diff options
Diffstat (limited to 'drivers/infiniband')
84 files changed, 2740 insertions, 2035 deletions
diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig index 85e103b147cc..b44b1c322ec8 100644 --- a/drivers/infiniband/Kconfig +++ b/drivers/infiniband/Kconfig @@ -55,6 +55,7 @@ config INFINIBAND_ON_DEMAND_PAGING bool "InfiniBand on-demand paging support" depends on INFINIBAND_USER_MEM select MMU_NOTIFIER + select INTERVAL_TREE default y ---help--- On demand paging support for the InfiniBand subsystem. diff --git a/drivers/infiniband/core/addr.c b/drivers/infiniband/core/addr.c index 9b76a8fcdd24..1dd467bed8fc 100644 --- a/drivers/infiniband/core/addr.c +++ b/drivers/infiniband/core/addr.c @@ -183,7 +183,7 @@ static int ib_nl_ip_send_msg(struct rdma_dev_addr *dev_addr, /* Repair the nlmsg header length */ nlmsg_end(skb, nlh); - rdma_nl_multicast(skb, RDMA_NL_GROUP_LS, GFP_KERNEL); + rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, GFP_KERNEL); /* Make the request retry, so when we get the response from userspace * we will have something. diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index 18e476b3ced0..00fb3eacda19 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -810,6 +810,7 @@ static void release_gid_table(struct ib_device *device, if (leak) return; + mutex_destroy(&table->lock); kfree(table->data_vec); kfree(table); } diff --git a/drivers/infiniband/core/cma_configfs.c b/drivers/infiniband/core/cma_configfs.c index 3ec2c415bb70..8b0b5ae22e4c 100644 --- a/drivers/infiniband/core/cma_configfs.c +++ b/drivers/infiniband/core/cma_configfs.c @@ -342,12 +342,18 @@ static struct configfs_subsystem cma_subsys = { int __init cma_configfs_init(void) { + int ret; + config_group_init(&cma_subsys.su_group); mutex_init(&cma_subsys.su_mutex); - return configfs_register_subsystem(&cma_subsys); + ret = configfs_register_subsystem(&cma_subsys); + if (ret) + mutex_destroy(&cma_subsys.su_mutex); + return ret; } void __exit cma_configfs_exit(void) { configfs_unregister_subsystem(&cma_subsys); + mutex_destroy(&cma_subsys.su_mutex); } diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h index 888d89ce81df..3a8b0911c3bc 100644 --- a/drivers/infiniband/core/core_priv.h +++ b/drivers/infiniband/core/core_priv.h @@ -36,6 +36,8 @@ #include <linux/list.h> #include <linux/spinlock.h> #include <linux/cgroup_rdma.h> +#include <net/net_namespace.h> +#include <net/netns/generic.h> #include <rdma/ib_verbs.h> #include <rdma/opa_addr.h> @@ -54,8 +56,26 @@ struct pkey_index_qp_list { struct list_head qp_list; }; +/** + * struct rdma_dev_net - rdma net namespace metadata for a net + * @nl_sock: Pointer to netlink socket + * @net: Pointer to owner net namespace + * @id: xarray id to identify the net namespace. + */ +struct rdma_dev_net { + struct sock *nl_sock; + possible_net_t net; + u32 id; +}; + extern const struct attribute_group ib_dev_attr_group; extern bool ib_devices_shared_netns; +extern unsigned int rdma_dev_net_id; + +static inline struct rdma_dev_net *rdma_net_to_dev_net(struct net *net) +{ + return net_generic(net, rdma_dev_net_id); +} int ib_device_register_sysfs(struct ib_device *device); void ib_device_unregister_sysfs(struct ib_device *device); @@ -179,7 +199,6 @@ void ib_mad_cleanup(void); int ib_sa_init(void); void ib_sa_cleanup(void); -int rdma_nl_init(void); void rdma_nl_exit(void); int ib_nl_handle_resolve_resp(struct sk_buff *skb, @@ -302,7 +321,9 @@ static inline struct ib_qp *_ib_create_qp(struct ib_device *dev, struct ib_udata *udata, struct ib_uobject *uobj) { + enum ib_qp_type qp_type = attr->qp_type; struct ib_qp *qp; + bool is_xrc; if (!dev->ops.create_qp) return ERR_PTR(-EOPNOTSUPP); @@ -320,7 +341,8 @@ static inline struct ib_qp *_ib_create_qp(struct ib_device *dev, * and more importantly they are created internaly by driver, * see mlx5 create_dev_resources() as an example. */ - if (attr->qp_type < IB_QPT_XRC_INI) { + is_xrc = qp_type == IB_QPT_XRC_INI || qp_type == IB_QPT_XRC_TGT; + if ((qp_type < IB_QPT_MAX && !is_xrc) || qp_type == IB_QPT_DRIVER) { qp->res.type = RDMA_RESTRACK_QP; if (uobj) rdma_restrack_uadd(&qp->res); @@ -362,4 +384,7 @@ void ib_port_unregister_module_stat(struct kobject *kobj); int ib_device_set_netns_put(struct sk_buff *skb, struct ib_device *dev, u32 ns_fd); + +int rdma_nl_net_init(struct rdma_dev_net *rnet); +void rdma_nl_net_exit(struct rdma_dev_net *rnet); #endif /* _CORE_PRIV_H */ diff --git a/drivers/infiniband/core/counters.c b/drivers/infiniband/core/counters.c index 01faef7bc061..61fcb3a31340 100644 --- a/drivers/infiniband/core/counters.c +++ b/drivers/infiniband/core/counters.c @@ -38,6 +38,9 @@ int rdma_counter_set_auto_mode(struct ib_device *dev, u8 port, int ret; port_counter = &dev->port_data[port].port_counter; + if (!port_counter->hstats) + return -EOPNOTSUPP; + mutex_lock(&port_counter->lock); if (on) { ret = __counter_set_mode(&port_counter->mode, @@ -393,6 +396,9 @@ u64 rdma_counter_get_hwstat_value(struct ib_device *dev, u8 port, u32 index) u64 sum; port_counter = &dev->port_data[port].port_counter; + if (!port_counter->hstats) + return 0; + sum = get_running_counters_hwstat_sum(dev, port, index); sum += port_counter->hstats->value[index]; @@ -506,6 +512,9 @@ int rdma_counter_bind_qpn_alloc(struct ib_device *dev, u8 port, if (!rdma_is_port_valid(dev, port)) return -EINVAL; + if (!dev->port_data[port].port_counter.hstats) + return -EOPNOTSUPP; + qp = rdma_counter_get_qp(dev, qp_num); if (!qp) return -ENOENT; @@ -592,9 +601,9 @@ int rdma_counter_get_mode(struct ib_device *dev, u8 port, void rdma_counter_init(struct ib_device *dev) { struct rdma_port_counter *port_counter; - u32 port; + u32 port, i; - if (!dev->ops.alloc_hw_stats || !dev->port_data) + if (!dev->port_data) return; rdma_for_each_port(dev, port) { @@ -602,6 +611,9 @@ void rdma_counter_init(struct ib_device *dev) port_counter->mode.mode = RDMA_COUNTER_MODE_NONE; mutex_init(&port_counter->lock); + if (!dev->ops.alloc_hw_stats) + continue; + port_counter->hstats = dev->ops.alloc_hw_stats(dev, port); if (!port_counter->hstats) goto fail; @@ -610,13 +622,12 @@ void rdma_counter_init(struct ib_device *dev) return; fail: - rdma_for_each_port(dev, port) { + for (i = port; i >= rdma_start_port(dev); i--) { port_counter = &dev->port_data[port].port_counter; kfree(port_counter->hstats); port_counter->hstats = NULL; + mutex_destroy(&port_counter->lock); } - - return; } void rdma_counter_release(struct ib_device *dev) @@ -624,11 +635,9 @@ void rdma_counter_release(struct ib_device *dev) struct rdma_port_counter *port_counter; u32 port; - if (!dev->ops.alloc_hw_stats) - return; - rdma_for_each_port(dev, port) { port_counter = &dev->port_data[port].port_counter; kfree(port_counter->hstats); + mutex_destroy(&port_counter->lock); } } diff --git a/drivers/infiniband/core/cq.c b/drivers/infiniband/core/cq.c index 7c599878ccf7..bbfded6d5d3d 100644 --- a/drivers/infiniband/core/cq.c +++ b/drivers/infiniband/core/cq.c @@ -253,6 +253,34 @@ out_free_cq: EXPORT_SYMBOL(__ib_alloc_cq_user); /** + * __ib_alloc_cq_any - allocate a completion queue + * @dev: device to allocate the CQ for + * @private: driver private data, accessible from cq->cq_context + * @nr_cqe: number of CQEs to allocate + * @poll_ctx: context to poll the CQ from + * @caller: module owner name + * + * Attempt to spread ULP Completion Queues over each device's interrupt + * vectors. A simple best-effort mechanism is used. + */ +struct ib_cq *__ib_alloc_cq_any(struct ib_device *dev, void *private, + int nr_cqe, enum ib_poll_context poll_ctx, + const char *caller) +{ + static atomic_t counter; + int comp_vector = 0; + + if (dev->num_comp_vectors > 1) + comp_vector = + atomic_inc_return(&counter) % + min_t(int, dev->num_comp_vectors, num_online_cpus()); + + return __ib_alloc_cq_user(dev, private, nr_cqe, comp_vector, poll_ctx, + caller, NULL); +} +EXPORT_SYMBOL(__ib_alloc_cq_any); + +/** * ib_free_cq_user - free a completion queue * @cq: completion queue to free. * @udata: User data or NULL for kernel object diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index 9773145dee09..99c4a55545cf 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -39,7 +39,6 @@ #include <linux/init.h> #include <linux/netdevice.h> #include <net/net_namespace.h> -#include <net/netns/generic.h> #include <linux/security.h> #include <linux/notifier.h> #include <linux/hashtable.h> @@ -94,28 +93,24 @@ static DEFINE_XARRAY_FLAGS(devices, XA_FLAGS_ALLOC); static DECLARE_RWSEM(devices_rwsem); #define DEVICE_REGISTERED XA_MARK_1 -static LIST_HEAD(client_list); +static u32 highest_client_id; #define CLIENT_REGISTERED XA_MARK_1 static DEFINE_XARRAY_FLAGS(clients, XA_FLAGS_ALLOC); static DECLARE_RWSEM(clients_rwsem); +static void ib_client_put(struct ib_client *client) +{ + if (refcount_dec_and_test(&client->uses)) + complete(&client->uses_zero); +} + /* * If client_data is registered then the corresponding client must also still * be registered. */ #define CLIENT_DATA_REGISTERED XA_MARK_1 -/** - * struct rdma_dev_net - rdma net namespace metadata for a net - * @net: Pointer to owner net namespace - * @id: xarray id to identify the net namespace. - */ -struct rdma_dev_net { - possible_net_t net; - u32 id; -}; - -static unsigned int rdma_dev_net_id; +unsigned int rdma_dev_net_id; /* * A list of net namespaces is maintained in an xarray. This is necessary @@ -508,6 +503,9 @@ static void ib_device_release(struct device *device) rcu_head); } + mutex_destroy(&dev->unregistration_lock); + mutex_destroy(&dev->compat_devs_mutex); + xa_destroy(&dev->compat_devs); xa_destroy(&dev->client_data); kfree_rcu(dev, rcu_head); @@ -661,6 +659,14 @@ static int add_client_context(struct ib_device *device, down_write(&device->client_data_rwsem); /* + * So long as the client is registered hold both the client and device + * unregistration locks. + */ + if (!refcount_inc_not_zero(&client->uses)) + goto out_unlock; + refcount_inc(&device->refcount); + + /* * Another caller to add_client_context got here first and has already * completely initialized context. */ @@ -683,6 +689,9 @@ static int add_client_context(struct ib_device *device, return 0; out: + ib_device_put(device); + ib_client_put(client); +out_unlock: up_write(&device->client_data_rwsem); return ret; } @@ -702,7 +711,7 @@ static void remove_client_context(struct ib_device *device, client_data = xa_load(&device->client_data, client_id); xa_clear_mark(&device->client_data, client_id, CLIENT_DATA_REGISTERED); client = xa_load(&clients, client_id); - downgrade_write(&device->client_data_rwsem); + up_write(&device->client_data_rwsem); /* * Notice we cannot be holding any exclusive locks when calling the @@ -712,17 +721,13 @@ static void remove_client_context(struct ib_device *device, * * For this reason clients and drivers should not call the * unregistration functions will holdling any locks. - * - * It tempting to drop the client_data_rwsem too, but this is required - * to ensure that unregister_client does not return until all clients - * are completely unregistered, which is required to avoid module - * unloading races. */ if (client->remove) client->remove(device, client_data); xa_erase(&device->client_data, client_id); - up_read(&device->client_data_rwsem); + ib_device_put(device); + ib_client_put(client); } static int alloc_port_data(struct ib_device *device) @@ -1047,7 +1052,7 @@ int rdma_compatdev_set(u8 enable) static void rdma_dev_exit_net(struct net *net) { - struct rdma_dev_net *rnet = net_generic(net, rdma_dev_net_id); + struct rdma_dev_net *rnet = rdma_net_to_dev_net(net); struct ib_device *dev; unsigned long index; int ret; @@ -1081,25 +1086,32 @@ static void rdma_dev_exit_net(struct net *net) } up_read(&devices_rwsem); + rdma_nl_net_exit(rnet); xa_erase(&rdma_nets, rnet->id); } static __net_init int rdma_dev_init_net(struct net *net) { - struct rdma_dev_net *rnet = net_generic(net, rdma_dev_net_id); + struct rdma_dev_net *rnet = rdma_net_to_dev_net(net); unsigned long index; struct ib_device *dev; int ret; + write_pnet(&rnet->net, net); + + ret = rdma_nl_net_init(rnet); + if (ret) + return ret; + /* No need to create any compat devices in default init_net. */ if (net_eq(net, &init_net)) return 0; - write_pnet(&rnet->net, net); - ret = xa_alloc(&rdma_nets, &rnet->id, rnet, xa_limit_32b, GFP_KERNEL); - if (ret) + if (ret) { + rdma_nl_net_exit(rnet); return ret; + } down_read(&devices_rwsem); xa_for_each_marked (&devices, index, dev, DEVICE_REGISTERED) { @@ -1224,7 +1236,7 @@ static int setup_device(struct ib_device *device) static void disable_device(struct ib_device *device) { - struct ib_client *client; + u32 cid; WARN_ON(!refcount_read(&device->refcount)); @@ -1232,10 +1244,19 @@ static void disable_device(struct ib_device *device) xa_clear_mark(&devices, device->index, DEVICE_REGISTERED); up_write(&devices_rwsem); + /* + * Remove clients in LIFO order, see assign_client_id. This could be + * more efficient if xarray learns to reverse iterate. Since no new + * clients can be added to this ib_device past this point we only need + * the maximum possible client_id value here. + */ down_read(&clients_rwsem); - list_for_each_entry_reverse(client, &client_list, list) - remove_client_context(device, client->client_id); + cid = highest_client_id; up_read(&clients_rwsem); + while (cid) { + cid--; + remove_client_context(device, cid); + } /* Pairs with refcount_set in enable_device */ ib_device_put(device); @@ -1662,30 +1683,31 @@ static int assign_client_id(struct ib_client *client) /* * The add/remove callbacks must be called in FIFO/LIFO order. To * achieve this we assign client_ids so they are sorted in - * registration order, and retain a linked list we can reverse iterate - * to get the LIFO order. The extra linked list can go away if xarray - * learns to reverse iterate. + * registration order. */ - if (list_empty(&client_list)) { - client->client_id = 0; - } else { - struct ib_client *last; - - last = list_last_entry(&client_list, struct ib_client, list); - client->client_id = last->client_id + 1; - } + client->client_id = highest_client_id; ret = xa_insert(&clients, client->client_id, client, GFP_KERNEL); if (ret) goto out; + highest_client_id++; xa_set_mark(&clients, client->client_id, CLIENT_REGISTERED); - list_add_tail(&client->list, &client_list); out: up_write(&clients_rwsem); return ret; } +static void remove_client_id(struct ib_client *client) +{ + down_write(&clients_rwsem); + xa_erase(&clients, client->client_id); + for (; highest_client_id; highest_client_id--) + if (xa_load(&clients, highest_client_id - 1)) + break; + up_write(&clients_rwsem); +} + /** * ib_register_client - Register an IB client * @client:Client to register @@ -1705,6 +1727,8 @@ int ib_register_client(struct ib_client *client) unsigned long index; int ret; + refcount_set(&client->uses, 1); + init_completion(&client->uses_zero); ret = assign_client_id(client); if (ret) return ret; @@ -1740,21 +1764,30 @@ void ib_unregister_client(struct ib_client *client) unsigned long index; down_write(&clients_rwsem); + ib_client_put(client); xa_clear_mark(&clients, client->client_id, CLIENT_REGISTERED); up_write(&clients_rwsem); - /* - * Every device still known must be serialized to make sure we are - * done with the client callbacks before we return. - */ - down_read(&devices_rwsem); - xa_for_each (&devices, index, device) + + /* We do not want to have locks while calling client->remove() */ + rcu_read_lock(); + xa_for_each (&devices, index, device) { + if (!ib_device_try_get(device)) + continue; + rcu_read_unlock(); + remove_client_context(device, client->client_id); - up_read(&devices_rwsem); - down_write(&clients_rwsem); - list_del(&client->list); - xa_erase(&clients, client->client_id); - up_write(&clients_rwsem); + ib_device_put(device); + rcu_read_lock(); + } + rcu_read_unlock(); + + /* + * remove_client_context() is not a fence, it can return even though a + * removal is ongoing. Wait until all removals are completed. + */ + wait_for_completion(&client->uses_zero); + remove_client_id(client); } EXPORT_SYMBOL(ib_unregister_client); @@ -1940,31 +1973,64 @@ void ib_dispatch_event(struct ib_event *event) } EXPORT_SYMBOL(ib_dispatch_event); -/** - * ib_query_port - Query IB port attributes - * @device:Device to query - * @port_num:Port number to query - * @port_attr:Port attributes - * - * ib_query_port() returns the attributes of a port through the - * @port_attr pointer. - */ -int ib_query_port(struct ib_device *device, - u8 port_num, - struct ib_port_attr *port_attr) +static int iw_query_port(struct ib_device *device, + u8 port_num, + struct ib_port_attr *port_attr) { - union ib_gid gid; + struct in_device *inetdev; + struct net_device *netdev; int err; - if (!rdma_is_port_valid(device, port_num)) - return -EINVAL; + memset(port_attr, 0, sizeof(*port_attr)); + + netdev = ib_device_get_netdev(device, port_num); + if (!netdev) + return -ENODEV; + + dev_put(netdev); + + port_attr->max_mtu = IB_MTU_4096; + port_attr->active_mtu = ib_mtu_int_to_enum(netdev->mtu); + + if (!netif_carrier_ok(netdev)) { + port_attr->state = IB_PORT_DOWN; + port_attr->phys_state = IB_PORT_PHYS_STATE_DISABLED; + } else { + inetdev = in_dev_get(netdev); + + if (inetdev && inetdev->ifa_list) { + port_attr->state = IB_PORT_ACTIVE; + port_attr->phys_state = IB_PORT_PHYS_STATE_LINK_UP; + in_dev_put(inetdev); + } else { + port_attr->state = IB_PORT_INIT; + port_attr->phys_state = + IB_PORT_PHYS_STATE_PORT_CONFIGURATION_TRAINING; + } + } + + err = device->ops.query_port(device, port_num, port_attr); + if (err) + return err; + + return 0; +} + +static int __ib_query_port(struct ib_device *device, + u8 port_num, + struct ib_port_attr *port_attr) +{ + union ib_gid gid = {}; + int err; memset(port_attr, 0, sizeof(*port_attr)); + err = device->ops.query_port(device, port_num, port_attr); if (err || port_attr->subnet_prefix) return err; - if (rdma_port_get_link_layer(device, port_num) != IB_LINK_LAYER_INFINIBAND) + if (rdma_port_get_link_layer(device, port_num) != + IB_LINK_LAYER_INFINIBAND) return 0; err = device->ops.query_gid(device, port_num, 0, &gid); @@ -1974,6 +2040,28 @@ int ib_query_port(struct ib_device *device, port_attr->subnet_prefix = be64_to_cpu(gid.global.subnet_prefix); return 0; } + +/** + * ib_query_port - Query IB port attributes + * @device:Device to query + * @port_num:Port number to query + * @port_attr:Port attributes + * + * ib_query_port() returns the attributes of a port through the + * @port_attr pointer. + */ +int ib_query_port(struct ib_device *device, + u8 port_num, + struct ib_port_attr *port_attr) +{ + if (!rdma_is_port_valid(device, port_num)) + return -EINVAL; + + if (rdma_protocol_iwarp(device, port_num)) + return iw_query_port(device, port_num, port_attr); + else + return __ib_query_port(device, port_num, port_attr); +} EXPORT_SYMBOL(ib_query_port); static void add_ndev_hash(struct ib_port_data *pdata) @@ -2528,6 +2616,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, get_vf_config); SET_DEVICE_OP(dev_ops, get_vf_stats); SET_DEVICE_OP(dev_ops, init_port); + SET_DEVICE_OP(dev_ops, invalidate_range); SET_DEVICE_OP(dev_ops, iw_accept); SET_DEVICE_OP(dev_ops, iw_add_ref); SET_DEVICE_OP(dev_ops, iw_connect); @@ -2626,12 +2715,6 @@ static int __init ib_core_init(void) goto err_comp_unbound; } - ret = rdma_nl_init(); - if (ret) { - pr_warn("Couldn't init IB netlink interface: err %d\n", ret); - goto err_sysfs; - } - ret = addr_init(); if (ret) { pr_warn("Could't init IB address resolution\n"); @@ -2677,8 +2760,6 @@ err_mad: err_addr: addr_cleanup(); err_ibnl: - rdma_nl_exit(); -err_sysfs: class_unregister(&ib_class); err_comp_unbound: destroy_workqueue(ib_comp_unbound_wq); diff --git a/drivers/infiniband/core/fmr_pool.c b/drivers/infiniband/core/fmr_pool.c index 7d841b689a1e..e08aec427027 100644 --- a/drivers/infiniband/core/fmr_pool.c +++ b/drivers/infiniband/core/fmr_pool.c @@ -148,13 +148,6 @@ static void ib_fmr_batch_release(struct ib_fmr_pool *pool) hlist_del_init(&fmr->cache_node); fmr->remap_count = 0; list_add_tail(&fmr->fmr->list, &fmr_list); - -#ifdef DEBUG - if (fmr->ref_count !=0) { - pr_warn(PFX "Unmapping FMR 0x%08x with ref count %d\n", - fmr, fmr->ref_count); - } -#endif } list_splice_init(&pool->dirty_list, &unmap_list); @@ -496,12 +489,6 @@ void ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) } } -#ifdef DEBUG - if (fmr->ref_count < 0) - pr_warn(PFX "FMR %p has ref count %d < 0\n", - fmr, fmr->ref_count); -#endif - spin_unlock_irqrestore(&pool->pool_lock, flags); } EXPORT_SYMBOL(ib_fmr_pool_unmap); diff --git a/drivers/infiniband/core/iwpm_msg.c b/drivers/infiniband/core/iwpm_msg.c index 2452b0ddcf0d..46686990a827 100644 --- a/drivers/infiniband/core/iwpm_msg.c +++ b/drivers/infiniband/core/iwpm_msg.c @@ -112,7 +112,7 @@ int iwpm_register_pid(struct iwpm_dev_data *pm_msg, u8 nl_client) pr_debug("%s: Multicasting a nlmsg (dev = %s ifname = %s iwpm = %s)\n", __func__, pm_msg->dev_name, pm_msg->if_name, iwpm_ulib_name); - ret = rdma_nl_multicast(skb, RDMA_NL_GROUP_IWPM, GFP_KERNEL); + ret = rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_IWPM, GFP_KERNEL); if (ret) { skb = NULL; /* skb is freed in the netlink send-op handling */ iwpm_user_pid = IWPM_PID_UNAVAILABLE; @@ -124,8 +124,7 @@ int iwpm_register_pid(struct iwpm_dev_data *pm_msg, u8 nl_client) return ret; pid_query_error: pr_info("%s: %s (client = %d)\n", __func__, err_str, nl_client); - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); if (nlmsg_request) iwpm_free_nlmsg_request(&nlmsg_request->kref); return ret; @@ -202,7 +201,7 @@ int iwpm_add_mapping(struct iwpm_sa_data *pm_msg, u8 nl_client) nlmsg_end(skb, nlh); nlmsg_request->req_buffer = pm_msg; - ret = rdma_nl_unicast_wait(skb, iwpm_user_pid); + ret = rdma_nl_unicast_wait(&init_net, skb, iwpm_user_pid); if (ret) { skb = NULL; /* skb is freed in the netlink send-op handling */ iwpm_user_pid = IWPM_PID_UNDEFINED; @@ -214,8 +213,7 @@ int iwpm_add_mapping(struct iwpm_sa_data *pm_msg, u8 nl_client) add_mapping_error: pr_info("%s: %s (client = %d)\n", __func__, err_str, nl_client); add_mapping_error_nowarn: - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); if (nlmsg_request) iwpm_free_nlmsg_request(&nlmsg_request->kref); return ret; @@ -297,7 +295,7 @@ int iwpm_add_and_query_mapping(struct iwpm_sa_data *pm_msg, u8 nl_client) nlmsg_end(skb, nlh); nlmsg_request->req_buffer = pm_msg; - ret = rdma_nl_unicast_wait(skb, iwpm_user_pid); + ret = rdma_nl_unicast_wait(&init_net, skb, iwpm_user_pid); if (ret) { skb = NULL; /* skb is freed in the netlink send-op handling */ err_str = "Unable to send a nlmsg"; @@ -308,8 +306,7 @@ int iwpm_add_and_query_mapping(struct iwpm_sa_data *pm_msg, u8 nl_client) query_mapping_error: pr_info("%s: %s (client = %d)\n", __func__, err_str, nl_client); query_mapping_error_nowarn: - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); if (nlmsg_request) iwpm_free_nlmsg_request(&nlmsg_request->kref); return ret; @@ -364,7 +361,7 @@ int iwpm_remove_mapping(struct sockaddr_storage *local_addr, u8 nl_client) nlmsg_end(skb, nlh); - ret = rdma_nl_unicast_wait(skb, iwpm_user_pid); + ret = rdma_nl_unicast_wait(&init_net, skb, iwpm_user_pid); if (ret) { skb = NULL; /* skb is freed in the netlink send-op handling */ iwpm_user_pid = IWPM_PID_UNDEFINED; diff --git a/drivers/infiniband/core/iwpm_util.c b/drivers/infiniband/core/iwpm_util.c index 41929bb83739..13495b43dbc1 100644 --- a/drivers/infiniband/core/iwpm_util.c +++ b/drivers/infiniband/core/iwpm_util.c @@ -645,7 +645,7 @@ static int send_mapinfo_num(u32 mapping_num, u8 nl_client, int iwpm_pid) nlmsg_end(skb, nlh); - ret = rdma_nl_unicast(skb, iwpm_pid); + ret = rdma_nl_unicast(&init_net, skb, iwpm_pid); if (ret) { skb = NULL; err_str = "Unable to send a nlmsg"; @@ -655,8 +655,7 @@ static int send_mapinfo_num(u32 mapping_num, u8 nl_client, int iwpm_pid) return 0; mapinfo_num_error: pr_info("%s: %s\n", __func__, err_str); - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); return ret; } @@ -674,7 +673,7 @@ static int send_nlmsg_done(struct sk_buff *skb, u8 nl_client, int iwpm_pid) return -ENOMEM; } nlh->nlmsg_type = NLMSG_DONE; - ret = rdma_nl_unicast(skb, iwpm_pid); + ret = rdma_nl_unicast(&init_net, skb, iwpm_pid); if (ret) pr_warn("%s Unable to send a nlmsg\n", __func__); return ret; @@ -778,8 +777,7 @@ send_mapping_info_unlock: send_mapping_info_exit: if (ret) { pr_warn("%s: %s (ret = %d)\n", __func__, err_str, ret); - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); return ret; } send_nlmsg_done(skb, nl_client, iwpm_pid); @@ -824,7 +822,7 @@ int iwpm_send_hello(u8 nl_client, int iwpm_pid, u16 abi_version) goto hello_num_error; nlmsg_end(skb, nlh); - ret = rdma_nl_unicast(skb, iwpm_pid); + ret = rdma_nl_unicast(&init_net, skb, iwpm_pid); if (ret) { skb = NULL; err_str = "Unable to send a nlmsg"; @@ -834,7 +832,6 @@ int iwpm_send_hello(u8 nl_client, int iwpm_pid, u16 abi_version) return 0; hello_num_error: pr_info("%s: %s\n", __func__, err_str); - if (skb) - dev_kfree_skb(skb); + dev_kfree_skb(skb); return ret; } diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index cc99479b2c09..9947d16edef2 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -3224,18 +3224,18 @@ static int ib_mad_port_open(struct ib_device *device, if (has_smi) cq_size *= 2; + port_priv->pd = ib_alloc_pd(device, 0); + if (IS_ERR(port_priv->pd)) { + dev_err(&device->dev, "Couldn't create ib_mad PD\n"); + ret = PTR_ERR(port_priv->pd); + goto error3; + } + port_priv->cq = ib_alloc_cq(port_priv->device, port_priv, cq_size, 0, IB_POLL_UNBOUND_WORKQUEUE); if (IS_ERR(port_priv->cq)) { dev_err(&device->dev, "Couldn't create ib_mad CQ\n"); ret = PTR_ERR(port_priv->cq); - goto error3; - } - - port_priv->pd = ib_alloc_pd(device, 0); - if (IS_ERR(port_priv->pd)) { - dev_err(&device->dev, "Couldn't create ib_mad PD\n"); - ret = PTR_ERR(port_priv->pd); goto error4; } @@ -3278,11 +3278,11 @@ error8: error7: destroy_mad_qp(&port_priv->qp_info[0]); error6: - ib_dealloc_pd(port_priv->pd); -error4: ib_free_cq(port_priv->cq); cleanup_recv_queue(&port_priv->qp_info[1]); cleanup_recv_queue(&port_priv->qp_info[0]); +error4: + ib_dealloc_pd(port_priv->pd); error3: kfree(port_priv); @@ -3312,8 +3312,8 @@ static int ib_mad_port_close(struct ib_device *device, int port_num) destroy_workqueue(port_priv->wq); destroy_mad_qp(&port_priv->qp_info[1]); destroy_mad_qp(&port_priv->qp_info[0]); - ib_dealloc_pd(port_priv->pd); ib_free_cq(port_priv->cq); + ib_dealloc_pd(port_priv->pd); cleanup_recv_queue(&port_priv->qp_info[1]); cleanup_recv_queue(&port_priv->qp_info[0]); /* XXX: Handle deallocation of MAD registration tables */ diff --git a/drivers/infiniband/core/netlink.c b/drivers/infiniband/core/netlink.c index eecfc0b377c9..81dbd5f41bed 100644 --- a/drivers/infiniband/core/netlink.c +++ b/drivers/infiniband/core/netlink.c @@ -36,20 +36,22 @@ #include <linux/export.h> #include <net/netlink.h> #include <net/net_namespace.h> +#include <net/netns/generic.h> #include <net/sock.h> #include <rdma/rdma_netlink.h> #include <linux/module.h> #include "core_priv.h" static DEFINE_MUTEX(rdma_nl_mutex); -static struct sock *nls; static struct { const struct rdma_nl_cbs *cb_table; } rdma_nl_types[RDMA_NL_NUM_CLIENTS]; bool rdma_nl_chk_listeners(unsigned int group) { - return netlink_has_listeners(nls, group); + struct rdma_dev_net *rnet = rdma_net_to_dev_net(&init_net); + + return netlink_has_listeners(rnet->nl_sock, group); } EXPORT_SYMBOL(rdma_nl_chk_listeners); @@ -73,13 +75,21 @@ static bool is_nl_msg_valid(unsigned int type, unsigned int op) return (op < max_num_ops[type]) ? true : false; } -static bool is_nl_valid(unsigned int type, unsigned int op) +static bool +is_nl_valid(const struct sk_buff *skb, unsigned int type, unsigned int op) { const struct rdma_nl_cbs *cb_table; if (!is_nl_msg_valid(type, op)) return false; + /* + * Currently only NLDEV client is supporting netlink commands in + * non init_net net namespace. + */ + if (sock_net(skb->sk) != &init_net && type != RDMA_NL_NLDEV) + return false; + if (!rdma_nl_types[type].cb_table) { mutex_unlock(&rdma_nl_mutex); request_module("rdma-netlink-subsys-%d", type); @@ -161,7 +171,7 @@ static int rdma_nl_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh, unsigned int op = RDMA_NL_GET_OP(type); const struct rdma_nl_cbs *cb_table; - if (!is_nl_valid(index, op)) + if (!is_nl_valid(skb, index, op)) return -EINVAL; cb_table = rdma_nl_types[index].cb_table; @@ -185,7 +195,7 @@ static int rdma_nl_rcv_msg(struct sk_buff *skb, struct nlmsghdr *nlh, .dump = cb_table[op].dump, }; if (c.dump) - return netlink_dump_start(nls, skb, nlh, &c); + return netlink_dump_start(skb->sk, skb, nlh, &c); return -EINVAL; } @@ -258,52 +268,65 @@ static void rdma_nl_rcv(struct sk_buff *skb) mutex_unlock(&rdma_nl_mutex); } -int rdma_nl_unicast(struct sk_buff *skb, u32 pid) +int rdma_nl_unicast(struct net *net, struct sk_buff *skb, u32 pid) { + struct rdma_dev_net *rnet = rdma_net_to_dev_net(net); int err; - err = netlink_unicast(nls, skb, pid, MSG_DONTWAIT); + err = netlink_unicast(rnet->nl_sock, skb, pid, MSG_DONTWAIT); return (err < 0) ? err : 0; } EXPORT_SYMBOL(rdma_nl_unicast); -int rdma_nl_unicast_wait(struct sk_buff *skb, __u32 pid) +int rdma_nl_unicast_wait(struct net *net, struct sk_buff *skb, __u32 pid) { + struct rdma_dev_net *rnet = rdma_net_to_dev_net(net); int err; - err = netlink_unicast(nls, skb, pid, 0); + err = netlink_unicast(rnet->nl_sock, skb, pid, 0); return (err < 0) ? err : 0; } EXPORT_SYMBOL(rdma_nl_unicast_wait); -int rdma_nl_multicast(struct sk_buff *skb, unsigned int group, gfp_t flags) +int rdma_nl_multicast(struct net *net, struct sk_buff *skb, + unsigned int group, gfp_t flags) { - return nlmsg_multicast(nls, skb, 0, group, flags); + struct rdma_dev_net *rnet = rdma_net_to_dev_net(net); + + return nlmsg_multicast(rnet->nl_sock, skb, 0, group, flags); } EXPORT_SYMBOL(rdma_nl_multicast); -int __init rdma_nl_init(void) +void rdma_nl_exit(void) +{ + int idx; + + for (idx = 0; idx < RDMA_NL_NUM_CLIENTS; idx++) + WARN(rdma_nl_types[idx].cb_table, + "Netlink client %d wasn't released prior to unloading %s\n", + idx, KBUILD_MODNAME); +} + +int rdma_nl_net_init(struct rdma_dev_net *rnet) { + struct net *net = read_pnet(&rnet->net); struct netlink_kernel_cfg cfg = { .input = rdma_nl_rcv, }; + struct sock *nls; - nls = netlink_kernel_create(&init_net, NETLINK_RDMA, &cfg); + nls = netlink_kernel_create(net, NETLINK_RDMA, &cfg); if (!nls) return -ENOMEM; nls->sk_sndtimeo = 10 * HZ; + rnet->nl_sock = nls; return 0; } -void rdma_nl_exit(void) +void rdma_nl_net_exit(struct rdma_dev_net *rnet) { - int idx; - - for (idx = 0; idx < RDMA_NL_NUM_CLIENTS; idx++) - rdma_nl_unregister(idx); - - netlink_kernel_release(nls); + netlink_kernel_release(rnet->nl_sock); } MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_RDMA); diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c index 783e465e7c41..cc08218f1ef7 100644 --- a/drivers/infiniband/core/nldev.c +++ b/drivers/infiniband/core/nldev.c @@ -832,7 +832,7 @@ static int nldev_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); @@ -972,7 +972,7 @@ static int nldev_port_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); @@ -1074,7 +1074,7 @@ static int nldev_res_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); @@ -1251,7 +1251,7 @@ static int res_get_common_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); @@ -1596,7 +1596,7 @@ static int nldev_get_chardev(struct sk_buff *skb, struct nlmsghdr *nlh, put_device(data.cdev); if (ibdev) ib_device_put(ibdev); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); out_data: put_device(data.cdev); @@ -1636,7 +1636,7 @@ static int nldev_sys_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, return err; } nlmsg_end(msg, nlh); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); } static int nldev_set_sys_set_doit(struct sk_buff *skb, struct nlmsghdr *nlh, @@ -1734,7 +1734,7 @@ static int nldev_stat_set_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_fill: rdma_counter_unbind_qpn(device, port, qpn, cntn); @@ -1802,7 +1802,7 @@ static int nldev_stat_del_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_fill: rdma_counter_bind_qpn(device, port, qpn, cntn); @@ -1893,7 +1893,7 @@ static int stat_get_doit_default_counter(struct sk_buff *skb, mutex_unlock(&stats->lock); nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_table: nla_nest_cancel(msg, table_attr); @@ -1952,16 +1952,20 @@ static int stat_get_doit_qp(struct sk_buff *skb, struct nlmsghdr *nlh, if (fill_nldev_handle(msg, device) || nla_put_u32(msg, RDMA_NLDEV_ATTR_PORT_INDEX, port) || - nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_MODE, mode)) + nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_MODE, mode)) { + ret = -EMSGSIZE; goto err_msg; + } if ((mode == RDMA_COUNTER_MODE_AUTO) && - nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_AUTO_MODE_MASK, mask)) + nla_put_u32(msg, RDMA_NLDEV_ATTR_STAT_AUTO_MODE_MASK, mask)) { + ret = -EMSGSIZE; goto err_msg; + } nlmsg_end(msg, nlh); ib_device_put(device); - return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); + return rdma_nl_unicast(sock_net(skb->sk), msg, NETLINK_CB(skb).portid); err_msg: nlmsg_free(msg); diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index 7d8071c7e564..17fc2936c077 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -860,7 +860,7 @@ static int ib_nl_send_msg(struct ib_sa_query *query, gfp_t gfp_mask) /* Repair the nlmsg header length */ nlmsg_end(skb, nlh); - return rdma_nl_multicast(skb, RDMA_NL_GROUP_LS, gfp_mask); + return rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, gfp_mask); } static int ib_nl_make_request(struct ib_sa_query *query, gfp_t gfp_mask) diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c index b477295a96c2..7a50cedcef1f 100644 --- a/drivers/infiniband/core/sysfs.c +++ b/drivers/infiniband/core/sysfs.c @@ -289,6 +289,24 @@ static ssize_t rate_show(struct ib_port *p, struct port_attribute *unused, ib_width_enum_to_int(attr.active_width), speed); } +static const char *phys_state_to_str(enum ib_port_phys_state phys_state) +{ + static const char * phys_state_str[] = { + "<unknown>", + "Sleep", + "Polling", + "Disabled", + "PortConfigurationTraining", + "LinkUp", + "LinkErrorRecovery", + "Phy Test", + }; + + if (phys_state < ARRAY_SIZE(phys_state_str)) + return phys_state_str[phys_state]; + return "<unknown>"; +} + static ssize_t phys_state_show(struct ib_port *p, struct port_attribute *unused, char *buf) { @@ -300,16 +318,8 @@ static ssize_t phys_state_show(struct ib_port *p, struct port_attribute *unused, if (ret) return ret; - switch (attr.phys_state) { - case 1: return sprintf(buf, "1: Sleep\n"); - case 2: return sprintf(buf, "2: Polling\n"); - case 3: return sprintf(buf, "3: Disabled\n"); - case 4: return sprintf(buf, "4: PortConfigurationTraining\n"); - case 5: return sprintf(buf, "5: LinkUp\n"); - case 6: return sprintf(buf, "6: LinkErrorRecovery\n"); - case 7: return sprintf(buf, "7: Phy Test\n"); - default: return sprintf(buf, "%d: <unknown>\n", attr.phys_state); - } + return sprintf(buf, "%d: %s\n", attr.phys_state, + phys_state_to_str(attr.phys_state)); } static ssize_t link_layer_show(struct ib_port *p, struct port_attribute *unused, diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c index 08da840ed7ee..37eb8643ec29 100644 --- a/drivers/infiniband/core/umem.c +++ b/drivers/infiniband/core/umem.c @@ -184,9 +184,6 @@ EXPORT_SYMBOL(ib_umem_find_best_pgsz); /** * ib_umem_get - Pin and DMA map userspace memory. * - * If access flags indicate ODP memory, avoid pinning. Instead, stores - * the mm for future page fault handling in conjunction with MMU notifiers. - * * @udata: userspace context to pin memory for * @addr: userspace virtual address to start at * @size: length of region to pin @@ -231,17 +228,12 @@ struct ib_umem *ib_umem_get(struct ib_udata *udata, unsigned long addr, if (!can_do_mlock()) return ERR_PTR(-EPERM); - if (access & IB_ACCESS_ON_DEMAND) { - umem = kzalloc(sizeof(struct ib_umem_odp), GFP_KERNEL); - if (!umem) - return ERR_PTR(-ENOMEM); - umem->is_odp = 1; - } else { - umem = kzalloc(sizeof(*umem), GFP_KERNEL); - if (!umem) - return ERR_PTR(-ENOMEM); - } + if (access & IB_ACCESS_ON_DEMAND) + return ERR_PTR(-EOPNOTSUPP); + umem = kzalloc(sizeof(*umem), GFP_KERNEL); + if (!umem) + return ERR_PTR(-ENOMEM); umem->context = context; umem->length = size; umem->address = addr; @@ -249,18 +241,6 @@ struct ib_umem *ib_umem_get(struct ib_udata *udata, unsigned long addr, umem->owning_mm = mm = current->mm; mmgrab(mm); - if (access & IB_ACCESS_ON_DEMAND) { - if (WARN_ON_ONCE(!context->invalidate_range)) { - ret = -EINVAL; - goto umem_kfree; - } - - ret = ib_umem_odp_get(to_ib_umem_odp(umem), access); - if (ret) - goto umem_kfree; - return umem; - } - page_list = (struct page **) __get_free_page(GFP_KERNEL); if (!page_list) { ret = -ENOMEM; @@ -346,15 +326,6 @@ umem_kfree: } EXPORT_SYMBOL(ib_umem_get); -static void __ib_umem_release_tail(struct ib_umem *umem) -{ - mmdrop(umem->owning_mm); - if (umem->is_odp) - kfree(to_ib_umem_odp(umem)); - else - kfree(umem); -} - /** * ib_umem_release - release memory pinned with ib_umem_get * @umem: umem struct to release @@ -363,30 +334,22 @@ void ib_umem_release(struct ib_umem *umem) { if (!umem) return; - - if (umem->is_odp) { - ib_umem_odp_release(to_ib_umem_odp(umem)); - __ib_umem_release_tail(umem); - return; - } + if (umem->is_odp) + return ib_umem_odp_release(to_ib_umem_odp(umem)); __ib_umem_release(umem->context->device, umem, 1); atomic64_sub(ib_umem_num_pages(umem), &umem->owning_mm->pinned_vm); - __ib_umem_release_tail(umem); + mmdrop(umem->owning_mm); + kfree(umem); } EXPORT_SYMBOL(ib_umem_release); int ib_umem_page_count(struct ib_umem *umem) { - int i; - int n; + int i, n = 0; struct scatterlist *sg; - if (umem->is_odp) - return ib_umem_num_pages(umem); - - n = 0; for_each_sg(umem->sg_head.sgl, sg, umem->nmap, i) n += sg_dma_len(sg) >> PAGE_SHIFT; diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c index 2a75c6f8d827..32fb0b579dec 100644 --- a/drivers/infiniband/core/umem_odp.c +++ b/drivers/infiniband/core/umem_odp.c @@ -39,44 +39,14 @@ #include <linux/export.h> #include <linux/vmalloc.h> #include <linux/hugetlb.h> -#include <linux/interval_tree_generic.h> +#include <linux/interval_tree.h> #include <linux/pagemap.h> #include <rdma/ib_verbs.h> #include <rdma/ib_umem.h> #include <rdma/ib_umem_odp.h> -/* - * The ib_umem list keeps track of memory regions for which the HW - * device request to receive notification when the related memory - * mapping is changed. - * - * ib_umem_lock protects the list. - */ - -static u64 node_start(struct umem_odp_node *n) -{ - struct ib_umem_odp *umem_odp = - container_of(n, struct ib_umem_odp, interval_tree); - - return ib_umem_start(umem_odp); -} - -/* Note that the representation of the intervals in the interval tree - * considers the ending point as contained in the interval, while the - * function ib_umem_end returns the first address which is not contained - * in the umem. - */ -static u64 node_last(struct umem_odp_node *n) -{ - struct ib_umem_odp *umem_odp = - container_of(n, struct ib_umem_odp, interval_tree); - - return ib_umem_end(umem_odp) - 1; -} - -INTERVAL_TREE_DEFINE(struct umem_odp_node, rb, u64, __subtree_last, - node_start, node_last, static, rbt_ib_umem) +#include "uverbs.h" static void ib_umem_notifier_start_account(struct ib_umem_odp *umem_odp) { @@ -104,35 +74,34 @@ static void ib_umem_notifier_end_account(struct ib_umem_odp *umem_odp) mutex_unlock(&umem_odp->umem_mutex); } -static int ib_umem_notifier_release_trampoline(struct ib_umem_odp *umem_odp, - u64 start, u64 end, void *cookie) -{ - /* - * Increase the number of notifiers running, to - * prevent any further fault handling on this MR. - */ - ib_umem_notifier_start_account(umem_odp); - umem_odp->dying = 1; - /* Make sure that the fact the umem is dying is out before we release - * all pending page faults. */ - smp_wmb(); - complete_all(&umem_odp->notifier_completion); - umem_odp->umem.context->invalidate_range( - umem_odp, ib_umem_start(umem_odp), ib_umem_end(umem_odp)); - return 0; -} - static void ib_umem_notifier_release(struct mmu_notifier *mn, struct mm_struct *mm) { struct ib_ucontext_per_mm *per_mm = container_of(mn, struct ib_ucontext_per_mm, mn); + struct rb_node *node; down_read(&per_mm->umem_rwsem); - if (per_mm->active) - rbt_ib_umem_for_each_in_range( - &per_mm->umem_tree, 0, ULLONG_MAX, - ib_umem_notifier_release_trampoline, true, NULL); + if (!per_mm->active) + goto out; + + for (node = rb_first_cached(&per_mm->umem_tree); node; + node = rb_next(node)) { + struct ib_umem_odp *umem_odp = + rb_entry(node, struct ib_umem_odp, interval_tree.rb); + + /* + * Increase the number of notifiers running, to prevent any + * further fault handling on this MR. + */ + ib_umem_notifier_start_account(umem_odp); + complete_all(&umem_odp->notifier_completion); + umem_odp->umem.context->device->ops.invalidate_range( + umem_odp, ib_umem_start(umem_odp), + ib_umem_end(umem_odp)); + } + +out: up_read(&per_mm->umem_rwsem); } @@ -140,7 +109,7 @@ static int invalidate_range_start_trampoline(struct ib_umem_odp *item, u64 start, u64 end, void *cookie) { ib_umem_notifier_start_account(item); - item->umem.context->invalidate_range(item, start, end); + item->umem.context->device->ops.invalidate_range(item, start, end); return 0; } @@ -204,27 +173,13 @@ static const struct mmu_notifier_ops ib_umem_notifiers = { .invalidate_range_end = ib_umem_notifier_invalidate_range_end, }; -static void add_umem_to_per_mm(struct ib_umem_odp *umem_odp) -{ - struct ib_ucontext_per_mm *per_mm = umem_odp->per_mm; - - down_write(&per_mm->umem_rwsem); - if (likely(ib_umem_start(umem_odp) != ib_umem_end(umem_odp))) - rbt_ib_umem_insert(&umem_odp->interval_tree, - &per_mm->umem_tree); - up_write(&per_mm->umem_rwsem); -} - static void remove_umem_from_per_mm(struct ib_umem_odp *umem_odp) { struct ib_ucontext_per_mm *per_mm = umem_odp->per_mm; down_write(&per_mm->umem_rwsem); - if (likely(ib_umem_start(umem_odp) != ib_umem_end(umem_odp))) - rbt_ib_umem_remove(&umem_odp->interval_tree, - &per_mm->umem_tree); + interval_tree_remove(&umem_odp->interval_tree, &per_mm->umem_tree); complete_all(&umem_odp->notifier_completion); - up_write(&per_mm->umem_rwsem); } @@ -267,33 +222,23 @@ out_pid: return ERR_PTR(ret); } -static int get_per_mm(struct ib_umem_odp *umem_odp) +static struct ib_ucontext_per_mm *get_per_mm(struct ib_umem_odp *umem_odp) { struct ib_ucontext *ctx = umem_odp->umem.context; struct ib_ucontext_per_mm *per_mm; + lockdep_assert_held(&ctx->per_mm_list_lock); + /* * Generally speaking we expect only one or two per_mm in this list, * so no reason to optimize this search today. */ - mutex_lock(&ctx->per_mm_list_lock); list_for_each_entry(per_mm, &ctx->per_mm_list, ucontext_list) { if (per_mm->mm == umem_odp->umem.owning_mm) - goto found; + return per_mm; } - per_mm = alloc_per_mm(ctx, umem_odp->umem.owning_mm); - if (IS_ERR(per_mm)) { - mutex_unlock(&ctx->per_mm_list_lock); - return PTR_ERR(per_mm); - } - -found: - umem_odp->per_mm = per_mm; - per_mm->odp_mrs_count++; - mutex_unlock(&ctx->per_mm_list_lock); - - return 0; + return alloc_per_mm(ctx, umem_odp->umem.owning_mm); } static void free_per_mm(struct rcu_head *rcu) @@ -334,76 +279,218 @@ static void put_per_mm(struct ib_umem_odp *umem_odp) mmu_notifier_call_srcu(&per_mm->rcu, free_per_mm); } -struct ib_umem_odp *ib_alloc_odp_umem(struct ib_umem_odp *root, - unsigned long addr, size_t size) +static inline int ib_init_umem_odp(struct ib_umem_odp *umem_odp, + struct ib_ucontext_per_mm *per_mm) +{ + struct ib_ucontext *ctx = umem_odp->umem.context; + int ret; + + umem_odp->umem.is_odp = 1; + if (!umem_odp->is_implicit_odp) { + size_t page_size = 1UL << umem_odp->page_shift; + size_t pages; + + umem_odp->interval_tree.start = + ALIGN_DOWN(umem_odp->umem.address, page_size); + if (check_add_overflow(umem_odp->umem.address, + umem_odp->umem.length, + &umem_odp->interval_tree.last)) + return -EOVERFLOW; + umem_odp->interval_tree.last = + ALIGN(umem_odp->interval_tree.last, page_size); + if (unlikely(umem_odp->interval_tree.last < page_size)) + return -EOVERFLOW; + + pages = (umem_odp->interval_tree.last - + umem_odp->interval_tree.start) >> + umem_odp->page_shift; + if (!pages) + return -EINVAL; + + /* + * Note that the representation of the intervals in the + * interval tree considers the ending point as contained in + * the interval. + */ + umem_odp->interval_tree.last--; + + umem_odp->page_list = kvcalloc( + pages, sizeof(*umem_odp->page_list), GFP_KERNEL); + if (!umem_odp->page_list) + return -ENOMEM; + + umem_odp->dma_list = kvcalloc( + pages, sizeof(*umem_odp->dma_list), GFP_KERNEL); + if (!umem_odp->dma_list) { + ret = -ENOMEM; + goto out_page_list; + } + } + + mutex_lock(&ctx->per_mm_list_lock); + if (!per_mm) { + per_mm = get_per_mm(umem_odp); + if (IS_ERR(per_mm)) { + ret = PTR_ERR(per_mm); + goto out_unlock; + } + } + umem_odp->per_mm = per_mm; + per_mm->odp_mrs_count++; + mutex_unlock(&ctx->per_mm_list_lock); + + mutex_init(&umem_odp->umem_mutex); + init_completion(&umem_odp->notifier_completion); + + if (!umem_odp->is_implicit_odp) { + down_write(&per_mm->umem_rwsem); + interval_tree_insert(&umem_odp->interval_tree, + &per_mm->umem_tree); + up_write(&per_mm->umem_rwsem); + } + mmgrab(umem_odp->umem.owning_mm); + + return 0; + +out_unlock: + mutex_unlock(&ctx->per_mm_list_lock); + kvfree(umem_odp->dma_list); +out_page_list: + kvfree(umem_odp->page_list); + return ret; +} + +/** + * ib_umem_odp_alloc_implicit - Allocate a parent implicit ODP umem + * + * Implicit ODP umems do not have a VA range and do not have any page lists. + * They exist only to hold the per_mm reference to help the driver create + * children umems. + * + * @udata: udata from the syscall being used to create the umem + * @access: ib_reg_mr access flags + */ +struct ib_umem_odp *ib_umem_odp_alloc_implicit(struct ib_udata *udata, + int access) +{ + struct ib_ucontext *context = + container_of(udata, struct uverbs_attr_bundle, driver_udata) + ->context; + struct ib_umem *umem; + struct ib_umem_odp *umem_odp; + int ret; + + if (access & IB_ACCESS_HUGETLB) + return ERR_PTR(-EINVAL); + + if (!context) + return ERR_PTR(-EIO); + if (WARN_ON_ONCE(!context->device->ops.invalidate_range)) + return ERR_PTR(-EINVAL); + + umem_odp = kzalloc(sizeof(*umem_odp), GFP_KERNEL); + if (!umem_odp) + return ERR_PTR(-ENOMEM); + umem = &umem_odp->umem; + umem->context = context; + umem->writable = ib_access_writable(access); + umem->owning_mm = current->mm; + umem_odp->is_implicit_odp = 1; + umem_odp->page_shift = PAGE_SHIFT; + + ret = ib_init_umem_odp(umem_odp, NULL); + if (ret) { + kfree(umem_odp); + return ERR_PTR(ret); + } + return umem_odp; +} +EXPORT_SYMBOL(ib_umem_odp_alloc_implicit); + +/** + * ib_umem_odp_alloc_child - Allocate a child ODP umem under an implicit + * parent ODP umem + * + * @root: The parent umem enclosing the child. This must be allocated using + * ib_alloc_implicit_odp_umem() + * @addr: The starting userspace VA + * @size: The length of the userspace VA + */ +struct ib_umem_odp *ib_umem_odp_alloc_child(struct ib_umem_odp *root, + unsigned long addr, size_t size) { - struct ib_ucontext_per_mm *per_mm = root->per_mm; - struct ib_ucontext *ctx = per_mm->context; + /* + * Caller must ensure that root cannot be freed during the call to + * ib_alloc_odp_umem. + */ struct ib_umem_odp *odp_data; struct ib_umem *umem; - int pages = size >> PAGE_SHIFT; int ret; + if (WARN_ON(!root->is_implicit_odp)) + return ERR_PTR(-EINVAL); + odp_data = kzalloc(sizeof(*odp_data), GFP_KERNEL); if (!odp_data) return ERR_PTR(-ENOMEM); umem = &odp_data->umem; - umem->context = ctx; + umem->context = root->umem.context; umem->length = size; umem->address = addr; - odp_data->page_shift = PAGE_SHIFT; umem->writable = root->umem.writable; - umem->is_odp = 1; - odp_data->per_mm = per_mm; - umem->owning_mm = per_mm->mm; - mmgrab(umem->owning_mm); - - mutex_init(&odp_data->umem_mutex); - init_completion(&odp_data->notifier_completion); - - odp_data->page_list = - vzalloc(array_size(pages, sizeof(*odp_data->page_list))); - if (!odp_data->page_list) { - ret = -ENOMEM; - goto out_odp_data; - } + umem->owning_mm = root->umem.owning_mm; + odp_data->page_shift = PAGE_SHIFT; - odp_data->dma_list = - vzalloc(array_size(pages, sizeof(*odp_data->dma_list))); - if (!odp_data->dma_list) { - ret = -ENOMEM; - goto out_page_list; + ret = ib_init_umem_odp(odp_data, root->per_mm); + if (ret) { + kfree(odp_data); + return ERR_PTR(ret); } - - /* - * Caller must ensure that the umem_odp that the per_mm came from - * cannot be freed during the call to ib_alloc_odp_umem. - */ - mutex_lock(&ctx->per_mm_list_lock); - per_mm->odp_mrs_count++; - mutex_unlock(&ctx->per_mm_list_lock); - add_umem_to_per_mm(odp_data); - return odp_data; - -out_page_list: - vfree(odp_data->page_list); -out_odp_data: - mmdrop(umem->owning_mm); - kfree(odp_data); - return ERR_PTR(ret); } -EXPORT_SYMBOL(ib_alloc_odp_umem); +EXPORT_SYMBOL(ib_umem_odp_alloc_child); -int ib_umem_odp_get(struct ib_umem_odp *umem_odp, int access) +/** + * ib_umem_odp_get - Create a umem_odp for a userspace va + * + * @udata: userspace context to pin memory for + * @addr: userspace virtual address to start at + * @size: length of region to pin + * @access: IB_ACCESS_xxx flags for memory being pinned + * + * The driver should use when the access flags indicate ODP memory. It avoids + * pinning, instead, stores the mm for future page fault handling in + * conjunction with MMU notifiers. + */ +struct ib_umem_odp *ib_umem_odp_get(struct ib_udata *udata, unsigned long addr, + size_t size, int access) { - struct ib_umem *umem = &umem_odp->umem; - /* - * NOTE: This must called in a process context where umem->owning_mm - * == current->mm - */ - struct mm_struct *mm = umem->owning_mm; - int ret_val; + struct ib_umem_odp *umem_odp; + struct ib_ucontext *context; + struct mm_struct *mm; + int ret; + + if (!udata) + return ERR_PTR(-EIO); + + context = container_of(udata, struct uverbs_attr_bundle, driver_udata) + ->context; + if (!context) + return ERR_PTR(-EIO); + + if (WARN_ON_ONCE(!(access & IB_ACCESS_ON_DEMAND)) || + WARN_ON_ONCE(!context->device->ops.invalidate_range)) + return ERR_PTR(-EINVAL); + + umem_odp = kzalloc(sizeof(struct ib_umem_odp), GFP_KERNEL); + if (!umem_odp) + return ERR_PTR(-ENOMEM); + + umem_odp->umem.context = context; + umem_odp->umem.length = size; + umem_odp->umem.address = addr; + umem_odp->umem.writable = ib_access_writable(access); + umem_odp->umem.owning_mm = mm = current->mm; umem_odp->page_shift = PAGE_SHIFT; if (access & IB_ACCESS_HUGETLB) { @@ -414,46 +501,24 @@ int ib_umem_odp_get(struct ib_umem_odp *umem_odp, int access) vma = find_vma(mm, ib_umem_start(umem_odp)); if (!vma || !is_vm_hugetlb_page(vma)) { up_read(&mm->mmap_sem); - return -EINVAL; + ret = -EINVAL; + goto err_free; } h = hstate_vma(vma); umem_odp->page_shift = huge_page_shift(h); up_read(&mm->mmap_sem); } - mutex_init(&umem_odp->umem_mutex); - - init_completion(&umem_odp->notifier_completion); + ret = ib_init_umem_odp(umem_odp, NULL); + if (ret) + goto err_free; + return umem_odp; - if (ib_umem_odp_num_pages(umem_odp)) { - umem_odp->page_list = - vzalloc(array_size(sizeof(*umem_odp->page_list), - ib_umem_odp_num_pages(umem_odp))); - if (!umem_odp->page_list) - return -ENOMEM; - - umem_odp->dma_list = - vzalloc(array_size(sizeof(*umem_odp->dma_list), - ib_umem_odp_num_pages(umem_odp))); - if (!umem_odp->dma_list) { - ret_val = -ENOMEM; - goto out_page_list; - } - } - - ret_val = get_per_mm(umem_odp); - if (ret_val) - goto out_dma_list; - add_umem_to_per_mm(umem_odp); - - return 0; - -out_dma_list: - vfree(umem_odp->dma_list); -out_page_list: - vfree(umem_odp->page_list); - return ret_val; +err_free: + kfree(umem_odp); + return ERR_PTR(ret); } +EXPORT_SYMBOL(ib_umem_odp_get); void ib_umem_odp_release(struct ib_umem_odp *umem_odp) { @@ -463,14 +528,18 @@ void ib_umem_odp_release(struct ib_umem_odp *umem_odp) * It is the driver's responsibility to ensure, before calling us, * that the hardware will not attempt to access the MR any more. */ - ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp), - ib_umem_end(umem_odp)); - - remove_umem_from_per_mm(umem_odp); + if (!umem_odp->is_implicit_odp) { + ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp), + ib_umem_end(umem_odp)); + remove_umem_from_per_mm(umem_odp); + kvfree(umem_odp->dma_list); + kvfree(umem_odp->page_list); + } put_per_mm(umem_odp); - vfree(umem_odp->dma_list); - vfree(umem_odp->page_list); + mmdrop(umem_odp->umem.owning_mm); + kfree(umem_odp); } +EXPORT_SYMBOL(ib_umem_odp_release); /* * Map for DMA and insert a single page into the on-demand paging page tables. @@ -538,7 +607,7 @@ out: if (remove_existing_mapping) { ib_umem_notifier_start_account(umem_odp); - context->invalidate_range( + dev->ops.invalidate_range( umem_odp, ib_umem_start(umem_odp) + (page_index << umem_odp->page_shift), @@ -765,35 +834,21 @@ int rbt_ib_umem_for_each_in_range(struct rb_root_cached *root, void *cookie) { int ret_val = 0; - struct umem_odp_node *node, *next; + struct interval_tree_node *node, *next; struct ib_umem_odp *umem; if (unlikely(start == last)) return ret_val; - for (node = rbt_ib_umem_iter_first(root, start, last - 1); + for (node = interval_tree_iter_first(root, start, last - 1); node; node = next) { /* TODO move the blockable decision up to the callback */ if (!blockable) return -EAGAIN; - next = rbt_ib_umem_iter_next(node, start, last - 1); + next = interval_tree_iter_next(node, start, last - 1); umem = container_of(node, struct ib_umem_odp, interval_tree); ret_val = cb(umem, start, last, cookie) || ret_val; } return ret_val; } -EXPORT_SYMBOL(rbt_ib_umem_for_each_in_range); - -struct ib_umem_odp *rbt_ib_umem_lookup(struct rb_root_cached *root, - u64 addr, u64 length) -{ - struct umem_odp_node *node; - - node = rbt_ib_umem_iter_first(root, addr, addr + length - 1); - if (node) - return container_of(node, struct ib_umem_odp, interval_tree); - return NULL; - -} -EXPORT_SYMBOL(rbt_ib_umem_lookup); diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c index 9f8a48016b41..d1407fa378e8 100644 --- a/drivers/infiniband/core/user_mad.c +++ b/drivers/infiniband/core/user_mad.c @@ -49,6 +49,7 @@ #include <linux/sched.h> #include <linux/semaphore.h> #include <linux/slab.h> +#include <linux/nospec.h> #include <linux/uaccess.h> @@ -884,11 +885,14 @@ static int ib_umad_unreg_agent(struct ib_umad_file *file, u32 __user *arg) if (get_user(id, arg)) return -EFAULT; + if (id >= IB_UMAD_MAX_AGENTS) + return -EINVAL; mutex_lock(&file->port->file_mutex); mutex_lock(&file->mutex); - if (id >= IB_UMAD_MAX_AGENTS || !__get_agent(file, id)) { + id = array_index_nospec(id, IB_UMAD_MAX_AGENTS); + if (!__get_agent(file, id)) { ret = -EINVAL; goto out; } @@ -1038,7 +1042,7 @@ static int ib_umad_close(struct inode *inode, struct file *filp) ib_unregister_mad_agent(file->agent[i]); mutex_unlock(&file->port->file_mutex); - + mutex_destroy(&file->mutex); kfree(file); return 0; } diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 7ddd0e5bc6b3..8f4fd4fac159 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -275,8 +275,6 @@ static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs) ret = ib_dev->ops.alloc_ucontext(ucontext, &attrs->driver_udata); if (ret) goto err_file; - if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING)) - ucontext->invalidate_range = NULL; rdma_restrack_uadd(&ucontext->res); diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 11c13c1381cf..02b57240176c 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -120,6 +120,8 @@ static void ib_uverbs_release_dev(struct device *device) uverbs_destroy_api(dev->uapi); cleanup_srcu_struct(&dev->disassociate_srcu); + mutex_destroy(&dev->lists_mutex); + mutex_destroy(&dev->xrcd_tree_mutex); kfree(dev); } @@ -212,6 +214,8 @@ void ib_uverbs_release_file(struct kref *ref) if (file->disassociate_page) __free_pages(file->disassociate_page, 0); + mutex_destroy(&file->umap_lock); + mutex_destroy(&file->ucontext_lock); kfree(file); } diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 92349bf37589..f974b6854224 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -2259,6 +2259,7 @@ int ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata) if (ret) return ret; } + mutex_destroy(&xrcd->tgt_qp_mutex); return xrcd->device->ops.dealloc_xrcd(xrcd, udata); } diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index a91653aabf38..f9e97d0cc459 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -220,10 +220,10 @@ int bnxt_re_query_port(struct ib_device *ibdev, u8 port_num, if (netif_running(rdev->netdev) && netif_carrier_ok(rdev->netdev)) { port_attr->state = IB_PORT_ACTIVE; - port_attr->phys_state = 5; + port_attr->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } else { port_attr->state = IB_PORT_DOWN; - port_attr->phys_state = 3; + port_attr->phys_state = IB_PORT_PHYS_STATE_DISABLED; } port_attr->max_mtu = IB_MTU_4096; port_attr->active_mtu = iboe_get_mtu(rdev->netdev->mtu); @@ -308,6 +308,7 @@ int bnxt_re_del_gid(const struct ib_gid_attr *attr, void **context) struct bnxt_re_dev *rdev = to_bnxt_re_dev(attr->device, ibdev); struct bnxt_qplib_sgid_tbl *sgid_tbl = &rdev->qplib_res.sgid_tbl; struct bnxt_qplib_gid *gid_to_del; + u16 vlan_id = 0xFFFF; /* Delete the entry from the hardware */ ctx = *context; @@ -317,7 +318,8 @@ int bnxt_re_del_gid(const struct ib_gid_attr *attr, void **context) if (sgid_tbl && sgid_tbl->active) { if (ctx->idx >= sgid_tbl->max) return -EINVAL; - gid_to_del = &sgid_tbl->tbl[ctx->idx]; + gid_to_del = &sgid_tbl->tbl[ctx->idx].gid; + vlan_id = sgid_tbl->tbl[ctx->idx].vlan_id; /* DEL_GID is called in WQ context(netdevice_event_work_handler) * or via the ib_unregister_device path. In the former case QP1 * may not be destroyed yet, in which case just return as FW @@ -335,7 +337,8 @@ int bnxt_re_del_gid(const struct ib_gid_attr *attr, void **context) } ctx->refcnt--; if (!ctx->refcnt) { - rc = bnxt_qplib_del_sgid(sgid_tbl, gid_to_del, true); + rc = bnxt_qplib_del_sgid(sgid_tbl, gid_to_del, + vlan_id, true); if (rc) { dev_err(rdev_to_dev(rdev), "Failed to remove GID: %#x", rc); diff --git a/drivers/infiniband/hw/bnxt_re/main.c b/drivers/infiniband/hw/bnxt_re/main.c index 029babe713f3..30a54f8aa42c 100644 --- a/drivers/infiniband/hw/bnxt_re/main.c +++ b/drivers/infiniband/hw/bnxt_re/main.c @@ -1473,7 +1473,6 @@ static int bnxt_re_ib_reg(struct bnxt_re_dev *rdev) &rdev->active_width); set_bit(BNXT_RE_FLAG_ISSUE_ROCE_STATS, &rdev->flags); bnxt_re_dispatch_event(&rdev->ibdev, NULL, 1, IB_EVENT_PORT_ACTIVE); - bnxt_re_dispatch_event(&rdev->ibdev, NULL, 1, IB_EVENT_GID_CHANGE); return 0; free_sctx: diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.c b/drivers/infiniband/hw/bnxt_re/qplib_res.c index 37928b1111df..bdbde8e22420 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.c @@ -488,7 +488,7 @@ static int bnxt_qplib_alloc_sgid_tbl(struct bnxt_qplib_res *res, struct bnxt_qplib_sgid_tbl *sgid_tbl, u16 max) { - sgid_tbl->tbl = kcalloc(max, sizeof(struct bnxt_qplib_gid), GFP_KERNEL); + sgid_tbl->tbl = kcalloc(max, sizeof(*sgid_tbl->tbl), GFP_KERNEL); if (!sgid_tbl->tbl) return -ENOMEM; @@ -526,9 +526,10 @@ static void bnxt_qplib_cleanup_sgid_tbl(struct bnxt_qplib_res *res, for (i = 0; i < sgid_tbl->max; i++) { if (memcmp(&sgid_tbl->tbl[i], &bnxt_qplib_gid_zero, sizeof(bnxt_qplib_gid_zero))) - bnxt_qplib_del_sgid(sgid_tbl, &sgid_tbl->tbl[i], true); + bnxt_qplib_del_sgid(sgid_tbl, &sgid_tbl->tbl[i].gid, + sgid_tbl->tbl[i].vlan_id, true); } - memset(sgid_tbl->tbl, 0, sizeof(struct bnxt_qplib_gid) * sgid_tbl->max); + memset(sgid_tbl->tbl, 0, sizeof(*sgid_tbl->tbl) * sgid_tbl->max); memset(sgid_tbl->hw_id, -1, sizeof(u16) * sgid_tbl->max); memset(sgid_tbl->vlan, 0, sizeof(u8) * sgid_tbl->max); sgid_tbl->active = 0; @@ -537,7 +538,11 @@ static void bnxt_qplib_cleanup_sgid_tbl(struct bnxt_qplib_res *res, static void bnxt_qplib_init_sgid_tbl(struct bnxt_qplib_sgid_tbl *sgid_tbl, struct net_device *netdev) { - memset(sgid_tbl->tbl, 0, sizeof(struct bnxt_qplib_gid) * sgid_tbl->max); + u32 i; + + for (i = 0; i < sgid_tbl->max; i++) + sgid_tbl->tbl[i].vlan_id = 0xffff; + memset(sgid_tbl->hw_id, -1, sizeof(u16) * sgid_tbl->max); } diff --git a/drivers/infiniband/hw/bnxt_re/qplib_res.h b/drivers/infiniband/hw/bnxt_re/qplib_res.h index 30c42c92fac7..fbda11a7ab1a 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_res.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_res.h @@ -111,7 +111,7 @@ struct bnxt_qplib_pd_tbl { }; struct bnxt_qplib_sgid_tbl { - struct bnxt_qplib_gid *tbl; + struct bnxt_qplib_gid_info *tbl; u16 *hw_id; u16 max; u16 active; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c index 48793d3512ac..40296b97d21e 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c @@ -213,12 +213,12 @@ int bnxt_qplib_get_sgid(struct bnxt_qplib_res *res, index, sgid_tbl->max); return -EINVAL; } - memcpy(gid, &sgid_tbl->tbl[index], sizeof(*gid)); + memcpy(gid, &sgid_tbl->tbl[index].gid, sizeof(*gid)); return 0; } int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, - struct bnxt_qplib_gid *gid, bool update) + struct bnxt_qplib_gid *gid, u16 vlan_id, bool update) { struct bnxt_qplib_res *res = to_bnxt_qplib(sgid_tbl, struct bnxt_qplib_res, @@ -236,7 +236,8 @@ int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, return -ENOMEM; } for (index = 0; index < sgid_tbl->max; index++) { - if (!memcmp(&sgid_tbl->tbl[index], gid, sizeof(*gid))) + if (!memcmp(&sgid_tbl->tbl[index].gid, gid, sizeof(*gid)) && + vlan_id == sgid_tbl->tbl[index].vlan_id) break; } if (index == sgid_tbl->max) { @@ -262,8 +263,9 @@ int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, if (rc) return rc; } - memcpy(&sgid_tbl->tbl[index], &bnxt_qplib_gid_zero, + memcpy(&sgid_tbl->tbl[index].gid, &bnxt_qplib_gid_zero, sizeof(bnxt_qplib_gid_zero)); + sgid_tbl->tbl[index].vlan_id = 0xFFFF; sgid_tbl->vlan[index] = 0; sgid_tbl->active--; dev_dbg(&res->pdev->dev, @@ -296,7 +298,8 @@ int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, } free_idx = sgid_tbl->max; for (i = 0; i < sgid_tbl->max; i++) { - if (!memcmp(&sgid_tbl->tbl[i], gid, sizeof(*gid))) { + if (!memcmp(&sgid_tbl->tbl[i], gid, sizeof(*gid)) && + sgid_tbl->tbl[i].vlan_id == vlan_id) { dev_dbg(&res->pdev->dev, "SGID entry already exist in entry %d!\n", i); *index = i; @@ -351,6 +354,7 @@ int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, } /* Add GID to the sgid_tbl */ memcpy(&sgid_tbl->tbl[free_idx], gid, sizeof(*gid)); + sgid_tbl->tbl[free_idx].vlan_id = vlan_id; sgid_tbl->active++; if (vlan_id != 0xFFFF) sgid_tbl->vlan[free_idx] = 1; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h index 0ec3b12b0bcd..13d9432d5ce2 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h @@ -84,6 +84,11 @@ struct bnxt_qplib_gid { u8 data[16]; }; +struct bnxt_qplib_gid_info { + struct bnxt_qplib_gid gid; + u16 vlan_id; +}; + struct bnxt_qplib_ah { struct bnxt_qplib_gid dgid; struct bnxt_qplib_pd *pd; @@ -221,7 +226,7 @@ int bnxt_qplib_get_sgid(struct bnxt_qplib_res *res, struct bnxt_qplib_sgid_tbl *sgid_tbl, int index, struct bnxt_qplib_gid *gid); int bnxt_qplib_del_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, - struct bnxt_qplib_gid *gid, bool update); + struct bnxt_qplib_gid *gid, u16 vlan_id, bool update); int bnxt_qplib_add_sgid(struct bnxt_qplib_sgid_tbl *sgid_tbl, struct bnxt_qplib_gid *gid, u8 *mac, u16 vlan_id, bool update, u32 *index); diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c index e775c1a1a450..dcf02ec02810 100644 --- a/drivers/infiniband/hw/cxgb3/iwch_provider.c +++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c @@ -991,33 +991,8 @@ static int iwch_query_device(struct ib_device *ibdev, struct ib_device_attr *pro static int iwch_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { - struct iwch_dev *dev; - struct net_device *netdev; - struct in_device *inetdev; - pr_debug("%s ibdev %p\n", __func__, ibdev); - dev = to_iwch_dev(ibdev); - netdev = dev->rdev.port_info.lldevs[port-1]; - - /* props being zeroed by the caller, avoid zeroing it here */ - props->max_mtu = IB_MTU_4096; - props->active_mtu = ib_mtu_int_to_enum(netdev->mtu); - - if (!netif_carrier_ok(netdev)) - props->state = IB_PORT_DOWN; - else { - inetdev = in_dev_get(netdev); - if (inetdev) { - if (inetdev->ifa_list) - props->state = IB_PORT_ACTIVE; - else - props->state = IB_PORT_INIT; - in_dev_put(inetdev); - } else - props->state = IB_PORT_INIT; - } - props->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_SNMP_TUNNEL_SUP | @@ -1273,8 +1248,24 @@ static const struct ib_device_ops iwch_dev_ops = { INIT_RDMA_OBJ_SIZE(ib_ucontext, iwch_ucontext, ibucontext), }; +static int set_netdevs(struct ib_device *ib_dev, struct cxio_rdev *rdev) +{ + int ret; + int i; + + for (i = 0; i < rdev->port_info.nports; i++) { + ret = ib_device_set_netdev(ib_dev, rdev->port_info.lldevs[i], + i + 1); + if (ret) + return ret; + } + return 0; +} + int iwch_register_device(struct iwch_dev *dev) { + int err; + pr_debug("%s iwch_dev %p\n", __func__, dev); memset(&dev->ibdev.node_guid, 0, sizeof(dev->ibdev.node_guid)); memcpy(&dev->ibdev.node_guid, dev->rdev.t3cdev_p->lldev->dev_addr, 6); @@ -1315,6 +1306,10 @@ int iwch_register_device(struct iwch_dev *dev) rdma_set_device_sysfs_group(&dev->ibdev, &iwch_attr_group); ib_set_device_ops(&dev->ibdev, &iwch_dev_ops); + err = set_netdevs(&dev->ibdev, &dev->rdev); + if (err) + return err; + return ib_register_device(&dev->ibdev, "cxgb3_%d"); } diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c index 5e59c5708729..d373ac0fe2cb 100644 --- a/drivers/infiniband/hw/cxgb4/provider.c +++ b/drivers/infiniband/hw/cxgb4/provider.c @@ -305,32 +305,8 @@ static int c4iw_query_device(struct ib_device *ibdev, struct ib_device_attr *pro static int c4iw_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { - struct c4iw_dev *dev; - struct net_device *netdev; - struct in_device *inetdev; - pr_debug("ibdev %p\n", ibdev); - dev = to_c4iw_dev(ibdev); - netdev = dev->rdev.lldi.ports[port-1]; - /* props being zeroed by the caller, avoid zeroing it here */ - props->max_mtu = IB_MTU_4096; - props->active_mtu = ib_mtu_int_to_enum(netdev->mtu); - - if (!netif_carrier_ok(netdev)) - props->state = IB_PORT_DOWN; - else { - inetdev = in_dev_get(netdev); - if (inetdev) { - if (inetdev->ifa_list) - props->state = IB_PORT_ACTIVE; - else - props->state = IB_PORT_INIT; - in_dev_put(inetdev); - } else - props->state = IB_PORT_INIT; - } - props->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_SNMP_TUNNEL_SUP | diff --git a/drivers/infiniband/hw/efa/efa.h b/drivers/infiniband/hw/efa/efa.h index 119f8efec564..2283e432693e 100644 --- a/drivers/infiniband/hw/efa/efa.h +++ b/drivers/infiniband/hw/efa/efa.h @@ -156,5 +156,8 @@ int efa_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, int qp_attr_mask, struct ib_udata *udata); enum rdma_link_layer efa_port_link_layer(struct ib_device *ibdev, u8 port_num); +struct rdma_hw_stats *efa_alloc_hw_stats(struct ib_device *ibdev, u8 port_num); +int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats, + u8 port_num, int index); #endif /* _EFA_H_ */ diff --git a/drivers/infiniband/hw/efa/efa_com.c b/drivers/infiniband/hw/efa/efa_com.c index 2cb42484b0f8..3c412bc5b94f 100644 --- a/drivers/infiniband/hw/efa/efa_com.c +++ b/drivers/infiniband/hw/efa/efa_com.c @@ -109,17 +109,19 @@ static u32 efa_com_reg_read32(struct efa_com_dev *edev, u16 offset) } while (time_is_after_jiffies(exp_time)); if (read_resp->req_id != mmio_read->seq_num) { - ibdev_err(edev->efa_dev, - "Reading register timed out. expected: req id[%u] offset[%#x] actual: req id[%u] offset[%#x]\n", - mmio_read->seq_num, offset, read_resp->req_id, - read_resp->reg_off); + ibdev_err_ratelimited( + edev->efa_dev, + "Reading register timed out. expected: req id[%u] offset[%#x] actual: req id[%u] offset[%#x]\n", + mmio_read->seq_num, offset, read_resp->req_id, + read_resp->reg_off); err = EFA_MMIO_READ_INVALID; goto out; } if (read_resp->reg_off != offset) { - ibdev_err(edev->efa_dev, - "Reading register failed: wrong offset provided\n"); + ibdev_err_ratelimited( + edev->efa_dev, + "Reading register failed: wrong offset provided\n"); err = EFA_MMIO_READ_INVALID; goto out; } @@ -293,9 +295,10 @@ static struct efa_comp_ctx *efa_com_get_comp_ctx(struct efa_com_admin_queue *aq, u16 ctx_id = cmd_id & (aq->depth - 1); if (aq->comp_ctx[ctx_id].occupied && capture) { - ibdev_err(aq->efa_dev, - "Completion context for command_id %#x is occupied\n", - cmd_id); + ibdev_err_ratelimited( + aq->efa_dev, + "Completion context for command_id %#x is occupied\n", + cmd_id); return NULL; } @@ -401,7 +404,7 @@ static struct efa_comp_ctx *efa_com_submit_admin_cmd(struct efa_com_admin_queue spin_lock(&aq->sq.lock); if (!test_bit(EFA_AQ_STATE_RUNNING_BIT, &aq->state)) { - ibdev_err(aq->efa_dev, "Admin queue is closed\n"); + ibdev_err_ratelimited(aq->efa_dev, "Admin queue is closed\n"); spin_unlock(&aq->sq.lock); return ERR_PTR(-ENODEV); } @@ -519,8 +522,9 @@ static int efa_com_wait_and_process_admin_cq_polling(struct efa_comp_ctx *comp_c break; if (time_is_before_jiffies(timeout)) { - ibdev_err(aq->efa_dev, - "Wait for completion (polling) timeout\n"); + ibdev_err_ratelimited( + aq->efa_dev, + "Wait for completion (polling) timeout\n"); /* EFA didn't have any completion */ atomic64_inc(&aq->stats.no_completion); @@ -561,17 +565,19 @@ static int efa_com_wait_and_process_admin_cq_interrupts(struct efa_comp_ctx *com atomic64_inc(&aq->stats.no_completion); if (comp_ctx->status == EFA_CMD_COMPLETED) - ibdev_err(aq->efa_dev, - "The device sent a completion but the driver didn't receive any MSI-X interrupt for admin cmd %s(%d) status %d (ctx: 0x%p, sq producer: %d, sq consumer: %d, cq consumer: %d)\n", - efa_com_cmd_str(comp_ctx->cmd_opcode), - comp_ctx->cmd_opcode, comp_ctx->status, - comp_ctx, aq->sq.pc, aq->sq.cc, aq->cq.cc); + ibdev_err_ratelimited( + aq->efa_dev, + "The device sent a completion but the driver didn't receive any MSI-X interrupt for admin cmd %s(%d) status %d (ctx: 0x%p, sq producer: %d, sq consumer: %d, cq consumer: %d)\n", + efa_com_cmd_str(comp_ctx->cmd_opcode), + comp_ctx->cmd_opcode, comp_ctx->status, + comp_ctx, aq->sq.pc, aq->sq.cc, aq->cq.cc); else - ibdev_err(aq->efa_dev, - "The device didn't send any completion for admin cmd %s(%d) status %d (ctx 0x%p, sq producer: %d, sq consumer: %d, cq consumer: %d)\n", - efa_com_cmd_str(comp_ctx->cmd_opcode), - comp_ctx->cmd_opcode, comp_ctx->status, - comp_ctx, aq->sq.pc, aq->sq.cc, aq->cq.cc); + ibdev_err_ratelimited( + aq->efa_dev, + "The device didn't send any completion for admin cmd %s(%d) status %d (ctx 0x%p, sq producer: %d, sq consumer: %d, cq consumer: %d)\n", + efa_com_cmd_str(comp_ctx->cmd_opcode), + comp_ctx->cmd_opcode, comp_ctx->status, + comp_ctx, aq->sq.pc, aq->sq.cc, aq->cq.cc); clear_bit(EFA_AQ_STATE_RUNNING_BIT, &aq->state); err = -ETIME; @@ -633,10 +639,11 @@ int efa_com_cmd_exec(struct efa_com_admin_queue *aq, cmd->aq_common_descriptor.opcode); comp_ctx = efa_com_submit_admin_cmd(aq, cmd, cmd_size, comp, comp_size); if (IS_ERR(comp_ctx)) { - ibdev_err(aq->efa_dev, - "Failed to submit command %s (opcode %u) err %ld\n", - efa_com_cmd_str(cmd->aq_common_descriptor.opcode), - cmd->aq_common_descriptor.opcode, PTR_ERR(comp_ctx)); + ibdev_err_ratelimited( + aq->efa_dev, + "Failed to submit command %s (opcode %u) err %ld\n", + efa_com_cmd_str(cmd->aq_common_descriptor.opcode), + cmd->aq_common_descriptor.opcode, PTR_ERR(comp_ctx)); up(&aq->avail_cmds); return PTR_ERR(comp_ctx); @@ -644,11 +651,12 @@ int efa_com_cmd_exec(struct efa_com_admin_queue *aq, err = efa_com_wait_and_process_admin_cq(comp_ctx, aq); if (err) - ibdev_err(aq->efa_dev, - "Failed to process command %s (opcode %u) comp_status %d err %d\n", - efa_com_cmd_str(cmd->aq_common_descriptor.opcode), - cmd->aq_common_descriptor.opcode, - comp_ctx->comp_status, err); + ibdev_err_ratelimited( + aq->efa_dev, + "Failed to process command %s (opcode %u) comp_status %d err %d\n", + efa_com_cmd_str(cmd->aq_common_descriptor.opcode), + cmd->aq_common_descriptor.opcode, comp_ctx->comp_status, + err); up(&aq->avail_cmds); diff --git a/drivers/infiniband/hw/efa/efa_com_cmd.c b/drivers/infiniband/hw/efa/efa_com_cmd.c index 62345d8abf3c..501dce89f275 100644 --- a/drivers/infiniband/hw/efa/efa_com_cmd.c +++ b/drivers/infiniband/hw/efa/efa_com_cmd.c @@ -44,7 +44,8 @@ int efa_com_create_qp(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to create qp [%d]\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to create qp [%d]\n", err); return err; } @@ -82,9 +83,10 @@ int efa_com_modify_qp(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, - "Failed to modify qp-%u modify_mask[%#x] [%d]\n", - cmd.qp_handle, cmd.modify_mask, err); + ibdev_err_ratelimited( + edev->efa_dev, + "Failed to modify qp-%u modify_mask[%#x] [%d]\n", + cmd.qp_handle, cmd.modify_mask, err); return err; } @@ -109,8 +111,9 @@ int efa_com_query_qp(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to query qp-%u [%d]\n", - cmd.qp_handle, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to query qp-%u [%d]\n", + cmd.qp_handle, err); return err; } @@ -139,8 +142,9 @@ int efa_com_destroy_qp(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to destroy qp-%u [%d]\n", - qp_cmd.qp_handle, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to destroy qp-%u [%d]\n", + qp_cmd.qp_handle, err); return err; } @@ -173,7 +177,8 @@ int efa_com_create_cq(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to create cq[%d]\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to create cq[%d]\n", err); return err; } @@ -201,8 +206,9 @@ int efa_com_destroy_cq(struct efa_com_dev *edev, sizeof(destroy_resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to destroy CQ-%u [%d]\n", - params->cq_idx, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to destroy CQ-%u [%d]\n", + params->cq_idx, err); return err; } @@ -250,7 +256,8 @@ int efa_com_register_mr(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to register mr [%d]\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to register mr [%d]\n", err); return err; } @@ -277,9 +284,9 @@ int efa_com_dereg_mr(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, - "Failed to de-register mr(lkey-%u) [%d]\n", - mr_cmd.l_key, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to de-register mr(lkey-%u) [%d]\n", + mr_cmd.l_key, err); return err; } @@ -306,8 +313,9 @@ int efa_com_create_ah(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to create ah for %pI6 [%d]\n", - ah_cmd.dest_addr, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to create ah for %pI6 [%d]\n", + ah_cmd.dest_addr, err); return err; } @@ -334,8 +342,9 @@ int efa_com_destroy_ah(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&cmd_completion, sizeof(cmd_completion)); if (err) { - ibdev_err(edev->efa_dev, "Failed to destroy ah-%d pd-%d [%d]\n", - ah_cmd.ah, ah_cmd.pd, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to destroy ah-%d pd-%d [%d]\n", + ah_cmd.ah, ah_cmd.pd, err); return err; } @@ -367,8 +376,9 @@ static int efa_com_get_feature_ex(struct efa_com_dev *edev, int err; if (!efa_com_check_supported_feature_id(edev, feature_id)) { - ibdev_err(edev->efa_dev, "Feature %d isn't supported\n", - feature_id); + ibdev_err_ratelimited(edev->efa_dev, + "Feature %d isn't supported\n", + feature_id); return -EOPNOTSUPP; } @@ -396,9 +406,10 @@ static int efa_com_get_feature_ex(struct efa_com_dev *edev, sizeof(*get_resp)); if (err) { - ibdev_err(edev->efa_dev, - "Failed to submit get_feature command %d [%d]\n", - feature_id, err); + ibdev_err_ratelimited( + edev->efa_dev, + "Failed to submit get_feature command %d [%d]\n", + feature_id, err); return err; } @@ -421,8 +432,9 @@ int efa_com_get_network_attr(struct efa_com_dev *edev, err = efa_com_get_feature(edev, &resp, EFA_ADMIN_NETWORK_ATTR); if (err) { - ibdev_err(edev->efa_dev, - "Failed to get network attributes %d\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to get network attributes %d\n", + err); return err; } @@ -441,8 +453,9 @@ int efa_com_get_device_attr(struct efa_com_dev *edev, err = efa_com_get_feature(edev, &resp, EFA_ADMIN_DEVICE_ATTR); if (err) { - ibdev_err(edev->efa_dev, "Failed to get device attributes %d\n", - err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to get device attributes %d\n", + err); return err; } @@ -456,9 +469,10 @@ int efa_com_get_device_attr(struct efa_com_dev *edev, result->db_bar = resp.u.device_attr.db_bar; if (result->admin_api_version < 1) { - ibdev_err(edev->efa_dev, - "Failed to get device attr api version [%u < 1]\n", - result->admin_api_version); + ibdev_err_ratelimited( + edev->efa_dev, + "Failed to get device attr api version [%u < 1]\n", + result->admin_api_version); return -EINVAL; } @@ -466,8 +480,9 @@ int efa_com_get_device_attr(struct efa_com_dev *edev, err = efa_com_get_feature(edev, &resp, EFA_ADMIN_QUEUE_ATTR); if (err) { - ibdev_err(edev->efa_dev, - "Failed to get network attributes %d\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to get network attributes %d\n", + err); return err; } @@ -497,7 +512,8 @@ int efa_com_get_hw_hints(struct efa_com_dev *edev, err = efa_com_get_feature(edev, &resp, EFA_ADMIN_HW_HINTS); if (err) { - ibdev_err(edev->efa_dev, "Failed to get hw hints %d\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to get hw hints %d\n", err); return err; } @@ -520,8 +536,9 @@ static int efa_com_set_feature_ex(struct efa_com_dev *edev, int err; if (!efa_com_check_supported_feature_id(edev, feature_id)) { - ibdev_err(edev->efa_dev, "Feature %d isn't supported\n", - feature_id); + ibdev_err_ratelimited(edev->efa_dev, + "Feature %d isn't supported\n", + feature_id); return -EOPNOTSUPP; } @@ -545,9 +562,10 @@ static int efa_com_set_feature_ex(struct efa_com_dev *edev, sizeof(*set_resp)); if (err) { - ibdev_err(edev->efa_dev, - "Failed to submit set_feature command %d error: %d\n", - feature_id, err); + ibdev_err_ratelimited( + edev->efa_dev, + "Failed to submit set_feature command %d error: %d\n", + feature_id, err); return err; } @@ -574,8 +592,9 @@ int efa_com_set_aenq_config(struct efa_com_dev *edev, u32 groups) err = efa_com_get_feature(edev, &get_resp, EFA_ADMIN_AENQ_CONFIG); if (err) { - ibdev_err(edev->efa_dev, "Failed to get aenq attributes: %d\n", - err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to get aenq attributes: %d\n", + err); return err; } @@ -585,9 +604,10 @@ int efa_com_set_aenq_config(struct efa_com_dev *edev, u32 groups) get_resp.u.aenq.enabled_groups); if ((get_resp.u.aenq.supported_groups & groups) != groups) { - ibdev_err(edev->efa_dev, - "Trying to set unsupported aenq groups[%#x] supported[%#x]\n", - groups, get_resp.u.aenq.supported_groups); + ibdev_err_ratelimited( + edev->efa_dev, + "Trying to set unsupported aenq groups[%#x] supported[%#x]\n", + groups, get_resp.u.aenq.supported_groups); return -EOPNOTSUPP; } @@ -595,8 +615,9 @@ int efa_com_set_aenq_config(struct efa_com_dev *edev, u32 groups) err = efa_com_set_feature(edev, &set_resp, &cmd, EFA_ADMIN_AENQ_CONFIG); if (err) { - ibdev_err(edev->efa_dev, "Failed to set aenq attributes: %d\n", - err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to set aenq attributes: %d\n", + err); return err; } @@ -619,7 +640,8 @@ int efa_com_alloc_pd(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to allocate pd[%d]\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to allocate pd[%d]\n", err); return err; } @@ -645,8 +667,9 @@ int efa_com_dealloc_pd(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to deallocate pd-%u [%d]\n", - cmd.pd, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to deallocate pd-%u [%d]\n", + cmd.pd, err); return err; } @@ -669,7 +692,8 @@ int efa_com_alloc_uar(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to allocate uar[%d]\n", err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to allocate uar[%d]\n", err); return err; } @@ -695,10 +719,47 @@ int efa_com_dealloc_uar(struct efa_com_dev *edev, (struct efa_admin_acq_entry *)&resp, sizeof(resp)); if (err) { - ibdev_err(edev->efa_dev, "Failed to deallocate uar-%u [%d]\n", - cmd.uar, err); + ibdev_err_ratelimited(edev->efa_dev, + "Failed to deallocate uar-%u [%d]\n", + cmd.uar, err); return err; } return 0; } + +int efa_com_get_stats(struct efa_com_dev *edev, + struct efa_com_get_stats_params *params, + union efa_com_get_stats_result *result) +{ + struct efa_com_admin_queue *aq = &edev->aq; + struct efa_admin_aq_get_stats_cmd cmd = {}; + struct efa_admin_acq_get_stats_resp resp; + int err; + + cmd.aq_common_descriptor.opcode = EFA_ADMIN_GET_STATS; + cmd.type = params->type; + cmd.scope = params->scope; + cmd.scope_modifier = params->scope_modifier; + + err = efa_com_cmd_exec(aq, + (struct efa_admin_aq_entry *)&cmd, + sizeof(cmd), + (struct efa_admin_acq_entry *)&resp, + sizeof(resp)); + if (err) { + ibdev_err_ratelimited( + edev->efa_dev, + "Failed to get stats type-%u scope-%u.%u [%d]\n", + cmd.type, cmd.scope, cmd.scope_modifier, err); + return err; + } + + result->basic_stats.tx_bytes = resp.basic_stats.tx_bytes; + result->basic_stats.tx_pkts = resp.basic_stats.tx_pkts; + result->basic_stats.rx_bytes = resp.basic_stats.rx_bytes; + result->basic_stats.rx_pkts = resp.basic_stats.rx_pkts; + result->basic_stats.rx_drops = resp.basic_stats.rx_drops; + + return 0; +} diff --git a/drivers/infiniband/hw/efa/efa_com_cmd.h b/drivers/infiniband/hw/efa/efa_com_cmd.h index a1174380462c..7f6c13052f49 100644 --- a/drivers/infiniband/hw/efa/efa_com_cmd.h +++ b/drivers/infiniband/hw/efa/efa_com_cmd.h @@ -225,6 +225,26 @@ struct efa_com_dealloc_uar_params { u16 uarn; }; +struct efa_com_get_stats_params { + /* see enum efa_admin_get_stats_type */ + u8 type; + /* see enum efa_admin_get_stats_scope */ + u8 scope; + u16 scope_modifier; +}; + +struct efa_com_basic_stats { + u64 tx_bytes; + u64 tx_pkts; + u64 rx_bytes; + u64 rx_pkts; + u64 rx_drops; +}; + +union efa_com_get_stats_result { + struct efa_com_basic_stats basic_stats; +}; + void efa_com_set_dma_addr(dma_addr_t addr, u32 *addr_high, u32 *addr_low); int efa_com_create_qp(struct efa_com_dev *edev, struct efa_com_create_qp_params *params, @@ -266,5 +286,8 @@ int efa_com_alloc_uar(struct efa_com_dev *edev, struct efa_com_alloc_uar_result *result); int efa_com_dealloc_uar(struct efa_com_dev *edev, struct efa_com_dealloc_uar_params *params); +int efa_com_get_stats(struct efa_com_dev *edev, + struct efa_com_get_stats_params *params, + union efa_com_get_stats_result *result); #endif /* _EFA_COM_CMD_H_ */ diff --git a/drivers/infiniband/hw/efa/efa_main.c b/drivers/infiniband/hw/efa/efa_main.c index dd1c6d49466f..83858f7e83d0 100644 --- a/drivers/infiniband/hw/efa/efa_main.c +++ b/drivers/infiniband/hw/efa/efa_main.c @@ -201,6 +201,7 @@ static const struct ib_device_ops efa_dev_ops = { .driver_id = RDMA_DRIVER_EFA, .uverbs_abi_ver = EFA_UVERBS_ABI_VERSION, + .alloc_hw_stats = efa_alloc_hw_stats, .alloc_pd = efa_alloc_pd, .alloc_ucontext = efa_alloc_ucontext, .create_ah = efa_create_ah, @@ -212,6 +213,7 @@ static const struct ib_device_ops efa_dev_ops = { .destroy_ah = efa_destroy_ah, .destroy_cq = efa_destroy_cq, .destroy_qp = efa_destroy_qp, + .get_hw_stats = efa_get_hw_stats, .get_link_layer = efa_port_link_layer, .get_port_immutable = efa_get_port_immutable, .mmap = efa_mmap, diff --git a/drivers/infiniband/hw/efa/efa_verbs.c b/drivers/infiniband/hw/efa/efa_verbs.c index df77bc312a25..4edae89e8e3c 100644 --- a/drivers/infiniband/hw/efa/efa_verbs.c +++ b/drivers/infiniband/hw/efa/efa_verbs.c @@ -41,6 +41,33 @@ static inline u64 get_mmap_key(const struct efa_mmap_entry *efa) ((u64)efa->mmap_page << PAGE_SHIFT); } +#define EFA_DEFINE_STATS(op) \ + op(EFA_TX_BYTES, "tx_bytes") \ + op(EFA_TX_PKTS, "tx_pkts") \ + op(EFA_RX_BYTES, "rx_bytes") \ + op(EFA_RX_PKTS, "rx_pkts") \ + op(EFA_RX_DROPS, "rx_drops") \ + op(EFA_SUBMITTED_CMDS, "submitted_cmds") \ + op(EFA_COMPLETED_CMDS, "completed_cmds") \ + op(EFA_NO_COMPLETION_CMDS, "no_completion_cmds") \ + op(EFA_KEEP_ALIVE_RCVD, "keep_alive_rcvd") \ + op(EFA_ALLOC_PD_ERR, "alloc_pd_err") \ + op(EFA_CREATE_QP_ERR, "create_qp_err") \ + op(EFA_REG_MR_ERR, "reg_mr_err") \ + op(EFA_ALLOC_UCONTEXT_ERR, "alloc_ucontext_err") \ + op(EFA_CREATE_AH_ERR, "create_ah_err") + +#define EFA_STATS_ENUM(ename, name) ename, +#define EFA_STATS_STR(ename, name) [ename] = name, + +enum efa_hw_stats { + EFA_DEFINE_STATS(EFA_STATS_ENUM) +}; + +static const char *const efa_stats_names[] = { + EFA_DEFINE_STATS(EFA_STATS_STR) +}; + #define EFA_CHUNK_PAYLOAD_SHIFT 12 #define EFA_CHUNK_PAYLOAD_SIZE BIT(EFA_CHUNK_PAYLOAD_SHIFT) #define EFA_CHUNK_PAYLOAD_PTR_SIZE 8 @@ -121,7 +148,7 @@ static inline struct efa_ah *to_eah(struct ib_ah *ibah) } #define field_avail(x, fld, sz) (offsetof(typeof(x), fld) + \ - sizeof(((typeof(x) *)0)->fld) <= (sz)) + FIELD_SIZEOF(typeof(x), fld) <= (sz)) #define is_reserved_cleared(reserved) \ !memchr_inv(reserved, 0, sizeof(reserved)) @@ -306,7 +333,7 @@ int efa_query_port(struct ib_device *ibdev, u8 port, props->lmc = 1; props->state = IB_PORT_ACTIVE; - props->phys_state = 5; + props->phys_state = IB_PORT_PHYS_STATE_LINK_UP; props->gid_tbl_len = 1; props->pkey_tbl_len = 1; props->active_speed = IB_SPEED_EDR; @@ -1473,14 +1500,12 @@ int efa_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata) ibdev_dbg(&dev->ibdev, "Deregister mr[%d]\n", ibmr->lkey); - if (mr->umem) { - params.l_key = mr->ibmr.lkey; - err = efa_com_dereg_mr(&dev->edev, ¶ms); - if (err) - return err; - } - ib_umem_release(mr->umem); + params.l_key = mr->ibmr.lkey; + err = efa_com_dereg_mr(&dev->edev, ¶ms); + if (err) + return err; + ib_umem_release(mr->umem); kfree(mr); return 0; @@ -1727,6 +1752,54 @@ void efa_destroy_ah(struct ib_ah *ibah, u32 flags) efa_ah_destroy(dev, ah); } +struct rdma_hw_stats *efa_alloc_hw_stats(struct ib_device *ibdev, u8 port_num) +{ + return rdma_alloc_hw_stats_struct(efa_stats_names, + ARRAY_SIZE(efa_stats_names), + RDMA_HW_STATS_DEFAULT_LIFESPAN); +} + +int efa_get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats, + u8 port_num, int index) +{ + struct efa_com_get_stats_params params = {}; + union efa_com_get_stats_result result; + struct efa_dev *dev = to_edev(ibdev); + struct efa_com_basic_stats *bs; + struct efa_com_stats_admin *as; + struct efa_stats *s; + int err; + + params.type = EFA_ADMIN_GET_STATS_TYPE_BASIC; + params.scope = EFA_ADMIN_GET_STATS_SCOPE_ALL; + + err = efa_com_get_stats(&dev->edev, ¶ms, &result); + if (err) + return err; + + bs = &result.basic_stats; + stats->value[EFA_TX_BYTES] = bs->tx_bytes; + stats->value[EFA_TX_PKTS] = bs->tx_pkts; + stats->value[EFA_RX_BYTES] = bs->rx_bytes; + stats->value[EFA_RX_PKTS] = bs->rx_pkts; + stats->value[EFA_RX_DROPS] = bs->rx_drops; + + as = &dev->edev.aq.stats; + stats->value[EFA_SUBMITTED_CMDS] = atomic64_read(&as->submitted_cmd); + stats->value[EFA_COMPLETED_CMDS] = atomic64_read(&as->completed_cmd); + stats->value[EFA_NO_COMPLETION_CMDS] = atomic64_read(&as->no_completion); + + s = &dev->stats; + stats->value[EFA_KEEP_ALIVE_RCVD] = atomic64_read(&s->keep_alive_rcvd); + stats->value[EFA_ALLOC_PD_ERR] = atomic64_read(&s->sw_stats.alloc_pd_err); + stats->value[EFA_CREATE_QP_ERR] = atomic64_read(&s->sw_stats.create_qp_err); + stats->value[EFA_REG_MR_ERR] = atomic64_read(&s->sw_stats.reg_mr_err); + stats->value[EFA_ALLOC_UCONTEXT_ERR] = atomic64_read(&s->sw_stats.alloc_ucontext_err); + stats->value[EFA_CREATE_AH_ERR] = atomic64_read(&s->sw_stats.create_ah_err); + + return ARRAY_SIZE(efa_stats_names); +} + enum rdma_link_layer efa_port_link_layer(struct ib_device *ibdev, u8 port_num) { diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c index d5b643a1d9fd..67052dc3100c 100644 --- a/drivers/infiniband/hw/hfi1/chip.c +++ b/drivers/infiniband/hw/hfi1/chip.c @@ -14452,7 +14452,7 @@ void hfi1_deinit_vnic_rsm(struct hfi1_devdata *dd) clear_rcvctrl(dd, RCV_CTRL_RCV_RSM_ENABLE_SMASK); } -static void init_rxe(struct hfi1_devdata *dd) +static int init_rxe(struct hfi1_devdata *dd) { struct rsm_map_table *rmt; u64 val; @@ -14461,6 +14461,9 @@ static void init_rxe(struct hfi1_devdata *dd) write_csr(dd, RCV_ERR_MASK, ~0ull); rmt = alloc_rsm_map_table(dd); + if (!rmt) + return -ENOMEM; + /* set up QOS, including the QPN map table */ init_qos(dd, rmt); init_fecn_handling(dd, rmt); @@ -14487,6 +14490,7 @@ static void init_rxe(struct hfi1_devdata *dd) val |= ((4ull & RCV_BYPASS_HDR_SIZE_MASK) << RCV_BYPASS_HDR_SIZE_SHIFT); write_csr(dd, RCV_BYPASS, val); + return 0; } static void init_other(struct hfi1_devdata *dd) @@ -15024,7 +15028,10 @@ int hfi1_init_dd(struct hfi1_devdata *dd) goto bail_cleanup; /* set initial RXE CSRs */ - init_rxe(dd); + ret = init_rxe(dd); + if (ret) + goto bail_cleanup; + /* set initial TXE CSRs */ init_txe(dd); /* set initial non-RXE, non-TXE CSRs */ diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c index 0477c14633ab..024a7c2b6124 100644 --- a/drivers/infiniband/hw/hfi1/rc.c +++ b/drivers/infiniband/hw/hfi1/rc.c @@ -1835,7 +1835,6 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct hfi1_opa_header *opah) cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) <= 0) break; trdma_clean_swqe(qp, wqe); - rvt_qp_wqe_unreserve(qp, wqe); trace_hfi1_qp_send_completion(qp, wqe, qp->s_last); rvt_qp_complete_swqe(qp, wqe, @@ -1882,7 +1881,6 @@ struct rvt_swqe *do_rc_completion(struct rvt_qp *qp, if (cmp_psn(wqe->lpsn, qp->s_sending_psn) < 0 || cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) > 0) { trdma_clean_swqe(qp, wqe); - rvt_qp_wqe_unreserve(qp, wqe); trace_hfi1_qp_send_completion(qp, wqe, qp->s_last); rvt_qp_complete_swqe(qp, wqe, diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c index 92acccaaaa86..996fc298207e 100644 --- a/drivers/infiniband/hw/hfi1/tid_rdma.c +++ b/drivers/infiniband/hw/hfi1/tid_rdma.c @@ -1620,6 +1620,7 @@ static int hfi1_kern_exp_rcv_alloc_flows(struct tid_rdma_request *req, flows[i].req = req; flows[i].npagesets = 0; flows[i].pagesets[0].mapped = 0; + flows[i].resync_npkts = 0; } req->flows = flows; return 0; @@ -1673,34 +1674,6 @@ static struct tid_rdma_flow *find_flow_ib(struct tid_rdma_request *req, return NULL; } -static struct tid_rdma_flow * -__find_flow_ranged(struct tid_rdma_request *req, u16 head, u16 tail, - u32 psn, u16 *fidx) -{ - for ( ; CIRC_CNT(head, tail, MAX_FLOWS); - tail = CIRC_NEXT(tail, MAX_FLOWS)) { - struct tid_rdma_flow *flow = &req->flows[tail]; - u32 spsn, lpsn; - - spsn = full_flow_psn(flow, flow->flow_state.spsn); - lpsn = full_flow_psn(flow, flow->flow_state.lpsn); - - if (cmp_psn(psn, spsn) >= 0 && cmp_psn(psn, lpsn) <= 0) { - if (fidx) - *fidx = tail; - return flow; - } - } - return NULL; -} - -static struct tid_rdma_flow *find_flow(struct tid_rdma_request *req, - u32 psn, u16 *fidx) -{ - return __find_flow_ranged(req, req->setup_head, req->clear_tail, psn, - fidx); -} - /* TID RDMA READ functions */ u32 hfi1_build_tid_rdma_read_packet(struct rvt_swqe *wqe, struct ib_other_headers *ohdr, u32 *bth1, @@ -2788,19 +2761,7 @@ static bool handle_read_kdeth_eflags(struct hfi1_ctxtdata *rcd, * to prevent continuous Flow Sequence errors for any * packets that could be still in the fabric. */ - flow = find_flow(req, psn, NULL); - if (!flow) { - /* - * We can't find the IB PSN matching the - * received KDETH PSN. The only thing we can - * do at this point is report the error to - * the QP. - */ - hfi1_kern_read_tid_flow_free(qp); - spin_unlock(&qp->s_lock); - rvt_rc_error(qp, IB_WC_LOC_QP_OP_ERR); - return ret; - } + flow = &req->flows[req->clear_tail]; if (priv->s_flags & HFI1_R_TID_SW_PSN) { diff = cmp_psn(psn, flow->flow_state.r_next_psn); diff --git a/drivers/infiniband/hw/hfi1/user_sdma.h b/drivers/infiniband/hw/hfi1/user_sdma.h index 4d8510b0fc38..9972e0e6545e 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.h +++ b/drivers/infiniband/hw/hfi1/user_sdma.h @@ -110,12 +110,6 @@ enum pkt_q_sdma_state { SDMA_PKT_Q_DEFERRED, }; -/* - * Maximum retry attempts to submit a TX request - * before putting the process to sleep. - */ -#define MAX_DEFER_RETRY_COUNT 1 - #define SDMA_IOWAIT_TIMEOUT 1000 /* in milliseconds */ #define SDMA_DBG(req, fmt, ...) \ diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index c4b243f50c76..9f53f63b1453 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -54,6 +54,7 @@ #include <linux/mm.h> #include <linux/vmalloc.h> #include <rdma/opa_addr.h> +#include <linux/nospec.h> #include "hfi.h" #include "common.h" @@ -873,16 +874,17 @@ int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps, else pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT); - if (unlikely(hfi1_dbg_should_fault_tx(qp, ps->opcode))) - pbc = hfi1_fault_tx(qp, ps->opcode, pbc); pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen); - /* Update HCRC based on packet opcode */ - pbc = update_hcrc(ps->opcode, pbc); + if (unlikely(hfi1_dbg_should_fault_tx(qp, ps->opcode))) + pbc = hfi1_fault_tx(qp, ps->opcode, pbc); + else + /* Update HCRC based on packet opcode */ + pbc = update_hcrc(ps->opcode, pbc); } tx->wqe = qp->s_wqe; ret = build_verbs_tx_desc(tx->sde, len, tx, ahg_info, pbc); @@ -1029,12 +1031,12 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps, else pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT); + pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen); if (unlikely(hfi1_dbg_should_fault_tx(qp, ps->opcode))) pbc = hfi1_fault_tx(qp, ps->opcode, pbc); - pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen); - - /* Update HCRC based on packet opcode */ - pbc = update_hcrc(ps->opcode, pbc); + else + /* Update HCRC based on packet opcode */ + pbc = update_hcrc(ps->opcode, pbc); } if (cb) iowait_pio_inc(&priv->s_iowait); @@ -1536,6 +1538,7 @@ static int hfi1_check_ah(struct ib_device *ibdev, struct rdma_ah_attr *ah_attr) sl = rdma_ah_get_sl(ah_attr); if (sl >= ARRAY_SIZE(ibp->sl_to_sc)) return -EINVAL; + sl = array_index_nospec(sl, ARRAY_SIZE(ibp->sl_to_sc)); sc5 = ibp->sl_to_sc[sl]; if (sc_to_vlt(dd, sc5) > num_vls && sc_to_vlt(dd, sc5) != 0xf) diff --git a/drivers/infiniband/hw/hns/Kconfig b/drivers/infiniband/hw/hns/Kconfig index 8bf847bcd8d3..d602b698b57e 100644 --- a/drivers/infiniband/hw/hns/Kconfig +++ b/drivers/infiniband/hw/hns/Kconfig @@ -1,6 +1,6 @@ # SPDX-License-Identifier: GPL-2.0-only config INFINIBAND_HNS - tristate "HNS RoCE Driver" + bool "HNS RoCE Driver" depends on NET_VENDOR_HISILICON depends on ARM64 || (COMPILE_TEST && 64BIT) ---help--- @@ -8,24 +8,16 @@ config INFINIBAND_HNS is used in Hisilicon Hip06 and more further ICT SoC based on platform device. - To compile HIP06 or HIP08 driver as module, choose M here. - config INFINIBAND_HNS_HIP06 - bool "Hisilicon Hip06 Family RoCE support" + tristate "Hisilicon Hip06 Family RoCE support" depends on INFINIBAND_HNS && HNS && HNS_DSAF && HNS_ENET ---help--- RoCE driver support for Hisilicon RoCE engine in Hisilicon Hip06 and Hip07 SoC. These RoCE engines are platform devices. - To compile this driver, choose Y here: if INFINIBAND_HNS is m, this - module will be called hns-roce-hw-v1 - config INFINIBAND_HNS_HIP08 - bool "Hisilicon Hip08 Family RoCE support" + tristate "Hisilicon Hip08 Family RoCE support" depends on INFINIBAND_HNS && PCI && HNS3 ---help--- RoCE driver support for Hisilicon RoCE engine in Hisilicon Hip08 SoC. The RoCE engine is a PCI device. - - To compile this driver, choose Y here: if INFINIBAND_HNS is m, this - module will be called hns-roce-hw-v2. diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile index e105945b94a1..449a2d81319d 100644 --- a/drivers/infiniband/hw/hns/Makefile +++ b/drivers/infiniband/hw/hns/Makefile @@ -9,12 +9,8 @@ hns-roce-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 -ifdef CONFIG_INFINIBAND_HNS_HIP06 hns-roce-hw-v1-objs := hns_roce_hw_v1.o $(hns-roce-objs) -obj-$(CONFIG_INFINIBAND_HNS) += hns-roce-hw-v1.o -endif +obj-$(CONFIG_INFINIBAND_HNS_HIP06) += hns-roce-hw-v1.o -ifdef CONFIG_INFINIBAND_HNS_HIP08 hns-roce-hw-v2-objs := hns_roce_hw_v2.o hns_roce_hw_v2_dfx.o $(hns-roce-objs) -obj-$(CONFIG_INFINIBAND_HNS) += hns-roce-hw-v2.o -endif +obj-$(CONFIG_INFINIBAND_HNS_HIP08) += hns-roce-hw-v2.o diff --git a/drivers/infiniband/hw/hns/hns_roce_cmd.c b/drivers/infiniband/hw/hns/hns_roce_cmd.c index 0cd09bf4d7ea..ade26faade8d 100644 --- a/drivers/infiniband/hw/hns/hns_roce_cmd.c +++ b/drivers/infiniband/hw/hns/hns_roce_cmd.c @@ -211,7 +211,6 @@ int hns_roce_cmd_init(struct hns_roce_dev *hr_dev) mutex_init(&hr_dev->cmd.hcr_mutex); sema_init(&hr_dev->cmd.poll_sem, 1); hr_dev->cmd.use_events = 0; - hr_dev->cmd.toggle = 1; hr_dev->cmd.max_cmds = CMD_MAX_NUM; hr_dev->cmd.pool = dma_pool_create("hns_roce_cmd", dev, HNS_ROCE_MAILBOX_SIZE, diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c index 4e50c22a2da4..22541d19cd09 100644 --- a/drivers/infiniband/hw/hns/hns_roce_cq.c +++ b/drivers/infiniband/hw/hns/hns_roce_cq.c @@ -83,7 +83,6 @@ static int hns_roce_sw2hw_cq(struct hns_roce_dev *dev, static int hns_roce_cq_alloc(struct hns_roce_dev *hr_dev, int nent, struct hns_roce_mtt *hr_mtt, - struct hns_roce_uar *hr_uar, struct hns_roce_cq *hr_cq, int vector) { struct hns_roce_cmd_mailbox *mailbox; @@ -154,7 +153,6 @@ static int hns_roce_cq_alloc(struct hns_roce_dev *hr_dev, int nent, hr_cq->cons_index = 0; hr_cq->arm_sn = 1; - hr_cq->uar = hr_uar; atomic_set(&hr_cq->refcount, 1); init_completion(&hr_cq->free); @@ -298,21 +296,127 @@ static void hns_roce_ib_free_cq_buf(struct hns_roce_dev *hr_dev, &buf->hr_buf); } +static int create_user_cq(struct hns_roce_dev *hr_dev, + struct hns_roce_cq *hr_cq, + struct ib_udata *udata, + struct hns_roce_ib_create_cq_resp *resp, + int cq_entries) +{ + struct hns_roce_ib_create_cq ucmd; + struct device *dev = hr_dev->dev; + int ret; + struct hns_roce_ucontext *context = rdma_udata_to_drv_context( + udata, struct hns_roce_ucontext, ibucontext); + + if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { + dev_err(dev, "Failed to copy_from_udata.\n"); + return -EFAULT; + } + + /* Get user space address, write it into mtt table */ + ret = hns_roce_ib_get_cq_umem(hr_dev, udata, &hr_cq->hr_buf, + &hr_cq->umem, ucmd.buf_addr, + cq_entries); + if (ret) { + dev_err(dev, "Failed to get_cq_umem.\n"); + return ret; + } + + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) && + (udata->outlen >= sizeof(*resp))) { + ret = hns_roce_db_map_user(context, udata, ucmd.db_addr, + &hr_cq->db); + if (ret) { + dev_err(dev, "cq record doorbell map failed!\n"); + goto err_mtt; + } + hr_cq->db_en = 1; + resp->cap_flags |= HNS_ROCE_SUPPORT_CQ_RECORD_DB; + } + + return 0; + +err_mtt: + hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt); + ib_umem_release(hr_cq->umem); + + return ret; +} + +static int create_kernel_cq(struct hns_roce_dev *hr_dev, + struct hns_roce_cq *hr_cq, int cq_entries) +{ + struct device *dev = hr_dev->dev; + struct hns_roce_uar *uar; + int ret; + + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) { + ret = hns_roce_alloc_db(hr_dev, &hr_cq->db, 1); + if (ret) + return ret; + + hr_cq->set_ci_db = hr_cq->db.db_record; + *hr_cq->set_ci_db = 0; + hr_cq->db_en = 1; + } + + /* Init mtt table and write buff address to mtt table */ + ret = hns_roce_ib_alloc_cq_buf(hr_dev, &hr_cq->hr_buf, cq_entries); + if (ret) { + dev_err(dev, "Failed to alloc_cq_buf.\n"); + goto err_db; + } + + uar = &hr_dev->priv_uar; + hr_cq->cq_db_l = hr_dev->reg_base + hr_dev->odb_offset + + DB_REG_OFFSET * uar->index; + + return 0; + +err_db: + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) + hns_roce_free_db(hr_dev, &hr_cq->db); + + return ret; +} + +static void destroy_user_cq(struct hns_roce_dev *hr_dev, + struct hns_roce_cq *hr_cq, + struct ib_udata *udata, + struct hns_roce_ib_create_cq_resp *resp) +{ + struct hns_roce_ucontext *context = rdma_udata_to_drv_context( + udata, struct hns_roce_ucontext, ibucontext); + + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) && + (udata->outlen >= sizeof(*resp))) + hns_roce_db_unmap_user(context, &hr_cq->db); + + hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt); + ib_umem_release(hr_cq->umem); +} + +static void destroy_kernel_cq(struct hns_roce_dev *hr_dev, + struct hns_roce_cq *hr_cq) +{ + hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt); + hns_roce_ib_free_cq_buf(hr_dev, &hr_cq->hr_buf, hr_cq->ib_cq.cqe); + + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) + hns_roce_free_db(hr_dev, &hr_cq->db); +} + int hns_roce_ib_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(ib_cq->device); struct device *dev = hr_dev->dev; - struct hns_roce_ib_create_cq ucmd; struct hns_roce_ib_create_cq_resp resp = {}; struct hns_roce_cq *hr_cq = to_hr_cq(ib_cq); - struct hns_roce_uar *uar = NULL; int vector = attr->comp_vector; int cq_entries = attr->cqe; int ret; - struct hns_roce_ucontext *context = rdma_udata_to_drv_context( - udata, struct hns_roce_ucontext, ibucontext); if (cq_entries < 1 || cq_entries > hr_dev->caps.max_cqes) { dev_err(dev, "Creat CQ failed. entries=%d, max=%d\n", @@ -328,61 +432,21 @@ int hns_roce_ib_create_cq(struct ib_cq *ib_cq, spin_lock_init(&hr_cq->lock); if (udata) { - if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { - dev_err(dev, "Failed to copy_from_udata.\n"); - ret = -EFAULT; - goto err_cq; - } - - /* Get user space address, write it into mtt table */ - ret = hns_roce_ib_get_cq_umem(hr_dev, udata, &hr_cq->hr_buf, - &hr_cq->umem, ucmd.buf_addr, - cq_entries); + ret = create_user_cq(hr_dev, hr_cq, udata, &resp, cq_entries); if (ret) { - dev_err(dev, "Failed to get_cq_umem.\n"); + dev_err(dev, "Create cq failed in user mode!\n"); goto err_cq; } - - if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) && - (udata->outlen >= sizeof(resp))) { - ret = hns_roce_db_map_user(context, udata, ucmd.db_addr, - &hr_cq->db); - if (ret) { - dev_err(dev, "cq record doorbell map failed!\n"); - goto err_mtt; - } - hr_cq->db_en = 1; - resp.cap_flags |= HNS_ROCE_SUPPORT_CQ_RECORD_DB; - } - - /* Get user space parameters */ - uar = &context->uar; } else { - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) { - ret = hns_roce_alloc_db(hr_dev, &hr_cq->db, 1); - if (ret) - goto err_cq; - - hr_cq->set_ci_db = hr_cq->db.db_record; - *hr_cq->set_ci_db = 0; - hr_cq->db_en = 1; - } - - /* Init mmt table and write buff address to mtt table */ - ret = hns_roce_ib_alloc_cq_buf(hr_dev, &hr_cq->hr_buf, - cq_entries); + ret = create_kernel_cq(hr_dev, hr_cq, cq_entries); if (ret) { - dev_err(dev, "Failed to alloc_cq_buf.\n"); - goto err_db; + dev_err(dev, "Create cq failed in kernel mode!\n"); + goto err_cq; } - - uar = &hr_dev->priv_uar; - hr_cq->cq_db_l = hr_dev->reg_base + hr_dev->odb_offset + - DB_REG_OFFSET * uar->index; } /* Allocate cq index, fill cq_context */ - ret = hns_roce_cq_alloc(hr_dev, cq_entries, &hr_cq->hr_buf.hr_mtt, uar, + ret = hns_roce_cq_alloc(hr_dev, cq_entries, &hr_cq->hr_buf.hr_mtt, hr_cq, vector); if (ret) { dev_err(dev, "Creat CQ .Failed to cq_alloc.\n"); @@ -416,20 +480,10 @@ err_cqc: hns_roce_free_cq(hr_dev, hr_cq); err_dbmap: - if (udata && (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB) && - (udata->outlen >= sizeof(resp))) - hns_roce_db_unmap_user(context, &hr_cq->db); - -err_mtt: - hns_roce_mtt_cleanup(hr_dev, &hr_cq->hr_buf.hr_mtt); - ib_umem_release(hr_cq->umem); - if (!udata) - hns_roce_ib_free_cq_buf(hr_dev, &hr_cq->hr_buf, - hr_cq->ib_cq.cqe); - -err_db: - if (!udata && (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RECORD_DB)) - hns_roce_free_db(hr_dev, &hr_cq->db); + if (udata) + destroy_user_cq(hr_dev, hr_cq, udata, &resp); + else + destroy_kernel_cq(hr_dev, hr_cq); err_cq: return ret; diff --git a/drivers/infiniband/hw/hns/hns_roce_db.c b/drivers/infiniband/hw/hns/hns_roce_db.c index 627aa46ef683..c00714c2f16a 100644 --- a/drivers/infiniband/hw/hns/hns_roce_db.c +++ b/drivers/infiniband/hw/hns/hns_roce_db.c @@ -12,13 +12,15 @@ int hns_roce_db_map_user(struct hns_roce_ucontext *context, struct ib_udata *udata, unsigned long virt, struct hns_roce_db *db) { + unsigned long page_addr = virt & PAGE_MASK; struct hns_roce_user_db_page *page; + unsigned int offset; int ret = 0; mutex_lock(&context->page_mutex); list_for_each_entry(page, &context->page_list, list) - if (page->user_virt == (virt & PAGE_MASK)) + if (page->user_virt == page_addr) goto found; page = kmalloc(sizeof(*page), GFP_KERNEL); @@ -28,8 +30,8 @@ int hns_roce_db_map_user(struct hns_roce_ucontext *context, } refcount_set(&page->refcount, 1); - page->user_virt = (virt & PAGE_MASK); - page->umem = ib_umem_get(udata, virt & PAGE_MASK, PAGE_SIZE, 0, 0); + page->user_virt = page_addr; + page->umem = ib_umem_get(udata, page_addr, PAGE_SIZE, 0, 0); if (IS_ERR(page->umem)) { ret = PTR_ERR(page->umem); kfree(page); @@ -39,10 +41,9 @@ int hns_roce_db_map_user(struct hns_roce_ucontext *context, list_add(&page->list, &context->page_list); found: - db->dma = sg_dma_address(page->umem->sg_head.sgl) + - (virt & ~PAGE_MASK); - page->umem->sg_head.sgl->offset = virt & ~PAGE_MASK; - db->virt_addr = sg_virt(page->umem->sg_head.sgl); + offset = virt - page_addr; + db->dma = sg_dma_address(page->umem->sg_head.sgl) + offset; + db->virt_addr = sg_virt(page->umem->sg_head.sgl) + offset; db->u.user_page = page; refcount_inc(&page->refcount); diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h index a548b28aab63..c7bf738fb137 100644 --- a/drivers/infiniband/hw/hns/hns_roce_device.h +++ b/drivers/infiniband/hw/hns/hns_roce_device.h @@ -84,7 +84,6 @@ #define HNS_ROCE_CEQ_ENTRY_SIZE 0x4 #define HNS_ROCE_AEQ_ENTRY_SIZE 0x10 -/* 4G/4K = 1M */ #define HNS_ROCE_SL_SHIFT 28 #define HNS_ROCE_TCLASS_SHIFT 20 #define HNS_ROCE_FLOW_LABEL_MASK 0xfffff @@ -322,7 +321,7 @@ struct hns_roce_hem_table { unsigned long num_hem; /* HEM entry record obj total num */ unsigned long num_obj; - /*Single obj size */ + /* Single obj size */ unsigned long obj_size; unsigned long table_chunk_size; int lowmem; @@ -343,7 +342,7 @@ struct hns_roce_mtt { struct hns_roce_buf_region { int offset; /* page offset */ - u32 count; /* page count*/ + u32 count; /* page count */ int hopnum; /* addressing hop num */ }; @@ -384,25 +383,25 @@ struct hns_roce_mr { u64 size; /* Address range of MR */ u32 key; /* Key of MR */ u32 pd; /* PD num of MR */ - u32 access;/* Access permission of MR */ + u32 access; /* Access permission of MR */ u32 npages; int enabled; /* MR's active status */ int type; /* MR's register type */ - u64 *pbl_buf;/* MR's PBL space */ + u64 *pbl_buf; /* MR's PBL space */ dma_addr_t pbl_dma_addr; /* MR's PBL space PA */ - u32 pbl_size;/* PA number in the PBL */ - u64 pbl_ba;/* page table address */ - u32 l0_chunk_last_num;/* L0 last number */ - u32 l1_chunk_last_num;/* L1 last number */ - u64 **pbl_bt_l2;/* PBL BT L2 */ - u64 **pbl_bt_l1;/* PBL BT L1 */ - u64 *pbl_bt_l0;/* PBL BT L0 */ - dma_addr_t *pbl_l2_dma_addr;/* PBL BT L2 dma addr */ - dma_addr_t *pbl_l1_dma_addr;/* PBL BT L1 dma addr */ - dma_addr_t pbl_l0_dma_addr;/* PBL BT L0 dma addr */ - u32 pbl_ba_pg_sz;/* BT chunk page size */ - u32 pbl_buf_pg_sz;/* buf chunk page size */ - u32 pbl_hop_num;/* multi-hop number */ + u32 pbl_size; /* PA number in the PBL */ + u64 pbl_ba; /* page table address */ + u32 l0_chunk_last_num; /* L0 last number */ + u32 l1_chunk_last_num; /* L1 last number */ + u64 **pbl_bt_l2; /* PBL BT L2 */ + u64 **pbl_bt_l1; /* PBL BT L1 */ + u64 *pbl_bt_l0; /* PBL BT L0 */ + dma_addr_t *pbl_l2_dma_addr; /* PBL BT L2 dma addr */ + dma_addr_t *pbl_l1_dma_addr; /* PBL BT L1 dma addr */ + dma_addr_t pbl_l0_dma_addr; /* PBL BT L0 dma addr */ + u32 pbl_ba_pg_sz; /* BT chunk page size */ + u32 pbl_buf_pg_sz; /* buf chunk page size */ + u32 pbl_hop_num; /* multi-hop number */ }; struct hns_roce_mr_table { @@ -425,16 +424,16 @@ struct hns_roce_wq { u32 max_post; int max_gs; int offset; - int wqe_shift;/* WQE size */ + int wqe_shift; /* WQE size */ u32 head; u32 tail; void __iomem *db_reg_l; }; struct hns_roce_sge { - int sge_cnt; /* SGE num */ + int sge_cnt; /* SGE num */ int offset; - int sge_shift;/* SGE size */ + int sge_shift; /* SGE size */ }; struct hns_roce_buf_list { @@ -618,7 +617,6 @@ struct hns_roce_cmdq { * close device, switch into poll mode(non event mode) */ u8 use_events; - u8 toggle; }; struct hns_roce_cmd_mailbox { @@ -654,8 +652,6 @@ struct hns_roce_qp { u32 doorbell_qpn; __le32 sq_signal_bits; u32 sq_next_wqe; - int sq_max_wqes_per_wr; - int sq_spare_wqes; struct hns_roce_wq sq; struct ib_umem *umem; @@ -752,7 +748,7 @@ struct hns_roce_eq { struct hns_roce_dev *hr_dev; void __iomem *doorbell; - int type_flag;/* Aeq:1 ceq:0 */ + int type_flag; /* Aeq:1 ceq:0 */ int eqn; u32 entries; int log_entries; @@ -798,22 +794,22 @@ struct hns_roce_caps { int local_ca_ack_delay; int num_uars; u32 phy_num_uars; - u32 max_sq_sg; /* 2 */ - u32 max_sq_inline; /* 32 */ - u32 max_rq_sg; /* 2 */ + u32 max_sq_sg; + u32 max_sq_inline; + u32 max_rq_sg; u32 max_extend_sg; - int num_qps; /* 256k */ + int num_qps; int reserved_qps; int num_qpc_timer; int num_cqc_timer; u32 max_srq_sg; int num_srqs; - u32 max_wqes; /* 16k */ + u32 max_wqes; u32 max_srqs; u32 max_srq_wrs; u32 max_srq_sges; - u32 max_sq_desc_sz; /* 64 */ - u32 max_rq_desc_sz; /* 64 */ + u32 max_sq_desc_sz; + u32 max_rq_desc_sz; u32 max_srq_desc_sz; int max_qp_init_rdma; int max_qp_dest_rdma; @@ -824,7 +820,7 @@ struct hns_roce_caps { int reserved_cqs; int reserved_srqs; u32 max_srqwqes; - int num_aeq_vectors; /* 1 */ + int num_aeq_vectors; int num_comp_vectors; int num_other_vectors; int num_mtpts; @@ -905,7 +901,7 @@ struct hns_roce_caps { u32 sl_num; u32 tsq_buf_pg_sz; u32 tpq_buf_pg_sz; - u32 chunk_sz; /* chunk size in non multihop mode*/ + u32 chunk_sz; /* chunk size in non multihop mode */ u64 flags; }; @@ -991,16 +987,6 @@ struct hns_roce_hw { const struct ib_device_ops *hns_roce_dev_srq_ops; }; -enum hns_phy_state { - HNS_ROCE_PHY_SLEEP = 1, - HNS_ROCE_PHY_POLLING = 2, - HNS_ROCE_PHY_DISABLED = 3, - HNS_ROCE_PHY_TRAINING = 4, - HNS_ROCE_PHY_LINKUP = 5, - HNS_ROCE_PHY_LINKERR = 6, - HNS_ROCE_PHY_TEST = 7 -}; - struct hns_roce_dev { struct ib_device ib_dev; struct platform_device *pdev; @@ -1045,8 +1031,8 @@ struct hns_roce_dev { int loop_idc; u32 sdb_offset; u32 odb_offset; - dma_addr_t tptr_dma_addr; /*only for hw v1*/ - u32 tptr_size; /*only for hw v1*/ + dma_addr_t tptr_dma_addr; /* only for hw v1 */ + u32 tptr_size; /* only for hw v1 */ const struct hns_roce_hw *hw; void *priv; struct workqueue_struct *irq_workq; diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c b/drivers/infiniband/hw/hns/hns_roce_hem.c index f4da5bd2884f..f2c4fef0b70e 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.c +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c @@ -41,29 +41,57 @@ bool hns_roce_check_whether_mhop(struct hns_roce_dev *hr_dev, u32 type) { - if ((hr_dev->caps.qpc_hop_num && type == HEM_TYPE_QPC) || - (hr_dev->caps.mpt_hop_num && type == HEM_TYPE_MTPT) || - (hr_dev->caps.cqc_hop_num && type == HEM_TYPE_CQC) || - (hr_dev->caps.srqc_hop_num && type == HEM_TYPE_SRQC) || - (hr_dev->caps.sccc_hop_num && type == HEM_TYPE_SCCC) || - (hr_dev->caps.qpc_timer_hop_num && type == HEM_TYPE_QPC_TIMER) || - (hr_dev->caps.cqc_timer_hop_num && type == HEM_TYPE_CQC_TIMER) || - (hr_dev->caps.cqe_hop_num && type == HEM_TYPE_CQE) || - (hr_dev->caps.mtt_hop_num && type == HEM_TYPE_MTT) || - (hr_dev->caps.srqwqe_hop_num && type == HEM_TYPE_SRQWQE) || - (hr_dev->caps.idx_hop_num && type == HEM_TYPE_IDX)) - return true; - - return false; + int hop_num = 0; + + switch (type) { + case HEM_TYPE_QPC: + hop_num = hr_dev->caps.qpc_hop_num; + break; + case HEM_TYPE_MTPT: + hop_num = hr_dev->caps.mpt_hop_num; + break; + case HEM_TYPE_CQC: + hop_num = hr_dev->caps.cqc_hop_num; + break; + case HEM_TYPE_SRQC: + hop_num = hr_dev->caps.srqc_hop_num; + break; + case HEM_TYPE_SCCC: + hop_num = hr_dev->caps.sccc_hop_num; + break; + case HEM_TYPE_QPC_TIMER: + hop_num = hr_dev->caps.qpc_timer_hop_num; + break; + case HEM_TYPE_CQC_TIMER: + hop_num = hr_dev->caps.cqc_timer_hop_num; + break; + case HEM_TYPE_CQE: + hop_num = hr_dev->caps.cqe_hop_num; + break; + case HEM_TYPE_MTT: + hop_num = hr_dev->caps.mtt_hop_num; + break; + case HEM_TYPE_SRQWQE: + hop_num = hr_dev->caps.srqwqe_hop_num; + break; + case HEM_TYPE_IDX: + hop_num = hr_dev->caps.idx_hop_num; + break; + default: + return false; + } + + return hop_num ? true : false; } static bool hns_roce_check_hem_null(struct hns_roce_hem **hem, u64 start_idx, - u32 bt_chunk_num) + u32 bt_chunk_num, u64 hem_max_num) { - int i; + u64 check_max_num = start_idx + bt_chunk_num; + u64 i; - for (i = 0; i < bt_chunk_num; i++) - if (hem[start_idx + i]) + for (i = start_idx; (i < check_max_num) && (i < hem_max_num); i++) + if (hem[i]) return false; return true; @@ -92,17 +120,13 @@ static int hns_roce_get_bt_num(u32 table_type, u32 hop_num) return 0; } -int hns_roce_calc_hem_mhop(struct hns_roce_dev *hr_dev, - struct hns_roce_hem_table *table, unsigned long *obj, - struct hns_roce_hem_mhop *mhop) +static int get_hem_table_config(struct hns_roce_dev *hr_dev, + struct hns_roce_hem_mhop *mhop, + u32 type) { struct device *dev = hr_dev->dev; - u32 chunk_ba_num; - u32 table_idx; - u32 bt_num; - u32 chunk_size; - switch (table->type) { + switch (type) { case HEM_TYPE_QPC: mhop->buf_chunk_size = 1 << (hr_dev->caps.qpc_buf_pg_sz + PAGE_SHIFT); @@ -193,10 +217,26 @@ int hns_roce_calc_hem_mhop(struct hns_roce_dev *hr_dev, break; default: dev_err(dev, "Table %d not support multi-hop addressing!\n", - table->type); + type); return -EINVAL; } + return 0; +} + +int hns_roce_calc_hem_mhop(struct hns_roce_dev *hr_dev, + struct hns_roce_hem_table *table, unsigned long *obj, + struct hns_roce_hem_mhop *mhop) +{ + struct device *dev = hr_dev->dev; + u32 chunk_ba_num; + u32 table_idx; + u32 bt_num; + u32 chunk_size; + + if (get_hem_table_config(hr_dev, mhop, table->type)) + return -EINVAL; + if (!obj) return 0; @@ -324,7 +364,7 @@ static int hns_roce_set_hem(struct hns_roce_dev *hr_dev, { spinlock_t *lock = &hr_dev->bt_cmd_lock; struct device *dev = hr_dev->dev; - unsigned long end = 0; + long end; unsigned long flags; struct hns_roce_hem_iter iter; void __iomem *bt_cmd; @@ -375,7 +415,7 @@ static int hns_roce_set_hem(struct hns_roce_dev *hr_dev, bt_cmd = hr_dev->reg_base + ROCEE_BT_CMD_H_REG; end = HW_SYNC_TIMEOUT_MSECS; - while (end) { + while (end > 0) { if (!(readl(bt_cmd) >> BT_CMD_SYNC_SHIFT)) break; @@ -457,6 +497,12 @@ static int hns_roce_table_mhop_get(struct hns_roce_dev *hr_dev, return -EINVAL; } + if (unlikely(hem_idx >= table->num_hem)) { + dev_err(dev, "Table %d exceed hem limt idx = %llu,max = %lu!\n", + table->type, hem_idx, table->num_hem); + return -EINVAL; + } + mutex_lock(&table->mutex); if (table->hem[hem_idx]) { @@ -693,7 +739,7 @@ static void hns_roce_table_mhop_put(struct hns_roce_dev *hr_dev, if (check_whether_bt_num_2(table->type, hop_num)) { start_idx = mhop.l0_idx * chunk_ba_num; if (hns_roce_check_hem_null(table->hem, start_idx, - chunk_ba_num)) { + chunk_ba_num, table->num_hem)) { if (table->type < HEM_TYPE_MTT && hr_dev->hw->clear_hem(hr_dev, table, obj, 0)) dev_warn(dev, "Clear HEM base address failed.\n"); @@ -707,7 +753,7 @@ static void hns_roce_table_mhop_put(struct hns_roce_dev *hr_dev, start_idx = mhop.l0_idx * chunk_ba_num * chunk_ba_num + mhop.l1_idx * chunk_ba_num; if (hns_roce_check_hem_null(table->hem, start_idx, - chunk_ba_num)) { + chunk_ba_num, table->num_hem)) { if (hr_dev->hw->clear_hem(hr_dev, table, obj, 1)) dev_warn(dev, "Clear HEM base address failed.\n"); @@ -791,7 +837,8 @@ void *hns_roce_table_find(struct hns_roce_dev *hr_dev, } else { u32 seg_size = 64; /* 8 bytes per BA and 8 BA per segment */ - hns_roce_calc_hem_mhop(hr_dev, table, &mhop_obj, &mhop); + if (hns_roce_calc_hem_mhop(hr_dev, table, &mhop_obj, &mhop)) + goto out; /* mtt mhop */ i = mhop.l0_idx; j = mhop.l1_idx; @@ -840,11 +887,13 @@ int hns_roce_table_get_range(struct hns_roce_dev *hr_dev, { struct hns_roce_hem_mhop mhop; unsigned long inc = table->table_chunk_size / table->obj_size; - unsigned long i; + unsigned long i = 0; int ret; if (hns_roce_check_whether_mhop(hr_dev, table->type)) { - hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop); + ret = hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop); + if (ret) + goto fail; inc = mhop.bt_chunk_size / table->obj_size; } @@ -874,7 +923,8 @@ void hns_roce_table_put_range(struct hns_roce_dev *hr_dev, unsigned long i; if (hns_roce_check_whether_mhop(hr_dev, table->type)) { - hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop); + if (hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop)) + return; inc = mhop.bt_chunk_size / table->obj_size; } @@ -887,7 +937,6 @@ int hns_roce_init_hem_table(struct hns_roce_dev *hr_dev, unsigned long obj_size, unsigned long nobj, int use_lowmem) { - struct device *dev = hr_dev->dev; unsigned long obj_per_chunk; unsigned long num_hem; @@ -900,99 +949,21 @@ int hns_roce_init_hem_table(struct hns_roce_dev *hr_dev, if (!table->hem) return -ENOMEM; } else { + struct hns_roce_hem_mhop mhop = {}; unsigned long buf_chunk_size; unsigned long bt_chunk_size; unsigned long bt_chunk_num; unsigned long num_bt_l0 = 0; u32 hop_num; - switch (type) { - case HEM_TYPE_QPC: - buf_chunk_size = 1 << (hr_dev->caps.qpc_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.qpc_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.qpc_bt_num; - hop_num = hr_dev->caps.qpc_hop_num; - break; - case HEM_TYPE_MTPT: - buf_chunk_size = 1 << (hr_dev->caps.mpt_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.mpt_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.mpt_bt_num; - hop_num = hr_dev->caps.mpt_hop_num; - break; - case HEM_TYPE_CQC: - buf_chunk_size = 1 << (hr_dev->caps.cqc_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.cqc_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.cqc_bt_num; - hop_num = hr_dev->caps.cqc_hop_num; - break; - case HEM_TYPE_SCCC: - buf_chunk_size = 1 << (hr_dev->caps.sccc_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.sccc_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.sccc_bt_num; - hop_num = hr_dev->caps.sccc_hop_num; - break; - case HEM_TYPE_QPC_TIMER: - buf_chunk_size = 1 << (hr_dev->caps.qpc_timer_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.qpc_timer_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.qpc_timer_bt_num; - hop_num = hr_dev->caps.qpc_timer_hop_num; - break; - case HEM_TYPE_CQC_TIMER: - buf_chunk_size = 1 << (hr_dev->caps.cqc_timer_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.cqc_timer_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.cqc_timer_bt_num; - hop_num = hr_dev->caps.cqc_timer_hop_num; - break; - case HEM_TYPE_SRQC: - buf_chunk_size = 1 << (hr_dev->caps.srqc_buf_pg_sz - + PAGE_SHIFT); - bt_chunk_size = 1 << (hr_dev->caps.srqc_ba_pg_sz - + PAGE_SHIFT); - num_bt_l0 = hr_dev->caps.srqc_bt_num; - hop_num = hr_dev->caps.srqc_hop_num; - break; - case HEM_TYPE_MTT: - buf_chunk_size = 1 << (hr_dev->caps.mtt_ba_pg_sz - + PAGE_SHIFT); - bt_chunk_size = buf_chunk_size; - hop_num = hr_dev->caps.mtt_hop_num; - break; - case HEM_TYPE_CQE: - buf_chunk_size = 1 << (hr_dev->caps.cqe_ba_pg_sz - + PAGE_SHIFT); - bt_chunk_size = buf_chunk_size; - hop_num = hr_dev->caps.cqe_hop_num; - break; - case HEM_TYPE_SRQWQE: - buf_chunk_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz - + PAGE_SHIFT); - bt_chunk_size = buf_chunk_size; - hop_num = hr_dev->caps.srqwqe_hop_num; - break; - case HEM_TYPE_IDX: - buf_chunk_size = 1 << (hr_dev->caps.idx_ba_pg_sz - + PAGE_SHIFT); - bt_chunk_size = buf_chunk_size; - hop_num = hr_dev->caps.idx_hop_num; - break; - default: - dev_err(dev, - "Table %d not support to init hem table here!\n", - type); + if (get_hem_table_config(hr_dev, &mhop, type)) return -EINVAL; - } + + buf_chunk_size = mhop.buf_chunk_size; + bt_chunk_size = mhop.bt_chunk_size; + num_bt_l0 = mhop.ba_l0_num; + hop_num = mhop.hop_num; + obj_per_chunk = buf_chunk_size / obj_size; num_hem = (nobj + obj_per_chunk - 1) / obj_per_chunk; bt_chunk_num = bt_chunk_size / BA_BYTE_LEN; @@ -1075,7 +1046,8 @@ static void hns_roce_cleanup_mhop_hem_table(struct hns_roce_dev *hr_dev, int i; u64 obj; - hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop); + if (hns_roce_calc_hem_mhop(hr_dev, table, NULL, &mhop)) + return; buf_chunk_size = table->type < HEM_TYPE_MTT ? mhop.buf_chunk_size : mhop.bt_chunk_size; diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h b/drivers/infiniband/hw/hns/hns_roce_hem.h index f1ccb8f35fe5..86783276fb1f 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.h +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h @@ -102,9 +102,9 @@ struct hns_roce_hem_mhop { u32 buf_chunk_size; u32 bt_chunk_size; u32 ba_l0_num; - u32 l0_idx;/* level 0 base address table index */ - u32 l1_idx;/* level 1 base address table index */ - u32 l2_idx;/* level 2 base address table index */ + u32 l0_idx; /* level 0 base address table index */ + u32 l1_idx; /* level 1 base address table index */ + u32 l2_idx; /* level 2 base address table index */ }; void hns_roce_free_hem(struct hns_roce_dev *hr_dev, struct hns_roce_hem *hem); diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c index 81e6dedb1e02..4c3ac2b75966 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c @@ -750,8 +750,10 @@ static int hns_roce_v1_rsv_lp_qp(struct hns_roce_dev *hr_dev) atomic_set(&free_mr->mr_free_cq->ib_cq.usecnt, 0); pd = rdma_zalloc_drv_obj(ibdev, ib_pd); - if (!pd) + if (!pd) { + ret = -ENOMEM; goto alloc_mem_failed; + } pd->device = ibdev; ret = hns_roce_alloc_pd(pd, NULL); @@ -969,7 +971,7 @@ static int hns_roce_v1_recreate_lp_qp(struct hns_roce_dev *hr_dev) struct hns_roce_free_mr *free_mr; struct hns_roce_v1_priv *priv; struct completion comp; - unsigned long end = HNS_ROCE_V1_RECREATE_LP_QP_TIMEOUT_MSECS; + long end = HNS_ROCE_V1_RECREATE_LP_QP_TIMEOUT_MSECS; priv = (struct hns_roce_v1_priv *)hr_dev->priv; free_mr = &priv->free_mr; @@ -989,7 +991,7 @@ static int hns_roce_v1_recreate_lp_qp(struct hns_roce_dev *hr_dev) queue_work(free_mr->free_mr_wq, &(lp_qp_work->work)); - while (end) { + while (end > 0) { if (try_wait_for_completion(&comp)) return 0; msleep(HNS_ROCE_V1_RECREATE_LP_QP_WAIT_VALUE); @@ -1107,7 +1109,7 @@ static int hns_roce_v1_dereg_mr(struct hns_roce_dev *hr_dev, struct hns_roce_free_mr *free_mr; struct hns_roce_v1_priv *priv; struct completion comp; - unsigned long end = HNS_ROCE_V1_FREE_MR_TIMEOUT_MSECS; + long end = HNS_ROCE_V1_FREE_MR_TIMEOUT_MSECS; unsigned long start = jiffies; int npages; int ret = 0; @@ -1137,7 +1139,7 @@ static int hns_roce_v1_dereg_mr(struct hns_roce_dev *hr_dev, queue_work(free_mr->free_mr_wq, &(mr_work->work)); - while (end) { + while (end > 0) { if (try_wait_for_completion(&comp)) goto free_mr; msleep(HNS_ROCE_V1_FREE_MR_WAIT_VALUE); @@ -2428,7 +2430,8 @@ static int hns_roce_v1_clear_hem(struct hns_roce_dev *hr_dev, { struct device *dev = &hr_dev->pdev->dev; struct hns_roce_v1_priv *priv; - unsigned long end = 0, flags = 0; + unsigned long flags = 0; + long end = HW_SYNC_TIMEOUT_MSECS; __le32 bt_cmd_val[2] = {0}; void __iomem *bt_cmd; u64 bt_ba = 0; @@ -2466,7 +2469,6 @@ static int hns_roce_v1_clear_hem(struct hns_roce_dev *hr_dev, bt_cmd = hr_dev->reg_base + ROCEE_BT_CMD_H_REG; - end = HW_SYNC_TIMEOUT_MSECS; while (1) { if (readl(bt_cmd) >> BT_CMD_SYNC_SHIFT) { if (!end) { @@ -4637,10 +4639,8 @@ static int hns_roce_get_cfg(struct hns_roce_dev *hr_dev) /* fetch the interrupt numbers */ for (i = 0; i < HNS_ROCE_V1_MAX_IRQ_NUM; i++) { hr_dev->irq[i] = platform_get_irq(hr_dev->pdev, i); - if (hr_dev->irq[i] <= 0) { - dev_err(dev, "platform get of irq[=%d] failed!\n", i); + if (hr_dev->irq[i] <= 0) return -EINVAL; - } } return 0; diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index b76e3beeafb8..206dfdb16cd5 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -239,7 +239,7 @@ static int hns_roce_v2_post_send(struct ib_qp *ibqp, struct device *dev = hr_dev->dev; struct hns_roce_v2_db sq_db; struct ib_qp_attr attr; - unsigned int sge_ind = 0; + unsigned int sge_ind; unsigned int owner_bit; unsigned long flags; unsigned int ind; @@ -887,8 +887,7 @@ static void hns_roce_cmq_init_regs(struct hns_roce_dev *hr_dev, bool ring_type) roce_write(hr_dev, ROCEE_TX_CMQ_BASEADDR_H_REG, upper_32_bits(dma)); roce_write(hr_dev, ROCEE_TX_CMQ_DEPTH_REG, - (ring->desc_num >> HNS_ROCE_CMQ_DESC_NUM_S) | - HNS_ROCE_CMQ_ENABLE); + ring->desc_num >> HNS_ROCE_CMQ_DESC_NUM_S); roce_write(hr_dev, ROCEE_TX_CMQ_HEAD_REG, 0); roce_write(hr_dev, ROCEE_TX_CMQ_TAIL_REG, 0); } else { @@ -896,8 +895,7 @@ static void hns_roce_cmq_init_regs(struct hns_roce_dev *hr_dev, bool ring_type) roce_write(hr_dev, ROCEE_RX_CMQ_BASEADDR_H_REG, upper_32_bits(dma)); roce_write(hr_dev, ROCEE_RX_CMQ_DEPTH_REG, - (ring->desc_num >> HNS_ROCE_CMQ_DESC_NUM_S) | - HNS_ROCE_CMQ_ENABLE); + ring->desc_num >> HNS_ROCE_CMQ_DESC_NUM_S); roce_write(hr_dev, ROCEE_RX_CMQ_HEAD_REG, 0); roce_write(hr_dev, ROCEE_RX_CMQ_TAIL_REG, 0); } @@ -1310,7 +1308,7 @@ static int hns_roce_set_vf_switch_param(struct hns_roce_dev *hr_dev, cpu_to_le16(HNS_ROCE_CMD_FLAG_NO_INTR | HNS_ROCE_CMD_FLAG_IN); desc.flag &= cpu_to_le16(~HNS_ROCE_CMD_FLAG_WR); roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_LPBK_S, 1); - roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_LCL_LPBK_S, 1); + roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_LCL_LPBK_S, 0); roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_DST_OVRD_S, 1); return hns_roce_cmq_send(hr_dev, &desc, 1); @@ -1888,7 +1886,7 @@ static int hns_roce_v2_init(struct hns_roce_dev *hr_dev) goto err_tpq_init_failed; } - /* Alloc memory for QPC Timer buffer space chunk*/ + /* Alloc memory for QPC Timer buffer space chunk */ for (qpc_count = 0; qpc_count < hr_dev->caps.qpc_timer_bt_num; qpc_count++) { ret = hns_roce_table_get(hr_dev, &hr_dev->qpc_timer_table, @@ -1899,7 +1897,7 @@ static int hns_roce_v2_init(struct hns_roce_dev *hr_dev) } } - /* Alloc memory for CQC Timer buffer space chunk*/ + /* Alloc memory for CQC Timer buffer space chunk */ for (cqc_count = 0; cqc_count < hr_dev->caps.cqc_timer_bt_num; cqc_count++) { ret = hns_roce_table_get(hr_dev, &hr_dev->cqc_timer_table, @@ -2409,7 +2407,7 @@ static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, for (prod_index = hr_cq->cons_index; get_sw_cqe_v2(hr_cq, prod_index); ++prod_index) { - if (prod_index == hr_cq->cons_index + hr_cq->ib_cq.cqe) + if (prod_index > hr_cq->cons_index + hr_cq->ib_cq.cqe) break; } @@ -2862,15 +2860,16 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, wc->smac[5] = roce_get_field(cqe->byte_28, V2_CQE_BYTE_28_SMAC_5_M, V2_CQE_BYTE_28_SMAC_5_S); + wc->wc_flags |= IB_WC_WITH_SMAC; if (roce_get_bit(cqe->byte_28, V2_CQE_BYTE_28_VID_VLD_S)) { wc->vlan_id = (u16)roce_get_field(cqe->byte_28, V2_CQE_BYTE_28_VID_M, V2_CQE_BYTE_28_VID_S); + wc->wc_flags |= IB_WC_WITH_VLAN; } else { wc->vlan_id = 0xffff; } - wc->wc_flags |= (IB_WC_WITH_VLAN | IB_WC_WITH_SMAC); wc->network_hdr_type = roce_get_field(cqe->byte_28, V2_CQE_BYTE_28_PORT_TYPE_M, V2_CQE_BYTE_28_PORT_TYPE_S); @@ -2905,11 +2904,49 @@ static int hns_roce_v2_poll_cq(struct ib_cq *ibcq, int num_entries, return npolled; } +static int get_op_for_set_hem(struct hns_roce_dev *hr_dev, u32 type, + int step_idx) +{ + int op; + + if (type == HEM_TYPE_SCCC && step_idx) + return -EINVAL; + + switch (type) { + case HEM_TYPE_QPC: + op = HNS_ROCE_CMD_WRITE_QPC_BT0; + break; + case HEM_TYPE_MTPT: + op = HNS_ROCE_CMD_WRITE_MPT_BT0; + break; + case HEM_TYPE_CQC: + op = HNS_ROCE_CMD_WRITE_CQC_BT0; + break; + case HEM_TYPE_SRQC: + op = HNS_ROCE_CMD_WRITE_SRQC_BT0; + break; + case HEM_TYPE_SCCC: + op = HNS_ROCE_CMD_WRITE_SCCC_BT0; + break; + case HEM_TYPE_QPC_TIMER: + op = HNS_ROCE_CMD_WRITE_QPC_TIMER_BT0; + break; + case HEM_TYPE_CQC_TIMER: + op = HNS_ROCE_CMD_WRITE_CQC_TIMER_BT0; + break; + default: + dev_warn(hr_dev->dev, + "Table %d not to be written by mailbox!\n", type); + return -EINVAL; + } + + return op + step_idx; +} + static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, struct hns_roce_hem_table *table, int obj, int step_idx) { - struct device *dev = hr_dev->dev; struct hns_roce_cmd_mailbox *mailbox; struct hns_roce_hem_iter iter; struct hns_roce_hem_mhop mhop; @@ -2922,7 +2959,7 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, u64 bt_ba = 0; u32 chunk_ba_num; u32 hop_num; - u16 op = 0xff; + int op; if (!hns_roce_check_whether_mhop(hr_dev, table->type)) return 0; @@ -2944,38 +2981,9 @@ static int hns_roce_v2_set_hem(struct hns_roce_dev *hr_dev, hem_idx = i; } - switch (table->type) { - case HEM_TYPE_QPC: - op = HNS_ROCE_CMD_WRITE_QPC_BT0; - break; - case HEM_TYPE_MTPT: - op = HNS_ROCE_CMD_WRITE_MPT_BT0; - break; - case HEM_TYPE_CQC: - op = HNS_ROCE_CMD_WRITE_CQC_BT0; - break; - case HEM_TYPE_SRQC: - op = HNS_ROCE_CMD_WRITE_SRQC_BT0; - break; - case HEM_TYPE_SCCC: - op = HNS_ROCE_CMD_WRITE_SCCC_BT0; - break; - case HEM_TYPE_QPC_TIMER: - op = HNS_ROCE_CMD_WRITE_QPC_TIMER_BT0; - break; - case HEM_TYPE_CQC_TIMER: - op = HNS_ROCE_CMD_WRITE_CQC_TIMER_BT0; - break; - default: - dev_warn(dev, "Table %d not to be written by mailbox!\n", - table->type); + op = get_op_for_set_hem(hr_dev, table->type, step_idx); + if (op == -EINVAL) return 0; - } - - if (table->type == HEM_TYPE_SCCC && step_idx) - return 0; - - op += step_idx; mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); if (IS_ERR(mailbox)) @@ -3118,6 +3126,43 @@ static void set_access_flags(struct hns_roce_qp *hr_qp, roce_set_bit(qpc_mask->byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S, 0); } +static void set_qpc_wqe_cnt(struct hns_roce_qp *hr_qp, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) +{ + if (hr_qp->ibqp.qp_type == IB_QPT_GSI) + roce_set_field(context->byte_4_sqpn_tst, + V2_QPC_BYTE_4_SGE_SHIFT_M, + V2_QPC_BYTE_4_SGE_SHIFT_S, + ilog2((unsigned int)hr_qp->sge.sge_cnt)); + else + roce_set_field(context->byte_4_sqpn_tst, + V2_QPC_BYTE_4_SGE_SHIFT_M, + V2_QPC_BYTE_4_SGE_SHIFT_S, + hr_qp->sq.max_gs > + HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE ? + ilog2((unsigned int)hr_qp->sge.sge_cnt) : 0); + + roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_SGE_SHIFT_M, + V2_QPC_BYTE_4_SGE_SHIFT_S, 0); + + roce_set_field(context->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, + ilog2((unsigned int)hr_qp->sq.wqe_cnt)); + roce_set_field(qpc_mask->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, 0); + + roce_set_field(context->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, + (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || + hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || + hr_qp->ibqp.srq) ? 0 : + ilog2((unsigned int)hr_qp->rq.wqe_cnt)); + + roce_set_field(qpc_mask->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, 0); +} + static void modify_qp_reset_to_init(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, @@ -3138,21 +3183,6 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp, roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_TST_M, V2_QPC_BYTE_4_TST_S, 0); - if (ibqp->qp_type == IB_QPT_GSI) - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - ilog2((unsigned int)hr_qp->sge.sge_cnt)); - else - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - hr_qp->sq.max_gs > 2 ? - ilog2((unsigned int)hr_qp->sge.sge_cnt) : 0); - - roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, 0); - roce_set_field(context->byte_4_sqpn_tst, V2_QPC_BYTE_4_SQPN_M, V2_QPC_BYTE_4_SQPN_S, hr_qp->qpn); roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_SQPN_M, @@ -3168,19 +3198,7 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp, roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQWS_M, V2_QPC_BYTE_20_RQWS_S, 0); - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, - ilog2((unsigned int)hr_qp->sq.wqe_cnt)); - roce_set_field(qpc_mask->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, 0); - - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, - (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || - hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || ibqp->srq) ? 0 : - ilog2((unsigned int)hr_qp->rq.wqe_cnt)); - roce_set_field(qpc_mask->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, 0); + set_qpc_wqe_cnt(hr_qp, context, qpc_mask); /* No VLAN need to set 0xFFF */ roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_VLAN_ID_M, @@ -3456,22 +3474,6 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp, roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_TST_M, V2_QPC_BYTE_4_TST_S, 0); - if (ibqp->qp_type == IB_QPT_GSI) - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - ilog2((unsigned int)hr_qp->sge.sge_cnt)); - else - roce_set_field(context->byte_4_sqpn_tst, - V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, - hr_qp->sq.max_gs > - HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE ? - ilog2((unsigned int)hr_qp->sge.sge_cnt) : 0); - - roce_set_field(qpc_mask->byte_4_sqpn_tst, V2_QPC_BYTE_4_SGE_SHIFT_M, - V2_QPC_BYTE_4_SGE_SHIFT_S, 0); - if (attr_mask & IB_QP_ACCESS_FLAGS) { roce_set_bit(context->byte_76_srqn_op_en, V2_QPC_BYTE_76_RRE_S, !!(attr->qp_access_flags & IB_ACCESS_REMOTE_READ)); @@ -3506,20 +3508,6 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp, 0); } - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, - ilog2((unsigned int)hr_qp->sq.wqe_cnt)); - roce_set_field(qpc_mask->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SQ_SHIFT_M, V2_QPC_BYTE_20_SQ_SHIFT_S, 0); - - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, - (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || - hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || ibqp->srq) ? 0 : - ilog2((unsigned int)hr_qp->rq.wqe_cnt)); - roce_set_field(qpc_mask->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, 0); - roce_set_field(context->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_PD_M, V2_QPC_BYTE_16_PD_S, to_hr_pd(ibqp->pd)->pdn); roce_set_field(qpc_mask->byte_16_buf_ba_pg_sz, V2_QPC_BYTE_16_PD_M, @@ -3974,30 +3962,119 @@ static inline bool hns_roce_v2_check_qp_stat(enum ib_qp_state cur_state, } -static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, - const struct ib_qp_attr *attr, - int attr_mask, enum ib_qp_state cur_state, - enum ib_qp_state new_state) +static int hns_roce_v2_set_path(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, + int attr_mask, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) { + const struct ib_global_route *grh = rdma_ah_read_grh(&attr->ah_attr); struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); - struct hns_roce_v2_qp_context *context; - struct hns_roce_v2_qp_context *qpc_mask; - struct device *dev = hr_dev->dev; - int ret = -EINVAL; + const struct ib_gid_attr *gid_attr = NULL; + int is_roce_protocol; + bool is_udp = false; + u16 vlan = 0xffff; + u8 ib_port; + u8 hr_port; + int ret; - context = kcalloc(2, sizeof(*context), GFP_ATOMIC); - if (!context) - return -ENOMEM; + ib_port = (attr_mask & IB_QP_PORT) ? attr->port_num : hr_qp->port + 1; + hr_port = ib_port - 1; + is_roce_protocol = rdma_cap_eth_ah(&hr_dev->ib_dev, ib_port) && + rdma_ah_get_ah_flags(&attr->ah_attr) & IB_AH_GRH; + + if (is_roce_protocol) { + gid_attr = attr->ah_attr.grh.sgid_attr; + ret = rdma_read_gid_l2_fields(gid_attr, &vlan, NULL); + if (ret) + return ret; + + if (gid_attr) + is_udp = (gid_attr->gid_type == + IB_GID_TYPE_ROCE_UDP_ENCAP); + } + + if (vlan < VLAN_CFI_MASK) { + roce_set_bit(context->byte_76_srqn_op_en, + V2_QPC_BYTE_76_RQ_VLAN_EN_S, 1); + roce_set_bit(qpc_mask->byte_76_srqn_op_en, + V2_QPC_BYTE_76_RQ_VLAN_EN_S, 0); + roce_set_bit(context->byte_168_irrl_idx, + V2_QPC_BYTE_168_SQ_VLAN_EN_S, 1); + roce_set_bit(qpc_mask->byte_168_irrl_idx, + V2_QPC_BYTE_168_SQ_VLAN_EN_S, 0); + } + + roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_VLAN_ID_M, + V2_QPC_BYTE_24_VLAN_ID_S, vlan); + roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_VLAN_ID_M, + V2_QPC_BYTE_24_VLAN_ID_S, 0); + + if (grh->sgid_index >= hr_dev->caps.gid_table_len[hr_port]) { + dev_err(hr_dev->dev, "sgid_index(%u) too large. max is %d\n", + grh->sgid_index, hr_dev->caps.gid_table_len[hr_port]); + return -EINVAL; + } + + if (attr->ah_attr.type != RDMA_AH_ATTR_TYPE_ROCE) { + dev_err(hr_dev->dev, "ah attr is not RDMA roce type\n"); + return -EINVAL; + } + + roce_set_field(context->byte_52_udpspn_dmac, V2_QPC_BYTE_52_UDPSPN_M, + V2_QPC_BYTE_52_UDPSPN_S, + is_udp ? 0x12b7 : 0); + + roce_set_field(qpc_mask->byte_52_udpspn_dmac, V2_QPC_BYTE_52_UDPSPN_M, + V2_QPC_BYTE_52_UDPSPN_S, 0); + + roce_set_field(context->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S, + grh->sgid_index); + + roce_set_field(qpc_mask->byte_20_smac_sgid_idx, + V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S, 0); + + roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_HOP_LIMIT_M, + V2_QPC_BYTE_24_HOP_LIMIT_S, grh->hop_limit); + roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_HOP_LIMIT_M, + V2_QPC_BYTE_24_HOP_LIMIT_S, 0); + + if (hr_dev->pci_dev->revision == 0x21 && is_udp) + roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, + V2_QPC_BYTE_24_TC_S, grh->traffic_class >> 2); + else + roce_set_field(context->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, + V2_QPC_BYTE_24_TC_S, grh->traffic_class); + roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, + V2_QPC_BYTE_24_TC_S, 0); + roce_set_field(context->byte_28_at_fl, V2_QPC_BYTE_28_FL_M, + V2_QPC_BYTE_28_FL_S, grh->flow_label); + roce_set_field(qpc_mask->byte_28_at_fl, V2_QPC_BYTE_28_FL_M, + V2_QPC_BYTE_28_FL_S, 0); + memcpy(context->dgid, grh->dgid.raw, sizeof(grh->dgid.raw)); + memset(qpc_mask->dgid, 0, sizeof(grh->dgid.raw)); + roce_set_field(context->byte_28_at_fl, V2_QPC_BYTE_28_SL_M, + V2_QPC_BYTE_28_SL_S, rdma_ah_get_sl(&attr->ah_attr)); + roce_set_field(qpc_mask->byte_28_at_fl, V2_QPC_BYTE_28_SL_M, + V2_QPC_BYTE_28_SL_S, 0); + hr_qp->sl = rdma_ah_get_sl(&attr->ah_attr); + + return 0; +} + +static int hns_roce_v2_set_abs_fields(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, + int attr_mask, + enum ib_qp_state cur_state, + enum ib_qp_state new_state, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + int ret = 0; - qpc_mask = context + 1; - /* - * In v2 engine, software pass context and context mask to hardware - * when modifying qp. If software need modify some fields in context, - * we should set all bits of the relevant fields in context mask to - * 0 at the same time, else set them to 0x1. - */ - memset(qpc_mask, 0xff, sizeof(*qpc_mask)); if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { memset(qpc_mask, 0, sizeof(*qpc_mask)); modify_qp_reset_to_init(ibqp, attr, attr_mask, context, @@ -4019,134 +4096,30 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, /* Nothing */ ; } else { - dev_err(dev, "Illegal state for QP!\n"); + dev_err(hr_dev->dev, "Illegal state for QP!\n"); ret = -EINVAL; goto out; } - /* When QP state is err, SQ and RQ WQE should be flushed */ - if (new_state == IB_QPS_ERR) { - roce_set_field(context->byte_160_sq_ci_pi, - V2_QPC_BYTE_160_SQ_PRODUCER_IDX_M, - V2_QPC_BYTE_160_SQ_PRODUCER_IDX_S, - hr_qp->sq.head); - roce_set_field(qpc_mask->byte_160_sq_ci_pi, - V2_QPC_BYTE_160_SQ_PRODUCER_IDX_M, - V2_QPC_BYTE_160_SQ_PRODUCER_IDX_S, 0); +out: + return ret; +} - if (!ibqp->srq) { - roce_set_field(context->byte_84_rq_ci_pi, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, - hr_qp->rq.head); - roce_set_field(qpc_mask->byte_84_rq_ci_pi, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, - V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, 0); - } - } +static int hns_roce_v2_set_opt_fields(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, + int attr_mask, + struct hns_roce_v2_qp_context *context, + struct hns_roce_v2_qp_context *qpc_mask) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); + int ret = 0; if (attr_mask & IB_QP_AV) { - const struct ib_global_route *grh = - rdma_ah_read_grh(&attr->ah_attr); - const struct ib_gid_attr *gid_attr = NULL; - int is_roce_protocol; - u16 vlan = 0xffff; - u8 ib_port; - u8 hr_port; - - ib_port = (attr_mask & IB_QP_PORT) ? attr->port_num : - hr_qp->port + 1; - hr_port = ib_port - 1; - is_roce_protocol = rdma_cap_eth_ah(&hr_dev->ib_dev, ib_port) && - rdma_ah_get_ah_flags(&attr->ah_attr) & IB_AH_GRH; - - if (is_roce_protocol) { - gid_attr = attr->ah_attr.grh.sgid_attr; - ret = rdma_read_gid_l2_fields(gid_attr, &vlan, NULL); - if (ret) - goto out; - } - - if (vlan < VLAN_CFI_MASK) { - roce_set_bit(context->byte_76_srqn_op_en, - V2_QPC_BYTE_76_RQ_VLAN_EN_S, 1); - roce_set_bit(qpc_mask->byte_76_srqn_op_en, - V2_QPC_BYTE_76_RQ_VLAN_EN_S, 0); - roce_set_bit(context->byte_168_irrl_idx, - V2_QPC_BYTE_168_SQ_VLAN_EN_S, 1); - roce_set_bit(qpc_mask->byte_168_irrl_idx, - V2_QPC_BYTE_168_SQ_VLAN_EN_S, 0); - } - - roce_set_field(context->byte_24_mtu_tc, - V2_QPC_BYTE_24_VLAN_ID_M, - V2_QPC_BYTE_24_VLAN_ID_S, vlan); - roce_set_field(qpc_mask->byte_24_mtu_tc, - V2_QPC_BYTE_24_VLAN_ID_M, - V2_QPC_BYTE_24_VLAN_ID_S, 0); - - if (grh->sgid_index >= hr_dev->caps.gid_table_len[hr_port]) { - dev_err(hr_dev->dev, - "sgid_index(%u) too large. max is %d\n", - grh->sgid_index, - hr_dev->caps.gid_table_len[hr_port]); - ret = -EINVAL; - goto out; - } - - if (attr->ah_attr.type != RDMA_AH_ATTR_TYPE_ROCE) { - dev_err(hr_dev->dev, "ah attr is not RDMA roce type\n"); - ret = -EINVAL; - goto out; - } - - roce_set_field(context->byte_52_udpspn_dmac, - V2_QPC_BYTE_52_UDPSPN_M, V2_QPC_BYTE_52_UDPSPN_S, - (gid_attr->gid_type != IB_GID_TYPE_ROCE_UDP_ENCAP) ? - 0 : 0x12b7); - - roce_set_field(qpc_mask->byte_52_udpspn_dmac, - V2_QPC_BYTE_52_UDPSPN_M, - V2_QPC_BYTE_52_UDPSPN_S, 0); - - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SGID_IDX_M, - V2_QPC_BYTE_20_SGID_IDX_S, grh->sgid_index); - - roce_set_field(qpc_mask->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SGID_IDX_M, - V2_QPC_BYTE_20_SGID_IDX_S, 0); - - roce_set_field(context->byte_24_mtu_tc, - V2_QPC_BYTE_24_HOP_LIMIT_M, - V2_QPC_BYTE_24_HOP_LIMIT_S, grh->hop_limit); - roce_set_field(qpc_mask->byte_24_mtu_tc, - V2_QPC_BYTE_24_HOP_LIMIT_M, - V2_QPC_BYTE_24_HOP_LIMIT_S, 0); - - if (hr_dev->pci_dev->revision == 0x21 && - gid_attr->gid_type == IB_GID_TYPE_ROCE_UDP_ENCAP) - roce_set_field(context->byte_24_mtu_tc, - V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S, - grh->traffic_class >> 2); - else - roce_set_field(context->byte_24_mtu_tc, - V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S, - grh->traffic_class); - roce_set_field(qpc_mask->byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, - V2_QPC_BYTE_24_TC_S, 0); - roce_set_field(context->byte_28_at_fl, V2_QPC_BYTE_28_FL_M, - V2_QPC_BYTE_28_FL_S, grh->flow_label); - roce_set_field(qpc_mask->byte_28_at_fl, V2_QPC_BYTE_28_FL_M, - V2_QPC_BYTE_28_FL_S, 0); - memcpy(context->dgid, grh->dgid.raw, sizeof(grh->dgid.raw)); - memset(qpc_mask->dgid, 0, sizeof(grh->dgid.raw)); - roce_set_field(context->byte_28_at_fl, V2_QPC_BYTE_28_SL_M, - V2_QPC_BYTE_28_SL_S, - rdma_ah_get_sl(&attr->ah_attr)); - roce_set_field(qpc_mask->byte_28_at_fl, V2_QPC_BYTE_28_SL_M, - V2_QPC_BYTE_28_SL_S, 0); - hr_qp->sl = rdma_ah_get_sl(&attr->ah_attr); + ret = hns_roce_v2_set_path(ibqp, attr, attr_mask, context, + qpc_mask); + if (ret) + return ret; } if (attr_mask & IB_QP_TIMEOUT) { @@ -4158,7 +4131,8 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, V2_QPC_BYTE_28_AT_M, V2_QPC_BYTE_28_AT_S, 0); } else { - dev_warn(dev, "Local ACK timeout shall be 0 to 30.\n"); + dev_warn(hr_dev->dev, + "Local ACK timeout shall be 0 to 30.\n"); } } @@ -4196,6 +4170,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, V2_QPC_BYTE_244_RNR_CNT_S, 0); } + /* RC&UC&UD required attr */ if (attr_mask & IB_QP_SQ_PSN) { roce_set_field(context->byte_172_sq_psn, V2_QPC_BYTE_172_SQ_CUR_PSN_M, @@ -4295,6 +4270,80 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, hr_qp->qkey = attr->qkey; } + return ret; +} + +static void hns_roce_v2_record_opt_fields(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, + int attr_mask) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); + + if (attr_mask & IB_QP_ACCESS_FLAGS) + hr_qp->atomic_rd_en = attr->qp_access_flags; + + if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) + hr_qp->resp_depth = attr->max_dest_rd_atomic; + if (attr_mask & IB_QP_PORT) { + hr_qp->port = attr->port_num - 1; + hr_qp->phy_port = hr_dev->iboe.phy_port[hr_qp->port]; + } +} + +static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, + const struct ib_qp_attr *attr, + int attr_mask, enum ib_qp_state cur_state, + enum ib_qp_state new_state) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); + struct hns_roce_v2_qp_context ctx[2]; + struct hns_roce_v2_qp_context *context = ctx; + struct hns_roce_v2_qp_context *qpc_mask = ctx + 1; + struct device *dev = hr_dev->dev; + int ret; + + /* + * In v2 engine, software pass context and context mask to hardware + * when modifying qp. If software need modify some fields in context, + * we should set all bits of the relevant fields in context mask to + * 0 at the same time, else set them to 0x1. + */ + memset(context, 0, sizeof(*context)); + memset(qpc_mask, 0xff, sizeof(*qpc_mask)); + ret = hns_roce_v2_set_abs_fields(ibqp, attr, attr_mask, cur_state, + new_state, context, qpc_mask); + if (ret) + goto out; + + /* When QP state is err, SQ and RQ WQE should be flushed */ + if (new_state == IB_QPS_ERR) { + roce_set_field(context->byte_160_sq_ci_pi, + V2_QPC_BYTE_160_SQ_PRODUCER_IDX_M, + V2_QPC_BYTE_160_SQ_PRODUCER_IDX_S, + hr_qp->sq.head); + roce_set_field(qpc_mask->byte_160_sq_ci_pi, + V2_QPC_BYTE_160_SQ_PRODUCER_IDX_M, + V2_QPC_BYTE_160_SQ_PRODUCER_IDX_S, 0); + + if (!ibqp->srq) { + roce_set_field(context->byte_84_rq_ci_pi, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, + hr_qp->rq.head); + roce_set_field(qpc_mask->byte_84_rq_ci_pi, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_M, + V2_QPC_BYTE_84_RQ_PRODUCER_IDX_S, 0); + } + } + + /* Configure the optional fields */ + ret = hns_roce_v2_set_opt_fields(ibqp, attr, attr_mask, context, + qpc_mask); + if (ret) + goto out; + roce_set_bit(context->byte_108_rx_reqepsn, V2_QPC_BYTE_108_INV_CREDIT_S, ibqp->srq ? 1 : 0); roce_set_bit(qpc_mask->byte_108_rx_reqepsn, @@ -4307,8 +4356,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, V2_QPC_BYTE_60_QP_ST_S, 0); /* SW pass context to HW */ - ret = hns_roce_v2_qp_modify(hr_dev, cur_state, new_state, - context, hr_qp); + ret = hns_roce_v2_qp_modify(hr_dev, cur_state, new_state, ctx, hr_qp); if (ret) { dev_err(dev, "hns_roce_qp_modify failed(%d)\n", ret); goto out; @@ -4316,15 +4364,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, hr_qp->state = new_state; - if (attr_mask & IB_QP_ACCESS_FLAGS) - hr_qp->atomic_rd_en = attr->qp_access_flags; - - if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC) - hr_qp->resp_depth = attr->max_dest_rd_atomic; - if (attr_mask & IB_QP_PORT) { - hr_qp->port = attr->port_num - 1; - hr_qp->phy_port = hr_dev->iboe.phy_port[hr_qp->port]; - } + hns_roce_v2_record_opt_fields(ibqp, attr, attr_mask); if (new_state == IB_QPS_RESET && !ibqp->uobject) { hns_roce_v2_cq_clean(to_hr_cq(ibqp->recv_cq), hr_qp->qpn, @@ -4344,7 +4384,6 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, } out: - kfree(context); return ret; } @@ -4395,16 +4434,12 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, { struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); - struct hns_roce_v2_qp_context *context; + struct hns_roce_v2_qp_context context = {}; struct device *dev = hr_dev->dev; int tmp_qp_state; int state; int ret; - context = kzalloc(sizeof(*context), GFP_KERNEL); - if (!context) - return -ENOMEM; - memset(qp_attr, 0, sizeof(*qp_attr)); memset(qp_init_attr, 0, sizeof(*qp_init_attr)); @@ -4416,14 +4451,14 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, goto done; } - ret = hns_roce_v2_query_qpc(hr_dev, hr_qp, context); + ret = hns_roce_v2_query_qpc(hr_dev, hr_qp, &context); if (ret) { dev_err(dev, "query qpc error\n"); ret = -EINVAL; goto out; } - state = roce_get_field(context->byte_60_qpst_tempid, + state = roce_get_field(context.byte_60_qpst_tempid, V2_QPC_BYTE_60_QP_ST_M, V2_QPC_BYTE_60_QP_ST_S); tmp_qp_state = to_ib_qp_st((enum hns_roce_v2_qp_state)state); if (tmp_qp_state == -1) { @@ -4433,7 +4468,7 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, } hr_qp->state = (u8)tmp_qp_state; qp_attr->qp_state = (enum ib_qp_state)hr_qp->state; - qp_attr->path_mtu = (enum ib_mtu)roce_get_field(context->byte_24_mtu_tc, + qp_attr->path_mtu = (enum ib_mtu)roce_get_field(context.byte_24_mtu_tc, V2_QPC_BYTE_24_MTU_M, V2_QPC_BYTE_24_MTU_S); qp_attr->path_mig_state = IB_MIG_ARMED; @@ -4441,20 +4476,20 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, if (hr_qp->ibqp.qp_type == IB_QPT_UD) qp_attr->qkey = V2_QKEY_VAL; - qp_attr->rq_psn = roce_get_field(context->byte_108_rx_reqepsn, + qp_attr->rq_psn = roce_get_field(context.byte_108_rx_reqepsn, V2_QPC_BYTE_108_RX_REQ_EPSN_M, V2_QPC_BYTE_108_RX_REQ_EPSN_S); - qp_attr->sq_psn = (u32)roce_get_field(context->byte_172_sq_psn, + qp_attr->sq_psn = (u32)roce_get_field(context.byte_172_sq_psn, V2_QPC_BYTE_172_SQ_CUR_PSN_M, V2_QPC_BYTE_172_SQ_CUR_PSN_S); - qp_attr->dest_qp_num = (u8)roce_get_field(context->byte_56_dqpn_err, + qp_attr->dest_qp_num = (u8)roce_get_field(context.byte_56_dqpn_err, V2_QPC_BYTE_56_DQPN_M, V2_QPC_BYTE_56_DQPN_S); - qp_attr->qp_access_flags = ((roce_get_bit(context->byte_76_srqn_op_en, + qp_attr->qp_access_flags = ((roce_get_bit(context.byte_76_srqn_op_en, V2_QPC_BYTE_76_RRE_S)) << V2_QP_RWE_S) | - ((roce_get_bit(context->byte_76_srqn_op_en, + ((roce_get_bit(context.byte_76_srqn_op_en, V2_QPC_BYTE_76_RWE_S)) << V2_QP_RRE_S) | - ((roce_get_bit(context->byte_76_srqn_op_en, + ((roce_get_bit(context.byte_76_srqn_op_en, V2_QPC_BYTE_76_ATE_S)) << V2_QP_ATE_S); if (hr_qp->ibqp.qp_type == IB_QPT_RC || @@ -4463,43 +4498,43 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, rdma_ah_retrieve_grh(&qp_attr->ah_attr); rdma_ah_set_sl(&qp_attr->ah_attr, - roce_get_field(context->byte_28_at_fl, + roce_get_field(context.byte_28_at_fl, V2_QPC_BYTE_28_SL_M, V2_QPC_BYTE_28_SL_S)); - grh->flow_label = roce_get_field(context->byte_28_at_fl, + grh->flow_label = roce_get_field(context.byte_28_at_fl, V2_QPC_BYTE_28_FL_M, V2_QPC_BYTE_28_FL_S); - grh->sgid_index = roce_get_field(context->byte_20_smac_sgid_idx, + grh->sgid_index = roce_get_field(context.byte_20_smac_sgid_idx, V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S); - grh->hop_limit = roce_get_field(context->byte_24_mtu_tc, + grh->hop_limit = roce_get_field(context.byte_24_mtu_tc, V2_QPC_BYTE_24_HOP_LIMIT_M, V2_QPC_BYTE_24_HOP_LIMIT_S); - grh->traffic_class = roce_get_field(context->byte_24_mtu_tc, + grh->traffic_class = roce_get_field(context.byte_24_mtu_tc, V2_QPC_BYTE_24_TC_M, V2_QPC_BYTE_24_TC_S); - memcpy(grh->dgid.raw, context->dgid, sizeof(grh->dgid.raw)); + memcpy(grh->dgid.raw, context.dgid, sizeof(grh->dgid.raw)); } qp_attr->port_num = hr_qp->port + 1; qp_attr->sq_draining = 0; - qp_attr->max_rd_atomic = 1 << roce_get_field(context->byte_208_irrl, + qp_attr->max_rd_atomic = 1 << roce_get_field(context.byte_208_irrl, V2_QPC_BYTE_208_SR_MAX_M, V2_QPC_BYTE_208_SR_MAX_S); - qp_attr->max_dest_rd_atomic = 1 << roce_get_field(context->byte_140_raq, + qp_attr->max_dest_rd_atomic = 1 << roce_get_field(context.byte_140_raq, V2_QPC_BYTE_140_RR_MAX_M, V2_QPC_BYTE_140_RR_MAX_S); - qp_attr->min_rnr_timer = (u8)roce_get_field(context->byte_80_rnr_rx_cqn, + qp_attr->min_rnr_timer = (u8)roce_get_field(context.byte_80_rnr_rx_cqn, V2_QPC_BYTE_80_MIN_RNR_TIME_M, V2_QPC_BYTE_80_MIN_RNR_TIME_S); - qp_attr->timeout = (u8)roce_get_field(context->byte_28_at_fl, + qp_attr->timeout = (u8)roce_get_field(context.byte_28_at_fl, V2_QPC_BYTE_28_AT_M, V2_QPC_BYTE_28_AT_S); - qp_attr->retry_cnt = roce_get_field(context->byte_212_lsn, + qp_attr->retry_cnt = roce_get_field(context.byte_212_lsn, V2_QPC_BYTE_212_RETRY_CNT_M, V2_QPC_BYTE_212_RETRY_CNT_S); - qp_attr->rnr_retry = context->rq_rnr_timer; + qp_attr->rnr_retry = context.rq_rnr_timer; done: qp_attr->cur_qp_state = qp_attr->qp_state; @@ -4518,7 +4553,6 @@ done: out: mutex_unlock(&hr_qp->mutex); - kfree(context); return ret; } @@ -4527,7 +4561,7 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev, struct ib_udata *udata) { struct hns_roce_cq *send_cq, *recv_cq; - struct device *dev = hr_dev->dev; + struct ib_device *ibdev = &hr_dev->ib_dev; int ret; if (hr_qp->ibqp.qp_type == IB_QPT_RC && hr_qp->state != IB_QPS_RESET) { @@ -4535,8 +4569,7 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev, ret = hns_roce_v2_modify_qp(&hr_qp->ibqp, NULL, 0, hr_qp->state, IB_QPS_RESET); if (ret) { - dev_err(dev, "modify QP %06lx to ERR failed.\n", - hr_qp->qpn); + ibdev_err(ibdev, "modify QP to Reset failed.\n"); return ret; } } @@ -4605,7 +4638,8 @@ static int hns_roce_v2_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) ret = hns_roce_v2_destroy_qp_common(hr_dev, hr_qp, udata); if (ret) { - dev_err(hr_dev->dev, "Destroy qp failed(%d)\n", ret); + ibdev_err(&hr_dev->ib_dev, "Destroy qp 0x%06lx failed(%d)\n", + hr_qp->qpn, ret); return ret; } @@ -4904,7 +4938,7 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq) { struct device *dev = hr_dev->dev; - struct hns_roce_aeqe *aeqe; + struct hns_roce_aeqe *aeqe = next_aeqe_sw_v2(eq); int aeqe_found = 0; int event_type; int sub_type; @@ -4912,8 +4946,7 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, u32 qpn; u32 cqn; - while ((aeqe = next_aeqe_sw_v2(eq))) { - + while (aeqe) { /* Make sure we read AEQ entry after we have checked the * ownership bit */ @@ -4977,11 +5010,12 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, ++eq->cons_index; aeqe_found = 1; - if (eq->cons_index > (2 * eq->entries - 1)) { - dev_warn(dev, "cons_index overflow, set back to 0.\n"); + if (eq->cons_index > (2 * eq->entries - 1)) eq->cons_index = 0; - } + hns_roce_v2_init_irq_work(hr_dev, eq, qpn, cqn); + + aeqe = next_aeqe_sw_v2(eq); } set_eq_cons_index_v2(eq); @@ -5034,12 +5068,11 @@ static int hns_roce_v2_ceq_int(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq) { struct device *dev = hr_dev->dev; - struct hns_roce_ceqe *ceqe; + struct hns_roce_ceqe *ceqe = next_ceqe_sw_v2(eq); int ceqe_found = 0; u32 cqn; - while ((ceqe = next_ceqe_sw_v2(eq))) { - + while (ceqe) { /* Make sure we read CEQ entry after we have checked the * ownership bit */ @@ -5058,6 +5091,8 @@ static int hns_roce_v2_ceq_int(struct hns_roce_dev *hr_dev, dev_warn(dev, "cons_index overflow, set back to 0.\n"); eq->cons_index = 0; } + + ceqe = next_ceqe_sw_v2(eq); } set_eq_cons_index_v2(eq); @@ -5202,14 +5237,12 @@ static void hns_roce_mhop_free_eq(struct hns_roce_dev *hr_dev, buf_chk_sz = 1 << (hr_dev->caps.eqe_buf_pg_sz + PAGE_SHIFT); bt_chk_sz = 1 << (hr_dev->caps.eqe_ba_pg_sz + PAGE_SHIFT); - /* hop_num = 0 */ if (mhop_num == HNS_ROCE_HOP_NUM_0) { dma_free_coherent(dev, (unsigned int)(eq->entries * eq->eqe_size), eq->bt_l0, eq->l0_dma); return; } - /* hop_num = 1 or hop = 2 */ dma_free_coherent(dev, bt_chk_sz, eq->bt_l0, eq->l0_dma); if (mhop_num == 1) { for (i = 0; i < eq->l0_last_num; i++) { @@ -5449,7 +5482,6 @@ static int hns_roce_mhop_alloc_eq(struct hns_roce_dev *hr_dev, buf_chk_sz); bt_num = DIV_ROUND_UP(ba_num, bt_chk_sz / BA_BYTE_LEN); - /* hop_num = 0 */ if (mhop_num == HNS_ROCE_HOP_NUM_0) { if (eq->entries > buf_chk_sz / eq->eqe_size) { dev_err(dev, "eq entries %d is larger than buf_pg_sz!", @@ -5515,7 +5547,8 @@ static int hns_roce_mhop_alloc_eq(struct hns_roce_dev *hr_dev, break; } eq->cur_eqe_ba = eq->buf_dma[0]; - eq->nxt_eqe_ba = eq->buf_dma[1]; + if (ba_num > 1) + eq->nxt_eqe_ba = eq->buf_dma[1]; } else if (mhop_num == 2) { /* alloc L1 BT and buf */ @@ -5556,7 +5589,8 @@ static int hns_roce_mhop_alloc_eq(struct hns_roce_dev *hr_dev, break; } eq->cur_eqe_ba = eq->buf_dma[0]; - eq->nxt_eqe_ba = eq->buf_dma[1]; + if (ba_num > 1) + eq->nxt_eqe_ba = eq->buf_dma[1]; } eq->l0_last_num = i + 1; @@ -5699,6 +5733,95 @@ free_cmd_mbox: return ret; } +static int __hns_roce_request_irq(struct hns_roce_dev *hr_dev, int irq_num, + int comp_num, int aeq_num, int other_num) +{ + struct hns_roce_eq_table *eq_table = &hr_dev->eq_table; + int i, j; + int ret; + + for (i = 0; i < irq_num; i++) { + hr_dev->irq_names[i] = kzalloc(HNS_ROCE_INT_NAME_LEN, + GFP_KERNEL); + if (!hr_dev->irq_names[i]) { + ret = -ENOMEM; + goto err_kzalloc_failed; + } + } + + /* irq contains: abnormal + AEQ + CEQ */ + for (j = 0; j < other_num; j++) + snprintf((char *)hr_dev->irq_names[j], + HNS_ROCE_INT_NAME_LEN, "hns-abn-%d", j); + + for (j = other_num; j < (other_num + aeq_num); j++) + snprintf((char *)hr_dev->irq_names[j], + HNS_ROCE_INT_NAME_LEN, "hns-aeq-%d", + j - other_num); + + for (j = (other_num + aeq_num); j < irq_num; j++) + snprintf((char *)hr_dev->irq_names[j], + HNS_ROCE_INT_NAME_LEN, "hns-ceq-%d", + j - other_num - aeq_num); + + for (j = 0; j < irq_num; j++) { + if (j < other_num) + ret = request_irq(hr_dev->irq[j], + hns_roce_v2_msix_interrupt_abn, + 0, hr_dev->irq_names[j], hr_dev); + + else if (j < (other_num + comp_num)) + ret = request_irq(eq_table->eq[j - other_num].irq, + hns_roce_v2_msix_interrupt_eq, + 0, hr_dev->irq_names[j + aeq_num], + &eq_table->eq[j - other_num]); + else + ret = request_irq(eq_table->eq[j - other_num].irq, + hns_roce_v2_msix_interrupt_eq, + 0, hr_dev->irq_names[j - comp_num], + &eq_table->eq[j - other_num]); + if (ret) { + dev_err(hr_dev->dev, "Request irq error!\n"); + goto err_request_failed; + } + } + + return 0; + +err_request_failed: + for (j -= 1; j >= 0; j--) + if (j < other_num) + free_irq(hr_dev->irq[j], hr_dev); + else + free_irq(eq_table->eq[j - other_num].irq, + &eq_table->eq[j - other_num]); + +err_kzalloc_failed: + for (i -= 1; i >= 0; i--) + kfree(hr_dev->irq_names[i]); + + return ret; +} + +static void __hns_roce_free_irq(struct hns_roce_dev *hr_dev) +{ + int irq_num; + int eq_num; + int i; + + eq_num = hr_dev->caps.num_comp_vectors + hr_dev->caps.num_aeq_vectors; + irq_num = eq_num + hr_dev->caps.num_other_vectors; + + for (i = 0; i < hr_dev->caps.num_other_vectors; i++) + free_irq(hr_dev->irq[i], hr_dev); + + for (i = 0; i < eq_num; i++) + free_irq(hr_dev->eq_table.eq[i].irq, &hr_dev->eq_table.eq[i]); + + for (i = 0; i < irq_num; i++) + kfree(hr_dev->irq_names[i]); +} + static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) { struct hns_roce_eq_table *eq_table = &hr_dev->eq_table; @@ -5710,7 +5833,7 @@ static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) int other_num; int comp_num; int aeq_num; - int i, j, k; + int i; int ret; other_num = hr_dev->caps.num_other_vectors; @@ -5724,27 +5847,18 @@ static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) if (!eq_table->eq) return -ENOMEM; - for (i = 0; i < irq_num; i++) { - hr_dev->irq_names[i] = kzalloc(HNS_ROCE_INT_NAME_LEN, - GFP_KERNEL); - if (!hr_dev->irq_names[i]) { - ret = -ENOMEM; - goto err_failed_kzalloc; - } - } - /* create eq */ - for (j = 0; j < eq_num; j++) { - eq = &eq_table->eq[j]; + for (i = 0; i < eq_num; i++) { + eq = &eq_table->eq[i]; eq->hr_dev = hr_dev; - eq->eqn = j; - if (j < comp_num) { + eq->eqn = i; + if (i < comp_num) { /* CEQ */ eq_cmd = HNS_ROCE_CMD_CREATE_CEQC; eq->type_flag = HNS_ROCE_CEQ; eq->entries = hr_dev->caps.ceqe_depth; eq->eqe_size = HNS_ROCE_CEQ_ENTRY_SIZE; - eq->irq = hr_dev->irq[j + other_num + aeq_num]; + eq->irq = hr_dev->irq[i + other_num + aeq_num]; eq->eq_max_cnt = HNS_ROCE_CEQ_DEFAULT_BURST_NUM; eq->eq_period = HNS_ROCE_CEQ_DEFAULT_INTERVAL; } else { @@ -5753,7 +5867,7 @@ static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) eq->type_flag = HNS_ROCE_AEQ; eq->entries = hr_dev->caps.aeqe_depth; eq->eqe_size = HNS_ROCE_AEQ_ENTRY_SIZE; - eq->irq = hr_dev->irq[j - comp_num + other_num]; + eq->irq = hr_dev->irq[i - comp_num + other_num]; eq->eq_max_cnt = HNS_ROCE_AEQ_DEFAULT_BURST_NUM; eq->eq_period = HNS_ROCE_AEQ_DEFAULT_INTERVAL; } @@ -5768,40 +5882,11 @@ static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) /* enable irq */ hns_roce_v2_int_mask_enable(hr_dev, eq_num, EQ_ENABLE); - /* irq contains: abnormal + AEQ + CEQ*/ - for (k = 0; k < irq_num; k++) - if (k < other_num) - snprintf((char *)hr_dev->irq_names[k], - HNS_ROCE_INT_NAME_LEN, "hns-abn-%d", k); - else if (k < (other_num + aeq_num)) - snprintf((char *)hr_dev->irq_names[k], - HNS_ROCE_INT_NAME_LEN, "hns-aeq-%d", - k - other_num); - else - snprintf((char *)hr_dev->irq_names[k], - HNS_ROCE_INT_NAME_LEN, "hns-ceq-%d", - k - other_num - aeq_num); - - for (k = 0; k < irq_num; k++) { - if (k < other_num) - ret = request_irq(hr_dev->irq[k], - hns_roce_v2_msix_interrupt_abn, - 0, hr_dev->irq_names[k], hr_dev); - - else if (k < (other_num + comp_num)) - ret = request_irq(eq_table->eq[k - other_num].irq, - hns_roce_v2_msix_interrupt_eq, - 0, hr_dev->irq_names[k + aeq_num], - &eq_table->eq[k - other_num]); - else - ret = request_irq(eq_table->eq[k - other_num].irq, - hns_roce_v2_msix_interrupt_eq, - 0, hr_dev->irq_names[k - comp_num], - &eq_table->eq[k - other_num]); - if (ret) { - dev_err(dev, "Request irq error!\n"); - goto err_request_irq_fail; - } + ret = __hns_roce_request_irq(hr_dev, irq_num, comp_num, + aeq_num, other_num); + if (ret) { + dev_err(dev, "Request irq failed.\n"); + goto err_request_irq_fail; } hr_dev->irq_workq = @@ -5809,26 +5894,20 @@ static int hns_roce_v2_init_eq_table(struct hns_roce_dev *hr_dev) if (!hr_dev->irq_workq) { dev_err(dev, "Create irq workqueue failed!\n"); ret = -ENOMEM; - goto err_request_irq_fail; + goto err_create_wq_fail; } return 0; +err_create_wq_fail: + __hns_roce_free_irq(hr_dev); + err_request_irq_fail: - for (k -= 1; k >= 0; k--) - if (k < other_num) - free_irq(hr_dev->irq[k], hr_dev); - else - free_irq(eq_table->eq[k - other_num].irq, - &eq_table->eq[k - other_num]); + hns_roce_v2_int_mask_enable(hr_dev, eq_num, EQ_DISABLE); err_create_eq_fail: - for (j -= 1; j >= 0; j--) - hns_roce_v2_free_eq(hr_dev, &eq_table->eq[j]); - -err_failed_kzalloc: for (i -= 1; i >= 0; i--) - kfree(hr_dev->irq_names[i]); + hns_roce_v2_free_eq(hr_dev, &eq_table->eq[i]); kfree(eq_table->eq); return ret; @@ -5837,30 +5916,22 @@ err_failed_kzalloc: static void hns_roce_v2_cleanup_eq_table(struct hns_roce_dev *hr_dev) { struct hns_roce_eq_table *eq_table = &hr_dev->eq_table; - int irq_num; int eq_num; int i; eq_num = hr_dev->caps.num_comp_vectors + hr_dev->caps.num_aeq_vectors; - irq_num = eq_num + hr_dev->caps.num_other_vectors; /* Disable irq */ hns_roce_v2_int_mask_enable(hr_dev, eq_num, EQ_DISABLE); - for (i = 0; i < hr_dev->caps.num_other_vectors; i++) - free_irq(hr_dev->irq[i], hr_dev); + __hns_roce_free_irq(hr_dev); for (i = 0; i < eq_num; i++) { hns_roce_v2_destroy_eqc(hr_dev, i); - free_irq(eq_table->eq[i].irq, &eq_table->eq[i]); - hns_roce_v2_free_eq(hr_dev, &eq_table->eq[i]); } - for (i = 0; i < irq_num; i++) - kfree(hr_dev->irq_names[i]); - kfree(eq_table->eq); flush_workqueue(hr_dev->irq_workq); diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h index 478f5a5b7aa1..58931b5399f8 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h @@ -126,8 +126,6 @@ #define HNS_ROCE_CMD_FLAG_ERR_INTR BIT(HNS_ROCE_CMD_FLAG_ERR_INTR_SHIFT) #define HNS_ROCE_CMQ_DESC_NUM_S 3 -#define HNS_ROCE_CMQ_EN_B 16 -#define HNS_ROCE_CMQ_ENABLE BIT(HNS_ROCE_CMQ_EN_B) #define HNS_ROCE_CMQ_SCC_CLR_DONE_CNT 5 diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c index 1e4ba48f5613..1b757cc924c3 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -262,7 +262,8 @@ static int hns_roce_query_port(struct ib_device *ib_dev, u8 port_num, props->state = (netif_running(net_dev) && netif_carrier_ok(net_dev)) ? IB_PORT_ACTIVE : IB_PORT_DOWN; props->phys_state = (props->state == IB_PORT_ACTIVE) ? - HNS_ROCE_PHY_LINKUP : HNS_ROCE_PHY_DISABLED; + IB_PORT_PHYS_STATE_LINK_UP : + IB_PORT_PHYS_STATE_DISABLED; spin_unlock_irqrestore(&hr_dev->iboe.lock, flags); diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c index 549e1a38dfe0..8157679021b9 100644 --- a/drivers/infiniband/hw/hns/hns_roce_mr.c +++ b/drivers/infiniband/hw/hns/hns_roce_mr.c @@ -347,155 +347,207 @@ static void hns_roce_loop_free(struct hns_roce_dev *hr_dev, mr->pbl_bt_l0 = NULL; mr->pbl_l0_dma_addr = 0; } +static int pbl_1hop_alloc(struct hns_roce_dev *hr_dev, int npages, + struct hns_roce_mr *mr, u32 pbl_bt_sz) +{ + struct device *dev = hr_dev->dev; -/* PBL multi hop addressing */ -static int hns_roce_mhop_alloc(struct hns_roce_dev *hr_dev, int npages, - struct hns_roce_mr *mr) + if (npages > pbl_bt_sz / 8) { + dev_err(dev, "npages %d is larger than buf_pg_sz!", + npages); + return -EINVAL; + } + mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, + &(mr->pbl_dma_addr), + GFP_KERNEL); + if (!mr->pbl_buf) + return -ENOMEM; + + mr->pbl_size = npages; + mr->pbl_ba = mr->pbl_dma_addr; + mr->pbl_hop_num = 1; + mr->pbl_ba_pg_sz = hr_dev->caps.pbl_ba_pg_sz; + mr->pbl_buf_pg_sz = hr_dev->caps.pbl_buf_pg_sz; + return 0; + +} + + +static int pbl_2hop_alloc(struct hns_roce_dev *hr_dev, int npages, + struct hns_roce_mr *mr, u32 pbl_bt_sz) { struct device *dev = hr_dev->dev; - int mr_alloc_done = 0; int npages_allocated; - int i = 0, j = 0; - u32 pbl_bt_sz; - u32 mhop_num; u64 pbl_last_bt_num; u64 pbl_bt_cnt = 0; - u64 bt_idx; u64 size; + int i; - mhop_num = (mr->type == MR_TYPE_FRMR ? 1 : hr_dev->caps.pbl_hop_num); - pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); pbl_last_bt_num = (npages + pbl_bt_sz / 8 - 1) / (pbl_bt_sz / 8); - if (mhop_num == HNS_ROCE_HOP_NUM_0) - return 0; - - /* hop_num = 1 */ - if (mhop_num == 1) { - if (npages > pbl_bt_sz / 8) { - dev_err(dev, "npages %d is larger than buf_pg_sz!", - npages); - return -EINVAL; + /* alloc L1 BT */ + for (i = 0; i < pbl_bt_sz / 8; i++) { + if (pbl_bt_cnt + 1 < pbl_last_bt_num) { + size = pbl_bt_sz; + } else { + npages_allocated = i * (pbl_bt_sz / 8); + size = (npages - npages_allocated) * 8; } - mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, - &(mr->pbl_dma_addr), - GFP_KERNEL); - if (!mr->pbl_buf) + mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, size, + &(mr->pbl_l1_dma_addr[i]), + GFP_KERNEL); + if (!mr->pbl_bt_l1[i]) { + hns_roce_loop_free(hr_dev, mr, 1, i, 0); return -ENOMEM; + } - mr->pbl_size = npages; - mr->pbl_ba = mr->pbl_dma_addr; - mr->pbl_hop_num = mhop_num; - mr->pbl_ba_pg_sz = hr_dev->caps.pbl_ba_pg_sz; - mr->pbl_buf_pg_sz = hr_dev->caps.pbl_buf_pg_sz; - return 0; + *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; + + pbl_bt_cnt++; + if (pbl_bt_cnt >= pbl_last_bt_num) + break; } - mr->pbl_l1_dma_addr = kcalloc(pbl_bt_sz / 8, - sizeof(*mr->pbl_l1_dma_addr), + mr->l0_chunk_last_num = i + 1; + + return 0; +} + +static int pbl_3hop_alloc(struct hns_roce_dev *hr_dev, int npages, + struct hns_roce_mr *mr, u32 pbl_bt_sz) +{ + struct device *dev = hr_dev->dev; + int mr_alloc_done = 0; + int npages_allocated; + u64 pbl_last_bt_num; + u64 pbl_bt_cnt = 0; + u64 bt_idx; + u64 size; + int i; + int j = 0; + + pbl_last_bt_num = (npages + pbl_bt_sz / 8 - 1) / (pbl_bt_sz / 8); + + mr->pbl_l2_dma_addr = kcalloc(pbl_last_bt_num, + sizeof(*mr->pbl_l2_dma_addr), GFP_KERNEL); - if (!mr->pbl_l1_dma_addr) + if (!mr->pbl_l2_dma_addr) return -ENOMEM; - mr->pbl_bt_l1 = kcalloc(pbl_bt_sz / 8, sizeof(*mr->pbl_bt_l1), + mr->pbl_bt_l2 = kcalloc(pbl_last_bt_num, + sizeof(*mr->pbl_bt_l2), GFP_KERNEL); - if (!mr->pbl_bt_l1) - goto err_kcalloc_bt_l1; - - if (mhop_num == 3) { - mr->pbl_l2_dma_addr = kcalloc(pbl_last_bt_num, - sizeof(*mr->pbl_l2_dma_addr), - GFP_KERNEL); - if (!mr->pbl_l2_dma_addr) - goto err_kcalloc_l2_dma; + if (!mr->pbl_bt_l2) + goto err_kcalloc_bt_l2; + + /* alloc L1, L2 BT */ + for (i = 0; i < pbl_bt_sz / 8; i++) { + mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, pbl_bt_sz, + &(mr->pbl_l1_dma_addr[i]), + GFP_KERNEL); + if (!mr->pbl_bt_l1[i]) { + hns_roce_loop_free(hr_dev, mr, 1, i, 0); + goto err_dma_alloc_l0; + } - mr->pbl_bt_l2 = kcalloc(pbl_last_bt_num, - sizeof(*mr->pbl_bt_l2), - GFP_KERNEL); - if (!mr->pbl_bt_l2) - goto err_kcalloc_bt_l2; - } + *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; - /* alloc L0 BT */ - mr->pbl_bt_l0 = dma_alloc_coherent(dev, pbl_bt_sz, - &(mr->pbl_l0_dma_addr), - GFP_KERNEL); - if (!mr->pbl_bt_l0) - goto err_dma_alloc_l0; + for (j = 0; j < pbl_bt_sz / 8; j++) { + bt_idx = i * pbl_bt_sz / 8 + j; - if (mhop_num == 2) { - /* alloc L1 BT */ - for (i = 0; i < pbl_bt_sz / 8; i++) { if (pbl_bt_cnt + 1 < pbl_last_bt_num) { size = pbl_bt_sz; } else { - npages_allocated = i * (pbl_bt_sz / 8); + npages_allocated = bt_idx * + (pbl_bt_sz / 8); size = (npages - npages_allocated) * 8; } - mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, size, - &(mr->pbl_l1_dma_addr[i]), - GFP_KERNEL); - if (!mr->pbl_bt_l1[i]) { - hns_roce_loop_free(hr_dev, mr, 1, i, 0); + mr->pbl_bt_l2[bt_idx] = dma_alloc_coherent( + dev, size, + &(mr->pbl_l2_dma_addr[bt_idx]), + GFP_KERNEL); + if (!mr->pbl_bt_l2[bt_idx]) { + hns_roce_loop_free(hr_dev, mr, 2, i, j); goto err_dma_alloc_l0; } - *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; + *(mr->pbl_bt_l1[i] + j) = + mr->pbl_l2_dma_addr[bt_idx]; pbl_bt_cnt++; - if (pbl_bt_cnt >= pbl_last_bt_num) + if (pbl_bt_cnt >= pbl_last_bt_num) { + mr_alloc_done = 1; break; - } - } else if (mhop_num == 3) { - /* alloc L1, L2 BT */ - for (i = 0; i < pbl_bt_sz / 8; i++) { - mr->pbl_bt_l1[i] = dma_alloc_coherent(dev, pbl_bt_sz, - &(mr->pbl_l1_dma_addr[i]), - GFP_KERNEL); - if (!mr->pbl_bt_l1[i]) { - hns_roce_loop_free(hr_dev, mr, 1, i, 0); - goto err_dma_alloc_l0; } + } - *(mr->pbl_bt_l0 + i) = mr->pbl_l1_dma_addr[i]; + if (mr_alloc_done) + break; + } - for (j = 0; j < pbl_bt_sz / 8; j++) { - bt_idx = i * pbl_bt_sz / 8 + j; + mr->l0_chunk_last_num = i + 1; + mr->l1_chunk_last_num = j + 1; - if (pbl_bt_cnt + 1 < pbl_last_bt_num) { - size = pbl_bt_sz; - } else { - npages_allocated = bt_idx * - (pbl_bt_sz / 8); - size = (npages - npages_allocated) * 8; - } - mr->pbl_bt_l2[bt_idx] = dma_alloc_coherent( - dev, size, - &(mr->pbl_l2_dma_addr[bt_idx]), - GFP_KERNEL); - if (!mr->pbl_bt_l2[bt_idx]) { - hns_roce_loop_free(hr_dev, mr, 2, i, j); - goto err_dma_alloc_l0; - } - *(mr->pbl_bt_l1[i] + j) = - mr->pbl_l2_dma_addr[bt_idx]; + return 0; - pbl_bt_cnt++; - if (pbl_bt_cnt >= pbl_last_bt_num) { - mr_alloc_done = 1; - break; - } - } +err_dma_alloc_l0: + kfree(mr->pbl_bt_l2); + mr->pbl_bt_l2 = NULL; - if (mr_alloc_done) - break; - } +err_kcalloc_bt_l2: + kfree(mr->pbl_l2_dma_addr); + mr->pbl_l2_dma_addr = NULL; + + return -ENOMEM; +} + + +/* PBL multi hop addressing */ +static int hns_roce_mhop_alloc(struct hns_roce_dev *hr_dev, int npages, + struct hns_roce_mr *mr) +{ + struct device *dev = hr_dev->dev; + u32 pbl_bt_sz; + u32 mhop_num; + + mhop_num = (mr->type == MR_TYPE_FRMR ? 1 : hr_dev->caps.pbl_hop_num); + pbl_bt_sz = 1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT); + + if (mhop_num == HNS_ROCE_HOP_NUM_0) + return 0; + + if (mhop_num == 1) + return pbl_1hop_alloc(hr_dev, npages, mr, pbl_bt_sz); + + mr->pbl_l1_dma_addr = kcalloc(pbl_bt_sz / 8, + sizeof(*mr->pbl_l1_dma_addr), + GFP_KERNEL); + if (!mr->pbl_l1_dma_addr) + return -ENOMEM; + + mr->pbl_bt_l1 = kcalloc(pbl_bt_sz / 8, sizeof(*mr->pbl_bt_l1), + GFP_KERNEL); + if (!mr->pbl_bt_l1) + goto err_kcalloc_bt_l1; + + /* alloc L0 BT */ + mr->pbl_bt_l0 = dma_alloc_coherent(dev, pbl_bt_sz, + &(mr->pbl_l0_dma_addr), + GFP_KERNEL); + if (!mr->pbl_bt_l0) + goto err_kcalloc_l2_dma; + + if (mhop_num == 2) { + if (pbl_2hop_alloc(hr_dev, npages, mr, pbl_bt_sz)) + goto err_kcalloc_l2_dma; + } + + if (mhop_num == 3) { + if (pbl_3hop_alloc(hr_dev, npages, mr, pbl_bt_sz)) + goto err_kcalloc_l2_dma; } - mr->l0_chunk_last_num = i + 1; - if (mhop_num == 3) - mr->l1_chunk_last_num = j + 1; mr->pbl_size = npages; mr->pbl_ba = mr->pbl_l0_dma_addr; @@ -505,14 +557,6 @@ static int hns_roce_mhop_alloc(struct hns_roce_dev *hr_dev, int npages, return 0; -err_dma_alloc_l0: - kfree(mr->pbl_bt_l2); - mr->pbl_bt_l2 = NULL; - -err_kcalloc_bt_l2: - kfree(mr->pbl_l2_dma_addr); - mr->pbl_l2_dma_addr = NULL; - err_kcalloc_l2_dma: kfree(mr->pbl_bt_l1); mr->pbl_bt_l1 = NULL; @@ -1161,6 +1205,83 @@ err_free: return ERR_PTR(ret); } +static int rereg_mr_trans(struct ib_mr *ibmr, int flags, + u64 start, u64 length, + u64 virt_addr, int mr_access_flags, + struct hns_roce_cmd_mailbox *mailbox, + u32 pdn, struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibmr->device); + struct hns_roce_mr *mr = to_hr_mr(ibmr); + struct device *dev = hr_dev->dev; + int npages; + int ret; + + if (mr->size != ~0ULL) { + npages = ib_umem_page_count(mr->umem); + + if (hr_dev->caps.pbl_hop_num) + hns_roce_mhop_free(hr_dev, mr); + else + dma_free_coherent(dev, npages * 8, + mr->pbl_buf, mr->pbl_dma_addr); + } + ib_umem_release(mr->umem); + + mr->umem = ib_umem_get(udata, start, length, mr_access_flags, 0); + if (IS_ERR(mr->umem)) { + ret = PTR_ERR(mr->umem); + mr->umem = NULL; + return -ENOMEM; + } + npages = ib_umem_page_count(mr->umem); + + if (hr_dev->caps.pbl_hop_num) { + ret = hns_roce_mhop_alloc(hr_dev, npages, mr); + if (ret) + goto release_umem; + } else { + mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, + &(mr->pbl_dma_addr), + GFP_KERNEL); + if (!mr->pbl_buf) { + ret = -ENOMEM; + goto release_umem; + } + } + + ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, + mr_access_flags, virt_addr, + length, mailbox->buf); + if (ret) + goto release_umem; + + + ret = hns_roce_ib_umem_write_mr(hr_dev, mr, mr->umem); + if (ret) { + if (mr->size != ~0ULL) { + npages = ib_umem_page_count(mr->umem); + + if (hr_dev->caps.pbl_hop_num) + hns_roce_mhop_free(hr_dev, mr); + else + dma_free_coherent(dev, npages * 8, + mr->pbl_buf, + mr->pbl_dma_addr); + } + + goto release_umem; + } + + return 0; + +release_umem: + ib_umem_release(mr->umem); + return ret; + +} + + int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_pd *pd, struct ib_udata *udata) @@ -1171,7 +1292,6 @@ int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, struct device *dev = hr_dev->dev; unsigned long mtpt_idx; u32 pdn = 0; - int npages; int ret; if (!mr->enabled) @@ -1198,73 +1318,25 @@ int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, pdn = to_hr_pd(pd)->pdn; if (flags & IB_MR_REREG_TRANS) { - if (mr->size != ~0ULL) { - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) - hns_roce_mhop_free(hr_dev, mr); - else - dma_free_coherent(dev, npages * 8, mr->pbl_buf, - mr->pbl_dma_addr); - } - ib_umem_release(mr->umem); - - mr->umem = - ib_umem_get(udata, start, length, mr_access_flags, 0); - if (IS_ERR(mr->umem)) { - ret = PTR_ERR(mr->umem); - mr->umem = NULL; + ret = rereg_mr_trans(ibmr, flags, + start, length, + virt_addr, mr_access_flags, + mailbox, pdn, udata); + if (ret) goto free_cmd_mbox; - } - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) { - ret = hns_roce_mhop_alloc(hr_dev, npages, mr); - if (ret) - goto release_umem; - } else { - mr->pbl_buf = dma_alloc_coherent(dev, npages * 8, - &(mr->pbl_dma_addr), - GFP_KERNEL); - if (!mr->pbl_buf) { - ret = -ENOMEM; - goto release_umem; - } - } - } - - ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, - mr_access_flags, virt_addr, - length, mailbox->buf); - if (ret) { - if (flags & IB_MR_REREG_TRANS) - goto release_umem; - else + } else { + ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, + mr_access_flags, virt_addr, + length, mailbox->buf); + if (ret) goto free_cmd_mbox; } - if (flags & IB_MR_REREG_TRANS) { - ret = hns_roce_ib_umem_write_mr(hr_dev, mr, mr->umem); - if (ret) { - if (mr->size != ~0ULL) { - npages = ib_umem_page_count(mr->umem); - - if (hr_dev->caps.pbl_hop_num) - hns_roce_mhop_free(hr_dev, mr); - else - dma_free_coherent(dev, npages * 8, - mr->pbl_buf, - mr->pbl_dma_addr); - } - - goto release_umem; - } - } - ret = hns_roce_sw2hw_mpt(hr_dev, mailbox, mtpt_idx); if (ret) { dev_err(dev, "SW2HW_MPT failed (%d)\n", ret); - goto release_umem; + ib_umem_release(mr->umem); + goto free_cmd_mbox; } mr->enabled = 1; @@ -1275,9 +1347,6 @@ int hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, u64 length, return 0; -release_umem: - ib_umem_release(mr->umem); - free_cmd_mbox: hns_roce_free_cmd_mailbox(hr_dev, mailbox); diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c index e0424029b058..ba8176869f51 100644 --- a/drivers/infiniband/hw/hns/hns_roce_qp.c +++ b/drivers/infiniband/hw/hns/hns_roce_qp.c @@ -324,31 +324,46 @@ static int hns_roce_set_rq_size(struct hns_roce_dev *hr_dev, return 0; } -static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev, - struct ib_qp_cap *cap, - struct hns_roce_qp *hr_qp, - struct hns_roce_ib_create_qp *ucmd) +static int check_sq_size_with_integrity(struct hns_roce_dev *hr_dev, + struct ib_qp_cap *cap, + struct hns_roce_ib_create_qp *ucmd) { u32 roundup_sq_stride = roundup_pow_of_two(hr_dev->caps.max_sq_desc_sz); u8 max_sq_stride = ilog2(roundup_sq_stride); - u32 ex_sge_num; - u32 page_size; - u32 max_cnt; /* Sanity check SQ size before proceeding */ if ((u32)(1 << ucmd->log_sq_bb_count) > hr_dev->caps.max_wqes || ucmd->log_sq_stride > max_sq_stride || ucmd->log_sq_stride < HNS_ROCE_IB_MIN_SQ_STRIDE) { - dev_err(hr_dev->dev, "check SQ size error!\n"); + ibdev_err(&hr_dev->ib_dev, "check SQ size error!\n"); return -EINVAL; } if (cap->max_send_sge > hr_dev->caps.max_sq_sg) { - dev_err(hr_dev->dev, "SQ sge error! max_send_sge=%d\n", - cap->max_send_sge); + ibdev_err(&hr_dev->ib_dev, "SQ sge error! max_send_sge=%d\n", + cap->max_send_sge); return -EINVAL; } + return 0; +} + +static int hns_roce_set_user_sq_size(struct hns_roce_dev *hr_dev, + struct ib_qp_cap *cap, + struct hns_roce_qp *hr_qp, + struct hns_roce_ib_create_qp *ucmd) +{ + u32 ex_sge_num; + u32 page_size; + u32 max_cnt; + int ret; + + ret = check_sq_size_with_integrity(hr_dev, cap, ucmd); + if (ret) { + ibdev_err(&hr_dev->ib_dev, "Sanity check sq size failed\n"); + return ret; + } + hr_qp->sq.wqe_cnt = 1 << ucmd->log_sq_bb_count; hr_qp->sq.wqe_shift = ucmd->log_sq_stride; @@ -501,6 +516,35 @@ static int calc_wqe_bt_page_shift(struct hns_roce_dev *hr_dev, return bt_pg_shift - PAGE_SHIFT; } +static int set_extend_sge_param(struct hns_roce_dev *hr_dev, + struct hns_roce_qp *hr_qp) +{ + struct device *dev = hr_dev->dev; + + if (hr_qp->sq.max_gs > 2) { + hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * + (hr_qp->sq.max_gs - 2)); + hr_qp->sge.sge_shift = 4; + } + + /* ud sqwqe's sge use extend sge */ + if (hr_dev->caps.max_sq_sg > 2 && hr_qp->ibqp.qp_type == IB_QPT_GSI) { + hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * + hr_qp->sq.max_gs); + hr_qp->sge.sge_shift = 4; + } + + if ((hr_qp->sq.max_gs > 2) && hr_dev->pci_dev->revision == 0x20) { + if (hr_qp->sge.sge_cnt > hr_dev->caps.max_extend_sg) { + dev_err(dev, "The extended sge cnt error! sge_cnt=%d\n", + hr_qp->sge.sge_cnt); + return -EINVAL; + } + } + + return 0; +} + static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, struct hns_roce_qp *hr_qp) @@ -509,6 +553,7 @@ static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev, u32 page_size; u32 max_cnt; int size; + int ret; if (cap->max_send_wr > hr_dev->caps.max_wqes || cap->max_send_sge > hr_dev->caps.max_sq_sg || @@ -518,8 +563,6 @@ static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev, } hr_qp->sq.wqe_shift = ilog2(hr_dev->caps.max_sq_desc_sz); - hr_qp->sq_max_wqes_per_wr = 1; - hr_qp->sq_spare_wqes = 0; if (hr_dev->caps.min_wqes) max_cnt = max(cap->max_send_wr, hr_dev->caps.min_wqes); @@ -539,25 +582,10 @@ static int hns_roce_set_kernel_sq_size(struct hns_roce_dev *hr_dev, else hr_qp->sq.max_gs = max_cnt; - if (hr_qp->sq.max_gs > 2) { - hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * - (hr_qp->sq.max_gs - 2)); - hr_qp->sge.sge_shift = 4; - } - - /* ud sqwqe's sge use extend sge */ - if (hr_dev->caps.max_sq_sg > 2 && hr_qp->ibqp.qp_type == IB_QPT_GSI) { - hr_qp->sge.sge_cnt = roundup_pow_of_two(hr_qp->sq.wqe_cnt * - hr_qp->sq.max_gs); - hr_qp->sge.sge_shift = 4; - } - - if ((hr_qp->sq.max_gs > 2) && hr_dev->pci_dev->revision == 0x20) { - if (hr_qp->sge.sge_cnt > hr_dev->caps.max_extend_sg) { - dev_err(dev, "The extended sge cnt error! sge_cnt=%d\n", - hr_qp->sge.sge_cnt); - return -EINVAL; - } + ret = set_extend_sge_param(hr_dev, hr_qp); + if (ret) { + dev_err(dev, "set extend sge parameters fail\n"); + return ret; } /* Get buf size, SQ and RQ are aligned to PAGE_SIZE */ @@ -826,11 +854,18 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev, hr_qp->sq.wrid = kcalloc(hr_qp->sq.wqe_cnt, sizeof(u64), GFP_KERNEL); - hr_qp->rq.wrid = kcalloc(hr_qp->rq.wqe_cnt, sizeof(u64), - GFP_KERNEL); - if (!hr_qp->sq.wrid || !hr_qp->rq.wrid) { + if (ZERO_OR_NULL_PTR(hr_qp->sq.wrid)) { ret = -ENOMEM; - goto err_wrid; + goto err_get_bufs; + } + + if (hr_qp->rq.wqe_cnt) { + hr_qp->rq.wrid = kcalloc(hr_qp->rq.wqe_cnt, sizeof(u64), + GFP_KERNEL); + if (ZERO_OR_NULL_PTR(hr_qp->rq.wrid)) { + ret = -ENOMEM; + goto err_sq_wrid; + } } } @@ -916,8 +951,8 @@ err_wrid: hns_roce_qp_has_rq(init_attr)) hns_roce_db_unmap_user(uctx, &hr_qp->rdb); } else { - kfree(hr_qp->sq.wrid); - kfree(hr_qp->rq.wrid); + if (hr_qp->rq.wqe_cnt) + kfree(hr_qp->rq.wrid); } err_sq_dbmap: @@ -928,6 +963,10 @@ err_sq_dbmap: hns_roce_qp_has_sq(init_attr)) hns_roce_db_unmap_user(uctx, &hr_qp->sdb); +err_sq_wrid: + if (!udata) + kfree(hr_qp->sq.wrid); + err_get_bufs: hns_roce_free_buf_list(buf_list, hr_qp->region_cnt); @@ -942,11 +981,13 @@ err_db: hns_roce_free_db(hr_dev, &hr_qp->rdb); err_rq_sge_list: - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && + hns_roce_qp_has_rq(init_attr)) kfree(hr_qp->rq_inl_buf.wqe_list[0].sg_list); err_wqe_list: - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && + hns_roce_qp_has_rq(init_attr)) kfree(hr_qp->rq_inl_buf.wqe_list); err_out: @@ -958,7 +999,7 @@ struct ib_qp *hns_roce_create_qp(struct ib_pd *pd, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); - struct device *dev = hr_dev->dev; + struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_sqp *hr_sqp; struct hns_roce_qp *hr_qp; int ret; @@ -972,7 +1013,8 @@ struct ib_qp *hns_roce_create_qp(struct ib_pd *pd, ret = hns_roce_create_qp_common(hr_dev, pd, init_attr, udata, 0, hr_qp); if (ret) { - dev_err(dev, "Create RC QP failed\n"); + ibdev_err(ibdev, "Create RC QP 0x%06lx failed(%d)\n", + hr_qp->qpn, ret); kfree(hr_qp); return ERR_PTR(ret); } @@ -984,7 +1026,7 @@ struct ib_qp *hns_roce_create_qp(struct ib_pd *pd, case IB_QPT_GSI: { /* Userspace is not allowed to create special QPs: */ if (udata) { - dev_err(dev, "not support usr space GSI\n"); + ibdev_err(ibdev, "not support usr space GSI\n"); return ERR_PTR(-EINVAL); } @@ -1006,7 +1048,7 @@ struct ib_qp *hns_roce_create_qp(struct ib_pd *pd, ret = hns_roce_create_qp_common(hr_dev, pd, init_attr, udata, hr_qp->ibqp.qp_num, hr_qp); if (ret) { - dev_err(dev, "Create GSI QP failed!\n"); + ibdev_err(ibdev, "Create GSI QP failed!\n"); kfree(hr_sqp); return ERR_PTR(ret); } @@ -1014,7 +1056,8 @@ struct ib_qp *hns_roce_create_qp(struct ib_pd *pd, break; } default:{ - dev_err(dev, "not support QP type %d\n", init_attr->qp_type); + ibdev_err(ibdev, "not support QP type %d\n", + init_attr->qp_type); return ERR_PTR(-EINVAL); } } @@ -1040,23 +1083,88 @@ int to_hr_qp_type(int qp_type) return transport_type; } +static int check_mtu_validate(struct hns_roce_dev *hr_dev, + struct hns_roce_qp *hr_qp, + struct ib_qp_attr *attr, int attr_mask) +{ + enum ib_mtu active_mtu; + int p; + + p = attr_mask & IB_QP_PORT ? (attr->port_num - 1) : hr_qp->port; + active_mtu = iboe_get_mtu(hr_dev->iboe.netdevs[p]->mtu); + + if ((hr_dev->caps.max_mtu >= IB_MTU_2048 && + attr->path_mtu > hr_dev->caps.max_mtu) || + attr->path_mtu < IB_MTU_256 || attr->path_mtu > active_mtu) { + ibdev_err(&hr_dev->ib_dev, + "attr path_mtu(%d)invalid while modify qp", + attr->path_mtu); + return -EINVAL; + } + + return 0; +} + +static int hns_roce_check_qp_attr(struct ib_qp *ibqp, struct ib_qp_attr *attr, + int attr_mask) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); + struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); + int p; + + if ((attr_mask & IB_QP_PORT) && + (attr->port_num == 0 || attr->port_num > hr_dev->caps.num_ports)) { + ibdev_err(&hr_dev->ib_dev, + "attr port_num invalid.attr->port_num=%d\n", + attr->port_num); + return -EINVAL; + } + + if (attr_mask & IB_QP_PKEY_INDEX) { + p = attr_mask & IB_QP_PORT ? (attr->port_num - 1) : hr_qp->port; + if (attr->pkey_index >= hr_dev->caps.pkey_table_len[p]) { + ibdev_err(&hr_dev->ib_dev, + "attr pkey_index invalid.attr->pkey_index=%d\n", + attr->pkey_index); + return -EINVAL; + } + } + + if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && + attr->max_rd_atomic > hr_dev->caps.max_qp_init_rdma) { + ibdev_err(&hr_dev->ib_dev, + "attr max_rd_atomic invalid.attr->max_rd_atomic=%d\n", + attr->max_rd_atomic); + return -EINVAL; + } + + if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && + attr->max_dest_rd_atomic > hr_dev->caps.max_qp_dest_rdma) { + ibdev_err(&hr_dev->ib_dev, + "attr max_dest_rd_atomic invalid.attr->max_dest_rd_atomic=%d\n", + attr->max_dest_rd_atomic); + return -EINVAL; + } + + if (attr_mask & IB_QP_PATH_MTU) + return check_mtu_validate(hr_dev, hr_qp, attr, attr_mask); + + return 0; +} + int hns_roce_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); enum ib_qp_state cur_state, new_state; - struct device *dev = hr_dev->dev; int ret = -EINVAL; - int p; - enum ib_mtu active_mtu; mutex_lock(&hr_qp->mutex); cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : (enum ib_qp_state)hr_qp->state; - new_state = attr_mask & IB_QP_STATE ? - attr->qp_state : cur_state; + new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state; if (ibqp->uobject && (attr_mask & IB_QP_STATE) && new_state == IB_QPS_ERR) { @@ -1066,67 +1174,27 @@ int hns_roce_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, if (hr_qp->rdb_en == 1) hr_qp->rq.head = *(int *)(hr_qp->rdb.virt_addr); } else { - dev_warn(dev, "flush cqe is not supported in userspace!\n"); + ibdev_warn(&hr_dev->ib_dev, + "flush cqe is not supported in userspace!\n"); goto out; } } if (!ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type, attr_mask)) { - dev_err(dev, "ib_modify_qp_is_ok failed\n"); + ibdev_err(&hr_dev->ib_dev, "ib_modify_qp_is_ok failed\n"); goto out; } - if ((attr_mask & IB_QP_PORT) && - (attr->port_num == 0 || attr->port_num > hr_dev->caps.num_ports)) { - dev_err(dev, "attr port_num invalid.attr->port_num=%d\n", - attr->port_num); - goto out; - } - - if (attr_mask & IB_QP_PKEY_INDEX) { - p = attr_mask & IB_QP_PORT ? (attr->port_num - 1) : hr_qp->port; - if (attr->pkey_index >= hr_dev->caps.pkey_table_len[p]) { - dev_err(dev, "attr pkey_index invalid.attr->pkey_index=%d\n", - attr->pkey_index); - goto out; - } - } - - if (attr_mask & IB_QP_PATH_MTU) { - p = attr_mask & IB_QP_PORT ? (attr->port_num - 1) : hr_qp->port; - active_mtu = iboe_get_mtu(hr_dev->iboe.netdevs[p]->mtu); - - if ((hr_dev->caps.max_mtu == IB_MTU_4096 && - attr->path_mtu > IB_MTU_4096) || - (hr_dev->caps.max_mtu == IB_MTU_2048 && - attr->path_mtu > IB_MTU_2048) || - attr->path_mtu < IB_MTU_256 || - attr->path_mtu > active_mtu) { - dev_err(dev, "attr path_mtu(%d)invalid while modify qp", - attr->path_mtu); - goto out; - } - } - - if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && - attr->max_rd_atomic > hr_dev->caps.max_qp_init_rdma) { - dev_err(dev, "attr max_rd_atomic invalid.attr->max_rd_atomic=%d\n", - attr->max_rd_atomic); - goto out; - } - - if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && - attr->max_dest_rd_atomic > hr_dev->caps.max_qp_dest_rdma) { - dev_err(dev, "attr max_dest_rd_atomic invalid.attr->max_dest_rd_atomic=%d\n", - attr->max_dest_rd_atomic); + ret = hns_roce_check_qp_attr(ibqp, attr, attr_mask); + if (ret) goto out; - } if (cur_state == new_state && cur_state == IB_QPS_RESET) { if (hr_dev->caps.min_wqes) { ret = -EPERM; - dev_err(dev, "cur_state=%d new_state=%d\n", cur_state, + ibdev_err(&hr_dev->ib_dev, + "cur_state=%d new_state=%d\n", cur_state, new_state); } else { ret = 0; diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c index 38bb548eaa6d..c011422112b2 100644 --- a/drivers/infiniband/hw/hns/hns_roce_srq.c +++ b/drivers/infiniband/hw/hns/hns_roce_srq.c @@ -175,6 +175,91 @@ static void hns_roce_srq_free(struct hns_roce_dev *hr_dev, hns_roce_bitmap_free(&srq_table->bitmap, srq->srqn, BITMAP_NO_RR); } +static int create_user_srq(struct hns_roce_srq *srq, struct ib_udata *udata, + int srq_buf_size) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); + struct hns_roce_ib_create_srq ucmd; + u32 page_shift; + u32 npages; + int ret; + + if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) + return -EFAULT; + + srq->umem = ib_umem_get(udata, ucmd.buf_addr, srq_buf_size, 0, 0); + if (IS_ERR(srq->umem)) + return PTR_ERR(srq->umem); + + if (hr_dev->caps.srqwqe_buf_pg_sz) { + npages = (ib_umem_page_count(srq->umem) + + (1 << hr_dev->caps.srqwqe_buf_pg_sz) - 1) / + (1 << hr_dev->caps.srqwqe_buf_pg_sz); + page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; + ret = hns_roce_mtt_init(hr_dev, npages, page_shift, &srq->mtt); + } else + ret = hns_roce_mtt_init(hr_dev, ib_umem_page_count(srq->umem), + PAGE_SHIFT, &srq->mtt); + if (ret) + goto err_user_buf; + + ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->mtt, srq->umem); + if (ret) + goto err_user_srq_mtt; + + /* config index queue BA */ + srq->idx_que.umem = ib_umem_get(udata, ucmd.que_addr, + srq->idx_que.buf_size, 0, 0); + if (IS_ERR(srq->idx_que.umem)) { + dev_err(hr_dev->dev, "ib_umem_get error for index queue\n"); + ret = PTR_ERR(srq->idx_que.umem); + goto err_user_srq_mtt; + } + + if (hr_dev->caps.idx_buf_pg_sz) { + npages = (ib_umem_page_count(srq->idx_que.umem) + + (1 << hr_dev->caps.idx_buf_pg_sz) - 1) / + (1 << hr_dev->caps.idx_buf_pg_sz); + page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; + ret = hns_roce_mtt_init(hr_dev, npages, page_shift, + &srq->idx_que.mtt); + } else { + ret = hns_roce_mtt_init(hr_dev, + ib_umem_page_count(srq->idx_que.umem), + PAGE_SHIFT, + &srq->idx_que.mtt); + } + + if (ret) { + dev_err(hr_dev->dev, "hns_roce_mtt_init error for idx que\n"); + goto err_user_idx_mtt; + } + + ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->idx_que.mtt, + srq->idx_que.umem); + if (ret) { + dev_err(hr_dev->dev, + "hns_roce_ib_umem_write_mtt error for idx que\n"); + goto err_user_idx_buf; + } + + return 0; + +err_user_idx_buf: + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + +err_user_idx_mtt: + ib_umem_release(srq->idx_que.umem); + +err_user_srq_mtt: + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + +err_user_buf: + ib_umem_release(srq->umem); + + return ret; +} + static int hns_roce_create_idx_que(struct ib_pd *pd, struct hns_roce_srq *srq, u32 page_shift) { @@ -196,6 +281,93 @@ static int hns_roce_create_idx_que(struct ib_pd *pd, struct hns_roce_srq *srq, return 0; } +static int create_kernel_srq(struct hns_roce_srq *srq, int srq_buf_size) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); + u32 page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; + int ret; + + if (hns_roce_buf_alloc(hr_dev, srq_buf_size, (1 << page_shift) * 2, + &srq->buf, page_shift)) + return -ENOMEM; + + srq->head = 0; + srq->tail = srq->max - 1; + + ret = hns_roce_mtt_init(hr_dev, srq->buf.npages, srq->buf.page_shift, + &srq->mtt); + if (ret) + goto err_kernel_buf; + + ret = hns_roce_buf_write_mtt(hr_dev, &srq->mtt, &srq->buf); + if (ret) + goto err_kernel_srq_mtt; + + page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; + ret = hns_roce_create_idx_que(srq->ibsrq.pd, srq, page_shift); + if (ret) { + dev_err(hr_dev->dev, "Create idx queue fail(%d)!\n", ret); + goto err_kernel_srq_mtt; + } + + /* Init mtt table for idx_que */ + ret = hns_roce_mtt_init(hr_dev, srq->idx_que.idx_buf.npages, + srq->idx_que.idx_buf.page_shift, + &srq->idx_que.mtt); + if (ret) + goto err_kernel_create_idx; + + /* Write buffer address into the mtt table */ + ret = hns_roce_buf_write_mtt(hr_dev, &srq->idx_que.mtt, + &srq->idx_que.idx_buf); + if (ret) + goto err_kernel_idx_buf; + + srq->wrid = kvmalloc_array(srq->max, sizeof(u64), GFP_KERNEL); + if (!srq->wrid) { + ret = -ENOMEM; + goto err_kernel_idx_buf; + } + + return 0; + +err_kernel_idx_buf: + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + +err_kernel_create_idx: + hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, + &srq->idx_que.idx_buf); + kfree(srq->idx_que.bitmap); + +err_kernel_srq_mtt: + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + +err_kernel_buf: + hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); + + return ret; +} + +static void destroy_user_srq(struct hns_roce_dev *hr_dev, + struct hns_roce_srq *srq) +{ + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + ib_umem_release(srq->idx_que.umem); + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + ib_umem_release(srq->umem); +} + +static void destroy_kernel_srq(struct hns_roce_dev *hr_dev, + struct hns_roce_srq *srq, int srq_buf_size) +{ + kvfree(srq->wrid); + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, &srq->idx_que.idx_buf); + kfree(srq->idx_que.bitmap); + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); +} + int hns_roce_create_srq(struct ib_srq *ib_srq, struct ib_srq_init_attr *srq_init_attr, struct ib_udata *udata) @@ -205,9 +377,7 @@ int hns_roce_create_srq(struct ib_srq *ib_srq, struct hns_roce_srq *srq = to_hr_srq(ib_srq); int srq_desc_size; int srq_buf_size; - u32 page_shift; int ret = 0; - u32 npages; u32 cqn; /* Check the actual SRQ wqe and SRQ sge num */ @@ -233,115 +403,16 @@ int hns_roce_create_srq(struct ib_srq *ib_srq, srq->idx_que.mtt.mtt_type = MTT_TYPE_IDX; if (udata) { - struct hns_roce_ib_create_srq ucmd; - - if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) - return -EFAULT; - - srq->umem = - ib_umem_get(udata, ucmd.buf_addr, srq_buf_size, 0, 0); - if (IS_ERR(srq->umem)) - return PTR_ERR(srq->umem); - - if (hr_dev->caps.srqwqe_buf_pg_sz) { - npages = (ib_umem_page_count(srq->umem) + - (1 << hr_dev->caps.srqwqe_buf_pg_sz) - 1) / - (1 << hr_dev->caps.srqwqe_buf_pg_sz); - page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; - ret = hns_roce_mtt_init(hr_dev, npages, - page_shift, - &srq->mtt); - } else - ret = hns_roce_mtt_init(hr_dev, - ib_umem_page_count(srq->umem), - PAGE_SHIFT, &srq->mtt); - if (ret) - goto err_buf; - - ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->mtt, srq->umem); - if (ret) - goto err_srq_mtt; - - /* config index queue BA */ - srq->idx_que.umem = ib_umem_get(udata, ucmd.que_addr, - srq->idx_que.buf_size, 0, 0); - if (IS_ERR(srq->idx_que.umem)) { - dev_err(hr_dev->dev, - "ib_umem_get error for index queue\n"); - ret = PTR_ERR(srq->idx_que.umem); - goto err_srq_mtt; - } - - if (hr_dev->caps.idx_buf_pg_sz) { - npages = (ib_umem_page_count(srq->idx_que.umem) + - (1 << hr_dev->caps.idx_buf_pg_sz) - 1) / - (1 << hr_dev->caps.idx_buf_pg_sz); - page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; - ret = hns_roce_mtt_init(hr_dev, npages, - page_shift, &srq->idx_que.mtt); - } else { - ret = hns_roce_mtt_init( - hr_dev, ib_umem_page_count(srq->idx_que.umem), - PAGE_SHIFT, &srq->idx_que.mtt); - } - - if (ret) { - dev_err(hr_dev->dev, - "hns_roce_mtt_init error for idx que\n"); - goto err_idx_mtt; - } - - ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->idx_que.mtt, - srq->idx_que.umem); + ret = create_user_srq(srq, udata, srq_buf_size); if (ret) { - dev_err(hr_dev->dev, - "hns_roce_ib_umem_write_mtt error for idx que\n"); - goto err_idx_buf; + dev_err(hr_dev->dev, "Create user srq failed\n"); + goto err_srq; } } else { - page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; - if (hns_roce_buf_alloc(hr_dev, srq_buf_size, - (1 << page_shift) * 2, &srq->buf, - page_shift)) - return -ENOMEM; - - srq->head = 0; - srq->tail = srq->max - 1; - - ret = hns_roce_mtt_init(hr_dev, srq->buf.npages, - srq->buf.page_shift, &srq->mtt); - if (ret) - goto err_buf; - - ret = hns_roce_buf_write_mtt(hr_dev, &srq->mtt, &srq->buf); - if (ret) - goto err_srq_mtt; - - page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; - ret = hns_roce_create_idx_que(ib_srq->pd, srq, page_shift); + ret = create_kernel_srq(srq, srq_buf_size); if (ret) { - dev_err(hr_dev->dev, "Create idx queue fail(%d)!\n", - ret); - goto err_srq_mtt; - } - - /* Init mtt table for idx_que */ - ret = hns_roce_mtt_init(hr_dev, srq->idx_que.idx_buf.npages, - srq->idx_que.idx_buf.page_shift, - &srq->idx_que.mtt); - if (ret) - goto err_create_idx; - - /* Write buffer address into the mtt table */ - ret = hns_roce_buf_write_mtt(hr_dev, &srq->idx_que.mtt, - &srq->idx_que.idx_buf); - if (ret) - goto err_idx_buf; - - srq->wrid = kvmalloc_array(srq->max, sizeof(u64), GFP_KERNEL); - if (!srq->wrid) { - ret = -ENOMEM; - goto err_idx_buf; + dev_err(hr_dev->dev, "Create kernel srq failed\n"); + goto err_srq; } } @@ -373,27 +444,12 @@ err_srqc_alloc: hns_roce_srq_free(hr_dev, srq); err_wrid: - kvfree(srq->wrid); - -err_idx_buf: - hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); - -err_idx_mtt: - ib_umem_release(srq->idx_que.umem); - -err_create_idx: - hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, - &srq->idx_que.idx_buf); - bitmap_free(srq->idx_que.bitmap); - -err_srq_mtt: - hns_roce_mtt_cleanup(hr_dev, &srq->mtt); - -err_buf: - ib_umem_release(srq->umem); - if (!udata) - hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); + if (udata) + destroy_user_srq(hr_dev, srq); + else + destroy_kernel_srq(hr_dev, srq, srq_buf_size); +err_srq: return ret; } diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index d169a8031375..8056930bbe2c 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -97,18 +97,7 @@ static int i40iw_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { - struct i40iw_device *iwdev = to_iwdev(ibdev); - struct net_device *netdev = iwdev->netdev; - - /* props being zeroed by the caller, avoid zeroing it here */ - props->max_mtu = IB_MTU_4096; - props->active_mtu = ib_mtu_int_to_enum(netdev->mtu); - props->lid = 1; - if (netif_carrier_ok(iwdev->netdev)) - props->state = IB_PORT_ACTIVE; - else - props->state = IB_PORT_DOWN; props->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_REINIT_SUP | IB_PORT_VENDOR_CLASS_SUP | IB_PORT_BOOT_MGMT_SUP; props->gid_tbl_len = 1; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 8790101facb7..8d2f1e38b891 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -734,7 +734,8 @@ out: static u8 state_to_phys_state(enum ib_port_state state) { - return state == IB_PORT_ACTIVE ? 5 : 3; + return state == IB_PORT_ACTIVE ? + IB_PORT_PHYS_STATE_LINK_UP : IB_PORT_PHYS_STATE_DISABLED; } static int eth_link_query_port(struct ib_device *ibdev, u8 port, diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c index 753479285ce9..6ae503cfc526 100644 --- a/drivers/infiniband/hw/mlx4/mr.c +++ b/drivers/infiniband/hw/mlx4/mr.c @@ -377,6 +377,7 @@ static struct ib_umem *mlx4_get_umem_mr(struct ib_udata *udata, u64 start, * again */ if (!ib_access_writable(access_flags)) { + unsigned long untagged_start = untagged_addr(start); struct vm_area_struct *vma; down_read(¤t->mm->mmap_sem); @@ -385,9 +386,9 @@ static struct ib_umem *mlx4_get_umem_mr(struct ib_udata *udata, u64 start, * cover the memory, but for now it requires a single vma to * entirely cover the MR to support RO mappings. */ - vma = find_vma(current->mm, start); - if (vma && vma->vm_end >= start + length && - vma->vm_start <= start) { + vma = find_vma(current->mm, untagged_start); + if (vma && vma->vm_end >= untagged_start + length && + vma->vm_start <= untagged_start) { if (vma->vm_flags & VM_WRITE) access_flags |= IB_ACCESS_LOCAL_WRITE; } else { diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index 82aff2f2fdc2..bd4aa04416c6 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -325,7 +325,7 @@ static int send_wqe_overhead(enum mlx4_ib_qp_type type, u32 flags) } static int set_rq_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap, - bool is_user, int has_rq, struct mlx4_ib_qp *qp, + bool is_user, bool has_rq, struct mlx4_ib_qp *qp, u32 inl_recv_sz) { /* Sanity check RQ size before proceeding */ @@ -506,10 +506,10 @@ static void free_proxy_bufs(struct ib_device *dev, struct mlx4_ib_qp *qp) kfree(qp->sqp_proxy_rcv); } -static int qp_has_rq(struct ib_qp_init_attr *attr) +static bool qp_has_rq(struct ib_qp_init_attr *attr) { if (attr->qp_type == IB_QPT_XRC_INI || attr->qp_type == IB_QPT_XRC_TGT) - return 0; + return false; return !attr->srq; } @@ -855,12 +855,143 @@ static void mlx4_ib_release_wqn(struct mlx4_ib_ucontext *context, mutex_unlock(&context->wqn_ranges_mutex); } -static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, - enum mlx4_ib_source_type src, - struct ib_qp_init_attr *init_attr, +static int create_rq(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, + struct ib_udata *udata, struct mlx4_ib_qp *qp) +{ + struct mlx4_ib_dev *dev = to_mdev(pd->device); + int qpn; + int err; + struct mlx4_ib_ucontext *context = rdma_udata_to_drv_context( + udata, struct mlx4_ib_ucontext, ibucontext); + struct mlx4_ib_cq *mcq; + unsigned long flags; + int range_size; + struct mlx4_ib_create_wq wq; + size_t copy_len; + int shift; + int n; + + qp->mlx4_ib_qp_type = MLX4_IB_QPT_RAW_PACKET; + + mutex_init(&qp->mutex); + spin_lock_init(&qp->sq.lock); + spin_lock_init(&qp->rq.lock); + INIT_LIST_HEAD(&qp->gid_list); + INIT_LIST_HEAD(&qp->steering_rules); + + qp->state = IB_QPS_RESET; + + copy_len = min(sizeof(struct mlx4_ib_create_wq), udata->inlen); + + if (ib_copy_from_udata(&wq, udata, copy_len)) { + err = -EFAULT; + goto err; + } + + if (wq.comp_mask || wq.reserved[0] || wq.reserved[1] || + wq.reserved[2]) { + pr_debug("user command isn't supported\n"); + err = -EOPNOTSUPP; + goto err; + } + + if (wq.log_range_size > ilog2(dev->dev->caps.max_rss_tbl_sz)) { + pr_debug("WQN range size must be equal or smaller than %d\n", + dev->dev->caps.max_rss_tbl_sz); + err = -EOPNOTSUPP; + goto err; + } + range_size = 1 << wq.log_range_size; + + if (init_attr->create_flags & IB_QP_CREATE_SCATTER_FCS) + qp->flags |= MLX4_IB_QP_SCATTER_FCS; + + err = set_rq_size(dev, &init_attr->cap, true, true, qp, qp->inl_recv_sz); + if (err) + goto err; + + qp->sq_no_prefetch = 1; + qp->sq.wqe_cnt = 1; + qp->sq.wqe_shift = MLX4_IB_MIN_SQ_STRIDE; + qp->buf_size = (qp->rq.wqe_cnt << qp->rq.wqe_shift) + + (qp->sq.wqe_cnt << qp->sq.wqe_shift); + + qp->umem = ib_umem_get(udata, wq.buf_addr, qp->buf_size, 0, 0); + if (IS_ERR(qp->umem)) { + err = PTR_ERR(qp->umem); + goto err; + } + + n = ib_umem_page_count(qp->umem); + shift = mlx4_ib_umem_calc_optimal_mtt_size(qp->umem, 0, &n); + err = mlx4_mtt_init(dev->dev, n, shift, &qp->mtt); + + if (err) + goto err_buf; + + err = mlx4_ib_umem_write_mtt(dev, &qp->mtt, qp->umem); + if (err) + goto err_mtt; + + err = mlx4_ib_db_map_user(udata, wq.db_addr, &qp->db); + if (err) + goto err_mtt; + qp->mqp.usage = MLX4_RES_USAGE_USER_VERBS; + + err = mlx4_ib_alloc_wqn(context, qp, range_size, &qpn); + if (err) + goto err_wrid; + + err = mlx4_qp_alloc(dev->dev, qpn, &qp->mqp); + if (err) + goto err_qpn; + + /* + * Hardware wants QPN written in big-endian order (after + * shifting) for send doorbell. Precompute this value to save + * a little bit when posting sends. + */ + qp->doorbell_qpn = swab32(qp->mqp.qpn << 8); + + qp->mqp.event = mlx4_ib_wq_event; + + spin_lock_irqsave(&dev->reset_flow_resource_lock, flags); + mlx4_ib_lock_cqs(to_mcq(init_attr->send_cq), + to_mcq(init_attr->recv_cq)); + /* Maintain device to QPs access, needed for further handling + * via reset flow + */ + list_add_tail(&qp->qps_list, &dev->qp_list); + /* Maintain CQ to QPs access, needed for further handling + * via reset flow + */ + mcq = to_mcq(init_attr->send_cq); + list_add_tail(&qp->cq_send_list, &mcq->send_qp_list); + mcq = to_mcq(init_attr->recv_cq); + list_add_tail(&qp->cq_recv_list, &mcq->recv_qp_list); + mlx4_ib_unlock_cqs(to_mcq(init_attr->send_cq), + to_mcq(init_attr->recv_cq)); + spin_unlock_irqrestore(&dev->reset_flow_resource_lock, flags); + return 0; + +err_qpn: + mlx4_ib_release_wqn(context, qp, 0); +err_wrid: + mlx4_ib_db_unmap_user(context, &qp->db); + +err_mtt: + mlx4_mtt_cleanup(dev->dev, &qp->mtt); +err_buf: + ib_umem_release(qp->umem); +err: + return err; +} + +static int create_qp_common(struct ib_pd *pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata, int sqpn, struct mlx4_ib_qp **caller_qp) { + struct mlx4_ib_dev *dev = to_mdev(pd->device); int qpn; int err; struct mlx4_ib_sqp *sqp = NULL; @@ -870,7 +1001,6 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, enum mlx4_ib_qp_type qp_type = (enum mlx4_ib_qp_type) init_attr->qp_type; struct mlx4_ib_cq *mcq; unsigned long flags; - int range_size = 0; /* When tunneling special qps, we use a plain UD qp */ if (sqpn) { @@ -921,15 +1051,13 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (!sqp) return -ENOMEM; qp = &sqp->qp; - qp->pri.vid = 0xFFFF; - qp->alt.vid = 0xFFFF; } else { qp = kzalloc(sizeof(struct mlx4_ib_qp), GFP_KERNEL); if (!qp) return -ENOMEM; - qp->pri.vid = 0xFFFF; - qp->alt.vid = 0xFFFF; } + qp->pri.vid = 0xFFFF; + qp->alt.vid = 0xFFFF; } else qp = *caller_qp; @@ -941,48 +1069,24 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, INIT_LIST_HEAD(&qp->gid_list); INIT_LIST_HEAD(&qp->steering_rules); - qp->state = IB_QPS_RESET; + qp->state = IB_QPS_RESET; if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR) qp->sq_signal_bits = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE); - if (udata) { - union { - struct mlx4_ib_create_qp qp; - struct mlx4_ib_create_wq wq; - } ucmd; + struct mlx4_ib_create_qp ucmd; size_t copy_len; int shift; int n; - copy_len = (src == MLX4_IB_QP_SRC) ? - sizeof(struct mlx4_ib_create_qp) : - min(sizeof(struct mlx4_ib_create_wq), udata->inlen); + copy_len = sizeof(struct mlx4_ib_create_qp); if (ib_copy_from_udata(&ucmd, udata, copy_len)) { err = -EFAULT; goto err; } - if (src == MLX4_IB_RWQ_SRC) { - if (ucmd.wq.comp_mask || ucmd.wq.reserved[0] || - ucmd.wq.reserved[1] || ucmd.wq.reserved[2]) { - pr_debug("user command isn't supported\n"); - err = -EOPNOTSUPP; - goto err; - } - - if (ucmd.wq.log_range_size > - ilog2(dev->dev->caps.max_rss_tbl_sz)) { - pr_debug("WQN range size must be equal or smaller than %d\n", - dev->dev->caps.max_rss_tbl_sz); - err = -EOPNOTSUPP; - goto err; - } - range_size = 1 << ucmd.wq.log_range_size; - } else { - qp->inl_recv_sz = ucmd.qp.inl_recv_sz; - } + qp->inl_recv_sz = ucmd.inl_recv_sz; if (init_attr->create_flags & IB_QP_CREATE_SCATTER_FCS) { if (!(dev->dev->caps.flags & @@ -1000,30 +1104,14 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, if (err) goto err; - if (src == MLX4_IB_QP_SRC) { - qp->sq_no_prefetch = ucmd.qp.sq_no_prefetch; + qp->sq_no_prefetch = ucmd.sq_no_prefetch; - err = set_user_sq_size(dev, qp, - (struct mlx4_ib_create_qp *) - &ucmd); - if (err) - goto err; - } else { - qp->sq_no_prefetch = 1; - qp->sq.wqe_cnt = 1; - qp->sq.wqe_shift = MLX4_IB_MIN_SQ_STRIDE; - /* Allocated buffer expects to have at least that SQ - * size. - */ - qp->buf_size = (qp->rq.wqe_cnt << qp->rq.wqe_shift) + - (qp->sq.wqe_cnt << qp->sq.wqe_shift); - } + err = set_user_sq_size(dev, qp, &ucmd); + if (err) + goto err; qp->umem = - ib_umem_get(udata, - (src == MLX4_IB_QP_SRC) ? ucmd.qp.buf_addr : - ucmd.wq.buf_addr, - qp->buf_size, 0, 0); + ib_umem_get(udata, ucmd.buf_addr, qp->buf_size, 0, 0); if (IS_ERR(qp->umem)) { err = PTR_ERR(qp->umem); goto err; @@ -1041,11 +1129,7 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, goto err_mtt; if (qp_has_rq(init_attr)) { - err = mlx4_ib_db_map_user(udata, - (src == MLX4_IB_QP_SRC) ? - ucmd.qp.db_addr : - ucmd.wq.db_addr, - &qp->db); + err = mlx4_ib_db_map_user(udata, ucmd.db_addr, &qp->db); if (err) goto err_mtt; } @@ -1115,10 +1199,6 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, goto err_wrid; } } - } else if (src == MLX4_IB_RWQ_SRC) { - err = mlx4_ib_alloc_wqn(context, qp, range_size, &qpn); - if (err) - goto err_wrid; } else { /* Raw packet QPNs may not have bits 6,7 set in their qp_num; * otherwise, the WQE BlueFlame setup flow wrongly causes @@ -1157,8 +1237,7 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd, */ qp->doorbell_qpn = swab32(qp->mqp.qpn << 8); - qp->mqp.event = (src == MLX4_IB_QP_SRC) ? mlx4_ib_qp_event : - mlx4_ib_wq_event; + qp->mqp.event = mlx4_ib_qp_event; if (!*caller_qp) *caller_qp = qp; @@ -1186,8 +1265,6 @@ err_qpn: if (!sqpn) { if (qp->flags & MLX4_IB_QP_NETIF) mlx4_ib_steer_qp_free(dev, qpn, 1); - else if (src == MLX4_IB_RWQ_SRC) - mlx4_ib_release_wqn(context, qp, 0); else mlx4_qp_release_range(dev->dev, qpn, 1); } @@ -1518,8 +1595,7 @@ static struct ib_qp *_mlx4_ib_create_qp(struct ib_pd *pd, /* fall through */ case IB_QPT_UD: { - err = create_qp_common(to_mdev(pd->device), pd, MLX4_IB_QP_SRC, - init_attr, udata, 0, &qp); + err = create_qp_common(pd, init_attr, udata, 0, &qp); if (err) { kfree(qp); return ERR_PTR(err); @@ -1549,8 +1625,7 @@ static struct ib_qp *_mlx4_ib_create_qp(struct ib_pd *pd, sqpn = get_sqp_num(to_mdev(pd->device), init_attr); } - err = create_qp_common(to_mdev(pd->device), pd, MLX4_IB_QP_SRC, - init_attr, udata, sqpn, &qp); + err = create_qp_common(pd, init_attr, udata, sqpn, &qp); if (err) return ERR_PTR(err); @@ -4047,8 +4122,8 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd, struct ib_wq_init_attr *init_attr, struct ib_udata *udata) { - struct mlx4_ib_dev *dev; - struct ib_qp_init_attr ib_qp_init_attr; + struct mlx4_dev *dev = to_mdev(pd->device)->dev; + struct ib_qp_init_attr ib_qp_init_attr = {}; struct mlx4_ib_qp *qp; struct mlx4_ib_create_wq ucmd; int err, required_cmd_sz; @@ -4073,14 +4148,13 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd, if (udata->outlen) return ERR_PTR(-EOPNOTSUPP); - dev = to_mdev(pd->device); - if (init_attr->wq_type != IB_WQT_RQ) { pr_debug("unsupported wq type %d\n", init_attr->wq_type); return ERR_PTR(-EOPNOTSUPP); } - if (init_attr->create_flags & ~IB_WQ_FLAGS_SCATTER_FCS) { + if (init_attr->create_flags & ~IB_WQ_FLAGS_SCATTER_FCS || + !(dev->caps.flags & MLX4_DEV_CAP_FLAG_FCS_KEEP)) { pr_debug("unsupported create_flags %u\n", init_attr->create_flags); return ERR_PTR(-EOPNOTSUPP); @@ -4093,7 +4167,6 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd, qp->pri.vid = 0xFFFF; qp->alt.vid = 0xFFFF; - memset(&ib_qp_init_attr, 0, sizeof(ib_qp_init_attr)); ib_qp_init_attr.qp_context = init_attr->wq_context; ib_qp_init_attr.qp_type = IB_QPT_RAW_PACKET; ib_qp_init_attr.cap.max_recv_wr = init_attr->max_wr; @@ -4104,8 +4177,7 @@ struct ib_wq *mlx4_ib_create_wq(struct ib_pd *pd, if (init_attr->create_flags & IB_WQ_FLAGS_SCATTER_FCS) ib_qp_init_attr.create_flags |= IB_QP_CREATE_SCATTER_FCS; - err = create_qp_common(dev, pd, MLX4_IB_RWQ_SRC, &ib_qp_init_attr, - udata, 0, &qp); + err = create_rq(pd, &ib_qp_init_attr, udata, qp); if (err) { kfree(qp); return ERR_PTR(err); diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index 48968b4d561d..59022b744144 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -233,6 +233,8 @@ static bool is_legacy_obj_event_num(u16 event_num) case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: case MLX5_EVENT_TYPE_DCT_DRAINED: case MLX5_EVENT_TYPE_COMP: + case MLX5_EVENT_TYPE_DCT_KEY_VIOLATION: + case MLX5_EVENT_TYPE_XRQ_ERROR: return true; default: return false; @@ -315,8 +317,10 @@ static u16 get_event_obj_type(unsigned long event_type, struct mlx5_eqe *eqe) case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: return eqe->data.qp_srq.type; case MLX5_EVENT_TYPE_CQ_ERROR: + case MLX5_EVENT_TYPE_XRQ_ERROR: return 0; case MLX5_EVENT_TYPE_DCT_DRAINED: + case MLX5_EVENT_TYPE_DCT_KEY_VIOLATION: return MLX5_EVENT_QUEUE_TYPE_DCT; default: return MLX5_GET(affiliated_event_header, &eqe->data, obj_type); @@ -542,6 +546,8 @@ static u64 devx_get_obj_id(const void *in) break; case MLX5_CMD_OP_ARM_XRQ: case MLX5_CMD_OP_SET_XRQ_DC_PARAMS_ENTRY: + case MLX5_CMD_OP_RELEASE_XRQ_ERROR: + case MLX5_CMD_OP_MODIFY_XRQ: obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_XRQ, MLX5_GET(arm_xrq_in, in, xrqn)); break; @@ -776,6 +782,14 @@ static bool devx_is_obj_create_cmd(const void *in, u16 *opcode) return true; return false; } + case MLX5_CMD_OP_CREATE_PSV: + { + u8 num_psv = MLX5_GET(create_psv_in, in, num_psv); + + if (num_psv == 1) + return true; + return false; + } default: return false; } @@ -810,6 +824,8 @@ static bool devx_is_obj_modify_cmd(const void *in) case MLX5_CMD_OP_ARM_DCT_FOR_KEY_VIOLATION: case MLX5_CMD_OP_ARM_XRQ: case MLX5_CMD_OP_SET_XRQ_DC_PARAMS_ENTRY: + case MLX5_CMD_OP_RELEASE_XRQ_ERROR: + case MLX5_CMD_OP_MODIFY_XRQ: return true; case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY: { @@ -1216,6 +1232,12 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, case MLX5_CMD_OP_ALLOC_XRCD: MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DEALLOC_XRCD); break; + case MLX5_CMD_OP_CREATE_PSV: + MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_CMD_OP_DESTROY_PSV); + MLX5_SET(destroy_psv_in, din, psvn, + MLX5_GET(create_psv_out, out, psv0_index)); + break; default: /* The entry must match to one of the devx_is_obj_create_cmd */ WARN_ON(true); @@ -2027,7 +2049,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_SUBSCRIBE_EVENT)( event_sub->eventfd = eventfd_ctx_fdget(redirect_fd); - if (IS_ERR(event_sub)) { + if (IS_ERR(event_sub->eventfd)) { err = PTR_ERR(event_sub->eventfd); event_sub->eventfd = NULL; goto err; @@ -2286,7 +2308,11 @@ static u32 devx_get_obj_id_from_event(unsigned long event_type, void *data) case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: obj_id = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; break; + case MLX5_EVENT_TYPE_XRQ_ERROR: + obj_id = be32_to_cpu(eqe->data.xrq_err.type_xrqn) & 0xffffff; + break; case MLX5_EVENT_TYPE_DCT_DRAINED: + case MLX5_EVENT_TYPE_DCT_KEY_VIOLATION: obj_id = be32_to_cpu(eqe->data.dct.dctn) & 0xffffff; break; case MLX5_EVENT_TYPE_CQ_ERROR: @@ -2645,12 +2671,13 @@ static int devx_async_event_close(struct inode *inode, struct file *filp) struct devx_async_event_file *ev_file = filp->private_data; struct devx_event_subscription *event_sub, *event_sub_tmp; struct devx_async_event_data *entry, *tmp; + struct mlx5_ib_dev *dev = ev_file->dev; - mutex_lock(&ev_file->dev->devx_event_table.event_xa_lock); + mutex_lock(&dev->devx_event_table.event_xa_lock); /* delete the subscriptions which are related to this FD */ list_for_each_entry_safe(event_sub, event_sub_tmp, &ev_file->subscribed_events_list, file_list) { - devx_cleanup_subscription(ev_file->dev, event_sub); + devx_cleanup_subscription(dev, event_sub); if (event_sub->eventfd) eventfd_ctx_put(event_sub->eventfd); @@ -2659,7 +2686,7 @@ static int devx_async_event_close(struct inode *inode, struct file *filp) kfree_rcu(event_sub, rcu); } - mutex_unlock(&ev_file->dev->devx_event_table.event_xa_lock); + mutex_unlock(&dev->devx_event_table.event_xa_lock); /* free the pending events allocation */ if (!ev_file->omit_data) { @@ -2671,7 +2698,7 @@ static int devx_async_event_close(struct inode *inode, struct file *filp) } uverbs_close_fd(filp); - put_device(&ev_file->dev->ib_dev.dev); + put_device(&dev->ib_dev.dev); return 0; } diff --git a/drivers/infiniband/hw/mlx5/flow.c b/drivers/infiniband/hw/mlx5/flow.c index b8841355fcd5..571bb8539cdb 100644 --- a/drivers/infiniband/hw/mlx5/flow.c +++ b/drivers/infiniband/hw/mlx5/flow.c @@ -32,6 +32,9 @@ mlx5_ib_ft_type_to_namespace(enum mlx5_ib_uapi_flow_table_type table_type, case MLX5_IB_UAPI_FLOW_TABLE_TYPE_FDB: *namespace = MLX5_FLOW_NAMESPACE_FDB; break; + case MLX5_IB_UAPI_FLOW_TABLE_TYPE_RDMA_RX: + *namespace = MLX5_FLOW_NAMESPACE_RDMA_RX; + break; default: return -EINVAL; } @@ -101,6 +104,11 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && !dest_devx) return -EINVAL; + /* Allow only DEVX object or QP as dest when inserting to RDMA_RX */ + if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && + ((!dest_devx && !dest_qp) || (dest_devx && dest_qp))) + return -EINVAL; + if (dest_devx) { devx_obj = uverbs_attr_get_obj( attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); @@ -112,8 +120,9 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( */ if (!mlx5_ib_devx_is_flow_dest(devx_obj, &dest_id, &dest_type)) return -EINVAL; - /* Allow only flow table as dest when inserting to FDB */ - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB && + /* Allow only flow table as dest when inserting to FDB or RDMA_RX */ + if ((fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB || + fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) && dest_type != MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) return -EINVAL; } else if (dest_qp) { diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index c2a5780cb394..07aecba16019 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -535,7 +535,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, props->max_msg_sz = 1 << MLX5_CAP_GEN(dev->mdev, log_max_msg); props->pkey_tbl_len = 1; props->state = IB_PORT_DOWN; - props->phys_state = 3; + props->phys_state = IB_PORT_PHYS_STATE_DISABLED; mlx5_query_nic_vport_qkey_viol_cntr(mdev, &qkey_viol_cntr); props->qkey_viol_cntr = qkey_viol_cntr; @@ -561,7 +561,7 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, if (netif_running(ndev) && netif_carrier_ok(ndev)) { props->state = IB_PORT_ACTIVE; - props->phys_state = 5; + props->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } ndev_ib_mtu = iboe_get_mtu(ndev->mtu); @@ -1867,10 +1867,6 @@ static int mlx5_ib_alloc_ucontext(struct ib_ucontext *uctx, if (err) goto out_sys_pages; - if (ibdev->attrs.device_cap_flags & IB_DEVICE_ON_DEMAND_PAGING) - context->ibucontext.invalidate_range = - &mlx5_ib_invalidate_range; - if (req.flags & MLX5_IB_ALLOC_UCTX_DEVX) { err = mlx5_ib_devx_create(dev, true); if (err < 0) @@ -3967,6 +3963,11 @@ _get_flow_table(struct mlx5_ib_dev *dev, esw_encap) flags |= MLX5_FLOW_TABLE_TUNNEL_EN_REFORMAT; priority = FDB_BYPASS_PATH; + } else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) { + max_table_size = + BIT(MLX5_CAP_FLOWTABLE_RDMA_RX(dev->mdev, + log_max_ft_size)); + priority = fs_matcher->priority; } max_table_size = min_t(int, max_table_size, MLX5_FS_MAX_ENTRIES); @@ -3981,6 +3982,8 @@ _get_flow_table(struct mlx5_ib_dev *dev, prio = &dev->flow_db->egress_prios[priority]; else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_FDB) prio = &dev->flow_db->fdb; + else if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_RDMA_RX) + prio = &dev->flow_db->rdma_rx[priority]; if (!prio) return ERR_PTR(-EINVAL); @@ -5322,11 +5325,21 @@ static const struct mlx5_ib_counter ext_ppcnt_cnts[] = { INIT_EXT_PPCNT_COUNTER(rx_icrc_encapsulated), }; +static bool is_mdev_switchdev_mode(const struct mlx5_core_dev *mdev) +{ + return MLX5_ESWITCH_MANAGER(mdev) && + mlx5_ib_eswitch_mode(mdev->priv.eswitch) == + MLX5_ESWITCH_OFFLOADS; +} + static void mlx5_ib_dealloc_counters(struct mlx5_ib_dev *dev) { + int num_cnt_ports; int i; - for (i = 0; i < dev->num_ports; i++) { + num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; + + for (i = 0; i < num_cnt_ports; i++) { if (dev->port[i].cnts.set_id_valid) mlx5_core_dealloc_q_counter(dev->mdev, dev->port[i].cnts.set_id); @@ -5428,13 +5441,15 @@ static void mlx5_ib_fill_counters(struct mlx5_ib_dev *dev, static int mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev) { + int num_cnt_ports; int err = 0; int i; bool is_shared; is_shared = MLX5_CAP_GEN(dev->mdev, log_max_uctx) != 0; + num_cnt_ports = is_mdev_switchdev_mode(dev->mdev) ? 1 : dev->num_ports; - for (i = 0; i < dev->num_ports; i++) { + for (i = 0; i < num_cnt_ports; i++) { err = __mlx5_ib_alloc_counters(dev, &dev->port[i].cnts); if (err) goto err_alloc; @@ -5454,7 +5469,6 @@ static int mlx5_ib_alloc_counters(struct mlx5_ib_dev *dev) } dev->port[i].cnts.set_id_valid = true; } - return 0; err_alloc: @@ -5462,25 +5476,50 @@ err_alloc: return err; } +static const struct mlx5_ib_counters *get_counters(struct mlx5_ib_dev *dev, + u8 port_num) +{ + return is_mdev_switchdev_mode(dev->mdev) ? &dev->port[0].cnts : + &dev->port[port_num].cnts; +} + +/** + * mlx5_ib_get_counters_id - Returns counters id to use for device+port + * @dev: Pointer to mlx5 IB device + * @port_num: Zero based port number + * + * mlx5_ib_get_counters_id() Returns counters set id to use for given + * device port combination in switchdev and non switchdev mode of the + * parent device. + */ +u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num) +{ + const struct mlx5_ib_counters *cnts = get_counters(dev, port_num); + + return cnts->set_id; +} + static struct rdma_hw_stats *mlx5_ib_alloc_hw_stats(struct ib_device *ibdev, u8 port_num) { struct mlx5_ib_dev *dev = to_mdev(ibdev); - struct mlx5_ib_port *port = &dev->port[port_num - 1]; + const struct mlx5_ib_counters *cnts; + bool is_switchdev = is_mdev_switchdev_mode(dev->mdev); - /* We support only per port stats */ - if (port_num == 0) + if ((is_switchdev && port_num) || (!is_switchdev && !port_num)) return NULL; - return rdma_alloc_hw_stats_struct(port->cnts.names, - port->cnts.num_q_counters + - port->cnts.num_cong_counters + - port->cnts.num_ext_ppcnt_counters, + cnts = get_counters(dev, port_num - 1); + + return rdma_alloc_hw_stats_struct(cnts->names, + cnts->num_q_counters + + cnts->num_cong_counters + + cnts->num_ext_ppcnt_counters, RDMA_HW_STATS_DEFAULT_LIFESPAN); } static int mlx5_ib_query_q_counters(struct mlx5_core_dev *mdev, - struct mlx5_ib_port *port, + const struct mlx5_ib_counters *cnts, struct rdma_hw_stats *stats, u16 set_id) { @@ -5497,8 +5536,8 @@ static int mlx5_ib_query_q_counters(struct mlx5_core_dev *mdev, if (ret) goto free; - for (i = 0; i < port->cnts.num_q_counters; i++) { - val = *(__be32 *)(out + port->cnts.offsets[i]); + for (i = 0; i < cnts->num_q_counters; i++) { + val = *(__be32 *)(out + cnts->offsets[i]); stats->value[i] = (u64)be32_to_cpu(val); } @@ -5508,10 +5547,10 @@ free: } static int mlx5_ib_query_ext_ppcnt_counters(struct mlx5_ib_dev *dev, - struct mlx5_ib_port *port, - struct rdma_hw_stats *stats) + const struct mlx5_ib_counters *cnts, + struct rdma_hw_stats *stats) { - int offset = port->cnts.num_q_counters + port->cnts.num_cong_counters; + int offset = cnts->num_q_counters + cnts->num_cong_counters; int sz = MLX5_ST_SZ_BYTES(ppcnt_reg); int ret, i; void *out; @@ -5524,12 +5563,10 @@ static int mlx5_ib_query_ext_ppcnt_counters(struct mlx5_ib_dev *dev, if (ret) goto free; - for (i = 0; i < port->cnts.num_ext_ppcnt_counters; i++) { + for (i = 0; i < cnts->num_ext_ppcnt_counters; i++) stats->value[i + offset] = be64_to_cpup((__be64 *)(out + - port->cnts.offsets[i + offset])); - } - + cnts->offsets[i + offset])); free: kvfree(out); return ret; @@ -5540,7 +5577,7 @@ static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, u8 port_num, int index) { struct mlx5_ib_dev *dev = to_mdev(ibdev); - struct mlx5_ib_port *port = &dev->port[port_num - 1]; + const struct mlx5_ib_counters *cnts = get_counters(dev, port_num - 1); struct mlx5_core_dev *mdev; int ret, num_counters; u8 mdev_port_num; @@ -5548,18 +5585,17 @@ static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, if (!stats) return -EINVAL; - num_counters = port->cnts.num_q_counters + - port->cnts.num_cong_counters + - port->cnts.num_ext_ppcnt_counters; + num_counters = cnts->num_q_counters + + cnts->num_cong_counters + + cnts->num_ext_ppcnt_counters; /* q_counters are per IB device, query the master mdev */ - ret = mlx5_ib_query_q_counters(dev->mdev, port, stats, - port->cnts.set_id); + ret = mlx5_ib_query_q_counters(dev->mdev, cnts, stats, cnts->set_id); if (ret) return ret; if (MLX5_CAP_PCAM_FEATURE(dev->mdev, rx_icrc_encapsulated_counter)) { - ret = mlx5_ib_query_ext_ppcnt_counters(dev, port, stats); + ret = mlx5_ib_query_ext_ppcnt_counters(dev, cnts, stats); if (ret) return ret; } @@ -5576,10 +5612,10 @@ static int mlx5_ib_get_hw_stats(struct ib_device *ibdev, } ret = mlx5_lag_query_cong_counters(dev->mdev, stats->value + - port->cnts.num_q_counters, - port->cnts.num_cong_counters, - port->cnts.offsets + - port->cnts.num_q_counters); + cnts->num_q_counters, + cnts->num_cong_counters, + cnts->offsets + + cnts->num_q_counters); mlx5_ib_put_native_port_mdev(dev, port_num); if (ret) @@ -5594,20 +5630,22 @@ static struct rdma_hw_stats * mlx5_ib_counter_alloc_stats(struct rdma_counter *counter) { struct mlx5_ib_dev *dev = to_mdev(counter->device); - struct mlx5_ib_port *port = &dev->port[counter->port - 1]; + const struct mlx5_ib_counters *cnts = + get_counters(dev, counter->port - 1); /* Q counters are in the beginning of all counters */ - return rdma_alloc_hw_stats_struct(port->cnts.names, - port->cnts.num_q_counters, + return rdma_alloc_hw_stats_struct(cnts->names, + cnts->num_q_counters, RDMA_HW_STATS_DEFAULT_LIFESPAN); } static int mlx5_ib_counter_update_stats(struct rdma_counter *counter) { struct mlx5_ib_dev *dev = to_mdev(counter->device); - struct mlx5_ib_port *port = &dev->port[counter->port - 1]; + const struct mlx5_ib_counters *cnts = + get_counters(dev, counter->port - 1); - return mlx5_ib_query_q_counters(dev->mdev, port, + return mlx5_ib_query_q_counters(dev->mdev, cnts, counter->stats, counter->id); } @@ -5784,7 +5822,6 @@ static void init_delay_drop(struct mlx5_ib_dev *dev) mlx5_ib_warn(dev, "Failed to init delay drop debugfs\n"); } -/* The mlx5_ib_multiport_mutex should be held when calling this function */ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, struct mlx5_ib_multiport_info *mpi) { @@ -5794,6 +5831,8 @@ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, int err; int i; + lockdep_assert_held(&mlx5_ib_multiport_mutex); + mlx5_ib_cleanup_cong_debugfs(ibdev, port_num); spin_lock(&port->mp.mpi_lock); @@ -5802,13 +5841,12 @@ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, return; } - if (mpi->mdev_events.notifier_call) - mlx5_notifier_unregister(mpi->mdev, &mpi->mdev_events); - mpi->mdev_events.notifier_call = NULL; - mpi->ibdev = NULL; spin_unlock(&port->mp.mpi_lock); + if (mpi->mdev_events.notifier_call) + mlx5_notifier_unregister(mpi->mdev, &mpi->mdev_events); + mpi->mdev_events.notifier_call = NULL; mlx5_remove_netdev_notifier(ibdev, port_num); spin_lock(&port->mp.mpi_lock); @@ -5844,13 +5882,14 @@ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, ibdev->port[port_num].roce.last_port_state = IB_PORT_DOWN; } -/* The mlx5_ib_multiport_mutex should be held when calling this function */ static bool mlx5_ib_bind_slave_port(struct mlx5_ib_dev *ibdev, struct mlx5_ib_multiport_info *mpi) { u8 port_num = mlx5_core_native_port_num(mpi->mdev) - 1; int err; + lockdep_assert_held(&mlx5_ib_multiport_mutex); + spin_lock(&ibdev->port[port_num].mp.mpi_lock); if (ibdev->port[port_num].mp.mpi) { mlx5_ib_dbg(ibdev, "port %d already affiliated.\n", @@ -6933,7 +6972,7 @@ static void *mlx5_ib_add(struct mlx5_core_dev *mdev) dev->port = kcalloc(num_ports, sizeof(*dev->port), GFP_KERNEL); if (!dev->port) { - ib_dealloc_device((struct ib_device *)dev); + ib_dealloc_device(&dev->ib_dev); return NULL; } diff --git a/drivers/infiniband/hw/mlx5/mem.c b/drivers/infiniband/hw/mlx5/mem.c index fe1a76d8531c..b5aece786b36 100644 --- a/drivers/infiniband/hw/mlx5/mem.c +++ b/drivers/infiniband/hw/mlx5/mem.c @@ -56,18 +56,6 @@ void mlx5_ib_cont_pages(struct ib_umem *umem, u64 addr, struct scatterlist *sg; int entry; - if (umem->is_odp) { - unsigned int page_shift = to_ib_umem_odp(umem)->page_shift; - - *ncont = ib_umem_page_count(umem); - *count = *ncont << (page_shift - PAGE_SHIFT); - *shift = page_shift; - if (order) - *order = ilog2(roundup_pow_of_two(*ncont)); - - return; - } - addr = addr >> PAGE_SHIFT; tmp = (unsigned long)addr; m = find_first_bit(&tmp, BITS_PER_LONG); diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index c482f19958b3..6abfbf3a69b7 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -200,6 +200,7 @@ struct mlx5_ib_flow_db { struct mlx5_ib_flow_prio sniffer[MLX5_IB_NUM_SNIFFER_FTS]; struct mlx5_ib_flow_prio egress[MLX5_IB_NUM_EGRESS_FTS]; struct mlx5_ib_flow_prio fdb; + struct mlx5_ib_flow_prio rdma_rx[MLX5_IB_NUM_FLOW_FT]; struct mlx5_flow_table *lag_demux_ft; /* Protect flow steering bypass flow tables * when add/del flow rules. @@ -481,6 +482,7 @@ struct mlx5_umr_wr { u64 length; int access_flags; u32 mkey; + u8 ignore_free_state:1; }; static inline const struct mlx5_umr_wr *umr_wr(const struct ib_send_wr *wr) @@ -1474,4 +1476,5 @@ int bfregn_to_uar_index(struct mlx5_ib_dev *dev, bool dyn_bfreg); int mlx5_ib_qp_set_counter(struct ib_qp *qp, struct rdma_counter *counter); +u16 mlx5_ib_get_counters_id(struct mlx5_ib_dev *dev, u8 port_num); #endif /* MLX5_IB_H */ diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index 20ece6e0b2fc..b7da619614e4 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -51,22 +51,12 @@ static void clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr); static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr); static int mr_cache_max_order(struct mlx5_ib_dev *dev); static int unreg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr); -static bool umr_can_modify_entity_size(struct mlx5_ib_dev *dev) -{ - return !MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled); -} static bool umr_can_use_indirect_mkey(struct mlx5_ib_dev *dev) { return !MLX5_CAP_GEN(dev->mdev, umr_indirect_mkey_disabled); } -static bool use_umr(struct mlx5_ib_dev *dev, int order) -{ - return order <= mr_cache_max_order(dev) && - umr_can_modify_entity_size(dev); -} - static int destroy_mkey(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) { int err = mlx5_core_destroy_mkey(dev->mdev, &mr->mmkey); @@ -545,13 +535,16 @@ void mlx5_mr_cache_free(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) return; c = order2idx(dev, mr->order); - if (c < 0 || c >= MAX_MR_CACHE_ENTRIES) { - mlx5_ib_warn(dev, "order %d, cache index %d\n", mr->order, c); - return; - } + WARN_ON(c < 0 || c >= MAX_MR_CACHE_ENTRIES); - if (unreg_umr(dev, mr)) + if (unreg_umr(dev, mr)) { + mr->allocated_from_cache = false; + destroy_mkey(dev, mr); + ent = &cache->ent[c]; + if (ent->cur < ent->limit) + queue_work(cache->wq, &ent->work); return; + } ent = &cache->ent[c]; spin_lock_irq(&ent->lock); @@ -791,19 +784,37 @@ static int mr_umem_get(struct mlx5_ib_dev *dev, struct ib_udata *udata, int *ncont, int *order) { struct ib_umem *u; - int err; *umem = NULL; - u = ib_umem_get(udata, start, length, access_flags, 0); - err = PTR_ERR_OR_ZERO(u); - if (err) { - mlx5_ib_dbg(dev, "umem get failed (%d)\n", err); - return err; + if (access_flags & IB_ACCESS_ON_DEMAND) { + struct ib_umem_odp *odp; + + odp = ib_umem_odp_get(udata, start, length, access_flags); + if (IS_ERR(odp)) { + mlx5_ib_dbg(dev, "umem get failed (%ld)\n", + PTR_ERR(odp)); + return PTR_ERR(odp); + } + + u = &odp->umem; + + *page_shift = odp->page_shift; + *ncont = ib_umem_odp_num_pages(odp); + *npages = *ncont << (*page_shift - PAGE_SHIFT); + if (order) + *order = ilog2(roundup_pow_of_two(*ncont)); + } else { + u = ib_umem_get(udata, start, length, access_flags, 0); + if (IS_ERR(u)) { + mlx5_ib_dbg(dev, "umem get failed (%ld)\n", PTR_ERR(u)); + return PTR_ERR(u); + } + + mlx5_ib_cont_pages(u, start, MLX5_MKEY_PAGE_SHIFT_MASK, npages, + page_shift, ncont, order); } - mlx5_ib_cont_pages(u, start, MLX5_MKEY_PAGE_SHIFT_MASK, npages, - page_shift, ncont, order); if (!*npages) { mlx5_ib_warn(dev, "avoid zero region\n"); ib_umem_release(u); @@ -1268,7 +1279,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_ib_mr *mr = NULL; - bool populate_mtts = false; + bool use_umr; struct ib_umem *umem; int page_shift; int npages; @@ -1300,29 +1311,30 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, if (err < 0) return ERR_PTR(err); - if (use_umr(dev, order)) { + use_umr = !MLX5_CAP_GEN(dev->mdev, umr_modify_entity_size_disabled) && + (!MLX5_CAP_GEN(dev->mdev, umr_modify_atomic_disabled) || + !MLX5_CAP_GEN(dev->mdev, atomic)); + + if (order <= mr_cache_max_order(dev) && use_umr) { mr = alloc_mr_from_cache(pd, umem, virt_addr, length, ncont, page_shift, order, access_flags); if (PTR_ERR(mr) == -EAGAIN) { mlx5_ib_dbg(dev, "cache empty for order %d\n", order); mr = NULL; } - populate_mtts = false; } else if (!MLX5_CAP_GEN(dev->mdev, umr_extended_translation_offset)) { if (access_flags & IB_ACCESS_ON_DEMAND) { err = -EINVAL; pr_err("Got MR registration for ODP MR > 512MB, not supported for Connect-IB\n"); goto error; } - populate_mtts = true; + use_umr = false; } if (!mr) { - if (!umr_can_modify_entity_size(dev)) - populate_mtts = true; mutex_lock(&dev->slow_path_mutex); mr = reg_create(NULL, pd, virt_addr, length, umem, ncont, - page_shift, access_flags, populate_mtts); + page_shift, access_flags, !use_umr); mutex_unlock(&dev->slow_path_mutex); } @@ -1338,7 +1350,7 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, update_odp_mr(mr); - if (!populate_mtts) { + if (use_umr) { int update_xlt_flags = MLX5_IB_UPD_XLT_ENABLE; if (access_flags & IB_ACCESS_ON_DEMAND) @@ -1373,9 +1385,11 @@ static int unreg_umr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) return 0; umrwr.wr.send_flags = MLX5_IB_SEND_UMR_DISABLE_MR | - MLX5_IB_SEND_UMR_FAIL_IF_FREE; + MLX5_IB_SEND_UMR_UPDATE_PD_ACCESS; umrwr.wr.opcode = MLX5_IB_WR_UMR; + umrwr.pd = dev->umrc.pd; umrwr.mkey = mr->mmkey.key; + umrwr.ignore_free_state = 1; return mlx5_ib_post_send_wait(dev, &umrwr); } @@ -1577,10 +1591,10 @@ static void clean_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) mr->sig = NULL; } - mlx5_free_priv_descs(mr); - - if (!allocated_from_cache) + if (!allocated_from_cache) { destroy_mkey(dev, mr); + mlx5_free_priv_descs(mr); + } } static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) @@ -1604,7 +1618,7 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) /* Wait for all running page-fault handlers to finish. */ synchronize_srcu(&dev->mr_srcu); /* Destroy all page mappings */ - if (umem_odp->page_list) + if (!umem_odp->is_implicit_odp) mlx5_ib_invalidate_range(umem_odp, ib_umem_start(umem_odp), ib_umem_end(umem_odp)); @@ -1615,7 +1629,7 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) * so that there will not be any invalidations in * flight, looking at the *mr struct. */ - ib_umem_release(umem); + ib_umem_odp_release(umem_odp); atomic_sub(npages, &dev->mdev->priv.reg_pages); /* Avoid double-freeing the umem. */ diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 429c72407d7f..905936423a03 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -184,7 +184,7 @@ void mlx5_odp_populate_klm(struct mlx5_klm *pklm, size_t offset, for (i = 0; i < nentries; i++, pklm++) { pklm->bcount = cpu_to_be32(MLX5_IMR_MTT_SIZE); va = (offset + i) * MLX5_IMR_MTT_SIZE; - if (odp && odp->umem.address == va) { + if (odp && ib_umem_start(odp) == va) { struct mlx5_ib_mr *mtt = odp->private; pklm->key = cpu_to_be32(mtt->ibmr.lkey); @@ -206,7 +206,7 @@ static void mr_leaf_free_action(struct work_struct *work) mr->parent = NULL; synchronize_srcu(&mr->dev->mr_srcu); - ib_umem_release(&odp->umem); + ib_umem_odp_release(odp); if (imr->live) mlx5_ib_update_xlt(imr, idx, 1, 0, MLX5_IB_UPD_XLT_INDIRECT | @@ -246,7 +246,7 @@ void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start, * overwrite the same MTTs. Concurent invalidations might race us, * but they will write 0s as well, so no difference in the end result. */ - + mutex_lock(&umem_odp->umem_mutex); for (addr = start; addr < end; addr += BIT(umem_odp->page_shift)) { idx = (addr - ib_umem_start(umem_odp)) >> umem_odp->page_shift; /* @@ -278,6 +278,7 @@ void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start, idx - blk_start_idx + 1, 0, MLX5_IB_UPD_XLT_ZAP | MLX5_IB_UPD_XLT_ATOMIC); + mutex_unlock(&umem_odp->umem_mutex); /* * We are now sure that the device will not access the * memory. We can safely unmap it, and mark it as dirty if @@ -383,7 +384,7 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev, } static struct mlx5_ib_mr *implicit_mr_alloc(struct ib_pd *pd, - struct ib_umem *umem, + struct ib_umem_odp *umem_odp, bool ksm, int access_flags) { struct mlx5_ib_dev *dev = to_mdev(pd->device); @@ -401,7 +402,7 @@ static struct mlx5_ib_mr *implicit_mr_alloc(struct ib_pd *pd, mr->dev = dev; mr->access_flags = access_flags; mr->mmkey.iova = 0; - mr->umem = umem; + mr->umem = &umem_odp->umem; if (ksm) { err = mlx5_ib_update_xlt(mr, 0, @@ -461,18 +462,17 @@ next_mr: if (nentries) nentries++; } else { - odp = ib_alloc_odp_umem(odp_mr, addr, - MLX5_IMR_MTT_SIZE); + odp = ib_umem_odp_alloc_child(odp_mr, addr, MLX5_IMR_MTT_SIZE); if (IS_ERR(odp)) { mutex_unlock(&odp_mr->umem_mutex); return ERR_CAST(odp); } - mtt = implicit_mr_alloc(mr->ibmr.pd, &odp->umem, 0, + mtt = implicit_mr_alloc(mr->ibmr.pd, odp, 0, mr->access_flags); if (IS_ERR(mtt)) { mutex_unlock(&odp_mr->umem_mutex); - ib_umem_release(&odp->umem); + ib_umem_odp_release(odp); return ERR_CAST(mtt); } @@ -494,7 +494,7 @@ next_mr: addr += MLX5_IMR_MTT_SIZE; if (unlikely(addr < io_virt + bcnt)) { odp = odp_next(odp); - if (odp && odp->umem.address != addr) + if (odp && ib_umem_start(odp) != addr) odp = NULL; goto next_mr; } @@ -518,19 +518,19 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd, int access_flags) { struct mlx5_ib_mr *imr; - struct ib_umem *umem; + struct ib_umem_odp *umem_odp; - umem = ib_umem_get(udata, 0, 0, access_flags, 0); - if (IS_ERR(umem)) - return ERR_CAST(umem); + umem_odp = ib_umem_odp_alloc_implicit(udata, access_flags); + if (IS_ERR(umem_odp)) + return ERR_CAST(umem_odp); - imr = implicit_mr_alloc(&pd->ibpd, umem, 1, access_flags); + imr = implicit_mr_alloc(&pd->ibpd, umem_odp, 1, access_flags); if (IS_ERR(imr)) { - ib_umem_release(umem); + ib_umem_odp_release(umem_odp); return ERR_CAST(imr); } - imr->umem = umem; + imr->umem = &umem_odp->umem; init_waitqueue_head(&imr->q_leaf_free); atomic_set(&imr->num_leaf_free, 0); atomic_set(&imr->num_pending_prefetch, 0); @@ -538,34 +538,31 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd, return imr; } -static int mr_leaf_free(struct ib_umem_odp *umem_odp, u64 start, u64 end, - void *cookie) +void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr) { - struct mlx5_ib_mr *mr = umem_odp->private, *imr = cookie; - - if (mr->parent != imr) - return 0; + struct ib_ucontext_per_mm *per_mm = mr_to_per_mm(imr); + struct rb_node *node; - ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp), - ib_umem_end(umem_odp)); + down_read(&per_mm->umem_rwsem); + for (node = rb_first_cached(&per_mm->umem_tree); node; + node = rb_next(node)) { + struct ib_umem_odp *umem_odp = + rb_entry(node, struct ib_umem_odp, interval_tree.rb); + struct mlx5_ib_mr *mr = umem_odp->private; - if (umem_odp->dying) - return 0; + if (mr->parent != imr) + continue; - WRITE_ONCE(umem_odp->dying, 1); - atomic_inc(&imr->num_leaf_free); - schedule_work(&umem_odp->work); + ib_umem_odp_unmap_dma_pages(umem_odp, ib_umem_start(umem_odp), + ib_umem_end(umem_odp)); - return 0; -} - -void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr) -{ - struct ib_ucontext_per_mm *per_mm = mr_to_per_mm(imr); + if (umem_odp->dying) + continue; - down_read(&per_mm->umem_rwsem); - rbt_ib_umem_for_each_in_range(&per_mm->umem_tree, 0, ULLONG_MAX, - mr_leaf_free, true, imr); + WRITE_ONCE(umem_odp->dying, 1); + atomic_inc(&imr->num_leaf_free); + schedule_work(&umem_odp->work); + } up_read(&per_mm->umem_rwsem); wait_event(imr->q_leaf_free, !atomic_read(&imr->num_leaf_free)); @@ -578,7 +575,6 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, u32 flags) { int npages = 0, current_seq, page_shift, ret, np; - bool implicit = false; struct ib_umem_odp *odp_mr = to_ib_umem_odp(mr->umem); bool downgrade = flags & MLX5_PF_FLAGS_DOWNGRADE; bool prefetch = flags & MLX5_PF_FLAGS_PREFETCH; @@ -587,13 +583,12 @@ static int pagefault_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, struct ib_umem_odp *odp; size_t size; - if (!odp_mr->page_list) { + if (odp_mr->is_implicit_odp) { odp = implicit_mr_get_data(mr, io_virt, bcnt); if (IS_ERR(odp)) return PTR_ERR(odp); mr = odp->private; - implicit = true; } else { odp = odp_mr; } @@ -606,7 +601,7 @@ next_mr: start_idx = (io_virt - (mr->mmkey.iova & page_mask)) >> page_shift; access_mask = ODP_READ_ALLOWED_BIT; - if (prefetch && !downgrade && !mr->umem->writable) { + if (prefetch && !downgrade && !odp->umem.writable) { /* prefetch with write-access must * be supported by the MR */ @@ -614,7 +609,7 @@ next_mr: goto out; } - if (mr->umem->writable && !downgrade) + if (odp->umem.writable && !downgrade) access_mask |= ODP_WRITE_ALLOWED_BIT; current_seq = READ_ONCE(odp->notifiers_seq); @@ -624,8 +619,8 @@ next_mr: */ smp_rmb(); - ret = ib_umem_odp_map_dma_pages(to_ib_umem_odp(mr->umem), io_virt, size, - access_mask, current_seq); + ret = ib_umem_odp_map_dma_pages(odp, io_virt, size, access_mask, + current_seq); if (ret < 0) goto out; @@ -633,8 +628,7 @@ next_mr: np = ret; mutex_lock(&odp->umem_mutex); - if (!ib_umem_mmu_notifier_retry(to_ib_umem_odp(mr->umem), - current_seq)) { + if (!ib_umem_mmu_notifier_retry(odp, current_seq)) { /* * No need to check whether the MTTs really belong to * this MR, since ib_umem_odp_map_dma_pages already @@ -667,7 +661,7 @@ next_mr: io_virt += size; next = odp_next(odp); - if (unlikely(!next || next->umem.address != io_virt)) { + if (unlikely(!next || ib_umem_start(next) != io_virt)) { mlx5_ib_dbg(dev, "next implicit leaf removed at 0x%llx. got %p\n", io_virt, next); return -EAGAIN; @@ -681,19 +675,15 @@ next_mr: out: if (ret == -EAGAIN) { - if (implicit || !odp->dying) { - unsigned long timeout = - msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT); - - if (!wait_for_completion_timeout( - &odp->notifier_completion, - timeout)) { - mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n", - current_seq, odp->notifiers_seq, odp->notifiers_count); - } - } else { - /* The MR is being killed, kill the QP as well. */ - ret = -EFAULT; + unsigned long timeout = msecs_to_jiffies(MMU_NOTIFIER_TIMEOUT); + + if (!wait_for_completion_timeout(&odp->notifier_completion, + timeout)) { + mlx5_ib_warn( + dev, + "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n", + current_seq, odp->notifiers_seq, + odp->notifiers_count); } } @@ -1002,9 +992,6 @@ static int mlx5_ib_mr_initiator_pfault_handler( u16 wqe_index = pfault->wqe.wqe_index; struct mlx5_base_av *av; unsigned ds, opcode; -#if defined(DEBUG) - u32 ctrl_wqe_index, ctrl_qpn; -#endif u32 qpn = qp->trans_qp.base.mqp.qpn; ds = be32_to_cpu(ctrl->qpn_ds) & MLX5_WQE_CTRL_DS_MASK; @@ -1020,27 +1007,6 @@ static int mlx5_ib_mr_initiator_pfault_handler( return -EFAULT; } -#if defined(DEBUG) - ctrl_wqe_index = (be32_to_cpu(ctrl->opmod_idx_opcode) & - MLX5_WQE_CTRL_WQE_INDEX_MASK) >> - MLX5_WQE_CTRL_WQE_INDEX_SHIFT; - if (wqe_index != ctrl_wqe_index) { - mlx5_ib_err(dev, "Got WQE with invalid wqe_index. wqe_index=0x%x, qpn=0x%x ctrl->wqe_index=0x%x\n", - wqe_index, qpn, - ctrl_wqe_index); - return -EFAULT; - } - - ctrl_qpn = (be32_to_cpu(ctrl->qpn_ds) & MLX5_WQE_CTRL_QPN_MASK) >> - MLX5_WQE_CTRL_QPN_SHIFT; - if (qpn != ctrl_qpn) { - mlx5_ib_err(dev, "Got WQE with incorrect QP number. wqe_index=0x%x, qpn=0x%x ctrl->qpn=0x%x\n", - wqe_index, qpn, - ctrl_qpn); - return -EFAULT; - } -#endif /* DEBUG */ - *wqe_end = *wqe + ds * MLX5_WQE_DS_UNITS; *wqe += sizeof(*ctrl); @@ -1576,6 +1542,7 @@ void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) static const struct ib_device_ops mlx5_ib_dev_odp_ops = { .advise_mr = mlx5_ib_advise_mr, + .invalidate_range = mlx5_ib_invalidate_range, }; int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) @@ -1726,7 +1693,7 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *work) num_pending_prefetch_dec(to_mdev(w->pd->device), w->sg_list, w->num_sge, 0); - kfree(w); + kvfree(w); } int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, @@ -1768,7 +1735,7 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, if (valid_req) queue_work(system_unbound_wq, &work->work); else - kfree(work); + kvfree(work); srcu_read_unlock(&dev->mr_srcu, srcu_key); diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 2a97619ed603..e96dcdfd6c3a 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -1713,7 +1713,6 @@ static int create_rss_raw_qp_tir(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, } MLX5_SET(tirc, tirc, rx_hash_fn, MLX5_RX_HASH_FN_TOEPLITZ); - MLX5_SET(tirc, tirc, rx_hash_symmetric, 1); memcpy(rss_key, ucmd.rx_hash_key, len); break; } @@ -3387,19 +3386,16 @@ static int __mlx5_ib_qp_set_counter(struct ib_qp *qp, struct mlx5_ib_dev *dev = to_mdev(qp->device); struct mlx5_ib_qp *mqp = to_mqp(qp); struct mlx5_qp_context context = {}; - struct mlx5_ib_port *mibport = NULL; struct mlx5_ib_qp_base *base; u32 set_id; if (!MLX5_CAP_GEN(dev->mdev, rts2rts_qp_counters_set_id)) return 0; - if (counter) { + if (counter) set_id = counter->id; - } else { - mibport = &dev->port[mqp->port - 1]; - set_id = mibport->cnts.set_id; - } + else + set_id = mlx5_ib_get_counters_id(dev, mqp->port - 1); base = &mqp->trans_qp.base; context.qp_counter_set_usr_page &= cpu_to_be32(0xffffff); @@ -3460,7 +3456,6 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, struct mlx5_ib_cq *send_cq, *recv_cq; struct mlx5_qp_context *context; struct mlx5_ib_pd *pd; - struct mlx5_ib_port *mibport = NULL; enum mlx5_qp_state mlx5_cur, mlx5_new; enum mlx5_qp_optpar optpar; u32 set_id = 0; @@ -3625,11 +3620,10 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, if (qp->flags & MLX5_IB_QP_UNDERLAY) port_num = 0; - mibport = &dev->port[port_num]; if (ibqp->counter) set_id = ibqp->counter->id; else - set_id = mibport->cnts.set_id; + set_id = mlx5_ib_get_counters_id(dev, port_num); context->qp_counter_set_usr_page |= cpu_to_be32(set_id << 24); } @@ -3818,6 +3812,8 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, dctc = MLX5_ADDR_OF(create_dct_in, qp->dct.in, dct_context_entry); if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { + u16 set_id; + required |= IB_QP_ACCESS_FLAGS | IB_QP_PKEY_INDEX | IB_QP_PORT; if (!is_valid_mask(attr_mask, required, 0)) return -EINVAL; @@ -3844,7 +3840,9 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, } MLX5_SET(dctc, dctc, pkey_index, attr->pkey_index); MLX5_SET(dctc, dctc, port, attr->port_num); - MLX5_SET(dctc, dctc, counter_set_id, dev->port[attr->port_num - 1].cnts.set_id); + + set_id = mlx5_ib_get_counters_id(dev, attr->port_num - 1); + MLX5_SET(dctc, dctc, counter_set_id, set_id); } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { struct mlx5_ib_modify_qp_resp resp = {}; @@ -4295,10 +4293,14 @@ static int set_reg_umr_segment(struct mlx5_ib_dev *dev, memset(umr, 0, sizeof(*umr)); - if (wr->send_flags & MLX5_IB_SEND_UMR_FAIL_IF_FREE) - umr->flags = MLX5_UMR_CHECK_FREE; /* fail if free */ - else - umr->flags = MLX5_UMR_CHECK_NOT_FREE; /* fail if not free */ + if (!umrwr->ignore_free_state) { + if (wr->send_flags & MLX5_IB_SEND_UMR_FAIL_IF_FREE) + /* fail if free */ + umr->flags = MLX5_UMR_CHECK_FREE; + else + /* fail if not free */ + umr->flags = MLX5_UMR_CHECK_NOT_FREE; + } umr->xlt_octowords = cpu_to_be16(get_xlt_octo(umrwr->xlt_size)); if (wr->send_flags & MLX5_IB_SEND_UMR_UPDATE_XLT) { @@ -6328,11 +6330,13 @@ int mlx5_ib_modify_wq(struct ib_wq *wq, struct ib_wq_attr *wq_attr, } if (curr_wq_state == IB_WQS_RESET && wq_state == IB_WQS_RDY) { + u16 set_id; + + set_id = mlx5_ib_get_counters_id(dev, 0); if (MLX5_CAP_GEN(dev->mdev, modify_rq_counter_set_id)) { MLX5_SET64(modify_rq_in, in, modify_bitmask, MLX5_MODIFY_RQ_IN_MODIFY_BITMASK_RQ_COUNTER_SET_ID); - MLX5_SET(rqc, rqc, counter_set_id, - dev->port->cnts.set_id); + MLX5_SET(rqc, rqc, counter_set_id, set_id); } else dev_info_once( &dev->ib_dev.dev, diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c index bccc11378109..e8267e590772 100644 --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c @@ -163,10 +163,10 @@ int ocrdma_query_port(struct ib_device *ibdev, netdev = dev->nic_info.netdev; if (netif_running(netdev) && netif_oper_up(netdev)) { port_state = IB_PORT_ACTIVE; - props->phys_state = 5; + props->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } else { port_state = IB_PORT_DOWN; - props->phys_state = 3; + props->phys_state = IB_PORT_PHYS_STATE_DISABLED; } props->max_mtu = IB_MTU_4096; props->active_mtu = iboe_get_mtu(netdev->mtu); diff --git a/drivers/infiniband/hw/qedr/main.c b/drivers/infiniband/hw/qedr/main.c index 533157a2a3be..5136b835e1ba 100644 --- a/drivers/infiniband/hw/qedr/main.c +++ b/drivers/infiniband/hw/qedr/main.c @@ -125,14 +125,20 @@ static ssize_t hw_rev_show(struct device *device, struct device_attribute *attr, struct qedr_dev *dev = rdma_device_to_drv_device(device, struct qedr_dev, ibdev); - return scnprintf(buf, PAGE_SIZE, "0x%x\n", dev->pdev->vendor); + return scnprintf(buf, PAGE_SIZE, "0x%x\n", dev->attr.hw_ver); } static DEVICE_ATTR_RO(hw_rev); static ssize_t hca_type_show(struct device *device, struct device_attribute *attr, char *buf) { - return scnprintf(buf, PAGE_SIZE, "%s\n", "HCA_TYPE_TO_SET"); + struct qedr_dev *dev = + rdma_device_to_drv_device(device, struct qedr_dev, ibdev); + + return scnprintf(buf, PAGE_SIZE, "FastLinQ QL%x %s\n", + dev->pdev->device, + rdma_protocol_iwarp(&dev->ibdev, 1) ? + "iWARP" : "RoCE"); } static DEVICE_ATTR_RO(hca_type); @@ -820,7 +826,7 @@ static int qedr_init_hw(struct qedr_dev *dev) if (rc) goto out; - dev->db_addr = (void __iomem *)(uintptr_t)out_params.dpi_addr; + dev->db_addr = out_params.dpi_addr; dev->db_phys_addr = out_params.dpi_phys_addr; dev->db_size = out_params.dpi_size; dev->dpi = out_params.dpi; diff --git a/drivers/infiniband/hw/qedr/qedr.h b/drivers/infiniband/hw/qedr/qedr.h index a92ca22e5de1..0cfd849b13d6 100644 --- a/drivers/infiniband/hw/qedr/qedr.h +++ b/drivers/infiniband/hw/qedr/qedr.h @@ -229,7 +229,7 @@ struct qedr_ucontext { struct ib_ucontext ibucontext; struct qedr_dev *dev; struct qedr_pd *pd; - u64 dpi_addr; + void __iomem *dpi_addr; u64 dpi_phys_addr; u32 dpi_size; u16 dpi; diff --git a/drivers/infiniband/hw/qedr/verbs.c b/drivers/infiniband/hw/qedr/verbs.c index 27d90a84ea01..6f3ce86019b7 100644 --- a/drivers/infiniband/hw/qedr/verbs.c +++ b/drivers/infiniband/hw/qedr/verbs.c @@ -221,10 +221,10 @@ int qedr_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *attr) /* *attr being zeroed by the caller, avoid zeroing it here */ if (rdma_port->port_state == QED_RDMA_PORT_UP) { attr->state = IB_PORT_ACTIVE; - attr->phys_state = 5; + attr->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } else { attr->state = IB_PORT_DOWN; - attr->phys_state = 3; + attr->phys_state = IB_PORT_PHYS_STATE_DISABLED; } attr->max_mtu = IB_MTU_4096; attr->active_mtu = iboe_get_mtu(dev->ndev->mtu); @@ -2451,7 +2451,6 @@ int qedr_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) struct qedr_dev *dev = qp->dev; struct ib_qp_attr attr; int attr_mask = 0; - int rc = 0; DP_DEBUG(dev, QEDR_MSG_QP, "destroy qp: destroying %p, qp type=%d\n", qp, qp->qp_type); @@ -2496,7 +2495,7 @@ int qedr_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) xa_erase_irq(&dev->qps, qp->qp_id); kfree(qp); } - return rc; + return 0; } int qedr_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *attr, u32 flags, diff --git a/drivers/infiniband/hw/qib/qib_file_ops.c b/drivers/infiniband/hw/qib/qib_file_ops.c index 27b6e664e59d..b0144229cf3b 100644 --- a/drivers/infiniband/hw/qib/qib_file_ops.c +++ b/drivers/infiniband/hw/qib/qib_file_ops.c @@ -1789,7 +1789,6 @@ static void unlock_expected_tids(struct qib_ctxtdata *rcd) static int qib_close(struct inode *in, struct file *fp) { - int ret = 0; struct qib_filedata *fd; struct qib_ctxtdata *rcd; struct qib_devdata *dd; @@ -1873,7 +1872,7 @@ static int qib_close(struct inode *in, struct file *fp) bail: kfree(fd); - return ret; + return 0; } static int qib_ctxt_info(struct file *fp, struct qib_ctxt_info __user *uinfo) diff --git a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c index eeb07b245ef9..556b8e44a51c 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_verbs.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_verbs.c @@ -194,7 +194,7 @@ find_free_vf_and_create_qp_grp(struct usnic_ib_dev *us_ibdev, return ERR_CAST(dev_list); for (i = 0; dev_list[i]; i++) { dev = dev_list[i]; - vf = pci_get_drvdata(to_pci_dev(dev)); + vf = dev_get_drvdata(dev); spin_lock(&vf->lock); vnic = vf->vnic; if (!usnic_vnic_check_room(vnic, res_spec)) { @@ -356,13 +356,14 @@ int usnic_ib_query_port(struct ib_device *ibdev, u8 port, if (!us_ibdev->ufdev->link_up) { props->state = IB_PORT_DOWN; - props->phys_state = 3; + props->phys_state = IB_PORT_PHYS_STATE_DISABLED; } else if (!us_ibdev->ufdev->inaddr) { props->state = IB_PORT_INIT; - props->phys_state = 4; + props->phys_state = + IB_PORT_PHYS_STATE_PORT_CONFIGURATION_TRAINING; } else { props->state = IB_PORT_ACTIVE; - props->phys_state = 5; + props->phys_state = IB_PORT_PHYS_STATE_LINK_UP; } props->port_cap_flags = 0; diff --git a/drivers/infiniband/sw/rxe/rxe.h b/drivers/infiniband/sw/rxe/rxe.h index ecf6e659c0da..fb07eed9e402 100644 --- a/drivers/infiniband/sw/rxe/rxe.h +++ b/drivers/infiniband/sw/rxe/rxe.h @@ -65,10 +65,6 @@ */ #define RXE_UVERBS_ABI_VERSION 2 -#define RDMA_LINK_PHYS_STATE_LINK_UP (5) -#define RDMA_LINK_PHYS_STATE_DISABLED (3) -#define RDMA_LINK_PHYS_STATE_POLLING (2) - #define RXE_ROCE_V2_SPORT (0xc000) static inline u32 rxe_crc32(struct rxe_dev *rxe, diff --git a/drivers/infiniband/sw/rxe/rxe_param.h b/drivers/infiniband/sw/rxe/rxe_param.h index 1abed47ca221..fe5207386700 100644 --- a/drivers/infiniband/sw/rxe/rxe_param.h +++ b/drivers/infiniband/sw/rxe/rxe_param.h @@ -154,7 +154,7 @@ enum rxe_port_param { RXE_PORT_ACTIVE_WIDTH = IB_WIDTH_1X, RXE_PORT_ACTIVE_SPEED = 1, RXE_PORT_PKEY_TBL_LEN = 64, - RXE_PORT_PHYS_STATE = 2, + RXE_PORT_PHYS_STATE = IB_PORT_PHYS_STATE_POLLING, RXE_PORT_SUBNET_PREFIX = 0xfe80000000000000ULL, }; diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c index 4ebdfcf4d33e..623129f27f5a 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.c +++ b/drivers/infiniband/sw/rxe/rxe_verbs.c @@ -69,11 +69,11 @@ static int rxe_query_port(struct ib_device *dev, &attr->active_width); if (attr->state == IB_PORT_ACTIVE) - attr->phys_state = RDMA_LINK_PHYS_STATE_LINK_UP; + attr->phys_state = IB_PORT_PHYS_STATE_LINK_UP; else if (dev_get_flags(rxe->ndev) & IFF_UP) - attr->phys_state = RDMA_LINK_PHYS_STATE_POLLING; + attr->phys_state = IB_PORT_PHYS_STATE_POLLING; else - attr->phys_state = RDMA_LINK_PHYS_STATE_DISABLED; + attr->phys_state = IB_PORT_PHYS_STATE_DISABLED; mutex_unlock(&rxe->usdev_lock); diff --git a/drivers/infiniband/sw/siw/Kconfig b/drivers/infiniband/sw/siw/Kconfig index dace276aea14..b622fc62f2cd 100644 --- a/drivers/infiniband/sw/siw/Kconfig +++ b/drivers/infiniband/sw/siw/Kconfig @@ -1,6 +1,6 @@ config RDMA_SIW tristate "Software RDMA over TCP/IP (iWARP) driver" - depends on INET && INFINIBAND && LIBCRC32C && 64BIT + depends on INET && INFINIBAND && LIBCRC32C select DMA_VIRT_OPS help This driver implements the iWARP RDMA transport over diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h index 03fd7b2f595f..77b1aabf6ff3 100644 --- a/drivers/infiniband/sw/siw/siw.h +++ b/drivers/infiniband/sw/siw/siw.h @@ -214,7 +214,7 @@ struct siw_wqe { struct siw_cq { struct ib_cq base_cq; spinlock_t lock; - u64 *notify; + struct siw_cq_ctrl *notify; struct siw_cqe *queue; u32 cq_put; u32 cq_get; diff --git a/drivers/infiniband/sw/siw/siw_cm.c b/drivers/infiniband/sw/siw/siw_cm.c index a7cde98e73e8..9ce8a1b925d2 100644 --- a/drivers/infiniband/sw/siw/siw_cm.c +++ b/drivers/infiniband/sw/siw/siw_cm.c @@ -220,13 +220,12 @@ static void siw_put_work(struct siw_cm_work *work) static void siw_cep_set_inuse(struct siw_cep *cep) { unsigned long flags; - int rv; retry: spin_lock_irqsave(&cep->lock, flags); if (cep->in_use) { spin_unlock_irqrestore(&cep->lock, flags); - rv = wait_event_interruptible(cep->waitq, !cep->in_use); + wait_event_interruptible(cep->waitq, !cep->in_use); if (signal_pending(current)) flush_signals(current); goto retry; diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c index f55c4e80aea4..05a92f997f60 100644 --- a/drivers/infiniband/sw/siw/siw_main.c +++ b/drivers/infiniband/sw/siw/siw_main.c @@ -160,10 +160,8 @@ static int siw_init_cpulist(void) out_err: siw_cpu_info.num_nodes = 0; - while (i) { + while (--i >= 0) kfree(siw_cpu_info.tx_valid_cpus[i]); - siw_cpu_info.tx_valid_cpus[i--] = NULL; - } kfree(siw_cpu_info.tx_valid_cpus); siw_cpu_info.tx_valid_cpus = NULL; @@ -612,6 +610,7 @@ static __init int siw_init_module(void) if (!siw_create_tx_threads()) { pr_info("siw: Could not start any TX thread\n"); + rv = -ENOMEM; goto out_error; } /* diff --git a/drivers/infiniband/sw/siw/siw_qp.c b/drivers/infiniband/sw/siw/siw_qp.c index 11383d9f95ef..0990307c5d2c 100644 --- a/drivers/infiniband/sw/siw/siw_qp.c +++ b/drivers/infiniband/sw/siw/siw_qp.c @@ -220,12 +220,14 @@ static int siw_qp_enable_crc(struct siw_qp *qp) { struct siw_rx_stream *c_rx = &qp->rx_stream; struct siw_iwarp_tx *c_tx = &qp->tx_ctx; - int size = crypto_shash_descsize(siw_crypto_shash) + - sizeof(struct shash_desc); + int size; if (siw_crypto_shash == NULL) return -ENOENT; + size = crypto_shash_descsize(siw_crypto_shash) + + sizeof(struct shash_desc); + c_tx->mpa_crc_hd = kzalloc(size, GFP_KERNEL); c_rx->mpa_crc_hd = kzalloc(size, GFP_KERNEL); if (!c_tx->mpa_crc_hd || !c_rx->mpa_crc_hd) { @@ -1011,18 +1013,24 @@ out: */ static bool siw_cq_notify_now(struct siw_cq *cq, u32 flags) { - u64 cq_notify; + u32 cq_notify; if (!cq->base_cq.comp_handler) return false; - cq_notify = READ_ONCE(*cq->notify); + /* Read application shared notification state */ + cq_notify = READ_ONCE(cq->notify->flags); if ((cq_notify & SIW_NOTIFY_NEXT_COMPLETION) || ((cq_notify & SIW_NOTIFY_SOLICITED) && (flags & SIW_WQE_SOLICITED))) { - /* dis-arm CQ */ - smp_store_mb(*cq->notify, SIW_NOTIFY_NOT); + /* + * CQ notification is one-shot: Since the + * current CQE causes user notification, + * the CQ gets dis-aremd and must be re-aremd + * by the user for a new notification. + */ + WRITE_ONCE(cq->notify->flags, SIW_NOTIFY_NOT); return true; } diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c index 32dc79d0e898..03176a3d1e18 100644 --- a/drivers/infiniband/sw/siw/siw_verbs.c +++ b/drivers/infiniband/sw/siw/siw_verbs.c @@ -206,7 +206,8 @@ int siw_query_port(struct ib_device *base_dev, u8 port, attr->gid_tbl_len = 1; attr->max_msg_sz = -1; attr->max_mtu = ib_mtu_int_to_enum(sdev->netdev->mtu); - attr->phys_state = sdev->state == IB_PORT_ACTIVE ? 5 : 3; + attr->phys_state = sdev->state == IB_PORT_ACTIVE ? + IB_PORT_PHYS_STATE_LINK_UP : IB_PORT_PHYS_STATE_DISABLED; attr->pkey_tbl_len = 1; attr->port_cap_flags = IB_PORT_CM_SUP | IB_PORT_DEVICE_MGMT_SUP; attr->state = sdev->state; @@ -1049,7 +1050,7 @@ int siw_create_cq(struct ib_cq *base_cq, const struct ib_cq_init_attr *attr, spin_lock_init(&cq->lock); - cq->notify = &((struct siw_cq_ctrl *)&cq->queue[size])->notify; + cq->notify = (struct siw_cq_ctrl *)&cq->queue[size]; if (udata) { struct siw_uresp_create_cq uresp = {}; @@ -1141,11 +1142,17 @@ int siw_req_notify_cq(struct ib_cq *base_cq, enum ib_cq_notify_flags flags) siw_dbg_cq(cq, "flags: 0x%02x\n", flags); if ((flags & IB_CQ_SOLICITED_MASK) == IB_CQ_SOLICITED) - /* CQ event for next solicited completion */ - smp_store_mb(*cq->notify, SIW_NOTIFY_SOLICITED); + /* + * Enable CQ event for next solicited completion. + * and make it visible to all associated producers. + */ + smp_store_mb(cq->notify->flags, SIW_NOTIFY_SOLICITED); else - /* CQ event for any signalled completion */ - smp_store_mb(*cq->notify, SIW_NOTIFY_ALL); + /* + * Enable CQ event for any signalled completion. + * and make it visible to all associated producers. + */ + smp_store_mb(cq->notify->flags, SIW_NOTIFY_ALL); if (flags & IB_CQ_REPORT_MISSED_EVENTS) return cq->cq_put - cq->cq_get; diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index 1a039f16d315..e25c70a56be6 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -1767,8 +1767,8 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch) goto out; retry: - ch->cq = ib_alloc_cq(sdev->device, ch, ch->rq_size + sq_size, - 0 /* XXX: spread CQs */, IB_POLL_WORKQUEUE); + ch->cq = ib_alloc_cq_any(sdev->device, ch, ch->rq_size + sq_size, + IB_POLL_WORKQUEUE); if (IS_ERR(ch->cq)) { ret = PTR_ERR(ch->cq); pr_err("failed to create CQ cqe= %d ret= %d\n", |