Date: Tue, 11 Mar 2003 09:26:03 -0800 (PST) From: John Baldwin <jhb@FreeBSD.org> To: Perforce Change Reviews <perforce@freebsd.org> Subject: PERFORCE change 26687 for review Message-ID: <200303111726.h2BHQ309059526@repoman.freebsd.org>
next in thread | raw e-mail | index | archive | help
http://perforce.freebsd.org/chv.cgi?CH=26687 Change 26687 by jhb@jhb_laptop on 2003/03/11 09:25:58 Bah, I think my problem with kldunload was a race condition where the next thread destory event was posted before thread 1 had a chance to reacquire the event_mtx lock. Instead, use a cv to close this race and similar races when sending events via the sysctl. In evtest_postevent() we wait on this cv after we send an event and in event_thread() we signal this cv after setting event back to 0. In thread_destroy() we wait on this cv only if the event is not zero after we resume from msleep(). Affected files ... .. //depot/projects/smpng/sys/modules/evtest/evtest.c#11 edit Differences ... ==== //depot/projects/smpng/sys/modules/evtest/evtest.c#11 (text+ko) ==== @@ -58,7 +58,7 @@ #define EVENT_TYPE_BROADCAST 0x1 static eventhandler_tag first_tag; -static struct cv event_cv, broadcast_cv; +static struct cv event_cv, broadcast_cv, event_recvd; static struct mtx event_mtx; static struct sema evtest_sema; static int event, broadcast_count; @@ -137,6 +137,7 @@ event = new_event; cv_signal(&event_cv); } + cv_wait(&event_recvd, &event_mtx); mtx_unlock(&event_mtx); return (0); } @@ -245,6 +246,7 @@ } } event = 0; + cv_signal(&event_recvd); while ((ev = event) == 0) cv_wait(&event_cv, &event_mtx); mtx_unlock(&event_mtx); @@ -351,6 +353,8 @@ cv_broadcast(&event_cv); msleep(threads[i].ti_p, &event_mtx, PWAIT, "evtstun", 0); threads[i].ti_p = NULL; + if (event != 0) + cv_wait(&event_recvd, &event_mtx); } static void To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe p4-projects" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200303111726.h2BHQ309059526>