#
✕ nick:janneg
#asahi-dev · 2023-12-17 08:57
janneglet me try your config. if feasible you could try the fedora kernel. rpms are here https://download.copr.fedorainfracloud.org/results/@asahi/kernel/fedora-39-aarch64/06764550-kernel/
#asahi-dev · 2023-12-17 07:08
jannegleio: can you upload your kernel config and binaries? the error doesn't feel like it is m2 ultra specific. I want to try to reproduce it on the m2 pro mac mini
#asahi-dev · 2023-12-16 17:34
jannegusually framebuffer not mapped correctly in disp*-dart stream 0
#asahi-dev · 2023-12-16 14:23
jannegdcp has two interfaces to control the display. one simplified used by iboot/m1n1 and the full one used by macos and linux
#asahi-dev · 2023-12-16 14:13
jannegso something the dcp driver does "wrong" as 1920x1080 works with iboot. I've seen the "FIFO error interrupt" before. Unfortunately I can't remember the circumstances
#asahi-dev · 2023-12-16 13:59
janneglet's try to limit the resolution to 1920x1080: https://paste.centos.org/view/76266583
#asahi-dev · 2023-12-16 13:56
jannegI wouldn't think so as you see the same or a very similar error on the 4k monitor
#asahi-dev · 2023-12-16 13:50
jannegleio: doesn't matter, I think it reached the timeout due to the verbose logging
#asahi-dev · 2023-12-16 13:49
jannegnot really, the HV/dcp trace will look ok. the error is dcp internal and we won't get more than the syslog line
#asahi-dev · 2023-12-16 13:48
jannegleio: can you try https://paste.centos.org/view/64ee3c0d with the 6k display
#asahi-dev · 2023-12-16 13:40
jannegthis is so strange. I have basically the same log except that I don't get "FIFO error interrupt" messages
#asahi-dev · 2023-12-16 13:26
jannegbased on the 6k monitor dmesg the patch will not help
#asahi-dev · 2023-12-16 13:25
jannegprobably my fault and tabs were lost in the paste, sorry
#asahi-dev · 2023-12-16 13:19
jannegno, that "error" is for reasons beyond me expected
#asahi-dev · 2023-12-16 13:18
jannegleio: please try https://paste.debian.net/1301284/
#asahi-dev · 2023-12-16 13:17
jannegOh, just seeing something
#asahi-dev · 2023-12-16 13:13
jannegI assume still no luck with unpluging and replugging the display
#asahi-dev · 2023-12-16 13:12
jannegplease remove apple_dcp.enable_verbose_logging=1 there's nothing helpful in it and obscures just the other messages
#asahi-dev · 2023-12-16 13:11
jannegso the same "AppleDCPDPTXController::handleInterrupt(): FIFO error interrupt COMMON_INT_STA_3=0x00000010" issue as with dcpext5. I suspect this might a display specific with the current dcp implementation
#asahi-dev · 2023-12-16 13:03
jannegok, you could use the installer's upgrade m1n1 routine if you replace m1n1.bin in the installer's temp directory after it has started
#asahi-dev · 2023-12-16 12:58
janneglet me chack how the installer's upgrade routine works
#asahi-dev · 2023-12-16 12:58
jannegyou have to append options from where to chainload the 2nd stage
#asahi-dev · 2023-12-16 12:55
janneghow did you install the 1st m1n1?
#asahi-dev · 2023-12-16 12:54
jannegcan you build a 2nd stage m1n1 without payload?
#asahi-dev · 2023-12-16 12:53
jannegor asahi-sources 6.6-p12 from the overlay
#asahi-dev · 2023-12-16 12:49
jannegyou don't need to build an entire new kernel, `make dtbs` is enough
#asahi-dev · 2023-12-16 12:49
jannegleio: yes, this is instead of the dcpext5 changes in kernel and m1n1
#asahi-dev · 2023-12-16 12:45
jannegleio: not sure, I'd expect it to work but I'm not sure what your 2nd stage does
#asahi-dev · 2023-12-16 08:18
jannegleio: please test https://github.com/AsahiLinux/m1n1/pull/375 - needs to be applied to both 1st and 2nd stage m1n1
#asahi-dev · 2023-12-15 18:47
jannegso sway apparently has a red screen of death
#asahi-dev · 2023-12-15 15:09
jannegmillenialhacker: amarioguy
#asahi-dev · 2023-12-15 09:28
jannegeiln: at the bottom of https://www.webmproject.org/vp9/levels/ are vp9 performance test streams. not as nice as the h264/h265 conformance suite
#asahi-dev · 2023-12-15 08:25
jannegdcp supports uploading edid so we should be able to provide a way for custom edids without anything extra
#asahi-dev · 2023-12-15 08:22
jannegmarcan: I have the edid, it reports a 3840x2160 59.996625 Hz mode
#asahi-dev · 2023-12-11 19:30
jannegcan you try adding apple_dcp.enable_verbose_logging=1 to the kernel commandline
#asahi-dev · 2023-12-11 19:29
jannegleio: yes. I supposed at the first "DPTXController.cpp:3227: AppleDCPDPTXController::handleInterrupt(): FIFO error interrupt COMMON_INT_STA_3=0x00000010" the display is broken
#asahi-dev · 2023-12-11 18:59
jannegleio: yes, edid from a different computer is good. thanks
#asahi-dev · 2023-12-11 18:47
jannegwhich monitor was this? have you tried a different one?
#asahi-dev · 2023-12-11 18:47
jannegleio: that looks like it should have worked at 4.444890. any idea why it could select 4k30 as mode? do you happen to have the edid for this display?
#asahi-dev · 2023-12-11 00:38
jannegleio: if you got a moment please check if https://github.com/jannau/m1n1/tree/t6022_use_dcpext5 and https://github.com/jannau/linux/tree/dcp/dptx-fixes-12-10 result in working HDMI on the M2 Ultra Studio
#asahi-dev · 2023-12-09 11:33
jannegaustriancoder: povik does. see https://github.com/AsahiLinux/linux/pull/151 but this is probably not the way will end up using. it circumvents one of the embedded co-processors and there doubts if that is a good idea
#asahi-dev · 2023-12-05 12:39
janneglast good tag is v1.3.2
#asahi-dev · 2023-12-05 12:39
jannegHV usb breakage bisected to 7f94c7f "tools.chainload.py: Fixes for 13.5", I don't see immediately how though
#asahi-dev · 2023-12-04 23:21
jannegsven: we seem to have broken USB (hotplug) in the HV. macos still talks to the HW but on hotplug nothing happens (mac mini m1,m2,m2 pro). I think that was still working last week on j314
#asahi-dev · 2023-12-03 15:36
jannegkettenis: I suppose marcan will update the u-boot/uefi-only binary when M2 desktops are switched to supported
#asahi-dev · 2023-12-02 12:19
jannegtracks for n in [5, 7] as well in macos traces
#asahi-dev · 2023-12-02 12:06
jannegsven: https://gist.github.com/jannau/7a6e5c25c2c356a05a0b4b40226c1961
#asahi-dev · 2023-12-02 11:29
jannegyes, LN_TXA_CAL_CTRL and LN_TX_CAL_CODE are based on the same value for my traces
#asahi-dev · 2023-12-02 11:09
jannegyes, I have t6020 macos traces with PMA_TXA_CAL_CTRL=0xf on LN1_AUSPMA_TX_SHM_TXA_IMP_REG0 (does it depend on the type-c orientation which lane pair is used for DP?)
#asahi-dev · 2023-12-02 11:02
jannegfor LN1_AUSPMA_TX_SHM_TXA_IMP_REG0 I see 0xf

plus ›