Messages

376

Operating Systems and Porting / Re: Linux kernel packages 5.9.6 for Debian buster with 4k page size available

« on: January 03, 2021, 04:10:46 am »

Pocock, however, I also need to know how to remove a Kernel of these that is not in the list of kernels used by Debian.  If I query the system, it tells me it only has the standard 4.19 kernel and that's it.  How do I remove the 5.9 version I installed at the beginning, the one not posted by you?  And how should I try to install your Kernel on Fedora to do the test?  Is the procedure the same as in Debian?  Thanks

377

Operating Systems and Porting / Re: Linux kernel packages 5.9.6 for Debian buster with 4k page size available

« on: January 02, 2021, 02:45:06 pm »

Yes, if I'm not mistaken it takes mesa 19.2 minimum, I think I've read it around.  Yes, you can provide me with this package as long as you usually give me the exact procedure to install it.  Use this Kernel for Fedora?  I don't know how to install it though, to install a kernel on Fedora I usually use the "rpm" command and install the kernel package and the kernel core package which then carries its dependencies.  Here on Debian I have seen that it doesn't work like that and I don't know how to do it and I'm not able to create a custom Kernel, I'm not a developer like you unfortunately ...

378

Operating Systems and Porting / Re: Linux kernel packages 5.9.6 for Debian buster with 4k page size available

« on: January 02, 2021, 10:25:27 am »

Hi Pocock, I did the tests with this variant with 4k pages that you entered today. After installing the Kernel, I placed the usual GPU enable file in the usr / share / X11 / xorg.conf folder. The file just to be precise in everything has this content:
# AST2500
Section "Device"
    Identifier "GPU0"
    "Modesetting" driver
    BusID "PCI: 2 @ 5: 0: 0"
    VendorName "ASpeed ​​Corporation"
EndSection

# RX5700 XT
Section "Device"
    Identifier "GPU1"
    Driver "modesetting" # or amdgpu if you have xf86-video-amdgpu installed
    BusID "PCI: 3 @ 0: 0: 0"
    VendorName "AMD Corporation"
EndSection

# this is absolutely necessary, it tells xorg which GPU to use for the screen
Section "Screen"
    Identifier "Screen0"
    Device "GPU1"
EndSection

I use it regularly on Fedora too so I don't think there are any variables when working under Xorg.
After this I restart the computer, I start the kernel you posted, when the operating system loads, it first gave me that output that I posted earlier, that error that was seen during loading and consequently it blocked everything, now instead it no longer gives that error output, it also seems to want to activate the Radeon as it does a sort of restart by turning off and restarting, I see it from the Radeon writing that turns off and then after a few seconds it turns on again. It gets to the gray screen that usually appears before the mouse pointer and login but instead of making all these things appear it stays in that screen and everything crashes irreversibly. So unfortunately even if it seems that we have taken a step forward, we still can't get everything to work with the Navi 10. It seems as if that famous bug that continues to persist in the Fedora Kernel from version 5.7 onwards, is also present in this Debian Kernel. I don't know if there are files that the various Kernel of the various distros share in the same way, obviously apart from the GPU drivers, the fact is that even here we have a very similar problem. Having said that I keep pointing out that even when you load this kernel 5.9, during system loading, in the output list that Debian does when it is loading, it always keeps giving me that error of failure to load the Kernel modules. It does this with both the standard 4.19 and this 5.9. I do assume that Debian installed it more than once to always improve installation details, like some options I forgot or whatever, but in general the main installation has always been the standard one that everyone does. I don't know if it's a 10.7 issue or something else, I just know that that output has always been, so there are obvious Kernel problems as a whole on Debian ...

379

Operating Systems and Porting / Re: [NEWS] Debian

« on: January 01, 2021, 09:15:22 am »

The command to bring up these UUID files should be "blkid" from what I see on the web. In any case, it is better for me to open them from the operating system directly and then find the folder with these files and once the contents of the end have been launched through a text edit make the change you tell me, both on Fedora and on Debian, but first I have to find those righteous ... I'll try thanks

380

Applications and Porting / Re: Wayland support/port for PPC64el

« on: January 01, 2021, 07:05:55 am »

Even on Ubuntu now you can choose whether to start with wayland, just do it at the time of login .. Too bad that nothing works, everything crashes and you have to restart manually unfortunately ... I don't know if it depends on the 4k / 64k pages or other  ...

381

Operating Systems and Porting / Re: [NEWS] Debian

« on: January 01, 2021, 05:28:02 am »

The file you say is this, we are talking about Fedora in this case and as you can see it has no partition other than its own and in this moment the hard drive containing Debian is completely deactivated and unmounted from the computer...

382

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 31, 2020, 04:46:03 pm »

I don't know honestly, you obviously speak from inside Debian I suppose ... In any case, let's see if Pcock tells us something for further help it would not be bad ... Meanwhile, I take this opportunity to wish all of you Open Power friends! Happy New Year, Happy New Year to the Raptor staff!

383

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 31, 2020, 11:58:17 am »

yes I used the graphical installer but honestly I have not seen this option, strange ... now do you think I can do something to avoid this small inconvenience?

384

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 31, 2020, 07:44:32 am »

Guys one thing I didn't understand about the installation of this Debian Buster is that when I prepared the 4 partitions to install it, it touched the swap partitions of the other 2 hard drives where I installed Fedora and I don't understand why. I have regularly selected the empty hard disk where I had to install Debian, why I have to touch the swap partitions of both one and the other hard disk that have nothing to do with it! Now both hard disks with Fedora boot with a 2o seconds delay and this is very annoying honestly and I didn't think that doing this touch on the swap partition of those, would create this problem... I've tried reformatting the swap partitions on both and but the problem persists, they keep starting with delay...

385

Blackbird / Re: RAM slot B1 not showing

« on: December 30, 2020, 06:23:19 pm »

guys this problem just happened to me too.  I had 16GB with 2 banks of 8 each, lately I got a 16GB one and wanted to use it additionally but as soon as I put it in slot 1b, it stopped seeing me that slot and now it only detects slot A1, not seeing the slot anymore  B1 ... do you think I also have to execute that command by entering the Blackbird BMC?

386

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 29, 2020, 10:28:35 am »

Okay, then I'm waiting for your publication, let me know how to install it, the command you need or where to download it and tell me if I have to delete at this point, the current kernel 5.9.6 that does not work even if I think say yes at this point. I know the procedure on Fedora for deleting kernels, I always delete the kernel entry and the Kernel modules entry and he deletes the kernel in question. In case it's different let me know this too ... Thanks.
So you think it's just a page size problem, it would be amazing that a simple thing like this would affect the Kernel so badly that it crashes. This perhaps would also mean that the other distros could settle down the same way at this point ... Hopefully, I trust you Pocock, you're all of us ...

387

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 29, 2020, 09:49:53 am »

Xlinder no I didn't give that command, I should try to do it and see what you say but apart from that, in my opinion the underlying problem is in the Kernel after version 5.6. I also tried to install Ubuntu, up to version 5.4 of the Kernel, with Ubuntu 20.0.4, the internal gpu mode works and with AMD Navi 10, I tried to enable the card by inserting the usual file in xorg.conf, as reported in the guide on wiki and that made me enable Navi 10 on Fedora, but the operating system at that point doesn't start anymore. In any case, before inserting this file on Xorg, I updated Ubuntu to the latest version or the 20.10 which has in Kernel 5.8 and coincidentally, even in AST mode and without enabling the Navi 10, the computer just after it loads the system, yes it crashes irreversibly and must be shut down manually, just like with kernels from 5.7 onwards on Fedora. So in my opinion there is an underlying problem of the Kernel from version 5.7 onwards on Power that goes into constant conflict with the Navi 10 and that no one has yet unearthed for our misfortune. It is not possible that all distros that exceed version 5.6 of the Kernel suddenly stop working! You have to work on this first thing, you have to eliminate this damn bug that causes conflict between the drivers of Navi 10 and the new files contained by the Kernel 5.7 onwards on Power! Now I want to try to disassemble the graphics card and start Ubuntu version 20.10, do you want to see that if I remove the graphics card it starts? We'll see...

388

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 28, 2020, 10:29:47 am »

Guys then update, I was able to understand the problem of the failure to download the kernel 5.9, pocock forgot to tell me that first you had to run another command to be able to prepare the download of the package, I do not know the steps unfortunately, not being familiar with Debian. In any case I found a site that gave a mini-guide for the commands to do, including what I posted pocock, after as the last command also performs the installation of firmware for this new Kernel. At reboot I also put the file on usr/share/X11/xorg.conf.d, to enable the amd gpu but at reboot unfortunately gives me an error and tells me that it can not load the gpu... I send you the photos of both the error for the gpu and the error that has always given me at boot Debian but this time I speak of Debian itself and its Kernel. In fact I've noticed that I always get error loading Kernel modules, both with the old one that is installed by default, and with this new one. I don't know if it's related to the problem but it's a fact that it always gives it and it's very strange considering that it does it right after the installation and then without me touching or modifying anything, I've noticed since the first boot, that this error does it always. In fact I noticed that in normal operation with kernel 4.19, sometimes the video disappears and returns to server mode for example. This Debian seems to have its own underlying problems, regardless of the 5.9 kernel in my opinion ...I attach pictures of what gives me as errors both kernel and gpu...

389

GPU Compute / Accelerators / Re: nVidia official proprietary and open source code for POWER

« on: December 28, 2020, 02:50:47 am »

Too bad, I thought that at least those GPUs could be used in full.  I did not know they also did this type of operations or that they could use a GPU only for one purpose, inhibiting everything else, great good rip off I would say ...

390

Operating Systems and Porting / Re: [NEWS] Debian

« on: December 27, 2020, 05:01:07 am »

I didn't know they were doing this kind of settings for operating system 3D management now, thanks for the MPC info.  In any case I managed to get it to work normally with a trick.  If you quit with ctrl alt F2 and then re-enter with startx, disable 3D handling and go back to 2D and you can use it.  But my problem is that the command that Pocock gave me doesn't work.  When I give it it gives me an error telling me that it is not valid for APT :: Default-Release since this release is not available in the sources ... I tried to look for a site where I could directly download Kernel and Kernel modules to install it manually but I cannot  find it ... How can I solve?