(1) How to automount a 2nd hdd using systemd (2) How to fix video screen tear (3) how to nomodeset in grub - Manjaro KDE Plasma 20.2.1

GCandy77 · January 27, 2021

Last night I installed Manjaro KDE Plasma 20.2.1 after using PopOs for just over a year as a noob in order to get Vulkan/dxvk working properly with my GPU. I have managed to get that working with my R9 280, but im still outta luck with the 290 (using the guides on the Archwiki) .. but in essence everything works and Im happy I made the jump. I dont know if my 1 year experience of using linux kicked in when I was setting Manjaro up, but I found the experience a lot easier than Pop and have got my pc back to where it was b4 + i can use vulkan

I have ran into a few things I cannot figure out myself and am in fear of bricking the Os completely like I have done already 3 times by being a donut ( I fixed it myself this time tho using the live usb + terminal route, so it make a nice change ).

I could not figure how to "no-modeset" in the grub tho as the normal "amdgpu.cik" wording is not there? If someone can point out what im missing it would be helpful (3)

(1) I found this guide https://forum.manjaro.org/t/root-tip-use-systemd-to-mount-any-device/1185 and managed to set it up so that 1 of my linux gaming partitions that is on a separate hdd loads no problems every time I boot, however the issue for me is I do not know if I need to create another "data-backup.mount" file and call it "data-backup.mount2" or to add it within the same file, in order to get a separate hdd with another linux gaming partition to also auto mount upon boot? I did try just adding it to the original file and it black screened the boot.

Im hesitant in just adding and creating a file as Im writing this just after bricked my Os again by adding this file "/etc/X11/xorg.conf.d/20-amdgpu.conf" ( by mistake again without thinking( i think )) and its configuration in order to get rid of screen tear (2).... I followed this guide https://wiki.archlinux.org/index.php/AMDGPU#Tear_free_rendering

Rebooted and got the black screen of death & then only did it dawn on me that I am using the Amdgpu experimental driver (installed via Pacman), which probably isnt the same as the one this guide is intended for? Is that correct? & does anyone know if this is fixable when using these drivers?

A cheeky 4th question n this may sound stupid, but remember Im a linux noob - I resolved the problem by booting in with the live usb, opening a terminal window from the directory that contained the 20-amdgpu.conf file and then went into "sudo su" mode and then ran rm 20-amdgpu.conf to delete it and then exit & reboot .... That fixed the issue, but is that the kinda correct way that it would normally be done or am I still going about it in a long route?

I have taken a look at some other guide but they seem to conflict with each other in slight ways that make me unsure if its the correct way, so I thought Id ask the questions here

Thx for reading

Wild Penquin · January 27, 2021

Hi GCandy77,

I'll try to answer your questions as you've numbered them.

1. dynamic mount via systemd units:

First question: why do you need to do this? Wouldn't listing the partitions in fstab suffice? This is also in the grey box right in the beginning of the Manjaro forum post you've linked.

(The OP there says the manual in in contradiction; I disagree with that, it's just a wrong interpretation the OP is making. I argue it makes sense to configure system so that the configuration is easy to read and change for humans. Systemd can have it's own logic and write unit files (and whatnot) dynamically if it makes sense from the software point of view. I believe this is rationale on the claim in the man page, although it is not explicitly mentioned: fstab is much easier to configure and more human readable).

However, in case you are determined you want to create these units (instead of using fstab) then yes, you need to create separate units (files) for every separate filesystem. In other words: copy the first unit file (you have working) to another file - which should be named [somedatanamewhichmakessensetome].mount. Change What and Where stanzas according to your partitions. If that does not make sense to you, post the outputs of 'lsblk' and 'blkid' commands here (as this heavily on your partition layout) and the contents of the current unit mount file you have (and which seems to be working).

(You could possibly mount both partitions in the same unit, one way or another, but it does not make sense usually since the partitions are separate filesystems; they might and should be allowed to fail separately in most sane scenarios I can fathom; there could be reasons, though, to deliberately make some unit file fail in case any combination of filesystems has failed).

2. tearfree rendering on amdgpu:

To remove guesswork from the repliers, post the configuration which you have made and which fails. Now it is not clear at all what you have actually done or if you've understood the wiki page correctly!

You should have something similar to this:


Section "Device"
       Identifier "AMD"
       Driver "amdgpu"
       Option "VariableRefresh" "true"
EndSection

In a file in /etc/X11/xorg.conf.d/XX-something.conf (where XX is a number and something is an arbitrary name; this is the convention, I'm not sure X.org actually requires it, though I'm pretty sure .conf is required).

You should probably not be using the amdgpu experimental driver from the AUR unless you have some specific problem you know is fixed in that driver (and know your way around in Linux systems and can fix it when it breaks!) or are developing the driver (frankly, from the looks of it it does sound like you have no idea why you have installed it from AUR). However this should not be a factor since VariableRefresh is a very old option in the amdgpu driver, unless that experimental branch is lagging way behind.

It seems that PKGBUILD installs a few extra configuration files on it's own, which I'm too lazy to check what they contain (a protip: you might want to post any configuration files contents deviating from a normal installation, such as these, in any question you ask).

3. Adding Kernel parameters (nomodeset)

Does this page help? Here it is told how to add default Kernel parameters to grub (or any other boot loader).

But this does raise some additional questions: what problems you are facing so that you think you need to use nomodeset?

Also, where do you get the idea amdgpu.cik is "the normal wording"? (btw. these are called Kernel parameters, and are space separated strings - their order is irrelevant. Btw: you can see your currently running Kernel command line by ´cat /proc/cmdline´.

4. The "cheeky question" about removing the configuration file

If all you did (before you got the "black screen") was adding this file, then the obvious way to revert the configuration is to delete the file. Generally, if you are not sure, you could move the file away somewhere (say, some folder in user home directory) so that X.org does not use the file anymore. That way you can revert your reversion (after deletion, the file is just gone).

In general, post as much information as you can. In this case, things that would be helpful in answering your question would be for example your partition layout (lsblk, blkid or similar information...), contents of all files in /etc/X11/xorg.conf.d (as you are using an experimental driver which has, it seems, installed extra configuration files here) - especially the file which caused the black screen, and, finally, what your kernel command line is currently set to, and how you've done that. EDIT: also the systemd mount unit file you have working, and the other one which didn't).

Also, as these questions are all mostly not related to each other, instead of one post you might like to create several posts, each with one question only (otherwise, thread might get quite convoluted and diffícult to follow, in case the problems are not simple, require several posts to solve, and all are being solved in one single thread, in parallel...).

Edited January 27, 2021 by Wild Penquin
TYPOs

GCandy77 · January 28, 2021

Hi Wild Penguin & thx for taking the time to answer all that

1 - I think u nailed the explanation bang on as its working all good, thx ... Im not against using FStab ... I just have a habit of bricking my system and I assumed this method was suggesting that if it did it wrong, it just wouldnt work, rather than brick the boot n would rather learn a bit more about FStab b4 going into it.

Heres my lsblk output ... blkid didnt output anything?

https://pastebin.com/Xa2XJ077

2 - This is what I did,

sudo nano/etc/X11/xorg.conf.d/20-amdgpu.conf

Then added

/etc/X11/xorg.conf.d/20-amdgpu.conf

Section "Device"

Identifier "AMD"

Driver "amdgpu"

Option "TearFree" "true" < I added this as per the guide (I assumed it went at the end like this?)

EndSection

I then rebooted and got the black screen of doom and then went about reversing it as mentioned.

Im using this driver as Im currently running a R9 280 (volcanic islands, GCN1) GPU & I want to use Dxvk/ Vulkan properly (especially the DX11/OpenGl 4.6 games) and I cannot do this using the Radeon drivers that come pre-installed as they always blackscreen the boot when the grub parameters are changed. Afaik this is a long term running issue for these cards & the 380/390`s too

Ironically even tho ive read the lastest kernel comes with built in support for the Volcanic Island cards, I still could not get Vulkan running until I installed the "amdgpu-experimental 20180518-1" drivers. As said, even tho these cards do a great job still in 2021 ... they are a right pain compared to a lot of other GPU`s to get working fully in linux .. But I got there in the end

I have to admit, I have no Idea what PKGBUILD is or what it does but I anticipate some interesting reading.

3 - It kinda makes sense, I think the bit that confused me is that whereas I was using the Radeon driver in Pop to boot there would normally be a line like

" GRUB_CMDLINE_LINUX_DEFAULT="quiet splash radeon.si_support=0 radeon.cik_support=0 amdgpu.si_support=1 amdgpu.cik_support=1"

I just couldnt see any line with the words "quiet splash" in it .... Im am dyslexic .. so there is a high chance I read it, but my brain didnt register it, so I will take another look. The reason I want to be aware of how to do this was just in case I run into another scenario with these lovely AMD cards & I saves time rather than using the LiveUsb method.

4 - Thx for those tips, il keep them in mind - very good!

Pretty much every thing is working for me now, I can even run Sea of thieves @1080p @ 60fps which totally blows me away. All I need to do now is figure out how to configure JACK so that I can use Alberton-Live 8 Via WINE ... Oh and fix the screen tear

Im so impressed with Linux ... yes I had pains setting it up but its been worth it in the end + its been a good learning experience and now this Os has helped me take my HW even further than I ever Imagined it could .. mental

Thx again for the time you put into explaining that in easy to understand method, its not only helped me fix my issue but explained the exercise in a way that makes sense for me to figure out how that works. Really grateful!

Also .. Point taken on about keeping the issues in separate posts

jojito · February 18, 2021

On 1/29/2021 at 1:12 AM, GCandy77 said:

1 - I think u nailed the explanation bang on as its working all good, thx ... Im not against using FStab ... I just have a habit of bricking my system and I assumed this method was suggesting that if it did it wrong, it just wouldnt work, rather than brick the boot n would rather learn a bit more about FStab b4 going into it.

I would also suggest using fstab, just see what needs to be mounted at startup and where it needs to be mounted. I would also suggest that you use UUIDs or Labels in fstab because all of the sdas can get mixed up and mess up all of the automounting. To see how to get labels and UUIDs you can check this out and if you need and this if you need more information about what exactly to put into fstab and how to format it. Hope this helps

GCandy77 · February 19, 2021

15 hours ago, jojito said:

I would also suggest using fstab, just see what needs to be mounted at startup and where it needs to be mounted. I would also suggest that you use UUIDs or Labels in fstab because all of the sdas can get mixed up and mess up all of the automounting. To see how to get labels and UUIDs you can check this out and if you need and this if you need more information about what exactly to put into fstab and how to format it. Hope this helps

Hi Jojito & thx for that ... I got it setup fine using systemd in the end, but I am trying to learn how to use terminal a bit more and those links are grea t thx, Kudos