Merge pull request #83534 from eeckstein/fix-string-serialization

Serialization: fix serialization of strings containing a 0-byte

Author: eeckstein

lib/Serialization/ModuleFileSharedCore.cpp

@@ -1422,11 +1422,13 @@ StringRef ModuleFileSharedCore::getIdentifierText(IdentifierID IID) const {
 
   assert(!IdentifierData.empty() && "no identifier data in module");
 
-  StringRef rawStrPtr = IdentifierData.substr(offset);
-  size_t terminatorOffset = rawStrPtr.find('\0');
-  assert(terminatorOffset != StringRef::npos &&
+  size_t endOffset = ((rawID + 1 < Identifiers.size()) ?
+    Identifiers[rawID + 1] : IdentifierData.size()) - 1;
+
+  ASSERT(IdentifierData[endOffset] == '\0' &&
          "unterminated identifier string data");
-  return rawStrPtr.slice(0, terminatorOffset);
+
+  return IdentifierData.slice(offset, endOffset);
 }
 
 ModuleFileSharedCore::ModuleFileSharedCore(

test/SIL/Serialization/literals.sil

@@ -14,6 +14,10 @@ bb0:
   %4 = string_literal bytes "ABCD"
 // CHECK: string_literal oslog "os log section"
   %5 = string_literal oslog "os log section"
-  %6 = tuple ()
-  return %6 : $()
+// CHECK: string_literal utf8 "abc\0def"
+  %6 = string_literal utf8 "abc\0def"
+// CHECK: string_literal utf8 "xyz\0"
+  %7 = string_literal utf8 "xyz\0"
+  %8 = tuple ()
+  return %8 : $()
 }