GPU (driver?) issue on AMD integrated graphics and GNOME: freezes and severe glitches
Linux - DesktopThis forum is for the discussion of all Linux Software used in a desktop context.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
GPU (driver?) issue on AMD integrated graphics and GNOME: freezes and severe glitches
Hi!
I have been experiencing GPU issues on my laptop lately (severe graphical glitches and full GPU resets).
Laptop: Thinkpad E15 gen 4, R5 5625U + Vega 7 integrated graphics. Running latest amdgpu driver. EndeavourOS.
The first time I experienced these issues I noticed severe graphical glitches every time something moved on the screen, flickering, polygon rendering just broke. I have a video but unfortunately it seems like I can't insert attachments yet. logs
Second time was yesterday: this time, no glitches, just a complete freeze with flickering every 5 seconds. I have logs indicating a page fault and then the GPU constantly resetting but failing: crash occurring and GPU failing to reset
Both crashes happened when I was running graphical intensive operations (Minecraft for the first and a browser game for the second)
Considering the nature of the crash (page fault) I immediately tested my hardware using UEFI utilities. All tests passed without any problem detected. Thus, I believe (and strongly hope) this is a driver issue or bug. For some reason during this second crash, I couldn't even access TTY but my music kept playing in the background.
Thank you in advance!
[EDIT] - the second crash is reproducible with a 100% success rate by loading this page. Works for a few seconds then freezes.
Last edited by lol-; 09-16-2023 at 05:06 AM.
Reason: added details about when crashes happen
Hi!
I have been experiencing GPU issues on my laptop lately (severe graphical glitches and full GPU resets).
Laptop: Thinkpad E15 gen 4, R5 5625U + Vega 7 integrated graphics. Running latest amdgpu driver. EndeavourOS.
The first time I experienced these issues I noticed severe graphical glitches every time something moved on the screen, flickering, polygon rendering just broke. I have a video but unfortunately it seems like I can't insert attachments yet. logs
Second time was yesterday: this time, no glitches, just a complete freeze with flickering every 5 seconds. I have logs indicating a page fault and then the GPU constantly resetting but failing: crash occurring and GPU failing to reset
Both crashes happened when I was running graphical intensive operations (Minecraft for the first and a browser game for the second)
Considering the nature of the crash (page fault) I immediately tested my hardware using UEFI utilities. All tests passed without any problem detected. Thus, I believe (and strongly hope) this is a driver issue or bug. For some reason during this second crash, I couldn't even access TTY but my music kept playing in the background.
Thank you in advance!
[EDIT] - the second crash is reproducible with a 100% success rate by loading this page. Works for a few seconds then freezes.
Hello, lol- & welcome to LQ.
I've an external AMD card and no issues. I wouldn't have recommended integrated graphics for gaming, but it should work, or fail predictably.
I loaded your "100% fault" page and it's unspectacular but fine.
You say 'page fault.' Do you mean a 'Segmentation fault?' If so, that's a memory error, brought about by dodgy memory or software. So you'd try with memtest86, fsck the disks, and then get suspicious of software.
We need to see the exact fault and output. The big "#" icon gives you and paste the vital log bits bang in the middle. The button to the left (of "#") gives a quote pair. The quotes have word wrap, the CODE tags don't, so lines are preserved. Look at the tutorials to find these things out.
There's lot of problems with amdgpu and/or Mesa on Lenovo. I've had these problems for years and can't find a solution. The older 'radeon' kernel module was more stable, but now recent Mesa won't work with it. Like you, random crashes and screen freezing. I can login remote, so only the display is shot. If you search, you'll find dozens of people with this problem.
The reason I'm telling you about this is so that you don't go crazy trying solutions that are unrelated to the problem. I'm guessing this is something that has to be fixed by the kernel/amdgpu/Mesa people. The bug reports I've looked at just get dismissed or closed after some time with no fixes.
@business_kid I only play Minecraft, integrated graphics have not failed me and performance is completely satisfactory.
No I didn't mean segmentation fault. The error happening is `Sep 16 00:43:38 lol kernel: amdgpu 0000:05:00.0: amdgpu: [gfxhub0] no-retry page fault (src_id:0 ring:24 vmid:6 pasid:32776, for process brave pid 1325117 thread brave:cs0 pid 1325143)` as per the logs I linked to in my original post. A page fault is indeed memory access related as well, and I have ran memory diagnostic tools, no errors have been detected. The specific page I linked to makes my system freeze after about 10 seconds on it. I have already given logs of the event, could you please be more specific as to what system logs would be useful? Would it be preferable to directly insert logs in my posts instead of inserting paste.rs links? I have done the latter in this thread because some of the lot snippets are lengthy. Thanks for the reply.https://stackoverflow.com/questions/...-vs-page-fault indicates that a page fault shouldn't raise an exception. In my case it does though.
There's lot of problems with amdgpu and/or Mesa on Lenovo. I've had these problems for years and can't find a solution. The older 'radeon' kernel module was more stable, but now recent Mesa won't work with it. Like you, random crashes and screen freezing. I can login remote, so only the display is shot. If you search, you'll find dozens of people with this problem.
The reason I'm telling you about this is so that you don't go crazy trying solutions that are unrelated to the problem. I'm guessing this is something that has to be fixed by the kernel/amdgpu/Mesa people. The bug reports I've looked at just get dismissed or closed after some time with no fixes.
If I understood correctly, the AMD drivers are buggy, and my system isn't at fault here? Well, it this happens to be the case, I guess I'll just live with some crashes...
I'm not sure if it's amdgpu or in Mesa, but yes, that's what I'm implying if indeed you do have the same issue as I. Sometimes I get weeks of uptime, sometimes it crashes in a day or two. I'm not sure what new GPU I could get that would fit and not be a problem so I just live with it. Save work often!
Well then, I guess I can live with it, I got to 8 days of uptime lately. I can't know for sure but I have the feeling a newer GPU would make the system even less stable @jayjwa
It's grabbing 1 Gig of memory. Phoronix speaks very poorly of cards that "only" have 8G of memory. By comparison, my 10 year old laptop had Intel HD4000 graphics, and that only grabbed 512MB of memory. When they describe those integrated GPUs there's much fewer of this and much fewer of that, so no wonder it's slower.
I now have my first half decent graphics card, and don't really want more than hdmi out of it. Once you get you're in the slow lane, be conservative in what you expect from it with games. Modern games are inclined to tell the box on startup: "Whatever you've got, I'll have all of it!" So, setting a lower video mode (e.g. 1280x720) might ease a lot of issues. You'd do that with a "PreferredMode" or "Virtual" setting (Or preferably both) in your xorg.conf.d/ settings.
EDIT: Don't be afraid of a new cpu, but don't be bullied either. It's entirely a choice of what you want to do with your hard earned money.The fancy graphics were never worth the price for me, but I'm not a gamer. I wanted decent 3d when I bought my RX6600XT, but google Earth still goes 2D on me in street view sometimes.
Last edited by business_kid; 09-17-2023 at 04:23 AM.
@business_kid
1. If all you say is true and considering most laptops don't have dedicated graphics how can 75% of people get away without any glitches?
2. iGPUs use shared system memory after their VRAM (which is also system ram but a reserved location is full, as far as I know.
3. Why are you talking about poor performance when I have already said I am getting several times the frame rate I need in games I play. FYI system monitor indicates 2gb of VRAM, which is what I have set in the BIOS and used never exceeds 1.7.
4. In your edit you mention a new CPU. I absolutely do not need this, my R5 5625u is fast enough for tasks like compiling or hosting Minecraft servers.
5. A slow system, low VRAM, or anything performance related cannot possibly explain this.
EDIT - I didn't realize this file sharing site was one-time download, my apologies
Sorry about typos. My edit meant to refer to a gpu, not as a cpu.
The video looks like a broken driver, but I guess not as it's a kernel issue. I would keep the framerate down as that lightens the workload on the GPU. I get the impression something's running out of time or space. I have a RazPi 4 which has poor gpu support. halving the frame rate improves it x2. Halving the dots per inch improves things x4.
Some (usually manufacturers) count megs & Gigs as even numbers 1,000,000 = 1Meg, and 1000 of those = 1 Gig. Others count them as powers of 1024, which is what is digitally addressed. So 2GB not being 2GB actually is a fact of life.
Lastly, I'd say that driving standards is actually an extremely complicated business and years of development go into the software. You absolutely should not be seeing that, but I've no sure fire cure for you.
@business_kid ah makes more sense. If I wanted serious GPU performance I'd definitely buy a desktop, but my budget doesn't allow for that right now. I have verified my 2G framebuffer is 2048 megabytes, so I guess the line you copied in your before-last message is indeed something going wrong . I am however extremely curious as to what sort of corruption would cause polygon rendering errors like that. Looks like math errors in the rendering pipeline to me.
The minimum gnome settings allows me to select is 60 fps. If my external monitor is plugged in I have 60 and 75. I have noticed no difference between both framerates. If you were talking about game FPS, haven't tested hut I'm almost sure it'll make little difference. Capped at 80 FPS my GPU usage is 35%, CPU load is 8% though, so definitely not maxed out
I have noticed a correlation between high system load and bugs of that nature though (running a VM dramatically increases the probability, especially if I play MC bedrock edition in the VM and Java on the host simultaneously.
Glitches of this nature have also occured without any game running, but with a lower severity (one or two widgets not drawn correctly)
Looks like stresses all right if you were able to effect some improvement. Sometimes there's a bottleneck slowing things down, like a pcie-4.x bus being held at pcie-3.x. I don't have your specs but if there's anything not new gone into your fairly new laptop, make sure it's up to speed. Even check that access timing for your ram is right. I think this should be a walk in the park for your box, but it's obviously not. I'd run htop in a terminal to see if it throws up any clues.
I've had these kind of faults before and you get an education while sorting them. You'll solve it, or somebody online will. I haven't much chance. The downside is that it's one class you can't drop out of :-P.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.
