extend footnote

[web.git] / personal / _posts / 2018-07-13-arc-synchronization.md

diff --git a/personal/_posts/2018-07-13-arc-synchronization.md b/personal/_posts/2018-07-13-arc-synchronization.md
index 7c78cd0dedf3a8d243fea6669332ae63171c6dc1..37f3a31048caf93655140020e581d7baa7f81beb 100644 (file)
--- a/personal/_posts/2018-07-13-arc-synchronization.md
+++ b/personal/_posts/2018-07-13-arc-synchronization.md
@@ -178,7 +178,7 @@ Discussing the exact details of the bug would probably fill another blog post (`
 With `Relaxed`, no such ordering is induced, so we have a data race.
 To be fair, it is very unlikely that this race could lead to real misbehavior, but I am still happy to know that we now have a proof that `Arc` is mostly[^1] correctly synchronized.
 
-[^1]: "Mostly", you may wonder?  Unfortunately it turns out that there is [one `Relaxed` access in `make_mut`](https://github.com/rust-lang/rust/blob/c0955a34bcb17f0b31d7b86522a520ebe7fa93ac/src/liballoc/sync.rs#L793) that Hai and Jacques-Henri have not yet been able to prove correct.  However, this is likely fixable by making the entire proof more complicated.  The version where that `Relaxed` is also replaced by `Acquire` *has* been proven correct.
+[^1]: "Mostly", you may wonder?  Unfortunately it turns out that there is [one `Relaxed` access in `make_mut`](https://github.com/rust-lang/rust/blob/c0955a34bcb17f0b31d7b86522a520ebe7fa93ac/src/liballoc/sync.rs#L793) that Hai and Jacques-Henri have not yet been able to prove correct.  However, this is likely fixable by making the entire proof more complicated.  The version where that `Relaxed` is also replaced by `Acquire` *has* been proven correct, albeit still against a model of relaxed memory accesses that is not quite as weak as C11.
 
 One last thing:
 I have previously claimed that our [first RustBelt paper]({{ site.baseurl }}{% post_url 2017-07-08-rustbelt %}) already verified correctness of `Arc`, how can there still be bugs?