Richard Earnshaw <rearn...@arm.com> writes:
>> gcc/
>> * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Restrict FPA_REGS
>> case to VFPv1.
>>
>
> I think adding a comment that VFPv1 would require a restriction but that
> that isn't supported would be useful. Then if the worst happens it will
> be one less bug to go track down.
>
> OK with that change.
Thanks. Here's what I installed.
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 2011-03-25 18:19:45.000000000 +0000
+++ gcc/config/arm/arm.h 2011-03-25 18:26:13.000000000 +0000
@@ -1167,12 +1167,12 @@ #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. In VFPv1, VFP registers could only be accessed in the mode
+ they were set, so subregs would be invalid there too. However, we don't
+ support VFPv1 at the moment, and the restriction was lifted in VFPv2. */
+#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. */
