116
Wayland-Proxy Load Balancer Helping Firefox Cope With Wayland Issues
(www.phoronix.com)
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 23 Dec 2023
116 points (97.5% liked)
Linux
48186 readers
1386 users here now
From Wikipedia, the free encyclopedia
Linux is a family of open source Unix-like operating systems based on the Linux kernel, an operating system kernel first released on September 17, 1991 by Linus Torvalds. Linux is typically packaged in a Linux distribution (or distro for short).
Distributions include the Linux kernel and supporting system software and libraries, many of which are provided by the GNU Project. Many Linux distributions use the word "Linux" in their name, but the Free Software Foundation uses the name GNU/Linux to emphasize the importance of GNU software, causing some controversy.
Rules
- Posts must be relevant to operating systems running the Linux kernel. GNU/Linux or otherwise.
- No misinformation
- No NSFW content
- No hate speech, bigotry, etc
Related Communities
Community icon by Alpár-Etele Méder, licensed under CC BY 3.0
founded 5 years ago
MODERATORS
Would be interesting if this is more on Firefox side, or on compositor side. I've been running Firefox in Wayland for about 9 months now, without any issues.
this is a wayland issue. Due to how wayland works, it cannot drop messages, this means if the messages stop being accepted (IE. the program becomes very slow and not very responsive) the application will wind up dying. EEVDF helped resolve a lot of these issues. but they arent gone yet.
a fairly easy replication cause is to start a large rust project compile since cargo will thread to oblivion if it gets the chance, then use the PC on wayland. Applications can frequently die, Firefox, MPV, Kate, gnome web, chromium, games, etc. it also doesn't matter what compositor you use right now as gnome, kde sway all share the issue
EEVDF really does help stop a lot of these crashing though
BeOS solved the issue of unresponsive GUIs in the 1990s. The GUI just must never run in the same thread as the logic.
while this is good on theory, when your CPU is being absolutely hammered, you need to re-adjust priorities to make a system responsive again, it's actually not a simple thing to do without a context aware scheduler. Even though EEVDF is pretty good, it still struggles some times
My PC with a 133MHz Pentium 1 processor was pretty responsive all the time back in the day. It's definitely a solved problem.
I agree. The proxy solution they're proposing seems like a band-aid on a fundamental design issue to me. It's easier to just tack yet another library onto a big project than to refactor large amounts of code. This is exactly why a lot of software is getting more and more shit.
Also this is the kind of issues Wayland will be facing now that it's starting to see widespread adoption, issues that arise from more and more complex situations created by interconnecting more apps with it in more ways.
How the devs handle this will be crucial and imo it can make or break the project in the long run. It's one thing to successfully run a hobby project at a small scale, it's another to shoulder the entire Linux desktop for the foreseeable future. That's the bar that X had to meet; if Wayland intends to be the Linux desktop it has to step up. "Not our problem, deal with it outside Wayland" will not do.
You're describing Wayland running into issues due to overall high system load, and not been given enough scheduler time to accept messages?
edit: This issue? https://gitlab.freedesktop.org/wayland/wayland/-/issues/159 - didn't find anything else matching the description, and personally have never seen that, both on my low specs notebook or my workstation, which probably counts as higher spec.
correct, this is the same issue, this generally really only happens with a sustained all core workload that will consistently leave you cpu at 100%, since if it's not sustained, the kernel will allot some time to the programs, and the crash wont happen
I guess that explains why I'm not seeing it - my workstation has 64 threads and more than enough memory, and on my notebook I'm scheduling load intensive stuff to not interfere with interactive device usage.