audio: dai-zephyr: enable user-space use of DAI, part 1 by kv2019i · Pull Request #10801 · thesofproject/sof

kv2019i · 2026-05-21T18:22:29Z

A series of of patches to enable dai-zephyr component to be used in user-space. This includes:

addition of new sof_umutex.h rtos interface that is used to implement DAI property locking (instead of spinlocks)
fixes to sof_dma.h (ability to use SG lists with loops)
make dai_get() work in user-space
fix use of dai_get_properties() to be safe to be used from user-space
modify heap allocations in dai-zephyr so that they will work also when building for user-space LL

Similar changes to host-zephyr are in #10799

Allow a non-null pointer at the end of the DMA transfer block list, if and only if it points to the first entry in the block list. The SOF DAI module sets the DMA transfers blocks like this and this change is required to use DAI module from user-space. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

The dai_get()/dai_put() provide a helper to access DAI devices. When used in user-space, the wrapper struct should be created in user-space memory. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2026-05-21T18:23:13Z

For context, part of #10558

Copilot

Pull request overview

This PR enables the dai-zephyr component (and related infrastructure) to be usable from Zephyr user-space threads by introducing a user-space mutex abstraction, updating syscall verification, and adjusting allocations to work with user-space heaps/memory domains.

Changes:

Add sof_umutex API (Zephyr + POSIX) with syscall plumbing and a user-space test.
Update dai-zephyr to use sof_umutex for property protection and to use heap-aware allocation/free paths.
Update the SOF DMA syscall deep-copy logic to support cyclic (looping) SG block lists.

Reviewed changes

Copilot reviewed 16 out of 16 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
zephyr/test/userspace/test_umutex.c	Adds a user-space thread test for `sof_umutex` init/lock/unlock/free.
zephyr/test/CMakeLists.txt	Wires new user-space tests into the Zephyr test build.
zephyr/syscall/umutex.c	Adds syscall verification handlers for `sof_umutex_*`.
zephyr/syscall/sof_dma.c	Extends DMA block list deep-copy to allow looping SG lists.
zephyr/lib/umutex.c	Implements kernel-side `sof_umutex` using dynamically allocated `k_mutex`.
zephyr/Kconfig	Introduces Kconfig switches for userspace alloc + mutex interfaces and selects mutex for userspace LL.
zephyr/include/rtos/umutex.h	Declares the `sof_umutex` API and syscall wrappers.
zephyr/include/rtos/mutex.h	Exposes `sof_umutex` via the RTOS mutex umbrella header.
zephyr/CMakeLists.txt	Adds `umutex` implementation and syscall sources/headers to the Zephyr build.
src/lib/dai.c	Updates `dai_get()`/`dai_put()` allocation/free to use `sof_heap_alloc/free` (userspace LL aware).
src/ipc/ipc4/dai.c	Uses heap-aware allocation/free for `dd->dai_spec_config`.
src/include/sof/lib/dai-zephyr.h	Switches DAI property lock to `sof_umutex` and adds `mod_alloc_ctx` to `dai_data`.
src/include/sof/audio/component.h	Moves `mod_alloc_ctx` definition earlier to satisfy DAI header needs.
src/audio/dai-zephyr.c	Replaces spinlock-based property protection with `sof_umutex` and updates multiple allocations to use `alloc_ctx.heap`.
posix/include/rtos/umutex.h	Adds POSIX header wrapper for `sof_umutex`.
posix/include/rtos/mutex.h	Adds a POSIX stub implementation of `sof_umutex`.

kv2019i · 2026-05-21T18:44:51Z

+static inline int z_vrfy_sof_umutex_lock(struct sof_umutex *umutex,
+					    k_timeout_t timeout)
+{
+	K_OOPS(K_SYSCALL_MEMORY_WRITE(umutex, sizeof(struct sof_umutex)));
+	return z_impl_sof_umutex_lock(umutex, timeout);
+}
+#include <zephyr/syscalls/sof_umutex_lock_mrsh.c>
+
+static inline int z_vrfy_sof_umutex_unlock(struct sof_umutex *umutex)
+{
+	K_OOPS(K_SYSCALL_MEMORY_WRITE(umutex, sizeof(struct sof_umutex)));
+	return z_impl_sof_umutex_unlock(umutex);
+}
+#include <zephyr/syscalls/sof_umutex_unlock_mrsh.c>
+
+static inline void z_vrfy_sof_umutex_free(struct sof_umutex *umutex)
+{
+	K_OOPS(K_SYSCALL_MEMORY_WRITE(umutex, sizeof(struct sof_umutex)));
+	z_impl_sof_umutex_free(umutex);


This is a valid concern. The verification is similar to zephyr sys_mutex() but it has additional step to look up the k_object() that is not here. Let me mark as draft to ensure this is addressed before merge.

+int z_impl_sof_umutex_init(struct sof_umutex *umutex)
+{
+	struct k_mutex *m;
+	int ret;
+
+	m = k_object_alloc(K_OBJ_MUTEX);
+	if (m == NULL) {
+		return -ENOMEM;
+	}
+
+	ret = k_mutex_init(m);
+	if (ret) {
+		k_object_free(m);
+		return ret;
+	}
+
+	umutex->mutex = m;
+	return 0;


kv2019i · 2026-05-22T11:48:50Z

 		dd->dma_buffer = NULL;
-		dma_sg_free(NULL, &config->elem_array);
-		rfree(dd->z_config);
+		dma_sg_free(dd->alloc_ctx.heap, &config->elem_array);


This is a valid problem, but problem exists in existing code, so I prefer to handle this case in a follow-up PR. If I need to revise this series, I will add a new commit to address this.
UPDATE: actually, this is not a real leak. Only error case where dd->z_config is allocated and non-zero, is the case where dai_set_dma_config() fails. But if that fails, nothing is allocated to dd->z_config->head_block either. So current code in main is safe.

lyakh

Yes, if possible, the kobject should be checked

lyakh · 2026-05-22T06:51:39Z

 	dev->ipc_config = *config;

-	dd = rzalloc(SOF_MEM_FLAG_USER | SOF_MEM_FLAG_COHERENT, sizeof(*dd));
+	dd = sof_heap_alloc(heap, SOF_MEM_FLAG_USER | SOF_MEM_FLAG_COHERENT, sizeof(*dd), 0);


just NULL

I went back and forth between "heap" variable and just putting NULL, but in the end I think this is more readable and makes it explicit same value needs to be passed to alloc and free in this function. Compilation output should be the same in both cases.

Reorder redefinitions to ensure "struct mod_alloc_ctx" is defined before lib/dai.h is included. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

Convert all memory allocations to use the sof_heap_alloc() interface and pass the dai_data specific heap object. This makes dai-zephyr code compatible with use from user-space, but does not affect kernel space use. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2026-05-22T11:42:47Z

V2 pushed:

drop the changes to DAI properties locking, this requires some more thought'n'work (now reviewing the origina commit in copier: fix allocation and configuration size #6930 where this locking was dded)
also dropped umutex.h as it was needed for the properties
keep the remaining patches

kv2019i

Replied to code review comments. V2 should be ready.

kv2019i · 2026-05-22T11:44:16Z

 	dev->ipc_config = *config;

-	dd = rzalloc(SOF_MEM_FLAG_USER | SOF_MEM_FLAG_COHERENT, sizeof(*dd));
+	dd = sof_heap_alloc(heap, SOF_MEM_FLAG_USER | SOF_MEM_FLAG_COHERENT, sizeof(*dd), 0);


I went back and forth between "heap" variable and just putting NULL, but in the end I think this is more readable and makes it explicit same value needs to be passed to alloc and free in this function. Compilation output should be the same in both cases.

kv2019i · 2026-05-22T11:48:50Z

 		dd->dma_buffer = NULL;
-		dma_sg_free(NULL, &config->elem_array);
-		rfree(dd->z_config);
+		dma_sg_free(dd->alloc_ctx.heap, &config->elem_array);


This is a valid problem, but problem exists in existing code, so I prefer to handle this case in a follow-up PR. If I need to revise this series, I will add a new commit to address this.
UPDATE: actually, this is not a real leak. Only error case where dd->z_config is allocated and non-zero, is the case where dai_set_dma_config() fails. But if that fails, nothing is allocated to dd->z_config->head_block either. So current code in main is safe.

kv2019i added 2 commits May 21, 2026 21:14

lib: dai: make dai_get() and dai_put() compatible with user-space

ff0bc0c

The dai_get()/dai_put() provide a helper to access DAI devices. When used in user-space, the wrapper struct should be created in user-space memory. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

Copilot AI review requested due to automatic review settings May 21, 2026 18:22

kv2019i requested review from abonislawski, dbaluta, iuliana-prodan, lbetlej, lgirdwood, lyakh, mmaka1 and plbossart as code owners May 21, 2026 18:22

kv2019i requested a review from jsarha May 21, 2026 18:22

Copilot started reviewing on behalf of kv2019i May 21, 2026 18:26 View session

kv2019i commented May 21, 2026

View reviewed changes

Comment thread zephyr/Kconfig Outdated

Comment thread zephyr/test/CMakeLists.txt Outdated

Copilot AI reviewed May 21, 2026

View reviewed changes

kv2019i marked this pull request as draft May 21, 2026 18:44

lyakh reviewed May 22, 2026

View reviewed changes

kv2019i added 2 commits May 22, 2026 14:00

audio: component: reorder include order for lib/dai.h

e6d1761

Reorder redefinitions to ensure "struct mod_alloc_ctx" is defined before lib/dai.h is included. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i force-pushed the 202605-host-dai-user branch from 9fdefeb to 37702da Compare May 22, 2026 11:26

kv2019i changed the title ~~audio: dai-zephyr: enable user-space use of DAI~~ audio: dai-zephyr: enable user-space use of DAI, part 1 May 22, 2026

kv2019i marked this pull request as ready for review May 22, 2026 11:43

kv2019i commented May 22, 2026

View reviewed changes

kv2019i mentioned this pull request May 22, 2026

audio: host-zephyr: make component usable from user-space #10799

Open

Conversation

kv2019i commented May 21, 2026

Uh oh!

kv2019i commented May 21, 2026

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

kv2019i May 21, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

kv2019i May 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lyakh left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

lyakh May 22, 2026

Choose a reason for hiding this comment

Uh oh!

kv2019i May 22, 2026

Choose a reason for hiding this comment

Uh oh!

kv2019i commented May 22, 2026

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

kv2019i May 22, 2026

Choose a reason for hiding this comment

Uh oh!

kv2019i May 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

kv2019i May 22, 2026 •

edited

Loading

kv2019i May 22, 2026 •

edited

Loading