[InstCombine] Fix FMF propagation in foldFCmpFSubIntoFCmp

Author: dtcxzyw

llvm/lib/Transforms/InstCombine/InstCombineCompares.cpp

@@ -8527,6 +8527,9 @@ static Instruction *foldFCmpFSubIntoFCmp(FCmpInst &I, Instruction *LHSI,
             DenormalMode::getIEEE()) {
       CI.replaceOperand(I, 0, X);
       CI.replaceOperand(I, 1, Y);
+      I.setHasNoInfs(LHSI->hasNoInfs());
+      if (LHSI->hasNoNaNs())
+        I.setHasNoNaNs(true);
       return &I;
     }
     break;

llvm/test/Transforms/InstCombine/fcmp.ll

@@ -1814,7 +1814,7 @@ define i1 @fcmp_ule_fsub_const(float %x, float %y) {
 
 define i1 @fcmp_ninf_ule_fsub_const(float %x, float %y) {
 ; CHECK-LABEL: @fcmp_ninf_ule_fsub_const(
-; CHECK-NEXT:    [[CMP:%.*]] = fcmp ninf ule float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp ule float [[X:%.*]], [[Y:%.*]]
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %fs = fsub float %x, %y
@@ -1834,7 +1834,7 @@ define i1 @fcmp_nnan_ule_fsub_const(float %x, float %y) {
 
 define i1 @fcmp_ule_fsub_ninf_const(float %x, float %y) {
 ; CHECK-LABEL: @fcmp_ule_fsub_ninf_const(
-; CHECK-NEXT:    [[CMP:%.*]] = fcmp ule float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp ninf ule float [[X:%.*]], [[Y:%.*]]
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %fs = fsub ninf float %x, %y
@@ -1844,7 +1844,7 @@ define i1 @fcmp_ule_fsub_ninf_const(float %x, float %y) {
 
 define i1 @fcmp_ule_fsub_nnan_const(float %x, float %y) {
 ; CHECK-LABEL: @fcmp_ule_fsub_nnan_const(
-; CHECK-NEXT:    [[CMP:%.*]] = fcmp ule float [[X:%.*]], [[Y:%.*]]
+; CHECK-NEXT:    [[CMP:%.*]] = fcmp nnan ule float [[X:%.*]], [[Y:%.*]]
 ; CHECK-NEXT:    ret i1 [[CMP]]
 ;
   %fs = fsub nnan float %x, %y