#
✕ nick:marcan
#asahi · 2021-10-03 05:10
marcanthat requires the other cores to be in deep ("down") sleep, which requires a cpuidle driver since wfi in that mode is non-state-preserving
#asahi · 2021-10-03 05:09
marcanit can only boost to 3.2GHz on a single core at once
#asahi · 2021-10-03 05:09
marcankdrag0n[m]: that is not a problem with the cpufreq driver, it's a problem with the lack of a cpuidle driver
#asahi · 2021-10-03 01:04
marcanthat one might be something to move to m1n1 tbh
#asahi · 2021-10-03 01:03
marcan(I don't get any of this because it's unnecessary on 12.0...)
#asahi · 2021-10-03 01:02
marcando you get "Initializing cluster (DVMR: %d)" lines on boot?
#asahi · 2021-10-03 01:00
marcanif that still works then it's definitely the DVMR thing
#asahi · 2021-10-03 01:00
marcanactually, can you try with cpu_pstates but commenting out set_pstate(1, 15)?
#asahi · 2021-10-03 00:59
marcanI wonder if I botched the DVMR thing
#asahi · 2021-10-03 00:59
marcanalyssa: can you benchmark an e-core and a p-core (maybe all of them) with/without cpu_pstates?
#asahi · 2021-10-03 00:58
marcanbut the point is this shouldn't be necessary
#asahi · 2021-10-03 00:58
marcannot really, pretty sure there's failsafes for that
#asahi · 2021-10-03 00:56
marcanif you have a big difference in userspace perf then something is weird
#asahi · 2021-10-03 00:56
marcanalyssa: userspace should boot equally fast, and the kernel *shouldn't* spend too much time before it does the pstates thing? it's like <1 second for me
#asahi · 2021-10-03 00:52
marcanthey should be at least clock gated while waiting
#asahi · 2021-10-03 00:52
marcanthere is a wfe in the "spin"loop
#asahi · 2021-10-03 00:52
marcankettenis: the cores don't spin
#asahi · 2021-10-02 16:17
marcanbut I hear it works well on a desk
#asahi · 2021-10-02 16:17
marcanonly problem is it might cook your lap ;)
#asahi · 2021-10-02 16:17
marcanyeah, it's actually easy, you remove a thermal insulator and replace it with a thermal pad
#asahi · 2021-10-02 15:36
marcan(and left behind bad state)
#asahi · 2021-10-02 15:36
marcanthis looks like I or someone else botched something with kobjects/sysfs, or possibly that something else exploded first and the failure path is broken
#asahi · 2021-10-02 15:36
marcanI suspect there's probably a warning or something useful before the big explosion
#asahi · 2021-10-02 15:35
marcanalyssa: yeah you're going to have to run the HV, I think the important bits scrolled off the screen
#asahi · 2021-10-02 15:33
marcanI'm going to sleep soon :p
#asahi · 2021-10-02 15:33
marcanok
#asahi · 2021-10-02 15:33
marcanthat makes pasting the backtrace harder :p
#asahi · 2021-10-02 15:33
marcanok
#asahi · 2021-10-02 15:32
marcan*bare
#asahi · 2021-10-02 15:32
marcanah, you're running this base metal for now?
#asahi · 2021-10-02 15:32
marcanbtw, you know about the "lower" thing in the hv, right?
#asahi · 2021-10-02 15:32
marcanI mean backtrace
#asahi · 2021-10-02 15:32
marcanpaste?
#asahi · 2021-10-02 15:32
marcanheh
#asahi · 2021-10-02 15:31
marcansame error?
#asahi · 2021-10-02 15:28
marcansorry, forgot about that one
#asahi · 2021-10-02 15:28
marcanalyssa: pushed m1n1 update
#asahi · 2021-10-02 15:27
marcanno, that's me
#asahi · 2021-10-02 15:27
marcanoh wait hold on
#asahi · 2021-10-02 15:27
marcanyeah that looks like a bad merge of some sort
#asahi · 2021-10-02 15:21
marcandid you get any conflicts?
#asahi · 2021-10-02 15:21
marcanjust force pushed with the obsolete change removed, but you don't have to care
#asahi · 2021-10-02 15:20
marcanjust apply it if you want to try it functionally
#asahi · 2021-10-02 15:20
marcanyeah, exactly
#asahi · 2021-10-02 15:20
marcan(it was for a prior attempt)
#asahi · 2021-10-02 15:19
marcanjust an API variant I was missing, and actually some of it is obsolete and can be removed I think
#asahi · 2021-10-02 15:17
marcanheh yeah
#asahi · 2021-10-02 15:17
marcanlol
#asahi · 2021-10-02 15:16
marcanand adds some devices at the top of the devices part
#asahi · 2021-10-02 15:16
marcanit changes cpus which haven't changed since forever

plus ›