Again.

Previously it was proposed to drop the `mbuf_mem_limit' and to do
calculations in run-time. I see no reason for that. The only `nmbclust'
user is the DIOCSETLIMIT case of pfioctl(), which only checks passed
fragments reference count against it and sets the hard limit on
corresponding pf(4) pool. Not on mbuf(9) pool. It also does not touch
the `mbuf_mem_limit' calculated within nmbclust_update(). Meanwhile the
kern_sysctl() thread only do pool_wakeup() on mbuf(9) and mbuf(9)
clusters pools, but does not touch the pf(4) pools. So I can say the
usage paths of `nmbclust' and `buf_mem_limit' are the different path and
we don't need to serialize the load of `nmbclust' and the
`mbuf_mem_limit' update. However, the update within the sysctl(2) path
should be serialized.

ok?

Index: sys/conf/param.c
===================================================================
RCS file: /cvs/src/sys/conf/param.c,v
diff -u -p -r1.52 param.c
--- sys/conf/param.c	20 Aug 2024 13:29:25 -0000	1.52
+++ sys/conf/param.c	5 Aug 2025 09:22:57 -0000
@@ -75,7 +75,7 @@ int	initialvnodes = NVNODE;
 int	maxprocess = NPROCESS;				/* [a] */
 int	maxthread = 2 * NPROCESS;			/* [a] */
 int	maxfiles = 5 * (NPROCESS + MAXUSERS) + 80;	/* [a] */
-long	nmbclust = NMBCLUSTERS;
+long	nmbclust = NMBCLUSTERS;				/* [a] */
 
 #ifndef BUFCACHEPERCENT
 #define BUFCACHEPERCENT	20
Index: sys/kern/kern_sysctl.c
===================================================================
RCS file: /cvs/src/sys/kern/kern_sysctl.c,v
diff -u -p -r1.481 kern_sysctl.c
--- sys/kern/kern_sysctl.c	24 Jul 2025 19:42:41 -0000	1.481
+++ sys/kern/kern_sysctl.c	5 Aug 2025 09:22:58 -0000
@@ -610,6 +610,20 @@ kern_sysctl(int *name, u_int namelen, vo
 		microboottime(&bt);
 		return (sysctl_rdstruct(oldp, oldlenp, newp, &bt, sizeof bt));
 	}
+	case KERN_MAXCLUSTERS: {
+		int oldval, newval;
+
+		oldval = newval = atomic_load_long(&nmbclust);
+		error = sysctl_int(oldp, oldlenp, newp, newlen, &newval);
+
+		if (error == 0 && oldval != newval) {
+			rw_enter_write(&sysctl_lock);
+			error = nmbclust_update(newval);
+			rw_exit_write(&sysctl_lock);
+		}
+
+		return (error);
+	}
 	case KERN_MBSTAT: {
 		uint64_t counters[mbs_ncounters];
 		struct mbstat mbs;
@@ -765,13 +779,6 @@ kern_sysctl_locked(int *name, u_int name
 		stackgap_random = stackgap;
 		return (0);
 	    }
-	case KERN_MAXCLUSTERS: {
-		int val = nmbclust;
-		error = sysctl_int(oldp, oldlenp, newp, newlen, &val);
-		if (error == 0 && val != nmbclust)
-			error = nmbclust_update(val);
-		return (error);
-	}
 	case KERN_CACHEPCT: {
 		u_int64_t dmapages;
 		int opct, pgs;
Index: sys/kern/uipc_mbuf.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_mbuf.c,v
diff -u -p -r1.301 uipc_mbuf.c
--- sys/kern/uipc_mbuf.c	17 Jul 2025 17:30:47 -0000	1.301
+++ sys/kern/uipc_mbuf.c	5 Aug 2025 09:22:58 -0000
@@ -217,8 +217,8 @@ nmbclust_update(long newval)
 	if (newval <= 0 || newval > LONG_MAX / MCLBYTES)
 		return ERANGE;
 	/* update the global mbuf memory limit */
-	nmbclust = newval;
-	atomic_store_long(&mbuf_mem_limit, nmbclust * MCLBYTES);
+	atomic_store_long(&nmbclust, newval);
+	atomic_store_long(&mbuf_mem_limit, newval * MCLBYTES);
 
 	pool_wakeup(&mbpool);
 	for (i = 0; i < nitems(mclsizes); i++)
Index: sys/net/pf_ioctl.c
===================================================================
RCS file: /cvs/src/sys/net/pf_ioctl.c,v
diff -u -p -r1.422 pf_ioctl.c
--- sys/net/pf_ioctl.c	7 Jul 2025 02:28:50 -0000	1.422
+++ sys/net/pf_ioctl.c	5 Aug 2025 09:22:58 -0000
@@ -2129,7 +2129,8 @@ pfioctl(dev_t dev, u_long cmd, caddr_t a
 			goto fail;
 		}
 		/* Fragments reference mbuf clusters. */
-		if (pl->index == PF_LIMIT_FRAGS && pl->limit > nmbclust) {
+		if (pl->index == PF_LIMIT_FRAGS &&
+		    pl->limit > atomic_load_long(&nmbclust)) {
 			error = EINVAL;
 			PF_UNLOCK();
 			goto fail;