[SOLVED] LM 20.2 (all versions tested) random freezes

[SMG]

So far, so good! Everything works fine, I even played 2 games in STK. I really hope that this will be the solution. Thank you very much.

That's good news.

I read that the C states relate to some power saving for the CPU. I wonder if this makes any difference in everyday work, performance wise and also in power consumption of the PC. Anyway, I will post again with the results.

This thread Gigabyte Z390: Disabling C-States is a different motherboard than yours (and asking about a different issue), but user terabyte explains, "Intel CPUs have a built in feature to disable some of its cores when they are not being used and this is exactly what C-states do. If you disable C-states none of the cores will go to sleep and there will be a lot less fluctuation in the amount of crrent flowing through VRMs resulting in less or no coil whine. Some people disable C-states because they think these are meant to be used in laptop where power saving features matter and there seems to be a general understanding in Overclocking community that C-states should be disabled on Desktop PCs to get better performance."

I would not think there would be any significant difference in power consumption, but that may also depend on the type of work you do.

[ylm]

Up to now, everything was working fine, with no freezes at all (I even played STK a couple of times). But, today, I have another freeze. Just Firefox and Libreoffice open. I restarted the PC via the power button, and after 10 minutes, the screen shown many small rectangular pixelized parts for a couple of seconds, then they disappeared, but the system was not responding.

New journal log: https://termbin.com/z3d1. I am still in the dark... I installed kernel 5.13 again, to see if I see any difference with this one and the C states disabled.

[SMG]

I restarted the PC via the power button, and after 10 minutes, the screen shown many small rectangular pixelized parts for a couple of seconds, then they disappeared, but the system was not responding.

That sounds hardware related. Either the hardware itself or the driver for the hardware.

New journal log

I noticed two items in this log.

The below was also in the 2nd one you posted, but not in the first. When I searched, I found this mentioned related to intel P states. I found one person who added a kernel parameter for it, but that person had many instances of these two lines in their log. Your log just has the two lines mentioned once.

Code: Select all

Δεκ 06 17:19:27 lm-desktop kernel: at24 0-0050: supply vcc not found, using dummy regulator
Δεκ 06 17:19:27 lm-desktop kernel: at24 0-0050: 256 byte spd EEPROM, read-only

The other item I noticed was the below which is only in this last log. I do not know what might have happened to your sdb1 that it needed to recover.

Code: Select all

Δεκ 06 17:19:29 lm-desktop kernel: EXT4-fs (sdb1): recovery complete
Δεκ 06 17:19:29 lm-desktop kernel: EXT4-fs (sdb1): mounted filesystem with ordered data mode. Opts: (null). Quota mode: none.
Δεκ 06 17:19:29 lm-desktop systemd[1]: Mounted /home/lm/WD_1TB.

There did not seem to be anything specific near the end of the log which might indicate a freeze reason unless somehow there is something squirrelly in your LibreOffice install. If there is, I have no idea what it might be.

We'll see how 5.13 works for you. There is also the option to try 5.4 again now that you have the C States disabled.

[ylm]

That sounds hardware related. Either the hardware itself or the driver for the hardware.

You are right, but I cannot find where exactly there is the faulty part.

The other item I noticed was the below which is only in this last log. I do not know what might have happened to your sdb1 that it needed to recover.

Code: Select all

Δεκ 06 17:19:29 lm-desktop kernel: EXT4-fs (sdb1): recovery complete
Δεκ 06 17:19:29 lm-desktop kernel: EXT4-fs (sdb1): mounted filesystem with ordered data mode. Opts: (null). Quota mode: none.
Δεκ 06 17:19:29 lm-desktop systemd[1]: Mounted /home/lm/WD_1TB.

sbc is a data disk. Maybe the hard reset caused errors in the filesystem?

We'll see how 5.13 works for you. There is also the option to try 5.4 again now that you have the C States disabled.

Now I have another problem. Although I installed kernel 5.13, it is not shown in the grub menu. I updated grub, but still it is not shown. The default LM menu boots kernel 5.11!!!!!

[SMG]

sbc is a data disk. Maybe the hard reset caused errors in the filesystem?

That is a possibility.

Now I have another problem. Although I installed kernel 5.13, it is not shown in the grub menu. I updated grub, but still it is not shown. The default LM menu boots kernel 5.11!!!!!

If you check in Update Manager, is it showing the 5.13 kernel as installed?

Usually Mint boots to the highest installed kernel. If, for some reason, grub has been changed one would run sudo update-grub to have the grub change take effect.

The only reason of which I am aware that might cause the system to not update the kernel is if an installed operating system other than Mint is in control of grub. I think you only have Mint installed so that should not be a reason.

[ylm]

Now I have another problem. Although I installed kernel 5.13, it is not shown in the grub menu. I updated grub, but still it is not shown. The default LM menu boots kernel 5.11!!!!!

If you check in Update Manager, is it showing the 5.13 kernel as installed?

Yes...

Usually Mint boots to the highest installed kernel. If, for some reason, grub has been changed one would run sudo update-grub to have the grub change take effect.

I did that already...

The only reason of which I am aware that might cause the system to not update the kernel is if an installed operating system other than Mint is in control of grub. I think you only have Mint installed so that should not be a reason.

No, I also have MX 21 installed. But I think that I can update grub from any distro installed. Or am I wrong here?

[SMG]

No, I also have MX 21 installed. But I think that I can update grub from any distro installed. Or am I wrong here?

Only one distro can have control of grub. You need to update grub in the distro which has control of grub for changes to take effect. Since it sounds like the change did not happen for Mint, I would suggest going into MX21 and updating grub there. That may be what is needed to get Mint to boot the 5.13 kernel.

[ylm]

No, I also have MX 21 installed. But I think that I can update grub from any distro installed. Or am I wrong here?

Only one distro can have control of grub. You need to update grub in the distro which has control of grub for changes to take effect. Since it sounds like the change did not happen for Mint, I would suggest going into MX21 and updating grub there. That may be what is needed to get Mint to boot the 5.13 kernel.

You are right. I updated grub from mx21, and now I am running kernel 5.13. I will update the thread with any news.
Thanks again!

[oldgranola]

A slightly different thing to look at with random crashes is to monitor your temp sensors. Overtemp's can easily cause crashes. I keep "psensors" app handy but I like to always have the sensor data in view as its harder to come back to check afterwards. For that I have the applet https://cinnamon-spices.linuxmint.com/applets/view/337sensor monitory. Also pay attention to free memory as you play. LM still has gradual progressive memory leak problems.

[ylm]

A slightly different thing to look at with random crashes is to monitor your temp sensors. Overtemp's can easily cause crashes. I keep "psensors" app handy but I like to always have the sensor data in view as its harder to come back to check afterwards. For that I have the applet https://cinnamon-spices.linuxmint.com/applets/view/337sensor monitory. Also pay attention to free memory as you play. LM still has gradual progressive memory leak problems.

When I had the Mate version install, up to few days ago, I had an applet showing temperatures in real time. I seldom had over 45oC. The limit for my CPU is about 66oC. Now, during winter is difficult for the CPU to reach 50oC. I will install to applet though, to be sure. Also, I have 8GB of RAM and a 4GB swap space. The latter is almost never used! Usually my used RAM is up to 50% of total. I have an applet as well for this, to monitor it.

Thanks!

[ylm]

Right now I have C states disabled in UEFI, and kernel 5.13. If the problem is not solved, I am thinking to try the following steps:
a) C states disabled and revert to kernel 5.4
b) (a) + "i915.enable_dc=0" at /etc/default/grub in order to disable GPU power management.

If nothing of the above works, I think that I will rest my case, and try another distro! Also, at the same time I will install windows 10, in order to rule out any HW problem. If everything works fine under windows 10, then I will be sure that it is not a HW problem, but a strange incompatibility of my HW with Linux Mint (and I am afraid not only this distro, since I have the same problems under MX Linux. I am thinking to try another distro, not debian based, e.g., Manjaro).

But let's wait and see how things will be in the following days.

[SMG]

b) (a) + "i915.enable_dc=0" at /etc/default/grub in order to disable GPU power management.

Usually, if that is needed there would be i915 errors in journalctl. I did not see any errors related to the graphics driver.

[ylm]

b) (a) + "i915.enable_dc=0" at /etc/default/grub in order to disable GPU power management.

Usually, if that is needed there would be i915 errors in journalctl. I did not see any errors related to the graphics driver.

I see, I did not know that. Thanks again

[ylm]

Unfortunately this morning froze again. This time I was watching a video through FreeTube! I will try to revert back to kernel 5.4, but I am not optimistic anymore!

[SMG]

Video freeze issues can sometimes relate to settings in the app one is using to watch them. If watching in a browser, sometimes turning off/disabling hardware acceleration can help. Changing that setting isn't so much an issue of the hardware per se, but of how the software/app uses or interacts with the hardware settings (which means the issue could relate to the app).

However, it seems you have had a variety of situations where this has happened, so it could relate to hardware. Unfortunately, I do not have enough experience troubleshooting hardware issues to make any more suggestions than I already have. Sorry.

Maybe someone else will have ideas.

[ylm]

Thank you very much for all your help anyway!
I am using 5.4 now, just to be sure I checked all options
I also disabled hw acceleration in firefox and thunderbird.

[ylm]

Yesterday I had two freezes. Both of them after about 10 minutes of work! I do not have the will to try it any further with LM, I give up.
I reset the UEFI at the defaults settings, and I am testing MX Linux 21 ahs (Xfce, based on Debian 11, kernel 5.14 and some more updates for newer HW). Up to now everything works fine. But it is too early to tell.

Thanks again for all your help.

[SMG]

Thanks again for all your help.

You're welcome. Wishing you the best to find something that works with your hardware.

[ylm]

Well, finally a problem free combo with my HW

Last Saturday I installed MX Linux 21 AHS (from the MX site: MX-21 “AHS” (Advanced Hardware Support) comeswith a 5.14 kernel and updated mesa, xorg and vulkan drivers to produce a graphic stack more in line with newer "hardware, along with a few recompiled applications that utilize the later kernel.)

I also reverted all UEFI settings to their default values (this means that CS are enabled now).

For the whole week the only problem I encountered is the lack of resuming normally from a standby. All in all, everything works fine, and I am quite happy to find out that there is no HW problem on my PC. I still have not found what caused all these freezes with all Ubuntu based distros (not only with LM). But since everything is working fine that is OK for me.

[SMG]

Last Saturday I installed MX Linux 21 AHS (from the MX site: MX-21 “AHS” (Advanced Hardware Support) comeswith a 5.14 kernel and updated mesa, xorg and vulkan drivers to produce a graphic stack more in line with newer "hardware, along with a few recompiled applications that utilize the later kernel.)

I mean no offence (my hardware is older than yours), but you do not have newer "hardware" so perhaps there is some other difference which is more compatible with your hardware or the programs you are running.

There is a 5.14-oem kernal available for use with Mint. And Mint just updated the Xorg to 1.20.13. I do not know how that might compare to what you have now. I tried finding the versions being used, but all it said was "newer".

Good to hear you found something that is working well for you.

For the whole week the only problem I encountered is the lack of resuming normally from a standby.

That apparently is a known problem according to MX-21 “AHS” iso now available:

One known issue for older systems, particularly with intel-based graphics, is with suspend-to-ram, or at least resuming after suspend. If that feature is important to you and your PC can otherwise use the 5.10 kernel from the main releases, then sticking with the main releases or at least the 5.10 kernels is probably the way to go. (This is how the KDE/plasma release of MX-21 is set up right now).

---

Edited to add: from that same article:

Keep in mind, users of the mainline MX releases can always enable the ahs repository (via mx-repo-manager) and do regular updates. You’ll keep your 5.10 kernel, but you’ll get the later mesa , xorg and vulkan drivers.

Maybe you can do that and get your suspend working again.