#
✕ nick:jannau
#asahi-dev · 2022-10-01 08:45
jannauI'm testing on the mini
#asahi-dev · 2022-10-01 08:42
jannausven: did you test trace_dcp.py on dcp? I get subtype 0xc0 commands with a length of 17. trace_dcp.py:321ff
#asahi-dev · 2022-10-01 08:22
jannauok
#asahi-dev · 2022-10-01 07:04
jannauI plan to work on dcp this weekend but I guess if I move the the dcpep code first to iomfb.c there will be little conflict if we work on it at the same time
#asahi-dev · 2022-10-01 07:00
jannaumost convenient way to check if everything works is /sys/kernel/debug/dri/2/state, number might differ, its 2 for me since I have vgem enabled
#asahi-dev · 2022-10-01 06:58
jannauthere's are 2 vblank related warning during initialization from drm but hdmi works
#asahi-dev · 2022-10-01 06:56
jannausven: pushed updates to my dcp branches which boots dcpext and uses disp0/dispext0 darts in a single domain
#asahi-dev · 2022-09-30 06:46
jannaulet's start with one dcpext and assing both stream 0 of the disp dart to the same domain and see if that is what we want
#asahi-dev · 2022-09-30 06:39
jannaunot sure what to do about the disp* darts. It would be convenient to share the page tables over all of them for multiple displays/overlays shared between two displays
#asahi-dev · 2022-09-29 17:47
jannauso it's preferable to merge it through asahi-soc/dt
#asahi-dev · 2022-09-29 17:47
jannauhttps://github.com/AsahiLinux/linux/commit/fffa78e31be6f95576e078e19287a541151267e7 might have minor conflicts if "dt-bindings: dma: apple,admac: Add iommus and power-domains properties" is dropped
#asahi-dev · 2022-09-29 17:35
jannaumarcan: Vinod queued "dt-bindings: dma: apple,admac: Add iommus and power-domains properties" via dmaengine, please drop it from the PR for arnd
#asahi-dev · 2022-09-28 19:22
jannauat least it's easy to check if dcp is used
#asahi-dev · 2022-09-28 19:21
jannaudecided to use/test dcp on the mbp14 and forgot about the notch
#asahi-dev · 2022-09-28 19:20
jannaudo I want my panel at the top of the screen?
#asahi-dev · 2022-09-27 07:20
jannauI tried using a cpufreq instance per die but that did not work. Could be easily my fault. I did not look into why it failed as I was doing something else
#asahi-dev · 2022-09-27 07:18
jannaufixes are here https://github.com/AsahiLinux/linux/pull/54 - the cpufreq fix is hackish for now
#asahi-dev · 2022-09-27 07:16
jannautime for sending the pull request today
#asahi-dev · 2022-09-27 07:16
jannaumarcan: except for 2 downstream patch issues no problems with asahi-6.0-rc6-1 on t8103, t6002 and t8112
#asahi-dev · 2022-09-26 22:56
jannaucrash doesn't seem to break anything. log in via Browser seems to work. Reported as annoyance: https://bugzilla.redhat.com/show_bug.cgi?id=2130009
#asahi-dev · 2022-09-26 22:54
jannauI wonder if has a switch to disable JIT or javascript globally
#asahi-dev · 2022-09-26 22:54
jannauGnome Web in Fedora is basicly rendered useless by this
#asahi-dev · 2022-09-26 22:52
jannaulol, missing BTI handling in Apple's Webkit javascript JIT compiler breaks on Linus' M2 Macbook Air wifi portal logins
#asahi-dev · 2022-09-26 21:26
jannauoh nice. 'fdt_path_offset(dt, "$alias")' works directly
#asahi-dev · 2022-09-26 19:13
jannauyou're right, it's started at boot
#asahi-dev · 2022-09-26 19:01
jannauif you connect dcpext after the system has booted there should be less pmgr noise
#asahi-dev · 2022-09-26 19:00
jannaudcpext get a hotPlug_notify_gated() CB here
#asahi-dev · 2022-09-26 18:56
jannaudifferences looks like either timing or capability related
#asahi-dev · 2022-09-26 18:55
jannauexcept for the rt_bandwidth_setup_ap data no suspicious differences before the first set_digital_out_mode
#asahi-dev · 2022-09-26 18:52
jannaubut that doesn't help either when the first modeset fails
#asahi-dev · 2022-09-26 18:51
jannauother obvious difference I see is that it calls set_digital_out_mode() twice for dcpext with different modes
#asahi-dev · 2022-09-26 18:50
jannaumatches disp0.clk[0] and dispext0.clk[0] values in pmgr's boot clocks
#asahi-dev · 2022-09-26 18:48
jannauI didn't either until now. comparing dcp and dcpext traces
#asahi-dev · 2022-09-26 18:47
jannausven: did you change the return value of ServiceRelay::sr_getClockFrequency? 533333328 for dcp and 0 for dcpext
#asahi-dev · 2022-09-26 17:41
jannausven: it's rather unlikely as those mappings are in pairs for disp and dcp. i.e. if you miss them you most likely will crash dcp
#asahi-dev · 2022-09-26 15:47
jannauI haven't yet monitored what those initial mappings are used for
#asahi-dev · 2022-09-26 15:46
jannauI think it's possible to break disp0 without crashing dcp. I would image that a modeset could fail under that circumstances
#asahi-dev · 2022-09-25 21:55
jannaudevice_link might not wotk between parent and child
#asahi-dev · 2022-09-25 21:55
jannauhow was this working at all previously? I would expect that the parent is probed before the child
#asahi-dev · 2022-09-25 21:32
jannauI guess the proper fix would be a device_link like in 8fe952efd56f64dd30d19c236a6b9c0a6a05c1b7 for dockchannel
#asahi-dev · 2022-09-25 21:29
jannauor the probe order changed
#asahi-dev · 2022-09-25 21:27
jannauof_find_device_by_node() apparently doesn't (anymore)? ensure that the device is probed
#asahi-dev · 2022-09-25 21:19
jannaunot sure, "fixed" by 'if (!piodma_dev->dma_ops) return -EPROBE_DEFER;'
#asahi-dev · 2022-09-25 20:55
jannauyes, the problem is that dma_ops is not yet since piodma attaches too late to the iommu
#asahi-dev · 2022-09-25 20:29
jannaunot sure that's a problem
#asahi-dev · 2022-09-25 20:15
jannaupiodma device ends up without dma_range_map
#asahi-dev · 2022-09-25 19:42
jannauhmm, no. what seems to happen is that we somehow end up in the !dma_capable() code path
#asahi-dev · 2022-09-25 18:56
jannauI assume it's a new build of PKGBUILDs for the asahi repo instead of the asahi-dev repo but I have no insight in arch packaging either
#asahi-dev · 2022-09-25 18:52
jannauit's manual
#asahi-dev · 2022-09-25 18:44
jannauon m1 ultra, I saw that on m1 too occasionally with DMA addr inside its physical memeory as well

plus ›