Point out that spurious wake-ups are possible, and that an LWP should

check for resource availability after being awoken.
This commit is contained in:
pgoyette 2019-12-12 02:34:55 +00:00
parent 8a031a1d1e
commit 08f76bd663
1 changed files with 9 additions and 2 deletions

View File

@ -1,4 +1,4 @@
.\" $NetBSD: condvar.9,v 1.20 2019/04/12 11:02:22 abhinav Exp $
.\" $NetBSD: condvar.9,v 1.21 2019/12/12 02:34:55 pgoyette Exp $
.\"
.\" Copyright (c) 2006, 2007, 2008 The NetBSD Foundation, Inc.
.\" All rights reserved.
@ -27,7 +27,7 @@
.\" ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
.\" POSSIBILITY OF SUCH DAMAGE.
.\"
.Dd November 12, 2017
.Dd December 12, 2019
.Dt CONDVAR 9
.Os
.Sh NAME
@ -144,6 +144,13 @@ as available until the calling LWP has begun to wait for it.
.Pp
Non-interruptable waits have the potential to deadlock the system, and so must
be kept short (typically, under one second).
.Pp
Upon being awakened, the calling LWP should verify the availability
of the resource (or other condition).
It should not blindly assume that the resource is now available.
If the resource is still not available, the calling LWP may call
.Fn cv_wait
again to continue waiting.
.It Fn cv_wait_sig "cv" "mtx"
.Pp
As per