#
#asahi-dev · 2026-10-07 10:17
JamesCalligeros[m]Nope that kills the whole system lol
#asahi-dev · 2026-10-07 10:15
chadmedone sec
#asahi-dev · 2026-10-07 10:15
MichaelReeves[m]p3s_4k=on
#asahi-dev · 2026-10-07 10:15
chadmedwhich one is that
#asahi-dev · 2026-10-07 10:14
MichaelReeves[m]if you enable the force 4k kernel flag does it fix it?
#asahi-dev · 2026-10-07 10:14
chadmedi assume because the host wine process is being mapped as 16k
#asahi-dev · 2026-10-07 10:14
chaos_princessatcphy - as soon as the dp lands in our tree, i intend to send it out. spmi - ¯\_(ツ)_/¯ - up to the kernel maintainers to do their thing, spmi is 6 days since last resend, tipd is 27 since last resend
#asahi-dev · 2026-10-07 10:14
chadmedfex in wine does not work
#asahi-dev · 2026-10-07 10:12
yukachaos_princess: what is the upstream status of the M3 atcphy & spmi patches? Would a Tested-by from M3 Max help?
#asahi-dev · 2026-10-07 10:11
yukaalso the M3 USB4 support stuff breaks the kboot_atc setup on M4
#asahi-dev · 2026-10-07 10:11
MichaelReeves[m]oh cool
#asahi-dev · 2026-10-07 10:11
MichaelReeves[m]thats ok nvm, its rly low priority compared to the other stuff
#asahi-dev · 2026-10-07 10:11
yukain other news I figured out why one of the usb ports was not working on M4. was a nasty typo in the device tree.
#asahi-dev · 2026-10-07 10:11
yukadon't think I will be able to do that today
#asahi-dev · 2026-10-07 10:09
MichaelReeves[m]I think WIP is only 3 commits ahead of that
#asahi-dev · 2026-10-07 10:09
MichaelReeves[m]chadmed: its based on asahi-7.1.13-3 tag
#asahi-dev · 2026-10-07 10:09
MichaelReeves[m]WTS is the program I made to test 4K edge cases automatically, https://github.com/IntegralPilot/wts
#asahi-dev · 2026-10-07 10:08
MichaelReeves[m]s/it/I - I keep making that typo idk why
#asahi-dev · 2026-10-07 10:08
MichaelReeves[m]yuka: sorry I should have been clearer, I meant while you've got it M4 out if you could try to see if its reproducible or if it forgot needed bits or overfitted to fedora DL (and in would need changes), I don't expect anything specifically different but I guess with apple you never know
#asahi-dev · 2026-10-07 10:07
tryhard[m]surprising, mine is 7.3gb and i didn't do as much
#asahi-dev · 2026-10-07 09:56
chadmedmy kernel .git is over 10gb..
#asahi-dev · 2026-10-07 09:55
chadmedalso is the 7.1 rebase on top of asahi-wip?
#asahi-dev · 2026-10-07 09:54
yukawhat is WTS?
#asahi-dev · 2026-10-07 09:54
yukaI don't see why it would need any changes?
#asahi-dev · 2026-10-07 09:54
chadmedits all architected arm features, i would be very very very very very shocked if anything needs changing for a specific soc
#asahi-dev · 2026-10-07 09:53
MichaelReeves[m]If you’ve got time, could you also try my 4K branch and WTS on M4, I wonder if it works or if anything needs changing for that?
#asahi-dev · 2026-10-07 09:50
MichaelReeves[m]Thank you also for replying to Rob for me last night after I went to bed
#asahi-dev · 2026-10-07 09:50
MichaelReeves[m](DT)
#asahi-dev · 2026-10-07 09:50
MichaelReeves[m]I will get upstream dchid ready for you also
#asahi-dev · 2026-10-07 09:49
MichaelReeves[m]awesome, thank you!!
#asahi-dev · 2026-10-07 08:51
yukalet me see if I can make M5 Pro work as well
#asahi-dev · 2026-10-07 08:50
yukahttps://yuyu.yuka.dev/public/m4-kvm.png
#asahi-dev · 2026-10-07 08:50
yukaMichaelReeves[m] the kvm diff I last sent also works on M4
#asahi-dev · 2026-10-07 07:14
mmediouni[m]But apple doesn't do that afaik - at least not on any released product
#asahi-dev · 2026-10-07 07:13
mmediouni[m]However the vendor can make a SoC where 4KiB pages are only supported at stage1 and stage2 has to use 16KiB pages
#asahi-dev · 2026-10-07 07:13
mmediouni[m]*used/not used
#asahi-dev · 2026-10-07 07:13
mmediouni[m]If stage1 translation supports 4KB pages, it supports it both when stage2 is supported or not supported
#asahi-dev · 2026-10-07 07:12
mmediouni[m]That doesn't change things
#asahi-dev · 2026-10-07 07:11
NickChan[m]<MichaelReeves[m]> "they'd probably just say QEMU/..." <- this was true but in more recent versions of the architecture there are seperate feature reg for stage1 and stage2 translation granule supoort
#asahi-dev · 2026-10-07 06:43
linkmauveThat’s what muvm is, yeah.
#asahi-dev · 2026-10-07 06:41
MichaelReeves[m]they'd probably just say QEMU/KVM can be used
#asahi-dev · 2026-10-07 06:33
linkmauveYou have the opposite argument, that it will allow us devs who previously couldn’t try 16k or 64k page sizes due to using some programs requiring 4k to test them now.
#asahi-dev · 2026-10-07 06:22
JamesCalligeros[m]Namely that this will be used as a crutch by lazy devs to not bother making their software work properly on non-4k systems
#asahi-dev · 2026-10-07 06:21
JamesCalligeros[m]My biggest concern re upstreaming would be arch folks using the same excuse they used to nack tso
#asahi-dev · 2026-10-07 06:20
MichaelReeves[m]oh also forgot to say but if you want to force everything into 4K like I did before to test, its p3s_4k=on in kernel command line (you can run WTS with or without this)
#asahi-dev · 2026-10-07 06:19
MichaelReeves[m]oh cool!
#asahi-dev · 2026-10-07 06:15
JamesCalligeros[m]Already being taken care of
#asahi-dev · 2026-10-07 06:15
MichaelReeves[m]by the way, should we (or I) contact Google? it seems the relevant person would be Kalesh
#asahi-dev · 2026-10-07 05:48
MichaelReeves[m]for the kernel, you need CONFIG_ARM64_PER_PROCESS_PAGE_SIZE=y
#asahi-dev · 2026-10-07 05:46
MichaelReeves[m]executable here https://github.com/IntegralPilot/wts/releases/tag/main-21e53d1ac2e20ba1a01cfc8073c04bad1c361415

plus ›