Richard Earnshaw <rearn...@arm.com> writes:
> On Thu, 2011-03-24 at 15:40 +0000, Richard Sandiford wrote:
>> gcc/
>> * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Restrict FPA_REGS
>> case to VFPv1.
>>
>
> GCC doesn't support VFPv1 (see the all_fpus table), and I don't think
> many chips based on that ever escaped into the wild world, so I'm not
> worried about trying to add that now.
>
> So it's probably safe to just kill that check for VFP entirely.
Even better. :-) How does this look? Tested on arm-linux-gnueabi,
no regressions.
Richard
gcc/
* config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Restrict FPA_REGS
case to VFPv1.
Index: gcc/config/arm/arm.h
===================================================================
--- gcc/config/arm/arm.h 2011-03-24 17:21:52.000000000 +0000
+++ gcc/config/arm/arm.h 2011-03-24 18:39:09.000000000 +0000
@@ -1167,12 +1167,10 @@ #define IRA_COVER_CLASSES
\
}
/* FPA registers can't do subreg as all values are reformatted to internal
- precision. VFP registers may only be accessed in the mode they
- were set. */
-#define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
- (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
- ? reg_classes_intersect_p (FPA_REGS, (CLASS)) \
- || reg_classes_intersect_p (VFP_REGS, (CLASS)) \
+ precision. */
+#define CANNOT_CHANGE_MODE_CLASS(FROM, TO, CLASS) \
+ (GET_MODE_SIZE (FROM) != GET_MODE_SIZE (TO) \
+ ? reg_classes_intersect_p (FPA_REGS, (CLASS)) \
: 0)
/* The class value for index registers, and the one for base regs. */
