[SOLVED] Problem at reboot with kernel 4.19.4

[kalo86]

Hello dear Community of Porteus!

It is not the first time that I compile and install a Linux Kernel in Porteus, but in this last week I compiled maybe 10 times the Linux Kernel v4.19.1 and 4.19.2 applying only the AUFS4 patch.
To avoid corrupted patches I downloaded every time the AUFS4 patch using the git clone command and the output file patch is always the same (I am following the official guide from the Porteus website).
The AUFS4 patch is applied without errors.
The compilation of the kernel is fine, I can take the bzImage (vmlinuz).
After the "make modules_install" command I can copy the /lib/modules folder in order to make the new 000-kernel.xzm module.
For the firmware folder I use the huge firmware package from the Slackware Repository (yes, I use all the included firmware because I really don't care of the space, it is not a problem).
I can also make the crippled sources without issues.
But there is a problem after the boot process: I can see all commands lines and the system will block just before showing the LXDE desktop. Why? I still don't know
With CTRL+ALT+F3 I can login again (with root/toor) and then I can send the command "startx". Then I am able to work without problems.

Here I can attach the dmesg output:

https://www.mediafire.com/file/0cbvsung ... g.txt/file

I hope that someone can help me.

Best regards,
kalo86

_ _ _ _ _ _ _ _ _ _ _ _

Cause and Solution

Cause: I changed the hostname "porteus" to "kalo"
CONFIG_DEFAULT_HOSTNAME="porteus"
The bzImage was OK. The kernel compilation was OK. The issue was only at the boot after loading all modules.

Solution: I recompiled the latest kernel leaving the hostname "porteus" and now Porteus is starting the Desktop environment without problems.
Using the older 001-core.xzm and 002-xorg.xzm modules, I can boot Porteus with the latest kernel and the modified hostname.

[brokenman]

There may be a more usable error in /var/log/Xorg

Have you tried booting into text mode and running cache-module before running startx?

[kalo86]

Hi brokenman, nice to hear you again!
This evening I will post the output of /var/log/Xorg of course.
I need to figure out why the manual compilation of the kernel is giving problems at the boot and no problems if I use the compiled modules (for example from neko user).
Best regards!

kalo

[kalo86]

Hello brokenman,

This is the first Xorg log 0:
https://www.mediafire.com/file/kijdbn2k ... 0.log/file

And this is the Xorg log 1 because agter 80 seconds (approx.) there is a new login in the LXDE environment (it seems like this, I don't know what is happening):
https://www.mediafire.com/file/y8h5dmm1 ... 1.log/file

/var/log/messages:
https://www.mediafire.com/file/ad6y3ln4 ... s.txt/file

/var/log/syslog:
https://www.mediafire.com/file/c86ntyoz ... g.txt/file

I hope to understand what is the problem...

[kalo86]

I tried to compile the linux kernel v4.19.2 with the GCC v8.2.0 and I hav always the same result!
I am tired

[Hamza]

Regarding AUFS, I think you should give a try to 4.14.X which is a TLS version. Just be sure that AUFS is enabled as it's being disabled by default.

[kalo86]

Hi Hamza,
I applied the AUFS patch to the linux kernel v4.19.1 and v4.19.2 and at the boot I see that modules are mounted correctly.
I made also a comparison of the AUFS patch between the mine and the one which is found on the Neko kernel package: they are identical. So this is not caused by the AUFS patch, correct?
If I use the compiled kernel by Neko, the system will boot correctly (startx is run automatically without problem). If I compile the kernel by myself, I have that issue.
For this reason I am understanding that I have a problem during the compilation maybe.
Yesterday I compiled two kernels v4.19.2 (one starting from the Porteus 4.0 config file and the other from the Neko config file): the result is the same with the black screen after the boot process.
I updated the GCC compiler and the problem is the same.
There should be something that I am missing...
I will try to compile the kernel in text mode.

[Hamza]

Get his .config and compare if with diff tool. That's the most viable way to get sorted.

[kalo86]

Yes, but I compiled the kernel with the Neko's config file and the problem of the black screen after boot is the same.
The issue is not caused by the config file. Anyway Neko uses another version of GCC compiler which is different than the stock one built in Porteus v4.0. Updating to the GCC v8.3.0 did not solve the issue.

[kalo86]

I can add a detail (maybe it helps to find the issue):

1. I compile the kernel 4.19.2
2. I create the vmlinuz, 000-kernel.xzm and crippled_sources.xzm and place these new files on the right path of my USB drive
3. I reboot in text mode
4. I login with root/toor
5. I lunch the command "init 4"
6. No problems at all.

If I boot in always fresh mode, the system will boot and after the init 4 auto-command the screen becomes black

[Hamza]

Don't you have a large storage ? Had a similar issue at work where systemd was timing out when mounting all partitions in fstab.

[kalo86]

16GB is enough I think.
Anyway the problem is not the USB drive because if I boot with the Neko's kernel v4.19.2 I have no problems.
If I boot with my compiled kernel (000-kernel.xzm and vmlinuz) I have the black screen after init 4 command.

[Hamza]

Would you mind posting us some logs like kern.org, messages, syslog of a failed/successful boot ?

[kalo86]

Hello brokenman,

This is the first Xorg log 0:
https://www.mediafire.com/file/kijdbn2k ... 0.log/file

And this is the Xorg log 1 because agter 80 seconds (approx.) there is a new login in the LXDE environment (it seems like this, I don't know what is happening):
https://www.mediafire.com/file/y8h5dmm1 ... 1.log/file

/var/log/messages:
https://www.mediafire.com/file/ad6y3ln4 ... s.txt/file

/var/log/syslog:
https://www.mediafire.com/file/c86ntyoz ... g.txt/file

I hope to understand what is the problem...

These are the logs of days ago, thanks for helping me!

[Hamza]

I think this comes from the fact that working kernel has been compiled with a patched GCC (with retpoline patch) and yours doesn't have that patch.

Source: https://askubuntu.com/questions/1009725 ... reen-black

Did you try with a patched GCC compiler ?