2 boot work-around

[xilinder]

I have a minimal Debian install that I use for kernel building. It does not have X.
After tinkering around for a bit I noticed I had warm rebooted more than 3 times. So just for grins I did four more warm reboots. No problems.

I then started Debian 10.2 (with X) and killed the display manager and was able to warm reboot that 4 times, that is all I tried. The same for Ubuntu 18.04

While in X open a terminal, log in as root or use sudo,

root@deb /etc/init.d/lightdm stop ---or whatever display manager is on your system like xdm, gdm3, etc..

Then CTRL+ALT+F2, login as root(or sudo) and type reboot.

It seems to take a bit of time for X to shutdown in Ubuntu.

Please let us know if this works for you more than 3 reboots.

[madscientist159]

If this is in relation to the AMD GPU hang, the actual fix for the GPU ASIC not reloading firmware is in kernel 5.1.  Once we update the skiroot kernel to 5.4 (with the Navi patches) hopefully this problem will go away for good.

[MauryG5]

As for the Kernel, now having the version with the patch you are talking about,(5.5.0.1 or 5.4.0.2) how should we behave in front of the official updates that for example periodically releases Fedora? Do they affect our patched kernel? If so what should we do?

[MPC7500]

KLICK

[xilinder]

I'm not a Fedora expert.
Since Fedora will update a complete kernel and initramfs it should not delete a existing kernel but rather move it up the list in grub.
On Debian I rename my custom kernels so they always appear at the bottom of the grub list in petitboot. For example, I may rename them vmlinux-9x3, initramfs-9x3 and SystemMap-9x3.
The real name for these kernels is still within the kernel itself and the new name is only for grub.
I do not know if this works on fedora or not, and you would need to update grub which is different on Fedora then from Debian. Please consult the fans of Fedora for a proper way of doing this to keep your custom kernel as the first or default boot kernel.

[MauryG5]

KLICK

CONFIG_LOCALVERSION="amdgpu" but I have to run this command in the terminal before updating the kernel and then updating or I have to put it elsewhere I don't understand ...

[xilinder]

What is the full name of your custom kernel right now?

You can do, uname -r, in a terminal to find out.

[MauryG5]

Hi Xilinder, you know that command, I actually use 3 hardisks, one with Ubuntu 19.10 and the other 2 with Fedora 31 on both. In one of the two Fedora I have the Kernel 5.4.0.2, downloaded and installed with the changes already made, in the other I have the modified one that I installed through the Wiki procedure and it is the 5.5.0.rc1 +
Now I have seen that Fedora is sending Kernel updates, they update often, they are very efficient in this, the last official Fedora Kernel that I have is 5.3.16 and I don't know if now we go to 5.3.17 in any case c 'and' this update to do and before doing it I would like to know how to not affect the 2 modified Kernels that I currently have in the 2 Fedora ...

[xilinder]

@ MauryG5
Offiicel Fedora kernels are usually named 'vmlinux-5.5.0-.fc31.300-powerpc64le' or something like that.
Your custom kernel for the amdgpu fix should have a different complete name.
Please open a terminal and enter "uname -r" (without the quotes) and post it here for your fixed kernel(s).

The new kernels from Fedora should not overwrite your currently installed kernels, but they will add kernels to your GRUB selections for booting.

[MauryG5]

Then the one modified with the patch on a hardisk is this: (5.4.0-2.fc32.ppc64le)
In the other Hardisk I have this: (5.5.0-rc1 + 31)
In the second it does not carry the ppc64le extension. Both are patched to run AMD's Navi 10 GPU in my case the RAdeon 5700 XT 50 th anniversary editon. Usually Fedora does what you say or adds a version of Kernel without touching the previous one and deletes only the older one but after 3 or 4 versions ahead. I just know that there are parts of the Kernel that are in common and I'm afraid that updating I have to change the modified Kernels for that I ask how to behave ... Merklot for example said that you have to carry out that little procedure CONFIG_LOCALVERSION = " amdgpu "but I don't understand what you mean, if you have to do it as a command in the terminal or otherwise ...

[xilinder]

CONFIG_LOCALVERSION= is a compilier command.
If you re-compile the kernel you can add something to the end of the version string.
If you are going to do that you really should talk with Fedora guys.

It sounds like the patches may be in the new kernels very soon and what you are worried about may not be a concern.

[MauryG5]

In fact, for that I am cautious at the moment. I want to make sure I don't harm before updating. Let's see what happens in the next few days, in fact at the moment I don't think there is a need to update the Kernel since I have 2 versions that in the end are more advanced than what they have by default. As I said the last version was 5.3.16-ppc64le which still resides in my system, I currently have 2 versions to make the GPU work, which are 5.4 and 5.5 so I think I am ahead of them ... Having said that, do you or somebody else know about the software to connect the Android phone via USB with our Linux Power?

[MauryG5]

Nothing I think I found it, Android file transfer is also available for linux Power. Thank you