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-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
<stemnic>
That does not look like it's based on their previous attempt with sandcastle
liur[m] has joined #asahi
bylaws has joined #asahi
<dottedmag>
Does not look upstreamable in this form either (disc: I'm an amateur in this) — changes all over generic arm64 code. Am I wrong, do such patches routinely get merged without pushback?
ephe_meral has joined #asahi
<Shiz>
the changes over gic look a bit sus, but the rest doesn't look terribly objectionable to me
<jix>
I'm also an amateur regarding this, but I think it's not that unusual and in general shouldn't be an issue if justified, no idea if these changes are though
<Shiz>
but yeah, this is also 'just' bringup with a minimal device tree and enabling SMP :)
<Shiz>
it's good work, but not earth-shattering
<tbodt>
this is probably patches taken from sandcastle and cleaned up a bit
<tbodt>
first step towards upstreaming
<hthh_>
the wfi and nGnRE quirks sound interesting
ephe_meral has quit [Ping timeout: 264 seconds]
<sven>
yeah, I think it's a great step in the right direction :)
mogery has quit [Remote host closed the connection]
mogery has joined #asahi
_whitelogger has joined #asahi
zkrx has joined #asahi
ephe_meral has joined #asahi
raster has joined #asahi
Axenntio has joined #asahi
raster has quit [Remote host closed the connection]
raster has joined #asahi
mogery has quit [Quit: Leaving]
mogery has joined #asahi
ant__ has joined #asahi
marvin24 has joined #asahi
marvin24_ has quit [Ping timeout: 260 seconds]
bgianf has joined #asahi
ephe_meral has quit [Read error: Connection reset by peer]
ant__ has quit [*.net *.split]
bgianf has quit [*.net *.split]
Swant has quit [*.net *.split]
superk has quit [*.net *.split]
browzing_ has quit [*.net *.split]
djhuk has quit [*.net *.split]
nyx_o has quit [*.net *.split]
balrog has quit [*.net *.split]
Hexagon has quit [*.net *.split]
spikebike has quit [*.net *.split]
dshep has quit [*.net *.split]
superk has joined #asahi
ant__ has joined #asahi
mogeryy has joined #asahi
Hexagon has joined #asahi
spikebike has joined #asahi
browzing_ has joined #asahi
dshep has joined #asahi
bgianf has joined #asahi
nyx_o has joined #asahi
bostjan has quit [Read error: Connection reset by peer]
crucifix has joined #asahi
balrog has joined #asahi
Swant has joined #asahi
justMaku has joined #asahi
djhuk has joined #asahi
mogery has quit [Ping timeout: 256 seconds]
mogeryy is now known as mogery
plainbits has joined #asahi
q3k|m has quit [Ping timeout: 246 seconds]
bgianf has quit [Quit: leaving]
bgianf has joined #asahi
suskun has joined #asahi
suskun is now known as lucid_0x80
lucid_0x80 has quit [Remote host closed the connection]
suskun has joined #asahi
suskun is now known as lucid_0x80
lucid_0x80 has quit [Remote host closed the connection]
suskun has joined #asahi
suskun is now known as lucid_0x80
q3k|m has joined #asahi
lucid_0x80 has quit [Ping timeout: 264 seconds]
ant__ is now known as ayehavgunne
<rcombs>
marcan: just write a 5-line malloc() and a no-op free() lol, you've got 16GB to burn
<jn__>
the ballistic missile approach to garbage collection
<opticron>
gross, but not wrong
suskun has joined #asahi
Necrosporus is now known as Guest35351
Necrosporus has joined #asahi
suskun has quit [Read error: Connection reset by peer]
suskun has joined #asahi
suskun is now known as lucid_0x80
lucid_0x80 has quit [Remote host closed the connection]
Guest35351 has quit [Read error: Connection reset by peer]
suskun has joined #asahi
suskun has quit [Remote host closed the connection]
marvin24 has quit [Ping timeout: 244 seconds]
marvin24 has joined #asahi
opticron has quit [Ping timeout: 264 seconds]
Axenntio has quit [Quit: Leaving]
opticron has joined #asahi
ayehavgunne has quit [Quit: Lost terminal]
modwizcode has joined #asahi
jamadazi has joined #asahi
Axenntio has joined #asahi
jamadazi has quit [Ping timeout: 264 seconds]
<dff>
what is the current guestimate on when the first build for public testing is out?
<dff>
months, years, decades?
<dff>
:)
frode_0xa has quit [Quit: leaving]
<Necrosporus>
dff, -1 day
<jn__>
Necrosporus: if you mean the Corellium kernel, well… it's not from Asahi
<Necrosporus>
Shiz, I guess if mach-o becomes more common, generating code will be in kernel tree rather than as separate program... Though mkimage for u-boot is still separate program
<Necrosporus>
I guess kernel tree has copy of mkimage, right?
<j`ey>
Necrosporus: the kernel wont have any MachO stuff
suskun has joined #asahi
<Necrosporus>
Why does it have efi stub then?
<Necrosporus>
IT could just as well be a converter taking image zImage
<Necrosporus>
and giving uefi image out
<j`ey>
fair, but if it does gain MachO, it might be in the same style as UEFI stub
<Necrosporus>
jn__, then maybe +1 day. there is already attempt to bring the kernel up, I guess it shouldn't be too hard to get as far as Corellium did
<Necrosporus>
( linux boots but no peripherals at all, seems )
* jn__
nods
balrog has quit [Remote host closed the connection]
suskun has quit [Remote host closed the connection]
<jn__>
adding a simple framebuffer in the DT should probably be easy enough, too once the MMU stuff is out of the way
balrog has joined #asahi
<Necrosporus>
What is mar[]can (don't want to ping and distract him) doing right now in the stream? Trying to dump RAM just after iBoot finished loading m1n1?
<j`ey>
trying to understand the framebuffer / mmu a bit
<jn__>
figuring out MMU mappings, i think
<Necrosporus>
He already has a picture in FB seems
<Necrosporus>
so it kinda works?
<j`ey>
yeah, with a physical address
<mogery>
it's more MMU mapping figuring out
<mogery>
to load linux into
<mogery>
but there was this beautiful idea to load linux into the framebuffer since its already mapped
<Necrosporus>
And did it work? Can you give me timecode?
<Necrosporus>
It reminds me of what I was trying to do
<Necrosporus>
I have a laptop which runs MS-DOS 3.3 out of ROM chip and has COM port and no storage at all except ramdrive of 64000 bytes
<mogery>
he didnt actually dry it (yet) if i understand correctly
<Necrosporus>
I have managed to dump entire C drive image and extract files from it
<mogery>
hes dumping ram rn
<modwizcode>
Necrosporus: sounds like a waste of both memory and a large rom chip lol'
<Necrosporus>
What do you mean?
<modwizcode>
Just seems like a lot to use a large ramdisk in the era of MS-DOS
rjzak_ has quit [Quit: Leaving]
<modwizcode>
*not large but larger than 0
<Necrosporus>
D: drive with 64000 bytes is quite useful when you don't have anything else to write files to
<Necrosporus>
no floppy, no cdrom, no hard drive
<modwizcode>
Yeah but it's... idk that's a strange way to structure your computer
<Necrosporus>
It has external floppy but I don't have the actual device
<Necrosporus>
modwizcode, it is 8086 (not even 286) computer from 1988
<modwizcode>
fair enough, I don't think anything I used had less than a 286 in it.
<modwizcode>
Still I thought memory would be pricey enough that that wouldn't make sense to do a system like tha
<Necrosporus>
Actually NEC v30, but it's something between 8086 and 80286 clone
<Necrosporus>
modwizcode, and if it didn't have ramdrive I wouldn't be able to run my own executable files on it
<modwizcode>
I mean theoretically they didn't have to go with a ramdisk though right? I've looked at some of DOS's source it theoretically had like io buffering so it could have run stuff off the external drive like that. Anyway this is all very offtopic :p
<Necrosporus>
But since it did then it was possible to throw in a small executable which didn't have EOF bytes in the middle via copy aux file.com (and append EOF byte to finish file). Then this file can be executed to dump arbitrary files encoded in base64
<modwizcode>
ah that's clever lol
<Necrosporus>
modwizcode, then I wrote some C code which reads hard drive and prints hexdump onto the stdout
<Necrosporus>
for 512 kb hd it's feasbile
<Necrosporus>
even with 9600 baud
<Necrosporus>
I have also copied files one by one and they matched those found on filesystem image proving it was correct
<Necrosporus>
copy file aux on dos side and cat /dev/ttyUSB0 > file on linux side
snalty has joined #asahi
<rcombs>
I don't expect the kernel to end up doing mach-o
<rcombs>
too much work to update, much easier for the user to just have a small mach-o bootloader they flash, which then loads the linux image
PdimH has joined #asahi
<Necrosporus>
rcombs, then why linux efi stub is supported?
<Necrosporus>
It would make easier to have small efi bootloader which loads regular zImage
<jn__>
Necrosporus: no need to argue for this now. asahi will probably first go the way rcombs describes, and if you like the other way you can (find someone to) implement it later
<Necrosporus>
I'm not quite arguing. The important point is whenever it boots or not. How is not as important
<rcombs>
it's much easier to update an EFI executable on an intel system than the Mach-O image on an M1
<jn__>
linux will boot, no matter whether it boots via m1n1, or directly from iboot
<rcombs>
since updating the Mach-O image requires rebooting into recovery mode, signing the new file, regenerating boot policy…
<rcombs>
not possible to do remotely either
<modwizcode>
it would be nice if we could do that somehow but I guess the signed image is more like loading the signing keys into your UEFI firmware
<modwizcode>
The thing we load can do it's own signing if you like
<rcombs>
as far as I'm aware, only 1TR has access to the relevant signing keys
<narmstrong>
Necrosporus: wow, I actually hacked on the V32 on the Casio Fx2/graph100 calculator like 20y ago, was funny how they had their DOS clone in ROM
<narmstrong>
*v#2
<narmstrong>
damn *v30
<narmstrong>
so they also did computers with it
<jix>
I did hack on the v30mz of the bandai wonderswan color a few years ago :)
jamadazi has joined #asahi
Tokamak has joined #asahi
<rcombs>
meanwhile the corellium people are probably currently violating GPL
<Necrosporus>
Maybe not
<Necrosporus>
they did upload some sources
<jix>
those do not match their binaries at all though
<rcombs>
I haven't seen any of the tooling required to actually build the macho
<Necrosporus>
What somebody is supposed to do if they uploaded some binaries then edited the sources and overwritten it with new (and perhaps better) code?