[InstCombine] propagate 'nsw' on pointer difference of 'inbounds' geps

This is a retry of 324a53205. I cautiously reverted that at 6aa3fc4
because the rules about gep math were not clear. Since then, we
have added this line to LangRef for gep inbounds:
"The successive addition of offsets (without adding the base address)
does not wrap the pointer index type in a signed sense (nsw)."

See D90708 and post-commit comments on the revert patch for more details.

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp b/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
index b8431a5..9a6a790 100644 (file)
--- a/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineAddSub.cpp
@@ -1678,11 +1678,12 @@ Value *InstCombinerImpl::OptimizePointerDifference(Value *LHS, Value *RHS,
         I->getOpcode() == Instruction::Mul)
       I->setHasNoUnsignedWrap();
 
-  // If we had a constant expression GEP on the other side offsetting the
-  // pointer, subtract it from the offset we have.
+  // If we have a 2nd GEP of the same base pointer, subtract the offsets.
+  // If both GEPs are inbounds, then the subtract does not have signed overflow.
   if (GEP2) {
     Value *Offset = EmitGEPOffset(GEP2);
-    Result = Builder.CreateSub(Result, Offset, "gepdiff");
+    Result = Builder.CreateSub(Result, Offset, "gepdiff", /* NUW */ false,
+                               GEP1->isInBounds() && GEP2->isInBounds());
   }
 
   // If we have p - gep(p, ...)  then we have to negate the result.

diff --git a/llvm/test/Transforms/InstCombine/sub-gep.ll b/llvm/test/Transforms/InstCombine/sub-gep.ll
index 9868ed1..2389b70 100644 (file)
--- a/llvm/test/Transforms/InstCombine/sub-gep.ll
+++ b/llvm/test/Transforms/InstCombine/sub-gep.ll
@@ -245,7 +245,7 @@ define i64 @test24b(i8* %P, i64 %A){
 define i64 @test25(i8* %P, i64 %A){
 ; CHECK-LABEL: @test25(
 ; CHECK-NEXT:    [[B_IDX:%.*]] = shl nsw i64 [[A:%.*]], 1
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = add i64 [[B_IDX]], -84
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = add nsw i64 [[B_IDX]], -84
 ; CHECK-NEXT:    ret i64 [[GEPDIFF]]
 ;
   %B = getelementptr inbounds [42 x i16], [42 x i16]* @Arr, i64 0, i64 %A
@@ -260,7 +260,7 @@ define i16 @test25_as1(i8 addrspace(1)* %P, i64 %A) {
 ; CHECK-LABEL: @test25_as1(
 ; CHECK-NEXT:    [[TMP1:%.*]] = trunc i64 [[A:%.*]] to i16
 ; CHECK-NEXT:    [[B_IDX:%.*]] = shl nsw i16 [[TMP1]], 1
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = add i16 [[B_IDX]], -84
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = add nsw i16 [[B_IDX]], -84
 ; CHECK-NEXT:    ret i16 [[GEPDIFF]]
 ;
   %B = getelementptr inbounds [42 x i16], [42 x i16] addrspace(1)* @Arr_as1, i64 0, i64 %A
@@ -272,7 +272,7 @@ define i16 @test25_as1(i8 addrspace(1)* %P, i64 %A) {
 define i64 @test30(i8* %foo, i64 %i, i64 %j) {
 ; CHECK-LABEL: @test30(
 ; CHECK-NEXT:    [[GEP1_IDX:%.*]] = shl nsw i64 [[I:%.*]], 2
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i64 [[GEP1_IDX]], [[J:%.*]]
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub nsw i64 [[GEP1_IDX]], [[J:%.*]]
 ; CHECK-NEXT:    ret i64 [[GEPDIFF]]
 ;
   %bit = bitcast i8* %foo to i32*
@@ -287,7 +287,7 @@ define i64 @test30(i8* %foo, i64 %i, i64 %j) {
 define i16 @test30_as1(i8 addrspace(1)* %foo, i16 %i, i16 %j) {
 ; CHECK-LABEL: @test30_as1(
 ; CHECK-NEXT:    [[GEP1_IDX:%.*]] = shl nsw i16 [[I:%.*]], 2
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i16 [[GEP1_IDX]], [[J:%.*]]
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub nsw i16 [[GEP1_IDX]], [[J:%.*]]
 ; CHECK-NEXT:    ret i16 [[GEPDIFF]]
 ;
   %bit = bitcast i8 addrspace(1)* %foo to i32 addrspace(1)*
@@ -299,9 +299,11 @@ define i16 @test30_as1(i8 addrspace(1)* %foo, i16 %i, i16 %j) {
   ret i16 %sub
 }
 
+; Inbounds translates to 'nsw' on sub
+
 define i64 @gep_diff_both_inbounds(i8* %foo, i64 %i, i64 %j) {
 ; CHECK-LABEL: @gep_diff_both_inbounds(
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i64 [[I:%.*]], [[J:%.*]]
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub nsw i64 [[I:%.*]], [[J:%.*]]
 ; CHECK-NEXT:    ret i64 [[GEPDIFF]]
 ;
   %gep1 = getelementptr inbounds i8, i8* %foo, i64 %i
@@ -312,6 +314,8 @@ define i64 @gep_diff_both_inbounds(i8* %foo, i64 %i, i64 %j) {
   ret i64 %sub
 }
 
+; Negative test for 'nsw' - both geps must be inbounds
+
 define i64 @gep_diff_first_inbounds(i8* %foo, i64 %i, i64 %j) {
 ; CHECK-LABEL: @gep_diff_first_inbounds(
 ; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i64 [[I:%.*]], [[J:%.*]]
@@ -325,6 +329,8 @@ define i64 @gep_diff_first_inbounds(i8* %foo, i64 %i, i64 %j) {
   ret i64 %sub
 }
 
+; Negative test for 'nsw' - both geps must be inbounds
+
 define i64 @gep_diff_second_inbounds(i8* %foo, i64 %i, i64 %j) {
 ; CHECK-LABEL: @gep_diff_second_inbounds(
 ; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i64 [[I:%.*]], [[J:%.*]]

diff --git a/llvm/test/Transforms/InstCombine/sub.ll b/llvm/test/Transforms/InstCombine/sub.ll
index d9c67d0..7809a93 100644 (file)
--- a/llvm/test/Transforms/InstCombine/sub.ll
+++ b/llvm/test/Transforms/InstCombine/sub.ll
@@ -1077,7 +1077,7 @@ define i64 @test58([100 x [100 x i8]]* %foo, i64 %i, i64 %j) {
 ; CHECK-LABEL: @test58(
 ; CHECK-NEXT:    [[GEP1_OFFS:%.*]] = add nsw i64 [[I:%.*]], 4200
 ; CHECK-NEXT:    [[GEP2_OFFS:%.*]] = add nsw i64 [[J:%.*]], 4200
-; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub i64 [[GEP1_OFFS]], [[GEP2_OFFS]]
+; CHECK-NEXT:    [[GEPDIFF:%.*]] = sub nsw i64 [[GEP1_OFFS]], [[GEP2_OFFS]]
 ; CHECK-NEXT:    ret i64 [[GEPDIFF]]
 ;
   %gep1 = getelementptr inbounds [100 x [100 x i8]], [100 x [100 x i8]]* %foo, i64 0, i64 42, i64 %i