This is an archive of the discontinued LLVM Phabricator instance.

Differential D55223

[SelectionDAG] Don't pass on DemandedElts when handling SCALAR_TO_VECTOR
ClosedPublic

Authored by pendingchaos on Dec 3 2018, 8:01 AM.

Details

Reviewers
rampitec
RKSimon
craig.topper
Commits
rGd498dee7a28c: [SelectionDAG] Don't pass on DemandedElts when handling SCALAR_TO_VECTOR
rL348574: [SelectionDAG] Don't pass on DemandedElts when handling SCALAR_TO_VECTOR
Summary

Fixes an assertion:

llc: lib/CodeGen/SelectionDAG/SelectionDAG.cpp:2200: llvm::KnownBits llvm::SelectionDAG::computeKnownBits(llvm::SDValue, const llvm::APInt&, unsigned int) const: Assertion `(!Op.getValueType().isVector() || NumElts == Op.getValueType().getVectorNumElements()) && "Unexpected vector size"' failed.

Diff Detail

Event Timeline

pendingchaos created this revision.Dec 3 2018, 8:01 AM
Herald added subscribers: nhaehnle, jvesely. · View Herald TranscriptDec 3 2018, 8:01 AM
lebedev.ri added a subscriber: lebedev.ri.Dec 3 2018, 8:03 AM
RKSimon added inline comments.Dec 3 2018, 8:09 AM
test/CodeGen/AMDGPU/computeKnownBits-scalar-to-vector-crash.ll
1

Even though this is a crash test its probably worth adding at least some basic FileCheck CHECKS (or just run update_llc_test_checks)

pendingchaos updated this revision to Diff 176985.Dec 6 2018, 8:31 AM

Makes the test case use FileCheck.

I don't have commit access.

pendingchaos marked an inline comment as done.Dec 6 2018, 8:32 AM
RKSimon accepted this revision.Dec 6 2018, 9:52 AM

LGTM - I'll commit shortly but please make sure you've requested commit access

This revision is now accepted and ready to land.Dec 6 2018, 9:52 AM
Closed by commit rL348574: [SelectionDAG] Don't pass on DemandedElts when handling SCALAR_TO_VECTOR (authored by RKSimon). · Explain WhyDec 7 2018, 1:21 AM
This revision was automatically updated to reflect the committed changes.

Revision Contents

PathSize
lib/
CodeGen/
SelectionDAG/
2 lines
test/
CodeGen/
AMDGPU/
10 lines

Diff 176410

lib/CodeGen/SelectionDAG/SelectionDAG.cpp

Context not available.
break; break;
SDValue N0 = Op.getOperand(0); SDValue N0 = Op.getOperand(0);
Known = computeKnownBits(N0, DemandedElts, Depth + 1); Known = computeKnownBits(N0, Depth + 1);
if (N0.getValueSizeInBits() != BitWidth) if (N0.getValueSizeInBits() != BitWidth)
Known = Known.trunc(BitWidth); Known = Known.trunc(BitWidth);
Context not available.

test/CodeGen/AMDGPU/computeKnownBits-scalar-to-vector-crash.ll

  • This file was added.
; RUN: llc -march=amdgcn -mcpu=gfx802 -verify-machineinstrs < %s
RKSimonUnsubmitted
Done
ReplyInline Actions

Even though this is a crash test its probably worth adding at least some basic FileCheck CHECKS (or just run update_llc_test_checks)

RKSimon: Even though this is a crash test its probably worth adding at least some basic FileCheck CHECKS…
define <2 x i16> @main(<2 x float>) #0 {
%2 = bitcast <2 x float> %0 to <4 x i16>
%3 = shufflevector <4 x i16> %2, <4 x i16> undef, <2 x i32> <i32 0, i32 undef>
%4 = extractelement <4 x i16> %2, i32 0
%5 = insertelement <2 x i16> %3, i16 %4, i32 0
ret <2 x i16> %5
}