Why post twice ??
Can you explain what one needs to do to resolve the issues you are reporting here.
It can only help users if you explain what you are trying to do and what the results of your experiments are.
Please explain more details, so that I can repeat your problem on my ASUS Ryzen 5 3500U ASUS laptop!
My first post, I hit reply to myself instead of reply to the person who was answering my questions about Clear. It was not meant to be repetitious. I removed it.
I was loading a current build of Clear Linux onto a Lenovo IdeaPad S145-15API which has a AMD Ryzen 3500U and noticed KVM refused to work. As noted above, at first it seemed to be missing some KVM required modules and then I edited it to start at boot.
I am going back and installing an earlier LTS build of another spin based on Ubuntu and see if the KVM behavior is different. I seem to recall it worked, but I want to verify it before posting again.
I suspect it is something about the 5.4 kernel, but I will have to check.
I installed a Ubuntu LTS spin (Bionic) kernel (5.3.0-40-generic).
KVM runs just fine.
I do see the same IOMMU perf counter error, immediately followed by PCI INT A: not connected.
Then it gives a AMD-Vi: Found IOMMU cap 0x40 followed by a AMD-Vi feature enablements I didn’t see this in the Clear Linux logs. Like AMD-Vi: Virtual APIC enabled. AMD-Vi: Extended Features.
Also I did not see the mcelog errors about the AMD Processor Family 23, CPU is unsupported.
So I guess I will try another rolling release with a kernel past 5.4 and see if it duplicates Clear’s problem with AMD-Vi and the IOMMU.
I loaded the next spin of Ubuntu (Eoan). Installed KVM and confirmed it worked. Pushed the kernel to 5.4.18 which is same as the Clear Linux kernel tested prior.
The only changes in the boot logs here was the first appearance of the “AMD-Vi Unable to read/write IOMMU perf counter”. But all other AMD-Vi and IOMMU functions came up normally.
KVM worked just fine.
I also tried to install a rolling release of Manjaro with kernel 5.5 and the keyboard and touchpad failed to work. So I removed it.
I will continue to push the kernel forward, but it appears that Clear Linux in its current release configuration cannot properly enable the AMD-Vi and the IOMMU on the AMD Ryzen 5 3500U.
Update:
I now have the latest kernel available running with Ubuntu and still no issues.
So Clear Linux 32480 was installed on the AMD Ryzen 3500U yesterday and the results are the same as prior builds. At this point one should review the build README file on the Clear Linux repository for each build and keep an eye open for any AMD-Vi or AMD IOMMU notations.
Unless someone is aware of another platform that can confirm Ryzen 3500U support with Clear Linux, I would mark it as incompatible for now.
Hi,
I’m obviously a third party to this discussion, but I’m also concerned with Clear Linux on AMD hardware. Has @Edwaleni’s issue been fixed?
Also, does Clear Linux use any sort of preinstalled software-based OOBM?
I am planning to test the latest build for Clear shortly on the same Lenovo Ryzen 3500U. I will post back the results.
I had set the laptop aside and am just getting back to it.
Update: I am testing with build 36010 and I am not seeing the AMD-Vi errors I saw previously. All the AMD-Vi feature enablements were detected and enabled.
Also I noted that the IOMMU is mapping correctly now, I can see the PCI ID for it w/o the error as before.
4 Likes
Thanks for the follow up, and great to hear that it’s working for you!
The AMD microcode appears to be loading. But reports the following.
microcode: CPU0: new patch_level=0x08301055
microcode: CPUx: ...
x86/CPU: CPU features have changed after loading microcode, but might not take effect.
x86/CPU: Please consider either early loading through ignited/built-in or a potential BIOS update.
microcode: Reload completed, microcode revision: 0x8301055
1 Like