__radix_tree_lookup errors with 4.14.x and Firefox
Greetings, LQ community. I've been having this problem since 4.14.0 and Firefox Quantum. It happens almost every time I use Firefox. 99% of the time it only happens on ad-heavy Internet sites such as Yahoo and Amazon, although it has also happened within a minute of launching a Google search.
I never had this problem before Firefox changed to its Quantum engine, and never had it in kernels 4.13.x and earlier. I can use Chromium or Seamonkey all day with no problems. First I thought it might be extensions, since they had to be replaced with the new format. I uninstalled down to only NoScript, AdBlock, Persona, and VideoDownloadHelper. I also tried with a fresh profile. No luck. I also thought it might be the nvidia 340.x legacy driver. I upgraded my GTX260 to a GTX750Ti using the latest 390.x - still no luck. I also thought it might be related to system monitors, so I disabled conky and gkrellm, but it still happens. When the error happens, I get a single pc speaker beep then the browser will no longer load any pages. I can open a different browser, though, and work from there. Also, occasionally, after the error, when shutting down it will get stuck umounting filesystems. I'm running Slackware 14.2 32-bit with kernel 4.14.39, on a ASRock 970M with 16GB RAM (using PAE). CPU is AMD Phenom II x2 unlocked to x4 using the ASRock's core unlocker. Nvidia driver is version 390.48 and I don't have any other 3rd party kernel modules. I dual boot with Win7 Enterprise 32-bit, and the same Firefox version with the same addons has no problems. I have searched several times for information about radix_tree in kernel errors but found nothing. The error output contains references to swap but I do not use swap. Below is a the most recent such error, copied from a rxvt window that was open at the time: Code:
Message from syslogd@Chimera at Fri Apr 27 07:56:40 2018 ... |
There have been some fixes in this area in later 4.14. Would you mind to get the latest 4.14 kernel (4.14.40 as of today) and check if this got fixed?
|
Thank you for the reply. I installed .40 yesterday and upgraded Firefox to 60.0 this morning. I have had Firefox open for 3 hours now using Amazon, Yahoo and Facebook and so far so good. I'll keep checking and hoping it is fixed.
|
It happened again this morning. Twice. First, I had used Firefox for a few minutes to download the files to build the mono-5.0.1.1 SlackBuild. When the downloads finished I closed Firefox and proceeded to begin the SlackBuild. All was fine for several minutes until the kernel crashed in the middle of compiling mono. I rebooted and remained at the CLI - no X running. I went straight to the directory where the mono SlackBuild files are located and started the build. Less than five minutes into it - crash!
This time I was able to see some lines at the top of the crash log that I did not get previously. The references to smp #1 make me think it might be a compatibility issue between the kernel and my cpu. Something that is triggered and then blows up under cpu load? I guess it had previously been coincidence that it always crashed during/after using Firefox. Code:
May 13 08:26:02 Chimera kernel: [ 6050.815209] BUG: unable to handle kernel NULL pointer dereference at 00000008 |
I've been using my computer most of the day today, booted into kernel 4.4.118, and using Firefox 60.0. No problems. I've pretty much decided to use the 4.4 kernel as my primary and keep 4.14.40 as a backup. At least until Pat moves Slackware to a newer kernel. I'm still curious if the error sounds like a bug in the kernel source or a problem in my kernel config.
|
All times are GMT -5. The time now is 01:18 PM. |