#
✕ nick:marcan
#asahi-dev · 2021-05-15 08:07
marcanyou need to reduce boot security first at least
#asahi-dev · 2021-05-15 08:07
marcansame way you can't "pop a USB stick in and everything goes brrr" on Intel Macs
#asahi-dev · 2021-05-15 08:06
marcanit's like having only the recovery portion of UEFI
#asahi-dev · 2021-05-15 08:06
marcanthe machines are not fully functional without 1TR
#asahi-dev · 2021-05-15 08:06
marcanthe *boot menu* is 1TR
#asahi-dev · 2021-05-15 08:06
marcanno, you really need to think of it as device firmware
#asahi-dev · 2021-05-15 08:05
marcanso this is, in context, like UEFI firmware on a PC
#asahi-dev · 2021-05-15 08:05
marcanyou need recovery mode, which is macos, but should be considered part of machine firmware
#asahi-dev · 2021-05-15 08:05
marcanwhich means booting into 1TR and entering user credentials
#asahi-dev · 2021-05-15 08:05
marcanupdating it requires physical attestation
#asahi-dev · 2021-05-15 08:05
marcanthis is due to the machine's secureboot design
#asahi-dev · 2021-05-15 08:05
marcanso you can never ship it as a package
#asahi-dev · 2021-05-15 08:05
marcanit's also worth noting that you cannot, and never will be able to, update m1n1 from Linux itself
#asahi-dev · 2021-05-15 08:05
marcan*bootloader
#asahi-dev · 2021-05-15 08:04
marcanso in that sense you could think of m1n1 as a distro-agnostic
#asahi-dev · 2021-05-15 08:04
marcanso our goal with our installer is that people who want to install Asahi can pick that, and people who want to use another distro can just install the m1n1+u-boot part and then boot a standard installer from that
#asahi-dev · 2021-05-15 08:04
marcanu-boot provides the UEFI environment, and will be embedded into m1n1 as an in-line payload
#asahi-dev · 2021-05-15 08:03
marcanthe second stage will be u-boot, and the third stage will be up to the distro but presumably many will use GRUB
#asahi-dev · 2021-05-15 08:03
marcanit's the first stage
#asahi-dev · 2021-05-15 08:03
marcanyes and no :-)
#asahi-dev · 2021-05-15 08:03
marcanthe thing is, you don't care about m1n1
#asahi-dev · 2021-05-15 08:03
marcannot really, because m1n1 doesn't have disk drivers so it can't load anything else
#asahi-dev · 2021-05-15 08:02
marcanthere are other blobs, font etc
#asahi-dev · 2021-05-15 08:02
marcanoh IMG is the logo, yes
#asahi-dev · 2021-05-15 08:01
marcan"image"?
#asahi-dev · 2021-05-15 08:01
marcanI'll be doing our equivalent for Arch Linux ARM (which is what asahi will be, just Arch with an extra repo and some installer glue)
#asahi-dev · 2021-05-15 08:01
marcanyeah, if you have access to aarch64 infra then that's no problem
#asahi-dev · 2021-05-15 08:00
marcanwell you need to at least build for ARM, which probably would best be done natively on ARM machines, but it could be any ARM64 box
#asahi-dev · 2021-05-15 08:00
marcanright *now* though there isn't much to do other than perhaps investigate whatever infra that would require on your side
#asahi-dev · 2021-05-15 08:00
marcanif you're interested in Fedora support and you really want to be in the bleeding edge, I would suggest bringing up and maintaining an RPM repo with our bleeding-edge package forks as they become available. That will get you support sooner than upstream (though hopefully not by much, since we want to keep the upstreaming pipeline going)
#asahi-dev · 2021-05-15 07:58
marcangoing to start a stream in fact
#asahi-dev · 2021-05-15 07:57
marcanI'm working on those right now :-)
#asahi-dev · 2021-05-15 07:57
marcanthere is no GPU kernel driver either, nor even a design for its interface
#asahi-dev · 2021-05-15 07:56
marcan5.13 only has basic CPU bring-up, no drivers
#asahi-dev · 2021-05-15 07:56
marcanlots
#asahi-dev · 2021-05-15 07:54
marcanthank you! :)
#asahi-dev · 2021-05-15 06:48
marcananyway, you all know where our website and twitter account are, so I'm not particularly concerned about contingency plans here; let's wait and watch for a bit, it won't be excessively disruptive to move if we have to
#asahi-dev · 2021-05-15 06:46
marcanI don't know what's going on, but clearly whatever *is* going on isn't healthy
#asahi-dev · 2021-05-15 06:45
marcanhttp://techrights.org/2021/05/14/how-freenode-works/ yay drama
#asahi-dev · 2021-05-15 06:41
marcannow I'm confused
#asahi-dev · 2021-05-15 06:41
marcanI thought this was all coming from the PIA acquisition some time back
#asahi-dev · 2021-05-15 06:41
marcanoh huh, so tomaw from OFTC is involved somehow?
#asahi-dev · 2021-05-15 01:22
marcanlet's wait and see how this plays out for now, but moving to OFTC would be an option yes
#asahi-dev · 2021-05-04 18:36
marcanhttps://mrcn.st/p/LCVn7jMf <- basic virtual UART (output only routed into the main CDC-ACM for now, will move it to a secondary later)
#asahi-dev · 2021-05-04 18:33
marcanlike gdbserver
#asahi-dev · 2021-05-04 18:33
marcanarnd: re gdbstub, I'll make python implement that (so you can dual wield gdb and the python debug shell), converting to my protocol
#asahi-dev · 2021-05-04 13:16
marcanhvc_dcc would work too, but I need this to work for macos so it makes sense to virtualize serial (and serial is easy anyway)
#asahi-dev · 2021-05-04 13:15
marcanright now there are no interrupts, so the HV only processes USB data when an exception happens. I'll do something about that soon, but for a m1n1 guest or just serial prints there is no need since those are triggered by the guest
#asahi-dev · 2021-05-04 13:14
marcanarnd: the HV goes over USB and the guest over serial right now, but the next thing i'm scheming is virtualizing serial into a secondary cdc-acm interface
#asahi-dev · 2021-05-04 11:00
marcanyou can also see/change context registers, e.g. ctx.regs[0] == x0 (and any changes are committed on return back to the guest)

plus ›