Bug 1405474 - No sound on the Dell XPS 13 9343 (early 2015 version)
Summary: No sound on the Dell XPS 13 9343 (early 2015 version)
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Fedora
Classification: Fedora
Component: kernel
Version: 27
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: ---
Assignee: Kernel Maintainer List
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2016-12-16 15:25 UTC by Niklas Wenzel
Modified: 2019-01-09 12:54 UTC (History)
14 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-02-21 15:54:09 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
dmesg (61.54 KB, text/plain)
2016-12-16 15:25 UTC, Niklas Wenzel
no flags Details

Description Niklas Wenzel 2016-12-16 15:25:27 UTC
Created attachment 1232561 [details]
dmesg

Description of problem:

Ever since the release of kernel version 4.4 on Fedora 23, I and a few other users haven't had sound output on the Dell XPS 13 9343 (early 2015 version), that ships with Linux (Ubuntu) preinstalled.

Here is the old bug for Fedora 23: https://bugzilla.redhat.com/show_bug.cgi?id=1313434

The same issue still persists on Fedora 25 with Kernel version 4.8.13.

As of version 4.4, the Fedora kernel has no longer been compiled with the option CONFIG_ACPI_REV_OVERRIDE_POSSIBLE=y and since then the sound has been broken on the above machine because the hardware now runs in I2S mode instead of HDA mode. The implementation of the former seems to be faulty for the laptop.

The current workaround is to use kernels compiled with the above-mentioned flag that force the hardware to run in HDA mode, but to be honest, that's not fun, as we have to recompile each and every kernel.

Herald van der Breggen hosts his builds here: https://xps13.landweg.nl/ (Reboot twice after installing them.)


How reproducible:

Always


Steps to Reproduce:

1) Install Fedora 25 with kernel version >=4.4 on said machine.
2) Do anything that plays sound.


Actual results:

No sound is heard.


Expected results:

Sound can be heard.


Additional info:

Gnome settings does not see any sound output devices.

dmesg output attached.

Comment 1 Michael 2016-12-17 13:59:56 UTC
Product Name: XPS 13 9343
4.8.6-300.fc25.x86_64
broadwell-rt286

Loads fine, speakers are fine
Headset has constant clicking noise
External mic is fine
Inline mic is always in "unplugged" and nothing is recorder

It is not exactly "no sound" but very close.

Back to HDA mode.

Comment 2 Niklas Wenzel 2016-12-19 12:49:58 UTC
@Michael: That seems like a different issue to me. I would suggest that you file another bug report for that. It's just the workaround (reintroducing a hack to use HDA mode) that is the same.

Comment 3 Michael 2016-12-19 13:21:40 UTC
Sorry Niklas, I am pretty sure we can stop creating useless bug reports all over the Internet. This issue has an unique upstream reported (with zero meaningful feedback) dating back to 4.4 kernel and still persists in 4.9. The problem is somewhere in the "codecs" configs that don't map things correctly. The actual behaviour (see the fc23 report) may differ from user to user as it seems to be a matter of "playing" with mixer and pulse audio.

And the only workaround that ALWAYS works is go back to ugly HDA.

Comment 4 Niklas Wenzel 2016-12-21 19:54:27 UTC
@Michael: Wow, that's interesting!

Do you have any further information about the incorrect mapping in the "codecs" configs? Any link?

Comment 5 Ferry 2017-01-03 12:26:49 UTC
Hi,

you could try booting with the additional parameters "noapic acpi=off". Not sure if it would help in this case though.

We're running into issues with several pieces of (mostly Dell) hardware. For example, networking won't work on a Dell Latitude 3350.

https://patchwork.kernel.org/patch/9423907/

The above kernel patch is not in Fedora (not yet anyway, it's not in the vanilla kernels either at this moment).

On Ubuntu we can get around it with acpi_rev_override and we're wondering as well why this option has been removed from the Fedora kernels. I would help us get around several hardware issues quite easily for now.

Frequently the "noapic acpi=off" options will work in this (network on the 3350) case but it disables much more.

Basically we'd really like Fedora to add CONFIG_ACPI_REV_OVERRIDE_POSSIBLE back to the kernel.

Comment 6 Justin M. Forbes 2017-04-11 14:38:17 UTC
*********** MASS BUG UPDATE **************

We apologize for the inconvenience.  There is a large number of bugs to go through and several of them have gone stale.  Due to this, we are doing a mass bug update across all of the Fedora 25 kernel bugs.

Fedora 25 has now been rebased to 4.10.9-200.fc25.  Please test this kernel update (or newer) and let us know if you issue has been resolved or if it is still present with the newer kernel.

If you have moved on to Fedora 26, and are still experiencing this issue, please change the version to Fedora 26.

If you experience different issues, please open a new bug report for those.

Comment 7 Herald van der Breggen 2017-04-15 09:50:05 UTC
Thanks for the info. Great to hear a massive update is done which might fix the problem.

Unfortunately the problem is not solved. However, there is an improvement. With the new kernel it is possible to play sound via HDMI output, which was not possible before. Also the delays/hangs are gone when a sound was about to play.

But still, the local speakers are not recognized, neither is the microphone.

So I started again a kernel rebuild job with the CONFIG_ACPI_REV_OVERRIDE_POSSIBLE=y option. Hoping that will still work.

Thanks,
Herald

Comment 8 Herald van der Breggen 2017-04-15 12:45:15 UTC
The odd thing is that once I booted the 4.10.9 kernel, the new behavior looks permanent, even when I boot (multiple times) the kernels that used to work with local speakers/microphone. Do I need to clean up something up to get the old situation back?

Comment 9 Sebastian Plamauer 2017-05-26 14:42:52 UTC
It's not just that the audio crashes and stops working, it kills the whole OS. Audio works for a time, but then crashes and the display freezes. All you can do is switch to tty and look at dmesg spamming haswell-pcm messages.

Comment 10 Kyrre Ness Sjøbæk 2017-05-29 21:56:35 UTC
Confirmed on F25; in many cases it really freezes the whole machine for up to 30 seconds. The device sometimes works just after system boot, then crashes. On F21 it worked flawlessly.

I'm trying to build a kernel following https://fedoraproject.org/wiki/Building_a_custom_kernel setting CONFIG_ACPI_REV_OVERRIDE_POSSIBLE=y in kernel-x86_64.config ; I'll hopefully report back as soon as I know the result...

Comment 11 Herald van der Breggen 2017-05-30 08:04:42 UTC
Yes, CONFIG_ACPI_REV_OVERRIDE_POSSIBLE=y still works, that is what I am doing for every kernel that is released for fedora in the last 6 months. 

See https://xps13.landweg.nl/ for README.txt how to do it and for kernels that are rebuild with this config setting.

Comment 12 Kyrre Ness Sjøbæk 2017-06-04 14:02:18 UTC
Herald,
just saw your reply, and confirm that everything works after recompiling with this flag like you did.

Since you have already published your RPM's I won't bother.

Cheers,
Kyrre

Comment 13 Herald van der Breggen 2017-07-20 06:30:07 UTC
Since the problem continues in Fedora 26, I would like to update the fedora version flag of this bug report to 26. But I have no clue how to do that. Can somebody do that please?

Regarding https://xps13.landweg.nl/ , from now on I'll rebuild the kernels for fc26.

Comment 14 Niklas Wenzel 2017-07-20 09:52:16 UTC
@Herald: Thanks for confirming it on fc26. I updated the bug's fedora version. :)

Comment 15 Michael 2017-11-16 22:02:19 UTC
Unsurprisingly fc27 - still the same.

Comment 16 Niklas Wenzel 2017-11-16 22:14:35 UTC
@Michaeli: Thanks for reporting. Updated the version in the bug report to 27.

Comment 17 Sebastian Plamauer 2017-11-30 17:17:25 UTC
This problem has not gotten better but worse. Due to whatever the audio bug is, the whole system starts to hang. Even on the livedisk. Which means I can not install Fedora 27 on my new SSD - the system freezes before I can finish the install.

Comment 18 Laura Abbott 2018-02-20 19:54:27 UTC
We apologize for the inconvenience.  There is a large number of bugs to go through and several of them have gone stale.  As kernel maintainers, we try to keep up with bugzilla but due the rate at which the upstream kernel project moves, bugs may be fixed without any indication to us. Due to this, we are doing a mass bug update across all of the Fedora 27 kernel bugs.
 
Fedora 27 has now been rebased to 4.15.3-300.f27.  Please test this kernel update (or newer) and let us know if you issue has been resolved or if it is still present with the newer kernel.
 
If you experience different issues, please open a new bug report for those.

Comment 19 Herald van der Breggen 2018-02-21 14:30:45 UTC
Hi Laura,

Thanks for the heads up!

I had already installed a rebuilded version of 4.15.3, but because of your question I switched back to the kernel from the fedora repository.

After two times poweroff/poweron the sound still worked. I could not believe it.

I powered off a third time and booted again. I double checked the kernel version ("uname -a" confirmed 4.15.3-300) and build host ("rpm -qi kernel-4.15.3" yes: bkernel02.phx2.fedoraproject.org).

I still can not believe it, but it looks like the problem is fixed in 4.15.3!

Thanks for asking this and thanks to everyone involved to solve this.

Comment 20 Justin M. Forbes 2018-02-21 15:54:09 UTC
Glad to hear this, feel free to re-open if it is not fixed for everyone.

Comment 21 Niklas Wenzel 2018-02-21 18:09:47 UTC
Thanks a lot for your answer, Herald! :)

I guess I'll finally update to fc27 on my xps13.

Comment 22 Niklas Wenzel 2018-02-21 19:02:10 UTC
I can confirm that upgrading to kernel 4.15.3 also solves this issue on Fedora 26.


Note You need to log in before you can comment on or make changes to this bug.