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
<jmr2>
OK, I'm NOP'ing 582.
<alyssa>
582 should be TRUE'd or maybe FALSE'd if that doesn't work, couldnt tell you without a hypervisor trace
bps has joined #asahi
<jmr2>
Oops... missed that. NOP doesn't work - it fails to boot. Trying true.
<jmr2>
Is there an easy way to capture dmesg at the other end of the USB link? Currently, the TTY link closes during linux boot.
<alyssa>
fails to boot how?
<alyssa>
I say this as someone who needs a UART -- you *need* a UART ...
<jmr2>
Fails to boot: never reaches the login prompt.
<jmr2>
With TRUE instead of NOP, the last line is "fb0: switching to apple from simple".
<jmr2>
For context: the USB-C to USB converter I'm using to connect the rootfs has a screen attached to it. Might mess things up.
<jmr2>
Let me try with an initrd instead.
<jmr2>
Seems to work better - I get to Debian Installer's first screen in that context.
<jmr2>
Allright. Unless you have another quick one for me, I'll be signing out for now. I need to improve my setup before I can assist properly.
bps has quit [Ping timeout: 480 seconds]
<alyssa>
no quick ones
<jmr2>
OK. Thanks for all your work alyssa.
<alyssa>
:]
<alyssa>
marcan: ^^ if you want to get dcp.py going on an air lmk :_p
<alyssa>
looks like I only get a single link up / link down interrupt and then I need a reboot :|
<alyssa>
oh it's ratelimited in the pcie driver derp
<alyssa>
hm no it's still unhappy with me
<alyssa>
guessing something's screwed up on the SMC side of the equation
eric_engestrom has quit [Ping timeout: 480 seconds]
guan has joined #asahi
ZLSA has joined #asahi
saintdev has joined #asahi
cptcobalt has joined #asahi
aeroraptor has joined #asahi
PhilippvK has joined #asahi
nkaretnikov has joined #asahi
kendfinger has joined #asahi
phiologe has quit [Ping timeout: 480 seconds]
cptcobalt has quit [Ping timeout: 480 seconds]
aeroraptor has quit [Ping timeout: 480 seconds]
yuyichao has joined #asahi
Chainsaw has joined #asahi
robher has joined #asahi
aeroraptor has joined #asahi
cptcobalt has joined #asahi
kenjigashu has joined #asahi
amw has quit [Ping timeout: 480 seconds]
kenjigashu has quit []
eichin_ has quit []
eichin has joined #asahi
ovf has joined #asahi
kov has quit [Quit: Coyote finally caught me]
roxfan has quit [Ping timeout: 480 seconds]
steev has joined #asahi
roxfan has joined #asahi
kov has joined #asahi
eric_engestrom has joined #asahi
jkkm has joined #asahi
marvin24 has joined #asahi
marvin24_ has quit [Ping timeout: 480 seconds]
<Glanzmann>
rross101: Recently the debian installer never works for me, so I end up running debootstrap or copying the rootfs over. But whatever works for you. Copying the rootfs over is much faster.
<Glanzmann>
alyssa: Where can I get the x hotplug patch?
<Glanzmann>
jmr2: Good to hear.
<Glanzmann>
rross101: About the other question, You need one partitions, one for m1n1 and one for the rootfs. And go with copying over the rootfs, it will be much faster.
<Glanzmann>
alyssa: If you have not found anyone else by then, I can get you a hypervisor trace from a macbook air. If you talk me through it. To my knowledge I ahve an uart because I have three m1 devices.
<Glanzmann>
I might avaiable on the weekend, but next week for sure.
<Glanzmann>
I can give you also remote access if that is any help.
rohin has quit [Quit: Konversation terminated!]
aleasto has joined #asahi
nuh^ has quit [Remote host closed the connection]
nsklaus has joined #asahi
nskl has quit [Ping timeout: 480 seconds]
bps has joined #asahi
<sven>
j_ey: i think the pinctrl binding also needs the #interrupt-cells = <2> property. can you add that as well when you submit the apple,npins change?
bps has quit [Ping timeout: 480 seconds]
<j_ey>
sven: I suppose that is forward compatible with making it <3> in the future if we want to put a group there
<sven>
i think you'd need a custom irq_domain_ops then to default the group to zero when only two cells are present.
<sven>
but it might make more sense to just round robin distribute the interrupts across the groups or something anyway
<sven>
but we need #interrupt-cells = <2>; in the device tree with the current code to make it work
<sven>
i'm trying to clean up the device tree mess we have a little bit :D
<j_ey>
I'm going to send out v3 today, probably in a few hours
<j_ey>
sven: some of those dtbs are huuuge, the m1s are quite short in comparison, I wonder if they'll eventually be as bad
<j_ey>
err as big
<sven>
probably :D
<sven>
adding usb with the pd chips is already a lot
<j_ey>
those can be added now right? linux-next has everything for those?
<sven>
we'll probably need that fun quirk for dwc3 first
<sven>
otherwise it's kinda broken
<j_ey>
i guess
Nspace has joined #asahi
Nspace has quit [Quit: Nspace]
psykose has quit [Remote host closed the connection]
psykose has joined #asahi
yuyichao has quit [Ping timeout: 480 seconds]
Nspace has joined #asahi
___nick___ has joined #asahi
___nick___ has quit []
___nick___ has joined #asahi
Nspace has quit [Quit: Nspace]
Nspace has joined #asahi
Nspace has quit []
<alyssa>
Glanzmann: downstream/20211015 has fixes in the DCP driver
<alyssa>
it also fixes the build
<alyssa>
j_ey: Adding the display subsystem (DCP x2 plus all the type-C crossbar) will make it even worse :')
<j_ey>
I've seen that others seem to split the dts up a bit
<j_ey>
so maybe we could have t8031-usb.dtsi etc
<j_ey>
might also help if future computers have USB being identical but something else is different?
<alyssa>
j_ey: that reminds me, we might need to mess with the PCIe dts a bit so we don't try to bring up type-A/Ethernet on the macbooks
<alyssa>
(It's harmless to do so, but it's also.... kinda wrong. And with the current PCIe driver adds a delay at boot.)
<j_ey>
well we can put them as status=disabled in the t8103.dtsi and only =okay them in j274.dts
<j_ey>
are you sure maz's dts change doesnt do that already?
<alyssa>
ah, sure
<alyssa>
we don't have a non-j274 dts in mainline so there's why ;-)
<j_ey>
maybe we can add j313 and whatever the mbp is called for 5.16
<alyssa>
j_ey: I'm holding out for the M1X release to drive down the prices of used M1 macbooks ;-p
<Glanzmann>
alyssa: I'm currently traveling, will try it tomorrow and report back.
<j_ey>
alyssa: ;)
<alyssa>
j_ey: It's either that or text marcan "hey I need a MacBook Air for DCP driver dev" and make it Patreon's problem ;-p
<j_ey>
alyssa: just quit school and get a full time job :P
<chadmed>
uni is fun though
<chadmed>
its a massive privilege to be able to work, study AND pursue a hobby all at one and you should absolutely do it while you can
<chadmed>
especially if the three overlap
* j_ey
hated uni
<chadmed>
i go through depressive episodes where i come dangerously close to dropping out but going in and doing meaningful research digs me out
<chadmed>
those opportunities are becoming rare as funding gets slashed though :(
<j_ey>
uh oh, trying to send the pinctrl v3 and the machine I use to git send-email is down..
<j_ey>
ok nvm, just the dns is borked, IP address worked
rross101 has quit [Remote host closed the connection]
Nspace has joined #asahi
Nspace has quit []
<marcan>
chadmed: uni can be a... very different experience for different people
<j_ey>
after running git send-email --dry-run 10 times, pinctrl v3 is finally out
<sven>
\o/
Nspace has joined #asahi
Nspace has quit [Quit: Nspace]
sirn has quit [Remote host closed the connection]
method_ has quit [Ping timeout: 480 seconds]
Method has joined #asahi
sirn has joined #asahi
doggkruse has joined #asahi
ZestOfBeans has joined #asahi
ZestOfBeans has quit []
ZestOfBeans has joined #asahi
method_ has joined #asahi
Method has quit [Ping timeout: 480 seconds]
<alyssa>
j_ey: nice!
<j_ey>
\m/
jmr2 has joined #asahi
zopieux has quit [Quit: zopieux]
zopieux has joined #asahi
<jmr2>
alyssa: I believe that for the D582 callback, what I tried yesterday was not what you were asking for.
<jmr2>
I tried "[582] = trampoline_nop" and "[582] = trampoline_true" (which both hang during the boot process), but I guess you meant "[582] = dcpep_cb_true" ?
<jmr2>
Let me know if there's something else you want me to try. FWIW, I'm now on yout 20211015 branch and wifi is working (no X yet, though).
<povik>
"[index] = " is in kernel? i thought that was plan 9 dialect :-p
<sorear>
that's a C99 designated initializer, not a plan9ism
<povik>
i guess it was a plan9ism before that, or i am mistaken
<alyssa>
jmr2: oh, trampoline_true is right, I forgot I did that refactor
<jmr2>
OK. Let me see how I can get you some output of that. Can't connect either on the console or through ssh, so it'll probably need to be a screen picture.
<nsklaus>
i believed m1 soc had an original gpu, not an integrated mali ?
<nsklaus>
i must have missed some bits of infos on this it seems
<sven>
just also needs my minimal atc phy that i'll clean up tomorrow and then unplugging a cable when the mac is in device mode should also no longer break the port
<sven>
nsklaus: she works on panfrost as well
<j_ey>
it's not related to the m1
<alyssa>
I stop working on panfrost ===> I stop having money ===> I have to get a real job ===> No more M1
<alyssa>
:-p
<nsklaus>
"device mode" is when you have the macbook lid closed and plugged in keyboard,mouse and external screen ?
<nsklaus>
alyssa: i see only one course of action there, you need to clone yourself
<sven>
no, device mode is when the usb controller pretends to be a device
<nsklaus>
marcan may need to that too
<nsklaus>
come to think about it, everyone in the team should
<alyssa>
I call dibs on the clone of sven
<nsklaus>
sven: that usb thing seems so much more quirky than i would have supposed it to be
<psykose>
wtf 2 svens
<sven>
nsklaus: yeah, and we haven't even started on usb 3 or thunderbolt yet. i fear it's just gonna get worse from here
<alyssa>
ooh, or external DCP 😉
doggkruse has joined #asahi
<sven>
ugh. yeah. that too.
<nsklaus>
for a v1 having working basic usb2 mode would already be nice
<sven>
usb2 "just" requires that quirk + a pretty minimal atc-phy driver
<sven>
the rest (mainly the iommu driver) is already upstream :)
<nsklaus>
i saw mentions of this "atc-phy" driver a few times by now, but i have no clear idea of what it is exactly ?
<nsklaus>
i thought OSI model was just a tcp/ip thing
<nsklaus>
is that what's required to have ethernet dongle through usb plug ? (network over usb)
<sven>
it's the hw block that translates between the analog usb line signals (e.g. D+/D- for usb2) and the internal controller signals
<milek7>
I don't know why OSI is still referred to, it is dead
<nsklaus>
sven: how could usb even work at all without that ? (i saw you people reporting usb was working already, but didn't support hotplug) from what you say, it seems pretty important
<sven>
m1n1 already initializes the phy
<alyssa>
^^ this is also how I already have wi-fi working with no SMC driver :-p
<nsklaus>
ah i see.. these boot loader become more and more proeminent these days, they seem to be responsible for many things i would have thought to be the responsability of the kernel only, from a few years ago (decades)
<nsklaus>
*from the stand point of a feww years ago, i meant
<sven>
firmware might be a better term for m1n1 fwiw
<sven>
and the usb phy initialization insider there is more of a hack that works by accident
<nsklaus>
they initialize and handle more and more stuffs
<sven>
it essentially configured the PHY in device mode but for some reason that's also enough to make it work in host mode inside linux
<nsklaus>
m1n1 seems like a bootloader/firmware handler/device init daemon/ and some more. on x86 too, grub seems like taking care of many things too
<nsklaus>
it seemed lighter and easier in the lilo days
<nsklaus>
better "compartimentalized" too
<j_ey>
nsklaus: i guess that m1n1 will eventually not need to initialise the phy, but m1n1 will contain some special apple silicon bits that the kernel will not accept
doggkruse has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sven>
eh... the stuff that runs long before grub is also going to do a lot of hw initialization/configuration you just don't see
<alyssa>
jmr2: "failed to parse display attribs"
<alyssa>
that's your problem i guess. I can write a patch to help debug that in a bit but currently doing mali debug.
<alyssa>
still, progress!
<nsklaus>
it was more like a general thought about bootloaders across the spectrum of available archs, not a critic specificaly about m1n1
<jmr2>
alyssa: when you deem it the right time. I'm just providing feedback, not asking for support.
<alyssa>
thanks
<alyssa>
meanwhile about AFBC being broken..
<alyssa>
Wait. R32F RGB1 AFBC???
<alyssa>
It's a thing apparently.
<alyssa>
encoded like R8G8B8A8
<jmr2>
Final tidbit: the monitor I'm using as a USB hub to convert USB-C to USB-A (for keyboard and rootfs) also supports DisplayPort over USB-C. I think this external display is what messes stuff up in my case.
<alyssa>
No, it's not
<alyssa>
Linux will completely ignore the DisplayPort for now
<alyssa>
(If you were having USB trouble, different story)
<jmr2>
OK. But I think the reason I don't reach the login prompt is because it doesn't see the USB key at this time. Anyway. I'll stop distracting you.
ZestOfBe_ has joined #asahi
ZestOfBeans has quit [Ping timeout: 480 seconds]
jmr2 has quit [Remote host closed the connection]
Nspace has quit [Quit: Nspace]
ZestOfBe_ has left #asahi [Leaving...]
V_ is now known as V
aleasto has quit [Quit: Konversation terminated!]
crabbedhaloablut has joined #asahi
loki_val has quit [Ping timeout: 480 seconds]
<povik>
sorear: ah, you're right. i knew there was something plan-9-specific about it but confused the details