diff options
| -rw-r--r-- | lib/CodeGen/CGCall.cpp | 10 | ||||
| -rw-r--r-- | test/CodeGenOpenCL/addr-space-struct-arg.cl | 23 |
2 files changed, 31 insertions, 2 deletions
diff --git a/lib/CodeGen/CGCall.cpp b/lib/CodeGen/CGCall.cpp index b6ec67d83c..fa5ce7fdca 100644 --- a/lib/CodeGen/CGCall.cpp +++ b/lib/CodeGen/CGCall.cpp @@ -2199,17 +2199,23 @@ RValue CodeGenFunction::EmitCall(const CGFunctionInfo &CallInfo, checkArgMatches(AI, IRArgNo, IRFuncTy); } else { // We want to avoid creating an unnecessary temporary+copy here; - // however, we need one in two cases: + // however, we need one in three cases: // 1. If the argument is not byval, and we are required to copy the // source. (This case doesn't occur on any common architecture.) // 2. If the argument is byval, RV is not sufficiently aligned, and // we cannot force it to be sufficiently aligned. + // 3. If the argument is byval, but RV is located in an address space + // different than that of the argument (0). llvm::Value *Addr = RV.getAggregateAddr(); unsigned Align = ArgInfo.getIndirectAlign(); const llvm::DataLayout *TD = &CGM.getDataLayout(); + const unsigned RVAddrSpace = Addr->getType()->getPointerAddressSpace(); + const unsigned ArgAddrSpace = (IRArgNo < IRFuncTy->getNumParams() ? + IRFuncTy->getParamType(IRArgNo)->getPointerAddressSpace() : 0); if ((!ArgInfo.getIndirectByVal() && I->NeedsCopy) || (ArgInfo.getIndirectByVal() && TypeAlign.getQuantity() < Align && - llvm::getOrEnforceKnownAlignment(Addr, Align, TD) < Align)) { + llvm::getOrEnforceKnownAlignment(Addr, Align, TD) < Align) || + (ArgInfo.getIndirectByVal() && (RVAddrSpace != ArgAddrSpace))) { // Create an aligned temporary, and copy to it. llvm::AllocaInst *AI = CreateMemTemp(I->Ty); if (Align > AI->getAlignment()) diff --git a/test/CodeGenOpenCL/addr-space-struct-arg.cl b/test/CodeGenOpenCL/addr-space-struct-arg.cl new file mode 100644 index 0000000000..f04923d39c --- /dev/null +++ b/test/CodeGenOpenCL/addr-space-struct-arg.cl @@ -0,0 +1,23 @@ +// RUN: %clang_cc1 %s -emit-llvm -o - -O0 -ffake-address-space-map -triple i686-pc-darwin | FileCheck %s
+
+typedef struct {
+ int cells[9];
+} Mat3X3;
+
+typedef struct {
+ int cells[16];
+} Mat4X4;
+
+Mat4X4 __attribute__((noinline)) foo(Mat3X3 in) {
+ Mat4X4 out;
+ return out;
+}
+
+kernel void ker(global Mat3X3 *in, global Mat4X4 *out) {
+ out[0] = foo(in[1]);
+}
+
+// Expect two mem copies: one for the argument "in", and one for
+// the return value.
+// CHECK: call void @llvm.memcpy.p0i8.p1i8.i32(i8*
+// CHECK: call void @llvm.memcpy.p1i8.p0i8.i32(i8 addrspace(1)*
|