From: Sanjay Patel <spatel@rotateright.com>
Date: Tue, 27 Jun 2017 21:46:34 +0000 (+0000)
Subject: [CGP] eliminate a sub instruction in memcmp expansion
X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=ca9df19568d44e4a5058ce809bcea83e33476e3c;p=llvm

[CGP] eliminate a sub instruction in memcmp expansion

As noted in D34071, there are some IR optimization opportunities that could be
handled by normal IR passes if this expansion wasn't happening so late in CGP.

Regardless of that, it seems wasteful to knowingly produce suboptimal IR here,
so I'm proposing this change:
  %s = sub i32 %x, %y
  %r = icmp ne %s, 0
    =>
  %r = icmp ne %x, %y

Changing the predicate to 'eq' mimics what InstCombine would do, so that's just
an efficiency improvement if we decide this expansion should happen sooner.

The fact that the PowerPC backend doesn't eliminate the 'subf.' might be
something for PPC folks to investigate separately.

Differential Revision: https://reviews.llvm.org/D34416


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@306471 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/CodeGen/CodeGenPrepare.cpp b/lib/CodeGen/CodeGenPrepare.cpp
index 7c9f0b2106f..474a214723e 100644
--- a/lib/CodeGen/CodeGenPrepare.cpp
+++ b/lib/CodeGen/CodeGenPrepare.cpp
@@ -1987,16 +1987,13 @@ void MemCmpExpansion::emitLoadCompareBlock(unsigned Index, unsigned LoadSize,
     ResBlock.PhiSrc2->addIncoming(LoadSrc2, LoadCmpBlocks[Index]);
   }
 
-  Value *Diff = Builder.CreateSub(LoadSrc1, LoadSrc2);
-
-  Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_NE, Diff,
-                                  ConstantInt::get(Diff->getType(), 0));
+  Value *Cmp = Builder.CreateICmp(ICmpInst::ICMP_EQ, LoadSrc1, LoadSrc2);
   BasicBlock *NextBB = (Index == (LoadCmpBlocks.size() - 1))
                            ? EndBlock
                            : LoadCmpBlocks[Index + 1];
   // Early exit branch if difference found to ResultBlock. Otherwise, continue
   // to next LoadCmpBlock or EndBlock.
-  BranchInst *CmpBr = BranchInst::Create(ResBlock.BB, NextBB, Cmp);
+  BranchInst *CmpBr = BranchInst::Create(NextBB, ResBlock.BB, Cmp);
   Builder.Insert(CmpBr);
 
   // Add a phi edge for the last LoadCmpBlock to Endblock with a value of 0
diff --git a/test/CodeGen/PowerPC/memCmpUsedInZeroEqualityComparison.ll b/test/CodeGen/PowerPC/memCmpUsedInZeroEqualityComparison.ll
index d398dfe7fc9..059665adc35 100644
--- a/test/CodeGen/PowerPC/memCmpUsedInZeroEqualityComparison.ll
+++ b/test/CodeGen/PowerPC/memCmpUsedInZeroEqualityComparison.ll
@@ -100,28 +100,26 @@ define signext i32 @zeroEqualityTest04() {
 ; CHECK-NEXT:    addi 5, 4, .LzeroEqualityTest02.buffer2@toc@l
 ; CHECK-NEXT:    ldbrx 3, 0, 6
 ; CHECK-NEXT:    ldbrx 4, 0, 5
-; CHECK-NEXT:    subf. 7, 4, 3
+; CHECK-NEXT:    cmpld 3, 4
 ; CHECK-NEXT:    bne 0, .LBB3_2
 ; CHECK-NEXT:  # BB#1: # %loadbb1
 ; CHECK-NEXT:    li 4, 8
 ; CHECK-NEXT:    ldbrx 3, 6, 4
 ; CHECK-NEXT:    ldbrx 4, 5, 4
-; CHECK-NEXT:    subf. 5, 4, 3
-; CHECK-NEXT:    beq 0, .LBB3_4
+; CHECK-NEXT:    li 5, 0
+; CHECK-NEXT:    cmpld 3, 4
+; CHECK-NEXT:    beq 0, .LBB3_3
 ; CHECK-NEXT:  .LBB3_2: # %res_block
 ; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    li 3, 1
+; CHECK-NEXT:    li 11, 1
 ; CHECK-NEXT:    li 12, -1
-; CHECK-NEXT:    isel 3, 12, 3, 0
+; CHECK-NEXT:    isel 5, 12, 11, 0
 ; CHECK-NEXT:  .LBB3_3: # %endblock
-; CHECK-NEXT:    cmpwi 3, 1
+; CHECK-NEXT:    cmpwi 5, 1
 ; CHECK-NEXT:    li 3, 0
 ; CHECK-NEXT:    li 4, 1
 ; CHECK-NEXT:    isel 3, 4, 3, 0
 ; CHECK-NEXT:    blr
-; CHECK-NEXT:  .LBB3_4:
-; CHECK-NEXT:    li 3, 0
-; CHECK-NEXT:    b .LBB3_3
   %call = tail call signext i32 @memcmp(i8* bitcast ([4 x i32]* @zeroEqualityTest02.buffer1 to i8*), i8* bitcast ([4 x i32]* @zeroEqualityTest02.buffer2 to i8*), i64 16)
   %not.cmp = icmp slt i32 %call, 1
   %. = zext i1 %not.cmp to i32
@@ -138,27 +136,25 @@ define signext i32 @zeroEqualityTest05() {
 ; CHECK-NEXT:    addi 5, 4, .LzeroEqualityTest03.buffer2@toc@l
 ; CHECK-NEXT:    ldbrx 3, 0, 6
 ; CHECK-NEXT:    ldbrx 4, 0, 5
-; CHECK-NEXT:    subf. 7, 4, 3
+; CHECK-NEXT:    cmpld 3, 4
 ; CHECK-NEXT:    bne 0, .LBB4_2
 ; CHECK-NEXT:  # BB#1: # %loadbb1
 ; CHECK-NEXT:    li 4, 8
 ; CHECK-NEXT:    ldbrx 3, 6, 4
 ; CHECK-NEXT:    ldbrx 4, 5, 4
-; CHECK-NEXT:    subf. 5, 4, 3
-; CHECK-NEXT:    beq 0, .LBB4_4
+; CHECK-NEXT:    li 5, 0
+; CHECK-NEXT:    cmpld 3, 4
+; CHECK-NEXT:    beq 0, .LBB4_3
 ; CHECK-NEXT:  .LBB4_2: # %res_block
 ; CHECK-NEXT:    cmpld 3, 4
-; CHECK-NEXT:    li 3, 1
+; CHECK-NEXT:    li 11, 1
 ; CHECK-NEXT:    li 12, -1
-; CHECK-NEXT:    isel 3, 12, 3, 0
+; CHECK-NEXT:    isel 5, 12, 11, 0
 ; CHECK-NEXT:  .LBB4_3: # %endblock
-; CHECK-NEXT:    srwi 3, 3, 31
+; CHECK-NEXT:    srwi 3, 5, 31
 ; CHECK-NEXT:    xori 3, 3, 1
 ; CHECK-NEXT:    clrldi 3, 3, 32
 ; CHECK-NEXT:    blr
-; CHECK-NEXT:  .LBB4_4:
-; CHECK-NEXT:    li 3, 0
-; CHECK-NEXT:    b .LBB4_3
   %call = tail call signext i32 @memcmp(i8* bitcast ([4 x i32]* @zeroEqualityTest03.buffer1 to i8*), i8* bitcast ([4 x i32]* @zeroEqualityTest03.buffer2 to i8*), i64 16)
   %call.lobit = lshr i32 %call, 31
   %call.lobit.not = xor i32 %call.lobit, 1
diff --git a/test/CodeGen/PowerPC/memcmp.ll b/test/CodeGen/PowerPC/memcmp.ll
index bae713cb207..039c48b2a96 100644
--- a/test/CodeGen/PowerPC/memcmp.ll
+++ b/test/CodeGen/PowerPC/memcmp.ll
@@ -13,10 +13,9 @@ entry:
 ; CHECK: ldbrx [[LOAD1:[0-9]+]]
 ; CHECK-NEXT: ldbrx [[LOAD2:[0-9]+]]
 ; CHECK-NEXT: li [[LI:[0-9]+]], 1
-; CHECK-NEXT: cmpld [[CMPLD:[0-9]+]], [[LOAD1]], [[LOAD2]]
-; CHECK-NEXT: subf. [[SUB:[0-9]+]], [[LOAD2]], [[LOAD1]]
+; CHECK-NEXT: cmpld [[LOAD1]], [[LOAD2]]
 ; CHECK-NEXT: li [[LI2:[0-9]+]], -1
-; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 4
+; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 0
 ; CHECK-NEXT: isel [[ISEL2:[0-9]+]], 0, [[ISEL]], 2
 ; CHECK-NEXT: extsw 3, [[ISEL2]]
 ; CHECK-NEXT: blr
@@ -35,10 +34,9 @@ entry:
 ; CHECK: lwbrx [[LOAD1:[0-9]+]]
 ; CHECK-NEXT: lwbrx [[LOAD2:[0-9]+]]
 ; CHECK-NEXT: li [[LI:[0-9]+]], 1
-; CHECK-NEXT: cmpld [[CMPLD:[0-9]+]], [[LOAD1]], [[LOAD2]]
-; CHECK-NEXT: subf. [[SUB:[0-9]+]], [[LOAD2]], [[LOAD1]]
+; CHECK-NEXT: cmpld [[LOAD1]], [[LOAD2]]
 ; CHECK-NEXT: li [[LI2:[0-9]+]], -1
-; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 4
+; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 0
 ; CHECK-NEXT: isel [[ISEL2:[0-9]+]], 0, [[ISEL]], 2
 ; CHECK-NEXT: extsw 3, [[ISEL2]]
 ; CHECK-NEXT: blr
@@ -57,10 +55,9 @@ entry:
 ; CHECK: lhbrx [[LOAD1:[0-9]+]]
 ; CHECK-NEXT: lhbrx [[LOAD2:[0-9]+]]
 ; CHECK-NEXT: li [[LI:[0-9]+]], 1
-; CHECK-NEXT: cmpld [[CMPLD:[0-9]+]], [[LOAD1]], [[LOAD2]]
-; CHECK-NEXT: subf. [[SUB:[0-9]+]], [[LOAD2]], [[LOAD1]]
+; CHECK-NEXT: cmpld [[LOAD1]], [[LOAD2]]
 ; CHECK-NEXT: li [[LI2:[0-9]+]], -1
-; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 4
+; CHECK-NEXT: isel [[ISEL:[0-9]+]], [[LI2]], [[LI]], 0
 ; CHECK-NEXT: isel [[ISEL2:[0-9]+]], 0, [[ISEL]], 2
 ; CHECK-NEXT: extsw 3, [[ISEL2]]
 ; CHECK-NEXT: blr
diff --git a/test/CodeGen/PowerPC/memcmpIR.ll b/test/CodeGen/PowerPC/memcmpIR.ll
index f052cc258df..044c8d5aa33 100644
--- a/test/CodeGen/PowerPC/memcmpIR.ll
+++ b/test/CodeGen/PowerPC/memcmpIR.ll
@@ -3,48 +3,47 @@
 
 define signext i32 @test1(i32* nocapture readonly %buffer1, i32* nocapture readonly %buffer2)  {
 entry:
+  ; CHECK-LABEL: @test1(
   ; CHECK: [[LOAD1:%[0-9]+]] = load i64, i64*
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64*
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD2]])
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[BSWAP1]], [[BSWAP2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[BSWAP1]], [[BSWAP2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %loadbb1, label %res_block
 
   ; CHECK-LABEL: res_block:{{.*}}
   ; CHECK: [[ICMP2:%[0-9]+]] = icmp ult i64
   ; CHECK-NEXT: [[SELECT:%[0-9]+]] = select i1 [[ICMP2]], i32 -1, i32 1
   ; CHECK-NEXT: br label %endblock
 
+  ; CHECK-LABEL: loadbb1:{{.*}}
   ; CHECK: [[GEP1:%[0-9]+]] = getelementptr i64, i64* {{.*}}, i64 1
   ; CHECK-NEXT: [[GEP2:%[0-9]+]] = getelementptr i64, i64* {{.*}}, i64 1
   ; CHECK-NEXT: [[LOAD1:%[0-9]+]] = load i64, i64* [[GEP1]]
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64* [[GEP2]]
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD2]])
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[BSWAP1]], [[BSWAP2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label %endblock
-
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[BSWAP1]], [[BSWAP2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %endblock, label %res_block
 
+  ; CHECK-BE-LABEL: @test1(
   ; CHECK-BE: [[LOAD1:%[0-9]+]] = load i64, i64*
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64*
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[LOAD1]], [[LOAD2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[LOAD1]], [[LOAD2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %loadbb1, label %res_block
 
   ; CHECK-BE-LABEL: res_block:{{.*}}
   ; CHECK-BE: [[ICMP2:%[0-9]+]] = icmp ult i64
   ; CHECK-BE-NEXT: [[SELECT:%[0-9]+]] = select i1 [[ICMP2]], i32 -1, i32 1
   ; CHECK-BE-NEXT: br label %endblock
 
+  ; CHECK-BE-LABEL: loadbb1:{{.*}}
   ; CHECK-BE: [[GEP1:%[0-9]+]] = getelementptr i64, i64* {{.*}}, i64 1
   ; CHECK-BE-NEXT: [[GEP2:%[0-9]+]] = getelementptr i64, i64* {{.*}}, i64 1
   ; CHECK-BE-NEXT: [[LOAD1:%[0-9]+]] = load i64, i64* [[GEP1]]
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64* [[GEP2]]
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[LOAD1]], [[LOAD2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label %endblock
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[LOAD1]], [[LOAD2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %endblock, label %res_block
 
   %0 = bitcast i32* %buffer1 to i8*
   %1 = bitcast i32* %buffer2 to i8*
@@ -55,28 +54,28 @@ entry:
 declare signext i32 @memcmp(i8* nocapture, i8* nocapture, i64) local_unnamed_addr #1
 
 define signext i32 @test2(i32* nocapture readonly %buffer1, i32* nocapture readonly %buffer2)  {
+  ; CHECK-LABEL: @test2(
   ; CHECK: [[LOAD1:%[0-9]+]] = load i32, i32*
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i32, i32*
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i32 @llvm.bswap.i32(i32 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i32 @llvm.bswap.i32(i32 [[LOAD2]])
   ; CHECK-NEXT: [[ZEXT1:%[0-9]+]] = zext i32 [[BSWAP1]] to i64
   ; CHECK-NEXT: [[ZEXT2:%[0-9]+]] = zext i32 [[BSWAP2]] to i64
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label %endblock
-
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %endblock, label %res_block
+  
   ; CHECK-LABEL: res_block:{{.*}}
   ; CHECK: [[ICMP2:%[0-9]+]] = icmp ult i64
   ; CHECK-NEXT: [[SELECT:%[0-9]+]] = select i1 [[ICMP2]], i32 -1, i32 1
   ; CHECK-NEXT: br label %endblock
 
+  ; CHECK-BE-LABEL: @test2(
   ; CHECK-BE: [[LOAD1:%[0-9]+]] = load i32, i32*
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i32, i32*
   ; CHECK-BE-NEXT: [[ZEXT1:%[0-9]+]] = zext i32 [[LOAD1]] to i64
   ; CHECK-BE-NEXT: [[ZEXT2:%[0-9]+]] = zext i32 [[LOAD2]] to i64
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label %endblock
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %endblock, label %res_block
 
   ; CHECK-BE-LABEL: res_block:{{.*}}
   ; CHECK-BE: [[ICMP2:%[0-9]+]] = icmp ult i64
@@ -95,35 +94,35 @@ define signext i32 @test3(i32* nocapture readonly %buffer1, i32* nocapture reado
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64*
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i64 @llvm.bswap.i64(i64 [[LOAD2]])
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[BSWAP1]], [[BSWAP2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[BSWAP1]], [[BSWAP2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %loadbb1, label %res_block
 
   ; CHECK-LABEL: res_block:{{.*}}
   ; CHECK: [[ICMP2:%[0-9]+]] = icmp ult i64
   ; CHECK-NEXT: [[SELECT:%[0-9]+]] = select i1 [[ICMP2]], i32 -1, i32 1
   ; CHECK-NEXT: br label %endblock
 
+  ; CHECK-LABEL: loadbb1:{{.*}}
   ; CHECK: [[LOAD1:%[0-9]+]] = load i32, i32*
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i32, i32*
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i32 @llvm.bswap.i32(i32 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i32 @llvm.bswap.i32(i32 [[LOAD2]])
   ; CHECK-NEXT: [[ZEXT1:%[0-9]+]] = zext i32 [[BSWAP1]] to i64
   ; CHECK-NEXT: [[ZEXT2:%[0-9]+]] = zext i32 [[BSWAP2]] to i64
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %loadbb2, label %res_block
 
+  ; CHECK-LABEL: loadbb2:{{.*}}
   ; CHECK: [[LOAD1:%[0-9]+]] = load i16, i16*
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i16, i16*
   ; CHECK-NEXT: [[BSWAP1:%[0-9]+]] = call i16 @llvm.bswap.i16(i16 [[LOAD1]])
   ; CHECK-NEXT: [[BSWAP2:%[0-9]+]] = call i16 @llvm.bswap.i16(i16 [[LOAD2]])
   ; CHECK-NEXT: [[ZEXT1:%[0-9]+]] = zext i16 [[BSWAP1]] to i64
   ; CHECK-NEXT: [[ZEXT2:%[0-9]+]] = zext i16 [[BSWAP2]] to i64
-  ; CHECK-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-NEXT:  br i1 [[ICMP]], label %loadbb3, label %res_block
 
+  ; CHECK-LABEL: loadbb3:{{.*}}
   ; CHECK: [[LOAD1:%[0-9]+]] = load i8, i8*
   ; CHECK-NEXT: [[LOAD2:%[0-9]+]] = load i8, i8*
   ; CHECK-NEXT: [[ZEXT1:%[0-9]+]] = zext i8 [[LOAD1]] to i32
@@ -133,9 +132,8 @@ define signext i32 @test3(i32* nocapture readonly %buffer1, i32* nocapture reado
 
   ; CHECK-BE: [[LOAD1:%[0-9]+]] = load i64, i64*
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i64, i64*
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[LOAD1]], [[LOAD2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[LOAD1]], [[LOAD2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %loadbb1, label %res_block
 
   ; CHECK-BE-LABEL: res_block:{{.*}}
   ; CHECK-BE: [[ICMP2:%[0-9]+]] = icmp ult i64
@@ -146,17 +144,15 @@ define signext i32 @test3(i32* nocapture readonly %buffer1, i32* nocapture reado
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i32, i32*
   ; CHECK-BE-NEXT: [[ZEXT1:%[0-9]+]] = zext i32 [[LOAD1]] to i64
   ; CHECK-BE-NEXT: [[ZEXT2:%[0-9]+]] = zext i32 [[LOAD2]] to i64
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %loadbb2, label %res_block
 
   ; CHECK-BE: [[LOAD1:%[0-9]+]] = load i16, i16*
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i16, i16*
   ; CHECK-BE-NEXT: [[ZEXT1:%[0-9]+]] = zext i16 [[LOAD1]] to i64
   ; CHECK-BE-NEXT: [[ZEXT2:%[0-9]+]] = zext i16 [[LOAD2]] to i64
-  ; CHECK-BE-NEXT: [[SUB:%[0-9]+]] = sub i64 [[ZEXT1]], [[ZEXT2]]
-  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp ne i64 [[SUB]], 0
-  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %res_block, label
+  ; CHECK-BE-NEXT: [[ICMP:%[0-9]+]] = icmp eq i64 [[ZEXT1]], [[ZEXT2]]
+  ; CHECK-BE-NEXT:  br i1 [[ICMP]], label %loadbb3, label %res_block
 
   ; CHECK-BE: [[LOAD1:%[0-9]+]] = load i8, i8*
   ; CHECK-BE-NEXT: [[LOAD2:%[0-9]+]] = load i8, i8*
diff --git a/test/CodeGen/X86/memcmp.ll b/test/CodeGen/X86/memcmp.ll
index 9d26aee2e8b..608f68fb479 100644
--- a/test/CodeGen/X86/memcmp.ll
+++ b/test/CodeGen/X86/memcmp.ll
@@ -20,18 +20,16 @@ define i32 @length2(i8* %X, i8* %Y) nounwind {
 ; X32-NEXT:    rolw $8, %cx
 ; X32-NEXT:    rolw $8, %ax
 ; X32-NEXT:    movzwl %cx, %ecx
-; X32-NEXT:    movzwl %ax, %eax
-; X32-NEXT:    cmpl %eax, %ecx
-; X32-NEXT:    je .LBB0_1
-; X32-NEXT:  # BB#2: # %res_block
+; X32-NEXT:    movzwl %ax, %edx
+; X32-NEXT:    xorl %eax, %eax
+; X32-NEXT:    cmpl %edx, %ecx
+; X32-NEXT:    je .LBB0_3
+; X32-NEXT:  # BB#1: # %res_block
 ; X32-NEXT:    movl $-1, %eax
-; X32-NEXT:    jb .LBB0_4
-; X32-NEXT:  # BB#3: # %res_block
+; X32-NEXT:    jb .LBB0_3
+; X32-NEXT:  # BB#2: # %res_block
 ; X32-NEXT:    movl $1, %eax
-; X32-NEXT:  .LBB0_4: # %endblock
-; X32-NEXT:    retl
-; X32-NEXT:  .LBB0_1:
-; X32-NEXT:    xorl %eax, %eax
+; X32-NEXT:  .LBB0_3: # %endblock
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: length2:
@@ -40,17 +38,16 @@ define i32 @length2(i8* %X, i8* %Y) nounwind {
 ; X64-NEXT:    movzwl (%rsi), %ecx
 ; X64-NEXT:    rolw $8, %ax
 ; X64-NEXT:    rolw $8, %cx
-; X64-NEXT:    movzwl %ax, %eax
+; X64-NEXT:    movzwl %ax, %edx
 ; X64-NEXT:    movzwl %cx, %ecx
-; X64-NEXT:    cmpq %rcx, %rax
-; X64-NEXT:    je .LBB0_1
-; X64-NEXT:  # BB#2: # %res_block
+; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:    cmpq %rcx, %rdx
+; X64-NEXT:    je .LBB0_2
+; X64-NEXT:  # BB#1: # %res_block
 ; X64-NEXT:    movl $-1, %ecx
 ; X64-NEXT:    movl $1, %eax
 ; X64-NEXT:    cmovbl %ecx, %eax
-; X64-NEXT:    retq
-; X64-NEXT:  .LBB0_1:
-; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:  .LBB0_2: # %endblock
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* %Y, i64 2) nounwind
   ret i32 %m
@@ -176,37 +173,34 @@ define i32 @length4(i8* %X, i8* %Y) nounwind {
 ; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X32-NEXT:    movl {{[0-9]+}}(%esp), %ecx
 ; X32-NEXT:    movl (%ecx), %ecx
-; X32-NEXT:    movl (%eax), %eax
+; X32-NEXT:    movl (%eax), %edx
 ; X32-NEXT:    bswapl %ecx
-; X32-NEXT:    bswapl %eax
-; X32-NEXT:    cmpl %eax, %ecx
-; X32-NEXT:    je .LBB6_1
-; X32-NEXT:  # BB#2: # %res_block
+; X32-NEXT:    bswapl %edx
+; X32-NEXT:    xorl %eax, %eax
+; X32-NEXT:    cmpl %edx, %ecx
+; X32-NEXT:    je .LBB6_3
+; X32-NEXT:  # BB#1: # %res_block
 ; X32-NEXT:    movl $-1, %eax
-; X32-NEXT:    jb .LBB6_4
-; X32-NEXT:  # BB#3: # %res_block
+; X32-NEXT:    jb .LBB6_3
+; X32-NEXT:  # BB#2: # %res_block
 ; X32-NEXT:    movl $1, %eax
-; X32-NEXT:  .LBB6_4: # %endblock
-; X32-NEXT:    retl
-; X32-NEXT:  .LBB6_1:
-; X32-NEXT:    xorl %eax, %eax
+; X32-NEXT:  .LBB6_3: # %endblock
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: length4:
 ; X64:       # BB#0: # %loadbb
-; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    movl (%rsi), %ecx
-; X64-NEXT:    bswapl %eax
+; X64-NEXT:    movl (%rdi), %ecx
+; X64-NEXT:    movl (%rsi), %edx
 ; X64-NEXT:    bswapl %ecx
-; X64-NEXT:    cmpq %rcx, %rax
-; X64-NEXT:    je .LBB6_1
-; X64-NEXT:  # BB#2: # %res_block
+; X64-NEXT:    bswapl %edx
+; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:    cmpq %rdx, %rcx
+; X64-NEXT:    je .LBB6_2
+; X64-NEXT:  # BB#1: # %res_block
 ; X64-NEXT:    movl $-1, %ecx
 ; X64-NEXT:    movl $1, %eax
 ; X64-NEXT:    cmovbl %ecx, %eax
-; X64-NEXT:    retq
-; X64-NEXT:  .LBB6_1:
-; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:  .LBB6_2: # %endblock
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* %Y, i64 4) nounwind
   ret i32 %m
@@ -310,19 +304,18 @@ define i32 @length8(i8* %X, i8* %Y) nounwind {
 ;
 ; X64-LABEL: length8:
 ; X64:       # BB#0: # %loadbb
-; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    movq (%rsi), %rcx
-; X64-NEXT:    bswapq %rax
+; X64-NEXT:    movq (%rdi), %rcx
+; X64-NEXT:    movq (%rsi), %rdx
 ; X64-NEXT:    bswapq %rcx
-; X64-NEXT:    cmpq %rcx, %rax
-; X64-NEXT:    je .LBB11_1
-; X64-NEXT:  # BB#2: # %res_block
+; X64-NEXT:    bswapq %rdx
+; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:    cmpq %rdx, %rcx
+; X64-NEXT:    je .LBB11_2
+; X64-NEXT:  # BB#1: # %res_block
 ; X64-NEXT:    movl $-1, %ecx
 ; X64-NEXT:    movl $1, %eax
 ; X64-NEXT:    cmovbl %ecx, %eax
-; X64-NEXT:    retq
-; X64-NEXT:  .LBB11_1:
-; X64-NEXT:    xorl %eax, %eax
+; X64-NEXT:  .LBB11_2: # %endblock
 ; X64-NEXT:    retq
   %m = tail call i32 @memcmp(i8* %X, i8* %Y, i64 8) nounwind
   ret i32 %m
diff --git a/test/Transforms/CodeGenPrepare/X86/memcmp.ll b/test/Transforms/CodeGenPrepare/X86/memcmp.ll
index 690e714af26..b0335ee3450 100644
--- a/test/Transforms/CodeGenPrepare/X86/memcmp.ll
+++ b/test/Transforms/CodeGenPrepare/X86/memcmp.ll
@@ -14,15 +14,14 @@ define i32 @cmp2(i8* nocapture readonly %x, i8* nocapture readonly %y)  {
 ; X32-NEXT:    [[TMP5:%.*]] = call i16 @llvm.bswap.i16(i16 [[TMP3]])
 ; X32-NEXT:    [[TMP6:%.*]] = zext i16 [[TMP4]] to i32
 ; X32-NEXT:    [[TMP7:%.*]] = zext i16 [[TMP5]] to i32
-; X32-NEXT:    [[TMP8:%.*]] = sub i32 [[TMP6]], [[TMP7]]
-; X32-NEXT:    [[TMP9:%.*]] = icmp ne i32 [[TMP8]], 0
-; X32-NEXT:    br i1 [[TMP9]], label %res_block, label %endblock
+; X32-NEXT:    [[TMP8:%.*]] = icmp eq i32 [[TMP6]], [[TMP7]]
+; X32-NEXT:    br i1 [[TMP8]], label %endblock, label %res_block
 ; X32:       res_block:
-; X32-NEXT:    [[TMP10:%.*]] = icmp ult i32 [[TMP6]], [[TMP7]]
-; X32-NEXT:    [[TMP11:%.*]] = select i1 [[TMP10]], i32 -1, i32 1
+; X32-NEXT:    [[TMP9:%.*]] = icmp ult i32 [[TMP6]], [[TMP7]]
+; X32-NEXT:    [[TMP10:%.*]] = select i1 [[TMP9]], i32 -1, i32 1
 ; X32-NEXT:    br label %endblock
 ; X32:       endblock:
-; X32-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP11]], %res_block ]
+; X32-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP10]], %res_block ]
 ; X32-NEXT:    ret i32 [[PHI_RES]]
 ;
 ; X64-LABEL: @cmp2(
@@ -35,15 +34,14 @@ define i32 @cmp2(i8* nocapture readonly %x, i8* nocapture readonly %y)  {
 ; X64-NEXT:    [[TMP5:%.*]] = call i16 @llvm.bswap.i16(i16 [[TMP3]])
 ; X64-NEXT:    [[TMP6:%.*]] = zext i16 [[TMP4]] to i64
 ; X64-NEXT:    [[TMP7:%.*]] = zext i16 [[TMP5]] to i64
-; X64-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP6]], [[TMP7]]
-; X64-NEXT:    [[TMP9:%.*]] = icmp ne i64 [[TMP8]], 0
-; X64-NEXT:    br i1 [[TMP9]], label %res_block, label %endblock
+; X64-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[TMP6]], [[TMP7]]
+; X64-NEXT:    br i1 [[TMP8]], label %endblock, label %res_block
 ; X64:       res_block:
-; X64-NEXT:    [[TMP10:%.*]] = icmp ult i64 [[TMP6]], [[TMP7]]
-; X64-NEXT:    [[TMP11:%.*]] = select i1 [[TMP10]], i32 -1, i32 1
+; X64-NEXT:    [[TMP9:%.*]] = icmp ult i64 [[TMP6]], [[TMP7]]
+; X64-NEXT:    [[TMP10:%.*]] = select i1 [[TMP9]], i32 -1, i32 1
 ; X64-NEXT:    br label %endblock
 ; X64:       endblock:
-; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP11]], %res_block ]
+; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP10]], %res_block ]
 ; X64-NEXT:    ret i32 [[PHI_RES]]
 ;
   %call = tail call i32 @memcmp(i8* %x, i8* %y, i64 2)
@@ -68,15 +66,14 @@ define i32 @cmp4(i8* nocapture readonly %x, i8* nocapture readonly %y)  {
 ; X32-NEXT:    [[TMP3:%.*]] = load i32, i32* [[TMP1]]
 ; X32-NEXT:    [[TMP4:%.*]] = call i32 @llvm.bswap.i32(i32 [[TMP2]])
 ; X32-NEXT:    [[TMP5:%.*]] = call i32 @llvm.bswap.i32(i32 [[TMP3]])
-; X32-NEXT:    [[TMP6:%.*]] = sub i32 [[TMP4]], [[TMP5]]
-; X32-NEXT:    [[TMP7:%.*]] = icmp ne i32 [[TMP6]], 0
-; X32-NEXT:    br i1 [[TMP7]], label %res_block, label %endblock
+; X32-NEXT:    [[TMP6:%.*]] = icmp eq i32 [[TMP4]], [[TMP5]]
+; X32-NEXT:    br i1 [[TMP6]], label %endblock, label %res_block
 ; X32:       res_block:
-; X32-NEXT:    [[TMP8:%.*]] = icmp ult i32 [[TMP4]], [[TMP5]]
-; X32-NEXT:    [[TMP9:%.*]] = select i1 [[TMP8]], i32 -1, i32 1
+; X32-NEXT:    [[TMP7:%.*]] = icmp ult i32 [[TMP4]], [[TMP5]]
+; X32-NEXT:    [[TMP8:%.*]] = select i1 [[TMP7]], i32 -1, i32 1
 ; X32-NEXT:    br label %endblock
 ; X32:       endblock:
-; X32-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP9]], %res_block ]
+; X32-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP8]], %res_block ]
 ; X32-NEXT:    ret i32 [[PHI_RES]]
 ;
 ; X64-LABEL: @cmp4(
@@ -89,15 +86,14 @@ define i32 @cmp4(i8* nocapture readonly %x, i8* nocapture readonly %y)  {
 ; X64-NEXT:    [[TMP5:%.*]] = call i32 @llvm.bswap.i32(i32 [[TMP3]])
 ; X64-NEXT:    [[TMP6:%.*]] = zext i32 [[TMP4]] to i64
 ; X64-NEXT:    [[TMP7:%.*]] = zext i32 [[TMP5]] to i64
-; X64-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP6]], [[TMP7]]
-; X64-NEXT:    [[TMP9:%.*]] = icmp ne i64 [[TMP8]], 0
-; X64-NEXT:    br i1 [[TMP9]], label %res_block, label %endblock
+; X64-NEXT:    [[TMP8:%.*]] = icmp eq i64 [[TMP6]], [[TMP7]]
+; X64-NEXT:    br i1 [[TMP8]], label %endblock, label %res_block
 ; X64:       res_block:
-; X64-NEXT:    [[TMP10:%.*]] = icmp ult i64 [[TMP6]], [[TMP7]]
-; X64-NEXT:    [[TMP11:%.*]] = select i1 [[TMP10]], i32 -1, i32 1
+; X64-NEXT:    [[TMP9:%.*]] = icmp ult i64 [[TMP6]], [[TMP7]]
+; X64-NEXT:    [[TMP10:%.*]] = select i1 [[TMP9]], i32 -1, i32 1
 ; X64-NEXT:    br label %endblock
 ; X64:       endblock:
-; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP11]], %res_block ]
+; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP10]], %res_block ]
 ; X64-NEXT:    ret i32 [[PHI_RES]]
 ;
   %call = tail call i32 @memcmp(i8* %x, i8* %y, i64 4)
@@ -144,15 +140,14 @@ define i32 @cmp8(i8* nocapture readonly %x, i8* nocapture readonly %y)  {
 ; X64-NEXT:    [[TMP3:%.*]] = load i64, i64* [[TMP1]]
 ; X64-NEXT:    [[TMP4:%.*]] = call i64 @llvm.bswap.i64(i64 [[TMP2]])
 ; X64-NEXT:    [[TMP5:%.*]] = call i64 @llvm.bswap.i64(i64 [[TMP3]])
-; X64-NEXT:    [[TMP6:%.*]] = sub i64 [[TMP4]], [[TMP5]]
-; X64-NEXT:    [[TMP7:%.*]] = icmp ne i64 [[TMP6]], 0
-; X64-NEXT:    br i1 [[TMP7]], label %res_block, label %endblock
+; X64-NEXT:    [[TMP6:%.*]] = icmp eq i64 [[TMP4]], [[TMP5]]
+; X64-NEXT:    br i1 [[TMP6]], label %endblock, label %res_block
 ; X64:       res_block:
-; X64-NEXT:    [[TMP8:%.*]] = icmp ult i64 [[TMP4]], [[TMP5]]
-; X64-NEXT:    [[TMP9:%.*]] = select i1 [[TMP8]], i32 -1, i32 1
+; X64-NEXT:    [[TMP7:%.*]] = icmp ult i64 [[TMP4]], [[TMP5]]
+; X64-NEXT:    [[TMP8:%.*]] = select i1 [[TMP7]], i32 -1, i32 1
 ; X64-NEXT:    br label %endblock
 ; X64:       endblock:
-; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP9]], %res_block ]
+; X64-NEXT:    [[PHI_RES:%.*]] = phi i32 [ 0, %loadbb ], [ [[TMP8]], %res_block ]
 ; X64-NEXT:    ret i32 [[PHI_RES]]
 ;
   %call = tail call i32 @memcmp(i8* %x, i8* %y, i64 8)