#asahi-dev · 2026-06-14 11:38
nickchanah, I see it's replacing wfi with nop#asahi-dev · 2026-06-14 11:29
NickChan[m]so M1+ appears to be the right call #asahi-dev · 2026-06-14 11:28
NickChan[m]technically applies to older socs too, but on them the deep wfi loses all state, so it's pointless to try to make force power down mode this way#asahi-dev · 2026-06-14 11:10
yukaI will run the wfi experiment once more on my M1#asahi-dev · 2026-06-14 11:04
yukahttps://cyberchaos.dev/yuka/linux/-/commit/5f9907223013a7fad6ae089bf0212295c041e2fe#asahi-dev · 2026-06-14 11:04
yukaThx#asahi-dev · 2026-06-14 10:35
yukathx#asahi-dev · 2026-06-14 10:34
NickChan[m]"Efficiency cores" are still, well, e-Core#asahi-dev · 2026-06-14 10:34
yukaYes I already noticed that in the adts#asahi-dev · 2026-06-14 10:34
NickChan[m]and the "Super Cores" are p-Cores in the header#asahi-dev · 2026-06-14 10:34
NickChan[m]Note the "new performance cores" in M5 Pro/Max is m-Core in that header#asahi-dev · 2026-06-14 10:31
NickChan[m]yuka: download xcode 27 or xcode 27 command line tools, nagivate to the MacOSX.sdk, then look at System/Library/Frameworks/Kernel.framework/Versions/A/Headers/arm/cpuid.h#asahi-dev · 2026-06-14 10:31
jannauyuka: does the kernel boot with "nohlt" in the kernel command line#asahi-dev · 2026-06-14 10:29
jannauonly the ones mentioned in Documentation/arch/arm64/cpu-feature-registers.rst. I can add the remaining ones but those are half 0 anyway#asahi-dev · 2026-06-14 10:28
jannaumaz: A18 Pro feature registers: https://gist.github.com/jannau/130b792789647554405dff506ba3cc56#asahi-dev · 2026-06-14 10:18
yukaI'm strongly guessing M4* and M5* are all affected. Is there some other place I could find the MIDR part numbers?#asahi-dev · 2026-06-14 10:13
yukaDoes anyone have an M4 Max and M5* to pull the MIDR part no from and run the WFI experiment?#asahi-dev · 2026-06-14 10:00
yukaNope, wfe does not work, not even for booting a tiny kernel. It simply gets stuck and never wakes up again once the secondaries are initialized.#asahi-dev · 2026-06-14 09:45
yukajust not because of losing state#asahi-dev · 2026-06-14 09:45
yukaAnd btw, this still does not fully make M4 boot reliably, it still crashes#asahi-dev · 2026-06-14 09:45
chaos_princessoh#asahi-dev · 2026-06-14 09:45
yukaI mean, I can try it. It led to problems in the past.#asahi-dev · 2026-06-14 09:43
chaos_princessput a wfe there#asahi-dev · 2026-06-14 09:43
yukahttps://cyberchaos.dev/yuka/linux/-/commit/481caf9cb9c1a114056cb1833958713dbd5bf86d#asahi-dev · 2026-06-14 09:36
svenit’s effectively a cpu errata :/#asahi-dev · 2026-06-14 09:36
svenfwiw, we essentially got confirmation that wfi losing state on m4+ is expected because that cyc_ovrd thing we set just doesn’t exist anymore #asahi-dev · 2026-06-14 06:44
yukaThere is some other race condition on M4 which causes the secondaries to crash with odd (seemingly random) stack traces#asahi-dev · 2026-06-14 06:44
yukaThis CYC_OVRD setting makes the M3 cores intentionally lose state on WFI, and this way I was able to confirm that replacing wfi and wfit with nops works!#asahi-dev · 2026-06-14 06:43
yukaI now have a clean environment to test patches for WFI losing state in, by running it on M3 and setting CYC_OVRD_WFI_MODE(0) or CYC_OVRD_WFI_MODE(3)#asahi-dev · 2026-06-14 06:41
yukaThere is something other than WFI losing state causing linux crashes with smp enabled on M4+#asahi-dev · 2026-06-14 05:49
StanfordAppliedCyber[m]yep, i did that thing where i load xnu, don't block the cpustart write, see if xnu can start its own cores, and no, xnu panics with the core didn't start message, nothing in my spin table#asahi-dev · 2026-06-14 05:32
StanfordAppliedCyber[m]but i doubt it'll work#asahi-dev · 2026-06-14 05:32
StanfordAppliedCyber[m]it's the only debugging step i haven't done#asahi-dev · 2026-06-14 05:32
StanfordAppliedCyber[m]yall think it's worth it to do a physical battery pull of this thing?#asahi-dev · 2026-06-14 03:18
StanfordAppliedCyber[m]can someone send me the full sysreg dump from a neo? specifically one where smp works#asahi-dev · 2026-06-14 02:48
StanfordAppliedCyber[m]also apparently iboot is identical between 26.5 and 26.5.1#asahi-dev · 2026-06-14 02:40
StanfordAppliedCyber[m]i've done the most work on the neo but no one else hit as many problems as i did, everyone else i talked to had everything just work#asahi-dev · 2026-06-14 02:39
StanfordAppliedCyber[m]i mean, possibly? but why then does everyone else's device work, but mine doesn't#asahi-dev · 2026-06-14 02:34
readyee[m]Do you think the chickens could be doing something with it?#asahi-dev · 2026-06-14 02:33
StanfordAppliedCyber[m]and from inside the shell running p.smp_start_secondaries() and proceed to see a wall of fail#asahi-dev · 2026-06-14 02:33
StanfordAppliedCyber[m]also for context all i'm doing is macvdmtool reboot serial, waiting a few seconds, then M1N1DEVICE=/dev/cu.debug-console python3 proxyclient/tools/shell.py, with the only patch being setting the baud to 115200 because anything higher breaks#asahi-dev · 2026-06-14 02:25
StanfordAppliedCyber[m]everything looks totally normal as far as i can tell#asahi-dev · 2026-06-14 02:24
StanfordAppliedCyber[m]ofc#asahi-dev · 2026-06-14 02:24
readyee[m]In the Shell.PY while the system booting up it tells you like what core you’re booted on what firmware you’re on did you double check that is the system and firmware is a 26.5.1#asahi-dev · 2026-06-14 02:21
StanfordAppliedCyber[m]no#asahi-dev · 2026-06-14 02:21
readyee[m]By any chance did you enable CPU frequency before SMP?#asahi-dev · 2026-06-14 02:19
StanfordAppliedCyber[m]then smp just.... doesn't work, the pstate is correct but the cores just never drop out of reset, same as 26.4.1#asahi-dev · 2026-06-14 02:19
StanfordAppliedCyber[m]i reboot into it, do a bunch of stuff, read logs, etc, then reflash m1n1#asahi-dev · 2026-06-14 02:18
StanfordAppliedCyber[m]it only broke yesterday after i install the regular macos kernelcache#asahi-dev · 2026-06-14 02:18
StanfordAppliedCyber[m]actually does anyone have ideas what underlying thing could be wrong with my machine? on 26.5, the moment i installed it then smp worked in raw mode, and it kept working for a while