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
raster has joined #asahi
raster- has joined #asahi
raster has quit [Ping timeout: 246 seconds]
raster- has quit [Client Quit]
raster has joined #asahi
mixfix41 has joined #asahi
raster- has joined #asahi
raster has quit [Ping timeout: 245 seconds]
raster- has quit [Client Quit]
jottr has quit [Ping timeout: 252 seconds]
jottr has joined #asahi
jottr has quit [Ping timeout: 268 seconds]
scooby2 has quit [Read error: Connection reset by peer]
<marcan>
jannau: that was a nice thing to wake up to, thanks :)
<marcan>
almost exactly what I had in mind, I just want to change the dwc side API to take a pipe parameter instead of having two versions, and then put wrappers to call those in usb.c (this also gets rid of the ugly casts there)
<marcan>
no worries if you don't get to it by the time I start working on the other stuff, I'll just merge it and tack that on as a separate commit
<jannau>
marcan: yes, that's cleaner. is there a reason for the ssize_t / size_t return type mismatch between iodev .write/.read and usb_dwc3_read/write?
<marcan>
iodev can return errors as negative numbers
<marcan>
jannau: I was going to stream now but I spent some time on #-dev, so if you want to tweak that soonish I can merge it then :)
<jannau>
PR already refreshed with using wrappers in usb.c
Eighth_Doctor is now known as Conan_Kudo
<jannau>
usb_dwc.c returns 0 on errors
Conan_Kudo is now known as Eighth_Doctor
<marcan>
jannau: it's kind of ambiguous what to return on errors, depending on what kind of error, might be worth reviewing the failure paths, but not a priority right now
<marcan>
(I should document this too...)
<marcan>
thanks!
<marcan>
jannau: you have a clang-format issue :)
<jannau>
marcan: fixed
<marcan>
jannau: can you drop the USAGE_VUART thing? I'm just going to make the python side pick the right one
<marcan>
not point in trying to write to both, especially since one will be handed off to the guest
<marcan>
you can leave the hv_vuart change as is, I'm going to mess with that anyway
<marcan>
just no need for the usage flag, it can just be 0
<jannau>
sure proably faster if you edit to the state you like
<marcan>
sure
<marcan>
I'll just edit and push :)
<jannau>
ok
<jannau>
enum is in usb_dwc3.h
<jannau>
marcan: in usb_dwc3.h is still CDC_ACM_PIPE_VUART
<j`ey>
what does this second vuart add?
<j`ey>
(I mean, whats the usage, the purpose etc)
<j`ey>
ohhh (answered by marcan on stream)
<jannau>
marcan: the acm interfaces are 1 and 3
<jannau>
0 and 2 are the control ones
aquijoule_ has quit [Ping timeout: 240 seconds]
richbridger has joined #asahi
<jannau>
marcan: running m1n1 as guest worked here without errors
<jannau>
i.e. without USB errors
odmir_ has joined #asahi
odmir_ has quit [Ping timeout: 268 seconds]
<jannau>
my change rebased on main still works here
<j`ey>
marcan: what was the bug, i missed it
<j`ey>
lol woops
jeffmiw has joined #asahi
jeffmiw has quit [Ping timeout: 246 seconds]
zkrx has quit [Ping timeout: 268 seconds]
zkrx has joined #asahi
raster has quit [Quit: Gettin' stinky!]
jabashque has quit [Quit: Connection closed for inactivity]
agnem has quit [Ping timeout: 252 seconds]
VinDuv has quit [Quit: Leaving.]
ephe_meral1 has joined #asahi
emptynine has quit [Ping timeout: 260 seconds]
odmir has joined #asahi
odmir has quit [Ping timeout: 260 seconds]
VinDuv has joined #asahi
ephe_meral1 has quit [Ping timeout: 265 seconds]
kettenis has quit [Ping timeout: 252 seconds]
maknho_ has joined #asahi
maknho has quit [Ping timeout: 245 seconds]
kettenis has joined #asahi
AONeiLL[m] has quit [Quit: Idle for 30+ days]
zkrx has quit [Ping timeout: 260 seconds]
herbas has joined #asahi
zkrx has joined #asahi
herbas has quit [Quit: herbas]
kettenis has quit [Ping timeout: 252 seconds]
kettenis has joined #asahi
<kettenis>
marcan: thanks for merging that pcie PR
<marcan>
sorry it took so long, it was really silly of me to leave it on the backburner like that
<marcan>
wasn't even much of a review
<marcan>
(feel free to ping me in the future if I do this again...)
<kettenis>
heads up to folks using u-boot: you need the code on the apple-m1-mini-pcie branch if you update m1n1 to a newer version
<marcan>
I should try it, now that I have sven's hack merged too
<marcan>
I want to see the pcie stuff work :)
<marcan>
wonder if it all works under the hypervisor too!
<kettenis>
]
<kettenis>
not sure what will happen if you try to boot macos after running that code
ephe_meral1 has joined #asahi
<kettenis>
I've been experimenting a bit with better pinctrl and pcie DT bindings
<kettenis>
sumitted the pinctrl bindings to the linux mailing list last weekend
<kettenis>
I'll probably do a v2 to fix a typo tomorrow and submit the pcie series as well
<_alice>
What’s up?
<kettenis>
been distracted a bit as some folks threw me a RISC-V curveball
<kettenis>
but we need proper clock and DART DT bindings as well before I can think about upstreaming the U-Boot stuff
* sven
hides
<sven>
i could just submit the clock stuff which maps a single page for each clock
<sven>
regarding the DART stuff I wanted to make sure this DAPF or the SMMU instances don't need special treatment first. i think they don't.
<sven>
now i just need to fix that 16k/4k pages issue and then i can submit v3
<kettenis>
I still think it is a mistake to have the darts and sub-darts as separate device nodes, but I guess I'll cope
<sven>
i still don't have a strong opinion on that
<sven>
but the maintainers seemed to prefer the separate node thing *shrug*
<sven>
it's still weird that some parts of the dwc controller have to use the first dart and others the second dart though
rjeffman has joined #asahi
<marcan>
that just sounds like a massive hack
<marcan>
like they had two DMA masters that get randomly used
<marcan>
and couldn't funnel them into a single DART so they just yolo'd it and threw two on there
<svenpeter>
The truly weird part is that a single DART should support up to 16 masters
<kettenis>
hmm
<marcan>
is that really 16 masters, or is it 16 tags on a single master?
<kettenis>
AttributeError: 'super' object has no attribute '_reset_input_buffer'
<marcan>
pyserial version difference?
<marcan>
that's from sven's PR
<kettenis>
this is version 3.4
<marcan>
tty_dev.reset_input_buffer()
<marcan>
I had it without the _hm
<marcan>
er
<marcan>
sorry
<marcan>
I have 3.5
<marcan>
kettenis: you can just comment that stuff out, it's just about managing messages in the uart buffer at script startup time
<marcan>
going to sleep now but I'll look tomorrow
choozy has joined #asahi
<svenpeter>
Hmmm… good point, it might be 16 tags for a single master
<svenpeter>
That would also explain why literally every single device has its own dart
<kettenis>
so there is a separate dart for each pcie root port but devices behind those ports each use their own tag?
<kettenis>
in other news, with the updated m1n1, the timeout of the OpenBSD bootloader prompt suddenly behaves as it should
<kettenis>
wonder what changed there
<pipcet[m]>
in hypervisor mode or just chain-loaded?
<kettenis>
just chain loaded
<kettenis>
wonder if that fixes the slowness of netbooting that was reported the other day
<pugguu[m]>
Odd question but would installing asahi linux void warrenty
<pipcet[m]>
svenpeter: maybe they're just dart cheap
<pipcet[m]>
I do think it's weird that even the touchbar display goes through a dart of its own
<pugguu[m]>
Also will there be a a debian based image once we have everything working
<kettenis>
once evrything works, any linux distro should just work
<pugguu[m]>
But wouldn't the distro creator need to make a image first
<pugguu[m]>
Or could i use a raspberry pi image or dose it have to be a generic aarch64 image
<kettenis>
a standard arm64 image should just work
<pugguu[m]>
Just intrested
<pugguu[m]>
Hmm ok
<j`ey>
pugguu[m]: depends if the rpi kernel disables features
<svenpeter>
Eventually most Linux installers should just work once all kernel patches have propagated downstream
<pipcet[m]>
(Debian works fine with the Corellium kernel, with the exception of Emacs which had a 4KB page size assumption)
<kettenis>
I've already loaded a Linux kernel from a standard arm64 debian installer written to a usb stick
<kettenis>
U-Boot boots GRUB and GRUB boots the kernel
<kettenis>
of course the kernel didn't work because it doesn't have any support for the M1
<kettenis>
but once debian ships with a 5.13 kernel it should at least print some stuff on the console
<pipcet[m]>
I've actually put together a debian-installer-in-a-macho monster
<pugguu[m]>
Sorry if i do get abit annoying with all the questions, its because i havent the slightest of clue on how this all works
<pipcet[m]>
too large for stock m1n1 to load, but it works fine with kmutil :-)
<kettenis>
well, that's the problem with you linux people
<kettenis>
you can't agree on a standard way to boot an OS ;)
<pugguu[m]>
I no joke just hit install and use something like synaptic or software center to get what i need and use it
<agraf>
kettenis: just keep up the great work :D. And thanks a lot for the U-Boot port!
<pugguu[m]>
Only started using linux full time last year untill i got my m1 mini and am using mac os untill this project is redy
<svenpeter>
Uboot -> grub sounds totally fine to
<svenpeter>
boot Linux to me ;)
<pugguu[m]>
I will probs do some user testing once we get some sort of gui or de going
ephe_meral1 has quit [Ping timeout: 240 seconds]
<pugguu[m]>
All i would need is 1 list of things to test with a list of commands and im good to go
<pugguu[m]>
But we are talking alot later
<pugguu[m]>
Anyways im gonna go see if mum has finished her movie so i can get my cookie dough icecream
<kettenis>
sven's PR also broke loading a kernel without explicitly specifying a tty on the command line
<kettenis>
issued a PR to fix that
maknho__ has joined #asahi
maknho_ has quit [Ping timeout: 246 seconds]
<svenpeter>
Weird, I remember testing that. Looks like many things changed in the past month. Thanks!
<kettenis>
yeah, probably the universe changed around you ;)
<svenpeter>
Damnit, I hoped that wouldn’t happen again ;)