[InstCombine] add folds for icmp + sub (PR36969)

(A - B) >u A --> A <u B
C <u (C - D) --> C <u D

https://rise4fun.com/Alive/e7j

Name: ugt
  %sub = sub i8 %x, %y
  %cmp = icmp ugt i8 %sub, %x
=>
  %cmp = icmp ult i8 %x, %y
  
Name: ult
  %sub = sub i8 %x, %y
  %cmp = icmp ult i8 %x, %sub
=>
  %cmp = icmp ult i8 %x, %y

This should fix:
https://bugs.llvm.org/show_bug.cgi?id=36969

llvm-svn: 329011

2 files changed, 9 insertions, 6 deletions

diff --git a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
index c0876a27594..cea3a1d705b 100644
--- a/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
+++ b/llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp
@@ -3005,17 +3005,22 @@ Instruction *InstCombiner::foldICmpBinOp(ICmpInst &I) {
   // icmp (X-Y), X -> icmp 0, Y for equalities or if there is no overflow.
   if (A == Op1 && NoOp0WrapProblem)
     return new ICmpInst(Pred, Constant::getNullValue(Op1->getType()), B);
-
   // icmp X, (X-Y) -> icmp Y, 0 for equalities or if there is no overflow.
   if (C == Op0 && NoOp1WrapProblem)
     return new ICmpInst(Pred, D, Constant::getNullValue(Op0->getType()));
 
+  // (A - B) >u A --> A <u B
+  if (A == Op1 && Pred == ICmpInst::ICMP_UGT)
+    return new ICmpInst(ICmpInst::ICMP_ULT, A, B);
+  // C <u (C - D) --> C <u D
+  if (C == Op0 && Pred == ICmpInst::ICMP_ULT)
+    return new ICmpInst(ICmpInst::ICMP_ULT, C, D);
+
   // icmp (Y-X), (Z-X) -> icmp Y, Z for equalities or if there is no overflow.
   if (B && D && B == D && NoOp0WrapProblem && NoOp1WrapProblem &&
       // Try not to increase register pressure.
       BO0->hasOneUse() && BO1->hasOneUse())
     return new ICmpInst(Pred, A, C);
-
   // icmp (X-Y), (X-Z) -> icmp Z, Y for equalities or if there is no overflow.
   if (A && C && A == C && NoOp0WrapProblem && NoOp1WrapProblem &&
       // Try not to increase register pressure.
diff --git a/llvm/test/Transforms/InstCombine/icmp.ll b/llvm/test/Transforms/InstCombine/icmp.ll
index 6e6a9135350..34fdf32adf1 100644
--- a/llvm/test/Transforms/InstCombine/icmp.ll
+++ b/llvm/test/Transforms/InstCombine/icmp.ll
@@ -550,8 +550,7 @@ define i1 @test36(i32 %x, i32 %y) {
 define i1 @ugt_sub(i32 %xsrc, i32 %y) {
 ; CHECK-LABEL: @ugt_sub(
 ; CHECK-NEXT:    [[X:%.*]] = udiv i32 [[XSRC:%.*]], 42
-; CHECK-NEXT:    [[SUB:%.*]] = sub i32 [[X]], [[Y:%.*]]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp ugt i32 [[SUB]], [[X]]
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ult i32 [[X]], [[Y:%.*]]
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %x = udiv i32 %xsrc, 42 ; thwart complexity-based canonicalization
@@ -565,8 +564,7 @@ define i1 @ugt_sub(i32 %xsrc, i32 %y) {
 define <2 x i1> @ult_sub(<2 x i8> %xsrc, <2 x i8> %y) {
 ; CHECK-LABEL: @ult_sub(
 ; CHECK-NEXT:    [[X:%.*]] = udiv <2 x i8> [[XSRC:%.*]], <i8 42, i8 -42>
-; CHECK-NEXT:    [[SUB:%.*]] = sub <2 x i8> [[X]], [[Y:%.*]]
-; CHECK-NEXT:    [[CMP:%.*]] = icmp ult <2 x i8> [[X]], [[SUB]]
+; CHECK-NEXT:    [[CMP:%.*]] = icmp ult <2 x i8> [[X]], [[Y:%.*]]
 ; CHECK-NEXT:    ret <2 x i1> [[CMP]]
 ;
   %x = udiv <2 x i8> %xsrc, <i8 42, i8 -42> ; thwart complexity-based canonicalization