Both 5.5 and 5.6 work though, as if we are missing a set of dependencies? I am running out of time this evening, but I'd be good to see if anyone else reproduces my findings.
I saw the same thing yesterday, also when testing UCM changes :)
I was planning to bisect this myself today. But given your findings I'm not sure if that is useful. I would count the intermediate errors as also being bad I guess and try to find the first bad commit and see from there.
Can you send me your git bisect log ?
I didn't keep it but it pointed to v5.6-rc1 as the first bad commit - rather useless.
I'll try to spend more time on this later today.
The good thing is that it's not an isolated hardware problem, so we will likely reproduce this and converge to a solution faster. I only have 2 devices that work reliably.