qed: Chain support for external PBL

iWARP would require the chains to allocate/free their PBL memory
independently, so add the infrastructure to provide it externally.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/infiniband/hw/qedr/main.c

@@ -276,7 +276,7 @@ static int qedr_alloc_resources(struct qedr_dev *dev)
 						   QED_CHAIN_CNT_TYPE_U16,
 						   n_entries,
 						   sizeof(struct regpair *),
-						   &cnq->pbl);
+						   &cnq->pbl, NULL);
 		if (rc)
 			goto err4;
 

drivers/infiniband/hw/qedr/verbs.c

@@ -925,7 +925,7 @@ struct ib_cq *qedr_create_cq(struct ib_device *ibdev,
 						   QED_CHAIN_CNT_TYPE_U32,
 						   chain_entries,
 						   sizeof(union rdma_cqe),
-						   &cq->pbl);
+						   &cq->pbl, NULL);
 		if (rc)
 			goto err1;
 
@@ -1413,7 +1413,7 @@ qedr_roce_create_kernel_qp(struct qedr_dev *dev,
 					   QED_CHAIN_CNT_TYPE_U32,
 					   n_sq_elems,
 					   QEDR_SQE_ELEMENT_SIZE,
-					   &qp->sq.pbl);
+					   &qp->sq.pbl, NULL);
 
 	if (rc)
 		return rc;
@@ -1427,7 +1427,7 @@ qedr_roce_create_kernel_qp(struct qedr_dev *dev,
 					   QED_CHAIN_CNT_TYPE_U32,
 					   n_rq_elems,
 					   QEDR_RQE_ELEMENT_SIZE,
-					   &qp->rq.pbl);
+					   &qp->rq.pbl, NULL);
 	if (rc)
 		return rc;
 

drivers/net/ethernet/qlogic/qed/qed_dev.c

@@ -3075,12 +3075,15 @@ static void qed_chain_free_pbl(struct qed_dev *cdev, struct qed_chain *p_chain)
 	}
 
 	pbl_size = page_cnt * QED_CHAIN_PBL_ENTRY_SIZE;
-	dma_free_coherent(&cdev->pdev->dev,
-			  pbl_size,
-			  p_chain->pbl_sp.p_virt_table,
-			  p_chain->pbl_sp.p_phys_table);
+
+	if (!p_chain->b_external_pbl)
+		dma_free_coherent(&cdev->pdev->dev,
+				  pbl_size,
+				  p_chain->pbl_sp.p_virt_table,
+				  p_chain->pbl_sp.p_phys_table);
 out:
 	vfree(p_chain->pbl.pp_virt_addr_tbl);
+	p_chain->pbl.pp_virt_addr_tbl = NULL;
 }
 
 void qed_chain_free(struct qed_dev *cdev, struct qed_chain *p_chain)
@@ -3174,7 +3177,10 @@ qed_chain_alloc_single(struct qed_dev *cdev, struct qed_chain *p_chain)
 	return 0;
 }
 
-static int qed_chain_alloc_pbl(struct qed_dev *cdev, struct qed_chain *p_chain)
+static int
+qed_chain_alloc_pbl(struct qed_dev *cdev,
+		    struct qed_chain *p_chain,
+		    struct qed_chain_ext_pbl *ext_pbl)
 {
 	u32 page_cnt = p_chain->page_cnt, size, i;
 	dma_addr_t p_phys = 0, p_pbl_phys = 0;
@@ -3194,8 +3200,16 @@ static int qed_chain_alloc_pbl(struct qed_dev *cdev, struct qed_chain *p_chain)
 	 * should be saved to allow its freeing during the error flow.
 	 */
 	size = page_cnt * QED_CHAIN_PBL_ENTRY_SIZE;
-	p_pbl_virt = dma_alloc_coherent(&cdev->pdev->dev,
-					size, &p_pbl_phys, GFP_KERNEL);
+
+	if (!ext_pbl) {
+		p_pbl_virt = dma_alloc_coherent(&cdev->pdev->dev,
+						size, &p_pbl_phys, GFP_KERNEL);
+	} else {
+		p_pbl_virt = ext_pbl->p_pbl_virt;
+		p_pbl_phys = ext_pbl->p_pbl_phys;
+		p_chain->b_external_pbl = true;
+	}
+
 	qed_chain_init_pbl_mem(p_chain, p_pbl_virt, p_pbl_phys,
 			       pp_virt_addr_tbl);
 	if (!p_pbl_virt)
@@ -3228,7 +3242,10 @@ int qed_chain_alloc(struct qed_dev *cdev,
 		    enum qed_chain_use_mode intended_use,
 		    enum qed_chain_mode mode,
 		    enum qed_chain_cnt_type cnt_type,
-		    u32 num_elems, size_t elem_size, struct qed_chain *p_chain)
+		    u32 num_elems,
+		    size_t elem_size,
+		    struct qed_chain *p_chain,
+		    struct qed_chain_ext_pbl *ext_pbl)
 {
 	u32 page_cnt;
 	int rc = 0;
@@ -3259,7 +3276,7 @@ int qed_chain_alloc(struct qed_dev *cdev,
 		rc = qed_chain_alloc_single(cdev, p_chain);
 		break;
 	case QED_CHAIN_MODE_PBL:
-		rc = qed_chain_alloc_pbl(cdev, p_chain);
+		rc = qed_chain_alloc_pbl(cdev, p_chain, ext_pbl);
 		break;
 	}
 	if (rc)

drivers/net/ethernet/qlogic/qed/qed_dev_api.h

@@ -307,6 +307,7 @@ int qed_dmae_host2host(struct qed_hwfn *p_hwfn,
  * @param num_elems
  * @param elem_size
  * @param p_chain
+ * @param ext_pbl - a possible external PBL
  *
  * @return int
  */
@@ -315,7 +316,9 @@ qed_chain_alloc(struct qed_dev *cdev,
 		enum qed_chain_use_mode intended_use,
 		enum qed_chain_mode mode,
 		enum qed_chain_cnt_type cnt_type,
-		u32 num_elems, size_t elem_size, struct qed_chain *p_chain);
+		u32 num_elems,
+		size_t elem_size,
+		struct qed_chain *p_chain, struct qed_chain_ext_pbl *ext_pbl);
 
 /**
  * @brief qed_chain_free - Free chain DMA memory

drivers/net/ethernet/qlogic/qed/qed_iscsi.c

@@ -752,7 +752,7 @@ static int qed_iscsi_allocate_connection(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_USE_TO_CONSUME_PRODUCE,
 			     QED_CHAIN_MODE_PBL,
 			     QED_CHAIN_CNT_TYPE_U16,
-			     r2tq_num_elements, 0x80, &p_conn->r2tq);
+			     r2tq_num_elements, 0x80, &p_conn->r2tq, NULL);
 	if (rc)
 		goto nomem_r2tq;
 
@@ -763,7 +763,7 @@ static int qed_iscsi_allocate_connection(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_MODE_PBL,
 			     QED_CHAIN_CNT_TYPE_U16,
 			     uhq_num_elements,
-			     sizeof(struct iscsi_uhqe), &p_conn->uhq);
+			     sizeof(struct iscsi_uhqe), &p_conn->uhq, NULL);
 	if (rc)
 		goto nomem_uhq;
 
@@ -773,7 +773,7 @@ static int qed_iscsi_allocate_connection(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_MODE_PBL,
 			     QED_CHAIN_CNT_TYPE_U16,
 			     xhq_num_elements,
-			     sizeof(struct iscsi_xhqe), &p_conn->xhq);
+			     sizeof(struct iscsi_xhqe), &p_conn->xhq, NULL);
 	if (rc)
 		goto nomem;
 

drivers/net/ethernet/qlogic/qed/qed_ll2.c

@@ -1056,7 +1056,7 @@ qed_ll2_acquire_connection_rx(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_CNT_TYPE_U16,
 			     p_ll2_info->input.rx_num_desc,
 			     sizeof(struct core_rx_bd),
-			     &p_ll2_info->rx_queue.rxq_chain);
+			     &p_ll2_info->rx_queue.rxq_chain, NULL);
 	if (rc) {
 		DP_NOTICE(p_hwfn, "Failed to allocate ll2 rxq chain\n");
 		goto out;
@@ -1078,7 +1078,7 @@ qed_ll2_acquire_connection_rx(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_CNT_TYPE_U16,
 			     p_ll2_info->input.rx_num_desc,
 			     sizeof(struct core_rx_fast_path_cqe),
-			     &p_ll2_info->rx_queue.rcq_chain);
+			     &p_ll2_info->rx_queue.rcq_chain, NULL);
 	if (rc) {
 		DP_NOTICE(p_hwfn, "Failed to allocate ll2 rcq chain\n");
 		goto out;
@@ -1108,7 +1108,7 @@ static int qed_ll2_acquire_connection_tx(struct qed_hwfn *p_hwfn,
 			     QED_CHAIN_CNT_TYPE_U16,
 			     p_ll2_info->input.tx_num_desc,
 			     sizeof(struct core_tx_bd),
-			     &p_ll2_info->tx_queue.txq_chain);
+			     &p_ll2_info->tx_queue.txq_chain, NULL);
 	if (rc)
 		goto out;
 

drivers/net/ethernet/qlogic/qed/qed_spq.c

@@ -419,7 +419,7 @@ int qed_eq_alloc(struct qed_hwfn *p_hwfn, u16 num_elem)
 			    QED_CHAIN_CNT_TYPE_U16,
 			    num_elem,
 			    sizeof(union event_ring_element),
-			    &p_eq->chain))
+			    &p_eq->chain, NULL))
 		goto eq_allocate_fail;
 
 	/* register EQ completion on the SP SB */
@@ -547,7 +547,7 @@ int qed_spq_alloc(struct qed_hwfn *p_hwfn)
 			    QED_CHAIN_CNT_TYPE_U16,
 			    0,   /* N/A when the mode is SINGLE */
 			    sizeof(struct slow_path_element),
-			    &p_spq->chain))
+			    &p_spq->chain, NULL))
 		goto spq_allocate_fail;
 
 	/* allocate and fill the SPQ elements (incl. ramrod data list) */
@@ -953,7 +953,7 @@ int qed_consq_alloc(struct qed_hwfn *p_hwfn)
 			    QED_CHAIN_MODE_PBL,
 			    QED_CHAIN_CNT_TYPE_U16,
 			    QED_CHAIN_PAGE_SIZE / 0x80,
-			    0x80, &p_consq->chain))
+			    0x80, &p_consq->chain, NULL))
 		goto consq_allocate_fail;
 
 	p_hwfn->p_consq = p_consq;

drivers/net/ethernet/qlogic/qede/qede_main.c

@@ -1317,8 +1317,7 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev, struct qede_rx_queue *rxq)
 					    QED_CHAIN_CNT_TYPE_U16,
 					    RX_RING_SIZE,
 					    sizeof(struct eth_rx_bd),
-					    &rxq->rx_bd_ring);
-
+					    &rxq->rx_bd_ring, NULL);
 	if (rc)
 		goto err;
 
@@ -1329,7 +1328,7 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev, struct qede_rx_queue *rxq)
 					    QED_CHAIN_CNT_TYPE_U16,
 					    RX_RING_SIZE,
 					    sizeof(union eth_rx_cqe),
-					    &rxq->rx_comp_ring);
+					    &rxq->rx_comp_ring, NULL);
 	if (rc)
 		goto err;
 
@@ -1387,7 +1386,8 @@ static int qede_alloc_mem_txq(struct qede_dev *edev, struct qede_tx_queue *txq)
 					    QED_CHAIN_MODE_PBL,
 					    QED_CHAIN_CNT_TYPE_U16,
 					    txq->num_tx_buffers,
-					    sizeof(*p_virt), &txq->tx_pbl);
+					    sizeof(*p_virt),
+					    &txq->tx_pbl, NULL);
 	if (rc)
 		goto err;
 

include/linux/qed/qed_chain.h

@@ -80,6 +80,11 @@ struct qed_chain_pbl_u32 {
 	u32 cons_page_idx;
 };
 
+struct qed_chain_ext_pbl {
+	dma_addr_t p_pbl_phys;
+	void *p_pbl_virt;
+};
+
 struct qed_chain_u16 {
 	/* Cyclic index of next element to produce/consme */
 	u16 prod_idx;
@@ -155,6 +160,8 @@ struct qed_chain {
 	u32 size;
 
 	u8 intended_use;
+
+	bool b_external_pbl;
 };
 
 #define QED_CHAIN_PBL_ENTRY_SIZE        (8)

include/linux/qed/qed_if.h

@@ -634,7 +634,8 @@ struct qed_common_ops {
 				       enum qed_chain_cnt_type cnt_type,
 				       u32 num_elems,
 				       size_t elem_size,
-				       struct qed_chain *p_chain);
+				       struct qed_chain *p_chain,
+				       struct qed_chain_ext_pbl *ext_pbl);
 
 	void		(*chain_free)(struct qed_dev *cdev,
 				      struct qed_chain *p_chain);