This patch fixes SingleSource/UnitTests/2003-05-07-VarArgs.c test on
arm-linux-gnueabi.
Lauro
Index: gcc/llvm-abi.h
===================================================================
--- gcc/llvm-abi.h (revision 285)
+++ gcc/llvm-abi.h (working copy)
@@ -32,6 +32,7 @@
#include "llvm-internal.h"
#include "llvm/Constants.h"
#include "llvm/DerivedTypes.h"
+#include "llvm/Target/TargetData.h"
namespace llvm {
class BasicBlock;
@@ -274,17 +275,32 @@
/// of the struct elements in.
void PassInIntegerRegisters(tree type, const Type *Ty) {
unsigned Size = TREE_INT_CST_LOW(TYPE_SIZE(type))/8;
-
+
+ // FIXME: We should preserve all aggregate value alignment information.
+
+ // Work around to preserve some aggregate value alignment information:
+ // don't bitcast aggregate value to Int64 if its alignment is different
+ // from Int64 alignment. ARM backend needs this.
+ unsigned Align = TYPE_ALIGN(type)/8;
+ unsigned Int64Align = getTargetData().getABITypeAlignment(Type::Int64Ty);
+ bool UseInt64 = (Int64Align == Align);
+
// FIXME: In cases where we can, we should use the original struct.
// Consider cases like { int, int } and {int, short} for example! This will
// produce far better LLVM code!
std::vector<const Type*> Elts;
- for (; Size >= 8; Size -= 8)
- Elts.push_back(Type::Int64Ty);
- if (Size >= 4) {
- Elts.push_back(Type::Int32Ty);
- Size -= 4;
+ if (UseInt64) {
+ for (; Size >= 8; Size -= 8)
+ Elts.push_back(Type::Int64Ty);
+ if (Size >= 4) {
+ Elts.push_back(Type::Int32Ty);
+ Size -= 4;
+ }
+ } else {
+ for (; Size >= 4; Size -= 4)
+ Elts.push_back(Type::Int32Ty);
}
+
if (Size >= 2) {
Elts.push_back(Type::Int16Ty);
Size -= 2;
_______________________________________________
llvm-commits mailing list
llvm-commits@cs.uiuc.edu
http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
