[BUG] [ICL] SOF codec#0 probe fails on Acer Switch SF314-57G

Describe the bug
The user cannot load signed firmware v1.3 on Acer Swift SF314-57G:

Manufacturer:      Acer
Product Name:      Swift SF314-57G
Product Version:   V1.07
Firmware Version:  V1.07
Board Vendor:      IL
Board Name:        Floris_IL

https://bugzilla.redhat.com/show_bug.cgi?id=1772498#c16

Discussion:

What is the expected behaviour, if the DSP has own (vendor) authentication key? Basically, how we can determine the right firmware for the given DSP in the generic Linux distribution? We need detect this automatically somehow. Can DSP give any hints to the driver before the fw upload?

@perexg

@perexg thanks for bringing our attention on this

The executive summary is:

All platforms use the firmware signed with the production Intel key.
The exceptions are: Up2 and GLK+ chromebooks which are signed with the community key.
I am not aware of OEMs signing the base firmware, only extensions which are currently not supported by SOF.

we cannot query the key used, that's unfortunate but that's the reality. When the firmware is provided through linux-firmware, I'll add a quirk for these exceptions to point to the firmware signed with the community key.

I have been able to use the signed firmware on Acer devices since 1.3

In this case, I would think it's a kernel issue (either a missing patch or a bad commit from Intel).

can you share the git tree used?

Yep, its 5.4-rc8 with extra backports from 5.5: https://github.com/perexg/linux/tree/perex/fedora-test

Could you try removing all S0ix patches, e.g. starting with c13c962

That series seems to introduce problems and there is on-going work to fix it.

Edit: actually trying plain vanilla 5.4-rc8 would be even better, just to make sure there's a reference to bisect from.

Per this kernel log, I don't think that fw is active, it appears like that the upload fails. The PM is not active (I think) in this case.

[    5.793409] sof-audio-pci 0000:00:1f.3: DSP detected with PCI class/subclass/prog-if info 0x040100
[    5.793702] sof-audio-pci 0000:00:1f.3: Digital mics found on Skylake+ platform, using SOF driver
[    5.793718] sof-audio-pci 0000:00:1f.3: enabling device (0000 -> 0002)
[    5.794122] sof-audio-pci 0000:00:1f.3: warning: No matching ASoC machine driver found
[    5.794148] sof-audio-pci 0000:00:1f.3: DSP detected with PCI class/subclass/prog-if 0x040100
[    5.794416] sof-audio-pci 0000:00:1f.3: use msi interrupt mode
[    5.794775] sof-audio-pci 0000:00:1f.3: bound 0000:00:02.0 (ops i915_audio_component_bind_ops [i915])
[    5.799105] sof-audio-pci 0000:00:1f.3: hda codecs found, mask 5
[    5.799109] sof-audio-pci 0000:00:1f.3: using HDA machine driver skl_hda_dsp_generic now
[    6.804322] sof-audio-pci 0000:00:1f.3: error: codec thesofproject/sof#2 probe error, ret: -5
[    6.808731] sof-audio-pci 0000:00:1f.3: unexpected ipc interrupt raised!
[    6.808735] sof-audio-pci 0000:00:1f.3: error: no reply expected, received 0x0
[    9.877648] sof-audio-pci 0000:00:1f.3: error: cl_copy_fw: timeout HDA_DSP_SRAM_REG_ROM_STATUS read
[    9.878144] sof-audio-pci 0000:00:1f.3: error: load fw failed ret: -110
[    9.878161] sof-audio-pci 0000:00:1f.3: error: status = 0x0000002c panic = 0x00000000
[    9.878166] sof-audio-pci 0000:00:1f.3: error: failed to reset DSP
[    9.878167] sof-audio-pci 0000:00:1f.3: error: failed to boot DSP firmware -110
[    9.878167] sof-audio-pci 0000:00:1f.3: error: sof_probe_work failed err: -110

The first error appears related to the HDaudio codec (or hdmi), can you check what happens if you select the FORCE_NOCODEC mode and use a nocodec topology? That would remove dependencies on the codecs and check if the firmware load works or not.
sof-cnl-nocodec.tplg.gz

@perexg

@perexg another point to verify is the PCI ID of the device. We recently split CFL, CNL and CML firmwares instead of using sof-cnl.ri across the board due to additional firmware manifest/signature differences. If you are using the CNL file on CML, it's likely not going to work.

IIRC correctly CFL and CNL are still identical, but CML has a different manifest. @jajanusz please correct me if I am wrong.

The complete alsa-info output is in the Fedora bugzilla (URL provided in the bug description):

00:1f.3 0401: 8086:34c8 (rev 30)
	Subsystem: 1025:1407

@perexg

@perexg I've tagged v1.4.1, @lbetlej can you sign.

@perexg

@perexg @lgirdwood PCI ID 34c8 is IceLake so that's a new one.

@perexg

@perexg can you please with the file attached (to be copied as /lib/firmware/intel/sof/sof-icl.ri)
I'd like to verify this is solved with the upcoming release, and if not to figure out the issue.

sof-icl-signed.ri.gz

I am just a "proxy" for this from the Fedora's bugzilla. I redirected the user here, so hopefully, he will test this.

new signed binaries now on v1.4.1 release page.

Original reporter here. Unfortunately new firmware makes no difference at all

@omg The updated binaries (CNL and ICL) for SOF 1.4.1 release are available for download. Could you test them and report if issue can be closed ? Thank you.

Sadly, no luck -- I tried the Hdmi.conf from upstream, reverted my default.pa change, removed my ~/.config/pulse, and rebooted to be safe, but I get dummy output and no input.

Pulseaudio debug logging is available here: https://gist.github.com/samkline/2aff0c71db990e09ad4bdf083666fc9b (lines 115-118 seem potentially relevant)

BTW, this machine does have an Intel GPU:

$ sudo lshw -C display
  *-display                 
       description: VGA compatible controller
       product: Iris Plus Graphics G7
       vendor: Intel Corporation
       physical id: 2
       bus info: pci@0000:00:02.0
       logical name: /dev/fb0
       version: 07
       width: 64 bits
       clock: 33MHz
       capabilities: pciexpress msi pm vga_controller bus_master cap_list rom fb
       configuration: depth=32 driver=i915 latency=0 mode=1920x1080 visual=truecolor xres=1920 yres=1080
       resources: iomemory:600-5ff iomemory:400-3ff irq:163 memory:601c000000-601cffffff memory:4000000000-400fffffff ioport:3000(size=64) memory:c0000-dffff

Oh, good news: I just installed pulseaudio-git (13.99.1-31-g1fe37), and it now works with the upstream Hdmi.conf without any changes to default.pa.

@RiderALT

Patch to fix the original bug has been merged upstream now, so I will proceed and close the bug:
https://mailman.alsa-project.org/pipermail/alsa-devel/2020-May/168417.html

Thank you @ltune , @RiderALT , @perexg and @samkline for reporting and helping with the debug. If there are new issues, please open a new bug and let's track any new issues separately.

That's good news. It may be good to add here which kernel version will contain the patch, once it is released, so I can load the corresponding Ubuntu kernel and have the internal microphone working🙂

@adam-fu

@adam-fu wrote:

That's good news. It may be good to add here which kernel version will contain the patch, once it is released, so I can load the corresponding Ubuntu kernel and have the internal microphone working🙂

It's currently queued for 5.8, so will be in 5.8.0. It may get backported to older stable kernels, and/or Ubuntu kernel package, so you probably don't have to wait that long.

I already apply this patch (commit 89d73cc) for my local build of the kernel 5.7.0 and it works great on my Acer SF314-57 without any additional setting (like kernel module options or pulseaudio settings).

The only problem I discovered with 5.7.0 on this machine is 3 HDMI audio outputs in pulseaudio but one HDMI and one thunderbolt output in hardware.i attached Alsa-info output to save a history.
alsa-info.txt.gz

@RiderALT

@RiderALT wrote:

I already apply this patch (commit 89d73cc) for my local build of the kernel 5.7.0 and it works great on my Acer SF314-57 without any additional setting (like kernel module options or pulseaudio

That's good to hear, thanks for testing!

The only problem I discovered with 5.7.0 on this machine is 3 HDMI audio outputs in pulseaudio but one HDMI and one thunderbolt output in hardware.i attached Alsa-info output to save a history.

This is to be expected. This hardware supports up to 3 concurrent HDMI/DP audio streams. On most systems like this, you can connect upto 3 monitors to the USB-C/thunderbolt port (using DP-MST), plus one to the HDMI connector. Pulseaudio should show which of the three ALSA HDMI PCMs has a connected monitor (in your alsa-info it shows no monitors w/ audio capability connected).

ASoC: SOF: Intel: hda: fix generic hda codec support

UPSTREAM: ASoC: SOF: Intel: hda: fix generic hda codec support

ASoC: SOF: Intel: hda: fix generic hda codec support

UPSTREAM: ASoC: SOF: Intel: hda: fix generic hda codec support

ASoC: SOF: Intel: hda: fix generic hda codec support

Navigation Menu

Repository navigation

[BUG] [ICL] SOF codec#0 probe fails on Acer Switch SF314-57G #1807

86 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

[BUG] [ICL] SOF codec#0 probe fails on Acer Switch SF314-57G #1807

Description

Activity

plbossart commented on Nov 22, 2019

perexg commented on Nov 22, 2019

plbossart commented on Nov 22, 2019

perexg commented on Nov 22, 2019

plbossart commented on Nov 22, 2019

plbossart commented on Nov 23, 2019

perexg commented on Nov 23, 2019

lgirdwood commented on Nov 23, 2019

plbossart commented on Nov 25, 2019

plbossart commented on Nov 25, 2019

perexg commented on Nov 25, 2019

lgirdwood commented on Nov 25, 2019

containmentbreach commented on Nov 25, 2019

lbetlej commented on Dec 5, 2019

86 remaining items

samkline commented on Jun 2, 2020

samkline commented on Jun 2, 2020

kv2019i commented on Jun 2, 2020

adam-fu commented on Jun 3, 2020

kv2019i commented on Jun 3, 2020

RiderALT commented on Jun 3, 2020

kv2019i commented on Jun 3, 2020

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions