Repository navigation
Revert "Fuzz waitqueue instructions (#9012)" - #9230
Conversation
This reverts commit 6c1a3cb.
kripken
left a comment
kripken
left a comment
There was a problem hiding this comment.
lgtm to revert, though if this is a matter of another day or two, waiting is also ok for me. (In my ping I mainly wanted to make sure this wasn't forgotten.)
| AndInt32, arrayNew->size, builder.makeConst(int32_t(1024 - 1))); | ||
| } | ||
| } | ||
| if (!ATOMIC_WAITS) { |
There was a problem hiding this comment.
Side note, should this have been on FuzzParams and not on the main class?
There was a problem hiding this comment.
I guess so. This was here before my PR (actually 9 years ago), and it looks like it's a compile-time constant; there's no code path that sets it to true.
I can see if we can refactor this part too.
There was a problem hiding this comment.
Oh, sorry, I thought it was part of the diff here...
Got it. I want to clean up some of the code as part of my fix anyway, so I'll revert to give us more time. |
|
Ran ~300k fuzzer iterations with no issues. |
2 participants
Reverts #9012. The fuzzing code for waitqueues is correct but it exposes a bug with how waitqueues are handled in ctor-eval. Revert for now to disable fuzzing while the underlying issue is fixed. The only flag guarding waitqueue instructions is
--shared-everythingwhich we don't want to disable since it covers other features too.Part of #8315.