Rebased PNaCl localmods in LLVM to 223109

test/NaCl/Bitcode/ptrtoint-elide.ll

+; Test how we handle eliding ptrtoint instructions.
+
+; RUN: llvm-as < %s | pnacl-freeze \
+; RUN:              | pnacl-bcanalyzer -dump-records \
+; RUN:              | FileCheck %s -check-prefix=PF2
+
+; RUN: llvm-as < %s | pnacl-freeze -allow-local-symbol-tables \
+; RUN:              | pnacl-thaw -allow-local-symbol-tables \
+; RUN:              | llvm-dis - | FileCheck %s -check-prefix=TD2
+
+; ------------------------------------------------------
+
+declare i32 @bar(i32)
+
+@bytes = internal global [4 x i8] c"abcd"
+
+; ------------------------------------------------------
+
+; Show simple case where we use ptrtoint
+define void @AllocCastSimple() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = bitcast [4 x i8]* @bytes to i32*
+  store i32 %2, i32* %3, align 1
+  ret void
+}
+
+; TD2:      define void @AllocCastSimple() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %3 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=3 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Same as above, but with the cast order changed. Shows
+; that we always inject casts back in a fixed order. Hence,
+; the casts will be reversed.
+define void @AllocCastSimpleReversed() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = bitcast [4 x i8]* @bytes to i32*
+  %3 = ptrtoint i8* %1 to i32
+  store i32 %3, i32* %2, align 1
+  ret void
+}
+
+; TD2:      define void @AllocCastSimpleReversed() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %3 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=3 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show case where we delete ptrtoint because they aren't used.
+define void @AllocCastDelete() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = alloca i8, i32 4, align 8
+  %4 = ptrtoint i8* %3 to i32
+  ret void
+}
+
+; TD2:      define void @AllocCastDelete() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = alloca i8, i32 4, align 8
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show case where we have optimized the ptrtoint (and bitcast) into a
+; single instruction, and will only be inserted before the first use
+; in the block.
+define void @AllocCastOpt() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = bitcast [4 x i8]* @bytes to i32*
+  %3 = ptrtoint i8* %1 to i32
+  store i32 %3, i32* %2, align 1
+  store i32 %3, i32* %2, align 1
+  ret void
+}
+
+; TD2:      define void @AllocCastOpt() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %3 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=3 op1=1 op2=1/>
+; PF2-NEXT:     <INST_STORE op0=3 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show case where ptrtoint (and bitcast) for store are not immediately
+; before the store, the casts will be moved to the store.
+define void @AllocCastMove(i32) {
+  %2 = alloca i8, i32 4, align 8
+  %3 = bitcast [4 x i8]* @bytes to i32*
+  %4 = ptrtoint i8* %2 to i32
+  %5 = add i32 %0, 1
+  store i32 %4, i32* %3, align 1
+  ret void
+}
+
+; TD2:      define void @AllocCastMove(i32) {
+; TD2-NEXT:   %2 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %3 = add i32 %0, 1
+; TD2-NEXT:   %4 = ptrtoint i8* %2 to i32
+; TD2-NEXT:   %5 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %4, i32* %5, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=2 op2=0/>
+; PF2-NEXT:     <INST_STORE op0=6 op1=2 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show case where ptrtoint on global variable is merged in a store, and
+; order is kept.
+define void @StoreGlobal() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint [4 x i8]* @bytes to i32
+  %3 = bitcast i8* %1 to i32*
+  store i32 %2, i32* %3, align 1
+  ret void
+}
+
+; TD2:      define void @StoreGlobal() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %3 = bitcast i8* %1 to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=1 op1=3 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Same as above, but with cast order reversed.
+define void @StoreGlobalCastsReversed() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = bitcast i8* %1 to i32*
+  %3 = ptrtoint [4 x i8]* @bytes to i32
+  store i32 %3, i32* %2, align 1
+  ret void
+}
+
+; TD2:      define void @StoreGlobalCastsReversed() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %3 = bitcast i8* %1 to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=1 op1=3 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we will move the ptrtoint of a global to the use.
+define i32 @StoreGlobalMovePtr2Int() {
+  %1 = ptrtoint [4 x i8]* @bytes to i32
+  %2 = alloca i8, i32 4, align 8
+  %3 = bitcast i8* %2 to i32*
+  store i32 %1, i32* %3, align 1
+  ret i32 0
+}
+
+; TD2:      define i32 @StoreGlobalMovePtr2Int() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %3 = bitcast i8* %1 to i32*
+; TD2-NEXT:   store i32 %2, i32* %3, align 1
+; TD2-NEXT:   ret i32 0
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_STORE op0=1 op1=4 op2=1/>
+; PF2-NEXT:     <INST_RET op0=2/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we handle add instructions with pointer casts.
+define void @CastAddAlloca() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+
+  ; Simple add.
+  %3 = add i32 1, 2
+
+  ; Cast first.
+  %4 = add i32 %2, 2
+
+  ; Cast second.
+  %5 = add i32 1, %2
+
+  ; Cast both.
+  %6 = add i32 %2, %2
+
+  ret void
+}
+
+; TD2:      define void @CastAddAlloca() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = add i32 1, 2
+; TD2-NEXT:   %3 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %4 = add i32 %3, 2
+; TD2-NEXT:   %5 = add i32 1, %3
+; TD2-NEXT:   %6 = add i32 %3, %3
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=3 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=2 op1=4 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=6 op1=3 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=4 op2=0/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we handle add instructions with pointer casts.
+define void @CastAddGlobal() {
+  %1 = ptrtoint [4 x i8]* @bytes to i32
+
+  ; Simple Add.
+  %2 = add i32 1, 2
+
+  ; Cast first.
+  %3 = add i32 %1, 2
+
+  ; Cast Second.
+  %4 = add i32 1, %1
+
+  ; Cast both.
+  %5 = add i32 %1, %1
+  ret void
+}
+
+; TD2:      define void @CastAddGlobal() {
+; TD2-NEXT:   %1 = add i32 1, 2
+; TD2-NEXT:   %2 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %3 = add i32 %2, 2
+; TD2-NEXT:   %4 = add i32 1, %2
+; TD2-NEXT:   %5 = add i32 %2, %2
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_BINOP op0=2 op1=1 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=2 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=5 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=6 op1=6 op2=0/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we can handle pointer conversions for other scalar binary operators.
+define void @CastBinop() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = ptrtoint [4 x i8]* @bytes to i32
+  %4 = sub i32 %2, %3
+  %5 = mul i32 %2, %3
+  %6 = udiv i32 %2, %3
+  %7 = urem i32 %2, %3
+  %8 = srem i32 %2, %3
+  %9 = shl i32 %2, %3
+  %10 = lshr i32 %2, %3
+  %11 = ashr i32 %2, %3
+  %12 = and i32 %2, %3
+  %13 = or i32 %2, %3
+  %14 = xor i32 %2, %3
+  ret void
+}
+
+; TD2:      define void @CastBinop() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %3 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %4 = sub i32 %2, %3
+; TD2-NEXT:   %5 = mul i32 %2, %3
+; TD2-NEXT:   %6 = udiv i32 %2, %3
+; TD2-NEXT:   %7 = urem i32 %2, %3
+; TD2-NEXT:   %8 = srem i32 %2, %3
+; TD2-NEXT:   %9 = shl i32 %2, %3
+; TD2-NEXT:   %10 = lshr i32 %2, %3
+; TD2-NEXT:   %11 = ashr i32 %2, %3
+; TD2-NEXT:   %12 = and i32 %2, %3
+; TD2-NEXT:   %13 = or i32 %2, %3
+; TD2-NEXT:   %14 = xor i32 %2, %3
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_BINOP op0=1 op1=3 op2=1/>
+; PF2-NEXT:     <INST_BINOP op0=2 op1=4 op2=2/>
+; PF2-NEXT:     <INST_BINOP op0=3 op1=5 op2=3/>
+; PF2-NEXT:     <INST_BINOP op0=4 op1=6 op2=5/>
+; PF2-NEXT:     <INST_BINOP op0=5 op1=7 op2=6/>
+; PF2-NEXT:     <INST_BINOP op0=6 op1=8 op2=7/>
+; PF2-NEXT:     <INST_BINOP op0=7 op1=9 op2=8/>
+; PF2-NEXT:     <INST_BINOP op0=8 op1=10 op2=9/>
+; PF2-NEXT:     <INST_BINOP op0=9 op1=11 op2=10/>
+; PF2-NEXT:     <INST_BINOP op0=10 op1=12 op2=11/>
+; PF2-NEXT:     <INST_BINOP op0=11 op1=13 op2=12/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we handle (non-special) bitcasts by converting pointer
+; casts to integer.
+define void @TestCasts() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+
+  %3 = trunc i32 257 to i8
+  %4 = trunc i32 %2 to i8
+
+  %5 = zext i32 257 to i64
+  %6 = zext i32 %2 to i64
+
+  %7 = sext i32 -1 to i64
+  %8 = sext i32 %2 to i64
+
+  %9 = uitofp i32 1 to float
+  %10 = uitofp i32 %2 to float
+
+  %11 = sitofp i32 -1 to float
+  %12 = sitofp i32 %2 to float
+  ret void
+}
+
+; TD2:      define void @TestCasts() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = trunc i32 257 to i8
+; TD2-NEXT:   %3 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %4 = trunc i32 %3 to i8
+; TD2-NEXT:   %5 = zext i32 257 to i64
+; TD2-NEXT:   %6 = zext i32 %3 to i64
+; TD2-NEXT:   %7 = sext i32 -1 to i64
+; TD2-NEXT:   %8 = sext i32 %3 to i64
+; TD2-NEXT:   %9 = uitofp i32 1 to float
+; TD2-NEXT:   %10 = uitofp i32 %3 to float
+; TD2-NEXT:   %11 = sitofp i32 -1 to float
+; TD2-NEXT:   %12 = sitofp i32 %3 to float
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_CAST op0=5 op1={{.*}} op2=0/>
+; PF2-NEXT:     <INST_CAST op0=2 op1={{.*}} op2=0/>
+; PF2-NEXT:     <INST_CAST op0=7 op1={{.*}} op2=1/>
+; PF2-NEXT:     <INST_CAST op0=4 op1={{.*}} op2=1/>
+; PF2-NEXT:     <INST_CAST op0=8 op1={{.*}} op2=2/>
+; PF2-NEXT:     <INST_CAST op0=6 op1={{.*}} op2=2/>
+; PF2-NEXT:     <INST_CAST op0=8 op1={{.*}} op2=5/>
+; PF2-NEXT:     <INST_CAST op0=8 op1={{.*}} op2=5/>
+; PF2-NEXT:     <INST_CAST op0=12 op1={{.*}} op2=6/>
+; PF2-NEXT:     <INST_CAST op0=10 op1={{.*}} op2=6/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we elide a ptrtoint cast for a call.
+define void @TestSavedPtrToInt() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = add i32 %2, 0
+  %4 = call i32 @bar(i32 %2)
+  ret void
+}
+
+; TD2:      define void @TestSavedPtrToInt() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %3 = add i32 %2, 0
+; TD2-NEXT:   %4 = call i32 @bar(i32 %2)
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_BINOP op0=1 op1=2 op2=0/>
+; PF2-NEXT:     <INST_CALL op0=0 op1=25 op2=2/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we can handle pointer conversions for icmp.
+define void @CastIcmp() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = ptrtoint [4 x i8]* @bytes to i32
+  %4 = icmp eq i32 1, 2
+  %5 = icmp eq i32 %2, 2
+  %6 = icmp eq i32 1, %3
+  %7 = icmp eq i32 %2, %3
+  %8 = icmp eq i32 %3, %2
+  ret void
+}
+
+; TD2:      define void @CastIcmp() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = icmp eq i32 1, 2
+; TD2-NEXT:   %3 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %4 = icmp eq i32 %3, 2
+; TD2-NEXT:   %5 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %6 = icmp eq i32 1, %5
+; TD2-NEXT:   %7 = icmp eq i32 %3, %5
+; TD2-NEXT:   %8 = icmp eq i32 %5, %3
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_CMP2 op0=4 op1=3 op2=32/>
+; PF2-NEXT:     <INST_CMP2 op0=2 op1=4 op2=32/>
+; PF2-NEXT:     <INST_CMP2 op0=6 op1=7 op2=32/>
+; PF2-NEXT:     <INST_CMP2 op0=4 op1=8 op2=32/>
+; PF2-NEXT:     <INST_CMP2 op0=9 op1=5 op2=32/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we can handle pointer conversions for Select.
+define void @CastSelect() {
+  %1 = alloca i8, i32 4, align 8
+  %2 = ptrtoint i8* %1 to i32
+  %3 = ptrtoint [4 x i8]* @bytes to i32
+  %4 = select i1 true, i32 1, i32 2
+  %5 = select i1 true, i32 %2, i32 2
+  %6 = select i1 true, i32 1, i32 %3
+  %7 = select i1 true, i32 %2, i32 %3
+  %8 = select i1 true, i32 %3, i32 %2
+  ret void
+}
+
+; TD2:      define void @CastSelect() {
+; TD2-NEXT:   %1 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %2 = select i1 true, i32 1, i32 2
+; TD2-NEXT:   %3 = ptrtoint i8* %1 to i32
+; TD2-NEXT:   %4 = select i1 true, i32 %3, i32 2
+; TD2-NEXT:   %5 = ptrtoint [4 x i8]* @bytes to i32
+; TD2-NEXT:   %6 = select i1 true, i32 1, i32 %5
+; TD2-NEXT:   %7 = select i1 true, i32 %3, i32 %5
+; TD2-NEXT:   %8 = select i1 true, i32 %5, i32 %3
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_VSELECT op0=5 op1=4 op2=2/>
+; PF2-NEXT:     <INST_VSELECT op0=2 op1=5 op2=3/>
+; PF2-NEXT:     <INST_VSELECT op0=7 op1=8 op2=4/>
+; PF2-NEXT:     <INST_VSELECT op0=4 op1=9 op2=5/>
+; PF2-NEXT:     <INST_VSELECT op0=10 op1=5 op2=6/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:   </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that if a phi node refers to a pointer cast, we add
+; them at the end of the incoming block.
+define void @PhiBackwardRefs(i1) {
+  %2 = alloca i8, i32 4, align 8
+  %3 = bitcast i8* %2 to i32*
+  %4 = alloca i8, i32 4, align 8
+  %5 = ptrtoint i8* %4 to i32
+  br i1 %0, label %true, label %false
+
+true:
+  %6 = load i32* %3
+  br label %merge
+
+false:
+  %7 = load i32* %3
+  br label %merge
+
+merge:
+  %8 = phi i32 [%5, %true], [%5, %false]
+  %9 = phi i32 [%6, %true], [%7, %false]
+  ret void
+}
+
+; TD2:      define void @PhiBackwardRefs(i1) {
+; TD2-NEXT:   %2 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %3 = alloca i8, i32 4, align 8
+; TD2-NEXT:   br i1 %0, label %true, label %false
+; TD2:      true:
+; TD2-NEXT:   %4 = bitcast i8* %2 to i32*
+; TD2-NEXT:   %5 = load i32* %4
+; TD2-NEXT:   %6 = ptrtoint i8* %3 to i32
+; TD2-NEXT:   br label %merge
+; TD2:      false:
+; TD2-NEXT:   %7 = bitcast i8* %2 to i32*
+; TD2-NEXT:   %8 = load i32* %7
+; TD2-NEXT:   %9 = ptrtoint i8* %3 to i32
+; TD2-NEXT:   br label %merge
+; TD2:      merge:
+; TD2-NEXT:   %10 = phi i32 [ %6, %true ], [ %9, %false ]
+; TD2-NEXT:   %11 = phi i32 [ %5, %true ], [ %8, %false ]
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_BR op0=1 op1=2 op2=4/>
+; PF2-NEXT:     <INST_LOAD op0=2 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BR op0=3/>
+; PF2-NEXT:     <INST_LOAD op0=3 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BR op0=3/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=6 op2=1 op3=6 op4=2/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=6 op2=1 op3=4 op4=2/>
+; PF2-NEXT:     <INST_RET/>
+; PF2:        </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Like PhiBackwardRefs except the phi nodes forward reference
+; instructions instead of backwards references.
+define void @PhiForwardRefs(i1) {
+  br label %start
+
+merge:
+  %2 = phi i32 [%9, %true], [%9, %false]
+  %3 = phi i32 [%4, %true], [%5, %false]
+  ret void
+
+true:
+  %4 = load i32* %7
+  br label %merge
+
+false:
+  %5 = load i32* %7
+  br label %merge
+
+start:
+  %6 = alloca i8, i32 4, align 8
+  %7 = bitcast i8* %6 to i32*
+  %8 = alloca i8, i32 4, align 8
+  %9 = ptrtoint i8* %8 to i32
+  br i1 %0, label %true, label %false
+}
+
+; TD2:      define void @PhiForwardRefs(i1) {
+; TD2-NEXT:   br label %start
+; TD2:      merge
+; TD2-NEXT:   %2 = phi i32 [ %11, %true ], [ %11, %false ]
+; TD2-NEXT:   %3 = phi i32 [ %5, %true ], [ %7, %false ]
+; TD2-NEXT:   ret void
+; TD2:      true:
+; TD2-NEXT:   %4 = inttoptr i32 %9 to i32*
+; TD2-NEXT:   %5 = load i32* %4
+; TD2-NEXT:   br label %merge
+; TD2:      false:
+; TD2-NEXT:   %6 = inttoptr i32 %9 to i32*
+; TD2-NEXT:   %7 = load i32* %6
+; TD2-NEXT:   br label %merge
+; TD2:      start:
+; TD2-NEXT:   %8 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %9 = ptrtoint i8* %8 to i32
+; TD2-NEXT:   %10 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %11 = ptrtoint i8* %10 to i32
+; TD2-NEXT:   br i1 %0, label %true, label %false
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_BR op0=4/>
+; PF2-NEXT:     <FORWARDTYPEREF op0=28 op1=0/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=11 op2=2 op3=11 op4=3/>
+; PF2-NEXT:     <FORWARDTYPEREF op0=25 op1=0/>
+; PF2-NEXT:     <FORWARDTYPEREF op0=26 op1=0/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=3 op2=2 op3=5 op4=3/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:     <FORWARDTYPEREF op0=27 op1=0/>
+; PF2-NEXT:     <INST_LOAD op0=4294967294 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BR op0=1/>
+; PF2-NEXT:     <INST_LOAD op0=4294967295 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BR op0=1/>
+; PF2-NEXT:     <INST_ALLOCA op0=5 op1=4/>
+; PF2-NEXT:     <INST_ALLOCA op0=6 op1=4/>
+; PF2-NEXT:     <INST_BR op0=2 op1=3 op2=8/>
+; PF2:        </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that if a phi node incoming block already has a pointer cast,
+; we use it instead of adding one at the end of the block. In this
+; example, we reuse instruction %7 in block true for phi node %10.
+define void @PhiMergeCast(i1) {
+  %2 = alloca i8, i32 4, align 8
+  %3 = bitcast i8* %2 to i32*
+  %4 = alloca i8, i32 4, align 8
+  %5 = ptrtoint i8* %4 to i32
+  br i1 %0, label %true, label %false
+
+true:
+  %6 = load i32* %3
+  %7 = ptrtoint i8* %4 to i32
+  %8 = add i32 %6, %7
+  br label %merge
+
+false:
+  %9 = load i32* %3
+  br label %merge
+
+merge:
+  %10 = phi i32 [%5, %true], [%5, %false]
+  %11 = phi i32 [%6, %true], [%9, %false]
+  ret void
+}
+
+; TD2:      define void @PhiMergeCast(i1) {
+; TD2-NEXT:   %2 = alloca i8, i32 4, align 8
+; TD2-NEXT:   %3 = alloca i8, i32 4, align 8
+; TD2-NEXT:   br i1 %0, label %true, label %false
+; TD2:      true:
+; TD2-NEXT:   %4 = bitcast i8* %2 to i32*
+; TD2-NEXT:   %5 = load i32* %4
+; TD2-NEXT:   %6 = ptrtoint i8* %3 to i32
+; TD2-NEXT:   %7 = add i32 %5, %6
+; TD2-NEXT:   br label %merge
+; TD2:      false:
+; TD2-NEXT:   %8 = bitcast i8* %2 to i32*
+; TD2-NEXT:   %9 = load i32* %8
+; TD2-NEXT:   %10 = ptrtoint i8* %3 to i32
+; TD2-NEXT:   br label %merge
+; TD2:      merge:
+; TD2-NEXT:   %11 = phi i32 [ %6, %true ], [ %10, %false ]
+; TD2-NEXT:   %12 = phi i32 [ %5, %true ], [ %9, %false ]
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_ALLOCA op0=2 op1=4/>
+; PF2-NEXT:     <INST_BR op0=1 op1=2 op2=4/>
+; PF2-NEXT:     <INST_LOAD op0=2 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BINOP op0=1 op1=2 op2=0/>
+; PF2-NEXT:     <INST_BR op0=3/>
+; PF2-NEXT:     <INST_LOAD op0=4 op1=0 op2=0/>
+; PF2-NEXT:     <INST_BR op0=3/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=8 op2=1 op3=8 op4=2/>
+; PF2-NEXT:     <INST_PHI op0=0 op1=8 op2=1 op3=4 op4=2/>
+; PF2-NEXT:     <INST_RET/>
+; PF2:        </FUNCTION_BLOCK>
+
+; ------------------------------------------------------
+
+; Show that we must introduce a cast reference for each
+; reachable block, but one is sufficient.
+define void @LongReachingCasts(i1) {
+  %2 = alloca i8, i32 4, align 8
+  %3 = ptrtoint i8* %2 to i32
+  %4 = bitcast [4 x i8]* @bytes to i32*
+  br i1 %0, label %Split1, label %Split2
+
+Split1:
+  br i1 %0, label %b1, label %b2
+
+Split2:
+  br i1 %0, label %b3, label %b4
+
+b1:
+  store i32 %3, i32* %4, align 1
+  store i32 %3, i32* %4, align 1
+  ret void
+
+b2:
+  store i32 %3, i32* %4, align 1
+  store i32 %3, i32* %4, align 1
+  ret void
+
+b3:
+  store i32 %3, i32* %4, align 1
+  store i32 %3, i32* %4, align 1
+  ret void
+
+b4:
+  store i32 %3, i32* %4, align 1
+  store i32 %3, i32* %4, align 1
+  ret void
+}
+
+; TD2:      define void @LongReachingCasts(i1) {
+; TD2-NEXT:   %2 = alloca i8, i32 4, align 8
+; TD2-NEXT:   br i1 %0, label %Split1, label %Split2
+; TD2:      Split1:
+; TD2-NEXT:   br i1 %0, label %b1, label %b2
+; TD2:      Split2:
+; TD2-NEXT:   br i1 %0, label %b3, label %b4
+; TD2:      b1:
+; TD2-NEXT:   %3 = ptrtoint i8* %2 to i32
+; TD2-NEXT:   %4 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %3, i32* %4, align 1
+; TD2-NEXT:   store i32 %3, i32* %4, align 1
+; TD2-NEXT:   ret void
+; TD2:      b2:
+; TD2-NEXT:   %5 = ptrtoint i8* %2 to i32
+; TD2-NEXT:   %6 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %5, i32* %6, align 1
+; TD2-NEXT:   store i32 %5, i32* %6, align 1
+; TD2-NEXT:   ret void
+; TD2:      b3:
+; TD2-NEXT:   %7 = ptrtoint i8* %2 to i32
+; TD2-NEXT:   %8 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %7, i32* %8, align 1
+; TD2-NEXT:   store i32 %7, i32* %8, align 1
+; TD2-NEXT:   ret void
+; TD2:      b4:
+; TD2-NEXT:   %9 = ptrtoint i8* %2 to i32
+; TD2-NEXT:   %10 = bitcast [4 x i8]* @bytes to i32*
+; TD2-NEXT:   store i32 %9, i32* %10, align 1
+; TD2-NEXT:   store i32 %9, i32* %10, align 1
+; TD2-NEXT:   ret void
+; TD2-NEXT: }
+
+; PF2:        <FUNCTION_BLOCK>
+; PF2:          </CONSTANTS_BLOCK>
+; PF2-NEXT:     <INST_ALLOCA op0=1 op1=4/>
+; PF2-NEXT:     <INST_BR op0=1 op1=2 op2=3/>
+; PF2-NEXT:     <INST_BR op0=3 op1=4 op2=3/>
+; PF2-NEXT:     <INST_BR op0=5 op1=6 op2=3/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_STORE op0=4 op1=1 op2=1/>
+; PF2-NEXT:     <INST_RET/>
+; PF2:        </FUNCTION_BLOCK>