cancel
Showing results for 
Search instead for 
Did you mean: 

Drivers & Software

baszczer
Adept III

Can't install amdgpu drivers on Ubuntu 20.04.1 (5.4.0-56-generic)

Hi,

System: Ubuntu 20.04.1, Gpu: Sapphire RX 5700 XT, Drivers tested (20.30, 20.40, 20.45)

When I try to install drivers, I get this:

Building initial module for 5.4.0-56-generic
ERROR: Cannot create report: [Errno 17] File exists: '/var/crash/amdgpu-dkms-firmware.0.crash'
Error! Bad return status for module build on kernel: 5.4.0-56-generic (x86_64)
Consult /var/lib/dkms/amdgpu/5.6.20.906300-1164792/build/make.log for more information.
dpkg: error processing package amdgpu-dkms (--configure):
installed amdgpu-dkms package post-installation script subprocess returned error exit status 10
dpkg: dependency problems prevent configuration of amdgpu:
amdgpu depends on amdgpu-dkms (= 1:5.6.20.906300-1164792); however:
Package amdgpu-dkms is not configured yet.

dpkg: error processing package amdgpu (--configure):
dependency problems - leaving unconfigured
Setting up xserver-xorg-amdgpu-video-amdgpu (1:19.1.0-1164792) ...
No apport report written because the error message indicates its a followup error from a previous failure.
Setting up mesa-amdgpu-omx-drivers:amd64 (1:20.1.6-1164792) ...
Setting up libegl1-amdgpu-mesa:amd64 (1:20.1.6-1164792) ...
Setting up libegl1-amdgpu-mesa:i386 (1:20.1.6-1164792) ...
Setting up libgl1-amdgpu-mesa-glx:amd64 (1:20.1.6-1164792) ...
Setting up libgl1-amdgpu-mesa-glx:i386 (1:20.1.6-1164792) ...
Setting up amdgpu-pro-core (20.45-1164792) ...
Setting up libgles2-amdgpu-mesa:amd64 (1:20.1.6-1164792) ...
Setting up libgles2-amdgpu-mesa:i386 (1:20.1.6-1164792) ...
Setting up libgl1-amdgpu-mesa-dri:amd64 (1:20.1.6-1164792) ...
Setting up libgl1-amdgpu-mesa-dri:i386 (1:20.1.6-1164792) ...
dpkg: dependency problems prevent configuration of amdgpu-pro-rocr-opencl:
amdgpu-pro-rocr-opencl depends on amdgpu-dkms (= 1:5.6.20.906300-1164792); however:
Package amdgpu-dkms is not configured yet.

dpkg: error processing package amdgpu-pro-rocr-opencl (--configure):
dependency problems - leaving unconfigured
Setting up vulkan-amdgpu-pro:amd64 (20.45-1164792) ...
No apport report written because the error message indicates its a followup error from a previous failure.
Setting up libosmesa6-amdgpu:amd64 (1:20.1.6-1164792) ...
Setting up libosmesa6-amdgpu:i386 (1:20.1.6-1164792) ...
dpkg: dependency problems prevent configuration of amdgpu-pro:
amdgpu-pro depends on amdgpu (= 20.45-1164792); however:
Package amdgpu is not configured yet.

dpkg: error processing package amdgpu-pro (--configure):
dependency problems - leaving unconfigured
Setting up comgr-amdgpu-pro:amd64 (1.7.0-1164792) ...
No apport report written because MaxReports is reached already
Setting up hsa-runtime-rocr-amdgpu:amd64 (1.2.0-1164792) ...
Setting up ocl-icd-libopencl1-amdgpu-pro:amd64 (20.45-1164792) ...
Setting up clinfo-amdgpu-pro (20.45-1164792) ...
Setting up hip-rocr-amdgpu-pro (20.45-1164792) ...
dpkg: dependency problems prevent configuration of amdgpu-pro-lib32:
amdgpu-pro-lib32 depends on amdgpu (= 20.45-1164792) | amdgpu-hwe (= 20.45-1164792); however:
Package amdgpu is not configured yet.
Package amdgpu-hwe is not installed.
amdgpu-pro-lib32 depends on amdgpu-pro (= 20.45-1164792) | amdgpu-pro-hwe (= 20.45-1164792); however:
Package amdgpu-pro is not configured yet.
Package amdgpu-pro-hwe is not installed.

dpkg: error processing package amdgpu-pro-lib32 (--configure):
dependency problems - leaving unconfigured
Setting up libglapi1-amdgpu-pro:amd64 (20.45-1164792) ...
No apport report written because MaxReports is reached already
Setting up libglapi1-amdgpu-pro:i386 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-dri:amd64 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-dri:i386 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-appprofiles (20.45-1164792) ...
Setting up libegl1-amdgpu-pro:amd64 (20.45-1164792) ...
Setting up libegl1-amdgpu-pro:i386 (20.45-1164792) ...
Setting up libegl1-amdgpu-mesa-drivers:amd64 (1:20.1.6-1164792) ...
Setting up libegl1-amdgpu-mesa-drivers:i386 (1:20.1.6-1164792) ...
Setting up libgles2-amdgpu-pro:amd64 (20.45-1164792) ...
Setting up libgles2-amdgpu-pro:i386 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-glx:amd64 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-glx:i386 (20.45-1164792) ...
Setting up opencl-rocr-amdgpu-pro:amd64 (20.45-1164792) ...
Setting up libgl1-amdgpu-pro-ext:amd64 (20.45-1164792) ...
Setting up amdgpu-lib (20.45-1164792) ...
Setting up amdgpu-lib32 (20.45-1164792) ...
Processing triggers for libc-bin (2.31-0ubuntu9.1) ...
Errors were encountered while processing:
amdgpu-dkms
amdgpu
amdgpu-pro-rocr-opencl
amdgpu-pro
amdgpu-pro-lib32
E: Sub-process /usr/bin/dpkg returned an error code (1)

1 Solution

well, i'm not baszczer, but still.

Thanks to baszczer, I've tryed to use 5.4.0-54 (the previous one) and it just works!

To install 5.4.0-54 do next:

sudo apt install linux-image-5.4.0-54-generic linux-headers-5.4.0-54-generic linux-modules-extra-5.4.0-54-generic

then reboot to 5.4.0-54

then

sudo apt remove linux-image-5.4.0-56-generic linux-image-unsigned-5.4.0-56-generic

(unsigned mentioned here because it tryed to install it for me by some reason.. it just helped)

then do

sudo apt autoremove

then go to /usr/src and remove linux-headers-5.4.0-56-generic and inux-headers-5.4.0-56 (if any of them are there)

now you can do

sudo ./amdgpu-install -y

View solution in original post

200 Replies


@bridgman wrote:

Going back to the 20.04 discussion, can anyone comment re: whether 20.45 is working on 20.04.1 for them ? Trying to figure out if we have a problem there as well, since it apparently worked OK on our internal testing.



I installed the Radeon 20.45 package (amdgpu-pro-20.45-1188099-ubuntu-20.04) on Ubuntu 20.04 after a fresh install a few weeks ago, which then broke last week following some automatic kernel upgrades.  I was on 5.8.0-36 and Ubuntu reports itself as 20.04.1.  Just today I found this post (several hours ago) and reinstalled my older 5.4.0-54 kernel just to get the amdgpu-install script to work again.  I just realized you're still posting and asking this question.  Is there anything I can do to help?  I think my answer is "yes I was having a problem with 20.04.1".  Earlier I collected some troubleshooting information preparing to submit a bug report that I never did, let me know if it would help you or if there is anything else I could do to assist.

0 Likes

has anyone managed to get amdgpu-pro drivers installed on ubuntu 20 LTS yet?

only thing I was able to do was go back to a manually backed up hard drive I cloned with 'gparted' last year... that SSD clone has ubuntu 20 lts with amdgpu-pro 20.40 on it and it works... just as long as I 'NOT run any updates'

that said, "I find it fascinating and strange Captain", after installation of a FRESH ubuntu 20 LTS onto a wiped ssd, amdgpu-pro 20.40-45 fails. [eyebrow raise] it's as if no one tested this before releasing it? 

https://community.amd.com/t5/drivers-software/amdgpu-20-40-fails-on-fresh-install-of-ubuntu-20-04-1/...

 

 I admit there could be something I'm doing and I hate to sound like I'm blaming AMD developers. Clearly they have all of us by the balls and there's nothing we can do except for grovel and beg on the support Forum. so...

"please. please help us. please sir. AMD, please, we needs the helps." [on knees]

 

0 Likes

OK, I'm not surprised that 20.40 is failing on latest 20.04.1 since we only updated 20.45 when Canonical updated the 20.04.1 kernel and removed the pci bios symbol, but 20.45 should be working - @mountkidd and I both confirmed that 20.45 includes a fix for the removed symbol.

We did test the 1188099 build on the latest 20.04.1 before posting but I'm wondering if there have been additional kernel changes that broke something else.

Would you happen to have a build log from the 20.45 install that shows what the build error was ?

 

0 Likes


@c_zagarskas wrote:

has anyone managed to get amdgpu-pro drivers installed on ubuntu 20 LTS yet?

only thing I was able to do was go back to a manually backed up hard drive I cloned with 'gparted' last year... that SSD clone has ubuntu 20 lts with amdgpu-pro 20.40 on it and it works... just as long as I 'NOT run any updates'

that said, "I find it fascinating and strange Captain", after installation of a FRESH ubuntu 20 LTS onto a wiped ssd, amdgpu-pro 20.40-45 fails. [eyebrow raise] it's as if no one tested this before releasing it? 

https://community.amd.com/t5/drivers-software/amdgpu-20-40-fails-on-fresh-install-of-ubuntu-20-04-1/...

 

 I admit there could be something I'm doing and I hate to sound like I'm blaming AMD developers. Clearly they have all of us by the balls and there's nothing we can do except for grovel and beg on the support Forum. so...

"please. please help us. please sir. AMD, please, we needs the helps." [on knees]

 



it's new, the best we've produced, looks good, but will it work - who cares, quick, quick, we have to make it before Nvidia, we'll fix the drivers later.

User is a loser, can buy and then he can cry. When the drivers will be fine-tuned, the card will be obsolete.

Maybe it's beyond human capacity, making good drivers, and it will always be like that, or it is like that because that's how corporations work.

 

0 Likes

As I mentioned previously, 20.45 (earlier ones also) does not install cleanly with kernels 5.8, 5.9, 5.10. The dkms module fails to build. It seems to be due to missing/changed symbols. I've put the log on pastebin but the forum seems to block my post with external url.

20.45 and 20.40 seem to build successfully with 5.4.0, the kernel that Ubuntu 20.04 and 20.04.1 ships with. It's been announced that 20.04.2, about to be released in february 2021, is going to be shipped with kernel 5.8 and that will introduce amdgpu install/build failures for many people.

Despite dkms failing to build, the included packages are installed and therefore come with their benefits and drawbacks:

  • VA-API video encoding works (ffmpeg, Shotcut, OBS Studio and others)
  • old version of Mesa (I think) which causes weird texture lighting errors observed in Daggerfall Unity with DREAM hd-texture mod
0 Likes


@baszczer wrote:
it's new, the best we've produced, looks good, but will it work - who cares, quick, quick, we have to make it before Nvidia, we'll fix the drivers later. User is a loser, can buy and then he can cry. When the drivers will be fine-tuned, the card will be obsolete.

Maybe it's beyond human capacity, making good drivers, and it will always be like that, or it is like that because that's how corporations work.


With respect, your message only works if you assume the possibility of time travel.

The issue here is that the drivers *were* tested and *did* work well with Ubuntu 20.04.1 at the time we shipped.

Unfortunately some time after that Canonical pushed updates to 20.04.1 which broke the driver (by making larger changes than normal for a .1 release, including removing kernel symbols). We fixed the driver with an update to build 1188099, but it seems that there may have been additional changes to 20.04.1 since then which broke the install again.

NVidia seems to have had similar problems with 20.04.1, only they broke with the 5.4.0-48 kernel while we broke with 5.4.0-56.

If you are saying that we should have travelled forward in time and tested the drivers against future versions of 20.04.1 before releasing and that we (and NVidia) callously refused  to do so, then I guess I can agree with that.

What we are looking at in order to prevent this happening in the future is regular testing of driver/OS combinations after release so that if a distro update breaks the driver then we have a good chance of finding out about it before it impacts users. This was not required in the past because enterprise distros tightly controlled the changes they made in already-shipped releases, but that no longer seems to be the case.

0 Likes


@bridgman wrote:

@baszczer wrote:
it's new, the best we've produced, looks good, but will it work - who cares, quick, quick, we have to make it before Nvidia, we'll fix the drivers later. User is a loser, can buy and then he can cry. When the drivers will be fine-tuned, the card will be obsolete.

Maybe it's beyond human capacity, making good drivers, and it will always be like that, or it is like that because that's how corporations work.


With respect, your message only works if you assume the possibility of time travel.

The issue here is that the drivers *were* tested and *did* work well with Ubuntu 20.04.1 at the time we shipped.

Unfortunately some time after that Canonical pushed updates to 20.04.1 which broke the driver (by making larger changes than normal for a .1 release, including removing kernel symbols). We fixed the driver with an update to build 1188099, but it seems that there may have been additional changes to 20.04.1 since then which broke the install again.

NVidia seems to have had similar problems with 20.04.1, only they broke with the 5.4.0-48 kernel while we broke with 5.4.0-56.

If you are saying that we should have travelled forward in time and tested the drivers against future versions of 20.04.1 before releasing and that we (and NVidia) callously refused  to do so, then I guess I can agree with that.

What we are looking at in order to prevent this happening in the future is regular testing of driver/OS combinations after release so that if a distro update breaks the driver then we have a good chance of finding out about it before it impacts users. This was not required in the past because enterprise distros tightly controlled the changes they made in already-shipped releases, but that no longer seems to be the case.


...no one came to Hawking's party, but who knows.

I know it adds nothing to the discussion, I wrote that because I'm angry and frustrated, but thanks for the info.

0 Likes


@baszczer wrote:


...no one came to Hawking's party, but who knows.

I know it adds nothing to the discussion, I wrote that because I'm angry and frustrated, but thanks for the info.


Yeah, I totally understand the frustration.

As far as I can see anyone on the HWE update path (whether they were knowingly on it or not) has been getting 5.8 kernels over the last couple of weeks while remaining on 20.04.1. Normally an HWE update like that would be part of a *.04.2 release in February. Don't know from Canonical yet if this was accidental or deliberate.

https://www.reddit.com/r/Ubuntu/comments/ks9lb9/was_the_hwe_kernel_upgrade_to_version_58_supposed/

There was a useful comment in that thread about an easier way to prevent the kernel updates:

"I chose to stay on GA kernel 5.4 for now, since I have no reason to fix what ain't broken. And 5.8 is evidently causing issues for a number of people. Removing linux-generic-hwe-20.04 and installing linux-generic successfully stopped it from wanting to upgrade.

AFAIK earlier LTS releases had HWE affect not only the kernel but xorg and mesa as well, but now I see that 20.04 HWE applies only to the kernel - I'm getting the mesa updates despite disabling HWE. All existing *-hwe-20.04 package names only start with linux."

This is an awkward change from Canonical - at first glance it seems that the new update policy is not sufficiently aggressive to handle new hardware (eg 5.9 kernel minimum required for our RX 68xx/69xx cards) but too aggressive and unpredictable for anyone trying to use out-of-tree drivers. We'll work out a solution ASAP but this has been a surprise for everyone AFAICS.

EDIT - I was just told that Canonical is now making their OEM kernel (currently at 5.10) available to all users, not just OEMs, and that is supposed to be their solution for new hardware support. That helps for gaming but not for users who require the -PRO packages... and I'm not sure if anyone has actually heard about this change.

0 Likes

What's the best way to stay notified for when there's a patch/update?

0 Likes

Good question, and something we are discussing. In the short term we might be able to do something with this forum, ie set up a thread for each release that people can subscribe to.

Before Canonical changed the way they handle Ubuntu LTS we didn't really need anything like that - very occasionally we would do a post-release fix but that was a rare exception rather than the rule.

My first thought would be to decouple OpenCL from the packaged drivers and otherwise toss the packaged drivers except for brief periods around new product launches - that would make finding out about updates moot - but as long as we have customers who prefer installing "AMD drivers" rather than picking up latest packages from their distro packager we may not be able to do that.

I just did a fresh install of Ubuntu 20.04.2 and amdgpu-pro with --opencl=par or --opencl=rocr still fails to install.

When are we to expect a fix???

Auggh, just realized we have two different threads going on this so different people are seeing different answers.

The Ubuntu wiki has been updated with instructions for moving from HWE to GA kernels - in the short term that is probably the best way to get the driver installed.

If you have already tried installing there will be some lock files on your system - do an uninstall first for anything you tried to install (eg pro vs non-pro have different uninstall commands), then move back to GA kernel following the instructions, then install the driver.

We are trying to pull in the date for 20.04.2 support since it seems that is being forced out to all users and the same files are being pushed to 20.04.1 users, but don't have ETA yet.

0 Likes

> just realized we have two different threads going on this so different people are seeing different answers.

Could you link the thread?

> The Ubuntu wiki has been updated with instructions for moving from HWE to GA kernels

Could you link it here?

> - in the short term that is probably the best way to get the driver installed

I just did the clean install of Ubuntu 20.04.2 that tracks GA kernel and neither AMDGPU-PRO+rocr nor ROCm worked.

Did you mean that I need to install the HWE kernel?

 

 

0 Likes

Whoops, sorry... I meant to include the links:

Other thread: https://community.amd.com/t5/drivers-software/amdgpu-20-40-fails-on-fresh-install-of-ubuntu-20-04-1/...

Ubuntu wiki: https://wiki.ubuntu.com/Kernel/LTSEnablementStack#Desktop

My understanding is that 20.04.2 now always tracks the HWE kernel unless you explicitly switch it back to GA. AFAIK the GA kernel should be 5.4.something while the HWE kernel is 5.8.something. The latest ROCm stack and packaged AMDGPU drivers will install on 5.4.x but not on 5.8.x.

Which kernel version did your fresh install of 20.04.2 give you ?

0 Likes

> My understanding is that 20.04.2 now always tracks the HWE kernel unless you explicitly switch it back to GA. AFAIK the GA kernel should be 5.4.something while the HWE kernel is 5.8.something. The latest ROCm stack and packaged AMDGPU drivers will install on 5.4.x but not on 5.8.x.

Actually you're right! I thought the LTS came with the GA kernel by default but it seams it now comes with the HWE.

> Which kernel version did your fresh install of 20.04.2 give you ?

It was 5.8.0-41-generic

 
0 Likes

Does anyone have a recipe that works??? I've read this thread all the way through and I am lost. As far as I can tell:

1. If you have 5.8.whatever you can MANUALLY patch it with some code that's about 9 pages back maybe???? but I cannot tell if that's for 5.4 which wasn't working prior to AMD PRO 20.45??? Which I'd rather not manually patch it anyway.

2. 5.4.whatever now works with AMD PRO 20.45 ???? But there seems to be no consensus on this or definitive statement on 5.4.watever works with AMD PRO 20.45

3. If AMD PRO 20.45 does work with 5.4.whatever Ubuntu has a recipe for switching from HWE to GA but no reference to doing that AFTER 5.8 was already distributed AND says ummm... don't do that.

4. A guy somewhere out on the inter webs seems to think THIS works...  https://math.dartmouth.edu/~sarunas/amdgpu.html  as of 2/21/2021??? I don't know but I've forked my system twice and spent a few yours un-un-re-un-doing amd pro-drivers to no avail. Kind of ridiculous given 20.04 is fairly old and this issue has no real answer or bottom line since December. 

0 Likes

re: #1 and #3 we are recommending following the Ubuntu instructions to switch from HWE to GA

re: #2, we fixed the problem introduced with 5.4.56 (removal of the pci_platform_rom() symbol) in an update to 20.45 so latest 20.45 should work with any 5.4.x Ubuntu kernel.

re: #4, the problem is that Canonical/Ubuntu recently pushed out a 20.04 update (after the Jan 20 post you linked) which makes it nothing like the original 20.04 (5.8.x kernel, new X.org and drivers etc...) and that is what is breaking driver install. Going from HWE back to GA updates should fix this.

I'm not seeing any pages relevant to Ubuntu 20.04 that discuss moving from hwe to ga. Is there a particular page you are referencing that you can link? If it's obvious and I should've found it I'm sorry. New to PC

0 Likes

These instructions return the error message that has been troubling me for hours, and I think is the crux of my problem. amdgpu and amdgpu dkms are not being found which returns the dpkg errorcode (1). This pops up when I try to install the driver from the amd page, and just about every workaround I can find. It may be the issue is then a different one to that which is being discussed, but I am unable to say if so. Thank you for the reply.

0 Likes

I'm a bit confused because I don't see any instructions there related to installing the driver. Which of the commands in the instructions I linked are giving you the DKMS messages ?

Are you religiously doing an uninstall after a failed driver install ? If you don't then every subsequent install will fail anyways because you have a half-completed install on your system.

Can you run "uname -a" and report what the output is ?

0 Likes

Linux JustFunctionPls 5.4.0-54-generic #60-Ubuntu SMP Fri Nov 6 10:37:59 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux

 

The linked page you provided does not mention installing a driver but that is what I need to achieve ultimately as

sudo lshw -c video 

returns

*-display UNCLAIMED
description: VGA compatible controller
product: Navi 14 [Radeon RX 5500/5500M / Pro 5500M]
vendor: Advanced Micro Devices, Inc. [AMD/ATI]
physical id: 0
bus info: pci@0000:0b:00.0
version: c5
width: 64 bits
clock: 33MHz
capabilities: pm pciexpress msi vga_controller bus_master cap_list
configuration: latency=0
resources: memory:e0000000-efffffff memory:f0000000-f01fffff ioport:e000(size=256) memory:fcc00000-fcc7ffff memory:c0000-dffff

And I was under the impression that this thread was addressing a solution to problems in then installing the amdgpu driver

0 Likes

I've made a fair bit of progress since yesterday. Now running on header 5.4.0-65-Generic. Also using oibaf PPA currently and that has fixed resolution. Trying to install the driver from the amd website still returns

Errors were encountered while processing:
amdgpu-dkms
amdgpu
E: Sub-process /usr/bin/dpkg returned an error code (1)

Which is the same as it said when resolution was poor. I checked that amdgpu file in usr/src/(...)65-Generic and it has 3 files which are all 0 bytes. If this is not meant to be the case, I have yet to see a solution unfortunately. Even though resolution is fixed, processing is super slow even in basic games like terraria which are running at 1/3 speed roughly.

0 Likes

@pupha   I just checked & /usr/src/amdgpu...1188099 has 2313 files and a size of 232MB.  Your installation seems to be messed up.  Also it seems you are mixing things from the oibaf PPA and amdgpu - do you really expect this to work?

0 Likes

Whether I am trying the 20.40 revision or the 20.45 revision the error code is the same. When I run sudo apt upgrade while the failed install is still there, it also creates the same error over resolving it. I've likely skipped past a working solution somewhere along the line so  basic comments won't be unappreciated, especially as I am very new to this and every failed process teaches me something.

0 Likes

I've tried installing without oibaf to receive the same error code. I have no idea what's going to work and what's not. I've uninstalled, deleted, downloaded, extracted and installed several times. There was a spell where I didn't wait long enough before extracting and that is probably what you see here else the download is funky. I'll check

0 Likes

@mountkiddMy current download of 1188099 has listed properties of 175 items, totalling 825.9 MB

0 Likes

Seems to be something wierd going on with the forum - none of my recent replies are visible, and mountkidd's post seems to have disappeared as well.

I'll go back through the last page or two, figure out which of my posts disappeared, and try to recreate where useful.

0 Likes

@pupha   Back home now...  I will outline the steps to follow for your install and you can cross-check this against what you have tried.

You have kernel 5.4.0-65 and a RX5500.  Can you confirm the installed version of Ubuntu & kernel?  In a terminal, use command lsb_release -sd   and  uname -r   to confirm the kernel

Assuming you are on Ubuntu 20.04.1 and given kernel 5.4.0-65, you must now use amdgpu 20.45 (1188099). and I assume you tried to install the -pro version. Confirm this as well...  amdgpu 20.40 only works on kernels 5.4.0-54 and earlier and probably doesn't support your card either, so it's a non-starter!

You have previously downloaded the amdgpu 20.45 bundle and extracted the contents to a convenient local directory.  cd to that directory.  There should be about 127 files, mostly .deb's and several install scripts.  

Uninstall previous amdgpu attempts with sudo ./amdgpu-pro-install --uninstall  and rebootYou can log all this using the script command - very useful for tracing things back...  If you also tried to install the generic version of amdgpu, then sudo ./amdgpu-install --uninstall  and reboot to ensure it all gets cleaned up.

Freshen up the packages list with sudo apt update, apt list --upgradable  and update anything that shows up.

Then, sudo ./amdgpu-pro-install -y --opencl=rocmrocm replaces the previous pal option and is required to support cards newer than Vega10, ie your RX5500 (navi 14).  script this as well so you can trace back through the install process.  Then reboot, one more time.

After the install, recheck /usr/src/amdgpu...1188099/ for number of files and size.

Let me know how this compares to what you have been doing...

"Then, sudo ./amdgpu-pro-install -y --opencl=rocmrocm replaces the previous pal option..."

rocm->rocr

Just spent the last few days banging my head against this. To summarise the previous hundreds of posts with what got me working.....


Fresh install of ubuntu 20.4.1 updated to 20.4.2 with Vulcan sdk 1.2148 and OIBAF graphic-drivers ppa

AS OF DRIVER 20.45 & UBUNTU 20.04.2- driver won't play with hwe stack, solution :- downgrade to GA kernel as ubuntu wiki  and this prev. post 

 

sudo add-apt-repository ppa:oibaf/graphics-drivers
sudo apt-get update
sudo apt-get upgrade
wget -qO - https://packages.lunarg.com/lunarg-signing-key-pub.asc | sudo apt-key add -
sudo wget -qO /etc/apt/sources.list.d/lunarg-vulkan-1.2.148-focal.list https://packages.lunarg.com/vulkan/1.2.148/lunarg-vulkan-1.2.148-focal.list
sudo apt update
sudo apt install vulkan-sdk
sudo apt install --install-recommends linux-generic
reboot
sudo apt remove --purge linux-generic-hwe-20.04 linux-oem-20.04 linux-hwe-* linux-oem-* linux-modules-5.1* linux-modules-5.8.0-* linux-modules-5.6.0-*
./amdgpu-pro-install -y --opencl=rocr,legacy
sudo apt install clinfo
clinfo
sudo apt-get install vulkan-utils
vulkaninfo

 

I'm not sure how much of that is nescessary (and I am not about to waste any more time finding out),but I now have a working driver

Note if the installation fails at the dkms part, uninstall with ./amdgpu-pro-install --uninstall then sudo rm /etc/modprobe.d/blacklist-amdgpu.conf to reset your graphics driver.

0 Likes

Back myself.

After going line by line I still got this:

Errors were encountered while processing:
amdgpu-dkms
amdgpu
amdgpu-pro-rocr-opencl
amdgpu-pro
amdgpu-pro-lib32
E: Sub-process /usr/bin/dpkg returned an error code (1)

0 Likes

@pupha   Are you installing from a terminal /commandline or using package managers like Discover or Synaptic?

Please post the commandline output from:

lsb_release -sd

uname -r

your complete  amdgpu-*   install command string

That's quite the mix of error messages - from both amdgpu & amdgpu-pro ???  Can you pick out & post specific errors that were reported?

E: Sub-process /usr/bin/dpkg returned an error code (1) looks like a Windows message.  Are you running under an emulator or vm? 

This is becoming an episode from Stranger Things... 

0 Likes

I was a dunce and did this on 20.04.2 and now can't see a 20.04.1 archive download. Going to try the other listed response above my reply since I'm on this version for now.

0 Likes

The 20.04.2 installation method listed above fails due to nomodeset (which is required to prevent black screen) interfering with the installation. Given that it is installing a driver, drivers being explicitly deactivated unsurprisingly interferes with that. I shall continue looking for the 20.04.1 download unless I also find a workaround for this. A solution is hinted at here but is unhelpfully vague: https://www.phoronix.com/forums/forum/linux-graphics-x-org-drivers/open-source-amd-linux/1087162-hel...

0 Likes

The archive of 20.04.1 is available here for others following the directions copied below:

http://old-releases.ubuntu.com/releases/20.04.1/

 

You have kernel 5.4.0-65 and a RX5500.  Can you confirm the installed version of Ubuntu & kernel?  In a terminal, use command lsb_release -sd   and  uname -r   to confirm the kernel

Assuming you are on Ubuntu 20.04.1 and given kernel 5.4.0-65, you must now use amdgpu 20.45 (1188099). and I assume you tried to install the -pro version. Confirm this as well...  amdgpu 20.40 only works on kernels 5.4.0-54 and earlier and probably doesn't support your card either, so it's a non-starter!

You have previously downloaded the amdgpu 20.45 bundle and extracted the contents to a convenient local directory.  cd to that directory.  There should be about 127 files, mostly .deb's and several install scripts.  

Uninstall previous amdgpu attempts with sudo ./amdgpu-pro-install --uninstall  and rebootYou can log all this using the script command - very useful for tracing things back...  If you also tried to install the generic version of amdgpu, then sudo ./amdgpu-install --uninstall  and reboot to ensure it all gets cleaned up.

Freshen up the packages list with sudo apt update, apt list --upgradable  and update anything that shows up.

Then, sudo ./amdgpu-pro-install -y --opencl=rocmrocm replaces the previous pal option and is required to support cards newer than Vega10, ie your RX5500 (navi 14).  script this as well so you can trace back through the install process.  Then reboot, one more time.

After the install, recheck /usr/src/amdgpu...1188099/ for number of files and size.

Let me know how this compares to what you have been doing...

 

--> Another commenter changed sudo ./amdgpu-pro-install -y --opencl=rocm to sudo ./amdgpu-pro-install -y --opencl=rocr

Will try both and state which is successful.

0 Likes

There was a typo in the original post - should have been rocr

Either 20.04.1 or 20.04.2 will be ok for the OS portion with amdgpu 20.45.  More important is the kernel.  With amdgpu 20.45, kernel 5.4.0 (other than -56) will be the easiest route for your install.  If you are on kernel 5.8.0 it might be wise to downgrade to 5.4.0-65 until reported issues with the 5.8.0 installs are resolved.

Same error output of:

WARNING: nomodeset detected in kernel parameters, amdgpu requires KMS

as with the 20.04.02 attempt. Did this happen for you? Will keep looking for workarounds for this.

0 Likes

What happens if you boot without nomodeset ?

You might have needed it for the inbox Ubuntu drivers (although I hadn't heard that before) but you definitely don't need it for the packaged drivers you presumably just installed.

0 Likes

@bridgmanBlack screen without nomodeset. It is not saved in the grub itself, I do a manual edit each boot (and when I forget the monitor outputs "no signal" after a moment). I expect it will not be necessary whatsoever when the packaged drivers are installed, but this error pops up when I do that: https://pastebin.com/qwXuRfGA

0 Likes