diff mbox series

[v4,04/14] util/dsa: Implement DSA task enqueue and dequeue.

Message ID 20240425022117.4035031-5-hao.xiang@linux.dev (mailing list archive)
State New, archived
Headers show
Series Use Intel DSA accelerator to offload zero page checking in multifd live migration. | expand

Commit Message

Hao Xiang April 25, 2024, 2:21 a.m. UTC 
* Use a safe thread queue for DSA task enqueue/dequeue.
* Implement DSA task submission.
* Implement DSA batch task submission.

Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
---
 include/qemu/dsa.h |  28 +++++++
 util/dsa.c         | 201 +++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 229 insertions(+)

Comments

Fabiano Rosas April 25, 2024, 8:55 p.m. UTC | #1 
Hao Xiang <hao.xiang@linux.dev> writes:

> * Use a safe thread queue for DSA task enqueue/dequeue.
> * Implement DSA task submission.
> * Implement DSA batch task submission.
>
> Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
> ---
>  include/qemu/dsa.h |  28 +++++++
>  util/dsa.c         | 201 +++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 229 insertions(+)
>
> diff --git a/include/qemu/dsa.h b/include/qemu/dsa.h
> index f15c05ee85..37cae8d9d2 100644
> --- a/include/qemu/dsa.h
> +++ b/include/qemu/dsa.h
> @@ -13,6 +13,34 @@
>  #include <linux/idxd.h>
>  #include "x86intrin.h"
>  
> +typedef enum DsaTaskType {
> +    DSA_TASK = 0,
> +    DSA_BATCH_TASK
> +} DsaTaskType;
> +
> +typedef enum DsaTaskStatus {
> +    DSA_TASK_READY = 0,
> +    DSA_TASK_PROCESSING,
> +    DSA_TASK_COMPLETION
> +} DsaTaskStatus;
> +
> +typedef void (*dsa_completion_fn)(void *);
> +
> +typedef struct dsa_batch_task {
> +    struct dsa_hw_desc batch_descriptor;
> +    struct dsa_hw_desc *descriptors;
> +    struct dsa_completion_record batch_completion __attribute__((aligned(32)));
> +    struct dsa_completion_record *completions;
> +    struct dsa_device_group *group;
> +    struct dsa_device *device;
> +    dsa_completion_fn completion_callback;
> +    QemuSemaphore sem_task_complete;
> +    DsaTaskType task_type;
> +    DsaTaskStatus status;
> +    int batch_size;
> +    QSIMPLEQ_ENTRY(dsa_batch_task) entry;
> +} dsa_batch_task;
> +
>  /**
>   * @brief Initializes DSA devices.
>   *
> diff --git a/util/dsa.c b/util/dsa.c
> index 05bbf8e31a..75739a1af6 100644
> --- a/util/dsa.c
> +++ b/util/dsa.c
> @@ -244,6 +244,205 @@ dsa_device_group_get_next_device(struct dsa_device_group *group)
>      return &group->dsa_devices[current];
>  }
>  
> +/**
> + * @brief Empties out the DSA task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + */
> +static void
> +dsa_empty_task_queue(struct dsa_device_group *group)
> +{
> +    qemu_mutex_lock(&group->task_queue_lock);
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    while (!QSIMPLEQ_EMPTY(task_queue)) {
> +        QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
> +    }
> +    qemu_mutex_unlock(&group->task_queue_lock);
> +}
> +
> +/**
> + * @brief Adds a task to the DSA task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + * @param context A pointer to the DSA task to enqueue.

This is wrong^

> + *
> + * @return int Zero if successful, otherwise a proper error code.
> + */
> +static int
> +dsa_task_enqueue(struct dsa_device_group *group,
> +                 struct dsa_batch_task *task)
> +{
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    QemuMutex *task_queue_lock = &group->task_queue_lock;
> +    QemuCond *task_queue_cond = &group->task_queue_cond;

It's more idiomatic to not hold any of these in a variable, just access
them directly.

> +
> +    bool notify = false;
> +
> +    qemu_mutex_lock(task_queue_lock);
> +
> +    if (!group->running) {
> +        error_report("DSA: Tried to queue task to stopped device queue.");
> +        qemu_mutex_unlock(task_queue_lock);
> +        return -1;
> +    }
> +
> +    /* The queue is empty. This enqueue operation is a 0->1 transition. */
> +    if (QSIMPLEQ_EMPTY(task_queue)) {
> +        notify = true;
> +    }
> +
> +    QSIMPLEQ_INSERT_TAIL(task_queue, task, entry);
> +
> +    /* We need to notify the waiter for 0->1 transitions. */
> +    if (notify) {
> +        qemu_cond_signal(task_queue_cond);
> +    }
> +
> +    qemu_mutex_unlock(task_queue_lock);
> +
> +    return 0;
> +}
> +
> +/**
> + * @brief Takes a DSA task out of the task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + * @return dsa_batch_task* The DSA task being dequeued.
> + */
> +__attribute__((unused))
> +static struct dsa_batch_task *
> +dsa_task_dequeue(struct dsa_device_group *group)
> +{
> +    struct dsa_batch_task *task = NULL;
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    QemuMutex *task_queue_lock = &group->task_queue_lock;
> +    QemuCond *task_queue_cond = &group->task_queue_cond;

Same here.

> +
> +    qemu_mutex_lock(task_queue_lock);
> +
> +    while (true) {
> +        if (!group->running) {
> +            goto exit;
> +        }
> +        task = QSIMPLEQ_FIRST(task_queue);
> +        if (task != NULL) {
> +            break;
> +        }
> +        qemu_cond_wait(task_queue_cond, task_queue_lock);
> +    }
> +
> +    QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
> +
> +exit:
> +    qemu_mutex_unlock(task_queue_lock);
> +    return task;
> +}
> +
> +/**
> + * @brief Submits a DSA work item to the device work queue.
> + *
> + * @param wq A pointer to the DSA work queue's device memory.
> + * @param descriptor A pointer to the DSA work item descriptor.
> + *
> + * @return Zero if successful, non-zero otherwise.
> + */
> +static int
> +submit_wi_int(void *wq, struct dsa_hw_desc *descriptor)
> +{
> +    uint64_t retry = 0;
> +
> +    _mm_sfence();
> +
> +    while (true) {
> +        if (_enqcmd(wq, descriptor) == 0) {
> +            break;
> +        }
> +        retry++;
> +        if (retry > max_retry_count) {
> +            error_report("Submit work retry %lu times.", retry);
> +            return -1;
> +        }
> +    }
> +
> +    return 0;
> +}
> +
> +/**
> + * @brief Synchronously submits a DSA work item to the
> + *        device work queue.
> + *
> + * @param wq A pointer to the DSA worjk queue's device memory.

s/worjk/work/

> + * @param descriptor A pointer to the DSA work item descriptor.
> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_wi(void *wq, struct dsa_hw_desc *descriptor)
> +{
> +    return submit_wi_int(wq, descriptor);
> +}
> +
> +/**
> + * @brief Asynchronously submits a DSA work item to the
> + *        device work queue.
> + *
> + * @param task A pointer to the buffer zero task.
> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_wi_async(struct dsa_batch_task *task)
> +{
> +    struct dsa_device_group *device_group = task->group;
> +    struct dsa_device *device_instance = task->device;
> +    int ret;
> +
> +    assert(task->task_type == DSA_TASK);
> +
> +    task->status = DSA_TASK_PROCESSING;
> +
> +    ret = submit_wi_int(device_instance->work_queue,
> +                        &task->descriptors[0]);
> +    if (ret != 0) {
> +        return ret;
> +    }
> +
> +    return dsa_task_enqueue(device_group, task);
> +}
> +
> +/**
> + * @brief Asynchronously submits a DSA batch work item to the
> + *        device work queue.
> + *
> + * @param dsa_batch_task A pointer to the batch buffer zero task.

s/buffer zero //

> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_batch_wi_async(struct dsa_batch_task *batch_task)
> +{
> +    struct dsa_device_group *device_group = batch_task->group;
> +    struct dsa_device *device_instance = batch_task->device;
> +    int ret;
> +
> +    assert(batch_task->task_type == DSA_BATCH_TASK);
> +    assert(batch_task->batch_descriptor.desc_count <= batch_task->batch_size);
> +    assert(batch_task->status == DSA_TASK_READY);
> +
> +    batch_task->status = DSA_TASK_PROCESSING;
> +
> +    ret = submit_wi_int(device_instance->work_queue,
> +                        &batch_task->batch_descriptor);
> +    if (ret != 0) {
> +        return ret;
> +    }
> +
> +    return dsa_task_enqueue(device_group, batch_task);
> +}
> +
>  /**
>   * @brief Check if DSA is running.
>   *
> @@ -300,6 +499,8 @@ void dsa_stop(void)
>      if (!group->running) {
>          return;
>      }
> +
> +    dsa_empty_task_queue(group);
>  }
>  
>  /**
Fabiano Rosas April 25, 2024, 9:48 p.m. UTC | #2 
Hao Xiang <hao.xiang@linux.dev> writes:

> * Use a safe thread queue for DSA task enqueue/dequeue.
> * Implement DSA task submission.
> * Implement DSA batch task submission.
>
> Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
> ---
>  include/qemu/dsa.h |  28 +++++++
>  util/dsa.c         | 201 +++++++++++++++++++++++++++++++++++++++++++++
>  2 files changed, 229 insertions(+)
>
> diff --git a/include/qemu/dsa.h b/include/qemu/dsa.h
> index f15c05ee85..37cae8d9d2 100644
> --- a/include/qemu/dsa.h
> +++ b/include/qemu/dsa.h
> @@ -13,6 +13,34 @@
>  #include <linux/idxd.h>
>  #include "x86intrin.h"
>  
> +typedef enum DsaTaskType {
> +    DSA_TASK = 0,
> +    DSA_BATCH_TASK
> +} DsaTaskType;
> +
> +typedef enum DsaTaskStatus {
> +    DSA_TASK_READY = 0,
> +    DSA_TASK_PROCESSING,
> +    DSA_TASK_COMPLETION
> +} DsaTaskStatus;
> +
> +typedef void (*dsa_completion_fn)(void *);
> +
> +typedef struct dsa_batch_task {
> +    struct dsa_hw_desc batch_descriptor;
> +    struct dsa_hw_desc *descriptors;
> +    struct dsa_completion_record batch_completion __attribute__((aligned(32)));
> +    struct dsa_completion_record *completions;
> +    struct dsa_device_group *group;
> +    struct dsa_device *device;
> +    dsa_completion_fn completion_callback;
> +    QemuSemaphore sem_task_complete;
> +    DsaTaskType task_type;
> +    DsaTaskStatus status;
> +    int batch_size;
> +    QSIMPLEQ_ENTRY(dsa_batch_task) entry;
> +} dsa_batch_task;
> +
>  /**
>   * @brief Initializes DSA devices.
>   *
> diff --git a/util/dsa.c b/util/dsa.c
> index 05bbf8e31a..75739a1af6 100644
> --- a/util/dsa.c
> +++ b/util/dsa.c
> @@ -244,6 +244,205 @@ dsa_device_group_get_next_device(struct dsa_device_group *group)
>      return &group->dsa_devices[current];
>  }
>  
> +/**
> + * @brief Empties out the DSA task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + */
> +static void
> +dsa_empty_task_queue(struct dsa_device_group *group)
> +{
> +    qemu_mutex_lock(&group->task_queue_lock);
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    while (!QSIMPLEQ_EMPTY(task_queue)) {
> +        QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
> +    }
> +    qemu_mutex_unlock(&group->task_queue_lock);
> +}
> +
> +/**
> + * @brief Adds a task to the DSA task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + * @param context A pointer to the DSA task to enqueue.
> + *
> + * @return int Zero if successful, otherwise a proper error code.
> + */
> +static int
> +dsa_task_enqueue(struct dsa_device_group *group,
> +                 struct dsa_batch_task *task)
> +{
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    QemuMutex *task_queue_lock = &group->task_queue_lock;
> +    QemuCond *task_queue_cond = &group->task_queue_cond;
> +
> +    bool notify = false;
> +
> +    qemu_mutex_lock(task_queue_lock);
> +
> +    if (!group->running) {
> +        error_report("DSA: Tried to queue task to stopped device queue.");
> +        qemu_mutex_unlock(task_queue_lock);
> +        return -1;
> +    }
> +
> +    /* The queue is empty. This enqueue operation is a 0->1 transition. */
> +    if (QSIMPLEQ_EMPTY(task_queue)) {
> +        notify = true;
> +    }
> +
> +    QSIMPLEQ_INSERT_TAIL(task_queue, task, entry);
> +
> +    /* We need to notify the waiter for 0->1 transitions. */
> +    if (notify) {
> +        qemu_cond_signal(task_queue_cond);
> +    }
> +
> +    qemu_mutex_unlock(task_queue_lock);
> +
> +    return 0;
> +}
> +
> +/**
> + * @brief Takes a DSA task out of the task queue.
> + *
> + * @param group A pointer to the DSA device group.
> + * @return dsa_batch_task* The DSA task being dequeued.
> + */
> +__attribute__((unused))
> +static struct dsa_batch_task *
> +dsa_task_dequeue(struct dsa_device_group *group)
> +{
> +    struct dsa_batch_task *task = NULL;
> +    dsa_task_queue *task_queue = &group->task_queue;
> +    QemuMutex *task_queue_lock = &group->task_queue_lock;
> +    QemuCond *task_queue_cond = &group->task_queue_cond;
> +
> +    qemu_mutex_lock(task_queue_lock);
> +
> +    while (true) {
> +        if (!group->running) {
> +            goto exit;
> +        }
> +        task = QSIMPLEQ_FIRST(task_queue);
> +        if (task != NULL) {
> +            break;
> +        }
> +        qemu_cond_wait(task_queue_cond, task_queue_lock);
> +    }
> +
> +    QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
> +
> +exit:
> +    qemu_mutex_unlock(task_queue_lock);
> +    return task;
> +}
> +
> +/**
> + * @brief Submits a DSA work item to the device work queue.
> + *
> + * @param wq A pointer to the DSA work queue's device memory.
> + * @param descriptor A pointer to the DSA work item descriptor.
> + *
> + * @return Zero if successful, non-zero otherwise.
> + */
> +static int
> +submit_wi_int(void *wq, struct dsa_hw_desc *descriptor)
> +{
> +    uint64_t retry = 0;
> +
> +    _mm_sfence();
> +
> +    while (true) {
> +        if (_enqcmd(wq, descriptor) == 0) {
> +            break;
> +        }
> +        retry++;
> +        if (retry > max_retry_count) {

You missed my comment in v2 that max_retry_count is UINT64_MAX.

> +            error_report("Submit work retry %lu times.", retry);
> +            return -1;
> +        }
> +    }
> +
> +    return 0;
> +}
> +
> +/**
> + * @brief Synchronously submits a DSA work item to the
> + *        device work queue.
> + *
> + * @param wq A pointer to the DSA worjk queue's device memory.
> + * @param descriptor A pointer to the DSA work item descriptor.
> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_wi(void *wq, struct dsa_hw_desc *descriptor)
> +{
> +    return submit_wi_int(wq, descriptor);
> +}
> +
> +/**
> + * @brief Asynchronously submits a DSA work item to the
> + *        device work queue.
> + *
> + * @param task A pointer to the buffer zero task.
> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_wi_async(struct dsa_batch_task *task)
> +{
> +    struct dsa_device_group *device_group = task->group;
> +    struct dsa_device *device_instance = task->device;
> +    int ret;
> +
> +    assert(task->task_type == DSA_TASK);
> +
> +    task->status = DSA_TASK_PROCESSING;
> +
> +    ret = submit_wi_int(device_instance->work_queue,
> +                        &task->descriptors[0]);
> +    if (ret != 0) {
> +        return ret;
> +    }
> +
> +    return dsa_task_enqueue(device_group, task);
> +}
> +
> +/**
> + * @brief Asynchronously submits a DSA batch work item to the
> + *        device work queue.
> + *
> + * @param dsa_batch_task A pointer to the batch buffer zero task.
> + *
> + * @return int Zero if successful, non-zero otherwise.
> + */
> +__attribute__((unused))
> +static int
> +submit_batch_wi_async(struct dsa_batch_task *batch_task)
> +{
> +    struct dsa_device_group *device_group = batch_task->group;
> +    struct dsa_device *device_instance = batch_task->device;
> +    int ret;
> +
> +    assert(batch_task->task_type == DSA_BATCH_TASK);
> +    assert(batch_task->batch_descriptor.desc_count <= batch_task->batch_size);
> +    assert(batch_task->status == DSA_TASK_READY);
> +
> +    batch_task->status = DSA_TASK_PROCESSING;
> +
> +    ret = submit_wi_int(device_instance->work_queue,
> +                        &batch_task->batch_descriptor);
> +    if (ret != 0) {
> +        return ret;
> +    }
> +
> +    return dsa_task_enqueue(device_group, batch_task);
> +}
> +
>  /**
>   * @brief Check if DSA is running.
>   *
> @@ -300,6 +499,8 @@ void dsa_stop(void)
>      if (!group->running) {
>          return;
>      }
> +
> +    dsa_empty_task_queue(group);
>  }
>  
>  /**
diff mbox series

Patch

diff --git a/include/qemu/dsa.h b/include/qemu/dsa.h
index f15c05ee85..37cae8d9d2 100644
--- a/include/qemu/dsa.h
+++ b/include/qemu/dsa.h
@@ -13,6 +13,34 @@ 
 #include <linux/idxd.h>
 #include "x86intrin.h"
 
+typedef enum DsaTaskType {
+    DSA_TASK = 0,
+    DSA_BATCH_TASK
+} DsaTaskType;
+
+typedef enum DsaTaskStatus {
+    DSA_TASK_READY = 0,
+    DSA_TASK_PROCESSING,
+    DSA_TASK_COMPLETION
+} DsaTaskStatus;
+
+typedef void (*dsa_completion_fn)(void *);
+
+typedef struct dsa_batch_task {
+    struct dsa_hw_desc batch_descriptor;
+    struct dsa_hw_desc *descriptors;
+    struct dsa_completion_record batch_completion __attribute__((aligned(32)));
+    struct dsa_completion_record *completions;
+    struct dsa_device_group *group;
+    struct dsa_device *device;
+    dsa_completion_fn completion_callback;
+    QemuSemaphore sem_task_complete;
+    DsaTaskType task_type;
+    DsaTaskStatus status;
+    int batch_size;
+    QSIMPLEQ_ENTRY(dsa_batch_task) entry;
+} dsa_batch_task;
+
 /**
  * @brief Initializes DSA devices.
  *
diff --git a/util/dsa.c b/util/dsa.c
index 05bbf8e31a..75739a1af6 100644
--- a/util/dsa.c
+++ b/util/dsa.c
@@ -244,6 +244,205 @@  dsa_device_group_get_next_device(struct dsa_device_group *group)
     return &group->dsa_devices[current];
 }
 
+/**
+ * @brief Empties out the DSA task queue.
+ *
+ * @param group A pointer to the DSA device group.
+ */
+static void
+dsa_empty_task_queue(struct dsa_device_group *group)
+{
+    qemu_mutex_lock(&group->task_queue_lock);
+    dsa_task_queue *task_queue = &group->task_queue;
+    while (!QSIMPLEQ_EMPTY(task_queue)) {
+        QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
+    }
+    qemu_mutex_unlock(&group->task_queue_lock);
+}
+
+/**
+ * @brief Adds a task to the DSA task queue.
+ *
+ * @param group A pointer to the DSA device group.
+ * @param context A pointer to the DSA task to enqueue.
+ *
+ * @return int Zero if successful, otherwise a proper error code.
+ */
+static int
+dsa_task_enqueue(struct dsa_device_group *group,
+                 struct dsa_batch_task *task)
+{
+    dsa_task_queue *task_queue = &group->task_queue;
+    QemuMutex *task_queue_lock = &group->task_queue_lock;
+    QemuCond *task_queue_cond = &group->task_queue_cond;
+
+    bool notify = false;
+
+    qemu_mutex_lock(task_queue_lock);
+
+    if (!group->running) {
+        error_report("DSA: Tried to queue task to stopped device queue.");
+        qemu_mutex_unlock(task_queue_lock);
+        return -1;
+    }
+
+    /* The queue is empty. This enqueue operation is a 0->1 transition. */
+    if (QSIMPLEQ_EMPTY(task_queue)) {
+        notify = true;
+    }
+
+    QSIMPLEQ_INSERT_TAIL(task_queue, task, entry);
+
+    /* We need to notify the waiter for 0->1 transitions. */
+    if (notify) {
+        qemu_cond_signal(task_queue_cond);
+    }
+
+    qemu_mutex_unlock(task_queue_lock);
+
+    return 0;
+}
+
+/**
+ * @brief Takes a DSA task out of the task queue.
+ *
+ * @param group A pointer to the DSA device group.
+ * @return dsa_batch_task* The DSA task being dequeued.
+ */
+__attribute__((unused))
+static struct dsa_batch_task *
+dsa_task_dequeue(struct dsa_device_group *group)
+{
+    struct dsa_batch_task *task = NULL;
+    dsa_task_queue *task_queue = &group->task_queue;
+    QemuMutex *task_queue_lock = &group->task_queue_lock;
+    QemuCond *task_queue_cond = &group->task_queue_cond;
+
+    qemu_mutex_lock(task_queue_lock);
+
+    while (true) {
+        if (!group->running) {
+            goto exit;
+        }
+        task = QSIMPLEQ_FIRST(task_queue);
+        if (task != NULL) {
+            break;
+        }
+        qemu_cond_wait(task_queue_cond, task_queue_lock);
+    }
+
+    QSIMPLEQ_REMOVE_HEAD(task_queue, entry);
+
+exit:
+    qemu_mutex_unlock(task_queue_lock);
+    return task;
+}
+
+/**
+ * @brief Submits a DSA work item to the device work queue.
+ *
+ * @param wq A pointer to the DSA work queue's device memory.
+ * @param descriptor A pointer to the DSA work item descriptor.
+ *
+ * @return Zero if successful, non-zero otherwise.
+ */
+static int
+submit_wi_int(void *wq, struct dsa_hw_desc *descriptor)
+{
+    uint64_t retry = 0;
+
+    _mm_sfence();
+
+    while (true) {
+        if (_enqcmd(wq, descriptor) == 0) {
+            break;
+        }
+        retry++;
+        if (retry > max_retry_count) {
+            error_report("Submit work retry %lu times.", retry);
+            return -1;
+        }
+    }
+
+    return 0;
+}
+
+/**
+ * @brief Synchronously submits a DSA work item to the
+ *        device work queue.
+ *
+ * @param wq A pointer to the DSA worjk queue's device memory.
+ * @param descriptor A pointer to the DSA work item descriptor.
+ *
+ * @return int Zero if successful, non-zero otherwise.
+ */
+__attribute__((unused))
+static int
+submit_wi(void *wq, struct dsa_hw_desc *descriptor)
+{
+    return submit_wi_int(wq, descriptor);
+}
+
+/**
+ * @brief Asynchronously submits a DSA work item to the
+ *        device work queue.
+ *
+ * @param task A pointer to the buffer zero task.
+ *
+ * @return int Zero if successful, non-zero otherwise.
+ */
+__attribute__((unused))
+static int
+submit_wi_async(struct dsa_batch_task *task)
+{
+    struct dsa_device_group *device_group = task->group;
+    struct dsa_device *device_instance = task->device;
+    int ret;
+
+    assert(task->task_type == DSA_TASK);
+
+    task->status = DSA_TASK_PROCESSING;
+
+    ret = submit_wi_int(device_instance->work_queue,
+                        &task->descriptors[0]);
+    if (ret != 0) {
+        return ret;
+    }
+
+    return dsa_task_enqueue(device_group, task);
+}
+
+/**
+ * @brief Asynchronously submits a DSA batch work item to the
+ *        device work queue.
+ *
+ * @param dsa_batch_task A pointer to the batch buffer zero task.
+ *
+ * @return int Zero if successful, non-zero otherwise.
+ */
+__attribute__((unused))
+static int
+submit_batch_wi_async(struct dsa_batch_task *batch_task)
+{
+    struct dsa_device_group *device_group = batch_task->group;
+    struct dsa_device *device_instance = batch_task->device;
+    int ret;
+
+    assert(batch_task->task_type == DSA_BATCH_TASK);
+    assert(batch_task->batch_descriptor.desc_count <= batch_task->batch_size);
+    assert(batch_task->status == DSA_TASK_READY);
+
+    batch_task->status = DSA_TASK_PROCESSING;
+
+    ret = submit_wi_int(device_instance->work_queue,
+                        &batch_task->batch_descriptor);
+    if (ret != 0) {
+        return ret;
+    }
+
+    return dsa_task_enqueue(device_group, batch_task);
+}
+
 /**
  * @brief Check if DSA is running.
  *
@@ -300,6 +499,8 @@  void dsa_stop(void)
     if (!group->running) {
         return;
     }
+
+    dsa_empty_task_queue(group);
 }
 
 /**