Comment 44 for bug 1973167

@Robin Seidel, @Johannès Jahan,

that was the risk, that there might be other drivers in the old 4.15 kernel that also still request a synchronous module load from an asynchronous context. The general change in the module loader exposes any such driver. My hope was that intel-lpss would be the only one, but apparently that is not the case.

What you could do is run capture the output of "lsmod" in a working kernel (176) and then in this proposed kernel (189) and diff them. The 189 output should be missing some modules (or show them with fewer or 0 references). Those are the ones which failed to load due to the "wrong" way of doing things.

BTW, once the system is up, you can do "sudo modprobe <module name>" to load them. I have to do that in the 188 kernel (after waiting the 3 minutes) with "snd_hda_intel" to make the sound work.