spectre_v2=off seems to fix the problem, at least I could boot our main Dell 540 with it that paniced consequently before. Before I came here I've opened a SR with SuSE support (no. 00355361) which contains a dmesg output from a HP AIO that I could collect before it crashed, as well as a screenlog from the full boot process of the Dell 540. I won't post them here in case they contain sensitive information, but you should be able to access them internally I guess.