[PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[Thierry Reding]

From: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>

The mfd_add_devices() parameter takes a struct resource * as fifth
argument, but the nvec driver passes in a void __iomem *. The driver
gets away with it because none of the subdevices ever directly access
the registers.

While at it, use platform_get_irq() instead of platform_get_resource()
to get the device's interrupt. This makes it easier to pass in the
register region since the variable is no longer reused.

Signed-off-by: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
---
Alternatively we could simply pass NULL into mfd_add_devices(), which
might be a slightly more accurate representation of what's going on.

Marc, Greg, any preferences?

 drivers/staging/nvec/nvec.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/staging/nvec/nvec.c b/drivers/staging/nvec/nvec.c
index d32504844896..11f9e1c3447c 100644
--- a/drivers/staging/nvec/nvec.c
+++ b/drivers/staging/nvec/nvec.c
@@ -821,8 +821,8 @@ static int tegra_nvec_probe(struct platform_device *pdev)
 	if (IS_ERR(base))
 		return PTR_ERR(base);
 
-	res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
-	if (!res) {
+	nvec->irq = platform_get_irq(pdev, 0);
+	if (nvec->irq < 0) {
 		dev_err(&pdev->dev, "no irq resource?\n");
 		return -ENODEV;
 	}
@@ -840,7 +840,6 @@ static int tegra_nvec_probe(struct platform_device *pdev)
 	}
 
 	nvec->base = base;
-	nvec->irq = res->start;
 	nvec->i2c_clk = i2c_clk;
 	nvec->rx = &nvec->msg_pool[0];
 
@@ -893,7 +892,7 @@ static int tegra_nvec_probe(struct platform_device *pdev)
 	}
 
 	ret = mfd_add_devices(nvec->dev, -1, nvec_devices,
-			      ARRAY_SIZE(nvec_devices), base, 0, NULL);
+			      ARRAY_SIZE(nvec_devices), res, 0, NULL);
 	if (ret)
 		dev_err(nvec->dev, "error adding subdevices\n");
 
-- 
2.0.1

Re: [PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[Marc Dietrich]

Hi Thierry,

On Mon, 21 Jul 2014 13:52:41 +0200
Thierry Reding <thierry.reding@gmail.com> wrote:

> From: Thierry Reding <treding@nvidia.com>
> 
> The mfd_add_devices() parameter takes a struct resource * as fifth
> argument, but the nvec driver passes in a void __iomem *. The driver
> gets away with it because none of the subdevices ever directly access
> the registers.

you are right, this one looked bogus. 

> While at it, use platform_get_irq() instead of platform_get_resource()
> to get the device's interrupt. This makes it easier to pass in the
> register region since the variable is no longer reused.
> 
> Signed-off-by: Thierry Reding <treding@nvidia.com>
> ---
> Alternatively we could simply pass NULL into mfd_add_devices(), which
> might be a slightly more accurate representation of what's going on.
>
> Marc, Greg, any preferences?

I just tested with NULL as base and it seems to produce no harm. Even
if we may get rid of mfd_add_devices in the future, passing NULL is
cleaner for now and no children will ever make use of it. So please send
a new version with this change.

Otherwise, thanks for looking and cleaning up.

Marc

>  drivers/staging/nvec/nvec.c | 7 +++----
>  1 file changed, 3 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/staging/nvec/nvec.c b/drivers/staging/nvec/nvec.c
> index d32504844896..11f9e1c3447c 100644
> --- a/drivers/staging/nvec/nvec.c
> +++ b/drivers/staging/nvec/nvec.c
> @@ -821,8 +821,8 @@ static int tegra_nvec_probe(struct platform_device *pdev)
>  	if (IS_ERR(base))
>  		return PTR_ERR(base);
>  
> -	res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
> -	if (!res) {
> +	nvec->irq = platform_get_irq(pdev, 0);
> +	if (nvec->irq < 0) {
>  		dev_err(&pdev->dev, "no irq resource?\n");
>  		return -ENODEV;
>  	}
> @@ -840,7 +840,6 @@ static int tegra_nvec_probe(struct platform_device *pdev)
>  	}
>  
>  	nvec->base = base;
> -	nvec->irq = res->start;
>  	nvec->i2c_clk = i2c_clk;
>  	nvec->rx = &nvec->msg_pool[0];
>  
> @@ -893,7 +892,7 @@ static int tegra_nvec_probe(struct platform_device *pdev)
>  	}
>  
>  	ret = mfd_add_devices(nvec->dev, -1, nvec_devices,
> -			      ARRAY_SIZE(nvec_devices), base, 0, NULL);
> +			      ARRAY_SIZE(nvec_devices), res, 0, NULL);
>  	if (ret)
>  		dev_err(nvec->dev, "error adding subdevices\n");
>  
> -- 
> 2.0.1
> 


-- 
Marc Dietrich <marvin24@gmx.de>

Re: [PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[Greg Kroah-Hartman]

On Mon, Jul 21, 2014 at 10:01:36PM +0200, Marc Dietrich wrote:
> Hi Thierry,
> 
> On Mon, 21 Jul 2014 13:52:41 +0200
> Thierry Reding <thierry.reding@gmail.com> wrote:
> 
> > From: Thierry Reding <treding@nvidia.com>
> > 
> > The mfd_add_devices() parameter takes a struct resource * as fifth
> > argument, but the nvec driver passes in a void __iomem *. The driver
> > gets away with it because none of the subdevices ever directly access
> > the registers.
> 
> you are right, this one looked bogus. 
> 
> > While at it, use platform_get_irq() instead of platform_get_resource()
> > to get the device's interrupt. This makes it easier to pass in the
> > register region since the variable is no longer reused.
> > 
> > Signed-off-by: Thierry Reding <treding@nvidia.com>
> > ---
> > Alternatively we could simply pass NULL into mfd_add_devices(), which
> > might be a slightly more accurate representation of what's going on.
> >
> > Marc, Greg, any preferences?
> 
> I just tested with NULL as base and it seems to produce no harm. Even
> if we may get rid of mfd_add_devices in the future, passing NULL is
> cleaner for now and no children will ever make use of it. So please send
> a new version with this change.
> 
> Otherwise, thanks for looking and cleaning up.

Ok, can I get your ack for this patch then?

thanks,

greg k-h

Re: [PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[Thierry Reding]

[-- Attachment #1: Type: text/plain, Size: 1669 bytes --]

On Mon, Jul 21, 2014 at 02:39:15PM -0700, Greg Kroah-Hartman wrote:
> On Mon, Jul 21, 2014 at 10:01:36PM +0200, Marc Dietrich wrote:
> > Hi Thierry,
> > 
> > On Mon, 21 Jul 2014 13:52:41 +0200
> > Thierry Reding <thierry.reding-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org> wrote:
> > 
> > > From: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> > > 
> > > The mfd_add_devices() parameter takes a struct resource * as fifth
> > > argument, but the nvec driver passes in a void __iomem *. The driver
> > > gets away with it because none of the subdevices ever directly access
> > > the registers.
> > 
> > you are right, this one looked bogus. 
> > 
> > > While at it, use platform_get_irq() instead of platform_get_resource()
> > > to get the device's interrupt. This makes it easier to pass in the
> > > register region since the variable is no longer reused.
> > > 
> > > Signed-off-by: Thierry Reding <treding-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
> > > ---
> > > Alternatively we could simply pass NULL into mfd_add_devices(), which
> > > might be a slightly more accurate representation of what's going on.
> > >
> > > Marc, Greg, any preferences?
> > 
> > I just tested with NULL as base and it seems to produce no harm. Even
> > if we may get rid of mfd_add_devices in the future, passing NULL is
> > cleaner for now and no children will ever make use of it. So please send
> > a new version with this change.
> > 
> > Otherwise, thanks for looking and cleaning up.
> 
> Ok, can I get your ack for this patch then?

I'll resend a patch which passes in NULL as base tomorrow as requested
by Marc.

Thierry

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

Re: [PATCH] staging/nvec: Pass proper resource to mfd_add_devices()

[Greg Kroah-Hartman]

On Mon, Jul 21, 2014 at 01:52:41PM +0200, Thierry Reding wrote:
> From: Thierry Reding <treding@nvidia.com>
> 
> The mfd_add_devices() parameter takes a struct resource * as fifth
> argument, but the nvec driver passes in a void __iomem *. The driver
> gets away with it because none of the subdevices ever directly access
> the registers.
> 
> While at it, use platform_get_irq() instead of platform_get_resource()
> to get the device's interrupt. This makes it easier to pass in the
> register region since the variable is no longer reused.
> 
> Signed-off-by: Thierry Reding <treding@nvidia.com>
> ---
> Alternatively we could simply pass NULL into mfd_add_devices(), which
> might be a slightly more accurate representation of what's going on.
> 
> Marc, Greg, any preferences?

None from me, Marc?