
 	/* wait for the pll to lock */
-	ret = rockchip_pll_wait_lock(pll);
+	ret = rockchip_rk3036_pll_wait_lock(pll);
 	if (ret) {
 		pr_warn("%s: pll update unsuccessful, trying to restore old params\n",
 			__func__);
@@ -260,7 +278,7 @@ static int rockchip_rk3036_pll_enable(struct clk_hw *hw)
 
 	writel(HIWORD_UPDATE(0, RK3036_PLLCON1_PWRDOWN, 0),
 	       pll->reg_base + RK3036_PLLCON(1));
-	rockchip_pll_wait_lock(pll);
+	rockchip_rk3036_pll_wait_lock(pll);
 
 	return 0;
 }
-- 
2.24.1

Re: [PATCH] clk: rockchip: convert rk3036 pll type to use internal lock status

[Stephen Boyd]

Quoting Heiko Stuebner (2020-01-13 07:26:56)
> diff --git a/drivers/clk/rockchip/clk-pll.c b/drivers/clk/rockchip/clk-pll.c
> index 198417d56300..37378ded0993 100644
> --- a/drivers/clk/rockchip/clk-pll.c
> +++ b/drivers/clk/rockchip/clk-pll.c
> @@ -118,12 +118,30 @@ static int rockchip_pll_wait_lock(struct rockchip_clk_pll *pll)
>  #define RK3036_PLLCON1_REFDIV_SHIFT            0
>  #define RK3036_PLLCON1_POSTDIV2_MASK           0x7
>  #define RK3036_PLLCON1_POSTDIV2_SHIFT          6
> +#define RK3036_PLLCON1_LOCK_STATUS             BIT(10)
>  #define RK3036_PLLCON1_DSMPD_MASK              0x1
>  #define RK3036_PLLCON1_DSMPD_SHIFT             12
> +#define RK3036_PLLCON1_PWRDOWN                 BIT(13)
>  #define RK3036_PLLCON2_FRAC_MASK               0xffffff
>  #define RK3036_PLLCON2_FRAC_SHIFT              0
>  
> -#define RK3036_PLLCON1_PWRDOWN                 (1 << 13)
> +static int rockchip_rk3036_pll_wait_lock(struct rockchip_clk_pll *pll)
> +{
> +       u32 pllcon;
> +       int delay = 24000000;
> +
> +       /* poll check the lock status in rk3399 xPLLCON2 */
> +       while (delay > 0) {
> +               pllcon = readl_relaxed(pll->reg_base + RK3036_PLLCON(1));
> +               if (pllcon & RK3036_PLLCON1_LOCK_STATUS)
> +                       return 0;
> +
> +               delay--;

There isn't any udelay here. So the timeout is just as fast as the CPU
can churn through this? Why not use an actual time? Or use the
readl_poll_timeout() APIs?

> +       }
> +
> +       pr_err("%s: timeout waiting for pll to lock\n", __func__);
> +       return -ETIMEDOUT;
> +}
>  
>  static void rockchip_rk3036_pll_get_params(struct rockchip_clk_pll *pll,
>                                         struct rockchip_pll_rate_table *rate)

Re: [PATCH] clk: rockchip: convert rk3036 pll type to use internal lock status

[Heiko Stuebner]

Am Dienstag, 14. Januar 2020, 06:05:18 CET schrieb Stephen Boyd:
> Quoting Heiko Stuebner (2020-01-13 07:26:56)
> > diff --git a/drivers/clk/rockchip/clk-pll.c b/drivers/clk/rockchip/clk-pll.c
> > index 198417d56300..37378ded0993 100644
> > --- a/drivers/clk/rockchip/clk-pll.c
> > +++ b/drivers/clk/rockchip/clk-pll.c
> > @@ -118,12 +118,30 @@ static int rockchip_pll_wait_lock(struct rockchip_clk_pll *pll)
> >  #define RK3036_PLLCON1_REFDIV_SHIFT            0
> >  #define RK3036_PLLCON1_POSTDIV2_MASK           0x7
> >  #define RK3036_PLLCON1_POSTDIV2_SHIFT          6
> > +#define RK3036_PLLCON1_LOCK_STATUS             BIT(10)
> >  #define RK3036_PLLCON1_DSMPD_MASK              0x1
> >  #define RK3036_PLLCON1_DSMPD_SHIFT             12
> > +#define RK3036_PLLCON1_PWRDOWN                 BIT(13)
> >  #define RK3036_PLLCON2_FRAC_MASK               0xffffff
> >  #define RK3036_PLLCON2_FRAC_SHIFT              0
> >  
> > -#define RK3036_PLLCON1_PWRDOWN                 (1 << 13)
> > +static int rockchip_rk3036_pll_wait_lock(struct rockchip_clk_pll *pll)
> > +{
> > +       u32 pllcon;
> > +       int delay = 24000000;
> > +
> > +       /* poll check the lock status in rk3399 xPLLCON2 */
> > +       while (delay > 0) {
> > +               pllcon = readl_relaxed(pll->reg_base + RK3036_PLLCON(1));
> > +               if (pllcon & RK3036_PLLCON1_LOCK_STATUS)
> > +                       return 0;
> > +
> > +               delay--;
> 
> There isn't any udelay here. So the timeout is just as fast as the CPU
> can churn through this? Why not use an actual time? Or use the
> readl_poll_timeout() APIs?

Done in 
http://lore.kernel.org/r/20200128100204.1318450-3-heiko@sntech.de

and to keep things similar, I did a conversion to iopoll helpers for the
other variants too.

Heiko