[PATCH] sh: mach-ecovec24: Enable SPU2 clock when boot

[PATCH] sh: mach-ecovec24: Enable SPU2 clock when boot

[Kuninori Morimoto]

Signed-off-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
---
 arch/sh/boards/mach-ecovec24/setup.c |    7 +++++++
 1 files changed, 7 insertions(+), 0 deletions(-)

diff --git a/arch/sh/boards/mach-ecovec24/setup.c b/arch/sh/boards/mach-ecovec24/setup.c
index 2274985..d5d36fe 100644
--- a/arch/sh/boards/mach-ecovec24/setup.c
+++ b/arch/sh/boards/mach-ecovec24/setup.c
@@ -535,6 +535,8 @@ extern char ecovec24_sdram_leave_end;
 
 static int __init arch_setup(void)
 {
+	struct clk *clk;
+
 	/* register board specific self-refresh code */
 	sh_mobile_register_self_refresh(SUSP_SH_STANDBY | SUSP_SH_SF,
 					&ecovec24_sdram_enter_start,
@@ -780,6 +782,11 @@ static int __init arch_setup(void)
 	i2c_register_board_info(1, i2c1_devices,
 				ARRAY_SIZE(i2c1_devices));
 
+	/* enable SPU2 */
+	clk = clk_get(NULL, "spu_clk");
+	clk_enable(clk); /* start SPU2 */
+	clk_put(clk);
+
 	return platform_add_devices(ecovec_devices,
 				    ARRAY_SIZE(ecovec_devices));
 }
-- 
1.6.3.3

Re: [PATCH] sh: mach-ecovec24: Enable SPU2 clock when boot

[Paul Mundt]

On Wed, Nov 04, 2009 at 07:34:43PM +0900, Kuninori Morimoto wrote:
> 
> Signed-off-by: Kuninori Morimoto <morimoto.kuninori@renesas.com>
> ---
>  arch/sh/boards/mach-ecovec24/setup.c |    7 +++++++
>  1 files changed, 7 insertions(+), 0 deletions(-)
> 
> diff --git a/arch/sh/boards/mach-ecovec24/setup.c b/arch/sh/boards/mach-ecovec24/setup.c
> index 2274985..d5d36fe 100644
> --- a/arch/sh/boards/mach-ecovec24/setup.c
> +++ b/arch/sh/boards/mach-ecovec24/setup.c
> @@ -535,6 +535,8 @@ extern char ecovec24_sdram_leave_end;
>  
>  static int __init arch_setup(void)
>  {
> +	struct clk *clk;
> +
>  	/* register board specific self-refresh code */
>  	sh_mobile_register_self_refresh(SUSP_SH_STANDBY | SUSP_SH_SF,
>  					&ecovec24_sdram_enter_start,
> @@ -780,6 +782,11 @@ static int __init arch_setup(void)
>  	i2c_register_board_info(1, i2c1_devices,
>  				ARRAY_SIZE(i2c1_devices));
>  
> +	/* enable SPU2 */
> +	clk = clk_get(NULL, "spu_clk");
> +	clk_enable(clk); /* start SPU2 */
> +	clk_put(clk);
> +
>  	return platform_add_devices(ecovec_devices,
>  				    ARRAY_SIZE(ecovec_devices));
>  }

This is just doing an enable on init, so it's better to just flag this in
the CPU code. ie:

diff --git a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
index dfe9192..9db7438 100644
--- a/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/clock-sh7724.c
@@ -152,7 +152,7 @@ struct clk div6_clks[] = {
 	SH_CLK_DIV6("fsia_clk", &div3_clk, FCLKACR, 0),
 	SH_CLK_DIV6("fsib_clk", &div3_clk, FCLKBCR, 0),
 	SH_CLK_DIV6("irda_clk", &div3_clk, IRDACLKCR, 0),
-	SH_CLK_DIV6("spu_clk", &div3_clk, SPUCLKCR, 0),
+	SH_CLK_DIV6("spu_clk", &div3_clk, SPUCLKCR, CLK_ENABLE_ON_INIT),
 };
 
 #define R_CLK (&r_clk)

Although this is going to be a nuisance for power management.

Re: [PATCH] sh: mach-ecovec24: Enable SPU2 clock when boot

[Kuninori Morimoto]

Dear Paul and Magnus

> This is just doing an enable on init, so it's better to just flag this in
> the CPU code. ie:
(snip)
> Although this is going to be a nuisance for power management.

I'm 100% not sure about power management.
Is spu clock operation unnecessary if there is following ?

+	.archdata = {
+		.hwblk_id = HWBLK_SPU,
+	},

If so, please throw away this patch.

Best regards
--
Kuninori Morimoto
 

Re: [PATCH] sh: mach-ecovec24: Enable SPU2 clock when boot

[Magnus Damm]

Hey Morimoto-san,

On Thu, Nov 5, 2009 at 5:01 PM, Kuninori Morimoto
<morimoto.kuninori@renesas.com> wrote:
>> This is just doing an enable on init, so it's better to just flag this in
>> the CPU code. ie:
> (snip)
>> Although this is going to be a nuisance for power management.
>
> I'm 100% not sure about power management.
> Is spu clock operation unnecessary if there is following ?
>
> +       .archdata = {
> +               .hwblk_id = HWBLK_SPU,
> +       },
>
> If so, please throw away this patch.

No, your code is correct. And Paul has a nicer solution for the spu_clk.

At this point you need to enable the parent clock and use hwblk_id for
module stop bits.

Cheers,

/ magnus