Pavel Machek wrote: > Find out what is wrong with the patch, fix it, resubmit it. I would like to, but unfortunately, I've no skills in programming so I'm not able to do that. > Oh. It appears to register trigger per block device. That's no-no. Hmm, why is that a no-no? It seems it works that way for CPUs already. I can see triggers from cpu0 to cpu39 on our servers with two CPUs, 10 cores and 2 threads per core. So we even have one trigger for every cpu thread. I've never seen a host with 40 nvme disks, so why is this problematic if having 40 cpu triggers is not? Or does "trigger per block device" mean sth. else? Michael
