qed: Revisit chain implementation

RoCE driver is going to need a 32-bit chain [current chain implementation
for qed* currently supports only 16-bit producer/consumer chains].

This patch adds said support, as well as doing other slight tweaks and
modifications to qed's chain API.

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

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

@@ -17,6 +17,7 @@
 #include <linux/pci.h>
 #include <linux/slab.h>
 #include <linux/string.h>
+#include <linux/vmalloc.h>
 #include <linux/etherdevice.h>
 #include <linux/qed/qed_chain.h>
 #include <linux/qed/qed_if.h>
@@ -1779,92 +1780,285 @@ void qed_hw_remove(struct qed_dev *cdev)
 	qed_iov_free_hw_info(cdev);
 }
 
+static void qed_chain_free_next_ptr(struct qed_dev *cdev,
+				    struct qed_chain *p_chain)
+{
+	void *p_virt = p_chain->p_virt_addr, *p_virt_next = NULL;
+	dma_addr_t p_phys = p_chain->p_phys_addr, p_phys_next = 0;
+	struct qed_chain_next *p_next;
+	u32 size, i;
+
+	if (!p_virt)
+		return;
+
+	size = p_chain->elem_size * p_chain->usable_per_page;
+
+	for (i = 0; i < p_chain->page_cnt; i++) {
+		if (!p_virt)
+			break;
+
+		p_next = (struct qed_chain_next *)((u8 *)p_virt + size);
+		p_virt_next = p_next->next_virt;
+		p_phys_next = HILO_DMA_REGPAIR(p_next->next_phys);
+
+		dma_free_coherent(&cdev->pdev->dev,
+				  QED_CHAIN_PAGE_SIZE, p_virt, p_phys);
+
+		p_virt = p_virt_next;
+		p_phys = p_phys_next;
+	}
+}
+
+static void qed_chain_free_single(struct qed_dev *cdev,
+				  struct qed_chain *p_chain)
+{
+	if (!p_chain->p_virt_addr)
+		return;
+
+	dma_free_coherent(&cdev->pdev->dev,
+			  QED_CHAIN_PAGE_SIZE,
+			  p_chain->p_virt_addr, p_chain->p_phys_addr);
+}
+
+static void qed_chain_free_pbl(struct qed_dev *cdev, struct qed_chain *p_chain)
+{
+	void **pp_virt_addr_tbl = p_chain->pbl.pp_virt_addr_tbl;
+	u32 page_cnt = p_chain->page_cnt, i, pbl_size;
+	u8 *p_pbl_virt = p_chain->pbl.p_virt_table;
+
+	if (!pp_virt_addr_tbl)
+		return;
+
+	if (!p_chain->pbl.p_virt_table)
+		goto out;
+
+	for (i = 0; i < page_cnt; i++) {
+		if (!pp_virt_addr_tbl[i])
+			break;
+
+		dma_free_coherent(&cdev->pdev->dev,
+				  QED_CHAIN_PAGE_SIZE,
+				  pp_virt_addr_tbl[i],
+				  *(dma_addr_t *)p_pbl_virt);
+
+		p_pbl_virt += QED_CHAIN_PBL_ENTRY_SIZE;
+	}
+
+	pbl_size = page_cnt * QED_CHAIN_PBL_ENTRY_SIZE;
+	dma_free_coherent(&cdev->pdev->dev,
+			  pbl_size,
+			  p_chain->pbl.p_virt_table, p_chain->pbl.p_phys_table);
+out:
+	vfree(p_chain->pbl.pp_virt_addr_tbl);
+}
+
+void qed_chain_free(struct qed_dev *cdev, struct qed_chain *p_chain)
+{
+	switch (p_chain->mode) {
+	case QED_CHAIN_MODE_NEXT_PTR:
+		qed_chain_free_next_ptr(cdev, p_chain);
+		break;
+	case QED_CHAIN_MODE_SINGLE:
+		qed_chain_free_single(cdev, p_chain);
+		break;
+	case QED_CHAIN_MODE_PBL:
+		qed_chain_free_pbl(cdev, p_chain);
+		break;
+	}
+}
+
+static int
+qed_chain_alloc_sanity_check(struct qed_dev *cdev,
+			     enum qed_chain_cnt_type cnt_type,
+			     size_t elem_size, u32 page_cnt)
+{
+	u64 chain_size = ELEMS_PER_PAGE(elem_size) * page_cnt;
+
+	/* The actual chain size can be larger than the maximal possible value
+	 * after rounding up the requested elements number to pages, and after
+	 * taking into acount the unusuable elements (next-ptr elements).
+	 * The size of a "u16" chain can be (U16_MAX + 1) since the chain
+	 * size/capacity fields are of a u32 type.
+	 */
+	if ((cnt_type == QED_CHAIN_CNT_TYPE_U16 &&
+	     chain_size > 0x10000) ||
+	    (cnt_type == QED_CHAIN_CNT_TYPE_U32 &&
+	     chain_size > 0x100000000ULL)) {
+		DP_NOTICE(cdev,
+			  "The actual chain size (0x%llx) is larger than the maximal possible value\n",
+			  chain_size);
+		return -EINVAL;
+	}
+
+	return 0;
+}
+
+static int
+qed_chain_alloc_next_ptr(struct qed_dev *cdev, struct qed_chain *p_chain)
+{
+	void *p_virt = NULL, *p_virt_prev = NULL;
+	dma_addr_t p_phys = 0;
+	u32 i;
+
+	for (i = 0; i < p_chain->page_cnt; i++) {
+		p_virt = dma_alloc_coherent(&cdev->pdev->dev,
+					    QED_CHAIN_PAGE_SIZE,
+					    &p_phys, GFP_KERNEL);
+		if (!p_virt) {
+			DP_NOTICE(cdev, "Failed to allocate chain memory\n");
+			return -ENOMEM;
+		}
+
+		if (i == 0) {
+			qed_chain_init_mem(p_chain, p_virt, p_phys);
+			qed_chain_reset(p_chain);
+		} else {
+			qed_chain_init_next_ptr_elem(p_chain, p_virt_prev,
+						     p_virt, p_phys);
+		}
+
+		p_virt_prev = p_virt;
+	}
+	/* Last page's next element should point to the beginning of the
+	 * chain.
+	 */
+	qed_chain_init_next_ptr_elem(p_chain, p_virt_prev,
+				     p_chain->p_virt_addr,
+				     p_chain->p_phys_addr);
+
+	return 0;
+}
+
+static int
+qed_chain_alloc_single(struct qed_dev *cdev, struct qed_chain *p_chain)
+{
+	dma_addr_t p_phys = 0;
+	void *p_virt = NULL;
+
+	p_virt = dma_alloc_coherent(&cdev->pdev->dev,
+				    QED_CHAIN_PAGE_SIZE, &p_phys, GFP_KERNEL);
+	if (!p_virt) {
+		DP_NOTICE(cdev, "Failed to allocate chain memory\n");
+		return -ENOMEM;
+	}
+
+	qed_chain_init_mem(p_chain, p_virt, p_phys);
+	qed_chain_reset(p_chain);
+
+	return 0;
+}
+
+static int qed_chain_alloc_pbl(struct qed_dev *cdev, struct qed_chain *p_chain)
+{
+	u32 page_cnt = p_chain->page_cnt, size, i;
+	dma_addr_t p_phys = 0, p_pbl_phys = 0;
+	void **pp_virt_addr_tbl = NULL;
+	u8 *p_pbl_virt = NULL;
+	void *p_virt = NULL;
+
+	size = page_cnt * sizeof(*pp_virt_addr_tbl);
+	pp_virt_addr_tbl = vmalloc(size);
+	if (!pp_virt_addr_tbl) {
+		DP_NOTICE(cdev,
+			  "Failed to allocate memory for the chain virtual addresses table\n");
+		return -ENOMEM;
+	}
+	memset(pp_virt_addr_tbl, 0, size);
+
+	/* The allocation of the PBL table is done with its full size, since it
+	 * is expected to be successive.
+	 * qed_chain_init_pbl_mem() is called even in a case of an allocation
+	 * failure, since pp_virt_addr_tbl was previously allocated, and it
+	 * 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);
+	qed_chain_init_pbl_mem(p_chain, p_pbl_virt, p_pbl_phys,
+			       pp_virt_addr_tbl);
+	if (!p_pbl_virt) {
+		DP_NOTICE(cdev, "Failed to allocate chain pbl memory\n");
+		return -ENOMEM;
+	}
+
+	for (i = 0; i < page_cnt; i++) {
+		p_virt = dma_alloc_coherent(&cdev->pdev->dev,
+					    QED_CHAIN_PAGE_SIZE,
+					    &p_phys, GFP_KERNEL);
+		if (!p_virt) {
+			DP_NOTICE(cdev, "Failed to allocate chain memory\n");
+			return -ENOMEM;
+		}
+
+		if (i == 0) {
+			qed_chain_init_mem(p_chain, p_virt, p_phys);
+			qed_chain_reset(p_chain);
+		}
+
+		/* Fill the PBL table with the physical address of the page */
+		*(dma_addr_t *)p_pbl_virt = p_phys;
+		/* Keep the virtual address of the page */
+		p_chain->pbl.pp_virt_addr_tbl[i] = p_virt;
+
+		p_pbl_virt += QED_CHAIN_PBL_ENTRY_SIZE;
+	}
+
+	return 0;
+}
+
 int qed_chain_alloc(struct qed_dev *cdev,
 		    enum qed_chain_use_mode intended_use,
 		    enum qed_chain_mode mode,
-		    u16 num_elems,
+		    enum qed_chain_cnt_type cnt_type,
-		    size_t elem_size,
+		    u32 num_elems, size_t elem_size, struct qed_chain *p_chain)
-		    struct qed_chain *p_chain)
 {
-	dma_addr_t p_pbl_phys = 0;
+	u32 page_cnt;
-	void *p_pbl_virt = NULL;
+	int rc = 0;
-	dma_addr_t p_phys = 0;
-	void *p_virt = NULL;
-	u16 page_cnt = 0;
-	size_t size;
 
 	if (mode == QED_CHAIN_MODE_SINGLE)
 		page_cnt = 1;
 	else
 		page_cnt = QED_CHAIN_PAGE_CNT(num_elems, elem_size, mode);
 
-	size = page_cnt * QED_CHAIN_PAGE_SIZE;
+	rc = qed_chain_alloc_sanity_check(cdev, cnt_type, elem_size, page_cnt);
-	p_virt = dma_alloc_coherent(&cdev->pdev->dev,
+	if (rc) {
-				    size, &p_phys, GFP_KERNEL);
+		DP_NOTICE(cdev,
-	if (!p_virt) {
+			  "Cannot allocate a chain with the given arguments:\n"
-		DP_NOTICE(cdev, "Failed to allocate chain mem\n");
+			  "[use_mode %d, mode %d, cnt_type %d, num_elems %d, elem_size %zu]\n",
+			  intended_use, mode, cnt_type, num_elems, elem_size);
+		return rc;
+	}
+
+	qed_chain_init_params(p_chain, page_cnt, (u8) elem_size, intended_use,
+			      mode, cnt_type);
+
+	switch (mode) {
+	case QED_CHAIN_MODE_NEXT_PTR:
+		rc = qed_chain_alloc_next_ptr(cdev, p_chain);
+		break;
+	case QED_CHAIN_MODE_SINGLE:
+		rc = qed_chain_alloc_single(cdev, p_chain);
+		break;
+	case QED_CHAIN_MODE_PBL:
+		rc = qed_chain_alloc_pbl(cdev, p_chain);
+		break;
+	}
+	if (rc)
 		goto nomem;
-	}
-
-	if (mode == QED_CHAIN_MODE_PBL) {
-		size = page_cnt * QED_CHAIN_PBL_ENTRY_SIZE;
-		p_pbl_virt = dma_alloc_coherent(&cdev->pdev->dev,
-						size, &p_pbl_phys,
-						GFP_KERNEL);
-		if (!p_pbl_virt) {
-			DP_NOTICE(cdev, "Failed to allocate chain pbl mem\n");
-			goto nomem;
-		}
-
-		qed_chain_pbl_init(p_chain, p_virt, p_phys, page_cnt,
-				   (u8)elem_size, intended_use,
-				   p_pbl_phys, p_pbl_virt);
-	} else {
-		qed_chain_init(p_chain, p_virt, p_phys, page_cnt,
-			       (u8)elem_size, intended_use, mode);
-	}
 
 	return 0;
 
 nomem:
-	dma_free_coherent(&cdev->pdev->dev,
+	qed_chain_free(cdev, p_chain);
-			  page_cnt * QED_CHAIN_PAGE_SIZE,
+	return rc;
-			  p_virt, p_phys);
-	dma_free_coherent(&cdev->pdev->dev,
-			  page_cnt * QED_CHAIN_PBL_ENTRY_SIZE,
-			  p_pbl_virt, p_pbl_phys);
-
-	return -ENOMEM;
 }
 
-void qed_chain_free(struct qed_dev *cdev,
+int qed_fw_l2_queue(struct qed_hwfn *p_hwfn, u16 src_id, u16 *dst_id)
-		    struct qed_chain *p_chain)
-{
-	size_t size;
-
-	if (!p_chain->p_virt_addr)
-		return;
-
-	if (p_chain->mode == QED_CHAIN_MODE_PBL) {
-		size = p_chain->page_cnt * QED_CHAIN_PBL_ENTRY_SIZE;
-		dma_free_coherent(&cdev->pdev->dev, size,
-				  p_chain->pbl.p_virt_table,
-				  p_chain->pbl.p_phys_table);
-	}
-
-	size = p_chain->page_cnt * QED_CHAIN_PAGE_SIZE;
-	dma_free_coherent(&cdev->pdev->dev, size,
-			  p_chain->p_virt_addr,
-			  p_chain->p_phys_addr);
-}
-
-int qed_fw_l2_queue(struct qed_hwfn *p_hwfn,
-		    u16 src_id, u16 *dst_id)
 {
 	if (src_id >= RESC_NUM(p_hwfn, QED_L2_QUEUE)) {
 		u16 min, max;
 
-		min = (u16)RESC_START(p_hwfn, QED_L2_QUEUE);
+		min = (u16) RESC_START(p_hwfn, QED_L2_QUEUE);
 		max = min + RESC_NUM(p_hwfn, QED_L2_QUEUE);
 		DP_NOTICE(p_hwfn,
 			  "l2_queue id [%d] is not valid, available indices [%d - %d]\n",

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

@@ -245,9 +245,8 @@ int
 qed_chain_alloc(struct qed_dev *cdev,
 		enum qed_chain_use_mode intended_use,
 		enum qed_chain_mode mode,
-		u16 num_elems,
+		enum qed_chain_cnt_type cnt_type,
-		size_t elem_size,
+		u32 num_elems, size_t elem_size, struct qed_chain *p_chain);
-		struct qed_chain *p_chain);
 
 /**
  * @brief qed_chain_free - Free chain DMA memory
@@ -255,8 +254,7 @@ qed_chain_alloc(struct qed_dev *cdev,
  * @param p_hwfn
  * @param p_chain
  */
-void qed_chain_free(struct qed_dev *cdev,
+void qed_chain_free(struct qed_dev *cdev, struct qed_chain *p_chain);
-		    struct qed_chain *p_chain);
 
 /**
  * @@brief qed_fw_l2_queue - Get absolute L2 queue ID

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

@@ -308,6 +308,7 @@ int qed_sp_pf_start(struct qed_hwfn *p_hwfn,
 	struct qed_spq_entry *p_ent = NULL;
 	struct qed_sp_init_data init_data;
 	int rc = -EINVAL;
+	u8 page_cnt;
 
 	/* update initial eq producer */
 	qed_eq_prod_update(p_hwfn,
@@ -350,8 +351,8 @@ int qed_sp_pf_start(struct qed_hwfn *p_hwfn,
 	/* Place EQ address in RAMROD */
 	DMA_REGPAIR_LE(p_ramrod->event_ring_pbl_addr,
 		       p_hwfn->p_eq->chain.pbl.p_phys_table);
-	p_ramrod->event_ring_num_pages = (u8)p_hwfn->p_eq->chain.page_cnt;
+	page_cnt = (u8)qed_chain_get_page_cnt(&p_hwfn->p_eq->chain);
-
+	p_ramrod->event_ring_num_pages = page_cnt;
 	DMA_REGPAIR_LE(p_ramrod->consolid_q_pbl_addr,
 		       p_hwfn->p_consq->chain.pbl.p_phys_table);
 

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

@@ -343,6 +343,7 @@ struct qed_eq *qed_eq_alloc(struct qed_hwfn *p_hwfn,
 	if (qed_chain_alloc(p_hwfn->cdev,
 			    QED_CHAIN_USE_TO_PRODUCE,
 			    QED_CHAIN_MODE_PBL,
+			    QED_CHAIN_CNT_TYPE_U16,
 			    num_elem,
 			    sizeof(union event_ring_element),
 			    &p_eq->chain)) {
@@ -416,10 +417,10 @@ int qed_eth_cqe_completion(struct qed_hwfn *p_hwfn,
 ***************************************************************************/
 void qed_spq_setup(struct qed_hwfn *p_hwfn)
 {
-	struct qed_spq		*p_spq	= p_hwfn->p_spq;
+	struct qed_spq *p_spq = p_hwfn->p_spq;
-	struct qed_spq_entry	*p_virt = NULL;
+	struct qed_spq_entry *p_virt = NULL;
-	dma_addr_t		p_phys	= 0;
+	dma_addr_t p_phys = 0;
-	unsigned int		i	= 0;
+	u32 i, capacity;
 
 	INIT_LIST_HEAD(&p_spq->pending);
 	INIT_LIST_HEAD(&p_spq->completion_pending);
@@ -431,7 +432,8 @@ void qed_spq_setup(struct qed_hwfn *p_hwfn)
 	p_phys	= p_spq->p_phys + offsetof(struct qed_spq_entry, ramrod);
 	p_virt	= p_spq->p_virt;
 
-	for (i = 0; i < p_spq->chain.capacity; i++) {
+	capacity = qed_chain_get_capacity(&p_spq->chain);
+	for (i = 0; i < capacity; i++) {
 		DMA_REGPAIR_LE(p_virt->elem.data_ptr, p_phys);
 
 		list_add_tail(&p_virt->list, &p_spq->free_pool);
@@ -459,9 +461,10 @@ void qed_spq_setup(struct qed_hwfn *p_hwfn)
 
 int qed_spq_alloc(struct qed_hwfn *p_hwfn)
 {
-	struct qed_spq		*p_spq	= NULL;
+	struct qed_spq_entry *p_virt = NULL;
-	dma_addr_t		p_phys	= 0;
+	struct qed_spq *p_spq = NULL;
-	struct qed_spq_entry	*p_virt = NULL;
+	dma_addr_t p_phys = 0;
+	u32 capacity;
 
 	/* SPQ struct */
 	p_spq =
@@ -475,6 +478,7 @@ int qed_spq_alloc(struct qed_hwfn *p_hwfn)
 	if (qed_chain_alloc(p_hwfn->cdev,
 			    QED_CHAIN_USE_TO_PRODUCE,
 			    QED_CHAIN_MODE_SINGLE,
+			    QED_CHAIN_CNT_TYPE_U16,
 			    0,   /* N/A when the mode is SINGLE */
 			    sizeof(struct slow_path_element),
 			    &p_spq->chain)) {
@@ -483,11 +487,11 @@ int qed_spq_alloc(struct qed_hwfn *p_hwfn)
 	}
 
 	/* allocate and fill the SPQ elements (incl. ramrod data list) */
+	capacity = qed_chain_get_capacity(&p_spq->chain);
 	p_virt = dma_alloc_coherent(&p_hwfn->cdev->pdev->dev,
-				    p_spq->chain.capacity *
+				    capacity *
 				    sizeof(struct qed_spq_entry),
-				    &p_phys,
+				    &p_phys, GFP_KERNEL);
-				    GFP_KERNEL);
 
 	if (!p_virt)
 		goto spq_allocate_fail;
@@ -507,16 +511,18 @@ spq_allocate_fail:
 void qed_spq_free(struct qed_hwfn *p_hwfn)
 {
 	struct qed_spq *p_spq = p_hwfn->p_spq;
+	u32 capacity;
 
 	if (!p_spq)
 		return;
 
-	if (p_spq->p_virt)
+	if (p_spq->p_virt) {
+		capacity = qed_chain_get_capacity(&p_spq->chain);
 		dma_free_coherent(&p_hwfn->cdev->pdev->dev,
-				  p_spq->chain.capacity *
+				  capacity *
 				  sizeof(struct qed_spq_entry),
-				  p_spq->p_virt,
+				  p_spq->p_virt, p_spq->p_phys);
-				  p_spq->p_phys);
+	}
 
 	qed_chain_free(p_hwfn->cdev, &p_spq->chain);
 	;
@@ -871,9 +877,9 @@ struct qed_consq *qed_consq_alloc(struct qed_hwfn *p_hwfn)
 	if (qed_chain_alloc(p_hwfn->cdev,
 			    QED_CHAIN_USE_TO_PRODUCE,
 			    QED_CHAIN_MODE_PBL,
+			    QED_CHAIN_CNT_TYPE_U16,
 			    QED_CHAIN_PAGE_SIZE / 0x80,
-			    0x80,
+			    0x80, &p_consq->chain)) {
-			    &p_consq->chain)) {
 		DP_NOTICE(p_hwfn, "Failed to allocate consq chain");
 		goto consq_allocate_fail;
 	}

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

@@ -2817,6 +2817,7 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev,
 	rc = edev->ops->common->chain_alloc(edev->cdev,
 					    QED_CHAIN_USE_TO_CONSUME_PRODUCE,
 					    QED_CHAIN_MODE_NEXT_PTR,
+					    QED_CHAIN_CNT_TYPE_U16,
 					    RX_RING_SIZE,
 					    sizeof(struct eth_rx_bd),
 					    &rxq->rx_bd_ring);
@@ -2828,6 +2829,7 @@ static int qede_alloc_mem_rxq(struct qede_dev *edev,
 	rc = edev->ops->common->chain_alloc(edev->cdev,
 					    QED_CHAIN_USE_TO_CONSUME,
 					    QED_CHAIN_MODE_PBL,
+					    QED_CHAIN_CNT_TYPE_U16,
 					    RX_RING_SIZE,
 					    sizeof(union eth_rx_cqe),
 					    &rxq->rx_comp_ring);
@@ -2879,9 +2881,9 @@ static int qede_alloc_mem_txq(struct qede_dev *edev,
 	rc = edev->ops->common->chain_alloc(edev->cdev,
 					    QED_CHAIN_USE_TO_CONSUME_PRODUCE,
 					    QED_CHAIN_MODE_PBL,
+					    QED_CHAIN_CNT_TYPE_U16,
 					    NUM_TX_BDS_MAX,
-					    sizeof(*p_virt),
+					    sizeof(*p_virt), &txq->tx_pbl);
-					    &txq->tx_pbl);
 	if (rc)
 		goto err;
 

include/linux/qed/qed_if.h

@@ -325,7 +325,8 @@ struct qed_common_ops {
 	int		(*chain_alloc)(struct qed_dev *cdev,
 				       enum qed_chain_use_mode intended_use,
 				       enum qed_chain_mode mode,
-				       u16 num_elems,
+				       enum qed_chain_cnt_type cnt_type,
+				       u32 num_elems,
 				       size_t elem_size,
 				       struct qed_chain *p_chain);