Avoid reading past end of archive looking for long file name

[inglorion]

Summary

GNU-style archives store long file names separated by newlines. The
last file name is not followed by a newline. This change stops the
search for the newline at the end of the string table, which avoids
a crash.

Fixes PR37244.

[pcc]

Is it possible to construct a test case?

[• espindola]

llvm/lib/Object/Archive.cpp

186

This should be an error, no?

[inglorion]

llvm/lib/Object/Archive.cpp

186

Not as I understand it. If I understand the GNU format correctly, all long filenames end in a slash, but newline is only used as a separator, meaning the last file doesn't get one. So the name is delimited either by "/\n" or by "/" <EOF>.

[inglorion]

A test case for this would be an archive with a long file name at the end of the string table. It may have to be a specific size to tickle the problem I'm trying to fix here, to prevent MemoryBuffer or the OS from padding the archive with 0 bytes, which would avoid the problem.

Unfortunately, I have to catch a plane and I'll be gone for a week. If anyone wants to commandeer this and land it, please go ahead.

[inglorion]

We landed D46527 instead.

llvm/lib/Object/Archive.cpp

if (Name.substr(1).rtrim(' ').getAsInteger(10, StringOffset)) {
                                Parent->getData().data();
       return malformedError("long name offset characters after the '/' are "
                             "not all decimal numbers: '" + Buf + "' for "
                             "archive member header at offset " +
                             Twine(ArchiveOffset));
     }
 
     // Verify it.
-    if (StringOffset >= Parent->getStringTable().size()) {
+    StringRef ST = Parent->getStringTable();
+    if (StringOffset >= ST.size()) {
       uint64_t ArchiveOffset = reinterpret_cast<const char *>(ArMemHdr) -
                                Parent->getData().data();
       return malformedError("long name offset " + Twine(StringOffset) + " past "
                             "the end of the string table for archive member "
                             "header at offset " + Twine(ArchiveOffset));
     }
-    const char *addr = Parent->getStringTable().begin() + StringOffset;
 
+    const char *Start = ST.begin() + StringOffset;
     // GNU long file names end with a "/\n".
     if (Parent->kind() == Archive::K_GNU ||
         Parent->kind() == Archive::K_GNU64) {
-      StringRef::size_type End = StringRef(addr).find('\n');
-      return StringRef(addr, End - 1);
+      StringRef::size_type End = ST.find('\n', StringOffset);
+      if (End == StringRef::npos)
+        End = ST.size();
+      return StringRef(Start, End - StringOffset - 1);
     }
-    return addr;
+    return Start;
   }
 
   if (Name.startswith("#1/")) {
     uint64_t NameLength;
     if (Name.substr(3).rtrim(' ').getAsInteger(10, NameLength)) {
       std::string Buf;
       raw_string_ostream OS(Buf);
       OS.write_escaped(Name.substr(3).rtrim(' '));