#
✕ nick:jannau
#asahi-dev · 2026-10-19 12:28
jannauerr, 1088 is 64 * 17
#asahi-dev · 2026-10-19 12:26
jannauthe coded size might differ. ffmpeg/mpv might want 64 pixel aligned width/height for hevc
#asahi-dev · 2026-10-19 12:24
jannauwhat resolution has the hevc video?
#asahi-dev · 2026-10-18 11:46
jannaubtw asahi "leaks" kernel objects (probably keeps them alive by holding refenrences). I'd recommend rebooting often to avoid OOM
#asahi-dev · 2026-10-18 11:43
jannausofus: on the m3 pro macbook pro? on m2 pro it does not work with USB4=y
#asahi-dev · 2026-10-18 11:41
jannaucan you try without interchange support
#asahi-dev · 2026-10-18 11:37
jannauI'm not surprised if it broke. kmsro isn't quite multi-gpu
#asahi-dev · 2026-10-18 11:34
jannaupossibly related to multi-gpu work in kwin 6.8
#asahi-dev · 2026-10-18 11:28
jannauwhere? the planes report NV12,... with MOD_LINEAR
#asahi-dev · 2026-10-18 11:13
jannauI'd look at (i2c based) tipd
#asahi-dev · 2026-10-18 11:12
jannaupossibly only on m3* though
#asahi-dev · 2026-10-18 11:11
jannausame here for connecting a mac with macos for tb. I think I tested usb3 on v7.2.10 based asai-wip
#asahi-dev · 2026-10-18 11:08
jannauI did not test usb4 after rebasing from v7.2.2 to 7.2.10 which contained tb fixes from sv_en
#asahi-dev · 2026-10-18 11:04
jannauplausible that import of external buffers (or possible export from asahi) broke
#asahi-dev · 2026-10-17 09:44
jannauyou can only use it though if the base branches of the stacked PRs are in the target repository
#asahi-dev · 2026-10-16 09:55
jannauI'd expect it's ..._APPLE_IX
#asahi-dev · 2026-10-16 09:29
jannau"ix" in the DCP is my change. "interchange" is just too long. I used it to spell layout fully instead of using "l"
#asahi-dev · 2026-10-16 07:26
jannauyes, not reproducible on 7.1
#asahi-dev · 2026-10-16 07:20
jannauyes, starting and stopping sddm/kwin_wayland increases Unevictable by 40MB, that's unexpected
#asahi-dev · 2026-10-15 22:41
jannaugpu might leak memory in asahi-wip
#asahi-dev · 2026-10-15 22:10
jannauI've seen that the upstream patch is saner, doing the same downstream
#asahi-dev · 2026-10-15 22:10
jannauwhy does dockchannel-hid use WQ_MEM_RECLAIM? why does it even use its own workqueue for interface initialization
#asahi-dev · 2026-10-15 21:41
jannauI'd really like to have critical configs in kconfig
#asahi-dev · 2026-10-15 21:40
jannauI only noticed because my mesa test script did not work
#asahi-dev · 2026-10-15 21:37
jannaudid KASAN always disable rust? I vaguely rememeber a patch to that effect
#asahi-dev · 2026-10-15 21:19
jannaunot yet, that's the next on my list
#asahi-dev · 2026-10-15 21:15
jannauI only said something because it was annoying to detect. it didn't even print reliably the issue on the serial console
#asahi-dev · 2026-10-15 21:13
jannauyes, I mostly wanted to notify people who might have built asahi wip in the last 1-2h about the fix
#asahi-dev · 2026-10-15 21:11
jannauand then I moved the single buffer check for multi-planar (which now used fb) before the fb assignment
#asahi-dev · 2026-10-15 21:10
jannauthen I updated all the code after that to use fb instead of new_plane_state->fb
#asahi-dev · 2026-10-15 21:09
jannauentirely my fault. I moved setting fb in apple_plane_atomic_check() to later in the file just before it was used in the interchange code
#asahi-dev · 2026-10-15 21:05
jannaucarelessly moving code around resulting in dereferencing an uninitialized pointer
#asahi-dev · 2026-10-15 21:04
jannaudcp was briefly broken in asahi-wip after merging / pushing the interchange support
#asahi-dev · 2026-10-15 18:12
jannauI see a deadlock on t6031 during shutdown with usb storage connected
#asahi-dev · 2026-10-15 17:39
jannauwent through now
#asahi-dev · 2026-10-15 17:39
jannaugithub git operations down for anyone else? both https fetch and ssh push are very slow and making no visible progress
#asahi-dev · 2026-10-15 17:34
jannauis everyone else seeing "WARNING: net/wireless/sme.c:1302 at __cfg80211_port_authorized+0x100/0x110 [cfg80211]" as well? caused by commit 52f8916587e7 ("wifi: brcmfmac: cfg80211: Report port_authorized for 4-way HS offload") but I don't see why
#asahi-dev · 2026-10-15 17:30
jannaubut the interrupt controller uses early initialization so it could behave odd wrt its power-domains
#asahi-dev · 2026-10-15 17:29
jannauunder which conditions? I never saw that happen on t6030 or t6031
#asahi-dev · 2026-10-15 17:02
jannauwhat issue is the missing (compared to the other SoCs) always-on on ps_aic causing?
#asahi-dev · 2026-10-15 17:01
jannauyuka: works for me under HV. can you reproduce the usb issue reliably?
#asahi-dev · 2026-10-15 11:43
jannausofus: not sure about linux firmware. we can either submit 2 binaries for each variant or it is feasible to "disable" codecs during packaging
#asahi-dev · 2026-10-15 11:30
jannauat the moment just hexedit the last 64 bytes of the firmware
#asahi-dev · 2026-10-15 11:28
jannauchanging the default supported video codecs in avd to av1 + vp9
#asahi-dev · 2026-10-15 11:28
jannauwhere I'm going to annoy almost everyone: https://github.com/AsahiLinux/avd-fw/pull/6
#asahi-dev · 2026-10-14 22:19
jannauchadmed: the interchange commit is in asahi-wip. do you have rebased branch of https://github.com/AsahiLinux/linux/pull/449 ?
#asahi-dev · 2026-10-14 20:00
jannauworks on t6030, maybe iboot reports disp already as on and so framebuffer0 can keep its power-domains
#asahi-dev · 2026-10-14 19:47
jannaut8112 has "dvfs-disp-min-state" in the adt
#asahi-dev · 2026-10-14 19:46
jannauhas anyone looked at the t8112 pmp issue?
#asahi-dev · 2026-10-14 19:46
jannaupmp fw ask for tunable-dvfs-disp-min-state which doesn't exists in the ADT. maybe it needs to be parsed from "dvfs-domain"

plus ›