#
✕ nick:maz
#asahi · 2021-12-06 10:38
mazi.e. the machine is throwaway, the data isn't.
#asahi · 2021-12-06 10:37
mazmy setup would be marginally different. I have no intention to ever trust anything but m1n1/u-boot to the internal nvme device, and I definitely want my EFI partition to live on external storage.
#asahi · 2021-12-01 17:10
mazwouter: Debian knows about EFI, and u-boot provides just enough EFI for debian to be happy.
#asahi · 2021-12-01 17:08
mazall it needs is a kernel that knows about the M1.
#asahi · 2021-12-01 17:08
mazthe debian installer works out of the box.
#asahi · 2021-11-25 13:59
mazand please report things over email. I'm not scanning this channel looking for problems!
#asahi · 2021-11-25 13:59
mazcould also be that the DT hasn't been updated to take care of the polarity.
#asahi · 2021-11-25 13:56
mazmarcan: where was that reported? 100ms is quite a long time, but I wonder if it could be that the 100us delay before releasing #PERST is too short.
#asahi · 2021-11-25 13:48
mazsven: I like that I don't have to say anything anymore. the _relaxed police is now doing its job! :D
#asahi · 2021-11-19 10:08
mazbgb: 5.16-rc1.
#asahi · 2021-11-19 09:36
mazhere's a working (and stupidly large) config: https://paste.debian.net/1220037/
#asahi · 2021-11-19 09:36
maz(give or take)
#asahi · 2021-11-19 09:36
mazotherwise, no DART support. no DART support, no PCIe. No PCIe, nothing at all.
#asahi · 2021-11-19 09:35
mazbgb: yes.
#asahi · 2021-11-19 09:29
mazyou need 16k pages, ARCH_APPLE, DART support, pinctrl, and pcie at the very least.
#asahi · 2021-11-19 09:28
mazyou'll need far more than just defconfig (which builds for 4k pages, so not much will work by default).
#asahi · 2021-11-17 16:04
mazah... with a bit of luck, this is the same 3 letter passwd that I use on everything else...
#asahi · 2021-11-17 16:02
mazmarcan: shit, I need to remember my macos password...
#asahi · 2021-11-17 15:22
mazI mean, this is already 30% faster than the best KVM/arm64 host. anything faster would just be illegal.
#asahi · 2021-11-17 15:20
mazmarcan: Debian still installs in 2.23, so no changes so far. but I'm chainloading, so I don't know if I would see it.
#asahi · 2021-11-17 14:42
mazmarcan: works perfectly, thanks a lot. now only one step away from being able to permanently use u-boot!
#asahi · 2021-11-17 14:36
mazbuilds
#asahi · 2021-11-17 14:35
mazmarcan: right. that's the 100% I'm after! ;-)
#asahi · 2021-11-17 14:34
mazmarcan: btw, is there a way to get m1n1 to do the same thing as cpu_pstates.py by default? somehow, that doesn't play well with a chainloaded u-boot.
#asahi · 2021-11-17 14:24
mazthanks!
#asahi · 2021-11-17 14:06
mazfire away!
#asahi · 2021-11-17 14:03
mazmarcan: I have a patch that allows the kernel to survive u-boot (as expected, we're missing the tickling of #PERST).
#asahi · 2021-11-17 14:02
mazmarcan: if you are confident that it does something useful and doesn't break existing stuff, I'll take it!
#asahi · 2021-11-17 13:53
mazmarcan: cool.
#asahi · 2021-11-17 13:40
mazmarcan: I'm collecting PCIe fixes. I remember that you have a patch fixing some clock stuff, so if you want me to carry it, let me know.
#asahi · 2021-11-16 14:10
mazas400: you are missing the point. I don't want to add anything. all I'm doing is 'dpkg -i mykernel.deb', and get the standard distro tools to do their job.
#asahi · 2021-11-16 14:07
mazas400: imagine you have a zoo of about 40 machine, not two of them the same. you deploy a test kernel on that many boxes. EFI+grub is the only way to do that sanely.
#asahi · 2021-11-16 14:05
mazsome of us actually need grub.
#asahi · 2021-11-16 12:30
mazbgb_: I really don't think you want to adopt my installation method... I installed a complete Debian VM on a USB device, plugged it onto the mini. The kernel gets loaded over USB as a m1n1 payload.
#asahi · 2021-11-16 12:21
mazbgb_: Debian.
#asahi · 2021-11-16 12:17
maz(or rather, I cross-compile to x86/ppc/mips when I really need it)
#asahi · 2021-11-16 12:16
mazbgb_: that's up to you. for me, this is an 'apt install' away (and I stopped cross-compiling many years ago...).
#asahi · 2021-11-16 12:02
mazperf stat*
#asahi · 2021-11-16 12:02
mazbgb_: something like "perf state -e PMUNAME/event=0xXX/ELs SOMEPROGRAM", where PMUNAME is one of apple_{ice,fire}storm_pmu, XX is the event number you want, ELs is one of {u,k,uk}, and SOMEPROGRAM the stuff you want to profile. you can count multiple events on multiple PMUs (comma separated).
#asahi · 2021-11-12 17:11
mazmkf727: np.
#asahi · 2021-11-12 17:07
mazI used the Asahi documentation. I'm not keen on reading code that comes with strings attached.
#asahi · 2021-11-12 17:05
mazPMCR0/1 are pretty well understood. that's how you enable interrupt signalling and per-counter enabling (PMCR0), as well as per EL filtering (PMCR1).
#asahi · 2021-11-12 16:36
mazmkf727: they certainly counted on all 3 for me. that's how I re'd the mask.
#asahi · 2021-11-10 13:30
mazrainbow coloured.
#asahi · 2021-11-09 18:19
mazit is possible, but still requires at least m1n1 on the internal storage.
#asahi · 2021-11-09 13:46
mazkettenis: if the SMC access can be modelled as such, I'm good with it.
#asahi · 2021-11-09 11:51
mazOK, so this needs to be modelled as a power gate, or something along those lines, I guess.
#asahi · 2021-11-09 11:49
mazmarcan: depends on what events the PCIe RC gets. if we can react to an interrupt and talk to SMC, that's fine. otherwise, we need to prod it before enabling the link...
#asahi · 2021-11-09 10:48
mazwhich is fine for USB and Ethernet, less so for wifi, and definitely bad for TB.
#asahi · 2021-11-09 10:48
mazj_ey: well, we take an interrupt, and do nothing except on the first probe.

plus ›