It looks like a really long backtrace, and there is something about stack corruption, so I wonder if the stack is actually overflowing here. Can you see if the same thing happens with Ard's vmap-stack branch from [1] or if that shows a different output?
[1] https://git.kernel.org/pub/scm/linux/kernel/git/ardb/linux.git/log/?h=arm-vm...
I do not see any crash on this tree and arm-vmap-stacks branch.
- Naresh