Audio: Level multiplier: Add new component #10176

singalsu · 2025-08-19T17:28:36Z

No description provided.

singalsu · 2025-08-19T17:30:46Z

Still draft, I'll add HiFi3/4/5 versions using the code from volume component. However I can't use same as library since volume uses Q1.31 gain build while this is Q9.23 to have possibility for amplify.

Edit - this version contains the HiFi optimizations. It can be run in sof-testbench4 and real device with generated test topologies for HDA and SDW. The default blob sets 0 dB gain and pass-through processing. The processing gain can be changed in runtime with sof-ctl with example blobs for -40 .. +40 dB.

lyakh · 2025-08-25T07:16:00Z

src/audio/level_multiplier/level_multiplier-ipc3.c

+ * is used to separate the controls instances of same type. In control payload
+ * the num_elems defines to how many channels the control is applied to.
+ */
+__cold int level_multiplier_set_config(struct processing_module *mod, uint32_t param_id,


I don't think anybody is using __cold with IPC3, but it probably doesn't hurt either. It's only defined to something non-trivial on ACE, and ACE doesn't use IPC3

lyakh · 2025-08-25T07:20:24Z

src/audio/level_multiplier/level_multiplier-ipc3.c

+	assert_can_be_cold();
+
+	comp_warn(mod->dev, "Not supported set control, type %d", cdata->cmd);
+	return 0;


would it work to just

#if IPC3 #define level_multiplier_set_config NULL #define level_multiplier_get_config NULL #endif

in a private header?

Need to think this, my first try gave this:

src/audio/level_multiplier/level_multiplier.h:25:37: error: expected identifier or ‘(’ before ‘void’ 25 | #define level_multiplier_set_config NULL /* No implementation for IPC3 set/get config */

lyakh · 2025-08-25T07:30:05Z

src/audio/level_multiplier/level_multiplier-ipc4.c

+	}
+
+	memcpy_s(&cd->gain, sizeof(int32_t), fragment, sizeof(int32_t));
+	comp_info(mod->dev, "Gain set to %d", cd->gain);


would a comp_dbg() be enough?

lyakh · 2025-08-25T07:30:32Z

src/audio/level_multiplier/level_multiplier-ipc4.c

+{
+	assert_can_be_cold();
+	return 0;
+}


just don't use .get_config?

Yep I removed it and also removed level_multiplier-ipc3.c. I defined set_config for IPC3 build as inline dummy function. The define trick didn't build.

Sorry, not following, why do we not have a get config here ? i.e. userspace will probably find it useful to know any gain or attenuation in the pipeline.

lyakh · 2025-08-25T07:31:00Z

src/audio/level_multiplier/level_multiplier.c

+
+	comp_info(dev, "level_multiplier_init()");
+
+	cd = rzalloc(SOF_MEM_FLAG_USER, sizeof(*cd));


mod_alloc()?

Thank, I did this and next

lyakh · 2025-08-25T07:35:34Z

src/audio/level_multiplier/level_multiplier.c

+	assert_can_be_cold();
+
+	comp_dbg(mod->dev, "level_multiplier_free()");
+	rfree(cd);


mod_free then

lgirdwood · 2025-08-25T16:35:43Z

src/audio/level_multiplier/level_multiplier.c

+
+	comp_info(dev, "level_multiplier_init()");
+
+	cd = rzalloc(SOF_MEM_FLAG_USER, sizeof(*cd));


src/audio/level_multiplier/tune/sof_level_multiplier_blobs.m

Copilot

Pull Request Overview

This PR adds a new level multiplier component to the audio processing pipeline that applies a configurable gain to audio signals. The component supports multiple sample formats (S16, S24, S32) with optimized implementations for different DSP architectures.

Adds level multiplier audio component with gain configuration control
Provides generic, HiFi3, and HiFi5 optimized implementations
Includes pre-configured gain settings from -40dB to +40dB in 10dB increments

Reviewed Changes

Copilot reviewed 43 out of 43 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
uuid-registry.txt	Adds UUID for level multiplier component
tools/topology/topology2/include/components/level_multiplier.conf	Component definition for topology integration
tools/topology/topology2/include/components/level_multiplier/*.conf	Pre-configured gain settings (-40dB to +40dB)
tools/topology/topology2/include/bench/*.conf	Benchmark configuration files
tools/topology/topology2/cavs-benchmark-*.conf	Integration with benchmark topologies
tools/rimage/config/.toml	Module configuration for firmware images
src/audio/level_multiplier/*.c	Core implementation with format-specific processing
src/audio/level_multiplier/*.h	Component interfaces and data structures

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

lgirdwood

LGTM, but just one question.

lgirdwood · 2025-08-29T13:09:00Z

src/audio/level_multiplier/level_multiplier-ipc4.c

+{
+	assert_can_be_cold();
+	return 0;
+}


Sorry, not following, why do we not have a get config here ? i.e. userspace will probably find it useful to know any gain or attenuation in the pipeline.

singalsu · 2025-09-01T08:38:39Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

lgirdwood · 2025-09-01T09:34:45Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

The reading of kcontrols is user initiated, so expectation is that level multiplier would expose a RO kcontrol for the gain/attenuation dB value.

kv2019i · 2025-09-04T10:28:32Z

src/audio/level_multiplier/level_multiplier.c

+	struct level_multiplier_comp_data *cd =  module_get_private_data(mod);
+	struct comp_dev *dev = mod->dev;
+	struct sof_source *source = sources[0]; /* One input in this example */
+	struct sof_sink *sink = sinks[0]; /* One output in this example */


Oops, fixing now.

singalsu · 2025-09-09T08:33:04Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

The reading of kcontrols is user initiated, so expectation is that level multiplier would expose a RO kcontrol for the gain/attenuation dB value.

Peter is reworking the the bytes control in kernel to make the notification possible. I'll check the impacts to this component also. The FW component should work with all meaningful control access rights defined in topology. Though we may not currently support all of them in kernel.

Level multiplier is used to set a fixed gain to stream, e.g. add +20 dB sensitivity to voice capture compared to media audio capture from the same microphone. The gain is defined by boot time set bytes control. Usually it would be set by topology. Runtime set by ALSA UCM2 is also possible usage. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

This patch adds script sof_level_multiplier_blobs.m It exports blobs for -40 to +40 dB gain configuration. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

This patch adds the HiFi3 and HiFi4 optimized processing. The MCPS performance reported for stereo 48 kHz audio by: process_test('level_multiplier', N, N, 48e3, 0, 1, 'xt-run --mem_model'); is 2.50 / 2.68 / 2.68 for S16_LE / S24_LE / S32_LE formats when processing with non-unity gain. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

This patch adds the HiFi5 optimized processing. The MCPS performance reported for stereo 48 kHz audio by: process_test('level_multiplier', N, N, 48e3, 0, 1, 'xt-run --mem_model'); is 2.21 / 2.41 / 2.41 for S16_LE / S24_LE / S32_LE formats when processing with non-unity gain. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

This patch adds build of e.g. sof-hda-benchmark-level_multiplier32.tplg and sof-mtl-sdw-benchmark-level_multiplier32.tplg to test run the component. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

lgirdwood · 2025-09-09T11:58:01Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

The reading of kcontrols is user initiated, so expectation is that level multiplier would expose a RO kcontrol for the gain/attenuation dB value.

Peter is reworking the the bytes control in kernel to make the notification possible. I'll check the impacts to this component also. The FW component should work with all meaningful control access rights defined in topology. Though we may not currently support all of them in kernel.

@singalsu @ujfalusi ok, we can follow up with the kcontrol in another PR, but this wont need a bytes kconrol. It will just need a RO volume kcontrol that shows a dB gain/attenuation.

lgirdwood · 2025-09-09T12:00:20Z

src/audio/level_multiplier/Kconfig

+	  Select for Level multiplier component. This component
+          applies a fixed gain to audio. The amount of gain
+          is configured in the bytes control that is typically
+          set in boot time from topology. It can e.g. increase
+          capture sensitivity of voice applications by 20 dB
+          compared to media capture.


btw - mix of tabs/spaces ? Not a blocker.

ujfalusi · 2025-09-09T12:57:50Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

The reading of kcontrols is user initiated, so expectation is that level multiplier would expose a RO kcontrol for the gain/attenuation dB value.

Peter is reworking the the bytes control in kernel to make the notification possible. I'll check the impacts to this component also. The FW component should work with all meaningful control access rights defined in topology. Though we may not currently support all of them in kernel.

@singalsu @ujfalusi ok, we can follow up with the kcontrol in another PR, but this wont need a bytes kconrol. It will just need a RO volume kcontrol that shows a dB gain/attenuation.

We don't have support for volume control change notification either. We only have Switch and Enum change support.
Volume controls are legacy IPC4 things and they are basically just a blob without much sanity - I'm not sure how to do the notification with them, but neither looked at it.
Bytes is almost done now, few rough edges here and there - mainly because binary is also legacy IPC4 style of just a blob w/o any information sent down and hope that it is handled as it should (and reaches what it should reach)

lgirdwood · 2025-09-10T10:44:27Z

LGTM, but just one question.

In IPC4 the kernel newer gets control value from firmware unless the firmware sends a notification to read it (currently for enum and switch, bytes support is coming). This component isn't using notification. I don't think we support the volatile control type without doing the notification.

The reading of kcontrols is user initiated, so expectation is that level multiplier would expose a RO kcontrol for the gain/attenuation dB value.

Peter is reworking the the bytes control in kernel to make the notification possible. I'll check the impacts to this component also. The FW component should work with all meaningful control access rights defined in topology. Though we may not currently support all of them in kernel.

@singalsu @ujfalusi ok, we can follow up with the kcontrol in another PR, but this wont need a bytes kconrol. It will just need a RO volume kcontrol that shows a dB gain/attenuation.

We don't have support for volume control change notification either. We only have Switch and Enum change support. Volume controls are legacy IPC4 things and they are basically just a blob without much sanity - I'm not sure how to do the notification with them, but neither looked at it. Bytes is almost done now, few rough edges here and there - mainly because binary is also legacy IPC4 style of just a blob w/o any information sent down and hope that it is handled as it should (and reaches what it should reach)

We dont need a notification here as this gain will never change. This is more about telling userspace via ALSA kcontrol API that we have a gain/attenuation in this pipeline. Not a blocker to merge now, but we need this when we have the IPC4 kcontrol updates merged.

singalsu force-pushed the add_level_multiply_comp branch 2 times, most recently from fb716e9 to d06f950 Compare August 22, 2025 14:27

singalsu marked this pull request as ready for review August 22, 2025 14:28

Copilot AI review requested due to automatic review settings August 22, 2025 14:28

singalsu requested review from dbaluta, kv2019i, lbetlej, lgirdwood, mmaka1, plbossart and ranj063 as code owners August 22, 2025 14:28

This comment was marked as outdated.

Sign in to view

singalsu force-pushed the add_level_multiply_comp branch from d06f950 to be9b290 Compare August 22, 2025 14:42

singalsu requested a review from jsarha as a code owner August 22, 2025 14:42

singalsu requested a review from Copilot August 22, 2025 14:45

This comment was marked as outdated.

Sign in to view

singalsu force-pushed the add_level_multiply_comp branch from be9b290 to 84f81b4 Compare August 22, 2025 15:44

singalsu requested a review from Copilot August 22, 2025 15:46

This comment was marked as outdated.

Sign in to view

lyakh reviewed Aug 25, 2025

View reviewed changes

lgirdwood reviewed Aug 25, 2025

View reviewed changes

singalsu force-pushed the add_level_multiply_comp branch 2 times, most recently from 4d9952c to 9548b83 Compare August 29, 2025 10:47

singalsu requested review from Copilot, lgirdwood and lyakh August 29, 2025 11:56

Copilot AI reviewed Aug 29, 2025

View reviewed changes

lgirdwood reviewed Aug 29, 2025

View reviewed changes

kv2019i mentioned this pull request Sep 3, 2025

rtos: remove remains of old XTOS spinlock.h implementation #10214

Merged

kv2019i approved these changes Sep 4, 2025

View reviewed changes

singalsu added 5 commits September 9, 2025 13:41

Audio: Level Multiplier: Add blobs generate script

99637fa

This patch adds script sof_level_multiplier_blobs.m It exports blobs for -40 to +40 dB gain configuration. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

Tools: Topology: Add test topology for Level Multiplier

36bd4b2

This patch adds build of e.g. sof-hda-benchmark-level_multiplier32.tplg and sof-mtl-sdw-benchmark-level_multiplier32.tplg to test run the component. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

singalsu force-pushed the add_level_multiply_comp branch from 9548b83 to 36bd4b2 Compare September 9, 2025 10:42

lgirdwood approved these changes Sep 9, 2025

View reviewed changes

lgirdwood merged commit 35345e4 into thesofproject:main Sep 10, 2025
37 of 47 checks passed


		comp_info(dev, "level_multiplier_init()");

		cd = rzalloc(SOF_MEM_FLAG_USER, sizeof(*cd));

Audio: Level multiplier: Add new component #10176

Audio: Level multiplier: Add new component #10176

Uh oh!

Conversation

singalsu commented Aug 19, 2025

Uh oh!

singalsu commented Aug 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

Uh oh!

This comment was marked as outdated.

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!

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!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

lgirdwood left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

singalsu commented Sep 1, 2025

Uh oh!

lgirdwood commented Sep 1, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

singalsu commented Sep 9, 2025

Uh oh!

lgirdwood commented Sep 9, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

ujfalusi commented Sep 9, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lgirdwood commented Sep 10, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

singalsu commented Aug 19, 2025 •

edited

Loading

ujfalusi commented Sep 9, 2025 •

edited

Loading