#asahi-dev · 2026-05-29 09:13
yuka(no fuses either though)#asahi-dev · 2026-05-29 09:12
yukasorry, phy-ip tunables#asahi-dev · 2026-05-29 09:12
yukaeven though they exist#asahi-dev · 2026-05-29 09:12
yukaOn A18 Pro, none of the phy-ip fuses are applied#asahi-dev · 2026-05-29 08:56
yukaAnd thanks to StanfordAppliedCyber for the traces. I was more confident to give it another try.#asahi-dev · 2026-05-29 08:55
yukaThe pcie code I had for A18 Pro since several weeks was working all along, I just didn't find the correct pwren pin#asahi-dev · 2026-05-29 08:55
yuka 01:00.0 Class 0280: 14c3:7932#asahi-dev · 2026-05-29 08:55
yuka~ # lspci#asahi-dev · 2026-05-28 21:29
yukahttps://github.com/AsahiLinux/linux/pull/504#asahi-dev · 2026-05-28 21:29
yukahttps://github.com/AsahiLinux/m1n1/pull/581#asahi-dev · 2026-05-28 21:07
yukayep, tried that#asahi-dev · 2026-05-28 21:04
yukait's very weird#asahi-dev · 2026-05-28 21:04
yukabtw the devices still don't reach their enabled state immediately after, and mac os also doesn't poll for them#asahi-dev · 2026-05-28 21:03
yukaSo Linux unconditionally clears these flags on each ps_set...?#asahi-dev · 2026-05-28 21:00
yukaI'm not yet sure#asahi-dev · 2026-05-28 21:00
yukaIt might be more appropriate to do the clearing PMGR_WAS_{CLK,PWR}GATED in "Cleaning up device states"#asahi-dev · 2026-05-28 20:57
yukaOk, changed the variant's name to T6031.#asahi-dev · 2026-05-28 20:51
yukaI have no idea how similar apcie,t6030 and apcie,t6031 are. For now I called the new variant T603X, but if I could get my hands on t6030 pcie traces I would double check if this needs to be handled separately.#asahi-dev · 2026-05-28 20:42
yukaOk. If we ignore the apcie1 on the M3 Ultra, I would put the pcie nodes in t6031-die0.dtsi#asahi-dev · 2026-05-28 20:41
yukaBelieve me, I stared at that for the last two hours#asahi-dev · 2026-05-28 20:39
yukaAlso is the apcie1 on j575d used for anything? it doesn't have any bridges#asahi-dev · 2026-05-28 20:39
yukaM3 Ultra Mac Studio#asahi-dev · 2026-05-28 20:38
yukaI don't know how to wire this up properly with the multi die stuff#asahi-dev · 2026-05-28 20:37
yukabut then it only has the dart-apcie* on die0#asahi-dev · 2026-05-28 20:37
yukaI assume this means it has an apcie on both dies?#asahi-dev · 2026-05-28 20:37
yukaj575d has apcie0@580000000 and apcie1@2580000000#asahi-dev · 2026-05-28 20:36
yukaThe regs of apcie{,0} seem to be similar / the same for all of t603{0,1,2,4}. t6030 has an apcie,t6030 compatible and one less per-port reg (which we don't use, but having it means we interpret the last reg as intr2axi in m1n1)#asahi-dev · 2026-05-28 20:34
yukavery unsure how to structure the pcie nodes in the linux dt#asahi-dev · 2026-05-28 20:33
yukaT6034#asahi-dev · 2026-05-28 20:33
yukasorry#asahi-dev · 2026-05-28 20:33
yukaOn T6040#asahi-dev · 2026-05-28 20:33
yukaI see a wifi device in linux lspci :-)#asahi-dev · 2026-05-28 17:57
yuka"pcie: PHY clock enable timed out"#asahi-dev · 2026-05-28 17:56
yukasure, but the code implies causality here#asahi-dev · 2026-05-28 17:51
yukaOn t6031 pcie, the polling of state->phy_base[0] + 0x8 happens _before_ APCIE_PHYCMN_CLK is set (unless the order got messed up in tracing?)#asahi-dev · 2026-05-28 17:19
yukamore like it clears PMGR_WAS_CLKGATED and PMGR_WAS_PWRGATED#asahi-dev · 2026-05-28 17:18
yukaSo I'm not sure PMGR_AUTO_ENABLE is involved?#asahi-dev · 2026-05-28 17:18
yukalater, it reads it back again and it has value 0x1f0003ff#asahi-dev · 2026-05-28 17:17
yukaRe pmgr, mac os reads APCIE_SYS_GP as 0x1000030f, writes it back as 0x1000000f#asahi-dev · 2026-05-28 15:23
yuka(used your switch_boot_cpu branch to get it)#asahi-dev · 2026-05-28 15:20
yuka0x48 is the other one#asahi-dev · 2026-05-28 15:17
yukadon't know from the top of my head how to get it in m1n1#asahi-dev · 2026-05-28 15:16
yukaI strongly suspect the E-Core part number is also the same as on T6031#asahi-dev · 2026-05-28 15:16
yukajannau: indeed, my boot core says CPU part: 0x49 M3 Max Everest#asahi-dev · 2026-05-28 15:03
yukaOh, cool#asahi-dev · 2026-05-28 15:02
yukaI added --cpus 4#asahi-dev · 2026-05-28 15:01
yukayes#asahi-dev · 2026-05-28 15:00
yukaregs apcie[5-7] are not accessed at all#asahi-dev · 2026-05-28 14:57
yukaas in mac os booted in the hv after a bit of fiddling and a few more tries#asahi-dev · 2026-05-28 14:57
yukaThis thing pretty much worked: https://gist.github.com/IntegralPilot/85c355124fe54e670e6d5586c7a07826