Arch only boots with kernel parameter amdgpu.dpm=0

puzzled_newbie · 2025-10-26 15:41:37

Hello everyone,

first of all my specs:
Motherboard: Gigabyte B450M S2H
CPU: AMD Ryzen 5 2600
GPU: AMD Radeon RX 580
on which I'm running Arch (kernel 6.17.5-arch1-1) with KDE Plasma 6.5 Wayland session, bootloader is grub.

The problem:
Some weeks ago I tried to install Arch and couldn't even get the install medium to work. The same thing happened with a swaybang and manjaro iso (bodhi linux did work).

I updated my uefi firmware
checked and replugged monitor and GPU cables

Only thing that worked was setting nomodeset as kernel parameter. I hoped to resolve this issue with GPU drivers afterwards.

I tried manually configuring the amdgpu driver like described on the Arch Wiki and many forum posts
reinstalled linux-firmware

Nothing worked.
Then I found a post in which amdgpu.dc=0 and amdgpu.dpm=0 were suggested. Setting amdgpu.dpm=0 as kernel parameter let's me boot without nomodeset and I finally had a system running where amdgpu even gets loaded.
The only downside (besides being solely a workaround as far as I understand) is, that my GPU fans are quite loud. Which seems to make perfectly sense with dpm disabled? But this is not a solution.
I tried using the linux-lts kernel, which also did not boot without amdgpu.dpm=0.

So here I am seeking for some help, as I suspect this being a problem with the specific hardware I'm using and since I have read in another post where a similar problem was discussed, that turning to the kernel people would be a good idea and maybe even reporting a bug. So maybe NewbieCorner is not the right place for this, please let me know if that is the case.

When I boot without amdgpu.dpm=0 as kernel parameter, the boot process freezes at

:: running early hook [udev]
Starting systemd-udevd version 258.1-1-arch
[    --------] [drm] amdgpu kernel modesetting enabled.
[    --------] amdgpu: Virtual CRAT table created for CPU
[    --------] amdgpu: Topology: Add CPU node
[    --------] amdgpu 0000:09:00.0: enabling device (0006 -> 0007)
[    --------] amdgpu 0000:09:00.0: amdgpu: initializing kernel modesetting (POLARIS10 0x1002:0x67DF 0x1458:0x22F1 0xE7)
[    --------] amdgpu 0000:09:00.0: amdgpu: register mmio base: 0xFCF00000
[    --------] amdgpu 0000:09:00.0: amdgpu: register mmio size: 262144
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 0 <vi_common>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 1 <gmc_v8_0>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 3 <gfx_v8_0>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 4 <sdma_v3_0>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 5 <powerplay>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 6 <dm>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 7 <uvd_v6_0>
[    --------] amdgpu 0000:09:00.0: amdgpu: detected ip block number 8 <vce_v3_0>
[    --------] amdgpu 0000:09:00.0: amdgpu: Fetched VBIOS from VFCT
[    --------] amdgpu: ATOM BIOS: xxx-xxx-xxx
[    --------] [drm] UVD is enabled in VM mode
[    --------] [drm] UVD ENC is enabled in VM mode
[    --------] [drm] VCE enabled in VM mode

(I had to copy this by hand, since I don't get any boot logs from booting like this, that's why this weird [ -----], I did not copy all these timestamps or whatever they are).
When I boot with amdgpu.dpm=0 set as parameter,

journalctl -b

gives

Oct 26 15:29:50 archdesktop kernel: [drm] amdgpu kernel modesetting enabled.
Oct 26 15:29:50 archdesktop kernel: amdgpu: Virtual CRAT table created for CPU
Oct 26 15:29:50 archdesktop kernel: amdgpu: Topology: Add CPU node
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: enabling device (0006 -> 0007)
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: initializing kernel modesetting (POLARIS10 0x1002:0x67DF 0x1458:0x22F1 0xE7).
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: register mmio base: 0xFCF00000
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: register mmio size: 262144
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 0 <vi_common>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 1 <gmc_v8_0>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 2 <tonga_ih>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 3 <gfx_v8_0>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 4 <sdma_v3_0>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 5 <powerplay>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 6 <dm>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 7 <uvd_v6_0>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: detected ip block number 8 <vce_v3_0>
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: Fetched VBIOS from VFCT
Oct 26 15:29:50 archdesktop kernel: amdgpu: ATOM BIOS: xxx-xxx-xxx
Oct 26 15:29:50 archdesktop kernel: [drm] UVD is enabled in VM mode
Oct 26 15:29:50 archdesktop kernel: [drm] UVD ENC is enabled in VM mode
Oct 26 15:29:50 archdesktop kernel: [drm] VCE enabled in VM mode
Oct 26 15:29:50 archdesktop kernel: Console: switching to colour dummy device 80x25
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: vgaarb: deactivate vga console
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: Trusted Memory Zone (TMZ) feature not supported
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: vm size is 64 GB, 2 levels, block size is 10-bit, fragment size is 9-bit
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: VRAM: 8192M 0x000000F400000000 - 0x000000F5FFFFFFFF (8192M used)
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: GART: 256M 0x000000FF00000000 - 0x000000FF0FFFFFFF
Oct 26 15:29:50 archdesktop kernel: [drm] Detected VRAM RAM=8192M, BAR=256M
Oct 26 15:29:50 archdesktop kernel: [drm] RAM width 256bits GDDR5
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: amdgpu: 8192M of VRAM memory ready
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: amdgpu: 7954M of GTT memory ready.
Oct 26 15:29:50 archdesktop kernel: [drm] GART: num cpu pages 65536, num gpu pages 65536
Oct 26 15:29:50 archdesktop kernel: [drm] PCIE GART of 256M enabled (table at 0x000000F400800000).
Oct 26 15:29:50 archdesktop kernel: [drm] Chained IB support enabled!
Oct 26 15:29:50 archdesktop kernel: amdgpu: hwmgr_sw_init smu backed is polaris10_smu
Oct 26 15:29:50 archdesktop kernel: [drm] Found UVD firmware Version: 1.130 Family ID: 16
Oct 26 15:29:50 archdesktop kernel: [drm] Found VCE firmware Version: 53.26 Binary ID: 3
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: [drm] Display Core v3.2.340 initialized on DCE 11.2
Oct 26 15:29:50 archdesktop kernel: [drm] UVD and UVD ENC initialized successfully.
Oct 26 15:29:50 archdesktop kernel: [drm] VCE initialized successfully.
Oct 26 15:29:50 archdesktop kernel: kfd kfd: amdgpu: Allocated 3969056 bytes on gart
Oct 26 15:29:50 archdesktop kernel: kfd kfd: amdgpu: Total number of KFD nodes to be created: 1
Oct 26 15:29:50 archdesktop kernel: amdgpu: Virtual CRAT table created for GPU
Oct 26 15:29:50 archdesktop kernel: amdgpu: Topology: Add dGPU node [0x67df:0x1002]
Oct 26 15:29:50 archdesktop kernel: kfd kfd: amdgpu: added device 1002:67df
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: SE 4, SH per SE 1, CU per SH 9, active_cu_number 36
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: Runtime PM not available
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: [drm] Registered 6 planes with drm panic
Oct 26 15:29:50 archdesktop kernel: [drm] Initialized amdgpu 3.64.0 for 0000:09:00.0 on minor 1
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: [drm] Failed to setup vendor infoframe on connector HDMI-A-1: -22 
Oct 26 15:29:50 archdesktop kernel: fbcon: amdgpudrmfb (fb0) is primary device
Oct 26 15:29:50 archdesktop kernel: Console: switching to colour frame buffer device 240x67
Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: [drm] fb0: amdgpudrmfb frame buffer device
Oct 26 15:29:50 archdesktop kernel: EXT4-fs (sdb3): mounted filesystem 4310e1d9-b012-4077-abfa-55f3ca23e28e r/w with ordered data mode. Quota mode: none.

I hope it makes sense to give you this passage of the jouranlctl, I don't want to hit you with unneccessary amounts of lines if you did not explicitly ask for them.

Other outputs are

lspci -k -d ::03xx

which gives

09:00.0 VGA compatible controller: Advanced Micro Devices, Inc. [AMD/ATI] Ellesmere [Radeon RX 470/480/570/570X/580/580X/590] (rev e7)
        Subsystem: Gigabyte Technology Co., Ltd Device 22f1
        Kernel driver in use: amdgpu
        Kernel modules: amdgpu

and

dmesg | grep amdgpu

gives

[    0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-linux root=UUID=4310e1d9-b012-4077-abfa-55f3ca23e28e rw amdgpu.dpm=0
[    0.048549] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-linux root=UUID=4310e1d9-b012-4077-abfa-55f3ca23e28e rw amdgpu.dpm=0
[    7.369024] [drm] amdgpu kernel modesetting enabled.
[    7.369161] amdgpu: Virtual CRAT table created for CPU
[    7.369180] amdgpu: Topology: Add CPU node
[    7.369289] amdgpu 0000:09:00.0: enabling device (0006 -> 0007)
[    7.369384] amdgpu 0000:09:00.0: amdgpu: initializing kernel modesetting (POLARIS10 0x1002:0x67DF 0x1458:0x22F1 0xE7).
[    7.369399] amdgpu 0000:09:00.0: amdgpu: register mmio base: 0xFCF00000
[    7.369401] amdgpu 0000:09:00.0: amdgpu: register mmio size: 262144
[    7.369593] amdgpu 0000:09:00.0: amdgpu: detected ip block number 0 <vi_common>
[    7.369596] amdgpu 0000:09:00.0: amdgpu: detected ip block number 1 <gmc_v8_0>
[    7.369599] amdgpu 0000:09:00.0: amdgpu: detected ip block number 2 <tonga_ih>
[    7.369601] amdgpu 0000:09:00.0: amdgpu: detected ip block number 3 <gfx_v8_0>
[    7.369604] amdgpu 0000:09:00.0: amdgpu: detected ip block number 4 <sdma_v3_0>
[    7.369606] amdgpu 0000:09:00.0: amdgpu: detected ip block number 5 <powerplay>
[    7.369608] amdgpu 0000:09:00.0: amdgpu: detected ip block number 6 <dm>
[    7.369610] amdgpu 0000:09:00.0: amdgpu: detected ip block number 7 <uvd_v6_0>
[    7.369613] amdgpu 0000:09:00.0: amdgpu: detected ip block number 8 <vce_v3_0>
[    7.369629] amdgpu 0000:09:00.0: amdgpu: Fetched VBIOS from VFCT
[    7.369632] amdgpu: ATOM BIOS: xxx-xxx-xxx
[    7.399096] amdgpu 0000:09:00.0: vgaarb: deactivate vga console
[    7.399099] amdgpu 0000:09:00.0: amdgpu: Trusted Memory Zone (TMZ) feature not supported
[    7.399158] amdgpu 0000:09:00.0: amdgpu: vm size is 64 GB, 2 levels, block size is 10-bit, fragment size is 9-bit
[    7.399321] amdgpu 0000:09:00.0: amdgpu: VRAM: 8192M 0x000000F400000000 - 0x000000F5FFFFFFFF (8192M used)
[    7.399325] amdgpu 0000:09:00.0: amdgpu: GART: 256M 0x000000FF00000000 - 0x000000FF0FFFFFFF
[    7.399462] amdgpu 0000:09:00.0: amdgpu: amdgpu: 8192M of VRAM memory ready
[    7.399465] amdgpu 0000:09:00.0: amdgpu: amdgpu: 7954M of GTT memory ready.
[    7.403318] amdgpu: hwmgr_sw_init smu backed is polaris10_smu
[    7.455936] amdgpu 0000:09:00.0: amdgpu: [drm] Display Core v3.2.340 initialized on DCE 11.2
[    7.648617] kfd kfd: amdgpu: Allocated 3969056 bytes on gart
[    7.648631] kfd kfd: amdgpu: Total number of KFD nodes to be created: 1
[    7.648727] amdgpu: Virtual CRAT table created for GPU
[    7.648879] amdgpu: Topology: Add dGPU node [0x67df:0x1002]
[    7.648882] kfd kfd: amdgpu: added device 1002:67df
[    7.648895] amdgpu 0000:09:00.0: amdgpu: SE 4, SH per SE 1, CU per SH 9, active_cu_number 36
[    7.649157] amdgpu 0000:09:00.0: amdgpu: Runtime PM not available
[    7.649790] amdgpu 0000:09:00.0: [drm] Registered 6 planes with drm panic
[    7.649793] [drm] Initialized amdgpu 3.64.0 for 0000:09:00.0 on minor 1
[    7.655800] amdgpu 0000:09:00.0: amdgpu: [drm] Failed to setup vendor infoframe on connector HDMI-A-1: -22 
[    7.664614] fbcon: amdgpudrmfb (fb0) is primary device
[    7.697542] amdgpu 0000:09:00.0: [drm] fb0: amdgpudrmfb frame buffer device
[    9.657181] snd_hda_intel 0000:09:00.1: bound 0000:09:00.0 (ops amdgpu_dm_audio_component_bind_ops [amdgpu])

Unfourtunately I don't understand too much of these, so I have to turn to more experienced linux users. Let me know if I can get more information for you.

I'm struggling with this issue for a couple of weeks now. On that very machine I ran manjaro for a couple of years without any problems. I wanted to do a reinstall anyways and migrate to Arch. So some weeks ago, when I had a weird firefox-not-playing-videos-bug which I couldn't solve, I took the step towards installing Arch after encountering this problem with the manjaro iso.

It seems to me, creating a new post here is my last resort. I apologise in advance, if I overlooked something on the internet I could have tried. But so far I did not have any luck with all of the posts on similar topics.

Thanks so much in advance!

Last edited by puzzled_newbie (2025-10-26 15:44:09)

seth · 2025-10-26 21:33:17

Setting amdgpu.dpm=0 as kernel parameter let's me boot without nomodeset

Interesting. Last thing I knew was that it broke booting altogether…

Oct 26 15:29:50 archdesktop kernel: amdgpu 0000:09:00.0: amdgpu: Runtime PM not available

Does "amdgpu.runpm=0" suffice?

puzzled_newbie · 2025-10-26 22:04:21

No, unfortunately not.
I just booted with "amdgpu.runpm=0" instead of "amdgpu.dpm=0" and let it sit for a while, but the boot freezes like before.

But I just saw that I forgot to mention following error message in the first post:

integrity: Problem loading X.509 certificate -22

But I have seen in an other thread, that this is not too much to worry about? It's not the last line before freeze as well...Booting with "loglevel=7" gives the same line as in the first post (the ones I had to copy by hand) and stops at "VCE enabled in VM mode".
(Secure boot is disabled, I have read somewhere that the certificate problems arise when secure boot is enabled?)

Edit: I have not mentioned it so far and I don't really know if it's not obvious, but "amdgpu.dpm=1" does not work!

Last edited by puzzled_newbie (2025-10-27 19:45:26)

schard · 2025-11-05 19:53:03

Mod note: Moving to Kernel Issues by implied request.

seth · 2025-11-05 20:31:20

Sorry for the long delay.
https://en.wikipedia.org/wiki/X.509 isn't gonna be related to this.

"amdgpu.dpm=1" is the default, so no surprises there either.

Does "amdgpu.ppfeaturemask=0xfcff8" work?
ppfeaturemask:
* PP_SCLK_DPM_MASK 0x1
* PP_MCLK_DPM_MASK 0x2
* PP_PCIE_DPM_MASK 0x4
* PP_SOCCLK_DPM_MASK 0x1000
* PP_DCEFCLK_DPM_MASK 0x2000

puzzled_newbie · 2025-11-06 17:56:25

Hi,

no problem, I just wasn't sure if it fit the Newbie Corner. Thanks for your reply!

Unfortunately your suggestion doesn't work. I booted with "amdgpu.ppfeaturemask=0xfcff8", but the boot process froze again.
The lines on which it got stuck are

Starting systemd-udevd version 258.1-1-arch
amdgpu: Overdrive is enabled, please disable it before reporting any bugs unrelated to overdrive

When I boot with loglevel=7 the last lines before freeze are, as before,

[    --------] amdgpu: ATOM BIOS: xxx-xxx-xxx
[    --------] [drm] UVD is enabled in VM mode
[    --------] [drm] UVD ENC is enabled in VM mode
[    --------] [drm] VCE enabled in VM mode

Is it possible to get some useful information out of a live bodhi linux iso, why this does work, but not any other distro?

topcat01 · 2025-11-06 19:04:27

Does linux-lts work any better? I'm also intrigued by the "Overdrive" message. Can you disable overdrive?

Last edited by topcat01 (2025-11-06 19:05:47)

puzzled_newbie · 2025-11-06 20:11:32

I did try linux-lts in the past without any success but just tried booting lts with ppfeaturemask which didn't boot either.

I'm sorry, I have to deal the first time with all of this, but from a quick google search on how to disable overdrive I just found to not use "ppfeaturemask" as kernel parameter, if I enabled it that way. So I don't quite understand how to disable it otherwise.

seth · 2025-11-06 20:30:26

"amdgpu.ppfeaturemask=0xf8ff8" - probably not the cause.

puzzled_newbie · 2025-11-06 21:00:55

No, it doesn't boot either with "0xf8ff8", though I don't get the "overdrive" line anymore. But I guess that's what you anticipated

seth · 2025-11-06 21:21:25

But I guess that's what you anticipated

Yup

I had hoped we could maybe fine-grain the DPM a bit more.
https://gitlab.freedesktop.org/drm/amd/-/issues/

There's only the RX 580 using amdgpu, no APU from the ryzen?

topcat01 · 2025-11-06 21:49:58

I don't really have any useful ideas, but sometimes it is advised to try linux-mainline if everything else fails. It might be a regression. Also curious if there is a last known good kernel.

puzzled_newbie · 2025-11-08 17:16:30

I'm not quite sure when I did the last kernel upgrade on the manjaro system I had installed prior to arch.
On the other hand, as I mentioned in the first post, bodhi linux did work when I tried some weeks ago, which uses 5.15 currently if I read that correctly of their download page? But bodhi also is compatible with a lot of hardware out of the box, right?

seth wrote:

I had hoped we could maybe fine-grain the DPM a bit more.

Any other suggestions where I could try to have a look and what topics I could read up on?

seth wrote:

There's only the RX 580 using amdgpu, no APU from the ryzen?

I'm not sure, sorry. I have to check on monday, I'm currently not at home. I will do that right away and update you.

Thanks for you help so far!

seth · 2025-11-08 20:16:19

Which topic?
https://bbs.archlinux.org/viewtopic.php?id=302858 ?

puzzled_newbie · 2025-11-10 17:41:39

Wikipedia says, the Ryzen 5 2600 does not have any integrated graphics unit, so none could interfere with the GPU...

seth wrote:

Which topic?

I just mean in general, but thanks for the link. I will try everything I can find.

How likely is it to be a bug? Does filing a bug report with the amdgpu or kernel people may be a good idea? Or do we not have enough information/certainty for that?

seth · 2025-11-10 19:54:44

How likely is it to be a bug?

Very - dpm is the default setting and supposed to work.

The question is what the cause is (since this is the first of a kind) and what can be done about it (if the GPU is just underpowered and dies when entering power saving, if this is a temporary condition and the system just runs into a power spike at boot but later on the GPU could power down fine, or it's a bug with the specific firmware or piece of hardware or everything above combined …)

Arch Linux

#1 2025-10-26 15:41:37

Arch only boots with kernel parameter amdgpu.dpm=0

#2 2025-10-26 21:33:17

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#3 2025-10-26 22:04:21

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#4 2025-11-05 19:53:03

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#5 2025-11-05 20:31:20

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#6 2025-11-06 17:56:25

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#7 2025-11-06 19:04:27

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#8 2025-11-06 20:11:32

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#9 2025-11-06 20:30:26

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#10 2025-11-06 21:00:55

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#11 2025-11-06 21:21:25

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#12 2025-11-06 21:49:58

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#13 2025-11-08 17:16:30

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#14 2025-11-08 20:16:19

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#15 2025-11-10 17:41:39

Re: Arch only boots with kernel parameter amdgpu.dpm=0

#16 2025-11-10 19:54:44

Re: Arch only boots with kernel parameter amdgpu.dpm=0

Board footer