marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | General project discussion | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
amw has joined #asahi
bgb has joined #asahi
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bgb has quit [Read error: Connection reset by peer]
bgb has joined #asahi
bgb has quit [Ping timeout: 480 seconds]
psykose has quit [Remote host closed the connection]
psykose has joined #asahi
phiologe has joined #asahi
user982492 has joined #asahi
PhilippvK has quit [Ping timeout: 480 seconds]
<marcan>
640x1136 is what you get without a display IIRC
<marcan>
that sounds like the iBoot DCP bring-up fails with your display
<marcan>
probably a regression on apple's side :/
rkt has joined #asahi
bgb has joined #asahi
jmr2 has joined #asahi
jmr2 has left #asahi [#asahi]
bgb has quit [Ping timeout: 480 seconds]
kenjigashu has joined #asahi
bgb has joined #asahi
rkt has quit [Quit: rkt]
<bgb>
marcan: for the cu.debug-console, we probably cannot fix this, right?
<marcan>
not sure... I mean if it's a bootarg or something you could downgrade security and set it
<bgb>
ok
marvin24_ has joined #asahi
rkt has joined #asahi
marvin24 has quit [Ping timeout: 480 seconds]
<marcan>
new MBP just arrived :-)
<i509vcb[m]>
this will be an interesting weekend :)
<chadmed>
theres some reviewer claiming that the m1 pro has 32 memory channels, but i cannot for the life of me work out how he got 32 channels from a 512 bit bus. ddr5 data bus width is 64 bits with two 40 bit channels interleaved over that, so (512/64)*2 is 16 channels, not 32
<chadmed>
s/pro/max
kov has quit [Quit: Coyote finally caught me]
kov has joined #asahi
<marcan>
i509vcb[m]: monday :)
<marcan>
busy weekend :p
<marcan>
chadmed: LPDDR channels are 16 bits
<marcan>
32 is correct
<marcan>
M1 has 8
<marcan>
(you can see that in my cpufreq patch series)
<marcan>
I don't know where you got the *2, DDR doesn't work like that, it's still one channel
<chadmed>
ohhhhh ok. somehow that information never came up when i was looking to verify the claim
<marcan>
so M1 Max has 32 channels but is equivalent, width-wise, to 8 channels of typical DDR
bgb_ has joined #asahi
<sorear>
64 byte cache blocks? does a LLC miss occupy 1 channel for 32 transfers or is it somehow strided?
bgb has quit [Ping timeout: 480 seconds]
<chadmed>
yeah nvm i just had a look at the spec again. it only mentions that each dimm houses two completely separate channels. i misinterpreted that part.
rkt has quit [Quit: rkt]
bgb_ has quit [Ping timeout: 480 seconds]
rkt has joined #asahi
rkt has left #asahi [#asahi]
rohin2 has joined #asahi
kenjigashu has quit [Ping timeout: 480 seconds]
<marcan>
sorear: it would be strided
<marcan>
chadmed: I think you mean ranks, not channels
<marcan>
one DIMM is one channel
<marcan>
unless this changed with DDR5?
<chadmed>
yeah each ddr5 dimm has two CA lines now
<marcan>
ah wait, 2x32
<marcan>
so they split the 64b into two halves
<marcan>
LPDDR5 is 16b or 32b so I guess that aligns more with that (M1s use 16b mode)
<chadmed>
yeah thats why i was a bit confused
<chadmed>
with ECC its two 40 bit channels per dimm, thats where i got that number from
<marcan>
yeah, so there's more ECC bits now
<marcan>
(used to be 72b total, now 80b)
<marcan>
which I guess you need to have usable per-bank ECC
rohin2 has quit [Ping timeout: 480 seconds]
kenzie has quit [Quit: Ping timeout (120 seconds)]
kenzie has joined #asahi
Hotswap has quit [Ping timeout: 480 seconds]
Hotswap has joined #asahi
darkapex3 is now known as darkapex
rkt has joined #asahi
<jannau>
marcan: the 12.0.1 ipsw is also required after updating to 12.0.1. recovery os pairing fails with 12.0 beta8
<marcan>
I've seen apple break the betas on release before
<marcan>
I think that's expected?
<marcan>
anyway, I'm going to push everyone towards 12.0.1 or later
<jannau>
yes, not really a problem. more a remark that the installer MR is also required on M1
<marcan>
yes, the first part anyway
<jannau>
the display regression is annoying. disconnecting the display results in the same values. do we know how to reenable the display?
<marcan>
alyssa's DCP driver should bring it up the same way macOS does
<marcan>
but other than that, no
<marcan>
that sounds like a bug to be reported to apple
rkt is now known as Guest4338
rkt_ has joined #asahi
rkt_ is now known as rkt
<marcan>
jannau: you don't see the apple boot logo at all, right?
<marcan>
(neither when booting macos nor 1TR)
<marcan>
well, macos should show it with the progress bar already started
<_jannau_>
I think so, not sure about booting to mac os but when booting into 1tr I don't see "hold the power button ..." The first thing I see is the "Options" icon in 1tr
<marcan>
yeah, definitely that then
<marcan>
iBoot isn't initializing the screen at all
Guest4338 has quit [Ping timeout: 480 seconds]
<marcan>
you should file a feedback report :)
<_jannau_>
I will
chadmed has quit [Quit: Konversation terminated!]
bgb_ has joined #asahi
chadmed has joined #asahi
<_jannau_>
done
<j_ey>
_jannau_: so would booting linux w/o DCP be broken on the laptops?
<_jannau_>
define broken. I would expect booting to work. it would just not display anything
aleasto has joined #asahi
<j_ey>
yeah, thats what I meant
<_jannau_>
I don't think it will be a problem on the laptops though. I doubt apple will ship a regression in the initialisation of the built-in displays
bgb_ has quit [Ping timeout: 480 seconds]
bgb_ has joined #asahi
sailorek1234 has joined #asahi
<marcan>
FWIW it works fine for me on my Mini, so it's not like it's broken overall
<marcan>
well, at least with the betas
<marcan>
(haven't tried 12.0.1 yet)
<marcan>
but I doubt they broke this for Minis overall either
<marcan>
it's probably just broken for some subset of displays including _jannau_'s
linearcannon has quit [Read error: Connection reset by peer]
linearcannon has joined #asahi
linearcannon is now known as Guest4346
linearcannon has joined #asahi
Guest4346 has quit [Ping timeout: 480 seconds]
marvin24 has joined #asahi
marvin24_ has quit [Ping timeout: 480 seconds]
klaus has quit [Quit: leaving]
nsklaus has joined #asahi
rkt has quit [Quit: rkt]
bgb_ has quit [Ping timeout: 480 seconds]
rkt has joined #asahi
<jannau>
my macbook pro has arrived
user982492 has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<povik>
same here
rkt has quit [Quit: rkt]
rkt has joined #asahi
<chadmed>
im still torn on whether i get a macbook after my final exam or hold out for the next slew of minis. i dont do much (any) mobile computing so it would be quite wasteful to get a laptop but im chomping at the bit to get a new machine
chadmed has quit [Quit: Konversation terminated!]
<povik>
hmm, i thought ssh keys are held by SEP out of the box on macos now
<povik>
that doesnt seem to be the case, and from googling i am not even sure if theres any support for it
<marcan>
apple don't themselves integrate it with SSH but you can with that
<povik>
marcan: it is!
<marcan>
it only does ECDSA though, but I think the SEP does ed25519 too internally (it's just not part of Apple's API), so I hope on Linux we can support that
<_alice>
there's also `ssh-keychain(8)`, which in theory should be able to do SEP-resident ssh keys, even though it was designed for smartcard PIV applets (the SEP APIs on macOS are mainly tied to the keychain)
__xqsl has quit [Remote host closed the connection]
rkt has quit [Quit: rkt]
rkt has joined #asahi
<sven>
povik: i can clean up and push some python code to bring up the SEP and parse this gigalocker structure (required to bring up the endpoint that handles machine-specific keys) if you want to look into how that all works
<povik>
sven: i am interested in how that all works but i think i wont be looking into it anytime soon
<povik>
so no need to hurry
<povik>
that was more of a passer-by comment up there :-p
<sven>
ok. i don't really know how any of this works yet. just barely enough to maybe bring SEPOS up and running :D
<povik>
that's something
<sven>
which already required APFS read and write afaict :/
<povik>
yeah, i am curious how well that will work out in interop between linux and macos
<povik>
after all work is done (if ever)
<sven>
i think the only issue is that APFS partition. but luckily the gigalocker seems to be almost the only thing on there and so far it even had a static filesize
<sven>
so i hope we won't run into any issues
vivg[m] has joined #asahi
mikebeaton[m] has joined #asahi
c10l has joined #asahi
c10l0 has quit [Read error: Connection reset by peer]
mikebeaton[m] has quit [Quit: issued !quit command]
<marcan>
I sure hope linux-apfs can handle at least that much :-)