#
#asahi-dev · 2026-01-02 00:37
fl0_idmmh, so I updated the kernel, updated m1n1, used the device tree from integralpilot, but still get an el2h exception seemingly before the kernel loads. one thing that maybe could be different - the type is not subscriptabl error when reading the adt. I thought it wasn't that important, but if it is and integralpilot and others ffix/bypass it in a different way, then maybe this is important...
#asahi-dev · 2026-01-01 15:45
oliverGot it working, thanks
#asahi-dev · 2026-01-01 14:14
oliverjannau: drm_gem_dma_object->vaddr is null, which looks like it is not. From the documentation:
#asahi-dev · 2026-01-01 14:14
oliver space this simply returns the cached virtual address.
#asahi-dev · 2026-01-01 14:14
oliver Since the DMA buffers are already mapped into the kernel virtual address
#asahi-dev · 2026-01-01 12:07
integralpilotonly have M3 so can’t see how it’s changed
#asahi-dev · 2026-01-01 12:06
integralpilotdebugging a tricky NVMe issue now, and if anyone could send the NVMe trace (from start of trace till to first nvme_cmd_read) on M1 or M2 under hypervised XNU and hypervised Linux I would appreciate it a lot!
#asahi-dev · 2026-01-01 11:18
oliverWouldn't do it any other way
#asahi-dev · 2026-01-01 11:17
oliverprintf based debugging?
#asahi-dev · 2026-01-01 10:55
chadmedrefer to the latest progress report for a foolproof method for debugging allocation errors :P
#asahi-dev · 2026-01-01 10:55
chadmedabsolutely indispensible especially given how useless clangd still seems to be with the kernel tree
#asahi-dev · 2026-01-01 10:54
chadmedelixir my beloved
#asahi-dev · 2026-01-01 10:53
chaos_princesshttps://elixir.bootlin.com/ - invaluable tool for that
#asahi-dev · 2026-01-01 10:53
chaos_princessat this point just follow all code paths that could result in an enomem and see which one you hit
#asahi-dev · 2026-01-01 10:52
oliverjannau: not working on a tty either
#asahi-dev · 2026-01-01 10:52
chadmedi will clean up the commit messages and explain stuff better at some point
#asahi-dev · 2026-01-01 10:51
chadmedleio: it was educational for me too. i knew nothing about drm/kms two weeks ago. now i know very very slightly more than nothing!
#asahi-dev · 2026-01-01 10:50
integralpilotI’m not sure it really works though cause I’ve used it once to ask them to upload a KDK for a release they forgot to do it for and now that’s not even the current release and still no KDK
#asahi-dev · 2026-01-01 10:50
integralpilotoh interesting, that’s weird!
#asahi-dev · 2026-01-01 10:40
chaos_princessrepro steps: open partition menu, add, then remove random partition, then recovery partition becomes deletable. confirmed with a 26.something vm
#asahi-dev · 2026-01-01 10:39
chaos_princessif you are into playing with the feedback assistant, mind telling them that you can nuke the recovery partition with gui disk utility
#asahi-dev · 2026-01-01 10:35
chaos_princessioreg is reading yet-another-plist-like-format from the kernel and printing it
#asahi-dev · 2026-01-01 10:34
chaos_princessbingo
#asahi-dev · 2026-01-01 10:31
integralpilotI can’t imagine the reason it does that though? Maybe they use some standard deserialisation thing that defaults to decimal probably
#asahi-dev · 2026-01-01 10:29
integralpilottempted to open feedback assistant (they accept developer tools feedback) and put something about the ioreg memory addresses being in DECIMAL not hex cause it’s annoying but I doubt they’d actually change it
#asahi-dev · 2026-01-01 10:11
leioread through the dcp/mergable commits, very nice! Would have been educational to read what userspace is doing with the info in commit ba87883176f and there's a "makling" typo in commit 57df9f1198
#asahi-dev · 2026-01-01 09:19
jannauoliver: can you test whether the mapping works on a tty (I think you'll only see swaps when switching ttys)
#asahi-dev · 2026-01-01 09:18
jannaubut the gem objects the kms driver uses (grm_gem_dma) are already mapped so that should not fail. except maybe when using imported objects from the GPU
#asahi-dev · 2026-01-01 09:16
jannaumap into the kernel virtual address space, nothing to do with iova
#asahi-dev · 2026-01-01 09:05
chadmed_oliver: wdym? the framebuffer for every plane is already mapped since we have to pass the iova to dcp. see https://github.com/chadmed/linux/blob/f62174f90e4617d27a2ac5d22854e6efd13709e3/drivers/gpu/drm/apple/plane.c#L251
#asahi-dev · 2026-01-01 08:30
oliverI am back with dma mapping again. I tried to map the frame buffer (just for science) in the dcp submit using drm_gem_vmap, but that returns errnomem. Any idea how I can read the submitted surface in the driver?
#asahi-dev · 2026-01-01 07:28
integralpilotlooks really cool! :)
#asahi-dev · 2026-01-01 07:11
chadmedbasically just the stuff for YUV scanout and some fixes for things we were doing wrong
#asahi-dev · 2026-01-01 07:10
chadmedive pulled some of the more experimental stuff out of my dcp dev branch and have a branch with just what can be merged now http://github.com/chadmed/linux/tree/dcp/mergable
#asahi-dev · 2025-12-31 15:22
svenintegralpilot: ouch, that sucks! what i usually do is commit everything i do and push it to a private github repo if i'm not ready to share it just yet
#asahi-dev · 2025-12-31 14:48
fl0_idjannau yeah it's in context of m3 bringup. it is from around october I think. Have to get a different host machine set up anyway, then I might as well reinstall m1m1.
#asahi-dev · 2025-12-31 14:32
jannauI think this is in the context of m3 bringup. usually the 1st stage m1n1 doesn't need to be updated but it's possible that the installed version does something broken. then you have to either undo it in the chainloaded m1n1 or reinstall a fixed version
#asahi-dev · 2025-12-31 14:27
chadmedonly very very old stage1 m1n1s need updating. if your install is ~recent (last 18 months or so) you should be fine
#asahi-dev · 2025-12-31 13:58
fl0_idbtw just to rule out errors - if I chainload a new m1n1, the original m1n1 boot object doesn't matter right? or should I update that too, just in case?
#asahi-dev · 2025-12-31 13:55
fl0_idI could use a pinebook pro to build a kernel, but prob would take forever...
#asahi-dev · 2025-12-31 13:54
fl0_idkernel build fails with input output error, and most commands too... at least there were errors
#asahi-dev · 2025-12-31 13:53
LarstiQI had a .git corruption I've never seen before, only thing I could think of was some disk failure
#asahi-dev · 2025-12-31 13:53
fl0_idyeah I was like f... rly now ^^
#asahi-dev · 2025-12-31 13:53
LarstiQfl0_id: ai :/
#asahi-dev · 2025-12-31 13:53
fl0_idit even restarted, but then shortly failed again...
#asahi-dev · 2025-12-31 13:52
fl0_idintegralpilot it seems to be the season of faulty drives. let's hope my problems (ok unlikely) are also just from that, as my ssd failed while recompiling the kernel. you (and sven) are prob right that it fails before the kernel, but let's hope it was just the faulty ssd..
#asahi-dev · 2025-12-31 08:12
integralpilotanyway sorry for boring you guys with x86 RE!
#asahi-dev · 2025-12-31 08:09
integralpilottaped out CLKREQ# and WAKE# too and still no luck, guess it's happening at the link training level :(
#asahi-dev · 2025-12-31 07:34
integralpilotand i thought apple hardware was frustating at a low level lol...
#asahi-dev · 2025-12-31 07:34
integralpilotmy 2015 BIOS (even updated it to the 2017 version but didn't fix) refuses to POST with the PCIe to M.2. adapter i got so now i need to physically tape over the SMBus pins and hope it works lol. :(

plus ›