On Fri, Sep 8, 2023 at 12:20 PM Shashank Sharma <shashank.sharma@xxxxxxx> wrote:
>
> This patch adds new functions to map/unmap a usermode queue into
> the FW, using the MES ring. As soon as this mapping is done, the
> queue would be considered ready to accept the workload.
>
> V1: Addressed review comments from Alex on the RFC patch series
> - Map/Unmap should be IP specific.
> V2:
> Addressed review comments from Christian:
> - Fix the wptr_mc_addr calculation (moved into another patch)
> Addressed review comments from Alex:
> - Do not add fptrs for map/unmap
>
> V3: Integration with doorbell manager
> V4: Rebase
> V5: Use gfx_v11_0 for function names (Alex)
> V6: Removed queue->proc/gang/fw_ctx_address variables and doing the
> address calculations locally to keep the queue structure GEN
> independent (Alex)
>
> Cc: Alex Deucher <alexander.deucher@xxxxxxx>
> Cc: Christian Koenig <christian.koenig@xxxxxxx>
> Signed-off-by: Shashank Sharma <shashank.sharma@xxxxxxx>
> ---
> drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 72 ++++++++++++++++++++++++++
> 1 file changed, 72 insertions(+)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> index 8ffb5dee72a9..e266674e0d44 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
> @@ -6427,6 +6427,67 @@ const struct amdgpu_ip_block_version gfx_v11_0_ip_block =
> .funcs = &gfx_v11_0_ip_funcs,
> };
>
> +static void gfx_v11_0_userq_unmap(struct amdgpu_userq_mgr *uq_mgr,
> + struct amdgpu_usermode_queue *queue)
> +{
> + struct amdgpu_device *adev = uq_mgr->adev;
> + struct mes_remove_queue_input queue_input;
> + struct amdgpu_userq_obj *ctx = &queue->fw_obj;
> + int r;
> +
> + memset(&queue_input, 0x0, sizeof(struct mes_remove_queue_input));
> + queue_input.doorbell_offset = queue->doorbell_index;
> + queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
> +
> + amdgpu_mes_lock(&adev->mes);
> + r = adev->mes.funcs->remove_hw_queue(&adev->mes, &queue_input);
> + amdgpu_mes_unlock(&adev->mes);
> + if (r)
> + DRM_ERROR("Failed to unmap queue in HW, err (%d)\n", r);
> +}
> +
> +static int gfx_v11_0_userq_map(struct amdgpu_userq_mgr *uq_mgr,
> + struct amdgpu_usermode_queue *queue,
> + struct amdgpu_mqd_prop *userq_props)
> +{
> + struct amdgpu_device *adev = uq_mgr->adev;
> + struct amdgpu_userq_obj *ctx = &queue->fw_obj;
> + struct mes_add_queue_input queue_input;
> + int r;
> +
> + memset(&queue_input, 0x0, sizeof(struct mes_add_queue_input));
> +
> + queue_input.process_va_start = 0;
> + queue_input.process_va_end = (adev->vm_manager.max_pfn - 1) << AMDGPU_GPU_PAGE_SHIFT;
> + queue_input.process_quantum = 100000; /* 10ms */
> + queue_input.gang_quantum = 10000; /* 1ms */
> + queue_input.paging = false;
> +
> + queue_input.process_context_addr = ctx->gpu_addr;
> + queue_input.gang_context_addr = ctx->gpu_addr + AMDGPU_USERQ_PROC_CTX_SZ;
> + queue_input.inprocess_gang_priority = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
> + queue_input.gang_global_priority_level = AMDGPU_MES_PRIORITY_LEVEL_NORMAL;
I can't remember, did we have a plan for priority handling?
Compositors would want high priority queues for example.
Alex
> +
> + queue_input.process_id = queue->vm->pasid;
> + queue_input.queue_type = queue->queue_type;
> + queue_input.mqd_addr = queue->mqd.gpu_addr;
> + queue_input.wptr_addr = userq_props->wptr_gpu_addr;
> + queue_input.queue_size = userq_props->queue_size >> 2;
> + queue_input.doorbell_offset = userq_props->doorbell_index;
> + queue_input.page_table_base_addr = amdgpu_gmc_pd_addr(queue->vm->root.bo);
> +
> + amdgpu_mes_lock(&adev->mes);
> + r = adev->mes.funcs->add_hw_queue(&adev->mes, &queue_input);
> + amdgpu_mes_unlock(&adev->mes);
> + if (r) {
> + DRM_ERROR("Failed to map queue in HW, err (%d)\n", r);
> + return r;
> + }
> +
> + DRM_DEBUG_DRIVER("Queue (doorbell:%d) mapped successfully\n", userq_props->doorbell_index);
> + return 0;
> +}
> +
> static void gfx_v11_0_userq_destroy_ctx_space(struct amdgpu_userq_mgr *uq_mgr,
> struct amdgpu_usermode_queue *queue)
> {
> @@ -6540,8 +6601,18 @@ static int gfx_v11_0_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr,
> goto free_mqd;
> }
>
> + /* Map userqueue into FW using MES */
> + r = gfx_v11_0_userq_map(uq_mgr, queue, &userq_props);
> + if (r) {
> + DRM_ERROR("Failed to init MQD\n");
> + goto free_ctx;
> + }
> +
> return 0;
>
> +free_ctx:
> + gfx_v11_0_userq_destroy_ctx_space(uq_mgr, queue);
> +
> free_mqd:
> amdgpu_bo_free_kernel(&queue->mqd.obj, &queue->mqd.gpu_addr, &queue->mqd.cpu_ptr);
> return r;
> @@ -6552,6 +6623,7 @@ gfx_v11_0_userq_mqd_destroy(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_userm
> {
> struct amdgpu_userq_obj *mqd = &queue->mqd;
>
> + gfx_v11_0_userq_unmap(uq_mgr, queue);
> gfx_v11_0_userq_destroy_ctx_space(uq_mgr, queue);
> amdgpu_bo_free_kernel(&mqd->obj, &mqd->gpu_addr, &mqd->cpu_ptr);
> }
> --
> 2.42.0
>
