[libcxx-commits] [libcxx] [libc++] Fix UB in <expected> related to "has value" flag (#68552) (PR #68733)

Richard Smith via libcxx-commits libcxx-commits at lists.llvm.org
Thu Oct 19 13:22:20 PDT 2023


Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>,
Jan =?utf-8?q?Kokemüller?= <jan.kokemueller at gmail.com>
Message-ID:
In-Reply-To: <llvm/llvm-project/pull/68733/libcxx at github.com>


================
@@ -440,9 +399,10 @@ class expected {
       std::destroy_at(std::addressof(__union_.__val_));
     } else {
       std::destroy_at(std::addressof(__union_.__unex_));
-      __has_val_ = true;
     }
-    return *std::construct_at(std::addressof(__union_.__val_), std::forward<_Args>(__args)...);
+    std::construct_at(std::addressof(__union_.__val_), std::forward<_Args>(__args)...);
----------------
zygoloid wrote:

I think this case is still broken, as this `construct_at` can write into the tail padding of the `expected` object, which might be nested within another object. We might need a compiler extension to make this work. :-(

https://github.com/llvm/llvm-project/pull/68733


More information about the libcxx-commits mailing list