maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[Hexagon] Fix reuse bug in Vector Loop Carried Reuse pass 

The Hexagon Vector Loop Carried Reuse pass was allowing reuse between
two shufflevectors with different masks. The reason is that the masks
are not instruction objects, so the code that checks each operand
just skipped over the operands.

This patch fixes the bug by checking if the operands are the same
when they are not instruction objects. If the objects are not the
same, then the code assumes that reuse cannot occur.

Differential Revision: https://reviews.llvm.org/D60019

llvm-svn: 358292
This commit is contained in:
Brendon Cahoon 2019-04-12 16:37:12 +00:00
parent 5e4ad39af7
commit 4df216cd62
2 changed files with 38 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
llvm/lib/Target/Hexagon/HexagonVectorLoopCarriedReuse.cpp
View File

@ -429,11 +429,18 @@ void HexagonVectorLoopCarriedReuse::findValueToReuse() {
for (int OpNo = 0; OpNo < NumOperands; ++OpNo) {
Value *Op = I->getOperand(OpNo);
Instruction *OpInst = dyn_cast<Instruction>(Op);
if (!OpInst)
continue;
Value *BEOp = BEUser->getOperand(OpNo);
Instruction *OpInst = dyn_cast<Instruction>(Op);
if (!OpInst) {
if (Op == BEOp)
continue;
// Do not allow reuse to occur when the operands may be different
// values.
BEUser = nullptr;
break;
}
Instruction *BEOpInst = dyn_cast<Instruction>(BEOp);
if (!isDepChainBtwn(OpInst, BEOpInst, Iters)) {

27
llvm/test/CodeGen/Hexagon/hexagon_vector_loop_carried_reuse_invalid.ll Normal file
View File

@ -0,0 +1,27 @@
; RUN: opt -hexagon-vlcr < %s -S | FileCheck %s
; Test that reuse doesn't occur due to two shufflevectors with different masks.
; CHECK-NOT: extract.h.hexagon.vlcr
define dso_local void @test() local_unnamed_addr #0 {
entry:
%0 = tail call <64 x i32> @llvm.hexagon.V6.vunpackuh.128B(<32 x i32> undef)
br i1 undef, label %for.end, label %for.body
for.body:
%a = phi <64 x i32> [ %1, %for.body ], [ %0, %entry ]
%extract.h = shufflevector <64 x i32> %a, <64 x i32> undef, <32 x i32> <i32 32, i32 33, i32 34, i32 35, i32 36, i32 37, i32 38, i32 39, i32 40, i32 41, i32 42, i32 43, i32 44, i32 45, i32 46, i32 47, i32 48, i32 49, i32 50, i32 51, i32 52, i32 53, i32 54, i32 55, i32 56, i32 57, i32 58, i32 59, i32 60, i32 61, i32 62, i32 63>
%1 = tail call <64 x i32> @llvm.hexagon.V6.vaddw.dv.128B(<64 x i32> undef, <64 x i32> undef)
%extract.l = shufflevector <64 x i32> %1, <64 x i32> undef, <32 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
br label %for.body
for.end:
ret void
}
declare <64 x i32> @llvm.hexagon.V6.vunpackuh.128B(<32 x i32>) #1
declare <64 x i32> @llvm.hexagon.V6.vaddw.dv.128B(<64 x i32>, <64 x i32>) #1
attributes #0 = { "use-soft-float"="false" }
attributes #1 = { nounwind readnone }