maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGenCXX/nrvo.cpp

262 lines
6.8 KiB
C++
Raw Normal View History

A number of test cases assume that an "int" parameter or return value will be represented in the IR as a plain "i32" type. This causes the tests to spuriously fail on platforms where int is not a 32-bit type, or where the ABI requires attributes like "signext" or "zeroext" to be used. This patch adds -triple or -target parameters to force those tests to use the i386-unknown-unknown target. llvm-svn: 166551
2012-10-24 20:22:56 +08:00
// RUN: %clang_cc1 -triple i386-unknown-unknown -emit-llvm -O1 -o - %s | FileCheck %s
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// RUN: %clang_cc1 -triple i386-unknown-unknown -emit-llvm -O1 -fcxx-exceptions -fexceptions -std=c++03 -o - %s | FileCheck --check-prefixes=CHECK-EH,CHECK-EH-03 %s
// RUN: %clang_cc1 -triple i386-unknown-unknown -emit-llvm -O1 -fcxx-exceptions -fexceptions -std=c++11 -o - %s | FileCheck --check-prefixes=CHECK-EH,CHECK-EH-11 %s
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
// Test code generation for the named return value optimization.
class X {
public:
X();
X(const X&);
~X();
};
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
template<typename T> struct Y {
Y();
static Y f() {
Y y;
return y;
}
};
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-LABEL: define void @_Z5test0v
// CHECK-EH-LABEL: define void @_Z5test0v
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X test0() {
X x;
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: call {{.*}} @_ZN1XC1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-NEXT: ret void
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH: call {{.*}} @_ZN1XC1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-EH-NEXT: ret void
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
return x;
}
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-LABEL: define void @_Z5test1b(
// CHECK-EH-LABEL: define void @_Z5test1b(
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X test1(bool B) {
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-NEXT: ret void
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X x;
if (B)
return (x);
return x;
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH: tail call {{.*}} @_ZN1XC1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-EH-NEXT: ret void
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
}
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-LABEL: define void @_Z5test2b
// CHECK-EH-LABEL: define void @_Z5test2b
Update clang to take into account the changes to personality fns llvm-svn: 239941
2015-06-18 04:53:19 +08:00
// CHECK-EH-SAME: personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*)
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X test2(bool B) {
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// No NRVO.
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X x;
X y;
if (B)
return y;
return x;
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: call {{.*}} @_ZN1XC1Ev
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: {{.*}} getelementptr inbounds %class.X, %class.X* %y, i32 0, i32 0
// CHECK-NEXT: call void @llvm.lifetime.start
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-NEXT: call {{.*}} @_ZN1XC1Ev
// CHECK: call {{.*}} @_ZN1XC1ERKS_
// CHECK: call {{.*}} @_ZN1XC1ERKS_
// CHECK: call {{.*}} @_ZN1XD1Ev
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: call void @llvm.lifetime.end
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: call {{.*}} @_ZN1XD1Ev
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: call void @llvm.lifetime.end
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
// CHECK: ret void
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// The block ordering in the -fexceptions IR is unfortunate.
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-EH: call void @llvm.lifetime.start
// CHECK-EH-NEXT: call {{.*}} @_ZN1XC1Ev
// CHECK-EH: call void @llvm.lifetime.start
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH-NEXT: invoke {{.*}} @_ZN1XC1Ev
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// -> %invoke.cont, %lpad
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// %invoke.cont:
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-EH: br i1
// -> %if.then, %if.end
// %if.then: returning 'x'
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH: invoke {{.*}} @_ZN1XC1ERKS_
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// -> %cleanup, %lpad1
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// %lpad: landing pad for ctor of 'y', dtor of 'y'
Update clang to take into account the changes to personality fns llvm-svn: 239941
2015-06-18 04:53:19 +08:00
// CHECK-EH: [[CAUGHTVAL:%.*]] = landingpad { i8*, i32 }
Throw the switch to convert clang to the new exception handling model! This model uses the 'landingpad' instruction, which is pinned to the top of the landing pad. (A landing pad is defined as the destination of the unwind branch of an invoke instruction.) All of the information needed to generate the correct exception handling metadata during code generation is encoded into the landingpad instruction. The new 'resume' instruction takes the place of the llvm.eh.resume intrinsic call. It's lowered in much the same way as the intrinsic is. llvm-svn: 140049
2011-09-20 04:31:14 +08:00
// CHECK-EH-NEXT: cleanup
// CHECK-EH-NEXT: extractvalue { i8*, i32 } [[CAUGHTVAL]], 0
// CHECK-EH-NEXT: extractvalue { i8*, i32 } [[CAUGHTVAL]], 1
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// CHECK-EH-NEXT: br label
// -> %eh.cleanup
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// %lpad1: landing pad for return copy ctors, EH cleanup for 'y'
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-03: invoke {{.*}} @_ZN1XD1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// -> %eh.cleanup, %terminate.lpad
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-11: call {{.*}} @_ZN1XD1Ev
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// %if.end: returning 'y'
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH: invoke {{.*}} @_ZN1XC1ERKS_
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// -> %cleanup, %lpad1
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// %cleanup: normal cleanup for 'y'
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-03: invoke {{.*}} @_ZN1XD1Ev
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// -> %invoke.cont11, %lpad
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-11: call {{.*}} @_ZN1XD1Ev
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// %invoke.cont11: normal cleanup for 'x'
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-EH: call void @llvm.lifetime.end
// CHECK-EH-NEXT: call {{.*}} @_ZN1XD1Ev
// CHECK-EH-NEXT: call void @llvm.lifetime.end
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// CHECK-EH-NEXT: ret void
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// %eh.cleanup: EH cleanup for 'x'
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-03: invoke {{.*}} @_ZN1XD1Ev
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// -> %invoke.cont17, %terminate.lpad
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-11: call {{.*}} @_ZN1XD1Ev
Teach IR generation how to lazily emit cleanups. This has a lot of advantages, mostly in avoiding unnecessary work at compile time but also in producing more sensible block orderings. Move the destructor cleanups for local variables over to use lazy cleanups. Eventually all cleanups will do this; for now we have some awkward code duplication. Tell IR generation just to never produce landing pads in -fno-exceptions. This is a much more comprehensive solution to a problem which previously was half-solved by checks in most cleanup-generation spots. llvm-svn: 108270
2010-07-14 04:32:21 +08:00
// %invoke.cont17: rethrow block for %eh.cleanup.
// This really should be elsewhere in the function.
Throw the switch to convert clang to the new exception handling model! This model uses the 'landingpad' instruction, which is pinned to the top of the landing pad. (A landing pad is defined as the destination of the unwind branch of an invoke instruction.) All of the information needed to generate the correct exception handling metadata during code generation is encoded into the landingpad instruction. The new 'resume' instruction takes the place of the llvm.eh.resume intrinsic call. It's lowered in much the same way as the intrinsic is. llvm-svn: 140049
2011-09-20 04:31:14 +08:00
// CHECK-EH: resume { i8*, i32 }
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
// %terminate.lpad: terminate landing pad.
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-03: [[T0:%.*]] = landingpad { i8*, i32 }
// CHECK-EH-03-NEXT: catch i8* null
// CHECK-EH-03-NEXT: [[T1:%.*]] = extractvalue { i8*, i32 } [[T0]], 0
// CHECK-EH-03-NEXT: call void @__clang_call_terminate(i8* [[T1]]) [[NR_NUW:#[0-9]+]]
// CHECK-EH-03-NEXT: unreachable
Validated by nightly-test runs on x86 and x86-64 darwin, including after self-host. Hopefully these results hold up on different platforms. I tried to keep the GNU ObjC runtime happy, but it's hard for me to test. Reimplement how clang generates IR for exceptions. Instead of creating new invoke destinations which sequentially chain to the previous destination, push a more semantic representation of *why* we need the cleanup/catch/filter behavior, then collect that information into a single landing pad upon request. Also reorganizes how normal cleanups (i.e. cleanups triggered by non-exceptional control flow) are generated, since it's actually fairly closely tied in with the former. Remove the need to track which cleanup scope a block is associated with. Document a lot of previously poorly-understood (by me, at least) behavior. The new framework implements the Horrible Hack (tm), which requires every landing pad to have a catch-all so that inlining will work. Clang no longer requires the Horrible Hack just to make exceptions flow correctly within a function, however. The HH is an unfortunate requirement of LLVM's EH IR. llvm-svn: 107631
2010-07-06 09:34:17 +08:00
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
}
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
// CHECK-LABEL: define void @_Z5test3b
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
X test3(bool B, X x) {
I fixed this fixme in r208017. llvm-svn: 208027
2014-05-06 09:09:50 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1Ev
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
if (B) {
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
X y;
return y;
}
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1ERKS_
Implement a simple form of the C++ named return value optimization for return statements. We perform NRVO only when all of the return statements in the function return the same variable. Fixes some link failures in Boost.Interprocess (which is relying on NRVO), and probably improves performance for some C++ applications. llvm-svn: 103867
2010-05-15 14:46:45 +08:00
return x;
}
Ensure that destructors are called for NRVO'd objects when the function does not return. Thanks to Eli for pointing out this corner case. llvm-svn: 103941
2010-05-17 23:52:46 +08:00
extern "C" void exit(int) throw();
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-LABEL: define void @_Z5test4b
Ensure that destructors are called for NRVO'd objects when the function does not return. Thanks to Eli for pointing out this corner case. llvm-svn: 103941
2010-05-17 23:52:46 +08:00
X test4(bool B) {
{
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1Ev
Ensure that destructors are called for NRVO'd objects when the function does not return. Thanks to Eli for pointing out this corner case. llvm-svn: 103941
2010-05-17 23:52:46 +08:00
X x;
// CHECK: br i1
if (B)
return x;
}
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK: tail call {{.*}} @_ZN1XD1Ev
Ensure that destructors are called for NRVO'd objects when the function does not return. Thanks to Eli for pointing out this corner case. llvm-svn: 103941
2010-05-17 23:52:46 +08:00
// CHECK: tail call void @exit(i32 1)
exit(1);
}
Promote the static getNRVOCandidate() function, which computed the NRVO candidate for a return statement, to Sema::getCopyElisionCandidate(), and teach it enough to also determine the NRVO candidate for a throw expression. We still don't use the latter information, however. Along the way, implement core issue 1148, which eliminates copy elision from catch parameters and clarifies that copy elision cannot occur from function parameters (which we already implemented). llvm-svn: 123982
2011-01-22 02:05:27 +08:00
Pass -fexceptions to all tests that use try/catch/throw. llvm-svn: 126037
2011-02-20 03:23:03 +08:00
#ifdef __EXCEPTIONS
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-EH-LABEL: define void @_Z5test5
Promote the static getNRVOCandidate() function, which computed the NRVO candidate for a return statement, to Sema::getCopyElisionCandidate(), and teach it enough to also determine the NRVO candidate for a throw expression. We still don't use the latter information, however. Along the way, implement core issue 1148, which eliminates copy elision from catch parameters and clarifies that copy elision cannot occur from function parameters (which we already implemented). llvm-svn: 123982
2011-01-22 02:05:27 +08:00
void may_throw();
X test5() {
try {
may_throw();
} catch (X x) {
Whack a bunch of tests in CodeGenCXX to work on ARM (using ARM ABI). Batch 2 of 3. llvm-svn: 133011
2011-06-15 05:20:53 +08:00
// CHECK-EH: invoke {{.*}} @_ZN1XC1ERKS_
Promote the static getNRVOCandidate() function, which computed the NRVO candidate for a return statement, to Sema::getCopyElisionCandidate(), and teach it enough to also determine the NRVO candidate for a throw expression. We still don't use the latter information, however. Along the way, implement core issue 1148, which eliminates copy elision from catch parameters and clarifies that copy elision cannot occur from function parameters (which we already implemented). llvm-svn: 123982
2011-01-22 02:05:27 +08:00
// CHECK-EH: call void @__cxa_end_catch()
// CHECK-EH: ret void
return x;
}
}
Pass -fexceptions to all tests that use try/catch/throw. llvm-svn: 126037
2011-02-20 03:23:03 +08:00
#endif
Don't apply NRVO to over-aligned variables. The caller only guarantees alignment up to the ABI alignment of the return type. llvm-svn: 144364
2011-11-11 11:57:31 +08:00
// rdar://problem/10430868
CHECK-LABEL-ify some code gen tests to improve diagnostic experience when tests fail. llvm-svn: 188447
2013-08-15 14:47:53 +08:00
// CHECK-LABEL: define void @_Z5test6v
Don't apply NRVO to over-aligned variables. The caller only guarantees alignment up to the ABI alignment of the return type. llvm-svn: 144364
2011-11-11 11:57:31 +08:00
X test6() {
X a __attribute__((aligned(8)));
return a;
// CHECK: [[A:%.*]] = alloca [[X:%.*]], align 8
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: [[PTR:%.*]] = getelementptr inbounds %class.X, %class.X* [[A]], i32 0, i32 0
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 1, i8* nonnull [[PTR]])
Adjust clang side tests effected by 239795 before reapplying said change llvm-svn: 239848
2015-06-17 04:24:06 +08:00
// CHECK-NEXT: call {{.*}} @_ZN1XC1Ev([[X]]* nonnull [[A]])
// CHECK-NEXT: call {{.*}} @_ZN1XC1ERKS_([[X]]* {{%.*}}, [[X]]* nonnull dereferenceable({{[0-9]+}}) [[A]])
// CHECK-NEXT: call {{.*}} @_ZN1XD1Ev([[X]]* nonnull [[A]])
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 1, i8* nonnull [[PTR]])
Don't apply NRVO to over-aligned variables. The caller only guarantees alignment up to the ABI alignment of the return type. llvm-svn: 144364
2011-11-11 11:57:31 +08:00
// CHECK-NEXT: ret void
}
Update to use references to attribute groups instead of listing the attributes on the call/invoke instructions. llvm-svn: 175878
2013-02-22 17:10:20 +08:00
Err, wait. Those new test functions in r207890 need CHECK lines. llvm-svn: 207891
2014-05-03 09:19:39 +08:00
// CHECK-LABEL: define void @_Z5test7b
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK-EH-LABEL: define void @_Z5test7b
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
X test7(bool b) {
Err, wait. Those new test functions in r207890 need CHECK lines. llvm-svn: 207891
2014-05-03 09:19:39 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1Ev
// CHECK-NEXT: ret
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK-EH: tail call {{.*}} @_ZN1XC1Ev
// CHECK-EH-NEXT: ret
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
if (b) {
X x;
return x;
}
return X();
}
Err, wait. Those new test functions in r207890 need CHECK lines. llvm-svn: 207891
2014-05-03 09:19:39 +08:00
// CHECK-LABEL: define void @_Z5test8b
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK-EH-LABEL: define void @_Z5test8b
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
X test8(bool b) {
Err, wait. Those new test functions in r207890 need CHECK lines. llvm-svn: 207891
2014-05-03 09:19:39 +08:00
// CHECK: tail call {{.*}} @_ZN1XC1Ev
// CHECK-NEXT: ret
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK-EH: tail call {{.*}} @_ZN1XC1Ev
// CHECK-EH-NEXT: ret
if (b) {
Rewrite NRVO determination. Track NRVO candidates on the parser Scope and apply the NRVO candidate flag to all possible NRVO candidates here, and remove the flags in computeNRVO or upon template instantiation. A variable now has NRVO applied if and only if every return statement in that scope returns that variable. This is nearly optimal. Performs NRVO roughly 7% more often in a bootstrap build of clang. Patch co-authored by Richard Smith. llvm-svn: 207890
2014-05-03 08:41:18 +08:00
X x;
return x;
} else {
X y;
return y;
}
}
Y<int> test9() {
Y<int>::f();
}
// CHECK-LABEL: define linkonce_odr void @_ZN1YIiE1fEv
// CHECK: tail call {{.*}} @_ZN1YIiEC1Ev
Update NRVO logic to support early return (Attempt 2) Summary: This is the second attempt of r333500 (Update NRVO logic to support early return). The previous one was reverted for a miscompilation for an incorrect NRVO set up on templates such as: ``` struct Foo {}; template <typename T> T bar() { T t; if (false) return T(); return t; } ``` Where, `t` is marked as non-NRVO variable before its instantiation. However, while its instantiation, it's left an NRVO candidate, turned into an NRVO variable later. Reviewers: rsmith Reviewed By: rsmith Subscribers: cfe-commits Differential Revision: https://reviews.llvm.org/D47586 llvm-svn: 335019
2018-06-19 12:39:07 +08:00
// CHECK-LABEL: define void @_Z6test10b
X test10(bool B, X x) {
if (B) {
// CHECK: tail call {{.*}} @_ZN1XC1ERKS_
// CHECK-EH: tail call {{.*}} @_ZN1XC1ERKS_
return x;
}
// CHECK: tail call {{.*}} @_ZN1XC1Ev
// CHECK-NOT: call {{.*}} @_ZN1XC1ERKS_
// CHECK-EH: tail call {{.*}} @_ZN1XC1Ev
// CHECK-EH-NOT: call {{.*}} @_ZN1XC1ERKS_
X y;
return y;
}
// CHECK-LABEL: define {{.*}} void @_Z6test11I1XET_v
// CHECK-EH-LABEL: define {{.*}} void @_Z6test11I1XET_v
template <typename T>
T test11() {
// CHECK: tail call {{.*}} @_ZN1XC1Ev
// CHECK-NEXT: ret void
// CHECK-EH: tail call {{.*}} @_ZN1XC1Ev
// CHECK-EH-NEXT: ret void
T t;
return t;
}
void test12() {
test11<X>();
}
C++11 test cleanup: nonthrowing destructors If a dtor has no interesting members, then it ends up being nothrow, which affects the generated IR. Modify some tests to tolerate this difference between C++03 and C++11. Differential Revision: http://reviews.llvm.org/D27936 llvm-svn: 290207
2016-12-21 06:26:11 +08:00
// CHECK-EH-03: attributes [[NR_NUW]] = { noreturn nounwind }