Date: Wed, 09 Oct 2013 22:15:58 +0000 From: Craig Butler <craig001@lerwick.hopto.org> To: Marius Strobl <marius@alchemy.franken.de> Cc: freebsd-sparc64@freebsd.org Subject: Re: 9.2-RELEASE broken on a netra X1 box. Message-ID: <1381356958.95378.18.camel@zbox.lerwick.hopto.org> In-Reply-To: <1381353370.95378.8.camel@zbox.lerwick.hopto.org> References: <1381318325.93904.3.camel@zbox.lerwick.hopto.org> <20131009192719.GA14057@alchemy.franken.de> <1381353370.95378.8.camel@zbox.lerwick.hopto.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, 2013-10-09 at 21:16 +0000, Craig Butler wrote: > On Wed, 2013-10-09 at 21:27 +0200, Marius Strobl wrote: > > On Wed, Oct 09, 2013 at 11:32:05AM +0000, Craig Butler wrote: > > > Hello Folks > > > > > > Trying out 9.2 RELEASE on one of our Netra X1 servers and something > > > broke. > > > > > > Freshly compiled and installed world and kernel. > > > > > > Booting [/boot/kernel/kernel]... > > > jumping to kernel entry at 0xc0098000. > > > Copyright (c) 1992-2013 The FreeBSD Project. > > > Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994 > > > The Regents of the University of California. All rights > > > reserved. > > > FreeBSD is a registered trademark of The FreeBSD Foundation. > > > FreeBSD 9.2-RELEASE #0: Tue Oct 8 16:09:50 BST 2013 > > > root@blueeye.lerwick.hopto.org:/usr/obj/usr/src/sys/GENERIC sparc64 > > > gcc version 4.2.1 20070831 patched [FreeBSD] > > > real memory = 1073741824 (1024 MB) > > > panic: kmem_suballoc: bad status return of 3 > > > cpuid = 0 > > > KDB: stack backtrace: > > > #0 0xc07d855c at kmem_suballoc+0x7c > > > #1 0xc07d79ac at vm_ksubmap_init+0x12c > > > #2 0xc0846f24 at cpu_startup+0xc4 > > > #3 0xc04b3e70 at mi_startup+0x1d0 > > > #4 0xc0098028 at btext+0x28 > > > Uptime: 1s > > > > > > > Typically, this is caused by a bug in MI code which only triggers with > > specific memory layouts depending on the exact machine model and certain > > amounts of RAM, though. That's why it would be helpful if people would at > > least test booting BETAs and RCs. It probably can be worked around by > > limiting the memory employed via the hw.physmem tunable to something like > > 512 MB for that machine. I need to see whether I can upgrade my X1 to 1 GB > > so I can reproduce it. > > > > Marius > > > Hi Marius > > Thank you for the reply. I have been doing a little digging and it > maybe related to ZFS and some of the loader.conf tunables as you have > suggested. > > It is however a change because various X1 servers have been rock stable > on 8.4 and 9.1 with gmirror /boot and zfs root with the same > configuration. > > loader.conf currently looks like; > craig@blueeye:~> cat /boot/loader.conf > wlan_wep_load="YES" > wlan_tkip_load="YES" > wlan_ccmp_load="YES" > vm.kmem_size="512M" > vm.kmem_size_max="512M" > geom_mirror_load="YES" > zfs_load="YES" > vfs.root.mountfrom="zfs:tank" > > When I get a few cycles I intend to change one of our dev Netra X1's to > ufs only (or disable the tunables) and 9.2-RELEASE to see if it'll boot > further. > > Will keep you informed. > > Kind Regards > > Craig Butler > > _______________________________________________ > freebsd-sparc64@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-sparc64 > To unsubscribe, send any mail to "freebsd-sparc64-unsubscribe@freebsd.org" Hi Again It boots if I hash out the vm.kmem_size and vm.kmem_size_max. Not sure how that's going to effect zfs stability (IIRC I added those a while back so that zfs would work with less than 4GB reasonably) I'll keep the system up a couple of days and see if any further issues appear. I wonder what changed between 9.1-RELEASE and 9.2-RELEASE to cause kmem_size to break things. Kind Regards Craig Butler.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1381356958.95378.18.camel>