#asahi-dev · 2024-08-07 11:56
chadmed_smc reports them as doing ~6000 rpm each#asahi-dev · 2024-08-07 11:56
chadmed_so i guess the macos "high power" mode might just do what it says on the tin - clamp the fans to ~3500-4000 rpm and just unclamp that if you enable it#asahi-dev · 2024-08-07 11:54
chadmed_perfect#asahi-dev · 2024-08-07 11:53
chadmed_looks like j314 just throttles if you clamp the fan speed#asahi-dev · 2024-08-07 05:58
chadmedmacs are expensive! #asahi-dev · 2024-08-07 05:58
chadmedokay im not _that_ game :p#asahi-dev · 2024-08-07 05:58
svenlol#asahi-dev · 2024-08-07 05:58
nicolas17sven: I think I did that on my desktop PC once, trying to make RAM errors more reproducible (it didn't work)#asahi-dev · 2024-08-07 05:57
svenfor extra credit use a hair dryer to really heat it up :>#asahi-dev · 2024-08-07 05:57
nicolas17(unless you live in arizona)#asahi-dev · 2024-08-07 05:56
chadmedi might dust off the j314 and test it#asahi-dev · 2024-08-07 05:56
chadmedim hoping thats what happens #asahi-dev · 2024-08-07 05:56
nicolas17wouldn't turning the fans off just "turn it into a MacBook Air"? CPU throttling should keep it at safe temps long before it reaches emergency-shutdown levels?#asahi-dev · 2024-08-07 05:54
chadmed(now that we bounds-check the values anyway hehe i almost broke my studio yesterday) #asahi-dev · 2024-08-07 05:53
chadmedyeah im almost 100% sure it would fail safe#asahi-dev · 2024-08-07 05:53
svenI would be surprised if there’s no thermal watchdog running somewhere which will just power off the machine if it gets too hot#asahi-dev · 2024-08-07 05:51
nicolas17wonder what this does, at a low level#asahi-dev · 2024-08-07 05:51
nicolas17https://support.apple.com/en-us/101613#asahi-dev · 2024-08-07 05:49
chadmedsince its unsafe it taints the kernel and we can check for that#asahi-dev · 2024-08-07 05:49
chadmedim happy to keep manual control gated behind the parameter forever fwiw#asahi-dev · 2024-08-07 05:48
chadmedyeah but thats a cop-out :p#asahi-dev · 2024-08-07 05:48
chadmedlike realistically no static manual fan speed is fully safe except the max#asahi-dev · 2024-08-07 05:48
jannauwe do right now by offering no fan control#asahi-dev · 2024-08-07 05:47
chadmedjannau: i know but we cant really outright deny that #asahi-dev · 2024-08-07 05:45
jannauchadmed_: allowing manual control to set the fab speed to minimal speed isn't necessarily a stable configuration either#asahi-dev · 2024-08-06 22:51
chadmed__pushed#asahi-dev · 2024-08-06 22:32
chadmed__oh nvm skill issue#asahi-dev · 2024-08-06 22:29
chadmed__of course this is _easily_ avoidable with an extra arm off the big if/else block but thats ugly :p#asahi-dev · 2024-08-06 22:27
chadmed__im mainly trying to avoid unnecessary writes to the SMC by having this arming mechanism since theyre slow. like if we unconditionally allow 0 writes then we are always re-arming automatic control through the SMC which is slow#asahi-dev · 2024-08-06 22:27
nicolas17but I have no idea how that meshes with existing userspace expectations#asahi-dev · 2024-08-06 22:26
chadmed__ill rejig that later today and see where we end up#asahi-dev · 2024-08-06 22:26
chadmed__that does make sense#asahi-dev · 2024-08-06 22:26
nicolas17I was just thinking if you want 0 to mean automatic, then setting it to 0 when it's already automatic should just do nothing#asahi-dev · 2024-08-06 22:26
chadmed__im thinking that if turning the fans off proves safe we can just allow a 0 write unconditionally#asahi-dev · 2024-08-06 22:25
nicolas17yeah that's going way deeper than what I know#asahi-dev · 2024-08-06 22:25
chadmed__theres just HWMON_F_TARGET which we already have wired up#asahi-dev · 2024-08-06 22:25
chadmed__maybe im dense but i cant find a hwmon attr that flags "this fan can be toggled between auto/manual control" that we can use instead of doing it this way#asahi-dev · 2024-08-06 22:24
nicolas17hrm#asahi-dev · 2024-08-06 22:23
chadmed__unfortunately theres no really "good" way to express what we need to :/ #asahi-dev · 2024-08-06 22:23
chadmed__i also dont want userspace tools thinking they did the right thing#asahi-dev · 2024-08-06 22:23
chadmed__nicolas17: the reason i want it to return -EINVAL is so that people don't get the impression that it should turn the fan off or do something special when manual mode is not armed#asahi-dev · 2024-08-06 16:46
nicolas17chadmed_: sounds like if it's in manual mode, writing 0/nothing once will take it out of manual mode, and doing it again will return -EINVAL, to me it seems better if that was idempotent...#asahi-dev · 2024-08-06 16:45
nicolas17<chadmed_> trying to write 0/nothing outside of manual mode will also return -EINVAL #asahi-dev · 2024-08-06 16:43
hermitfeatherOn a M1 Max MBP.#asahi-dev · 2024-08-06 16:42
hermitfeatherHey there, I'm new to linux development but wanted to take a shot at shaving the first yak. Before starting I wanted to get some clarification. On gnome, my current running sink is called audio_effect.j316-convolver. What it needs to be by default is alsa_output.platform-sound.HiFi__hw_AppleJ316_0__sink with the other hidden. Is this correct?#asahi-dev · 2024-08-06 15:43
svenif it was only phy bindings I wouldn’t mind, but we’re still at the bikeshedding for basic stuff like platform_device so I think I’ll just stick to C#asahi-dev · 2024-08-06 15:41
svenbut it’s very minimal so far unfortunately #asahi-dev · 2024-08-06 15:41
svenright, I thought there was more upstream already #asahi-dev · 2024-08-06 15:35
kettenisright, make upstreaming even harder since now you have rust bindings to bikeshed about ;)#asahi-dev · 2024-08-06 15:27
svenmight be useful as a template for phy/ bindings though