#asahi-dev · 2026-07-05 11:32
chadmedif m1n1 just kills it and we dont care about it in the kernel we can add the t8103 compat and be done with it then #asahi-dev · 2026-07-05 11:32
yukanot that I know of#asahi-dev · 2026-07-05 11:32
chaos_princessdo we need the secondary for anything?#asahi-dev · 2026-07-05 11:32
chadmedyeah, point being do we want to actually use the secondary or just kill it and leave it alone #asahi-dev · 2026-07-05 11:31
yukaand then linux re-enabled it if it had the driver#asahi-dev · 2026-07-05 11:31
yukam1n1 has always disabled the primary watchdog#asahi-dev · 2026-07-05 11:31
chadmedforgot about that whole thing :P #asahi-dev · 2026-07-05 11:31
chadmedyeah no generic #asahi-dev · 2026-07-05 11:31
yukaSo this would be "apple,t8132-wdt", "apple,t8103-wdt"; either way?#asahi-dev · 2026-07-05 11:30
yukat8122 has "apple,t8122-wdt", "apple,t8103-wdt"#asahi-dev · 2026-07-05 11:30
yukaah nevermind#asahi-dev · 2026-07-05 11:30
yukaI thought we were not adding the generic "apple,wdt" anymore#asahi-dev · 2026-07-05 11:29
chadmedimo it should get one, and then it should just be compatible = "apple,t8132-wdt", "apple,t8103-wdt", "apple,wdt"; or minus the t8103 compat if we want linux to deal with the secondary watchdog rather than just killing it from m1n1 #asahi-dev · 2026-07-05 11:25
yukashould the M4 watchdog get a new compatible? iirc once the secondary watchdog is disabled, it can be used like the previous watchdogs#asahi-dev · 2026-07-05 11:16
sofusops#asahi-dev · 2026-07-05 07:14
LarstiQah, had not read all of all backlog yet#asahi-dev · 2026-07-05 07:14
LarstiQdifference between iboot versions?#asahi-dev · 2026-07-05 06:54
sofuschadmed: yeah just annoying#asahi-dev · 2026-07-05 06:50
sofusbut still the vp9 driver is pretty lean compared to the others i looked at#asahi-dev · 2026-07-05 06:50
chadmedcould be a lot worse #asahi-dev · 2026-07-05 06:48
sofuslooks like this https://pastebin.com/8jjPB5bb#asahi-dev · 2026-07-05 06:47
sofusv4l2 wants pointers to all the symbol counts, but it expects a certain memory layout that avd does not use, so after every frame we also need to manually copy counts :(#asahi-dev · 2026-07-05 06:44
sofusso the first link is setting up pointers to all the coeff/eob symbols in the symbol counts struct#asahi-dev · 2026-07-05 06:43
sofuseiln did that in software, but avd actually exposes that also#asahi-dev · 2026-07-05 06:43
sofusin vp9 there are the prob blobs, but also another larger struct that counts how many times each value in the prob blob is used#asahi-dev · 2026-07-05 06:42
chadmedO(maybe)#asahi-dev · 2026-07-05 06:42
chadmedoh thats revolting#asahi-dev · 2026-07-05 06:39
chaos_princessno, i was thinking specifically of https://github.com/eiln/avd/blob/main/avid/vp9/probs.py#L72#asahi-dev · 2026-07-05 06:37
sofusthats actually another part, you are thinking of https://pastebin.com/HRgC5jJx#asahi-dev · 2026-07-05 06:36
chaos_princessheh, that part#asahi-dev · 2026-07-05 06:36
sofusalso vp9 is gonna be super ugly https://pastebin.com/iXPh5uMi#asahi-dev · 2026-07-05 06:35
sofusso we might need to make the cm3 do more work#asahi-dev · 2026-07-05 06:34
sofusim having timing related issues with stuff apple does not seem to check#asahi-dev · 2026-07-05 06:32
chadmednice! #asahi-dev · 2026-07-05 06:32
sofusi think the fast ap cores are giving me problems#asahi-dev · 2026-07-05 06:31
sofusi went from 0 tests passing to ~150 in two days :p#asahi-dev · 2026-07-05 06:31
sofusimplementing vp9 is going faster than i thought it would#asahi-dev · 2026-07-05 01:47
integralpilotenverbalalic: have you set up the other end of the dockchannel correctly? iirc if you havent it might lead to that, but not 100% sure. could be another issue with dc-uart (i did make it very rushed as my focus was HID)#asahi-dev · 2026-07-05 01:46
integralpilotyuka: yay! perfect, thank you for fixing that. such a silly typo from me lol, sorry for making it. will integrate into v2 :)#asahi-dev · 2026-07-05 00:30
StanfordAppliedCyber[m]depending on how messed up the mac neo is i may even try to daily drive that system#asahi-dev · 2026-07-05 00:29
StanfordAppliedCyber[m]i'll be at defcon#asahi-dev · 2026-07-05 00:16
nicolas17any asahi devs attending Akademy? (KDE conference, Graz, Austria)#asahi-dev · 2026-07-04 22:45
enverbalalicyuka: which issue of uart does writel fix ? if it's the hanging after 1 message part, that's still present on 6041#asahi-dev · 2026-07-04 22:16
yukabut the commit messages are to some extent documentation of the state and some of the things I ran into#asahi-dev · 2026-07-04 22:16
yukathis may or may not be useful in its current state#asahi-dev · 2026-07-04 22:16
yukahttps://cyberchaos.dev/yuka/linux/-/commits/feature/m4-m5-minimal-device-trees#asahi-dev · 2026-07-04 22:15
StanfordAppliedCyber[m]yeah fair, not quite there yet, this accursed gpu#asahi-dev · 2026-07-04 22:15
yukaanyways I'm leaving for today#asahi-dev · 2026-07-04 22:15
yukabut this means it should probably already be at the login screen at least