On Sun, 3 Jul 2016, [email protected] wrote:
> >Synopsis: savecore complains on every boot about bad namelist
...
> >Description:
> With "savecore_flags=-z /var/crash" in /etc/rc.conf.local, on every
> boot, I see these two errors appear on the console:
> Jul 2 22:48:55 mail savecore: /var/crash: kvm_nlist: bad namelist
> Jul 2 22:48:55 mail savecore: /var/crash: _time_second not in namelist
> I see the identical behaviour on bare metal, this VM just happens to be
> the easiest place to run "sendbug" from.
This is a misunderstanding of use of the savecore_flags variable, combined
with an undocumented compatibility behavior. You do *not* need to include
the savecore directory in the savecore_flags value; /etc/rc *always*
supplies /var/crash as the last argument to savecore. If you change your
/etc/rc.conf.local to say
savecore_flags=-z
the error will go away.
The undocumented compat behavior is that if there's a positional argument
after the crash directory, it's used as the kernel namelist file, ala the
-N option. The effect is as if you were invoking
savecore -N /var/crash -z /var/crash
...which doesn't work, obviously.
I think we should get rid of that old compat behavior:
Index: savecore.c
===================================================================
RCS file: /data/src/openbsd/src/sbin/savecore/savecore.c,v
retrieving revision 1.55
diff -u -p -r1.55 savecore.c
--- savecore.c 18 Oct 2015 03:17:48 -0000 1.55
+++ savecore.c 4 Jul 2016 03:10:55 -0000
@@ -161,12 +161,10 @@ main(int argc, char *argv[])
argv += optind;
if (!clear) {
- if (argc != 1 && argc != 2)
+ if (argc != 1)
usage();
dirn = argv[0];
}
- if (argc == 2)
- kernel = argv[1];
(void)time(&now);
kmem_setup();
