[clang]Fix Handle structs exceeding 1EB size limit (#146032)

When declaring multiple arrays of 1 ExaByte in a struct, the offset can
exceed 2EB, causing incorrect struct size reporting (only 1EB). This fix
ensures an error is thrown, preventing the generation of incorrect
assembly. #60272

Author: Karthikdhondi

clang/include/clang/Basic/DiagnosticASTKinds.td

@@ -999,6 +999,8 @@ def note_module_odr_violation_mismatch_decl_unknown : Note<
   "different friend declaration|different function template|different method|"
   "different instance variable|different property|another unexpected decl}2">;
 
+def err_struct_too_large : Error<
+  "structure '%0' is too large, which exceeds maximum allowed size of %1 bytes">;
 
 def remark_sanitize_address_insert_extra_padding_accepted : Remark<
     "-fsanitize-address-field-padding applied to %0">, ShowInSystemHeader,

clang/lib/AST/RecordLayoutBuilder.cpp

@@ -3463,6 +3463,13 @@ ASTContext::getASTRecordLayout(const RecordDecl *D) const {
 
   ASTRecordLayouts[D] = NewEntry;
 
+  constexpr uint64_t MaxStructSizeInBytes = 1ULL << 60;
+  CharUnits StructSize = NewEntry->getSize();
+  if (static_cast<uint64_t>(StructSize.getQuantity()) >= MaxStructSizeInBytes) {
+    getDiagnostics().Report(D->getLocation(), diag::err_struct_too_large)
+        << D->getName() << MaxStructSizeInBytes;
+  }
+
   if (getLangOpts().DumpRecordLayouts) {
     llvm::outs() << "\n*** Dumping AST Record Layout\n";
     DumpRecordLayout(D, llvm::outs(), getLangOpts().DumpRecordLayoutsSimple);

clang/test/AST/absurdly_big_struct.cpp

@@ -0,0 +1,13 @@
+// RUN: %clang_cc1 -fsyntax-only -verify %s -triple x86_64-linux-gnu
+
+struct a { // expected-error {{structure 'a' is too large, which exceeds maximum allowed size of 1152921504606846976 bytes}}
+  char x[1ull<<60]; 
+  char x2[1ull<<60]; 
+};
+
+a z[1];
+long long x() { return sizeof(a); }
+long long x2() { return sizeof(a::x); }
+long long x3() { return sizeof(a::x2); }
+long long x4() { return sizeof(z); }
+

clang/test/Sema/offsetof-64.c

@@ -2,15 +2,15 @@
 
 // PR15216
 // Don't crash when taking computing the offset of structs with large arrays.
-const unsigned long Size = (1l << 60);
+const unsigned long Size = (1l << 58);
 
 struct Chunk1 {
   char padding[Size]; // expected-warning {{folded to constant}}
   char more_padding[1][Size]; // expected-warning {{folded to constant}}
   char data;
 };
 
-int test1 = __builtin_offsetof(struct Chunk1, data);
+unsigned long test1 = __builtin_offsetof(struct Chunk1, data);
 
 struct Chunk2 {
   char padding[Size][Size][Size];  // expected-error {{array is too large}}