[v7 04/11] drm/amdgpu: Add helper functions for user queue preemption and restoration
Alex Deucher
alexdeucher at gmail.com
Fri Aug 8 19:49:25 UTC 2025
On Tue, Aug 5, 2025 at 10:39 PM Jesse.Zhang <Jesse.Zhang at amd.com> wrote:
>
> This patch introduces two new helper functions to manage user queue
> preemption and restoration:
>
> 1. amdgpu_userqueue_preempt_helper() - Handles preempting a user queue
> by calling the appropriate IP-specific preempt function and updating
> the queue state. If preemption fails, it marks the queue as hung and
> triggers queue detection/reset.
>
> 2. amdgpu_userqueue_restore_helper() - Handles restoring a preempted
> user queue by calling the IP-specific restore function. On failure,
> it marks the queue as hung and triggers a GPU reset.
>
> The helpers properly manage queue state transitions between MAPPED,
> PREEMPTED, and HUNG states, and handle error cases by initiating
> appropriate recovery actions.
>
> These functions will be used by subsequent patches to implement
> user queue preemption support in the driver.
>
> Signed-off-by: Alex Deucher <alexander.deucher at amd.com>
> Signed-off-by: Jesse Zhang <Jesse.Zhang at amd.com>
> ---
> drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c | 48 +++++++++++++++++++++++
> 1 file changed, 48 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
> index aac0de86f3e8..1496544cd1c2 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c
> @@ -44,6 +44,54 @@ u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev)
> return userq_ip_mask;
> }
>
> +static int
> +amdgpu_userqueue_preempt_helper(struct amdgpu_userq_mgr *uq_mgr,
> + struct amdgpu_usermode_queue *queue)
> +{
> + struct amdgpu_device *adev = uq_mgr->adev;
> + const struct amdgpu_userq_funcs *userq_funcs =
> + adev->userq_funcs[queue->queue_type];
> + int r = 0;
> +
> + if (queue->state == AMDGPU_USERQ_STATE_MAPPED) {
> + r = userq_funcs->preempt(uq_mgr, queue);
> + if (r) {
> + amdgpu_userq_detect_and_reset_queues(uq_mgr);
This function isn't available yet. This should be part of a later patch.
> + queue->state = AMDGPU_USERQ_STATE_HUNG;
> + } else {
> + queue->state = AMDGPU_USERQ_STATE_PREEMPTED;
> + }
> + }
> +
> + return r;
> +}
> +
> +static int
> +amdgpu_userqueue_restore_helper(struct amdgpu_userq_mgr *uq_mgr,
> + struct amdgpu_usermode_queue *queue)
> +{
> + struct amdgpu_device *adev = uq_mgr->adev;
> + const struct amdgpu_userq_funcs *userq_funcs =
> + adev->userq_funcs[queue->queue_type];
> + bool gpu_reset = false;
> + int r = 0;
> +
> + if (queue->state == AMDGPU_USERQ_STATE_PREEMPTED) {
> + r = userq_funcs->restore(uq_mgr, queue);
> + if (r) {
> + queue->state = AMDGPU_USERQ_STATE_HUNG;
> + gpu_reset = true;
> + } else {
> + queue->state = AMDGPU_USERQ_STATE_MAPPED;
> + }
> + }
> +
> + if (gpu_reset)
> + amdgpu_userq_gpu_reset(adev);
Same here.
Alex
> +
> + return r;
> +}
> +
> static int
> amdgpu_userq_unmap_helper(struct amdgpu_userq_mgr *uq_mgr,
> struct amdgpu_usermode_queue *queue)
> --
> 2.49.0
>
More information about the amd-gfx
mailing list