RDMA/umem: Use ib_umem_odp in all function signatures connected to ODP

All of these functions already require the ODP version of the umem struct,
make this very clear by having the signature require it. This paves the
way to using the container_of() pattern to link umem_odp and umem
together.
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
Signed-off-by: Doug Ledford <dledford@redhat.com>

drivers/infiniband/core/umem.c

@@ -262,7 +262,7 @@ void ib_umem_release(struct ib_umem *umem)
 	struct ib_ucontext *context = umem->context;
 
 	if (umem->odp_data) {
-		ib_umem_odp_release(umem);
+		ib_umem_odp_release(to_ib_umem_odp(umem));
 		return;
 	}
 

drivers/infiniband/hw/mlx5/mlx5_ib.h

@@ -1150,7 +1150,7 @@ void mlx5_ib_pfault(struct mlx5_core_dev *mdev, void *context,
 int mlx5_ib_odp_init_one(struct mlx5_ib_dev *ibdev);
 int __init mlx5_ib_odp_init(void);
 void mlx5_ib_odp_cleanup(void);
-void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
+void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start,
 			      unsigned long end);
 void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent);
 void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset,

drivers/infiniband/hw/mlx5/mr.c

@@ -1631,7 +1631,8 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr)
 		synchronize_srcu(&dev->mr_srcu);
 		/* Destroy all page mappings */
 		if (umem->odp_data->page_list)
-			mlx5_ib_invalidate_range(umem, ib_umem_start(umem),
+			mlx5_ib_invalidate_range(to_ib_umem_odp(umem),
+						 ib_umem_start(umem),
 						 ib_umem_end(umem));
 		else
 			mlx5_ib_free_implicit_mr(mr);

drivers/infiniband/hw/mlx5/odp.c

@@ -170,22 +170,24 @@ static void mr_leaf_free_action(struct work_struct *work)
 		wake_up(&imr->q_leaf_free);
 }
 
-void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
+void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start,
 			      unsigned long end)
 {
 	struct mlx5_ib_mr *mr;
 	const u64 umr_block_mask = (MLX5_UMR_MTT_ALIGNMENT /
 				    sizeof(struct mlx5_mtt)) - 1;
 	u64 idx = 0, blk_start_idx = 0;
+	struct ib_umem *umem;
 	int in_block = 0;
 	u64 addr;
 
-	if (!umem || !umem->odp_data) {
+	if (!umem_odp) {
 		pr_err("invalidation called on NULL umem or non-ODP umem\n");
 		return;
 	}
+	umem = umem_odp->umem;
 
-	mr = umem->odp_data->private;
+	mr = umem_odp->private;
 
 	if (!mr || !mr->ibmr.pd)
 		return;
@@ -208,7 +210,7 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 		 * estimate the cost of another UMR vs. the cost of bigger
 		 * UMR.
 		 */
-		if (umem->odp_data->dma_list[idx] &
+		if (umem_odp->dma_list[idx] &
 		    (ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT)) {
 			if (!in_block) {
 				blk_start_idx = idx;
@@ -237,13 +239,13 @@ void mlx5_ib_invalidate_range(struct ib_umem *umem, unsigned long start,
 	 * needed.
 	 */
 
-	ib_umem_odp_unmap_dma_pages(umem, start, end);
+	ib_umem_odp_unmap_dma_pages(umem_odp, start, end);
 
 	if (unlikely(!umem->npages && mr->parent &&
-		     !umem->odp_data->dying)) {
-		WRITE_ONCE(umem->odp_data->dying, 1);
+		     !umem_odp->dying)) {
+		WRITE_ONCE(umem_odp->dying, 1);
 		atomic_inc(&mr->parent->num_leaf_free);
-		schedule_work(&umem->odp_data->work);
+		schedule_work(&umem_odp->work);
 	}
 }
 
@@ -372,7 +374,6 @@ static struct ib_umem_odp *implicit_mr_get_data(struct mlx5_ib_mr *mr,
 	u64 addr = io_virt & MLX5_IMR_MTT_MASK;
 	int nentries = 0, start_idx = 0, ret;
 	struct mlx5_ib_mr *mtt;
-	struct ib_umem *umem;
 
 	mutex_lock(&mr->umem->odp_data->umem_mutex);
 	odp = odp_lookup(ctx, addr, 1, mr);
@@ -385,22 +386,22 @@ static struct ib_umem_odp *implicit_mr_get_data(struct mlx5_ib_mr *mr,
 		if (nentries)
 			nentries++;
 	} else {
-		umem = ib_alloc_odp_umem(ctx, addr, MLX5_IMR_MTT_SIZE);
-		if (IS_ERR(umem)) {
+		odp = ib_alloc_odp_umem(ctx, addr, MLX5_IMR_MTT_SIZE);
+		if (IS_ERR(odp)) {
 			mutex_unlock(&mr->umem->odp_data->umem_mutex);
-			return ERR_CAST(umem);
+			return ERR_CAST(odp);
 		}
 
-		mtt = implicit_mr_alloc(mr->ibmr.pd, umem, 0, mr->access_flags);
+		mtt = implicit_mr_alloc(mr->ibmr.pd, odp->umem, 0,
+					mr->access_flags);
 		if (IS_ERR(mtt)) {
 			mutex_unlock(&mr->umem->odp_data->umem_mutex);
-			ib_umem_release(umem);
+			ib_umem_release(odp->umem);
 			return ERR_CAST(mtt);
 		}
 
-		odp = umem->odp_data;
 		odp->private = mtt;
-		mtt->umem = umem;
+		mtt->umem = odp->umem;
 		mtt->mmkey.iova = addr;
 		mtt->parent = mr;
 		INIT_WORK(&odp->work, mr_leaf_free_action);
@@ -460,24 +461,24 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd,
 	return imr;
 }
 
-static int mr_leaf_free(struct ib_umem *umem, u64 start,
-			u64 end, void *cookie)
+static int mr_leaf_free(struct ib_umem_odp *umem_odp, u64 start, u64 end,
+			void *cookie)
 {
-	struct mlx5_ib_mr *mr = umem->odp_data->private, *imr = cookie;
+	struct mlx5_ib_mr *mr = umem_odp->private, *imr = cookie;
+	struct ib_umem *umem = umem_odp->umem;
 
 	if (mr->parent != imr)
 		return 0;
 
-	ib_umem_odp_unmap_dma_pages(umem,
-				    ib_umem_start(umem),
+	ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem),
 				    ib_umem_end(umem));
 
-	if (umem->odp_data->dying)
+	if (umem_odp->dying)
 		return 0;
 
-	WRITE_ONCE(umem->odp_data->dying, 1);
+	WRITE_ONCE(umem_odp->dying, 1);
 	atomic_inc(&imr->num_leaf_free);
-	schedule_work(&umem->odp_data->work);
+	schedule_work(&umem_odp->work);
 
 	return 0;
 }
@@ -533,7 +534,7 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
 	 */
 	smp_rmb();
 
-	ret = ib_umem_odp_map_dma_pages(mr->umem, io_virt, size,
+	ret = ib_umem_odp_map_dma_pages(to_ib_umem_odp(mr->umem), io_virt, size,
 					access_mask, current_seq);
 
 	if (ret < 0)
@@ -542,7 +543,8 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr,
 	np = ret;
 
 	mutex_lock(&odp->umem_mutex);
-	if (!ib_umem_mmu_notifier_retry(mr->umem, current_seq)) {
+	if (!ib_umem_mmu_notifier_retry(to_ib_umem_odp(mr->umem),
+					current_seq)) {
 		/*
 		 * No need to check whether the MTTs really belong to
 		 * this MR, since ib_umem_odp_map_dma_pages already

include/rdma/ib_umem_odp.h

@@ -82,15 +82,18 @@ struct ib_umem_odp {
 	struct work_struct	work;
 };
 
+static inline struct ib_umem_odp *to_ib_umem_odp(struct ib_umem *umem)
+{
+	return umem->odp_data;
+}
+
 #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING
 
 int ib_umem_odp_get(struct ib_ucontext *context, struct ib_umem *umem,
 		    int access);
-struct ib_umem *ib_alloc_odp_umem(struct ib_ucontext *context,
-				  unsigned long addr,
-				  size_t size);
-
-void ib_umem_odp_release(struct ib_umem *umem);
+struct ib_umem_odp *ib_alloc_odp_umem(struct ib_ucontext *context,
+				      unsigned long addr, size_t size);
+void ib_umem_odp_release(struct ib_umem_odp *umem_odp);
 
 /*
  * The lower 2 bits of the DMA address signal the R/W permissions for
@@ -105,13 +108,14 @@ void ib_umem_odp_release(struct ib_umem *umem);
 
 #define ODP_DMA_ADDR_MASK (~(ODP_READ_ALLOWED_BIT | ODP_WRITE_ALLOWED_BIT))
 
-int ib_umem_odp_map_dma_pages(struct ib_umem *umem, u64 start_offset, u64 bcnt,
-			      u64 access_mask, unsigned long current_seq);
+int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 start_offset,
+			      u64 bcnt, u64 access_mask,
+			      unsigned long current_seq);
 
-void ib_umem_odp_unmap_dma_pages(struct ib_umem *umem, u64 start_offset,
+void ib_umem_odp_unmap_dma_pages(struct ib_umem_odp *umem_odp, u64 start_offset,
 				 u64 bound);
 
-typedef int (*umem_call_back)(struct ib_umem *item, u64 start, u64 end,
+typedef int (*umem_call_back)(struct ib_umem_odp *item, u64 start, u64 end,
 			      void *cookie);
 /*
  * Call the callback on each ib_umem in the range. Returns the logical or of
@@ -129,25 +133,25 @@ int rbt_ib_umem_for_each_in_range(struct rb_root_cached *root,
 struct ib_umem_odp *rbt_ib_umem_lookup(struct rb_root_cached *root,
 				       u64 addr, u64 length);
 
-static inline int ib_umem_mmu_notifier_retry(struct ib_umem *item,
+static inline int ib_umem_mmu_notifier_retry(struct ib_umem_odp *umem_odp,
 					     unsigned long mmu_seq)
 {
 	/*
 	 * This code is strongly based on the KVM code from
 	 * mmu_notifier_retry. Should be called with
-	 * the relevant locks taken (item->odp_data->umem_mutex
+	 * the relevant locks taken (umem_odp->umem_mutex
 	 * and the ucontext umem_mutex semaphore locked for read).
 	 */
 
 	/* Do not allow page faults while the new ib_umem hasn't seen a state
 	 * with zero notifiers yet, and doesn't have its own valid set of
 	 * private counters. */
-	if (!item->odp_data->mn_counters_active)
+	if (!umem_odp->mn_counters_active)
 		return 1;
 
-	if (unlikely(item->odp_data->notifiers_count))
+	if (unlikely(umem_odp->notifiers_count))
 		return 1;
-	if (item->odp_data->notifiers_seq != mmu_seq)
+	if (umem_odp->notifiers_seq != mmu_seq)
 		return 1;
 	return 0;
 }
@@ -161,14 +165,13 @@ static inline int ib_umem_odp_get(struct ib_ucontext *context,
 	return -EINVAL;
 }
 
-static inline struct ib_umem *ib_alloc_odp_umem(struct ib_ucontext *context,
-						unsigned long addr,
-						size_t size)
+static inline struct ib_umem_odp *
+ib_alloc_odp_umem(struct ib_ucontext *context, unsigned long addr, size_t size)
 {
 	return ERR_PTR(-EINVAL);
 }
 
-static inline void ib_umem_odp_release(struct ib_umem *umem) {}
+static inline void ib_umem_odp_release(struct ib_umem_odp *umem_odp) {}
 
 #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */
 

include/rdma/ib_verbs.h

@@ -69,6 +69,8 @@
 
 #define IB_FW_VERSION_NAME_MAX	ETHTOOL_FWVERS_LEN
 
+struct ib_umem_odp;
+
 extern struct workqueue_struct *ib_wq;
 extern struct workqueue_struct *ib_comp_wq;
 extern struct workqueue_struct *ib_comp_unbound_wq;
@@ -1506,7 +1508,7 @@ struct ib_ucontext {
 	 * mmu notifiers registration.
 	 */
 	struct rw_semaphore	umem_rwsem;
-	void (*invalidate_range)(struct ib_umem *umem,
+	void (*invalidate_range)(struct ib_umem_odp *umem_odp,
 				 unsigned long start, unsigned long end);
 
 	struct mmu_notifier	mn;