Ryzen 2200G Still Not Reliable in 4.18rc2, when will it work?

[ControlD]

Ribz
Thanks for the information. My last attempt was actually with the Mint 19 beta + the kernel you had linked to. That resulted in a system that could not boot, but only after I applied the available Mint updates. Maybe I will give this one more try.

[LasseKongo]

Did you have to do anything special on the initial install to get things working? I have been trying to get various distros working with my 2200g + B350 and I have had no luck at all. Every so often I can get a system installed and get a good boot, but once I update the kernel and/or add AMD drivers the systrem is pretty much toast.

Not that I can remember, when I first installed Tumbleweed it had the 4.15 kernel which had a number of problems, both performance and stability, with the 2200G. Had to run a discrete GPU until 4.16 was released, then performance was much better but still lockups almost every day. After 4.17 upgrade it is running more stable with no lockups so far.
Using this motherboard https://www.asus.com/us/Motherboards...MING/overview/

[bryanbr]

I guess I'll have to fire up a GT 1030 or better to see if that fixes the boot issues.

So quite obviously installing the GT 1030 last night fixed the boot and stability issues with my X470 and 2200G.

The boot issues frequently, but not always occur before the GRUB menu. I have little understanding of how the kernel and BIOS work together. Does a boot freeze before the GRUB menu mean that it's a BIOS issue? I guess there must be a kernel stub that runs prior to GRUB.

Does anyone understand why AMD would want to fix this in the BIOS instead of the kernel?

Additionally, if they supposedly fixed these issues in the 2400G, why is it taking so long to fix the 2200G? I mean shouldn't the architecture and kernel initialization between the two chips be very similar?

Doesn't AMD sell more of the cheaper 2200G chips than the 2400G chips so they would have more incentive to get the 2200G working first?

[ControlD]

The boot issues frequently, but not always occur before the GRUB menu. I have little understanding of how the kernel and BIOS work together. Does a boot freeze before the GRUB menu mean that it's a BIOS issue? I guess there must be a kernel stub that runs prior to GRUB.

I have the exact same issue with my B350 + 2200g. Most boot attempts never even result in my GRUB menu appearing.

[bridgman]

So quite obviously installing the GT 1030 last night fixed the boot and stability issues with my X470 and 2200G.

It's actually not obvious at all, but it does help to narrow down the problem.

The boot issues frequently, but not always occur before the GRUB menu. I have little understanding of how the kernel and BIOS work together. Does a boot freeze before the GRUB menu mean that it's a BIOS issue? I guess there must be a kernel stub that runs prior to GRUB.

Nope, GRUB runs before any kernel code executes. It displays & accepts input via SBIOS/VBIOS only.

EDIT - when you say "the boot issues frequently, but not always occur before the GRUB menu" does that refer to the case where you are using integrated GPU rather than the 1030 ?

Does anyone understand why AMD would want to fix this in the BIOS instead of the kernel?

See above.

Additionally, if they supposedly fixed these issues in the 2400G, why is it taking so long to fix the 2200G? I mean shouldn't the architecture and kernel initialization between the two chips be very similar?

The two chips are not just similar, they are identical other than some blocks being fused off. If we are actually seeing different behaviour between them that is useful - maybe a problem in the harvest logic, not sure.

Doesn't AMD sell more of the cheaper 2200G chips than the 2400G chips so they would have more incentive to get the 2200G working first?

Again, it's exactly the same code in both cases, although SBIOS might be different... but if the problem goes away when you stop using the iGPU then at first glance it doesn't seem like an obvious SBIOS/AGESA issue (unless SBIOS/AGESA is misrepresenting the fusing/harvesting to the VBIOS and/or GPU driver).

If the boot problems happen only with iGPU, and with Linux but not Windows then first thought would be GRUB and VBIOS not playing well together.

[ControlD]

If the boot problems happen only with iGPU, and with Linux but not Windows then first thought would be GRUB and VBIOS not playing well together.

I'm not sure if this helps or not, but in my case Windows 10 boots every time with no issues. If I try to boot Linux, a good amount of the time I never reach the GRUB screen. I do not however have a dual boot solution, these are two independent hard drives. In both cases I am using the iGPU. I can try adding a discrete card tonight and see what happens.

[bryanbr]

The 4.17.2 kernel is this one: https://github.com/M-Bab/linux-kernel-amdgpu-binaries

My guess of changed BIOS options that helped:
South Bridge Onboard HD Audio = disabled (I do not care, using HDMI sound)
I noticed that sound seemed to freeze 1 sec before complete freeze
NBIO Common Options Power Supply Idle Control = Typical Current Idle

Check BIOS options, that may be the key.

I tried disabling HD Audio on both my X470 + 2200G machines last night. And the boot failures and instability were still there.

Just as bridgman indicates, the pre-GRUB boot failures are tied to BIOSes that don't handle the 2200G chip. I think the post-GRUB boot failures are related to graphics driver initialization. And those failures occur for me even with X disabled on boot.

[bryanbr]

Nope, GRUB runs before any kernel code executes. It displays & accepts input via SBIOS/VBIOS only.

EDIT - when you say "the boot issues frequently, but not always occur before the GRUB menu" does that refer to the case where you are using integrated GPU rather than the 1030 ?

I've experience no boot failures with the 2200G Vega 8 disabled. I experienced no boot failures when using Windows.

That's interesting about the 2200G and 2400G being the same chip. I briefly considered getting the 2400G because there was a report by Michael Larabel that the boot issues resolved with the 2400G. I'd have to assume that these same issues are occurring for 2400G owners as well.

There were boot failures after Grub with the 2200G and my X470 boards. Even with X disabled there were a couple places in the boot process where the kernel reliably panics.

And there was misbehavior inside of Grub: sometimes the selected line in Grub couldn't be moved from the first or last line.

[bridgman]

Thanks everyone. This is starting to smell like a 2200G VBIOS issue.

I'll see what I can learn re: how different the VBIOSes are supposed to be between 2200G and 2400G.

[zap12344]

Hi, I have an asrock ab350m board with the latest beta bios with AGESA 1.0.0.3b and Ubuntu 18.04, I haven't experienced a freeze in weeks after I have updated the firmware for the raven ridge AND the amd microcode to the latest version (https://bugs.freedesktop.org/show_bug.cgi?id=105251#c9). Make sure to check that.
I still have problem booting half of the time, it's probably due to multiple problem, sometime the screen goes black during boot (https://bugs.freedesktop.org/show_bug.cgi?id=106225) other time it boots but it crashes during the boot.