[kv2019i]

The speaker and jack echo reference configurations always add a route from alh-copier.Loopback_Virtual.25 to the echo reference siggen, but cavs-sdw.conf defines that DAI copier only with SDW_ECHO_REF_DAI=true.

sof-sdca-amp-ref is built with SDW_SPK_ECHO_REF=true and the default SDW_ECHO_REF_DAI=false, so it contains a route from a widget that no topology defines, and the kernel fails loading it in sof_route_load():

error: G001 route 'alh-copier.Loopback_Virtual.25' -> 'siggen.120.1':
source widget 'alh-copier.Loopback_Virtual.25' does not exist
(kernel: snd_soc_dapm_add_routes, sof_route_load)

Add the loopback route only with SDW_ECHO_REF_DAI=true. The jack echo reference has the same problem, although no current target uses it without the loopback DAI. Topologies built with SDW_ECHO_REF_DAI=true are unchanged apart from the order of routes.

[kv2019i]

Found with the new tplg2 validator tool #11344

[Copilot]

Warning

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Copilot review overview

No open issues 🎉

What changed in this PR

Conditionally adds the echo reference loopback route only when the loopback DAI copier is actually defined (SDW_ECHO_REF_DAI=true), preventing kernel topology load failures due to missing source widgets.

Changes:

• Gate the alh-copier.Loopback_Virtual.25 -> siggen.* route behind IncludeByKey.SDW_ECHO_REF_DAI { "true" { ... } } for both speaker and jack echo reference configurations.

File	Description
tools/​topology/​topology2/​platform/​intel/​sdw-jack-echo-ref.conf	Wraps the loopback echo-ref route in an SDW_ECHO_REF_DAI conditional to avoid referencing an undefined DAI copier.
tools/​topology/​topology2/​platform/​intel/​sdw-amp-echo-ref.conf	Same conditional gating for speaker echo reference to prevent missing-widget route errors when the loopback DAI isn’t present.

_{Review effort: Lite}

[intel-sofci]

PR 11359: test results

Run date: 2026-10-26 16:58 UTC

Tested commit: da6c8e47a8f189c38a9228a7feba6b093403d144

[bardliao]

This was intentional. We may have echo reference for jack and or amp and we need to make sure there is one and only one echo reference DAI. The sof-sdca-amp-ref.tplg is supposed to work with sof-sdca-jack-ref-dai.tplg and the echo reference DAI will be created by sof-sdca-jack-ref-dai.tplg.

[kv2019i]

Thanks @bardliao, you're right, and I misread this. The tool checked sof-sdca-amp-ref against sof-sdca-amp-ref-dai, which are alternatives, and missed that it's meant to pair with sof-sdca-jack-ref-dai, which provides the single loopback DAI. Validated together, the way the kernel loads them (sof-sdca-1amp-id2, sof-sdca-jack-id0, then sof-sdca-jack-ref-dai before sof-sdca-amp-ref in feature_topologies), the route resolves and there's no error. The jack side of this PR is the mirror case, so I'm closing the whole PR.

I've also updated the validator in #11344 to handle this. A new --base option checks a feature topology on top of the topologies loaded before it. It also flags a second definition of a base topology's widget, so a duplicate echo reference DAI would show up. The known-issues entry for sof-sdca-amp-ref now records that it depends on sof-sdca-jack-ref-dai.

One question while we're here. In that combination the validator still warns that alh-copier.Loopback_Virtual.25 has 0 output pins, and that siggen.120.1has two input routes but only one input pin. If the kernel binds Loopback_Virtual.25 -> siggen.120.1 when "Speaker Echo Reference" capture is opened, I'd expect sof_ipc4_get_queue_id() to fail with "invalid output num_pins: 0". Is that route skipped for the virtual loopback DAI or handled some other way?

[bardliao]

@kv2019i Not sure what combination you are referring to. Below are the tplg graphics of 1. sof-sdca-1amp-id2 + sof-sdca-amp-ref-dai 2. sof-sdca-jack-id0 + sof-sdca-jack-ref-dai and 3 sof-sdca-1amp-id2 + sof-sdca-amp-ref-dai + sof-sdca-jack-id0 + sof-sdca-jack-ref-dai

alh-copier.Loopback_Virtual.25 has 0 input but has output(s). And it has 0 input because it is a virtual DAI.