To expose the various capabilities via vaapi for encode/decode - fedora strips patent encumbered bits from mesa by default. IME people forget to run the dnf swap bits after adding rpmfusion .
You can check what is exposed by installing libva-utils and running vainfo.
NOTE: If you are installing to an AMD Ryzen 7040 Series Framework 13, we are recommending Fedora 39 at this time due to installation challenges with Fedora 38. Testing has indicated F38 “can” be installed, it’s not a reliable process and best skipped in my testing.
If you are on the pre-release, it’s time to update your F39 install!
Fedora Bugzilla - Framework-related bugs top-level ticket - rhbz#2240811: This top-level bug ticket manages all the Framework Laptop-related tickets. You can see the ticket list on the item: “Depends On” on the linked page. Note that the keyword “rhbz#NNNNNNN” is used to describe a Bugzilla ticket ID, and it is used by some Fedora infra systems such as Budhi (a document page) in Fedora project.
Create your Fedora Bugzilla account if you don’t have it.
Click New or File a Bug, then select Product: Fedora. If you don’t know the component, you can add the most possible component such as kernel and etc, then someone may change the component to a proper one.
When editing the ticket, add the Framework top-level ticket id 2240811 to the item: Blocks.
Shall we adjust the current thread title “[Fedora 39] Official Fedora 39 Thread” to align with existing thread titles?
I don’t know what is the intent of the “[Fedora 39]” prefix. The word “Fedora 39” also exists in the “Official Fedora 39 Thread”. The word “Thread” does not seem to be meaningful. I see some different prefixes are used for this kind of thread.
We have often aligned with “Something on the Framework Laptop” for the guide thread that started from Fedora 34 on the Framework Laptop . Framework’s Linux compatibility page’s title is also aligned with “Something on the Framework Laptop” (= “Linux on the Framework Laptop”).
I would suggest that “[OFFICIAL GUIDE] Fedora 39 on the Framework Laptop” or similar one.
So I noticed a couple of things after installing Fedora 39 beta on my AMD Framework 13. Running all the latest updates / fresh install as of 10/10/23.
The first time you log in after booting it defaults to X11, then if you log out and log back in it will use Wayland. Not sure why this is happening. Is it a F39 issue or something else going on?
The mouse / UI feels laggy especially compared to my desktop. I’m not really sure how to explain it otherwise. I realized only 512 mb of ram was being allocated to the iGPU and was full. In the BIOS I changed the iGPU Mode to Game Optimized and things seem to be much more fluid now. Is this a known issue? Or maybe fixed with BIOS update?
The default trackpad scroll speed is insanely fast. I was able to find this project and get it working Warning: Non-Potable Water / libinput-config · GitLab. Hopefully that helps other people here wondering the same thing.
One time I noticed some visual artifacts after waking from sleep, maybe this was related to #2 – will see if it persists.
Appreciate this as it gives me something to feedback to my Fedora team contacts.
I noticed this as well, and found that logging back into Wayland did work from the login screen.
I’ve been focused on the touchpad, is this happening with the touchpad or a Bluetooth or USB attached mouse? Anything showing up in the logs? Based on the next point below, assuming this is the touchpad.
Odd, I did not experience this at all, my touchpad was very smooth. During the most recent testing, nothing attached at all, just AC power or no power attached. No external displays as this is a separate thing to go over and done in secondary testing on my part.
Yes I have not tested it with an external mouse yet. Just using the built in touchpad. Hard to fully explain, but it feels like even though the screen is running at 60 hz that the responsiveness is closer to 30 hz. Like I said though, changing the iGPU setting in the BIOS seems to have improved it. I am just running 3.02 so maybe that’s the difference with your setup? Maybe since the VRAM was full, the desktop wasn’t being rendered at full speed? Happy to look into logs, just let me know what commands to run.
Using Wayland with fractional scaling enabled at 150%. Tested both on AC and on battery, no peripherals just like you. When you use 2 fingers to scroll by default, it is very sensitive (eg. moving your fingers a short distance results in a much longer scroll movement). It’s not an issue with smoothness, it was very smooth. So more a sensitivity thing.
@Aman_Chhabra Not seeing it here myself, but will toggle the iGPU setting in the BIOS to see if the behavior is different somehow.
What applications are you using when you experience this? Anything interesting appear if you run with a terminal open running journalctl -f as you navigate around?
Looking into the original post now. Vanilla kernel 5.6.5, Debian unstable. Okay, not seeing Fedora listed or the current kernel that comes with.
I am running Steam games (nothing AAA), but using Proton and whatnot. Using the Steam client provided in the Software Store (via rpm-fusion - flatpak also worked fine) set up during install, all looks good there. Touch pad was also super responsive.
Toggling this over to iGPU Mode to Game Optimized. I honestly did not see any difference. That may be me, it may be something with your specific laptop. Not sure.
What also strikes me is that you indicated you see visual artifacts on resume, I do not.
Updates today, always update as it’s a pre-release and updates are fast and furious right now.
I’m on kernel 6.5.6-300.fc39 which is current at the time I am writing this.
EDIT: Just to clarify that this testing is on battery vs plugged in. Am noticing now that there is a difference.
So I have been messing around with it myself. The perceived cursor choppiness seems to be more related to using Power Saver mode vs Balanced. Probably has something to do with the GPU clocks not ramping up fast enough. I tried installing a dynamic triple buffering patched mutter via COPR repo which seems to have made it better. Went back to stock to continue testing with the same setup as you.
Also the choppiness is only on Wayland, I noticed with X11 the cursor feels smoother closer to what you’d get on Windows. Currently I set up a dual boot to test back and forth.
The iGPU Auto vs Game Optimized seems to have been placebo affect or unrelated.
Here are some of the error messages I am getting with journalctl -f:
Oct 11 18:14:07 framework-13 kernel: [drm:amdgpu_dm_process_dmub_aux_transfer_sync [amdgpu]] *ERROR* wait_for_completion_timeout timeout!
Oct 11 18:14:17 framework-13 kernel: [drm:amdgpu_dm_process_dmub_aux_transfer_sync [amdgpu]] *ERROR* wait_for_completion_timeout timeout!
Oct 11 18:14:27 framework-13 kernel: [drm:amdgpu_dm_process_dmub_aux_transfer_sync [amdgpu]] *ERROR* wait_for_completion_timeout timeout!
Also I noticed some intermittent hard crashes / issues seemingly GPU related. One time trying to launch Obsidian and the other trying to launch Telegram (both installed via flatpak):
Oct 11 20:48:38 framework-13 md.obsidian.Obsidian.desktop[8041]: [67:1011/204838.575775:ERROR:shared_context_state.cc(898)] SharedContextState context lost via ARB/EXT_robustness. Reset status = GL_INNOCENT_CONTEXT_RESET_KHR
Oct 11 20:48:38 framework-13 md.obsidian.Obsidian.desktop[8041]: [67:1011/204838.576044:ERROR:gpu_service_impl.cc(1010)] Exiting GPU process because some drivers can't recover from errors. GPU process will restart shortly.
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 md.obsidian.Obsidian.desktop[8041]: [13:1011/204838.595552:ERROR:gpu_process_host.cc(954)] GPU process exited unexpectedly: exit_code=8704
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The CS has been rejected (-125). Recreate the context.
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The CS has been rejected (-125). Recreate the context.
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The CS has been rejected (-125). Recreate the context.
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The CS has been rejected (-125). Recreate the context.
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The CS has been rejected (-125), but the context isn't robust.
Oct 11 20:48:38 framework-13 flatpak[8249]: amdgpu: The process will be terminated.
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 kernel: [drm] Skip scheduling IBs!
Oct 11 20:48:38 framework-13 systemd[2609]: dbus-:1.1-org.telegram.desktop@0.service: Main process exited, code=exited, status=1/FAILURE
Oct 11 20:48:38 framework-13 systemd[2609]: dbus-:1.1-org.telegram.desktop@0.service: Failed with result 'exit-code'.
Oct 11 20:48:38 framework-13 systemd[2609]: app-flatpak-org.telegram.desktop-8235.scope: Consumed 3.068s CPU time.
Oct 11 20:48:39 framework-13 systemd[2609]: Starting tracker-extract-3.service - Tracker metadata extractor...
Oct 11 20:48:39 framework-13 systemd[2609]: Started tracker-extract-3.service - Tracker metadata extractor.
Oct 11 20:49:23 framework-13 kernel: i2c_designware AMDI0010:00: i2c_dw_handle_tx_abort: lost arbitration
Just to clarify your update status as I did not catch the kernel or how up to date you were?
Additionally, sounds like we ventured outside of vanilla installation which will be difficult reproduce. Let’s start off making sure you’re on the latest and greatest updates/kernel.
Guys who was able to reproduce this issue on Framework Laptop 13 AMD, if you have spare time, I hope you create your account on Fedora Bugzilla, and then file the bug to the Fedora Bugzilla, and please let us know the bug ticket URL in this thread.
https://bugzilla.redhat.com/ - Click New or File a Bug, then select Product: Fedora. If you don’t know the component, you can add the most possible component such as kernel and etc, then someone may change the component to a proper one.
I also had this issue, but on Fedora 38 (haven’t upgraded yet). I was able to slow it down with libinput-config (I had to set the factor to 0.2 to be comfortable) but that took a while to find, would be nice to have a Framework guide on this at least
No worries! So I have been experiencing some intermittent application crashes / GPU resets. Do you want me to file bugs for those too, or would I have to file them on the amdgpu gitlab?
If you can specify the upstream project (amdgpu gitlab), in my opinion, filing the bug on the upstream (amdgpu gitlab) first, then filing the bug on the Fedora Bugzilla referring to the filed upstream amdgpu bug ticket link is the best way. But I hope that you follow your intuition to decide what the best way is!
What’s very odd to me is other than the x11 → logout → wayland issue, I’m now not seeing anything like what I was having problems with to reproduce and report. Even the fwupdmgr issue resolved it self…no amdgpu issues…I didn’t update anything either… I’ll be filing bugs as I see them, though.
i am also experiencing this. very choppy and weird.
i’m also experiencing an issue with youtube. some videos play fine in firefox, but others give me this error "An error occurred. Please try again later. (Playback ID:xxxxxxxxxxx). this is my first time using fedora, and linux in general outside of my steamdeck, any help with this youtube issue would be great.
