RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2025-01-24 21:25:41 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

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

Browse Source 
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



git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@303213 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Sanjay Patel 2017-05-16 21:51:04 +00:00
parent 10dbbfe1c9
commit 25529caa1b
2 changed files with 22 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

18
lib/Analysis/InstructionSimplify.cpp
View File

@ -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)))) {

12
test/Transforms/InstSimplify/AndOrXor.ll
View File

@ -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