#
✕ nick:maz
#asahi-dev · 2023-09-10 15:02
mazmarcan: well, if you only plan to hit upstream after a few *years*, that's probably not something I should be concerned about, and you'll have to convince other people than me!
#asahi-dev · 2023-09-10 12:27
mazthat will give you the possibility to save/restore the state as long as there is an EL12 accessor.
#asahi-dev · 2023-09-10 12:25
mazwell, I'm willing to help. but I'm also not going to turn KVM upside down for that. my proposal is to allow a module to change the IMPDEF state at load/put time. for stuff such as AIDR, we could either expose it as writable to userspace, or trap it and forward that trap to userspace.
#asahi-dev · 2023-09-10 12:19
mazCPU*
#asahi-dev · 2023-09-10 12:19
mazshow me the spec of that COPU, and we'll talk! :D
#asahi-dev · 2023-09-10 12:18
mazwell, that's IMPDEF... so any behaviour is compliant!
#asahi-dev · 2023-09-10 12:17
mazyup.
#asahi-dev · 2023-09-10 12:17
mazKVM just exposes the host view. nothing else.
#asahi-dev · 2023-09-10 12:16
mazno, this can be overloaded with VMIDR_EL2.
#asahi-dev · 2023-09-10 12:15
mazKVM has no provision to cope with asymetric systems.
#asahi-dev · 2023-09-10 12:15
mazMIDR values as seen as the host's, directly from KVM. which means that if your vcpu thread migrates from one type to another, you see it.
#asahi-dev · 2023-09-10 12:13
mazas long as there is an _EL12/EL02 accessor
#asahi-dev · 2023-09-10 12:13
mazfor IMPDEF regs, you could disable TIDCP, and context switch that in the same location.
#asahi-dev · 2023-09-10 12:10
maz(well, migrating an apple guest to anything else is fscked for plenty of reasons)
#asahi-dev · 2023-09-10 12:10
mazbut this still breaks a lot of things: migration of such a guest is fscked.
#asahi-dev · 2023-09-10 12:09
mazquite.
#asahi-dev · 2023-09-10 12:09
mazthat's the distros problem.
#asahi-dev · 2023-09-10 12:09
mazand indeed, I don't plan to have a stable ABI.
#asahi-dev · 2023-09-10 12:08
mazthey've been around for 30 years.
#asahi-dev · 2023-09-10 12:06
mazif you don't need the dynamic flip of TSO inside the guest, then it 's pretty easy to do, and we could probably stick that in a module -- just hook whatever you need to do in the arch-specific vcpu_load/vcpu_put helpers (the module interface itself is to be created).
#asahi-dev · 2023-09-10 10:58
maz... I'd be interested in reviewing it.
#asahi-dev · 2023-09-10 10:57
mazmarcan: the one thing I don't want to create is some new form of ABI at the userspace or hypercall level upstream. I know you went down that way in the Asahi tree, and that's fine by me as long as you keep it there. *how* you make it work is interesting though. If you have a bit of a spec/write-up that describes the various controls in the AUX regs at both ELs.
#asahi-dev · 2023-09-08 17:34
mazyup. the switches (which are absolutely tiny) are either off by about 1mm, or have some bad bridges. the only change I made in v3.1 was to pick hummer-sized ICs, and it's been pretty solid since.
#asahi-dev · 2023-09-08 17:18
mazI've been secretly thinking of buying a hot-plate to rework the handful of v3 that were butchered by jlcpcb... total nonsense from a cost perspective, but still...
#asahi-dev · 2023-09-08 16:52
mazsven: I know a few mad people... :D
#asahi-dev · 2023-09-08 16:46
mazkevans91: did you get a stencil as well, or just the boards?
#asahi-dev · 2023-09-08 16:28
mazkevans91: don't throw them away -- if someone asks, we can always direct them to you.
#asahi-dev · 2023-08-30 09:31
mazsven: quite possibly. v3.2 is using a good old trusted full fat Molex, which is a bit of a pain to route. but at least, you can it anywhere.
#asahi-dev · 2023-08-30 08:54
mazkevans91: yeah, I switched a couple of things for that exact reason -- the connectors were not easily sourced outside of the LCSC catalog.
#asahi-dev · 2023-08-30 07:07
mazkevans91: functional enough for me! it really is v3.1 with minor changes that I consider forward looking, and there is no SW-visible difference between any of the v3 revisions. if you're building it, let me know how it fares!
#asahi-dev · 2023-08-29 13:08
mazyeah, it's been a long time coming. good stuff.
#asahi-dev · 2023-08-23 08:56
mazyeah, that's what I have used too. I'll be doing another run with Molex connectors to see if that makes a difference.
#asahi-dev · 2023-08-23 08:55
mazhappy birthday to my M1 Ultra!
#asahi-dev · 2023-08-23 08:54
maz 09:53:38 up 371 days, 16:50, 6 users, load average: 0.00, 0.05, 0.08
#asahi-dev · 2023-08-23 08:54
mazHuh, just noticed:
#asahi-dev · 2023-08-23 08:54
mazsven: shitty connectors? I'm so surprised... :-(
#asahi-dev · 2023-08-21 15:25
mazwants "Green Professional"...
#asahi-dev · 2023-08-21 15:20
mazvery impressive.
#asahi-dev · 2023-08-20 15:32
mazI also don't think 'professionalism' applies to any of this. nobody has a contractual obligation to answer anything.
#asahi-dev · 2023-08-20 15:00
mazmarcan: make sure you checkout the right tag,
#asahi-dev · 2023-08-20 15:00
mazno.
#asahi-dev · 2023-08-20 14:59
mazbut if a v1 fails (with more or less nothing on the signal path), I'm a bit worried for v3+, which do have switches on that signal.
#asahi-dev · 2023-08-20 14:57
mazmarcan: nah, I keep them to myself! ;-)
#asahi-dev · 2023-08-20 14:57
mazI think only kettenis has ever seen a v0.
#asahi-dev · 2023-08-20 14:56
mazv1
#asahi-dev · 2023-08-20 14:53
mazsven: that. or try to use the test points that are right by the micro-USB connector, and see if you get something there.
#asahi-dev · 2023-08-20 14:34
mazsven: at some point, it'd be good to understand what part is screwing up the signal. I wouldn't be surprised if the connectors themselves acted as capacitors if they were one way or another oxidised.
#asahi-dev · 2023-08-20 10:12
mazyup. ${job-3} had a huge f'in scope capable of capturing an eye diagram, and all of a sudden I miss that environment...
#asahi-dev · 2023-08-20 10:08
mazI guess going with a full ground plane (which requires 4 layer board) would help.
#asahi-dev · 2023-08-20 10:08
mazyeah, that's a bit worrying. I've bean thinking of moving the uUSB connector much closer to the USB-C, but that's hard (specially with the newer versions that actively switch the D+/D- pair between serial and USB).

plus ›