#AMDGPU Phoenix1

1 messages · Page 1 of 1 (latest)

coral cave
#

CPU: 8-core AMD Ryzen 7 7840U with Radeon 780M Graphics (-MT MCP-)

lspci:

62:00.0 VGA compatible controller [0300]: Advanced Micro Devices, Inc. [AMD/ATI] Phoenix1 [1002:15bf] (rev c9)
    Subsystem: Lenovo Device [17aa:382d]
    Kernel driver in use: amdgpu
    Kernel modules: amdgpu

CONFIG_EXTRA_FIRMWARE=

dcn_3_1_4_dmcub.bin

gc_11_0_1_imu.bin
gc_11_0_1_mec.bin
gc_11_0_1_mes_2.bin
gc_11_0_1_pfp.bin
gc_11_0_1_me.bin
gc_11_0_1_mes1.bin
gc_11_0_1_mes.bin
gc_11_0_1_rlc.bin

vcn_4_0_2.bin

sdma_6_0_1.bin

psp_13_0_4_ta.bin
psp_13_0_4_toc.bin

smu_13_0_0.bin
smu_13_0_0_kicker.bin
#

dmesg | rg amdgpu: https://privatebin.net/?6aa9f81803701713#J98KTCfihCwxVmep77JqRxBpF5uXz2DwQp2suqTkSBVm

QUESTIONS:

  1. Do I need smu_13_0_0.bin or smu_13_0_0_kicker.bin?
    as it's mentioned in dmesg:
    [ 0.306583] amdgpu 0000:62:00.0: amdgpu: detected ip block number 4 <smu_v13_0_0> (smu)
    therefore I thought I'd need smu_13_0_0*.bin, but do I really need it and which varient of it and what does it do?
    because according to Gentoo's AMDGPU page only, NAVI31 NAVI32 NAVI44 NAVI48 need it.

How to fix those warnings:
2. securedisplay even though I disabled it (CONFIG_DRM_AMD_SECURE_DISPLAY), warning still persists...

[    1.103517] amdgpu 0000:62:00.0: amdgpu: SECUREDISPLAY: optional securedisplay ta ucode is not available
  1. RAS and RAP, no ta ucode warning
[    1.095116] amdgpu 0000:62:00.0: amdgpu: RAS: optional ras ta ucode is not available
[    1.103511] amdgpu 0000:62:00.0: amdgpu: RAP: optional rap ta ucode is not available
  1. Runtime PM not available warning
    [ 1.212633] amdgpu 0000:62:00.0: amdgpu: Runtime PM not available
red tinsel
#

what's the problem

coral cave
red tinsel
coral cave
west shard
#

Are you having any performance issues specifically which lead you to ask about these warnings? None of these messages really give much indication to a specific problem. With some more context I might be able to help more.

Also, do you have the linux-firmware package installed on your system? Can you run lsmod to show what modules are being loaded by your GPU?

  1. The _kicker fw package is for RDNA4/RX9000 series, you should be perfectly fine using smu_13_0_0.bin

  2. Secure display is a microcode-level support feature for security-sensitive and trusted applications (realistically, it's only needed on chromebooks, iot, and some enterprise security stuff). This warning can be safely ignored. As for why you are still getting a warning after disabling the kernel option, I am not sure.

  3. The absence of RAS & RAP ucode would likely not be causing any noticeable issues with your GPU. These provide feature support typically only needed in enterprise systems.

  4. This warning is stating that AMD Runtime Power Management is not being used. Once again though, without context this could really be an ignorable issue. I've only seen this feature used on SoC devices, I'm not sure if your GPU even supports it. I'd check to make sure Suspend to RAM is enabled in your kernel just in case though.

Power management options
    [*] Suspend to RAM and standby
#

@coral cave

coral cave
# west shard Are you having any performance issues specifically which lead you to ask about t...
  1. okay kicker one i don't need but do I really need: smu_13_0_0.bin
    [ 0.305190] amdgpu 0000:62:00.0: amdgpu: detected ip block number 4 <smu_v13_0_0> (smu)

  2. I understand but I still don't get why it still logs while CONFIG_DRM_AMD_SECURE_DISPLAY is turned off

  3. I see but how to eliminate that warning

  4. yes it's enabled: CONFIG_SUSPEND, why though?

additional questions:

should I enable:

  • CONFIG_DRM_AMD_ISP
  • CONFIG_DRM_AMD_ACP
  • CONFIG_DRM_AMD_DC
  • CONFIG_HSA_AMD
west shard
#
  1. I suppose you don't really need either, since smu_13_0_0.bin is really only needed for Navi31/Navi32 GPU's (which you do not have either). I'd say you're good to ignore this :)

  2. Honestly no clue why it's still giving the warning, possibly a bug or kernel is still compiling with this =Y for some reason. Either way, likely safe to ignore. As long as it isn't flooding your logs, it doesn't really matter.

  3. You might not be able to tbh. This warning is thrown by amdgpu_psp which is included in the amdgpu package by default. If you REALLY **REALLY ** need to disable this, either you will need to add the fw it's asking for (which is kinda pointless), or just patch out the warning from the kernel. I recommend just letting it throw the warning and just ignore it.

  4. I was just wondering, some systems will throw this warning if CONFIG_SUSPEND isn't enabled.

CONFIG_DRM_AMD_ISP: This is needed for some cameras connected to the APU (eg: mipi camera). It's not likely that you would need this feature on your laptop.
CONFIG_DRM_AMD_ACP: Could be helpful for HDMI audio, I'd keep it enabled just cuz why not :))
CONFIG_DRM_AMD_DC: Also helpful for full-featured display support. You'd probably be fine without it, but I'd enable it just in case.
CONFIG_HSA_AMD: Yeah, I'd enable this. It'll probably help a bit with performance since you're using an APU.

#

@coral cave

coral cave
west shard
# coral cave 1. Yes I didn't need `smu_13_0_0.bin`. 2. Yes but I really don't want to see it....

PM Runtime message is not an error, just a kernel msg.

If the warnings/messages are annoying, I think your best bet is just to adjust your kernel log suppression level to something a bit more minimal (as defined in printk).
(eg: https://superuser.com/a/793692)

A lot of these messages & warnings do not have an easy way to disable them without either writing a patch or disabling larger features. Unless these messages are flooding your logs, I'd recommend just ignoring them.