[BUG] Intel Corporation Tiger Lake error: hda_dsp_core_reset_enter: timeout on HDA_DSP_REG_ADSPCS read #9339
Labels
bug
Something isn't working as expected
Comments
|
@real-t0mg can you add this file |
|
Attaching dmesg output as requested. Let me know if this works for debugging on your end: interesting bit here: |
|
if you don't have dynamic debug enabled, chances are you are already in a rather specific configuration out of the beaten path... Please make sure you enable the ME engine and/or try with another distro with a USB live image. |
|
@plbossart ME engine is enabled, and in current level per the latest lenovo available, it was even reinstalled as some old bug report recommended re-installation of the ME engine. Suggestions other than "options snd_intel_dspcfg dsp_driver=1" are welcome. |
|
@mrhpearson FYI, not sure how to deal with this issue. |
|
for the record, as this will hurt other ThinkPad users, I could only make it work with release version v2023.12.1, no other combination of sof-tgl.ri and sof-tplg/sof-hda-generic-2ch.tplg works for me. If you are in pain with a ThinkPad T14s Gen 2i , use this release (https://github.com/thesofproject/sof-bin/releases/tag/v2023.12.1), Next time, buy AMD. dmesg for the bug record for the record: |
|
I created internal ticket LO-3221 for tracking - but a few questions @real-t0mg - Can you let me know what distro you're running please? Also it would be good to know the BIOS and EC versions (likely not important here - but having it handy in case the FW team ask is good) Would you be able to try with Fedora 40 Live USB - that should have 2024.03 (I don't think anybody has 2024.06 packaged yet - I know I'm still fighting with it for the Debian packaging). Unfortunately I don't have a T14s G2 intel system on hand - but once we have more details we'll see if we can reproduce in the lab. |
|
@mrhpearson, sure: Slackware 15.0 (http://www.slackware.com/). Thanks for tracking the issue. BIOS: EC: Regarding the live cd boot with Fedora, I may later today and report back. Full time running a distro with systemd is not a viable alternative for me, but I can download and compile my own kernels to test if that helps. I prefer running long-term support kernel alternatives if at all possible help to debug this. I'm reluctant to attach a full dmidecode output as it has the laptop's serial number that I registered to hava a valid warranty, but I can email it if you need it. |
|
The topology on 2024.03 does not work on my hardware: dmesg output: interesting bit: topology fails: |
|
ok so the good news is that there's at least one version that works, thanks for trying older versions @real-t0mg The last try shows that the firmware version 2024.03 boots as expected, if that is confirmed then only the last version has a problem. I am not sure what changed between the firmware v2023.12.1/v2024.3 and v2024.06, in theory the keys are identical. The only thing that comes to my mind is compatibility between tools used for the signature by Intel and tools used by Lenovo for the authentication in the ME, or one of the 'Feature' flags that changed on the target. We've seen both before. Wrt to the topology problem, IIRC this is a kernel bug that was corrected. The additional trailing characters in the widget names, e.g. "sink HDA4.OUT\xd7\x07" remind me of something, I just can't find the actual fix. @ujfalusi, @bardliao @ranj063 does this ring a bell?. |
|
the bad news is that reverting to the version that worked (sof-bin-2023.12.1.tar.gz), no longer works and now it fails with a different error. Let me know what additional info I may provide, now no version works for me :-( . dmesg log: interesting bit: fail: |
|
without changing the firmware nor the topology files, the error messages regarding the topology feel random (i.e. different HDA#.OUT accross reboots): dmesg log: sample: |
|
Now I remember, it was a fairly recent thread in https://lore.kernel.org/linux-sound/171812236450.201359.3019210915105428447.b4-ty@kernel.org/T/#m8c4bd5abf453960fde6f826c4b7f84881da63e9d Maybe you have a bad patch that was added to your kernel, best to try with 6.10 if you can to rule out a backport issue. |
|
@plbossart you were right about kernel 6.10. Running out of the box with v2024.06 (https://github.com/thesofproject/sof-bin/releases/tag/v2024.06). Running as expected on non-debug 6.10.2 kernel. With the added benefit I can run the Xe drivers too... dmesg output for completeness:
code: |
|
well that's good news @real-t0mg, at least you are in business with the latest firmware and kernel. I am still surprised that this 6.6 kernel would prevent the latest firmware from booting. The TigerLake kernel stuff is a couple of years old, we haven't really touched it, same for the firmware which comes from a 2.2 stable branch. In general we don't expect users to update both kernel and firmware at the same time. Releases should be backwards-compatible. @ujfalusi any idea on what could have caused this boot problem? |
|
@plbossart, fwiw on my TGL laptop 2024.06 release works fine: The error looks strange: The ROM fails with Invalid feature mask (0x14 / 20) status code. The feature_mask is identical between 2024.03 and 2024.06 (0xffff, according to These are form the sof-bin release tarballs. |
|
@plbossart, I think I know why the firmware was not booting: It is the IPC4 firmware file loaded as ipc3 image. In this case the header parsing is skipped and the downloaded firmware can be incorrect for booting. @real-t0mg, can you re-check this with the 2024.06 release: and confirm that you have the Which distribution you are using and have you installed the sof-bin (2024.06) manually or via distro package? Thanks |
|
Thanks for all the input. @ujfalusi I tried both, IPC4 & IPC3 firmwares from several binary releases. After the latest round of longterm support kernel upgrades soured from kernel.org (https://lwn.net/Articles/983652/), the firmware boot fail started happening, no matter what binaries I sourced put in my /lib/firmware path. It was hit or miss, apparently random.
Right now I'm sourcing sof-bin-2024.06.tar.gz, as mentioned here #9339 (comment). I understand that for my setup I only need two files to make it work sof/intel-signed/sof-tgl.ri and the topology intel/sof-tplg/sof-hda-generic-2ch.tplg. I use Slackware and, even though I can build my own package, for this particular case I decompress the bin distribution from SOF and get the two files I need out of the binary distribution tar (https://github.com/thesofproject/sof-bin/releases/download/v2024.06/sof-bin-2024.06.tar.gz) For your refrence, from my setup: |
|
@real-t0mg, you cannot just use IPC4 firmware when the kernel is expecting IPC3 setup [1]. For the boot error I can only say that the ipc4 firmware was placed where the ipc3 version should be and that caused the boot failure. The IPC3 firmware has not changed for a while: It is easy to mix them when doing manual installs (done that myself several times). [1] For IPC4 you need different firmware and topology file. If the sof-bin is properly installed then |
|
@ujfalusi thanks! Yes, I figured out that part yesterday re-reading https://thesofproject.github.io/latest/getting_started/intel_debug/introduction.html#user-space-and-filesystem-requirements My chip/codec is a Realtek ALC257. What @plbossart says here #9339 (comment) applies to my issue after having placed the correct IPC3 firmware and topology to my setup. My issue started after upgrading from 6.1.101 to 6.1.102. In the kernel. In the changelog I see these were changed: I the proceeded to upgrade to 6.6.43, which you see in my comments above, which in turn logs these changes: If I reboot into 6.6.43, I get the error I logged in the issue with the topology file. |
It's "easy" only if you move and rename individual files - but why would you do that? Some firmware developers might do that sometimes as a costly "shortcut" but there's really no reason why any end user would. Same issue with overwriting files from any .deb or .rpm package; not just sof-bin.tgz. |
|
@real-t0mg, I think the subject of the issue has been rootcaused and fixed and now we are discussing unrelated issue(s) to iron out. |
|
closing bug, the recommended fix is to run a current kernel. Thank you! |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
On ThinkPad T14s Gen 2i, kernel 6.6.43, Tiger Lake-LP Smart Sound Technology Audio Controller firmware loading fails with the following message:
SOF version: https://github.com/thesofproject/sof-bin/releases/tag/v2024.06
The text was updated successfully, but these errors were encountered: