marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | Not ready for end users / self contained install yet. Soon. | 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
axboe has quit [Ping timeout: 480 seconds]
axboe has joined #asahi
yuyichao has quit [Ping timeout: 480 seconds]
a_fellow_human[m] has joined #asahi
yuyichao has joined #asahi
refi64 has quit [Remote host closed the connection]
refi64 has joined #asahi
mmarchini has joined #asahi
refi64 has quit []
refi64 has joined #asahi
ciggi_ has joined #asahi
ciggi has quit [Ping timeout: 480 seconds]
phiologe has joined #asahi
kov has quit [Quit: Coyote finally caught me]
PhilippvK has quit [Ping timeout: 480 seconds]
axboe has quit [Quit: leaving]
wasd has joined #asahi
ciggi_ has quit [Ping timeout: 480 seconds]
jmr2 has joined #asahi
<jmr2>
Regarding the "Unable to determine primary OS" issue: At 11:45 in the original "Installer research" video, marcan mentions that the Volume UUID is the same as the APFS volume group UUID.
<jmr2>
This isn't the case on my setup.
wasd has quit [Quit: Page closed]
<marcan>
jmr2: diskutil apfs listVolumeGroups shows a different VGUUID from the data role volume?
<jmr2>
Errr... Sorry... I don't understand...
<marcan>
the "Volume Groups <foo>" should be the same as "Volume UUID: <foo>" for the data volume under it
Major_Biscuit has quit [Ping timeout: 480 seconds]
<David[m]123456789>
Welp! I booted on the righoust most asahi startup disk (though it would be the most recent one, so the one where installation finished), but now I'n in a boot loop!
<David[m]123456789>
I have no guarantee
<Glanzmann>
David[m]123456789: that is normal.
<Glanzmann>
So you need to power off the machine.
<Glanzmann>
than you press and hole the power button and select options
<Glanzmann>
Than from utilities > terminal you open a terminal and type: /Volumes/Linux/step2.sh
<Glanzmann>
Than you will be promoted for password, y, username, password.
<Glanzmann>
Afterwards you should have m1n1 on your system and no longer a boot loop.
<Glanzmann>
If you want u-boot that, you need to install boot loop
<David[m]123456789>
I'm in a boot loop when I go in 1TR
<David[m]123456789>
* when I try go in, * in 1TR or boot normaly
<Glanzmann>
That is not 1tr.
<Glanzmann>
That is recovery.
<Glanzmann>
Shut it down. Completly.
<Glanzmann>
Than press and hold the power button for 15 seconds.
<Glanzmann>
Let it go.
<Glanzmann>
choose options. Than you're in 1tr.
<Glanzmann>
1tr and recovery look similiar, but are not.
<David[m]123456789>
I pressed power until the second options restant (ie. 1TR)
<David[m]123456789>
s/I pressed power until the second options restant (ie. 1TR)/I pressed power until the second options restart (1TR I think) before it goes in boot loop marcan @marcan:matrix.org /
<Glanzmann>
David[m]123456789: That is not normal.
<Glanzmann>
David[m]123456789: I assume you need to recovery your mac using a second mac or idevicerestore.
<Glanzmann>
I never tried idevicerestore only the other thing.
the_lanetly_052___ has quit [Ping timeout: 480 seconds]
<David[m]123456789>
Thanks for the info, how is the debian machine connected? I will also go to an apple store
<VinDuv>
idevicerestore works fine in my experience but you need to install the right parts from source, start usbmuxd beforehand and have sufficient disk space
<j`ey>
David[m]123456789: so did you delete a partition or something?
<David[m]123456789>
No I didn't, I must have boot on the wrong Asahi startup disk (the one who did not complete the install)
<j`ey>
you should be able to get to 1TR or something then
<VinDuv>
David[m]12345678: any USB cable which can be connected to the USB-C port on the Mac and to a USB-A or USB-C port on the other machine should work
<VinDuv>
The other machine doesn’t even needs to have USB3 ports, USB2 will work fine
<VinDuv>
(ok given that this is USB maybe saying “any USB cable” is a bit optimistic :)
<David[m]123456789>
<j`ey> "you should be able to get to 1TR..." <- The installer did not finished due to a network outage, not the 1st of the 2 step install
<j`ey>
yeah, I saw the vid now.. not sure how that could have caused a boot loop :(
<David[m]123456789>
I should have deleted the wrong partition and don't assume the one on the rigth i s the most recent/rigth one
<David[m]123456789>
None, I just restarted on the stub for which the install did not finish
<j`ey>
Im not sure why you changed the startup disk to that one, if you knew the install didnt finish
kov has joined #asahi
bps has joined #asahi
kloenk has quit [Remote host closed the connection]
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
roxfan2 has joined #asahi
jmr2 has joined #asahi
<jmr2>
David[m]123456789: when you selected Asahi as the boot disk, how many options were avaiable there? Can you describe them all?
<jmr2>
For info, the order in which those are displayed is random.
<David[m]123456789>
2 options
roxfan has quit [Ping timeout: 480 seconds]
<jmr2>
Macintosh HD and Asahi ?
<David[m]123456789>
s/2 options/Machintosh HD + 2 Asahi L/
<jmr2>
OK. Any idea why there were 2 Asahi listed? I though you only managed to get the installer to run with the latest fix from a few hours ago. Did you run the installer twice since? Did it jam at the same place both times?
axboe has joined #asahi
eroux has joined #asahi
jmr2 has quit [Remote host closed the connection]
jmr2 has joined #asahi
<jmr2>
marcan: tested successfully. Thank you!
jmr2 has quit [Remote host closed the connection]
jmr2 has joined #asahi
<jmr2>
David[m]123456789: in case you didn't try already, try other patterns with the power button. Hold it even longer, or release it as soon as "Chargement des options de démarrage" appears.
<David[m]123456789>
a 16gb stick should be enough? That leaves 12 CB for the downloads
<David[m]123456789>
s/CB/GB/
<j`ey>
well i guess check how much space is actually used
<j`ey>
VinDuv did say 'you have to have sufficient space', so it might be quite big..
<jmr2>
If you check (I suggest "du -h"), what does it says?
<jmr2>
Given that it's a "live" OS, I wonder if the available space isn't limited by your RAM instead of the USB stick size.
<David[m]123456789>
<jmr2> "OK. Any idea why there were 2..." <- Due to a network outage it failed later but still installed itself, I then later installer it properly and reboot on what I though was the correct install
<David[m]123456789>
* Due to a network outage it failed later but still installed itself, I then later installed it properly and rebooted on what I though was the correct install
<David[m]123456789>
* Due to a network outage it failed but still installed itself, I then later installed it properly and rebooted on what I though was the correct install
<David[m]123456789>
It's seems to be copied on udev and tmpfs 2 times
<David[m]123456789>
s/udev/overlay/
<jmr2>
I'm booting a live debian to check exactly, but in any case, it doesn't matter. There's no partition larger than 7.7GB, so no way to store a 14GB file anywhere.
<David[m]123456789>
s/on/in/, s/udev/overlay/
<VinDuv>
When I restored the mini, idevicerestore downloaded a 13.2 GiB ipsw and extracted 11.8 GiB of files out of it for a total of 24.9 GiB used
<VinDuv>
(yes idevicerestore has a -l option that automatically downloads the right ipsw… but if you happen to run out of disk space while it downloads the files, it produces confusing error messages, unfortunately)
jmr2 has quit [Remote host closed the connection]
<David[m]123456789>
I downloaded it manually
jmr2 has joined #asahi
<David[m]123456789>
<jmr2> "I'm booting a live debian to..." <- So no way to do it from a live system?
<jmr2>
Looks like you need 24GB of free space. From my test, the root filesystem is indeed mounted on RAM, but on this system with 16GB RAM, I'm offered ~8GB of rootfs space, just as you are.
<jmr2>
You might be able to use the machine's disk and mount that.
<jmr2>
But I'd recommend not doing it on a FAT/NTFS filesystem. So you might need to repartition its disk, which, again, isn't risk-free.
<jmr2>
If you have an unused USB backup hard disk, that might be another option.
<jmr2>
Full disclosure: I've never done an idevicerestore recovery.
<jmr2>
I also kind of remember you telling that you have a lot of data on the Air. I'm not sure if idevicerestore will preserve any of that, you might want to consider other options.
<David[m]123456789>
I did a timemachine backup before
<jmr2>
OK. I have to go for now. Keep us posted.
jmr2 has quit [Quit: Page closed]
ifthenelse2 has joined #asahi
Raqbit7 has joined #asahi
ifthenelse has quit [Quit: Ping timeout (120 seconds)]
Raqbit has quit [Quit: Ping timeout (120 seconds)]
<mps>
my advice: use alpine linux edge (aka development release) and you will have bleeding edge software
<mps>
not that I promote alpine
<mps>
alpine is bad distro (I know) but other ones are worse
<mps>
this is my POV
<patience[m]>
why is it bad?
<mps>
patience[m]: if I explain it will be off-topic here, but try to use it and you will see
<mps>
my msgs above were intended for Glanzmann
thinkofher has joined #asahi
kendfinger has quit [Remote host closed the connection]
robher has quit [Read error: No route to host]
kendfinger has joined #asahi
robher has joined #asahi
the_lanetly_052___ has joined #asahi
axboe has joined #asahi
<Glanzmann>
mps: I'll probably continue to use Debian, maybe I give alpine a go. In the early days, I tried a lot of distros and Operating Systems, but Debian is what got stuck.
<Glanzmann>
povik: When I need bleeding edge software in Debian, I either use Debian and wait a little while or I pull it from sid and build a deb that I use, or compile and install it locally.
<Glanzmann>
But actually I like Debian because it is not bleeding edge and well tested.
<marcan>
there should be *some* way to get into system recovery if the paired recovery doesn't work, but I don't know what it is
<mps>
Glanzmann: sure, use what you feel is best for you (I don't want to tell anyone what to do)
<marcan>
let me try some stuff on my Air...
<axboe>
my main gripe with trying a new distro is the packaging system
<axboe>
I don't need to learn Yet Another One
<Glanzmann>
axboe: I agree, but i packaged more rpms in my live than debs. :-)
<Glanzmann>
axboe: Btw. the new m1-test-nvme branch is that the the one with half the cpu but get everything possible?
<axboe>
Glanzmann: that one was mostly just for sven to see the apple-nvme changes to support some of the effiency features
<axboe>
Glanzmann: outside of that, it's just sitting on top of the stuff I currently run on it
<Glanzmann>
I see.
<axboe>
which does have the "kill dead cores on my m1 pro"
<Glanzmann>
I saw that. :-)
<j`ey>
axboe: how do you kill what is already dead? :P
<axboe>
mostly good stuff in there, a few work-arounds that are specific to my setup
<axboe>
j`ey: with fire!
<Glanzmann>
axboe: I saw you have a keyboard backlight patch, what is necessary from the userland in order to use that?
<axboe>
Glanzmann: nothing
<Glanzmann>
Perfect.
<landscape15[m]>
Glanzmann: thanks. It’s the same script you shared in the docs right?
<axboe>
Glanzmann: fwiw, I also run an m1n1 hack to have it work with the disabled cores
<axboe>
just for full disclosure
eroux has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Glanzmann>
axboe: And another thing I wanted to ask, you. I have a lot of Samsung SSDs with uncorrectable read errors that I bought last yaar with uncorrectable read errors, evo 2TB/4TB, qvo 4TB, NVMe pro 980 2 TB. Pattern is everywhere the same. Debian does a raid check every month, or I read a lot of data, and the uncorrectable read errors pop up. I did not loose anything, but I wonder if you heard something. I
<Glanzmann>
see more and more people on the internet seeing simliar things. For me the pattern is always the same barely used SSD ( 10 - 30 TB written in 3 months), uncorrectable read erros. Out of 20 disks I bought last year, 15 show these symptoms.
<Glanzmann>
landscape15[m]: Yeah, it is the same script, I test the new stuff and if I consider it stable, I put it in there.
<Glanzmann>
I saw one bug with tagged command queueing trims, I trimmed the drives, same issue pops up 4 weeks later during raid check.
kendfinger has quit [Read error: No route to host]
tardyp has quit [Remote host closed the connection]
mmarchini has quit [Remote host closed the connection]
robher has quit [Remote host closed the connection]
kendfinger has joined #asahi
mmarchini has joined #asahi
robher has joined #asahi
tardyp has joined #asahi
<mps>
where is the keyboard backlight patch?
<marcan>
ok, I found the override
<marcan>
David[m]123456789: when in the bootloop, double-press and hold the power button before the apple shows up
<Glanzmann>
mps: Let me know how it goes, maybe i try it, too.
<Glanzmann>
But need to get poviks macbook air speaker development setup ready first.
<mps>
Glanzmann: applied patch and looking in .config, though don't see changes for DTS
the_lanetly_052___ has quit [Remote host closed the connection]
<Glanzmann>
mps: I think they're already in. because in axboes tree there is patch for adding it to m1 max/pro but it that patch it appears they're already in: https://git.kernel.dk/cgit/linux-block/log/?h=m1-test Search for 'Andreas Heider' patches.
mmarchini has quit [Read error: No route to host]
mmarchini has joined #asahi
tardyp has quit [Read error: No route to host]
robher has quit [Read error: No route to host]
kendfinger has quit [Read error: No route to host]
kendfinger has joined #asahi
tardyp has joined #asahi
robher has joined #asahi
the_lanetly_052 has joined #asahi
<mps>
Glanzmann: I don't see backlight word in dts in spmi/work jannau's branch
<Glanzmann>
mps: I think you have to look for fpwm
<mps>
fpwm is there
<mps>
but not backlight
<j`ey>
jannau's branch doesnt have the backlight, that's only in that branch
<mps>
j`ey: ah, that is, thanks
<mps>
I think this will not work on t8103 without dts patching
<mps>
but will build it to see if it will not break something
<ChaosPrincess>
how do apple do kernel updates w/o a trip through 1tr? just by being apple and having special signatures or a mechanism that we can use?
<Glanzmann>
But the latest debian installer video had u-boot already installed. But in none of the videos I creaed a '3gb' partition. Btw. the stub is 2.5 GB.
<marcan>
ChaosPrincess: yes
<j`ey>
Glanzmann: so does the installer create the stub parition from the freespace?
<j`ey>
Glanzmann: and yes i was using 3gb as an estimate for 2.5 :P
<ChaosPrincess>
marcan: okay, s/or/xor/
<marcan>
the former
<marcan>
they don't use the fuOS mechanism which requires 1TR to provision a kernel
<Glanzmann>
j`ey: Yes, the installer creates the stub partition from free space, it is a prerequisit.
<j`ey>
Glanzmann: so if there's only 2.5g there it can just use all of that space?
<Glanzmann>
j`ey: 2.5GB is used by the asahi installer for the stub. Than you need extra free space if you want a esp and root partition.
<Glanzmann>
j`ey: Or if you have the stub, of course you can also boot from usb using u-boot or tethered using m1n1.
<axboe>
Glanzmann: an ssd should never throw uncorrectable read errors, that's a huge red flag
<Glanzmann>
axboe: I know. And I think I'm starting to return them all. I already started using Intel DC which are slower and more expansive, but hopefully are stable.
<axboe>
Glanzmann: I've been using samsung in other boxes and my other laptop, I've never seen that. I'm curious if it's some side effect of an issue with eg queued trim like you mentioned, because if not, the fw is pretty buggy
<Glanzmann>
axboe: After I told the issue a friend of mine, he did a dd if=/dev/sdX of=/dev/null bs=8M over is evo 4TB, guess what .... 17 MB were not readable. He returned the drive after evacuating the data and bought a micron.
the_lanetly_052__ has joined #asahi
<axboe>
the way ssds do error correction these days, it should rewrite the block if the read confidence is marginal
<axboe>
if after a straight overwrite, defective hw for sure
<Glanzmann>
axboe: In the past I had a lot of Samsung SSDs 840/850 EVOs. Actually 36 of them, I used them for 8 years+, no a single one had any issues whatever. I'm still using them today.
<axboe>
I generally find them pretty solid too, some models end up being dodgy but most are good
<Glanzmann>
That is what I thought, I'll replace them all, I guess. A lot of money ...
<axboe>
Glanzmann: yeah, that sucks...
<mps>
Glanzmann: bad cables or adapter (that is my experience)
<axboe>
cabling would usually show transport issues, not uncorrectable media errors
<axboe>
though of course who knows what'll happen with a marginal setup
the_lanetly_052 has quit [Ping timeout: 480 seconds]
<axboe>
should not show up in smartlog from the drive
jmr2 has joined #asahi
robher has quit [Read error: No route to host]
tardyp has quit [Read error: No route to host]
robher has joined #asahi
<jmr2>
marcan: that video is a great testimony to your old "Safe Hacking" blog post. Thanks for caring.
tardyp has joined #asahi
<marcan>
I'm going to fix the installer so this can't happen too
<marcan>
in fact it couldn't in 11.x, this is a 12.x regression due to the paired recovery
<kov>
now that I read marcan's thread on twitter I think I actually managed to get into a similar state once, but since I had another mac I used their Apple Configurator 2 thingy and just reflashed
<jmr2>
Out of curiosity, once booted in a recovery, how can you tell which one was used?
<j`ey>
I think the device tree has it?
<marcan>
it does
<marcan>
the installer will tell you too
<j`ey>
(hence why step2.sh errors out)
<jmr2>
OS Type in the output of bputil -d ?
<j`ey>
yeah
<j`ey>
or Current OS environment:
<jmr2>
Let me retry it. BRB.
jmr2 has quit [Quit: Page closed]
<sami-dev[m]>
Making the jump to a M1 Max, super excited to follow this project more closely. I'll be sure to drop some money on sponsors/patreon as well. Does anyone know if it's better to donate through sponsor or patreon (not sure if github takes a cut like patreon might)
jmr2 has joined #asahi
<jmr2>
So from MacOS, that says macos. From shutdown, booting with a long press returns "one true recovery", and double-click and hold returns "ordinary recovery os".
<jmr2>
Is the systemwide recovery the "ordinary recovery os", or is there a 4th option ?
<ChaosPrincess>
on 12.x, its always the paired recovery, unless it is broken, and then its systemwide recovery
<ChaosPrincess>
1tr vs normal recovery is the same image, just the security coprocessor configuration is different
<axboe>
curious if anyone have experienced: "Synchronous Exception at 0x664C" when running qemu?
<marcan>
sami-dev[m]: github has no fees :)
<marcan>
"ordinary recovery os" means any recovery OS that was not booted through the very special 1TR process
<j`ey>
axboe: that sounds like it's an error fro UEFI itself?
<marcan>
that field does not tell you what recovery OS was booted
<marcan>
the DT stuff does
<marcan>
that field tells you what *mode* the system is booted in
<axboe>
j`ey: good point, I was a bit puzzled in figuring out where it's coming from
<axboe>
I get it sporadically here
<axboe>
j`ey: will try and remove the efi bios argument, boots fine without it
<opticron>
hmmm...maybe I'll swap over to github, then
<jmr2>
Thanks.
<j`ey>
can you pay on paypal with github?
jmr2 has quit [Quit: Page closed]
<Glanzmann>
mps / axboe: This was with different cables, mothboards, power supplies, and datacenters. Also with downgraded link speed, disabled tagged command queuing and disabled trim. So I tried a lot of stuff. For now I'll keep my hands away from Samsung. But the Samsung datacenter edition doesn't seem to have the issues. At least only one drive failed.
<Glanzmann>
jmr: I also always end up in recovery os instead of 1tr. No idea why, but I'm certain that I don't let go of the power button and that it is off. Before I try. See here: https://tg.st/u/asahi-installer-u-boot.mp4
<Glanzmann>
jmr2: ^
<axboe>
15 second press always does it for me, the two minor blinks at 10 and 11 seconds help my synchronize it :)
<Glanzmann>
j`ey: Yes
<Glanzmann>
hrhr. I always count 21, 22, ...
<sami-dev[m]>
<marcan> "sami-dev: github has no fees :)" <- Great! Is there posts on github as well about progress etc?
<sami-dev[m]>
I know patreon has that functionality but unsure about github 😅
<marcan>
I do mail out things via GH/Patreon when I post an update, but the updates themselves go on the blog, yeah
<marcan>
and thank you for helping out :-)
<j`ey>
🎉 You sponsored @marcan for $100
<marcan>
\o/
<j`ey>
marcan: ty for the continious hard work!
<marcan>
j`ey: you really didn't have to, you're helping out tons already ^^
<j`ey>
marcan: :-)
* marcan
off to sleep
<tpw_rules>
i seem to recall the next revision of linux-asahi is coming soon?
<marcan>
yes, I did some git cleanup today
<marcan>
hope to wrap it up tomorrow
<tpw_rules>
excellent. sleep well
<mps>
and sweet dreams
<sami-dev[m]>
You're now sponsoring @marcan for $24 a month
<sami-dev[m]>
I'm just a simple webdev so while I can't help out with kernel things I can at least help out with something so people like you can help me run my favorite OS on some really nice hardware ;)
bisko has joined #asahi
<marcan>
thank you! :)
MajorBiscuit has quit [Ping timeout: 480 seconds]
axboe has quit [Quit: leaving]
<Glanzmann>
tpw_rules: I read a line in -dev that hinted that marcan thinks about doing the first release within a month.
<tpw_rules>
maybe my m1 max will finally be here by then
<David[m]123456789>
That worked! Thanks a lot marcan @_oftc_marcan:matrix.org
<kov>
removing the stub and re-running the install worked, but I ended up in a grub rescue prompt due to the numbering of the partitions changing, setting the prefix manually, loading the normal module and running it was enough to boot properly though \o/
<kov>
(and re-installing grub after that of course)
jmr2 has joined #asahi
<jmr2>
David[m]123456789: first time I see this one. Is it the same username that's displayed when step2.sh prompts you for your password the first time ?
jmr2 has quit [Quit: Page closed]
jmr2 has joined #asahi
<jmr2>
David[m]123456789: I believe your second screenshot is from the MacOS login screen. If this is the case, this is your full name, which may or may not be the same as your account name.
<jmr2>
Go back to MacOS, open System Preferences, Users & groups, unlock the lock, right click on your account, select advanced options.
<jmr2>
You're looking for the Account name field.
<jmr2>
I'm going to guess david, with lowercase D.