sof-icl-es8336-ssp0.tplg: HDMI PCM/pipeline ids (5/6/7) mismatch iDisp/HDA token IDs (3/4/5) — causes sof IPC hw_params errors

[marmabiswas123]

Summary
When booting with intel/sof/sof-icl.ri + topology intel/sof-tplg/sof-icl-es8336-ssp0.tplg, the SOF driver reports IPC errors for HDMI playback streams and ASoC hw_params failures (ipc tx error ... -5, sof_ipc3_pcm_hw_params), caused by a mismatch between HDMI PCM IDs / pipeline indices and the iDisp/HDA IDs encoded in the topology. This causes HDMI PCM hw_params IPC failures (HDMI streams fail to start). See dmesg excerpts below and attached decompiled topology file.

Impact

HDMI PCM (iDisp) playback streams fail to set hw_params → audio over HDMI unavailable.

IPC errors reported in dmesg: ipc tx error for 0x60010000 ... -5 and sof_ipc3_pcm_hw_params: pcm6 (HDMI 2) ... STREAM_PCM_PARAMS ipc failed.

Affects machines using this topology with ES8336 codec + ICL SOF firmware.

Environment / versions (from dmesg)

Linux dmesg excerpt (relevant lines):

[    0.068320] software IO TLB: area num 4.
[    0.679839] pps_core: Software ver. 5.3.6 - Copyright 2005-2007 Rodolfo Giometti <giometti@linux.it>
[    0.784715] PCI-DMA: Using software bounce buffering for IO (SWIOTLB)
[    0.784718] software IO TLB: mapped [mem 0x000000000b000000-0x000000000f000000] (64MB)
[    4.090941] sof-audio-pci-intel-icl 0000:00:1f.3: enabling device (0000 -> 0002)
[    4.091097] sof-audio-pci-intel-icl 0000:00:1f.3: DSP detected with PCI class/subclass/prog-if 0x040380
[    4.091161] sof-audio-pci-intel-icl 0000:00:1f.3: bound 0000:00:02.0 (ops intel_audio_component_bind_ops [i915])
[    4.098595] sof-audio-pci-intel-icl 0000:00:1f.3: use msi interrupt mode
[    4.113422] sof-audio-pci-intel-icl 0000:00:1f.3: hda codecs found, mask 4
[    4.113427] sof-audio-pci-intel-icl 0000:00:1f.3: BT link detected in NHLT tables: 0x0
[    4.113429] sof-audio-pci-intel-icl 0000:00:1f.3: DMICs detected in NHLT tables: 0
[    4.113431] sof-audio-pci-intel-icl 0000:00:1f.3: NHLT device I2S(4) detected, ssp_mask 0x1
[    4.113433] sof-audio-pci-intel-icl 0000:00:1f.3: Overriding topology with MCLK mask 0x1 from NHLT
[    4.120484] sof-audio-pci-intel-icl 0000:00:1f.3: Firmware paths/files for ipc type 0:
[    4.120490] sof-audio-pci-intel-icl 0000:00:1f.3:  Firmware file:     intel/sof/sof-icl.ri
[    4.120493] sof-audio-pci-intel-icl 0000:00:1f.3:  Topology file:     intel/sof-tplg/sof-icl-es8336-ssp0.tplg
[    4.121155] sof-audio-pci-intel-icl 0000:00:1f.3: Firmware info: version 2:2:0-57864
[    4.121160] sof-audio-pci-intel-icl 0000:00:1f.3: Firmware: ABI 3:22:1 Kernel ABI 3:23:1
[    4.210820] sof-audio-pci-intel-icl 0000:00:1f.3: Firmware info: version 2:2:0-57864
[    4.210827] sof-audio-pci-intel-icl 0000:00:1f.3: Firmware: ABI 3:22:1 Kernel ABI 3:23:1
[    4.220599] sof-essx8336 sof-essx8336: quirk mask 0x0
[    4.220604] sof-essx8336 sof-essx8336: quirk SSP0
[    4.221463] sof-audio-pci-intel-icl 0000:00:1f.3: Topology: ABI 3:22:1 Kernel ABI 3:23:1
[    4.221650] sof-essx8336 sof-essx8336: ASoC: Parent card not yet available, widget card binding deferred
[    4.551088] sof-essx8336 sof-essx8336: hda_dsp_hdmi_build_controls: no PCM in topology for HDMI converter 3
[    4.564017] input: sof-essx8336 Headset as /devices/pci0000:00/0000:00:1f.3/sof-essx8336/sound/card0/input12
[    4.564073] input: sof-essx8336 HDMI/DP,pcm=5 as /devices/pci0000:00/0000:00:1f.3/sof-essx8336/sound/card0/input13
[    4.564116] input: sof-essx8336 HDMI/DP,pcm=6 as /devices/pci0000:00/0000:00:1f.3/sof-essx8336/sound/card0/input14
[    4.564168] input: sof-essx8336 HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:1f.3/sof-essx8336/sound/card0/input15
[    8.620950] sof-audio-pci-intel-icl 0000:00:1f.3: ipc tx error for 0x60010000 (msg/reply size: 108/20): -5
[    8.620960] sof-audio-pci-intel-icl 0000:00:1f.3: sof_ipc3_pcm_hw_params: pcm6 (HDMI 2), dir 0: STREAM_PCM_PARAMS ipc failed for stream_tag 1
[    8.620964] sof-audio-pci-intel-icl 0000:00:1f.3: ASoC error (-5): at snd_soc_pcm_component_hw_params() on 0000:00:1f.3

Reproduction steps

Boot the machine with the included topology (intel/sof-tplg/sof-icl-es8336-ssp0.tplg) and SOF firmware intel/sof/sof-icl.ri.

Attempt to open an HDMI playback device (e.g., play audio to HDMI or run aplay -D hw:0,6 ... for PCM6).

Observe dmesg errors shown above and that HDMI audio is not available.

Raw evidence / logs

Key topology pieces (excerpts from the decompiled file):

sof-icl-es8336-ssp0-tplg.conf.txt

SectionPCM shows:

SectionPCM {
	'ES8336' {
		dai.'ES8336 0'.id 0
		pcm {
			playback.capabilities 'Passthrough Playback 0'
			capture.capabilities 'Low Latency Capture 0'
		}
	}
	'HDMI 1' {
		id 5
		dai.'HDMI 1 5'.id 5
		pcm.playback.capabilities 'Passthrough Playback 5'
	}
	'HDMI 2' {
		id 6
		dai.'HDMI 2 6'.id 6
		pcm.playback.capabilities 'Passthrough Playback 6'
	}
	'HDMI 3' {
		id 7
		dai.'HDMI 3 7'.id 7
		pcm.playback.capabilities 'Passthrough Playback 7'
	}
}

SectionBE shows:

SectionBE {
	SSP0-Codec.data 'SSP0-Codec:tuple0'
	iDisp1 {
		id 3
		default_hw_conf_id 3
		data 'iDisp1:tuple0'
	}
	iDisp2 {
		id 4
		default_hw_conf_id 4
		data 'iDisp2:tuple0'
	}
	iDisp3 {
		id 5
		default_hw_conf_id 5
		data 'iDisp3:tuple0'
	}
}

Vendor tuples (HDA3/HDA4/HDA5) encode token155 values:

HDA3.OUT:tuple0 → token155 3

HDA4.OUT:tuple0 → token155 4

HDA5.OUT:tuple0 → token155 5

Observation: PCM indexes for HDMI are 5,6,7 while iDisp / HDA token indices are 3,4,5. Pipelines are declared with indices 5/6/7 (for PCM5/6/7) but HDA tuple token155 values refer to 3/4/5. This mismatch between the pipeline/PCM IDs and the HDA/iDisp IDs leads to incorrect mapping when SOF builds the IPC objects for HDMI, causing STREAM_PCM_PARAMS IPC to be sent with wrong IDs/params and the firmware to return error (-5).

Root cause (analysis)

The topology assigns PCM / pipeline IDs 5, 6, 7 to HDMI PCMs (SectionPCM and pipeline sections), but the iDisp/HDA endpoints and their tuples (SectionBE VendorTuples token155) are encoded as 3, 4, 5. The mismatch results in the driver building IPC messages referencing pipelines/devices that do not match the HDA endpoint IDs encoded in the topology, and the firmware rejects the hw_params IPC (error -5). In short: HDA token/pipeline indices are inconsistent inside the topology (PCM/pipeline indices ≠ HDA/iDisp token indices).

Proposed fix (concrete)

There are two consistent fixes — pick one and apply to the topology:

Option A (recommended): Update the iDisp entries and the token155 values in the HDA tuples to match PCM/pipeline IDs (5/6/7). This is minimally invasive because the PCM/pipeline blocks already use 5/6/7.

Option B
(alternate): Renumber the PCM/pipeline entries to 3/4/5 to match current iDisp/HDA token indices. This would require renaming the PCM playback capability names (PCM5P/PCM6P/PCM7P etc.) and pipeline indices accordingly.

[marmabiswas123]

Requesting issue labels to help categorize this report:

• bug
• tplg
• hdmi
• intel-icl
• ES8336

This involves topology pipeline/HDA ID mismatch for HDMI PCM 5/6/7 vs iDisp 3/4/5. Thanks!
@marc-hb @kv2019i @lrgirdwo

[kv2019i]

@marmabiswas123 Thank you very much for the detailed bug report. Let me move this to the main SOF issue tracker, so it gets more attention. This sof-bin issue tracker is only meant for issues regarding how the binaries are pacakged to Linux distrobutions.

[marc-hb]

Just in case: https://github.com/thesofproject/linux/issues?q=label%3A%22codec%20ES8336%22

EDIT: I didn't see this label was requested. Added. Easier to remove if wrong than to miss if correct.

[lgirdwood]

@marmabiswas123 do you have a topology fix for option A in progress ?

[marmabiswas123]

@lgirdwood — thanks for looking into this.
I attempted the approach you mentioned, but I don’t currently have a properly configured environment to compile SOF topology files. I tried compiling the decompiled topology locally using alsatplg, however the resulting binary was rejected when the OS attempted to load it.

Could you please advise the recommended toolchain and exact commands/versions to compile this topology for ICL (for example, which alsatplg/tplgtool version or branch of the SOF tools I should use)?

Alternatively, if you can provide a prebuilt .tplg (or a patch that I can apply to the decompiled file), I’d be happy to test it and report results. For reference, the decompiled topology I used is attached here: /mnt/data/sof-tplg.txt.

I’m available to run additional logs or tests as required — thanks again for the help.

[lgirdwood]

@marmabiswas123 so for topology 1 used by ICL you dont need any git version of alsatpg, you can just use distro version of alsatplg.

Can you run ./scripts/build-tools -X -C from the SOF scripts directory. This should generate the Cmakefiles to build all topology1 binaries from source, e.g.

/scripts/build-tools.sh -X -C
-- The C compiler identification is GNU 15.2.0
-- Detecting C compiler ABI info
-- Detecting C compiler ABI info - done
-- Check for working C compiler: /usr/bin/cc - skipped
-- Detecting C compile features
-- Detecting C compile features - done
-- No CMAKE_BUILD_TYPE, defaulting to Debug
-- Looking for include file sys/inotify.h
-- Looking for include file sys/inotify.h - found
-- Configuring done (0.1s)
-- Generating done (0.1s)
-- Build files have been written to: /home/lrg/work/sof/sof/tools/build_tools

Build commands for respective tools:
        ctl:        ninja -C "/home/lrg/work/sof/sof/tools/build_tools" sof-ctl
        logger:     ninja -C "/home/lrg/work/sof/sof/tools/build_tools" sof-logger
        probes:     ninja -C "/home/lrg/work/sof/sof/tools/build_tools" sof-probes
        topologies: ninja -C "/home/lrg/work/sof/sof/tools/build_tools" topologies
        topologies1: ninja -C "/home/lrg/work/sof/sof/tools/build_tools" topologies1
        topologies2: ninja -C "/home/lrg/work/sof/sof/tools/build_tools" topologies2

        test tplgs: ninja -C "/home/lrg/work/sof/sof/tools/build_tools" tests
               (or ./tools/test/topology/tplg-build.sh directly)

        list of targets:
                    ninja -C "/home/lrg/work/sof/sof/tools/build_tools/" help

WARNING: building tools/ is now incremental by default!
         To build from scratch delete: /home/lrg/work/sof/sof/tools/build_tools
         or use the -C option.

You should then be able to build and test your topology change.

[marmabiswas123]

Hi @lgirdwood ,
Thank you for the guidance earlier.

I tried following your suggested workflow using ./scripts/build-tools.sh -X -C and the topology1 build structure. However, I ran into a practical issue: the SOF repo does not contain any existing .m4 topology1 file for sof-icl-es8336-ssp0 (or anything close to the ES8336 machine driver for ICL), so there was no reference implementation I could extend.

Since the required topology1 .m4 did not exist in the tree, I attempted the following approach:

1. Decompiling the existing firmware topology

I decompiled the system-provided topology:

/lib/firmware/intel/sof-tplg/sof-icl-es8336-ssp0.tplg

by using : alsatplg -d sof-icl-es8336-ssp0.tplg -o sof-icl-es8336-ssp0.conf

into a .conf, manually corrected the HDMI → pipeline ID mismatches, and recompiled it using alsatplg.

alsatplg -c sof-icl-es8336-ssp0.conf -o sof-icl-es8336-ssp0.tplg

But the kernel rejected the recompiled .tplg with:

No space in tuples array for new tokens
failed to parse SSP tokens
ASoC: topology load failed -22

2. Writing a clean topology1 .m4 file

Since editing raw .conf was unreliable, I attempted to create a proper topology1 .m4 source and compile it through the SOF build system.

However, even after cleaning cached tool outputs, the topology1 build kept pulling in macros that expand to topology2-style pipelines (pipe-host-playback.m4, pipe-volume-playback.m4), which led to:

bad expression in eval: PCM_MAX_RATE / 100 * SCHEDULE_PERIOD / 10000
bad expression in eval: 2 * 4 * PIPELINE_CHANNELS *
fatal error: Invalid parameters to PIPELINE_PCM_ADD

Below is the last version of the topology1 .m4 file I attempted to compile:

dnl ---- includes (use paths that exist in this repo)
include(`common/abi.m4')dnl
include(`common/tlv.m4')dnl
include(`utils.m4')dnl

dnl include the tokens file from the `sof` folder
include(`sof/tokens.m4')dnl

dnl standard helper macros
include(`dai.m4')dnl
include(`pcm.m4')dnl
include(`pipeline.m4')dnl
include(`buffer.m4')dnl

dnl platform helpers
include(`platform/common/hda.m4')dnl
include(`platform/common/ssp.m4')dnl

dnl -----------------------------------------------------------------
dnl SSP hardware blocks
dnl -----------------------------------------------------------------

DECLARE_SSP_HW_CONFIG(5, SSP0, SSP_HS, SSP_CONFIG_FORMAT_I2S)
DECLARE_SSP_HW_CONFIG(6, SSP0, SSP_HS, SSP_CONFIG_FORMAT_I2S)
DECLARE_SSP_HW_CONFIG(7, SSP0, SSP_HS, SSP_CONFIG_FORMAT_I2S)

dnl -----------------------------------------------------------------
dnl PCM + Pipelines
dnl -----------------------------------------------------------------

PIPELINE_ADD_PCM(5, 0, playback, 48000, 2)
PIPELINE_ADD_PCM(6, 0, playback, 48000, 2)
PIPELINE_ADD_PCM(7, 0, playback, 48000, 2)

dnl -----------------------------------------------------------------
dnl HDMI iDisp configs
dnl -----------------------------------------------------------------

DAI_CONFIG(HDA, 5, HDA_OUT, HDA5_conf)
DAI(HDA, 5, 5, HDA5_conf, 0)

DAI_CONFIG(HDA, 6, HDA_OUT, HDA6_conf)
DAI(HDA, 6, 6, HDA6_conf, 0)

DAI_CONFIG(HDA, 7, HDA_OUT, HDA7_conf)
DAI(HDA, 7, 7, HDA7_conf, 0)

dnl -----------------------------------------------------------------
dnl ES8336 Codec on SSP0
dnl -----------------------------------------------------------------

DAI_CONFIG(SSP, 0, ES8336, SSP0_conf)
DAI_CONFIG_DATA(SSP0_conf,
    SSP_CONFIG(SSP0, SSP_MODE_I2S,
        SSP_CLOCK_MASTER,
        48000, 48000,
        SSP_SAMPLE_16BIT, SSP_SAMPLE_16BIT,
        2, 2))

DAI(SSP, 0, 5, SSP0_conf, 0)
DAI(SSP, 0, 6, SSP0_conf, 0)
DAI(SSP, 0, 7, SSP0_conf, 0)

dnl =================================================================
dnl Minimal PCM widgets
dnl =================================================================

WIDGET_PCM_PIPELINE(pcm5_widget, 5, playback)
WIDGET_PCM_PIPELINE(pcm6_widget, 6, playback)
WIDGET_PCM_PIPELINE(pcm7_widget, 7, playback)

Result

Even with a minimal .m4, the topology1 build still fails because the generated configuration pulls in pipeline fragments containing undefined variables (PIPELINE_CHANNELS, SCHEDULE_PERIOD, etc.), which seem to only exist in topology2.

Can you please write a m4 code so that I can test ?
I'm attaching the decompiled and modified .conf file here for your reference.

sof-icl-es8336-ssp0.conf.txt

[lgirdwood]

@marmabiswas123 I think we can go a little easier here, I've dug out the es8336 topologies from git log

git log | grep -C 4 es8336
commit 57e976ec3854cfe4ca34211f6edc041909c9c24a
Author: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Date:   Wed Dec 21 10:24:50 2022 -0600

    topology1: add missing ADL-es8336 topologies.
    
    The ADL integration was botched with missing topologies and errors in
    topology names in the kernel.
    
--
commit 9709ef81dcb8e9630c55419d3fe2d3e65a422fe9
Author: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Date:   Wed Mar 2 11:50:30 2022 -0600

    topology1: sof-glk-es8336: use mclk ID0
    
    It's not clear why the initial contribution used mclk ID1, mclk ID0 is
    used by 99% of devices - and a manual inspection of NHLT supports that
    claim. We will have to deal with the MCLK ID1 case with quirks.
--
commit 04a0d3c1b584acc77c110e12379c15998c2275b2
Author: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Date:   Tue Feb 15 20:43:33 2022 -0600

    topology1: sof-cml-es8336: use correct 24MHz MCLK
    
    Rookie mistake, the clock can't possibly be 19.2 MHz
    
    Reported-by: Nikolai Kostrigin <nickel@altlinux.org>
--
commit 24ad7ea0fa2ea33687c2ef3f7eb85efe75430c60
Author: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Date:   Fri Jul 30 08:21:08 2021 -0500

    topology: sof-glk-es8336: initial support
    
    This topology is used on GLK and TGL. variations include:
    - DMIC supported on TGL and not on GLK

ICL and older Intel devices have not been supported in main branch for some time as they are very old.

Are you able to checkout an older SOF tag that contains the M4 based topologies you need to edit or use as a template. This tooling in this version should be able to build the legacy M4 topologies for you too.

Once you have something that works on your device, we can the upstream into the older stable version tag and the binary into sof-bin.

[marmabiswas123]

@lgirdwood

After an extensive search across all branches, tags, and historical commits of the SOF repository, I was unable to find any .m4 topology source file for ES8336 on Ice Lake (ICL).
The only related files upstream are:

sof-icl-nocodec.m4

ICL topologies for other codecs (RT700, RT711, etc.)

None of these can serve as a template for ES8336 on ICL, because they lack SSP0-Codec definitions, ES8336-specific tokens/tuples, and the associated backend configuration.

Given that no .m4 or .conf source exists upstream, it appears that the following topology was shipped from out-of-tree or unpublished/internal SOF sources:

sof-icl-es8336-ssp0.tplg

Since this binary is present in sof-bin, I would like to kindly request @mengdonglin — who contributed the commit containing this topology — to check whether the original topology source (.m4 or .conf) still exists internally or was part of an unpublished branch.

Having access to the source would allow debugging, reproducibility, and potentially upstreaming proper ES8336 support on Ice Lake.

Thank you very much for looking into this.