[1/2] clk: change clk_ops' ->round_rate() prototype

Commit Message

Heiko Stübner April 19, 2015, 12:13 p.m. UTC

Hi Boris,

Am Freitag, 17. April 2015, 09:29:28 schrieb Boris Brezillon:
> Clock rates are stored in an unsigned long field, but ->round_rate()
> (which returns a rounded rate from a requested one) returns a long
> value (errors are reported using negative error codes), which can lead
> to long overflow if the clock rate exceed 2Ghz.
> 
> Change ->round_rate() prototype to return 0 or an error code, and pass the
> requested rate as a pointer so that it can be adjusted depending on
> hardware capabilities.
> 
> Signed-off-by: Boris Brezillon <[e-mail]>
> ---

On a rk3288-veyron-pinky with the fix described below:
Tested-by: Heiko Stuebner <[e-mail]>


> diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> index fa5a00e..1462ddc 100644
> --- a/drivers/clk/clk.c
> +++ b/drivers/clk/clk.c
> @@ -1640,8 +1643,10 @@ static struct clk_core *clk_calc_new_rates(struct
> clk_core *clk, &parent_hw);
>  		parent = parent_hw ? parent_hw->core : NULL;
>  	} else if (clk->ops->round_rate) {
> -		new_rate = clk->ops->round_rate(clk->hw, rate,
> -						&best_parent_rate);
> +		if (clk->ops->round_rate(clk->hw, &new_rate,
> +					 &best_parent_rate))
> +			return NULL;
> +
>  		if (new_rate < min_rate || new_rate > max_rate)
>  			return NULL;
>  	} else if (!parent || !(clk->flags & CLK_SET_RATE_PARENT)) {

This is using new_rate uninitialized when calling into the round_rate
callback. Which in turn pushed my PLLs up to 2.2GHz :-)

I guess you'll need something like the following:





> diff --git a/drivers/clk/rockchip/clk-pll.c b/drivers/clk/rockchip/clk-pll.c
> index f8d3baf..bd408ef 100644
> --- a/drivers/clk/rockchip/clk-pll.c
> +++ b/drivers/clk/rockchip/clk-pll.c
> @@ -63,8 +63,8 @@ static const struct rockchip_pll_rate_table
> *rockchip_get_pll_settings( return NULL;
>  }
> 
> -static long rockchip_pll_round_rate(struct clk_hw *hw,
> -			    unsigned long drate, unsigned long *prate)
> +static int rockchip_pll_round_rate(struct clk_hw *hw,
> +			    unsigned long *drate, unsigned long *prate)
>  {
>  	struct rockchip_clk_pll *pll = to_rockchip_clk_pll(hw);
>  	const struct rockchip_pll_rate_table *rate_table = pll->rate_table;
> @@ -72,12 +72,15 @@ static long rockchip_pll_round_rate(struct clk_hw *hw,
> 
>  	/* Assumming rate_table is in descending order */
>  	for (i = 0; i < pll->rate_count; i++) {
> -		if (drate >= rate_table[i].rate)
> -			return rate_table[i].rate;
> +		if (*drate >= rate_table[i].rate) {
> +			*drate = rate_table[i].rate;
> +			return 0;
> +		}
>  	}
> 
>  	/* return minimum supported value */
> -	return rate_table[i - 1].rate;
> +	*drate = rate_table[i - 1].rate;
> +	return 0;
>  }
> 
>  /*

The rockchip-part:
Reviewed-by: Heiko Stuebner <[e-mail]>


And as I've stumbled onto this recently too, the clock-maintainership has
expanded to Stephen Boyd and [e-mail] .


Heiko
--
To unsubscribe from this list: send the line "unsubscribe linux-arm-msm" in
the body of a message to [e-mail]
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Comments

Boris BREZILLON April 19, 2015, 3:30 p.m. UTC | #1

Hi Heiko,

On Sun, 19 Apr 2015 14:13:04 +0200
Heiko Stübner <[e-mail]> wrote:

> Hi Boris,
> 
> Am Freitag, 17. April 2015, 09:29:28 schrieb Boris Brezillon:
> > Clock rates are stored in an unsigned long field, but ->round_rate()
> > (which returns a rounded rate from a requested one) returns a long
> > value (errors are reported using negative error codes), which can lead
> > to long overflow if the clock rate exceed 2Ghz.
> > 
> > Change ->round_rate() prototype to return 0 or an error code, and pass the
> > requested rate as a pointer so that it can be adjusted depending on
> > hardware capabilities.
> > 
> > Signed-off-by: Boris Brezillon <[e-mail]>
> > ---
> 
> On a rk3288-veyron-pinky with the fix described below:
> Tested-by: Heiko Stuebner <[e-mail]>
> 
> 
> > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
> > index fa5a00e..1462ddc 100644
> > --- a/drivers/clk/clk.c
> > +++ b/drivers/clk/clk.c
> > @@ -1640,8 +1643,10 @@ static struct clk_core *clk_calc_new_rates(struct
> > clk_core *clk, &parent_hw);
> >  		parent = parent_hw ? parent_hw->core : NULL;
> >  	} else if (clk->ops->round_rate) {
> > -		new_rate = clk->ops->round_rate(clk->hw, rate,
> > -						&best_parent_rate);
> > +		if (clk->ops->round_rate(clk->hw, &new_rate,
> > +					 &best_parent_rate))
> > +			return NULL;
> > +
> >  		if (new_rate < min_rate || new_rate > max_rate)
> >  			return NULL;
> >  	} else if (!parent || !(clk->flags & CLK_SET_RATE_PARENT)) {
> 
> This is using new_rate uninitialized when calling into the round_rate
> callback. Which in turn pushed my PLLs up to 2.2GHz :-)

Indeed, thanks for the fix.

[...]

> 
> 
> And as I've stumbled onto this recently too, the clock-maintainership has
> expanded to Stephen Boyd and [e-mail] .

Noted. I'll add Stephen and the new linux-clk ML in the recipient list
next time.

Best Regards,

Boris

Patch

diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c
index db4e4b2..afc7733 100644
--- a/drivers/clk/clk.c
+++ b/drivers/clk/clk.c
@@ -1605,6 +1605,7 @@  static struct clk_core *clk_calc_new_rates(struct clk_core *clk,
 						    &parent_hw);
 		parent = parent_hw ? parent_hw->core : NULL;
 	} else if (clk->ops->round_rate) {
+		new_rate = rate;
 		if (clk->ops->round_rate(clk->hw, &new_rate,
 					 &best_parent_rate))
 			return NULL;