UCM2: Intel: sof-hda-dsp: Control SOF processing from UCM #419

singalsu · 2024-05-22T07:14:10Z

This is a set of patches to control with UCMv2 audio processing in Sound Open Firmware (SOF). The settings are generic, mostly pass-through mode, except for speaker where a generic high-pass filter and DRC boost are applied.

All the filters can be set up with a bespoke tuned version per product for a better audio experience.

perexg · 2024-05-28T08:47:14Z

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

+		}
+	}
+
+        Macro.headphone.SofControl "endpoint=Headphone drcswitch=off"


Please, try to keep only this line in SectionDevice."Headphones" subtree. The subtrees should be merged, so there is no requirement to redefine the whole block again.

This change works,

SectionDevice."Headphones" { Macro.headphone.SofControl "endpoint=Headphone drcswitch=off" }

perexg · 2024-05-28T08:47:26Z

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

+			}
+		}
+
+                Macro.speaker.SofControl "endpoint=Speaker drcswitch=on"


Just this line.

But here, both of these fail. @perexg What is the correct way?

SectionDevice."Speaker" { Macro.speaker.SofControl "endpoint=Speaker drcswitch=on" }

If.spk { Condition { Type ControlExists Control "name='Speaker Playback Switch'" } True.SectionDevice."Speaker" { Macro.speaker.SofControl "endpoint=Speaker drcswitch=on" } }

The speaker controls are not set, there's no print Speaker happening to /tmp/alsa-ucm.txt and I see in /var/log/syslog message " wireplumber[3991]: Failed to enable ucm device Speaker".

@pereg Sorry for noise, again my bad. In my testing version I had device specific customization in place but needed EQ blobs were not in place. I should get this now working!

perexg · 2024-05-28T08:51:41Z

ucm2/sof/ipc3/eq_fir/README.md

@@ -0,0 +1,7 @@
+# How to build
+
+These blobs were exported with example_fir_eq.m tool from


I would probably create one more level for those blobs like ucm2/blobs/sof/....

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

The example set contains passthrough configuration blobs with SOF IPC3 and IPC4 headers for DRC, FIR, and IIR. A few high-pass configurations are added for IIR to be used e.g. for speakers. A DRC blob is added that can be used to boost speaker playback loudness. The blobs are all in binary format. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

This example shows how to define IIR, FIR, and DRC processing for speaker and headphone endpoints. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

kv2019i

I tested an earlier version, but did another test with this version on a Intel IPC4 laptop. I tested specifying tuning blobs, switching between speaker and headphones, and also specifying the blob via a user config. All worked as advertise.

I can do a similar test when @singalsu you update the version.

kv2019i · 2024-05-30T17:23:59Z

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

+                                EnableSequence [
+                                        cset-tlv "name='${var:PostMixerAnalogPlaybackIIRBytes}' ${var:EndpointIirBlob}"
+                                        cset-tlv "name='${var:PostMixerAnalogPlaybackFIRBytes}' ${var:EndpointFirBlob}"
+                                        #shell "/bin/echo '${var:__endpoint} ${var:EndpointIirBlob} ${var:EndpointFirBlob}' >> /tmp/alsa-ucm.txt"


I know it's not common to leave such commented-out code, but I did find these immensely useful when debugging and I could imagine these are useful for anyone who wants to add tuning for new laptop configurations later one. So if possible, +1 to leave these as examples.

I moved the commented shell up before controls set to see the print before fails with them. There's now also a comment about use for debugging.

ujfalusi

@singalsu, this looks clean to my untrained eyes, I only have one nitpick and one question

ujfalusi · 2024-05-31T10:46:57Z

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

+	}
+}
+
+# Merge this add to Headpones subtree in HDA/HiFi-analog.conf


Drop the 'add' from the comment

ucm2/Intel/sof-hda-dsp/HiFi-sof.conf

perexg

Except the speaker block (missing condition) all looks really nice now.

This patch adds to Intel/sof-hda-dsp/HiFi.conf inclusion of HiFi-sof.conf that by redefine of headphone and speaker handling adds to UCM control of DRC and EQ SOF processing components. The modified setting are applied in case of SOF processing components' controls are detected. There is no change to operation if no controls are present e.g. with legacy SOF topology. If DRC control is found, it is assumed that also FIR and IIR also exist. If there is no DRC but FIR is found, then it is assumed that IIR also exists. This matches SOF FW builds for IPC3 (FIR, IIR) and IPC4 (DRC, FIR, IIR). The controls names are different in IPC3 and IPC4 topologies. Also the configuration blobs differ. The speaker mode is by default set up with 100 Hz high-pass IIR. The DRC is set to a default speaker setting that boosts playback loudness. The FIR is bypassed. In the headphone mode all the processing is set to bypass and DRC switch is set off. The processing can be customized for products with UCM scripts placed into blobs/sof/product_configs. The file path should be <sys_vendor>/<product_name>.conf from DMI ID. An user configuration can be similarly placed into blobs/sof/user_configs directory to e.g. avoid it being overwritten by distribution. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

ujfalusi

@singalsu, looks clean to me.

This example shows how to define IIR, FIR, and DRC processing for speaker and headphone endpoints. Closes: #419 Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com> Signed-off-by: Jaroslav Kysela <perex@perex.cz>

This patch adds to Intel/sof-hda-dsp/HiFi.conf inclusion of HiFi-sof.conf that by redefine of headphone and speaker handling adds to UCM control of DRC and EQ SOF processing components. The modified setting are applied in case of SOF processing components' controls are detected. There is no change to operation if no controls are present e.g. with legacy SOF topology. If DRC control is found, it is assumed that also FIR and IIR also exist. If there is no DRC but FIR is found, then it is assumed that IIR also exists. This matches SOF FW builds for IPC3 (FIR, IIR) and IPC4 (DRC, FIR, IIR). The controls names are different in IPC3 and IPC4 topologies. Also the configuration blobs differ. The speaker mode is by default set up with 100 Hz high-pass IIR. The DRC is set to a default speaker setting that boosts playback loudness. The FIR is bypassed. In the headphone mode all the processing is set to bypass and DRC switch is set off. The processing can be customized for products with UCM scripts placed into blobs/sof/product_configs. The file path should be <sys_vendor>/<product_name>.conf from DMI ID. An user configuration can be similarly placed into blobs/sof/user_configs directory to e.g. avoid it being overwritten by distribution. Closes: #419 Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com> Signed-off-by: Jaroslav Kysela <perex@perex.cz>

perexg · 2024-05-31T12:25:51Z

Thanks. Applied now.

kv2019i mentioned this pull request May 27, 2024

[FEATURE] UCM support for DMI named based differentiation thesofproject/sof#9061

Closed

perexg requested changes May 28, 2024

View reviewed changes

singalsu added 2 commits May 30, 2024 18:25

UCM2: SOF: Add example blobs customization for AAEON UPX-TGL01 board

eabdf9f

This example shows how to define IIR, FIR, and DRC processing for speaker and headphone endpoints. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

kv2019i reviewed May 30, 2024

View reviewed changes

singalsu requested review from kv2019i and perexg May 31, 2024 10:26

ujfalusi reviewed May 31, 2024

View reviewed changes

perexg requested changes May 31, 2024

View reviewed changes

ujfalusi approved these changes May 31, 2024

View reviewed changes

perexg closed this in 4502b17 May 31, 2024

		@@ -0,0 +1,7 @@
		# How to build

		These blobs were exported with example_fir_eq.m tool from

UCM2: Intel: sof-hda-dsp: Control SOF processing from UCM #419

UCM2: Intel: sof-hda-dsp: Control SOF processing from UCM #419

Uh oh!

Conversation

singalsu commented May 22, 2024

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ujfalusi left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

perexg left a comment

Choose a reason for hiding this comment

Uh oh!

ujfalusi left a comment

Choose a reason for hiding this comment

Uh oh!

perexg commented May 31, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants