#asahi-dev · 2025-05-17 07:32
svencan take a brief look later today if you want to wait until then #asahi-dev · 2025-05-15 15:54
svenadjusted the reg sizes as well now#asahi-dev · 2025-05-15 15:53
svenwill wait for after the merge window to send this out anyway but i'll use it as a base to continue preparing that usb3 series#asahi-dev · 2025-05-15 15:53
sventhanks!#asahi-dev · 2025-05-15 15:48
svendidn't notice the hilighted line#asahi-dev · 2025-05-15 15:48
svenohhh#asahi-dev · 2025-05-15 15:47
svenchaos_princess: node name?#asahi-dev · 2025-05-15 15:42
svenhrm... or maybe not#asahi-dev · 2025-05-15 15:42
sveni think i can even reduce that to 1 for all of them#asahi-dev · 2025-05-15 15:41
svenabsolutely :)#asahi-dev · 2025-05-15 15:41
svenduh, yes#asahi-dev · 2025-05-15 15:34
svenhttps://github.com/AsahiLinux/linux/commits/sven/nvmem/ would be nice if someone could double check if I got the fixup dt commits correct. they covert the 4 byte aligned fuses to byte-aligned ones#asahi-dev · 2025-05-13 18:04
svenand also dev_err to dmesg in the gpu driver when the required properties aren't found so that we know if we get bug reports what's going on#asahi-dev · 2025-05-13 18:03
svenwe can't prevent that but we need to make it very obvious when that happens#asahi-dev · 2025-05-13 18:02
svenwe just have to make sure we don't shoot ourselves in the foot because distros compile m1n1 without rust and then the gpu doesn't work#asahi-dev · 2025-05-13 18:02
svenworks for me#asahi-dev · 2025-05-13 18:02
svenwe either need to make that rust part optional or convert it to C#asahi-dev · 2025-05-13 18:01
svenyes, but not in m1n1 stage 2 because we break *all* openbsd users then#asahi-dev · 2025-05-13 18:00
svenand let's not make rust mandatory for m1n1 stage 2 then#asahi-dev · 2025-05-13 17:57
svenmight be worth trying to fix it before 6.16 is out (not -rc1 but the final release weeks from now)#asahi-dev · 2025-05-13 16:46
svenalso, i vaguely remember that compiling m1n1 w/ rust was kinda painful on bsd. if my memory's not wrong: is that still true? i'm asking because we're considering copy/pasting some rust code into m1n1 which would require rust even for stage2 with the current implementation (https://github.com/AsahiLinux/m1n1/pull/466)#asahi-dev · 2025-05-13 16:44
svenkettenis: how do we continue with that compatible? try to get it changed upstream or add a quirk to m1n1 that appends the old compatible for at least the next year at least to the device trees?#asahi-dev · 2025-05-12 17:59
svenI think the current solution (allow offset >8) makes more sense but I’m not gonna fight for it #asahi-dev · 2025-05-12 17:57
sventhe other alternative ofc is to do the unaligned u8 reading inside our driver and just adjust all the dt nodes *shrug*#asahi-dev · 2025-05-12 17:57
svenI’ve only adjusted the reading but not the writing path #asahi-dev · 2025-05-12 17:56
sventhe revert needs to be changed anyway I think #asahi-dev · 2025-05-12 17:56
svenbut makes sense #asahi-dev · 2025-05-12 17:55
svenlol, ouch #asahi-dev · 2025-05-12 16:33
svenhttps://lore.kernel.org/asahi/174705404483.2941395.9971423828224597027.robh@kernel.org/T/#m0b44c56cee7ba4c437543136a14bd9574e280575 huh... why? did I mess the original apple,efuses.yaml up or am I overlooking something dumb?#asahi-dev · 2025-05-12 05:22
sven*while#asahi-dev · 2025-05-12 05:22
svenyeah, that’s what I was thinking. just add it there and leave that code around for long whilen#asahi-dev · 2025-05-11 20:06
svenor sneakily adjust it in m1n1#asahi-dev · 2025-05-11 20:06
svenugh. maybe we can still fix that but we should at least ship a downstream device tree with the old compatible for quite a while #asahi-dev · 2025-05-11 10:42
svensmc errors from the kext#asahi-dev · 2025-05-11 10:42
svenhttps://gist.github.com/svenpeter42/c9467235de8e8f35471b686658fc77a3#asahi-dev · 2025-05-11 10:33
svenjannau: probably obvious, but macOS calls error 132 from smc kSMCKeyNotFound#asahi-dev · 2025-05-11 08:21
svenhttps://lore.kernel.org/asahi/20250511-smc-6-15-v5-0-f5980bdb18bd@svenpeter.dev/ smc v5#asahi-dev · 2025-05-10 15:30
svenack, i'll wait for your review then. was planning to send out v5 tomorrow probably#asahi-dev · 2025-05-10 15:28
svenmostly lee's and alyssa's comments addressed plus some drive-by fixes#asahi-dev · 2025-05-10 15:28
svenhttps://github.com/AsahiLinux/linux/commits/sven/smc-v5/ is what i have for v5 right now fwiw#asahi-dev · 2025-05-10 15:23
sven return -EIO;#asahi-dev · 2025-05-10 15:23
sven if (result)#asahi-dev · 2025-05-10 15:23
sven result = FIELD_GET(SMC_RESULT, smc->cmd_ret);#asahi-dev · 2025-05-10 15:22
svenI think i fixed that in v5#asahi-dev · 2025-05-10 15:00
svenlol, yeah, I noticed those as well when I compiled smc with W=1 like lee requested earlier today :D#asahi-dev · 2025-05-10 14:56
sventhen i'm not gonna add that to that PR#asahi-dev · 2025-05-10 14:55
svenhave we cleared the before?#asahi-dev · 2025-05-10 14:20
svennickchan: that entire sleep function looks suspicious to me anyway. the isb; wfi in a loop makes no sense to me e.g.#asahi-dev · 2025-05-10 14:15
svenhttps://github.com/AsahiLinux/linux/commits/sven/smc-v5/ smc v5 with review comments addressed