[patch] Width of volatile bitfields (PR23623)

[Richard Earnshaw]

On Tue, 2006-03-28 at 17:20, Paul Brook wrote:

> Tested with cross to arm-none-eabi, and spot check on i686-inux.
> Ok?
> 
> Paul
> 
> 2006-03-28  Paul Brook  <paul@codesourcery.com>
> 
> 	* targhooks.c (default_narrow_bitfield): New fuction.
> 	* targhooks.h (default_narrow_bitfield): add prototype.
> 	* target-def.h (TARGET_NARROW_VOLATILE_BITFIELD): Define.
> 	* doc/tm.texi: Document TARGET_NARROW_VOLATILE_BITFIELDS.
> 	* config/arm/arm.c (TARGET_NARROW_VOLATILE_BITFIELD): Define.
> 
> +++ gcc/doc/tm.texi	(working copy)
> @@ -1212,6 +1212,14 @@ structure.  The hook should return true 
>  the alignment requirements of an unnamed bitfield's type.
>  @end deftypefn
>  
> +@deftypefn {Target Hook} bool TARGET_NARROW_VOLATILE_BITFIELDS (void)
> +This target hook should return @code{true} if assesses to volatile bitfields

accesses

> Index: gcc/targhooks.c
> ===================================================================
> --- gcc/targhooks.c	(revision 112423)
> +++ gcc/targhooks.c	(working copy)
> @@ -463,4 +463,18 @@ default_internal_arg_pointer (void)
>      return virtual_incoming_args_rtx;
>  }
>  
> +
> +/* If STRICT_ALIGNMENT is true we use the container type for accessing
> +   volatile bitfields.  This is generally the prefered behavior for memory

preferred (doubled 'r')


> Index: gcc/config/arm/vfp.md

Nothing to do with this patch... :-)

Otherwise OK.

R.