arm64: dts: apple: t8112-j493: Enable SIO for DP audio - #572
Conversation
On bootup, an Apple MacBook Pro (13-inch, M2, 2022) dmesg shows: dcp-dp-audio 238334000.audio-controller: No DMA device And: $ cat /proc/device-tree/soc/sio@236400000/status disabled This model's t8112-j493 hardware is able to encode this data through its SIO coprocessor, but the device tree doesn't configure it to be able to. Similar to t8112-j473, enable the node and assign the sio alias so m1n1 injects the correct firmware. With this applied the sink is available: ~ cat /proc/asound/cards 0 [DisplayPort ]: apple_dcp - Apple DisplayPort Apple DisplayPort ... And audio works correctly over DP and is stable. Signed-off-by: Lorenzo Stoakes <ljs@kernel.org>
|
I do not think we want it to be generally available yet, whether it is stable or not highly depends on which device is on the other side of the dp link. |
|
Ack fair enough, I have tested a Samsung LS49AG95 which is working fine, but I did notice some (minor) instability immediately after sending this (typical) - the audio dropped after I switched from 120 Hz -> 60 Hz requiring a pipewire restart that fixed it. I also noticed (not audio obviously but DP-related) it seems to have issues with 10-bit colour depth at 120 Hz, resulting in compression and visual artifacts, so have forced to 60 Hz, will try a hack locally to force 8 bit to see if that fixes that. Obviously intending this one for the fairydust branch but understood if you'd rather avoid possible instability with this even on that. Will report any other bugs I notice with this setup, is rapidly becoming a daily driver for me now... :) |
|
If i found the right monitor - isn't that expected? 5120x1440@120hz with 10bpp is just slightly above dp 1.4 speed, so it needs either subsampling or dsc, and i am not sure if we do that correctly. |
|
yeah it just seems the G9 isn't brilliant at rendering the compressed signal :) and it seems that 8 bit isn't available as an option as currently implemented? I've put in a hack to force 8 bit which seems to have fixed it. But obviously the real fix would be to make that otherwise available. |
2 participants
On bootup, an Apple MacBook Pro (13-inch, M2, 2022) dmesg shows:
And:
This model's t8112-j493 hardware is able to encode this data through its SIO coprocessor, but the device tree doesn't configure it to be able to.
Similar to t8112-j473, enable the node and assign the sio alias so m1n1 injects the correct firmware.
With this applied the sink is available:
And audio works correctly over DP and is stable.