Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Is the latest Client GPUs driver working as expected on meteor laker device? #811

Closed
ocean1ee1 opened this issue Feb 12, 2025 · 4 comments
Closed
Labels

Comments

@ocean1ee1
Copy link

I'm trying to run applications which used oneAPI on level zero device.

So from some guides, the GPU driver should be installed first.

And from hardware table, I can find my device is listed in that table

PCI IDs | Name | Architecture | Codename | Kernel | EU number
7D45 | Intel® Graphics | Xe-LPG | Meteor Lake | 6.8* | 64/48

The Kernel column represents the version of the upstream Linux kernel that first introduced support for each device. Kernel versions marked with an asterisk (*) require enabling support via the force_probe flag.


My environment is ubuntu 24.04 with a 6.12.13 kernel.

After installed these driver packages, sycl-ls can identify this igpu, however in dmesg, it shows it still use i915 related drivers.

  1. Tried to add xe.force_probe=7d45 and i915.force_probe=!7d45 to boot paras to start 6.12.13. I got a black screen and not able to login(even try to switch to command line mode), in this time I can use ctrl + alt + del to reboot.
  2. Tried to only add i915.force_probe=!7d45 to boot paras to start 6.12.13, wait a moment, it will let me login with a basic graphic driver. in dmesg, it shows my device is not officially support to use xe.
  3. Tried to add xe.force_probe=7d45 and i915.force_probe=!7d45 to boot paras to start 6.8.0, it will let me login, but in dmesg, it shows errors, and I cannot run application that required level zero device.
    6.8 kernel dmesg:
    [ 3.063764] xe 0000:00:02.0: vgaarb: deactivate vga console
    [ 3.076925] xe 0000:00:02.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
    [ 3.080849] xe 0000:00:02.0: [drm] HDCP support not yet implemented
    [ 3.090831] xe 0000:00:02.0: [drm] Finished loading DMC firmware i915/mtl_dmc.bin (v2.21)
    [ 3.353625] xe 0000:00:02.0: [drm] Using GuC firmware from i915/mtl_guc_70.bin version 70.20.0
    [ 3.459489] xe 0000:00:02.0: [drm] Using GuC firmware from i915/mtl_guc_70.bin version 70.20.0
    [ 3.466550] xe 0000:00:02.0: [drm] Using HuC firmware from i915/mtl_huc_gsc.bin version 8.5.4
    [ 3.476128] xe 0000:00:02.0: [drm] Using GSC firmware from i915/mtl_gsc_1.bin version 102.0.10.1878
    [ 3.784019] xe 0000:00:02.0: [drm] Selective fetch area calculation failed in pipe A
    [ 3.921435] xe 0000:00:02.0: [drm] GT1: found GSC cv1.0
    [ 5.640247] [drm] Initialized xe 1.1.0 20201103 for 0000:00:02.0 on minor 1
    [ 5.646419] xe 0000:00:02.0: [drm] Allocated fbdev into stolen
    [ 5.654215] xe 0000:00:02.0: [drm] fb1: xedrmfb frame buffer device
    [ 20.638486] xe 0000:00:02.0: [drm] ERROR [CRTC:80:pipe A] DSB 0 timed out waiting for idle (current head=0x4c0, head=0x0, tail=0x1080)

Since I'm new to this area, are these normal ? the meteor devices are only supported by i915?
Or is there other things I can try, or other information I need to provide ?

@JablonskiMateusz
Copy link
Contributor

Hi @ocean1ee1
Meteorlake platform is supported by upstream i915 KMD
Xe-KMD supports Lunarlake, Battlemage and onwards platforms

@ocean1ee1
Copy link
Author

ocean1ee1 commented Feb 12, 2025

Thanks. Mateusz

In this case, maybe the installation guide document need to be more specific about this. From that guide, I can see there are some gpus that supported by i915-kmd

May I know in future, will this change? Is there any plan or rumor to make Xe-KMD support to Meteor lake in sometime?

@eero-t
Copy link

eero-t commented Feb 12, 2025

May I know in future, will this change? Is there any plan or rumor to make Xe-KMD support to Meteor lake in sometime?

Upstream Linux policy prohibits ABI changes, such as changing the driver for a given HW after it has been already enabled for another driver, to avoid regressions (e.g. on upgrades).

I.e. while xe KMD may provide experimental support for earlier platforms, it's not going to be officially supporting platforms (like MeteorLake) that are already officially supported by the i915 KMD. This is/was documented in upstream: https://www.kernel.org/doc/html/v6.8/gpu/rfc/xe.html#xe-platforms

User-space drivers are naturally tested with the official KMD drivers. When force-probing unofficial drivers (for given HW), there should be no expectation that things work, at least not fully.

@ocean1ee1
Copy link
Author

Eero, thank you for explanation and time. I will close this

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

3 participants