[PATCH] i2c-s3c2410: move resume state flag to end of resume

[PATCH] i2c-s3c2410: move resume state flag to end of resume

[Huisung Kang]

From: Kisoo Yu <ksoo.yu@samsung.com>

need to set flag after finishing its work

Signed-off-by: Kisoo Yu <ksoo.yu@samsung.com>
Signed-off-by: Huisung Kang <hs1218.kang@samsung.com>
---
 drivers/i2c/busses/i2c-s3c2410.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/drivers/i2c/busses/i2c-s3c2410.c b/drivers/i2c/busses/i2c-s3c2410.c
index 4c17180..7d6d2b7 100644
--- a/drivers/i2c/busses/i2c-s3c2410.c
+++ b/drivers/i2c/busses/i2c-s3c2410.c
@@ -1082,10 +1082,10 @@ static int s3c24xx_i2c_resume(struct device *dev)
 	struct platform_device *pdev = to_platform_device(dev);
 	struct s3c24xx_i2c *i2c = platform_get_drvdata(pdev);
 
-	i2c->suspended = 0;
 	clk_enable(i2c->clk);
 	s3c24xx_i2c_init(i2c);
 	clk_disable(i2c->clk);
+	i2c->suspended = 0;
 
 	return 0;
 }
-- 
1.7.1

Re: [PATCH] i2c-s3c2410: move resume state flag to end of resume

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 398 bytes --]

On Sat, Feb 11, 2012 at 12:22:21PM +0900, Huisung Kang wrote:
> From: Kisoo Yu <ksoo.yu@samsung.com>

> need to set flag after finishing its work

> -	i2c->suspended = 0;
>  	clk_enable(i2c->clk);
>  	s3c24xx_i2c_init(i2c);
>  	clk_disable(i2c->clk);
> +	i2c->suspended = 0;

Why is this needed - given that there's no locking involved anywhere
here does this mean that we've got a race somewhere?

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] i2c-s3c2410: move resume state flag to end of resume

[Ben Dooks]

On Sat, Feb 11, 2012 at 12:22:21PM +0900, Huisung Kang wrote:
> From: Kisoo Yu <ksoo.yu-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>
> 
> need to set flag after finishing its work
> 
> Signed-off-by: Kisoo Yu <ksoo.yu-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>
> Signed-off-by: Huisung Kang <hs1218.kang-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>
> ---
>  drivers/i2c/busses/i2c-s3c2410.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/drivers/i2c/busses/i2c-s3c2410.c b/drivers/i2c/busses/i2c-s3c2410.c
> index 4c17180..7d6d2b7 100644
> --- a/drivers/i2c/busses/i2c-s3c2410.c
> +++ b/drivers/i2c/busses/i2c-s3c2410.c
> @@ -1082,10 +1082,10 @@ static int s3c24xx_i2c_resume(struct device *dev)
>  	struct platform_device *pdev = to_platform_device(dev);
>  	struct s3c24xx_i2c *i2c = platform_get_drvdata(pdev);
>  
> -	i2c->suspended = 0;
>  	clk_enable(i2c->clk);
>  	s3c24xx_i2c_init(i2c);
>  	clk_disable(i2c->clk);
> +	i2c->suspended = 0;

Yes, technically we should not mark ourselves as un-suspended until
the peripheral has been properly restarted. I will have a look at
Mark's comments before this is applied.

-- 
Ben

Re: [PATCH] i2c-s3c2410: move resume state flag to end of resume

[Ben Dooks]

On Sun, Feb 12, 2012 at 01:12:31PM +0000, Mark Brown wrote:
> On Sat, Feb 11, 2012 at 12:22:21PM +0900, Huisung Kang wrote:
> > From: Kisoo Yu <ksoo.yu-Sze3O3UU22JBDgjK7y7TUQ@public.gmane.org>
> 
> > need to set flag after finishing its work
> 
> > -	i2c->suspended = 0;
> >  	clk_enable(i2c->clk);
> >  	s3c24xx_i2c_init(i2c);
> >  	clk_disable(i2c->clk);
> > +	i2c->suspended = 0;
> 
> Why is this needed - given that there's no locking involved anywhere
> here does this mean that we've got a race somewhere?

It is in the wrong order, the lack of locking is a bit of a problem
but then again so is the whole suspend/resume framework at the
moment. I'm inclined to apply this to the fixes branch.

-- 
Ben