svn commit: r333801 - head/sys/contrib/dev/ath/ath_hal/ar9300

[Ian Lepore]

On Fri, 2018-05-18 at 17:23 +0000, Sean Bruno wrote:
> Author: sbruno
> Date: Fri May 18 17:23:23 2018
> New Revision: 333801
> URL: https://svnweb.freebsd.org/changeset/base/333801
> 
> Log:
>   Quiesce a couple pages of clang warnings with a cast.  Duplicates
>   linux maintainer commit:
>   
>   https://github.com/torvalds/linux/commit/627871b71c89a6ec12fbed75063f238e0c7127b2#diff-8c6ddb4c3ad69a6fb9f289475821db56
>   
>   ar9300template_aphrodite.h:575:40: warning: implicit conversion from 'int'
>     to 'u_int8_t' (aka 'unsigned char') changes value from 3495 to 167
>     [-Wconstant-conversion]
>               /* Data[8].ctl_edges[7].bChannel*/FREQ2FBIN(5795, 0)}
>                                                 ^~~~~~~~~~~~~~~~~~
>   ar9300eep.h:142:41: note: expanded from macro 'FREQ2FBIN'
>       (((y) == HAL_FREQ_BAND_2GHZ) ? ((x) - 2300) : (((x) - 4800) / 5))
>   
>   Reviewed by:	imp
>   MFC after:	1 week
>   Differential Revision:	https://reviews.freebsd.org/D15476
> 
> Modified:
>   head/sys/contrib/dev/ath/ath_hal/ar9300/ar9300eep.h
> 
> Modified: head/sys/contrib/dev/ath/ath_hal/ar9300/ar9300eep.h
> ==============================================================================
> --- head/sys/contrib/dev/ath/ath_hal/ar9300/ar9300eep.h	Fri May 18 17:07:59 2018	(r333800)
> +++ head/sys/contrib/dev/ath/ath_hal/ar9300/ar9300eep.h	Fri May 18 17:23:23 2018	(r333801)
> @@ -139,7 +139,7 @@ enum Ar9300EepromTemplate
>  #define OSPREY_CUSTOMER_DATA_SIZE    20
>  
>  #define FREQ2FBIN(x,y) \
> -    (((y) == HAL_FREQ_BAND_2GHZ) ? ((x) - 2300) : (((x) - 4800) / 5))
> +    (u_int8_t)(((y) == HAL_FREQ_BAND_2GHZ) ? ((x) - 2300) : (((x) - 4800) / 5))
>  #define FBIN2FREQ(x,y) \
>      (((y) == HAL_FREQ_BAND_2GHZ) ? (2300 + x) : (4800 + 5 * x))
>  #define OSPREY_MAX_CHAINS            3
> 

How can this change possibly be correct? All the values involved are
compile-time constants, the compiler did the math and tells you the
result is 3495. Cast or not, that value is never going to fit into a
uint8_t. Sure this supresses the warning, but what about actually
fixing the overflow?

-- Ian