bnxt_en: add support to enable VF-representors

This patch is a part of a patch-set that introduces support for
VF-reps in the bnxt_en driver. The driver registers eswitch mode
get/set methods with the devlink interface that allow a user to
enable SRIOV switchdev mode. When enabled, the driver registers
a VF-rep netdev object for each VF with the stack. This can
essentially bring the VFs unders the management perview of the
hypervisor and applications such as OVS.

The next patch in the series, adds the RX/TX routines and a slim
netdev implementation for the VF-reps.

Signed-off-by: Sathya Perla <sathya.perla@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/ethernet/broadcom/Kconfig

@@ -193,6 +193,7 @@ config SYSTEMPORT
 config BNXT
 	tristate "Broadcom NetXtreme-C/E support"
 	depends on PCI
+	depends on MAY_USE_DEVLINK
 	select FW_LOADER
 	select LIBCRC32C
 	---help---

drivers/net/ethernet/broadcom/bnxt/Makefile

@@ -1,3 +1,3 @@
 obj-$(CONFIG_BNXT) += bnxt_en.o
 
-bnxt_en-y := bnxt.o bnxt_sriov.o bnxt_ethtool.o bnxt_dcb.o bnxt_ulp.o bnxt_xdp.o
+bnxt_en-y := bnxt.o bnxt_sriov.o bnxt_ethtool.o bnxt_dcb.o bnxt_ulp.o bnxt_xdp.o bnxt_vfr.o

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

@@ -57,6 +57,7 @@
 #include "bnxt_ethtool.h"
 #include "bnxt_dcb.h"
 #include "bnxt_xdp.h"
+#include "bnxt_vfr.h"
 
 #define BNXT_TX_TIMEOUT		(5 * HZ)
 
@@ -7539,8 +7540,10 @@ static void bnxt_remove_one(struct pci_dev *pdev)
 	struct net_device *dev = pci_get_drvdata(pdev);
 	struct bnxt *bp = netdev_priv(dev);
 
-	if (BNXT_PF(bp))
+	if (BNXT_PF(bp)) {
 		bnxt_sriov_disable(bp);
+		bnxt_dl_unregister(bp);
+	}
 
 	pci_disable_pcie_error_reporting(pdev);
 	unregister_netdev(dev);
@@ -7843,6 +7846,7 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 
 #ifdef CONFIG_BNXT_SRIOV
 	init_waitqueue_head(&bp->sriov_cfg_wait);
+	mutex_init(&bp->sriov_lock);
 #endif
 	bp->gro_func = bnxt_gro_func_5730x;
 	if (BNXT_CHIP_P4_PLUS(bp))
@@ -7934,6 +7938,9 @@ static int bnxt_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
 	if (rc)
 		goto init_err_clr_int;
 
+	if (BNXT_PF(bp))
+		bnxt_dl_register(bp);
+
 	netdev_info(dev, "%s found at mem %lx, node addr %pM\n",
 		    board_info[ent->driver_data].name,
 		    (long)pci_resource_start(pdev, 0), dev->dev_addr);

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

@@ -19,6 +19,7 @@
 #define DRV_VER_UPD	0
 
 #include <linux/interrupt.h>
+#include <net/devlink.h>
 
 struct tx_bd {
 	__le32 tx_bd_len_flags_type;
@@ -618,6 +619,8 @@ struct bnxt_tpa_info {
 
 #define BNXT_TPA_OUTER_L3_OFF(hdr_info)	\
 	((hdr_info) & 0x1ff)
+
+	u16			cfa_code; /* cfa_code in TPA start compl */
 };
 
 struct bnxt_rx_ring_info {
@@ -928,6 +931,23 @@ struct bnxt_test_info {
 #define BNXT_CAG_REG_LEGACY_INT_STATUS	0x4014
 #define BNXT_CAG_REG_BASE		0x300000
 
+struct bnxt_vf_rep_stats {
+	u64			packets;
+	u64			bytes;
+	u64			dropped;
+};
+
+struct bnxt_vf_rep {
+	struct bnxt			*bp;
+	struct net_device		*dev;
+	u16				vf_idx;
+	u16				tx_cfa_action;
+	u16				rx_cfa_code;
+
+	struct bnxt_vf_rep_stats	rx_stats;
+	struct bnxt_vf_rep_stats	tx_stats;
+};
+
 struct bnxt {
 	void __iomem		*bar0;
 	void __iomem		*bar1;
@@ -1208,6 +1228,12 @@ struct bnxt {
 	wait_queue_head_t	sriov_cfg_wait;
 	bool			sriov_cfg;
 #define BNXT_SRIOV_CFG_WAIT_TMO	msecs_to_jiffies(10000)
+
+	/* lock to protect VF-rep creation/cleanup via
+	 * multiple paths such as ->sriov_configure() and
+	 * devlink ->eswitch_mode_set()
+	 */
+	struct mutex		sriov_lock;
 #endif
 
 #define BNXT_NTP_FLTR_MAX_FLTR	4096
@@ -1234,6 +1260,12 @@ struct bnxt {
 	struct bnxt_led_info	leds[BNXT_MAX_LED];
 
 	struct bpf_prog		*xdp_prog;
+
+	/* devlink interface and vf-rep structs */
+	struct devlink		*dl;
+	enum devlink_eswitch_mode eswitch_mode;
+	struct bnxt_vf_rep	**vf_reps; /* array of vf-rep ptrs */
+	u16			*cfa_code_map; /* cfa_code -> vf_idx map */
 };
 
 #define BNXT_RX_STATS_OFFSET(counter)			\

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

@@ -18,6 +18,7 @@
 #include "bnxt.h"
 #include "bnxt_ulp.h"
 #include "bnxt_sriov.h"
+#include "bnxt_vfr.h"
 #include "bnxt_ethtool.h"
 
 #ifdef CONFIG_BNXT_SRIOV
@@ -587,6 +588,10 @@ void bnxt_sriov_disable(struct bnxt *bp)
 	if (!num_vfs)
 		return;
 
+	/* synchronize VF and VF-rep create and destroy */
+	mutex_lock(&bp->sriov_lock);
+	bnxt_vf_reps_destroy(bp);
+
 	if (pci_vfs_assigned(bp->pdev)) {
 		bnxt_hwrm_fwd_async_event_cmpl(
 			bp, NULL, ASYNC_EVENT_CMPL_EVENT_ID_PF_DRVR_UNLOAD);
@@ -597,6 +602,7 @@ void bnxt_sriov_disable(struct bnxt *bp)
 		/* Free the HW resources reserved for various VF's */
 		bnxt_hwrm_func_vf_resource_free(bp, num_vfs);
 	}
+	mutex_unlock(&bp->sriov_lock);
 
 	bnxt_free_vf_resources(bp);
 

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

@@ -0,0 +1,244 @@
+/* Broadcom NetXtreme-C/E network driver.
+ *
+ * Copyright (c) 2016-2017 Broadcom Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation.
+ */
+#include <linux/pci.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/rtnetlink.h>
+#include <linux/jhash.h>
+
+#include "bnxt_hsi.h"
+#include "bnxt.h"
+#include "bnxt_vfr.h"
+
+#define CFA_HANDLE_INVALID		0xffff
+
+static void __bnxt_vf_reps_destroy(struct bnxt *bp)
+{
+	u16 num_vfs = pci_num_vf(bp->pdev);
+	struct bnxt_vf_rep *vf_rep;
+	int i;
+
+	for (i = 0; i < num_vfs; i++) {
+		vf_rep = bp->vf_reps[i];
+		if (vf_rep) {
+			if (vf_rep->dev) {
+				/* if register_netdev failed, then netdev_ops
+				 * would have been set to NULL
+				 */
+				if (vf_rep->dev->netdev_ops)
+					unregister_netdev(vf_rep->dev);
+				free_netdev(vf_rep->dev);
+			}
+		}
+	}
+
+	kfree(bp->vf_reps);
+	bp->vf_reps = NULL;
+}
+
+void bnxt_vf_reps_destroy(struct bnxt *bp)
+{
+	bool closed = false;
+
+	if (bp->eswitch_mode != DEVLINK_ESWITCH_MODE_SWITCHDEV)
+		return;
+
+	if (!bp->vf_reps)
+		return;
+
+	/* Ensure that parent PF's and VF-reps' RX/TX has been quiesced
+	 * before proceeding with VF-rep cleanup.
+	 */
+	rtnl_lock();
+	if (netif_running(bp->dev)) {
+		bnxt_close_nic(bp, false, false);
+		closed = true;
+	}
+	bp->eswitch_mode = DEVLINK_ESWITCH_MODE_LEGACY;
+
+	if (closed)
+		bnxt_open_nic(bp, false, false);
+	rtnl_unlock();
+
+	/* Need to call vf_reps_destroy() outside of rntl_lock
+	 * as unregister_netdev takes rtnl_lock
+	 */
+	__bnxt_vf_reps_destroy(bp);
+}
+
+/* Use the OUI of the PF's perm addr and report the same mac addr
+ * for the same VF-rep each time
+ */
+static void bnxt_vf_rep_eth_addr_gen(u8 *src_mac, u16 vf_idx, u8 *mac)
+{
+	u32 addr;
+
+	ether_addr_copy(mac, src_mac);
+
+	addr = jhash(src_mac, ETH_ALEN, 0) + vf_idx;
+	mac[3] = (u8)(addr & 0xFF);
+	mac[4] = (u8)((addr >> 8) & 0xFF);
+	mac[5] = (u8)((addr >> 16) & 0xFF);
+}
+
+static void bnxt_vf_rep_netdev_init(struct bnxt *bp, struct bnxt_vf_rep *vf_rep,
+				    struct net_device *dev)
+{
+	struct net_device *pf_dev = bp->dev;
+
+	/* Just inherit all the featues of the parent PF as the VF-R
+	 * uses the RX/TX rings of the parent PF
+	 */
+	dev->hw_features = pf_dev->hw_features;
+	dev->gso_partial_features = pf_dev->gso_partial_features;
+	dev->vlan_features = pf_dev->vlan_features;
+	dev->hw_enc_features = pf_dev->hw_enc_features;
+	dev->features |= pf_dev->features;
+	bnxt_vf_rep_eth_addr_gen(bp->pf.mac_addr, vf_rep->vf_idx,
+				 dev->perm_addr);
+	ether_addr_copy(dev->dev_addr, dev->perm_addr);
+}
+
+static int bnxt_vf_reps_create(struct bnxt *bp)
+{
+	u16 num_vfs = pci_num_vf(bp->pdev);
+	struct bnxt_vf_rep *vf_rep;
+	struct net_device *dev;
+	int rc, i;
+
+	bp->vf_reps = kcalloc(num_vfs, sizeof(vf_rep), GFP_KERNEL);
+	if (!bp->vf_reps)
+		return -ENOMEM;
+
+	for (i = 0; i < num_vfs; i++) {
+		dev = alloc_etherdev(sizeof(*vf_rep));
+		if (!dev) {
+			rc = -ENOMEM;
+			goto err;
+		}
+
+		vf_rep = netdev_priv(dev);
+		bp->vf_reps[i] = vf_rep;
+		vf_rep->dev = dev;
+		vf_rep->bp = bp;
+		vf_rep->vf_idx = i;
+		vf_rep->tx_cfa_action = CFA_HANDLE_INVALID;
+
+		bnxt_vf_rep_netdev_init(bp, vf_rep, dev);
+		rc = register_netdev(dev);
+		if (rc) {
+			/* no need for unregister_netdev in cleanup */
+			dev->netdev_ops = NULL;
+			goto err;
+		}
+	}
+
+	bp->eswitch_mode = DEVLINK_ESWITCH_MODE_SWITCHDEV;
+	return 0;
+
+err:
+	netdev_info(bp->dev, "%s error=%d", __func__, rc);
+	__bnxt_vf_reps_destroy(bp);
+	return rc;
+}
+
+/* Devlink related routines */
+static int bnxt_dl_eswitch_mode_get(struct devlink *devlink, u16 *mode)
+{
+	struct bnxt *bp = bnxt_get_bp_from_dl(devlink);
+
+	*mode = bp->eswitch_mode;
+	return 0;
+}
+
+static int bnxt_dl_eswitch_mode_set(struct devlink *devlink, u16 mode)
+{
+	struct bnxt *bp = bnxt_get_bp_from_dl(devlink);
+	int rc = 0;
+
+	mutex_lock(&bp->sriov_lock);
+	if (bp->eswitch_mode == mode) {
+		netdev_info(bp->dev, "already in %s eswitch mode",
+			    mode == DEVLINK_ESWITCH_MODE_LEGACY ?
+			    "legacy" : "switchdev");
+		rc = -EINVAL;
+		goto done;
+	}
+
+	switch (mode) {
+	case DEVLINK_ESWITCH_MODE_LEGACY:
+		bnxt_vf_reps_destroy(bp);
+		break;
+
+	case DEVLINK_ESWITCH_MODE_SWITCHDEV:
+		if (pci_num_vf(bp->pdev) == 0) {
+			netdev_info(bp->dev,
+				    "Enable VFs before setting swtichdev mode");
+			rc = -EPERM;
+			goto done;
+		}
+		rc = bnxt_vf_reps_create(bp);
+		break;
+
+	default:
+		rc = -EINVAL;
+		goto done;
+	}
+done:
+	mutex_unlock(&bp->sriov_lock);
+	return rc;
+}
+
+static const struct devlink_ops bnxt_dl_ops = {
+	.eswitch_mode_set = bnxt_dl_eswitch_mode_set,
+	.eswitch_mode_get = bnxt_dl_eswitch_mode_get
+};
+
+int bnxt_dl_register(struct bnxt *bp)
+{
+	struct devlink *dl;
+	int rc;
+
+	if (!pci_find_ext_capability(bp->pdev, PCI_EXT_CAP_ID_SRIOV))
+		return 0;
+
+	if (bp->hwrm_spec_code < 0x10800) {
+		netdev_warn(bp->dev, "Firmware does not support SR-IOV E-Switch SWITCHDEV mode.\n");
+		return -ENOTSUPP;
+	}
+
+	dl = devlink_alloc(&bnxt_dl_ops, sizeof(struct bnxt_dl));
+	if (!dl) {
+		netdev_warn(bp->dev, "devlink_alloc failed");
+		return -ENOMEM;
+	}
+
+	bnxt_link_bp_to_dl(dl, bp);
+	bp->eswitch_mode = DEVLINK_ESWITCH_MODE_LEGACY;
+	rc = devlink_register(dl, &bp->pdev->dev);
+	if (rc) {
+		bnxt_link_bp_to_dl(dl, NULL);
+		devlink_free(dl);
+		netdev_warn(bp->dev, "devlink_register failed. rc=%d", rc);
+		return rc;
+	}
+
+	return 0;
+}
+
+void bnxt_dl_unregister(struct bnxt *bp)
+{
+	struct devlink *dl = bp->dl;
+
+	if (!dl)
+		return;
+
+	devlink_unregister(dl);
+	devlink_free(dl);
+}

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

@@ -0,0 +1,38 @@
+/* Broadcom NetXtreme-C/E network driver.
+ *
+ * Copyright (c) 2016-2017 Broadcom Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation.
+ */
+
+#ifndef BNXT_VFR_H
+#define BNXT_VFR_H
+
+#define	MAX_CFA_CODE			65536
+
+/* Struct to hold housekeeping info needed by devlink interface */
+struct bnxt_dl {
+	struct bnxt *bp;	/* back ptr to the controlling dev */
+};
+
+static inline struct bnxt *bnxt_get_bp_from_dl(struct devlink *dl)
+{
+	return ((struct bnxt_dl *)devlink_priv(dl))->bp;
+}
+
+static inline void bnxt_link_bp_to_dl(struct devlink *dl, struct bnxt *bp)
+{
+	struct bnxt_dl *bp_dl = devlink_priv(dl);
+
+	bp_dl->bp = bp;
+	if (bp)
+		bp->dl = dl;
+}
+
+int bnxt_dl_register(struct bnxt *bp);
+void bnxt_dl_unregister(struct bnxt *bp);
+void bnxt_vf_reps_destroy(struct bnxt *bp);
+
+#endif /* BNXT_VFR_H */