#asahi-dev · 2022-10-21 20:23
jannaudcp suspend works and resume almost works. it doesn't crash dcp but it appears to need a mode set after resume#asahi-dev · 2022-10-20 16:32
jannaubut we have to clear the L1 table in io-pgtble-dart since apple-dart can't replace an exisitng mapping#asahi-dev · 2022-10-20 16:30
jannauthe main problem is that io-pgtable-dart currently has no knowledge for which device it initializes the page table. so installing the mappings from the DT there is weird#asahi-dev · 2022-10-20 16:20
jannauas long as nothing invalidates the mapping nothing bad happens and scanout/simpledrm work#asahi-dev · 2022-10-20 16:19
jannauio-pgtable-dart, memremaps the L1 table and clears the page. later apple-dart restores the mapping based on the DT provided mappings#asahi-dev · 2022-10-20 16:17
jannauin a completely unsafe way but in practice nothing happens due to caching of the mappings#asahi-dev · 2022-10-19 22:06
jannaucould it be that the problem is that the device is probed after probing the first dart and device iommu setup races with probing of additional darts?#asahi-dev · 2022-10-19 22:04
jannaudwc3 and admac directly, pcie one for each link#asahi-dev · 2022-10-19 22:03
jannauit looks like probing dart and devices using dart at the same causes problems for devices which in someway use more than 1 dart#asahi-dev · 2022-10-19 21:56
jannaurobher: reproduced after removing the dart module from the initramfs#asahi-dev · 2022-10-19 21:32
jannaurobher: are the drivers mostly built-in? I can't reproduce it here but I have a much nicer probe order. all darts are probed before the first device is attached to an iommu domain#asahi-dev · 2022-10-19 21:22
jannaunot new#asahi-dev · 2022-10-19 21:19
jannauI see "cacheinfo: Unable to detect cache hierarchy for CPU x" for every cpu core#asahi-dev · 2022-10-19 21:09
jannauyes. it was a first guess since we moved dart_sio to order it correctly by address which could have cause merge/rebase issues but dtc would have screamed#asahi-dev · 2022-10-19 20:55
jannauI guess I'll test on the ultra, more darts to trigger bugs#asahi-dev · 2022-10-19 20:54
jannauno dart probe errors m2 so far and I did quite a few reboots due to the tps6598x issue#asahi-dev · 2022-10-19 20:40
jannaubut yes, I would suspect an interaction with something else in v6.1-rc1. I've never seen that with the same devicetrees in v6.0-rc*#asahi-dev · 2022-10-19 20:39
jannaupcie darts fail as well#asahi-dev · 2022-10-19 20:35
jannauhmm, is that just a side effect of the dual darts in admac#asahi-dev · 2022-10-19 20:30
jannausysfs: cannot create duplicate filename '/devices/platform/soc/39b400000.dma-controller/iommu_group'#asahi-dev · 2022-10-19 20:30
jannaudid we somehow end up with duplicated device nodes: #asahi-dev · 2022-10-19 12:28
jannaubut it's in nvram/adt which is more convenient#asahi-dev · 2022-10-19 12:24
jannaujust checked: it's encoded as string in the form 12-34-56-78-AD-EC... if you klich show password#asahi-dev · 2022-10-19 12:18
jannauisn't there an option in keychain to display as hex? I can't remember that I had to convert the keys from base64#asahi-dev · 2022-10-19 07:00
jannauI'll look tonight how (if) MAC and link key are encoded in there#asahi-dev · 2022-10-19 06:59
jannauBluetoothUHEDevices is in nvram-proxy-data in the ADT. the link keys are probably sensitive enough to avoid stuffing them into the devicetree#asahi-dev · 2022-10-19 06:43
jannauI see BluetoothUHEDevices in nvram which has strings of the connected devices and enough binary looking additional data to include MAC and link key#asahi-dev · 2022-10-19 05:52
jannaudo we know hoow the wlan ssid/passphrase/key is shared with recovery? I would expect that bt device MACs and link keys are shared in the same way#asahi-dev · 2022-10-19 05:35
jannauthey are synced with recovery as well so maybe they are in nvram#asahi-dev · 2022-10-19 05:33
jannaumarcan: bluetooth link keys are stored in the keychain#asahi-dev · 2022-10-18 19:45
jannausmc might be easier as it's already used for waking on the lid/power button#asahi-dev · 2022-10-18 19:37
jannausmc needs to have a gpio or similar for any keypress for the annoying feature of powering the laptops on on a keypress#asahi-dev · 2022-10-18 19:36
jannauI think we're currently not not powering the keyboard down so we could just wakeup on a keypress#asahi-dev · 2022-10-18 19:34
jannauI was verifying that keyboard input doesn't wake and did tenths of keypresses which filled the password prompt after waking#asahi-dev · 2022-10-18 19:27
jannausleep working on the m2 macbook pro as well#asahi-dev · 2022-10-18 19:27
jannaunot using the keyboard as wakeup source but also not powring it down has has unexpected results#asahi-dev · 2022-10-18 11:26
jannaumarcan: sd reader was working at some point on the mac studio, I think#asahi-dev · 2022-10-18 11:25
jannauso no more excuse to skip rpm in new drivers#asahi-dev · 2022-10-17 22:02
jannausven: not reproducible with a single msleep(1); in the irq path so too fast looks like a valid theory#asahi-dev · 2022-10-17 21:57
jannauM2 has atcrt0 and atcrt1 on i2c2 with 'compatible = [atcrt]' but not much information to figure out what it is#asahi-dev · 2022-10-17 21:38
jannauat least for the audio codec it is working. setting the volume got noticeable faster#asahi-dev · 2022-10-17 21:23
jannauno longer reproducible with "i2c/pasemi: PASemi I2C controller IRQ enablement" reverted#asahi-dev · 2022-10-17 21:00
jannauI can't remember when I saw that the last time. obvious change in the kernel is i2c irq support#asahi-dev · 2022-10-17 20:59
jannaualso reproducible with an usb device on the UART enabled port, could be random#asahi-dev · 2022-10-17 20:51
jannaureproducible with disconnecting a usb-c charger#asahi-dev · 2022-10-17 20:49
jannauI'm seeing tps6598x issues with v6.1.0-rc1-asahi-1 http://paste.debian.net/1257408/#asahi-dev · 2022-10-17 19:57
jannauthe kcalc energy use is clearly a bug. It gets constantly resizeEvents and redraws itself. bug filed in kde's bugzilla#asahi-dev · 2022-10-17 12:49
jannaulooks like it re-renders constantly#asahi-dev · 2022-10-17 12:44
jannauI'm mostly astounded that there's still such broken user space#asahi-dev · 2022-10-17 12:42
jannaumore than 9W vs. 3W and a noticeable warm macbook