Merge branch 'bnxt_en-Updates-for-net-next'

Michael Chan says:

====================
bnxt_en: Updates for net-next.

This patchset includes these main changes:

1. Firmware spec. update.
2. Context memory sizing improvements for the hardware TQM block.
3. ethtool chip reset improvements and fixes for correctness.
4. Improve L2 doorbell mapping by mapping only up to the size specified
by firmware.  This allows the RoCE driver to map the remaining doorbell
space for its purpose, such as write-combining.
5. Improve ethtool -S channel statistics by showing only relevant ring
counters for non-combined channels.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/broadcom/bnxt/bnxt.c

@@ -1766,7 +1766,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 
 		rc = -EIO;
 		if (rx_err & RX_CMPL_ERRORS_BUFFER_ERROR_MASK) {
-			bnapi->cp_ring.rx_buf_errors++;
+			bnapi->cp_ring.sw_stats.rx.rx_buf_errors++;
 			if (!(bp->flags & BNXT_FLAG_CHIP_P5)) {
 				netdev_warn(bp->dev, "RX buffer error %x\n",
 					    rx_err);
@@ -1849,7 +1849,7 @@ static int bnxt_rx_pkt(struct bnxt *bp, struct bnxt_cp_ring_info *cpr,
 	} else {
 		if (rxcmp1->rx_cmp_cfa_code_errors_v2 & RX_CMP_L4_CS_ERR_BITS) {
 			if (dev->features & NETIF_F_RXCSUM)
-				bnapi->cp_ring.rx_l4_csum_errors++;
+				bnapi->cp_ring.sw_stats.rx.rx_l4_csum_errors++;
 		}
 	}
 
@@ -5045,8 +5045,7 @@ int bnxt_hwrm_vnic_cfg(struct bnxt *bp, u16 vnic_id)
 	req.dflt_ring_grp = cpu_to_le16(bp->grp_info[grp_idx].fw_grp_id);
 	req.lb_rule = cpu_to_le16(0xffff);
 vnic_mru:
-	req.mru = cpu_to_le16(bp->dev->mtu + ETH_HLEN + ETH_FCS_LEN +
-			      VLAN_HLEN);
+	req.mru = cpu_to_le16(bp->dev->mtu + ETH_HLEN + VLAN_HLEN);
 
 	req.vnic_id = cpu_to_le16(vnic->fw_vnic_id);
 #ifdef CONFIG_BNXT_SRIOV
@@ -5356,9 +5355,9 @@ static void bnxt_set_db(struct bnxt *bp, struct bnxt_db_info *db, u32 ring_type,
 {
 	if (bp->flags & BNXT_FLAG_CHIP_P5) {
 		if (BNXT_PF(bp))
-			db->doorbell = bp->bar1 + 0x10000;
+			db->doorbell = bp->bar1 + DB_PF_OFFSET_P5;
 		else
-			db->doorbell = bp->bar1 + 0x4000;
+			db->doorbell = bp->bar1 + DB_VF_OFFSET_P5;
 		switch (ring_type) {
 		case HWRM_RING_ALLOC_TX:
 			db->db_key64 = DBR_PATH_L2 | DBR_TYPE_SQ;
@@ -6365,6 +6364,7 @@ static int bnxt_hwrm_func_qcfg(struct bnxt *bp)
 {
 	struct hwrm_func_qcfg_input req = {0};
 	struct hwrm_func_qcfg_output *resp = bp->hwrm_cmd_resp_addr;
+	u32 min_db_offset = 0;
 	u16 flags;
 	int rc;
 
@@ -6413,6 +6413,21 @@ static int bnxt_hwrm_func_qcfg(struct bnxt *bp)
 	if (!bp->max_mtu)
 		bp->max_mtu = BNXT_MAX_MTU;
 
+	if (bp->db_size)
+		goto func_qcfg_exit;
+
+	if (bp->flags & BNXT_FLAG_CHIP_P5) {
+		if (BNXT_PF(bp))
+			min_db_offset = DB_PF_OFFSET_P5;
+		else
+			min_db_offset = DB_VF_OFFSET_P5;
+	}
+	bp->db_size = PAGE_ALIGN(le16_to_cpu(resp->l2_doorbell_bar_size_kb) *
+				 1024);
+	if (!bp->db_size || bp->db_size > pci_resource_len(bp->pdev, 2) ||
+	    bp->db_size <= min_db_offset)
+		bp->db_size = pci_resource_len(bp->pdev, 2);
+
 func_qcfg_exit:
 	mutex_unlock(&bp->hwrm_cmd_lock);
 	return rc;
@@ -6434,23 +6449,13 @@ static int bnxt_hwrm_func_backing_store_qcaps(struct bnxt *bp)
 	if (!rc) {
 		struct bnxt_ctx_pg_info *ctx_pg;
 		struct bnxt_ctx_mem_info *ctx;
-		int i;
+		int i, tqm_rings;
 
 		ctx = kzalloc(sizeof(*ctx), GFP_KERNEL);
 		if (!ctx) {
 			rc = -ENOMEM;
 			goto ctx_err;
 		}
-		ctx_pg = kzalloc(sizeof(*ctx_pg) * (bp->max_q + 1), GFP_KERNEL);
-		if (!ctx_pg) {
-			kfree(ctx);
-			rc = -ENOMEM;
-			goto ctx_err;
-		}
-		for (i = 0; i < bp->max_q + 1; i++, ctx_pg++)
-			ctx->tqm_mem[i] = ctx_pg;
-
-		bp->ctx = ctx;
 		ctx->qp_max_entries = le32_to_cpu(resp->qp_max_entries);
 		ctx->qp_min_qp1_entries = le16_to_cpu(resp->qp_min_qp1_entries);
 		ctx->qp_max_l2_entries = le16_to_cpu(resp->qp_max_l2_entries);
@@ -6483,6 +6488,20 @@ static int bnxt_hwrm_func_backing_store_qcaps(struct bnxt *bp)
 		ctx->tim_entry_size = le16_to_cpu(resp->tim_entry_size);
 		ctx->tim_max_entries = le32_to_cpu(resp->tim_max_entries);
 		ctx->ctx_kind_initializer = resp->ctx_kind_initializer;
+		ctx->tqm_fp_rings_count = resp->tqm_fp_rings_count;
+		if (!ctx->tqm_fp_rings_count)
+			ctx->tqm_fp_rings_count = bp->max_q;
+
+		tqm_rings = ctx->tqm_fp_rings_count + 1;
+		ctx_pg = kcalloc(tqm_rings, sizeof(*ctx_pg), GFP_KERNEL);
+		if (!ctx_pg) {
+			kfree(ctx);
+			rc = -ENOMEM;
+			goto ctx_err;
+		}
+		for (i = 0; i < tqm_rings; i++, ctx_pg++)
+			ctx->tqm_mem[i] = ctx_pg;
+		bp->ctx = ctx;
 	} else {
 		rc = 0;
 	}
@@ -6735,7 +6754,7 @@ static void bnxt_free_ctx_mem(struct bnxt *bp)
 		return;
 
 	if (ctx->tqm_mem[0]) {
-		for (i = 0; i < bp->max_q + 1; i++)
+		for (i = 0; i < ctx->tqm_fp_rings_count + 1; i++)
 			bnxt_free_ctx_pg_tbls(bp, ctx->tqm_mem[i]);
 		kfree(ctx->tqm_mem[0]);
 		ctx->tqm_mem[0] = NULL;
@@ -6756,6 +6775,7 @@ static int bnxt_alloc_ctx_mem(struct bnxt *bp)
 	struct bnxt_ctx_pg_info *ctx_pg;
 	struct bnxt_ctx_mem_info *ctx;
 	u32 mem_size, ena, entries;
+	u32 entries_sp, min;
 	u32 num_mr, num_ah;
 	u32 extra_srqs = 0;
 	u32 extra_qps = 0;
@@ -6845,14 +6865,17 @@ static int bnxt_alloc_ctx_mem(struct bnxt *bp)
 	ena |= FUNC_BACKING_STORE_CFG_REQ_ENABLES_TIM;
 
 skip_rdma:
-	entries = ctx->qp_max_l2_entries + extra_qps;
+	min = ctx->tqm_min_entries_per_ring;
+	entries_sp = ctx->vnic_max_vnic_entries + ctx->qp_max_l2_entries +
+		     2 * (extra_qps + ctx->qp_min_qp1_entries) + min;
+	entries_sp = roundup(entries_sp, ctx->tqm_entries_multiple);
+	entries = ctx->qp_max_l2_entries + extra_qps + ctx->qp_min_qp1_entries;
 	entries = roundup(entries, ctx->tqm_entries_multiple);
-	entries = clamp_t(u32, entries, ctx->tqm_min_entries_per_ring,
-			  ctx->tqm_max_entries_per_ring);
-	for (i = 0; i < bp->max_q + 1; i++) {
+	entries = clamp_t(u32, entries, min, ctx->tqm_max_entries_per_ring);
+	for (i = 0; i < ctx->tqm_fp_rings_count + 1; i++) {
 		ctx_pg = ctx->tqm_mem[i];
-		ctx_pg->entries = entries;
-		mem_size = ctx->tqm_entry_size * entries;
+		ctx_pg->entries = i ? entries : entries_sp;
+		mem_size = ctx->tqm_entry_size * ctx_pg->entries;
 		rc = bnxt_alloc_ctx_pg_tbls(bp, ctx_pg, mem_size, 1, false);
 		if (rc)
 			return rc;
@@ -10262,7 +10285,7 @@ static void bnxt_chk_missed_irq(struct bnxt *bp)
 			bnxt_dbg_hwrm_ring_info_get(bp,
 				DBG_RING_INFO_GET_REQ_RING_TYPE_L2_CMPL,
 				fw_ring_id, &val[0], &val[1]);
-			cpr->missed_irqs++;
+			cpr->sw_stats.cmn.missed_irqs++;
 		}
 	}
 }
@@ -10891,6 +10914,9 @@ static int bnxt_init_board(struct pci_dev *pdev, struct net_device *dev)
 	bp->dev = dev;
 	bp->pdev = pdev;
 
+	/* Doorbell BAR bp->bar1 is mapped after bnxt_fw_init_one_p2()
+	 * determines the BAR size.
+	 */
 	bp->bar0 = pci_ioremap_bar(pdev, 0);
 	if (!bp->bar0) {
 		dev_err(&pdev->dev, "Cannot map device registers, aborting\n");
@@ -10898,13 +10924,6 @@ static int bnxt_init_board(struct pci_dev *pdev, struct net_device *dev)
 		goto init_err_release;
 	}
 
-	bp->bar1 = pci_ioremap_bar(pdev, 2);
-	if (!bp->bar1) {
-		dev_err(&pdev->dev, "Cannot map doorbell registers, aborting\n");
-		rc = -ENOMEM;
-		goto init_err_release;
-	}
-
 	bp->bar2 = pci_ioremap_bar(pdev, 4);
 	if (!bp->bar2) {
 		dev_err(&pdev->dev, "Cannot map bar4 registers, aborting\n");
@@ -11826,6 +11845,16 @@ static int bnxt_pcie_dsn_get(struct bnxt *bp, u8 dsn[])
 	return 0;
 }
 
+static int bnxt_map_db_bar(struct bnxt *bp)
+{
+	if (!bp->db_size)
+		return -ENODEV;
+	bp->bar1 = pci_iomap(bp->pdev, 2, bp->db_size);
+	if (!bp->bar1)
+		return -ENOMEM;
+	return 0;
+}
+
 static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 {
 	struct net_device *dev;
@@ -11886,6 +11915,13 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (rc)
 		goto init_err_pci_clean;
 
+	rc = bnxt_map_db_bar(bp);
+	if (rc) {
+		dev_err(&pdev->dev, "Cannot map doorbell BAR rc = %d, aborting\n",
+			rc);
+		goto init_err_pci_clean;
+	}
+
 	dev->hw_features = NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM | NETIF_F_SG |
 			   NETIF_F_TSO | NETIF_F_TSO6 |
 			   NETIF_F_GSO_UDP_TUNNEL | NETIF_F_GSO_GRE |

drivers/net/ethernet/broadcom/bnxt/bnxt.h

@@ -537,6 +537,9 @@ struct nqe_cn {
 #define DBR_TYPE_NQ_ARM					(0xbULL << 60)
 #define DBR_TYPE_NULL					(0xfULL << 60)
 
+#define DB_PF_OFFSET_P5					0x10000
+#define DB_VF_OFFSET_P5					0x4000
+
 #define INVALID_HW_RING_ID	((u16)-1)
 
 /* The hardware supports certain page sizes.  Use the supported page sizes
@@ -907,6 +910,20 @@ struct bnxt_rx_ring_info {
 	struct page_pool	*page_pool;
 };
 
+struct bnxt_rx_sw_stats {
+	u64			rx_l4_csum_errors;
+	u64			rx_buf_errors;
+};
+
+struct bnxt_cmn_sw_stats {
+	u64			missed_irqs;
+};
+
+struct bnxt_sw_stats {
+	struct bnxt_rx_sw_stats rx;
+	struct bnxt_cmn_sw_stats cmn;
+};
+
 struct bnxt_cp_ring_info {
 	struct bnxt_napi	*bnapi;
 	u32			cp_raw_cons;
@@ -934,9 +951,8 @@ struct bnxt_cp_ring_info {
 	struct ctx_hw_stats	*hw_stats;
 	dma_addr_t		hw_stats_map;
 	u32			hw_stats_ctx_id;
-	u64			rx_l4_csum_errors;
-	u64			rx_buf_errors;
-	u64			missed_irqs;
+
+	struct bnxt_sw_stats	sw_stats;
 
 	struct bnxt_ring_struct	cp_ring_struct;
 
@@ -1357,6 +1373,7 @@ struct bnxt_ctx_mem_info {
 	u16	mrav_num_entries_units;
 	u8	tqm_entries_multiple;
 	u8	ctx_kind_initializer;
+	u8	tqm_fp_rings_count;
 
 	u32	flags;
 	#define BNXT_CTX_FLAG_INITED	0x01
@@ -1816,6 +1833,7 @@ struct bnxt {
 	/* ensure atomic 64-bit doorbell writes on 32-bit systems. */
 	spinlock_t		db_lock;
 #endif
+	int			db_size;
 
 #define BNXT_NTP_FLTR_MAX_FLTR	4096
 #define BNXT_NTP_FLTR_HASH_SIZE	512

drivers/net/ethernet/broadcom/bnxt/bnxt_ethtool.h

@@ -77,8 +77,12 @@ struct hwrm_dbg_cmn_output {
 #define BNXT_LED_DFLT_ENABLES(x)			\
 	cpu_to_le32(BNXT_LED_DFLT_ENA << (BNXT_LED_DFLT_ENA_SHIFT * (x)))
 
-#define BNXT_FW_RESET_AP	0xfffe
-#define BNXT_FW_RESET_CHIP	0xffff
+#define BNXT_FW_RESET_AP	(ETH_RESET_AP << ETH_RESET_SHARED_SHIFT)
+#define BNXT_FW_RESET_CHIP	((ETH_RESET_MGMT | ETH_RESET_IRQ |	\
+				  ETH_RESET_DMA | ETH_RESET_FILTER |	\
+				  ETH_RESET_OFFLOAD | ETH_RESET_MAC |	\
+				  ETH_RESET_PHY | ETH_RESET_RAM)	\
+				 << ETH_RESET_SHARED_SHIFT)
 
 extern const struct ethtool_ops bnxt_ethtool_ops;
 

drivers/net/ethernet/broadcom/bnxt/bnxt_sriov.c

@@ -651,7 +651,7 @@ static int bnxt_hwrm_func_cfg(struct bnxt *bp, int num_vfs)
 				  FUNC_CFG_REQ_ENABLES_NUM_VNICS |
 				  FUNC_CFG_REQ_ENABLES_NUM_HW_RING_GRPS);
 
-	mtu = bp->dev->mtu + ETH_HLEN + ETH_FCS_LEN + VLAN_HLEN;
+	mtu = bp->dev->mtu + ETH_HLEN + VLAN_HLEN;
 	req.mru = cpu_to_le16(mtu);
 	req.mtu = cpu_to_le16(mtu);
 

drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.c

@@ -104,7 +104,13 @@ static void bnxt_fill_msix_vecs(struct bnxt *bp, struct bnxt_msix_entry *ent)
 	for (i = 0; i < num_msix; i++) {
 		ent[i].vector = bp->irq_tbl[idx + i].vector;
 		ent[i].ring_idx = idx + i;
-		ent[i].db_offset = (idx + i) * 0x80;
+		if (bp->flags & BNXT_FLAG_CHIP_P5) {
+			ent[i].db_offset = DB_PF_OFFSET_P5;
+			if (BNXT_VF(bp))
+				ent[i].db_offset = DB_VF_OFFSET_P5;
+		} else {
+			ent[i].db_offset = (idx + i) * 0x80;
+		}
 	}
 }
 
@@ -475,6 +481,8 @@ struct bnxt_en_dev *bnxt_ulp_probe(struct net_device *dev)
 			edev->flags |= BNXT_EN_FLAG_ROCEV2_CAP;
 		edev->net = dev;
 		edev->pdev = bp->pdev;
+		edev->l2_db_size = bp->db_size;
+		edev->l2_db_size_nc = bp->db_size;
 		bp->edev = edev;
 	}
 	return bp->edev;

drivers/net/ethernet/broadcom/bnxt/bnxt_ulp.h

@@ -67,6 +67,14 @@ struct bnxt_en_dev {
 	#define BNXT_EN_FLAG_ULP_STOPPED	0x8
 	const struct bnxt_en_ops	*en_ops;
 	struct bnxt_ulp			ulp_tbl[BNXT_MAX_ULP];
+	int				l2_db_size;	/* Doorbell BAR size in
+							 * bytes mapped by L2
+							 * driver.
+							 */
+	int				l2_db_size_nc;	/* Doorbell BAR size in
+							 * bytes mapped as non-
+							 * cacheable.
+							 */
 };
 
 struct bnxt_en_ops {