marcan changed the topic of #asahi to: Asahi Linux: porting Linux to Apple Silicon macs | "Does XXX work yet?": https://alx.sh/fs | GitHub: https://alx.sh/g | Wiki: https://alx.sh/w | Topics: #asahi-dev #asahi-re #asahi-gpu #asahi-alt #asahi-stream #asahi-offtopic | Keep things on topic | Logs: https://alx.sh/l/asahi
lactose has quit [Read error: Connection reset by peer]
lactose has joined #asahi
novaandromeda has left #asahi [#asahi]
lactose has quit [Read error: Connection reset by peer]
lactose has joined #asahi
rvalue has quit [Ping timeout: 480 seconds]
lactose has quit [Read error: Connection reset by peer]
rvalue has joined #asahi
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
kidplayer666 has quit [Quit: Connection closed for inactivity]
SalimTerryLi has quit [Ping timeout: 480 seconds]
kitlith has joined #asahi
<kitlith>
I have two Asahi Linux installs on my macbook, which I use for different purposes. I needed to resize the install that was at the end of the disk, so I ended up moving the partitions of both installs closer to the beginning of the disk. Now, the boot entries for both installs appear to boot to the same installation: the one closer to the beginning of the disk, and not the one I resized the data
<kitlith>
partition on. Any ideas?
Brainium has quit [Ping timeout: 480 seconds]
hspak3 has joined #asahi
kitlith has quit [Remote host closed the connection]
kitlith has joined #asahi
hspak has quit [Ping timeout: 480 seconds]
hspak3 has quit []
hspak3 has joined #asahi
<tpw_rules>
you probably changed the partition GUIDs
<tpw_rules>
it's unclear what exactly broke, do they boot to the same grub or different grubs?
<kitlith>
u-boot says they both boot to nvme0:5, iirc
<kitlith>
I don't know how to tell which instance of u-boot it is.
<tpw_rules>
i'm not sure if that's possible
<tpw_rules>
do you have a backup of the partition table? that would be the easiest way to fix this
jakesjt has joined #asahi
jakesjt has quit [Quit: leaving]
jakesjt has joined #asahi
<nicolas17>
well, you can't just restore the partition table
<tpw_rules>
yes, but you can read the GUIDs out of it and re-apply them
<nicolas17>
ah yes
ciara has quit [Remote host closed the connection]
<kitlith>
UUIDs appear to be correct, at least according to /etc/fstab and lsblk -f
<kitlith>
is there another UUID that I'm blanking on?
<tpw_rules>
the PARTUUID in the GPT shown by gdisk
<tpw_rules>
i don't know if lsblk and blkid show that one, i think they show the one in the filesystem
<tpw_rules>
it's also the ESP that is wrong, not the rootfs
ciara has joined #asahi
<tpw_rules>
if you can't find it, the second easiest way is probably to back up your ESPs, destroy them and the corresponding stubs, then create two new UEFI-only ones using the installer and restore their contents
Freddy5Bear has joined #asahi
<tpw_rules>
the correct UUID is in the special blessed boot object and i'm not immediately sure if there's a way to pull that back out. it's certainly not changeable without apple's tools
<kitlith>
yeah, lsblk -f outputs the filesystem uuid, not the partuuid.
lactose has joined #asahi
bertrand has quit [Ping timeout: 480 seconds]
bertrand has joined #asahi
lactose has quit [Read error: Connection reset by peer]
malfunction54 has quit [Remote host closed the connection]
malfunction54 has joined #asahi
<kitlith>
I can read the boot.bin from the APFS partition, from when the installer was installing it. partuuid shows up in strings.
<tpw_rules>
ok, then use gdisk to set it back. i think you need x for the advanced menu then c and it will let you type it back in
<tpw_rules>
did you just strings the partition? it should be after chainload= and like asahi,efi-system-partition=
<kitlith>
I strings the boot.bin, and yes, that's where it was
<kitlith>
it appears to already be correct, though.
<tpw_rules>
for both partitions?
<tpw_rules>
both ESPs
<kitlith>
for the one I was having issues with at least, I'll go check the other now.
jakesjt has quit [Remote host closed the connection]
<kitlith>
yes, the partition UUIDs appear to be correct for both EFI partitions, at least according to macos.
<tpw_rules>
ok, i'm not sure what could be up then
<tpw_rules>
i'm not sure if apple's stuff can get confused about which stub to boot
<kitlith>
is there an argument m1n1 passes to u-boot that might help for seeing what's going on?
<tpw_rules>
that asahi,efi-system-partition should show up in the device tree under /chosen if you want to check in each u-boot and make sure it's different
<tpw_rules>
basically, apple's stuff does apple's stuff and loads m1n1 from a secret place somehow related to that APFS partition. that m1n1 searches the attached ESP (pointed to by chainload) for another copy of itself. that second copy has u-boot attached and that gets loaded. u-boot then checks /chosen/asahi,efi-system-partition to check which ESP to load a bootloader off. then that bootloader does whatever to find the rootfs and start linux
<tpw_rules>
(apple's stuff may even load m1n1 from the APFS partition, not 100% sure)
<tpw_rules>
so it's not really clear to me which step is getting mixed up
<tpw_rules>
i think you can also query u-boot for which ESP it's using
<kitlith>
I feel like I'm missing something in regards to how you query the device tree from u-boot. dm doesn't feel like the right command, and I can't seem to figure out how to set the FDT address to make fdt work.
<tpw_rules>
fdt should work by default, it shouldn't need an address
<tpw_rules>
but iirc there's an environment variable fdt_addr_r
<tpw_rules>
with the address
<kitlith>
apparently it was fdtcontroladdr instead
<tpw_rules>
so you're able to read /chosen ?
<kitlith>
yep. now time to do it again to check if it's different.
<kitlith>
They are different!
<tpw_rules>
so then it has to be a problem with grub or whichever bootloader you use
<tpw_rules>
and just to be clear, they are both different and also both correspond to the PARTUUID of a partition on your disk
<kitlith>
the scanning/booting logs come from u-boot though, right?
<tpw_rules>
i'm not sure, it depends which you're talking about. m1n1 won't put anything on the screen unless it can't find the second copy of itself. but it could also be grub or the kernel
<kitlith>
I didn't take an image of the second one, but the one I care about does, indeed, correspond to a partition on my disk
<tpw_rules>
the one you care about is the one that's booting the wrong linux?
<kitlith>
yes
<kitlith>
it's weird because that chosen partition is the 9th partition, but u-boot (or something) seems to keep booting the 5th partition (the working Linux install)
<tpw_rules>
do your ESPs have the EF00 type code in gparted?
<tpw_rules>
s/gparted/gdisk/
<kitlith>
yes, they have the EF00 type code.
<tpw_rules>
i'd have to see the log to be sure what you're talking about
<kitlith>
efi-system-partition lines up with partition 9, type lines up with what you would expect from an ESP, boot flag doesn't seem to be shown there
<tpw_rules>
yeah it all does
veloek has joined #asahi
<tpw_rules>
i'm sorry but i think i've about reached the limits of static analysis
<tpw_rules>
i return to my previous recommendation of back up the ESPs and wipe the stubs
bertrand has quit [Read error: Connection reset by peer]
bertrand has joined #asahi
<kitlith>
well, going back to my idea of modifying the boot order in eficonfig, that seems to do the trick for a single boot.
<kitlith>
it just doesn't persist.
<kitlith>
(which was to be expected)
<tpw_rules>
what if you do part list nvme 0 -bootable
<kitlith>
it lists partitions 5 and 9
<kitlith>
I don't see any reason why it *wouldn't* boot partition 5 anywhere
<kitlith>
since that'd be the first one that scan_dev_for_boot_part will find
<tpw_rules>
it shouldn't be going through that path
<tpw_rules>
it should be running the boot manager
<kitlith>
am I parsing something wrong in the path of commands from bootcmd -> distro_bootcmd -> bootcmd_nvme0 -> nvme_boot -> scan_dev_for_boot_part?
chadmed has quit [Ping timeout: 480 seconds]
chadmed has joined #asahi
chadmed_ has quit [Ping timeout: 480 seconds]
<tpw_rules>
that eventually does scan_dev_for_boot then scan_dev_for_efi then run boot_efi_bootmgr
<kitlith>
Going back to what you said about how the fdt should already be loaded -- I think you're right, and it's broken for some reason, which is messing up the overriding of efi boot order afterwards
<tpw_rules>
i checked on my machine and i'm not. fdt_addr_r doesn't point to a valid fdt if i interrupt the autoboot
<kitlith>
ah. and fdt list complains that no addr is set?
kitlith has quit [Remote host closed the connection]
kitlith has joined #asahi
kitlith has quit [Remote host closed the connection]
kitlith has joined #asahi
kitlith has quit [Remote host closed the connection]
kitlith has joined #asahi
bertrand has quit [Remote host closed the connection]
bertrand has joined #asahi
bertrand has quit [Remote host closed the connection]
bertrand has joined #asahi
bertrand has quit [Remote host closed the connection]
bertrand has joined #asahi
possiblemeatball has quit [Quit: Quit]
bertrand has quit [Remote host closed the connection]
bertrand has joined #asahi
bertrand has quit [Read error: Connection reset by peer]
bertrand has joined #asahi
<kitlith>
I think I'm done for the night, but I am somewhat interested in digging deeper at some point. will check logs tomorrow
kitlith has quit [Read error: Connection reset by peer]
crabbedhaloablut has joined #asahi
chadmed_ has joined #asahi
crabbedhaloablut has quit [Read error: Connection reset by peer]
crabbedhaloablut has joined #asahi
crabbedhaloablut has quit [Read error: Connection reset by peer]
crabbedhaloablut has joined #asahi
john-cabaj has quit [Ping timeout: 480 seconds]
crabbedhaloablut has quit [Ping timeout: 480 seconds]
bertrand has quit [Ping timeout: 480 seconds]
rvalue has quit [Ping timeout: 480 seconds]
crabbedhaloablut has joined #asahi
chadmed_ has quit [Ping timeout: 480 seconds]
chadmed has quit [Ping timeout: 480 seconds]
crabbedhaloablut has quit [Ping timeout: 480 seconds]
lactose has joined #asahi
zawad has joined #asahi
zawad has left #asahi [#asahi]
zawad has joined #asahi
kidplayer666 has joined #asahi
gotlou has joined #asahi
zawad has quit []
zawad has joined #asahi
<gotlou>
hello, I was looking at the great work the asahi team did on speakersafetyd and was wondering how much additional work would be required to get it running on intel macs
<gotlou>
specifically the last intel macs from 2020, which are the closest machines to the m1 macs
chadmed has joined #asahi
zawad has quit [Remote host closed the connection]
holiday has joined #asahi
<kidplayer666>
Hey, I’m trying to install PyQt6, but it is requiring an extremely high amount of RAM, and it is using up all my 8Gb of ram, plus the additional 24GB of swap, and it still crashes my laptop. Is this normal? How can I make it use less RAM?
<kidplayer666>
gotlou: I get a feeling that MacBook with touchbar probably is the easiest to port as they probably reused most stuff. Everything else probably needs proper testing as they changed the chassis if I’m not mistaken
bertrand has joined #asahi
<gotlou>
kidplayer666: I see, what about the macbook air? afaik the m1 version is just an intel mba with no fan and an m1 instead of an intel chip, and a slightly different keyboard
<kidplayer666>
gotlou: I am not sure, marcan and chamed probably know it a looot better than me
maximbaz has joined #asahi
crabbedhaloablut has joined #asahi
gotlou has quit [Remote host closed the connection]
<chadmed>
gotlou: the amount of work required will be quite significant as the intel macs are a weird mix of SOF, HDA and codec drivers
<chadmed>
and its not really clear how it's all supposed to mesh together post-2017-ish
<chadmed>
speakersafetyd expects/requires specific kernel level features which do not exist outside of our fork currently, and are kinda tightly bound to ASoC
rvalue has joined #asahi
rvalue has quit []
rvalue has joined #asahi
crabbedhaloablut has quit [Read error: No route to host]
crabbedhaloablut has joined #asahi
rvalue has quit []
crabbedhaloablut has quit []
rvalue has joined #asahi
rvalue has quit []
crabbedhaloablut has joined #asahi
rvalue has joined #asahi
rvalue- has joined #asahi
chadmed_ has joined #asahi
chndio has quit [Remote host closed the connection]
chndio has joined #asahi
rvalue- has quit []
rvalue has quit [Ping timeout: 480 seconds]
chndio has quit [Remote host closed the connection]
chndio has joined #asahi
AnuthaDev has joined #asahi
rvalue has joined #asahi
r0ni has quit [Ping timeout: 480 seconds]
capta1nt0ad has joined #asahi
chadmed has quit [Remote host closed the connection]
<capta1nt0ad>
Hello, I messaged here a few weeks ago (December 20th) regarding speaker support on the legacy Arch Linux ARM Asahi distribution (which I am currently unable to reinstall due to time constraints and the amount of packages/configuration I have installed and made.) I noticed that the new version of lsp-plugins was released on the 24th and has been merged into ALARM. Can I expect the speakers to be enabled in ALARM soon, or will
<capta1nt0ad>
I have to find a time to switch to Fedora before I can use the speakers?
lactose has quit [Read error: Connection reset by peer]
<capta1nt0ad>
Sorry if I am seeming a little impatient, that's not my intention at all.
rvalue has quit [Ping timeout: 480 seconds]
<kettenis>
kitlith: if the fw_dev_part u-boot env variable isn't set, you must be running a u-boot that doesn't have the required patches
rvalue has joined #asahi
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
nela has quit [Ping timeout: 480 seconds]
AnuthaDev has quit []
AnuthaDev has joined #asahi
AnuthaDev has quit []
AnuthaDev has joined #asahi
chadmed has joined #asahi
<chadmed>
capta1nt0ad: not sure when new packages will be built but you can build your own packages if you really need to. the pkgbuilds should be there in the repo
<capta1nt0ad>
chadmed: ok, thank you.
<j`ey>
and there's PRs with updated packages
<chadmed>
just be aware that at this point alarm is effectively an -alt distro and very little effort will probably but into its maintenance going forward
<j`ey>
I think joske or someone should just take that over
lactose has joined #asahi
Misthios has quit [Quit: Misthios]
lactose has quit [Read error: Connection reset by peer]
kidplayer666 has quit [Quit: Connection closed for inactivity]
gotlou has joined #asahi
<gotlou>
chadmed: thanks for the clarification! I guess something like easyeffects can get similar effect (albeit volume won't go as high as on macos)
kidplayer666 has joined #asahi
<chadmed>
gotlou: speakersafetyd has absolutely nothing to do with the DSP/EQ
<chadmed>
that is entirely driven by pipewire and LV2 plugins
<chadmed>
no porting required there, we developed everything upstream
farghal has joined #asahi
farghal has quit []
<chadmed>
pipewire 0.3.85 and newer will Just Work(tm) as long as you follow the conventions set in asahi-audio
<chadmed>
fwiw the speakers are ~90% safe with _just_ the DSP but i wouldnt risk it for your users if you dont have some sort of last-ditch SCRAM like speakersafetyd unless you seriously drop the max volume in the IRs
<gotlou>
thanks for the explanation. from what I recall the audio system on macos adjusts the voltage on the speakers to just under the limit of what they can handle for better output?
<gotlou>
that part is speakersafetyd? and the actual DSP is done by interfacing with pipewire similar to easyeffects?
<chadmed>
correct
<gotlou>
I see
<gotlou>
so for conventional PCs, the missing part is the "driving speakers to their limit" for better audio?
<chadmed>
no the whole stack is missing
<chadmed>
PCs dont do any of this
chaos_princess has quit [Ping timeout: 480 seconds]
<gotlou>
PCs can do DSP with easyeffects for instance, but the differentiator with macs is adjusting speaker voltage? that isn't something that you can do on a PC unless you tune it for each speaker?
<sven>
how are the speakers routed in the Intel Macs? Does the signal go through the t2 there?
<chadmed>
HDA and SOF do not implement anything that could communicate this back to Linux unless the codecs themselves on the t2 macs are exposed to the kernel
<chadmed>
and speakersafetyd doesnt "adjust voltage" it _measures_ the voltage and current through the voice coil and then just turns the kcontrol volume down for any speaker that is at risk of overheating
<chadmed>
from my experience with how sound works on pre-arm macs theres nothing you can really do to implement speakersafetyd-like functionality
<chadmed>
you would have to use, e.g. our "dry" model which just computes an estimate from some "sane" values
<chadmed>
fwiw this is also how macos does it on pre-arm macs and even a couple of the arm ones
capta1nt0ad has quit [Remote host closed the connection]
Brainium has joined #asahi
SalimTerryLi has joined #asahi
jadesoturi_ has quit [Quit: Even death needs to sleep sometimes...]
r0ni has joined #asahi
bertrand has joined #asahi
chadmed has quit [Remote host closed the connection]
chadmed_ is now known as chadmed
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
chir4gm8277 has quit [Ping timeout: 480 seconds]
jadesoturi has joined #asahi
knedlik has joined #asahi
<nicolas17>
chadmed: wait, what *is* the state of audio on the latest Intel Macs?
<chadmed>
a clusterfuck :p
<nicolas17>
is it bad sounding, and/or unsafe?
<sven>
that sounds like the state of most things on those machines :(
<chadmed>
some of them are "supported" (in the sense that the woofers will blare out the worst muddiest most disgusting thing youve ever heard), some of them have no support, some of them kinda work but only line out, etc
<chadmed>
and yes im like 80% sure if you wire up the tweeters via Pro Audio profile trickery you will melt them eventually
<chadmed>
the 13" touchbar era chassis has no tweeters though just two pairs of woofers
<chadmed>
and they sound awful even with processing
<nicolas17>
does the 13" touchbar apple-silicon have tweeters?
<chadmed>
nope
possiblemeatball has joined #asahi
chndio has quit [Remote host closed the connection]
chndio has joined #asahi
chndio has quit [Remote host closed the connection]
lactose has quit [Read error: Connection reset by peer]
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
knedlik has quit [Quit: knedlik]
knedlik has joined #asahi
possiblemeatball has quit [Quit: Quit]
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
Bogdan513 has joined #asahi
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
holiday has joined #asahi
chaos_princess has joined #asahi
<MichaelLong>
hm this is weird, has anybody tested USB-C hubs (not necessarly USB-C docks) with the current asahi-fedora? For me, in dmesg theres no USB device insert event happending when plugging in any kind of device, be it a USB-storage or HID into the hub. However when re-plugging the USB-C hub, attached devices are recognized. Can anybody reproduce this?
lactose has joined #asahi
lactose has quit [Read error: Connection reset by peer]
pjakobsson has quit [Remote host closed the connection]
holiday has quit [Quit: WeeChat 4.1.2]
knedlik has quit [Quit: knedlik]
julio7359 has joined #asahi
knedlik has joined #asahi
gspbirel56 has quit [Quit: Ping timeout (120 seconds)]
holiday has joined #asahi
gspbirel56 has joined #asahi
SalimTerryLi has quit [Remote host closed the connection]
lactose has joined #asahi
lactose has quit [Remote host closed the connection]
lactose has joined #asahi
Szadek636700707897 has quit []
lactose has quit []
Szadek636700707897 has joined #asahi
cow123 has quit [Remote host closed the connection]
possiblemeatball has joined #asahi
zkrx has quit []
zkrx has joined #asahi
bertrand has quit [Remote host closed the connection]
bertrand has joined #asahi
<holiday>
is there any way to run lutris under FEX?
<holiday>
or anything other than steam?
<j`ey>
cant you run FEXbash lutris?
<j`ey>
lutris seems to be packaged by fedora, so i guess you can just dnf install that in rhe vm
<holiday>
didnt want to work before
<holiday>
im retying it rn
rvalue has quit [Ping timeout: 480 seconds]
rvalue has joined #asahi
<holiday>
j`ey: tried "sudo dnf install lutris" but it asks me for a root pass
<j`ey>
try without sudo, you might be root already
<holiday>
tried, it tells me it needs sudo
<holiday>
i think just the guy who created the guide knows the password
<j`ey>
type `whoami`
<holiday>
"steam"
<holiday>
makes sense since the guide is made specifically for steam
<j`ey>
yeah this is far beyond what I know now, maybe you need an x86 rootfs that has python in.. or maybe you can run native lutris to download games and then fexbash the games directly..
<holiday>
i think i will try to understand how krunvm works without a prebuilt machine and then experiment with it further and see if i can get something running
<Nefsen402>
Might be worth installing a ssh server to see if the whole thing locks up
<Nefsen402>
If not, creating a seatd instance so you can launch a compositor from your ssh session
<Nefsen402>
maybe having a compositor boot back up will kick it out of the bad state
<zeromind>
yeah, that's a good idea -- also using sway on the work notebook, launching via tty, exiting sway makes the notebook go into standby for some reason
<zeromind>
*(intel) notebook
<Nefsen402>
I run sway on a old 2015 macbook pro with intel + amd discrete graphics and never saw that
<Nefsen402>
plus on a couple desktop computers
kidplayer666 has joined #asahi
<kidplayer666>
might be a weird question, but is it possible force flatpaks to use the system mesa isntead of what they're shipped with
<kidplayer666>
if not, anyone else who gets more than 30GB of ram use while trying to install PyQt6?
<kidplayer666>
because i am using 8gb and a 24gb swap and it is not enough
pthariensflame has joined #asahi
zawad has joined #asahi
pthariensflame has quit []
<i509vcb>
kidplayer666: I recall that kicad had an issue because it uses the sdk. I think there is something to do with the mesa platform flatpak uses doesn't actually have the asahi driver in it
<kidplayer666>
yeah, i know
<kidplayer666>
like most flatpak apps it uses the normal mesa version
<kidplayer666>
i am trying to use a whatsapp client
<kidplayer666>
it runs on python and has pyqt6 as a dependency
<i509vcb>
if you need whatsapp and are fine without using the specific electron frontend for it, you can use a pinned tab in firefox
<kidplayer666>
but installing that occupies more than 32gb of ram apparently
<i509vcb>
I'd be curious if the amount of ram is lowered if you limit how many jobs it's built with?
<kidplayer666>
i prefer to have a dedicated app to be honest
<kidplayer666>
idk how to do that
<kidplayer666>
i am just doing a "pip install pyqt6"
<kidplayer666>
and at some step it is using qmake
<kidplayer666>
because i had to install it before being able to even try to use pip install pyqt6
<i509vcb>
can you get qmake to emit a bunch of makefiles and then manually invoke it with a lower job count?
<i509vcb>
or is pip giving you no control over how qmake is called?