This is an archive of the discontinued LLVM Phabricator instance.

[LIR] Avoid turning non-temporal stores into memset.
ClosedPublic

Authored by haicheng on Feb 17 2016, 12:04 PM.

Download Raw Diff

Details

Reviewers

gberry
mssimpso
mcrosier

Summary

This is to fix PR26645.

Diff Detail

Repository: rL LLVM

Event Timeline

haicheng updated this revision to Diff 48227.Feb 17 2016, 12:04 PM

haicheng retitled this revision from to [LIR] Avoid turning non-temporal stores into memset..

haicheng updated this object.

haicheng added reviewers: mcrosier, gberry, mssimpso.

haicheng set the repository for this revision to rL LLVM.

haicheng added a subscriber: llvm-commits.

Herald added subscribers: mzolotukhin, mcrosier. · View Herald TranscriptFeb 17 2016, 12:04 PM

haicheng added a subscriber: wristow.Feb 17 2016, 12:09 PM

LGTM. Thanks for addressing this quickly, Haicheng.

This revision is now accepted and ready to land.Feb 17 2016, 12:11 PM

Committed in r261149.

Revision Contents

Path

Size

lib/

Transforms/

Scalar/

LoopIdiomRecognize.cpp

4 lines

test/

Transforms/

LoopIdiom/

nontemporal_store.ll

31 lines

Diff 48227

lib/Transforms/Scalar/LoopIdiomRecognize.cpp

	Show First 20 Lines • Show All 308 Lines • ▼ Show 20 Lines
	}			}

	bool LoopIdiomRecognize::isLegalStore(StoreInst *SI, bool &ForMemset,			bool LoopIdiomRecognize::isLegalStore(StoreInst *SI, bool &ForMemset,
	bool &ForMemsetPattern, bool &ForMemcpy) {			bool &ForMemsetPattern, bool &ForMemcpy) {
	// Don't touch volatile stores.			// Don't touch volatile stores.
	if (!SI->isSimple())			if (!SI->isSimple())
	return false;			return false;

				// Avoid merging nontemporal stores.
				if (SI->getMetadata(LLVMContext::MD_nontemporal))
				return false;

	Value *StoredVal = SI->getValueOperand();			Value *StoredVal = SI->getValueOperand();
	Value *StorePtr = SI->getPointerOperand();			Value *StorePtr = SI->getPointerOperand();

	// Reject stores that are so large that they overflow an unsigned.			// Reject stores that are so large that they overflow an unsigned.
	uint64_t SizeInBits = DL->getTypeSizeInBits(StoredVal->getType());			uint64_t SizeInBits = DL->getTypeSizeInBits(StoredVal->getType());
	if ((SizeInBits & 7) \|\| (SizeInBits >> 32) != 0)			if ((SizeInBits & 7) \|\| (SizeInBits >> 32) != 0)
	return false;			return false;

	▲ Show 20 Lines • Show All 933 Lines • Show Last 20 Lines

test/Transforms/LoopIdiom/nontemporal_store.ll

This file was added.

				; RUN: opt -loop-idiom < %s -S \| FileCheck %s
				target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
				target triple = "x86_64-unknown-linux-gnu"

				%struct.bigBlock_t = type { [256 x <4 x float>] }

				; CHECK-LABEL: @test(
				; CHECK-NOT: llvm.memset
				define void @test(%struct.bigBlock_t* %p) {
				entry:
				%0 = getelementptr inbounds %struct.bigBlock_t, %struct.bigBlock_t* %p, i64 0, i32 0, i64 0, i64 0
				br label %for.body

				for.body: ; preds = %entry, %for.body
				%index.02 = phi i32 [ 0, %entry ], [ %add, %for.body ]
				%dst.01 = phi float* [ %0, %entry ], [ %add.ptr2, %for.body ]
				%cast.i5 = bitcast float* %dst.01 to <4 x float>*
				store <4 x float> zeroinitializer, <4 x float>* %cast.i5, align 16, !nontemporal !0
				%add.ptr1 = getelementptr inbounds float, float* %dst.01, i64 4
				%cast.i = bitcast float* %add.ptr1 to <4 x float>*
				store <4 x float> zeroinitializer, <4 x float>* %cast.i, align 16, !nontemporal !0
				%add.ptr2 = getelementptr inbounds float, float* %dst.01, i64 8
				%add = add nuw nsw i32 %index.02, 32
				%cmp = icmp ult i32 %add, 4096
				br i1 %cmp, label %for.body, label %for.end

				for.end: ; preds = %for.body
				ret void
				}

				!0 = !{i32 1}