drm/xe: Kill xe_device_mem_access_{get*,put}

Let's simply convert all the current callers towards direct
xe_pm_runtime access and remove this extra layer of indirection.

No functional change is expected with this patch since
xe_mem_access_get was already using the xe_pm_runtime_get_noresume
at this point.

v2: Convert all the current callers instead of a big refactor
at once.

v3: - Rebased
    - Squashed the GSC/HDCP
    - Added a new case: sriov_pf_policy
    - Improved commit message to highlight that
      there's no functional change in this patch.

Reviewed-by: Matthew Auld <matthew.auld@intel.com> #v2
Cc: Suraj Kandpal <suraj.kandpal@intel.com>
Cc: Michal Wajdeczko <michal.wajdeczko@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Reviewed-by: Suraj Kandpal <suraj.kandpal@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20240418143049.43231-1-rodrigo.vivi@intel.com

drivers/gpu/drm/xe/display/xe_fb_pin.c

@@ -10,6 +10,7 @@
 #include "intel_fb_pin.h"
 #include "xe_ggtt.h"
 #include "xe_gt.h"
+#include "xe_pm.h"
 
 #include <drm/ttm/ttm_bo.h>
 
@@ -193,7 +194,7 @@ static int __xe_pin_fb_vma_ggtt(struct intel_framebuffer *fb,
 	/* TODO: Consider sharing framebuffer mapping?
 	 * embed i915_vma inside intel_framebuffer
 	 */
-	xe_device_mem_access_get(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_get_noresume(tile_to_xe(ggtt->tile));
 	ret = mutex_lock_interruptible(&ggtt->lock);
 	if (ret)
 		goto out;
@@ -244,7 +245,7 @@ static int __xe_pin_fb_vma_ggtt(struct intel_framebuffer *fb,
 out_unlock:
 	mutex_unlock(&ggtt->lock);
 out:
-	xe_device_mem_access_put(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_put(tile_to_xe(ggtt->tile));
 	return ret;
 }
 

drivers/gpu/drm/xe/display/xe_hdcp_gsc.c

@@ -215,7 +215,7 @@ ssize_t intel_hdcp_gsc_msg_send(struct xe_device *xe, u8 *msg_in,
 	addr_out_off = PAGE_SIZE;
 
 	host_session_id = xe_gsc_create_host_session_id();
-	xe_device_mem_access_get(xe);
+	xe_pm_runtime_get_noresume(xe);
 	addr_in_wr_off = xe_gsc_emit_header(xe, &hdcp_message->hdcp_bo->vmap,
 					    addr_in_wr_off, HECI_MEADDRESS_HDCP,
 					    host_session_id, msg_in_len);
@@ -247,6 +247,6 @@ ssize_t intel_hdcp_gsc_msg_send(struct xe_device *xe, u8 *msg_in,
 			   msg_out_len);
 
 out:
-	xe_device_mem_access_put(xe);
+	xe_pm_runtime_put(xe);
 	return ret;
 }

drivers/gpu/drm/xe/xe_bo.c

@@ -716,7 +716,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 
 	xe_assert(xe, migrate);
 	trace_xe_bo_move(bo, new_mem->mem_type, old_mem_type, move_lacks_source);
-	xe_device_mem_access_get(xe);
+	xe_pm_runtime_get_noresume(xe);
 
 	if (xe_bo_is_pinned(bo) && !xe_bo_is_user(bo)) {
 		/*
@@ -740,7 +740,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 
 				if (XE_WARN_ON(new_mem->start == XE_BO_INVALID_OFFSET)) {
 					ret = -EINVAL;
-					xe_device_mem_access_put(xe);
+					xe_pm_runtime_put(xe);
 					goto out;
 				}
 
@@ -758,7 +758,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 						new_mem, handle_system_ccs);
 		if (IS_ERR(fence)) {
 			ret = PTR_ERR(fence);
-			xe_device_mem_access_put(xe);
+			xe_pm_runtime_put(xe);
 			goto out;
 		}
 		if (!move_lacks_source) {
@@ -783,7 +783,7 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict,
 		dma_fence_put(fence);
 	}
 
-	xe_device_mem_access_put(xe);
+	xe_pm_runtime_put(xe);
 
 out:
 	return ret;

drivers/gpu/drm/xe/xe_device.c

@@ -729,42 +729,6 @@ void xe_device_assert_mem_access(struct xe_device *xe)
 	xe_assert(xe, !xe_pm_runtime_suspended(xe));
 }
 
-void xe_device_mem_access_get(struct xe_device *xe)
-{
-	int ref;
-
-	/*
-	 * This looks racy, but should be fine since the pm_callback_task only
-	 * transitions from NULL -> current (and back to NULL again), during the
-	 * runtime_resume() or runtime_suspend() callbacks, for which there can
-	 * only be a single one running for our device. We only need to prevent
-	 * recursively calling the runtime_get or runtime_put from those
-	 * callbacks, as well as preventing triggering any access_ongoing
-	 * asserts.
-	 */
-	if (xe_pm_read_callback_task(xe) == current)
-		return;
-
-	xe_pm_runtime_get_noresume(xe);
-	ref = atomic_inc_return(&xe->mem_access.ref);
-
-	xe_assert(xe, ref != S32_MAX);
-
-}
-
-void xe_device_mem_access_put(struct xe_device *xe)
-{
-	int ref;
-
-	if (xe_pm_read_callback_task(xe) == current)
-		return;
-
-	ref = atomic_dec_return(&xe->mem_access.ref);
-	xe_pm_runtime_put(xe);
-
-	xe_assert(xe, ref >= 0);
-}
-
 void xe_device_snapshot_print(struct xe_device *xe, struct drm_printer *p)
 {
 	struct xe_gt *gt;

drivers/gpu/drm/xe/xe_device.h

@@ -133,9 +133,6 @@ static inline struct xe_force_wake *gt_to_fw(struct xe_gt *gt)
 	return &gt->mmio.fw;
 }
 
-void xe_device_mem_access_get(struct xe_device *xe);
-void xe_device_mem_access_put(struct xe_device *xe);
-
 void xe_device_assert_mem_access(struct xe_device *xe);
 
 static inline bool xe_device_in_fault_mode(struct xe_device *xe)

drivers/gpu/drm/xe/xe_device_types.h

@@ -384,9 +384,6 @@ struct xe_device {
 	 * triggering additional actions when they occur.
 	 */
 	struct {
-		/** @mem_access.ref: ref count of memory accesses */
-		atomic_t ref;
-
 		/**
 		 * @mem_access.vram_userfault: Encapsulate vram_userfault
 		 * related stuff

drivers/gpu/drm/xe/xe_exec_queue.c

@@ -589,7 +589,7 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data,
 				return -EINVAL;
 
 			/* The migration vm doesn't hold rpm ref */
-			xe_device_mem_access_get(xe);
+			xe_pm_runtime_get_noresume(xe);
 
 			flags = EXEC_QUEUE_FLAG_VM | (id ? EXEC_QUEUE_FLAG_BIND_ENGINE_CHILD : 0);
 
@@ -598,7 +598,7 @@ int xe_exec_queue_create_ioctl(struct drm_device *dev, void *data,
 						   args->width, hwe, flags,
 						   args->extensions);
 
-			xe_device_mem_access_put(xe); /* now held by engine */
+			xe_pm_runtime_put(xe); /* now held by engine */
 
 			xe_vm_put(migrate_vm);
 			if (IS_ERR(new)) {

drivers/gpu/drm/xe/xe_ggtt.c

@@ -21,6 +21,7 @@
 #include "xe_gt_printk.h"
 #include "xe_gt_tlb_invalidation.h"
 #include "xe_map.h"
+#include "xe_pm.h"
 #include "xe_sriov.h"
 #include "xe_wopcm.h"
 
@@ -403,7 +404,7 @@ static int __xe_ggtt_insert_bo_at(struct xe_ggtt *ggtt, struct xe_bo *bo,
 	if (err)
 		return err;
 
-	xe_device_mem_access_get(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_get_noresume(tile_to_xe(ggtt->tile));
 	mutex_lock(&ggtt->lock);
 	err = drm_mm_insert_node_in_range(&ggtt->mm, &bo->ggtt_node, bo->size,
 					  alignment, 0, start, end, 0);
@@ -413,7 +414,7 @@ static int __xe_ggtt_insert_bo_at(struct xe_ggtt *ggtt, struct xe_bo *bo,
 
 	if (!err && bo->flags & XE_BO_FLAG_GGTT_INVALIDATE)
 		xe_ggtt_invalidate(ggtt);
-	xe_device_mem_access_put(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_put(tile_to_xe(ggtt->tile));
 
 	return err;
 }
@@ -432,7 +433,7 @@ int xe_ggtt_insert_bo(struct xe_ggtt *ggtt, struct xe_bo *bo)
 void xe_ggtt_remove_node(struct xe_ggtt *ggtt, struct drm_mm_node *node,
 			 bool invalidate)
 {
-	xe_device_mem_access_get(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_get_noresume(tile_to_xe(ggtt->tile));
 
 	mutex_lock(&ggtt->lock);
 	xe_ggtt_clear(ggtt, node->start, node->size);
@@ -443,7 +444,7 @@ void xe_ggtt_remove_node(struct xe_ggtt *ggtt, struct drm_mm_node *node,
 	if (invalidate)
 		xe_ggtt_invalidate(ggtt);
 
-	xe_device_mem_access_put(tile_to_xe(ggtt->tile));
+	xe_pm_runtime_put(tile_to_xe(ggtt->tile));
 }
 
 void xe_ggtt_remove_bo(struct xe_ggtt *ggtt, struct xe_bo *bo)

drivers/gpu/drm/xe/xe_gt_sriov_pf_policy.c

@@ -12,6 +12,7 @@
 #include "xe_gt_sriov_printk.h"
 #include "xe_guc_ct.h"
 #include "xe_guc_klv_helpers.h"
+#include "xe_pm.h"
 
 /*
  * Return: number of KLVs that were successfully parsed and saved,
@@ -368,7 +369,7 @@ int xe_gt_sriov_pf_policy_reprovision(struct xe_gt *gt, bool reset)
 {
 	int err = 0;
 
-	xe_device_mem_access_get(gt_to_xe(gt));
+	xe_pm_runtime_get_noresume(gt_to_xe(gt));
 
 	mutex_lock(xe_gt_sriov_pf_master_mutex(gt));
 	if (reset)
@@ -378,7 +379,7 @@ int xe_gt_sriov_pf_policy_reprovision(struct xe_gt *gt, bool reset)
 	err |= pf_reprovision_sample_period(gt);
 	mutex_unlock(xe_gt_sriov_pf_master_mutex(gt));
 
-	xe_device_mem_access_put(gt_to_xe(gt));
+	xe_pm_runtime_put(gt_to_xe(gt));
 
 	return err ? -ENXIO : 0;
 }

drivers/gpu/drm/xe/xe_sched_job.c

@@ -16,6 +16,7 @@
 #include "xe_hw_fence.h"
 #include "xe_lrc.h"
 #include "xe_macros.h"
+#include "xe_pm.h"
 #include "xe_sync_types.h"
 #include "xe_trace.h"
 #include "xe_vm.h"
@@ -159,7 +160,7 @@ struct xe_sched_job *xe_sched_job_create(struct xe_exec_queue *q,
 
 	/* All other jobs require a VM to be open which has a ref */
 	if (unlikely(q->flags & EXEC_QUEUE_FLAG_KERNEL))
-		xe_device_mem_access_get(job_to_xe(job));
+		xe_pm_runtime_get_noresume(job_to_xe(job));
 	xe_device_assert_mem_access(job_to_xe(job));
 
 	trace_xe_sched_job_create(job);
@@ -192,7 +193,7 @@ void xe_sched_job_destroy(struct kref *ref)
 		container_of(ref, struct xe_sched_job, refcount);
 
 	if (unlikely(job->q->flags & EXEC_QUEUE_FLAG_KERNEL))
-		xe_device_mem_access_put(job_to_xe(job));
+		xe_pm_runtime_put(job_to_xe(job));
 	xe_exec_queue_put(job->q);
 	dma_fence_put(job->fence);
 	drm_sched_job_cleanup(&job->drm);

drivers/gpu/drm/xe/xe_vm.c

@@ -1266,7 +1266,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags)
 	vm->pt_ops = &xelp_pt_ops;
 
 	if (!(flags & XE_VM_FLAG_MIGRATION))
-		xe_device_mem_access_get(xe);
+		xe_pm_runtime_get_noresume(xe);
 
 	vm_resv_obj = drm_gpuvm_resv_object_alloc(&xe->drm);
 	if (!vm_resv_obj) {
@@ -1376,7 +1376,7 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags)
 		xe_range_fence_tree_fini(&vm->rftree[id]);
 	kfree(vm);
 	if (!(flags & XE_VM_FLAG_MIGRATION))
-		xe_device_mem_access_put(xe);
+		xe_pm_runtime_put(xe);
 	return ERR_PTR(err);
 }
 
@@ -1507,7 +1507,7 @@ static void xe_vm_free(struct drm_gpuvm *gpuvm)
 	mutex_destroy(&vm->snap_mutex);
 
 	if (!(vm->flags & XE_VM_FLAG_MIGRATION))
-		xe_device_mem_access_put(xe);
+		xe_pm_runtime_put(xe);
 
 	for_each_tile(tile, xe, id)
 		XE_WARN_ON(vm->pt_root[id]);