i normally update my fedora every week or two. but over the last few months. firefox has been especially unstable. i normally have 5-10 tabs open, and 2-5 crash each day. (they never used to.) or is this normal-ish now? i also had a question about the reboot sequence; which forum would be the right one for that question?
Besides an issue with my NVIDIA driver, everything runs as usual on my machine
Please open a new topic here for your reboot question.
For your firefox issue can you provide some additional details about your hardware/configuration?
Fedora often runs ahead of third party drivers. Older hardware can have problems inadvertently introduced with new kernels (developers don’t have access to older hardware, so it is up to users to report issues). You don’t mention the RAM capacity of your system and whether you are using Wayland or Xorg, Gnome or ??. Firefox memory requirements have increased over time – I suspect largely due to changes web sites have made. Prime suspects for problems with Firefox are memory, buggy extensions, and websites using less common video formats.
When reporting problems, you should supply enough detail to allow others to reproduce the issue. Firefox is tricky as it has a huge list of extensions. You should review the extensions you use, and exclude ones that aren’t widely used one at a time to rule out issues caused by extensions. Journalctl should have details of memory problems and video format issues. There are sites that allow you to test browser support for new video formats.
thanks for the replies.
my hardware is an i9-12900 cpu, 32GB mem, RX-580. (and several hdd drives.) i use a stock standard fedora 39, which means i use wayland.
i used to use an old nvidia GPU (something 12-13 years old), but in an effort to resolve lingering doubts about nbvidia, switched to a somewhat recent AMD gpu (rx-580).
i use no extensions (that i can find via the firefox menus).
would you like any other information?
(i would point out that i always send in the dump reports that firefox prompts me for when it detects an issue.)
i should also add that in looking at dmesg output, i get corrected bus errors every few hundred seconds:
[74729.186389] nvme 0000:07:00.0: PCIe Bus Error: severity=Corrected, type=Physical Layer, (Receiver ID)
[74729.186390] nvme 0000:07:00.0: device [144d:a80a] error status/mask=00000001/0000e000
[74729.186391] nvme 0000:07:00.0: [ 0] RxErr (First)
i’ve not followed up on those yet; hopefully (because it says corrected) they shouldn’t affect my system.
Just some ideas that come to mind.
I would be curious to know what motherboard and nvme drive you have. Smartctl long test results on drive? Bios settings IRT storage? Any bios updates available with suspiciously related fixes in their changelog? You could also post a fuller log on pastebin or something. Hardware accel or something that might not be working 100% in browser settings? What is resource usage like in htop when the browser tabs crash? Is it just a coincidence that the problem is manifested in the browser because you are spending most time in the browser, or could you replicate the crashes when using other programs like the file browser?
I’m still on 38, but i doubt ff is that unstable, in general. You might want to make sure you’re not troubleshooting this issue without ruling out the specific websites in question. Perhaps you have some websites with badly behaving scripts that have been included in each “tab crash test”? They would cause resource usage/spikes in htop, i would imagine.
i am sure it is website unrelated; the most common tab is youtube, and sometimes, firefox itself crashes.
my motherboard is a gigabyte z690 aorus pro. i am a little puzzled about the nvme question. i have a mix of seagate 3.5in drives and samsung 1TB ssd’s.
i am rarely in top (i hope htop was a mis-spell of top?) so i have no idea about resource usage (but generally the system is miostly idle save for some downloads). i have seen no other messages for programs crashing.
IRT to websites: i would just try totally different sites to make sure it’s not relevant. I got the “nvme” from the log output you posted. By samsung 1TB ssds you mean 2.5" sata? htop is a fancier version of top: https://htop.dev/
You might run htop and surf and see what happens to resources before a crash. You could also run firefox from the terminal with “firefox” and then surf until crash and it might give you more info.
htop is top with scrolling and mouse support. It isn’t in the base Fedora Workstation install, but well worth having. There is also a
pcp version (see man page).
The PCIE recoverable errors are often related to APST power saving, see: NVME Controller failiure due to broken APST support. This is a rather old article, so it may be that errors mentioned in the article are now recoverable with more recent firmware or kernel changes. You may want to use
journalctl to check for correlation between the times of the PCIE recoverable errors and application crashes.