#asahi · 2021-03-14 02:03
marcanarnd: m1n1 is already kind-of a HV :-) (it can already run "guests" in EL1)#asahi · 2021-03-13 17:00
marcanwasn't expecting the *progress report* to end up on El Reg of all things#asahi · 2021-03-13 17:00
marcanhttps://www.theregister.com/2021/03/13/asahi_linux_apple/ woah#asahi · 2021-03-11 13:35
marcanI did invite them to our IRC (over twitter) and the corellium guy writing the linux patches was CCed on my upstream submissions#asahi · 2021-03-11 13:34
marcanthe goal is indeed to support all the hardware, including the GPU, and upstream it :)#asahi · 2021-03-11 13:34
marcanso lacking further info and a communications channel, we're basically working on our own#asahi · 2021-03-11 13:33
marcanI don't really know what corellium's plans are, since unfortunately they haven't interacted with the linux kernel community at all to date#asahi · 2021-03-11 13:33
marcanpipcet[m]: basically we're working on doing everything in an upstreamable way, so it takes a bit more time but it should be more efficient in the long term (since linux kernel forks aren't sustainable long-term)#asahi · 2021-03-11 08:43
marcanhttps://twitter.com/AsahiLinux/status/1369931997653561349#asahi · 2021-03-11 08:17
marcanI'll wait a bit before firing off the tweets#asahi · 2021-03-11 08:17
marcanwell, this was a long time coming: https://asahilinux.org/2021/03/progress-report-january-february-2021/ :)#asahi · 2021-03-09 06:27
marcanGuest56_: thank you! please don't feel obligated to :)#asahi · 2021-03-09 04:10
marcanGuest56_: the logs are public, having more copies does not hurt#asahi · 2021-03-08 19:20
marcanhopefully I can mostly wrap that up tomorrow#asahi · 2021-03-08 19:20
marcanwriting blog posts takes a while... I'm 3500 words into the progress report and I haven't even gotten to the linux kernel yet#asahi · 2021-03-08 17:35
marcanI mean ffs, the version of macos that shipped on these things had some hideous brokenness in the boot process; everyone was screaming "update before you even dare do anything"#asahi · 2021-03-08 17:35
marcanI still bet it works ~fine and they just didn't make shipping for macos#asahi · 2021-03-08 15:46
marcan:D#asahi · 2021-03-08 15:46
marcanwhat spec?#asahi · 2021-03-08 15:46
marcanOUT_OF_SPEC?#asahi · 2021-03-08 15:46
marcanha#asahi · 2021-03-08 15:05
marcanah, heh#asahi · 2021-03-08 15:04
marcanmaz: I'm really looking forward to this, running a few benchmarks side by side on qemu/macos vs qemu/kvm will make for good clickbait ;)#asahi · 2021-03-08 15:00
marcanagraf would know :)#asahi · 2021-03-08 14:59
marcanI think all the registers trap...#asahi · 2021-03-08 14:59
marcanthat's really all there is to it#asahi · 2021-03-08 14:59
marcanhttps://gist.github.com/ant4g0nist/9b1786b73293052ed17ffaa1a38e4608#file-hypervisor-rs-L1429 look at this#asahi · 2021-03-08 14:58
marcanIRQ, or FIQ#asahi · 2021-03-08 14:58
marcanthat#asahi · 2021-03-08 14:58
marcanrawValue being the enum value#asahi · 2021-03-08 14:58
marcanand it really is just IRQ and FIQ#asahi · 2021-03-08 14:58
marcanI think this is just an enum wrapper#asahi · 2021-03-08 14:58
marcanthe structure is called hv_interrupt_type_t#asahi · 2021-03-08 14:57
marcanwait, no, are you sure?#asahi · 2021-03-08 14:56
marcanand in that case they probably don't use that magic LR#asahi · 2021-03-08 14:56
marcanyeah#asahi · 2021-03-08 14:55
marcanI skimmed too quickly there#asahi · 2021-03-08 14:55
marcanah, I think you're right#asahi · 2021-03-08 14:54
marcanprobably HV_EXIT_REASON_EXCEPTION?#asahi · 2021-03-08 14:53
marcanbut there is nothing on IRQs in the API other than "IRQ, FIQ"#asahi · 2021-03-08 14:52
marcanand they probably never got around to implementing it#asahi · 2021-03-08 14:52
marcanbecause this launch was very, *very* obviously rushed#asahi · 2021-03-08 14:50
marcanmaz: I think they don't do vgic at all, and it's all emulated in userspace?#asahi · 2021-03-08 13:53
marcan"After receiving a HV_EXIT_REASON_VTIMER_ACTIVATED exit reason, the caller of hv_vcpu_run() needs to make the interrupts to the VTimer pending in the guest’s virtual interrupt controller."#asahi · 2021-03-08 13:53
marcanall you get is management of the pending IRQ state and the vtimer mask#asahi · 2021-03-08 13:52
marcanhttps://developer.apple.com/documentation/hypervisor/apple_silicon/vcpu_management#asahi · 2021-03-08 13:52
marcanso it should all be emulated in userspace#asahi · 2021-03-08 13:52
marcanarnd, maz: as far as I know hypervisor.framework only deals with IRQ/FIQ, it does not do vGIC#asahi · 2021-03-08 11:04
marcanwhen we have a use case#asahi · 2021-03-08 11:03
marcanwe can worry about micro-optimizing this later if we need to