#
✕ nick:yuka
#asahi-dev · 2026-07-09 17:17
yukaor was it the opposite?
#asahi-dev · 2026-07-09 17:17
yukaonly the hub needed resetting?
#asahi-dev · 2026-07-09 17:16
yukabut I think the usb2 repeater was always present on the i2c bus?
#asahi-dev · 2026-07-09 17:16
yukaAh, no
#asahi-dev · 2026-07-09 17:01
yukait's very possible some setup of the usb repeater over i2c is necessary
#asahi-dev · 2026-07-09 16:41
yukaI don't know yet how to tell which smc keys are gpios
#asahi-dev · 2026-07-09 16:41
yukaAnd then write an automated test
#asahi-dev · 2026-07-09 16:40
yukathe uart is too slow for smccli
#asahi-dev · 2026-07-09 16:40
yukaI should probably add a more specialized proxy command which is small enough to run over the broken/slow uart, takes a gpio name and state and toggles that gpio before turning on usb
#asahi-dev · 2026-07-09 16:38
yukathis pulls the port out of debugusb mode (I confirmed that this works in principle on a device without usb hub)
#asahi-dev · 2026-07-09 16:38
yukathen run the INIT_USB command and quickly re-plug the usb cable
#asahi-dev · 2026-07-09 16:38
yukaI start out in debugusb mode, enable/disable the gpio(s) via smccli
#asahi-dev · 2026-07-09 16:37
yukathe way I currently test is using a new INIT_USB proxy command which has 5 seconds of delay built in
#asahi-dev · 2026-07-09 16:37
yuka* by pass the usb hub
#asahi-dev · 2026-07-09 16:37
yukaOne of the following to assumptions is wrong: 1. toggling the gpio is enough to bypass the usb, 2. the gpio is one of gP?? on the SMC
#asahi-dev · 2026-07-09 16:36
yukano luck so far
#asahi-dev · 2026-07-09 16:35
yukaI took a stab at finding the gpio to bypass the usb hub on the NEo
#asahi-dev · 2026-07-09 15:37
yukaoh, true
#asahi-dev · 2026-07-09 15:27
yukaI don't know if it actually makes sense from a stability / power draw reduction point of view, but it's a good use case to think about when adding the pwren stuff
#asahi-dev · 2026-07-09 15:25
yukaSo modeling it well is difficult
#asahi-dev · 2026-07-09 15:25
yukathis needs to coordinate the sd card reader, pcie and smc/gpio drivers
#asahi-dev · 2026-07-09 15:25
yukaThe pcie reader is enabled on boot, probed, then shut down until an sd card is inserted
#asahi-dev · 2026-07-09 15:24
yukaYes
#asahi-dev · 2026-07-09 15:24
yukamarcan mentioned it as an example of what a good device tree /driver architecture should be able to do
#asahi-dev · 2026-07-09 15:23
yukapcie power management
#asahi-dev · 2026-07-09 15:21
yukado we know how to read it?
#asahi-dev · 2026-07-09 15:21
yukais smc-pmu something entirely different than the other smc gpios?
#asahi-dev · 2026-07-09 15:20
yukacould not find the sd card detection pin with smc_watcher.py
#asahi-dev · 2026-07-09 08:49
yukaI don't really care if we do dockchannel-uart over mailbox or not, but ideally the device tree should describe the hardware without making assumptions about the implementation, which is really difficult in this case
#asahi-dev · 2026-07-09 06:43
yuka*cores
#asahi-dev · 2026-07-09 06:42
yukaCan you make the exception handler quiet on all cotes except one?
#asahi-dev · 2026-07-09 06:25
yukaintegralpilot: do you have a suggestion for how to handle this?
#asahi-dev · 2026-07-09 06:17
yukaBut it feels wrong to follow the mailbox link, assume the mailbox is a dockchannel device and read its regs.
#asahi-dev · 2026-07-09 06:16
yukaNow I'm trying to come up with a way to have this driver run on the proposed upstream device trees.
#asahi-dev · 2026-07-09 06:16
yukaI have a dockchannel-uart driver which works for tx / direct register access (so what we need for earlycon). Currently it is independent and has its own regs (pointing to the same as adt dockchannel-uart[0]).
#asahi-dev · 2026-07-09 06:14
yukaI'm getting headaches from the dockchannel device tree layout
#asahi-dev · 2026-07-09 06:14
yukathe garbled exception handlers output
#asahi-dev · 2026-07-09 06:14
yukaI've seen that before
#asahi-dev · 2026-07-09 05:16
yukaon the same version which didn't do this before
#asahi-dev · 2026-07-09 05:15
yukaI got the SError on trying to access DISP_CPU on t8132 too
#asahi-dev · 2026-07-08 19:46
yukagot tx mostly working including multiple channels, console support and earlycon
#asahi-dev · 2026-07-08 19:46
yukaI decided I want to have a working dockchannel uart driver before trying to make it work with the mailbox
#asahi-dev · 2026-07-08 08:56
yukaI'm sure of it
#asahi-dev · 2026-07-08 08:56
yukaone of them was jannau testing it
#asahi-dev · 2026-07-08 08:56
yukalol, the two users with a Mac Pro
#asahi-dev · 2026-07-08 08:56
yuka> stats
#asahi-dev · 2026-07-08 08:30
yuka27.0b3 on t8132 is unchanged to 27.0b2: SYS_IMP_APL_VM_TMR_FIQ_ENA_EL2 is locked and iboot regularly panics
#asahi-dev · 2026-07-08 08:14
yuka"start" is judging from pmgr regs
#asahi-dev · 2026-07-08 08:13
yukathe cores start, but never enter our code
#asahi-dev · 2026-07-08 07:47
yukaprogress, yay

plus ›