Buster amdgpu graphics - Most stable approach

Getting your soundcard to work, using Debian on non-i386 hardware, etc

Buster amdgpu graphics - Most stable approach

Postby shep » 2020-01-25 18:13

I set up a neighbor with a new Ryzen 240GE gpu and Debian 10. I choose that GPU, mainly because my AMD 2200G system runs well with Buster and it is also Ryzen 3. With just Debian 10, the 240GE graphics would fail during kernel mode setting.
Based on this Debian wiki
https://wiki.debian.org/AMDGPUDriverOnStretchAndBuster2
I solved the issue by installing the Buster backports kernel which grub2 made the default.

My main question is whether the 240GE fix will persist through apt update/upgrade? The link suggests upgrading other components but so far it does not seem necessary.

A secondary question is if there is any benefit for my own system to upgrade to the newer kernel? It has been running well but not perfect.
Last edited by shep on 2020-01-25 18:50, edited 1 time in total.
shep
 
Posts: 394
Joined: 2011-03-15 15:22

Re: Buster amdgpu graphics - Most stable approach

Postby stevepusser » 2020-01-25 18:18

Since not everyone is running your same hardware, I think the best solution is for you to add a newer kernel and test it yourself. Don't remove the standard one--keep it as a fallback.

It's also quite easy to add and test a Liquorix kernel if you're looking for lower latency.
The MX Linux repositories: Backports galore! If we don't have something, just ask and we'll try--we like challenges. New packages: LibreOffice 7.0.0, QMPlay2 20.07.04, Pale Moon 28.12.0, KeepassXC 2.6.0, Arqiver 0.6.0+git, Foliate 2.4.2
User avatar
stevepusser
 
Posts: 11891
Joined: 2009-10-06 05:53

Re: Buster amdgpu graphics - Most stable approach

Postby Head_on_a_Stick » 2020-01-25 18:34

shep wrote:My main question is whether the 240GE fix will persist through apt update/upgrade?

Did you install the linux-image-amd64 metapackage from buster-backports? That should always depend on the latest kernel version.

I personally wouldn't install a newer kernel unless I had to because regressions are rife in Linux (it's not like OpenBSD, unfortunately) and the backported kernel lags behind slightly so you may be exposed to vulnerabilities for longer than with the stable kernel.

shep wrote:It has been running well but not perfect.

What problems are you having?
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: Buster amdgpu graphics - Most stable approach

Postby shep » 2020-01-25 18:50

Did you install the linux-image-amd64 metapackage from buster-backports? That should always depend on the latest kernel version.

I do not have the 240GE system in front of me but I believe it was a 5.2.x kernel and did not pull any other dependencies.

What problems are you having?

I have occasional browser sessions (chromium, firefox-esr and webkitgtk webkit2gtk with my vimb build) stop responding to mouse clicks on content url's. I can still click to close the session.
Last edited by shep on 2020-01-30 17:19, edited 2 times in total.
shep
 
Posts: 394
Joined: 2011-03-15 15:22

Re: Buster amdgpu graphics - Most stable approach

Postby Head_on_a_Stick » 2020-01-25 18:57

shep wrote:I believe it was a 5.2.x kernel

If it was a numbered version then it wasn't the metapackage and so won't be updated when a new major version becomes available.

So next time you have the system just run
Code: Select all
# apt install -t buster-backports linux-image-amd64

It's up to 5.4.0-0 now so you're already behind.

shep wrote:I have occasional browser sessions (chromium, firefox-esr and webkitgtk with my vimb build) stop responding to mouse clicks on content url's. I can still click to close the session.

That might be a compositor issue, which desktop is this? Have you tried starting the browsers from a terminal to see if they return any errors? Or check ~/.xsession-errors

And don't use the webkit-based browsers for anything important: https://www.debian.org/releases/stable/ ... r-security
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: Buster amdgpu graphics - Most stable approach

Postby shep » 2020-01-25 19:36

Thanks for the update syntax.

The webkit2gtk source package is covered by security support in Buster.


It currently has 3 cve's which is not bad:
Bug jessie stretch buster bullseye sid
CVE-2019-8846 vulnerable (no DSA, ignored) vulnerable (no DSA, ignored) vulnerable fixed fixed
CVE-2019-8844 vulnerable (no DSA, ignored) vulnerable (no DSA, ignored) vulnerable fixed fixed
CVE-2019-8835 vulnerable (no DSA, ignored) vulnerable (no DSA, ignored) vulnerable fixed fixed


The last Chromium update had many more cve's - some dated back months.

My system is running xcompmgr -n : simple client-side compositing.

I have these .xsession-errors concerning the ptr:

Code: Select all
(vimb:7023): Gtk-CRITICAL **: 11:22:34.768: gtk_widget_destroy: assertion 'GTK_IS_WIDGET (widget)' failed
Cannot connect to server socket err = No such file or directory
Cannot connect to server request channel
jack server is not running or cannot be started
JackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlock
JackShmReadWritePtr::~JackShmReadWritePtr - Init not done for -1, skipping unlock
shep
 
Posts: 394
Joined: 2011-03-15 15:22

Re: Buster amdgpu graphics - Most stable approach

Postby Head_on_a_Stick » 2020-01-25 20:15

So does the problem disappear if you kill xcompmgr?

You might want to try compton or picom instead, they are more capable than xcompmgr.
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12305
Joined: 2014-06-01 17:46
Location: /dev/chair


Return to Hardware

Who is online

Users browsing this forum: No registered users and 7 guests

fashionable