#
✕ nick:janneg
#asahi-dev · 2023-12-02 10:56
jannegsven: I see 0x1f and 0x7f for logs which should be macos (possibly over different devices and displays). 0x3f seems to appear only in linux logs
#asahi-dev · 2023-12-01 23:20
jannegretrieving the EDID from dcp seems to be surprisingly simple
#asahi-dev · 2023-12-01 23:16
jannegwasn't that a m2 max/ultra update at release with 13.x?
#asahi-dev · 2023-11-30 22:17
jannegso it looks more like the m1/m2 macbook pro 14"
#asahi-dev · 2023-11-30 22:16
jannegI suppose M3 macbook pro is an earlier design than the m3 pro/max. It was supposed to be announced in the summer but is late because there was not enough chip supply for macbook air 15"
#asahi-dev · 2023-11-30 20:23
jannegwe could try dcpext1 and see if that fixes the issue on the ultra as well (don't ask me why that doesn't depend on ps_afr)
#asahi-dev · 2023-11-30 20:21
jannegI think it's the same problem I see/saw with dcp on the m2 pro so until that's solved I can test myself
#asahi-dev · 2023-11-30 20:18
jannegI'm not sure about that. I tried horrible mdelay() crimes without any luck
#asahi-dev · 2023-11-30 20:15
jannegleio: I have a partial solution for the dcp/hdmi issue on linux. dcp at least wakes up but is not fully working except with tracing and verbose rtkit syslog logging
#asahi-dev · 2023-11-30 20:10
jannegthe second die has an offset of 0x2000000000
#asahi-dev · 2023-11-30 20:09
jannegleio: yes. m2 ultra always used dcpext0 (on the second die)
#asahi-dev · 2023-11-30 19:57
jannegdon't ask my why ps_dispext0_cpu0 has ps_afr as parent (via ps_dispext0_sys)
#asahi-dev · 2023-11-30 19:55
jannegu-boot accesses the whole simple-framebuffer power-domain tree and runs into a data abort on the unmapped 0x404e80000 + 0x100 with https://github.com/AsahiLinux/linux/commit/760ed8e3be2f4a8c550f197aeb135e7564e0ee2e
#asahi-dev · 2023-11-30 19:45
janneg0x480000000 is only for completeness without current use
#asahi-dev · 2023-11-30 19:40
jannegkettenis: partially, dcpext0/dispext0 on t602x which drives simple-framebuffer needs ps_afr from pmgr_gfx at 0x404e80000
#asahi-dev · 2023-11-30 19:14
jannegM3 Macbook Pro still uses cd3217 as type-c controller
#asahi-dev · 2023-11-30 17:54
jannegnot that either of those has any useful public information
#asahi-dev · 2023-11-30 17:48
janneghttps://www.ifixit.com/Guide/MacBook+Pro+14-Inch+Late+2023+M3+Pro+Chip+ID/167049
#asahi-dev · 2023-11-30 17:47
jannegbut the macbook pro teardown has Texas Instruments SN25A12 USB type-C controllers
#asahi-dev · 2023-11-30 17:46
jannegifixit has identifies a Texas Instruments SN2012017 battery charger on the iphone 15 pro
#asahi-dev · 2023-11-30 17:38
jannegDannyB: it is 1 die. the second number is the maximum numnber of dies the implemenation can handle. the theoretical limit is 8 but so far Mx Max can only handle irqs from 2 dies i.e. for Mx Max pairs in Mx Ultra (if aic3 in still compatible to aic2 in that regard)
#asahi-dev · 2023-11-29 19:37
jannegaccelrators means in this case decoding and encoding engines. those are in apr. the video decoding block supports h264, h265 and vp9
#asahi-dev · 2023-11-29 19:31
jannegwasn't the hw prores decoder/encoder added with m1 pro/max?
#asahi-dev · 2023-11-29 08:26
jannegkettenis: is there a reason why we split the main mmio range up in u-boot?
#asahi-dev · 2023-11-27 19:02
jannegI think this is the same issue I saw on m2 pro with quiesced vs. sleep
#asahi-dev · 2023-11-27 18:49
jannegleio: please try to revert the change to dcp_shutdown() in https://github.com/AsahiLinux/m1n1/commit/5006e083c754700d7fe79de9d6081133fa7a6b3e#diff-646589574896ce29a1bd75998aebe7224b21817ee9adb8631e3bd864c4b612e0 in the 2nd stage m1n1
#asahi-dev · 2023-11-27 18:46
jannegso it fails early enough to not tear down the boot frame buffer. that's one good bit
#asahi-dev · 2023-11-27 18:43
jannegbut stage 1 m1n1 is a previous version and not v1.4.8? Do not install v1.4.8 as 1st stage
#asahi-dev · 2023-11-27 18:42
jannegok, can you try chainloaing m1n1 v1.4.8 multiple times? This looks like the problem for which I switched the other devices to dcpext(0)
#asahi-dev · 2023-11-27 18:40
jannegleio: does m1n1 initialize the display correctly? I suppose one as 2nd stage
#asahi-dev · 2023-11-27 17:52
jannegno, dcp seems to be fused off on both dies
#asahi-dev · 2023-11-27 17:38
jannegleio: ack
#asahi-dev · 2023-11-27 01:04
jannegDSC seems to work automatically but is used earlier than expected. Code limits the pixel clock to 924 MHz (~5k 60Hz)
#asahi-dev · 2023-11-27 01:01
jannegleio: https://github.com/AsahiLinux/linux/pull/240 and https://github.com/AsahiLinux/m1n1/pull/361 if you get a chance to test on m2 ultra
#asahi-dev · 2023-11-26 15:38
jannegyou have to guess which dcpext macos will use but at least it's not random per boot
#asahi-dev · 2023-11-26 15:36
jannegyou need to tell it which dcp/dcpext to trace. see 'dcp_name' at the beginning
#asahi-dev · 2023-11-26 15:33
jannegsigh, why can't the dcp firmware behave consistently over all platforms
#asahi-dev · 2023-11-26 14:30
janneg€500 for 5 meters expensive
#asahi-dev · 2023-11-25 23:45
jannegtomorrow (UTC)
#asahi-dev · 2023-11-25 23:44
jannegleio: I probably figured out why the dcp/linux doesn't work on m2 ultra. I should have soomething to test tomorrow. the 6k display will likely not work though, at least not with 60Hz
#asahi-dev · 2023-11-25 18:19
jannegsven: https://gist.github.com/jannau/060b3f97ab7c4140850661515bd9fb58
#asahi-dev · 2023-11-25 16:52
jannegsven: old i2c trace https://gist.github.com/jannau/023f2c2a36b880636e037151b84ca626
#asahi-dev · 2023-11-24 22:54
jannegyes, I see the same oslog messages from mtp on j493
#asahi-dev · 2023-11-24 22:51
jannegkettenis: which coproc and which device? I don't see anything on j314c or j375d. should be logged as warning
#asahi-dev · 2023-11-24 17:30
jannegnicolas17: VP9? they supposedly added that for Google/Youtube on A* chips but didn't want to support it / let the world know
#asahi-dev · 2023-11-21 12:08
jannegnicolas17: still looking for macOS build 22G2074?
#asahi-dev · 2023-11-19 17:37
jannegeric_engestrom: yes, eiln usually checks logs if not present
#asahi-dev · 2023-11-19 16:09
jannegleio: it's strange that you don't see any "DP2HDMI HPD" messages if you boot with hdmi cable connected
#asahi-dev · 2023-11-19 14:05
jannegleio: sounds like oops on dcp unload/poweroff which blocks shutdown
#asahi-dev · 2023-11-18 21:22
jannegyes, for dcp, that's why I said fun. I would have expected that the dp-audio0 nodes are deleted from the laptop device trees. needed for mac studio

plus ›