Alder Lake init failed #1371
Comments
|
I have exactly the same error on Arch: |
|
We are seeing this on the Intel Alder Lake device Dell XPS 13 9315 with running Debian sid/unstable currently with libva2 2.15.0-1, and intel-media-driver-non-free 22.5.1+ds1-1. The issue was reported at the end of March, and almost five months have past without a roadmap. It’d be great, if you could give one. @mrhpearson, I am assuming Lenovo Thinkpads are also affected. Are you aware of this? |
|
There might be 2 rather than 1 issue here.
Patch for it is long available, see 1fefcd0. I've no idea what's wrong here. Someone needs to debug. In a meanwhile, @Brampie66, @aloise : do you still observe the issue on a more recent driver builds for this device?
|
|
Second issue (missing device ids) got addressed by 1350c5a (media-driver) and intel/vpl-gpu-rt@e790c7d (adding here for consistency). @paulmenzel : can you, please, help rebuild media-driver and try again on your side under Intel Alder Lake device Dell XPS 13 9315? Please, let us know the behavior. |
|
Hi, I did some updates and tried again: dmesg: packages: thx |
|
I do not see this issue running out-of-the-box media driver on Ubuntu 22.04: Also works with linux-image-unsigned-5.17.0-1015-oem, again on Ubuntu 22.04. Same out-of-the-box Ubuntu 22.04 user space stack. Can you help to collect output of From you dmesg log, this line somewhat bothers, thought later on I don't see errors from i915: Did you try to update linux-firmware and regenerate ramfs? |
|
Hi, I have the non K model, don't know if that makes a difference. cat /proc/cpuinfo | grep name I updated the firmware, dmesg is now: packages: xorg.0.log strace thx |
|
To completely wave away kernel setup issue you can try to run vainfo under ubuntu 22,04 docker. Something like this:
Thank you for the log. Problem is that I don't see any issue in it except that libva ultimately failed to initialize. (/etc/igfx_user_feature.txt don't count - that's just optional driver settings file, it's ok if it's not exist). At the moment the best I can think of is that something is wrong with the driver build. Can you highlight what's your system setup? If you will explain how you've configured system and especially user space - we can try to replicate on our side. |
|
Hi, I'll try on different hardware and see if I can add some more info. thx |
|
Hi, Using the docker method it does work |
|
Hi, I tried the ssd in a gigabyte brix with a J5005 and here it does work. dmesg thx |
|
Hi, I did some more testing, thx |
|
I'm having the same issue with on linux 6.0.2 and intel-media-driver 22.4.4 on archlinux and the following chipset: The only particular thing I did was adding those options to the i915 module: Here are the kernel logs: Not sure what the last line is about. Is there anything I can do to help moving forward with this issue? |
|
Seeing something similar on my Seen on a fresh install of Ubuntu 22.04. Tried to build the latest relevant packages myself and got this: |
|
Got it working! Submitted a patch: #1544 The fix is that not all PCI IDs are registered yet. I found a bug where this was fixed in MESA awhile ago. I'm not sure if it'll fix other's use of this driver, but it fixed mine with the latest builds. |
|
@Nexarian any idea how I can check what's the device id on my computer to see if it's the same you added or some more are missing? |
|
@victornoel Follow the steps here: https://dgpu-docs.intel.com/devices/hardware-table.html
Mine looks like this:
And the device ID that's relevant is Feel free to post the output and we can discuss! |
|
@victornoel Also note that #1544 was just merged into the media-driver, so you can try to build it and see if it helps! |
|
@Nexarian thanks, it looks like my device is in there, but I just realized I am using 22.4.4 of intel media driver so I suspect I should first try the latest version as you recommend :) For the record: |
|
It indeed works well for the Dell XPS 9315 with Alder Lake :) (something I could have know earlier if I had read a bit more the first comments at the top :P) |
|
@Brampie66 What about you? Were you ever able to resolve the reason you created this issue? |
|
Hey guys, I see you solved the issue. Can you help me with the same. Sorry, I new on Linux and didn't understand what you've exactly done to solve it.. Here is some info of my system: |
|
@ReshaSD we "simply" manually built the latest version (tag 22.6.4) of this repository (and I suspect you will also need latest version of https://github.com/intel/gmmlib/, 22.3.2) and installed it on our system. This is not very complex but each distribution will have its own way of doing it cleanly. Note that on Archlinux, finally the latest version 22.6.4 is now available in the package manager. You can check if packages for the latest version exist for your distribution. All of this is a bit out of scope of this discussion though, so I would recommend either waiting for the latest version be available in your distribution or investigate how to build the driver manually by yourself. See the README of the repository: https://github.com/intel/media-driver#building |
|
@victornoel I installed libva 2.16.0 and latest tags from gmmlib and media-driver. |
|
@ReshaSD, it’d be great if you reported the issue to your distribution (Ubuntu?), so they can backport the patches. |
|
@paulmenzel Yes, I am on Ubuntu 22.10. Thanks for suggestion. |
|
My Dell xps 9315 also works fine with |
|
@woodrow-shen, is the video engine used, when using mpv or another video player? Anyway, please create a separate issue, and you can check VAAPI things with Mozilla Firefox with the logging options below: |
|
@paulmenzel Thanks for reply, fix my words, I can confirm that decoding works on firefox, and meanwhile I can see the workload from intel_gpu_top as well. Great work. |
|
bumping this issue - Intel Alder Lake device Dell XPS 13 9315, vainfo returns Everything is up to date, so Dell didn't care about making the GPU of a 2k$ laptop work with the oem-shipped ubuntu in over one year? I'm asking this in order to figure out if this problem affects only us or if I should ask our purchase folks to ask dell for clarification. Anyway! Has the fix been included in some development build so we can get the GPU to work without the whole recompilation process? Thankyou! |
|
@selfdrivingmenee hello, I have the same laptop and as discussed in the comments above, this works as expected with the latest release version. Multiple people confirmed this, at least for the present issue. You may be hitting a problem of configuration or maybe not using the latest version. In any case, if you think this is a big, the best is to open a new issue with details of the versions of libva, intel-media-driver and gmmlib :) |
|
@victornoel Ciao Victor, first of all thank you since I managed to get it to work by manually building it following your advice from your previous post - I can confirm that the GPU is now working perfectly fine on my XPS13 9315! Seems then that the issue is not in the driver itself but somewhere in the "distribution chain" for the Dell-provided oem Ubuntu distro. I contacted them to inform of the issue and see if it's only on our side or it's affecting all users with same computer. |
|
@selfdrivingmenee hehe, yes, I looked quickly at their ubuntu oem version and was a bit scared by all the hacks they have in there to make things work. I really find it unacceptable that they sell a laptop saying it's linux-certified but in the end requires so much hacking around, either by ubuntu people or manually as in this issue… this is a bit sad :) I myself am on archlinux and still need to hack around to get the webcam working, all the rest (audio and video mainly) seems ok now. |
|
Hello guys, So if I understand correctly, it is a Dell-specific problem? |
Initial problem was that driver itself was missing GPU HW IDs for some of the newer ADL models, so it did not recognize the iGPU. After driver was updated, only problem people had was their Linux distributions shipping too old driver version. This apparently included Ubuntu version pre-installed on given DELL laptop. If you're using LTS version of Ubuntu or its derivative, you can install so called HWE (HW enabling) meta package that pulls in newer driver versions. That provides drivers from newer Ubuntu non-LTS releases, after they've been tested few months in those newer distro versions. (You can see the available HWE driver packages with Maybe MiNT has something similar? If that does help i.e. current HWE packages do not yet provide new enough drivers, you may need to wait until next Ubuntu release + few months, for HWE packages in LTS distro to be updated. PS. While one could always build & install drivers from sources, I would recommend that only for experienced Linux users, or somebody wanting to learn more about debugging. Things installed outside of the distro's own repositories have a tendency of creating problems much later, when other packages get updated (and you do not any more remember installing things from other sources). |
|
Thanks @eero-t to summarize the history and suggestions. I'm going to close this issue. Please feel free to re-open it if any concerns. |
Hi,
I'm trying to get vaapi to work on an intel i7 12700 (alder lake), the pc is running centos8
vainfo:
Found init function _vaDriverInit_1_14
/usr/lib64/dri/iHD_drv_video.so init failed
va_openDriver() returns 1
I've tried and checked the following:
Linux localhost.localdomain 5.17.0-1.el8.elrepo.x86_64 , I updated linux-firmware as well, this dmesg part looks ok to me
[ 1.516969] fb0: switching to i915 from simple
[ 1.517131] i915 0000:00:02.0: vgaarb: deactivate vga console
[ 1.517903] i915 0000:00:02.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=io+mem:owns=io+mem
[ 1.518405] i915 0000:00:02.0: [drm] Finished loading DMC firmware i915/adls_dmc_ver2_01.bin (v2.1)
[ 1.534970] i915 0000:00:02.0: [drm] GuC firmware i915/tgl_guc_62.0.0.bin version 62.0 submission:disabled
[ 1.534983] i915 0000:00:02.0: [drm] GuC SLPC: disabled
[ 1.534986] i915 0000:00:02.0: [drm] HuC firmware i915/tgl_huc_7.9.3.bin version 7.9 authenticated:yes
[ 1.539418] [drm] Initialized i915 1.6.0 20201103 for 0000:00:02.0 on minor 0
[ 1.554958] fbcon: i915drmfb (fb0) is primary device
[ 1.609166] i915 0000:00:02.0: [drm] fb0: i915drmfb frame buffer device
[ 3.046435] snd_hda_intel 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
media_driver is 22.2.2 , libva is 2.14
xorg log:
[ 7.085] (==) modeset(0): Backing store enabled
[ 7.085] (==) modeset(0): Silken mouse enabled
[ 7.215] (II) modeset(0): Initializing kms color map for depth 24, 8 bpc.
[ 7.215] (==) modeset(0): DPMS enabled
[ 7.215] (II) modeset(0): [DRI2] Setup complete
[ 7.215] (II) modeset(0): [DRI2] DRI driver: iris
[ 7.215] (II) modeset(0): [DRI2] VDPAU driver: va_gl
........
[ 7.217] (II) Initializing extension DRI3
I tried the same combo (libva,media-driver,gmmlib,kernel) on a much older J5005 and this works ok.
I was under the impression vaapi should work for this cpu on that kernel, or am I mistaken ?
How can I check to see why vainfo is failing ?
thx
Bram
The text was updated successfully, but these errors were encountered: