#
✕ nick:sven
#asahi-dev · 2026-01-23 19:45
svenI think when this came up the first time I looked at XNU and also didn't see it access SYS_APL_CYC_OVRD anymore
#asahi-dev · 2026-01-23 19:45
svenwe only needed the special one because we wanted the deeper wfi state that drops register state
#asahi-dev · 2026-01-23 19:44
sveniirc there's generic code (or a generic driver) that uses wfi
#asahi-dev · 2026-01-23 19:43
svenyeah, but no need for a special driver then (assuming wfi really keeps state now)
#asahi-dev · 2026-01-23 19:42
svenyes, that changes wfi behavior
#asahi-dev · 2026-01-23 19:42
sventhough i guess if linux runs for a while and doesn't run into weird crashes that's proof enough that it keeps state :D
#asahi-dev · 2026-01-23 19:41
sveni'd still double check if wfi truly never loses state on these machines, that aidr_el1 is impdef iirc. i don't expect them to change it but who knows
#asahi-dev · 2026-01-23 19:39
svenwhich is a problem because we can't implement psci on these machines because there's no EL3 or similar
#asahi-dev · 2026-01-23 19:38
svenarm64 doesn't accept custom cpuidle drivers, they want everything to go through PSCI
#asahi-dev · 2026-01-23 19:38
svenit can never be upstreamed
#asahi-dev · 2026-01-23 19:36
svenyuka: cpuidle can't be upstreamed anyway, does wfi retain full register state on your machine?
#asahi-dev · 2026-01-23 15:33
svenwe don’t really know how pmgr works there I think
#asahi-dev · 2026-01-23 15:33
svenit won’t work on m4 fwiw, iirc there are missing references in the adt
#asahi-dev · 2026-01-23 14:37
sven*install
#asahi-dev · 2026-01-23 14:37
sveni'll see if i can still that one on a third partition on my mac mini
#asahi-dev · 2026-01-23 14:36
svenso i guess we need a trace of macos >15 booting and see if it does anything different with nvme
#asahi-dev · 2026-01-23 14:33
sven:(
#asahi-dev · 2026-01-23 14:31
sventhat's gonna be very annoying to deal with
#asahi-dev · 2026-01-23 14:31
svenugh
#asahi-dev · 2026-01-23 14:29
svenah
#asahi-dev · 2026-01-23 14:29
svenhmmmm okay, weird
#asahi-dev · 2026-01-23 14:27
sveniirc someone mentioned that it breaks with 15+ (or was it 26+)?
#asahi-dev · 2026-01-23 14:26
svenwhich macos are you on? (the one where you replaced xnu with m1n1)
#asahi-dev · 2026-01-23 14:26
svencould be a different firmware version
#asahi-dev · 2026-01-23 14:23
svensounds like nvme works for you :D
#asahi-dev · 2026-01-23 14:11
svennoopwafel: sounds like some dependency of mbox (power domain maybe?) hasn't probed yet
#asahi-dev · 2026-01-23 14:11
svenkeep in mind that apple likes to lie in the ADT. so even though they claim aicv3 it may actually be aicv3.5 or aicv3 but soc-specific #ifdefs in the driver
#asahi-dev · 2026-01-23 13:24
svenand the nvme node ofc
#asahi-dev · 2026-01-23 13:24
svenyou need the pmgr nodes in the device tree, ./proxyclient/tools/pmgr_adt2dt.py should generate them
#asahi-dev · 2026-01-23 13:19
svenhave you checked if the rtkit sequence is the same? does it use the same target power state, etc?
#asahi-dev · 2026-01-23 13:16
svenweird
#asahi-dev · 2026-01-23 13:08
svenor maybe some additional power domain is missing?
#asahi-dev · 2026-01-23 13:07
svenmaybe we're not waiting until the co-processor is fully up or something
#asahi-dev · 2026-01-23 13:07
sventhat's what i meant to write :D
#asahi-dev · 2026-01-23 13:07
svenso it just wrote to that chicken but on t6030 there and it worked?
#asahi-dev · 2026-01-23 13:06
svenerm.
#asahi-dev · 2026-01-23 13:06
sven*writes
#asahi-dev · 2026-01-23 13:06
svenso it just write to that chicken bit on t6030 and it just wrote to that chicken bit there? huh
#asahi-dev · 2026-01-23 13:05
svenintegralpilot: on M1/M2, or have you traced M3 as well now?
#asahi-dev · 2026-01-23 13:05
svennoopwafel: if the hv is working running trace_nvme.py with verbose=3 until the first command would be helpful
#asahi-dev · 2026-01-23 12:58
sveneasiest way to achieve that is by disabling module support in general
#asahi-dev · 2026-01-23 12:58
svenduring early bringup I generally compile everything as built-in fwiw
#asahi-dev · 2026-01-23 12:38
svenhuh, kinda surprised just earlycon isn’t enough
#asahi-dev · 2026-01-22 09:40
sventhe PR looks like a good step towards m4 support to me then :)
#asahi-dev · 2026-01-22 08:59
svenhuh, i think it already tries to do that unless i'm reading the code wrong
#asahi-dev · 2026-01-22 08:57
sveni guess we could also adjust chainload.py to just move the next stage to the same base address the previous one was
#asahi-dev · 2026-01-22 08:48
sventrue, we could just put a "branch to current stage entry" stub there on chainload
#asahi-dev · 2026-01-22 08:43
svenif they’re locked by iboot we probably need to keep a page at rvbar „forever“ that catches them and forwards them to the currently running m1n1
#asahi-dev · 2026-01-22 08:42
sventhe cores come up at least? Nice, last time someone tried they couldn’t even get them up
#asahi-dev · 2026-01-22 07:39
svenalso sounds good to me

plus ›