#asahi · 2021-10-18 02:17
alyssaNo#asahi · 2021-10-18 02:17
alyssajmr2: Anwyay, so there's no problem parsing#asahi · 2021-10-18 02:15
alyssahttps://github.com/AsahiLinux/docs/wiki/FW:ADT#asahi · 2021-10-18 02:14
alyssaI think there are redacted ones floating around#asahi · 2021-10-18 02:14
alyssam1n1 has a script for it but there's PII in them by default IIRC#asahi · 2021-10-18 02:14
alyssaUhhh#asahi · 2021-10-18 02:10
alyssaI suppose I should take a look at a macbook air ADT next..#asahi · 2021-10-18 02:10
alyssa^ Here's the data you sent as JSON (from https://github.com/AsahiLinux/dcp-parser )#asahi · 2021-10-18 02:10
alyssahttps://rosenzweig.io/timing-blob.txt#asahi · 2021-10-18 02:10
alyssahttps://rosenzweig.io/product.txt#asahi · 2021-10-18 02:10
alyssahttps://rosenzweig.io/color-blob.txt#asahi · 2021-10-18 02:09
alyssaThank you so much for grabbing that unsolicited dump ;-)#asahi · 2021-10-18 02:09
alyssaI see what's happening now. #asahi · 2021-10-18 02:09
alyssajmr2: Yeah, this is right.#asahi · 2021-10-18 02:07
alyssaNo, it's not cut off. It's just short on the macbook air. nice.#asahi · 2021-10-18 01:56
alyssaunless it just be like that on the macbooks#asahi · 2021-10-18 01:56
alyssaI suppose that could explain the parse fail, though#asahi · 2021-10-18 01:55
alyssaThose all seem to be way too short..#asahi · 2021-10-18 01:54
alyssajmr2: Ah ha#asahi · 2021-10-17 21:21
alyssathe iBoot simple-framebuffer ought to be sufficient for the first stage bootloader anyway#asahi · 2021-10-17 21:20
alyssabut for the typical case it ... isn't necessary?#asahi · 2021-10-17 21:20
alyssaThere is a way to make all of this work of course (1TR -> macOS)#asahi · 2021-10-17 21:20
alyssapovik: It's worth noting I explicitly think using DCP in the bootloader is... a bit silly.#asahi · 2021-10-17 21:02
alyssaerasing the mac is definitely not necessary#asahi · 2021-10-17 20:17
alyssaYou did! :)#asahi · 2021-10-17 19:54
alyssajmr2: anyway, i'll need the actual dumps of what we're trying to parse#asahi · 2021-10-17 19:54
alyssa(skip is also destructive)#asahi · 2021-10-17 19:53
alyssajmr2: parse tag is destructive, that won't work#asahi · 2021-10-17 19:18
alyssathank you#asahi · 2021-10-17 19:18
alyssathat makes sense#asahi · 2021-10-17 19:18
alyssa?#asahi · 2021-10-17 19:18
alyssasven: so it's sae to DEFER on "another device has probe()d" ubut not safe to DEFER on "another device has booted after probing"#asahi · 2021-10-17 19:17
alyssaAhhhhh#asahi · 2021-10-17 19:17
alyssaj_ey: it just keeps trying#asahi · 2021-10-17 19:17
alyssaThis is racy-- the core logic will stop trying to probe after a while if you DEFER too many times. Depending how fast the DCPs come up, this can be racy#asahi · 2021-10-17 19:16
alyssaand then the core logic will try again later when the DCPs come up#asahi · 2021-10-17 19:16
alyssaand in the apple-display driver probe(), check if all the coprocessors from the DT are booted and if not return -EPROBE_DEFER#asahi · 2021-10-17 19:16
alyssaThe "obvious" way is to have an "is_booted()" call on the DCP driver#asahi · 2021-10-17 19:16
alyssabut then how do you ensure the display subsystem doesn't probe until both coprocessors boot?#asahi · 2021-10-17 19:15
alyssadisplay-subsystem { compatible="apple-display"; coprocessors = <&dcp, &dcpext> }#asahi · 2021-10-17 19:15
alyssadcpext { compatible="dcp" }#asahi · 2021-10-17 19:15
alyssadcp { compatible="dcp" }#asahi · 2021-10-17 19:15
alyssaIt's tempting to model this as 3 nodes in the device tree:#asahi · 2021-10-17 19:14
alyssaThat is, we want to present userspace the impression of dual monitors on one graphics card, in oldspeak.#asahi · 2021-10-17 19:14
alyssaThere are two DCPs. Each DCP has its own drm_crtc object, but we want a single drm_device object for both that contains both drm_crtc's.#asahi · 2021-10-17 19:13
alyssais typing#asahi · 2021-10-17 19:13
alyssai wish irc had typing indicators#asahi · 2021-10-17 19:12
alyssasven: subsystem stuff.#asahi · 2021-10-17 19:12
alyssai don't like that indeed#asahi · 2021-10-17 19:11
alyssaIn theory that sort of nonsense is supposed to be avoided with the mode_valid callback