Tools: Testbench: Track and print heap usage for modules by singalsu · Pull Request #10593 · thesofproject/sof

singalsu · 2026-03-03T17:03:24Z

This patch adds to end to sof-testbench4 run print of peak heap consumption for each module. The information is retrieved from heap_high_water_mark data that is tracked by module adapter for each module.

Example of output (sof-hda-generic.tplg):

Heap usage for module gain.1.1: 708 bytes
Heap usage for module mixin.1.1: 164 bytes
Heap usage for module gain.2.1: 708 bytes
Heap usage for module eqiir.2.1: 316 bytes
Heap usage for module eqfir.2.1: 388 bytes
Heap usage for module drc.2.1: 4316 bytes
Heap usage for module mixout.2.1: 252 bytes

Copilot

Pull request overview

This PR adds reporting of per-module peak heap consumption at the end of a sof-testbench4 run by reading each module’s heap_high_water_mark tracked by the module adapter.

Changes:

Collect IPC4 module heap high-water marks by iterating the IPC4 widget list and resolving component devices.
Print per-module heap usage in the test summary output.
Introduce a small record struct (tb_heap_usage_record) to pass heap usage data to the stats printer.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
tools/testbench/testbench.c	Adds IPC4-only collection of module heap HWM and prints the results in the pipeline stats output.
tools/testbench/include/testbench/utils.h	Introduces `tb_heap_usage_record` used to carry module name and heap HWM to the stats function.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-03T17:13:43Z

tools/testbench/testbench.c

+		}
+
+		records[count].module_name = info->name;
+		records[count].heap_max = (int)dev->mod->priv.resources.heap_high_water_mark;


heap_high_water_mark is a size_t (see src/include/sof/audio/module_adapter/module/generic.h), but it’s being cast to int and printed with %d. This can truncate/overflow and report incorrect (even negative) values on 64-bit hosts. Consider storing heap_max as size_t (or uint64_t) in tb_heap_usage_record and printing with the matching format specifier (e.g., %zu).

Suggested change

records[count].heap_max = (int)dev->mod->priv.resources.heap_high_water_mark;

/* heap_high_water_mark is size_t; clamp to INT_MAX to avoid overflow */

{

size_t heap_high_water_mark = dev->mod->priv.resources.heap_high_water_mark;

if (heap_high_water_mark > (size_t)INT_MAX)

records[count].heap_max = INT_MAX;

else

records[count].heap_max = (int)heap_high_water_mark;

}

Print of %zu isn't working with xt-clang/xt-run env. I can change the storage type and typecast the print.

Storage is now size_t and print is with typecast to uint32_t. We can't use more heap in DSP than what the uint32_t range is.

tools/testbench/testbench.c

tools/testbench/include/testbench/utils.h

This patch adds to end to sof-testbench4 run print of peak heap consumption for each module. The information is retrieved from heap_high_water_mark data that is tracked by module adapter for each module. Signed-off-by: Seppo Ingalsuo <seppo.ingalsuo@linux.intel.com>

Copilot

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-03-04T10:51:03Z

tools/testbench/testbench.c

+		if (count >= TB_NUM_WIDGETS_SUPPORTED) {
+			fprintf(stderr, "Error: Too many components for heap records, max %d.\n",
+				TB_NUM_WIDGETS_SUPPORTED);
+			break;


tb_collect_heap_usage() caps heap records at TB_NUM_WIDGETS_SUPPORTED (16) and breaks out, which will silently drop heap usage for remaining modules on larger topologies. Consider sizing the record buffer to the actual number of non-AIF/DAI components (e.g., first pass to count, then allocate), or introduce a dedicated "max components" constant that is large enough for real-world IPC4 topologies.

Suggested change

break;

*count_out = count;

return -EINVAL;

Copilot · 2026-03-04T10:51:03Z

tools/testbench/testbench.c

+		}
+
+		records[count].module_name = info->name;
+		records[count].heap_max = dev->mod->priv.resources.heap_high_water_mark;


tb_collect_heap_usage() reads dev->mod->priv.resources.heap_high_water_mark directly. Since module_adapter_heap_usage(mod, &hwm) exists in the module adapter API, prefer using that accessor to avoid coupling testbench to internal processing_module layout/details.

Suggested change

records[count].heap_max = dev->mod->priv.resources.heap_high_water_mark;

if (module_adapter_heap_usage(dev->mod, &records[count].heap_max) < 0)

continue;

Copilot · 2026-03-04T10:51:03Z

tools/testbench/testbench.c

 		delta_t = (td1.tv_sec - td0.tv_sec) * 1000000;
 		delta_t += (td1.tv_nsec - td0.tv_nsec) / 1000;
-		test_pipeline_stats(tp, delta_t);
+		test_pipeline_stats(tp, delta_t, heap_usage_records, heap_usage_records_count);


In error paths that goto out before tb_gettime(&td0) / tb_gettime(&td1) run, td0/td1 can be uninitialized, but delta_t is still computed from them. This is undefined behavior and can print garbage timing (or worse). Initialize td0/td1 and/or gate the delta_t calculation/printing behind a flag that indicates timing was captured.

Copilot · 2026-03-04T10:51:03Z

tools/testbench/testbench.c

-	printf("\n");
+	if (delta_t)
+		printf("Total execution time: %lld us, %.2f x realtime\n\n", delta_t,
+		       (float)frames_out / tp->fs_out * 1000000 / delta_t);


test_pipeline_stats() no longer prints a trailing blank line when delta_t == 0 (previously it always printed \n). If delta_t can be 0 for short runs or error cases, consider keeping an unconditional newline to avoid the next output running into the summary block.

Suggested change

(float)frames_out / tp->fs_out * 1000000 / delta_t);

(float)frames_out / tp->fs_out * 1000000 / delta_t);

else

printf("\n");

singalsu marked this pull request as ready for review March 3, 2026 17:08

Copilot AI review requested due to automatic review settings March 3, 2026 17:08

singalsu requested review from dbaluta, kv2019i, lbetlej, lgirdwood, mmaka1, plbossart and ranj063 as code owners March 3, 2026 17:08

Copilot started reviewing on behalf of singalsu March 3, 2026 17:10 View session

Copilot AI reviewed Mar 3, 2026

View reviewed changes

singalsu force-pushed the testbench_print_heap_usage branch from 215636e to 8fc72b9 Compare March 4, 2026 10:41

singalsu requested a review from Copilot March 4, 2026 10:41

Copilot started reviewing on behalf of singalsu March 4, 2026 10:45 View session

Copilot AI reviewed Mar 4, 2026

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Tools: Testbench: Track and print heap usage for modules#10593

Tools: Testbench: Track and print heap usage for modules#10593
singalsu wants to merge 1 commit intothesofproject:mainfrom
singalsu:testbench_print_heap_usage

singalsu commented Mar 3, 2026 •

edited

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 3, 2026

Uh oh!

singalsu Mar 4, 2026

Uh oh!

singalsu Mar 4, 2026

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Mar 4, 2026

Uh oh!

Copilot AI Mar 4, 2026

Uh oh!

Copilot AI Mar 4, 2026

Uh oh!

Copilot AI Mar 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

-		records[count].heap_max = (int)dev->mod->priv.resources.heap_high_water_mark;
+		/* heap_high_water_mark is size_t; clamp to INT_MAX to avoid overflow */
+		{
+			size_t heap_high_water_mark = dev->mod->priv.resources.heap_high_water_mark;
+			if (heap_high_water_mark > (size_t)INT_MAX)
+				records[count].heap_max = INT_MAX;
+			else
+				records[count].heap_max = (int)heap_high_water_mark;
+		}

	records[count].heap_max = dev->mod->priv.resources.heap_high_water_mark;
	if (module_adapter_heap_usage(dev->mod, &records[count].heap_max) < 0)
	continue;

Conversation

singalsu commented Mar 3, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Copilot AI Mar 3, 2026

Choose a reason for hiding this comment

Uh oh!

singalsu Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

singalsu Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Mar 4, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

singalsu commented Mar 3, 2026 •

edited

Loading