#asahi-dev · 2022-02-15 18:49
axboesven: oh, reminds me, didn't find any time to fiddle with shared tags yet#asahi-dev · 2022-02-15 18:45
axboewith a big backlog like that, useful to just try and flush out some of the independent ones. at least that makes you feel like you made progress ;)#asahi-dev · 2022-02-15 18:39
axboej`ey: I know! but can always get done ;)#asahi-dev · 2022-02-15 18:33
axboeI know you guys all know this, but I'm still blown away by the perf here. game changing imho#asahi-dev · 2022-02-15 18:33
axboehttps://twitter.com/axboe/status/1493646629953589248#asahi-dev · 2022-02-15 18:33
axboeI'm all for flushing out patches, hate carrying them#asahi-dev · 2022-02-15 18:22
axboeshipped#asahi-dev · 2022-02-15 18:20
axboethat your preferred email?#asahi-dev · 2022-02-15 18:20
axboeSven Peter <sven@svenpeter.dev>#asahi-dev · 2022-02-15 18:19
axboecredit where credit is due#asahi-dev · 2022-02-15 18:19
axboesven: do you want a suggested-by in there?#asahi-dev · 2022-02-15 18:18
axboesven: adding an actual changelog :)#asahi-dev · 2022-02-15 18:17
axboeok, will do so right now#asahi-dev · 2022-02-15 18:15
axboesven: either way is fine with me, just let me know what you prefer#asahi-dev · 2022-02-15 18:00
axboejannau: see the debug output I badly typed#asahi-dev · 2022-02-15 17:59
axboejannau: like you asked :)#asahi-dev · 2022-02-15 17:59
axboejannau: for your patch I used unmodified#asahi-dev · 2022-02-15 17:51
axboeso might work with stock dtb#asahi-dev · 2022-02-15 17:51
axboewhich the hack works around#asahi-dev · 2022-02-15 17:51
axboethe sparseness ends up being an issue for the spin_table#asahi-dev · 2022-02-15 17:50
axboehaven't tried it with un-edited dtsi#asahi-dev · 2022-02-15 17:50
axboefor reference, this is the m1n1 hack#asahi-dev · 2022-02-15 17:50
axboehttps://pastebin.com/cT08t1PT#asahi-dev · 2022-02-15 17:49
axboejannau: yep#asahi-dev · 2022-02-15 17:39
axboejannau: I tried your patch with stock dtsi#asahi-dev · 2022-02-15 17:38
axboejannau: so e(0 1) p(0 1 2) p(0 1 2)#asahi-dev · 2022-02-15 17:38
axboejannau: this is with the dtsi trimmed, didn't try without for that hack#asahi-dev · 2022-02-15 17:33
axboefixes capacity, and cpufreq then works on 2nd pcluster too#asahi-dev · 2022-02-15 17:32
axboeif I just hack smp.c to keep a separate cpu index rather than use i it works for me, on top of plain m1n1#asahi-dev · 2022-02-15 17:20
axboeDT CPU 1 MPIDR mishatc; 0x1 ~= 0x10100#asahi-dev · 2022-02-15 17:20
axboejannau: gives me a MPIDR mismatch#asahi-dev · 2022-02-15 17:14
axboejannau: ok, will try#asahi-dev · 2022-02-15 17:12
axboejannau: with or without editing the dtsi to remove the non-existent cores?#asahi-dev · 2022-02-15 17:12
axboejannau: I can give it a whirl#asahi-dev · 2022-02-15 16:43
axboenow it finds last cpu in p cluster 1 "not alive", and then hits a MPIDR mismatch on first cpu in cluster 2#asahi-dev · 2022-02-15 16:42
axboeI think m1n1 needs a fixup for that#asahi-dev · 2022-02-15 16:42
axboebooting now with those 2 removed#asahi-dev · 2022-02-15 16:39
axboelet me try on the new base, just in case I messed it up#asahi-dev · 2022-02-15 16:38
axboedid try that yesterday but didn't make a difference#asahi-dev · 2022-02-15 16:35
axboe456 here for E cluster, 1024 for first P cluster, 0 for second P cluster#asahi-dev · 2022-02-15 16:34
axboejannau: hmm#asahi-dev · 2022-02-15 16:32
axboejannau: curious on the m1 max, do you get the cpu capacity set for both p clusters?#asahi-dev · 2022-02-15 16:29
axboein the devicetree in sysfs, capacity looks correct for all CPUs#asahi-dev · 2022-02-15 16:23
axboegiven that the other clusters have it set right#asahi-dev · 2022-02-15 16:23
axboeso might be as simple as schedutil being borken with capacity == 0#asahi-dev · 2022-02-15 16:23
axboenew one does seem to set the right clock if I just force performance as the governor#asahi-dev · 2022-02-15 16:22
axboeand the fact that capacity isn't set either just made me assume that something is wrong with the setup of that second pcluster#asahi-dev · 2022-02-15 16:21
axboewhich again made little sense, I saw all three get registered#asahi-dev · 2022-02-15 16:21
axboeI poked a bit yesterday, and seems like it was using the wrong clk base for pcluster2, it was the pcluster1 address#asahi-dev · 2022-02-15 16:15
axboeI feel like this is something silly, but it eluded me yesterday