When building with W=1 we see the following warning:

|  drivers/i2c/busses/i2c-bcm-iproc.c:1039:3: error: cast to smaller \
|       integer type 'enum bcm_iproc_i2c_type' from 'const void *' [-Werror,-Wvoid-pointer-to-enum-cast]
|   1039 |                 (enum bcm_iproc_i2c_type)of_device_get_match_data(&pdev->dev);
|        |                 ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

This is due to the fact that the `bcm_iproc_i2c_type` enum members are
int-width and a cast from pointer-width down to int will cause
truncation and possible data loss. Although in this case `bcm_iproc_i2c_type`
has only a few enumerated fields and thus there is likely no data loss
occurring. Nonetheless, this patch is necessary to the goal of promoting
this warning out of W=1.

Link: https://github.com/ClangBuiltLinux/linux/issues/1903
Signed-off-by: Justin Stitt <justinstitt@google.com>
---
 drivers/i2c/busses/i2c-bcm-iproc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/i2c/busses/i2c-bcm-iproc.c b/drivers/i2c/busses/i2c-bcm-iproc.c
index 2d8342fdc25d..3464f3a376a5 100644
--- a/drivers/i2c/busses/i2c-bcm-iproc.c
+++ b/drivers/i2c/busses/i2c-bcm-iproc.c
@@ -1036,7 +1036,7 @@ static int bcm_iproc_i2c_probe(struct platform_device *pdev)
 	platform_set_drvdata(pdev, iproc_i2c);
 	iproc_i2c->device = &pdev->dev;
 	iproc_i2c->type =
-		(enum bcm_iproc_i2c_type)of_device_get_match_data(&pdev->dev);
+		(unsigned long) of_device_get_match_data(&pdev->dev);
 	init_completion(&iproc_i2c->done);
 
 	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);

-- 
2.41.0.640.ga95def55d0-goog

Hi Justin,

> @@ -1036,7 +1036,7 @@ static int bcm_iproc_i2c_probe(struct platform_device *pdev)
>  	platform_set_drvdata(pdev, iproc_i2c);
>  	iproc_i2c->device = &pdev->dev;
>  	iproc_i2c->type =
> -		(enum bcm_iproc_i2c_type)of_device_get_match_data(&pdev->dev);
> +		(unsigned long) of_device_get_match_data(&pdev->dev);

I think this should be uintptr_t, as defined in types.h:

    typedef unsigned long uintptr_t;

(I'm a bit puzzled to see a void *data cast to a 0/1 value.)

Andi

Hi Andi,

On Thu, Aug 10, 2023 at 10:56:32AM +0200, Andi Shyti wrote:
> Hi Justin,
> 
> > @@ -1036,7 +1036,7 @@ static int bcm_iproc_i2c_probe(struct platform_device *pdev)
> >  	platform_set_drvdata(pdev, iproc_i2c);
> >  	iproc_i2c->device = &pdev->dev;
> >  	iproc_i2c->type =
> > -		(enum bcm_iproc_i2c_type)of_device_get_match_data(&pdev->dev);
> > +		(unsigned long) of_device_get_match_data(&pdev->dev);
> 
> I think this should be uintptr_t, as defined in types.h:
> 
>     typedef unsigned long uintptr_t;

Yes, that is a valid comment, although I will say that I have seen some
developers/maintainers prefer 'unsigned long' over 'uintptr_t', so it is
really a crapshoot which one to pick :)

> (I'm a bit puzzled to see a void *data cast to a 0/1 value.)

It is an odd pattern but as far as I understand it, the kernel uses it a
lot to pass platform specific data around (just grepping for
'.data = (void *)' shows a lot of instances.

Cheers,
Nathan