Even though I speak no assembler, I guess that register x2 being 0x0 and the offset of "308" matching the offset of type in struct request, the rq->mq_hctx->type is the culprit, mq_hctx being NULL and stored in x2. Just guessing, though ;-) I rebooted with Kernel:HEAD 5.18.0 and am trying to reproduce again. I have all the previous debuginfo / debugsource still installed so I can investigate older crashdumps if needed.