Open delguoqing opened 1 month ago
Might be related: #396
Similar to #396 , If GC_ASSERTIONS is enabled in my project, the process will abort at that
sem_getvalue(&GC_suspend_ack_sem, &i);
GC_ASSERT(0 == i);
I have a fix for this problem and want to contribute, How I can do that?
I have a fix for this problem and want to contribute
Cool. Just make a PR and reference it here.
Probably the fix could be (not sure, I've not looked at it deeply):
} while (ao_load_acquire_async(&GC_stop_count) == my_stop_count
->
} while ((ao_load_acquire_async(&GC_stop_count) & ~(AO_t)THREAD_RESTARTED) == my_stop_count
Probably some other fix exist, let's see your version.
Hello, I have created a pull request. #631 I have tried to make my best and described this problem clearer. Can you take a look and make some comments?
Good, I will review it within a week. Could you please use your real name in commit (--author)? (Re-pushing the PR should also run the tests on Travis CI.)
Hi, I have changed user.name and re-open a PR #633 Although I have commited using my real name, Github displays my github account name anyway.
@ivmai anything I can do to move forward?
No-no, sorry about delay(( (I'm doing big refactoring of the code base these days, it takes me longer than planned, and I just don't want to switch to deeper review of your patch. I hope I will complete it within May.)
Hi, maintainer, I have encountered a problem of GC stuck in resend_lost_signals. Because it is a bit hard to reproduce this problem. I will describe my findings as following: