Part 1 of userspace DP subset PR #10302 by lyakh · Pull Request #10350 · thesofproject/sof

lyakh · 2025-11-03T16:26:20Z

The simpler part of #10302 - no integration with the driver heap API

Copilot

Pull Request Overview

This PR introduces new heap allocation APIs and refactors memory allocation in the module adapter to consolidate allocation/deallocation logic into helper functions.

Adds sof_heap_alloc() and sof_heap_free() functions for heap-specific memory allocation with flag support
Refactors module adapter memory allocation/deallocation into dedicated helper functions
Extends module allocation API with mod_alloc_ext() to support allocation flags (e.g., coherency control)

Reviewed Changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
zephyr/lib/alloc.c	Adds `sof_heap_alloc()` and `sof_heap_free()` functions for managing heap allocations with flag support
zephyr/include/rtos/alloc.h	Declares new heap allocation functions
test/cmocka/src/common_mocks.c	Provides mock implementations of heap allocation functions for testing
src/platform/library/lib/alloc.c	Adds stub implementations using standard malloc/free
src/lib/alloc.c	Adds stub implementations using standard malloc/free
src/include/sof/audio/module_adapter/module/generic.h	Adds `mod_alloc_ext()` declaration and converts `mod_alloc_align()` to inline wrapper; fixes typo
src/audio/module_adapter/module_adapter.c	Extracts memory allocation/free logic into helper functions
src/audio/module_adapter/module/generic.c	Renames `mod_alloc_align()` to `mod_alloc_ext()` with flags parameter; fixes typos
posix/include/rtos/kernel.h	Adds stub `k_heap` struct definition
posix/include/rtos/alloc.h	Adds forward declarations and stub functions for heap allocation

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

Copilot · 2025-11-03T16:31:01Z

src/platform/library/lib/alloc.c

+void *sof_heap_alloc(struct k_heap *heap, uint32_t flags, size_t bytes,
+		     size_t alignment)
+{
+	return malloc(bytes);


The alignment parameter is not used in this implementation. While this is a stub implementation, it should either align the allocated memory or document why alignment is not respected. Consider using aligned_alloc() or documenting this limitation.

Suggested change

return malloc(bytes);

/* Use aligned_alloc to respect the alignment parameter.

* Note: aligned_alloc requires that bytes is a multiple of alignment.

*/

if (alignment && (bytes % alignment != 0)) {

bytes = ((bytes + alignment - 1) / alignment) * alignment;

}

return aligned_alloc(alignment ? alignment : sizeof(void *), bytes);

Copilot · 2025-11-03T16:31:01Z

src/lib/alloc.c

+void *sof_heap_alloc(struct k_heap *heap, uint32_t flags, size_t bytes,
+		     size_t alignment)
+{
+	return malloc(bytes);


The alignment parameter is not used in this implementation. While this is a stub implementation, it should either align the allocated memory or document why alignment is not respected. Consider using aligned_alloc() or documenting this limitation.

Suggested change

return malloc(bytes);

return aligned_alloc(alignment, bytes);

Copilot · 2025-11-03T16:31:01Z

test/cmocka/src/common_mocks.c

+void WEAK *sof_heap_alloc(struct k_heap *heap, uint32_t flags, size_t bytes,
+		     size_t alignment)
+{
+	(void)heap;
+	(void)flags;
+	(void)alignment;
+
+	return malloc(bytes);
+}


The alignment parameter is marked as unused but is not respected by the underlying malloc() call. This could lead to misaligned memory in tests. Consider using aligned_alloc() to honor the alignment requirement or documenting that alignment is not supported in test mocks.

lgirdwood

@lyakh are you able to drop the 1st patch atm, and get the others that don't depend on patch 1 merged as part 1. Then we can follow up with heaps.

kv2019i

@lyakh can you add some human readable summary in the title? this is now part 1 of 1000, and the 1000 is another "part y of 1002" of PR. it would be nice for the summary to immediately describe what this PR is about.

lgirdwood · 2025-11-04T11:28:17Z

@lyakh are you able to drop the 1st patch atm, and get the others that don't depend on patch 1 merged as part 1. Then we can follow up with heaps.

Ok, I'm good after chat.

lgirdwood · 2025-11-04T11:31:08Z

SOFCI TEST

softwarecki · 2025-11-04T15:19:35Z

src/audio/module_adapter/module_adapter.c

+
+	if (!mod) {
+		comp_err(dev, "failed to allocate memory for module");
+		goto err;


Would you consider moving the code from the 'err' label to this location?

I'm usually all for "early returns," i.e. for

if (fail) return;

instead of

if (fail) goto err; ... return 0; err: return -EINVAL;

but if there's some cleanup to be done in (all) error cases, I usually prefer a goto. When there's only one error case - well, that's a grey zone for me... But since we're likely to extend this function in the future, I'd rather keep it this way.

Add sof_heap_alloc() and sof_heap_free() to allocate and free memory on a private heap. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Memory, allocated using the module_driver_heap_*() API, should be freed using module_driver_heap_free(), not rfree(). Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Allocate and free memory for module and device objects and for input pins to separate functions to support varying allocation methods. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Add a definition of struct k_heap and SOF k_heap wrappers sof_heap_alloc() and sof_heap_free() for host native builds. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

Add a new mod_alloc_ext() allocation function with support for allocator flags. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

s/allocatd/allocated/ Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

lyakh · 2025-11-05T10:32:12Z

SOFCI TEST

kv2019i · 2025-11-13T19:57:04Z

@lgirdwood @lyakh Hmm, you broke the doxygen CI run with this PR. Now we have this failing a required check in all PRs.

Copilot AI review requested due to automatic review settings November 3, 2025 16:26

lyakh requested review from abonislawski, dabekjakub, dbaluta, iuliana-prodan, jxstelter, kv2019i, lbetlej, lgirdwood, mmaka1, plbossart and ranj063 as code owners November 3, 2025 16:26

lyakh mentioned this pull request Nov 3, 2025

userspace: move further allocations to the module heap (part 2 of #10287) #10302

Merged

Copilot AI reviewed Nov 3, 2025

View reviewed changes

lgirdwood reviewed Nov 3, 2025

View reviewed changes

lyakh force-pushed the user-p1_1 branch 3 times, most recently from 8592072 to 15c2b47 Compare November 4, 2025 09:30

kv2019i reviewed Nov 4, 2025

View reviewed changes

softwarecki reviewed Nov 4, 2025

View reviewed changes

lyakh added 6 commits November 4, 2025 16:22

alloc: add functions to use a private heap

1df93bc

Add sof_heap_alloc() and sof_heap_free() to allocate and free memory on a private heap. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

module-adapter: fix wrong memory freeing

9b9132e

Memory, allocated using the module_driver_heap_*() API, should be freed using module_driver_heap_free(), not rfree(). Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

module-adapter: move allocating and freeing into functions

a564dc5

Allocate and free memory for module and device objects and for input pins to separate functions to support varying allocation methods. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

platform: library: add k_heap* types and functions

90ef190

Add a definition of struct k_heap and SOF k_heap wrappers sof_heap_alloc() and sof_heap_free() for host native builds. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

module: add an allocation function with flags

4fbccef

Add a new mod_alloc_ext() allocation function with support for allocator flags. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

audio: module-adapter: (cosmetic) fix two typos

aa4d87f

s/allocatd/allocated/ Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

softwarecki approved these changes Nov 4, 2025

View reviewed changes

lyakh force-pushed the user-p1_1 branch from 15c2b47 to aa4d87f Compare November 4, 2025 15:25

lyakh changed the title ~~Part 1 of #10302~~ Part 1 of userspace DP subset PR #10302 Nov 5, 2025

kv2019i approved these changes Nov 5, 2025

View reviewed changes

lgirdwood merged commit 9c32699 into thesofproject:main Nov 5, 2025
35 of 42 checks passed

lyakh deleted the user-p1_1 branch November 5, 2025 13:45

kv2019i mentioned this pull request Nov 13, 2025

llext: fix save-restore with no initialized modules #10360

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Part 1 of userspace DP subset PR #10302#10350

Part 1 of userspace DP subset PR #10302#10350
lgirdwood merged 6 commits intothesofproject:mainfrom
lyakh:user-p1_1

lyakh commented Nov 3, 2025

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Nov 3, 2025

Uh oh!

Copilot AI Nov 3, 2025

Uh oh!

Copilot AI Nov 3, 2025

Uh oh!

lgirdwood left a comment

Uh oh!

kv2019i left a comment

Uh oh!

lgirdwood commented Nov 4, 2025

Uh oh!

lgirdwood commented Nov 4, 2025

Uh oh!

softwarecki Nov 4, 2025

Uh oh!

lyakh Nov 5, 2025

Uh oh!

lyakh commented Nov 5, 2025

Uh oh!

Uh oh!

kv2019i commented Nov 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

-	return malloc(bytes);
+	/* Use aligned_alloc to respect the alignment parameter.
+	 * Note: aligned_alloc requires that bytes is a multiple of alignment.
+	 */
+	if (alignment && (bytes % alignment != 0)) {
+		bytes = ((bytes + alignment - 1) / alignment) * alignment;
+	}
+	return aligned_alloc(alignment ? alignment : sizeof(void *), bytes);

	return malloc(bytes);
	return aligned_alloc(alignment, bytes);

Conversation

lyakh commented Nov 3, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

Copilot AI Nov 3, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 3, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 3, 2025

Choose a reason for hiding this comment

Uh oh!

lgirdwood left a comment

Choose a reason for hiding this comment

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

lgirdwood commented Nov 4, 2025

Uh oh!

lgirdwood commented Nov 4, 2025

Uh oh!

softwarecki Nov 4, 2025

Choose a reason for hiding this comment

Uh oh!

lyakh Nov 5, 2025

Choose a reason for hiding this comment

Uh oh!

lyakh commented Nov 5, 2025

Uh oh!

Uh oh!

kv2019i commented Nov 13, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants