[FEATURE][CHT] Add SOF topology support for Lenovo Yoga Book RT5677

[MaurizioCasciano]

Is your feature request related to a problem? Please describe.

The Lenovo Yoga Book YB1-X91F/L uses an Intel Cherry Trail audio DSP (808622A8) with an RT5677 codec exposed as ACPI ID 10EC5677.

The Linux cht-yogabook ASoC machine driver works on this hardware with the legacy SST driver and intel/fw_sst_22a8.bin. When SOF is selected, however, the probe stops with:

sof-audio-acpi-intel-byt 808622A8:00: error: no matching ASoC machine driver found - aborting probe

The Yoga Book entry in the Cherry Trail machine table deliberately has no sof_tplg_filename, because neither the SOF source tree nor current sof-bin releases provide sof-cht-rt5677.tplg:

https://github.com/Yoga-Book/Yoga-Book-Linux-Kernel/blob/8a24dc4997fa8d6a69523fa3abc0ed5a15632309/sound/soc/intel/common/soc-acpi-intel-cht-match.c#L140-L146

The machine driver is here:

https://github.com/Yoga-Book/Yoga-Book-Linux-Kernel/blob/8a24dc4997fa8d6a69523fa3abc0ed5a15632309/sound/soc/intel/boards/cht_yogabook.c

Describe the solution you'd like

Add and publish a Cherry Trail RT5677 topology, tentatively named sof-cht-rt5677.tplg, so that Linux can reference it from the Yoga Book machine-table entry and instantiate the existing cht-yogabook machine driver under SOF.

The working SST machine configuration indicates these backend requirements:

• SSP2 connected to rt5677-aif1
• SND_SOC_DAIFMT_DSP_B | SND_SOC_DAIFMT_IB_NF
• codec and frame consumer (CBC_CFC)
• 48 kHz, stereo, S24_LE backend
• four TDM slots, slot width 25, TX/RX masks 0x0f
• 19.2 MHz MCLK
• RT5677 PLL output of 24.576 MHz (48000 * 512)
• playback and capture, including the existing media and deep-buffer front ends

I would also appreciate guidance on the accepted implementation path. SOF v2.2 generated the existing Cherry Trail codec topologies from the topology1 sof-byt-codec.m4 recipe, but current main no longer lists those CHT recipes in its production topology1 build, and I could not find an Atom/Cherry Trail topology2 equivalent. Should new support for this older platform:

1. extend the legacy topology1 recipe,
2. add Cherry Trail support to topology2, or
3. follow another maintained compatibility path?

I can implement the topology and test it on physical YB1-X91L hardware once the preferred upstream direction is known.

Describe alternatives you've considered

The currently working workaround is to force the legacy SST driver with:

snd_intel_dspcfg.dsp_driver=2

Using or renaming sof-cht-rt5670.tplg does not appear safe. The historical generic CHT RT5670 recipe configures I2S with two 25-bit slots and masks 0x03, whereas the Yoga Book RT5677 machine driver configures DSP_B with four 25-bit slots and masks 0x0f.

There is also a historical Broadwell RT5677 topology, but its DSP platform, SSP port, clocking, and PCM graph differ from Cherry Trail and it is not directly reusable.

Additional context

After a topology is available and tested, the corresponding Linux change should be small: add .sof_tplg_filename = "sof-cht-rt5677.tplg" to the 10EC5677 Yoga Book machine-table entry. Any machine-driver adjustments discovered during physical SOF testing would be submitted separately to Linux.

The compiled topology would subsequently need inclusion in sof-bin, followed by distribution firmware packaging and verification of the Yoga Book ALSA UCM profile under the SOF card identity.