Commit 1fb7f897 authored by Mark Bloch's avatar Mark Bloch Committed by Jason Gunthorpe
Browse files

RDMA: Support more than 255 rdma ports 

Current code uses many different types when dealing with a port of a RDMA
device: u8, unsigned int and u32. Switch to u32 to clean up the logic.

This allows us to make (at least) the core view consistent and use the
same type. Unfortunately not all places can be converted. Many uverbs
functions expect port to be u8 so keep those places in order not to break
UAPIs.  HW/Spec defined values must also not be changed.

With the switch to u32 we now can support devices with more than 255
ports. U32_MAX is reserved to make control logic a bit easier to deal
with. As a device with U32_MAX ports probably isn't going to happen any
time soon this seems like a non issue.

When a device with more than 255 ports is created uverbs will report the
RDMA device as having 255 ports as this is the max currently supported.

The verbs interface is not changed yet because the IBTA spec limits the
port size in too many places to be u8 and all applications that relies in
verbs won't be able to cope with this change. At this stage, we are
extending the interfaces that are using vendor channel solely

Once the limitation is lifted mlx5 in switchdev mode will be able to have
thousands of SFs created by the device. As the only instance of an RDMA
device that reports more than 255 ports will be a representor device and
it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other
ULPs aren't effected by this change and their sysfs/interfaces that are
exposes to userspace can remain unchanged.

While here cleanup some alignment issues and remove unneeded sanity
checks (mainly in rdmavt),

Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org


Signed-off-by: default avatarMark Bloch <mbloch@nvidia.com>
Signed-off-by: default avatarLeon Romanovsky <leonro@nvidia.com>
Signed-off-by: default avatarJason Gunthorpe <jgg@nvidia.com>
parent 847d19a4

drivers/infiniband/core/cache.c

+38 −46
Original line number Diff line number Diff line
@@ -121,7 +121,7 @@ struct ib_gid_table { 
	u32				default_gid_indices;

};



static void dispatch_gid_change_event(struct ib_device *ib_dev, u8 port)

static void dispatch_gid_change_event(struct ib_device *ib_dev, u32 port)

{

	struct ib_event event;
@@ -197,7 +197,7 @@ int ib_cache_gid_parse_type_str(const char *buf) 
}

EXPORT_SYMBOL(ib_cache_gid_parse_type_str);



static struct ib_gid_table *rdma_gid_table(struct ib_device *device, u8 port)

static struct ib_gid_table *rdma_gid_table(struct ib_device *device, u32 port)

{

	return device->port_data[port].cache.gid;

}
@@ -237,10 +237,10 @@ static void put_gid_ndev(struct rcu_head *head) 
static void free_gid_entry_locked(struct ib_gid_table_entry *entry)

{

	struct ib_device *device = entry->attr.device;

	u8 port_num = entry->attr.port_num;

	u32 port_num = entry->attr.port_num;

	struct ib_gid_table *table = rdma_gid_table(device, port_num);



	dev_dbg(&device->dev, "%s port=%d index=%d gid %pI6\n", __func__,

	dev_dbg(&device->dev, "%s port=%u index=%d gid %pI6\n", __func__,

		port_num, entry->attr.index, entry->attr.gid.raw);



	write_lock_irq(&table->rwlock);
@@ -282,7 +282,7 @@ static void free_gid_work(struct work_struct *work) 
	struct ib_gid_table_entry *entry =

		container_of(work, struct ib_gid_table_entry, del_work);

	struct ib_device *device = entry->attr.device;

	u8 port_num = entry->attr.port_num;

	u32 port_num = entry->attr.port_num;

	struct ib_gid_table *table = rdma_gid_table(device, port_num);



	mutex_lock(&table->lock);
@@ -379,7 +379,7 @@ static int add_roce_gid(struct ib_gid_table_entry *entry) 
 * @ix:		GID entry index to delete

 *

 */

static void del_gid(struct ib_device *ib_dev, u8 port,

static void del_gid(struct ib_device *ib_dev, u32 port,

		    struct ib_gid_table *table, int ix)

{

	struct roce_gid_ndev_storage *ndev_storage;
@@ -387,7 +387,7 @@ static void del_gid(struct ib_device *ib_dev, u8 port, 


	lockdep_assert_held(&table->lock);



	dev_dbg(&ib_dev->dev, "%s port=%d index=%d gid %pI6\n", __func__, port,

	dev_dbg(&ib_dev->dev, "%s port=%u index=%d gid %pI6\n", __func__, port,

		ix, table->data_vec[ix]->attr.gid.raw);



	write_lock_irq(&table->rwlock);
@@ -543,7 +543,7 @@ static void make_default_gid(struct net_device *dev, union ib_gid *gid) 
	addrconf_ifid_eui48(&gid->raw[8], dev);

}



static int __ib_cache_gid_add(struct ib_device *ib_dev, u8 port,

static int __ib_cache_gid_add(struct ib_device *ib_dev, u32 port,

			      union ib_gid *gid, struct ib_gid_attr *attr,

			      unsigned long mask, bool default_gid)

{
@@ -587,7 +587,7 @@ static int __ib_cache_gid_add(struct ib_device *ib_dev, u8 port, 
	return ret;

}



int ib_cache_gid_add(struct ib_device *ib_dev, u8 port,

int ib_cache_gid_add(struct ib_device *ib_dev, u32 port,

		     union ib_gid *gid, struct ib_gid_attr *attr)

{

	unsigned long mask = GID_ATTR_FIND_MASK_GID |
@@ -598,7 +598,7 @@ int ib_cache_gid_add(struct ib_device *ib_dev, u8 port, 
}



static int

_ib_cache_gid_del(struct ib_device *ib_dev, u8 port,

_ib_cache_gid_del(struct ib_device *ib_dev, u32 port,

		  union ib_gid *gid, struct ib_gid_attr *attr,

		  unsigned long mask, bool default_gid)

{
@@ -627,7 +627,7 @@ _ib_cache_gid_del(struct ib_device *ib_dev, u8 port, 
	return ret;

}



int ib_cache_gid_del(struct ib_device *ib_dev, u8 port,

int ib_cache_gid_del(struct ib_device *ib_dev, u32 port,

		     union ib_gid *gid, struct ib_gid_attr *attr)

{

	unsigned long mask = GID_ATTR_FIND_MASK_GID	  |
@@ -638,7 +638,7 @@ int ib_cache_gid_del(struct ib_device *ib_dev, u8 port, 
	return _ib_cache_gid_del(ib_dev, port, gid, attr, mask, false);

}



int ib_cache_gid_del_all_netdev_gids(struct ib_device *ib_dev, u8 port,

int ib_cache_gid_del_all_netdev_gids(struct ib_device *ib_dev, u32 port,

				     struct net_device *ndev)

{

	struct ib_gid_table *table;
@@ -683,7 +683,7 @@ const struct ib_gid_attr * 
rdma_find_gid_by_port(struct ib_device *ib_dev,

		      const union ib_gid *gid,

		      enum ib_gid_type gid_type,

		      u8 port, struct net_device *ndev)

		      u32 port, struct net_device *ndev)

{

	int local_index;

	struct ib_gid_table *table;
@@ -734,7 +734,7 @@ EXPORT_SYMBOL(rdma_find_gid_by_port); 
 *

 */

const struct ib_gid_attr *rdma_find_gid_by_filter(

	struct ib_device *ib_dev, const union ib_gid *gid, u8 port,

	struct ib_device *ib_dev, const union ib_gid *gid, u32 port,

	bool (*filter)(const union ib_gid *gid, const struct ib_gid_attr *,

		       void *),

	void *context)
@@ -818,7 +818,7 @@ static void release_gid_table(struct ib_device *device, 
	kfree(table);

}



static void cleanup_gid_table_port(struct ib_device *ib_dev, u8 port,

static void cleanup_gid_table_port(struct ib_device *ib_dev, u32 port,

				   struct ib_gid_table *table)

{

	int i;
@@ -834,7 +834,7 @@ static void cleanup_gid_table_port(struct ib_device *ib_dev, u8 port, 
	mutex_unlock(&table->lock);

}



void ib_cache_gid_set_default_gid(struct ib_device *ib_dev, u8 port,

void ib_cache_gid_set_default_gid(struct ib_device *ib_dev, u32 port,

				  struct net_device *ndev,

				  unsigned long gid_type_mask,

				  enum ib_cache_gid_default_mode mode)
@@ -867,7 +867,7 @@ void ib_cache_gid_set_default_gid(struct ib_device *ib_dev, u8 port, 
	}

}



static void gid_table_reserve_default(struct ib_device *ib_dev, u8 port,

static void gid_table_reserve_default(struct ib_device *ib_dev, u32 port,

				      struct ib_gid_table *table)

{

	unsigned int i;
@@ -884,7 +884,7 @@ static void gid_table_reserve_default(struct ib_device *ib_dev, u8 port, 


static void gid_table_release_one(struct ib_device *ib_dev)

{

	unsigned int p;

	u32 p;



	rdma_for_each_port (ib_dev, p) {

		release_gid_table(ib_dev, ib_dev->port_data[p].cache.gid);
@@ -895,7 +895,7 @@ static void gid_table_release_one(struct ib_device *ib_dev) 
static int _gid_table_setup_one(struct ib_device *ib_dev)

{

	struct ib_gid_table *table;

	unsigned int rdma_port;

	u32 rdma_port;



	rdma_for_each_port (ib_dev, rdma_port) {

		table = alloc_gid_table(
@@ -915,7 +915,7 @@ static int _gid_table_setup_one(struct ib_device *ib_dev) 


static void gid_table_cleanup_one(struct ib_device *ib_dev)

{

	unsigned int p;

	u32 p;



	rdma_for_each_port (ib_dev, p)

		cleanup_gid_table_port(ib_dev, p,
@@ -950,7 +950,7 @@ static int gid_table_setup_one(struct ib_device *ib_dev) 
 * Returns 0 on success or appropriate error code.

 *

 */

int rdma_query_gid(struct ib_device *device, u8 port_num,

int rdma_query_gid(struct ib_device *device, u32 port_num,

		   int index, union ib_gid *gid)

{

	struct ib_gid_table *table;
@@ -1014,7 +1014,7 @@ const struct ib_gid_attr *rdma_find_gid(struct ib_device *device, 
	unsigned long mask = GID_ATTR_FIND_MASK_GID |

			     GID_ATTR_FIND_MASK_GID_TYPE;

	struct ib_gid_attr gid_attr_val = {.ndev = ndev, .gid_type = gid_type};

	unsigned int p;

	u32 p;



	if (ndev)

		mask |= GID_ATTR_FIND_MASK_NETDEV;
@@ -1043,7 +1043,7 @@ const struct ib_gid_attr *rdma_find_gid(struct ib_device *device, 
EXPORT_SYMBOL(rdma_find_gid);



int ib_get_cached_pkey(struct ib_device *device,

		       u8                port_num,

		       u32               port_num,

		       int               index,

		       u16              *pkey)

{
@@ -1069,8 +1069,7 @@ int ib_get_cached_pkey(struct ib_device *device, 
}

EXPORT_SYMBOL(ib_get_cached_pkey);



int ib_get_cached_subnet_prefix(struct ib_device *device,

				u8                port_num,

int ib_get_cached_subnet_prefix(struct ib_device *device, u32 port_num,

				u64 *sn_pfx)

{

	unsigned long flags;
@@ -1086,10 +1085,8 @@ int ib_get_cached_subnet_prefix(struct ib_device *device, 
}

EXPORT_SYMBOL(ib_get_cached_subnet_prefix);



int ib_find_cached_pkey(struct ib_device *device,

			u8                port_num,

			u16               pkey,

			u16              *index)

int ib_find_cached_pkey(struct ib_device *device, u32 port_num,

			u16 pkey, u16 *index)

{

	struct ib_pkey_cache *cache;

	unsigned long flags;
@@ -1132,10 +1129,8 @@ int ib_find_cached_pkey(struct ib_device *device, 
}

EXPORT_SYMBOL(ib_find_cached_pkey);



int ib_find_exact_cached_pkey(struct ib_device *device,

			      u8                port_num,

			      u16               pkey,

			      u16              *index)

int ib_find_exact_cached_pkey(struct ib_device *device, u32 port_num,

			      u16 pkey, u16 *index)

{

	struct ib_pkey_cache *cache;

	unsigned long flags;
@@ -1169,9 +1164,7 @@ int ib_find_exact_cached_pkey(struct ib_device *device, 
}

EXPORT_SYMBOL(ib_find_exact_cached_pkey);



int ib_get_cached_lmc(struct ib_device *device,

		      u8                port_num,

		      u8                *lmc)

int ib_get_cached_lmc(struct ib_device *device, u32 port_num, u8 *lmc)

{

	unsigned long flags;

	int ret = 0;
@@ -1187,8 +1180,7 @@ int ib_get_cached_lmc(struct ib_device *device, 
}

EXPORT_SYMBOL(ib_get_cached_lmc);



int ib_get_cached_port_state(struct ib_device   *device,

			     u8                  port_num,

int ib_get_cached_port_state(struct ib_device *device, u32 port_num,

			     enum ib_port_state *port_state)

{

	unsigned long flags;
@@ -1222,7 +1214,7 @@ EXPORT_SYMBOL(ib_get_cached_port_state); 
 * code.

 */

const struct ib_gid_attr *

rdma_get_gid_attr(struct ib_device *device, u8 port_num, int index)

rdma_get_gid_attr(struct ib_device *device, u32 port_num, int index)

{

	const struct ib_gid_attr *attr = ERR_PTR(-ENODATA);

	struct ib_gid_table *table;
@@ -1263,7 +1255,7 @@ ssize_t rdma_query_gid_table(struct ib_device *device, 
	const struct ib_gid_attr *gid_attr;

	ssize_t num_entries = 0, ret;

	struct ib_gid_table *table;

	unsigned int port_num, i;

	u32 port_num, i;

	struct net_device *ndev;

	unsigned long flags;
@@ -1361,7 +1353,7 @@ struct net_device *rdma_read_gid_attr_ndev_rcu(const struct ib_gid_attr *attr) 
			container_of(attr, struct ib_gid_table_entry, attr);

	struct ib_device *device = entry->attr.device;

	struct net_device *ndev = ERR_PTR(-EINVAL);

	u8 port_num = entry->attr.port_num;

	u32 port_num = entry->attr.port_num;

	struct ib_gid_table *table;

	unsigned long flags;

	bool valid;
@@ -1441,7 +1433,7 @@ int rdma_read_gid_l2_fields(const struct ib_gid_attr *attr, 
EXPORT_SYMBOL(rdma_read_gid_l2_fields);



static int config_non_roce_gid_cache(struct ib_device *device,

				     u8 port, int gid_tbl_len)

				     u32 port, int gid_tbl_len)

{

	struct ib_gid_attr gid_attr = {};

	struct ib_gid_table *table;
@@ -1472,7 +1464,7 @@ static int config_non_roce_gid_cache(struct ib_device *device, 
}



static int

ib_cache_update(struct ib_device *device, u8 port, bool enforce_security)

ib_cache_update(struct ib_device *device, u32 port, bool enforce_security)

{

	struct ib_port_attr       *tprops = NULL;

	struct ib_pkey_cache      *pkey_cache = NULL, *old_pkey_cache;
@@ -1621,7 +1613,7 @@ EXPORT_SYMBOL(ib_dispatch_event); 


int ib_cache_setup_one(struct ib_device *device)

{

	unsigned int p;

	u32 p;

	int err;



	rwlock_init(&device->cache_lock);
@@ -1641,7 +1633,7 @@ int ib_cache_setup_one(struct ib_device *device) 


void ib_cache_release_one(struct ib_device *device)

{

	unsigned int p;

	u32 p;



	/*

	 * The release function frees all the cache elements.

drivers/infiniband/core/cm.c

+6 −6
Original line number Diff line number Diff line
@@ -202,7 +202,7 @@ static struct attribute *cm_counter_default_attrs[] = { 
struct cm_port {

	struct cm_device *cm_dev;

	struct ib_mad_agent *mad_agent;

	u8 port_num;

	u32 port_num;

	struct list_head cm_priv_prim_list;

	struct list_head cm_priv_altr_list;

	struct cm_counter_group counter_group[CM_COUNTER_GROUPS];
@@ -1631,7 +1631,7 @@ static bool cm_req_has_alt_path(struct cm_req_msg *req_msg) 
					       req_msg))));

}



static void cm_path_set_rec_type(struct ib_device *ib_device, u8 port_num,

static void cm_path_set_rec_type(struct ib_device *ib_device, u32 port_num,

				 struct sa_path_rec *path, union ib_gid *gid)

{

	if (ib_is_opa_gid(gid) && rdma_cap_opa_ah(ib_device, port_num))
@@ -1750,7 +1750,7 @@ static void cm_format_paths_from_req(struct cm_req_msg *req_msg, 
static u16 cm_get_bth_pkey(struct cm_work *work)

{

	struct ib_device *ib_dev = work->port->cm_dev->ib_device;

	u8 port_num = work->port->port_num;

	u32 port_num = work->port->port_num;

	u16 pkey_index = work->mad_recv_wc->wc->pkey_index;

	u16 pkey;

	int ret;
@@ -1778,7 +1778,7 @@ static void cm_opa_to_ib_sgid(struct cm_work *work, 
			      struct sa_path_rec *path)

{

	struct ib_device *dev = work->port->cm_dev->ib_device;

	u8 port_num = work->port->port_num;

	u32 port_num = work->port->port_num;



	if (rdma_cap_opa_ah(dev, port_num) &&

	    (ib_is_opa_gid(&path->sgid))) {
@@ -4334,7 +4334,7 @@ static int cm_add_one(struct ib_device *ib_device) 
	unsigned long flags;

	int ret;

	int count = 0;

	unsigned int i;

	u32 i;



	cm_dev = kzalloc(struct_size(cm_dev, port, ib_device->phys_port_cnt),

			 GFP_KERNEL);
@@ -4432,7 +4432,7 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data) 
		.clr_port_cap_mask = IB_PORT_CM_SUP

	};

	unsigned long flags;

	unsigned int i;

	u32 i;



	write_lock_irqsave(&cm.device_lock, flags);

	list_del(&cm_dev->list);

drivers/infiniband/core/cma.c

+9 −9
Original line number Diff line number Diff line
@@ -278,7 +278,7 @@ struct cma_device *cma_enum_devices_by_ibdev(cma_device_filter filter, 
}



int cma_get_default_gid_type(struct cma_device *cma_dev,

			     unsigned int port)

			     u32 port)

{

	if (!rdma_is_port_valid(cma_dev->device, port))

		return -EINVAL;
@@ -287,7 +287,7 @@ int cma_get_default_gid_type(struct cma_device *cma_dev, 
}



int cma_set_default_gid_type(struct cma_device *cma_dev,

			     unsigned int port,

			     u32 port,

			     enum ib_gid_type default_gid_type)

{

	unsigned long supported_gids;
@@ -310,7 +310,7 @@ int cma_set_default_gid_type(struct cma_device *cma_dev, 
	return 0;

}



int cma_get_default_roce_tos(struct cma_device *cma_dev, unsigned int port)

int cma_get_default_roce_tos(struct cma_device *cma_dev, u32 port)

{

	if (!rdma_is_port_valid(cma_dev->device, port))

		return -EINVAL;
@@ -318,7 +318,7 @@ int cma_get_default_roce_tos(struct cma_device *cma_dev, unsigned int port) 
	return cma_dev->default_roce_tos[port - rdma_start_port(cma_dev->device)];

}



int cma_set_default_roce_tos(struct cma_device *cma_dev, unsigned int port,

int cma_set_default_roce_tos(struct cma_device *cma_dev, u32 port,

			     u8 default_roce_tos)

{

	if (!rdma_is_port_valid(cma_dev->device, port))
@@ -553,7 +553,7 @@ static int cma_translate_addr(struct sockaddr *addr, struct rdma_dev_addr *dev_a 
}



static const struct ib_gid_attr *

cma_validate_port(struct ib_device *device, u8 port,

cma_validate_port(struct ib_device *device, u32 port,

		  enum ib_gid_type gid_type,

		  union ib_gid *gid,

		  struct rdma_id_private *id_priv)
@@ -611,7 +611,7 @@ static int cma_acquire_dev_by_src_ip(struct rdma_id_private *id_priv) 
	struct cma_device *cma_dev;

	enum ib_gid_type gid_type;

	int ret = -ENODEV;

	unsigned int port;

	u32 port;



	if (dev_addr->dev_type != ARPHRD_INFINIBAND &&

	    id_priv->id.ps == RDMA_PS_IPOIB)
@@ -702,8 +702,8 @@ static int cma_iw_acquire_dev(struct rdma_id_private *id_priv, 
	struct cma_device *cma_dev;

	enum ib_gid_type gid_type;

	int ret = -ENODEV;

	unsigned int port;

	union ib_gid gid;

	u32 port;



	if (dev_addr->dev_type != ARPHRD_INFINIBAND &&

	    id_priv->id.ps == RDMA_PS_IPOIB)
@@ -1572,7 +1572,7 @@ static bool cma_match_private_data(struct rdma_id_private *id_priv, 
static bool cma_protocol_roce(const struct rdma_cm_id *id)

{

	struct ib_device *device = id->device;

	const int port_num = id->port_num ?: rdma_start_port(device);

	const u32 port_num = id->port_num ?: rdma_start_port(device);



	return rdma_protocol_roce(device, port_num);

}
@@ -4861,9 +4861,9 @@ static int cma_add_one(struct ib_device *device) 
	struct rdma_id_private *to_destroy;

	struct cma_device *cma_dev;

	struct rdma_id_private *id_priv;

	unsigned int i;

	unsigned long supported_gids = 0;

	int ret;

	u32 i;



	cma_dev = kmalloc(sizeof(*cma_dev), GFP_KERNEL);

	if (!cma_dev)

drivers/infiniband/core/cma_configfs.c

+4 −4
Original line number Diff line number Diff line
@@ -43,7 +43,7 @@ struct cma_device; 
struct cma_dev_group;



struct cma_dev_port_group {

	unsigned int		port_num;

	u32			port_num;

	struct cma_dev_group	*cma_dev_group;

	struct config_group	group;

};
@@ -200,10 +200,10 @@ static const struct config_item_type cma_port_group_type = { 
static int make_cma_ports(struct cma_dev_group *cma_dev_group,

			  struct cma_device *cma_dev)

{

	struct ib_device *ibdev;

	unsigned int i;

	unsigned int ports_num;

	struct cma_dev_port_group *ports;

	struct ib_device *ibdev;

	u32 ports_num;

	u32 i;



	ibdev = cma_get_ib_dev(cma_dev);

drivers/infiniband/core/cma_priv.h

+4 −4
Original line number Diff line number Diff line
@@ -117,11 +117,11 @@ void cma_dev_put(struct cma_device *dev); 
typedef bool (*cma_device_filter)(struct ib_device *, void *);

struct cma_device *cma_enum_devices_by_ibdev(cma_device_filter filter,

					     void *cookie);

int cma_get_default_gid_type(struct cma_device *dev, unsigned int port);

int cma_set_default_gid_type(struct cma_device *dev, unsigned int port,

int cma_get_default_gid_type(struct cma_device *dev, u32 port);

int cma_set_default_gid_type(struct cma_device *dev, u32 port,

			     enum ib_gid_type default_gid_type);

int cma_get_default_roce_tos(struct cma_device *dev, unsigned int port);

int cma_set_default_roce_tos(struct cma_device *dev, unsigned int port,

int cma_get_default_roce_tos(struct cma_device *dev, u32 port);

int cma_set_default_roce_tos(struct cma_device *dev, u32 port,

			     u8 default_roce_tos);

struct ib_device *cma_get_ib_dev(struct cma_device *dev);