On Tue, Apr 17, 2018 at 02:40:35PM +0100, Stephen Borrill wrote:
I've recently updated our XenServers with the latest security patches and
had to set pv-linear-pt=true to get NetBSD domUs to boot. For historical
reasons, I have a couple of netbsd-5/i386 VMs which are now frequently
panicking:
xpq_flush_queue: 3 entries
0x000000060481bfe0: 0x0000000604819001
0x00000007eac9fd40: 0x0000000604819001
0x00000007eab73d40: 0x0000000604819001
panic: HYPERVISOR_mmu_update failed
fatal breakpoint trap in supervisor mode
trap type 1 code 0 eip c039e374 cs 9 eflags 246 cr2 bb75b1a0 ilevel 6
Stopped in pid 15.1 (fsck_ffs) at 0xc039e374: popl %ebp
Kernel is pretty old and also has no symbols:
NetBSD 5.2_STABLE (XEN3PAE_DOMU) #37: Fri Nov 27 12:02:02 UTC 2015
Does anyone have an idea of a quick way to fix this (yes, I know I could use
a newer kernel with the old userland)? If I build a kernel with symbols,
would a backtrace help?
Not sure. What could help it to boot with xen-debug.gz and see what's
in 'xl demsg' when this happends.