InstCombine: Turn (zext A) udiv (zext B) into (zext (A udiv B)). Same for urem or constant B.

This obviously helps a lot if the division would be turned into a libcall
(think i64 udiv on i386), but div is also one of the few remaining instructions
on modern CPUs that become more expensive when the bitwidth gets bigger.

This also helps register pressure on i386 when dividing chars, divb needs
two 8-bit parts of a 16 bit register as input where divl uses two registers.

int foo(unsigned char a) { return a/10; }
int bar(unsigned char a, unsigned char b) { return a/b; }

compiles into (x86_64)
_foo:
  imull $205, %edi, %eax
  shrl  $11, %eax
  ret
_bar:
  movzbl        %dil, %eax
  divb  %sil, %al
  movzbl        %al, %eax
  ret

llvm-svn: 130615

llvm/lib/Transforms/InstCombine/InstCombineMulDivRem.cpp

@@ -336,6 +336,19 @@ Instruction *InstCombiner::commonIDivTransforms(BinaryOperator &I) {
   return 0;
 }
 
+/// dyn_castZExtVal - Checks if V is a zext or constant that can
+/// be truncated to Ty without losing bits.
+static Value *dyn_castZExtVal(Value *V, const Type *Ty) {
+  if (ZExtInst *Z = dyn_cast<ZExtInst>(V)) {
+    if (Z->getSrcTy() == Ty)
+      return Z->getOperand(0);
+  } else if (ConstantInt *C = dyn_cast<ConstantInt>(V)) {
+    if (C->getValue().getActiveBits() <= cast<IntegerType>(Ty)->getBitWidth())
+      return ConstantExpr::getTrunc(C, Ty);
+  }
+  return 0;
+}
+
 Instruction *InstCombiner::visitUDiv(BinaryOperator &I) {
   Value *Op0 = I.getOperand(0), *Op1 = I.getOperand(1);
 
@@ -394,6 +407,14 @@ Instruction *InstCombiner::visitUDiv(BinaryOperator &I) {
       return SelectInst::Create(Cond, TSI, FSI);
     }
   }
+
+  // (zext A) udiv (zext B) --> zext (A udiv B)
+  if (ZExtInst *ZOp0 = dyn_cast<ZExtInst>(Op0))
+    if (Value *ZOp1 = dyn_castZExtVal(Op1, ZOp0->getSrcTy()))
+      return new ZExtInst(Builder->CreateUDiv(ZOp0->getOperand(0), ZOp1, "div",
+                                              I.isExact()),
+                          I.getType());
+
   return 0;
 }
 
@@ -568,7 +589,13 @@ Instruction *InstCombiner::visitURem(BinaryOperator &I) {
       return SelectInst::Create(Cond, TrueAnd, FalseAnd);
     }
   }
-  
+
+  // (zext A) urem (zext B) --> zext (A urem B)
+  if (ZExtInst *ZOp0 = dyn_cast<ZExtInst>(Op0))
+    if (Value *ZOp1 = dyn_castZExtVal(Op1, ZOp0->getSrcTy()))
+      return new ZExtInst(Builder->CreateURem(ZOp0->getOperand(0), ZOp1),
+                          I.getType());
+
   return 0;
 }
 

llvm/test/Transforms/InstCombine/udivrem-change-width.ll

-; RUN: opt < %s -instcombine -S | not grep zext
-; PR4548
+; RUN: opt < %s -instcombine -S | FileCheck %s
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128"
 
+; PR4548
 define i8 @udiv_i8(i8 %a, i8 %b) nounwind {
   %conv = zext i8 %a to i32       
   %conv2 = zext i8 %b to i32      
   %div = udiv i32 %conv, %conv2   
   %conv3 = trunc i32 %div to i8   
   ret i8 %conv3
+; CHECK: @udiv_i8
+; CHECK: udiv i8 %a, %b
 }
 
 define i8 @urem_i8(i8 %a, i8 %b) nounwind {
@@ -17,5 +19,44 @@ define i8 @urem_i8(i8 %a, i8 %b) nounwind {
   %div = urem i32 %conv, %conv2   
   %conv3 = trunc i32 %div to i8   
   ret i8 %conv3
+; CHECK: @urem_i8
+; CHECK: urem i8 %a, %b
 }
 
+define i32 @udiv_i32(i8 %a, i8 %b) nounwind {
+  %conv = zext i8 %a to i32
+  %conv2 = zext i8 %b to i32
+  %div = udiv i32 %conv, %conv2
+  ret i32 %div
+; CHECK: @udiv_i32
+; CHECK: udiv i8 %a, %b
+; CHECK: zext
+}
+
+define i32 @urem_i32(i8 %a, i8 %b) nounwind {
+  %conv = zext i8 %a to i32
+  %conv2 = zext i8 %b to i32
+  %div = urem i32 %conv, %conv2
+  ret i32 %div
+; CHECK: @urem_i32
+; CHECK: urem i8 %a, %b
+; CHECK: zext
+}
+
+define i32 @udiv_i32_c(i8 %a) nounwind {
+  %conv = zext i8 %a to i32
+  %div = udiv i32 %conv, 10
+  ret i32 %div
+; CHECK: @udiv_i32_c
+; CHECK: udiv i8 %a, 10
+; CHECK: zext
+}
+
+define i32 @urem_i32_c(i8 %a) nounwind {
+  %conv = zext i8 %a to i32
+  %div = urem i32 %conv, 10
+  ret i32 %div
+; CHECK: @urem_i32_c
+; CHECK: urem i8 %a, 10
+; CHECK: zext
+}