search

AsahiLinux / linux

Linux kernel source tree
Other
2.2k stars 85 forks source link

[TRACKER] linux-asahi-edge issues #70

Open marcan opened 1 year ago

marcan commented 1 year ago

This is a global tracker bug to group reports related to linux-asahi-edge. Since this is a bleeding-edge kernel, we know some things are broken and we make no promises that we'll get to reports in any amount of time. Please post any issues as comments on this bug, and we'll handle them or break them out into their own bugs as we deem appropriate. Please do not open new bugs for linux-asahi-edge issues that are not reproducible with the default linux-asahi kernel.

Only report issues here if they happen with linux-asahi-edge and NOT with linux-asahi. If you find a problem unrelated to the new features of linux-asahi-edge, you must test across at least 5 reboots on linux-asahi to make sure it never happens there and is truly a regression related to linux-asahi-edge.

Please make sure you have updated all your packages and have asahi-meta installed before reporting anything. Setups without asahi-meta installed are not supported and are expected to randomly break as we make changes to our packages. All Asahi installs since the first official alpha release come with asahi-meta preinstalled (and also asahi-desktop-meta for the desktop builds). If you choose to uninstall our infrastructure packages, you are on your own and we cannot provide support.

Known issues:

When writing a comment, please run asahi-diagnose and attach the resulting logfile.

aykevl commented 1 year ago

Maybe this is already known but I'm reporting this here just in case...

Brightness control does not seem to work for me with the 6.1.0-rc6-asahi-4-1-edge-ARCH kernel, on a Macbook Pro (16 inch, M1 Pro, 2021):

(If there is any way I can help I'd certainly be willing to: brightness control is my number one issue with Asahi Linux right now).

marcan commented 1 year ago

Maybe this is already known but I'm reporting this here just in case...

Brightness control does not seem to work for me with the 6.1.0-rc6-asahi-4-1-edge-ARCH kernel, on a Macbook Pro (16 inch, M1 Pro, 2021):

* It doesn't work with the normal ways of changing brightness (brightness keys or slider in the KDE power taskbar widget). I think it was reported as the lowest brightness setting in KDE on startup. Even going all the way to the lowest setting doesn't affect brightness (the screen is still on).

* When I change `/sys/class/backlight/apple-panel-bl/brightness` manually, there is no change in display brightness (but the value is kept in the sysfs file).

(If there is any way I can help I'd certainly be willing to: brightness control is my number one issue with Asahi Linux right now).

Your mention of gpio-bl before the edit made me realize that shouldn't be there. Please update your kernel and try again. I don't know whether this will fix your issue, but that was definitely wrong.

aykevl commented 1 year ago

Whoa, I updated the kernel and the issue seems to be gone now? I can now adjust my screen brightness. I hope it stays that way, if not I'll report back.

Interestingly, before the update I booted into asahi-linux-edge, but the screen didn't turn on (or rather, it turned off just before the login screen). This happened twice. I had to reboot into the regular asahi-linux kernel to be able to install the kernel update.

(One nit I still have is that the brightness change steps are a bit too large to my taste. But that's a KDE issue. I suppose KDE wasn't made for screens with such a large range of brightness values).

(Somehow the kernel update also fixed a different issue I was about to report: before the update Alt + arrow keys would switch between TTYs which is rather inconvenient when you use them to go back/forward in web browsers).

(EDIT: sadly DCP support makes yakuake far too slow to be usable for me so I'm afraid I'll switch back to a non-DCP kernel even if that means no brightness control).

ghost commented 1 year ago

Hi, I can not mount my USB2 USB key with both asahi-linux-6.1rc6 and asahi-linux-edge through a USB dongle. Is it a common problem or I'm the first one reporting it?

EDIT: last time I've tried AsahiLinux there was 6.1 rc4 and it used to works as expected.

EDIT 2: connecting the USB key to the dongle and then the dongle to my MacBook did the trick.

EDIT 3: during boot both u-boot and grub see the USB key while the initramfs' shell doesn't. That's a problem for these like me who want to install another OS, like gentoo, over AsahiLinux.

jannau commented 1 year ago

Restoring brightness after DPMS requires updated brightness to DAC mappings. The current code based on the tables in the ADT results in an increasing brightness for each cycle.

jannau commented 1 year ago
Capta1nT0ad commented 1 year ago

The 3.5mm jack is working in linux-asahi but not in linux-asahi-edge. I get no sound and KDE doesn't recognise my headphones. However, I do hear a crackle through the headphones when plugging them in our out.

EDIT: Forgot to mention I'm on M2 MBP EDIT 2: Randomly fixed itself ¯_(ツ)_/¯

acclaude commented 1 year ago

Bose mini II SE speakers 'crackles' in linux-asahi-edge : bluetooth link (sounds + crackles) I'm on MBP 14' M1

PS : 6.1.0-rc6-asahi-5-1-edge-ARCH

marcan commented 1 year ago

Please only report issues that are specific to the new features of linux-asahi-edge or regressions here. Unless you can reproduce a problem in edge and reliably never reproduce it in the regular kernel, it does not belong here.

aykevl commented 1 year ago

Update: for those reading this now, this is easily fixed by changing the resolution and logging out/in. You just need to change the resolution once (or twice) to trigger something that fixes the scaling issue. Somehow, the scaling is connected to DCP/no-DCP, and you need to redo this when you switch. (Or use Wayland instead of Xorg I guess).

Something is broken in desktop scaling under KDE, but only with linux-asahi-edge. Did not test other desktop environments. Text is rendered at the correct size, but all the other elements are much smaller than they should be. Take a look at the below screenshot for example:

Screenshot_20221124_170729

While buttons are mostly correct, other widgets (sliders, scrollbars, icons, etc) are way too small. The window also starts much smaller than with linux-asahi. (This could very well be a KDE/X11 issue, but since this is a regression compared to linux-asahi I'm reporting it just in case - I suspect DPI isn't reported correctly somewhere in the stack).

EDIT: no environment variables related to scaling changed between the two kernels. These are the same whether I use asahi-linux or asahi-linux-edge: GDK_DPI_SCALE=0.5, GDK_SCALE=2, QT_AUTO_SCREEN_SCALE_FACTOR=0, QT_SCREEN_SCALE_FACTORS=None-1=2.5;

marcan commented 1 year ago

Something is broken in desktop scaling under KDE, but only with linux-asahi-edge. Did not test other desktop environments. Text is rendered at the correct size, but all the other elements are much smaller than they should be.

Is this with an external screen, or otherwise what device? Did you manually change any scaling settings? Scaling on Xorg is... "interesting" to put it mildly. There are several sources of information, and the DCP driver in edge should be capable of reporting the physical screen size where available, which probably affects this.

jannau commented 1 year ago

DCP doesn't report DisplayAttributes for the integrated displays. We need to add those to the devicetree. Can be calculated from the resolution and the dot-pitch property in disp0. I don't think this results in this issue though.

aykevl commented 1 year ago

Is this with an external screen, or otherwise what device?

Just the laptop screen itself. It's a 2021 Macbook Pro 16" M1 Pro.

Did you manually change any scaling settings?

The only thing I changed was that I changed "Global scale" in Display Configuration to 250% (from the default of, I think, 200%). I rebooted to linux-asahi-edge with Global scale set to 200% but that doesn't fix the issue (just makes it slightly less noticeable as the difference in scale is smaller).

Scaling on Xorg is... "interesting" to put it mildly.

True that. But it works surprisingly well in KDE, even better than with Wayland IMHO.

There are several sources of information, and the DCP driver in edge should be capable of reporting the physical screen size where available, which probably affects this.

I checked some sources of information. Here is xrandr -q:

 diff -u tmp/xrandr-fb.txt tmp/xrandr-dcp.txt
--- tmp/xrandr-fb.txt   2022-11-24 18:47:35.720265529 +0100
+++ tmp/xrandr-dcp.txt  2022-11-24 19:10:34.371000021 +0100
@@ -1,3 +1,3 @@
-Screen 0: minimum 320 x 200, current 3456 x 2160, maximum 4096 x 4096
-None-1 connected primary 3456x2160+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
-   3456x2160     60.00*+
+Screen 0: minimum 320 x 200, current 3456 x 2160, maximum 16384 x 16384
+eDP-1 connected primary 3456x2160+0+0 (normal left inverted right x axis y axis) 0mm x 0mm
+   3456x2160     60.00*+  50.00    59.94    48.00    47.95

It's different, but I don't see anything wrong here. I also checked xdpyinfo but while there were differences, I didn't see anything related to DPI or resolution.

Still, this should not be the issue. KDE/Qt clearly knows at what resolution to render because text looks correct. It's just all the other things that are wrong for some reason. But apparently DCP did affect rendering somehow.

EDIT: huh, now this same issue also affects linux-asahi. This is very weird, it has worked fine on linux-asahi all this time.

aykevl commented 1 year ago

Ok, I think I figured it out - sort of. Apparently the text scale setting is kept between the framebuffer and DCP, but the scale for other UI elements is not. So if I change the global scale on asahi-linux-edge, global scale only works properly on asahi-linux-edge (but not on asahi-linux), and vice versa.

...with this new information, I don't think there is a linux-asahi-edge bug. Good thing I played with the global scale setting, otherwise I wouldn't have found this.

Perhaps KDE/Plasma/X11/whatever stores the display ID with the global scale configuration and only applies it to the same display? (Where DCP is seen as a different display).

marcan commented 1 year ago

Yeah, that sounds about right. As I said, the whole mechanism is... "interesting" to say the least. We're likely to switch to Wayland by default when the GPU stuff lands, which will hopefully make this a bit less crazy.

saghul commented 1 year ago

Hey there! Kudos to everyone for some amazing work. I am typing this from an M1 Max running the edge kernel. I am experiencing some very weird stuff when putting the machine to sleep, I'll try to explain it here. If there are any logs I can provide, I'll be happy to do so.

The machine seems to go to sleep ok when using KDE's "sleep" option.

When coming back from sleep, however, the keyboard and trackpack (I have no peripherals connected) don't seem to come back.

If I power cycle it, same thing, the keyboard will work in GRUB but not when KDE is started.

Rebooting to the non edge kernel seems to help somehow. After that I shutdown, boot the edge kernel again and I got the keyboard back.

Is this a known issue? Anything I can provide to help?

Once again, thanks Asahi team for the work!

russelltg commented 1 year ago

I'm running into an interesting issue with DCP with a monitor (LG 32UK550) connected via HDMI to my Mac Studio (M1 Max). My monitor displays indicates that it s in HDR mode, and the display looks rather washed out/dim. There is a setting in my monitor called "HDMI ULTRA HD Deep Color," which I can turn off to get nice bright colors, however it only works at 30 Hz after this change.

In macOS, the "High Dynamic Range" setting in settings can toggle the monitor between HDR and non-HDR modes, and when in HDR modes, the "ULTRA HD Deep Color" option is available. Turning it off when in HDR mode makes macOS only drive the display at 30Hz. Which indicates probably some quirk of the display hardware or monitor maybe?

It is notable that when plugged into my intel-based laptop on Linux, it sets the display to a non-HDR mode, which seems correct given the state of HDR on Linux.

In Windows, I can turn HDR on and it looks quite similar to how it looked in asahi (dim/washed out), but they have a slider to increase the baseline brightness (not really my field, I'm sure there's more precise terminology)...Very strangely, even with HDR on in windows, the "ULTRA HD Deep Color" support is still locked to off, which is disappointing as I could not test if windows could drive it at 60 Hz in this configuration.

If there is any info I can provide I'd be happy to. I think the most reasonable fix would be to mimic whatever macOS does when you turn HDR off in settings. I'm pretty technical and happy to collect any information if my setup happens to be useful.

thasti commented 1 year ago

Seen on M2 MBA, adding it here since I didn't find it above: After returning from s2idle, bluetooth devices can no longer be connected.

At resume time, dmesg shows:

[  259.346210] Bluetooth: hci0: command 0x0c01 tx timeout
[  259.346210] Bluetooth: hci0: Opcode 0x c01 failed: -110
[  261.394223] Bluetooth: hci0: command 0x0c1a tx timeout
[  261.394223] Bluetooth: hci0: Opcode 0x c1a failed: -110

Any manual connection attempts time out, and in dmesg I see:

[  678.866055] Bluetooth: hci0: command 0x0408 tx timeout
Capta1nT0ad commented 1 year ago

Something is broken in desktop scaling under KDE, but only with linux-asahi-edge. Did not test other desktop environments. Text is rendered at the correct size, but all the other elements are much smaller than they should be. Take a look at the below screenshot for example:

Screenshot_20221124_170729

While buttons are mostly correct, other widgets (sliders, scrollbars, icons, etc) are way too small. The window also starts much smaller than with linux-asahi. (This could very well be a KDE/X11 issue, but since this is a regression compared to linux-asahi I'm reporting it just in case - I suspect DPI isn't reported correctly somewhere in the stack).

EDIT: no environment variables related to scaling changed between the two kernels. These are the same whether I use asahi-linux or asahi-linux-edge: GDK_DPI_SCALE=0.5, GDK_SCALE=2, QT_AUTO_SCREEN_SCALE_FACTOR=0, QT_SCREEN_SCALE_FACTORS=None-1=2.5;

Can confirm on M2 with DCP in linux-asahi-edge. More notes: The desktop appears to be snappier with DCP but mouse cursor is choppy/flashes and there appears to be more screen tearing. The scaling issue doesn't seem to effect GTK apps.

marcan commented 1 year ago

You should change your display scale setting when switching to -edge. Under Xorg, changing that should change the DPI setting in the fonts dialog. It is likely that the display scale got reset to 100% since the display output changed, but not the DPI setting since that is global.

This is expected. Display scaling is all kinds of quirky and broken on Xorg. I preconfigured things to work right for new installs, but there's no way to make it work smoothly across changes like this, and approximately nobody is interested in working on Xorg bugs upstream. You should be able to play around with the scale/DPI settings and find something that works. But I don't think there's anything we can do our end here. This has nothing to do with the -edge kernel other than the fact that the new driver shows up as a new display device and therefore doesn't inherit the old scale setting.

I've added this to the known issues list; we really can't do anything about it.

CalcProgrammer1 commented 1 year ago

Just installed linux-asahi-edge on my M1 Mac Mini which I use as a TV PC. GPU seems to be running great, but the HDMI output brightness seems like it's set too high. Colors are very washed out. It was not this way before installing edge, nor is it this way on MacOS. My TV is an old Apex Digital 1080p from 2013.

russelltg commented 1 year ago

@CalcProgrammer1 any chance that this is a hdr tv? Could very well be related to https://github.com/AsahiLinux/linux/issues/70#issuecomment-1336280342

ezntek commented 1 year ago

Hi, SDDM crashes on boot, and it only happens when booting into linux-asahi-edge.

Xorg.0.log:

[     6.393] (--) Log file renamed from "/var/log/Xorg.pid-525.log" to "/var/log/Xorg.0.log"
[     6.393] 
X.Org X Server 1.21.1.4
X Protocol Version 11, Revision 0
[     6.393] Current Operating System: Linux easontek-mac 6.1.0-rc8-asahi-3-1-edge-ARCH #2 SMP PREEMPT_DYNAMIC Tue, 06 Dec 2022 11:30:42 +0000 aarch64
[     6.393] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-linux-asahi-edge root=UUID=dab2b306-53cf-4ff9-b505-f86e2659cdee rw loglevel=3 quiet
[     6.393]  
[     6.393] Current version of pixman: 0.42.2
[     6.393]    Before reporting problems, check http://wiki.x.org
        to make sure that you have the latest version.
[     6.393] Markers: (--) probed, (**) from config file, (==) default setting,
        (++) from command line, (!!) notice, (II) informational,
        (WW) warning, (EE) error, (NI) not implemented, (??) unknown.
[     6.393] (==) Log file: "/var/log/Xorg.0.log", Time: Thu Dec  8 13:00:05 2022
[     6.393] (==) Using config directory: "/etc/X11/xorg.conf.d"
[     6.393] (==) Using system config directory "/usr/share/X11/xorg.conf.d"
[     6.393] (==) No Layout section.  Using the first Screen section.
[     6.393] (==) No screen section available. Using defaults.
[     6.393] (**) |-->Screen "Default Screen Section" (0)
[     6.393] (**) |   |-->Monitor "<default monitor>"
[     6.393] (==) No monitor specified for screen "Default Screen Section".
        Using a default monitor configuration.
[     6.393] (==) Automatically adding devices
[     6.393] (==) Automatically enabling devices
[     6.393] (==) Automatically adding GPU devices
[     6.393] (==) Automatically binding GPU devices
[     6.393] (==) Max clients allowed: 256, resource mask: 0x1fffff
[     6.393] (WW) The directory "/usr/share/fonts/misc" does not exist.
[     6.393]    Entry deleted from font path.
[     6.393] (WW) `fonts.dir' not found (or not valid) in "/usr/share/fonts/TTF".
[     6.393]    Entry deleted from font path.
[     6.393]    (Run 'mkfontdir' on "/usr/share/fonts/TTF").
[     6.393] (WW) The directory "/usr/share/fonts/OTF" does not exist.
[     6.393]    Entry deleted from font path.
[     6.393] (WW) The directory "/usr/share/fonts/Type1" does not exist.
[     6.393]    Entry deleted from font path.
[     6.393] (WW) The directory "/usr/share/fonts/100dpi" does not exist.
[     6.393]    Entry deleted from font path.
[     6.393] (WW) The directory "/usr/share/fonts/75dpi" does not exist.
[     6.393]    Entry deleted from font path.
[     6.393] (==) FontPath set to:

[     6.393] (==) ModulePath set to "/usr/lib/xorg/modules"
[     6.393] (II) The server relies on udev to provide the list of input devices.
        If no devices become available, reconfigure udev or disable AutoAddDevices.
[     6.393] (II) Module ABI versions:
[     6.393]    X.Org ANSI C Emulation: 0.4
[     6.393]    X.Org Video Driver: 25.2
[     6.393]    X.Org XInput driver : 24.4
[     6.394]    X.Org Server Extension : 10.0
[     6.394] (++) using VT number 1

[     6.394] (II) systemd-logind: logind integration requires -keeptty and -keeptty was not provided, disabling logind integration
[     6.395] (II) xfree86: Adding drm device (/dev/dri/card1)
[     6.395] (II) Platform probe for /sys/devices/platform/soc/206400000.gpu/drm/card1
[     6.396] (II) xfree86: Adding drm device (/dev/dri/card2)
[     6.396] (II) Platform probe for /sys/devices/platform/soc/soc:display-subsystem/drm/card2
[     6.405] (II) no primary bus or device found
[     6.405]    falling back to /sys/devices/platform/soc/206400000.gpu/drm/card1
[     6.405] (II) LoadModule: "glx"
[     6.405] (II) Loading /usr/lib/xorg/modules/extensions/libglx.so
[     6.406] (II) Module glx: vendor="X.Org Foundation"
[     6.406]    compiled for 1.21.1.4, module version = 1.0.0
[     6.406]    ABI class: X.Org Server Extension, version 10.0
[     6.406] (==) Matched modesetting as autoconfigured driver 0
[     6.406] (==) Matched fbdev as autoconfigured driver 1
[     6.406] (==) Assigned the driver to the xf86ConfigLayout
[     6.406] (II) LoadModule: "modesetting"
[     6.406] (II) Loading /usr/lib/xorg/modules/drivers/modesetting_drv.so
[     6.407] (II) Module modesetting: vendor="X.Org Foundation"
[     6.407]    compiled for 1.21.1.4, module version = 1.21.1
[     6.407]    Module class: X.Org Video Driver
[     6.407]    ABI class: X.Org Video Driver, version 25.2
[     6.407] (II) LoadModule: "fbdev"
[     6.407] (WW) Warning, couldn't open module fbdev
[     6.407] (EE) Failed to load module "fbdev" (module does not exist, 0)
[     6.407] (II) modesetting: Driver for Modesetting Kernel Drivers: kms
[     6.429] (WW) Falling back to old probe method for modesetting
[     6.429] (EE) open /dev/dri/card0: No such file or directory
[     6.429] (WW) Falling back to old probe method for modesetting
[     6.429] (EE) open /dev/dri/card0: No such file or directory
[     6.438] (II) modeset(G0): using drv /dev/dri/card2
[     6.438] (EE) No devices detected.
[     6.438] (EE) 
Fatal server error:
[     6.438] (EE) no screens found(EE) 
[     6.438] (EE) 
Please consult the The X.Org Foundation support 
         at http://wiki.x.org
 for help. 
[     6.438] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[     6.438] (EE) 
[     6.502] (EE) Server terminated with error (1). Closing log file.

journalctl -xeu sddm.service:

Dec 08 13:00:02 easontek-mac sddm[507]: Initializing...
Dec 08 13:00:02 easontek-mac sddm[507]: Starting...
Dec 08 13:00:02 easontek-mac sddm[507]: Logind interface found
Dec 08 13:00:02 easontek-mac sddm[507]: Adding new display on vt 1 ...
Dec 08 13:00:02 easontek-mac sddm[507]: Loading theme configuration from ""
Dec 08 13:00:02 easontek-mac sddm[507]: Display server starting...
Dec 08 13:00:02 easontek-mac sddm[507]: Adding cookie to "/var/run/sddm/{a3543f8f-8a85-41b9-8129-2cd6>
Dec 08 13:00:02 easontek-mac sddm[507]: Running: /usr/bin/X -nolisten tcp -dpi 144 -background none ->
Dec 08 13:00:02 easontek-mac sddm[507]: Failed to read display number from pipe
Dec 08 13:00:02 easontek-mac sddm[507]: Display server stopping...
Dec 08 13:00:02 easontek-mac sddm[507]: Attempt 1 starting the Display server on vt 1 failed
Dec 08 13:00:03 easontek-mac sddm[507]: Display server starting...
Dec 08 13:00:03 easontek-mac sddm[507]: Adding cookie to "/var/run/sddm/{a3543f8f-8a85-41b9-8129-2cd6>
Dec 08 13:00:03 easontek-mac sddm[507]: Running: /usr/bin/X -nolisten tcp -dpi 144 -background none ->
Dec 08 13:00:03 easontek-mac sddm[507]: Failed to read display number from pipe
Dec 08 13:00:03 easontek-mac sddm[507]: Display server stopping...
Dec 08 13:00:03 easontek-mac sddm[507]: Attempt 2 starting the Display server on vt 1 failed
Dec 08 13:00:05 easontek-mac sddm[507]: Display server starting...
Dec 08 13:00:05 easontek-mac sddm[507]: Adding cookie to "/var/run/sddm/{a3543f8f-8a85-41b9-8129-2cd6>
Dec 08 13:00:05 easontek-mac sddm[507]: Running: /usr/bin/X -nolisten tcp -dpi 144 -background none ->
Dec 08 13:00:05 easontek-mac sddm[507]: Failed to read display number from pipe
Dec 08 13:00:05 easontek-mac sddm[507]: Display server stopping...
Dec 08 13:00:05 easontek-mac sddm[507]: Attempt 3 starting the Display server on vt 1 failed
Dec 08 13:00:05 easontek-mac sddm[507]: Could not start Display server on vt 1
marcan commented 1 year ago

@tek967 You uninstalled asahi-meta, or you did not follow the instructions and update everything. You also didn't attach asahi-diagnose output as requested (which would've made it evident you don't have several required packages).

Please don't uninstall our support packages. Things break without them.

CalcProgrammer1 commented 1 year ago

@CalcProgrammer1 any chance that this is a hdr tv? Could very well be related to #70 (comment)

I highly doubt it, it's a pretty low end TV from almost 10 years ago.

marcan commented 1 year ago

It's not about HDR, it's about color formats I believe. So yes, it can affect older TVs. Known issue.

accensi commented 1 year ago

With edge, after boot login with sddm or lightdm freezes keyboard and mouse. TTy terminals are working. alsa-diagnose attached asahi-diagnose-20221209-170127.txt

goonmandu commented 1 year ago

The keyboard modifier keys (shift, control, option, command, fn, etc) don't work inside the terminal or the login menu when using Wayland. This results in not being able to type uppercase letters, ~, |, or any other important terminal command flags.

They work with X11 just fine. asahi-diagnose-20221210-143159.txt

marcan commented 1 year ago

@goonmandu That's not a linux-asahi-edge issue.

jannau commented 1 year ago

@russelltg and @CalcProgrammer1 please attach the output of ioreg -r -c AppleCLCD2 -k ColorElements -x from macOS when the problematic display is connected (without other displays). I think I cann reproduce the LG Ultra HD Deep Color issue and should be able to resolve it.

CharlesYiu commented 1 year ago

just followed the installation steps and rebooted my m1 macbook air. got this message:

ieee80211 phy0: brcmf_p2p_send_action_frame: Unknown Frame: category 0x5, action 0x4

i also did not install the recommended wayland environment

marcan commented 1 year ago

@charlesyiu That's just an informational message. If you have an actual problem please attach an asahi-diagnose log.

CharlesYiu commented 1 year ago

@charlesyiu That's just an informational message. If you have an actual problem please attach an asahi-diagnose log.

i can't run asahi-diagnose because asahi-linux-edge does not launch the desktop enviorment. it just shows that message and does nothing else.

marcan commented 1 year ago

@charlesyiu Try option+control+F2 or just run asahi-diagnose while running the regular kernel.

andy9a9 commented 1 year ago

Not sure, if it's reproducible, but once I booted the MBP14, USB wasn't working. I realized that via USB mouse, which wasn't lighting. So I un/plugged it, no dmesg or journal change. After next boot it was fine.

marcan commented 1 year ago

@andy9a9 Please gather asahi-diagnose logs when you run into a problem like that. Otherwise we can't do anything about it.

kujeger commented 1 year ago

Seen on M2 MBA, adding it here since I didn't find it above: After returning from s2idle, bluetooth devices can no longer be connected.

At resume time, dmesg shows:

[  259.346210] Bluetooth: hci0: command 0x0c01 tx timeout
[  259.346210] Bluetooth: hci0: Opcode 0x c01 failed: -110
[  261.394223] Bluetooth: hci0: command 0x0c1a tx timeout
[  261.394223] Bluetooth: hci0: Opcode 0x c1a failed: -110

Any manual connection attempts time out, and in dmesg I see:

[  678.866055] Bluetooth: hci0: command 0x0408 tx timeout

Also seeing this on a M1 Pro 14". As a workaround I have found that powering bluetooth off ( bluetoothctl power off) before suspending allows you to power it on again after resume, with full functionality.

ghost commented 1 year ago

Seen on M2 MBA, adding it here since I didn't find it above: After returning from s2idle, bluetooth devices can no longer be connected. At resume time, dmesg shows:

[  259.346210] Bluetooth: hci0: command 0x0c01 tx timeout
[  259.346210] Bluetooth: hci0: Opcode 0x c01 failed: -110
[  261.394223] Bluetooth: hci0: command 0x0c1a tx timeout
[  261.394223] Bluetooth: hci0: Opcode 0x c1a failed: -110

Any manual connection attempts time out, and in dmesg I see:

[  678.866055] Bluetooth: hci0: command 0x0408 tx timeout

Also seeing this on a M1 Pro 14". As a workaround I have found that powering bluetooth off ( bluetoothctl power off) before suspending allows you to power it on again after resume, with full functionality.

bluetoothctl off/on works great as a workaround, added them to keyboard shortcuts

marcan commented 1 year ago

Yes, bluetooth is known to break after suspend/resume.

zaheeruddinbabur commented 1 year ago

Hi, I am using MBPro M1 and the boot process stops at the following prompt. 20230105_020500.jpg

I have restarted multiple times and reinstalled the edge kernel 2 times with no luck.

rdvm commented 1 year ago

Hi, I am using MBPro M1 and the boot process stops at the following prompt. 20230105_020500.jpg

I have restarted multiple times and reinstalled the edge kernel 2 times with no luck.

I had the same thing crop up the other day on my M1 MacBook Air (A2337/MacBookAir10,1) and found I could still access TTY to do some poking around and was also able to boot the linux-asahi kernel and things came up normally and allowed me to log in.

I just spent a little time digging to see if I could find anything, and ended up getting things working by downgrading the mesa package from mesa-22.3.2-3 to mesa-22.2.3-1. I tried mesa-22.3.2-1 first since I had that one in my pacman cache too, but the problem persisted there.

Asahi team, let me know if you'd like me to try the newer mesa package again and grab output from asahi-diagnose or anything else.

marcan commented 1 year ago

asahi-diagnose please. Even if you can only grab it while running the non-edge kernel, that's useful.

rdvm commented 1 year ago

Ran updates this morning to upgrade mesa to mesa-22.3.2-3 (FWIW that was the only update for the system so nothing else has changed). Booted edge, saw the same error as before, logged in via tty2, ran asahi-diagnose and also grabbed the output from manually running startx, which is what pointed me to mesa in the first place.

asahi-diagnose gist

startxoutput gist

h0m3 commented 1 year ago

Under KDE going to s2idle using the sleep functionality it drained 20% of battery in 10 hours. That is the expected battery drain?

marcan commented 1 year ago

Under KDE going to s2idle using the sleep functionality it drained 20% of battery in 10 hours. That is the expected battery drain?

Yes.

lebakassemmerl commented 1 year ago

Hi all! I think I hit a bug concerning the asahi GPU driver. I am running asahi-linux-edge on a Macbook Air M1 with 16GB RAM and 512GB SSD.

When opening a certain set of webpages with firefox, firefox crashes with a segfault. Before the segfault a lot of DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1) messages occur. The german part of the error messages means something like "Timout for the connection occurred" or "the connection timed out". This is not a literal translation tough.

I tried to run firefox with GDB to get at least an idea what is going but for me it seems the debug-output doesn't print much useful information:

DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
[Thread 0xffffc3eef0e0 (LWP 1721) exited]
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)
DRM_IOCTL_ASAHI_SUBMIT failed: Die Wartezeit für die Verbindung ist abgelaufen (437x437 tile 32x32 layers 1 samples 1)

Thread 39 "CanvasRenderer" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xffffd5eaf0e0 (LWP 1495)]
0x0000fffff7bc62c0 in memcpy () from /usr/lib/libc.so.6
(gdb) quit

When I try to open the pages the asahi-linux kernel without GPU driver, it works without problems. The pages where I can reproduce the crash every time are the following ones (link): Screenshot_20230110_202845

Below you find a zip for x11-cpurender, x11-agx and wayland-agx session with the following content:

firefox-crash.zip

Thanks for all your great work and if you need some additional information, please let me know!

Patronics commented 1 year ago

Hi, I am using MBPro M1 and the boot process stops at the following prompt. 20230105_020500.jpg

I have restarted multiple times and reinstalled the edge kernel 2 times with no luck.

I'm getting the same issue from a freshly updated asahi-linux-edge install on my M1 Max Macbook Pro, hanging at roughly that stage in booting, except without that last line being printed (instead stopping after printing the number of blocks). The first time I attempted booting after the update it ended up rebooting to macOS after hanging there for a while, the second time it just seemed to be stuck there. The non-edge version of asahi still works, and the output from running asahi-diagnose upon rebooting there is attached. asahi-diagnose-20230111-150112.txt

marcan commented 1 year ago

Xorg is broken with -edge unless you install the mesa-asahi-edge GPU drivers. This is a known bug, not a kernel problem.

captn3m0 commented 1 year ago

M2 Air, suspend then resume fails with the X server failing to start. Reproduces intermittently (only on long duration suspends), but tested across more than 5 reboots with the edge kernel.

Jan 25 17:03:53 submarine sddm[509]: Display server starting...
Jan 25 17:03:53 submarine sddm[509]: Adding cookie to "/var/run/sddm/{d11f3dc7-73ac-4523-a590-4991c21a01a1}"
Jan 25 17:03:53 submarine sddm[509]: Running: /usr/bin/X -nolisten tcp -dpi 144 -background none -seat seat0 vt1 -auth /var/run/sddm/{d11f3dc7-73ac-4523-a590-4991c21a01a1} -noreset -displayfd 19
Jan 25 17:03:53 submarine sddm[509]: Failed to read display number from pipe
Jan 25 17:03:53 submarine sddm[509]: Display server stopping...
Jan 25 17:03:53 submarine sddm[509]: Attempt 2 starting the Display server on vt 1 failed

Switching to a tty and restarting sddm fixes the issue.

https://gist.github.com/captn3m0/2d8ee1e4b3e7d12d80b49a8edc9809d0

cyrinux commented 1 year ago

On a m2 air with edge kernel, sometimes I need to reboot to be able to connect to my headset after I resume from suspend.

I got those kernel message when trying to connect, restart bluetooth.service don't help, and I can't unload bluetooth driver.

[81726.050282] Bluetooth: hci0: command 0x0408 tx timeout
[81762.914073] Bluetooth: hci0: command 0x0408 tx timeout
[81767.010042] Bluetooth: hci0: command 0x0408 tx timeout
[81812.065782] Bluetooth: hci0: command 0x0408 tx timeout