There's a bug that exposed this lack of testing: https://bugzilla.yoctoproject.org/show_bug.cgi?id=16132 There are sstate reuse tests that cover other scenarios, so they could perhaps be extended. We can't cover reuse between all combinations of target architectures, but at least one or two pairs should be covered.
We have existing tests in oe-selftest in sstatetests but they don't compare this combination - natives should never change hashes regardless of the target machine
Bulk move of unassigned 6.0-M* bugs to 6.1-M2.