From: bluhm <bluhm@openbsd.org>
Date: Mon, 3 Jul 2017 23:05:21 +0000 (+0000)
Subject: Do not grab the socket lock in doaccept() twice.  Pass NOTE_SUBMIT
X-Git-Url: http://artulab.com/gitweb/?a=commitdiff_plain;h=879c51079d0c11d6b8c1e4847af11e498717740d;p=openbsd

Do not grab the socket lock in doaccept() twice.  Pass NOTE_SUBMIT
to KNOTE() as we are already holding the lock.  Fixes "panic:
rw_enter: netlock locking against myself" reported by Gregor Best
and reproduced with src/regress/lib/libtls/gotls.
OK millert@
---

diff --git a/sys/kern/uipc_syscalls.c b/sys/kern/uipc_syscalls.c
index acf671a65f3..e346c161d90 100644
--- a/sys/kern/uipc_syscalls.c
+++ b/sys/kern/uipc_syscalls.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: uipc_syscalls.c,v 1.151 2017/03/27 11:45:49 bluhm Exp $	*/
+/*	$OpenBSD: uipc_syscalls.c,v 1.152 2017/07/03 23:05:21 bluhm Exp $	*/
 /*	$NetBSD: uipc_syscalls.c,v 1.19 1996/02/09 19:00:48 christos Exp $	*/
 
 /*
@@ -327,7 +327,7 @@ doaccept(struct proc *p, int sock, struct sockaddr *name, socklen_t *anamelen,
 	    : (flags & SOCK_NONBLOCK ? FNONBLOCK : 0);
 
 	/* connection has been removed from the listen queue */
-	KNOTE(&head->so_rcv.sb_sel.si_note, 0);
+	KNOTE(&head->so_rcv.sb_sel.si_note, NOTE_SUBMIT);
 
 	fp->f_type = DTYPE_SOCKET;
 	fp->f_flag = FREAD | FWRITE | nflag;