Merge branch 'main' into merge-functions

Author: Casperento

clang/include/clang/Basic/AttrDocs.td

@@ -3985,6 +3985,8 @@ The capturing entity ``X`` can be one of the following:
       std::set<std::string_view> s;
     };
 
+  Note: When applied to a constructor parameter, `[[clang::lifetime_capture_by(this)]]` is just an alias of `[[clang::lifetimebound]]`.
+
 - `global`, `unknown`.
 
   .. code-block:: c++

clang/include/clang/Basic/TargetBuiltins.h

@@ -336,6 +336,7 @@ namespace clang {
     bool isTupleSet() const { return Flags & IsTupleSet; }
     bool isReadZA() const { return Flags & IsReadZA; }
     bool isWriteZA() const { return Flags & IsWriteZA; }
+    bool setsFPMR() const { return Flags & SetsFPMR; }
     bool isReductionQV() const { return Flags & IsReductionQV; }
     uint64_t getBits() const { return Flags; }
     bool isFlagSet(uint64_t Flag) const { return Flags & Flag; }

clang/include/clang/Basic/arm_sve.td

@@ -2422,14 +2422,16 @@ let SVETargetGuard = InvalidMode, SMETargetGuard = "sme2" in {
   def SVUUNPK_X4 : SInst<"svunpk_{d}[_{3}_x4]", "42.h", "UsUiUl", MergeNone, "aarch64_sve_uunpk_x4", [IsStreaming], []>;
 }
 
-//
-// Multi-vector scaling
-//
-let  SVETargetGuard = InvalidMode, SMETargetGuard = "sme2,fp8" in {
+let SVETargetGuard = InvalidMode, SMETargetGuard = "sme2,fp8" in {
+  // Multi-vector scaling
   def FSCALE_SINGLE_X2 : Inst<"svscale[_single_{d}_x2]", "22x", "fhd", MergeNone, "aarch64_sme_fp8_scale_single_x2", [IsStreaming],[]>;
   def FSCALE_SINGLE_X4 : Inst<"svscale[_single_{d}_x4]", "44x", "fhd", MergeNone, "aarch64_sme_fp8_scale_single_x4", [IsStreaming],[]>;
   def FSCALE_X2 : Inst<"svscale[_{d}_x2]", "222.x", "fhd", MergeNone, "aarch64_sme_fp8_scale_x2", [IsStreaming],[]>;
   def FSCALE_X4 : Inst<"svscale[_{d}_x4]", "444.x", "fhd", MergeNone, "aarch64_sme_fp8_scale_x4", [IsStreaming],[]>;
+
+  // Convert from FP8 to deinterleaved half-precision/BFloat16 multi-vector
+  def SVF1CVTL : Inst<"svcvtl1_{d}[_mf8]_x2_fpm",  "2~>", "bh", MergeNone, "aarch64_sve_fp8_cvtl1_x2",  [IsStreaming, SetsFPMR], []>;
+  def SVF2CVTL : Inst<"svcvtl2_{d}[_mf8]_x2_fpm",  "2~>", "bh", MergeNone, "aarch64_sve_fp8_cvtl2_x2",  [IsStreaming, SetsFPMR], []>;
 }
 
 let SVETargetGuard = "sve2p1", SMETargetGuard = "sme2" in {

clang/include/clang/Basic/arm_sve_sme_incl.td

@@ -94,6 +94,7 @@ include "arm_immcheck_incl.td"
 // l: int64_t
 // m: uint32_t
 // n: uint64_t
+// >: fpm_t
 
 // [: svuint8_t
 // t: svint32_t
@@ -103,6 +104,7 @@ include "arm_immcheck_incl.td"
 // M: svfloat32_t
 // N: svfloat64_t
 // $: svbfloat16_t
+// ~: svmfloat8_t
 
 // J: Prefetch type (sv_prfop)
 
@@ -235,6 +237,7 @@ def IsInOutZA                       : FlagType<0x200000000000>;
 def IsInZT0                         : FlagType<0x400000000000>;
 def IsOutZT0                        : FlagType<0x800000000000>;
 def IsInOutZT0                      : FlagType<0x1000000000000>;
+def SetsFPMR                        : FlagType<0x2000000000000>;
 
 defvar InvalidMode = "";
 

clang/lib/CodeGen/CGBuiltin.cpp

@@ -10859,6 +10859,10 @@ Value *CodeGenFunction::EmitAArch64SVEBuiltinExpr(unsigned BuiltinID,
   else if (TypeFlags.isUndef())
     return UndefValue::get(Ty);
   else if (Builtin->LLVMIntrinsic != 0) {
+    // Emit set FPMR for intrinsics that require it
+    if (TypeFlags.setsFPMR())
+      Builder.CreateCall(CGM.getIntrinsic(Intrinsic::aarch64_set_fpmr),
+                         Ops.pop_back_val());
     if (TypeFlags.getMergeType() == SVETypeFlags::MergeZeroExp)
       InsertExplicitZeroOperand(Builder, Ty, Ops);
 

clang/lib/Sema/CheckExprLifetime.cpp

@@ -9,6 +9,7 @@
 #include "CheckExprLifetime.h"
 #include "clang/AST/Decl.h"
 #include "clang/AST/Expr.h"
+#include "clang/AST/Type.h"
 #include "clang/Basic/DiagnosticSema.h"
 #include "clang/Sema/Initialization.h"
 #include "clang/Sema/Sema.h"
@@ -253,9 +254,17 @@ static void visitLocalsRetainedByReferenceBinding(IndirectLocalPath &Path,
                                                   LocalVisitor Visit);
 
 template <typename T> static bool isRecordWithAttr(QualType Type) {
-  if (auto *RD = Type->getAsCXXRecordDecl())
-    return RD->hasAttr<T>();
-  return false;
+  auto *RD = Type->getAsCXXRecordDecl();
+  if (!RD)
+    return false;
+  if (auto *CTSD = dyn_cast<ClassTemplateSpecializationDecl>(RD))
+    RD = CTSD->getSpecializedTemplate()->getTemplatedDecl();
+  return RD->hasAttr<T>();
+}
+
+bool isPointerLikeType(QualType QT) {
+  return isRecordWithAttr<PointerAttr>(QT) || QT->isPointerType() ||
+         QT->isNullPtrType();
 }
 
 // Decl::isInStdNamespace will return false for iterators in some STL
@@ -276,11 +285,6 @@ static bool isInStlNamespace(const Decl *D) {
   return DC->isStdNamespace();
 }
 
-static bool isPointerLikeType(QualType Type) {
-  return isRecordWithAttr<PointerAttr>(Type) || Type->isPointerType() ||
-         Type->isNullPtrType();
-}
-
 // Returns true if the given Record decl is a form of `GSLOwner<Pointer>`
 // type, e.g. std::vector<string_view>, std::optional<string_view>.
 static bool isContainerOfPointer(const RecordDecl *Container) {
@@ -623,6 +627,26 @@ static void visitFunctionCallArguments(IndirectLocalPath &Path, Expr *Call,
     }
     if (CheckCoroCall || Callee->getParamDecl(I)->hasAttr<LifetimeBoundAttr>())
       VisitLifetimeBoundArg(Callee->getParamDecl(I), Arg);
+    else if (const auto *CaptureAttr =
+                 Callee->getParamDecl(I)->getAttr<LifetimeCaptureByAttr>();
+             CaptureAttr && isa<CXXConstructorDecl>(Callee) &&
+             llvm::any_of(CaptureAttr->params(), [](int ArgIdx) {
+               return ArgIdx == LifetimeCaptureByAttr::THIS;
+             }))
+      // `lifetime_capture_by(this)` in a class constructor has the same
+      // semantics as `lifetimebound`:
+      //
+      // struct Foo {
+      //   const int& a;
+      //   // Equivalent to Foo(const int& t [[clang::lifetimebound]])
+      //   Foo(const int& t [[clang::lifetime_capture_by(this)]]) : a(t) {}
+      // };
+      //
+      // In the implementation, `lifetime_capture_by` is treated as an alias for
+      // `lifetimebound` and shares the same code path. This implies the emitted
+      // diagnostics will be emitted under `-Wdangling`, not
+      // `-Wdangling-capture`.
+      VisitLifetimeBoundArg(Callee->getParamDecl(I), Arg);
     else if (EnableGSLAnalysis && I == 0) {
       // Perform GSL analysis for the first argument
       if (shouldTrackFirstArgument(Callee)) {

clang/lib/Sema/CheckExprLifetime.h

@@ -18,6 +18,10 @@
 
 namespace clang::sema {
 
+// Tells whether the type is annotated with [[gsl::Pointer]] or is a pointer
+// type.
+bool isPointerLikeType(QualType QT);
+
 /// Describes an entity that is being assigned.
 struct AssignedEntity {
   // The left-hand side expression of the assignment.

clang/lib/Sema/SemaAttr.cpp

@@ -269,18 +269,6 @@ void Sema::inferLifetimeBoundAttribute(FunctionDecl *FD) {
   }
 }
 
-static bool isPointerLikeType(QualType QT) {
-  QT = QT.getNonReferenceType();
-  if (QT->isPointerType())
-    return true;
-  auto *RD = QT->getAsCXXRecordDecl();
-  if (!RD)
-    return false;
-  if (auto *CTSD = dyn_cast<ClassTemplateSpecializationDecl>(RD))
-    RD = CTSD->getSpecializedTemplate()->getTemplatedDecl();
-  return RD->hasAttr<PointerAttr>();
-}
-
 void Sema::inferLifetimeCaptureByAttribute(FunctionDecl *FD) {
   if (!FD)
     return;
@@ -299,7 +287,7 @@ void Sema::inferLifetimeCaptureByAttribute(FunctionDecl *FD) {
     if (PVD->hasAttr<LifetimeCaptureByAttr>())
       return;
   for (ParmVarDecl *PVD : MD->parameters()) {
-    if (isPointerLikeType(PVD->getType())) {
+    if (sema::isPointerLikeType(PVD->getType().getNonReferenceType())) {
       int CaptureByThis[] = {LifetimeCaptureByAttr::THIS};
       PVD->addAttr(
           LifetimeCaptureByAttr::CreateImplicit(Context, CaptureByThis, 1));

clang/lib/Sema/SemaChecking.cpp

@@ -3240,8 +3240,14 @@ void Sema::checkLifetimeCaptureBy(FunctionDecl *FD, bool IsMemberFunction,
                                  unsigned ArgIdx) {
     if (!Attr)
       return;
+
     Expr *Captured = const_cast<Expr *>(GetArgAt(ArgIdx));
     for (int CapturingParamIdx : Attr->params()) {
+      // lifetime_capture_by(this) case is handled in the lifetimebound expr
+      // initialization codepath.
+      if (CapturingParamIdx == LifetimeCaptureByAttr::THIS &&
+          isa<CXXConstructorDecl>(FD))
+        continue;
       Expr *Capturing = const_cast<Expr *>(GetArgAt(CapturingParamIdx));
       CapturingEntity CE{Capturing};
       // Ensure that 'Captured' outlives the 'Capturing' entity.

clang/test/CodeGen/AArch64/fp8-intrinsics/acle_sme2_fp8_cvt.c

@@ -0,0 +1,81 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
+
+// REQUIRES: aarch64-registered-target
+
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme2 -target-feature +fp8 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme2 -target-feature +fp8 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme2 -target-feature +fp8 -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
+// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme2 -target-feature +fp8 -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
+// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme -target-feature +sme2 -target-feature +fp8 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
+
+#include <arm_sve.h>
+
+#ifdef SVE_OVERLOADED_FORMS
+#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3) A1##A3
+#else
+#define SVE_ACLE_FUNC(A1,A2,A3) A1##A2##A3
+#endif
+
+// CHECK-LABEL: @test_cvtl1_f16_x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.fp8.cvtl1.x2.nxv8f16(<vscale x 16 x i8> [[ZN:%.*]])
+// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
+//
+// CPP-CHECK-LABEL: @_Z17test_cvtl1_f16_x2u13__SVMfloat8_tm(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.fp8.cvtl1.x2.nxv8f16(<vscale x 16 x i8> [[ZN:%.*]])
+// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
+//
+svfloat16x2_t test_cvtl1_f16_x2(svmfloat8_t zn, fpm_t fpmr)  __arm_streaming {
+  return SVE_ACLE_FUNC(svcvtl1_f16,_mf8,_x2_fpm)(zn, fpmr);
+}
+
+// CHECK-LABEL: @test_cvtl2_f16_x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.fp8.cvtl2.x2.nxv8f16(<vscale x 16 x i8> [[ZN:%.*]])
+// CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
+//
+// CPP-CHECK-LABEL: @_Z17test_cvtl2_f16_x2u13__SVMfloat8_tm(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.fp8.cvtl2.x2.nxv8f16(<vscale x 16 x i8> [[ZN:%.*]])
+// CPP-CHECK-NEXT:    ret { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP0]]
+//
+svfloat16x2_t test_cvtl2_f16_x2(svmfloat8_t zn, fpm_t fpmr)  __arm_streaming {
+  return SVE_ACLE_FUNC(svcvtl2_f16,_mf8,_x2_fpm)(zn, fpmr);
+}
+
+// CHECK-LABEL: @test_cvtl1_bf16_x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.fp8.cvtl1.x2.nxv8bf16(<vscale x 16 x i8> [[ZN:%.*]])
+// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]]
+//
+// CPP-CHECK-LABEL: @_Z18test_cvtl1_bf16_x2u13__SVMfloat8_tm(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.fp8.cvtl1.x2.nxv8bf16(<vscale x 16 x i8> [[ZN:%.*]])
+// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]]
+//
+svbfloat16x2_t test_cvtl1_bf16_x2(svmfloat8_t zn, fpm_t fpmr)  __arm_streaming {
+  return SVE_ACLE_FUNC(svcvtl1_bf16,_mf8,_x2_fpm)(zn, fpmr);
+}
+
+// CHECK-LABEL: @test_cvtl2_bf16_x2(
+// CHECK-NEXT:  entry:
+// CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.fp8.cvtl2.x2.nxv8bf16(<vscale x 16 x i8> [[ZN:%.*]])
+// CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]]
+//
+// CPP-CHECK-LABEL: @_Z18test_cvtl2_bf16_x2u13__SVMfloat8_tm(
+// CPP-CHECK-NEXT:  entry:
+// CPP-CHECK-NEXT:    tail call void @llvm.aarch64.set.fpmr(i64 [[FPMR:%.*]])
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.fp8.cvtl2.x2.nxv8bf16(<vscale x 16 x i8> [[ZN:%.*]])
+// CPP-CHECK-NEXT:    ret { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP0]]
+//
+svbfloat16x2_t test_cvtl2_bf16_x2(svmfloat8_t zn, fpm_t fpmr)  __arm_streaming {
+  return SVE_ACLE_FUNC(svcvtl2_bf16,_mf8,_x2_fpm)(zn, fpmr);
+}