maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGenObjC/arc-precise-lifetime.m

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

323 lines
17 KiB
Mathematica
Raw Normal View History

Make '-disable-llvm-optzns' an alias for '-disable-llvm-passes'. Much to my surprise, '-disable-llvm-optzns' which I thought was the magical flag I wanted to get at the raw LLVM IR coming out of Clang deosn't do that. It still runs some passes over the IR. I don't want that, I really want the *raw* IR coming out of Clang and I strongly suspect everyone else using it is in the same camp. There is actually a flag that does what I want that I didn't know about called '-disable-llvm-passes'. I suspect many others don't know about it either. It both does what I want and is much simpler. This removes the confusing version and makes that spelling of the flag an alias for '-disable-llvm-passes'. I've also moved everything in Clang to use the 'passes' spelling as it seems both more accurate (*all* LLVM passes are disabled, not just optimizations) and much easier to remember and spell correctly. This is part of simplifying how Clang drives LLVM to make it cleaner to wire up to the new pass manager. Differential Revision: https://reviews.llvm.org/D28047 llvm-svn: 290392
2016-12-23 08:23:01 +08:00
// RUN: %clang_cc1 -triple x86_64-apple-darwin10 -emit-llvm -fblocks -fobjc-arc -fobjc-runtime-has-weak -O2 -disable-llvm-passes -o - %s | FileCheck %s
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
#define PRECISE_LIFETIME __attribute__((objc_precise_lifetime))
id test0_helper(void) __attribute__((ns_returns_retained));
void test0() {
PRECISE_LIFETIME id x = test0_helper();
x = 0;
// CHECK: [[X:%.*]] = alloca i8*
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[XPTR1:%.*]] = bitcast i8** [[X]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[XPTR1]])
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[CALL:%.*]] = call i8* @test0_helper()
// CHECK-NEXT: store i8* [[CALL]], i8** [[X]]
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T1:%.*]] = load i8*, i8** [[X]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: store i8* null, i8** [[X]]
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW:#[0-9]+]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NOT: clang.imprecise_release
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T1:%.*]] = load i8*, i8** [[X]]
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW:#[0-9]+]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NOT: clang.imprecise_release
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[XPTR2:%.*]] = bitcast i8** [[X]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[XPTR2]])
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: ret void
}
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// rdar://problem/9821110 - precise lifetime should suppress extension
// rdar://problem/22172983 - should work for calls via property syntax, too
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
@interface Test1
- (char*) interior __attribute__((objc_returns_inner_pointer));
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// Should we allow this on properties? Yes! see // rdar://14990439
@property (nonatomic, readonly) char * PropertyReturnsInnerPointer __attribute__((objc_returns_inner_pointer));
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
@end
extern Test1 *test1_helper(void);
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1a_message()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1a_message(void) {
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[C:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[CPTR1:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[CPTR1]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T2:%.*]] = call i8* @llvm.objc.retainAutorelease(i8* [[T1]])
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T3:%.*]] = bitcast i8* [[T2]] to [[TEST1]]*
// CHECK-NEXT: [[T5:%.*]] = bitcast [[TEST1]]* [[T3]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T4:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T6:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T6]], i8**
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[CPTR2:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[CPTR2]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]], !clang.imprecise_release
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: ret void
Test1 *ptr = test1_helper();
char *c = [(ptr) interior];
}
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1a_property()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1a_property(void) {
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[C:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[CPTR1:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[CPTR1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T2:%.*]] = call i8* @llvm.objc.retainAutorelease(i8* [[T1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T3:%.*]] = bitcast i8* [[T2]] to [[TEST1]]*
// CHECK-NEXT: [[T5:%.*]] = bitcast [[TEST1]]* [[T3]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T4:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T6:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T6]], i8**
// CHECK-NEXT: [[CPTR2:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[CPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]], !clang.imprecise_release
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: ret void
Test1 *ptr = test1_helper();
char *c = ptr.interior;
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1b_message()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1b_message(void) {
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[C:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[CPTR1:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[CPTR1]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T2:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T1:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T3:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T3]], i8**
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[CPTR2:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[CPTR2]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NOT: clang.imprecise_release
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: ret void
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
PRECISE_LIFETIME Test1 *ptr = test1_helper();
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
char *c = [ptr interior];
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1b_property()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1b_property(void) {
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[C:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[CPTR1:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[CPTR1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T2:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T1:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T3:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T3]], i8**
// CHECK-NEXT: [[CPTR2:%.*]] = bitcast i8** [[C]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[CPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]]
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NOT: clang.imprecise_release
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: ret void
PRECISE_LIFETIME Test1 *ptr = test1_helper();
char *c = ptr.interior;
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1c_message()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1c_message(void) {
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[PC:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[PCPTR1:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PCPTR1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T2:%.*]] = call i8* @llvm.objc.retainAutorelease(i8* [[T1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T3:%.*]] = bitcast i8* [[T2]] to [[TEST1]]*
// CHECK-NEXT: [[T5:%.*]] = bitcast [[TEST1]]* [[T3]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T4:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T6:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T6]], i8**
// CHECK-NEXT: [[PCPTR2:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PCPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]], !clang.imprecise_release
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: ret void
Test1 *ptr = test1_helper();
char *pc = [ptr PropertyReturnsInnerPointer];
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1c_property()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1c_property(void) {
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[PC:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PCPTR1:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PCPTR1]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T2:%.*]] = call i8* @llvm.objc.retainAutorelease(i8* [[T1]])
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: [[T3:%.*]] = bitcast i8* [[T2]] to [[TEST1]]*
// CHECK-NEXT: [[T5:%.*]] = bitcast [[TEST1]]* [[T3]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[T4:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: [[T6:%.*]] = call i8* bitcast
// CHECK-NEXT: store i8* [[T6]], i8**
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PCPTR2:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PCPTR2]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]], !clang.imprecise_release
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: ret void
Test1 *ptr = test1_helper();
char *pc = ptr.PropertyReturnsInnerPointer;
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1d_message()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1d_message(void) {
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[PC:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[PCPTR1:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PCPTR1]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[EIGHT:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[SEL:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
// CHECK-NEXT: [[CALL1:%.*]] = call i8* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to i8* (i8*, i8*)*)(i8* noundef [[EIGHT]], i8* noundef [[SEL]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: store i8* [[CALL1]], i8**
// CHECK-NEXT: [[PCPTR2:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PCPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[NINE:%.*]] = load [[TEST1]]*, [[TEST1]]**
// CHECK-NEXT: [[TEN:%.*]] = bitcast [[TEST1]]* [[NINE]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[TEN]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: ret void
PRECISE_LIFETIME Test1 *ptr = test1_helper();
char *pc = [ptr PropertyReturnsInnerPointer];
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test1d_property()
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
void test1d_property(void) {
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK: [[PTR:%.*]] = alloca [[PTR_T:%.*]]*, align 8
// CHECK: [[PC:%.*]] = alloca i8*, align 8
// CHECK: [[PTRPTR1:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PTRPTR1]])
[ObjC][ARC] Use operand bundle "clang.arc.attachedcall" on x86-64 https://reviews.llvm.org/D92808 made clang use the operand bundle instead of emitting retainRV/claimRV calls on arm64. This commit makes changes to clang that are needed to use the operand bundle on x86-64. Differential Revision: https://reviews.llvm.org/D111331
2021-11-09 10:01:48 +08:00
// CHECK: [[T0:%.*]] = call [[TEST1:%.*]]* @test1_helper() [ "clang.arc.attachedcall"(i8* (i8*)* @llvm.objc.retainAutoreleasedReturnValue) ]
// CHECK-NEXT: call void (...) @llvm.objc.clang.arc.noop.use([[TEST1]]* [[T0]])
// CHECK-NEXT: store [[TEST1]]* [[T0]]
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PCPTR1:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.start.p0i8(i64 8, i8* [[PCPTR1]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST1]]*, [[TEST1]]**
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
// CHECK-NEXT: [[EIGHT:%.*]] = bitcast [[TEST1]]* [[T0]] to i8*
[ObjC] Emit selector load right before msgSend call. We currently emit the selector load early, but only because we need it to compute the signature (so that we know which msgSend variant to call). We can prepare the signature with a plain undef, and replace it with the materialized selector value if (and only if) needed, later. Concretely, this usually doesn't have an effect, but tests need updating because we reordered the receiver bitcast and the selector load, which is always fine. There is one notable change: with this, when a msgSend needs a receiver null check, the selector is now loaded in the non-null block, instead of before the null check. That should be a mild improvement.
2022-01-27 09:31:26 +08:00
// CHECK-NEXT: [[SEL:%.*]] = load i8*, i8** @OBJC_SELECTOR_REFERENCES_
// CHECK-NEXT: [[CALL1:%.*]] = call i8* bitcast (i8* (i8*, i8*, ...)* @objc_msgSend to i8* (i8*, i8*)*)(i8* noundef [[EIGHT]], i8* noundef [[SEL]])
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: store i8* [[CALL1]], i8**
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PCPTR2:%.*]] = bitcast i8** [[PC]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PCPTR2]])
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[NINE:%.*]] = load [[TEST1]]*, [[TEST1]]**
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: [[TEN:%.*]] = bitcast [[TEST1]]* [[NINE]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[TEN]])
Revert "Revert r234581, it might have caused a few miscompiles in Chromium." This reverts commit r234700. It turns out that the lifetime markers were not the cause of Chromium failing but a bug which was uncovered by optimizations exposed by the markers. llvm-svn: 235553
2015-04-23 05:38:15 +08:00
// CHECK-NEXT: [[PTRPTR2:%.*]] = bitcast [[PTR_T]]** [[PTR]] to i8*
Update for lifetime intrinsic signature change llvm-svn: 299877
2017-04-11 04:18:45 +08:00
// CHECK-NEXT: call void @llvm.lifetime.end.p0i8(i64 8, i8* [[PTRPTR2]])
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
// CHECK-NEXT: ret void
ARC: Fix the precise-lifetime suppression of returns_inner_pointer receiver extension for message sends via property syntax. rdar://22172983 llvm-svn: 247209
2015-09-10 07:37:17 +08:00
PRECISE_LIFETIME Test1 *ptr = test1_helper();
ObjectiveC: Allow NS_RETURNS_INNER_POINTER annotation of ObjectiveC properties to mean annotation of NS_RETURNS_INNER_POINTER on its synthesized getter. This also facilitates more migration to properties when methods are annotated with NS_RETURNS_INNER_POINTER. // rdar://14990439 llvm-svn: 191009
2013-09-20 00:37:20 +08:00
char *pc = ptr.PropertyReturnsInnerPointer;
}
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
@interface Test2 {
@public
id ivar;
}
@end
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test2(
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
void test2(Test2 *x) {
x->ivar = 0;
// CHECK: [[X:%.*]] = alloca [[TEST2:%.*]]*
// CHECK-NEXT: [[T0:%.*]] = bitcast [[TEST2]]* {{%.*}} to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T1:%.*]] = call i8* @llvm.objc.retain(i8* [[T0]]) [[NUW]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T2:%.*]] = bitcast i8* [[T1]] to [[TEST2]]*
// CHECK-NEXT: store [[TEST2]]* [[T2]], [[TEST2]]** [[X]],
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST2]]*, [[TEST2]]** [[X]],
// CHECK-NEXT: [[OFFSET:%.*]] = load i64, i64* @"OBJC_IVAR_$_Test2.ivar"
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST2]]* [[T0]] to i8*
Update Clang tests to handle explicitly typed gep changes in LLVM. llvm-svn: 230783
2015-02-28 03:18:17 +08:00
// CHECK-NEXT: [[T2:%.*]] = getelementptr inbounds i8, i8* [[T1]], i64 [[OFFSET]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T3:%.*]] = bitcast i8* [[T2]] to i8**
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T4:%.*]] = load i8*, i8** [[T3]],
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: store i8* null, i8** [[T3]],
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T4]]) [[NUW]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NOT: imprecise
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load [[TEST2]]*, [[TEST2]]** [[X]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: [[T1:%.*]] = bitcast [[TEST2]]* [[T0]] to i8*
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T1]]) [[NUW]], !clang.imprecise_release
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: ret void
}
[test] Add {{.*}} to make tests immune to dso_local/dso_preemptable/(none) differences For a definition (of most linkage types), dso_local is set for ELF -fno-pic/-fpie and COFF, but not for Mach-O. This nuance causes unneeded binary format differences. This patch replaces (function) `define ` with `define{{.*}} `, (variable/constant/alias) `= ` with `={{.*}} `, or inserts appropriate `{{.*}} ` if there is an explicit linkage. * Clang will set dso_local for Mach-O, which is currently implied by TargetMachine.cpp. This will make COFF/Mach-O and executable ELF similar. * Eventually I hope we can make dso_local the textual LLVM IR default (write explicit "dso_preemptable" when applicable) and -fpic ELF will be similar to everything else. This patch helps move toward that goal.
2020-12-31 12:45:56 +08:00
// CHECK-LABEL: define{{.*}} void @test3(i8*
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
void test3(PRECISE_LIFETIME id x) {
// CHECK: [[X:%.*]] = alloca i8*,
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: [[T0:%.*]] = call i8* @llvm.objc.retain(i8* {{%.*}}) [[NUW]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NEXT: store i8* [[T0]], i8** [[X]],
Update Clang tests to handle explicitly typed load changes in LLVM. llvm-svn: 230795
2015-02-28 05:19:58 +08:00
// CHECK-NEXT: [[T0:%.*]] = load i8*, i8** [[X]]
Generate objc intrinsics instead of runtime calls as the ARC optimizer now works only on intrinsics Differential Revision: https://reviews.llvm.org/D55802 Reviewers: rjmccall llvm-svn: 349535
2018-12-19 04:33:00 +08:00
// CHECK-NEXT: call void @llvm.objc.release(i8* [[T0]]) [[NUW]]
Tighten up the rules for precise lifetime and document the requirements on the ARC optimizer. rdar://13407451 llvm-svn: 176924
2013-03-13 11:10:54 +08:00
// CHECK-NOT: imprecise_release
// CHECK-NEXT: ret void
}
// CHECK: attributes [[NUW]] = { nounwind }