[PATCH v2 0/2] pwm fixes for 3.8

[PATCH v2 0/2] pwm fixes for 3.8

[Tony Prisk]

Hi Theirry,

As discussed, here is the patch to correct the issue created by enabling
disabling the pwm clock during pwm_enable/_disable. A second patch is
including which fixes a build error introduced in the same patch.

These should be applied on top of the previous patch.

v2: Add error checking for clk_enable()

Tony Prisk (2):
  Fix build error in pwm-vt8500.c
  pwm: vt8500: Ensure pwm clock is enabled during pwm_config

 drivers/pwm/pwm-vt8500.c |   14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

-- 
1.7.9.5

[PATCH v2 1/2] Fix build error in pwm-vt8500.c

[Tony Prisk]

A missing '{' causes a build error in pwm-vt8500.c

Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
---
 drivers/pwm/pwm-vt8500.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/pwm/pwm-vt8500.c b/drivers/pwm/pwm-vt8500.c
index 970b0c6..806f72c 100644
--- a/drivers/pwm/pwm-vt8500.c
+++ b/drivers/pwm/pwm-vt8500.c
@@ -100,7 +100,7 @@ static int vt8500_pwm_enable(struct pwm_chip *chip, struct pwm_device *pwm)
 	struct vt8500_chip *vt8500 = to_vt8500_chip(chip);
 
 	err = clk_enable(vt8500->clk);
-	if (err < 0)
+	if (err < 0) {
 		dev_err(chip->dev, "failed to enable clock\n");
 		return err;
 	};
-- 
1.7.9.5

[PATCH v2 1/2] Fix build error in pwm-vt8500.c

[Sergei Shtylyov]

Hello.

On 19-11-2012 11:30, Tony Prisk wrote:

> A missing '{' causes a build error in pwm-vt8500.c

> Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
> ---
>   drivers/pwm/pwm-vt8500.c |    2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)

> diff --git a/drivers/pwm/pwm-vt8500.c b/drivers/pwm/pwm-vt8500.c
> index 970b0c6..806f72c 100644
> --- a/drivers/pwm/pwm-vt8500.c
> +++ b/drivers/pwm/pwm-vt8500.c
> @@ -100,7 +100,7 @@ static int vt8500_pwm_enable(struct pwm_chip *chip, struct pwm_device *pwm)
>   	struct vt8500_chip *vt8500 = to_vt8500_chip(chip);
>
>   	err = clk_enable(vt8500->clk);
> -	if (err < 0)
> +	if (err < 0) {
>   		dev_err(chip->dev, "failed to enable clock\n");
>   		return err;
>   	};

   Could as well remove the unneeded semicolon here.

WBR, Sergei

[PATCH v2 2/2] pwm: vt8500: Ensure pwm clock is enabled during pwm_config

[Tony Prisk]

This patch corrects a bug reported by Peter Vasil.

When all PWMs are disabled, PWM module may be disabled during
calls to pwm_config. This patch enables/disables the clock in
pwm_config to ensure the module is active before register read/
writes.

Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
Tested-by: Peter Vasil <petervasil@gmail.com>
---
 drivers/pwm/pwm-vt8500.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/pwm/pwm-vt8500.c b/drivers/pwm/pwm-vt8500.c
index 806f72c..9fdb70d 100644
--- a/drivers/pwm/pwm-vt8500.c
+++ b/drivers/pwm/pwm-vt8500.c
@@ -62,6 +62,13 @@ static int vt8500_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
 	struct vt8500_chip *vt8500 = to_vt8500_chip(chip);
 	unsigned long long c;
 	unsigned long period_cycles, prescale, pv, dc;
+	int err;
+
+	err = clk_enable(vt8500->clk);
+	if (err < 0) {
+		dev_err(chip->dev, "failed to enable clock in pwm_config()\n");
+		return err;
+	};
 
 	c = clk_get_rate(vt8500->clk);
 	c = c * period_ns;
@@ -75,8 +82,10 @@ static int vt8500_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
 	if (pv > 4095)
 		pv = 4095;
 
-	if (prescale > 1023)
+	if (prescale > 1023) {
+		clk_disable(vt8500->clk);
 		return -EINVAL;
+	}
 
 	c = (unsigned long long)pv * duty_ns;
 	do_div(c, period_ns);
@@ -91,6 +100,7 @@ static int vt8500_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
 	pwm_busy_wait(vt8500->base + 0x40 + pwm->hwpwm, (1 << 3));
 	writel(dc, vt8500->base + 0xc + (pwm->hwpwm << 4));
 
+	clk_disable(vt8500->clk);
 	return 0;
 }
 
-- 
1.7.9.5

[PATCH v2 2/2] pwm: vt8500: Ensure pwm clock is enabled during pwm_config

[Thierry Reding]

On Mon, Nov 19, 2012 at 08:30:54PM +1300, Tony Prisk wrote:
> This patch corrects a bug reported by Peter Vasil.
> 
> When all PWMs are disabled, PWM module may be disabled during
> calls to pwm_config. This patch enables/disables the clock in
> pwm_config to ensure the module is active before register read/
> writes.
> 
> Signed-off-by: Tony Prisk <linux@prisktech.co.nz>
> Tested-by: Peter Vasil <petervasil@gmail.com>
> ---
>  drivers/pwm/pwm-vt8500.c |   12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/pwm/pwm-vt8500.c b/drivers/pwm/pwm-vt8500.c
> index 806f72c..9fdb70d 100644
> --- a/drivers/pwm/pwm-vt8500.c
> +++ b/drivers/pwm/pwm-vt8500.c
> @@ -62,6 +62,13 @@ static int vt8500_pwm_config(struct pwm_chip *chip, struct pwm_device *pwm,
>  	struct vt8500_chip *vt8500 = to_vt8500_chip(chip);
>  	unsigned long long c;
>  	unsigned long period_cycles, prescale, pv, dc;
> +	int err;
> +
> +	err = clk_enable(vt8500->clk);
> +	if (err < 0) {
> +		dev_err(chip->dev, "failed to enable clock in pwm_config()\n");
> +		return err;
> +	};

I'm not entirely happy about this error message. What bugs me about it is
that the clock doesn't fail to enable in pwm_config() but rather in
vt8500_pwm_config(), so it might be slightly misleading. I'm not sure if
the failure location is needed at all. "failed to enable clock" sounds
useful enough. Furthermore I think that if it fails here it will likely
also fail elsewhere.

If you really absolutely want to specify where clk_enable() failed, then
I'd be slightly happier with "%s(): failed to enable clock", __func__.

Thierry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: not available
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20121119/4bb0c7a6/attachment.sig>