Corruption in xenstored tdb file?

[Brian Buhrow]

	hello.  Following up on this thread, I'm still having a problem with starting domains
under FreebSD-12.1 as dom0 after ensuring that /var/lib/xenstoredb/tdb is deleted on startup.
The first domu starts just fine, an old NetBSD-5.2 domain.  the second one, however, a
NetBSD-current as of January 19 or so, however, starts fine but doesn't have a network
interface by the time it gets to multiuser mode.  The errors on the back end look like:

xnb(xnb_probe:1129): Claiming device 1, xnb
xnb(xnb_attach:1273): Attaching to backend/vif/10/0
xnb(xnb_frontend_changed:1397): frontend_state=Initialising, xnb_state=InitWait
xnb10.0: link state changed to DOWN
xnb10.0: link state changed to UP
xnb10.0: link state changed to DOWN
xnb10.0: promiscuous mode enabled
xnb10.0: link state changed to UP
nd6_dad_timer: cancel DAD on xnb10.0 because of ND6_IFF_IFDISABLED.
xnb(xnb_frontend_changed:1397): frontend_state=Initialised, xnb_state=InitWait
xnb1: Error 2 Unable to retrieve ring information from frontend /local/domain/10/device/vif/0.  Unable to connect.
xnb1: Fatal error. Transitioning to Closing State
xnb(xnb_frontend_changed:1397): frontend_state=Connected, xnb_state=Closing
xnb(xnb_connect_comms:793): rings connected!
xnb(xnb_frontend_changed:1397): frontend_state=Closed, xnb_state=Connected

In looking at the code, it looks like this is failing somewhere in xs_gather() in 
syskj/dev/xen/xenstore/xenstore.c

I thought it was some kind of race condition at first, because I could stop the domains that
didn't come up with a network interface, wait a bit, restar them and find they worked.  
Now, however, having upgraded to 12.1-P13, I find that I'm consistently getting this failure
regardless of how often I destroy and create the domain.

Any ideas on what might be going on?
-thanks
-Brian