drm/amdkfd: Remove unaligned memory access

Unaligned atomic operations can cause problems on some CPU
architectures. Use simpler bitmask operations instead. Atomic bit
manipulations are not necessary since dqm->lock is held during these
operations.

Signed-off-by: Harish Kasiviswanathan <Harish.Kasiviswanathan@amd.com>
Signed-off-by: Felix Kuehling <Felix.Kuehling@amd.com>
Acked-by: Oded Gabbay <oded.gabbay@gmail.com>
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>

drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c

@@ -118,9 +118,8 @@ static int allocate_vmid(struct device_queue_manager *dqm,
 	if (dqm->vmid_bitmap == 0)
 		return -ENOMEM;
 
-	bit = find_first_bit((unsigned long *)&dqm->vmid_bitmap,
-				dqm->dev->vm_info.vmid_num_kfd);
-	clear_bit(bit, (unsigned long *)&dqm->vmid_bitmap);
+	bit = ffs(dqm->vmid_bitmap) - 1;
+	dqm->vmid_bitmap &= ~(1 << bit);
 
 	allocated_vmid = bit + dqm->dev->vm_info.first_vmid_kfd;
 	pr_debug("vmid allocation %d\n", allocated_vmid);
@@ -142,7 +141,7 @@ static void deallocate_vmid(struct device_queue_manager *dqm,
 	/* Release the vmid mapping */
 	set_pasid_vmid_mapping(dqm, 0, qpd->vmid);
 
-	set_bit(bit, (unsigned long *)&dqm->vmid_bitmap);
+	dqm->vmid_bitmap |= (1 << bit);
 	qpd->vmid = 0;
 	q->properties.vmid = 0;
 }
@@ -223,12 +222,8 @@ static int allocate_hqd(struct device_queue_manager *dqm, struct queue *q)
 			continue;
 
 		if (dqm->allocated_queues[pipe] != 0) {
-			bit = find_first_bit(
-				(unsigned long *)&dqm->allocated_queues[pipe],
-				get_queues_per_pipe(dqm));
-
-			clear_bit(bit,
-				(unsigned long *)&dqm->allocated_queues[pipe]);
+			bit = ffs(dqm->allocated_queues[pipe]) - 1;
+			dqm->allocated_queues[pipe] &= ~(1 << bit);
 			q->pipe = pipe;
 			q->queue = bit;
 			set = true;
@@ -249,7 +244,7 @@ static int allocate_hqd(struct device_queue_manager *dqm, struct queue *q)
 static inline void deallocate_hqd(struct device_queue_manager *dqm,
 				struct queue *q)
 {
-	set_bit(q->queue, (unsigned long *)&dqm->allocated_queues[q->pipe]);
+	dqm->allocated_queues[q->pipe] |= (1 << q->queue);
 }
 
 static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
@@ -589,10 +584,8 @@ static int allocate_sdma_queue(struct device_queue_manager *dqm,
 	if (dqm->sdma_bitmap == 0)
 		return -ENOMEM;
 
-	bit = find_first_bit((unsigned long *)&dqm->sdma_bitmap,
-				CIK_SDMA_QUEUES);
-
-	clear_bit(bit, (unsigned long *)&dqm->sdma_bitmap);
+	bit = ffs(dqm->sdma_bitmap) - 1;
+	dqm->sdma_bitmap &= ~(1 << bit);
 	*sdma_queue_id = bit;
 
 	return 0;
@@ -603,7 +596,7 @@ static void deallocate_sdma_queue(struct device_queue_manager *dqm,
 {
 	if (sdma_queue_id >= CIK_SDMA_QUEUES)
 		return;
-	set_bit(sdma_queue_id, (unsigned long *)&dqm->sdma_bitmap);
+	dqm->sdma_bitmap |= (1 << sdma_queue_id);
 }
 
 static int create_sdma_queue_nocpsch(struct device_queue_manager *dqm,