-
Notifications
You must be signed in to change notification settings - Fork 11.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[clang codegen] Add CreateRuntimeFunction overload that takes a clang type. #113506
base: main
Are you sure you want to change the base?
Conversation
852334c
to
bb58bdd
Compare
It would be great if this new method also called TargetLibraryInfo and set the extension attributes for any integer <=32 bits... Or are you maybe waiting with that as a separate patch? |
SetLLVMFunctionAttributes() should add the appropriate signext/zeroext attribute. At least, it appears to work in my testing. I'll add a regression test. |
Ah, I see. I ran this now on SystemZ and can confirm that this is now getting the proper extension attributes. A test might not even be needed, considering that this will (hopefully soon by default) be checked on SystemZ (general check in back-end), where it was discovered. This approach is making most sense to me as well now as there is already code that adds the attributes for clang types. |
What does it mean that this is a "draft"? It seems to me that this is ready for a regular review..? |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM, I think this is the correct way to fix the issue.
… type. Correctly computing the LLVM types/attributes is complicated in general, so add a variant which does that for you.
bb58bdd
to
5d789d1
Compare
@llvm/pr-subscribers-clang Author: Eli Friedman (efriedma-quic) ChangesCorrectly computing the LLVM types/attributes is complicated in general, so add a variant which does that for you. Full diff: https://github.com/llvm/llvm-project/pull/113506.diff 4 Files Affected:
diff --git a/clang/lib/CodeGen/CGBlocks.cpp b/clang/lib/CodeGen/CGBlocks.cpp
index bfa9b0a2f836bc8..1c46bac4bb232dd 100644
--- a/clang/lib/CodeGen/CGBlocks.cpp
+++ b/clang/lib/CodeGen/CGBlocks.cpp
@@ -2837,10 +2837,9 @@ llvm::FunctionCallee CodeGenModule::getBlockObjectDispose() {
if (BlockObjectDispose)
return BlockObjectDispose;
- llvm::Type *args[] = { Int8PtrTy, Int32Ty };
- llvm::FunctionType *fty
- = llvm::FunctionType::get(VoidTy, args, false);
- BlockObjectDispose = CreateRuntimeFunction(fty, "_Block_object_dispose");
+ QualType args[] = {Context.VoidPtrTy, Context.IntTy};
+ BlockObjectDispose =
+ CreateRuntimeFunction(Context.VoidTy, args, "_Block_object_dispose");
configureBlocksRuntimeObject(
*this, cast<llvm::Constant>(BlockObjectDispose.getCallee()));
return BlockObjectDispose;
@@ -2850,10 +2849,9 @@ llvm::FunctionCallee CodeGenModule::getBlockObjectAssign() {
if (BlockObjectAssign)
return BlockObjectAssign;
- llvm::Type *args[] = { Int8PtrTy, Int8PtrTy, Int32Ty };
- llvm::FunctionType *fty
- = llvm::FunctionType::get(VoidTy, args, false);
- BlockObjectAssign = CreateRuntimeFunction(fty, "_Block_object_assign");
+ QualType args[] = {Context.VoidPtrTy, Context.VoidPtrTy, Context.IntTy};
+ BlockObjectAssign =
+ CreateRuntimeFunction(Context.VoidTy, args, "_Block_object_assign");
configureBlocksRuntimeObject(
*this, cast<llvm::Constant>(BlockObjectAssign.getCallee()));
return BlockObjectAssign;
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index ba376f9ecfacde7..6a3af6890bae7de 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4903,6 +4903,52 @@ GetRuntimeFunctionDecl(ASTContext &C, StringRef Name) {
return nullptr;
}
+static void setWindowsItaniumDLLImport(CodeGenModule &CGM, bool Local,
+ llvm::Function *F, StringRef Name) {
+ // In Windows Itanium environments, try to mark runtime functions
+ // dllimport. For Mingw and MSVC, don't. We don't really know if the user
+ // will link their standard library statically or dynamically. Marking
+ // functions imported when they are not imported can cause linker errors
+ // and warnings.
+ if (!Local && CGM.getTriple().isWindowsItaniumEnvironment() &&
+ !CGM.getCodeGenOpts().LTOVisibilityPublicStd) {
+ const FunctionDecl *FD = GetRuntimeFunctionDecl(CGM.getContext(), Name);
+ if (!FD || FD->hasAttr<DLLImportAttr>()) {
+ F->setDLLStorageClass(llvm::GlobalValue::DLLImportStorageClass);
+ F->setLinkage(llvm::GlobalValue::ExternalLinkage);
+ }
+ }
+}
+
+llvm::FunctionCallee CodeGenModule::CreateRuntimeFunction(
+ QualType ReturnTy, ArrayRef<QualType> ArgTys, StringRef Name,
+ llvm::AttributeList ExtraAttrs, bool Local, bool AssumeConvergent) {
+ if (AssumeConvergent) {
+ ExtraAttrs =
+ ExtraAttrs.addFnAttribute(VMContext, llvm::Attribute::Convergent);
+ }
+
+ QualType FTy = Context.getFunctionType(ReturnTy, ArgTys,
+ FunctionProtoType::ExtProtoInfo());
+ const CGFunctionInfo &Info = getTypes().arrangeFreeFunctionType(
+ Context.getCanonicalType(FTy).castAs<FunctionProtoType>());
+ auto *ConvTy = getTypes().GetFunctionType(Info);
+ llvm::Constant *C = GetOrCreateLLVMFunction(
+ Name, ConvTy, GlobalDecl(), /*ForVTable=*/false,
+ /*DontDefer=*/false, /*IsThunk=*/false, ExtraAttrs);
+
+ if (auto *F = dyn_cast<llvm::Function>(C)) {
+ if (F->empty()) {
+ SetLLVMFunctionAttributes(GlobalDecl(), Info, F, /*IsThunk*/ false);
+ // FIXME: Set calling-conv properly in ExtProtoInfo
+ F->setCallingConv(getRuntimeCC());
+ setWindowsItaniumDLLImport(*this, Local, F, Name);
+ setDSOLocal(F);
+ }
+ }
+ return {ConvTy, C};
+}
+
/// CreateRuntimeFunction - Create a new runtime function with the specified
/// type and name.
llvm::FunctionCallee
@@ -4922,20 +4968,7 @@ CodeGenModule::CreateRuntimeFunction(llvm::FunctionType *FTy, StringRef Name,
if (auto *F = dyn_cast<llvm::Function>(C)) {
if (F->empty()) {
F->setCallingConv(getRuntimeCC());
-
- // In Windows Itanium environments, try to mark runtime functions
- // dllimport. For Mingw and MSVC, don't. We don't really know if the user
- // will link their standard library statically or dynamically. Marking
- // functions imported when they are not imported can cause linker errors
- // and warnings.
- if (!Local && getTriple().isWindowsItaniumEnvironment() &&
- !getCodeGenOpts().LTOVisibilityPublicStd) {
- const FunctionDecl *FD = GetRuntimeFunctionDecl(Context, Name);
- if (!FD || FD->hasAttr<DLLImportAttr>()) {
- F->setDLLStorageClass(llvm::GlobalValue::DLLImportStorageClass);
- F->setLinkage(llvm::GlobalValue::ExternalLinkage);
- }
- }
+ setWindowsItaniumDLLImport(*this, Local, F, Name);
setDSOLocal(F);
// FIXME: We should use CodeGenModule::SetLLVMFunctionAttributes() instead
// of trying to approximate the attributes using the LLVM function
diff --git a/clang/lib/CodeGen/CodeGenModule.h b/clang/lib/CodeGen/CodeGenModule.h
index 8d5787769382f68..a0586af9ad3e27f 100644
--- a/clang/lib/CodeGen/CodeGenModule.h
+++ b/clang/lib/CodeGen/CodeGenModule.h
@@ -1252,6 +1252,12 @@ class CodeGenModule : public CodeGenTypeCache {
llvm::AttributeList ExtraAttrs = llvm::AttributeList(),
bool Local = false, bool AssumeConvergent = false);
+ llvm::FunctionCallee
+ CreateRuntimeFunction(QualType ReturnTy, ArrayRef<QualType> ArgTys,
+ StringRef Name,
+ llvm::AttributeList ExtraAttrs = llvm::AttributeList(),
+ bool Local = false, bool AssumeConvergent = false);
+
/// Create a new runtime global variable with the specified type and name.
llvm::Constant *CreateRuntimeVariable(llvm::Type *Ty,
StringRef Name);
diff --git a/clang/test/CodeGen/blocks.c b/clang/test/CodeGen/blocks.c
index d508de9c9f80143..d104078a9816cce 100644
--- a/clang/test/CodeGen/blocks.c
+++ b/clang/test/CodeGen/blocks.c
@@ -1,5 +1,6 @@
// RUN: %clang_cc1 -triple i386-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks | FileCheck --check-prefix=CHECK --check-prefix=SIG_STR %s
// RUN: %clang_cc1 -triple i386-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks -fdisable-block-signature-string | FileCheck --check-prefix=CHECK --check-prefix=NO_SIG_STR %s
+// RUN: %clang_cc1 -triple s390x-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks | FileCheck --check-prefix=SYSTEMZ %s
// SIG_STR: @[[STR:.*]] = private unnamed_addr constant [6 x i8] c"v4@?0\00", align 1
// SIG_STR: @{{.*}} = internal constant { ptr, i32, i32, ptr, ptr } { ptr @_NSConcreteGlobalBlock, i32 1342177280, i32 0, ptr @f_block_invoke, ptr @{{.*}} }, align 4
@@ -50,6 +51,8 @@ void (^test1)(void) = ^(void) {
// CHECK-NEXT: call void @_Block_object_assign(ptr %[[V5]], ptr %[[BLOCKCOPY_SRC]], i32 8)
// CHECK-NEXT: ret void
+// SYSTEMZ: declare void @_Block_object_assign(ptr noundef, ptr noundef, i32 noundef signext)
+
// CHECK-LABEL: define linkonce_odr hidden void @__destroy_helper_block_4_20r(ptr noundef %0) unnamed_addr
// CHECK: %[[_ADDR:.*]] = alloca ptr, align 4
// CHECK-NEXT: store ptr %0, ptr %[[_ADDR]], align 4
@@ -59,6 +62,8 @@ void (^test1)(void) = ^(void) {
// CHECK-NEXT: call void @_Block_object_dispose(ptr %[[V3]], i32 8)
// CHECK-NEXT: ret void
+// SYSTEMZ: declare void @_Block_object_dispose(ptr noundef, i32 noundef signext)
+
typedef double ftype(double);
// It's not clear that we *should* support this syntax, but until that decision
// is made, we should support it properly and not crash.
|
@llvm/pr-subscribers-clang-codegen Author: Eli Friedman (efriedma-quic) ChangesCorrectly computing the LLVM types/attributes is complicated in general, so add a variant which does that for you. Full diff: https://github.com/llvm/llvm-project/pull/113506.diff 4 Files Affected:
diff --git a/clang/lib/CodeGen/CGBlocks.cpp b/clang/lib/CodeGen/CGBlocks.cpp
index bfa9b0a2f836bc8..1c46bac4bb232dd 100644
--- a/clang/lib/CodeGen/CGBlocks.cpp
+++ b/clang/lib/CodeGen/CGBlocks.cpp
@@ -2837,10 +2837,9 @@ llvm::FunctionCallee CodeGenModule::getBlockObjectDispose() {
if (BlockObjectDispose)
return BlockObjectDispose;
- llvm::Type *args[] = { Int8PtrTy, Int32Ty };
- llvm::FunctionType *fty
- = llvm::FunctionType::get(VoidTy, args, false);
- BlockObjectDispose = CreateRuntimeFunction(fty, "_Block_object_dispose");
+ QualType args[] = {Context.VoidPtrTy, Context.IntTy};
+ BlockObjectDispose =
+ CreateRuntimeFunction(Context.VoidTy, args, "_Block_object_dispose");
configureBlocksRuntimeObject(
*this, cast<llvm::Constant>(BlockObjectDispose.getCallee()));
return BlockObjectDispose;
@@ -2850,10 +2849,9 @@ llvm::FunctionCallee CodeGenModule::getBlockObjectAssign() {
if (BlockObjectAssign)
return BlockObjectAssign;
- llvm::Type *args[] = { Int8PtrTy, Int8PtrTy, Int32Ty };
- llvm::FunctionType *fty
- = llvm::FunctionType::get(VoidTy, args, false);
- BlockObjectAssign = CreateRuntimeFunction(fty, "_Block_object_assign");
+ QualType args[] = {Context.VoidPtrTy, Context.VoidPtrTy, Context.IntTy};
+ BlockObjectAssign =
+ CreateRuntimeFunction(Context.VoidTy, args, "_Block_object_assign");
configureBlocksRuntimeObject(
*this, cast<llvm::Constant>(BlockObjectAssign.getCallee()));
return BlockObjectAssign;
diff --git a/clang/lib/CodeGen/CodeGenModule.cpp b/clang/lib/CodeGen/CodeGenModule.cpp
index ba376f9ecfacde7..6a3af6890bae7de 100644
--- a/clang/lib/CodeGen/CodeGenModule.cpp
+++ b/clang/lib/CodeGen/CodeGenModule.cpp
@@ -4903,6 +4903,52 @@ GetRuntimeFunctionDecl(ASTContext &C, StringRef Name) {
return nullptr;
}
+static void setWindowsItaniumDLLImport(CodeGenModule &CGM, bool Local,
+ llvm::Function *F, StringRef Name) {
+ // In Windows Itanium environments, try to mark runtime functions
+ // dllimport. For Mingw and MSVC, don't. We don't really know if the user
+ // will link their standard library statically or dynamically. Marking
+ // functions imported when they are not imported can cause linker errors
+ // and warnings.
+ if (!Local && CGM.getTriple().isWindowsItaniumEnvironment() &&
+ !CGM.getCodeGenOpts().LTOVisibilityPublicStd) {
+ const FunctionDecl *FD = GetRuntimeFunctionDecl(CGM.getContext(), Name);
+ if (!FD || FD->hasAttr<DLLImportAttr>()) {
+ F->setDLLStorageClass(llvm::GlobalValue::DLLImportStorageClass);
+ F->setLinkage(llvm::GlobalValue::ExternalLinkage);
+ }
+ }
+}
+
+llvm::FunctionCallee CodeGenModule::CreateRuntimeFunction(
+ QualType ReturnTy, ArrayRef<QualType> ArgTys, StringRef Name,
+ llvm::AttributeList ExtraAttrs, bool Local, bool AssumeConvergent) {
+ if (AssumeConvergent) {
+ ExtraAttrs =
+ ExtraAttrs.addFnAttribute(VMContext, llvm::Attribute::Convergent);
+ }
+
+ QualType FTy = Context.getFunctionType(ReturnTy, ArgTys,
+ FunctionProtoType::ExtProtoInfo());
+ const CGFunctionInfo &Info = getTypes().arrangeFreeFunctionType(
+ Context.getCanonicalType(FTy).castAs<FunctionProtoType>());
+ auto *ConvTy = getTypes().GetFunctionType(Info);
+ llvm::Constant *C = GetOrCreateLLVMFunction(
+ Name, ConvTy, GlobalDecl(), /*ForVTable=*/false,
+ /*DontDefer=*/false, /*IsThunk=*/false, ExtraAttrs);
+
+ if (auto *F = dyn_cast<llvm::Function>(C)) {
+ if (F->empty()) {
+ SetLLVMFunctionAttributes(GlobalDecl(), Info, F, /*IsThunk*/ false);
+ // FIXME: Set calling-conv properly in ExtProtoInfo
+ F->setCallingConv(getRuntimeCC());
+ setWindowsItaniumDLLImport(*this, Local, F, Name);
+ setDSOLocal(F);
+ }
+ }
+ return {ConvTy, C};
+}
+
/// CreateRuntimeFunction - Create a new runtime function with the specified
/// type and name.
llvm::FunctionCallee
@@ -4922,20 +4968,7 @@ CodeGenModule::CreateRuntimeFunction(llvm::FunctionType *FTy, StringRef Name,
if (auto *F = dyn_cast<llvm::Function>(C)) {
if (F->empty()) {
F->setCallingConv(getRuntimeCC());
-
- // In Windows Itanium environments, try to mark runtime functions
- // dllimport. For Mingw and MSVC, don't. We don't really know if the user
- // will link their standard library statically or dynamically. Marking
- // functions imported when they are not imported can cause linker errors
- // and warnings.
- if (!Local && getTriple().isWindowsItaniumEnvironment() &&
- !getCodeGenOpts().LTOVisibilityPublicStd) {
- const FunctionDecl *FD = GetRuntimeFunctionDecl(Context, Name);
- if (!FD || FD->hasAttr<DLLImportAttr>()) {
- F->setDLLStorageClass(llvm::GlobalValue::DLLImportStorageClass);
- F->setLinkage(llvm::GlobalValue::ExternalLinkage);
- }
- }
+ setWindowsItaniumDLLImport(*this, Local, F, Name);
setDSOLocal(F);
// FIXME: We should use CodeGenModule::SetLLVMFunctionAttributes() instead
// of trying to approximate the attributes using the LLVM function
diff --git a/clang/lib/CodeGen/CodeGenModule.h b/clang/lib/CodeGen/CodeGenModule.h
index 8d5787769382f68..a0586af9ad3e27f 100644
--- a/clang/lib/CodeGen/CodeGenModule.h
+++ b/clang/lib/CodeGen/CodeGenModule.h
@@ -1252,6 +1252,12 @@ class CodeGenModule : public CodeGenTypeCache {
llvm::AttributeList ExtraAttrs = llvm::AttributeList(),
bool Local = false, bool AssumeConvergent = false);
+ llvm::FunctionCallee
+ CreateRuntimeFunction(QualType ReturnTy, ArrayRef<QualType> ArgTys,
+ StringRef Name,
+ llvm::AttributeList ExtraAttrs = llvm::AttributeList(),
+ bool Local = false, bool AssumeConvergent = false);
+
/// Create a new runtime global variable with the specified type and name.
llvm::Constant *CreateRuntimeVariable(llvm::Type *Ty,
StringRef Name);
diff --git a/clang/test/CodeGen/blocks.c b/clang/test/CodeGen/blocks.c
index d508de9c9f80143..d104078a9816cce 100644
--- a/clang/test/CodeGen/blocks.c
+++ b/clang/test/CodeGen/blocks.c
@@ -1,5 +1,6 @@
// RUN: %clang_cc1 -triple i386-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks | FileCheck --check-prefix=CHECK --check-prefix=SIG_STR %s
// RUN: %clang_cc1 -triple i386-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks -fdisable-block-signature-string | FileCheck --check-prefix=CHECK --check-prefix=NO_SIG_STR %s
+// RUN: %clang_cc1 -triple s390x-unknown-unknown %s -emit-llvm -Wno-strict-prototypes -o - -fblocks | FileCheck --check-prefix=SYSTEMZ %s
// SIG_STR: @[[STR:.*]] = private unnamed_addr constant [6 x i8] c"v4@?0\00", align 1
// SIG_STR: @{{.*}} = internal constant { ptr, i32, i32, ptr, ptr } { ptr @_NSConcreteGlobalBlock, i32 1342177280, i32 0, ptr @f_block_invoke, ptr @{{.*}} }, align 4
@@ -50,6 +51,8 @@ void (^test1)(void) = ^(void) {
// CHECK-NEXT: call void @_Block_object_assign(ptr %[[V5]], ptr %[[BLOCKCOPY_SRC]], i32 8)
// CHECK-NEXT: ret void
+// SYSTEMZ: declare void @_Block_object_assign(ptr noundef, ptr noundef, i32 noundef signext)
+
// CHECK-LABEL: define linkonce_odr hidden void @__destroy_helper_block_4_20r(ptr noundef %0) unnamed_addr
// CHECK: %[[_ADDR:.*]] = alloca ptr, align 4
// CHECK-NEXT: store ptr %0, ptr %[[_ADDR]], align 4
@@ -59,6 +62,8 @@ void (^test1)(void) = ^(void) {
// CHECK-NEXT: call void @_Block_object_dispose(ptr %[[V3]], i32 8)
// CHECK-NEXT: ret void
+// SYSTEMZ: declare void @_Block_object_dispose(ptr noundef, i32 noundef signext)
+
typedef double ftype(double);
// It's not clear that we *should* support this syntax, but until that decision
// is made, we should support it properly and not crash.
|
Correctly computing the LLVM types/attributes is complicated in general, so add a variant which does that for you.