diff --git a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
--- a/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/DAGCombiner.cpp
@@ -12640,10 +12640,37 @@
   if (SDValue ExtLoad = CombineExtLoad(N))
     return ExtLoad;
 
-  // fold (zext (and/or/xor (load x), cst)) ->
-  //      (and/or/xor (zextload x), (zext cst))
-  // Unless (and (load x) cst) will match as a zextload already and has
-  // additional users.
+  // (zext (xor/and/or (load x), (load y)))
+  // ->
+  // (xor/and/or (zextload x), (zextload y))
+  // May be possible to extend, currently limited to the simplest cases.
+  if (!VT.isVector() &&
+      (N0.getOpcode() == ISD::AND || N0.getOpcode() == ISD::OR ||
+       N0.getOpcode() == ISD::XOR)) {
+    SDValue N00 = N0.getOperand(0);
+    SDValue N01 = N0.getOperand(1);
+    LoadSDNode *L00 = dyn_cast<LoadSDNode>(N00);
+    LoadSDNode *L01 = dyn_cast<LoadSDNode>(N01);
+    auto CanExtendLoad = [&](LoadSDNode *Load, SDValue LoadValue) {
+      return (Load->hasOneUse() &&
+              Load->getExtensionType() == ISD::NON_EXTLOAD &&
+              Load->isUnindexed() && Load->isSimple() &&
+              TLI.isLoadExtLegal(ISD::ZEXTLOAD, VT, LoadValue.getValueType()));
+    };
+    if (L00 && L01 && CanExtendLoad(L00, N00) && CanExtendLoad(L01, N01)) {
+      auto GetExtLoad = [&](LoadSDNode *Load, SDValue LoadValue) {
+        return DAG.getExtLoad(ISD::ZEXTLOAD, SDLoc(LoadValue), VT,
+                              Load->getChain(), Load->getBasePtr(),
+                              LoadValue.getValueType(), Load->getMemOperand());
+      };
+      SDValue ExtL00 = GetExtLoad(L00, N00);
+      SDValue ExtL01 = GetExtLoad(L01, N01);
+      SDValue Op = DAG.getNode(N0.getOpcode(), SDLoc(N), VT, ExtL00, ExtL01);
+      CombineTo(N, Op);
+      return SDValue(N, 0);
+    }
+  }
+
   if ((N0.getOpcode() == ISD::AND || N0.getOpcode() == ISD::OR ||
        N0.getOpcode() == ISD::XOR) &&
       isa<LoadSDNode>(N0.getOperand(0)) &&
diff --git a/llvm/test/CodeGen/AArch64/bcmp.ll b/llvm/test/CodeGen/AArch64/bcmp.ll
--- a/llvm/test/CodeGen/AArch64/bcmp.ll
+++ b/llvm/test/CodeGen/AArch64/bcmp.ll
@@ -46,10 +46,8 @@
 ; CHECK-NEXT:    ldrh w9, [x1]
 ; CHECK-NEXT:    ldrb w10, [x0, #2]
 ; CHECK-NEXT:    ldrb w11, [x1, #2]
-; CHECK-NEXT:    eor w8, w8, w9
-; CHECK-NEXT:    eor w9, w10, w11
-; CHECK-NEXT:    orr w8, w8, w9
-; CHECK-NEXT:    cmp w8, #0
+; CHECK-NEXT:    cmp w8, w9
+; CHECK-NEXT:    ccmp w10, w11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 3)
@@ -77,10 +75,8 @@
 ; CHECK-NEXT:    ldr w9, [x1]
 ; CHECK-NEXT:    ldrb w10, [x0, #4]
 ; CHECK-NEXT:    ldrb w11, [x1, #4]
-; CHECK-NEXT:    eor w8, w8, w9
-; CHECK-NEXT:    eor w9, w10, w11
-; CHECK-NEXT:    orr w8, w8, w9
-; CHECK-NEXT:    cmp w8, #0
+; CHECK-NEXT:    cmp w8, w9
+; CHECK-NEXT:    ccmp w10, w11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 5)
@@ -95,10 +91,8 @@
 ; CHECK-NEXT:    ldr w9, [x1]
 ; CHECK-NEXT:    ldrh w10, [x0, #4]
 ; CHECK-NEXT:    ldrh w11, [x1, #4]
-; CHECK-NEXT:    eor w8, w8, w9
-; CHECK-NEXT:    eor w9, w10, w11
-; CHECK-NEXT:    orr w8, w8, w9
-; CHECK-NEXT:    cmp w8, #0
+; CHECK-NEXT:    cmp w8, w9
+; CHECK-NEXT:    ccmp w10, w11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 6)
@@ -138,15 +132,12 @@
 define i1 @bcmp9(ptr %a, ptr %b) {
 ; CHECK-LABEL: bcmp9:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    ldrb w9, [x0, #8]
-; CHECK-NEXT:    ldrb w10, [x1, #8]
 ; CHECK-NEXT:    ldr x8, [x0]
-; CHECK-NEXT:    ldr x11, [x1]
-; CHECK-NEXT:    eor w9, w9, w10
-; CHECK-NEXT:    and x9, x9, #0xff
-; CHECK-NEXT:    eor x8, x8, x11
-; CHECK-NEXT:    orr x8, x8, x9
-; CHECK-NEXT:    cmp x8, #0
+; CHECK-NEXT:    ldr x9, [x1]
+; CHECK-NEXT:    ldrb w10, [x0, #8]
+; CHECK-NEXT:    ldrb w11, [x1, #8]
+; CHECK-NEXT:    cmp x8, x9
+; CHECK-NEXT:    ccmp x10, x11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 9)
@@ -157,15 +148,12 @@
 define i1 @bcmp10(ptr %a, ptr %b) {
 ; CHECK-LABEL: bcmp10:
 ; CHECK:       // %bb.0:
-; CHECK-NEXT:    ldrh w9, [x0, #8]
-; CHECK-NEXT:    ldrh w10, [x1, #8]
 ; CHECK-NEXT:    ldr x8, [x0]
-; CHECK-NEXT:    ldr x11, [x1]
-; CHECK-NEXT:    eor w9, w9, w10
-; CHECK-NEXT:    and x9, x9, #0xffff
-; CHECK-NEXT:    eor x8, x8, x11
-; CHECK-NEXT:    orr x8, x8, x9
-; CHECK-NEXT:    cmp x8, #0
+; CHECK-NEXT:    ldr x9, [x1]
+; CHECK-NEXT:    ldrh w10, [x0, #8]
+; CHECK-NEXT:    ldrh w11, [x1, #8]
+; CHECK-NEXT:    cmp x8, x9
+; CHECK-NEXT:    ccmp x10, x11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 10)
@@ -196,10 +184,8 @@
 ; CHECK-NEXT:    ldr x9, [x1]
 ; CHECK-NEXT:    ldr w10, [x0, #8]
 ; CHECK-NEXT:    ldr w11, [x1, #8]
-; CHECK-NEXT:    eor x8, x8, x9
-; CHECK-NEXT:    eor w9, w10, w11
-; CHECK-NEXT:    orr x8, x8, x9
-; CHECK-NEXT:    cmp x8, #0
+; CHECK-NEXT:    cmp x8, x9
+; CHECK-NEXT:    ccmp x10, x11, #0, eq
 ; CHECK-NEXT:    cset w0, eq
 ; CHECK-NEXT:    ret
   %cr = call i32 @bcmp(ptr %a, ptr %b, i64 12)
@@ -278,7 +264,7 @@
 ; CHECK-NEXT:    ldr w13, [x1, #16]
 ; CHECK-NEXT:    eor x8, x8, x10
 ; CHECK-NEXT:    eor x9, x9, x11
-; CHECK-NEXT:    eor w10, w12, w13
+; CHECK-NEXT:    eor x10, x12, x13
 ; CHECK-NEXT:    orr x8, x8, x9
 ; CHECK-NEXT:    orr x8, x8, x10
 ; CHECK-NEXT:    cmp x8, #0
@@ -322,7 +308,7 @@
 ; CHECK-NEXT:    eor x9, x9, x11
 ; CHECK-NEXT:    eor x10, x12, x13
 ; CHECK-NEXT:    orr x8, x8, x9
-; CHECK-NEXT:    eor w11, w14, w15
+; CHECK-NEXT:    eor x11, x14, x15
 ; CHECK-NEXT:    orr x9, x10, x11
 ; CHECK-NEXT:    orr x8, x8, x9
 ; CHECK-NEXT:    cmp x8, #0
@@ -342,14 +328,13 @@
 ; CHECK-NEXT:    ldp x14, x15, [x1, #16]
 ; CHECK-NEXT:    eor x8, x8, x10
 ; CHECK-NEXT:    eor x9, x9, x11
-; CHECK-NEXT:    ldrb w16, [x0, #32]
+; CHECK-NEXT:    ldrb w10, [x0, #32]
 ; CHECK-NEXT:    orr x8, x8, x9
-; CHECK-NEXT:    ldrb w17, [x1, #32]
-; CHECK-NEXT:    eor x10, x12, x14
-; CHECK-NEXT:    eor x11, x13, x15
-; CHECK-NEXT:    eor w12, w16, w17
-; CHECK-NEXT:    orr x9, x10, x11
-; CHECK-NEXT:    and x10, x12, #0xff
+; CHECK-NEXT:    ldrb w11, [x1, #32]
+; CHECK-NEXT:    eor x12, x12, x14
+; CHECK-NEXT:    eor x13, x13, x15
+; CHECK-NEXT:    orr x9, x12, x13
+; CHECK-NEXT:    eor x10, x10, x11
 ; CHECK-NEXT:    orr x8, x8, x9
 ; CHECK-NEXT:    orr x8, x8, x10
 ; CHECK-NEXT:    cmp x8, #0
diff --git a/llvm/test/CodeGen/X86/memcmp-more-load-pairs-x32.ll b/llvm/test/CodeGen/X86/memcmp-more-load-pairs-x32.ll
--- a/llvm/test/CodeGen/X86/memcmp-more-load-pairs-x32.ll
+++ b/llvm/test/CodeGen/X86/memcmp-more-load-pairs-x32.ll
@@ -178,8 +178,8 @@
 ; X86-NEXT:    movzwl (%ecx), %edx
 ; X86-NEXT:    xorw (%eax), %dx
 ; X86-NEXT:    movzbl 2(%ecx), %ecx
-; X86-NEXT:    xorb 2(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 2(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orw %dx, %ax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
@@ -311,8 +311,8 @@
 ; X86-NEXT:    movl (%ecx), %edx
 ; X86-NEXT:    xorl (%eax), %edx
 ; X86-NEXT:    movzbl 4(%ecx), %ecx
-; X86-NEXT:    xorb 4(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 4(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %edx, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
@@ -514,8 +514,8 @@
 ; X86-NEXT:    xorl 4(%eax), %esi
 ; X86-NEXT:    orl %edx, %esi
 ; X86-NEXT:    movzbl 8(%ecx), %ecx
-; X86-NEXT:    xorb 8(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 8(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %esi, %eax
 ; X86-NEXT:    sete %al
 ; X86-NEXT:    popl %esi
@@ -537,8 +537,8 @@
 ; X86-NEXT:    xorl 4(%eax), %esi
 ; X86-NEXT:    orl %edx, %esi
 ; X86-NEXT:    movzwl 8(%ecx), %ecx
-; X86-NEXT:    xorw 8(%eax), %cx
-; X86-NEXT:    movzwl %cx, %eax
+; X86-NEXT:    movzwl 8(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %esi, %eax
 ; X86-NEXT:    sete %al
 ; X86-NEXT:    popl %esi
@@ -645,8 +645,8 @@
 ; X86-NEXT:    movl 8(%edx), %esi
 ; X86-NEXT:    xorl 8(%ecx), %esi
 ; X86-NEXT:    movzbl 12(%edx), %edx
-; X86-NEXT:    xorb 12(%ecx), %dl
-; X86-NEXT:    movzbl %dl, %ecx
+; X86-NEXT:    movzbl 12(%ecx), %ecx
+; X86-NEXT:    xorl %edx, %ecx
 ; X86-NEXT:    orl %esi, %ecx
 ; X86-NEXT:    orl %eax, %ecx
 ; X86-NEXT:    sete %al
@@ -671,8 +671,8 @@
 ; X86-NEXT:    movl 8(%edx), %esi
 ; X86-NEXT:    xorl 8(%ecx), %esi
 ; X86-NEXT:    movzwl 12(%edx), %edx
-; X86-NEXT:    xorw 12(%ecx), %dx
-; X86-NEXT:    movzwl %dx, %ecx
+; X86-NEXT:    movzwl 12(%ecx), %ecx
+; X86-NEXT:    xorl %edx, %ecx
 ; X86-NEXT:    orl %esi, %ecx
 ; X86-NEXT:    orl %eax, %ecx
 ; X86-NEXT:    sete %al
diff --git a/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll b/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
--- a/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
+++ b/llvm/test/CodeGen/X86/memcmp-more-load-pairs.ll
@@ -167,9 +167,9 @@
 ; X64-NEXT:    movzwl (%rdi), %eax
 ; X64-NEXT:    xorw (%rsi), %ax
 ; X64-NEXT:    movzbl 2(%rdi), %ecx
-; X64-NEXT:    xorb 2(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orw %ax, %cx
+; X64-NEXT:    movzbl 2(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orw %ax, %dx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 3) nounwind
@@ -284,9 +284,9 @@
 ; X64-NEXT:    movl (%rdi), %eax
 ; X64-NEXT:    xorl (%rsi), %eax
 ; X64-NEXT:    movzbl 4(%rdi), %ecx
-; X64-NEXT:    xorb 4(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orl %eax, %ecx
+; X64-NEXT:    movzbl 4(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orl %eax, %edx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 5) nounwind
@@ -443,9 +443,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movzbl 8(%rdi), %ecx
-; X64-NEXT:    xorb 8(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movzbl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 9) nounwind
@@ -459,9 +459,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movzwl 8(%rdi), %ecx
-; X64-NEXT:    xorw 8(%rsi), %cx
-; X64-NEXT:    movzwl %cx, %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movzwl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 10) nounwind
@@ -490,8 +490,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    xorl 8(%rsi), %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 12) nounwind
diff --git a/llvm/test/CodeGen/X86/memcmp-optsize-x32.ll b/llvm/test/CodeGen/X86/memcmp-optsize-x32.ll
--- a/llvm/test/CodeGen/X86/memcmp-optsize-x32.ll
+++ b/llvm/test/CodeGen/X86/memcmp-optsize-x32.ll
@@ -106,9 +106,9 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movzwl (%ecx), %edx
 ; X86-NEXT:    xorw (%eax), %dx
-; X86-NEXT:    movb 2(%ecx), %cl
-; X86-NEXT:    xorb 2(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 2(%ecx), %ecx
+; X86-NEXT:    movzbl 2(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orw %dx, %ax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
@@ -197,9 +197,9 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl (%ecx), %edx
 ; X86-NEXT:    xorl (%eax), %edx
-; X86-NEXT:    movb 4(%ecx), %cl
-; X86-NEXT:    xorb 4(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 4(%ecx), %ecx
+; X86-NEXT:    movzbl 4(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %edx, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
diff --git a/llvm/test/CodeGen/X86/memcmp-optsize.ll b/llvm/test/CodeGen/X86/memcmp-optsize.ll
--- a/llvm/test/CodeGen/X86/memcmp-optsize.ll
+++ b/llvm/test/CodeGen/X86/memcmp-optsize.ll
@@ -94,10 +94,10 @@
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
 ; X64-NEXT:    xorw (%rsi), %ax
-; X64-NEXT:    movb 2(%rdi), %cl
-; X64-NEXT:    xorb 2(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orw %ax, %cx
+; X64-NEXT:    movzbl 2(%rdi), %ecx
+; X64-NEXT:    movzbl 2(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orw %ax, %dx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 3) nounwind
@@ -173,10 +173,10 @@
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl (%rdi), %eax
 ; X64-NEXT:    xorl (%rsi), %eax
-; X64-NEXT:    movb 4(%rdi), %cl
-; X64-NEXT:    xorb 4(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orl %eax, %ecx
+; X64-NEXT:    movzbl 4(%rdi), %ecx
+; X64-NEXT:    movzbl 4(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orl %eax, %edx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 5) nounwind
@@ -230,8 +230,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    xorl 8(%rsi), %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 12) nounwind
diff --git a/llvm/test/CodeGen/X86/memcmp-pgso-x32.ll b/llvm/test/CodeGen/X86/memcmp-pgso-x32.ll
--- a/llvm/test/CodeGen/X86/memcmp-pgso-x32.ll
+++ b/llvm/test/CodeGen/X86/memcmp-pgso-x32.ll
@@ -106,9 +106,9 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movzwl (%ecx), %edx
 ; X86-NEXT:    xorw (%eax), %dx
-; X86-NEXT:    movb 2(%ecx), %cl
-; X86-NEXT:    xorb 2(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 2(%ecx), %ecx
+; X86-NEXT:    movzbl 2(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orw %dx, %ax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
@@ -197,9 +197,9 @@
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X86-NEXT:    movl (%ecx), %edx
 ; X86-NEXT:    xorl (%eax), %edx
-; X86-NEXT:    movb 4(%ecx), %cl
-; X86-NEXT:    xorb 4(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 4(%ecx), %ecx
+; X86-NEXT:    movzbl 4(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %edx, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
diff --git a/llvm/test/CodeGen/X86/memcmp-pgso.ll b/llvm/test/CodeGen/X86/memcmp-pgso.ll
--- a/llvm/test/CodeGen/X86/memcmp-pgso.ll
+++ b/llvm/test/CodeGen/X86/memcmp-pgso.ll
@@ -94,10 +94,10 @@
 ; X64:       # %bb.0:
 ; X64-NEXT:    movzwl (%rdi), %eax
 ; X64-NEXT:    xorw (%rsi), %ax
-; X64-NEXT:    movb 2(%rdi), %cl
-; X64-NEXT:    xorb 2(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orw %ax, %cx
+; X64-NEXT:    movzbl 2(%rdi), %ecx
+; X64-NEXT:    movzbl 2(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orw %ax, %dx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 3) nounwind
@@ -173,10 +173,10 @@
 ; X64:       # %bb.0:
 ; X64-NEXT:    movl (%rdi), %eax
 ; X64-NEXT:    xorl (%rsi), %eax
-; X64-NEXT:    movb 4(%rdi), %cl
-; X64-NEXT:    xorb 4(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orl %eax, %ecx
+; X64-NEXT:    movzbl 4(%rdi), %ecx
+; X64-NEXT:    movzbl 4(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orl %eax, %edx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 5) nounwind
@@ -230,8 +230,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    xorl 8(%rsi), %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 12) nounwind
diff --git a/llvm/test/CodeGen/X86/memcmp-x32.ll b/llvm/test/CodeGen/X86/memcmp-x32.ll
--- a/llvm/test/CodeGen/X86/memcmp-x32.ll
+++ b/llvm/test/CodeGen/X86/memcmp-x32.ll
@@ -206,8 +206,8 @@
 ; X86-NEXT:    movzwl (%ecx), %edx
 ; X86-NEXT:    xorw (%eax), %dx
 ; X86-NEXT:    movzbl 2(%ecx), %ecx
-; X86-NEXT:    xorb 2(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 2(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orw %dx, %ax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
@@ -339,8 +339,8 @@
 ; X86-NEXT:    movl (%ecx), %edx
 ; X86-NEXT:    xorl (%eax), %edx
 ; X86-NEXT:    movzbl 4(%ecx), %ecx
-; X86-NEXT:    xorb 4(%eax), %cl
-; X86-NEXT:    movzbl %cl, %eax
+; X86-NEXT:    movzbl 4(%eax), %eax
+; X86-NEXT:    xorl %ecx, %eax
 ; X86-NEXT:    orl %edx, %eax
 ; X86-NEXT:    setne %al
 ; X86-NEXT:    retl
diff --git a/llvm/test/CodeGen/X86/memcmp.ll b/llvm/test/CodeGen/X86/memcmp.ll
--- a/llvm/test/CodeGen/X86/memcmp.ll
+++ b/llvm/test/CodeGen/X86/memcmp.ll
@@ -193,9 +193,9 @@
 ; X64-NEXT:    movzwl (%rdi), %eax
 ; X64-NEXT:    xorw (%rsi), %ax
 ; X64-NEXT:    movzbl 2(%rdi), %ecx
-; X64-NEXT:    xorb 2(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orw %ax, %cx
+; X64-NEXT:    movzbl 2(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orw %ax, %dx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 3) nounwind
@@ -310,9 +310,9 @@
 ; X64-NEXT:    movl (%rdi), %eax
 ; X64-NEXT:    xorl (%rsi), %eax
 ; X64-NEXT:    movzbl 4(%rdi), %ecx
-; X64-NEXT:    xorb 4(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orl %eax, %ecx
+; X64-NEXT:    movzbl 4(%rsi), %edx
+; X64-NEXT:    xorl %ecx, %edx
+; X64-NEXT:    orl %eax, %edx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 5) nounwind
@@ -469,9 +469,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movzbl 8(%rdi), %ecx
-; X64-NEXT:    xorb 8(%rsi), %cl
-; X64-NEXT:    movzbl %cl, %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movzbl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 9) nounwind
@@ -485,9 +485,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movzwl 8(%rdi), %ecx
-; X64-NEXT:    xorw 8(%rsi), %cx
-; X64-NEXT:    movzwl %cx, %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movzwl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 10) nounwind
@@ -516,8 +516,9 @@
 ; X64-NEXT:    movq (%rdi), %rax
 ; X64-NEXT:    xorq (%rsi), %rax
 ; X64-NEXT:    movl 8(%rdi), %ecx
-; X64-NEXT:    xorl 8(%rsi), %ecx
-; X64-NEXT:    orq %rax, %rcx
+; X64-NEXT:    movl 8(%rsi), %edx
+; X64-NEXT:    xorq %rcx, %rdx
+; X64-NEXT:    orq %rax, %rdx
 ; X64-NEXT:    setne %al
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(ptr %X, ptr %Y, i64 12) nounwind