During boot system pauses about 10 secs

I have two 15.1 systems here, only one of them shows the problem.

I can also boot the one with the problem in 13.1, which does not show the problem.

The problem is that during the boot the sytem pauses, not doing anything. Even the screen goes sleeping after a few seconds.
After about 10 secs the system returns to life and finishes booting and all is fine then.

I tried o find out what is displayed before and/or after this delay (all is going very fast) and I have the idea that it is somewhere around this snippet from dmesg:

    4.211081] usbcore: registered new interface driver ums-realtek
    4.350006] [drm] ring test on 0 succeeded in 1 usecs
    4.350010] [drm] ring test on 1 succeeded in 1 usecs
    4.350014] [drm] ring test on 2 succeeded in 1 usecs
    4.350021] [drm] ring test on 3 succeeded in 3 usecs
    4.350026] [drm] ring test on 4 succeeded in 3 usecs
    5.526989] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
    6.546928] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
    7.566899] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
    8.586871] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
    9.606843] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   10.626814] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   11.646786] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   12.666757] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   13.686729] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   14.706700] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, trying to reset the VCPU!!!
   14.726700] [drm:uvd_v1_0_start [radeon]] *ERROR* UVD not responding, giving up!!!
   14.726764] radeon 0000:01:00.0: failed initializing UVD (-1).
   14.726910] clocksource: Switched to clocksource tsc
   14.727088] [drm] ib test on ring 0 succeeded in 0 usecs
   14.727184] [drm] ib test on ring 1 succeeded in 0 usecs
   14.727208] [drm] ib test on ring 2 succeeded in 0 usecs
   14.727233] [drm] ib test on ring 3 succeeded in 0 usecs
   14.727265] [drm] ib test on ring 4 succeeded in 0 usecs
   14.727537] [drm] Radeon Display Connectors

Journalctl shows the same messages about UVD and VCPU (no idea what it is), but even in read!

Any ideas on how to investigate further.

Hi
Upgrade to 15.2 :wink:

What is the graphics card in use?


/sbin/lspci -nnk | egrep -A3 "VGA|Display|3D"

Depending on age, may need to configure uvd off with the grub kernel option radeon.uvd=0 I would add temporarily at boot and see it that helps, then make permanent.

I normally do not go to the next version before ironing out the problems in the current version. Specially as the problems did not exist in the earlier (13.1) version.

beneden:/home/wij/pr # /sbin/lspci -nnk | egrep -A3 "VGA|Display|3D"
00:01.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Kaveri [Radeon R7 Graphics] [1002:130f]
        Subsystem: Hewlett-Packard Company Device [103c:2b1d]
        Kernel driver in use: radeon
        Kernel modules: radeon, amdgpu
--
01:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Oland [Radeon HD 8570 / R7 240/340 / Radeon 520 OEM] [1002:6611]
        Subsystem: Pegatron Radeon R7 240 OEM [1b0a:90d3]
        Kernel driver in use: radeon
        Kernel modules: radeon, amdgpu
beneden:/home/wij/pr #

As said, I have no idea what these UVDs etc. are, but when you say that they can be “switched off”, I will try that.

Back in 10 mins.

Ok, adding

radeon.uvd=0

to the kernel parameters helps.

So the problem is “switched off”, I have no idea whatever I have “switched off” with it.
Does having this set to 0 any consequences on the system or system performance?
Is it a bug that I now circumvent, but that should be reported?

An additional question.

The output I provided seems to show two graphical cards. Is this assumption correct? And what is the use of that?

Hi
Since they show both VGA, I would guess one is part of the CPU, the other an onboard card in a PCIe slot. You could switch to the amdgpu by blacklisting radeon, I think you would get better performance? If you switch, then you would need to add amdgpu.cik=0 amdgpu.si=1 to the boot options, both of those cards are southern islands ones (hence si support).

Maybe in the system BIOS you can disable the onboard GPU, likely setting the primary adapter to PCIe.

Hi
No, see other post, I would switch to the amdgpu driver as this is the one to use for the cards… you will get better performance :wink:

I do not understand much of this (I konjw New Zealand has a Southern Island, but there it stops with me). Nor thus the term “si support” say anything to me.

But I will think it over. When I add this two kernel parameters, do I remove thje radeon.vdu=0 one?

In any case the 10 secs delay is gone. The wife will be happy. Thank you.

Hi
Southern Islands in the Pacific Ocean, see: RadeonFeature :wink: I would switch for better support/performance.