IB/mlx4: Implement FMRs

Implement FMRs for mlx4.  This is an adaptation of code from mthca.
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>

drivers/infiniband/hw/mlx4/main.c

@@ -607,6 +607,11 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.detach_mcast	= mlx4_ib_mcg_detach;
 	ibdev->ib_dev.process_mad	= mlx4_ib_process_mad;
 
+	ibdev->ib_dev.alloc_fmr		= mlx4_ib_fmr_alloc;
+	ibdev->ib_dev.map_phys_fmr	= mlx4_ib_map_phys_fmr;
+	ibdev->ib_dev.unmap_fmr		= mlx4_ib_unmap_fmr;
+	ibdev->ib_dev.dealloc_fmr	= mlx4_ib_fmr_dealloc;
+
 	if (init_node_data(ibdev))
 		goto err_map;
 

drivers/infiniband/hw/mlx4/mlx4_ib.h

@@ -93,6 +93,11 @@ struct mlx4_ib_mr {
 	struct ib_umem	       *umem;
 };
 
+struct mlx4_ib_fmr {
+	struct ib_fmr           ibfmr;
+	struct mlx4_fmr         mfmr;
+};
+
 struct mlx4_ib_wq {
 	u64		       *wrid;
 	spinlock_t		lock;
@@ -199,6 +204,10 @@ static inline struct mlx4_ib_mr *to_mmr(struct ib_mr *ibmr)
 	return container_of(ibmr, struct mlx4_ib_mr, ibmr);
 }
 
+static inline struct mlx4_ib_fmr *to_mfmr(struct ib_fmr *ibfmr)
+{
+	return container_of(ibfmr, struct mlx4_ib_fmr, ibfmr);
+}
 static inline struct mlx4_ib_qp *to_mqp(struct ib_qp *ibqp)
 {
 	return container_of(ibqp, struct mlx4_ib_qp, ibqp);
@@ -284,6 +293,13 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
 int mlx4_ib_mad_init(struct mlx4_ib_dev *dev);
 void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev);
 
+struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int mr_access_flags,
+				  struct ib_fmr_attr *fmr_attr);
+int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list, int npages,
+			 u64 iova);
+int mlx4_ib_unmap_fmr(struct list_head *fmr_list);
+int mlx4_ib_fmr_dealloc(struct ib_fmr *fmr);
+
 static inline int mlx4_ib_ah_grh_present(struct mlx4_ib_ah *ah)
 {
 	return !!(ah->av.g_slid & 0x80);

drivers/infiniband/hw/mlx4/mr.c

@@ -181,3 +181,96 @@ int mlx4_ib_dereg_mr(struct ib_mr *ibmr)
 
 	return 0;
 }
+
+struct ib_fmr *mlx4_ib_fmr_alloc(struct ib_pd *pd, int acc,
+				 struct ib_fmr_attr *fmr_attr)
+{
+	struct mlx4_ib_dev *dev = to_mdev(pd->device);
+	struct mlx4_ib_fmr *fmr;
+	int err = -ENOMEM;
+
+	fmr = kmalloc(sizeof *fmr, GFP_KERNEL);
+	if (!fmr)
+		return ERR_PTR(-ENOMEM);
+
+	err = mlx4_fmr_alloc(dev->dev, to_mpd(pd)->pdn, convert_access(acc),
+			     fmr_attr->max_pages, fmr_attr->max_maps,
+			     fmr_attr->page_shift, &fmr->mfmr);
+	if (err)
+		goto err_free;
+
+	err = mlx4_mr_enable(to_mdev(pd->device)->dev, &fmr->mfmr.mr);
+	if (err)
+		goto err_mr;
+
+	fmr->ibfmr.rkey = fmr->ibfmr.lkey = fmr->mfmr.mr.key;
+
+	return &fmr->ibfmr;
+
+err_mr:
+	mlx4_mr_free(to_mdev(pd->device)->dev, &fmr->mfmr.mr);
+
+err_free:
+	kfree(fmr);
+
+	return ERR_PTR(err);
+}
+
+int mlx4_ib_map_phys_fmr(struct ib_fmr *ibfmr, u64 *page_list,
+		      int npages, u64 iova)
+{
+	struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+	struct mlx4_ib_dev *dev = to_mdev(ifmr->ibfmr.device);
+
+	return mlx4_map_phys_fmr(dev->dev, &ifmr->mfmr, page_list, npages, iova,
+				 &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey);
+}
+
+int mlx4_ib_unmap_fmr(struct list_head *fmr_list)
+{
+	struct ib_fmr *ibfmr;
+	int err;
+	struct mlx4_dev *mdev = NULL;
+
+	list_for_each_entry(ibfmr, fmr_list, list) {
+		if (mdev && to_mdev(ibfmr->device)->dev != mdev)
+			return -EINVAL;
+		mdev = to_mdev(ibfmr->device)->dev;
+	}
+
+	if (!mdev)
+		return 0;
+
+	list_for_each_entry(ibfmr, fmr_list, list) {
+		struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+
+		mlx4_fmr_unmap(mdev, &ifmr->mfmr, &ifmr->ibfmr.lkey, &ifmr->ibfmr.rkey);
+	}
+
+	/*
+	 * Make sure all MPT status updates are visible before issuing
+	 * SYNC_TPT firmware command.
+	 */
+	wmb();
+
+	err = mlx4_SYNC_TPT(mdev);
+	if (err)
+		printk(KERN_WARNING "mlx4_ib: SYNC_TPT error %d when "
+		       "unmapping FMRs\n", err);
+
+	return 0;
+}
+
+int mlx4_ib_fmr_dealloc(struct ib_fmr *ibfmr)
+{
+	struct mlx4_ib_fmr *ifmr = to_mfmr(ibfmr);
+	struct mlx4_ib_dev *dev = to_mdev(ibfmr->device);
+	int err;
+
+	err = mlx4_fmr_free(dev->dev, &ifmr->mfmr);
+
+	if (!err)
+		kfree(ifmr);
+
+	return err;
+}

drivers/net/mlx4/mr.c

@@ -68,6 +68,9 @@ struct mlx4_mpt_entry {
 
 #define MLX4_MTT_FLAG_PRESENT		1
 
+#define MLX4_MPT_STATUS_SW		0xF0
+#define MLX4_MPT_STATUS_HW		0x00
+
 static u32 mlx4_buddy_alloc(struct mlx4_buddy *buddy, int order)
 {
 	int o;
@@ -469,3 +472,165 @@ void mlx4_cleanup_mr_table(struct mlx4_dev *dev)
 	mlx4_buddy_cleanup(&mr_table->mtt_buddy);
 	mlx4_bitmap_cleanup(&mr_table->mpt_bitmap);
 }
+
+static inline int mlx4_check_fmr(struct mlx4_fmr *fmr, u64 *page_list,
+				  int npages, u64 iova)
+{
+	int i, page_mask;
+
+	if (npages > fmr->max_pages)
+		return -EINVAL;
+
+	page_mask = (1 << fmr->page_shift) - 1;
+
+	/* We are getting page lists, so va must be page aligned. */
+	if (iova & page_mask)
+		return -EINVAL;
+
+	/* Trust the user not to pass misaligned data in page_list */
+	if (0)
+		for (i = 0; i < npages; ++i) {
+			if (page_list[i] & ~page_mask)
+				return -EINVAL;
+		}
+
+	if (fmr->maps >= fmr->max_maps)
+		return -EINVAL;
+
+	return 0;
+}
+
+int mlx4_map_phys_fmr(struct mlx4_dev *dev, struct mlx4_fmr *fmr, u64 *page_list,
+		      int npages, u64 iova, u32 *lkey, u32 *rkey)
+{
+	u32 key;
+	int i, err;
+
+	err = mlx4_check_fmr(fmr, page_list, npages, iova);
+	if (err)
+		return err;
+
+	++fmr->maps;
+
+	key = key_to_hw_index(fmr->mr.key);
+	key += dev->caps.num_mpts;
+	*lkey = *rkey = fmr->mr.key = hw_index_to_key(key);
+
+	*(u8 *) fmr->mpt = MLX4_MPT_STATUS_SW;
+
+	/* Make sure MPT status is visible before writing MTT entries */
+	wmb();
+
+	for (i = 0; i < npages; ++i)
+		fmr->mtts[i] = cpu_to_be64(page_list[i] | MLX4_MTT_FLAG_PRESENT);
+
+	dma_sync_single(&dev->pdev->dev, fmr->dma_handle,
+			npages * sizeof(u64), DMA_TO_DEVICE);
+
+	fmr->mpt->key    = cpu_to_be32(key);
+	fmr->mpt->lkey   = cpu_to_be32(key);
+	fmr->mpt->length = cpu_to_be64(npages * (1ull << fmr->page_shift));
+	fmr->mpt->start  = cpu_to_be64(iova);
+
+	/* Make MTT entries are visible before setting MPT status */
+	wmb();
+
+	*(u8 *) fmr->mpt = MLX4_MPT_STATUS_HW;
+
+	/* Make sure MPT status is visible before consumer can use FMR */
+	wmb();
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_map_phys_fmr);
+
+int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages,
+		   int max_maps, u8 page_shift, struct mlx4_fmr *fmr)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	u64 mtt_seg;
+	int err = -ENOMEM;
+
+	if (page_shift < 12 || page_shift >= 32)
+		return -EINVAL;
+
+	/* All MTTs must fit in the same page */
+	if (max_pages * sizeof *fmr->mtts > PAGE_SIZE)
+		return -EINVAL;
+
+	fmr->page_shift = page_shift;
+	fmr->max_pages  = max_pages;
+	fmr->max_maps   = max_maps;
+	fmr->maps = 0;
+
+	err = mlx4_mr_alloc(dev, pd, 0, 0, access, max_pages,
+			    page_shift, &fmr->mr);
+	if (err)
+		return err;
+
+	mtt_seg = fmr->mr.mtt.first_seg * dev->caps.mtt_entry_sz;
+
+	fmr->mtts = mlx4_table_find(&priv->mr_table.mtt_table,
+				    fmr->mr.mtt.first_seg,
+				    &fmr->dma_handle);
+	if (!fmr->mtts) {
+		err = -ENOMEM;
+		goto err_free;
+	}
+
+	fmr->mpt = mlx4_table_find(&priv->mr_table.dmpt_table,
+				    key_to_hw_index(fmr->mr.key), NULL);
+	if (!fmr->mpt) {
+		err = -ENOMEM;
+		goto err_free;
+	}
+
+	return 0;
+
+err_free:
+	mlx4_mr_free(dev, &fmr->mr);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_fmr_alloc);
+
+int mlx4_fmr_enable(struct mlx4_dev *dev, struct mlx4_fmr *fmr)
+{
+	return mlx4_mr_enable(dev, &fmr->mr);
+}
+EXPORT_SYMBOL_GPL(mlx4_fmr_enable);
+
+void mlx4_fmr_unmap(struct mlx4_dev *dev, struct mlx4_fmr *fmr,
+		    u32 *lkey, u32 *rkey)
+{
+	u32 key;
+
+	if (!fmr->maps)
+		return;
+
+	key = key_to_hw_index(fmr->mr.key);
+	key &= dev->caps.num_mpts - 1;
+	*lkey = *rkey = fmr->mr.key = hw_index_to_key(key);
+
+	fmr->maps = 0;
+
+	*(u8 *) fmr->mpt = MLX4_MPT_STATUS_SW;
+}
+EXPORT_SYMBOL_GPL(mlx4_fmr_unmap);
+
+int mlx4_fmr_free(struct mlx4_dev *dev, struct mlx4_fmr *fmr)
+{
+	if (fmr->maps)
+		return -EBUSY;
+
+	fmr->mr.enabled = 0;
+	mlx4_mr_free(dev, &fmr->mr);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(mlx4_fmr_free);
+
+int mlx4_SYNC_TPT(struct mlx4_dev *dev)
+{
+	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_SYNC_TPT, 1000);
+}
+EXPORT_SYMBOL_GPL(mlx4_SYNC_TPT);

include/linux/mlx4/device.h

@@ -214,6 +214,17 @@ struct mlx4_mr {
 	int			enabled;
 };
 
+struct mlx4_fmr {
+	struct mlx4_mr		mr;
+	struct mlx4_mpt_entry  *mpt;
+	__be64		       *mtts;
+	dma_addr_t		dma_handle;
+	int			max_pages;
+	int			max_maps;
+	int			maps;
+	u8			page_shift;
+};
+
 struct mlx4_uar {
 	unsigned long		pfn;
 	int			index;
@@ -337,4 +348,14 @@ int mlx4_CLOSE_PORT(struct mlx4_dev *dev, int port);
 int mlx4_multicast_attach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16]);
 int mlx4_multicast_detach(struct mlx4_dev *dev, struct mlx4_qp *qp, u8 gid[16]);
 
+int mlx4_map_phys_fmr(struct mlx4_dev *dev, struct mlx4_fmr *fmr, u64 *page_list,
+		      int npages, u64 iova, u32 *lkey, u32 *rkey);
+int mlx4_fmr_alloc(struct mlx4_dev *dev, u32 pd, u32 access, int max_pages,
+		   int max_maps, u8 page_shift, struct mlx4_fmr *fmr);
+int mlx4_fmr_enable(struct mlx4_dev *dev, struct mlx4_fmr *fmr);
+void mlx4_fmr_unmap(struct mlx4_dev *dev, struct mlx4_fmr *fmr,
+		    u32 *lkey, u32 *rkey);
+int mlx4_fmr_free(struct mlx4_dev *dev, struct mlx4_fmr *fmr);
+int mlx4_SYNC_TPT(struct mlx4_dev *dev);
+
 #endif /* MLX4_DEVICE_H */