On 9/28/23 2:01 AM, Peter Zijlstra wrote: > All of this is quite insane, but the above is actually broken. You're > using wall-time to measure runtime of a preemptible thread. Would have to agree with that... wall-time specifics aside, this whole thing seems to attempt to solve an issue quite the wrong way around. > Now, I see what you're trying to do, but who actually uses this data? This is the important question - and if you need this data, then why not just account it in sqpoll itself and have some way to export it? Let's please not implicate the core kernel bits for that. -- Jens Axboe
