[InstCombine] Fix vscale zext/sext optimization when vscale_range is unbounded.

According to the LangRef, a (vscale_range) value of 0 means unbounded.

This patch additionally cleans up the test file vscale_sext_and_zext.ll.

Author: sdesmalen-arm

llvm/lib/Transforms/InstCombine/InstCombineCasts.cpp

@@ -1368,7 +1368,7 @@ Instruction *InstCombinerImpl::visitZExt(ZExtInst &CI) {
                                .getVScaleRangeArgs()
                                .second;
       unsigned TypeWidth = Src->getType()->getScalarSizeInBits();
-      if (Log2_32(MaxVScale) < TypeWidth) {
+      if (MaxVScale > 0 && Log2_32(MaxVScale) < TypeWidth) {
         Value *VScale = Builder.CreateVScale(ConstantInt::get(DestTy, 1));
         return replaceInstUsesWith(CI, VScale);
       }
@@ -1625,8 +1625,7 @@ Instruction *InstCombinerImpl::visitSExt(SExtInst &CI) {
                                ->getFnAttribute(Attribute::VScaleRange)
                                .getVScaleRangeArgs()
                                .second;
-      unsigned TypeWidth = Src->getType()->getScalarSizeInBits();
-      if (Log2_32(MaxVScale) < (TypeWidth - 1)) {
+      if (MaxVScale > 0 && Log2_32(MaxVScale) < (SrcBitSize - 1)) {
         Value *VScale = Builder.CreateVScale(ConstantInt::get(DestTy, 1));
         return replaceInstUsesWith(CI, VScale);
       }

llvm/test/Transforms/InstCombine/vscale_sext_and_zext.ll

@@ -1,85 +1,109 @@
 ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
 ; RUN: opt < %s -instcombine -S | FileCheck %s
 
-define i64 @vscale_SExt_i32toi64() #0 {
-; CHECK:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    ret i64 [[TMP0]]
-entry:
-  %0 = call i32 @llvm.vscale.i32()
-  %1 = sext i32 %0 to i64
-  ret i64 %1
-}
+;
+; Sign-extend
+;
 
-define i32 @vscale_SExt_i8toi32() #0 {
-; CHECK:  entry:
+define i32 @vscale_SExt_i8toi32() vscale_range(0, 127) {
+; CHECK-LABEL: @vscale_SExt_i8toi32(
+; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
 ; CHECK-NEXT:    ret i32 [[TMP0]]
+;
 entry:
   %0 = call i8 @llvm.vscale.i8()
   %1 = sext i8 %0 to i32
   ret i32 %1
 }
 
-
-define i32 @vscale_SExt_i8toi32_poison() vscale_range(0, 192) {
-; CHECK:  entry:
+define i32 @vscale_SExt_i8toi32_poison() vscale_range(0, 128) {
+; CHECK-LABEL: @vscale_SExt_i8toi32_poison(
+; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.vscale.i8()
 ; CHECK-NEXT:    [[TMP1:%.*]] = sext i8 [[TMP0]] to i32
 ; CHECK-NEXT:    ret i32 [[TMP1]]
+;
   entry:
   %0 = call i8 @llvm.vscale.i8()
   %1 = sext i8 %0 to i32
   ret i32 %1
 }
 
+;
+; Zero-extend
+;
 
-
-define i64 @vscale_ZExt_i32toi64() #0 {
-; CHECK:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    ret i64 [[TMP0]]
+define i32 @vscale_ZExt_i8toi32() vscale_range(0, 128) {
+; CHECK-LABEL: @vscale_ZExt_i8toi32(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = call i32 @llvm.vscale.i32()
+; CHECK-NEXT:    ret i32 [[TMP0]]
+;
 entry:
-  %0 = call i32 @llvm.vscale.i32()
-  %1 = zext i32 %0 to i64
-  ret i64 %1
+  %0 = call i8 @llvm.vscale.i8()
+  %1 = zext i8 %0 to i32
+  ret i32 %1
 }
 
-define i64 @vscale_ZExt_i1toi64() vscale_range(0, 1) {
-; CHECK:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i64 @llvm.vscale.i64()
-; CHECK-NEXT:    ret i64 [[TMP0]]
-entry:
-  %0 = call i1 @llvm.vscale.i1()
-  %1 = zext i1 %0 to i64
-  ret i64 %1
+define i32 @vscale_ZExt_i8toi32_poison() vscale_range(0, 256) {
+; CHECK-LABEL: @vscale_ZExt_i8toi32_poison(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.vscale.i8()
+; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[TMP0]] to i32
+; CHECK-NEXT:    ret i32 [[TMP1]]
+;
+  entry:
+  %0 = call i8 @llvm.vscale.i8()
+  %1 = zext i8 %0 to i32
+  ret i32 %1
 }
 
-define i32 @vscale_ZExt_i8toi32_poison() vscale_range(0, 1024) {
-; CHECK:  entry:
+;
+; No vscale_range attribute
+;
+
+define i32 @vscale_ZExt_i8toi32_unknown() {
+; CHECK-LABEL: @vscale_ZExt_i8toi32_unknown(
+; CHECK-NEXT:  entry:
 ; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.vscale.i8()
 ; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[TMP0]] to i32
 ; CHECK-NEXT:    ret i32 [[TMP1]]
+;
   entry:
   %0 = call i8 @llvm.vscale.i8()
   %1 = zext i8 %0 to i32
   ret i32 %1
 }
 
-define i32 @vscale_ZExt_i16toi32_unknown() {
-; CHECK:  entry:
-; CHECK-NEXT:    [[TMP0:%.*]] = call i16 @llvm.vscale.i16()
-; CHECK-NEXT:    [[TMP1:%.*]] = zext i16 [[TMP0]] to i32
+;
+; unbounded vscale_range maximum (0)
+;
+
+define i32 @vscale_SExt_i8toi32_unbounded() vscale_range(0, 0) {
+; CHECK-LABEL: @vscale_SExt_i8toi32_unbounded(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.vscale.i8()
+; CHECK-NEXT:    [[TMP1:%.*]] = sext i8 [[TMP0]] to i32
 ; CHECK-NEXT:    ret i32 [[TMP1]]
+;
   entry:
-  %0 = call i16 @llvm.vscale.i16()
-  %1 = zext i16 %0 to i32
+  %0 = call i8 @llvm.vscale.i8()
+  %1 = sext i8 %0 to i32
   ret i32 %1
 }
 
-attributes #0 = { vscale_range(0, 16) }
+define i32 @vscale_ZExt_i8toi32_unbounded() vscale_range(0, 0) {
+; CHECK-LABEL: @vscale_ZExt_i8toi32_unbounded(
+; CHECK-NEXT:  entry:
+; CHECK-NEXT:    [[TMP0:%.*]] = call i8 @llvm.vscale.i8()
+; CHECK-NEXT:    [[TMP1:%.*]] = zext i8 [[TMP0]] to i32
+; CHECK-NEXT:    ret i32 [[TMP1]]
+;
+  entry:
+  %0 = call i8 @llvm.vscale.i8()
+  %1 = zext i8 %0 to i32
+  ret i32 %1
+}
 
-declare i1  @llvm.vscale.i1()
 declare i8  @llvm.vscale.i8()
-declare i16 @llvm.vscale.i16()
-declare i32 @llvm.vscale.i32()