[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Alexey Dobriyan]

vid_to_reg() can return -1 and char can be unsigned.

Signed-off-by: Alexey Dobriyan <adobriyan at gmail.com>
---

 drivers/hwmon/atxp1.c |    3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

--- a/drivers/hwmon/atxp1.c
+++ b/drivers/hwmon/atxp1.c
@@ -116,8 +116,7 @@ static ssize_t atxp1_storevcore(struct d
 {
 	struct atxp1_data *data;
 	struct i2c_client *client;
-	char vid;
-	char cvid;
+	int vid, cvid;
 	unsigned int vcore;
 
 	client = to_i2c_client(dev);

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Jean Delvare]

Hi Alexey,

> vid_to_reg() can return -1 and char can be unsigned.

What does it take for char be unsigned? Never seen that.

> Signed-off-by: Alexey Dobriyan <adobriyan at gmail.com>
> ---
> 
>  drivers/hwmon/atxp1.c |    3 +--
>  1 file changed, 1 insertion(+), 2 deletions(-)
> 
> --- a/drivers/hwmon/atxp1.c
> +++ b/drivers/hwmon/atxp1.c
> @@ -116,8 +116,7 @@ static ssize_t atxp1_storevcore(struct d
>  {
>  	struct atxp1_data *data;
>  	struct i2c_client *client;
> -	char vid;
> -	char cvid;
> +	int vid, cvid;
>  	unsigned int vcore;
>  
>  	client = to_i2c_client(dev);

Whatever, this looks better with ints, I agree. I'll take that patch,
unless Sebastian objects.

-- 
Jean Delvare

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Alexey Dobriyan]

On Thu, Jul 27, 2006 at 08:20:53PM +0200, Jean Delvare wrote:
> > vid_to_reg() can return -1 and char can be unsigned.
>
> What does it take for char be unsigned? Never seen that.

Cross-gcc for arm here. That's how it was noticed. Or -funsigned-char
gcc flag.

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Sebastian Witt]

Jean Delvare wrote:
> 
> Whatever, this looks better with ints, I agree. I'll take that patch,
> unless Sebastian objects.

No, I'm fine with that.

Bye,
Sebastian

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Jean Delvare]

> On Thu, Jul 27, 2006 at 08:20:53PM +0200, Jean Delvare wrote:
> > > vid_to_reg() can return -1 and char can be unsigned.
> >
> > What does it take for char be unsigned? Never seen that.
> 
> Cross-gcc for arm here. That's how it was noticed.

I'd call it a bug.

> Or -funsigned-char gcc flag.

Irrelevant as far as the kernel compilation goes - but thanks for the
hint.

I take the patch anyway.

-- 
Jean Delvare

[lm-sensors] [PATCH] atxp1: signed/unsigned char bug

[Alexey Dobriyan]

On Thu, Jul 27, 2006 at 10:28:49PM +0200, Jean Delvare wrote:
> > On Thu, Jul 27, 2006 at 08:20:53PM +0200, Jean Delvare wrote:
> > > > vid_to_reg() can return -1 and char can be unsigned.
> > >
> > > What does it take for char be unsigned? Never seen that.
> >
> > Cross-gcc for arm here. That's how it was noticed.
>
> I'd call it a bug.

C standard explicitly says that "char" equivalence to either
"signed char" or "unsigned char" is implementation defined.
But, yes, it's confusing.