#
✕ nick:jannau
#asahi-dev · 2022-08-10 18:17
jannauI could do the device tree parsing there but it feels wrong to do that in alloc_io_pgtable_ops() instead of dart_probe()
#asahi-dev · 2022-08-10 18:15
jannausven: my problem with atomic updates of the L2 tables for locked darts is the availability of the pre-reserved mappings in io-pgtbl*
#asahi-dev · 2022-08-08 11:17
jannauprice would be more complicated fdt fixups in m1n1 and less information in the dts source
#asahi-dev · 2022-08-08 11:16
jannaujust looking at the dart and transfer the mappings to the fdt would be valid too and should avoid changes on new SoCs
#asahi-dev · 2022-08-08 11:14
jannauthe region-id-xy properties look stable over macos releases and different SoC. the only change I see is that m1x uses an additional region and m2 again one additional region over m1x
#asahi-dev · 2022-08-08 11:12
jannauone semi open question is how we want to handle the mapping transfer, I prefer the mostly static approach in https://github.com/jannau/m1n1/commit/950af163563c19ece1f28b4271e60ba887dd1e20 based on /chosen/carveout-memory-map/ properties and templated fdt nodes
#asahi-dev · 2022-08-08 11:10
jannauregarding iommu handoff, thierry has dt-bindings nobody complained about yet
#asahi-dev · 2022-08-08 11:04
jannaumost current dcp WIP tree is in https://github.com/jannau/linux/tree/asahi-dcp-july and https://github.com/jannau/m1n1/tree/dcp-july based on asahi / m1n1 from 2022-07-31, I expect it to rebase without problem
#asahi-dev · 2022-08-08 10:58
jannaulocking dart-disp0 is probably ok, we have to handle that case already for dcp
#asahi-dev · 2022-08-08 10:57
jannauhmm, something changed on the laptop as well since I wasn't been able to wake it up without the register poke
#asahi-dev · 2022-08-08 10:56
jannauyes, disp0 is the problem. I think dcp wouldn't be a problem with dcp been shutdown
#asahi-dev · 2022-08-08 10:23
jannauthis will take at least a couple of weeks
#asahi-dev · 2022-08-08 10:21
jannauI traced the dp-crossbar already and there was not going on much
#asahi-dev · 2022-08-08 10:20
jannauI was planning to start looking at hdmi port on the macbook pro 14/16 as that omits atc-dpin, no idea if that makes the atc handling any easier
#asahi-dev · 2022-08-08 06:25
jannaubiggest item is switching dart-dcp and stream 4 of dart-disp* to an unmanaged iommu domain
#asahi-dev · 2022-08-08 06:23
jannauwe need to come up with a new compatible string for dart-disp* to avoid the reset on probe
#asahi-dev · 2022-08-08 06:21
jannauexternal display support is important for the submission to get the device tree bindings right
#asahi-dev · 2022-08-08 06:20
jannaudcp should work on all machines except m2, I haven't looked at m2 since it needs further dart changes since "vm-base" is non-zero
#asahi-dev · 2022-08-08 06:18
jannaumarcan: backlight control seems to require the screen calibration data (set_block_dcp) or something else. Just setting the property like macos crashes dcp
#asahi-dev · 2022-08-05 20:57
jannauperf top accounts 20% to pthread_mutex_lock and 10% to clock_gettime (+10% unknown symbol in vdso)
#asahi-dev · 2022-08-05 20:55
jannauabmysal revocery rates for replicated pools (5 - 15 MB/s) with high CPU load
#asahi-dev · 2022-08-05 20:53
jannaumarcan: fyi ceph's threadpool looks badly broken on my m1 mini install (gentoo's ceph 17.2.3 compiled with "-O2 -march=armv8.4-a")
#asahi-dev · 2022-08-04 20:42
jannauthe keyboard brightness menu
#asahi-dev · 2022-08-04 20:42
jannautop right menu is in ventura replaced with "stage manager"
#asahi-dev · 2022-08-04 20:37
jannauor in system settings, which is only visible if "adjust keyboard brightness in low light" is not activated
#asahi-dev · 2022-08-04 20:33
jannaujust on/off. is there an option to dim the keyboard backlight manually in macos?
#asahi-dev · 2022-08-04 20:31
jannau+set
#asahi-dev · 2022-08-04 20:30
jannaubtw, there is a spi hid command to toggle the keyoard backlight. macos uses that when the keyboard backlight is to only while typing
#asahi-dev · 2022-08-04 16:46
jannauno, an usb2 usb-a/usb-c cable is enough
#asahi-dev · 2022-08-04 16:45
jannauwith an aarch64 (cross-)compiler
#asahi-dev · 2022-08-04 16:45
jannauno, any machine with usb2 should work
#asahi-dev · 2022-08-03 22:10
jannauone getting zapped by iboot is the only reason I can think of. still strange that there would be two logic board variants which differ in this aspect
#asahi-dev · 2022-08-03 22:09
jannaumacos 12.5 j413 template adt still has two identical cs42l84 codec nodes on i2c1 and i2c3
#asahi-dev · 2022-08-03 22:08
jannaucongrats povik
#asahi-dev · 2022-08-03 21:35
jannaudid you check the 12.5 adt as well?
#asahi-dev · 2022-08-03 21:27
jannauit's on i2c3 for j493
#asahi-dev · 2022-08-03 21:26
jannaumaybe there are two? ask marcan to look at the iboot processed adt
#asahi-dev · 2022-08-03 19:54
jannauand the cable being connected before booting m1n1
#asahi-dev · 2022-08-03 19:53
jannaudo you see anything usb releated in dmesg or lsusb after connecting the cable?
#asahi-dev · 2022-08-03 19:49
jannauall usb-c ports should provide acm over usb except on mac studio m1 max (front ports don't work) and the 4 port imac (I think only the 2 ports closest to the power connector work)
#asahi-dev · 2022-08-03 06:44
jannauI tested it only on the m2 macbook as I assumed m1 would continue to work
#asahi-dev · 2022-08-03 06:34
jannaualthough I can see the appeal of using ~arm64 in general. I'm using it so only where necessary but it's a little bit annoying
#asahi-dev · 2022-08-03 06:32
jannaunothing wrong with the ebuild, also it's ~arm64 keyworded for a reason
#asahi-dev · 2022-08-03 05:03
jannauchadmed: confirmed with asahi linux. spihid keyboard is broken with u-boot in asahi-dev
#asahi-dev · 2022-08-02 05:30
jannaumarcan: did you tests the installed asahi-dev u-boot on macbook air and pro?
#asahi-dev · 2022-08-01 17:58
jannauI'm not sure if I have time to look into it tonight. multiple Linux installs on the device
#asahi-dev · 2022-08-01 17:57
jannaumarcan: update to asahi-dev's latest kernel/u-boot on the m2 macbook pro still has no keyboard in u-boot
#asahi-dev · 2022-07-31 20:17
jannaulightly tested on mac mini and m1 max, major changes: should support DPMS now and I made some calls synchronous but waiting for the completion of the dcp reply
#asahi-dev · 2022-07-31 20:14
jannaualyssa: dcp rebase ontop of current asahi/asahi: https://github.com/jannau/m1n1/tree/dcp-july + https://github.com/jannau/linux/tree/asahi-dcp-july
#asahi-dev · 2022-07-31 19:56
jannauI guess that means we have set aperture_start/aperture_end from device tree based on vm-base and vm-size from the ADT

plus ›