[SelectionDAG] Fix return calling convention in expansion of ?MULO

Summary:
The SMULO/UMULO DAG nodes, when not directly supported by the target,
expand to a multiplication twice as wide. In case that the resulting
type is not legal, the legalizer cannot directly call the intrinsic
with the wide arguments; instead, it "pre-lowers" them by splitting
them in halves.

rL283203 made sure that on big endian targets, the legalizer passes
the argument halves in the correct order. It did not do the same
for the return value halves because the existing code used a hack;
it put an illegal type into DAG and hoped that nothing would break
and it would be correctly lowered elsewhere.

rL307207 fixed this, handling return value halves similar to how
argument handles are handled, but did not take big-endian targets
into account.

This commit fixes the expansion on big-endian targets, such as
the out-of-tree OR1K target.

Reviewers: eli.friedman, vadimcn

Subscribers: george-hopkins, efriedma, llvm-commits

Differential Revision: https://reviews.llvm.org/D45355

llvm-svn: 353854

llvm/lib/CodeGen/SelectionDAG/LegalizeDAG.cpp

@@ -3423,7 +3423,7 @@ bool SelectionDAGLegalize::ExpandNode(SDNode *Node) {
       // being a legal type for the architecture and thus has to be split to
       // two arguments.
       SDValue Ret;
-      if(DAG.getDataLayout().isLittleEndian()) {
+      if (DAG.getDataLayout().isLittleEndian()) {
         // Halves of WideVT are packed into registers in different order
         // depending on platform endianness. This is usually handled by
         // the C calling convention, but we can't defer to it in
@@ -3436,8 +3436,14 @@ bool SelectionDAGLegalize::ExpandNode(SDNode *Node) {
       }
       assert(Ret.getOpcode() == ISD::MERGE_VALUES &&
              "Ret value is a collection of constituent nodes holding result.");
-      BottomHalf = Ret.getOperand(0);
-      TopHalf = Ret.getOperand(1);
+      if (DAG.getDataLayout().isLittleEndian()) {
+        // Same as above.
+        BottomHalf = Ret.getOperand(0);
+        TopHalf = Ret.getOperand(1);
+      } else {
+        BottomHalf = Ret.getOperand(1);
+        TopHalf = Ret.getOperand(0);
+      }
     }
 
     if (isSigned) {

llvm/test/CodeGen/Thumb/umulo-32-legalisation-lowering.ll

@@ -0,0 +1,43 @@
+; RUN: llc < %s -mtriple=thumb-eabi   | FileCheck %s --check-prefixes CHECK,CHECK-LITTLE
+; RUN: llc < %s -mtriple=thumbeb-eabi | FileCheck %s --check-prefixes CHECK,CHECK-BIG
+
+define i1 @umulo32(i32 %l, i32 %r) unnamed_addr #0 {
+; CHECK-LABEL: umulo32:
+; CHECK:                @ %bb.0: @ %start
+; CHECK-NEXT:           .save {r7, lr}
+; CHECK-NEXT:           push {r7, lr}
+; CHECK-LITTLE-NEXT:    movs r2, r1
+; CHECK-LITTLE-NEXT:    movs r1, #0
+; CHECK-NEXT:           movs r3, r1
+; CHECK-BIG-NEXT:       movs r1, r0
+; CHECK-BIG-NEXT:       movs r0, #0
+; CHECK-BIG-NEXT:       movs r2, r0
+; CHECK-NEXT:           bl __aeabi_lmul
+; CHECK-LITTLE-NEXT:    cmp  r1, #0
+; CHECK-LITTLE-NEXT:    bne  .LBB0_2
+; CHECK-LITTLE-NEXT:    @ %bb.1:
+; CHECK-LITTLE-NEXT:    movs r0, r1
+; CHECK-LITTLE-NEXT:    b    .LBB0_3
+; CHECK-LITTLE-NEXT:    .LBB0_2:
+; CHECK-LITTLE-NEXT:    movs r0, #1
+; CHECK-LITTLE-NEXT:    .LBB0_3:
+; CHECK-BIG-NEXT:       cmp  r0, #0
+; CHECK-BIG-NEXT:       beq  .LBB0_2
+; CHECK-BIG-NEXT:       @ %bb.1:
+; CHECK-BIG-NEXT:       movs r0, #1
+; CHECK-BIG-NEXT:       .LBB0_2:
+; CHECK-NEXT:           pop {r7}
+; CHECK-NEXT:           pop {r1}
+; CHECK-NEXT:           bx r1
+start:
+  %0 = tail call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %l, i32 %r) #2
+  %1 = extractvalue { i32, i1 } %0, 1
+  ret i1 %1
+}
+
+; Function Attrs: nounwind readnone speculatable
+declare { i32, i1 } @llvm.umul.with.overflow.i32(i32, i32) #1
+
+attributes #0 = { nounwind readnone uwtable }
+attributes #1 = { nounwind readnone speculatable }
+attributes #2 = { nounwind }