RDMA/hns: Initialize bonding resources

Allocate bond_grp resources for each card when the first device in
this card is registered. Block the initialization of VF when its PF
is a bonded slave, as VF is not supported in this case due to HW
constraints.

Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com>
Link: https://patch.msgid.link/20251112093510.3696363-3-huangjunxian6@hisilicon.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>

6 files changed, 251 insertions, 1 deletions

diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile
index baf592e6f21b..d07ef02c5231 100644
--- a/drivers/infiniband/hw/hns/Makefile
+++ b/drivers/infiniband/hw/hns/Makefile
@@ -4,11 +4,13 @@
 #
 
 ccflags-y :=  -I $(srctree)/drivers/net/ethernet/hisilicon/hns3
+ccflags-y +=  -I $(srctree)/drivers/net/ethernet/hisilicon/hns3/hns3pf
+ccflags-y +=  -I $(srctree)/drivers/net/ethernet/hisilicon/hns3/hns3_common
 ccflags-y +=  -I $(src)
 
 hns-roce-hw-v2-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_pd.o \
 	hns_roce_ah.o hns_roce_hem.o hns_roce_mr.o hns_roce_qp.o \
 	hns_roce_cq.o hns_roce_alloc.o hns_roce_db.o hns_roce_srq.o hns_roce_restrack.o \
-	hns_roce_debugfs.o hns_roce_hw_v2.o
+	hns_roce_debugfs.o hns_roce_hw_v2.o hns_roce_bond.o
 
 obj-$(CONFIG_INFINIBAND_HNS_HIP08) += hns-roce-hw-v2.o
diff --git a/drivers/infiniband/hw/hns/hns_roce_bond.c b/drivers/infiniband/hw/hns/hns_roce_bond.c
new file mode 100644
index 000000000000..918c1382fa65
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_bond.c
@@ -0,0 +1,192 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * Copyright (c) 2025 Hisilicon Limited.
+ */
+
+#include "hns_roce_device.h"
+#include "hns_roce_hw_v2.h"
+#include "hns_roce_bond.h"
+
+static DEFINE_XARRAY(roce_bond_xa);
+
+static struct net_device *get_upper_dev_from_ndev(struct net_device *net_dev)
+{
+	struct net_device *upper_dev;
+
+	rcu_read_lock();
+	upper_dev = netdev_master_upper_dev_get_rcu(net_dev);
+	dev_hold(upper_dev);
+	rcu_read_unlock();
+
+	return upper_dev;
+}
+
+static int get_netdev_bond_slave_id(struct net_device *net_dev,
+				    struct hns_roce_bond_group *bond_grp)
+{
+	int i;
+
+	for (i = 0; i < ROCE_BOND_FUNC_MAX; i++)
+		if (net_dev == bond_grp->bond_func_info[i].net_dev)
+			return i;
+
+	return -ENOENT;
+}
+
+struct hns_roce_bond_group *hns_roce_get_bond_grp(struct net_device *net_dev,
+						  u8 bus_num)
+{
+	struct hns_roce_die_info *die_info = xa_load(&roce_bond_xa, bus_num);
+	struct hns_roce_bond_group *bond_grp;
+	struct net_device *upper_dev = NULL;
+	int i;
+
+	if (!die_info)
+		return NULL;
+
+	for (i = 0; i < ROCE_BOND_NUM_MAX; i++) {
+		bond_grp = die_info->bgrps[i];
+		if (!bond_grp)
+			continue;
+		if (get_netdev_bond_slave_id(net_dev, bond_grp) >= 0)
+			return bond_grp;
+		if (bond_grp->upper_dev) {
+			upper_dev = get_upper_dev_from_ndev(net_dev);
+			if (bond_grp->upper_dev == upper_dev) {
+				dev_put(upper_dev);
+				return bond_grp;
+			}
+			dev_put(upper_dev);
+		}
+	}
+
+	return NULL;
+}
+
+static struct hns_roce_die_info *alloc_die_info(int bus_num)
+{
+	struct hns_roce_die_info *die_info;
+	int ret;
+
+	die_info = kzalloc(sizeof(*die_info), GFP_KERNEL);
+	if (!die_info)
+		return NULL;
+
+	ret = xa_err(xa_store(&roce_bond_xa, bus_num, die_info, GFP_KERNEL));
+	if (ret) {
+		kfree(die_info);
+		return NULL;
+	}
+
+	return die_info;
+}
+
+static void dealloc_die_info(struct hns_roce_die_info *die_info, u8 bus_num)
+{
+	xa_erase(&roce_bond_xa, bus_num);
+	kfree(die_info);
+}
+
+static int alloc_bond_id(struct hns_roce_bond_group *bond_grp)
+{
+	u8 bus_num = bond_grp->bus_num;
+	struct hns_roce_die_info *die_info = xa_load(&roce_bond_xa, bus_num);
+	int i;
+
+	if (!die_info) {
+		die_info = alloc_die_info(bus_num);
+		if (!die_info)
+			return -ENOMEM;
+	}
+
+	for (i = 0; i < ROCE_BOND_NUM_MAX; i++) {
+		if (die_info->bond_id_mask & BOND_ID(i))
+			continue;
+
+		die_info->bond_id_mask |= BOND_ID(i);
+		die_info->bgrps[i] = bond_grp;
+		bond_grp->bond_id = i;
+
+		return 0;
+	}
+
+	return -ENOSPC;
+}
+
+static int remove_bond_id(int bus_num, u8 bond_id)
+{
+	struct hns_roce_die_info *die_info = xa_load(&roce_bond_xa, bus_num);
+
+	if (bond_id >= ROCE_BOND_NUM_MAX)
+		return -EINVAL;
+
+	if (!die_info)
+		return -ENODEV;
+
+	die_info->bond_id_mask &= ~BOND_ID(bond_id);
+	die_info->bgrps[bond_id] = NULL;
+	if (!die_info->bond_id_mask)
+		dealloc_die_info(die_info, bus_num);
+
+	return 0;
+}
+
+int hns_roce_alloc_bond_grp(struct hns_roce_dev *hr_dev)
+{
+	struct hns_roce_bond_group *bgrps[ROCE_BOND_NUM_MAX];
+	struct hns_roce_bond_group *bond_grp;
+	u8 bus_num = get_hr_bus_num(hr_dev);
+	int ret;
+	int i;
+
+	if (xa_load(&roce_bond_xa, bus_num))
+		return 0;
+
+	for (i = 0; i < ROCE_BOND_NUM_MAX; i++) {
+		bond_grp = kvzalloc(sizeof(*bond_grp), GFP_KERNEL);
+		if (!bond_grp) {
+			ret = -ENOMEM;
+			goto mem_err;
+		}
+
+		bond_grp->bus_num = bus_num;
+
+		ret = alloc_bond_id(bond_grp);
+		if (ret) {
+			dev_err(hr_dev->dev,
+				"failed to alloc bond ID, ret = %d.\n", ret);
+			goto alloc_id_err;
+		}
+
+		bgrps[i] = bond_grp;
+	}
+
+	return 0;
+
+alloc_id_err:
+	kvfree(bond_grp);
+mem_err:
+	for (i--; i >= 0; i--) {
+		remove_bond_id(bgrps[i]->bus_num, bgrps[i]->bond_id);
+		kvfree(bgrps[i]);
+	}
+	return ret;
+}
+
+void hns_roce_dealloc_bond_grp(void)
+{
+	struct hns_roce_bond_group *bond_grp;
+	struct hns_roce_die_info *die_info;
+	unsigned long id;
+	int i;
+
+	xa_for_each(&roce_bond_xa, id, die_info) {
+		for (i = 0; i < ROCE_BOND_NUM_MAX; i++) {
+			bond_grp = die_info->bgrps[i];
+			if (!bond_grp)
+				continue;
+			remove_bond_id(bond_grp->bus_num, bond_grp->bond_id);
+			kvfree(bond_grp);
+		}
+	}
+}
diff --git a/drivers/infiniband/hw/hns/hns_roce_bond.h b/drivers/infiniband/hw/hns/hns_roce_bond.h
new file mode 100644
index 000000000000..61c52135588e
--- /dev/null
+++ b/drivers/infiniband/hw/hns/hns_roce_bond.h
@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: GPL-2.0+ */
+/*
+ * Copyright (c) 2025 Hisilicon Limited.
+ */
+
+#ifndef _HNS_ROCE_BOND_H
+#define _HNS_ROCE_BOND_H
+
+#include <linux/netdevice.h>
+#include <net/bonding.h>
+
+#define ROCE_BOND_FUNC_MAX 4
+#define ROCE_BOND_NUM_MAX 2
+
+#define BOND_ID(id) BIT(id)
+
+struct hns_roce_func_info {
+	struct net_device *net_dev;
+};
+
+struct hns_roce_bond_group {
+	struct net_device *upper_dev;
+	u8 bond_id;
+	u8 bus_num;
+	struct hns_roce_func_info bond_func_info[ROCE_BOND_FUNC_MAX];
+};
+
+struct hns_roce_die_info {
+	u8 bond_id_mask;
+	struct hns_roce_bond_group *bgrps[ROCE_BOND_NUM_MAX];
+};
+
+struct hns_roce_bond_group *hns_roce_get_bond_grp(struct net_device *net_dev,
+						  u8 bus_num);
+int hns_roce_alloc_bond_grp(struct hns_roce_dev *hr_dev);
+void hns_roce_dealloc_bond_grp(void);
+
+#endif
diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h
index 5ae37832059f..cc1402fc8943 100644
--- a/drivers/infiniband/hw/hns/hns_roce_device.h
+++ b/drivers/infiniband/hw/hns/hns_roce_device.h
@@ -154,6 +154,7 @@ enum {
 	HNS_ROCE_CAP_FLAG_SDI_MODE		= BIT(14),
 	HNS_ROCE_CAP_FLAG_STASH			= BIT(17),
 	HNS_ROCE_CAP_FLAG_CQE_INLINE		= BIT(19),
+	HNS_ROCE_CAP_FLAG_BOND                  = BIT(21),
 	HNS_ROCE_CAP_FLAG_SRQ_RECORD_DB         = BIT(22),
 };
 
diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
index f82bdd46a917..6750bad9bb53 100644
--- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
+++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c
@@ -43,11 +43,13 @@
 #include <rdma/ib_umem.h>
 #include <rdma/uverbs_ioctl.h>
 
+#include "hclge_main.h"
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_cmd.h"
 #include "hns_roce_hem.h"
 #include "hns_roce_hw_v2.h"
+#include "hns_roce_bond.h"
 
 #define CREATE_TRACE_POINTS
 #include "hns_roce_trace.h"
@@ -2270,6 +2272,9 @@ static int hns_roce_query_caps(struct hns_roce_dev *hr_dev)
 	caps->flags |= le16_to_cpu(resp_d->cap_flags_ex) <<
 		       HNS_ROCE_CAP_FLAGS_EX_SHIFT;
 
+	if (hr_dev->is_vf)
+		caps->flags &= ~HNS_ROCE_CAP_FLAG_BOND;
+
 	caps->num_cqs = 1 << hr_reg_read(resp_c, PF_CAPS_C_NUM_CQS);
 	caps->gid_table_len[0] = hr_reg_read(resp_c, PF_CAPS_C_MAX_GID);
 	caps->max_cqes = 1 << hr_reg_read(resp_c, PF_CAPS_C_CQ_DEPTH);
@@ -7260,6 +7265,7 @@ static int __init hns_roce_hw_v2_init(void)
 
 static void __exit hns_roce_hw_v2_exit(void)
 {
+	hns_roce_dealloc_bond_grp();
 	hnae3_unregister_client(&hns_roce_hw_v2_client);
 	hns_roce_cleanup_debugfs();
 }
diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c
index 8bca0b10c69e..7fa25586ccd8 100644
--- a/drivers/infiniband/hw/hns/hns_roce_main.c
+++ b/drivers/infiniband/hw/hns/hns_roce_main.c
@@ -40,6 +40,7 @@
 #include "hns_roce_device.h"
 #include "hns_roce_hem.h"
 #include "hns_roce_hw_v2.h"
+#include "hns_roce_bond.h"
 
 static int hns_roce_set_mac(struct hns_roce_dev *hr_dev, u32 port,
 			    const u8 *addr)
@@ -744,6 +745,16 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev)
 	ib_set_device_ops(ib_dev, hr_dev->hw->hns_roce_dev_ops);
 	ib_set_device_ops(ib_dev, &hns_roce_dev_ops);
 	ib_set_device_ops(ib_dev, &hns_roce_dev_restrack_ops);
+
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_BOND) {
+		ret = hns_roce_alloc_bond_grp(hr_dev);
+		if (ret) {
+			dev_err(dev, "failed to alloc bond_grp for bus %u, ret = %d\n",
+				get_hr_bus_num(hr_dev), ret);
+			return ret;
+		}
+	}
+
 	for (i = 0; i < hr_dev->caps.num_ports; i++) {
 		net_dev = get_hr_netdev(hr_dev, i);
 		if (!net_dev)