[InstSimplify] add folds for constant mask of value shifted by constant

We would eventually catch these via demanded bits and computing known bits in InstCombine,
but I think it's better to handle the simple cases as soon as possible as a matter of efficiency.

This fold allows further simplifications based on distributed ops transforms. eg:
  %a = lshr i8 %x, 7
  %b = or i8 %a, 2
  %c = and i8 %b, 1

InstSimplify can directly fold this now:
  %a = lshr i8 %x, 7

Differential Revision: https://reviews.llvm.org/D33221

llvm-svn: 303213

2 files changed, 22 insertions, 8 deletions

diff --git a/llvm/lib/Analysis/InstructionSimplify.cpp b/llvm/lib/Analysis/InstructionSimplify.cpp
index 5728887cc1e..5652248a60c 100644
--- a/llvm/lib/Analysis/InstructionSimplify.cpp
+++ b/llvm/lib/Analysis/InstructionSimplify.cpp
@@ -1752,6 +1752,24 @@ static Value *SimplifyAndInst(Value *Op0, Value *Op1, const SimplifyQuery &Q,
       (A == Op0 || B == Op0))
     return Op0;
 
+  // A mask that only clears known zeros of a shifted value is a no-op.
+  Value *X;
+  const APInt *Mask;
+  const APInt *ShAmt;
+  if (match(Op1, m_APInt(Mask))) {
+    // If all bits in the inverted and shifted mask are clear:
+    // and (shl X, ShAmt), Mask --> shl X, ShAmt
+    if (match(Op0, m_Shl(m_Value(X), m_APInt(ShAmt))) &&
+        (~(*Mask)).lshr(*ShAmt).isNullValue())
+      return Op0;
+
+    // If all bits in the inverted and shifted mask are clear:
+    // and (lshr X, ShAmt), Mask --> lshr X, ShAmt
+    if (match(Op0, m_LShr(m_Value(X), m_APInt(ShAmt))) &&
+        (~(*Mask)).shl(*ShAmt).isNullValue())
+      return Op0;
+  }
+
   // A & (-A) = A if A is a power of two or zero.
   if (match(Op0, m_Neg(m_Specific(Op1))) ||
       match(Op1, m_Neg(m_Specific(Op0)))) {
diff --git a/llvm/test/Transforms/InstSimplify/AndOrXor.ll b/llvm/test/Transforms/InstSimplify/AndOrXor.ll
index 427ea655fcb..a9b4e4e5cfc 100644
--- a/llvm/test/Transforms/InstSimplify/AndOrXor.ll
+++ b/llvm/test/Transforms/InstSimplify/AndOrXor.ll
@@ -738,8 +738,7 @@ define i32 @test54(i32 %a, i32 %b) {
 define i8 @lshr_perfect_mask(i8 %x) {
 ; CHECK-LABEL: @lshr_perfect_mask(
 ; CHECK-NEXT:    [[SH:%.*]] = lshr i8 %x, 5
-; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[SH]], 7
-; CHECK-NEXT:    ret i8 [[MASK]]
+; CHECK-NEXT:    ret i8 [[SH]]
 ;
   %sh = lshr i8 %x, 5
   %mask = and i8 %sh, 7  ; 0x07
@@ -749,8 +748,7 @@ define i8 @lshr_perfect_mask(i8 %x) {
 define <2 x i8> @lshr_oversized_mask_splat(<2 x i8> %x) {
 ; CHECK-LABEL: @lshr_oversized_mask_splat(
 ; CHECK-NEXT:    [[SH:%.*]] = lshr <2 x i8> %x, <i8 5, i8 5>
-; CHECK-NEXT:    [[MASK:%.*]] = and <2 x i8> [[SH]], <i8 -121, i8 -121>
-; CHECK-NEXT:    ret <2 x i8> [[MASK]]
+; CHECK-NEXT:    ret <2 x i8> [[SH]]
 ;
   %sh = lshr <2 x i8> %x, <i8 5, i8 5>
   %mask = and <2 x i8> %sh, <i8 135, i8 135>  ; 0x87
@@ -771,8 +769,7 @@ define i8 @lshr_undersized_mask(i8 %x) {
 define <2 x i8> @shl_perfect_mask_splat(<2 x i8> %x) {
 ; CHECK-LABEL: @shl_perfect_mask_splat(
 ; CHECK-NEXT:    [[SH:%.*]] = shl <2 x i8> %x, <i8 6, i8 6>
-; CHECK-NEXT:    [[MASK:%.*]] = and <2 x i8> [[SH]], <i8 -64, i8 -64>
-; CHECK-NEXT:    ret <2 x i8> [[MASK]]
+; CHECK-NEXT:    ret <2 x i8> [[SH]]
 ;
   %sh = shl <2 x i8> %x, <i8 6, i8 6>
   %mask = and <2 x i8> %sh, <i8 192, i8 192>  ; 0xC0
@@ -782,8 +779,7 @@ define <2 x i8> @shl_perfect_mask_splat(<2 x i8> %x) {
 define i8 @shl_oversized_mask(i8 %x) {
 ; CHECK-LABEL: @shl_oversized_mask(
 ; CHECK-NEXT:    [[SH:%.*]] = shl i8 %x, 6
-; CHECK-NEXT:    [[MASK:%.*]] = and i8 [[SH]], -61
-; CHECK-NEXT:    ret i8 [[MASK]]
+; CHECK-NEXT:    ret i8 [[SH]]
 ;
   %sh = shl i8 %x, 6
   %mask = and i8 %sh, 195  ; 0xC3