Redox: Fix Condvar.wait(); do not lock mutex twice #43082
Conversation
alexcrichton
added
S-waiting-on-review
|
@bors: r+ rollup |
|
📌 Commit 86191f4 has been approved by |
|
🔒 Merge conflict |
|
Is there a test for this stuff? I'm assuming the previous behavior would be an immediate deadlock as soon as the convar wakes up? |
Yes. The implementation now matches https://locklessinc.com/articles/mutex_cv_futex/, which @jackpot51 says he used as a reference to implement it; it seems this line was added by accident. I don't see a test for |
shepmaster
added
the
S-waiting-on-author
|
Heads up, @ids1024; you've got a merge conflict. |
|
But what merge conflict? It seems to merge cleanly with master, and the only file changed here is a Redox-specific file that hasn't been changed in months. |
|
@ids1024 you'll want to just rebase against rust-lang/rust's master branch |
The atomic_xchg() loop locks the mutex, so the call to mutex_lock is incorrect, and blocks.
|
@alexcrichton it has been rebased. |
|
@bors: r+ |
|
📌 Commit 59981e4 has been approved by |
Redox: Fix Condvar.wait(); do not lock mutex twice The atomic_xchg() loop locks the mutex, so the call to mutex_lock is incorrect, and blocks.
|
☀️ Test successful - status-appveyor, status-travis |
The atomic_xchg() loop locks the mutex, so the call to mutex_lock is
incorrect, and blocks.