This is an archive of the discontinued LLVM Phabricator instance.

[Codegen] Updated test for D66158
ClosedPublic

Authored by xbolva00 on Aug 13 2019, 2:41 PM.

Download Raw Diff

Details

Reviewers

Commits

rGb0a8a2544207: [Codegen] Updated test for D66158
rL368809: [Codegen] Updated test for D66158
rC368809: [Codegen] Updated test for D66158

Diff Detail

Event Timeline

xbolva00 created this revision.Aug 13 2019, 2:41 PM

Herald added a project: Restricted Project. · View Herald TranscriptAug 13 2019, 2:41 PM

Herald added a subscriber: cfe-commits. · View Herald Transcript

xbolva00 added a reviewer: jdoerfert.Aug 13 2019, 2:41 PM

Might not be the best test given that this should arguably be a memset not a memcpy.

hm.. memset?

Explicit memcpy

In D66173#1628079, @xbolva00 wrote:

hm.. memset?

When I say memset, here and in the other review, I mean memmove.

In D66173#1628088, @xbolva00 wrote:

Explicit memcpy

thx. LGTM.

This revision is now accepted and ready to land.Aug 13 2019, 3:23 PM

xbolva00 updated this revision to Diff 215039.Aug 14 2019, 1:18 AM

Closed by commit rL368809: [Codegen] Updated test for D66158 (authored by xbolva00). · Explain WhyAug 14 2019, 1:33 AM

This revision was automatically updated to reflect the committed changes.

Herald added a project: Restricted Project. · View Herald TranscriptAug 14 2019, 1:33 AM

Herald added a subscriber: llvm-commits. · View Herald Transcript

Revision Contents

Path

Size

test/

CodeGen/

struct-copy.c

36 lines

Diff 215039

test/CodeGen/struct-copy.c

	// RUN: %clang_cc1 -emit-llvm %s -o - \| grep 'call.*llvm.memcpy'			// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
				// RUN: %clang_cc1 -emit-llvm %s -o - \| FileCheck %s
	struct x { int a[100]; };			struct x { int a[100]; };

				// CHECK-LABEL: @foo(
				// CHECK-NEXT: entry:
				// CHECK-NEXT: [[P_ADDR:%.]] = alloca %struct.x, align 8
				// CHECK-NEXT: [[Q_ADDR:%.]] = alloca %struct.x, align 8
				// CHECK-NEXT: store %struct.x* [[P:%.]], %struct.x* [[P_ADDR]], align 8
				// CHECK-NEXT: store %struct.x* [[Q:%.]], %struct.x* [[Q_ADDR]], align 8
				// CHECK-NEXT: [[TMP0:%.]] = load %struct.x, %struct.x** [[P_ADDR]], align 8
				// CHECK-NEXT: [[TMP1:%.]] = load %struct.x, %struct.x** [[Q_ADDR]], align 8
				// CHECK-NEXT: [[TMP2:%.]] = bitcast %struct.x [[TMP0]] to i8*
				// CHECK-NEXT: [[TMP3:%.]] = bitcast %struct.x [[TMP1]] to i8*
				// CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP2]], i8* align 4 [[TMP3]], i64 400, i1 false)
				// CHECK-NEXT: ret void
				//
	void foo(struct x P, struct x Q) {			void foo(struct x P, struct x Q) {
	P = Q;			P = Q;
	}			}

				// CHECK: declare void @llvm.memcpy.p0i8.p0i8.i64(i8* noalias nocapture writeonly, i8* noalias nocapture readonly, i64, i1 immarg)

				// CHECK-LABEL: @bar(
				// CHECK-NEXT: entry:
				// CHECK-NEXT: [[P_ADDR:%.]] = alloca %struct.x, align 8
				// CHECK-NEXT: [[Q_ADDR:%.]] = alloca %struct.x, align 8
				// CHECK-NEXT: store %struct.x* [[P:%.]], %struct.x* [[P_ADDR]], align 8
				// CHECK-NEXT: store %struct.x* [[Q:%.]], %struct.x* [[Q_ADDR]], align 8
				// CHECK-NEXT: [[TMP0:%.]] = load %struct.x, %struct.x** [[P_ADDR]], align 8
				// CHECK-NEXT: [[TMP1:%.]] = bitcast %struct.x [[TMP0]] to i8*
				// CHECK-NEXT: [[TMP2:%.]] = load %struct.x, %struct.x** [[Q_ADDR]], align 8
				// CHECK-NEXT: [[TMP3:%.]] = bitcast %struct.x [[TMP2]] to i8*
				// CHECK-NEXT: call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 4 [[TMP1]], i8* align 4 [[TMP3]], i64 400, i1 false)
				// CHECK-NEXT: ret void
				//
				void bar(struct x P, struct x Q) {
				__builtin_memcpy(P, Q, sizeof(struct x));
				}