summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--lib/Target/PowerPC/PPCCallingConv.td11
-rw-r--r--lib/Target/PowerPC/PPCISelLowering.cpp32
-rw-r--r--test/CodeGen/PowerPC/vsx-args.ll26
3 files changed, 62 insertions, 7 deletions
diff --git a/lib/Target/PowerPC/PPCCallingConv.td b/lib/Target/PowerPC/PPCCallingConv.td
index 5e3074bacc..d48164d621 100644
--- a/lib/Target/PowerPC/PPCCallingConv.td
+++ b/lib/Target/PowerPC/PPCCallingConv.td
@@ -36,7 +36,8 @@ def RetCC_PPC : CallingConv<[
CCIfType<[f64], CCAssignToReg<[F1, F2, F3, F4]>>,
// Vector types are always returned in V2.
- CCIfType<[v16i8, v8i16, v4i32, v4f32, v2f64, v2i64], CCAssignToReg<[V2]>>
+ CCIfType<[v16i8, v8i16, v4i32, v4f32], CCAssignToReg<[V2]>>,
+ CCIfType<[v2f64, v2i64], CCAssignToReg<[VSH2]>>
]>;
@@ -70,7 +71,8 @@ def RetCC_PPC64_ELF_FIS : CallingConv<[
CCIfType<[i128], CCAssignToReg<[X3, X4, X5, X6]>>,
CCIfType<[f32], CCAssignToReg<[F1, F2]>>,
CCIfType<[f64], CCAssignToReg<[F1, F2, F3, F4]>>,
- CCIfType<[v16i8, v8i16, v4i32, v4f32, v2f64, v2i64], CCAssignToReg<[V2]>>
+ CCIfType<[v16i8, v8i16, v4i32, v4f32], CCAssignToReg<[V2]>>,
+ CCIfType<[v2f64, v2i64], CCAssignToReg<[VSH2]>>
]>;
//===----------------------------------------------------------------------===//
@@ -118,8 +120,11 @@ def CC_PPC32_SVR4_VarArg : CallingConv<[
// put vector arguments in vector registers before putting them on the stack.
def CC_PPC32_SVR4 : CallingConv<[
// The first 12 Vector arguments are passed in AltiVec registers.
- CCIfType<[v16i8, v8i16, v4i32, v4f32, v2f64, v2i64],
+ CCIfType<[v16i8, v8i16, v4i32, v4f32],
CCAssignToReg<[V2, V3, V4, V5, V6, V7, V8, V9, V10, V11, V12, V13]>>,
+ CCIfType<[v2f64, v2i64],
+ CCAssignToReg<[VSH2, VSH3, VSH4, VSH5, VSH6, VSH7, VSH8, VSH9,
+ VSH10, VSH11, VSH12, VSH13]>>,
CCDelegateTo<CC_PPC32_SVR4_Common>
]>;
diff --git a/lib/Target/PowerPC/PPCISelLowering.cpp b/lib/Target/PowerPC/PPCISelLowering.cpp
index 1c19160c9c..6a43ecc250 100644
--- a/lib/Target/PowerPC/PPCISelLowering.cpp
+++ b/lib/Target/PowerPC/PPCISelLowering.cpp
@@ -2162,9 +2162,11 @@ PPCTargetLowering::LowerFormalArguments_32SVR4(
case MVT::v8i16:
case MVT::v4i32:
case MVT::v4f32:
+ RC = &PPC::VRRCRegClass;
+ break;
case MVT::v2f64:
case MVT::v2i64:
- RC = &PPC::VRRCRegClass;
+ RC = &PPC::VSHRCRegClass;
break;
}
@@ -2381,6 +2383,10 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
+ static const uint16_t VSRH[] = {
+ PPC::VSH2, PPC::VSH3, PPC::VSH4, PPC::VSH5, PPC::VSH6, PPC::VSH7, PPC::VSH8,
+ PPC::VSH9, PPC::VSH10, PPC::VSH11, PPC::VSH12, PPC::VSH13
+ };
const unsigned Num_GPR_Regs = array_lengthof(GPR);
const unsigned Num_FPR_Regs = 13;
@@ -2573,7 +2579,9 @@ PPCTargetLowering::LowerFormalArguments_64SVR4(
// Note that vector arguments in registers don't reserve stack space,
// except in varargs functions.
if (VR_idx != Num_VR_Regs) {
- unsigned VReg = MF.addLiveIn(VR[VR_idx], &PPC::VRRCRegClass);
+ unsigned VReg = (ObjectVT == MVT::v2f64 || ObjectVT == MVT::v2i64) ?
+ MF.addLiveIn(VSRH[VR_idx], &PPC::VSHRCRegClass) :
+ MF.addLiveIn(VR[VR_idx], &PPC::VRRCRegClass);
ArgVal = DAG.getCopyFromReg(Chain, dl, VReg, ObjectVT);
if (isVarArg) {
while ((ArgOffset % 16) != 0) {
@@ -4011,6 +4019,11 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
PPC::V2, PPC::V3, PPC::V4, PPC::V5, PPC::V6, PPC::V7, PPC::V8,
PPC::V9, PPC::V10, PPC::V11, PPC::V12, PPC::V13
};
+ static const uint16_t VSRH[] = {
+ PPC::VSH2, PPC::VSH3, PPC::VSH4, PPC::VSH5, PPC::VSH6, PPC::VSH7, PPC::VSH8,
+ PPC::VSH9, PPC::VSH10, PPC::VSH11, PPC::VSH12, PPC::VSH13
+ };
+
const unsigned NumGPRs = array_lengthof(GPR);
const unsigned NumFPRs = 13;
const unsigned NumVRs = array_lengthof(VR);
@@ -4242,7 +4255,13 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
MachinePointerInfo(),
false, false, false, 0);
MemOpChains.push_back(Load.getValue(1));
- RegsToPass.push_back(std::make_pair(VR[VR_idx++], Load));
+
+ unsigned VReg = (Arg.getSimpleValueType() == MVT::v2f64 ||
+ Arg.getSimpleValueType() == MVT::v2i64) ?
+ VSRH[VR_idx] : VR[VR_idx];
+ ++VR_idx;
+
+ RegsToPass.push_back(std::make_pair(VReg, Load));
}
ArgOffset += 16;
for (unsigned i=0; i<16; i+=PtrByteSize) {
@@ -4262,7 +4281,12 @@ PPCTargetLowering::LowerCall_64SVR4(SDValue Chain, SDValue Callee,
// stack space allocated at the end.
if (VR_idx != NumVRs) {
// Doesn't have GPR space allocated.
- RegsToPass.push_back(std::make_pair(VR[VR_idx++], Arg));
+ unsigned VReg = (Arg.getSimpleValueType() == MVT::v2f64 ||
+ Arg.getSimpleValueType() == MVT::v2i64) ?
+ VSRH[VR_idx] : VR[VR_idx];
+ ++VR_idx;
+
+ RegsToPass.push_back(std::make_pair(VReg, Arg));
} else {
LowerMemOpCallTo(DAG, MF, Chain, Arg, PtrOff, SPDiff, ArgOffset,
true, isTailCall, true, MemOpChains,
diff --git a/test/CodeGen/PowerPC/vsx-args.ll b/test/CodeGen/PowerPC/vsx-args.ll
new file mode 100644
index 0000000000..520aeb5fa9
--- /dev/null
+++ b/test/CodeGen/PowerPC/vsx-args.ll
@@ -0,0 +1,26 @@
+; RUN: llc < %s -mcpu=pwr7 -mattr=+vsx | FileCheck %s
+target datalayout = "E-m:e-i64:64-n32:64"
+target triple = "powerpc64-unknown-linux-gnu"
+
+declare <2 x double> @sv(<2 x double>, <2 x i64>, <4 x float>) #0
+
+define <2 x double> @main(<4 x float> %a, <2 x double> %b, <2 x i64> %c) #1 {
+entry:
+ %ca = tail call <2 x double> @sv(<2 x double> %b, <2 x i64> %c, <4 x float> %a)
+ %v = fadd <2 x double> %ca, <double 1.0, double 1.0>
+ ret <2 x double> %v
+
+; CHECK-LABEL: @main
+; CHECK-DAG: vor [[V:[0-9]+]], 2, 2
+; CHECK-DAG: xxlor 34, 35, 35
+; CHECK-DAG: xxlor 35, 36, 36
+; CHECK-DAG: vor 4, [[V]], [[V]]
+; CHECK-DAG: bl sv
+; CHECK-DAG: lxvd2x [[VC:[0-9]+]],
+; CHECK: xvadddp 34, 34, [[VC]]
+; CHECK: blr
+}
+
+attributes #0 = { noinline nounwind readnone }
+attributes #1 = { nounwind }
+
generated by cgit v1.2.3 (git 2.25.1) at 2024-06-11 16:44:06 +0000