From 1c7e337a3621ef67c5f7a41db42e9f96cd74d618 Mon Sep 17 00:00:00 2001 From: Roman Lebedev Date: Mon, 7 Oct 2019 20:52:52 +0000 Subject: [PATCH] [InstCombine] dropRedundantMaskingOfLeftShiftInput(): propagate undef shift amounts Summary: When we do `ConstantExpr::getZExt()`, that "extends" `undef` to `0`, which means that for patterns a/b we'd assume that we must not produce any bits for that channel, while in reality we simply didn't care about that channel - i.e. we don't need to mask it. Reviewers: spatel Reviewed By: spatel Subscribers: hiraditya, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D68239 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@373960 91177308-0d34-0410-b5e6-96231b3b80d8 --- .../InstCombine/InstCombineShifts.cpp | 33 +++++++++++++++++++ ...dant-left-shift-input-masking-variant-a.ll | 2 +- ...dant-left-shift-input-masking-variant-b.ll | 2 +- ...dant-left-shift-input-masking-variant-c.ll | 2 +- ...dant-left-shift-input-masking-variant-d.ll | 2 +- ...dant-left-shift-input-masking-variant-e.ll | 2 +- 6 files changed, 38 insertions(+), 5 deletions(-) diff --git a/lib/Transforms/InstCombine/InstCombineShifts.cpp b/lib/Transforms/InstCombine/InstCombineShifts.cpp index 8ab4aeb38be..6675ab12aee 100644 --- a/lib/Transforms/InstCombine/InstCombineShifts.cpp +++ b/lib/Transforms/InstCombine/InstCombineShifts.cpp @@ -117,6 +117,24 @@ reassociateShiftAmtsOfTwoSameDirectionShifts(BinaryOperator *Sh0, return Ret; } +// Try to replace `undef` constants in C with Replacement. +static Constant *replaceUndefsWith(Constant *C, Constant *Replacement) { + if (C && match(C, m_Undef())) + return Replacement; + + if (auto *CV = dyn_cast(C)) { + llvm::SmallVector NewOps(CV->getNumOperands()); + for (unsigned i = 0, NumElts = NewOps.size(); i != NumElts; ++i) { + Constant *EltC = CV->getOperand(i); + NewOps[i] = EltC && match(EltC, m_Undef()) ? Replacement : EltC; + } + return ConstantVector::get(NewOps); + } + + // Don't know how to deal with this constant. + return C; +} + // If we have some pattern that leaves only some low bits set, and then performs // left-shift of those bits, if none of the bits that are left after the final // shift are modified by the mask, we can omit the mask. @@ -177,6 +195,14 @@ dropRedundantMaskingOfLeftShiftInput(BinaryOperator *OuterShift, // The mask must be computed in a type twice as wide to ensure // that no bits are lost if the sum-of-shifts is wider than the base type. Type *ExtendedTy = Ty->getExtendedType(); + // An extend of an undef value becomes zero because the high bits are + // never completely unknown. Replace the the `undef` shift amounts with + // final shift bitwidth to ensure that the value remains undef when + // creating the subsequent shift op. + SumOfShAmts = replaceUndefsWith( + SumOfShAmts, + ConstantInt::get(SumOfShAmts->getType()->getScalarType(), + ExtendedTy->getScalarType()->getScalarSizeInBits())); auto *ExtendedSumOfShAmts = ConstantExpr::getZExt(SumOfShAmts, ExtendedTy); // And compute the mask as usual: ~(-1 << (SumOfShAmts)) @@ -212,6 +238,13 @@ dropRedundantMaskingOfLeftShiftInput(BinaryOperator *OuterShift, // The mask must be computed in a type twice as wide to ensure // that no bits are lost if the sum-of-shifts is wider than the base type. Type *ExtendedTy = Ty->getExtendedType(); + // An extend of an undef value becomes zero because the high bits are + // never completely unknown. Replace the the `undef` shift amounts with + // negated shift bitwidth to ensure that the value remains undef when + // creating the subsequent shift op. + ShAmtsDiff = replaceUndefsWith( + ShAmtsDiff, + ConstantInt::get(ShAmtsDiff->getType()->getScalarType(), -BitWidth)); auto *ExtendedNumHighBitsToClear = ConstantExpr::getZExt( ConstantExpr::getAdd( ConstantExpr::getNeg(ShAmtsDiff), diff --git a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-a.ll b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-a.ll index bcaf6440efc..205dcfbf22c 100644 --- a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-a.ll +++ b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-a.ll @@ -82,7 +82,7 @@ define <8 x i32> @t1_vec_splat_undef(<8 x i32> %x, <8 x i32> %nbits) { ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T2]]) ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T4]]) ; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i32> [[X:%.*]], [[T4]] -; CHECK-NEXT: [[T5:%.*]] = and <8 x i32> [[TMP1]], +; CHECK-NEXT: [[T5:%.*]] = and <8 x i32> [[TMP1]], ; CHECK-NEXT: ret <8 x i32> [[T5]] ; %t0 = add <8 x i32> %nbits, diff --git a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-b.ll b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-b.ll index faf069cfec9..4d3d5432375 100644 --- a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-b.ll +++ b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-b.ll @@ -82,7 +82,7 @@ define <8 x i32> @t1_vec_splat_undef(<8 x i32> %x, <8 x i32> %nbits) { ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T2]]) ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T4]]) ; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i32> [[X:%.*]], [[T4]] -; CHECK-NEXT: [[T5:%.*]] = and <8 x i32> [[TMP1]], +; CHECK-NEXT: [[T5:%.*]] = and <8 x i32> [[TMP1]], ; CHECK-NEXT: ret <8 x i32> [[T5]] ; %t0 = add <8 x i32> %nbits, diff --git a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-c.ll b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-c.ll index 2c6bb31ebf7..fc964f8725b 100644 --- a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-c.ll +++ b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-c.ll @@ -62,7 +62,7 @@ define <8 x i32> @t1_vec_splat_undef(<8 x i32> %x, <8 x i32> %nbits) { ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T0]]) ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T2]]) ; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i32> [[X:%.*]], [[T2]] -; CHECK-NEXT: [[T3:%.*]] = and <8 x i32> [[TMP1]], +; CHECK-NEXT: [[T3:%.*]] = and <8 x i32> [[TMP1]], ; CHECK-NEXT: ret <8 x i32> [[T3]] ; %t0 = lshr <8 x i32> , %nbits diff --git a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-d.ll b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-d.ll index fcbb7eb50d5..83c0df1d83e 100644 --- a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-d.ll +++ b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-d.ll @@ -72,7 +72,7 @@ define <8 x i32> @t2_vec_splat_undef(<8 x i32> %x, <8 x i32> %nbits) { ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T1]]) ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T3]]) ; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i32> [[X:%.*]], [[T3]] -; CHECK-NEXT: [[T4:%.*]] = and <8 x i32> [[TMP1]], +; CHECK-NEXT: [[T4:%.*]] = and <8 x i32> [[TMP1]], ; CHECK-NEXT: ret <8 x i32> [[T4]] ; %t0 = shl <8 x i32> , %nbits diff --git a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-e.ll b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-e.ll index 2b3fb5098aa..200d414d825 100644 --- a/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-e.ll +++ b/test/Transforms/InstCombine/partally-redundant-left-shift-input-masking-variant-e.ll @@ -62,7 +62,7 @@ define <8 x i32> @t1_vec_splat_undef(<8 x i32> %x, <8 x i32> %nbits) { ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T0]]) ; CHECK-NEXT: call void @use8xi32(<8 x i32> [[T2]]) ; CHECK-NEXT: [[TMP1:%.*]] = shl <8 x i32> [[X]], [[T2]] -; CHECK-NEXT: [[T3:%.*]] = and <8 x i32> [[TMP1]], +; CHECK-NEXT: [[T3:%.*]] = and <8 x i32> [[TMP1]], ; CHECK-NEXT: ret <8 x i32> [[T3]] ; %t0 = shl <8 x i32> %x, %nbits -- 2.40.0