[src/netbsd-8]: src/sys/net Pull up following revision(s) (requested by riast...

[martin <martin%NetBSD.org@localhost>]

details:   https://anonhg.NetBSD.org/src/rev/7685e84613fe
branches:  netbsd-8
changeset: 373635:7685e84613fe
user:      martin <martin%NetBSD.org@localhost>
date:      Wed Feb 22 18:55:06 2023 +0000

description:
Pull up following revision(s) (requested by riastradh in ticket #1801):

        sys/net/route.c: revision 1.236

route(4): Work around deadlock in rt_free wait path.
PR kern/56844

diffstat:

 sys/net/route.c |  15 ++++++++++++---
 1 files changed, 12 insertions(+), 3 deletions(-)

diffs (37 lines):

diff -r f13348edb638 -r 7685e84613fe sys/net/route.c
--- a/sys/net/route.c   Wed Feb 22 18:50:41 2023 +0000
+++ b/sys/net/route.c   Wed Feb 22 18:55:06 2023 +0000
@@ -1,4 +1,4 @@
-/*     $NetBSD: route.c,v 1.194.6.15 2019/10/04 11:26:35 martin Exp $  */
+/*     $NetBSD: route.c,v 1.194.6.16 2023/02/22 18:55:06 martin Exp $  */
 
 /*-
  * Copyright (c) 1998, 2008 The NetBSD Foundation, Inc.
@@ -97,7 +97,7 @@
 #endif
 
 #include <sys/cdefs.h>
-__KERNEL_RCSID(0, "$NetBSD: route.c,v 1.194.6.15 2019/10/04 11:26:35 martin Exp $");
+__KERNEL_RCSID(0, "$NetBSD: route.c,v 1.194.6.16 2023/02/22 18:55:06 martin Exp $");
 
 #include <sys/param.h>
 #ifdef RTFLUSH_DEBUG
@@ -644,8 +644,17 @@
 rt_wait_ok(void)
 {
 
+       /*
+        * This originally returned !cpu_softintr_p(), but that doesn't
+        * work: the caller may hold a lock (probably softnet lock)
+        * that a softint is waiting for, in which case waiting here
+        * would cause a deadlock.  See https://gnats.netbsd.org/56844
+        * for details.  For now, until the locking paths are sorted
+        * out, we just disable the waiting option altogether and
+        * always defer to workqueue.
+        */
        KASSERT(!cpu_intr_p());
-       return !cpu_softintr_p();
+       return false;
 }
 
 void