IB/iser: Revert commit bba7ebb "avoid recv buffer exhaustion"
authorOr Gerlitz <ogerlitz@voltaire.com>
Mon, 8 Feb 2010 13:17:11 +0000 (13:17 +0000)
committerRoland Dreier <rolandd@cisco.com>
Wed, 24 Feb 2010 17:41:09 +0000 (09:41 -0800)
We will make a major change in the recv buffer posting logic, after
which the problem commit bba7ebb "avoid recv buffer exhaustion caused
by unexpected PDUs" comes to solve doesn't exist any more, so revert it.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
drivers/infiniband/ulp/iser/iscsi_iser.h
drivers/infiniband/ulp/iser/iser_initiator.c
drivers/infiniband/ulp/iser/iser_verbs.c

index 9d529cae1f0d9677f65668cac966fe4ecbee405f..e8dfdcfa1daff56991d97b07ce21a9f920559c7c 100644 (file)
@@ -252,9 +252,6 @@ struct iser_conn {
        wait_queue_head_t            wait;          /* waitq for conn/disconn  */
        atomic_t                     post_recv_buf_count; /* posted rx count   */
        atomic_t                     post_send_buf_count; /* posted tx count   */
-       atomic_t                     unexpected_pdu_count;/* count of received *
-                                                          * unexpected pdus   *
-                                                          * not yet retired   */
        char                         name[ISER_OBJECT_NAME_SIZE];
        struct iser_page_vec         *page_vec;     /* represents SG to fmr maps*
                                                     * maps serialized as tx is*/
index 9de640200ad3b61d4b6c05ba9792998a1860faf2..5f42fbe3080c6876aeeb64272bf4e2a32f4642cc 100644 (file)
@@ -183,8 +183,14 @@ static int iser_post_receive_control(struct iscsi_conn *conn)
        struct iser_regd_buf *regd_data;
        struct iser_dto      *recv_dto = NULL;
        struct iser_device  *device = iser_conn->ib_conn->device;
-       int rx_data_size, err;
-       int posts, outstanding_unexp_pdus;
+       int rx_data_size, err = 0;
+
+       rx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
+       if (rx_desc == NULL) {
+               iser_err("Failed to alloc desc for post recv\n");
+               return -ENOMEM;
+       }
+       rx_desc->type = ISCSI_RX;
 
        /* for the login sequence we must support rx of upto 8K; login is done
         * after conn create/bind (connect) and conn stop/bind (reconnect),
@@ -195,80 +201,46 @@ static int iser_post_receive_control(struct iscsi_conn *conn)
        else /* FIXME till user space sets conn->max_recv_dlength correctly */
                rx_data_size = 128;
 
-       outstanding_unexp_pdus =
-               atomic_xchg(&iser_conn->ib_conn->unexpected_pdu_count, 0);
-
-       /*
-        * in addition to the response buffer, replace those consumed by
-        * unexpected pdus.
-        */
-       for (posts = 0; posts < 1 + outstanding_unexp_pdus; posts++) {
-               rx_desc = kmem_cache_alloc(ig.desc_cache, GFP_NOIO);
-               if (rx_desc == NULL) {
-                       iser_err("Failed to alloc desc for post recv %d\n",
-                                posts);
-                       err = -ENOMEM;
-                       goto post_rx_cache_alloc_failure;
-               }
-               rx_desc->type = ISCSI_RX;
-               rx_desc->data = kmalloc(rx_data_size, GFP_NOIO);
-               if (rx_desc->data == NULL) {
-                       iser_err("Failed to alloc data buf for post recv %d\n",
-                                posts);
-                       err = -ENOMEM;
-                       goto post_rx_kmalloc_failure;
-               }
+       rx_desc->data = kmalloc(rx_data_size, GFP_NOIO);
+       if (rx_desc->data == NULL) {
+               iser_err("Failed to alloc data buf for post recv\n");
+               err = -ENOMEM;
+               goto post_rx_kmalloc_failure;
+       }
 
-               recv_dto = &rx_desc->dto;
-               recv_dto->ib_conn = iser_conn->ib_conn;
-               recv_dto->regd_vector_len = 0;
+       recv_dto = &rx_desc->dto;
+       recv_dto->ib_conn = iser_conn->ib_conn;
+       recv_dto->regd_vector_len = 0;
 
-               regd_hdr = &rx_desc->hdr_regd_buf;
-               memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
-               regd_hdr->device  = device;
-               regd_hdr->virt_addr  = rx_desc; /* == &rx_desc->iser_header */
-               regd_hdr->data_size  = ISER_TOTAL_HEADERS_LEN;
+       regd_hdr = &rx_desc->hdr_regd_buf;
+       memset(regd_hdr, 0, sizeof(struct iser_regd_buf));
+       regd_hdr->device  = device;
+       regd_hdr->virt_addr  = rx_desc; /* == &rx_desc->iser_header */
+       regd_hdr->data_size  = ISER_TOTAL_HEADERS_LEN;
 
-               iser_reg_single(device, regd_hdr, DMA_FROM_DEVICE);
+       iser_reg_single(device, regd_hdr, DMA_FROM_DEVICE);
 
-               iser_dto_add_regd_buff(recv_dto, regd_hdr, 0, 0);
+       iser_dto_add_regd_buff(recv_dto, regd_hdr, 0, 0);
 
-               regd_data = &rx_desc->data_regd_buf;
-               memset(regd_data, 0, sizeof(struct iser_regd_buf));
-               regd_data->device  = device;
-               regd_data->virt_addr  = rx_desc->data;
-               regd_data->data_size  = rx_data_size;
+       regd_data = &rx_desc->data_regd_buf;
+       memset(regd_data, 0, sizeof(struct iser_regd_buf));
+       regd_data->device  = device;
+       regd_data->virt_addr  = rx_desc->data;
+       regd_data->data_size  = rx_data_size;
 
-               iser_reg_single(device, regd_data, DMA_FROM_DEVICE);
+       iser_reg_single(device, regd_data, DMA_FROM_DEVICE);
 
-               iser_dto_add_regd_buff(recv_dto, regd_data, 0, 0);
+       iser_dto_add_regd_buff(recv_dto, regd_data, 0, 0);
 
-               err = iser_post_recv(rx_desc);
-               if (err) {
-                       iser_err("Failed iser_post_recv for post %d\n", posts);
-                       goto post_rx_post_recv_failure;
-               }
-       }
-       /* all posts successful */
-       return 0;
+       err = iser_post_recv(rx_desc);
+       if (!err)
+               return 0;
 
-post_rx_post_recv_failure:
+       /* iser_post_recv failed */
        iser_dto_buffs_release(recv_dto);
        kfree(rx_desc->data);
 post_rx_kmalloc_failure:
        kmem_cache_free(ig.desc_cache, rx_desc);
-post_rx_cache_alloc_failure:
-       if (posts > 0) {
-               /*
-                * response buffer posted, but did not replace all unexpected
-                * pdu recv bufs. Ignore error, retry occurs next send
-                */
-               outstanding_unexp_pdus -= (posts - 1);
-               err = 0;
-       }
-       atomic_add(outstanding_unexp_pdus,
-                  &iser_conn->ib_conn->unexpected_pdu_count);
-
        return err;
 }
 
@@ -302,10 +274,8 @@ int iser_conn_set_full_featured_mode(struct iscsi_conn *conn)
        struct iscsi_iser_conn *iser_conn = conn->dd_data;
 
        int i;
-       /*
-        * FIXME this value should be declared to the target during login with
-        * the MaxOutstandingUnexpectedPDUs key when supported
-        */
+       /* no need to keep it in a var, we are after login so if this should
+        * be negotiated, by now the result should be available here */
        int initial_post_recv_bufs_num = ISER_MAX_RX_MISC_PDUS;
 
        iser_dbg("Initially post: %d\n", initial_post_recv_bufs_num);
@@ -507,7 +477,6 @@ int iser_send_control(struct iscsi_conn *conn,
        int err = 0;
        struct iser_regd_buf *regd_buf;
        struct iser_device *device;
-       unsigned char opcode;
 
        if (!iser_conn_state_comp(iser_conn->ib_conn, ISER_CONN_UP)) {
                iser_err("Failed to send, conn: 0x%p is not up\n", iser_conn->ib_conn);
@@ -542,15 +511,10 @@ int iser_send_control(struct iscsi_conn *conn,
                                       data_seg_len);
        }
 
-       opcode = task->hdr->opcode & ISCSI_OPCODE_MASK;
-
-       /* post recv buffer for response if one is expected */
-       if (!(opcode == ISCSI_OP_NOOP_OUT && task->hdr->itt == RESERVED_ITT)) {
-               if (iser_post_receive_control(conn) != 0) {
-                       iser_err("post_rcv_buff failed!\n");
-                       err = -ENOMEM;
-                       goto send_control_error;
-               }
+       if (iser_post_receive_control(conn) != 0) {
+               iser_err("post_rcv_buff failed!\n");
+               err = -ENOMEM;
+               goto send_control_error;
        }
 
        err = iser_post_send(mdesc);
@@ -621,20 +585,6 @@ void iser_rcv_completion(struct iser_desc *rx_desc,
         * parallel to the execution of iser_conn_term. So the code that waits *
         * for the posted rx bufs refcount to become zero handles everything   */
        atomic_dec(&conn->ib_conn->post_recv_buf_count);
-
-       /*
-        * if an unexpected PDU was received then the recv wr consumed must
-        * be replaced, this is done in the next send of a control-type PDU
-        */
-       if (opcode == ISCSI_OP_NOOP_IN && hdr->itt == RESERVED_ITT) {
-               /* nop-in with itt = 0xffffffff */
-               atomic_inc(&conn->ib_conn->unexpected_pdu_count);
-       }
-       else if (opcode == ISCSI_OP_ASYNC_EVENT) {
-               /* asyncronous message */
-               atomic_inc(&conn->ib_conn->unexpected_pdu_count);
-       }
-       /* a reject PDU consumes the recv buf posted for the response */
 }
 
 void iser_snd_completion(struct iser_desc *tx_desc)
index 8579f32ce38e6df739bceb270a0ab28623a8f76f..7092503a10e3cdc7be850e17daa300c4367bf179 100644 (file)
@@ -491,7 +491,6 @@ void iser_conn_init(struct iser_conn *ib_conn)
        init_waitqueue_head(&ib_conn->wait);
        atomic_set(&ib_conn->post_recv_buf_count, 0);
        atomic_set(&ib_conn->post_send_buf_count, 0);
-       atomic_set(&ib_conn->unexpected_pdu_count, 0);
        atomic_set(&ib_conn->refcount, 1);
        INIT_LIST_HEAD(&ib_conn->conn_list);
        spin_lock_init(&ib_conn->lock);