Kernel 4.8.x - problems and more problems

[carlosfm]

Hello,

I'm on Mageia Cauldron with XFCE with my "old" Acer Core2Duo laptop with integrated Intel video, it's been working fine and rolling for years.
But since kernel 4.8.0 came, lots of boot problems started.
Sometimes it does not boot (hangs early in the blank screen before going graphical mode in the kernel boot), sometimes it hangs in the login screen, or just always hangs in the login screen for plenty of time, not accepting keyboard entries or mouse.
Closing the lid of my laptop is an immediate hang (and I have it set to do nothing on lid close!), it gets completely frozen with a black screen and the power off button is the only way.

Too many problems, then.
I've had no time to investigate this lately but today I looked at the kernel logs.
In the errors.log file I found two lines with the same error:

Code: Select all

[   56.800253] [drm:drm_atomic_helper_commit_cleanup_done [drm_kms_helper]] *ERROR* [CRTC:26:pipe A] flip_done timed out

A web search sent me here:

https://bbs.archlinux.org/viewtopic.php?id=218581

Looks like it's happening with Arch too.
This has been happening since Kernel 4.8.0, today came 4.8.4 in cauldron and the problems persist.
I'm loosing faith.
Linus was mad about some problems with the 4.8 kernel but it seems like it's major problems...
Am I alone with these problems with Kernel 4.8 in Mageia? I didn't find anything here.
I'm only hoping that these problems don't persist with the next (4.9) kernel or else I'm loosing faith in Linux.

[syschuck]

I've been a Linux user for years so I know when there are issues, but reporting a linux bug can be tricky. This is one. I have Calderon loaded on one of my laptops (And older model, Gateway M-1631U AMD Turion X2 TL60 with Radeon M1200 graphics and RS690 Sound, 4GB RM. The last Calderon update brought in a new kernel and it has caused some challenging problems. First in the typical graphical boot screen, the system proceed to about 4 bubbles, froze up, and then the screen slowly turned to bleach white. Really odd. After forcing a power down, I restarted in text mode proceeded to "checking new hardware" where it locked up. After messing around with journalctl (worst tool ever), I found that it was locking up on loading the module snd_hda_intel. So snd_hda_intel has a problem with this chipset that it didn't have before. Putting snd_hda_intel in the /etc/modprobe.d/blacklist-compat.conf is the workaround to get it to the desktop. Once in graphics mode the system seemed fine (without sound of course) until after a period of an hours use, they system locked up, frozen, unresponsive to everything except a forced power down. So another few hours later and the lock-up happened again but now with the screen slowly bleaching out to solid white again. Because these are lockups, I'm not getting any diagnostics to help. But yeah, 4.8 is causing problems here too.

[carlosfm]

The kernel 4.8 is causing problems in Arch and in the Manjaro forums the recommendation is to stay away from 4.8, too many problems.
It is also causing problems in Ubuntu.

[gohlip]

I have kernel 4.8 in Mageia, no problem in any version of 4.8 (now 4.8.6).
As for Manjaro, I'm now on 4.9.
In Manjaro, sometimes earlier kernel version (in unstable branch of repository, like Cauldron in Mageia), there may be problems, yes.
But unstable repo is for brave people to test and help developers and sure, there may be issues and to resolve - and therefore in stable repo, people won't get into trouble.

I also have kernel 4.8 in Kubuntu. Same, no issue anywhere in any 4.8 versions.

Yes, I know this is a Mageia forum but we as a responsible distro should not spread unfounded accusations against other distros.
There are many things I like about Mageia (and some things I don't like). Same goes for other distros, and we should address issues in the relevant distro forum.
But we should never denigrate other distros with disinformation.

[carlosfm]

The 4.8 problems in Manjaro are not just the RC kernels or the unstable repository.
You can find it by googling "kernel 4.8 problem"
Like:

https://forum.manjaro.org/t/cant-boot-w ... 8/10717/12

or:
https://forum.manjaro.org/t/kernel-4-8- ... ze/10784/5

and...

https://forum.manjaro.org/t/kernel-4-8- ... work/11372

--
Maybe it is hardware related and may not happen to everyone but if going back to a previous kernel all those problems disappear, then the newer kernel has problems.

[syschuck]

I have 4.8.6 on my desktop at work and have had no problems. It's a newer machine AMD A10-6800K APU Radeon OC 4.2Ghz. So this makes me think it's having issues with older machines. Also I have to throw in my gripe about systemd. Because it initializes things in parallel, it really gets hard tracking what is being started and what is running in the log files. All kinds of systemd log entries are scattered all over the kernel messages. Regardless, on my laptop, it seems to break starting plymoth. Typically I can get to 4 bubbles and then the lockup. Booting into text mode seems to never reach multi-user mode so I'm stumped. I tried going back to 4.7 but I think something changed in systemd on the update and now it's stuck and drops to the emergency "Ctrl-D to continue, or login as root".

[gohlip]

The 4.8 problems in Manjaro are not just the RC kernels or the unstable repository.
You can find it by googling "kernel 4.8 problem"
Like:
>
>


--
Maybe it is hardware related and may not happen to everyone but if going back to a previous kernel all those problems disappear, then the newer kernel has problems.

Yes, I'm aware of these links you provide.
To date, there are no issues. These (links) may be at unstable repo, or faced by older machines ( I think now these older machines are okay now with 4.8).

Other than the 'bug on' mistake by a kernel contributor (fixed within a few days) on 4.8.....


FYI, 4.9 face more problems (again at unstable) but have been sorted out too.
Also, we might expect (for all distros, including Mageia) that at unstable (or Cauldron) some issues were not picked up and passed on to stable or 'release'.
Don't forget in Mageia. 4.8 is in Cauldron, and we do not need to mention the risk a user takes in using Cauldron,
And also, particularly with older machines, we can always use older kernels.
Even new machines too as with cases like skylake (resolved) or nvme disks, etc, may need some time for kernels to catch up and may need kernel parameters to help out temporarily.

All these should be expected, but of course, we should alert our distro of any difficulty we face.
The main thing I want to stress is that we should not make sweeping statements or jump to conclusions (happens in every distro forum including Manjaro, I admit; but should be discouraged)

Anyway, looks like we've got these things off our chests now, let's move on to make these work.


As I said, 4.8 works fine with me on a good desktop and with an old inspiron laptop.
Perhaps it is other things like the newest update that broke plasma 5.8? I got that fixed too, not kernel related.
Perhaps you want to come out with a new topic on your problem so we can go at it afresh?

Cheers and good luck.

[carlosfm]

Did you read this thread? The first sentence of the opening post?
I'm with XFCE.
I'm updating every day and there are still problems with 4.8.6.
I may just give up eventually, but I'll wait for 4.9.
These problems are hard to diagnose, and it's not even consistent. Sometimes the machine hangs on boot, sometimes it does not - but it takes a much longer time to boot since 4.8.

Oh, and I know the risks of Cauldron, sometimes we even get RC kernels. But these are not RC kernels. Problems started with 4.8.0 and are still present.
I know the risks very well but this machine is rolling with Cauldron for more than 3 years, without major problems - until now.
And notice: I'm not complaining about Mageia. That's not the case, really.

[tlmiller]

Yeah, I have issues with my 4.8.6 on Arch. Before arch I had tried the latest Kubuntu (4.8 also) and it had issues. Strangely, my Mageia laptop has no issues (I assume it's 4.8 being the 6sta1 just updated last night, I'm not near it to check). Not surprisingly, I have no issues on either my Fedora (4.7.9) or Debian (4.7.6) machines, although I dread the Fedora switch to 4.8 after my experience with Arch and Kubuntu.

[gohlip]

And notice: I'm not complaining about Mageia. That's not the case, really.

Yes. Understood. Point taken. <thumbs up>

All distros modifies kernels with patches and is documented somewhere (like in kernel<>.scr.rpm).
And so there will be kernel differences between distros.

That said, usually it is not the distro modification kernels that is causing issues, but the different way distros handle lots of other things like graphic cards, de, dm....
The recent Mageia crash I experienced not only affected plasma but all other DE's, for example.
One other distro I have have a bad problem in the upgrade of systemd.

Hopefully and thankfully, these issues are in Cauldron or in 'unstable' and reported (not saying 'complaining') so that 'stable' or 'release' users will not be affected.
[and hope distro developers take note in forums and not always insisting bug reports filed to have any action taken (hint-Magiea developers)]

Cheers to all.

[tlmiller]

So just installed the 4.8 kernel in Debian...same exact issues as Kubuntu and Arch.

[carlosfm]

Guys, check your kernel logs and see if you have that strange error. See the first post of this thread.
This is a killer kernel.

[ozky]

Sounds cstate problem what my laptop suffer too you can try this if it works and prevent crashes.

Code: Select all

vim /etc/default/grub

Add this without quotes "intel_idle.max_cstate=1" to iside quotes GRUB_CMDLINE_LINUX_DEFAULT="

update-grub or update-grub2 then reboot system

edited line should look like this:

GRUB_CMDLINE_LINUX_DEFAULT="splash quiet noiswmd resume=UUID=6982aed3-e406-4d1f-a613-a22497c6290c intel_idle.max_cstate=1"

https://wiki.archlinux.org/index.php/in ... tel_driver

https://bugzilla.kernel.org/show_bug.cgi?id=109051

[carlosfm]

Baytrail?
I'm on a much older Intel CPU, a core2duo, and problems only started with kernel 4.8, never had these problems before.

[ozky]

That will affect to many other than baytrail only if you really read that kernel bug report more carefully.

[carlosfm]

I did, but nothing that is said on that bug happens to me. And they are talking about older kernels, while this is happening only after kernel 4.8.0.
What happens to me is described on the first post of this thread.
No hangs after it boots (when it boots), except when I close the laptop lid - sometimes it doesn't turn on the screen (I have it set to just turn off the screen on lid close) or there's no mouse and keyboard.
But others here and elsewhere (in the Arch forums) are complaining of frequent hangs.