PC Drivers & Software

[wcleaver]

Greetings,

I have been trying to install the amdgpu driver for the AMD RX 6500XT video card on openSUSE LEAP 15.4 without success. The amdgpu driver for openSUSE Leap 15.3 works fine on openSUSE LEAP 15.3 (i.e., same physical system). The current driver package for openSUSE LEAP 15.4 is amdgpu-install-22.20.50200-1.noarch.rpm.

The openSUSE LEAP 15.4 installation is up to date:

~> uname -a
Linux 5.14.21-150400.24.11-default #1 SMP PREEMPT_DYNAMIC Sun Jul 17 20:46:33 UTC 2022 (be260ca) x86_64 x86_64 x86_64 GNU/Linux

~> cat /etc/os-release
NAME="openSUSE Leap"
VERSION="15.4"
ID="opensuse-leap"
ID_LIKE="suse opensuse"
VERSION_ID="15.4"
PRETTY_NAME="openSUSE Leap 15.4"
ANSI_COLOR="0;32"
CPE_NAME="cpe:/o:opensuse:leap:15.4"
BUG_REPORT_URL="https://bugs.opensuse.org"
HOME_URL="https://www.opensuse.org/"
DOCUMENTATION_URL="https://en.opensuse.org/Portal:Leap"
LOGO="distributor-logo-Leap"

The amdgpu-install package installes without any problem:

~> sudo zypper install /data0/REPOS/AMD-RX_6500XT/opensuse-15.4/amdgpu-install-22.20.50200-1.noarch.rpm
Loading repository data...
Reading installed packages...
Resolving package dependencies...

The following NEW package is going to be installed:
  amdgpu-install

1 new package to install.
Overall download size: 20.1 KiB. Already cached: 0 B. After the operation,
additional 32.3 KiB will be used.
Continue? [y/n/v/...? shows all options] (y): 
Retrieving package amdgpu-install-22.20.50200-1438746.noarch
                                           (1/1),  20.1 KiB ( 32.3 KiB unpacked)

Checking for file conflicts: .............................................[done]
(1/1) Installing: amdgpu-install-22.20.50200-1438746.noarch ..............[done]

The amdgpu-install package installes without any problem:

~> sudo zypper install /data0/REPOS/AMD-RX_6500XT/opensuse-15.4/amdgpu-install-22.20.50200-1.noarch.rpm
Loading repository data...
Reading installed packages...
Resolving package dependencies...

The following NEW package is going to be installed:
  amdgpu-install

1 new package to install.
Overall download size: 20.1 KiB. Already cached: 0 B. After the operation,
additional 32.3 KiB will be used.
Continue? [y/n/v/...? shows all options] (y): 
Retrieving package amdgpu-install-22.20.50200-1438746.noarch
                                           (1/1),  20.1 KiB ( 32.3 KiB unpacked)

Checking for file conflicts: .............................................[done]
(1/1) Installing: amdgpu-install-22.20.50200-1438746.noarch ..............[done]

 However, amdgpu-install fails to build the kernel module for any use case. The minimal use case needed is opencl which fails as:

~> amdgpu-install -y --usecase=opencl --opencl=rocr
Building repository 'AMDGPU 22.20 repository' cache ......................[done]
Building repository 'ROCm 5.2 repository' cache ..........................[done]
Loading repository data...
Reading installed packages...
Resolving package dependencies...

The following 13 NEW packages are going to be installed:
  amdgpu-dkms amdgpu-dkms-firmware comgr dkms hsakmt-roct-devel hsa-rocr
  hsa-rocr-devel rocm-core rocm-language-runtime rocm-ocl-icd rocm-opencl
  rocm-opencl-runtime suse-module-tools-legacy

13 new packages to install.
Overall download size: 52.3 MiB. Already cached: 0 B. After the operation,
additional 494.4 MiB will be used.
Continue? [y/n/v/...? shows all options] (y): y
Retrieving package amdgpu-dkms-firmware-1:5.16.9.22.20.50200-1438746.noarch
                                          (1/13),   8.6 MiB ( 62.3 MiB unpacked)
Retrieving: amdgpu-dkms-firmware-5.16.9.22.20.50200-1438746.noarch.rpm ...[done]
Retrieving package dkms-2.6.1-2.7.noarch  (2/13),  73.0 KiB (200.1 KiB unpacked)
Retrieving: dkms-2.6.1-2.7.noarch.rpm ....................................[done]
Retrieving package amdgpu-dkms-1:5.16.9.22.20.50200-1438746.noarch
                                          (3/13),   8.6 MiB (310.5 MiB unpacked)
Retrieving: amdgpu-dkms-5.16.9.22.20.50200-1438746.noarch.rpm ............[done]
Retrieving package suse-module-tools-legacy-15.4.12-150400.1.4.x86_64
                                          (4/13),  20.6 KiB (  8.9 KiB unpacked)
Retrieving: suse-module-tools-legacy-15.4.12-150400.1.4.x86_64.rpm .......[done]
Retrieving package rocm-core-5.2.0.50200-sles153.65.x86_64
                                          (5/13),  15.3 KiB ( 15.8 KiB unpacked)
Retrieving: rocm-core-5.2.0.50200-sles153.65.x86_64.rpm ..................[done]
Retrieving package rocm-ocl-icd-2.0.0.50200-sles153.65.x86_64
                                          (6/13),  20.5 KiB ( 41.4 KiB unpacked)
Retrieving: rocm-ocl-icd-2.0.0.50200-sles153.65.x86_64.rpm ...............[done]
Retrieving package hsakmt-roct-devel-20220426.0.86.50200-sles153.65.x86_64
                                          (7/13), 267.5 KiB (728.9 KiB unpacked)
Retrieving: hsakmt-roct-devel-20220426.0.86.50200-sles153.65.x86_64.rpm ..[done]
Retrieving package comgr-2.4.0.50200-sles153.65.x86_64
                                          (8/13),  33.8 MiB (116.5 MiB unpacked)
Retrieving: comgr-2.4.0.50200-sles153.65.x86_64.rpm ..........[done (1.7 MiB/s)]
Retrieving package hsa-rocr-1.5.0.50200-sles153.65.x86_64
                                          (9/13), 447.5 KiB (  2.3 MiB unpacked)
Retrieving: hsa-rocr-1.5.0.50200-sles153.65.x86_64.rpm ...................[done]
Retrieving package rocm-opencl-2.0.0.50200-sles153.65.x86_64
                                         (10/13), 440.2 KiB (  1.3 MiB unpacked)
Retrieving: rocm-opencl-2.0.0.50200-sles153.65.x86_64.rpm ................[done]
Retrieving package hsa-rocr-devel-1.5.0.50200-sles153.65.x86_64
                                         (11/13),  91.0 KiB (486.9 KiB unpacked)
Retrieving: hsa-rocr-devel-1.5.0.50200-sles153.65.x86_64.rpm .............[done]
Retrieving package rocm-language-runtime-5.2.0.50200-sles153.65.x86_64
                                         (12/13),   6.9 KiB (    9   B unpacked)
Retrieving: rocm-language-runtime-5.2.0.50200-sles153.65.x86_64.rpm ......[done]
Retrieving package rocm-opencl-runtime-5.2.0.50200-sles153.65.x86_64
                                         (13/13),   6.9 KiB (    9   B unpacked)
Retrieving: rocm-opencl-runtime-5.2.0.50200-sles153.65.x86_64.rpm ........[done]

Checking for file conflicts: .............................................[done]
( 1/13) Installing: amdgpu-dkms-firmware-1:5.16.9.22.20.50200-1438746.noar[done]
( 2/13) Installing: dkms-2.6.1-2.7.noarch ................................[done]
Loading new amdgpu-5.16.9.22.20-1438746 DKMS files...
Building for 5.14.21-150400.24.11-default
Building initial module for 5.14.21-150400.24.11-default
Error! Bad return status for module build on kernel: 5.14.21-150400.24.11-default (x86_64)
Consult /var/lib/dkms/amdgpu/5.16.9.22.20-1438746/build/make.log for more information.
warning: %post(amdgpu-dkms-1:5.16.9.22.20.50200-1438746.noarch) scriptlet failed, exit status 10
( 3/13) Installing: amdgpu-dkms-1:5.16.9.22.20.50200-1438746.noarch ......[done]
( 4/13) Installing: suse-module-tools-legacy-15.4.12-150400.1.4.x86_64 ...[done]
update-alternatives: using /opt/rocm-5.2.0 to provide /opt/rocm (rocm) in auto mode
( 5/13) Installing: rocm-core-5.2.0.50200-sles153.65.x86_64 ..............[done]
( 6/13) Installing: rocm-ocl-icd-2.0.0.50200-sles153.65.x86_64 ...........[done]
( 7/13) Installing: hsakmt-roct-devel-20220426.0.86.50200-sles153.65.x86_6[done]
( 8/13) Installing: comgr-2.4.0.50200-sles153.65.x86_64 ..................[done]
( 9/13) Installing: hsa-rocr-1.5.0.50200-sles153.65.x86_64 ...............[done]
(10/13) Installing: rocm-opencl-2.0.0.50200-sles153.65.x86_64 ............[done]
(11/13) Installing: hsa-rocr-devel-1.5.0.50200-sles153.65.x86_64 .........[done]
(12/13) Installing: rocm-language-runtime-5.2.0.50200-sles153.65.x86_64 ..[done]
(13/13) Installing: rocm-opencl-runtime-5.2.0.50200-sles153.65.x86_64 ....[done]
Executing %posttrans scripts .............................................[done]

Upon deinstalling a number of "no such file or directory warnings" are issued:

~> amdgpu-install --uninstall
[sudo] password for root: 
Reading installed packages...
Resolving package dependencies...

The following 11 packages are going to be REMOVED:
  amdgpu-dkms amdgpu-dkms-firmware comgr hsakmt-roct-devel hsa-rocr
  hsa-rocr-devel rocm-core rocm-language-runtime rocm-ocl-icd rocm-opencl
  rocm-opencl-runtime

11 packages to remove.
After the operation, 494.2 MiB will be freed.
Continue? [y/n/v/...? shows all options] (y): 
------------------------------
Deleting module version: 5.16.9.22.20-1438746
completely from the DKMS tree.
------------------------------
Done.
/var/tmp/rpm-tmp.YUhlTr: line 3: /var/tmp/amdgpu-dkms-kernels: No such file or directory
( 1/11) Removing amdgpu-dkms-1:5.16.9.22.20.50200-1438746.noarch .........[done]
( 2/11) Removing rocm-opencl-runtime-5.2.0.50200-sles153.65.x86_64 .......[done]
( 3/11) Removing amdgpu-dkms-firmware-1:5.16.9.22.20.50200-1438746.noarch [done]
( 4/11) Removing rocm-opencl-2.0.0.50200-sles153.65.x86_64 ...............[done]
( 5/11) Removing rocm-language-runtime-5.2.0.50200-sles153.65.x86_64 .....[done]
( 6/11) Removing rocm-ocl-icd-2.0.0.50200-sles153.65.x86_64 ..............[done]
( 7/11) Removing hsa-rocr-devel-1.5.0.50200-sles153.65.x86_64 ............[done]
( 8/11) Removing comgr-2.4.0.50200-sles153.65.x86_64 .....................[done]
warning: file /opt/rocm-5.2.0/hsa/lib/libhsa-runtime64.so.1: remove failed: No such file or directory
warning: file /opt/rocm-5.2.0/hsa/lib/libhsa-runtime64.so: remove failed: No such file or directory
warning: file /opt/rocm-5.2.0/hsa/lib: remove failed: No such file or directory
warning: file /opt/rocm-5.2.0/hsa: remove failed: No such file or directory
( 9/11) Removing hsa-rocr-1.5.0.50200-sles153.65.x86_64 ..................[done]
(10/11) Removing hsakmt-roct-devel-20220426.0.86.50200-sles153.65.x86_64 .[done]
update-alternatives: warning: alternative /opt/rocm-5.2.0 (part of link group rocm) doesn't exist; removing from list of alternatives
update-alternatives: warning: /etc/alternatives/rocm is dangling; it will be updated with best choice
(11/11) Removing rocm-core-5.2.0.50200-sles153.65.x86_64 .................[done]

It looks like the AMD installer is buggy or something is missing. However, there may be some change(s) to the AMD package that I've missed or is possibly lacking documentation. For example, older versions of amdgpu-install use the "amdgpu-uninstall" command to uninstall the amdgpu stack. The current 22.20 version uses the command "amdgpu-install --uninstall" to uninstall the amdgpu stack although the documentation still states to use the now deprecated "amdgpu-uninstall" command.

Does anyone have any info on how to get this correctly installed? The OpenCL component is desperately needed, other components would be nice.

Thanks much!