[openib-general] [PATCH 5/5] IB/iser: Do not use FMR for a single dma entry sg

[Erez Zilber]

Fast Memory Registration (fmr) is used to register for rdma an sg whose
elements are not linearly sequential after dma mapping.

The IB verbs layer provides an "all dma memory MR (memory region)" which
can be used for RDMA-ing a dma linearly sequential buffer.

Change the code to use the dma mr instead of doing fmr when dma mapping
produces a single dma entry sg.

Signed-off-by: Erez Zilber <erezz at voltaire.com>

---

 drivers/infiniband/ulp/iser/iscsi_iser.h  |    1 +
 drivers/infiniband/ulp/iser/iser_memory.c |   48 +++++++++++++++++++++--------
 drivers/infiniband/ulp/iser/iser_verbs.c  |    6 ++--
 3 files changed, 39 insertions(+), 16 deletions(-)

c403e930977afb2838588523d10819ce586951a2
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h
index 2c8bc67..2cf9ae0 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.h
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.h
@@ -175,6 +175,7 @@ struct iser_mem_reg {
 	u64  va;
 	u64  len;
 	void *mem_h;
+	int  is_fmr;
 };
 
 struct iser_regd_buf {
diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c
index 8fea0bc..e0d4347 100644
--- a/drivers/infiniband/ulp/iser/iser_memory.c
+++ b/drivers/infiniband/ulp/iser/iser_memory.c
@@ -56,7 +56,7 @@ int iser_regd_buff_release(struct iser_r
 	if ((atomic_read(&regd_buf->ref_count) == 0) ||
 	    atomic_dec_and_test(&regd_buf->ref_count)) {
 		/* if we used the dma mr, unreg is just NOP */
-		if (regd_buf->reg.rkey != 0)
+		if (regd_buf->reg.is_fmr)
 			iser_unreg_mem(&regd_buf->reg);
 
 		if (regd_buf->dma_addr) {
@@ -91,9 +91,9 @@ void iser_reg_single(struct iser_device 
 	BUG_ON(dma_mapping_error(dma_addr));
 
 	regd_buf->reg.lkey = device->mr->lkey;
-	regd_buf->reg.rkey = 0; /* indicate there's no need to unreg */
 	regd_buf->reg.len  = regd_buf->data_size;
 	regd_buf->reg.va   = dma_addr;
+	regd_buf->reg.is_fmr = 0;
 
 	regd_buf->dma_addr  = dma_addr;
 	regd_buf->direction = direction;
@@ -379,11 +379,13 @@ int iser_reg_rdma_mem(struct iscsi_iser_
 		      enum   iser_data_dir        cmd_dir)
 {
 	struct iser_conn     *ib_conn = iser_ctask->iser_conn->ib_conn;
+	struct iser_device   *device = ib_conn->device;
 	struct iser_data_buf *mem = &iser_ctask->data[cmd_dir];
 	struct iser_regd_buf *regd_buf;
 	int aligned_len;
 	int err;
 	int i;
+	struct scatterlist *sg;
 
 	regd_buf = &iser_ctask->rdma_regd[cmd_dir];
 
@@ -399,19 +401,37 @@ int iser_reg_rdma_mem(struct iscsi_iser_
 		mem = &iser_ctask->data_copy[cmd_dir];
 	}
 
-	iser_page_vec_build(mem, ib_conn->page_vec);
-	err = iser_reg_page_vec(ib_conn, ib_conn->page_vec, &regd_buf->reg);
-	if (err) {
-		iser_data_buf_dump(mem);
-		iser_err("mem->dma_nents = %d (dlength = 0x%x)\n", mem->dma_nents,
-			 ntoh24(iser_ctask->desc.iscsi_header.dlength));
-		iser_err("page_vec: data_size = 0x%x, length = %d, offset = 0x%x\n",
-			 ib_conn->page_vec->data_size, ib_conn->page_vec->length,
-			 ib_conn->page_vec->offset);
-		for (i=0 ; i<ib_conn->page_vec->length ; i++) {
-			iser_err("page_vec[%d] = 0x%lx\n", i, ib_conn->page_vec->pages[i]);
+	/* if there a single dma entry, FMR is not needed */
+	if (mem->dma_nents == 1) {
+		sg = (struct scatterlist *)mem->buf;
+
+		regd_buf->reg.lkey = device->mr->lkey;
+		regd_buf->reg.rkey = device->mr->rkey;
+		regd_buf->reg.len  = sg_dma_len(&sg[0]);
+		regd_buf->reg.va   = sg_dma_address(&sg[0]);
+		regd_buf->reg.is_fmr = 0;
+
+		iser_dbg("PHYSICAL Mem.register: lkey: 0x%08X rkey: 0x%08X  "
+			 "va: 0x%08lX sz: %ld]\n",
+			 (unsigned int)regd_buf->reg.lkey,
+			 (unsigned int)regd_buf->reg.rkey,
+			 (unsigned long)regd_buf->reg.va,
+			 (unsigned long)regd_buf->reg.len);
+	} else { /* use FMR for multiple dma entries */
+		iser_page_vec_build(mem, ib_conn->page_vec);
+		err = iser_reg_page_vec(ib_conn, ib_conn->page_vec, &regd_buf->reg);
+		if (err) {
+			iser_data_buf_dump(mem);
+			iser_err("mem->dma_nents = %d (dlength = 0x%x)\n", mem->dma_nents,
+				 ntoh24(iser_ctask->desc.iscsi_header.dlength));
+			iser_err("page_vec: data_size = 0x%x, length = %d, offset = 0x%x\n",
+				 ib_conn->page_vec->data_size, ib_conn->page_vec->length,
+				 ib_conn->page_vec->offset);
+			for (i=0 ; i<ib_conn->page_vec->length ; i++) {
+				iser_err("page_vec[%d] = 0x%lx\n", i, ib_conn->page_vec->pages[i]);
+			}
+			return err;
 		}
-		return err;
 	}
 
 	/* take a reference on this regd buf such that it will not be released *
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 9b27a7c..ecdca7f 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -88,8 +88,9 @@ static int iser_create_device_ib_res(str
 		     iser_cq_tasklet_fn,
 		     (unsigned long)device);
 
-	device->mr = ib_get_dma_mr(device->pd,
-				   IB_ACCESS_LOCAL_WRITE);
+	device->mr = ib_get_dma_mr(device->pd, IB_ACCESS_LOCAL_WRITE |
+				   IB_ACCESS_REMOTE_WRITE |
+				   IB_ACCESS_REMOTE_READ);
 	if (IS_ERR(device->mr))
 		goto dma_mr_err;
 
@@ -606,6 +607,7 @@ int iser_reg_page_vec(struct iser_conn  
 	mem_reg->rkey  = mem->fmr->rkey;
 	mem_reg->len   = page_vec->length * SIZE_4K;
 	mem_reg->va    = io_addr;
+	mem_reg->is_fmr = 1;
 	mem_reg->mem_h = (void *)mem;
 
 	mem_reg->va   += page_vec->offset;
-- 
1.2.6