-
Notifications
You must be signed in to change notification settings - Fork 12.6k
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][AArch64] Fixed incorrect _BitInt alignment #90602
Conversation
@llvm/pr-subscribers-clang @llvm/pr-subscribers-backend-aarch64 Author: None (Lukacma) ChangesThis patch makes determining alignment and width of BitInt to be target ABI specific and makes it consistent with Procedure Call Standard for the Arm® 64-bit Architecture (AArch64) for AArch64 targets. Full diff: https://github.com/llvm/llvm-project/pull/90602.diff 5 Files Affected:
diff --git a/clang/include/clang/Basic/TargetInfo.h b/clang/include/clang/Basic/TargetInfo.h
index 3ced2e7397a754..1b5efa488b6ded 100644
--- a/clang/include/clang/Basic/TargetInfo.h
+++ b/clang/include/clang/Basic/TargetInfo.h
@@ -518,6 +518,14 @@ class TargetInfo : public TransferrableTargetInfo,
/// getInt128Align() - Returns the alignment of Int128.
unsigned getInt128Align() const { return Int128Align; }
+ /// getBitIntAlign/Width - Return aligned size of '_BitInt' and
+ /// 'unsigned _BitInt' for this target, in bits.
+ unsigned getBitIntWidth(unsigned NumBits) const {
+ return llvm::alignTo(NumBits, getBitIntAlign(NumBits));}
+ virtual unsigned getBitIntAlign(unsigned NumBits) const {
+ return std::clamp<unsigned>(llvm::PowerOf2Ceil(NumBits), getCharWidth(),
+ getLongLongAlign());}
+
/// getShortAccumWidth/Align - Return the size of 'signed short _Accum' and
/// 'unsigned short _Accum' for this target, in bits.
unsigned getShortAccumWidth() const { return ShortAccumWidth; }
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index cbf4932aff9a6b..f440af50e08a49 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -2263,9 +2263,8 @@ TypeInfo ASTContext::getTypeInfoImpl(const Type *T) const {
}
case Type::BitInt: {
const auto *EIT = cast<BitIntType>(T);
- Align = std::clamp<unsigned>(llvm::PowerOf2Ceil(EIT->getNumBits()),
- getCharWidth(), Target->getLongLongAlign());
- Width = llvm::alignTo(EIT->getNumBits(), Align);
+ Align = Target->getBitIntAlign(EIT->getNumBits());
+ Width = Target->getBitIntWidth(EIT->getNumBits());
break;
}
case Type::Record:
diff --git a/clang/lib/Basic/Targets/AArch64.cpp b/clang/lib/Basic/Targets/AArch64.cpp
index c8d243a8fb7aea..9a6c197ff77b58 100644
--- a/clang/lib/Basic/Targets/AArch64.cpp
+++ b/clang/lib/Basic/Targets/AArch64.cpp
@@ -1473,6 +1473,10 @@ bool AArch64TargetInfo::validatePointerAuthKey(
bool AArch64TargetInfo::hasInt128Type() const { return true; }
+unsigned AArch64TargetInfo::getBitIntAlign(unsigned NumBits) const{
+ return std::clamp<unsigned>(llvm::PowerOf2Ceil(NumBits), getCharWidth(),
+ getInt128Align());}
+
AArch64leTargetInfo::AArch64leTargetInfo(const llvm::Triple &Triple,
const TargetOptions &Opts)
: AArch64TargetInfo(Triple, Opts) {}
@@ -1674,4 +1678,4 @@ void RenderScript64TargetInfo::getTargetDefines(const LangOptions &Opts,
MacroBuilder &Builder) const {
Builder.defineMacro("__RENDERSCRIPT__");
AArch64leTargetInfo::getTargetDefines(Opts, Builder);
-}
+}
\ No newline at end of file
diff --git a/clang/lib/Basic/Targets/AArch64.h b/clang/lib/Basic/Targets/AArch64.h
index 12fb50286f7511..d8cdc814b2a9ae 100644
--- a/clang/lib/Basic/Targets/AArch64.h
+++ b/clang/lib/Basic/Targets/AArch64.h
@@ -202,6 +202,9 @@ class LLVM_LIBRARY_VISIBILITY AArch64TargetInfo : public TargetInfo {
bool hasBitIntType() const override { return true; }
bool validateTarget(DiagnosticsEngine &Diags) const override;
+
+ unsigned getBitIntAlign(unsigned NumBits) const override;
+
};
class LLVM_LIBRARY_VISIBILITY AArch64leTargetInfo : public AArch64TargetInfo {
diff --git a/clang/test/CodeGen/aapcs64-align.cpp b/clang/test/CodeGen/aapcs64-align.cpp
index de231f2123b975..7a8151022852ea 100644
--- a/clang/test/CodeGen/aapcs64-align.cpp
+++ b/clang/test/CodeGen/aapcs64-align.cpp
@@ -1,7 +1,7 @@
// REQUIRES: arm-registered-target
// RUN: %clang_cc1 -triple aarch64-none-elf \
// RUN: -O2 \
-// RUN: -emit-llvm -o - %s | FileCheck %s
+// RUN: -emit-llvm -fexperimental-max-bitint-width=1024 -o - %s | FileCheck %s
extern "C" {
@@ -100,4 +100,66 @@ void f5m(int, int, int, int, int, P16);
// CHECK: declare void @f5(i32 noundef, [2 x i64])
// CHECK: declare void @f5m(i32 noundef, i32 noundef, i32 noundef, i32 noundef, i32 noundef, [2 x i64])
+//BitInt alignment
+struct BITINT129 {
+ char ch;
+ unsigned _BitInt(129) v;
+};
+
+int test_bitint129(){
+ return __builtin_offsetof(struct BITINT129, v);
}
+// CHECK: ret i32 16
+
+struct BITINT127 {
+ char ch;
+ _BitInt(127) v;
+};
+
+int test_bitint127(){
+ return __builtin_offsetof(struct BITINT127, v);
+}
+// CHECK: ret i32 16
+
+struct BITINT63 {
+ char ch;
+ _BitInt(63) v;
+};
+
+int test_bitint63(){
+ return __builtin_offsetof(struct BITINT63, v);
+}
+// CHECK: ret i32 8
+
+struct BITINT32 {
+ char ch;
+ unsigned _BitInt(32) v;
+};
+
+int test_bitint32(){
+ return __builtin_offsetof(struct BITINT32, v);
+}
+// CHECK: ret i32 4
+
+struct BITINT9 {
+ char ch;
+ unsigned _BitInt(9) v;
+};
+
+int test_bitint9(){
+ return __builtin_offsetof(struct BITINT9, v);
+}
+// CHECK: ret i32 2
+
+struct BITINT8 {
+ char ch;
+ unsigned _BitInt(8) v;
+};
+
+int test_bitint8(){
+ return __builtin_offsetof(struct BITINT8, v);
+}
+// CHECK: ret i32 1
+
+}
+
|
✅ With the latest revision this PR passed the C/C++ code formatter. |
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.
LG
clang/lib/Basic/Targets/AArch64.cpp
Outdated
@@ -1674,4 +1679,4 @@ void RenderScript64TargetInfo::getTargetDefines(const LangOptions &Opts, | |||
MacroBuilder &Builder) const { | |||
Builder.defineMacro("__RENDERSCRIPT__"); | |||
AArch64leTargetInfo::getTargetDefines(Opts, Builder); | |||
} | |||
} |
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.
Missing newline?
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.
Done
unsigned getBitIntWidth(unsigned NumBits) const { | ||
return llvm::alignTo(NumBits, getBitIntAlign(NumBits)); | ||
} | ||
virtual unsigned getBitIntAlign(unsigned NumBits) const { |
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.
Instead of making this virtual, maybe it makes sense to add a field std::optional<unsigned> BitIntMaxAlign
to TargetInfo? I expect the logic here besides the max width to be consistent across all targets.
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.
I am not sure I understand how this idea would work. Could you please elaborate on what do you mean ?
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.
unsigned getBitIntMaxAlign() {
return BitIntMaxAlign.or(LongLongAlign);
}
unsigned getBitIntAlign(unsigned NumBits) const {
return std::clamp<unsigned>(llvm::PowerOf2Ceil(NumBits), getCharWidth(),
getBitIntMaxAlign());
}
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.
Done
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
This patch makes determining alignment and width of BitInt to be target ABI specific and makes it consistent with Procedure Call Standard for the Arm® 64-bit Architecture (AArch64) for AArch64 targets.