Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -41014,6 +41014,7 @@
   assert((X86ISD::ADD == N->getOpcode() || X86ISD::SUB == N->getOpcode()) &&
          "Expected X86ISD::ADD or X86ISD::SUB");
 
+  SDLoc DL(N);
   SDValue LHS = N->getOperand(0);
   SDValue RHS = N->getOperand(1);
   MVT VT = LHS.getSimpleValueType();
@@ -41021,21 +41022,20 @@
 
   // If we don't use the flag result, simplify back to a generic ADD/SUB.
   if (!N->hasAnyUseOfValue(1)) {
-    SDLoc DL(N);
     SDValue Res = DAG.getNode(GenericOpc, DL, VT, LHS, RHS);
     return DAG.getMergeValues({Res, DAG.getConstant(0, DL, MVT::i32)}, DL);
   }
 
   // Fold any similar generic ADD/SUB opcodes to reuse this node.
   auto MatchGeneric = [&](SDValue N0, SDValue N1, bool Negate) {
-    // TODO: Add SUB(RHS, LHS) -> SUB(0, SUB(LHS, RHS)) negation support, this
-    // currently causes regressions as we don't have broad x86sub combines.
-    if (Negate)
-      return;
     SDValue Ops[] = {N0, N1};
     SDVTList VTs = DAG.getVTList(N->getValueType(0));
-    if (SDNode *GenericAddSub = DAG.getNodeIfExists(GenericOpc, VTs, Ops))
-      DCI.CombineTo(GenericAddSub, SDValue(N, 0));
+    if (SDNode *GenericAddSub = DAG.getNodeIfExists(GenericOpc, VTs, Ops)) {
+      SDValue Op(N, 0);
+      if (Negate)
+        Op = DAG.getNode(ISD::SUB, DL, VT, DAG.getConstant(0, DL, VT), Op);
+      DCI.CombineTo(GenericAddSub, Op);
+    }
   };
   MatchGeneric(LHS, RHS, false);
   MatchGeneric(RHS, LHS, X86ISD::SUB == N->getOpcode());
Index: test/CodeGen/X86/combine-sbb.ll
===================================================================
--- test/CodeGen/X86/combine-sbb.ll
+++ test/CodeGen/X86/combine-sbb.ll
@@ -199,26 +199,24 @@
 define i32 @PR40483_sub3(i32*, i32) nounwind {
 ; X86-LABEL: PR40483_sub3:
 ; X86:       # %bb.0:
-; X86-NEXT:    pushl %edi
 ; X86-NEXT:    pushl %esi
-; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
+; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-NEXT:    movl (%eax), %edx
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-NEXT:    movl (%esi), %edx
-; X86-NEXT:    movl %edx, %eax
-; X86-NEXT:    subl %ecx, %eax
-; X86-NEXT:    movl %edx, %edi
-; X86-NEXT:    subl %ecx, %edi
-; X86-NEXT:    movl %edi, (%esi)
+; X86-NEXT:    movl %edx, %ecx
+; X86-NEXT:    subl %esi, %ecx
+; X86-NEXT:    subl %esi, %edx
+; X86-NEXT:    movl %edx, (%eax)
 ; X86-NEXT:    jae .LBB5_1
 ; X86-NEXT:  # %bb.2:
 ; X86-NEXT:    xorl %eax, %eax
-; X86-NEXT:    jmp .LBB5_3
+; X86-NEXT:    popl %esi
+; X86-NEXT:    retl
 ; X86-NEXT:  .LBB5_1:
-; X86-NEXT:    subl %edx, %ecx
+; X86-NEXT:    movl %ecx, %eax
+; X86-NEXT:    negl %eax
 ; X86-NEXT:    orl %ecx, %eax
-; X86-NEXT:  .LBB5_3:
 ; X86-NEXT:    popl %esi
-; X86-NEXT:    popl %edi
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: PR40483_sub3:
@@ -226,8 +224,8 @@
 ; X64-NEXT:    movl (%rdi), %ecx
 ; X64-NEXT:    movl %ecx, %eax
 ; X64-NEXT:    subl %esi, %eax
-; X64-NEXT:    movl %esi, %edx
-; X64-NEXT:    subl %ecx, %edx
+; X64-NEXT:    movl %eax, %edx
+; X64-NEXT:    negl %edx
 ; X64-NEXT:    orl %eax, %edx
 ; X64-NEXT:    xorl %eax, %eax
 ; X64-NEXT:    subl %esi, %ecx
Index: test/CodeGen/X86/jump_sign.ll
===================================================================
--- test/CodeGen/X86/jump_sign.ll
+++ test/CodeGen/X86/jump_sign.ll
@@ -48,11 +48,16 @@
 define i32 @func_h(i32 %a, i32 %b) nounwind {
 ; CHECK-LABEL: func_h:
 ; CHECK:       # %bb.0:
+; CHECK-NEXT:    pushl %esi
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    subl %ecx, %eax
-; CHECK-NEXT:    cmovlel %edx, %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-NEXT:    movl %ecx, %esi
+; CHECK-NEXT:    subl %edx, %esi
+; CHECK-NEXT:    negl %esi
+; CHECK-NEXT:    xorl %eax, %eax
+; CHECK-NEXT:    cmpl %edx, %ecx
+; CHECK-NEXT:    cmovll %esi, %eax
+; CHECK-NEXT:    popl %esi
 ; CHECK-NEXT:    retl
   %cmp = icmp slt i32 %b, %a
   %sub = sub nsw i32 %a, %b
@@ -91,11 +96,16 @@
 define i32 @func_k(i32 %a, i32 %b) nounwind {
 ; CHECK-LABEL: func_k:
 ; CHECK:       # %bb.0:
+; CHECK-NEXT:    pushl %esi
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    subl %ecx, %eax
-; CHECK-NEXT:    cmovbel %edx, %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-NEXT:    movl %ecx, %esi
+; CHECK-NEXT:    subl %edx, %esi
+; CHECK-NEXT:    negl %esi
+; CHECK-NEXT:    xorl %eax, %eax
+; CHECK-NEXT:    cmpl %edx, %ecx
+; CHECK-NEXT:    cmovbl %esi, %eax
+; CHECK-NEXT:    popl %esi
 ; CHECK-NEXT:    retl
   %cmp = icmp ult i32 %b, %a
   %sub = sub i32 %a, %b
@@ -111,7 +121,9 @@
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
 ; CHECK-NEXT:    movl %edx, %eax
 ; CHECK-NEXT:    subl %ecx, %eax
-; CHECK-NEXT:    cmovlel %edx, %eax
+; CHECK-NEXT:    negl %eax
+; CHECK-NEXT:    cmpl %ecx, %edx
+; CHECK-NEXT:    cmovgel %ecx, %eax
 ; CHECK-NEXT:    retl
   %cmp = icmp slt i32 %b, %a
   %sub = sub nsw i32 %a, %b
@@ -138,10 +150,11 @@
 define i32 @func_l2(i32 %a, i32 %b) nounwind {
 ; CHECK-LABEL: func_l2:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    movl %eax, %ecx
-; CHECK-NEXT:    subl %edx, %ecx
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-NEXT:    movl %edx, %ecx
+; CHECK-NEXT:    subl %eax, %ecx
+; CHECK-NEXT:    negl %ecx
 ; CHECK-NEXT:    cmpl %eax, %edx
 ; CHECK-NEXT:    jne .LBB8_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
@@ -167,9 +180,12 @@
 ; CHECK-LABEL: func_l3:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    subl %ecx, %eax
-; CHECK-NEXT:    jge .LBB9_2
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-NEXT:    movl %ecx, %eax
+; CHECK-NEXT:    subl %edx, %eax
+; CHECK-NEXT:    negl %eax
+; CHECK-NEXT:    cmpl %edx, %ecx
+; CHECK-NEXT:    jle .LBB9_2
 ; CHECK-NEXT:  # %bb.1: # %if.then
 ; CHECK-NEXT:    retl
 ; CHECK-NEXT:  .LBB9_2: # %if.else
@@ -192,11 +208,16 @@
 define i32 @func_l4(i32 %a, i32 %b) nounwind {
 ; CHECK-LABEL: func_l4:
 ; CHECK:       # %bb.0:
+; CHECK-NEXT:    pushl %esi
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    xorl %edx, %edx
-; CHECK-NEXT:    subl %ecx, %eax
-; CHECK-NEXT:    cmovll %edx, %eax
+; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %edx
+; CHECK-NEXT:    movl %ecx, %esi
+; CHECK-NEXT:    subl %edx, %esi
+; CHECK-NEXT:    negl %esi
+; CHECK-NEXT:    xorl %eax, %eax
+; CHECK-NEXT:    cmpl %edx, %ecx
+; CHECK-NEXT:    cmovlel %esi, %eax
+; CHECK-NEXT:    popl %esi
 ; CHECK-NEXT:    retl
   %cmp = icmp sgt i32 %b, %a
   %sub = sub i32 %a, %b
@@ -318,12 +339,10 @@
 ; CHECK-LABEL: func_q:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; CHECK-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; CHECK-NEXT:    movl %ecx, %edx
-; CHECK-NEXT:    subl %eax, %edx
-; CHECK-NEXT:    cmpl %ecx, %eax
-; CHECK-NEXT:    sbbl %eax, %eax
-; CHECK-NEXT:    xorl %edx, %eax
+; CHECK-NEXT:    subl {{[0-9]+}}(%esp), %eax
+; CHECK-NEXT:    sbbl %ecx, %ecx
+; CHECK-NEXT:    negl %eax
+; CHECK-NEXT:    xorl %ecx, %eax
 ; CHECK-NEXT:    retl
   %t1 = icmp ult i32 %a0, %a1
   %t2 = sub i32 %a1, %a0