rnk updated this revision to Diff 245491.
rnk added a comment.
- rebase
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D74455/new/
https://reviews.llvm.org/D74455
Files:
clang/lib/CodeGen/MicrosoftCXXABI.cpp
clang/test/CodeGenCXX/inalloca-overaligned.cpp
Index: clang/test/CodeGenCXX/inalloca-overaligned.cpp
===================================================================
--- clang/test/CodeGenCXX/inalloca-overaligned.cpp
+++ clang/test/CodeGenCXX/inalloca-overaligned.cpp
@@ -4,10 +4,6 @@
// MSVC passes overaligned types indirectly since MSVC 2015. Make sure that
// works with inalloca.
-// FIXME: Pass non-trivial *and* overaligned types indirectly. Right now the
C++
-// ABI rules say to use inalloca, and they take precedence, so it's not easy to
-// implement this.
-
struct NonTrivial {
NonTrivial();
@@ -50,3 +46,27 @@
// CHECK: getelementptr inbounds <{ %struct.NonTrivial, %struct.OverAligned*
}>, <{ %struct.NonTrivial, %struct.OverAligned* }>* %{{.*}}, i32 0, i32 1
// CHECK: store %struct.OverAligned* [[TMP]], %struct.OverAligned** %{{.*}},
align 4
// CHECK: call i32 @"?receive_inalloca_overaligned@@Y{{.*}}"(<{
%struct.NonTrivial, %struct.OverAligned* }>* inalloca %argmem)
+
+struct __declspec(align(64)) Both {
+ Both();
+ Both(const Both &o);
+ int x;
+};
+
+int receive_both1(Both b) {
+ return b.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_both1@@Y{{.*}}"
+// CHECK-SAME: (%struct.Both* %b)
+
+int receive_both2(Both x, Both y) {
+ return x.x + y.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_both2@@Y{{.*}}"
+// CHECK-SAME: (%struct.Both* %x, %struct.Both* %y)
+
+int receive_nt_and_both(NonTrivial x, Both y) {
+ return x.x + y.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_nt_and_both@@Y{{.*}}"
+// CHECK-SAME: (<{ %struct.NonTrivial, %struct.Both* }>* inalloca %0)
Index: clang/lib/CodeGen/MicrosoftCXXABI.cpp
===================================================================
--- clang/lib/CodeGen/MicrosoftCXXABI.cpp
+++ clang/lib/CodeGen/MicrosoftCXXABI.cpp
@@ -821,18 +821,19 @@
return !RD->canPassInRegisters() ? RAA_Indirect : RAA_Default;
case llvm::Triple::x86:
- // All record arguments are passed in memory on x86. Decide whether to
- // construct the object directly in argument memory, or to construct the
- // argument elsewhere and copy the bytes during the call.
-
- // If C++ prohibits us from making a copy, construct the arguments directly
- // into argument memory.
- if (!RD->canPassInRegisters())
- return RAA_DirectInMemory;
-
- // Otherwise, construct the argument into a temporary and copy the bytes
- // into the outgoing argument memory.
- return RAA_Default;
+ // If C++ allows us to copy the struct bytes, use the normal C convention
+ // rules.
+ if (RD->canPassInRegisters())
+ return RAA_Default;
+
+ // If the type has a required alignment (alignas, __declspec(align)), MSVC
+ // passes it indirectly.
+ if (getContext().isAlignmentRequired(getContext().getTypeDeclType(RD)))
+ return RAA_Indirect;
+
+ // In the worst case, the argument must be constructed directly into the
+ // outgoing argument memory slots.
+ return RAA_DirectInMemory;
case llvm::Triple::x86_64:
case llvm::Triple::aarch64:
Index: clang/test/CodeGenCXX/inalloca-overaligned.cpp
===================================================================
--- clang/test/CodeGenCXX/inalloca-overaligned.cpp
+++ clang/test/CodeGenCXX/inalloca-overaligned.cpp
@@ -4,10 +4,6 @@
// MSVC passes overaligned types indirectly since MSVC 2015. Make sure that
// works with inalloca.
-// FIXME: Pass non-trivial *and* overaligned types indirectly. Right now the C++
-// ABI rules say to use inalloca, and they take precedence, so it's not easy to
-// implement this.
-
struct NonTrivial {
NonTrivial();
@@ -50,3 +46,27 @@
// CHECK: getelementptr inbounds <{ %struct.NonTrivial, %struct.OverAligned* }>, <{ %struct.NonTrivial, %struct.OverAligned* }>* %{{.*}}, i32 0, i32 1
// CHECK: store %struct.OverAligned* [[TMP]], %struct.OverAligned** %{{.*}}, align 4
// CHECK: call i32 @"?receive_inalloca_overaligned@@Y{{.*}}"(<{ %struct.NonTrivial, %struct.OverAligned* }>* inalloca %argmem)
+
+struct __declspec(align(64)) Both {
+ Both();
+ Both(const Both &o);
+ int x;
+};
+
+int receive_both1(Both b) {
+ return b.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_both1@@Y{{.*}}"
+// CHECK-SAME: (%struct.Both* %b)
+
+int receive_both2(Both x, Both y) {
+ return x.x + y.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_both2@@Y{{.*}}"
+// CHECK-SAME: (%struct.Both* %x, %struct.Both* %y)
+
+int receive_nt_and_both(NonTrivial x, Both y) {
+ return x.x + y.x;
+}
+// CHECK-LABEL: define dso_local i32 @"?receive_nt_and_both@@Y{{.*}}"
+// CHECK-SAME: (<{ %struct.NonTrivial, %struct.Both* }>* inalloca %0)
Index: clang/lib/CodeGen/MicrosoftCXXABI.cpp
===================================================================
--- clang/lib/CodeGen/MicrosoftCXXABI.cpp
+++ clang/lib/CodeGen/MicrosoftCXXABI.cpp
@@ -821,18 +821,19 @@
return !RD->canPassInRegisters() ? RAA_Indirect : RAA_Default;
case llvm::Triple::x86:
- // All record arguments are passed in memory on x86. Decide whether to
- // construct the object directly in argument memory, or to construct the
- // argument elsewhere and copy the bytes during the call.
-
- // If C++ prohibits us from making a copy, construct the arguments directly
- // into argument memory.
- if (!RD->canPassInRegisters())
- return RAA_DirectInMemory;
-
- // Otherwise, construct the argument into a temporary and copy the bytes
- // into the outgoing argument memory.
- return RAA_Default;
+ // If C++ allows us to copy the struct bytes, use the normal C convention
+ // rules.
+ if (RD->canPassInRegisters())
+ return RAA_Default;
+
+ // If the type has a required alignment (alignas, __declspec(align)), MSVC
+ // passes it indirectly.
+ if (getContext().isAlignmentRequired(getContext().getTypeDeclType(RD)))
+ return RAA_Indirect;
+
+ // In the worst case, the argument must be constructed directly into the
+ // outgoing argument memory slots.
+ return RAA_DirectInMemory;
case llvm::Triple::x86_64:
case llvm::Triple::aarch64:
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
