#
✕ nick:kettenis
#asahi-dev · 2022-01-14 11:08
kettenisu-boot nvme support sent upstream
#asahi-dev · 2022-01-13 09:05
kettenisanyway, I now have a fairly elegant hack in OpenBSD to power the machine down
#asahi-dev · 2022-01-13 09:03
kettenisthis suggests that the watchdog just activates a GPIO which then triggers a reset through the PMU
#asahi-dev · 2022-01-13 09:01
kettenissomewhat unexpected indeed
#asahi-dev · 2022-01-12 22:00
ketteniscorellium writes to two other bytes, but this is the only one that seems to be really necessary
#asahi-dev · 2022-01-12 21:53
kettenisor via the watchdog
#asahi-dev · 2022-01-12 21:53
ketteniseither via the SMC ("MBSE"/"off1")
#asahi-dev · 2022-01-12 21:52
kettenisso to make the M1 machines powerdown, I can get away with writing 0x08 at address 0x9f0f in the SPMI PMU's address space and trigger a reset
#asahi-dev · 2022-01-12 09:39
kettenisjmr2: best if I add a description myself
#asahi-dev · 2022-01-11 22:43
kettenisah, a fellow OpenBSD developer has figured out the SMC keys for battery time that's left and time to fully charge
#asahi-dev · 2022-01-09 22:07
kettenisthose values look plausible I guess
#asahi-dev · 2022-01-09 22:07
kettenisaplsmc0.temp4 26.00 degC palm rest
#asahi-dev · 2022-01-09 22:07
kettenisaplsmc0.temp3 28.00 degC palm rest
#asahi-dev · 2022-01-08 17:23
kettenisah yeah, please fix that
#asahi-dev · 2022-01-08 13:36
kettenisI've entered that into https://github.com/AsahiLinux/docs/wiki/SW:-DT-bindings
#asahi-dev · 2022-01-08 13:35
kettenisyup
#asahi-dev · 2022-01-08 13:33
kettenisthe ADT has no indication that there are multiple versions of it
#asahi-dev · 2022-01-08 13:33
kettenisspeaking about SMC, I suppose we'll give it a "apple,smc" compatible?
#asahi-dev · 2022-01-08 13:16
kettenismarcan: thanks
#asahi-dev · 2022-01-08 13:00
kettenisand the duplicate include
#asahi-dev · 2022-01-08 13:00
kettenismarcan: still has the incorrect apple,t6000-spi in the t8103.dtsi
#asahi-dev · 2022-01-06 13:38
kettenisrole = ANS2?
#asahi-dev · 2022-01-06 13:35
kettenisI've already added the compatible for that to OpenBSD ;)
#asahi-dev · 2022-01-06 13:35
kettenissven: let's stick to -ans2
#asahi-dev · 2022-01-06 13:30
kettenisand "-ans" vs. "-ans2"
#asahi-dev · 2022-01-06 13:29
kettenisthe current issues seem to be "-mailbox" vs "-mailbox-v4" (the latter being submitted upstream, but not used in any of the asahi trees yet)
#asahi-dev · 2022-01-06 13:28
kettenisyup
#asahi-dev · 2022-01-06 13:27
ketteniswhich means that I need stable DT bindings for nvme by mid march or so ;)
#asahi-dev · 2022-01-06 13:26
kettenisMay 1st in principle
#asahi-dev · 2022-01-06 13:25
kettenisnow that I fixed the MP issue on OpenBSD the plan is to "officially" support the M1 machines in OpenBSD 7.1
#asahi-dev · 2022-01-06 13:24
kettenisI hope to get at least 1-3 done for the 2022.04 release
#asahi-dev · 2022-01-06 13:21
ketteniswith m1 pro/max support thrown in somewhere once we figure out what the problem with my current diff is
#asahi-dev · 2022-01-06 13:20
kettenis4. pci
#asahi-dev · 2022-01-06 13:20
kettenis3. spi keyboard
#asahi-dev · 2022-01-06 13:20
kettenis2. nvme
#asahi-dev · 2022-01-06 13:19
kettenis1. get upstreamed what I already submitted (wdt, pmgr)
#asahi-dev · 2022-01-06 13:19
kettenisanyway, my priorities for u-boot at the moment are:
#asahi-dev · 2022-01-06 13:19
kettenisdon't think u-boot supports pci sdhc controllers on non-x86 systems right now
#asahi-dev · 2022-01-06 13:15
kettenisyeah, that would be a reason to keep it
#asahi-dev · 2022-01-06 13:03
kettenisnot sure if I'm going to upstream that bit though, since U-Boot doesn't really need it since I don't plan to add wifi support to u-boot
#asahi-dev · 2022-01-06 13:02
kettenismarcan: no it isn't, the smc support in u-boot has been there ever since I added PCIe support to u-boot
#asahi-dev · 2022-01-04 15:58
kettenisthat conclusion may actually be wrong
#asahi-dev · 2022-01-04 15:51
kettenis(forgotten all about that one, but my fix works there as well)
#asahi-dev · 2022-01-04 15:51
kettenisthe multifunction BCM5719 on the power9 machine I have seems to have a similar issue
#asahi-dev · 2022-01-04 15:50
kettenisI'm blaming broadcom
#asahi-dev · 2022-01-04 13:31
kettenisI think my solution is to let userland only read pci config space for devices actually probed by the kernel
#asahi-dev · 2022-01-04 13:30
kettenismanaged to reproduce it with pcidump(4) (OpenBSD's equivalent of lspci)
#asahi-dev · 2022-01-04 13:30
kettenisXorg "probes" the pci bus again through libpci (even though that's pointless on these machines)
#asahi-dev · 2022-01-04 13:29
ketteniswe noticed that starting X killed wifi
#asahi-dev · 2022-01-04 13:29
kettenisinitial probing by the kernel that is

plus ›