Mark setrtable(2) as NOLOCK

[Martin Pieuchot]

Now that suser() is no longer messing with a per-process field, we
can directly turn setrtable(2) as NOLOCK.

Apart from sanity checks this syscall writes an int-sized per-process
field.  Is a memory barrier enough?

ok?

Index: kern/syscalls.master
===================================================================
RCS file: /cvs/src/sys/kern/syscalls.master,v
retrieving revision 1.180
diff -u -p -r1.180 syscalls.master
--- kern/syscalls.master        12 Dec 2017 01:12:34 -0000      1.180
+++ kern/syscalls.master        19 Feb 2018 15:05:34 -0000
@@ -532,7 +532,7 @@
 307    OBSOL           statfs53
 308    OBSOL           fstatfs53
 309    OBSOL           fhstatfs53
-310    STD             { int sys_setrtable(int rtableid); }
+310    STD NOLOCK      { int sys_setrtable(int rtableid); }
 311    STD NOLOCK      { int sys_getrtable(void); }
 312    OBSOL           t32_getdirentries
 313    STD             { int sys_faccessat(int fd, const char *path, \
Index: kern/uipc_syscalls.c
===================================================================
RCS file: /cvs/src/sys/kern/uipc_syscalls.c,v
retrieving revision 1.165
diff -u -p -r1.165 uipc_syscalls.c
--- kern/uipc_syscalls.c        19 Feb 2018 08:59:52 -0000      1.165
+++ kern/uipc_syscalls.c        19 Feb 2018 15:19:35 -0000
@@ -1189,6 +1190,9 @@ sys_setrtable(struct proc *p, void *v, r
                return (EINVAL);
 
        p->p_p->ps_rtableid = (u_int)rtableid;
+       /* Force visibility */
+       membar_producer();
+
        return (0);
 }