diff options
| author | John McCall <rjmccall@apple.com> | 2018-04-01 21:04:30 +0000 |
|---|---|---|
| committer | John McCall <rjmccall@apple.com> | 2018-04-01 21:04:30 +0000 |
| commit | 4fcd9ef6736ebc9bd1cd2c7192579c469a54b958 (patch) | |
| tree | 9a32b0881bf884f8d29c23124cd8c3a60319ff15 | |
| parent | ba8033be4d422fea1a4117238038197f8db56469 (diff) | |
| download | bcm5719-llvm-4fcd9ef6736ebc9bd1cd2c7192579c469a54b958.tar.gz bcm5719-llvm-4fcd9ef6736ebc9bd1cd2c7192579c469a54b958.zip | |
Fix a major swiftcall ABI bug with trivial C++ class types.
The problem with the previous logic was that there might not be any
explicit copy/move constructor declarations, e.g. if the type is
trivial and we've never type-checked a copy of it. Relying on Sema's
computation seems much more reliable.
Also, I believe Richard's recommendation is exactly the rule we use
now on the Itanium ABI, modulo the trivial_abi attribute (which this
change of course fixes our handling of in Swift).
This does mean that we have a less portable rule for deciding
indirectness for swiftcall. I would prefer it if we just applied the
Itanium rule universally under swiftcall, but in the meantime, I need
to fix this bug.
This only arises when defining functions with class-type arguments
in C++, as we do in the Swift runtime. It doesn't affect normal Swift
operation because we don't import code as C++.
llvm-svn: 328942
| -rw-r--r-- | clang/lib/CodeGen/SwiftCallingConv.cpp | 20 | ||||
| -rw-r--r-- | clang/test/CodeGenCXX/arm-swiftcall.cpp | 10 |
2 files changed, 14 insertions, 16 deletions
diff --git a/clang/lib/CodeGen/SwiftCallingConv.cpp b/clang/lib/CodeGen/SwiftCallingConv.cpp index 2a12078e198..7dc2a1d089d 100644 --- a/clang/lib/CodeGen/SwiftCallingConv.cpp +++ b/clang/lib/CodeGen/SwiftCallingConv.cpp @@ -742,22 +742,10 @@ void swiftcall::legalizeVectorType(CodeGenModule &CGM, CharUnits origVectorSize, bool swiftcall::shouldPassCXXRecordIndirectly(CodeGenModule &CGM, const CXXRecordDecl *record) { - // Following a recommendation from Richard Smith, pass a C++ type - // indirectly only if the destructor is non-trivial or *all* of the - // copy/move constructors are deleted or non-trivial. - - if (record->hasNonTrivialDestructor()) - return true; - - // It would be nice if this were summarized on the CXXRecordDecl. - for (auto ctor : record->ctors()) { - if (ctor->isCopyOrMoveConstructor() && !ctor->isDeleted() && - ctor->isTrivial()) { - return false; - } - } - - return true; + // FIXME: should we not rely on the standard computation in Sema, just in + // case we want to diverge from the platform ABI (e.g. on targets where + // that uses the MSVC rule)? + return !record->canPassInRegisters(); } static ABIArgInfo classifyExpandedType(SwiftAggLowering &lowering, diff --git a/clang/test/CodeGenCXX/arm-swiftcall.cpp b/clang/test/CodeGenCXX/arm-swiftcall.cpp index 5c932c77b1d..36a5afad4c6 100644 --- a/clang/test/CodeGenCXX/arm-swiftcall.cpp +++ b/clang/test/CodeGenCXX/arm-swiftcall.cpp @@ -113,3 +113,13 @@ TEST(struct_indirect_1) // Should not be byval. // CHECK-LABEL: define {{.*}} void @take_struct_indirect_1({{.*}}*{{( %.*)?}}) + +// Do a simple standalone test here of a function definition to ensure that +// we don't have problems due to failure to eagerly synthesize a copy +// constructor declaration. +class struct_trivial { + int x; +}; +// CHECK-LABEL define void @test_struct_trivial(i32{{( %.*)?}}) +extern "C" SWIFTCALL +void test_struct_trivial(struct_trivial triv) {} |

