#
✕ nick:marcan
#asahi-dev · 2022-02-23 12:02
marcanactually checking for EL1 and skipping that might not be a good idea
#asahi-dev · 2022-02-23 12:01
marcanbut sure
#asahi-dev · 2022-02-23 12:01
marcanI still want serial on the laptops ;)
#asahi-dev · 2022-02-23 12:01
marcanheh
#asahi-dev · 2022-02-23 12:01
marcanbut then again if you're using earlycon you probably aren't using u-boot (unless you're debugging that)
#asahi-dev · 2022-02-23 12:00
marcanI need to test whether this actually works on linux though
#asahi-dev · 2022-02-23 11:59
marcan*it makes sense
#asahi-dev · 2022-02-23 11:59
marcanthen maybe it depends for u-boot to do this instead of m1n1? on m1n1, especially under the HV, you pretty much always want regular serial earlycon
#asahi-dev · 2022-02-23 11:57
marcanbut then that wouldn't be looking up the DT
#asahi-dev · 2022-02-23 11:57
marcanthere's an EFI earlycon, which I guess might work?
#asahi-dev · 2022-02-23 11:57
marcanI think there's an earlycon for FBs though? but no sure if it's a thing as simplefb
#asahi-dev · 2022-02-23 11:56
marcanno, earlycon is much earlier than a framebuffer
#asahi-dev · 2022-02-23 11:55
marcanyou just pass "earlycon" and then it uses stdout-path to decide what to use
#asahi-dev · 2022-02-23 11:55
marcannot earlycon
#asahi-dev · 2022-02-23 11:54
marcanhm, for linux serial always makes sense since that's largely for earlycon
#asahi-dev · 2022-02-23 11:50
marcanjannau: moved to PKGBUILDs, also invited you to a team to manage access that way
#asahi-dev · 2022-02-23 11:42
marcanjust transfer it to me then, then I can move it into asahi
#asahi-dev · 2022-02-23 11:42
marcanah
#asahi-dev · 2022-02-23 11:41
marcanspace?
#asahi-dev · 2022-02-23 11:39
marcanjannau: unless you prefer I just fork it
#asahi-dev · 2022-02-23 11:31
marcanwant to invoke the transfer?
#asahi-dev · 2022-02-23 11:30
marcanwe can rewrite the history if you think that makes sense, sure
#asahi-dev · 2022-02-23 11:24
marcanjannau: what do you think about moving AsahiLinux-PKGBUILD under AsahiLinux?
#asahi-dev · 2022-02-22 05:49
marcaninstead of trying to make it wholly generic
#asahi-dev · 2022-02-22 05:49
marcanconsidering it's a panic situation, I'm guessing it would make sense to do it in individual drivers which have an emergency shutdown mechanism (like NVMe)
#asahi-dev · 2022-02-22 05:49
marcando you think it'd make sense to hook that into nvme or the block layer as a whole, like macOS does?
#asahi-dev · 2022-02-22 05:48
marcanaxboe: I see linux does have a panic_notifier_list mechanism, but it isn't really used for much aside from some core stuff
#asahi-dev · 2022-02-21 16:18
marcanand we can't really do it for dcp
#asahi-dev · 2022-02-21 16:18
marcanit doesn't make any sense to do it for simplefb and not dcp
#asahi-dev · 2022-02-21 16:14
marcanwe don't have enough layers to fake that
#asahi-dev · 2022-02-21 16:14
marcanit'd have to be something the compositor does
#asahi-dev · 2022-02-21 16:14
marcanunlikely to be practical I think, at least at the DRM level
#asahi-dev · 2022-02-21 16:11
marcan(except the logo)
#asahi-dev · 2022-02-21 16:11
marcan(when loading something else)
#asahi-dev · 2022-02-21 16:10
marcanVinDuv: m1n1 clears the screen
#asahi-dev · 2022-02-21 16:10
marcanfwiw my margin patch was broken, it faults on scroll
#asahi-dev · 2022-02-21 15:59
marcanworks with linux and u-boot
#asahi-dev · 2022-02-21 15:58
marcan(since we know these screens are 16:10 notchless, and one would hope Apple will stick to even dimensions plus notch)
#asahi-dev · 2022-02-21 15:58
marcanthe code is a heuristic to avoid per-platform code (for now); we do have "is there a notch" in the ADT. it checks that, then rounds down the height to an integer 16:N aspect
#asahi-dev · 2022-02-21 15:57
marcanm1n1-m1n1 keeps the full fb, same with xnu/hv/etc
#asahi-dev · 2022-02-21 15:57
marcanI do it in kboot so it only affects the m1n1 -> u-boot (or linux) boundary
#asahi-dev · 2022-02-21 15:57
marcangood point on that one, this makes more sense
#asahi-dev · 2022-02-21 15:57
marcankettenis: pushed notch hiding to m1n1, removed the margin stuff
#asahi-dev · 2022-02-21 14:06
marcanbasically, tl;dr it's illegal to map read-sensitive or read-illegal memory as Normal memory into the guest if the guest ever plans to run with translation disabled, because there is no way for it to stop the CPU from speculating into it otherwise
#asahi-dev · 2022-02-21 14:02
marcanso if you have S2 and no S1, and S2 says it's normal memory, you end up with normal memory, and things explode randomly
#asahi-dev · 2022-02-21 14:02
marcanwhich limits speculation to allow software to guarantee the CPU does not speculate into memory it shouldn't, but *only* if *all* stages of translation are disabled
#asahi-dev · 2022-02-21 14:02
marcanthe only reason that doesn't explode all over the place is the later subsection "Behavior of instruction fetches when all associated stages of translation are disabled
#asahi-dev · 2022-02-21 14:01
marcantranslation regime
#asahi-dev · 2022-02-21 14:01
marcanshareability attributes determined by the value of the SCTLR_ELx.I bit for the
#asahi-dev · 2022-02-21 14:01
marcanThe stage 1 translation assigns the Normal memory attribute, with the cacheability and

plus ›