merge main into amd-staging (llvm#4218)

Author: z1-cciauto

bolt/test/X86/dwarf5-dwoid-no-dwoname.s

@@ -1,7 +1,7 @@
 ## Check that DWARF CU with a valid DWOId but missing a dwo_name is correctly detected.
 # RUN: rm -rf %t && mkdir -p %t && cd %t
 # RUN: llvm-mc -dwarf-version=5 -filetype=obj -triple x86_64-unknown-linux %s -split-dwarf-file=main.dwo -o main.o
-# RUN: %clang -O3 -g -gdwarf-5 -gsplit-dwarf -Wl,-q %t/main.o -o main.exe
+# RUN: %clang %cflags -O3 -g -gdwarf-5 -gsplit-dwarf -Wl,-q %t/main.o -o main.exe
 # RUN: llvm-bolt %t/main.exe -o %t/main.exe.bolt -update-debug-sections  2>&1 | FileCheck %s --check-prefix=PRECHECK
 # PRECHECK: BOLT-ERROR: broken DWARF found in CU at offset 0x3e (DWOId=0x0, missing DW_AT_dwo_name / DW_AT_GNU_dwo_name)
 

clang/docs/ReleaseNotes.rst

@@ -272,6 +272,9 @@ Attribute Changes in Clang
   attribute, allowing the attribute to only be attached to the declaration. Prior, this would be
   treated as an error where the definition and declaration would have differing types.
 
+- New format attributes ``gnu_printf``, ``gnu_scanf``, ``gnu_strftime`` and ``gnu_strfmon`` are added
+  as aliases for ``printf``, ``scanf``, ``strftime`` and ``strfmon``. (#GH16219)
+
 Improvements to Clang's diagnostics
 -----------------------------------
 

clang/include/clang/Sema/Sema.h

@@ -503,7 +503,6 @@ enum class FormatStringType {
   FreeBSDKPrintf,
   OSTrace,
   OSLog,
-  Syslog,
   Unknown
 };
 

clang/lib/AST/ByteCode/Context.cpp

@@ -237,6 +237,52 @@ bool Context::evaluateCharRange(State &Parent, const Expr *SizeExpr,
   return evaluateStringRepr(Parent, SizeExpr, PtrExpr, Result);
 }
 
+bool Context::evaluateString(State &Parent, const Expr *E,
+                             std::string &Result) {
+  assert(Stk.empty());
+  Compiler<EvalEmitter> C(*this, *P, Parent, Stk);
+
+  auto PtrRes = C.interpretAsPointer(E, [&](const Pointer &Ptr) {
+    const Descriptor *FieldDesc = Ptr.getFieldDesc();
+    if (!FieldDesc->isPrimitiveArray())
+      return false;
+
+    if (!Ptr.isConst())
+      return false;
+
+    unsigned N = Ptr.getNumElems();
+
+    if (Ptr.elemSize() == 1 /* bytes */) {
+      const char *Chars = reinterpret_cast<const char *>(Ptr.getRawAddress());
+      unsigned Length = strnlen(Chars, N);
+      // Wasn't null terminated.
+      if (N == Length)
+        return false;
+      Result.assign(Chars, Length);
+      return true;
+    }
+
+    PrimType ElemT = FieldDesc->getPrimType();
+    for (unsigned I = Ptr.getIndex(); I != N; ++I) {
+      INT_TYPE_SWITCH(ElemT, {
+        auto Elem = Ptr.elem<T>(I);
+        if (Elem.isZero())
+          return true;
+        Result.push_back(static_cast<char>(Elem));
+      });
+    }
+    // We didn't find a 0 byte.
+    return false;
+  });
+
+  if (PtrRes.isInvalid()) {
+    C.cleanup();
+    Stk.clear();
+    return false;
+  }
+  return true;
+}
+
 bool Context::evaluateStrlen(State &Parent, const Expr *E, uint64_t &Result) {
   assert(Stk.empty());
   Compiler<EvalEmitter> C(*this, *P, Parent, Stk);

clang/lib/AST/ByteCode/Context.h

@@ -67,6 +67,10 @@ class Context final {
   bool evaluateCharRange(State &Parent, const Expr *SizeExpr,
                          const Expr *PtrExpr, std::string &Result);
 
+  /// Evaluate \param E and if it can be evaluated to a null-terminated string,
+  /// copy the result into \param Result.
+  bool evaluateString(State &Parent, const Expr *E, std::string &Result);
+
   /// Evalute \param E and if it can be evaluated to a string literal,
   /// run strlen() on it.
   bool evaluateStrlen(State &Parent, const Expr *E, uint64_t &Result);

clang/lib/AST/ExprConstant.cpp

@@ -18905,9 +18905,15 @@ std::optional<std::string> Expr::tryEvaluateString(ASTContext &Ctx) const {
   uint64_t Result;
   std::string StringResult;
 
+  if (Info.EnableNewConstInterp) {
+    if (!Info.Ctx.getInterpContext().evaluateString(Info, this, StringResult))
+      return std::nullopt;
+    return StringResult;
+  }
+
   if (EvaluateBuiltinStrLen(this, Result, Info, &StringResult))
     return StringResult;
-  return {};
+  return std::nullopt;
 }
 
 template <typename T>

clang/lib/Format/Format.cpp

@@ -2185,47 +2185,68 @@ std::error_code parseConfiguration(llvm::MemoryBufferRef Config,
   if (Input.error())
     return Input.error();
 
-  for (unsigned i = 0; i < Styles.size(); ++i) {
-    // Ensures that only the first configuration can skip the Language option.
-    if (Styles[i].Language == FormatStyle::LK_None && i != 0)
+  assert(!Styles.empty());
+  const auto StyleCount = Styles.size();
+
+  // Start from the second style as (only) the first one may be the default.
+  for (unsigned I = 1; I < StyleCount; ++I) {
+    const auto Lang = Styles[I].Language;
+    if (Lang == FormatStyle::LK_None)
       return make_error_code(ParseError::Error);
     // Ensure that each language is configured at most once.
-    for (unsigned j = 0; j < i; ++j) {
-      if (Styles[i].Language == Styles[j].Language) {
+    for (unsigned J = 0; J < I; ++J) {
+      if (Lang == Styles[J].Language) {
         LLVM_DEBUG(llvm::dbgs()
                    << "Duplicate languages in the config file on positions "
-                   << j << " and " << i << "\n");
+                   << J << " and " << I << '\n');
         return make_error_code(ParseError::Error);
       }
     }
   }
-  // Look for a suitable configuration starting from the end, so we can
-  // find the configuration for the specific language first, and the default
-  // configuration (which can only be at slot 0) after it.
-  FormatStyle::FormatStyleSet StyleSet;
-  bool LanguageFound = false;
-  for (const FormatStyle &Style : llvm::reverse(Styles)) {
-    const auto Lang = Style.Language;
-    if (Lang != FormatStyle::LK_None)
-      StyleSet.Add(Style);
-    if (Lang == Language ||
-        // For backward compatibility.
-        (Lang == FormatStyle::LK_Cpp && Language == FormatStyle::LK_C)) {
-      LanguageFound = true;
-    } else if (IsDotHFile && Language == FormatStyle::LK_Cpp &&
-               (Lang == FormatStyle::LK_C || Lang == FormatStyle::LK_ObjC)) {
-      Language = Lang;
-      LanguageFound = true;
+
+  int LanguagePos = -1; // Position of the style for Language.
+  int CppPos = -1;      // Position of the style for C++.
+  int CPos = -1;        // Position of the style for C.
+
+  // Search Styles for Language and store the positions of C++ and C styles in
+  // case Language is not found.
+  for (unsigned I = 0; I < StyleCount; ++I) {
+    const auto Lang = Styles[I].Language;
+    if (Lang == Language) {
+      LanguagePos = I;
+      break;
     }
-  }
-  if (!LanguageFound) {
-    if (Styles.empty() || Styles[0].Language != FormatStyle::LK_None)
+    if (Lang == FormatStyle::LK_Cpp)
+      CppPos = I;
+    else if (Lang == FormatStyle::LK_C)
+      CPos = I;
+  }
+
+  // If Language is not found, use the default style if there is one. Otherwise,
+  // use the C style for C++ .h files and for backward compatibility, the C++
+  // style for .c files.
+  if (LanguagePos < 0) {
+    if (Styles[0].Language == FormatStyle::LK_None) // Default style.
+      LanguagePos = 0;
+    else if (IsDotHFile && Language == FormatStyle::LK_Cpp)
+      LanguagePos = CPos;
+    else if (!IsDotHFile && Language == FormatStyle::LK_C)
+      LanguagePos = CppPos;
+    if (LanguagePos < 0)
       return make_error_code(ParseError::Unsuitable);
-    FormatStyle DefaultStyle = Styles[0];
-    DefaultStyle.Language = Language;
-    StyleSet.Add(std::move(DefaultStyle));
   }
-  *Style = *StyleSet.Get(Language);
+
+  for (const auto &S : llvm::reverse(llvm::drop_begin(Styles)))
+    Style->StyleSet.Add(S);
+
+  *Style = Styles[LanguagePos];
+
+  if (LanguagePos == 0) {
+    if (Style->Language == FormatStyle::LK_None) // Default style.
+      Style->Language = Language;
+    Style->StyleSet.Add(*Style);
+  }
+
   if (Style->InsertTrailingCommas != FormatStyle::TCS_None &&
       Style->BinPackArguments) {
     // See comment on FormatStyle::TSC_Wrapped.
@@ -2256,14 +2277,8 @@ FormatStyle::FormatStyleSet::Get(FormatStyle::LanguageKind Language) const {
   if (!Styles)
     return std::nullopt;
   auto It = Styles->find(Language);
-  if (It == Styles->end()) {
-    if (Language != FormatStyle::LK_C)
-      return std::nullopt;
-    // For backward compatibility.
-    It = Styles->find(FormatStyle::LK_Cpp);
-    if (It == Styles->end())
-      return std::nullopt;
-  }
+  if (It == Styles->end())
+    return std::nullopt;
   FormatStyle Style = It->second;
   Style.StyleSet = *this;
   return Style;

clang/lib/Sema/SemaAMDGPU.cpp

@@ -59,9 +59,11 @@ bool SemaAMDGPU::CheckAMDGCNBuiltinFunctionCall(unsigned BuiltinID,
       [[fallthrough]];
     }
     default:
-      Diag(ArgExpr->getExprLoc(), diag::err_amdgcn_load_lds_size_invalid_value)
+      SemaRef.targetDiag(ArgExpr->getExprLoc(),
+                         diag::err_amdgcn_load_lds_size_invalid_value)
           << ArgExpr->getSourceRange();
-      Diag(ArgExpr->getExprLoc(), diag::note_amdgcn_load_lds_size_valid_value)
+      SemaRef.targetDiag(ArgExpr->getExprLoc(),
+                         diag::note_amdgcn_load_lds_size_valid_value)
           << HasGFX950Insts << ArgExpr->getSourceRange();
       return true;
     }

clang/lib/Sema/SemaChecking.cpp

@@ -6866,11 +6866,12 @@ StringRef Sema::GetFormatStringTypeName(FormatStringType FST) {
 
 FormatStringType Sema::GetFormatStringType(StringRef Flavor) {
   return llvm::StringSwitch<FormatStringType>(Flavor)
-      .Case("scanf", FormatStringType::Scanf)
-      .Cases("printf", "printf0", "syslog", FormatStringType::Printf)
+      .Cases("gnu_scanf", "scanf", FormatStringType::Scanf)
+      .Cases("gnu_printf", "printf", "printf0", "syslog",
+             FormatStringType::Printf)
       .Cases("NSString", "CFString", FormatStringType::NSString)
-      .Case("strftime", FormatStringType::Strftime)
-      .Case("strfmon", FormatStringType::Strfmon)
+      .Cases("gnu_strftime", "strftime", FormatStringType::Strftime)
+      .Cases("gnu_strfmon", "strfmon", FormatStringType::Strfmon)
       .Cases("kprintf", "cmn_err", "vcmn_err", "zcmn_err",
              FormatStringType::Kprintf)
       .Case("freebsd_kprintf", FormatStringType::FreeBSDKPrintf)
@@ -6990,7 +6991,6 @@ bool Sema::CheckFormatArguments(ArrayRef<const Expr *> Args,
     case FormatStringType::Kprintf:
     case FormatStringType::FreeBSDKPrintf:
     case FormatStringType::Printf:
-    case FormatStringType::Syslog:
       Diag(FormatLoc, diag::note_format_security_fixit)
         << FixItHint::CreateInsertion(FormatLoc, "\"%s\", ");
       break;
@@ -9120,8 +9120,7 @@ static void CheckFormatString(
   if (Type == FormatStringType::Printf || Type == FormatStringType::NSString ||
       Type == FormatStringType::Kprintf ||
       Type == FormatStringType::FreeBSDKPrintf ||
-      Type == FormatStringType::OSLog || Type == FormatStringType::OSTrace ||
-      Type == FormatStringType::Syslog) {
+      Type == FormatStringType::OSLog || Type == FormatStringType::OSTrace) {
     bool IsObjC =
         Type == FormatStringType::NSString || Type == FormatStringType::OSTrace;
     if (ReferenceFormatString == nullptr) {
@@ -9157,8 +9156,7 @@ bool Sema::CheckFormatStringsCompatible(
   if (Type != FormatStringType::Printf && Type != FormatStringType::NSString &&
       Type != FormatStringType::Kprintf &&
       Type != FormatStringType::FreeBSDKPrintf &&
-      Type != FormatStringType::OSLog && Type != FormatStringType::OSTrace &&
-      Type != FormatStringType::Syslog)
+      Type != FormatStringType::OSLog && Type != FormatStringType::OSTrace)
     return true;
 
   bool IsObjC =
@@ -9192,8 +9190,7 @@ bool Sema::ValidateFormatString(FormatStringType Type,
   if (Type != FormatStringType::Printf && Type != FormatStringType::NSString &&
       Type != FormatStringType::Kprintf &&
       Type != FormatStringType::FreeBSDKPrintf &&
-      Type != FormatStringType::OSLog && Type != FormatStringType::OSTrace &&
-      Type != FormatStringType::Syslog)
+      Type != FormatStringType::OSLog && Type != FormatStringType::OSTrace)
     return true;
 
   FormatStringLiteral RefLit = Str;

clang/lib/Sema/SemaDeclAttr.cpp

@@ -3629,10 +3629,11 @@ static FormatAttrKind getFormatAttrKind(StringRef Format) {
       // Check for formats that get handled specially.
       .Case("NSString", NSStringFormat)
       .Case("CFString", CFStringFormat)
-      .Case("strftime", StrftimeFormat)
+      .Cases("gnu_strftime", "strftime", StrftimeFormat)
 
       // Otherwise, check for supported formats.
-      .Cases("scanf", "printf", "printf0", "strfmon", SupportedFormat)
+      .Cases("gnu_scanf", "scanf", "gnu_printf", "printf", "printf0",
+             "gnu_strfmon", "strfmon", SupportedFormat)
       .Cases("cmn_err", "vcmn_err", "zcmn_err", SupportedFormat)
       .Cases("kprintf", "syslog", SupportedFormat) // OpenBSD.
       .Case("freebsd_kprintf", SupportedFormat)    // FreeBSD.