diff options
| author | Linus Torvalds | 2021-02-22 10:27:48 -0800 |
|---|---|---|
| committer | Linus Torvalds | 2021-02-22 10:27:48 -0800 |
| commit | 3672ac8ac0d8bece188f82c48770bbe40f234f1e (patch) | |
| tree | e596c4917eb6f0c138a3eaf27f0ad458013c25f2 | |
| parent | bdb39c9509e6d31943cb29dbb6ccd1b64013fb98 (diff) | |
| parent | 7289e26f395b583f68b676d4d12a0971e4f6f65c (diff) | |
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Pull rdma updates from Jason Gunthorpe:
"This is quite a small cycle, if not for Lee's 70 patches cleaning the
kdocs it would be well below typical for patch count.
Most of the interesting work here was in the HNS and rxe drivers which
got fairly major internal changes.
Summary:
- Driver updates and bug fixes: siw, hns, bnxt_re, mlx5, efa
- Significant rework in rxe to get it ready to have XRC support added
- Several rts bug fixes
- Big series to get to 'make W=1' cleanness, primarily updating kdocs
- Support for creating a RDMA MR from a DMABUF fd to allow PCI peer
to peer transfers to GPU VRAM
- Device disassociation now works properly with umad
- Work to support more than 255 ports on a RDMA device
- Further support for the new HNS HIP09 hardware
- Coding style cleanups: comma to semicolon, unneded semicolon/blank
lines, remove 'h' printk format, don't check for NULL before kfree,
use true/false for bool"
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (205 commits)
RDMA/rtrs-srv: Do not pass a valid pointer to PTR_ERR()
RDMA/srp: Fix support for unpopulated and unbalanced NUMA nodes
RDMA/mlx5: Fail QP creation if the device can not support the CQE TS
RDMA/mlx5: Allow CQ creation without attached EQs
RDMA/rtrs-srv-sysfs: fix missing put_device
RDMA/rtrs-srv: fix memory leak by missing kobject free
RDMA/rtrs: Only allow addition of path to an already established session
RDMA/rtrs-srv: Fix stack-out-of-bounds
RDMA/rxe: Remove unused pkt->offset
RDMA/ucma: Fix use-after-free bug in ucma_create_uevent
RDMA/core: Fix kernel doc warnings for ib_port_immutable_read()
RDMA/qedr: Use true and false for bool variable
RDMA/hns: Adjust definition of FRMR fields
RDMA/hns: Refactor process of posting CMDQ
RDMA/hns: Adjust fields and variables about CMDQ tail/head
RDMA/hns: Remove redundant operations on CMDQ
RDMA/hns: Fixes missing error code of CMDQ
RDMA/hns: Remove unused member and variable of CMDQ
RDMA/ipoib: Remove racy Subnet Manager sendonly join checks
RDMA/mlx5: Support 400Gbps IB rate in mlx5 driver
...
150 files changed, 3580 insertions, 2667 deletions
diff --git a/drivers/infiniband/Kconfig b/drivers/infiniband/Kconfig index 9325e189a215..04a78d9f8fe3 100644 --- a/drivers/infiniband/Kconfig +++ b/drivers/infiniband/Kconfig @@ -41,6 +41,7 @@ config INFINIBAND_USER_MEM bool depends on INFINIBAND_USER_ACCESS != n depends on MMU + select DMA_SHARED_BUFFER default y config INFINIBAND_ON_DEMAND_PAGING diff --git a/drivers/infiniband/core/Makefile b/drivers/infiniband/core/Makefile index ccf2670ef45e..8ab4eea5a0a5 100644 --- a/drivers/infiniband/core/Makefile +++ b/drivers/infiniband/core/Makefile @@ -40,5 +40,5 @@ ib_uverbs-y := uverbs_main.o uverbs_cmd.o uverbs_marshall.o \ uverbs_std_types_srq.o \ uverbs_std_types_wq.o \ uverbs_std_types_qp.o -ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o +ib_uverbs-$(CONFIG_INFINIBAND_USER_MEM) += umem.o umem_dmabuf.o ib_uverbs-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += umem_odp.o diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index 7989b7e1d1c0..5c9fac7cf420 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -669,11 +669,10 @@ int ib_cache_gid_del_all_netdev_gids(struct ib_device *ib_dev, u8 port, * rdma_find_gid_by_port - Returns the GID entry attributes when it finds * a valid GID entry for given search parameters. It searches for the specified * GID value in the local software cache. - * @device: The device to query. + * @ib_dev: The device to query. * @gid: The GID value to search for. * @gid_type: The GID type to search for. - * @port_num: The port number of the device where the GID value should be - * searched. + * @port: The port number of the device where the GID value should be searched. * @ndev: In RoCE, the net device of the device. NULL means ignore. * * Returns sgid attributes if the GID is found with valid reference or @@ -719,7 +718,7 @@ EXPORT_SYMBOL(rdma_find_gid_by_port); /** * rdma_find_gid_by_filter - Returns the GID table attribute where a * specified GID value occurs - * @device: The device to query. + * @ib_dev: The device to query. * @gid: The GID value to search for. * @port: The port number of the device where the GID value could be * searched. @@ -728,6 +727,7 @@ EXPORT_SYMBOL(rdma_find_gid_by_port); * otherwise, we continue searching the GID table. It's guaranteed that * while filter is executed, ndev field is valid and the structure won't * change. filter is executed in an atomic context. filter must not be NULL. + * @context: Private data to pass into the call-back. * * rdma_find_gid_by_filter() searches for the specified GID value * of which the filter function returns true in the port's GID table. @@ -1253,7 +1253,6 @@ EXPORT_SYMBOL(rdma_get_gid_attr); * @entries: Entries where GID entries are returned. * @max_entries: Maximum number of entries that can be returned. * Entries array must be allocated to hold max_entries number of entries. - * @num_entries: Updated to the number of entries that were successfully read. * * Returns number of entries on success or appropriate error code. */ diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index 98165589c8ab..be996dba040c 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -4333,7 +4333,7 @@ static int cm_add_one(struct ib_device *ib_device) unsigned long flags; int ret; int count = 0; - u8 i; + unsigned int i; cm_dev = kzalloc(struct_size(cm_dev, port, ib_device->phys_port_cnt), GFP_KERNEL); @@ -4345,7 +4345,7 @@ static int cm_add_one(struct ib_device *ib_device) cm_dev->going_down = 0; set_bit(IB_MGMT_METHOD_SEND, reg_req.method_mask); - for (i = 1; i <= ib_device->phys_port_cnt; i++) { + rdma_for_each_port (ib_device, i) { if (!rdma_cap_ib_cm(ib_device, i)) continue; @@ -4431,7 +4431,7 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data) .clr_port_cap_mask = IB_PORT_CM_SUP }; unsigned long flags; - int i; + unsigned int i; write_lock_irqsave(&cm.device_lock, flags); list_del(&cm_dev->list); @@ -4441,7 +4441,7 @@ static void cm_remove_one(struct ib_device *ib_device, void *client_data) cm_dev->going_down = 1; spin_unlock_irq(&cm.lock); - for (i = 1; i <= ib_device->phys_port_cnt; i++) { + rdma_for_each_port (ib_device, i) { if (!rdma_cap_ib_cm(ib_device, i)) continue; diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index c51b84b2d2f3..94096511599f 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -352,7 +352,13 @@ struct ib_device *cma_get_ib_dev(struct cma_device *cma_dev) struct cma_multicast { struct rdma_id_private *id_priv; - struct ib_sa_multicast *sa_mc; + union { + struct ib_sa_multicast *sa_mc; + struct { + struct work_struct work; + struct rdma_cm_event event; + } iboe_join; + }; struct list_head list; void *context; struct sockaddr_storage addr; @@ -1823,6 +1829,8 @@ static void destroy_mc(struct rdma_id_private *id_priv, cma_igmp_send(ndev, &mgid, false); dev_put(ndev); } + + cancel_work_sync(&mc->iboe_join.work); } kfree(mc); } @@ -2683,6 +2691,28 @@ static int cma_query_ib_route(struct rdma_id_private *id_priv, return (id_priv->query_id < 0) ? id_priv->query_id : 0; } +static void cma_iboe_join_work_handler(struct work_struct *work) +{ + struct cma_multicast *mc = + container_of(work, struct cma_multicast, iboe_join.work); + struct rdma_cm_event *event = &mc->iboe_join.event; + struct rdma_id_private *id_priv = mc->id_priv; + int ret; + + mutex_lock(&id_priv->handler_mutex); + if (READ_ONCE(id_priv->state) == RDMA_CM_DESTROYING || + READ_ONCE(id_priv->state) == RDMA_CM_DEVICE_REMOVAL) + goto out_unlock; + + ret = cma_cm_event_handler(id_priv, event); + WARN_ON(ret); + +out_unlock: + mutex_unlock(&id_priv->handler_mutex); + if (event->event == RDMA_CM_EVENT_MULTICAST_JOIN) + rdma_destroy_ah_attr(&event->param.ud.ah_attr); +} + static void cma_work_handler(struct work_struct *_work) { struct cma_work *work = container_of(_work, struct cma_work, work); @@ -4478,10 +4508,7 @@ static int cma_ib_mc_handler(int status, struct ib_sa_multicast *multicast) cma_make_mc_event(status, id_priv, multicast, &event, mc); ret = cma_cm_event_handler(id_priv, &event); rdma_destroy_ah_attr(&event.param.ud.ah_attr); - if (ret) { - destroy_id_handler_unlock(id_priv); - return 0; - } + WARN_ON(ret); out: mutex_unlock(&id_priv->handler_mutex); @@ -4542,17 +4569,6 @@ static int cma_join_ib_multicast(struct rdma_id_private *id_priv, rec.pkey = cpu_to_be16(ib_addr_get_pkey(dev_addr)); rec.join_state = mc->join_state; - if ((rec.join_state == BIT(SENDONLY_FULLMEMBER_JOIN)) && - (!ib_sa_sendonly_fullmem_support(&sa_client, - id_priv->id.device, - id_priv->id.port_num))) { - dev_warn( - &id_priv->id.device->dev, - "RDMA CM: port %u Unable to multicast join: SM doesn't support Send Only Full Member option\n", - id_priv->id.port_num); - return -EOPNOTSUPP; - } - comp_mask = IB_SA_MCMEMBER_REC_MGID | IB_SA_MCMEMBER_REC_PORT_GID | IB_SA_MCMEMBER_REC_PKEY | IB_SA_MCMEMBER_REC_JOIN_STATE | IB_SA_MCMEMBER_REC_QKEY | IB_SA_MCMEMBER_REC_SL | @@ -4604,7 +4620,6 @@ static void cma_iboe_set_mgid(struct sockaddr *addr, union ib_gid *mgid, static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, struct cma_multicast *mc) { - struct cma_work *work; struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr; int err = 0; struct sockaddr *addr = (struct sockaddr *)&mc->addr; @@ -4618,10 +4633,6 @@ static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, if (cma_zero_addr(addr)) return -EINVAL; - work = kzalloc(sizeof *work, GFP_KERNEL); - if (!work) - return -ENOMEM; - gid_type = id_priv->cma_dev->default_gid_type[id_priv->id.port_num - rdma_start_port(id_priv->cma_dev->device)]; cma_iboe_set_mgid(addr, &ib.rec.mgid, gid_type); @@ -4632,10 +4643,9 @@ static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, if (dev_addr->bound_dev_if) ndev = dev_get_by_index(dev_addr->net, dev_addr->bound_dev_if); - if (!ndev) { - err = -ENODEV; - goto err_free; - } + if (!ndev) + return -ENODEV; + ib.rec.rate = iboe_get_rate(ndev); ib.rec.hop_limit = 1; ib.rec.mtu = iboe_get_mtu(ndev->mtu); @@ -4653,24 +4663,15 @@ static int cma_iboe_join_multicast(struct rdma_id_private *id_priv, err = -ENOTSUPP; } dev_put(ndev); - if (err || !ib.rec.mtu) { - if (!err) - err = -EINVAL; - goto err_free; - } + if (err || !ib.rec.mtu) + return err ?: -EINVAL; + rdma_ip2gid((struct sockaddr *)&id_priv->id.route.addr.src_addr, &ib.rec.port_gid); - work->id = id_priv; - INIT_WORK(&work->work, cma_work_handler); - cma_make_mc_event(0, id_priv, &ib, &work->event, mc); - /* Balances with cma_id_put() in cma_work_handler */ - cma_id_get(id_priv); - queue_work(cma_wq, &work->work); + INIT_WORK(&mc->iboe_join.work, cma_iboe_join_work_handler); + cma_make_mc_event(0, id_priv, &ib, &mc->iboe_join.event, mc); + queue_work(cma_wq, &mc->iboe_join.work); return 0; - -err_free: - kfree(work); - return err; } int rdma_join_multicast(struct rdma_cm_id *id, struct sockaddr *addr, diff --git a/drivers/infiniband/core/cma_configfs.c b/drivers/infiniband/core/cma_configfs.c index 97a77ea8d3c9..e0d5e3bae458 100644 --- a/drivers/infiniband/core/cma_configfs.c +++ b/drivers/infiniband/core/cma_configfs.c @@ -204,7 +204,6 @@ static int make_cma_ports(struct cma_dev_group *cma_dev_group, unsigned int i; unsigned int ports_num; struct cma_dev_port_group *ports; - int err; ibdev = cma_get_ib_dev(cma_dev); @@ -215,10 +214,8 @@ static int make_cma_ports(struct cma_dev_group *cma_dev_group, ports = kcalloc(ports_num, sizeof(*cma_dev_group->ports), GFP_KERNEL); - if (!ports) { - err = -ENOMEM; - goto free; - } + if (!ports) + return -ENOMEM; for (i = 0; i < ports_num; i++) { char port_str[10]; @@ -234,12 +231,7 @@ static int make_cma_ports(struct cma_dev_group *cma_dev_group, } cma_dev_group->ports = ports; - return 0; -free: - kfree(ports); - cma_dev_group->ports = NULL; - return err; } static void release_cma_dev(struct config_item *item) diff --git a/drivers/infiniband/core/counters.c b/drivers/infiniband/core/counters.c index 92745522250e..f3a7c1f404af 100644 --- a/drivers/infiniband/core/counters.c +++ b/drivers/infiniband/core/counters.c @@ -10,30 +10,35 @@ #define ALL_AUTO_MODE_MASKS (RDMA_COUNTER_MASK_QP_TYPE | RDMA_COUNTER_MASK_PID) -static int __counter_set_mode(struct rdma_counter_mode *curr, +static int __counter_set_mode(struct rdma_port_counter *port_counter, enum rdma_nl_counter_mode new_mode, enum rdma_nl_counter_mask new_mask) { - if ((new_mode == RDMA_COUNTER_MODE_AUTO) && - ((new_mask & (~ALL_AUTO_MODE_MASKS)) || - (curr->mode != RDMA_COUNTER_MODE_NONE))) - return -EINVAL; + if (new_mode == RDMA_COUNTER_MODE_AUTO && port_counter->num_counters) + if (new_mask & ~ALL_AUTO_MODE_MASKS || + port_counter->mode.mode != RDMA_COUNTER_MODE_NONE) + return -EINVAL; - curr->mode = new_mode; - curr->mask = new_mask; + port_counter->mode.mode = new_mode; + port_counter->mode.mask = new_mask; return 0; } -/** +/* * rdma_counter_set_auto_mode() - Turn on/off per-port auto mode * - * When @on is true, the @mask must be set; When @on is false, it goes - * into manual mode if there's any counter, so that the user is able to - * manually access them. + * @dev: Device to operate + * @port: Port to use + * @mask: Mask to configure + * @extack: Message to the user + * + * Return 0 on success. */ int rdma_counter_set_auto_mode(struct ib_device *dev, u8 port, - bool on, enum rdma_nl_counter_mask mask) + enum rdma_nl_counter_mask mask, + struct netlink_ext_ack *extack) { + enum rdma_nl_counter_mode mode = RDMA_COUNTER_MODE_AUTO; struct rdma_port_counter *port_counter; int ret; @@ -42,23 +47,23 @@ int rdma_counter_set_auto_mode(struct ib_device *dev, u8 port, return -EOPNOTSUPP; mutex_lock(&port_counter->lock); - if (on) { - ret = __counter_set_mode(&port_counter->mode, - RDMA_COUNTER_MODE_AUTO, mask); - } else { - if (port_counter->mode.mode != RDMA_COUNTER_MODE_AUTO) { - ret = -EINVAL; - goto out; - } + if (mask) { + ret = __counter_set_mode(port_counter, mode, mask); + if (ret) + NL_SET_ERR_MSG( + extack, + "Turning on auto mode is not allowed when there is bound QP"); + goto out; + } - if (port_counter->num_counters) - ret = __counter_set_mode(&port_counter->mode, - RDMA_COUNTER_MODE_MANUAL, 0); - else - ret = __counter_set_mode(&port_counter->mode, - RDMA_COUNTER_MODE_NONE, 0); + if (port_counter->mode.mode != RDMA_COUNTER_MODE_AUTO) { + ret = -EINVAL; + goto out; } + mode = (port_counter->num_counters) ? RDMA_COUNTER_MODE_MANUAL : + RDMA_COUNTER_MODE_NONE; + ret = __counter_set_mode(port_counter, mode, 0); out: mutex_unlock(&port_counter->lock); return ret; @@ -122,8 +127,8 @@ static struct rdma_counter *alloc_and_bind(struct ib_device *dev, u8 port, mutex_lock(&port_counter->lock); switch (mode) { case RDMA_COUNTER_MODE_MANUAL: - ret = __counter_set_mode(&port_counter->mode, - RDMA_COUNTER_MODE_MANUAL, 0); + ret = __counter_set_mode(port_counter, RDMA_COUNTER_MODE_MANUAL, + 0); if (ret) { mutex_unlock(&port_counter->lock); goto err_mode; @@ -170,8 +175,7 @@ static void rdma_counter_free(struct rdma_counter *counter) port_counter->num_counters--; if (!port_counter->num_counters && (port_counter->mode.mode == RDMA_COUNTER_MODE_MANUAL)) - __counter_set_mode(&port_counter->mode, RDMA_COUNTER_MODE_NONE, - 0); + __counter_set_mode(port_counter, RDMA_COUNTER_MODE_NONE, 0); mutex_unlock(&port_counter->lock); @@ -227,7 +231,7 @@ static void counter_history_stat_update(struct rdma_counter *counter) port_counter->hstats->value[i] += counter->stats->value[i]; } -/** +/* * rdma_get_counter_auto_mode - Find the counter that @qp should be bound * with in auto mode * @@ -274,7 +278,7 @@ static void counter_release(struct kref *kref) rdma_counter_free(counter); } -/** +/* * rdma_counter_bind_qp_auto - Check and bind the QP to a counter base on * the auto-mode rule */ @@ -311,7 +315,7 @@ int rdma_counter_bind_qp_auto(struct ib_qp *qp, u8 port) return 0; } -/** +/* * rdma_counter_unbind_qp - Unbind a qp from a counter * @force: * true - Decrease the counter ref-count anyway (e.g., qp destroy) @@ -380,7 +384,7 @@ next: return sum; } -/** +/* * rdma_counter_get_hwstat_value() - Get the sum value of all counters on a * specific port, including the running ones and history data */ @@ -436,7 +440,7 @@ static struct rdma_counter *rdma_get_counter_by_id(struct ib_device *dev, return counter; } -/** +/* * rdma_counter_bind_qpn() - Bind QP @qp_num to counter @counter_id */ int rdma_counter_bind_qpn(struct ib_device *dev, u8 port, @@ -485,7 +489,7 @@ err: return ret; } -/** +/* * rdma_counter_bind_qpn_alloc() - Alloc a counter and bind QP @qp_num to it * The id of new counter is returned in @counter_id */ @@ -533,7 +537,7 @@ err: return ret; } -/** +/* * rdma_counter_unbind_qpn() - Unbind QP @qp_num from a counter */ int rdma_counter_unbind_qpn(struct ib_device *dev, u8 port, diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index e96f979e6d52..aac0fe14e1d9 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -848,6 +848,20 @@ static int setup_port_data(struct ib_device *device) return 0; } +/** + * ib_port_immutable_read() - Read rdma port's immutable data + * @dev: IB device + * @port: port number whose immutable data to read. It starts with index 1 and + * valid upto including rdma_end_port(). + */ +const struct ib_port_immutable* +ib_port_immutable_read(struct ib_device *dev, unsigned int port) +{ + WARN_ON(!rdma_is_port_valid(dev, port)); + return &dev->port_data[port].immutable; +} +EXPORT_SYMBOL(ib_port_immutable_read); + void ib_get_device_fw_str(struct ib_device *dev, char *str) { if (dev->ops.get_dev_fw_str) @@ -1887,9 +1901,9 @@ static int __ib_get_client_nl_info(struct ib_device *ibdev, /** * ib_get_client_nl_info - Fetch the nl_info from a client - * @device - IB device - * @client_name - Name of the client - * @res - Result of the query + * @ibdev: IB device + * @client_name: Name of the client + * @res: Result of the query */ int ib_get_client_nl_info(struct ib_device *ibdev, const char *client_name, struct ib_client_nl_info *res) @@ -2317,7 +2331,7 @@ void ib_enum_all_roce_netdevs(roce_netdev_filter filter, up_read(&devices_rwsem); } -/** +/* * ib_enum_all_devs - enumerate all ib_devices * @cb: Callback to call for each found ib_device * @@ -2681,6 +2695,7 @@ void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops) SET_DEVICE_OP(dev_ops, read_counters); SET_DEVICE_OP(dev_ops, reg_dm_mr); SET_DEVICE_OP(dev_ops, reg_user_mr); + SET_DEVICE_OP(dev_ops, reg_user_mr_dmabuf); SET_DEVICE_OP(dev_ops, req_ncomp_notif); SET_DEVICE_OP(dev_ops, req_notify_cq); SET_DEVICE_OP(dev_ops, rereg_user_mr); diff --git a/drivers/infiniband/core/iwpm_msg.c b/drivers/infiniband/core/iwpm_msg.c index 46686990a827..30a0ff76b332 100644 --- a/drivers/infiniband/core/iwpm_msg.c +++ b/drivers/infiniband/core/iwpm_msg.c @@ -392,7 +392,7 @@ static const struct nla_policy resp_reg_policy[IWPM_NLA_RREG_PID_MAX] = { /** * iwpm_register_pid_cb - Process the port mapper response to * iwpm_register_pid query - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) * * If successful, the function receives the userspace port mapper pid @@ -468,7 +468,7 @@ static const struct nla_policy resp_add_policy[IWPM_NLA_RMANAGE_MAPPING_MAX] = { /** * iwpm_add_mapping_cb - Process the port mapper response to * iwpm_add_mapping request - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) */ int iwpm_add_mapping_cb(struct sk_buff *skb, struct netlink_callback *cb) @@ -545,7 +545,7 @@ static const struct nla_policy resp_query_policy[IWPM_NLA_RQUERY_MAPPING_MAX] = /** * iwpm_add_and_query_mapping_cb - Process the port mapper response to * iwpm_add_and_query_mapping request - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) */ int iwpm_add_and_query_mapping_cb(struct sk_buff *skb, @@ -627,7 +627,7 @@ query_mapping_response_exit: /** * iwpm_remote_info_cb - Process remote connecting peer address info, which * the port mapper has received from the connecting peer - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) * * Stores the IPv4/IPv6 address info in a hash table @@ -706,7 +706,7 @@ static const struct nla_policy resp_mapinfo_policy[IWPM_NLA_MAPINFO_REQ_MAX] = { /** * iwpm_mapping_info_cb - Process a notification that the userspace * port mapper daemon is started - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) * * Using the received port mapper pid, send all the local mapping @@ -766,7 +766,7 @@ static const struct nla_policy ack_mapinfo_policy[IWPM_NLA_MAPINFO_NUM_MAX] = { /** * iwpm_ack_mapping_info_cb - Process the port mapper ack for * the provided local mapping info records - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) */ int iwpm_ack_mapping_info_cb(struct sk_buff *skb, struct netlink_callback *cb) @@ -796,7 +796,7 @@ static const struct nla_policy map_error_policy[IWPM_NLA_ERR_MAX] = { /** * iwpm_mapping_error_cb - Process port mapper notification for error * - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) */ int iwpm_mapping_error_cb(struct sk_buff *skb, struct netlink_callback *cb) @@ -841,7 +841,7 @@ static const struct nla_policy hello_policy[IWPM_NLA_HELLO_MAX] = { /** * iwpm_hello_cb - Process a hello message from iwpmd * - * @skb: + * @skb: The socket buffer * @cb: Contains the received message (payload and netlink header) * * Using the received port mapper pid, send the kernel's abi_version diff --git a/drivers/infiniband/core/iwpm_util.c b/drivers/infiniband/core/iwpm_util.c index 13495b43dbc1..f80e5550b51f 100644 --- a/drivers/infiniband/core/iwpm_util.c +++ b/drivers/infiniband/core/iwpm_util.c @@ -127,8 +127,8 @@ static struct hlist_head *get_mapinfo_hash_bucket(struct sockaddr_storage *, /** * iwpm_create_mapinfo - Store local and mapped IPv4/IPv6 address * info in a hash table - * @local_addr: Local ip/tcp address - * @mapped_addr: Mapped local ip/tcp address + * @local_sockaddr: Local ip/tcp address + * @mapped_sockaddr: Mapped local ip/tcp address * @nl_client: The index of the netlink client * @map_flags: IWPM mapping flags */ @@ -174,7 +174,7 @@ int iwpm_create_mapinfo(struct sockaddr_storage *local_sockaddr, /** * iwpm_remove_mapinfo - Remove local and mapped IPv4/IPv6 address * info from the hash table - * @local_addr: Local ip/tcp address + * @local_sockaddr: Local ip/tcp address * @mapped_local_addr: Mapped local ip/tcp address * * Returns err code if mapping info is not found in the hash table, diff --git a/drivers/infiniband/core/multicast.c b/drivers/infiniband/core/multicast.c index 740f03ecc05d..57519ca6cd2c 100644 --- a/drivers/infiniband/core/multicast.c +++ b/drivers/infiniband/core/multicast.c @@ -721,6 +721,7 @@ EXPORT_SYMBOL(ib_sa_get_mcmember_rec); * member record and gid of the device. * @device: RDMA device * @port_num: Port of the rdma device to consider + * @rec: Multicast member record to use * @ndev: Optional netdevice, applicable only for RoCE * @gid_type: GID type to consider * @ah_attr: AH attribute to fillup on successful completion diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c index 08366e254b1d..d306049c22a2 100644 --- a/drivers/infiniband/core/nldev.c +++ b/drivers/infiniband/core/nldev.c @@ -1768,9 +1768,7 @@ static int nldev_stat_set_doit(struct sk_buff *skb, struct nlmsghdr *nlh, if (tb[RDMA_NLDEV_ATTR_STAT_AUTO_MODE_MASK]) mask = nla_get_u32( tb[RDMA_NLDEV_ATTR_STAT_AUTO_MODE_MASK]); - - ret = rdma_counter_set_auto_mode(device, port, - mask ? true : false, mask); + ret = rdma_counter_set_auto_mode(device, port, mask, extack); if (ret) goto err_msg; } else { diff --git a/drivers/infiniband/core/restrack.c b/drivers/infiniband/core/restrack.c index ff1551b3cf61..ffabaf327242 100644 --- a/drivers/infiniband/core/restrack.c +++ b/drivers/infiniband/core/restrack.c @@ -201,8 +201,8 @@ EXPORT_SYMBOL(rdma_restrack_parent_name); /** * rdma_restrack_new() - Initializes new restrack entry to allow _put() interface * to release memory in fully automatic way. - * @res - Entry to initialize - * @type - REstrack type + * @res: Entry to initialize + * @type: REstrack type */ void rdma_restrack_new(struct rdma_restrack_entry *res, enum rdma_restrack_type type) diff --git a/drivers/infiniband/core/roce_gid_mgmt.c b/drivers/infiniband/core/roce_gid_mgmt.c index 6b8364bb032d..34fff94eaa38 100644 --- a/drivers/infiniband/core/roce_gid_mgmt.c +++ b/drivers/infiniband/core/roce_gid_mgmt.c @@ -505,7 +505,7 @@ static void enum_all_gids_of_dev_cb(struct ib_device *ib_dev, * rdma_roce_rescan_device - Rescan all of the network devices in the system * and add their gids, as needed, to the relevant RoCE devices. * - * @device: the rdma device + * @ib_dev: the rdma device */ void rdma_roce_rescan_device(struct ib_device *ib_dev) { diff --git a/drivers/infiniband/core/rw.c b/drivers/infiniband/core/rw.c index a96030b784eb..31156e22d3e7 100644 --- a/drivers/infiniband/core/rw.c +++ b/drivers/infiniband/core/rw.c @@ -410,7 +410,7 @@ int rdma_rw_ctx_signature_init(struct rdma_rw_ctx *ctx, struct ib_qp *qp, ctx->type = RDMA_RW_SIG_MR; ctx->nr_ops = 1; - ctx->reg = kcalloc(1, sizeof(*ctx->reg), GFP_KERNEL); + ctx->reg = kzalloc(sizeof(*ctx->reg), GFP_KERNEL); if (!ctx->reg) { ret = -ENOMEM; goto out_unmap_prot_sg; diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index 89a831fa1885..9ef1a355131b 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -1434,7 +1434,7 @@ enum opa_pr_supported { PR_IB_SUPPORTED }; -/** +/* * opa_pr_query_possible - Check if current PR query can be an OPA query. * * Retuns PR_NOT_SUPPORTED if a path record query is not @@ -1951,30 +1951,6 @@ err1: } EXPORT_SYMBOL(ib_sa_guid_info_rec_query); -bool ib_sa_sendonly_fullmem_support(struct ib_sa_client *client, - struct ib_device *device, - u8 port_num) -{ - struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client); - struct ib_sa_port *port; - bool ret = false; - unsigned long flags; - - if (!sa_dev) - return ret; - - port = &sa_dev->port[port_num - sa_dev->start_port]; - - spin_lock_irqsave(&port->classport_lock, flags); - if ((port->classport_info.valid) && - (port->classport_info.data.type == RDMA_CLASS_PORT_INFO_IB)) - ret = ib_get_cpi_capmask2(&port->classport_info.data.ib) - & IB_SA_CAP_MASK2_SENDONLY_FULL_MEM_SUPPORT; - spin_unlock_irqrestore(&port->classport_lock, flags); - return ret; -} -EXPORT_SYMBOL(ib_sa_sendonly_fullmem_support); - struct ib_classport_info_context { struct completion done; struct ib_sa_query *sa_query; diff --git a/drivers/infiniband/core/umem.c b/drivers/infiniband/core/umem.c index 917338db7ac1..2dde99a9ba07 100644 --- a/drivers/infiniband/core/umem.c +++ b/drivers/infiniband/core/umem.c @@ -2,6 +2,7 @@ * Copyright (c) 2005 Topspin Communications. All rights reserved. * Copyright (c) 2005 Cisco Systems. All rights reserved. * Copyright (c) 2005 Mellanox Technologies. All rights reserved. + * Copyright (c) 2020 Intel Corporation. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -278,6 +279,8 @@ void ib_umem_release(struct ib_umem *umem) { if (!umem) return; + if (umem->is_dmabuf) + return ib_umem_dmabuf_release(to_ib_umem_dmabuf(umem)); if (umem->is_odp) return ib_umem_odp_release(to_ib_umem_odp(umem)); diff --git a/drivers/infiniband/core/umem_dmabuf.c b/drivers/infiniband/core/umem_dmabuf.c new file mode 100644 index 000000000000..f9b5162d9260 --- /dev/null +++ b/drivers/infiniband/core/umem_dmabuf.c @@ -0,0 +1,174 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright (c) 2020 Intel Corporation. All rights reserved. + */ + +#include <linux/dma-buf.h> +#include <linux/dma-resv.h> +#include <linux/dma-mapping.h> + +#include "uverbs.h" + +int ib_umem_dmabuf_map_pages(struct ib_umem_dmabuf *umem_dmabuf) +{ + struct sg_table *sgt; + struct scatterlist *sg; + struct dma_fence *fence; + unsigned long start, end, cur = 0; + unsigned int nmap = 0; + int i; + + dma_resv_assert_held(umem_dmabuf->attach->dmabuf->resv); + + if (umem_dmabuf->sgt) + goto wait_fence; + + sgt = dma_buf_map_attachment(umem_dmabuf->attach, DMA_BIDIRECTIONAL); + if (IS_ERR(sgt)) + return PTR_ERR(sgt); + + /* modify the sg list in-place to match umem address and length */ + + start = ALIGN_DOWN(umem_dmabuf->umem.address, PAGE_SIZE); + end = ALIGN(umem_dmabuf->umem.address + umem_dmabuf->umem.length, + PAGE_SIZE); + for_each_sgtable_dma_sg(sgt, sg, i) { + if (start < cur + sg_dma_len(sg) && cur < end) + nmap++; + if (cur <= start && start < cur + sg_dma_len(sg)) { + unsigned long offset = start - cur; + + umem_dmabuf->first_sg = sg; + umem_dmabuf->first_sg_offset = offset; + sg_dma_address(sg) += offset; + sg_dma_len(sg) -= offset; + cur += offset; + } + if (cur < end && end <= cur + sg_dma_len(sg)) { + unsigned long trim = cur + sg_dma_len(sg) - end; + + umem_dmabuf->last_sg = sg; + umem_dmabuf->last_sg_trim = trim; + sg_dma_len(sg) -= trim; + break; + } + cur += sg_dma_len(sg); + } + + umem_dmabuf->umem.sg_head.sgl = umem_dmabuf->first_sg; + umem_dmabuf->umem.sg_head.nents = nmap; + umem_dmabuf->umem.nmap = nmap; + umem_dmabuf->sgt = sgt; + +wait_fence: + /* + * Although the sg list is valid now, the content of the pages + * may be not up-to-date. Wait for the exporter to finish + * the migration. + */ + fence = dma_resv_get_excl(umem_dmabuf->attach->dmabuf->resv); + if (fence) + return dma_fence_wait(fence, false); + + return 0; +} +EXPORT_SYMBOL(ib_umem_dmabuf_map_pages); + +void ib_umem_dmabuf_unmap_pages(struct ib_umem_dmabuf *umem_dmabuf) +{ + dma_resv_assert_held(umem_dmabuf->attach->dmabuf->resv); + + if (!umem_dmabuf->sgt) + return; + + /* retore the original sg list */ + if (umem_dmabuf->first_sg) { + sg_dma_address(umem_dmabuf->first_sg) -= + umem_dmabuf->first_sg_offset; + sg_dma_len(umem_dmabuf->first_sg) += + umem_dmabuf->first_sg_offset; + umem_dmabuf->first_sg = NULL; + umem_dmabuf->first_sg_offset = 0; + } + if (umem_dmabuf->last_sg) { + sg_dma_len(umem_dmabuf->last_sg) += + umem_dmabuf->last_sg_trim; + umem_dmabuf->last_sg = NULL; + umem_dmabuf->last_sg_trim = 0; + } + + dma_buf_unmap_attachment(umem_dmabuf->attach, umem_dmabuf->sgt, + DMA_BIDIRECTIONAL); + + umem_dmabuf->sgt = NULL; +} +EXPORT_SYMBOL(ib_umem_dmabuf_unmap_pages); + +struct ib_umem_dmabuf *ib_umem_dmabuf_get(struct ib_device *device, + unsigned long offset, size_t size, + int fd, int access, + const struct dma_buf_attach_ops *ops) +{ + struct dma_buf *dmabuf; + struct ib_umem_dmabuf *umem_dmabuf; + struct ib_umem *umem; + unsigned long end; + struct ib_umem_dmabuf *ret = ERR_PTR(-EINVAL); + + if (check_add_overflow(offset, (unsigned long)size, &end)) + return ret; + + if (unlikely(!ops || !ops->move_notify)) + return ret; + + dmabuf = dma_buf_get(fd); + if (IS_ERR(dmabuf)) + return ERR_CAST(dmabuf); + + if (dmabuf->size < end) + goto out_release_dmabuf; + + umem_dmabuf = kzalloc(sizeof(*umem_dmabuf), GFP_KERNEL); + if (!umem_dmabuf) { + ret = ERR_PTR(-ENOMEM); + goto out_release_dmabuf; + } + + umem = &umem_dmabuf->umem; + umem->ibdev = device; + umem->length = size; + umem->address = offset; + umem->writable = ib_access_writable(access); + umem->is_dmabuf = 1; + + if (!ib_umem_num_pages(umem)) + goto out_free_umem; + + umem_dmabuf->attach = dma_buf_dynamic_attach( + dmabuf, + device->dma_device, + ops, + umem_dmabuf); + if (IS_ERR(umem_dmabuf->attach)) { + ret = ERR_CAST(umem_dmabuf->attach); + goto out_free_umem; + } + return umem_dmabuf; + +out_free_umem: + kfree(umem_dmabuf); + +out_release_dmabuf: + dma_buf_put(dmabuf); + return ret; +} +EXPORT_SYMBOL(ib_umem_dmabuf_get); + +void ib_umem_dmabuf_release(struct ib_umem_dmabuf *umem_dmabuf) +{ + struct dma_buf *dmabuf = umem_dmabuf->attach->dmabuf; + + dma_buf_detach(dmabuf, umem_dmabuf->attach); + dma_buf_put(dmabuf); + kfree(umem_dmabuf); +} diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c index 19104a675691..dd7f3b437c6b 100644 --- a/drivers/infiniband/core/user_mad.c +++ b/drivers/infiniband/core/user_mad.c @@ -379,6 +379,11 @@ static ssize_t ib_umad_read(struct file *filp, char __user *buf, mutex_lock(&file->mutex); + if (file->agents_dead) { + mutex_unlock(&file->mutex); + return -EIO; + } + while (list_empty(&file->recv_list)) { mutex_unlock(&file->mutex); @@ -392,6 +397,11 @@ static ssize_t ib_umad_read(struct file *filp, char __user *buf, mutex_lock(&file->mutex); } + if (file->agents_dead) { + mutex_unlock(&file->mutex); + return -EIO; + } + packet = list_entry(file->recv_list.next, struct ib_umad_packet, list); list_del(&packet->list); @@ -524,7 +534,7 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf, agent = __get_agent(file, packet->mad.hdr.id); if (!agent) { - ret = -EINVAL; + ret = -EIO; goto err_up; } @@ -653,10 +663,14 @@ static __poll_t ib_umad_poll(struct file *filp, struct poll_table_struct *wait) /* we will always be able to post a MAD send */ __poll_t mask = EPOLLOUT | EPOLLWRNORM; + mutex_lock(&file->mutex); poll_wait(filp, &file->recv_wait, wait); if (!list_empty(&file->recv_list)) mask |= EPOLLIN | EPOLLRDNORM; + if (file->agents_dead) + mask = EPOLLERR; + mutex_unlock(&file->mutex); return mask; } @@ -1336,6 +1350,7 @@ static void ib_umad_kill_port(struct ib_umad_port *port) list_for_each_entry(file, &port->file_list, port_list) { mutex_lock(&file->mutex); file->agents_dead = 1; + wake_up_interruptible(&file->recv_wait); mutex_unlock(&file->mutex); for (id = 0; id < IB_UMAD_MAX_AGENTS; ++id) diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index 98a5d36813ff..f5b8be3bedde 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -1382,7 +1382,7 @@ static int create_qp(struct uverbs_attr_bundle *attrs, if (has_sq) scq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd->send_cq_handle, attrs); - if (!ind_tbl) + if (!ind_tbl && cmd->qp_type != IB_QPT_XRC_INI) rcq = rcq ?: scq; pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd->pd_handle, attrs); diff --git a/drivers/infiniband/core/uverbs_std_types_mr.c b/drivers/infiniband/core/uverbs_std_types_mr.c index dd4e76b26c74..f782d5e1aa25 100644 --- a/drivers/infiniband/core/uverbs_std_types_mr.c +++ b/drivers/infiniband/core/uverbs_std_types_mr.c @@ -1,5 +1,6 @@ /* * Copyright (c) 2018, Mellanox Technologies inc. All rights reserved. + * Copyright (c) 2020, Intel Corporation. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -182,6 +183,86 @@ static int UVERBS_HANDLER(UVERBS_METHOD_QUERY_MR)( return IS_UVERBS_COPY_ERR(ret) ? ret : 0; } +static int UVERBS_HANDLER(UVERBS_METHOD_REG_DMABUF_MR)( + struct uverbs_attr_bundle *attrs) +{ + struct ib_uobject *uobj = + uverbs_attr_get_uobject(attrs, UVERBS_ATTR_REG_DMABUF_MR_HANDLE); + struct ib_pd *pd = + uverbs_attr_get_obj(attrs, UVERBS_ATTR_REG_DMABUF_MR_PD_HANDLE); + struct ib_device *ib_dev = pd->device; + + u64 offset, length, iova; + u32 fd, access_flags; + struct ib_mr *mr; + int ret; + + if (!ib_dev->ops.reg_user_mr_dmabuf) + return -EOPNOTSUPP; + + ret = uverbs_copy_from(&offset, attrs, + UVERBS_ATTR_REG_DMABUF_MR_OFFSET); + if (ret) + return ret; + + ret = uverbs_copy_from(&length, attrs, + UVERBS_ATTR_REG_DMABUF_MR_LENGTH); + if (ret) + return ret; + + ret = uverbs_copy_from(&iova, attrs, + UVERBS_ATTR_REG_DMABUF_MR_IOVA); + if (ret) + return ret; + + if ((offset & ~PAGE_MASK) != (iova & ~PAGE_MASK)) + return -EINVAL; + + ret = uverbs_copy_from(&fd, attrs, + UVERBS_ATTR_REG_DMABUF_MR_FD); + if (ret) + return ret; + + ret = uverbs_get_flags32(&access_flags, attrs, + UVERBS_ATTR_REG_DMABUF_MR_ACCESS_FLAGS, + IB_ACCESS_LOCAL_WRITE | + IB_ACCESS_REMOTE_READ | + IB_ACCESS_REMOTE_WRITE | + IB_ACCESS_REMOTE_ATOMIC | + IB_ACCESS_RELAXED_ORDERING); + if (ret) + return ret; + + ret = ib_check_mr_access(ib_dev, access_flags); + if (ret) + return ret; + + mr = pd->device->ops.reg_user_mr_dmabuf(pd, offset, length, iova, fd, + access_flags, + &attrs->driver_udata); + if (IS_ERR(mr)) + return PTR_ERR(mr); + + mr->device = pd->device; + mr->pd = pd; + mr->type = IB_MR_TYPE_USER; + mr->uobject = uobj; + atomic_inc(&pd->usecnt); + + uobj->object = mr; + + uverbs_finalize_uobj_create(attrs, UVERBS_ATTR_REG_DMABUF_MR_HANDLE); + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_REG_DMABUF_MR_RESP_LKEY, + &mr->lkey, sizeof(mr->lkey)); + if (ret) + return ret; + + ret = uverbs_copy_to(attrs, UVERBS_ATTR_REG_DMABUF_MR_RESP_RKEY, + &mr->rkey, sizeof(mr->rkey)); + return ret; +} + DECLARE_UVERBS_NAMED_METHOD( UVERBS_METHOD_ADVISE_MR, UVERBS_ATTR_IDR(UVERBS_ATTR_ADVISE_MR_PD_HANDLE, @@ -247,6 +328,37 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_TYPE(u32), UA_MANDATORY)); +DECLARE_UVERBS_NAMED_METHOD( + UVERBS_METHOD_REG_DMABUF_MR, + UVERBS_ATTR_IDR(UVERBS_ATTR_REG_DMABUF_MR_HANDLE, + UVERBS_OBJECT_MR, + UVERBS_ACCESS_NEW, + UA_MANDATORY), + UVERBS_ATTR_IDR(UVERBS_ATTR_REG_DMABUF_MR_PD_HANDLE, + UVERBS_OBJECT_PD, + UVERBS_ACCESS_READ, + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_REG_DMABUF_MR_OFFSET, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_REG_DMABUF_MR_LENGTH, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_REG_DMABUF_MR_IOVA, + UVERBS_ATTR_TYPE(u64), + UA_MANDATORY), + UVERBS_ATTR_PTR_IN(UVERBS_ATTR_REG_DMABUF_MR_FD, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_FLAGS_IN(UVERBS_ATTR_REG_DMABUF_MR_ACCESS_FLAGS, + enum ib_access_flags), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_REG_DMABUF_MR_RESP_LKEY, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY), + UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_REG_DMABUF_MR_RESP_RKEY, + UVERBS_ATTR_TYPE(u32), + UA_MANDATORY)); + DECLARE_UVERBS_NAMED_METHOD_DESTROY( UVERBS_METHOD_MR_DESTROY, UVERBS_ATTR_IDR(UVERBS_ATTR_DESTROY_MR_HANDLE, @@ -257,10 +369,11 @@ DECLARE_UVERBS_NAMED_METHOD_DESTROY( DECLARE_UVERBS_NAMED_OBJECT( UVERBS_OBJECT_MR, UVERBS_TYPE_ALLOC_IDR(uverbs_free_mr), + &UVERBS_METHOD(UVERBS_METHOD_ADVISE_MR), &UVERBS_METHOD(UVERBS_METHOD_DM_MR_REG), &UVERBS_METHOD(UVERBS_METHOD_MR_DESTROY), - &UVERBS_METHOD(UVERBS_METHOD_ADVISE_MR), - &UVERBS_METHOD(UVERBS_METHOD_QUERY_MR)); + &UVERBS_METHOD(UVERBS_METHOD_QUERY_MR), + &UVERBS_METHOD(UVERBS_METHOD_REG_DMABUF_MR)); const struct uapi_definition uverbs_def_obj_mr[] = { UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MR, diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 9137a25bb521..28464c58738c 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -2248,7 +2248,7 @@ static bool is_valid_mcast_lid(struct ib_qp *qp, u16 lid) struct ib_qp_init_attr init_attr = {}; struct ib_qp_attr attr = {}; int num_eth_ports = 0; - int port; + unsigned int port; /* If QP state >= init, it is assigned to a port and we can check this * port only. @@ -2263,7 +2263,7 @@ static bool is_valid_mcast_lid(struct ib_qp *qp, u16 lid) } /* Can't get a quick answer, iterate over all ports */ - for (port = 0; port < qp->device->phys_port_cnt; port++) + rdma_for_each_port(qp->device, port) if (rdma_port_get_link_layer(qp->device, port) != IB_LINK_LAYER_INFINIBAND) num_eth_ports++; diff --git a/drivers/infiniband/hw/bnxt_re/ib_verbs.c b/drivers/infiniband/hw/bnxt_re/ib_verbs.c index 401bdc9e931e..ba515efd4fdc 100644 --- a/drivers/infiniband/hw/bnxt_re/ib_verbs.c +++ b/drivers/infiniband/hw/bnxt_re/ib_verbs.c @@ -469,7 +469,6 @@ static int bnxt_re_create_fence_mr(struct bnxt_re_pd *pd) struct bnxt_re_mr *mr = NULL; dma_addr_t dma_addr = 0; struct ib_mw *mw; - u64 pbl_tbl; int rc; dma_addr = dma_map_single(dev, fence->va, BNXT_RE_FENCE_BYTES, @@ -504,9 +503,8 @@ static int bnxt_re_create_fence_mr(struct bnxt_re_pd *pd) mr->ib_mr.lkey = mr->qplib_mr.lkey; mr->qplib_mr.va = (u64)(unsigned long)fence->va; mr->qplib_mr.total_size = BNXT_RE_FENCE_BYTES; - pbl_tbl = dma_addr; - rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, &pbl_tbl, - BNXT_RE_FENCE_PBL_SIZE, false, PAGE_SIZE); + rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, NULL, + BNXT_RE_FENCE_PBL_SIZE, PAGE_SIZE); if (rc) { ibdev_err(&rdev->ibdev, "Failed to register fence-MR\n"); goto fail; @@ -3589,7 +3587,6 @@ struct ib_mr *bnxt_re_get_dma_mr(struct ib_pd *ib_pd, int mr_access_flags) struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd); struct bnxt_re_dev *rdev = pd->rdev; struct bnxt_re_mr *mr; - u64 pbl = 0; int rc; mr = kzalloc(sizeof(*mr), GFP_KERNEL); @@ -3608,7 +3605,7 @@ struct ib_mr *bnxt_re_get_dma_mr(struct ib_pd *ib_pd, int mr_access_flags) mr->qplib_mr.hwq.level = PBL_LVL_MAX; mr->qplib_mr.total_size = -1; /* Infinte length */ - rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, &pbl, 0, false, + rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, NULL, 0, PAGE_SIZE); if (rc) goto fail_mr; @@ -3779,19 +3776,6 @@ int bnxt_re_dealloc_mw(struct ib_mw *ib_mw) return rc; } -static int fill_umem_pbl_tbl(struct ib_umem *umem, u64 *pbl_tbl_orig, - int page_shift) -{ - u64 *pbl_tbl = pbl_tbl_orig; - u64 page_size = BIT_ULL(page_shift); - struct ib_block_iter biter; - - rdma_umem_for_each_dma_block(umem, &biter, page_size) - *pbl_tbl++ = rdma_block_iter_dma_address(&biter); - - return pbl_tbl - pbl_tbl_orig; -} - /* uverbs */ struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *ib_pd, u64 start, u64 length, u64 virt_addr, int mr_access_flags, @@ -3801,7 +3785,6 @@ struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *ib_pd, u64 start, u64 length, struct bnxt_re_dev *rdev = pd->rdev; struct bnxt_re_mr *mr; struct ib_umem *umem; - u64 *pbl_tbl = NULL; unsigned long page_size; int umem_pgs, rc; @@ -3846,39 +3829,19 @@ struct ib_mr *bnxt_re_reg_user_mr(struct ib_pd *ib_pd, u64 start, u64 length, } mr->qplib_mr.total_size = length; - if (page_size == BNXT_RE_PAGE_SIZE_4K && - length > BNXT_RE_MAX_MR_SIZE_LOW) { - ibdev_err(&rdev->ibdev, "Requested MR Sz:%llu Max sup:%llu", - length, (u64)BNXT_RE_MAX_MR_SIZE_LOW); - rc = -EINVAL; - goto free_umem; - } - umem_pgs = ib_umem_num_dma_blocks(umem, page_size); - pbl_tbl = kcalloc(umem_pgs, sizeof(*pbl_tbl), GFP_KERNEL); - if (!pbl_tbl) { - rc = -ENOMEM; - goto free_umem; - } - - /* Map umem buf ptrs to the PBL */ - umem_pgs = fill_umem_pbl_tbl(umem, pbl_tbl, order_base_2(page_size)); - rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, pbl_tbl, - umem_pgs, false, page_size); + rc = bnxt_qplib_reg_mr(&rdev->qplib_res, &mr->qplib_mr, umem, + umem_pgs, page_size); if (rc) { ibdev_err(&rdev->ibdev, "Failed to register user MR"); - goto fail; + goto free_umem; } - kfree(pbl_tbl); - mr->ib_mr.lkey = mr->qplib_mr.lkey; mr->ib_mr.rkey = mr->qplib_mr.lkey; atomic_inc(&rdev->mr_count); return &mr->ib_mr; -fail: - kfree(pbl_tbl); free_umem: ib_umem_release(umem); free_mrw: diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.c b/drivers/infiniband/hw/bnxt_re/qplib_sp.c index 6316179583a6..049b3576302b 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.c @@ -650,42 +650,32 @@ int bnxt_qplib_dereg_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw, } int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, - u64 *pbl_tbl, int num_pbls, bool block, u32 buf_pg_size) + struct ib_umem *umem, int num_pbls, u32 buf_pg_size) { struct bnxt_qplib_rcfw *rcfw = res->rcfw; struct bnxt_qplib_hwq_attr hwq_attr = {}; struct bnxt_qplib_sg_info sginfo = {}; struct creq_register_mr_resp resp; struct cmdq_register_mr req; - int pg_ptrs, pages, i, rc; u16 cmd_flags = 0, level; - dma_addr_t **pbl_ptr; + int pages, rc; u32 pg_size; if (num_pbls) { + pages = roundup_pow_of_two(num_pbls); /* Allocate memory for the non-leaf pages to store buf ptrs. * Non-leaf pages always uses system PAGE_SIZE */ - pg_ptrs = roundup_pow_of_two(num_pbls); - pages = pg_ptrs >> MAX_PBL_LVL_1_PGS_SHIFT; - if (!pages) - pages++; - - if (pages > MAX_PBL_LVL_1_PGS) { - dev_err(&res->pdev->dev, - "SP: Reg MR: pages requested (0x%x) exceeded max (0x%x)\n", - pages, MAX_PBL_LVL_1_PGS); - return -ENOMEM; - } /* Free the hwq if it already exist, must be a rereg */ if (mr->hwq.max_elements) bnxt_qplib_free_hwq(res, &mr->hwq); /* Use system PAGE_SIZE */ hwq_attr.res = res; hwq_attr.depth = pages; - hwq_attr.stride = PAGE_SIZE; + hwq_attr.stride = buf_pg_size; hwq_attr.type = HWQ_TYPE_MR; hwq_attr.sginfo = &sginfo; + hwq_attr.sginfo->umem = umem; hwq_attr.sginfo->npages = pages; hwq_attr.sginfo->pgsize = PAGE_SIZE; hwq_attr.sginfo->pgshft = PAGE_SHIFT; @@ -695,11 +685,6 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, "SP: Reg MR memory allocation failed\n"); return -ENOMEM; } - /* Write to the hwq */ - pbl_ptr = (dma_addr_t **)mr->hwq.pbl_ptr; - for (i = 0; i < num_pbls; i++) - pbl_ptr[PTR_PG(i)][PTR_IDX(i)] = - (pbl_tbl[i] & PAGE_MASK) | PTU_PTE_VALID; } RCFW_CMD_PREP(req, REGISTER_MR, cmd_flags); @@ -711,7 +696,7 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, req.pbl = 0; pg_size = PAGE_SIZE; } else { - level = mr->hwq.level + 1; + level = mr->hwq.level; req.pbl = cpu_to_le64(mr->hwq.pbl[PBL_LVL_0].pg_map_arr[0]); } pg_size = buf_pg_size ? buf_pg_size : PAGE_SIZE; @@ -728,7 +713,7 @@ int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, req.mr_size = cpu_to_le64(mr->total_size); rc = bnxt_qplib_rcfw_send_message(rcfw, (void *)&req, - (void *)&resp, NULL, block); + (void *)&resp, NULL, false); if (rc) goto fail; diff --git a/drivers/infiniband/hw/bnxt_re/qplib_sp.h b/drivers/infiniband/hw/bnxt_re/qplib_sp.h index 967890cd81f2..bc228340684f 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_sp.h +++ b/drivers/infiniband/hw/bnxt_re/qplib_sp.h @@ -254,7 +254,7 @@ int bnxt_qplib_alloc_mrw(struct bnxt_qplib_res *res, int bnxt_qplib_dereg_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mrw, bool block); int bnxt_qplib_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, - u64 *pbl_tbl, int num_pbls, bool block, u32 buf_pg_size); + struct ib_umem *umem, int num_pbls, u32 buf_pg_size); int bnxt_qplib_free_mrw(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr); int bnxt_qplib_alloc_fast_reg_mr(struct bnxt_qplib_res *res, struct bnxt_qplib_mrw *mr, int max); diff --git a/drivers/infiniband/hw/cxgb4/restrack.c b/drivers/infiniband/hw/cxgb4/restrack.c index b32e6516d65f..ff645b955a08 100644 --- a/drivers/infiniband/hw/cxgb4/restrack.c +++ b/drivers/infiniband/hw/cxgb4/restrack.c @@ -209,7 +209,7 @@ int c4iw_fill_res_cm_id_entry(struct sk_buff *msg, epcp = (struct c4iw_ep_common *)iw_cm_id->provider_data; if (!epcp) return 0; - uep = kcalloc(1, sizeof(*uep), GFP_KERNEL); + uep = kzalloc(sizeof(*uep), GFP_KERNEL); if (!uep) return 0; diff --git a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h index b199e4ac6cf9..fa38b34eddb8 100644 --- a/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h +++ b/drivers/infiniband/hw/efa/efa_admin_cmds_defs.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */ /* - * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2021 Amazon.com, Inc. or its affiliates. All rights reserved. */ #ifndef _EFA_ADMIN_CMDS_H_ @@ -161,8 +161,8 @@ struct efa_admin_create_qp_resp { u32 qp_handle; /* - * QP number in the given EFA virtual device. Least-significant bits - * (as needed according to max_qp) carry unique QP ID + * QP number in the given EFA virtual device. Least-significant bits (as + * needed according to max_qp) carry unique QP ID */ u16 qp_num; @@ -465,7 +465,7 @@ struct efa_admin_create_cq_cmd { /* * number of sub cqs - must be equal to sub_cqs_per_cq of queue - * attributes. + * attributes. */ u16 num_sub_cqs; @@ -563,12 +563,8 @@ struct efa_admin_acq_get_stats_resp { }; struct efa_admin_get_set_feature_common_desc { - /* - * 1:0 : select - 0x1 - current value; 0x3 - default - * value - * 7:3 : reserved3 - MBZ - */ - u8 flags; + /* MBZ */ + u8 reserved0; /* as appears in efa_admin_aq_feature_id */ u8 feature_id; @@ -823,12 +819,6 @@ enum efa_admin_aenq_group { EFA_ADMIN_AENQ_GROUPS_NUM = 5, }; -enum efa_admin_aenq_notification_syndrom { - EFA_ADMIN_SUSPEND = 0, - EFA_ADMIN_RESUME = 1, - EFA_ADMIN_UPDATE_HINTS = 2, -}; - struct efa_admin_mmio_req_read_less_resp { u16 req_id; @@ -909,9 +899,6 @@ struct efa_admin_host_info { #define EFA_ADMIN_CREATE_CQ_CMD_VIRT_MASK BIT(6) #define EFA_ADMIN_CREATE_CQ_CMD_CQ_ENTRY_SIZE_WORDS_MASK GENMASK(4, 0) -/* get_set_feature_common_desc */ -#define EFA_ADMIN_GET_SET_FEATURE_COMMON_DESC_SELECT_MASK GENMASK(1, 0) - /* feature_device_attr_desc */ #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RDMA_READ_MASK BIT(0) #define EFA_ADMIN_FEATURE_DEVICE_ATTR_DESC_RNR_RETRY_MASK BIT(1) diff --git a/drivers/infiniband/hw/efa/efa_admin_defs.h b/drivers/infiniband/hw/efa/efa_admin_defs.h index 29d53ed63b3e..78ff9389ae25 100644 --- a/drivers/infiniband/hw/efa/efa_admin_defs.h +++ b/drivers/infiniband/hw/efa/efa_admin_defs.h @@ -1,6 +1,6 @@ /* SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause */ /* - * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2021 Amazon.com, Inc. or its affiliates. All rights reserved. */ #ifndef _EFA_ADMIN_H_ @@ -82,7 +82,7 @@ struct efa_admin_acq_common_desc { /* * indicates to the driver which AQ entry has been consumed by the - * device and could be reused + * device and could be reused */ u16 sq_head_indx; }; diff --git a/drivers/infiniband/hw/efa/efa_com.c b/drivers/infiniband/hw/efa/efa_com.c index 336bc2c57bb1..0d523ad736c7 100644 --- a/drivers/infiniband/hw/efa/efa_com.c +++ b/drivers/infiniband/hw/efa/efa_com.c @@ -1,6 +1,6 @@ // SPDX-License-Identifier: GPL-2.0 OR BSD-2-Clause /* - * Copyright 2018-2020 Amazon.com, Inc. or its affiliates. All rights reserved. + * Copyright 2018-2021 Amazon.com, Inc. or its affiliates. All rights reserved. */ #include "efa_com.h" @@ -20,9 +20,6 @@ #define EFA_CTRL_MINOR 0 #define EFA_CTRL_SUB_MINOR 1 -#define EFA_DMA_ADDR_TO_UINT32_LOW(x) ((u32)((u64)(x))) -#define EFA_DMA_ADDR_TO_UINT32_HIGH(x) ((u32)(((u64)(x)) >> 32)) - enum efa_cmd_status { EFA_CMD_SUBMITTED, EFA_CMD_COMPLETED, @@ -33,8 +30,6 @@ struct efa_comp_ctx { struct efa_admin_acq_entry *user_cqe; u32 comp_size; enum efa_cmd_status status; - /* status from the device */ - u8 comp_status; u8 cmd_opcode; u8 occupied; }; @@ -140,8 +135,8 @@ static int efa_com_admin_init_sq(struct efa_com_dev *edev) sq->db_addr = (u32 __iomem *)(edev->reg_bar + EFA_REGS_AQ_PROD_DB_OFF); - addr_high = EFA_DMA_ADDR_TO_UINT32_HIGH(sq->dma_addr); - addr_low = EFA_DMA_ADDR_TO_UINT32_LOW(sq->dma_addr); + addr_high = upper_32_bits(sq->dma_addr); + addr_low = lower_32_bits(sq->dma_addr); writel(addr_low, edev->reg_bar + EFA_REGS_AQ_BASE_LO_OFF); writel(addr_high, edev->reg_bar + EFA_REGS_AQ_BASE_HI_OFF); @@ -174,8 +169,8 @@ static int efa_com_admin_init_cq(struct efa_com_dev *edev) cq->cc = 0; cq->phase = 1; - addr_high = EFA_DMA_ADDR_TO_UINT32_HIGH(cq->dma_addr); - addr_low = EFA_DMA_ADDR_TO_UINT32_LOW(cq->dma_addr); + addr_high = upper_32_bits(cq->dma_addr); + addr_low = lower_32_bits(cq->dma_addr); writel(addr_low, edev->reg_bar + EFA_REGS_ACQ_BASE_LO_OFF); writel(addr_high, edev->reg_bar + EFA_REGS_ACQ_BASE_HI_OFF); @@ -215,8 +210,8 @@ static int efa_com_admin_init_aenq(struct efa_com_dev *edev, aenq->cc = 0; aenq->phase = 1; - addr_low = EFA_DMA_ADDR_TO_UINT32_LOW(aenq->dma_addr); - addr_high = EFA_DMA_ADDR_TO_UINT32_HIGH(aenq->dma_addr); + addr_low = lower_32_bits(aenq->dma_addr); + addr_high = upper_32_bits(aenq->dma_addr); writel(addr_low, edev->reg_bar + EFA_REGS_AENQ_BASE_LO_OFF); writel(addr_high, edev->reg_bar + EFA_REGS_AENQ_BASE_HI_OFF); @@ -421,9 +416,7 @@ static void efa_com_handle_single_admin_completion(struct efa_com_admin_queue *a } comp_ctx->status = EFA_CMD_COMPLETED; - comp_ctx->comp_status = cqe->acq_common_descriptor.status; - if (comp_ctx->user_cqe) - memcpy(comp_ctx->user_cqe, cqe, comp_ctx->comp_size); + memcpy(comp_ctx->user_cqe, cqe, comp_ctx->comp_size); if (!test_bit(EFA_AQ_STATE_POLLING_BIT, &aq->state)) complete(&comp_ctx->wait_event); @@ -521,7 +514,7 @@ static int efa_com_wait_and_process_admin_cq_polling(struct efa_comp_ctx *comp_c msleep(aq->poll_interval); } - err = efa_com_comp_status_to_errno(comp_ctx->comp_status); + err = efa_com_comp_status_to_errno(comp_ctx->user_cqe->acq_common_descriptor.status); out: efa_com_put_comp_ctx(aq, comp_ctx); return err; @@ -569,7 +562,7 @@ static int efa_com_wait_and_process_admin_cq_interrupts(struct efa_comp_ctx *com goto out; } - err = efa_com_comp_status_to_errno(comp_ctx->comp_status); + err = efa_com_comp_status_to_errno(comp_ctx->user_cqe->acq_common_descriptor.status); out: efa_com_put_comp_ctx(aq, comp_ctx); return err; @@ -641,8 +634,8 @@ int efa_com_cmd_exec(struct efa_com_admin_queue *aq, 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); + cmd->aq_common_descriptor.opcode, + comp_ctx->user_cqe->acq_common_descriptor.status, err); atomic64_inc(&aq->stats.cmd_err); } @@ -795,7 +788,7 @@ err_destroy_comp_ctxt: * This method goes over the admin completion queue and wakes up * all the pending threads that wait on the commands wait event. * - * @note: Should be called after MSI-X interrupt. + * Note: Should be called after MSI-X interrupt. */ void efa_com_admin_q_comp_intr_handler(struct efa_com_dev *edev) { diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c index c87b94ea2939..993cbf37e0b9 100644 --- a/drivers/infiniband/hw/hfi1/chip.c +++ b/drivers/infiniband/hw/hfi1/chip.c @@ -1323,8 +1323,8 @@ CNTR_ELEM(#name, \ /** * hfi_addr_from_offset - return addr for readq/writeq - * @dd - the dd device - * @offset - the offset of the CSR within bar0 + * @dd: the dd device + * @offset: the offset of the CSR within bar0 * * This routine selects the appropriate base address * based on the indicated offset. @@ -1340,8 +1340,8 @@ static inline void __iomem *hfi1_addr_from_offset( /** * read_csr - read CSR at the indicated offset - * @dd - the dd device - * @offset - the offset of the CSR within bar0 + * @dd: the dd device + * @offset: the offset of the CSR within bar0 * * Return: the value read or all FF's if there * is no mapping @@ -1355,9 +1355,9 @@ u64 read_csr(const struct hfi1_devdata *dd, u32 offset) /** * write_csr - write CSR at the indicated offset - * @dd - the dd device - * @offset - the offset of the CSR within bar0 - * @value - value to write + * @dd: the dd device + * @offset: the offset of the CSR within bar0 + * @value: value to write */ void write_csr(const struct hfi1_devdata *dd, u32 offset, u64 value) { @@ -1373,8 +1373,8 @@ void write_csr(const struct hfi1_devdata *dd, u32 offset, u64 value) /** * get_csr_addr - return te iomem address for offset - * @dd - the dd device - * @offset - the offset of the CSR within bar0 + * @dd: the dd device + * @offset: the offset of the CSR within bar0 * * Return: The iomem address to use in subsequent * writeq/readq operations. @@ -8433,7 +8433,7 @@ static inline int check_packet_present(struct hfi1_ctxtdata *rcd) return hfi1_rcd_head(rcd) != tail; } -/** +/* * Common code for receive contexts interrupt handlers. * Update traces, increment kernel IRQ counter and * setup ASPM when needed. @@ -8447,7 +8447,7 @@ static void receive_interrupt_common(struct hfi1_ctxtdata *rcd) aspm_ctx_disable(rcd); } -/** +/* * __hfi1_rcd_eoi_intr() - Make HW issue receive interrupt * when there are packets present in the queue. When calling * with interrupts enabled please use hfi1_rcd_eoi_intr. @@ -8484,8 +8484,8 @@ static void hfi1_rcd_eoi_intr(struct hfi1_ctxtdata *rcd) /** * hfi1_netdev_rx_napi - napi poll function to move eoi inline - * @napi - pointer to napi object - * @budget - netdev budget + * @napi: pointer to napi object + * @budget: netdev budget */ int hfi1_netdev_rx_napi(struct napi_struct *napi, int budget) { @@ -10142,7 +10142,7 @@ u32 lrh_max_header_bytes(struct hfi1_devdata *dd) /* * Set Send Length - * @ppd - per port data + * @ppd: per port data * * Set the MTU by limiting how many DWs may be sent. The SendLenCheck* * registers compare against LRH.PktLen, so use the max bytes included @@ -14200,9 +14200,9 @@ u8 hfi1_get_qp_map(struct hfi1_devdata *dd, u8 idx) /** * init_qpmap_table - * @dd - device data - * @first_ctxt - first context - * @last_ctxt - first context + * @dd: device data + * @first_ctxt: first context + * @last_ctxt: first context * * This return sets the qpn mapping table that * is indexed by qpn[8:1]. @@ -14383,8 +14383,8 @@ no_qos: /** * init_qos - init RX qos - * @dd - device data - * @rmt - RSM map table + * @dd: device data + * @rmt: RSM map table * * This routine initializes Rule 0 and the RSM map table to implement * quality of service (qos). @@ -15022,8 +15022,7 @@ err_exit: /** * hfi1_init_dd() - Initialize most of the dd structure. - * @dev: the pci_dev for hfi1_ib device - * @ent: pci_device_id struct for this dev + * @dd: the dd device * * This is global, and is called directly at init to set up the * chip-specific function pointers for later use. @@ -15378,10 +15377,11 @@ static u16 delay_cycles(struct hfi1_pportdata *ppd, u32 desired_egress_rate, /** * create_pbc - build a pbc for transmission + * @ppd: info of physical Hfi port * @flags: special case flags or-ed in built pbc - * @srate: static rate + * @srate_mbs: static rate * @vl: vl - * @dwlen: dword length (header words + data words + pbc words) + * @dw_len: dword length (header words + data words + pbc words) * * Create a PBC with the given flags, rate, VL, and length. * diff --git a/drivers/infiniband/hw/hfi1/exp_rcv.c b/drivers/infiniband/hw/hfi1/exp_rcv.c index e9d5cc8b771a..91f13140ddf2 100644 --- a/drivers/infiniband/hw/hfi1/exp_rcv.c +++ b/drivers/infiniband/hw/hfi1/exp_rcv.c @@ -50,7 +50,7 @@ /** * exp_tid_group_init - initialize exp_tid_set - * @set - the set + * @set: the set */ static void hfi1_exp_tid_set_init(struct exp_tid_set *set) { @@ -60,7 +60,7 @@ static void hfi1_exp_tid_set_init(struct exp_tid_set *set) /** * hfi1_exp_tid_group_init - initialize rcd expected receive - * @rcd - the rcd + * @rcd: the rcd */ void hfi1_exp_tid_group_init(struct hfi1_ctxtdata *rcd) { @@ -71,7 +71,7 @@ void hfi1_exp_tid_group_init(struct hfi1_ctxtdata *rcd) /** * alloc_ctxt_rcv_groups - initialize expected receive groups - * @rcd - the context to add the groupings to + * @rcd: the context to add the groupings to */ int hfi1_alloc_ctxt_rcv_groups(struct hfi1_ctxtdata *rcd) { @@ -101,7 +101,7 @@ int hfi1_alloc_ctxt_rcv_groups(struct hfi1_ctxtdata *rcd) /** * free_ctxt_rcv_groups - free expected receive groups - * @rcd - the context to free + * @rcd: the context to free * * The routine dismantles the expect receive linked * list and clears any tids associated with the receive diff --git a/drivers/infiniband/hw/hfi1/file_ops.c b/drivers/infiniband/hw/hfi1/file_ops.c index 329ee4f48d95..3b7bbc7b9d10 100644 --- a/drivers/infiniband/hw/hfi1/file_ops.c +++ b/drivers/infiniband/hw/hfi1/file_ops.c @@ -1522,7 +1522,7 @@ int hfi1_set_uevent_bits(struct hfi1_pportdata *ppd, const int evtbit) * manage_rcvq - manage a context's receive queue * @uctxt: the context * @subctxt: the sub-context - * @start_stop: action to carry out + * @arg: start/stop action to carry out * * start_stop == 0 disables receive on the context, for use in queue * overflow conditions. start_stop==1 re-enables, to be used to diff --git a/drivers/infiniband/hw/hfi1/intr.c b/drivers/infiniband/hw/hfi1/intr.c index 387305b768e9..5ba5c11459e7 100644 --- a/drivers/infiniband/hw/hfi1/intr.c +++ b/drivers/infiniband/hw/hfi1/intr.c @@ -91,9 +91,9 @@ static void add_full_mgmt_pkey(struct hfi1_pportdata *ppd) /** * format_hwmsg - format a single hwerror message - * @msg message buffer - * @msgl length of message buffer - * @hwmsg message to add to message buffer + * @msg: message buffer + * @msgl: length of message buffer + * @hwmsg: message to add to message buffer */ static void format_hwmsg(char *msg, size_t msgl, const char *hwmsg) { @@ -104,11 +104,11 @@ static void format_hwmsg(char *msg, size_t msgl, const char *hwmsg) /** * hfi1_format_hwerrors - format hardware error messages for display - * @hwerrs hardware errors bit vector - * @hwerrmsgs hardware error descriptions - * @nhwerrmsgs number of hwerrmsgs - * @msg message buffer - * @msgl message buffer length + * @hwerrs: hardware errors bit vector + * @hwerrmsgs: hardware error descriptions + * @nhwerrmsgs: number of hwerrmsgs + * @msg: message buffer + * @msgl: message buffer length */ void hfi1_format_hwerrors(u64 hwerrs, const struct hfi1_hwerror_msgs *hwerrmsgs, size_t nhwerrmsgs, char *msg, size_t msgl) diff --git a/drivers/infiniband/hw/hfi1/iowait.c b/drivers/infiniband/hw/hfi1/iowait.c index 5836fe7b2817..111489802614 100644 --- a/drivers/infiniband/hw/hfi1/iowait.c +++ b/drivers/infiniband/hw/hfi1/iowait.c @@ -26,7 +26,7 @@ inline void iowait_clear_flag(struct iowait *wait, u32 flag) clear_bit(flag, &wait->flags); } -/** +/* * iowait_init() - initialize wait structure * @wait: wait struct to initialize * @tx_limit: limit for overflow queuing @@ -88,7 +88,7 @@ void iowait_cancel_work(struct iowait *w) /** * iowait_set_work_flag - set work flag based on leg - * @w - the iowait work struct + * @w: the iowait work struct */ int iowait_set_work_flag(struct iowait_work *w) { diff --git a/drivers/infiniband/hw/hfi1/mad.c b/drivers/infiniband/hw/hfi1/mad.c index 3222e3acb79c..e2f2f7847aed 100644 --- a/drivers/infiniband/hw/hfi1/mad.c +++ b/drivers/infiniband/hw/hfi1/mad.c @@ -1341,7 +1341,7 @@ static int set_port_states(struct hfi1_pportdata *ppd, struct opa_smp *smp, return 0; } -/** +/* * subn_set_opa_portinfo - set port information * @smp: the incoming SM packet * @ibdev: the infiniband device @@ -4902,6 +4902,8 @@ static int hfi1_process_ib_mad(struct ib_device *ibdev, int mad_flags, u8 port, * @in_grh: the global route header for this packet * @in_mad: the incoming MAD * @out_mad: any outgoing MAD reply + * @out_mad_size: size of the outgoing MAD reply + * @out_mad_pkey_index: used to apss back the packet key index * * Returns IB_MAD_RESULT_SUCCESS if this is a MAD that we are not * interested in processing. diff --git a/drivers/infiniband/hw/hfi1/msix.c b/drivers/infiniband/hw/hfi1/msix.c index d61ee853d215..cf3040bb177f 100644 --- a/drivers/infiniband/hw/hfi1/msix.c +++ b/drivers/infiniband/hw/hfi1/msix.c @@ -103,8 +103,8 @@ int msix_initialize(struct hfi1_devdata *dd) * @arg: context information for the IRQ * @handler: IRQ handler * @thread: IRQ thread handler (could be NULL) - * @idx: zero base idx if multiple devices are needed * @type: affinty IRQ type + * @name: IRQ name * * Allocated an MSIx vector if available, and then create the appropriate * meta data needed to keep track of the pci IRQ request. diff --git a/drivers/infiniband/hw/hfi1/netdev_rx.c b/drivers/infiniband/hw/hfi1/netdev_rx.c index 6d263c9749b3..1fb6e1a0e4e1 100644 --- a/drivers/infiniband/hw/hfi1/netdev_rx.c +++ b/drivers/infiniband/hw/hfi1/netdev_rx.c @@ -467,7 +467,7 @@ void *hfi1_netdev_get_data(struct hfi1_devdata *dd, int id) * hfi1_netdev_get_first_dat - Gets first entry with greater or equal id. * * @dd: hfi1 dev data - * @id: requested integer id up to INT_MAX + * @start_id: requested integer id up to INT_MAX */ void *hfi1_netdev_get_first_data(struct hfi1_devdata *dd, int *start_id) { diff --git a/drivers/infiniband/hw/hfi1/pcie.c b/drivers/infiniband/hw/hfi1/pcie.c index 18d32f053d26..6f06e9920503 100644 --- a/drivers/infiniband/hw/hfi1/pcie.c +++ b/drivers/infiniband/hw/hfi1/pcie.c @@ -334,7 +334,7 @@ int pcie_speeds(struct hfi1_devdata *dd) return 0; } -/** +/* * Restore command and BARs after a reset has wiped them out * * Returns 0 on success, otherwise a negative error value @@ -393,7 +393,7 @@ error: return pcibios_err_to_errno(ret); } -/** +/* * Save BARs and command to rewrite after device reset * * Returns 0 on success, otherwise a negative error value diff --git a/drivers/infiniband/hw/hfi1/pio_copy.c b/drivers/infiniband/hw/hfi1/pio_copy.c index 4a4ec2397857..14bfd8287f4a 100644 --- a/drivers/infiniband/hw/hfi1/pio_copy.c +++ b/drivers/infiniband/hw/hfi1/pio_copy.c @@ -55,6 +55,7 @@ /** * pio_copy - copy data block to MMIO space + * @dd: hfi1 dev data * @pbuf: a number of blocks allocated within a PIO send context * @pbc: PBC to send * @from: source, must be 8 byte aligned diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c index 681bb4e918c9..e037df911512 100644 --- a/drivers/infiniband/hw/hfi1/qp.c +++ b/drivers/infiniband/hw/hfi1/qp.c @@ -186,7 +186,7 @@ static void flush_iowait(struct rvt_qp *qp) write_sequnlock_irqrestore(lock, flags); } -/** +/* * This function is what we would push to the core layer if we wanted to be a * "first class citizen". Instead we hide this here and rely on Verbs ULPs * to blindly pass the MTU enum value from the PathRecord to us. @@ -289,9 +289,9 @@ void hfi1_modify_qp(struct rvt_qp *qp, struct ib_qp_attr *attr, /** * hfi1_setup_wqe - set up the wqe - * @qp - The qp - * @wqe - The built wqe - * @call_send - Determine if the send should be posted or scheduled. + * @qp: The qp + * @wqe: The built wqe + * @call_send: Determine if the send should be posted or scheduled. * * Perform setup of the wqe. This is called * prior to inserting the wqe into the ring but after @@ -595,7 +595,7 @@ struct sdma_engine *qp_to_sdma_engine(struct rvt_qp *qp, u8 sc5) return sde; } -/* +/** * qp_to_send_context - map a qp to a send context * @qp: the QP * @sc5: the 5 bit sc @@ -912,8 +912,8 @@ void notify_error_qp(struct rvt_qp *qp) /** * hfi1_qp_iter_cb - callback for iterator - * @qp - the qp - * @v - the sl in low bits of v + * @qp: the qp + * @v: the sl in low bits of v * * This is called from the iterator callback to work * on an individual qp. diff --git a/drivers/infiniband/hw/hfi1/qsfp.c b/drivers/infiniband/hw/hfi1/qsfp.c index 8386c84c2d92..38f311f855b5 100644 --- a/drivers/infiniband/hw/hfi1/qsfp.c +++ b/drivers/infiniband/hw/hfi1/qsfp.c @@ -242,7 +242,7 @@ static int i2c_bus_write(struct hfi1_devdata *dd, struct hfi1_i2c_bus *i2c, msgs[0].buf = offset_bytes; msgs[1].addr = slave_addr; - msgs[1].flags = I2C_M_NOSTART, + msgs[1].flags = I2C_M_NOSTART; msgs[1].len = len; msgs[1].buf = data; break; @@ -290,7 +290,7 @@ static int i2c_bus_read(struct hfi1_devdata *dd, struct hfi1_i2c_bus *bus, msgs[0].buf = offset_bytes; msgs[1].addr = slave_addr; - msgs[1].flags = I2C_M_RD, + msgs[1].flags = I2C_M_RD; msgs[1].len = len; msgs[1].buf = data; break; diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c index 1bb5f57152d3..0174b8ee9f00 100644 --- a/drivers/infiniband/hw/hfi1/rc.c +++ b/drivers/infiniband/hw/hfi1/rc.c @@ -421,6 +421,7 @@ bail: /** * hfi1_make_rc_req - construct a request packet (SEND, RDMA r/w, ATOMIC) * @qp: a pointer to the QP + * @ps: the current packet state * * Assumes s_lock is held. * @@ -1375,9 +1376,8 @@ static const hfi1_make_rc_ack hfi1_make_rc_ack_tbl[2] = { [HFI1_PKT_TYPE_16B] = &hfi1_make_rc_ack_16B }; -/** +/* * hfi1_send_rc_ack - Construct an ACK packet and send it - * @qp: a pointer to the QP * * This is called from hfi1_rc_rcv() and handle_receive_interrupt(). * Note that RDMA reads and atomics are handled in the @@ -1992,7 +1992,7 @@ static void update_qp_retry_state(struct rvt_qp *qp, u32 psn, u32 spsn, } } -/** +/* * do_rc_ack - process an incoming RC ACK * @qp: the QP the ACK came in on * @psn: the packet sequence number of the ACK @@ -2541,6 +2541,7 @@ static inline void rc_cancel_ack(struct rvt_qp *qp) * @opcode: the opcode for this packet * @psn: the packet sequence number for this packet * @diff: the difference between the PSN and the expected PSN + * @rcd: the receive context * * This is called from hfi1_rc_rcv() to process an unexpected * incoming RC packet for the given QP. diff --git a/drivers/infiniband/hw/hfi1/ruc.c b/drivers/infiniband/hw/hfi1/ruc.c index 23ac6057b211..c3fa1814c6a8 100644 --- a/drivers/infiniband/hw/hfi1/ruc.c +++ b/drivers/infiniband/hw/hfi1/ruc.c @@ -260,6 +260,7 @@ static inline void hfi1_make_ruc_bth(struct rvt_qp *qp, * @qp: the queue pair * @ohdr: a pointer to the destination header memory * @bth0: bth0 passed in from the RC/UC builder + * @bth1: bth1 passed in from the RC/UC builder * @bth2: bth2 passed in from the RC/UC builder * @middle: non zero implies indicates ahg "could" be used * @ps: the current packet state @@ -348,6 +349,7 @@ static inline void hfi1_make_ruc_header_16B(struct rvt_qp *qp, * @qp: the queue pair * @ohdr: a pointer to the destination header memory * @bth0: bth0 passed in from the RC/UC builder + * @bth1: bth1 passed in from the RC/UC builder * @bth2: bth2 passed in from the RC/UC builder * @middle: non zero implies indicates ahg "could" be used * @ps: the current packet state @@ -455,11 +457,10 @@ void hfi1_make_ruc_header(struct rvt_qp *qp, struct ib_other_headers *ohdr, /** * hfi1_schedule_send_yield - test for a yield required for QP * send engine - * @timeout: Final time for timeout slice for jiffies * @qp: a pointer to QP * @ps: a pointer to a structure with commonly lookup values for * the the send engine progress - * @tid - true if it is the tid leg + * @tid: true if it is the tid leg * * This routine checks if the time slice for the QP has expired * for RC QPs, if so an additional work entry is queued. At this diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c index a307d4c8b15a..46b5290b2839 100644 --- a/drivers/infiniband/hw/hfi1/sdma.c +++ b/drivers/infiniband/hw/hfi1/sdma.c @@ -1740,7 +1740,7 @@ retry: sane = (hwhead == swhead); if (unlikely(!sane)) { - dd_dev_err(dd, "SDMA(%u) bad head (%s) hwhd=%hu swhd=%hu swtl=%hu cnt=%hu\n", + dd_dev_err(dd, "SDMA(%u) bad head (%s) hwhd=%u swhd=%u swtl=%u cnt=%u\n", sde->this_idx, use_dmahead ? "dma" : "kreg", hwhead, swhead, swtail, cnt); @@ -2448,11 +2448,11 @@ nodesc: * @sde: sdma engine to use * @wait: SE wait structure to use when full (may be NULL) * @tx_list: list of sdma_txreqs to submit - * @count: pointer to a u16 which, after return will contain the total number of - * sdma_txreqs removed from the tx_list. This will include sdma_txreqs - * whose SDMA descriptors are submitted to the ring and the sdma_txreqs - * which are added to SDMA engine flush list if the SDMA engine state is - * not running. + * @count_out: pointer to a u16 which, after return will contain the total number of + * sdma_txreqs removed from the tx_list. This will include sdma_txreqs + * whose SDMA descriptors are submitted to the ring and the sdma_txreqs + * which are added to SDMA engine flush list if the SDMA engine state is + * not running. * * The call submits the list into the ring. * diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c index 92aa2a9b3b5a..0b1f9e4d038b 100644 --- a/drivers/infiniband/hw/hfi1/tid_rdma.c +++ b/drivers/infiniband/hw/hfi1/tid_rdma.c @@ -309,7 +309,8 @@ int hfi1_kern_exp_rcv_init(struct hfi1_ctxtdata *rcd, int reinit) /** * qp_to_rcd - determine the receive context used by a qp - * @qp - the qp + * @rdi: rvt dev struct + * @qp: the qp * * This routine returns the receive context associated * with a a qp's qpn. @@ -484,6 +485,7 @@ static struct rvt_qp *first_qp(struct hfi1_ctxtdata *rcd, /** * kernel_tid_waiters - determine rcd wait * @rcd: the receive context + * @queue: the queue to operate on * @qp: the head of the qp being processed * * This routine will return false IFF @@ -517,7 +519,9 @@ static bool kernel_tid_waiters(struct hfi1_ctxtdata *rcd, /** * dequeue_tid_waiter - dequeue the qp from the list - * @qp - the qp to remove the wait list + * @rcd: the receive context + * @queue: the queue to operate on + * @qp: the qp to remove the wait list * * This routine removes the indicated qp from the * wait list if it is there. @@ -549,6 +553,7 @@ static void dequeue_tid_waiter(struct hfi1_ctxtdata *rcd, /** * queue_qp_for_tid_wait - suspend QP on tid space * @rcd: the receive context + * @queue: the queue to operate on * @qp: the qp * * The qp is inserted at the tail of the rcd @@ -593,7 +598,7 @@ static void __trigger_tid_waiter(struct rvt_qp *qp) /** * tid_rdma_schedule_tid_wakeup - schedule wakeup for a qp - * @qp - the qp + * @qp: the qp * * trigger a schedule or a waiting qp in a deadlock * safe manner. The qp reference is held prior @@ -630,7 +635,7 @@ static void tid_rdma_schedule_tid_wakeup(struct rvt_qp *qp) /** * tid_rdma_trigger_resume - field a trigger work request - * @work - the work item + * @work: the work item * * Complete the off qp trigger processing by directly * calling the progress routine. @@ -654,7 +659,7 @@ static void tid_rdma_trigger_resume(struct work_struct *work) rvt_put_qp(qp); } -/** +/* * tid_rdma_flush_wait - unwind any tid space wait * * This is called when resetting a qp to @@ -693,8 +698,8 @@ void hfi1_tid_rdma_flush_wait(struct rvt_qp *qp) /* Flow functions */ /** * kern_reserve_flow - allocate a hardware flow - * @rcd - the context to use for allocation - * @last - the index of the preferred flow. Use RXE_NUM_TID_FLOWS to + * @rcd: the context to use for allocation + * @last: the index of the preferred flow. Use RXE_NUM_TID_FLOWS to * signify "don't care". * * Use a bit mask based allocation to reserve a hardware @@ -860,9 +865,10 @@ static u8 trdma_pset_order(struct tid_rdma_pageset *s) /** * tid_rdma_find_phys_blocks_4k - get groups base on mr info - * @npages - number of pages - * @pages - pointer to an array of page structs - * @list - page set array to return + * @flow: overall info for a TID RDMA segment + * @pages: pointer to an array of page structs + * @npages: number of pages + * @list: page set array to return * * This routine returns the number of groups associated with * the current sge information. This implementation is based @@ -949,10 +955,10 @@ static u32 tid_rdma_find_phys_blocks_4k(struct tid_rdma_flow *flow, /** * tid_flush_pages - dump out pages into pagesets - * @list - list of pagesets - * @idx - pointer to current page index - * @pages - number of pages to dump - * @sets - current number of pagesset + * @list: list of pagesets + * @idx: pointer to current page index + * @pages: number of pages to dump + * @sets: current number of pagesset * * This routine flushes out accumuated pages. * @@ -990,9 +996,10 @@ static u32 tid_flush_pages(struct tid_rdma_pageset *list, /** * tid_rdma_find_phys_blocks_8k - get groups base on mr info - * @pages - pointer to an array of page structs - * @npages - number of pages - * @list - page set array to return + * @flow: overall info for a TID RDMA segment + * @pages: pointer to an array of page structs + * @npages: number of pages + * @list: page set array to return * * This routine parses an array of pages to compute pagesets * in an 8k compatible way. @@ -1064,7 +1071,7 @@ static u32 tid_rdma_find_phys_blocks_8k(struct tid_rdma_flow *flow, return sets; } -/** +/* * Find pages for one segment of a sge array represented by @ss. The function * does not check the sge, the sge must have been checked for alignment with a * prior call to hfi1_kern_trdma_ok. Other sge checking is done as part of @@ -1598,7 +1605,7 @@ void hfi1_kern_exp_rcv_clear_all(struct tid_rdma_request *req) /** * hfi1_kern_exp_rcv_free_flows - free priviously allocated flow information - * @req - the tid rdma request to be cleaned + * @req: the tid rdma request to be cleaned */ static void hfi1_kern_exp_rcv_free_flows(struct tid_rdma_request *req) { @@ -3435,7 +3442,7 @@ static u32 hfi1_compute_tid_rnr_timeout(struct rvt_qp *qp, u32 to_seg) return 0; } -/** +/* * Central place for resource allocation at TID write responder, * is called from write_req and write_data interrupt handlers as * well as the send thread when a queued QP is scheduled for diff --git a/drivers/infiniband/hw/hfi1/uc.c b/drivers/infiniband/hw/hfi1/uc.c index 1fb918399da0..5b0f536b34e0 100644 --- a/drivers/infiniband/hw/hfi1/uc.c +++ b/drivers/infiniband/hw/hfi1/uc.c @@ -55,6 +55,7 @@ /** * hfi1_make_uc_req - construct a request packet (SEND, RDMA write) * @qp: a pointer to the QP + * @ps: the current packet state * * Assume s_lock is held. * @@ -291,12 +292,7 @@ bail_no_tx: /** * hfi1_uc_rcv - handle an incoming UC packet - * @ibp: the port the packet came in on - * @hdr: the header of the packet - * @rcv_flags: flags relevant to rcv processing - * @data: the packet data - * @tlen: the length of the packet - * @qp: the QP for this packet. + * @packet: the packet structure * * This is called from qp_rcv() to process an incoming UC packet * for the given QP. diff --git a/drivers/infiniband/hw/hfi1/ud.c b/drivers/infiniband/hw/hfi1/ud.c index e804af71b629..6ecb984c85fa 100644 --- a/drivers/infiniband/hw/hfi1/ud.c +++ b/drivers/infiniband/hw/hfi1/ud.c @@ -468,6 +468,7 @@ void hfi1_make_ud_req_16B(struct rvt_qp *qp, struct hfi1_pkt_state *ps, /** * hfi1_make_ud_req - construct a UD request packet * @qp: the QP + * @ps: the current packet state * * Assume s_lock is held. * @@ -840,12 +841,7 @@ static int opa_smp_check(struct hfi1_ibport *ibp, u16 pkey, u8 sc5, /** * hfi1_ud_rcv - receive an incoming UD packet - * @ibp: the port the packet came in on - * @hdr: the packet header - * @rcv_flags: flags relevant to rcv processing - * @data: the packet data - * @tlen: the packet length - * @qp: the QP the packet came on + * @packet: the packet structure * * This is called from qp_rcv() to process an incoming UD packet * for the given QP. diff --git a/drivers/infiniband/hw/hfi1/user_exp_rcv.c b/drivers/infiniband/hw/hfi1/user_exp_rcv.c index b94fc7fd75a9..58dcab2679d9 100644 --- a/drivers/infiniband/hw/hfi1/user_exp_rcv.c +++ b/drivers/infiniband/hw/hfi1/user_exp_rcv.c @@ -154,12 +154,12 @@ void hfi1_user_exp_rcv_free(struct hfi1_filedata *fd) fd->entry_to_rb = NULL; } -/** +/* * Release pinned receive buffer pages. * - * @mapped - true if the pages have been DMA mapped. false otherwise. - * @idx - Index of the first page to unpin. - * @npages - No of pages to unpin. + * @mapped: true if the pages have been DMA mapped. false otherwise. + * @idx: Index of the first page to unpin. + * @npages: No of pages to unpin. * * If the pages have been DMA mapped (indicated by mapped parameter), their * info will be passed via a struct tid_rb_node. If they haven't been mapped, @@ -189,7 +189,7 @@ static void unpin_rcv_pages(struct hfi1_filedata *fd, fd->tid_n_pinned -= npages; } -/** +/* * Pin receive buffer pages. */ static int pin_rcv_pages(struct hfi1_filedata *fd, struct tid_user_buf *tidbuf) diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 3591923abebb..0dd4bb0a5a7e 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -729,7 +729,7 @@ bail_txadd: /** * update_tx_opstats - record stats by opcode - * @qp; the qp + * @qp: the qp * @ps: transmit packet state * @plen: the plen in dwords * @@ -1145,7 +1145,7 @@ static inline int egress_pkey_matches_entry(u16 pkey, u16 ent) * egress_pkey_check - check P_KEY of a packet * @ppd: Physical IB port data * @slid: SLID for packet - * @bkey: PKEY for header + * @pkey: PKEY for header * @sc5: SC for packet * @s_pkey_index: It will be used for look up optimization for kernel contexts * only. If it is negative value, then it means user contexts is calling this @@ -1206,7 +1206,7 @@ bad: return 1; } -/** +/* * get_send_routine - choose an egress routine * * Choose an egress routine based on QP type diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h index 5afee04fb02c..23c438cef40d 100644 --- a/drivers/infiniband/hw/hns/hns_roce_common.h +++ b/drivers/infiniband/hw/hns/hns_roce_common.h @@ -32,6 +32,7 @@ #ifndef _HNS_ROCE_COMMON_H #define _HNS_ROCE_COMMON_H +#include <linux/bitfield.h> #define roce_write(dev, reg, val) writel((val), (dev)->reg_base + (reg)) #define roce_read(dev, reg) readl((dev)->reg_base + (reg)) @@ -65,6 +66,27 @@ #define hr_reg_enable(ptr, field) _hr_reg_enable(ptr, field) +#define _hr_reg_clear(ptr, field_type, field_h, field_l) \ + ({ \ + const field_type *_ptr = ptr; \ + *((__le32 *)_ptr + (field_h) / 32) &= \ + cpu_to_le32( \ + ~GENMASK((field_h) % 32, (field_l) % 32)) + \ + BUILD_BUG_ON_ZERO(((field_h) / 32) != \ + ((field_l) / 32)); \ + }) + +#define hr_reg_clear(ptr, field) _hr_reg_clear(ptr, field) + +#define _hr_reg_write(ptr, field_type, field_h, field_l, val) \ + ({ \ + _hr_reg_clear(ptr, field_type, field_h, field_l); \ + *((__le32 *)ptr + (field_h) / 32) |= cpu_to_le32(FIELD_PREP( \ + GENMASK((field_h) % 32, (field_l) % 32), val)); \ + }) + +#define hr_reg_write(ptr, field, val) _hr_reg_write(ptr, field, val) + #define ROCEE_GLB_CFG_ROCEE_DB_SQ_MODE_S 3 #define ROCEE_GLB_CFG_ROCEE_DB_OTH_MODE_S 4 @@ -342,8 +364,8 @@ #define ROCEE_TX_CMQ_BASEADDR_L_REG 0x07000 #define ROCEE_TX_CMQ_BASEADDR_H_REG 0x07004 #define ROCEE_TX_CMQ_DEPTH_REG 0x07008 -#define ROCEE_TX_CMQ_TAIL_REG 0x07010 -#define ROCEE_TX_CMQ_HEAD_REG 0x07014 +#define ROCEE_TX_CMQ_HEAD_REG 0x07010 +#define ROCEE_TX_CMQ_TAIL_REG 0x07014 #define ROCEE_RX_CMQ_BASEADDR_L_REG 0x07018 #define ROCEE_RX_CMQ_BASEADDR_H_REG 0x0701c diff --git a/drivers/infiniband/hw/hns/hns_roce_cq.c b/drivers/infiniband/hw/hns/hns_roce_cq.c index 8533fc2d8df2..74fc4940b03a 100644 --- a/drivers/infiniband/hw/hns/hns_roce_cq.c +++ b/drivers/infiniband/hw/hns/hns_roce_cq.c @@ -38,11 +38,74 @@ #include "hns_roce_hem.h" #include "hns_roce_common.h" +static u8 get_least_load_bankid_for_cq(struct hns_roce_bank *bank) +{ + u32 least_load = bank[0].inuse; + u8 bankid = 0; + u32 bankcnt; + u8 i; + + for (i = 1; i < HNS_ROCE_CQ_BANK_NUM; i++) { + bankcnt = bank[i].inuse; + if (bankcnt < least_load) { + least_load = bankcnt; + bankid = i; + } + } + + return bankid; +} + +static int alloc_cqn(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) +{ + struct hns_roce_cq_table *cq_table = &hr_dev->cq_table; + struct hns_roce_bank *bank; + u8 bankid; + int id; + + mutex_lock(&cq_table->bank_mutex); + bankid = get_least_load_bankid_for_cq(cq_table->bank); + bank = &cq_table->bank[bankid]; + + id = ida_alloc_range(&bank->ida, bank->min, bank->max, GFP_KERNEL); + if (id < 0) { + mutex_unlock(&cq_table->bank_mutex); + return id; + } + + /* the lower 2 bits is bankid */ + hr_cq->cqn = (id << CQ_BANKID_SHIFT) | bankid; + bank->inuse++; + mutex_unlock(&cq_table->bank_mutex); + + return 0; +} + +static inline u8 get_cq_bankid(unsigned long cqn) +{ + /* The lower 2 bits of CQN are used to hash to different banks */ + return (u8)(cqn & GENMASK(1, 0)); +} + +static void free_cqn(struct hns_roce_dev *hr_dev, unsigned long cqn) +{ + struct hns_roce_cq_table *cq_table = &hr_dev->cq_table; + struct hns_roce_bank *bank; + + bank = &cq_table->bank[get_cq_bankid(cqn)]; + + ida_free(&bank->ida, cqn >> CQ_BANKID_SHIFT); + + mutex_lock(&cq_table->bank_mutex); + bank->inuse--; + mutex_unlock(&cq_table->bank_mutex); +} + static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) { + struct hns_roce_cq_table *cq_table = &hr_dev->cq_table; struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_cmd_mailbox *mailbox; - struct hns_roce_cq_table *cq_table; u64 mtts[MTT_MIN_COUNT] = { 0 }; dma_addr_t dma_handle; int ret; @@ -54,13 +117,6 @@ static int alloc_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) return -EINVAL; } - cq_table = &hr_dev->cq_table; - ret = hns_roce_bitmap_alloc(&cq_table->bitmap, &hr_cq->cqn); - if (ret) { - ibdev_err(ibdev, "failed to alloc CQ bitmap, ret = %d.\n", ret); - return ret; - } - /* Get CQC memory HEM(Hardware Entry Memory) table */ ret = hns_roce_table_get(hr_dev, &cq_table->table, hr_cq->cqn); if (ret) { @@ -110,7 +166,6 @@ err_put: hns_roce_table_put(hr_dev, &cq_table->table, hr_cq->cqn); err_out: - hns_roce_bitmap_free(&cq_table->bitmap, hr_cq->cqn, BITMAP_NO_RR); return ret; } @@ -138,7 +193,6 @@ static void free_cqc(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq) wait_for_completion(&hr_cq->free); hns_roce_table_put(hr_dev, &cq_table->table, hr_cq->cqn); - hns_roce_bitmap_free(&cq_table->bitmap, hr_cq->cqn, BITMAP_NO_RR); } static int alloc_cq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, @@ -152,7 +206,6 @@ static int alloc_cq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_cq *hr_cq, buf_attr.region[0].size = hr_cq->cq_depth * hr_cq->cqe_size; buf_attr.region[0].hopnum = hr_dev->caps.cqe_hop_num; buf_attr.region_count = 1; - buf_attr.fixed_page = true; ret = hns_roce_mtr_create(hr_dev, &hr_cq->mtr, &buf_attr, hr_dev->caps.cqe_ba_pg_sz + HNS_HW_PAGE_SHIFT, @@ -298,11 +351,17 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, goto err_cq_buf; } + ret = alloc_cqn(hr_dev, hr_cq); + if (ret) { + ibdev_err(ibdev, "failed to alloc CQN, ret = %d.\n", ret); + goto err_cq_db; + } + ret = alloc_cqc(hr_dev, hr_cq); if (ret) { ibdev_err(ibdev, "failed to alloc CQ context, ret = %d.\n", ret); - goto err_cq_db; + goto err_cqn; } /* @@ -326,6 +385,8 @@ int hns_roce_create_cq(struct ib_cq *ib_cq, const struct ib_cq_init_attr *attr, err_cqc: free_cqc(hr_dev, hr_cq); +err_cqn: + free_cqn(hr_dev, hr_cq->cqn); err_cq_db: free_cq_db(hr_dev, hr_cq, udata); err_cq_buf: @@ -341,9 +402,11 @@ int hns_roce_destroy_cq(struct ib_cq *ib_cq, struct ib_udata *udata) if (hr_dev->hw->destroy_cq) hr_dev->hw->destroy_cq(ib_cq, udata); - free_cq_buf(hr_dev, hr_cq); - free_cq_db(hr_dev, hr_cq, udata); free_cqc(hr_dev, hr_cq); + free_cqn(hr_dev, hr_cq->cqn); + free_cq_db(hr_dev, hr_cq, udata); + free_cq_buf(hr_dev, hr_cq); + return 0; } @@ -402,18 +465,33 @@ void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type) complete(&hr_cq->free); } -int hns_roce_init_cq_table(struct hns_roce_dev *hr_dev) +void hns_roce_init_cq_table(struct hns_roce_dev *hr_dev) { struct hns_roce_cq_table *cq_table = &hr_dev->cq_table; + unsigned int reserved_from_bot; + unsigned int i; + mutex_init(&cq_table->bank_mutex); xa_init(&cq_table->array); - return hns_roce_bitmap_init(&cq_table->bitmap, hr_dev->caps.num_cqs, - hr_dev->caps.num_cqs - 1, - hr_dev->caps.reserved_cqs, 0); + reserved_from_bot = hr_dev->caps.reserved_cqs; + + for (i = 0; i < reserved_from_bot; i++) { + cq_table->bank[get_cq_bankid(i)].inuse++; + cq_table->bank[get_cq_bankid(i)].min++; + } + + for (i = 0; i < HNS_ROCE_CQ_BANK_NUM; i++) { + ida_init(&cq_table->bank[i].ida); + cq_table->bank[i].max = hr_dev->caps.num_cqs / + HNS_ROCE_CQ_BANK_NUM - 1; + } } void hns_roce_cleanup_cq_table(struct hns_roce_dev *hr_dev) { - hns_roce_bitmap_cleanup(&hr_dev->cq_table.bitmap); + int i; + + for (i = 0; i < HNS_ROCE_CQ_BANK_NUM; i++) + ida_destroy(&hr_dev->cq_table.bank[i].ida); } diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h index ad8253245a85..3d6b7a2db496 100644 --- a/drivers/infiniband/hw/hns/hns_roce_device.h +++ b/drivers/infiniband/hw/hns/hns_roce_device.h @@ -54,6 +54,7 @@ /* Hardware specification only for v1 engine */ #define HNS_ROCE_MIN_CQE_NUM 0x40 #define HNS_ROCE_MIN_WQE_NUM 0x20 +#define HNS_ROCE_MIN_SRQ_WQE_NUM 1 /* Hardware specification only for v1 engine */ #define HNS_ROCE_MAX_INNER_MTPT_NUM 0x7 @@ -65,6 +66,8 @@ #define HNS_ROCE_CQE_WCMD_EMPTY_BIT 0x2 #define HNS_ROCE_MIN_CQE_CNT 16 +#define HNS_ROCE_RESERVED_SGE 1 + #define HNS_ROCE_MAX_IRQ_NUM 128 #define HNS_ROCE_SGE_IN_WQE 2 @@ -90,6 +93,7 @@ #define HNS_ROCE_MAX_PORTS 6 #define HNS_ROCE_GID_SIZE 16 #define HNS_ROCE_SGE_SIZE 16 +#define HNS_ROCE_DWQE_SIZE 65536 #define HNS_ROCE_HOP_NUM_0 0xff @@ -119,6 +123,9 @@ #define SRQ_DB_REG 0x230 #define HNS_ROCE_QP_BANK_NUM 8 +#define HNS_ROCE_CQ_BANK_NUM 4 + +#define CQ_BANKID_SHIFT 2 /* The chip implementation of the consumer index is calculated * according to twice the actual EQ depth @@ -163,44 +170,6 @@ enum hns_roce_event { HNS_ROCE_EVENT_TYPE_FLR = 0x15, }; -/* Local Work Queue Catastrophic Error,SUBTYPE 0x5 */ -enum { - HNS_ROCE_LWQCE_QPC_ERROR = 1, - HNS_ROCE_LWQCE_MTU_ERROR = 2, - HNS_ROCE_LWQCE_WQE_BA_ADDR_ERROR = 3, - HNS_ROCE_LWQCE_WQE_ADDR_ERROR = 4, - HNS_ROCE_LWQCE_SQ_WQE_SHIFT_ERROR = 5, - HNS_ROCE_LWQCE_SL_ERROR = 6, - HNS_ROCE_LWQCE_PORT_ERROR = 7, -}; - -/* Local Access Violation Work Queue Error,SUBTYPE 0x7 */ -enum { - HNS_ROCE_LAVWQE_R_KEY_VIOLATION = 1, - HNS_ROCE_LAVWQE_LENGTH_ERROR = 2, - HNS_ROCE_LAVWQE_VA_ERROR = 3, - HNS_ROCE_LAVWQE_PD_ERROR = 4, - HNS_ROCE_LAVWQE_RW_ACC_ERROR = 5, - HNS_ROCE_LAVWQE_KEY_STATE_ERROR = 6, - HNS_ROCE_LAVWQE_MR_OPERATION_ERROR = 7, -}; - -/* DOORBELL overflow subtype */ -enum { - HNS_ROCE_DB_SUBTYPE_SDB_OVF = 1, - HNS_ROCE_DB_SUBTYPE_SDB_ALM_OVF = 2, - HNS_ROCE_DB_SUBTYPE_ODB_OVF = 3, - HNS_ROCE_DB_SUBTYPE_ODB_ALM_OVF = 4, - HNS_ROCE_DB_SUBTYPE_SDB_ALM_EMP = 5, - HNS_ROCE_DB_SUBTYPE_ODB_ALM_EMP = 6, -}; - -enum { - /* RQ&SRQ related operations */ - HNS_ROCE_OPCODE_SEND_DATA_RECEIVE = 0x06, - HNS_ROCE_OPCODE_RDMA_WITH_IMM_RECEIVE = 0x07, -}; - #define HNS_ROCE_CAP_FLAGS_EX_SHIFT 12 enum { @@ -253,9 +222,6 @@ enum { #define HNS_ROCE_CMD_SUCCESS 1 -#define HNS_ROCE_PORT_DOWN 0 -#define HNS_ROCE_PORT_UP 1 - /* The minimum page size is 4K for hardware */ #define HNS_HW_PAGE_SHIFT 12 #define HNS_HW_PAGE_SIZE (1 << HNS_HW_PAGE_SHIFT) @@ -332,7 +298,6 @@ struct hns_roce_buf_attr { } region[HNS_ROCE_MAX_BT_REGION]; unsigned int region_count; /* valid region count */ unsigned int page_shift; /* buffer page shift */ - bool fixed_page; /* decide page shift is fixed-size or maximum size */ unsigned int user_access; /* umem access flag */ bool mtt_only; /* only alloc buffer-required MTT memory */ }; @@ -393,6 +358,7 @@ struct hns_roce_wq { spinlock_t lock; u32 wqe_cnt; /* WQE num */ u32 max_gs; + u32 rsv_sge; int offset; int wqe_shift; /* WQE size */ u32 head; @@ -489,6 +455,8 @@ struct hns_roce_idx_que { struct hns_roce_mtr mtr; int entry_shift; unsigned long *bitmap; + u32 head; + u32 tail; }; struct hns_roce_srq { @@ -496,7 +464,9 @@ struct hns_roce_srq { unsigned long srqn; u32 wqe_cnt; int max_gs; + u32 rsv_sge; int wqe_shift; + u32 cqn; void __iomem *db_reg_l; atomic_t refcount; @@ -507,8 +477,6 @@ struct hns_roce_srq { u64 *wrid; struct hns_roce_idx_que idx_que; spinlock_t lock; - u16 head; - u16 tail; struct mutex mutex; void (*event)(struct hns_roce_srq *srq, enum hns_roce_event event); }; @@ -536,9 +504,10 @@ struct hns_roce_qp_table { }; struct hns_roce_cq_table { - struct hns_roce_bitmap bitmap; struct xarray array; struct hns_roce_hem_table table; + struct hns_roce_bank bank[HNS_ROCE_CQ_BANK_NUM]; + struct mutex bank_mutex; }; struct hns_roce_srq_table { @@ -640,6 +609,10 @@ struct hns_roce_work { u32 queue_num; }; +enum { + HNS_ROCE_QP_CAP_DIRECT_WQE = BIT(5), +}; + struct hns_roce_qp { struct ib_qp ibqp; struct hns_roce_wq rq; @@ -647,7 +620,7 @@ struct hns_roce_qp { struct hns_roce_db sdb; unsigned long en_flags; u32 doorbell_qpn; - u32 sq_signal_bits; + enum ib_sig_type sq_signal_bits; struct hns_roce_wq sq; struct hns_roce_mtr mtr; @@ -779,7 +752,7 @@ struct hns_roce_caps { u32 max_cqes; u32 min_cqes; u32 min_wqes; - int reserved_cqs; + u32 reserved_cqs; int reserved_srqs; int num_aeq_vectors; int num_comp_vectors; @@ -911,8 +884,7 @@ struct hns_roce_hw { int (*write_mtpt)(struct hns_roce_dev *hr_dev, void *mb_buf, struct hns_roce_mr *mr, unsigned long mtpt_idx); int (*rereg_write_mtpt)(struct hns_roce_dev *hr_dev, - struct hns_roce_mr *mr, int flags, u32 pdn, - int mr_access_flags, u64 iova, u64 size, + struct hns_roce_mr *mr, int flags, void *mb_buf); int (*frmr_write_mtpt)(struct hns_roce_dev *hr_dev, void *mb_buf, struct hns_roce_mr *mr); @@ -945,11 +917,7 @@ struct hns_roce_hw { int (*modify_cq)(struct ib_cq *cq, u16 cq_count, u16 cq_period); int (*init_eq)(struct hns_roce_dev *hr_dev); void (*cleanup_eq)(struct hns_roce_dev *hr_dev); - void (*write_srqc)(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq, u32 pdn, u16 xrcd, u32 cqn, - void *mb_buf, u64 *mtts_wqe, u64 *mtts_idx, - dma_addr_t dma_handle_wqe, - dma_addr_t dma_handle_idx); + int (*write_srqc)(struct hns_roce_srq *srq, void *mb_buf); int (*modify_srq)(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr, enum ib_srq_attr_mask srq_attr_mask, struct ib_udata *udata); @@ -982,6 +950,7 @@ struct hns_roce_dev { struct mutex pgdir_mutex; int irq[HNS_ROCE_MAX_IRQ_NUM]; u8 __iomem *reg_base; + void __iomem *mem_base; struct hns_roce_caps caps; struct xarray qp_table_xa; @@ -1067,7 +1036,7 @@ static inline struct hns_roce_srq *to_hr_srq(struct ib_srq *ibsrq) static inline void hns_roce_write64_k(__le32 val[2], void __iomem *dest) { - __raw_writeq(*(u64 *) val, dest); + writeq(*(u64 *)val, dest); } static inline struct hns_roce_qp @@ -1164,7 +1133,7 @@ int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, int hns_roce_init_pd_table(struct hns_roce_dev *hr_dev); int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev); -int hns_roce_init_cq_table(struct hns_roce_dev *hr_dev); +void hns_roce_init_cq_table(struct hns_roce_dev *hr_dev); int hns_roce_init_qp_table(struct hns_roce_dev *hr_dev); int hns_roce_init_srq_table(struct hns_roce_dev *hr_dev); @@ -1281,7 +1250,6 @@ u8 hns_get_gid_index(struct hns_roce_dev *hr_dev, u8 port, int gid_index); void hns_roce_handle_device_err(struct hns_roce_dev *hr_dev); int hns_roce_init(struct hns_roce_dev *hr_dev); void hns_roce_exit(struct hns_roce_dev *hr_dev); - int hns_roce_fill_res_cq_entry(struct sk_buff *msg, struct ib_cq *ib_cq); #endif /* _HNS_ROCE_DEVICE_H */ diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c b/drivers/infiniband/hw/hns/hns_roce_hem.c index edc9d6b98d95..cfd2e1b60c7f 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.c +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c @@ -1075,9 +1075,8 @@ static struct roce_hem_item *hem_list_alloc_item(struct hns_roce_dev *hr_dev, return NULL; if (exist_bt) { - hem->addr = dma_alloc_coherent(hr_dev->dev, - count * BA_BYTE_LEN, - &hem->dma_addr, GFP_KERNEL); + hem->addr = dma_alloc_coherent(hr_dev->dev, count * BA_BYTE_LEN, + &hem->dma_addr, GFP_KERNEL); if (!hem->addr) { kfree(hem); return NULL; @@ -1336,6 +1335,10 @@ static int hem_list_alloc_root_bt(struct hns_roce_dev *hr_dev, if (ba_num < 1) return -ENOMEM; + if (ba_num > unit) + return -ENOBUFS; + + ba_num = min_t(int, ba_num, unit); INIT_LIST_HEAD(&temp_root); offset = r->offset; /* indicate to last region */ diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c index f68585ff8e8a..5346fdca9473 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.c @@ -43,6 +43,22 @@ #include "hns_roce_hem.h" #include "hns_roce_hw_v1.h" +/** + * hns_get_gid_index - Get gid index. + * @hr_dev: pointer to structure hns_roce_dev. + * @port: port, value range: 0 ~ MAX + * @gid_index: gid_index, value range: 0 ~ MAX + * Description: + * N ports shared gids, allocation method as follow: + * GID[0][0], GID[1][0],.....GID[N - 1][0], + * GID[0][0], GID[1][0],.....GID[N - 1][0], + * And so on + */ +u8 hns_get_gid_index(struct hns_roce_dev *hr_dev, u8 port, int gid_index) +{ + return gid_index * hr_dev->caps.num_ports + port; +} + static void set_data_seg(struct hns_roce_wqe_data_seg *dseg, struct ib_sge *sg) { dseg->lkey = cpu_to_le32(sg->lkey); @@ -314,8 +330,6 @@ out: /* Set DB return */ if (likely(nreq)) { qp->sq.head += nreq; - /* Memory barrier */ - wmb(); roce_set_field(sq_db.u32_4, SQ_DOORBELL_U32_4_SQ_HEAD_M, SQ_DOORBELL_U32_4_SQ_HEAD_S, @@ -395,8 +409,6 @@ static int hns_roce_v1_post_recv(struct ib_qp *ibqp, out: if (likely(nreq)) { hr_qp->rq.head += nreq; - /* Memory barrier */ - wmb(); if (ibqp->qp_type == IB_QPT_GSI) { __le32 tmp; @@ -1391,7 +1403,7 @@ static void hns_roce_free_mr_free(struct hns_roce_dev *hr_dev) /** * hns_roce_v1_reset - reset RoCE * @hr_dev: RoCE device struct pointer - * @enable: true -- drop reset, false -- reset + * @dereset: true -- drop reset, false -- reset * return 0 - success , negative --fail */ static int hns_roce_v1_reset(struct hns_roce_dev *hr_dev, bool dereset) @@ -1968,12 +1980,6 @@ static void __hns_roce_v1_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, if (nfreed) { hr_cq->cons_index += nfreed; - /* - * Make sure update of buffer contents is done before - * updating consumer index. - */ - wmb(); - hns_roce_v1_cq_set_ci(hr_cq, hr_cq->cons_index); } } @@ -2314,8 +2320,6 @@ int hns_roce_v1_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc) *hr_cq->tptr_addr = hr_cq->cons_index & ((hr_cq->cq_depth << 1) - 1); - /* Memroy barrier */ - wmb(); hns_roce_v1_cq_set_ci(hr_cq, hr_cq->cons_index); } @@ -3204,9 +3208,6 @@ static int hns_roce_v1_m_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, * need to hw to flash RQ HEAD by DB again */ if (cur_state == IB_QPS_INIT && new_state == IB_QPS_INIT) { - /* Memory barrier */ - wmb(); - roce_set_field(doorbell[0], RQ_DOORBELL_U32_4_RQ_HEAD_M, RQ_DOORBELL_U32_4_RQ_HEAD_S, hr_qp->rq.head); roce_set_field(doorbell[1], RQ_DOORBELL_U32_8_QPN_M, diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v1.h b/drivers/infiniband/hw/hns/hns_roce_hw_v1.h index 46ab0a321d21..84383236e47d 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v1.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v1.h @@ -193,6 +193,49 @@ #define HNS_ROCE_AEQE_EVENT_CE_EVENT_CEQE_CEQN_S 0 #define HNS_ROCE_AEQE_EVENT_CE_EVENT_CEQE_CEQN_M GENMASK(4, 0) +/* Local Work Queue Catastrophic Error,SUBTYPE 0x5 */ +enum { + HNS_ROCE_LWQCE_QPC_ERROR = 1, + HNS_ROCE_LWQCE_MTU_ERROR, + HNS_ROCE_LWQCE_WQE_BA_ADDR_ERROR, + HNS_ROCE_LWQCE_WQE_ADDR_ERROR, + HNS_ROCE_LWQCE_SQ_WQE_SHIFT_ERROR, + HNS_ROCE_LWQCE_SL_ERROR, + HNS_ROCE_LWQCE_PORT_ERROR, +}; + +/* Local Access Violation Work Queue Error,SUBTYPE 0x7 */ +enum { + HNS_ROCE_LAVWQE_R_KEY_VIOLATION = 1, + HNS_ROCE_LAVWQE_LENGTH_ERROR, + HNS_ROCE_LAVWQE_VA_ERROR, + HNS_ROCE_LAVWQE_PD_ERROR, + HNS_ROCE_LAVWQE_RW_ACC_ERROR, + HNS_ROCE_LAVWQE_KEY_STATE_ERROR, + HNS_ROCE_LAVWQE_MR_OPERATION_ERROR, +}; + +/* DOORBELL overflow subtype */ +enum { + HNS_ROCE_DB_SUBTYPE_SDB_OVF = 1, + HNS_ROCE_DB_SUBTYPE_SDB_ALM_OVF, + HNS_ROCE_DB_SUBTYPE_ODB_OVF, + HNS_ROCE_DB_SUBTYPE_ODB_ALM_OVF, + HNS_ROCE_DB_SUBTYPE_SDB_ALM_EMP, + HNS_ROCE_DB_SUBTYPE_ODB_ALM_EMP, +}; + +enum { + /* RQ&SRQ related operations */ + HNS_ROCE_OPCODE_SEND_DATA_RECEIVE = 0x06, + HNS_ROCE_OPCODE_RDMA_WITH_IMM_RECEIVE, +}; + +enum { + HNS_ROCE_PORT_DOWN = 0, + HNS_ROCE_PORT_UP, +}; + struct hns_roce_cq_context { __le32 cqc_byte_4; __le32 cq_bt_l; diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index 833e1f259936..c3934abeb260 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -48,8 +48,8 @@ #include "hns_roce_hem.h" #include "hns_roce_hw_v2.h" -static void set_data_seg_v2(struct hns_roce_v2_wqe_data_seg *dseg, - struct ib_sge *sg) +static inline void set_data_seg_v2(struct hns_roce_v2_wqe_data_seg *dseg, + struct ib_sge *sg) { dseg->lkey = cpu_to_le32(sg->lkey); dseg->addr = cpu_to_le64(sg->addr); @@ -99,16 +99,16 @@ static void set_frmr_seg(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, u64 pbl_ba; /* use ib_access_flags */ - roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_BIND_EN_S, - wr->access & IB_ACCESS_MW_BIND ? 1 : 0); - roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_ATOMIC_S, - wr->access & IB_ACCESS_REMOTE_ATOMIC ? 1 : 0); - roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_RR_S, - wr->access & IB_ACCESS_REMOTE_READ ? 1 : 0); - roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_RW_S, - wr->access & IB_ACCESS_REMOTE_WRITE ? 1 : 0); - roce_set_bit(rc_sq_wqe->byte_4, V2_RC_FRMR_WQE_BYTE_4_LW_S, - wr->access & IB_ACCESS_LOCAL_WRITE ? 1 : 0); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_BIND_EN_S, + !!(wr->access & IB_ACCESS_MW_BIND)); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_ATOMIC_S, + !!(wr->access & IB_ACCESS_REMOTE_ATOMIC)); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_RR_S, + !!(wr->access & IB_ACCESS_REMOTE_READ)); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_RW_S, + !!(wr->access & IB_ACCESS_REMOTE_WRITE)); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_LW_S, + !!(wr->access & IB_ACCESS_LOCAL_WRITE)); /* Data structure reuse may lead to confusion */ pbl_ba = mr->pbl_mtr.hem_cfg.root_ba; @@ -121,12 +121,10 @@ static void set_frmr_seg(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, rc_sq_wqe->va = cpu_to_le64(wr->mr->iova); fseg->pbl_size = cpu_to_le32(mr->npages); - roce_set_field(fseg->mode_buf_pg_sz, - V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_M, + roce_set_field(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_M, V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_S, to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.buf_pg_shift)); - roce_set_bit(fseg->mode_buf_pg_sz, - V2_RC_FRMR_WQE_BYTE_40_BLK_MODE_S, 0); + roce_set_bit(fseg->byte_40, V2_RC_FRMR_WQE_BYTE_40_BLK_MODE_S, 0); } static void set_atomic_seg(const struct ib_send_wr *wr, @@ -361,7 +359,7 @@ static int check_send_valid(struct hns_roce_dev *hr_dev, } else if (unlikely(hr_qp->state == IB_QPS_RESET || hr_qp->state == IB_QPS_INIT || hr_qp->state == IB_QPS_RTR)) { - ibdev_err(ibdev, "failed to post WQE, QP state %hhu!\n", + ibdev_err(ibdev, "failed to post WQE, QP state %u!\n", hr_qp->state); return -EINVAL; } else if (unlikely(hr_dev->state >= HNS_ROCE_DEVICE_STATE_RST_DOWN)) { @@ -469,7 +467,6 @@ static inline int set_ud_wqe(struct hns_roce_qp *qp, int ret; valid_num_sge = calc_wr_sge_num(wr, &msg_len); - memset(ud_sq_wqe, 0, sizeof(*ud_sq_wqe)); ret = set_ud_opcode(ud_sq_wqe, wr); if (WARN_ON(ret)) @@ -503,6 +500,8 @@ static inline int set_ud_wqe(struct hns_roce_qp *qp, if (ret) return ret; + qp->sl = to_hr_ah(ud_wr(wr)->ah)->av.sl; + set_extend_sge(qp, wr->sg_list, &curr_idx, valid_num_sge); /* @@ -521,10 +520,12 @@ static inline int set_ud_wqe(struct hns_roce_qp *qp, return 0; } -static int set_rc_opcode(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, +static int set_rc_opcode(struct hns_roce_dev *hr_dev, + struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, const struct ib_send_wr *wr) { u32 ib_op = wr->opcode; + int ret = 0; rc_sq_wqe->immtdata = get_immtdata(wr); @@ -544,7 +545,10 @@ static int set_rc_opcode(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, rc_sq_wqe->va = cpu_to_le64(atomic_wr(wr)->remote_addr); break; case IB_WR_REG_MR: - set_frmr_seg(rc_sq_wqe, reg_wr(wr)); + if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) + set_frmr_seg(rc_sq_wqe, reg_wr(wr)); + else + ret = -EOPNOTSUPP; break; case IB_WR_LOCAL_INV: roce_set_bit(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_SO_S, 1); @@ -553,19 +557,23 @@ static int set_rc_opcode(struct hns_roce_v2_rc_send_wqe *rc_sq_wqe, rc_sq_wqe->inv_key = cpu_to_le32(wr->ex.invalidate_rkey); break; default: - return -EINVAL; + ret = -EINVAL; } + if (unlikely(ret)) + return ret; + roce_set_field(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_OPCODE_M, V2_RC_SEND_WQE_BYTE_4_OPCODE_S, to_hr_opcode(ib_op)); - return 0; + return ret; } static inline int set_rc_wqe(struct hns_roce_qp *qp, const struct ib_send_wr *wr, void *wqe, unsigned int *sge_idx, unsigned int owner_bit) { + struct hns_roce_dev *hr_dev = to_hr_dev(qp->ibqp.device); struct hns_roce_v2_rc_send_wqe *rc_sq_wqe = wqe; unsigned int curr_idx = *sge_idx; unsigned int valid_num_sge; @@ -573,11 +581,10 @@ static inline int set_rc_wqe(struct hns_roce_qp *qp, int ret; valid_num_sge = calc_wr_sge_num(wr, &msg_len); - memset(rc_sq_wqe, 0, sizeof(*rc_sq_wqe)); rc_sq_wqe->msg_len = cpu_to_le32(msg_len); - ret = set_rc_opcode(rc_sq_wqe, wr); + ret = set_rc_opcode(hr_dev, rc_sq_wqe, wr); if (WARN_ON(ret)) return ret; @@ -635,6 +642,8 @@ static inline void update_sq_db(struct hns_roce_dev *hr_dev, V2_DB_BYTE_4_TAG_S, qp->doorbell_qpn); roce_set_field(sq_db.byte_4, V2_DB_BYTE_4_CMD_M, V2_DB_BYTE_4_CMD_S, HNS_ROCE_V2_SQ_DB); + /* indicates data on new BAR, 0 : SQ doorbell, 1 : DWQE */ + roce_set_bit(sq_db.byte_4, V2_DB_FLAG_S, 0); roce_set_field(sq_db.parameter, V2_DB_PARAMETER_IDX_M, V2_DB_PARAMETER_IDX_S, qp->sq.head); roce_set_field(sq_db.parameter, V2_DB_PARAMETER_SL_M, @@ -644,6 +653,38 @@ static inline void update_sq_db(struct hns_roce_dev *hr_dev, } } +static void hns_roce_write512(struct hns_roce_dev *hr_dev, u64 *val, + u64 __iomem *dest) +{ +#define HNS_ROCE_WRITE_TIMES 8 + struct hns_roce_v2_priv *priv = (struct hns_roce_v2_priv *)hr_dev->priv; + struct hnae3_handle *handle = priv->handle; + const struct hnae3_ae_ops *ops = handle->ae_algo->ops; + int i; + + if (!hr_dev->dis_db && !ops->get_hw_reset_stat(handle)) + for (i = 0; i < HNS_ROCE_WRITE_TIMES; i++) + writeq_relaxed(*(val + i), dest + i); +} + +static void write_dwqe(struct hns_roce_dev *hr_dev, struct hns_roce_qp *qp, + void *wqe) +{ + struct hns_roce_v2_rc_send_wqe *rc_sq_wqe = wqe; + + /* All kinds of DirectWQE have the same header field layout */ + roce_set_bit(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_FLAG_S, 1); + roce_set_field(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_DB_SL_L_M, + V2_RC_SEND_WQE_BYTE_4_DB_SL_L_S, qp->sl); + roce_set_field(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_DB_SL_H_M, + V2_RC_SEND_WQE_BYTE_4_DB_SL_H_S, qp->sl >> 2); + roce_set_field(rc_sq_wqe->byte_4, V2_RC_SEND_WQE_BYTE_4_WQE_INDEX_M, + V2_RC_SEND_WQE_BYTE_4_WQE_INDEX_S, qp->sq.head); + + hns_roce_write512(hr_dev, wqe, hr_dev->mem_base + + HNS_ROCE_DWQE_SIZE * qp->ibqp.qp_num); +} + static int hns_roce_v2_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, const struct ib_send_wr **bad_wr) @@ -708,9 +749,12 @@ out: if (likely(nreq)) { qp->sq.head += nreq; qp->next_sge = sge_idx; - /* Memory barrier */ - wmb(); - update_sq_db(hr_dev, qp); + + if (nreq == 1 && qp->sq.head == qp->sq.tail + 1 && + (qp->en_flags & HNS_ROCE_QP_CAP_DIRECT_WQE)) + write_dwqe(hr_dev, qp, wqe); + else + update_sq_db(hr_dev, qp); } spin_unlock_irqrestore(&qp->sq.lock, flags); @@ -721,14 +765,74 @@ out: static int check_recv_valid(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp) { + struct ib_device *ibdev = &hr_dev->ib_dev; + struct ib_qp *ibqp = &hr_qp->ibqp; + + if (unlikely(ibqp->qp_type != IB_QPT_RC && + ibqp->qp_type != IB_QPT_GSI && + ibqp->qp_type != IB_QPT_UD)) { + ibdev_err(ibdev, "unsupported qp type, qp_type = %d.\n", + ibqp->qp_type); + return -EOPNOTSUPP; + } + if (unlikely(hr_dev->state >= HNS_ROCE_DEVICE_STATE_RST_DOWN)) return -EIO; - else if (hr_qp->state == IB_QPS_RESET) + + if (hr_qp->state == IB_QPS_RESET) return -EINVAL; return 0; } +static void fill_recv_sge_to_wqe(const struct ib_recv_wr *wr, void *wqe, + u32 max_sge, bool rsv) +{ + struct hns_roce_v2_wqe_data_seg *dseg = wqe; + u32 i, cnt; + + for (i = 0, cnt = 0; i < wr->num_sge; i++) { + /* Skip zero-length sge */ + if (!wr->sg_list[i].length) + continue; + set_data_seg_v2(dseg + cnt, wr->sg_list + i); + cnt++; + } + + /* Fill a reserved sge to make hw stop reading remaining segments */ + if (rsv) { + dseg[cnt].lkey = cpu_to_le32(HNS_ROCE_INVALID_LKEY); + dseg[cnt].addr = 0; + dseg[cnt].len = cpu_to_le32(HNS_ROCE_INVALID_SGE_LENGTH); + } else { + /* Clear remaining segments to make ROCEE ignore sges */ + if (cnt < max_sge) + memset(dseg + cnt, 0, + (max_sge - cnt) * HNS_ROCE_SGE_SIZE); + } +} + +static void fill_rq_wqe(struct hns_roce_qp *hr_qp, const struct ib_recv_wr *wr, + u32 wqe_idx, u32 max_sge) +{ + struct hns_roce_rinl_sge *sge_list; + void *wqe = NULL; + u32 i; + + wqe = hns_roce_get_recv_wqe(hr_qp, wqe_idx); + fill_recv_sge_to_wqe(wr, wqe, max_sge, hr_qp->rq.rsv_sge); + + /* rq support inline data */ + if (hr_qp->rq_inl_buf.wqe_cnt) { + sge_list = hr_qp->rq_inl_buf.wqe_list[wqe_idx].sg_list; + hr_qp->rq_inl_buf.wqe_list[wqe_idx].sge_cnt = (u32)wr->num_sge; + for (i = 0; i < wr->num_sge; i++) { + sge_list[i].addr = (void *)(u64)wr->sg_list[i].addr; + sge_list[i].len = wr->sg_list[i].length; + } + } +} + static int hns_roce_v2_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr) @@ -736,14 +840,9 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, struct hns_roce_dev *hr_dev = to_hr_dev(ibqp->device); struct hns_roce_qp *hr_qp = to_hr_qp(ibqp); struct ib_device *ibdev = &hr_dev->ib_dev; - struct hns_roce_v2_wqe_data_seg *dseg; - struct hns_roce_rinl_sge *sge_list; + u32 wqe_idx, nreq, max_sge; unsigned long flags; - void *wqe = NULL; - u32 wqe_idx; - int nreq; int ret; - int i; spin_lock_irqsave(&hr_qp->rq.lock, flags); @@ -754,6 +853,7 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, goto out; } + max_sge = hr_qp->rq.max_gs - hr_qp->rq.rsv_sge; for (nreq = 0; wr; ++nreq, wr = wr->next) { if (unlikely(hns_roce_wq_overflow(&hr_qp->rq, nreq, hr_qp->ibqp.recv_cq))) { @@ -762,50 +862,22 @@ static int hns_roce_v2_post_recv(struct ib_qp *ibqp, goto out; } - wqe_idx = (hr_qp->rq.head + nreq) & (hr_qp->rq.wqe_cnt - 1); - - if (unlikely(wr->num_sge > hr_qp->rq.max_gs)) { + if (unlikely(wr->num_sge > max_sge)) { ibdev_err(ibdev, "num_sge = %d >= max_sge = %u.\n", - wr->num_sge, hr_qp->rq.max_gs); + wr->num_sge, max_sge); ret = -EINVAL; *bad_wr = wr; goto out; } - wqe = hns_roce_get_recv_wqe(hr_qp, wqe_idx); - dseg = (struct hns_roce_v2_wqe_data_seg *)wqe; - for (i = 0; i < wr->num_sge; i++) { - if (!wr->sg_list[i].length) - continue; - set_data_seg_v2(dseg, wr->sg_list + i); - dseg++; - } - - if (wr->num_sge < hr_qp->rq.max_gs) { - dseg->lkey = cpu_to_le32(HNS_ROCE_INVALID_LKEY); - dseg->addr = 0; - } - - /* rq support inline data */ - if (hr_qp->rq_inl_buf.wqe_cnt) { - sge_list = hr_qp->rq_inl_buf.wqe_list[wqe_idx].sg_list; - hr_qp->rq_inl_buf.wqe_list[wqe_idx].sge_cnt = - (u32)wr->num_sge; - for (i = 0; i < wr->num_sge; i++) { - sge_list[i].addr = - (void *)(u64)wr->sg_list[i].addr; - sge_list[i].len = wr->sg_list[i].length; - } - } - + wqe_idx = (hr_qp->rq.head + nreq) & (hr_qp->rq.wqe_cnt - 1); + fill_rq_wqe(hr_qp, wr, wqe_idx, max_sge); hr_qp->rq.wrid[wqe_idx] = wr->wr_id; } out: if (likely(nreq)) { hr_qp->rq.head += nreq; - /* Memory barrier */ - wmb(); /* * Hip08 hardware cannot flush the WQEs in RQ if the QP state @@ -829,41 +901,82 @@ out: return ret; } -static void *get_srq_wqe(struct hns_roce_srq *srq, int n) +static void *get_srq_wqe_buf(struct hns_roce_srq *srq, u32 n) { return hns_roce_buf_offset(srq->buf_mtr.kmem, n << srq->wqe_shift); } -static void *get_idx_buf(struct hns_roce_idx_que *idx_que, unsigned int n) +static void *get_idx_buf(struct hns_roce_idx_que *idx_que, u32 n) { return hns_roce_buf_offset(idx_que->mtr.kmem, n << idx_que->entry_shift); } -static void hns_roce_free_srq_wqe(struct hns_roce_srq *srq, int wqe_index) +static void hns_roce_free_srq_wqe(struct hns_roce_srq *srq, u32 wqe_index) { /* always called with interrupts disabled. */ spin_lock(&srq->lock); bitmap_clear(srq->idx_que.bitmap, wqe_index, 1); - srq->tail++; + srq->idx_que.tail++; spin_unlock(&srq->lock); } -static int find_empty_entry(struct hns_roce_idx_que *idx_que, - unsigned long size) +static int hns_roce_srqwq_overflow(struct hns_roce_srq *srq) { - int wqe_idx; + struct hns_roce_idx_que *idx_que = &srq->idx_que; - if (unlikely(bitmap_full(idx_que->bitmap, size))) + return idx_que->head - idx_que->tail >= srq->wqe_cnt; +} + +static int check_post_srq_valid(struct hns_roce_srq *srq, u32 max_sge, + const struct ib_recv_wr *wr) +{ + struct ib_device *ib_dev = srq->ibsrq.device; + + if (unlikely(wr->num_sge > max_sge)) { + ibdev_err(ib_dev, + "failed to check sge, wr->num_sge = %d, max_sge = %u.\n", + wr->num_sge, max_sge); + return -EINVAL; + } + + if (unlikely(hns_roce_srqwq_overflow(srq))) { + ibdev_err(ib_dev, + "failed to check srqwq status, srqwq is full.\n"); + return -ENOMEM; + } + + return 0; +} + +static int get_srq_wqe_idx(struct hns_roce_srq *srq, u32 *wqe_idx) +{ + struct hns_roce_idx_que *idx_que = &srq->idx_que; + u32 pos; + + pos = find_first_zero_bit(idx_que->bitmap, srq->wqe_cnt); + if (unlikely(pos == srq->wqe_cnt)) return -ENOSPC; - wqe_idx = find_first_zero_bit(idx_que->bitmap, size); + bitmap_set(idx_que->bitmap, pos, 1); + *wqe_idx = pos; + return 0; +} - bitmap_set(idx_que->bitmap, wqe_idx, 1); +static void fill_wqe_idx(struct hns_roce_srq *srq, unsigned int wqe_idx) +{ + struct hns_roce_idx_que *idx_que = &srq->idx_que; + unsigned int head; + __le32 *buf; - return wqe_idx; + head = idx_que->head & (srq->wqe_cnt - 1); + + buf = get_idx_buf(idx_que, head); + *buf = cpu_to_le32(wqe_idx); + + idx_que->head++; } static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq, @@ -872,77 +985,42 @@ static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq, { struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); struct hns_roce_srq *srq = to_hr_srq(ibsrq); - struct hns_roce_v2_wqe_data_seg *dseg; struct hns_roce_v2_db srq_db; unsigned long flags; - unsigned int ind; - __le32 *srq_idx; int ret = 0; - int wqe_idx; + u32 max_sge; + u32 wqe_idx; void *wqe; - int nreq; - int i; + u32 nreq; spin_lock_irqsave(&srq->lock, flags); - ind = srq->head & (srq->wqe_cnt - 1); - + max_sge = srq->max_gs - srq->rsv_sge; for (nreq = 0; wr; ++nreq, wr = wr->next) { - if (unlikely(wr->num_sge >= srq->max_gs)) { - ret = -EINVAL; - *bad_wr = wr; - break; - } - - if (unlikely(srq->head == srq->tail)) { - ret = -ENOMEM; + ret = check_post_srq_valid(srq, max_sge, wr); + if (ret) { *bad_wr = wr; break; } - wqe_idx = find_empty_entry(&srq->idx_que, srq->wqe_cnt); - if (unlikely(wqe_idx < 0)) { - ret = -ENOMEM; + ret = get_srq_wqe_idx(srq, &wqe_idx); + if (unlikely(ret)) { *bad_wr = wr; break; } - wqe = get_srq_wqe(srq, wqe_idx); - dseg = (struct hns_roce_v2_wqe_data_seg *)wqe; - - for (i = 0; i < wr->num_sge; ++i) { - dseg[i].len = cpu_to_le32(wr->sg_list[i].length); - dseg[i].lkey = cpu_to_le32(wr->sg_list[i].lkey); - dseg[i].addr = cpu_to_le64(wr->sg_list[i].addr); - } - - if (wr->num_sge < srq->max_gs) { - dseg[i].len = 0; - dseg[i].lkey = cpu_to_le32(0x100); - dseg[i].addr = 0; - } - - srq_idx = get_idx_buf(&srq->idx_que, ind); - *srq_idx = cpu_to_le32(wqe_idx); - + wqe = get_srq_wqe_buf(srq, wqe_idx); + fill_recv_sge_to_wqe(wr, wqe, max_sge, srq->rsv_sge); + fill_wqe_idx(srq, wqe_idx); srq->wrid[wqe_idx] = wr->wr_id; - ind = (ind + 1) & (srq->wqe_cnt - 1); } if (likely(nreq)) { - srq->head += nreq; - - /* - * Make sure that descriptors are written before - * doorbell record. - */ - wmb(); - srq_db.byte_4 = cpu_to_le32(HNS_ROCE_V2_SRQ_DB << V2_DB_BYTE_4_CMD_S | (srq->srqn & V2_DB_BYTE_4_TAG_M)); srq_db.parameter = - cpu_to_le32(srq->head & V2_DB_PARAMETER_IDX_M); + cpu_to_le32(srq->idx_que.head & V2_DB_PARAMETER_IDX_M); hns_roce_write64(hr_dev, (__le32 *)&srq_db, srq->db_reg_l); } @@ -1059,15 +1137,6 @@ static int hns_roce_v2_rst_process_cmd(struct hns_roce_dev *hr_dev) return 0; } -static int hns_roce_cmq_space(struct hns_roce_v2_cmq_ring *ring) -{ - int ntu = ring->next_to_use; - int ntc = ring->next_to_clean; - int used = (ntu - ntc + ring->desc_num) % ring->desc_num; - - return ring->desc_num - used - 1; -} - static int hns_roce_alloc_cmq_desc(struct hns_roce_dev *hr_dev, struct hns_roce_v2_cmq_ring *ring) { @@ -1107,8 +1176,7 @@ static int hns_roce_init_cmq_ring(struct hns_roce_dev *hr_dev, bool ring_type) &priv->cmq.csq : &priv->cmq.crq; ring->flag = ring_type; - ring->next_to_clean = 0; - ring->next_to_use = 0; + ring->head = 0; return hns_roce_alloc_cmq_desc(hr_dev, ring); } @@ -1207,34 +1275,10 @@ static void hns_roce_cmq_setup_basic_desc(struct hns_roce_cmq_desc *desc, static int hns_roce_cmq_csq_done(struct hns_roce_dev *hr_dev) { - u32 head = roce_read(hr_dev, ROCEE_TX_CMQ_HEAD_REG); + u32 tail = roce_read(hr_dev, ROCEE_TX_CMQ_TAIL_REG); struct hns_roce_v2_priv *priv = hr_dev->priv; - return head == priv->cmq.csq.next_to_use; -} - -static int hns_roce_cmq_csq_clean(struct hns_roce_dev *hr_dev) -{ - struct hns_roce_v2_priv *priv = hr_dev->priv; - struct hns_roce_v2_cmq_ring *csq = &priv->cmq.csq; - struct hns_roce_cmq_desc *desc; - u16 ntc = csq->next_to_clean; - u32 head; - int clean = 0; - - desc = &csq->desc[ntc]; - head = roce_read(hr_dev, ROCEE_TX_CMQ_HEAD_REG); - while (head != ntc) { - memset(desc, 0, sizeof(*desc)); - ntc++; - if (ntc == csq->desc_num) - ntc = 0; - desc = &csq->desc[ntc]; - clean++; - } - csq->next_to_clean = ntc; - - return clean; + return tail == priv->cmq.csq.head; } static int __hns_roce_cmq_send(struct hns_roce_dev *hr_dev, @@ -1242,42 +1286,26 @@ static int __hns_roce_cmq_send(struct hns_roce_dev *hr_dev, { struct hns_roce_v2_priv *priv = hr_dev->priv; struct hns_roce_v2_cmq_ring *csq = &priv->cmq.csq; - struct hns_roce_cmq_desc *desc_to_use; - bool complete = false; u32 timeout = 0; - int handle = 0; u16 desc_ret; - int ret = 0; - int ntc; + u32 tail; + int ret; + int i; spin_lock_bh(&csq->lock); - if (num > hns_roce_cmq_space(csq)) { - spin_unlock_bh(&csq->lock); - return -EBUSY; - } - - /* - * Record the location of desc in the cmq for this time - * which will be use for hardware to write back - */ - ntc = csq->next_to_use; + tail = csq->head; - while (handle < num) { - desc_to_use = &csq->desc[csq->next_to_use]; - *desc_to_use = desc[handle]; - dev_dbg(hr_dev->dev, "set cmq desc:\n"); - csq->next_to_use++; - if (csq->next_to_use == csq->desc_num) - csq->next_to_use = 0; - handle++; + for (i = 0; i < num; i++) { + csq->desc[csq->head++] = desc[i]; + if (csq->head == csq->desc_num) + csq->head = 0; } /* Write to hardware */ - roce_write(hr_dev, ROCEE_TX_CMQ_TAIL_REG, csq->next_to_use); + roce_write(hr_dev, ROCEE_TX_CMQ_HEAD_REG, csq->head); - /* - * If the command is sync, wait for the firmware to write back, + /* If the command is sync, wait for the firmware to write back, * if multi descriptors to be sent, use the first one to check */ if (le16_to_cpu(desc->flag) & HNS_ROCE_CMD_FLAG_NO_INTR) { @@ -1285,39 +1313,34 @@ static int __hns_roce_cmq_send(struct hns_roce_dev *hr_dev, if (hns_roce_cmq_csq_done(hr_dev)) break; udelay(1); - timeout++; - } while (timeout < priv->cmq.tx_timeout); + } while (++timeout < priv->cmq.tx_timeout); } if (hns_roce_cmq_csq_done(hr_dev)) { - complete = true; - handle = 0; - while (handle < num) { - /* get the result of hardware write back */ - desc_to_use = &csq->desc[ntc]; - desc[handle] = *desc_to_use; - dev_dbg(hr_dev->dev, "Get cmq desc:\n"); - desc_ret = le16_to_cpu(desc[handle].retval); - if (desc_ret == CMD_EXEC_SUCCESS) - ret = 0; - else - ret = -EIO; - priv->cmq.last_status = desc_ret; - ntc++; - handle++; - if (ntc == csq->desc_num) - ntc = 0; + for (ret = 0, i = 0; i < num; i++) { + /* check the result of hardware write back */ + desc[i] = csq->desc[tail++]; + if (tail == csq->desc_num) + tail = 0; + + desc_ret = le16_to_cpu(desc[i].retval); + if (likely(desc_ret == CMD_EXEC_SUCCESS)) + continue; + + dev_err_ratelimited(hr_dev->dev, + "Cmdq IO error, opcode = %x, return = %x\n", + desc->opcode, desc_ret); + ret = -EIO; } - } + } else { + /* FW/HW reset or incorrect number of desc */ + tail = roce_read(hr_dev, ROCEE_TX_CMQ_TAIL_REG); + dev_warn(hr_dev->dev, "CMDQ move tail from %d to %d\n", + csq->head, tail); + csq->head = tail; - if (!complete) ret = -EAGAIN; - - /* clean the command send queue */ - handle = hns_roce_cmq_csq_clean(hr_dev); - if (handle != num) - dev_warn(hr_dev->dev, "Cleaned %d, need to clean %d\n", - handle, num); + } spin_unlock_bh(&csq->lock); @@ -1530,7 +1553,8 @@ static int hns_roce_config_global_param(struct hns_roce_dev *hr_dev) CFG_GLOBAL_PARAM_DATA_0_ROCEE_TIME_1US_CFG_S, 0x3e8); roce_set_field(req->time_cfg_udp_port, CFG_GLOBAL_PARAM_DATA_0_ROCEE_UDP_PORT_M, - CFG_GLOBAL_PARAM_DATA_0_ROCEE_UDP_PORT_S, 0x12b7); + CFG_GLOBAL_PARAM_DATA_0_ROCEE_UDP_PORT_S, + ROCE_V2_UDP_DPORT); return hns_roce_cmq_send(hr_dev, &desc, 1); } @@ -1541,17 +1565,13 @@ static int hns_roce_query_pf_resource(struct hns_roce_dev *hr_dev) struct hns_roce_pf_res_a *req_a; struct hns_roce_pf_res_b *req_b; int ret; - int i; - for (i = 0; i < 2; i++) { - hns_roce_cmq_setup_basic_desc(&desc[i], - HNS_ROCE_OPC_QUERY_PF_RES, true); + hns_roce_cmq_setup_basic_desc(&desc[0], HNS_ROCE_OPC_QUERY_PF_RES, + true); + desc[0].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - if (i == 0) - desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - else - desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - } + hns_roce_cmq_setup_basic_desc(&desc[1], HNS_ROCE_OPC_QUERY_PF_RES, + true); ret = hns_roce_cmq_send(hr_dev, desc, 2); if (ret) @@ -1644,19 +1664,16 @@ static int hns_roce_alloc_vf_resource(struct hns_roce_dev *hr_dev) struct hns_roce_cmq_desc desc[2]; struct hns_roce_vf_res_a *req_a; struct hns_roce_vf_res_b *req_b; - int i; req_a = (struct hns_roce_vf_res_a *)desc[0].data; req_b = (struct hns_roce_vf_res_b *)desc[1].data; - for (i = 0; i < 2; i++) { - hns_roce_cmq_setup_basic_desc(&desc[i], - HNS_ROCE_OPC_ALLOC_VF_RES, false); - if (i == 0) - desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - else - desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - } + hns_roce_cmq_setup_basic_desc(&desc[0], HNS_ROCE_OPC_ALLOC_VF_RES, + false); + desc[0].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); + + hns_roce_cmq_setup_basic_desc(&desc[1], HNS_ROCE_OPC_ALLOC_VF_RES, + false); roce_set_field(req_a->vf_qpc_bt_idx_num, VF_RES_A_DATA_1_VF_QPC_BT_IDX_M, @@ -1866,7 +1883,6 @@ static void set_default_caps(struct hns_roce_dev *hr_dev) caps->flags = HNS_ROCE_CAP_FLAG_REREG_MR | HNS_ROCE_CAP_FLAG_ROCE_V1_V2 | - HNS_ROCE_CAP_FLAG_RQ_INLINE | HNS_ROCE_CAP_FLAG_RECORD_DB | HNS_ROCE_CAP_FLAG_SQ_RECORD_DB; @@ -1999,10 +2015,12 @@ static int hns_roce_query_pf_caps(struct hns_roce_dev *hr_dev) caps->max_sq_sg = le16_to_cpu(resp_a->max_sq_sg); caps->max_sq_inline = le16_to_cpu(resp_a->max_sq_inline); caps->max_rq_sg = le16_to_cpu(resp_a->max_rq_sg); + caps->max_rq_sg = roundup_pow_of_two(caps->max_rq_sg); caps->max_extend_sg = le32_to_cpu(resp_a->max_extend_sg); caps->num_qpc_timer = le16_to_cpu(resp_a->num_qpc_timer); caps->num_cqc_timer = le16_to_cpu(resp_a->num_cqc_timer); caps->max_srq_sges = le16_to_cpu(resp_a->max_srq_sges); + caps->max_srq_sges = roundup_pow_of_two(caps->max_srq_sges); caps->num_aeq_vectors = resp_a->num_aeq_vectors; caps->num_other_vectors = resp_a->num_other_vectors; caps->max_sq_desc_sz = resp_a->max_sq_desc_sz; @@ -2336,7 +2354,6 @@ static int hns_roce_config_link_table(struct hns_roce_dev *hr_dev, struct hns_roce_link_table_entry *entry; enum hns_roce_opcode_type opcode; u32 page_num; - int i; switch (type) { case TSQ_LINK_TABLE: @@ -2354,14 +2371,10 @@ static int hns_roce_config_link_table(struct hns_roce_dev *hr_dev, page_num = link_tbl->npages; entry = link_tbl->table.buf; - for (i = 0; i < 2; i++) { - hns_roce_cmq_setup_basic_desc(&desc[i], opcode, false); + hns_roce_cmq_setup_basic_desc(&desc[0], opcode, false); + desc[0].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - if (i == 0) - desc[i].flag |= cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - else - desc[i].flag &= ~cpu_to_le16(HNS_ROCE_CMD_FLAG_NEXT); - } + hns_roce_cmq_setup_basic_desc(&desc[1], opcode, false); req_a->base_addr_l = cpu_to_le32(link_tbl->table.map & 0xffffffff); req_a->base_addr_h = cpu_to_le32(link_tbl->table.map >> 32); @@ -2880,36 +2893,20 @@ static int hns_roce_v2_write_mtpt(struct hns_roce_dev *hr_dev, mpt_entry = mb_buf; memset(mpt_entry, 0, sizeof(*mpt_entry)); - roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_MPT_ST_M, - V2_MPT_BYTE_4_MPT_ST_S, V2_MPT_ST_VALID); - roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PBL_HOP_NUM_M, - V2_MPT_BYTE_4_PBL_HOP_NUM_S, mr->pbl_hop_num == - HNS_ROCE_HOP_NUM_0 ? 0 : mr->pbl_hop_num); - roce_set_field(mpt_entry->byte_4_pd_hop_st, - V2_MPT_BYTE_4_PBL_BA_PG_SZ_M, - V2_MPT_BYTE_4_PBL_BA_PG_SZ_S, - to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.ba_pg_shift)); - roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M, - V2_MPT_BYTE_4_PD_S, mr->pd); - - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_RA_EN_S, 0); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_R_INV_EN_S, 0); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_L_INV_EN_S, 1); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_BIND_EN_S, - (mr->access & IB_ACCESS_MW_BIND ? 1 : 0)); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_ATOMIC_EN_S, - mr->access & IB_ACCESS_REMOTE_ATOMIC ? 1 : 0); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_RR_EN_S, - (mr->access & IB_ACCESS_REMOTE_READ ? 1 : 0)); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_RW_EN_S, - (mr->access & IB_ACCESS_REMOTE_WRITE ? 1 : 0)); - roce_set_bit(mpt_entry->byte_8_mw_cnt_en, V2_MPT_BYTE_8_LW_EN_S, - (mr->access & IB_ACCESS_LOCAL_WRITE ? 1 : 0)); - - roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_PA_S, - mr->type == MR_TYPE_MR ? 0 : 1); - roce_set_bit(mpt_entry->byte_12_mw_pa, V2_MPT_BYTE_12_INNER_PA_VLD_S, - 1); + hr_reg_write(mpt_entry, MPT_ST, V2_MPT_ST_VALID); + hr_reg_write(mpt_entry, MPT_PD, mr->pd); + hr_reg_enable(mpt_entry, MPT_L_INV_EN); + + hr_reg_write(mpt_entry, MPT_BIND_EN, + !!(mr->access & IB_ACCESS_MW_BIND)); + hr_reg_write(mpt_entry, MPT_ATOMIC_EN, + !!(mr->access & IB_ACCESS_REMOTE_ATOMIC)); + hr_reg_write(mpt_entry, MPT_RR_EN, + !!(mr->access & IB_ACCESS_REMOTE_READ)); + hr_reg_write(mpt_entry, MPT_RW_EN, + !!(mr->access & IB_ACCESS_REMOTE_WRITE)); + hr_reg_write(mpt_entry, MPT_LW_EN, + !!((mr->access & IB_ACCESS_LOCAL_WRITE))); mpt_entry->len_l = cpu_to_le32(lower_32_bits(mr->size)); mpt_entry->len_h = cpu_to_le32(upper_32_bits(mr->size)); @@ -2917,9 +2914,19 @@ static int hns_roce_v2_write_mtpt(struct hns_roce_dev *hr_dev, mpt_entry->va_l = cpu_to_le32(lower_32_bits(mr->iova)); mpt_entry->va_h = cpu_to_le32(upper_32_bits(mr->iova)); + if (mr->type != MR_TYPE_MR) + hr_reg_enable(mpt_entry, MPT_PA); + if (mr->type == MR_TYPE_DMA) return 0; + if (mr->pbl_hop_num != HNS_ROCE_HOP_NUM_0) + hr_reg_write(mpt_entry, MPT_PBL_HOP_NUM, mr->pbl_hop_num); + + hr_reg_write(mpt_entry, MPT_PBL_BA_PG_SZ, + to_hr_hw_page_shift(mr->pbl_mtr.hem_cfg.ba_pg_shift)); + hr_reg_enable(mpt_entry, MPT_INNER_PA_VLD); + ret = set_mtpt_pbl(hr_dev, mpt_entry, mr); return ret; @@ -2927,20 +2934,17 @@ static int hns_roce_v2_write_mtpt(struct hns_roce_dev *hr_dev, static int hns_roce_v2_rereg_write_mtpt(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, int flags, - u32 pdn, int mr_access_flags, u64 iova, - u64 size, void *mb_buf) + void *mb_buf) { struct hns_roce_v2_mpt_entry *mpt_entry = mb_buf; + u32 mr_access_flags = mr->access; int ret = 0; roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_MPT_ST_M, V2_MPT_BYTE_4_MPT_ST_S, V2_MPT_ST_VALID); - if (flags & IB_MR_REREG_PD) { - roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M, - V2_MPT_BYTE_4_PD_S, pdn); - mr->pd = pdn; - } + roce_set_field(mpt_entry->byte_4_pd_hop_st, V2_MPT_BYTE_4_PD_M, + V2_MPT_BYTE_4_PD_S, mr->pd); if (flags & IB_MR_REREG_ACCESS) { roce_set_bit(mpt_entry->byte_8_mw_cnt_en, @@ -2958,13 +2962,10 @@ static int hns_roce_v2_rereg_write_mtpt(struct hns_roce_dev *hr_dev, } if (flags & IB_MR_REREG_TRANS) { - mpt_entry->va_l = cpu_to_le32(lower_32_bits(iova)); - mpt_entry->va_h = cpu_to_le32(upper_32_bits(iova)); - mpt_entry->len_l = cpu_to_le32(lower_32_bits(size)); - mpt_entry->len_h = cpu_to_le32(upper_32_bits(size)); - - mr->iova = iova; - mr->size = size; + mpt_entry->va_l = cpu_to_le32(lower_32_bits(mr->iova)); + mpt_entry->va_h = cpu_to_le32(upper_32_bits(mr->iova)); + mpt_entry->len_l = cpu_to_le32(lower_32_bits(mr->size)); + mpt_entry->len_h = cpu_to_le32(upper_32_bits(mr->size)); ret = set_mtpt_pbl(hr_dev, mpt_entry, mr); } @@ -3126,11 +3127,6 @@ static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, if (nfreed) { hr_cq->cons_index += nfreed; - /* - * Make sure update of buffer contents is done before - * updating consumer index. - */ - wmb(); hns_roce_v2_cq_set_ci(hr_cq, hr_cq->cons_index); } } @@ -3639,11 +3635,8 @@ static int hns_roce_v2_poll_cq(struct ib_cq *ibcq, int num_entries, break; } - if (npolled) { - /* Memory barrier */ - wmb(); + if (npolled) hns_roce_v2_cq_set_ci(hr_cq, hr_cq->cons_index); - } out: spin_unlock_irqrestore(&hr_cq->lock, flags); @@ -4235,7 +4228,6 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, 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 ib_device *ibdev = &hr_dev->ib_dev; @@ -4243,7 +4235,6 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, dma_addr_t irrl_ba; enum ib_mtu mtu; u8 lp_pktn_ini; - u8 port_num; u64 *mtts; u8 *dmac; u8 *smac; @@ -4324,15 +4315,6 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp, V2_QPC_BYTE_56_DQPN_M, V2_QPC_BYTE_56_DQPN_S, 0); } - /* Configure GID index */ - port_num = rdma_ah_get_port_num(&attr->ah_attr); - roce_set_field(context->byte_20_smac_sgid_idx, - V2_QPC_BYTE_20_SGID_IDX_M, V2_QPC_BYTE_20_SGID_IDX_S, - hns_get_gid_index(hr_dev, port_num - 1, - 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); - memcpy(&(context->dmac), dmac, sizeof(u32)); roce_set_field(context->byte_52_udpspn_dmac, V2_QPC_BYTE_52_DMAC_M, V2_QPC_BYTE_52_DMAC_S, *((u16 *)(&dmac[4]))); @@ -5083,7 +5065,7 @@ static int hns_roce_v2_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *qp_attr, done: qp_attr->cur_qp_state = qp_attr->qp_state; qp_attr->cap.max_recv_wr = hr_qp->rq.wqe_cnt; - qp_attr->cap.max_recv_sge = hr_qp->rq.max_gs; + qp_attr->cap.max_recv_sge = hr_qp->rq.max_gs - hr_qp->rq.rsv_sge; if (!ibqp->uobject) { qp_attr->cap.max_send_wr = hr_qp->sq.wqe_cnt; @@ -5174,6 +5156,9 @@ static int hns_roce_v2_qp_flow_control_init(struct hns_roce_dev *hr_dev, struct hns_roce_cmq_desc desc; int ret, i; + if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) + return 0; + mutex_lock(&hr_dev->qp_table.scc_mutex); /* set scc ctx clear done flag */ @@ -5220,98 +5205,96 @@ out: return ret; } -static void hns_roce_v2_write_srqc(struct hns_roce_dev *hr_dev, - struct hns_roce_srq *srq, u32 pdn, u16 xrcd, - u32 cqn, void *mb_buf, u64 *mtts_wqe, - u64 *mtts_idx, dma_addr_t dma_handle_wqe, - dma_addr_t dma_handle_idx) +#define DMA_IDX_SHIFT 3 +#define DMA_WQE_SHIFT 3 + +static int hns_roce_v2_write_srqc_index_queue(struct hns_roce_srq *srq, + struct hns_roce_srq_context *ctx) { - struct hns_roce_srq_context *srq_context; + struct hns_roce_idx_que *idx_que = &srq->idx_que; + struct ib_device *ibdev = srq->ibsrq.device; + struct hns_roce_dev *hr_dev = to_hr_dev(ibdev); + u64 mtts_idx[MTT_MIN_COUNT] = {}; + dma_addr_t dma_handle_idx = 0; + int ret; + + /* Get physical address of idx que buf */ + ret = hns_roce_mtr_find(hr_dev, &idx_que->mtr, 0, mtts_idx, + ARRAY_SIZE(mtts_idx), &dma_handle_idx); + if (ret < 1) { + ibdev_err(ibdev, "failed to find mtr for SRQ idx, ret = %d.\n", + ret); + return -ENOBUFS; + } + + hr_reg_write(ctx, SRQC_IDX_HOP_NUM, + to_hr_hem_hopnum(hr_dev->caps.idx_hop_num, srq->wqe_cnt)); + + hr_reg_write(ctx, SRQC_IDX_BT_BA_L, dma_handle_idx >> DMA_IDX_SHIFT); + hr_reg_write(ctx, SRQC_IDX_BT_BA_H, + upper_32_bits(dma_handle_idx >> DMA_IDX_SHIFT)); + + hr_reg_write(ctx, SRQC_IDX_BA_PG_SZ, + to_hr_hw_page_shift(idx_que->mtr.hem_cfg.ba_pg_shift)); + hr_reg_write(ctx, SRQC_IDX_BUF_PG_SZ, + to_hr_hw_page_shift(idx_que->mtr.hem_cfg.buf_pg_shift)); + + hr_reg_write(ctx, SRQC_IDX_CUR_BLK_ADDR_L, + to_hr_hw_page_addr(mtts_idx[0])); + hr_reg_write(ctx, SRQC_IDX_CUR_BLK_ADDR_H, + upper_32_bits(to_hr_hw_page_addr(mtts_idx[0]))); + + hr_reg_write(ctx, SRQC_IDX_NXT_BLK_ADDR_L, + to_hr_hw_page_addr(mtts_idx[1])); + hr_reg_write(ctx, SRQC_IDX_NXT_BLK_ADDR_H, + upper_32_bits(to_hr_hw_page_addr(mtts_idx[1]))); + + return 0; +} - srq_context = mb_buf; - memset(srq_context, 0, sizeof(*srq_context)); - - roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQ_ST_M, - SRQC_BYTE_4_SRQ_ST_S, 1); - - roce_set_field(srq_context->byte_4_srqn_srqst, - SRQC_BYTE_4_SRQ_WQE_HOP_NUM_M, - SRQC_BYTE_4_SRQ_WQE_HOP_NUM_S, - to_hr_hem_hopnum(hr_dev->caps.srqwqe_hop_num, - srq->wqe_cnt)); - roce_set_field(srq_context->byte_4_srqn_srqst, - SRQC_BYTE_4_SRQ_SHIFT_M, SRQC_BYTE_4_SRQ_SHIFT_S, - ilog2(srq->wqe_cnt)); - - roce_set_field(srq_context->byte_4_srqn_srqst, SRQC_BYTE_4_SRQN_M, - SRQC_BYTE_4_SRQN_S, srq->srqn); - - roce_set_field(srq_context->byte_8_limit_wl, SRQC_BYTE_8_SRQ_LIMIT_WL_M, - SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); - - roce_set_field(srq_context->byte_12_xrcd, SRQC_BYTE_12_SRQ_XRCD_M, - SRQC_BYTE_12_SRQ_XRCD_S, xrcd); - - srq_context->wqe_bt_ba = cpu_to_le32((u32)(dma_handle_wqe >> 3)); - - roce_set_field(srq_context->byte_24_wqe_bt_ba, - SRQC_BYTE_24_SRQ_WQE_BT_BA_M, - SRQC_BYTE_24_SRQ_WQE_BT_BA_S, - dma_handle_wqe >> 35); - - roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_PD_M, - SRQC_BYTE_28_PD_S, pdn); - roce_set_field(srq_context->byte_28_rqws_pd, SRQC_BYTE_28_RQWS_M, - SRQC_BYTE_28_RQWS_S, srq->max_gs <= 0 ? 0 : - fls(srq->max_gs - 1)); - - srq_context->idx_bt_ba = cpu_to_le32(dma_handle_idx >> 3); - roce_set_field(srq_context->rsv_idx_bt_ba, - SRQC_BYTE_36_SRQ_IDX_BT_BA_M, - SRQC_BYTE_36_SRQ_IDX_BT_BA_S, - dma_handle_idx >> 35); - - srq_context->idx_cur_blk_addr = - cpu_to_le32(to_hr_hw_page_addr(mtts_idx[0])); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_M, - SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_S, - upper_32_bits(to_hr_hw_page_addr(mtts_idx[0]))); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_HOP_NUM_M, - SRQC_BYTE_44_SRQ_IDX_HOP_NUM_S, - to_hr_hem_hopnum(hr_dev->caps.idx_hop_num, - srq->wqe_cnt)); - - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_M, - SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_S, - to_hr_hw_page_shift(srq->idx_que.mtr.hem_cfg.ba_pg_shift)); - roce_set_field(srq_context->byte_44_idxbufpgsz_addr, - SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_M, - SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_S, - to_hr_hw_page_shift(srq->idx_que.mtr.hem_cfg.buf_pg_shift)); - - srq_context->idx_nxt_blk_addr = - cpu_to_le32(to_hr_hw_page_addr(mtts_idx[1])); - roce_set_field(srq_context->rsv_idxnxtblkaddr, - SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_M, - SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_S, - upper_32_bits(to_hr_hw_page_addr(mtts_idx[1]))); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_XRC_CQN_M, SRQC_BYTE_56_SRQ_XRC_CQN_S, - cqn); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_M, - SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_S, - to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.ba_pg_shift)); - roce_set_field(srq_context->byte_56_xrc_cqn, - SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_M, - SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_S, - to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.buf_pg_shift)); - - roce_set_bit(srq_context->db_record_addr_record_en, - SRQC_BYTE_60_SRQ_RECORD_EN_S, 0); +static int hns_roce_v2_write_srqc(struct hns_roce_srq *srq, void *mb_buf) +{ + struct ib_device *ibdev = srq->ibsrq.device; + struct hns_roce_dev *hr_dev = to_hr_dev(ibdev); + struct hns_roce_srq_context *ctx = mb_buf; + u64 mtts_wqe[MTT_MIN_COUNT] = {}; + dma_addr_t dma_handle_wqe = 0; + int ret; + + memset(ctx, 0, sizeof(*ctx)); + + /* Get the physical address of srq buf */ + ret = hns_roce_mtr_find(hr_dev, &srq->buf_mtr, 0, mtts_wqe, + ARRAY_SIZE(mtts_wqe), &dma_handle_wqe); + if (ret < 1) { + ibdev_err(ibdev, "failed to find mtr for SRQ WQE, ret = %d.\n", + ret); + return -ENOBUFS; + } + + hr_reg_write(ctx, SRQC_SRQ_ST, 1); + hr_reg_write(ctx, SRQC_PD, to_hr_pd(srq->ibsrq.pd)->pdn); + hr_reg_write(ctx, SRQC_SRQN, srq->srqn); + hr_reg_write(ctx, SRQC_XRCD, 0); + hr_reg_write(ctx, SRQC_XRC_CQN, srq->cqn); + hr_reg_write(ctx, SRQC_SHIFT, ilog2(srq->wqe_cnt)); + hr_reg_write(ctx, SRQC_RQWS, + srq->max_gs <= 0 ? 0 : fls(srq->max_gs - 1)); + + hr_reg_write(ctx, SRQC_WQE_HOP_NUM, + to_hr_hem_hopnum(hr_dev->caps.srqwqe_hop_num, + srq->wqe_cnt)); + + hr_reg_write(ctx, SRQC_WQE_BT_BA_L, dma_handle_wqe >> DMA_WQE_SHIFT); + hr_reg_write(ctx, SRQC_WQE_BT_BA_H, + upper_32_bits(dma_handle_wqe >> DMA_WQE_SHIFT)); + + hr_reg_write(ctx, SRQC_WQE_BA_PG_SZ, + to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.ba_pg_shift)); + hr_reg_write(ctx, SRQC_WQE_BUF_PG_SZ, + to_hr_hw_page_shift(srq->buf_mtr.hem_cfg.buf_pg_shift)); + + return hns_roce_v2_write_srqc_index_queue(srq, ctx); } static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq, @@ -5331,7 +5314,7 @@ static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq, return -EINVAL; if (srq_attr_mask & IB_SRQ_LIMIT) { - if (srq_attr->srq_limit >= srq->wqe_cnt) + if (srq_attr->srq_limit > srq->wqe_cnt) return -EINVAL; mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); @@ -5394,8 +5377,8 @@ static int hns_roce_v2_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr) SRQC_BYTE_8_SRQ_LIMIT_WL_S); attr->srq_limit = limit_wl; - attr->max_wr = srq->wqe_cnt - 1; - attr->max_sge = srq->max_gs; + attr->max_wr = srq->wqe_cnt; + attr->max_sge = srq->max_gs - srq->rsv_sge; out: hns_roce_free_cmd_mailbox(hr_dev, mailbox); @@ -5626,9 +5609,6 @@ 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)) - eq->cons_index = 0; - hns_roce_v2_init_irq_work(hr_dev, eq, queue_num); aeqe = next_aeqe_sw_v2(eq); @@ -5671,9 +5651,6 @@ static int hns_roce_v2_ceq_int(struct hns_roce_dev *hr_dev, ++eq->cons_index; ceqe_found = 1; - if (eq->cons_index > (EQ_DEPTH_COEFF * eq->entries - 1)) - eq->cons_index = 0; - ceqe = next_ceqe_sw_v2(eq); } @@ -5948,7 +5925,6 @@ static int alloc_eq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_eq *eq) buf_attr.region[0].size = eq->entries * eq->eqe_size; buf_attr.region[0].hopnum = eq->hop_num; buf_attr.region_count = 1; - buf_attr.fixed_page = true; err = hns_roce_mtr_create(hr_dev, &eq->mtr, &buf_attr, hr_dev->caps.eqe_ba_pg_sz + @@ -6286,6 +6262,7 @@ static void hns_roce_hw_v2_get_cfg(struct hns_roce_dev *hr_dev, /* Get info from NIC driver. */ hr_dev->reg_base = handle->rinfo.roce_io_base; + hr_dev->mem_base = handle->rinfo.roce_mem_base; hr_dev->caps.num_ports = 1; hr_dev->iboe.netdevs[0] = handle->rinfo.netdev; hr_dev->iboe.phy_port[0] = 0; diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h index bdaccf86460d..39621fb6ec16 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h @@ -96,7 +96,8 @@ #define HNS_ROCE_V2_CQC_TIMER_ENTRY_SZ PAGE_SIZE #define HNS_ROCE_V2_PAGE_SIZE_SUPPORTED 0xFFFFF000 #define HNS_ROCE_V2_MAX_INNER_MTPT_NUM 2 -#define HNS_ROCE_INVALID_LKEY 0x100 +#define HNS_ROCE_INVALID_LKEY 0x0 +#define HNS_ROCE_INVALID_SGE_LENGTH 0x80000000 #define HNS_ROCE_CMQ_TX_TIMEOUT 30000 #define HNS_ROCE_V2_UC_RC_SGE_NUM_IN_WQE 2 #define HNS_ROCE_V2_RSV_QPS 8 @@ -366,24 +367,61 @@ struct hns_roce_v2_cq_context { #define CQC_STASH CQC_FIELD_LOC(63, 63) struct hns_roce_srq_context { - __le32 byte_4_srqn_srqst; - __le32 byte_8_limit_wl; - __le32 byte_12_xrcd; - __le32 byte_16_pi_ci; - __le32 wqe_bt_ba; - __le32 byte_24_wqe_bt_ba; - __le32 byte_28_rqws_pd; - __le32 idx_bt_ba; - __le32 rsv_idx_bt_ba; - __le32 idx_cur_blk_addr; - __le32 byte_44_idxbufpgsz_addr; - __le32 idx_nxt_blk_addr; - __le32 rsv_idxnxtblkaddr; - __le32 byte_56_xrc_cqn; - __le32 db_record_addr_record_en; - __le32 db_record_addr; + __le32 byte_4_srqn_srqst; + __le32 byte_8_limit_wl; + __le32 byte_12_xrcd; + __le32 byte_16_pi_ci; + __le32 wqe_bt_ba; + __le32 byte_24_wqe_bt_ba; + __le32 byte_28_rqws_pd; + __le32 idx_bt_ba; + __le32 rsv_idx_bt_ba; + __le32 idx_cur_blk_addr; + __le32 byte_44_idxbufpgsz_addr; + __le32 idx_nxt_blk_addr; + __le32 rsv_idxnxtblkaddr; + __le32 byte_56_xrc_cqn; + __le32 db_record_addr_record_en; + __le32 db_record_addr; }; +#define SRQC_FIELD_LOC(h, l) FIELD_LOC(struct hns_roce_srq_context, h, l) + +#define SRQC_SRQ_ST SRQC_FIELD_LOC(1, 0) +#define SRQC_WQE_HOP_NUM SRQC_FIELD_LOC(3, 2) +#define SRQC_SHIFT SRQC_FIELD_LOC(7, 4) +#define SRQC_SRQN SRQC_FIELD_LOC(31, 8) +#define SRQC_LIMIT_WL SRQC_FIELD_LOC(47, 32) +#define SRQC_RSV0 SRQC_FIELD_LOC(63, 48) +#define SRQC_XRCD SRQC_FIELD_LOC(87, 64) +#define SRQC_RSV1 SRQC_FIELD_LOC(95, 88) +#define SRQC_PRODUCER_IDX SRQC_FIELD_LOC(111, 96) +#define SRQC_CONSUMER_IDX SRQC_FIELD_LOC(127, 112) +#define SRQC_WQE_BT_BA_L SRQC_FIELD_LOC(159, 128) +#define SRQC_WQE_BT_BA_H SRQC_FIELD_LOC(188, 160) +#define SRQC_RSV2 SRQC_FIELD_LOC(191, 189) +#define SRQC_PD SRQC_FIELD_LOC(215, 192) +#define SRQC_RQWS SRQC_FIELD_LOC(219, 216) +#define SRQC_RSV3 SRQC_FIELD_LOC(223, 220) +#define SRQC_IDX_BT_BA_L SRQC_FIELD_LOC(255, 224) +#define SRQC_IDX_BT_BA_H SRQC_FIELD_LOC(284, 256) +#define SRQC_RSV4 SRQC_FIELD_LOC(287, 285) +#define SRQC_IDX_CUR_BLK_ADDR_L SRQC_FIELD_LOC(319, 288) +#define SRQC_IDX_CUR_BLK_ADDR_H SRQC_FIELD_LOC(339, 320) +#define SRQC_RSV5 SRQC_FIELD_LOC(341, 340) +#define SRQC_IDX_HOP_NUM SRQC_FIELD_LOC(343, 342) +#define SRQC_IDX_BA_PG_SZ SRQC_FIELD_LOC(347, 344) +#define SRQC_IDX_BUF_PG_SZ SRQC_FIELD_LOC(351, 348) +#define SRQC_IDX_NXT_BLK_ADDR_L SRQC_FIELD_LOC(383, 352) +#define SRQC_IDX_NXT_BLK_ADDR_H SRQC_FIELD_LOC(403, 384) +#define SRQC_RSV6 SRQC_FIELD_LOC(415, 404) +#define SRQC_XRC_CQN SRQC_FIELD_LOC(439, 416) +#define SRQC_WQE_BA_PG_SZ SRQC_FIELD_LOC(443, 440) +#define SRQC_WQE_BUF_PG_SZ SRQC_FIELD_LOC(447, 444) +#define SRQC_DB_RECORD_EN SRQC_FIELD_LOC(448, 448) +#define SRQC_DB_RECORD_ADDR_L SRQC_FIELD_LOC(479, 449) +#define SRQC_DB_RECORD_ADDR_H SRQC_FIELD_LOC(511, 480) + #define SRQC_BYTE_4_SRQ_ST_S 0 #define SRQC_BYTE_4_SRQ_ST_M GENMASK(1, 0) @@ -993,6 +1031,45 @@ struct hns_roce_v2_mpt_entry { __le32 byte_64_buf_pa1; }; +#define MPT_FIELD_LOC(h, l) FIELD_LOC(struct hns_roce_v2_mpt_entry, h, l) + +#define MPT_ST MPT_FIELD_LOC(1, 0) +#define MPT_PBL_HOP_NUM MPT_FIELD_LOC(3, 2) +#define MPT_PBL_BA_PG_SZ MPT_FIELD_LOC(7, 4) +#define MPT_PD MPT_FIELD_LOC(31, 8) +#define MPT_RA_EN MPT_FIELD_LOC(32, 32) +#define MPT_R_INV_EN MPT_FIELD_LOC(33, 33) +#define MPT_L_INV_EN MPT_FIELD_LOC(34, 34) +#define MPT_BIND_EN MPT_FIELD_LOC(35, 35) +#define MPT_ATOMIC_EN MPT_FIELD_LOC(36, 36) +#define MPT_RR_EN MPT_FIELD_LOC(37, 37) +#define MPT_RW_EN MPT_FIELD_LOC(38, 38) +#define MPT_LW_EN MPT_FIELD_LOC(39, 39) +#define MPT_MW_CNT MPT_FIELD_LOC(63, 40) +#define MPT_FRE MPT_FIELD_LOC(64, 64) +#define MPT_PA MPT_FIELD_LOC(65, 65) +#define MPT_ZBVA MPT_FIELD_LOC(66, 66) +#define MPT_SHARE MPT_FIELD_LOC(67, 67) +#define MPT_MR_MW MPT_FIELD_LOC(68, 68) +#define MPT_BPD MPT_FIELD_LOC(69, 69) +#define MPT_BQP MPT_FIELD_LOC(70, 70) +#define MPT_INNER_PA_VLD MPT_FIELD_LOC(71, 71) +#define MPT_MW_BIND_QPN MPT_FIELD_LOC(95, 72) +#define MPT_BOUND_LKEY MPT_FIELD_LOC(127, 96) +#define MPT_LEN MPT_FIELD_LOC(191, 128) +#define MPT_LKEY MPT_FIELD_LOC(223, 192) +#define MPT_VA MPT_FIELD_LOC(287, 224) +#define MPT_PBL_SIZE MPT_FIELD_LOC(319, 288) +#define MPT_PBL_BA MPT_FIELD_LOC(380, 320) +#define MPT_BLK_MODE MPT_FIELD_LOC(381, 381) +#define MPT_RSV0 MPT_FIELD_LOC(383, 382) +#define MPT_PA0 MPT_FIELD_LOC(441, 384) +#define MPT_BOUND_VA MPT_FIELD_LOC(447, 442) +#define MPT_PA1 MPT_FIELD_LOC(505, 448) +#define MPT_PERSIST_EN MPT_FIELD_LOC(506, 506) +#define MPT_RSV2 MPT_FIELD_LOC(507, 507) +#define MPT_PBL_BUF_PG_SZ MPT_FIELD_LOC(511, 508) + #define V2_MPT_BYTE_4_MPT_ST_S 0 #define V2_MPT_BYTE_4_MPT_ST_M GENMASK(1, 0) @@ -1059,6 +1136,8 @@ struct hns_roce_v2_mpt_entry { #define V2_DB_BYTE_4_CMD_S 24 #define V2_DB_BYTE_4_CMD_M GENMASK(27, 24) +#define V2_DB_FLAG_S 31 + #define V2_DB_PARAMETER_IDX_S 0 #define V2_DB_PARAMETER_IDX_M GENMASK(15, 0) @@ -1155,6 +1234,15 @@ struct hns_roce_v2_rc_send_wqe { #define V2_RC_SEND_WQE_BYTE_4_OPCODE_S 0 #define V2_RC_SEND_WQE_BYTE_4_OPCODE_M GENMASK(4, 0) +#define V2_RC_SEND_WQE_BYTE_4_DB_SL_L_S 5 +#define V2_RC_SEND_WQE_BYTE_4_DB_SL_L_M GENMASK(6, 5) + +#define V2_RC_SEND_WQE_BYTE_4_DB_SL_H_S 13 +#define V2_RC_SEND_WQE_BYTE_4_DB_SL_H_M GENMASK(14, 13) + +#define V2_RC_SEND_WQE_BYTE_4_WQE_INDEX_S 15 +#define V2_RC_SEND_WQE_BYTE_4_WQE_INDEX_M GENMASK(30, 15) + #define V2_RC_SEND_WQE_BYTE_4_OWNER_S 7 #define V2_RC_SEND_WQE_BYTE_4_CQE_S 8 @@ -1167,15 +1255,17 @@ struct hns_roce_v2_rc_send_wqe { #define V2_RC_SEND_WQE_BYTE_4_INLINE_S 12 -#define V2_RC_FRMR_WQE_BYTE_4_BIND_EN_S 19 +#define V2_RC_FRMR_WQE_BYTE_40_BIND_EN_S 10 + +#define V2_RC_FRMR_WQE_BYTE_40_ATOMIC_S 11 -#define V2_RC_FRMR_WQE_BYTE_4_ATOMIC_S 20 +#define V2_RC_FRMR_WQE_BYTE_40_RR_S 12 -#define V2_RC_FRMR_WQE_BYTE_4_RR_S 21 +#define V2_RC_FRMR_WQE_BYTE_40_RW_S 13 -#define V2_RC_FRMR_WQE_BYTE_4_RW_S 22 +#define V2_RC_FRMR_WQE_BYTE_40_LW_S 14 -#define V2_RC_FRMR_WQE_BYTE_4_LW_S 23 +#define V2_RC_SEND_WQE_BYTE_4_FLAG_S 31 #define V2_RC_SEND_WQE_BYTE_16_XRC_SRQN_S 0 #define V2_RC_SEND_WQE_BYTE_16_XRC_SRQN_M GENMASK(23, 0) @@ -1190,7 +1280,7 @@ struct hns_roce_v2_rc_send_wqe { struct hns_roce_wqe_frmr_seg { __le32 pbl_size; - __le32 mode_buf_pg_sz; + __le32 byte_40; }; #define V2_RC_FRMR_WQE_BYTE_40_PBL_BUF_PG_SZ_S 4 @@ -1786,12 +1876,8 @@ struct hns_roce_v2_cmq_ring { dma_addr_t desc_dma_addr; struct hns_roce_cmq_desc *desc; u32 head; - u32 tail; - u16 buf_size; u16 desc_num; - int next_to_use; - int next_to_clean; u8 flag; spinlock_t lock; /* command queue lock */ }; @@ -1800,7 +1886,6 @@ struct hns_roce_v2_cmq { struct hns_roce_v2_cmq_ring csq; struct hns_roce_v2_cmq_ring crq; u16 tx_timeout; - u16 last_status; }; enum hns_roce_link_table_type { diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c index d9179bae4989..c9c0836394a2 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -42,22 +42,6 @@ #include "hns_roce_device.h" #include "hns_roce_hem.h" -/** - * hns_get_gid_index - Get gid index. - * @hr_dev: pointer to structure hns_roce_dev. - * @port: port, value range: 0 ~ MAX - * @gid_index: gid_index, value range: 0 ~ MAX - * Description: - * N ports shared gids, allocation method as follow: - * GID[0][0], GID[1][0],.....GID[N - 1][0], - * GID[0][0], GID[1][0],.....GID[N - 1][0], - * And so on - */ -u8 hns_get_gid_index(struct hns_roce_dev *hr_dev, u8 port, int gid_index) -{ - return gid_index * hr_dev->caps.num_ports + port; -} - static int hns_roce_set_mac(struct hns_roce_dev *hr_dev, u8 port, u8 *addr) { u8 phy_port; @@ -217,7 +201,8 @@ static int hns_roce_query_device(struct ib_device *ib_dev, props->max_srq_sge = hr_dev->caps.max_srq_sges; } - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_FRMR) { + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_FRMR && + hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) { props->device_cap_flags |= IB_DEVICE_MEM_MGT_EXTENSIONS; props->max_fast_reg_page_list_len = HNS_ROCE_FRMR_MAX_PA; } @@ -748,11 +733,7 @@ static int hns_roce_setup_hca(struct hns_roce_dev *hr_dev) goto err_pd_table_free; } - ret = hns_roce_init_cq_table(hr_dev); - if (ret) { - dev_err(dev, "Failed to init completion queue table.\n"); - goto err_mr_table_free; - } + hns_roce_init_cq_table(hr_dev); ret = hns_roce_init_qp_table(hr_dev); if (ret) { @@ -772,13 +753,10 @@ static int hns_roce_setup_hca(struct hns_roce_dev *hr_dev) return 0; err_qp_table_free: - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) - hns_roce_cleanup_qp_table(hr_dev); + hns_roce_cleanup_qp_table(hr_dev); err_cq_table_free: hns_roce_cleanup_cq_table(hr_dev); - -err_mr_table_free: hns_roce_cleanup_mr_table(hr_dev); err_pd_table_free: diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c index 1bcffd93ff3e..79b3c3023fe7 100644 --- a/drivers/infiniband/hw/hns/hns_roce_mr.c +++ b/drivers/infiniband/hw/hns/hns_roce_mr.c @@ -66,8 +66,7 @@ int hns_roce_hw_destroy_mpt(struct hns_roce_dev *hr_dev, HNS_ROCE_CMD_TIMEOUT_MSECS); } -static int alloc_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, - u32 pd, u64 iova, u64 size, u32 access) +static int alloc_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr) { struct ib_device *ibdev = &hr_dev->ib_dev; unsigned long obj = 0; @@ -82,11 +81,6 @@ static int alloc_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, return -ENOMEM; } - mr->iova = iova; /* MR va starting addr */ - mr->size = size; /* MR addr range */ - mr->pd = pd; /* MR num */ - mr->access = access; /* MR access permit */ - mr->enabled = 0; /* MR active status */ mr->key = hw_index_to_key(obj); /* MR key */ err = hns_roce_table_get(hr_dev, &hr_dev->mr_table.mtpt_table, obj); @@ -110,8 +104,7 @@ static void free_mr_key(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr) } static int alloc_mr_pbl(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, - size_t length, struct ib_udata *udata, u64 start, - int access) + struct ib_udata *udata, u64 start) { struct ib_device *ibdev = &hr_dev->ib_dev; bool is_fast = mr->type == MR_TYPE_FRMR; @@ -121,11 +114,10 @@ static int alloc_mr_pbl(struct hns_roce_dev *hr_dev, struct hns_roce_mr *mr, mr->pbl_hop_num = is_fast ? 1 : hr_dev->caps.pbl_hop_num; buf_attr.page_shift = is_fast ? PAGE_SHIFT : hr_dev->caps.pbl_buf_pg_sz + PAGE_SHIFT; - buf_attr.region[0].size = length; + buf_attr.region[0].size = mr->size; buf_attr.region[0].hopnum = mr->pbl_hop_num; buf_attr.region_count = 1; - buf_attr.fixed_page = true; - buf_attr.user_access = access; + buf_attr.user_access = mr->access; /* fast MR's buffer is alloced before mapping, not at creation */ buf_attr.mtt_only = is_fast; @@ -197,9 +189,6 @@ static int hns_roce_mr_enable(struct hns_roce_dev *hr_dev, } mr->enabled = 1; - hns_roce_free_cmd_mailbox(hr_dev, mailbox); - - return 0; err_page: hns_roce_free_cmd_mailbox(hr_dev, mailbox); @@ -237,14 +226,16 @@ struct ib_mr *hns_roce_get_dma_mr(struct ib_pd *pd, int acc) return ERR_PTR(-ENOMEM); mr->type = MR_TYPE_DMA; + mr->pd = to_hr_pd(pd)->pdn; + mr->access = acc; /* Allocate memory region key */ hns_roce_hem_list_init(&mr->pbl_mtr.hem_list); - ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, 0, 0, acc); + ret = alloc_mr_key(hr_dev, mr); if (ret) goto err_free; - ret = hns_roce_mr_enable(to_hr_dev(pd->device), mr); + ret = hns_roce_mr_enable(hr_dev, mr); if (ret) goto err_mr; @@ -271,13 +262,17 @@ struct ib_mr *hns_roce_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, if (!mr) return ERR_PTR(-ENOMEM); + mr->iova = virt_addr; + mr->size = length; + mr->pd = to_hr_pd(pd)->pdn; + mr->access = access_flags; mr->type = MR_TYPE_MR; - ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, virt_addr, length, - access_flags); + + ret = alloc_mr_key(hr_dev, mr); if (ret) goto err_alloc_mr; - ret = alloc_mr_pbl(hr_dev, mr, length, udata, start, access_flags); + ret = alloc_mr_pbl(hr_dev, mr, udata, start); if (ret) goto err_alloc_key; @@ -299,35 +294,6 @@ err_alloc_mr: 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 ib_device *ibdev = &hr_dev->ib_dev; - struct hns_roce_mr *mr = to_hr_mr(ibmr); - int ret; - - free_mr_pbl(hr_dev, mr); - ret = alloc_mr_pbl(hr_dev, mr, length, udata, start, mr_access_flags); - if (ret) { - ibdev_err(ibdev, "failed to create mr PBL, ret = %d.\n", ret); - return ret; - } - - ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, - mr_access_flags, virt_addr, - length, mailbox->buf); - if (ret) { - ibdev_err(ibdev, "failed to write mtpt, ret = %d.\n", ret); - free_mr_pbl(hr_dev, mr); - } - - return ret; -} - struct ib_mr *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, @@ -338,7 +304,6 @@ struct ib_mr *hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, struct hns_roce_mr *mr = to_hr_mr(ibmr); struct hns_roce_cmd_mailbox *mailbox; unsigned long mtpt_idx; - u32 pdn = 0; int ret; if (!mr->enabled) @@ -360,23 +325,29 @@ struct ib_mr *hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, ibdev_warn(ib_dev, "failed to destroy MPT, ret = %d.\n", ret); mr->enabled = 0; + mr->iova = virt_addr; + mr->size = length; if (flags & IB_MR_REREG_PD) - pdn = to_hr_pd(pd)->pdn; + mr->pd = to_hr_pd(pd)->pdn; + + if (flags & IB_MR_REREG_ACCESS) + mr->access = mr_access_flags; if (flags & IB_MR_REREG_TRANS) { - ret = rereg_mr_trans(ibmr, flags, - start, length, - virt_addr, mr_access_flags, - mailbox, pdn, udata); - if (ret) - goto free_cmd_mbox; - } else { - ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, pdn, - mr_access_flags, virt_addr, - length, mailbox->buf); - if (ret) + free_mr_pbl(hr_dev, mr); + ret = alloc_mr_pbl(hr_dev, mr, udata, start); + if (ret) { + ibdev_err(ib_dev, "failed to alloc mr PBL, ret = %d.\n", + ret); goto free_cmd_mbox; + } + } + + ret = hr_dev->hw->rereg_write_mtpt(hr_dev, mr, flags, mailbox->buf); + if (ret) { + ibdev_err(ib_dev, "failed to write mtpt, ret = %d.\n", ret); + goto free_cmd_mbox; } ret = hns_roce_hw_create_mpt(hr_dev, mailbox, mtpt_idx); @@ -386,12 +357,6 @@ struct ib_mr *hns_roce_rereg_user_mr(struct ib_mr *ibmr, int flags, u64 start, } mr->enabled = 1; - if (flags & IB_MR_REREG_ACCESS) - mr->access = mr_access_flags; - - hns_roce_free_cmd_mailbox(hr_dev, mailbox); - - return NULL; free_cmd_mbox: hns_roce_free_cmd_mailbox(hr_dev, mailbox); @@ -421,7 +386,6 @@ struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); struct device *dev = hr_dev->dev; struct hns_roce_mr *mr; - u64 length; int ret; if (mr_type != IB_MR_TYPE_MEM_REG) @@ -438,14 +402,15 @@ struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, return ERR_PTR(-ENOMEM); mr->type = MR_TYPE_FRMR; + mr->pd = to_hr_pd(pd)->pdn; + mr->size = max_num_sg * (1 << PAGE_SHIFT); /* Allocate memory region key */ - length = max_num_sg * (1 << PAGE_SHIFT); - ret = alloc_mr_key(hr_dev, mr, to_hr_pd(pd)->pdn, 0, length, 0); + ret = alloc_mr_key(hr_dev, mr); if (ret) goto err_free; - ret = alloc_mr_pbl(hr_dev, mr, length, NULL, 0, 0); + ret = alloc_mr_pbl(hr_dev, mr, NULL, 0); if (ret) goto err_key; @@ -454,7 +419,7 @@ struct ib_mr *hns_roce_alloc_mr(struct ib_pd *pd, enum ib_mr_type mr_type, goto err_pbl; mr->ibmr.rkey = mr->ibmr.lkey = mr->key; - mr->ibmr.length = length; + mr->ibmr.length = mr->size; return &mr->ibmr; @@ -631,30 +596,26 @@ int hns_roce_dealloc_mw(struct ib_mw *ibmw) } static int mtr_map_region(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, - dma_addr_t *pages, struct hns_roce_buf_region *region) + struct hns_roce_buf_region *region, dma_addr_t *pages, + int max_count) { + int count, npage; + int offset, end; __le64 *mtts; - int offset; - int count; - int npage; u64 addr; - int end; int i; - /* if hopnum is 0, buffer cannot store BAs, so skip write mtt */ - if (!region->hopnum) - return 0; - offset = region->offset; end = offset + region->count; npage = 0; - while (offset < end) { + while (offset < end && npage < max_count) { + count = 0; mtts = hns_roce_hem_list_find_mtt(hr_dev, &mtr->hem_list, offset, &count, NULL); if (!mtts) return -ENOBUFS; - for (i = 0; i < count; i++) { + for (i = 0; i < count && npage < max_count; i++) { if (hr_dev->hw_rev == HNS_ROCE_HW_VER1) addr = to_hr_hw_page_addr(pages[npage]); else @@ -666,7 +627,7 @@ static int mtr_map_region(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, offset += count; } - return 0; + return npage; } static inline bool mtr_has_mtt(struct hns_roce_buf_attr *attr) @@ -729,25 +690,15 @@ static void mtr_free_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr) } static int mtr_alloc_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, - struct hns_roce_buf_attr *buf_attr, bool is_direct, + struct hns_roce_buf_attr *buf_attr, struct ib_udata *udata, unsigned long user_addr) { struct ib_device *ibdev = &hr_dev->ib_dev; - unsigned int best_pg_shift; - int all_pg_count = 0; size_t total_size; - int ret; total_size = mtr_bufs_size(buf_attr); - if (total_size < 1) { - ibdev_err(ibdev, "failed to check mtr size\n."); - return -EINVAL; - } if (udata) { - unsigned long pgsz_bitmap; - unsigned long page_size; - mtr->kmem = NULL; mtr->umem = ib_umem_get(ibdev, user_addr, total_size, buf_attr->user_access); @@ -756,76 +707,67 @@ static int mtr_alloc_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, PTR_ERR(mtr->umem)); return -ENOMEM; } - if (buf_attr->fixed_page) - pgsz_bitmap = 1 << buf_attr->page_shift; - else - pgsz_bitmap = GENMASK(buf_attr->page_shift, PAGE_SHIFT); - - page_size = ib_umem_find_best_pgsz(mtr->umem, pgsz_bitmap, - user_addr); - if (!page_size) - return -EINVAL; - best_pg_shift = order_base_2(page_size); - all_pg_count = ib_umem_num_dma_blocks(mtr->umem, page_size); - ret = 0; } else { mtr->umem = NULL; - mtr->kmem = - hns_roce_buf_alloc(hr_dev, total_size, - buf_attr->page_shift, - is_direct ? HNS_ROCE_BUF_DIRECT : 0); + mtr->kmem = hns_roce_buf_alloc(hr_dev, total_size, + buf_attr->page_shift, + mtr->hem_cfg.is_direct ? + HNS_ROCE_BUF_DIRECT : 0); if (IS_ERR(mtr->kmem)) { ibdev_err(ibdev, "failed to alloc kmem, ret = %ld.\n", PTR_ERR(mtr->kmem)); return PTR_ERR(mtr->kmem); } - - best_pg_shift = buf_attr->page_shift; - all_pg_count = mtr->kmem->npages; - } - - /* must bigger than minimum hardware page shift */ - if (best_pg_shift < HNS_HW_PAGE_SHIFT || all_pg_count < 1) { - ret = -EINVAL; - ibdev_err(ibdev, - "failed to check mtr, page shift = %u count = %d.\n", - best_pg_shift, all_pg_count); - goto err_alloc_mem; } - mtr->hem_cfg.buf_pg_shift = best_pg_shift; - mtr->hem_cfg.buf_pg_count = all_pg_count; - return 0; -err_alloc_mem: - mtr_free_bufs(hr_dev, mtr); - return ret; } -static int mtr_get_pages(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, - dma_addr_t *pages, int count, unsigned int page_shift) +static int mtr_map_bufs(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + int page_count, unsigned int page_shift) { struct ib_device *ibdev = &hr_dev->ib_dev; + dma_addr_t *pages; int npage; - int err; + int ret; + + /* alloc a tmp array to store buffer's dma address */ + pages = kvcalloc(page_count, sizeof(dma_addr_t), GFP_KERNEL); + if (!pages) + return -ENOMEM; if (mtr->umem) - npage = hns_roce_get_umem_bufs(hr_dev, pages, count, 0, + npage = hns_roce_get_umem_bufs(hr_dev, pages, page_count, 0, mtr->umem, page_shift); else - npage = hns_roce_get_kmem_bufs(hr_dev, pages, count, 0, + npage = hns_roce_get_kmem_bufs(hr_dev, pages, page_count, 0, mtr->kmem); + if (npage != page_count) { + ibdev_err(ibdev, "failed to get mtr page %d != %d.\n", npage, + page_count); + ret = -ENOBUFS; + goto err_alloc_list; + } + if (mtr->hem_cfg.is_direct && npage > 1) { - err = mtr_check_direct_pages(pages, npage, page_shift); - if (err) { - ibdev_err(ibdev, "Failed to check %s direct page-%d\n", - mtr->umem ? "user" : "kernel", err); - npage = err; + ret = mtr_check_direct_pages(pages, npage, page_shift); + if (ret) { + ibdev_err(ibdev, "failed to check %s mtr, idx = %d.\n", + mtr->umem ? "user" : "kernel", ret); + ret = -ENOBUFS; + goto err_alloc_list; } } - return npage; + ret = hns_roce_mtr_map(hr_dev, mtr, pages, page_count); + if (ret) + ibdev_err(ibdev, "failed to map mtr pages, ret = %d.\n", ret); + +err_alloc_list: + kvfree(pages); + + return ret; } int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, @@ -833,8 +775,8 @@ int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, { struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_buf_region *r; - unsigned int i; - int err; + unsigned int i, mapped_cnt; + int ret; /* * Only use the first page address as root ba when hopnum is 0, this @@ -845,26 +787,42 @@ int hns_roce_mtr_map(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, return 0; } - for (i = 0; i < mtr->hem_cfg.region_count; i++) { + for (i = 0, mapped_cnt = 0; i < mtr->hem_cfg.region_count && + mapped_cnt < page_cnt; i++) { r = &mtr->hem_cfg.region[i]; + /* if hopnum is 0, no need to map pages in this region */ + if (!r->hopnum) { + mapped_cnt += r->count; + continue; + } + if (r->offset + r->count > page_cnt) { - err = -EINVAL; + ret = -EINVAL; ibdev_err(ibdev, "failed to check mtr%u end %u + %u, max %u.\n", i, r->offset, r->count, page_cnt); - return err; + return ret; } - err = mtr_map_region(hr_dev, mtr, &pages[r->offset], r); - if (err) { + ret = mtr_map_region(hr_dev, mtr, r, &pages[r->offset], + page_cnt - mapped_cnt); + if (ret < 0) { ibdev_err(ibdev, "failed to map mtr%u offset %u, ret = %d.\n", - i, r->offset, err); - return err; + i, r->offset, ret); + return ret; } + mapped_cnt += ret; + ret = 0; } - return 0; + if (mapped_cnt < page_cnt) { + ret = -ENOBUFS; + ibdev_err(ibdev, "failed to map mtr pages count: %u < %u.\n", + mapped_cnt, page_cnt); + } + + return ret; } int hns_roce_mtr_find(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, @@ -928,68 +886,92 @@ done: static int mtr_init_buf_cfg(struct hns_roce_dev *hr_dev, struct hns_roce_buf_attr *attr, struct hns_roce_hem_cfg *cfg, - unsigned int *buf_page_shift) + unsigned int *buf_page_shift, int unalinged_size) { struct hns_roce_buf_region *r; + int first_region_padding; + int page_cnt, region_cnt; unsigned int page_shift; - int page_cnt = 0; size_t buf_size; - int region_cnt; + /* If mtt is disabled, all pages must be within a continuous range */ + cfg->is_direct = !mtr_has_mtt(attr); + buf_size = mtr_bufs_size(attr); if (cfg->is_direct) { - buf_size = cfg->buf_pg_count << cfg->buf_pg_shift; - page_cnt = DIV_ROUND_UP(buf_size, HNS_HW_PAGE_SIZE); - /* - * When HEM buffer use level-0 addressing, the page size equals - * the buffer size, and the the page size = 4K * 2^N. + /* When HEM buffer uses 0-level addressing, the page size is + * equal to the whole buffer size, and we split the buffer into + * small pages which is used to check whether the adjacent + * units are in the continuous space and its size is fixed to + * 4K based on hns ROCEE's requirement. */ - cfg->buf_pg_shift = HNS_HW_PAGE_SHIFT + order_base_2(page_cnt); - if (attr->region_count > 1) { - cfg->buf_pg_count = page_cnt; - page_shift = HNS_HW_PAGE_SHIFT; - } else { - cfg->buf_pg_count = 1; - page_shift = cfg->buf_pg_shift; - if (buf_size != 1 << page_shift) { - ibdev_err(&hr_dev->ib_dev, - "failed to check direct size %zu shift %d.\n", - buf_size, page_shift); - return -EINVAL; - } - } + page_shift = HNS_HW_PAGE_SHIFT; + + /* The ROCEE requires the page size to be 4K * 2 ^ N. */ + cfg->buf_pg_count = 1; + cfg->buf_pg_shift = HNS_HW_PAGE_SHIFT + + order_base_2(DIV_ROUND_UP(buf_size, HNS_HW_PAGE_SIZE)); + first_region_padding = 0; } else { - page_shift = cfg->buf_pg_shift; + page_shift = attr->page_shift; + cfg->buf_pg_count = DIV_ROUND_UP(buf_size + unalinged_size, + 1 << page_shift); + cfg->buf_pg_shift = page_shift; + first_region_padding = unalinged_size; } - /* convert buffer size to page index and page count */ - for (page_cnt = 0, region_cnt = 0; page_cnt < cfg->buf_pg_count && - region_cnt < attr->region_count && + /* Convert buffer size to page index and page count for each region and + * the buffer's offset needs to be appended to the first region. + */ + for (page_cnt = 0, region_cnt = 0; region_cnt < attr->region_count && region_cnt < ARRAY_SIZE(cfg->region); region_cnt++) { r = &cfg->region[region_cnt]; r->offset = page_cnt; - buf_size = hr_hw_page_align(attr->region[region_cnt].size); + buf_size = hr_hw_page_align(attr->region[region_cnt].size + + first_region_padding); r->count = DIV_ROUND_UP(buf_size, 1 << page_shift); + first_region_padding = 0; page_cnt += r->count; r->hopnum = to_hr_hem_hopnum(attr->region[region_cnt].hopnum, r->count); } - if (region_cnt < 1) { - ibdev_err(&hr_dev->ib_dev, - "failed to check mtr region count, pages = %d.\n", - cfg->buf_pg_count); - return -ENOBUFS; - } - cfg->region_count = region_cnt; *buf_page_shift = page_shift; return page_cnt; } +static int mtr_alloc_mtt(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, + unsigned int ba_page_shift) +{ + struct hns_roce_hem_cfg *cfg = &mtr->hem_cfg; + int ret; + + hns_roce_hem_list_init(&mtr->hem_list); + if (!cfg->is_direct) { + ret = hns_roce_hem_list_request(hr_dev, &mtr->hem_list, + cfg->region, cfg->region_count, + ba_page_shift); + if (ret) + return ret; + cfg->root_ba = mtr->hem_list.root_ba; + cfg->ba_pg_shift = ba_page_shift; + } else { + cfg->ba_pg_shift = cfg->buf_pg_shift; + } + + return 0; +} + +static void mtr_free_mtt(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr) +{ + hns_roce_hem_list_release(hr_dev, &mtr->hem_list); +} + /** * hns_roce_mtr_create - Create hns memory translate region. * + * @hr_dev: RoCE device struct pointer * @mtr: memory translate region * @buf_attr: buffer attribute for creating mtr * @ba_page_shift: page shift for multi-hop base address table @@ -1001,95 +983,51 @@ int hns_roce_mtr_create(struct hns_roce_dev *hr_dev, struct hns_roce_mtr *mtr, unsigned int ba_page_shift, struct ib_udata *udata, unsigned long user_addr) { - struct hns_roce_hem_cfg *cfg = &mtr->hem_cfg; struct ib_device *ibdev = &hr_dev->ib_dev; unsigned int buf_page_shift = 0; - dma_addr_t *pages = NULL; - int all_pg_cnt; - int get_pg_cnt; - int ret = 0; - - /* if disable mtt, all pages must in a continuous address range */ - cfg->is_direct = !mtr_has_mtt(buf_attr); - - /* if buffer only need mtt, just init the hem cfg */ - if (buf_attr->mtt_only) { - cfg->buf_pg_shift = buf_attr->page_shift; - cfg->buf_pg_count = mtr_bufs_size(buf_attr) >> - buf_attr->page_shift; - mtr->umem = NULL; - mtr->kmem = NULL; - } else { - ret = mtr_alloc_bufs(hr_dev, mtr, buf_attr, cfg->is_direct, - udata, user_addr); - if (ret) { - ibdev_err(ibdev, - "failed to alloc mtr bufs, ret = %d.\n", ret); - return ret; - } - } + int buf_page_cnt; + int ret; - all_pg_cnt = mtr_init_buf_cfg(hr_dev, buf_attr, cfg, &buf_page_shift); - if (all_pg_cnt < 1) { - ret = -ENOBUFS; - ibdev_err(ibdev, "failed to init mtr buf cfg.\n"); - goto err_alloc_bufs; + buf_page_cnt = mtr_init_buf_cfg(hr_dev, buf_attr, &mtr->hem_cfg, + &buf_page_shift, + udata ? user_addr & ~PAGE_MASK : 0); + if (buf_page_cnt < 1 || buf_page_shift < HNS_HW_PAGE_SHIFT) { + ibdev_err(ibdev, "failed to init mtr cfg, count %d shift %d.\n", + buf_page_cnt, buf_page_shift); + return -EINVAL; } - hns_roce_hem_list_init(&mtr->hem_list); - if (!cfg->is_direct) { - ret = hns_roce_hem_list_request(hr_dev, &mtr->hem_list, - cfg->region, cfg->region_count, - ba_page_shift); - if (ret) { - ibdev_err(ibdev, "failed to request mtr hem, ret = %d.\n", - ret); - goto err_alloc_bufs; - } - cfg->root_ba = mtr->hem_list.root_ba; - cfg->ba_pg_shift = ba_page_shift; - } else { - cfg->ba_pg_shift = cfg->buf_pg_shift; + ret = mtr_alloc_mtt(hr_dev, mtr, ba_page_shift); + if (ret) { + ibdev_err(ibdev, "failed to alloc mtr mtt, ret = %d.\n", ret); + return ret; } - /* no buffer to map */ - if (buf_attr->mtt_only) + /* The caller has its own buffer list and invokes the hns_roce_mtr_map() + * to finish the MTT configuration. + */ + if (buf_attr->mtt_only) { + mtr->umem = NULL; + mtr->kmem = NULL; return 0; - - /* alloc a tmp array to store buffer's dma address */ - pages = kvcalloc(all_pg_cnt, sizeof(dma_addr_t), GFP_KERNEL); - if (!pages) { - ret = -ENOMEM; - ibdev_err(ibdev, "failed to alloc mtr page list %d.\n", - all_pg_cnt); - goto err_alloc_hem_list; - } - - get_pg_cnt = mtr_get_pages(hr_dev, mtr, pages, all_pg_cnt, - buf_page_shift); - if (get_pg_cnt != all_pg_cnt) { - ibdev_err(ibdev, "failed to get mtr page %d != %d.\n", - get_pg_cnt, all_pg_cnt); - ret = -ENOBUFS; - goto err_alloc_page_list; } - /* write buffer's dma address to BA table */ - ret = hns_roce_mtr_map(hr_dev, mtr, pages, all_pg_cnt); + ret = mtr_alloc_bufs(hr_dev, mtr, buf_attr, udata, user_addr); if (ret) { - ibdev_err(ibdev, "failed to map mtr pages, ret = %d.\n", ret); - goto err_alloc_page_list; + ibdev_err(ibdev, "failed to alloc mtr bufs, ret = %d.\n", ret); + goto err_alloc_mtt; } - /* drop tmp array */ - kvfree(pages); - return 0; -err_alloc_page_list: - kvfree(pages); -err_alloc_hem_list: - hns_roce_hem_list_release(hr_dev, &mtr->hem_list); -err_alloc_bufs: + /* Write buffer's dma address to MTT */ + ret = mtr_map_bufs(hr_dev, mtr, buf_page_cnt, buf_page_shift); + if (ret) + ibdev_err(ibdev, "failed to map mtr bufs, ret = %d.\n", ret); + else + return 0; + mtr_free_bufs(hr_dev, mtr); +err_alloc_mtt: + mtr_free_mtt(hr_dev, mtr); return ret; } diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c index 1116371adf74..004aca9086ab 100644 --- a/drivers/infiniband/hw/hns/hns_roce_qp.c +++ b/drivers/infiniband/hw/hns/hns_roce_qp.c @@ -413,9 +413,32 @@ static void free_qpn(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp) mutex_unlock(&hr_dev->qp_table.bank_mutex); } +static u32 proc_rq_sge(struct hns_roce_dev *dev, struct hns_roce_qp *hr_qp, + bool user) +{ + u32 max_sge = dev->caps.max_rq_sg; + + if (dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) + return max_sge; + + /* Reserve SGEs only for HIP08 in kernel; The userspace driver will + * calculate number of max_sge with reserved SGEs when allocating wqe + * buf, so there is no need to do this again in kernel. But the number + * may exceed the capacity of SGEs recorded in the firmware, so the + * kernel driver should just adapt the value accordingly. + */ + if (user) + max_sge = roundup_pow_of_two(max_sge + 1); + else + hr_qp->rq.rsv_sge = 1; + + return max_sge; +} + static int set_rq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, - struct hns_roce_qp *hr_qp, int has_rq) + struct hns_roce_qp *hr_qp, int has_rq, bool user) { + u32 max_sge = proc_rq_sge(hr_dev, hr_qp, user); u32 cnt; /* If srq exist, set zero for relative number of rq */ @@ -431,8 +454,9 @@ static int set_rq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, /* Check the validity of QP support capacity */ if (!cap->max_recv_wr || cap->max_recv_wr > hr_dev->caps.max_wqes || - cap->max_recv_sge > hr_dev->caps.max_rq_sg) { - ibdev_err(&hr_dev->ib_dev, "RQ config error, depth=%u, sge=%d\n", + cap->max_recv_sge > max_sge) { + ibdev_err(&hr_dev->ib_dev, + "RQ config error, depth = %u, sge = %u\n", cap->max_recv_wr, cap->max_recv_sge); return -EINVAL; } @@ -444,7 +468,8 @@ static int set_rq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, return -EINVAL; } - hr_qp->rq.max_gs = roundup_pow_of_two(max(1U, cap->max_recv_sge)); + hr_qp->rq.max_gs = roundup_pow_of_two(max(1U, cap->max_recv_sge) + + hr_qp->rq.rsv_sge); if (hr_dev->caps.max_rq_sg <= HNS_ROCE_SGE_IN_WQE) hr_qp->rq.wqe_shift = ilog2(hr_dev->caps.max_rq_desc_sz); @@ -459,7 +484,7 @@ static int set_rq_size(struct hns_roce_dev *hr_dev, struct ib_qp_cap *cap, hr_qp->rq_inl_buf.wqe_cnt = 0; cap->max_recv_wr = cnt; - cap->max_recv_sge = hr_qp->rq.max_gs; + cap->max_recv_sge = hr_qp->rq.max_gs - hr_qp->rq.rsv_sge; return 0; } @@ -599,7 +624,6 @@ static int set_wqe_buf_attr(struct hns_roce_dev *hr_dev, return -EINVAL; buf_attr->page_shift = HNS_HW_PAGE_SHIFT + hr_dev->caps.mtt_buf_pg_sz; - buf_attr->fixed_page = true; buf_attr->region_count = idx; return 0; @@ -919,7 +943,7 @@ static int set_qp_param(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp, hr_qp->sq_signal_bits = IB_SIGNAL_REQ_WR; ret = set_rq_size(hr_dev, &init_attr->cap, hr_qp, - hns_roce_qp_has_rq(init_attr)); + hns_roce_qp_has_rq(init_attr), !!udata); if (ret) { ibdev_err(ibdev, "failed to set user RQ size, ret = %d.\n", ret); diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c index c4ae57e4173a..d5a6de0e7095 100644 --- a/drivers/infiniband/hw/hns/hns_roce_srq.c +++ b/drivers/infiniband/hw/hns/hns_roce_srq.c @@ -3,6 +3,7 @@ * Copyright (c) 2018 Hisilicon Limited. */ +#include <linux/pci.h> #include <rdma/ib_umem.h> #include "hns_roce_device.h" #include "hns_roce_cmd.h" @@ -76,40 +77,16 @@ static int hns_roce_hw_destroy_srq(struct hns_roce_dev *dev, HNS_ROCE_CMD_TIMEOUT_MSECS); } -static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, - u32 pdn, u32 cqn, u16 xrcd, u64 db_rec_addr) +static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) { struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_cmd_mailbox *mailbox; - u64 mtts_wqe[MTT_MIN_COUNT] = { 0 }; - u64 mtts_idx[MTT_MIN_COUNT] = { 0 }; - dma_addr_t dma_handle_wqe = 0; - dma_addr_t dma_handle_idx = 0; int ret; - /* Get the physical address of srq buf */ - ret = hns_roce_mtr_find(hr_dev, &srq->buf_mtr, 0, mtts_wqe, - ARRAY_SIZE(mtts_wqe), &dma_handle_wqe); - if (ret < 1) { - ibdev_err(ibdev, "failed to find mtr for SRQ WQE, ret = %d.\n", - ret); - return -ENOBUFS; - } - - /* Get physical address of idx que buf */ - ret = hns_roce_mtr_find(hr_dev, &srq->idx_que.mtr, 0, mtts_idx, - ARRAY_SIZE(mtts_idx), &dma_handle_idx); - if (ret < 1) { - ibdev_err(ibdev, "failed to find mtr for SRQ idx, ret = %d.\n", - ret); - return -ENOBUFS; - } - ret = hns_roce_bitmap_alloc(&srq_table->bitmap, &srq->srqn); if (ret) { - ibdev_err(ibdev, - "failed to alloc SRQ number, ret = %d.\n", ret); + ibdev_err(ibdev, "failed to alloc SRQ number.\n"); return -ENOMEM; } @@ -127,34 +104,36 @@ static int alloc_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); if (IS_ERR_OR_NULL(mailbox)) { - ret = -ENOMEM; ibdev_err(ibdev, "failed to alloc mailbox for SRQC.\n"); + ret = -ENOMEM; goto err_xa; } - hr_dev->hw->write_srqc(hr_dev, srq, pdn, xrcd, cqn, mailbox->buf, - mtts_wqe, mtts_idx, dma_handle_wqe, - dma_handle_idx); + ret = hr_dev->hw->write_srqc(srq, mailbox->buf); + if (ret) { + ibdev_err(ibdev, "failed to write SRQC.\n"); + goto err_mbox; + } ret = hns_roce_hw_create_srq(hr_dev, mailbox, srq->srqn); - hns_roce_free_cmd_mailbox(hr_dev, mailbox); if (ret) { ibdev_err(ibdev, "failed to config SRQC, ret = %d.\n", ret); - goto err_xa; + goto err_mbox; } - atomic_set(&srq->refcount, 1); - init_completion(&srq->free); - return ret; + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + + return 0; +err_mbox: + hns_roce_free_cmd_mailbox(hr_dev, mailbox); err_xa: xa_erase(&srq_table->xa, srq->srqn); - err_put: hns_roce_table_put(hr_dev, &srq_table->table, srq->srqn); - err_out: hns_roce_bitmap_free(&srq_table->bitmap, srq->srqn, BITMAP_NO_RR); + return ret; } @@ -178,46 +157,13 @@ static void free_srqc(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) hns_roce_bitmap_free(&srq_table->bitmap, srq->srqn, BITMAP_NO_RR); } -static int alloc_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, - struct ib_udata *udata, unsigned long addr) -{ - struct ib_device *ibdev = &hr_dev->ib_dev; - struct hns_roce_buf_attr buf_attr = {}; - int err; - - srq->wqe_shift = ilog2(roundup_pow_of_two(max(HNS_ROCE_SGE_SIZE, - HNS_ROCE_SGE_SIZE * - srq->max_gs))); - - buf_attr.page_shift = hr_dev->caps.srqwqe_buf_pg_sz + HNS_HW_PAGE_SHIFT; - buf_attr.region[0].size = to_hr_hem_entries_size(srq->wqe_cnt, - srq->wqe_shift); - buf_attr.region[0].hopnum = hr_dev->caps.srqwqe_hop_num; - buf_attr.region_count = 1; - buf_attr.fixed_page = true; - - err = hns_roce_mtr_create(hr_dev, &srq->buf_mtr, &buf_attr, - hr_dev->caps.srqwqe_ba_pg_sz + - HNS_HW_PAGE_SHIFT, udata, addr); - if (err) - ibdev_err(ibdev, - "failed to alloc SRQ buf mtr, ret = %d.\n", err); - - return err; -} - -static void free_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) -{ - hns_roce_mtr_destroy(hr_dev, &srq->buf_mtr); -} - static int alloc_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, struct ib_udata *udata, unsigned long addr) { struct hns_roce_idx_que *idx_que = &srq->idx_que; struct ib_device *ibdev = &hr_dev->ib_dev; struct hns_roce_buf_attr buf_attr = {}; - int err; + int ret; srq->idx_que.entry_shift = ilog2(HNS_ROCE_IDX_QUE_ENTRY_SZ); @@ -226,31 +172,33 @@ static int alloc_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, srq->idx_que.entry_shift); buf_attr.region[0].hopnum = hr_dev->caps.idx_hop_num; buf_attr.region_count = 1; - buf_attr.fixed_page = true; - err = hns_roce_mtr_create(hr_dev, &idx_que->mtr, &buf_attr, + ret = hns_roce_mtr_create(hr_dev, &idx_que->mtr, &buf_attr, hr_dev->caps.idx_ba_pg_sz + HNS_HW_PAGE_SHIFT, udata, addr); - if (err) { + if (ret) { ibdev_err(ibdev, - "failed to alloc SRQ idx mtr, ret = %d.\n", err); - return err; + "failed to alloc SRQ idx mtr, ret = %d.\n", ret); + return ret; } if (!udata) { idx_que->bitmap = bitmap_zalloc(srq->wqe_cnt, GFP_KERNEL); if (!idx_que->bitmap) { ibdev_err(ibdev, "failed to alloc SRQ idx bitmap.\n"); - err = -ENOMEM; + ret = -ENOMEM; goto err_idx_mtr; } } + idx_que->head = 0; + idx_que->tail = 0; + return 0; err_idx_mtr: hns_roce_mtr_destroy(hr_dev, &idx_que->mtr); - return err; + return ret; } static void free_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) @@ -262,10 +210,42 @@ static void free_srq_idx(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) hns_roce_mtr_destroy(hr_dev, &idx_que->mtr); } +static int alloc_srq_wqe_buf(struct hns_roce_dev *hr_dev, + struct hns_roce_srq *srq, + struct ib_udata *udata, unsigned long addr) +{ + struct ib_device *ibdev = &hr_dev->ib_dev; + struct hns_roce_buf_attr buf_attr = {}; + int ret; + + srq->wqe_shift = ilog2(roundup_pow_of_two(max(HNS_ROCE_SGE_SIZE, + HNS_ROCE_SGE_SIZE * + srq->max_gs))); + + buf_attr.page_shift = hr_dev->caps.srqwqe_buf_pg_sz + HNS_HW_PAGE_SHIFT; + buf_attr.region[0].size = to_hr_hem_entries_size(srq->wqe_cnt, + srq->wqe_shift); + buf_attr.region[0].hopnum = hr_dev->caps.srqwqe_hop_num; + buf_attr.region_count = 1; + + ret = hns_roce_mtr_create(hr_dev, &srq->buf_mtr, &buf_attr, + hr_dev->caps.srqwqe_ba_pg_sz + + HNS_HW_PAGE_SHIFT, udata, addr); + if (ret) + ibdev_err(ibdev, + "failed to alloc SRQ buf mtr, ret = %d.\n", ret); + + return ret; +} + +static void free_srq_wqe_buf(struct hns_roce_dev *hr_dev, + struct hns_roce_srq *srq) +{ + hns_roce_mtr_destroy(hr_dev, &srq->buf_mtr); +} + static int alloc_srq_wrid(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) { - srq->head = 0; - srq->tail = srq->wqe_cnt - 1; srq->wrid = kvmalloc_array(srq->wqe_cnt, sizeof(u64), GFP_KERNEL); if (!srq->wrid) return -ENOMEM; @@ -279,96 +259,171 @@ static void free_srq_wrid(struct hns_roce_srq *srq) srq->wrid = NULL; } -int hns_roce_create_srq(struct ib_srq *ib_srq, - struct ib_srq_init_attr *init_attr, - struct ib_udata *udata) +static u32 proc_srq_sge(struct hns_roce_dev *dev, struct hns_roce_srq *hr_srq, + bool user) +{ + u32 max_sge = dev->caps.max_srq_sges; + + if (dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) + return max_sge; + + /* Reserve SGEs only for HIP08 in kernel; The userspace driver will + * calculate number of max_sge with reserved SGEs when allocating wqe + * buf, so there is no need to do this again in kernel. But the number + * may exceed the capacity of SGEs recorded in the firmware, so the + * kernel driver should just adapt the value accordingly. + */ + if (user) + max_sge = roundup_pow_of_two(max_sge + 1); + else + hr_srq->rsv_sge = 1; + + return max_sge; +} + +static int set_srq_basic_param(struct hns_roce_srq *srq, + struct ib_srq_init_attr *init_attr, + struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); + struct ib_srq_attr *attr = &init_attr->attr; + u32 max_sge; + + max_sge = proc_srq_sge(hr_dev, srq, !!udata); + if (attr->max_wr > hr_dev->caps.max_srq_wrs || + attr->max_sge > max_sge) { + ibdev_err(&hr_dev->ib_dev, + "invalid SRQ attr, depth = %u, sge = %u.\n", + attr->max_wr, attr->max_sge); + return -EINVAL; + } + + attr->max_wr = max_t(u32, attr->max_wr, HNS_ROCE_MIN_SRQ_WQE_NUM); + srq->wqe_cnt = roundup_pow_of_two(attr->max_wr); + srq->max_gs = roundup_pow_of_two(attr->max_sge + srq->rsv_sge); + + attr->max_wr = srq->wqe_cnt; + attr->max_sge = srq->max_gs - srq->rsv_sge; + attr->srq_limit = 0; + + return 0; +} + +static void set_srq_ext_param(struct hns_roce_srq *srq, + struct ib_srq_init_attr *init_attr) +{ + srq->cqn = ib_srq_has_cq(init_attr->srq_type) ? + to_hr_cq(init_attr->ext.cq)->cqn : 0; +} + +static int set_srq_param(struct hns_roce_srq *srq, + struct ib_srq_init_attr *init_attr, + struct ib_udata *udata) { - struct hns_roce_dev *hr_dev = to_hr_dev(ib_srq->device); - struct hns_roce_ib_create_srq_resp resp = {}; - struct hns_roce_srq *srq = to_hr_srq(ib_srq); - struct ib_device *ibdev = &hr_dev->ib_dev; - struct hns_roce_ib_create_srq ucmd = {}; int ret; - u32 cqn; - if (init_attr->srq_type != IB_SRQT_BASIC && - init_attr->srq_type != IB_SRQT_XRC) - return -EOPNOTSUPP; + ret = set_srq_basic_param(srq, init_attr, udata); + if (ret) + return ret; - /* Check the actual SRQ wqe and SRQ sge num */ - if (init_attr->attr.max_wr >= hr_dev->caps.max_srq_wrs || - init_attr->attr.max_sge > hr_dev->caps.max_srq_sges) - return -EINVAL; + set_srq_ext_param(srq, init_attr); - mutex_init(&srq->mutex); - spin_lock_init(&srq->lock); + return 0; +} - srq->wqe_cnt = roundup_pow_of_two(init_attr->attr.max_wr + 1); - srq->max_gs = init_attr->attr.max_sge; +static int alloc_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq, + struct ib_udata *udata) +{ + struct hns_roce_ib_create_srq ucmd = {}; + int ret; if (udata) { ret = ib_copy_from_udata(&ucmd, udata, min(udata->inlen, sizeof(ucmd))); if (ret) { - ibdev_err(ibdev, "failed to copy SRQ udata, ret = %d.\n", + ibdev_err(&hr_dev->ib_dev, + "failed to copy SRQ udata, ret = %d.\n", ret); return ret; } } - ret = alloc_srq_buf(hr_dev, srq, udata, ucmd.buf_addr); - if (ret) { - ibdev_err(ibdev, - "failed to alloc SRQ buffer, ret = %d.\n", ret); + ret = alloc_srq_idx(hr_dev, srq, udata, ucmd.que_addr); + if (ret) return ret; - } - ret = alloc_srq_idx(hr_dev, srq, udata, ucmd.que_addr); - if (ret) { - ibdev_err(ibdev, "failed to alloc SRQ idx, ret = %d.\n", ret); - goto err_buf_alloc; - } + ret = alloc_srq_wqe_buf(hr_dev, srq, udata, ucmd.buf_addr); + if (ret) + goto err_idx; if (!udata) { ret = alloc_srq_wrid(hr_dev, srq); - if (ret) { - ibdev_err(ibdev, "failed to alloc SRQ wrid, ret = %d.\n", - ret); - goto err_idx_alloc; - } + if (ret) + goto err_wqe_buf; } - cqn = ib_srq_has_cq(init_attr->srq_type) ? - to_hr_cq(init_attr->ext.cq)->cqn : 0; - srq->db_reg_l = hr_dev->reg_base + SRQ_DB_REG; + return 0; - ret = alloc_srqc(hr_dev, srq, to_hr_pd(ib_srq->pd)->pdn, cqn, 0, 0); - if (ret) { - ibdev_err(ibdev, - "failed to alloc SRQ context, ret = %d.\n", ret); - goto err_wrid_alloc; - } +err_wqe_buf: + free_srq_wqe_buf(hr_dev, srq); +err_idx: + free_srq_idx(hr_dev, srq); - srq->event = hns_roce_ib_srq_event; - resp.srqn = srq->srqn; + return ret; +} + +static void free_srq_buf(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) +{ + free_srq_wrid(srq); + free_srq_wqe_buf(hr_dev, srq); + free_srq_idx(hr_dev, srq); +} + +int hns_roce_create_srq(struct ib_srq *ib_srq, + struct ib_srq_init_attr *init_attr, + struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ib_srq->device); + struct hns_roce_ib_create_srq_resp resp = {}; + struct hns_roce_srq *srq = to_hr_srq(ib_srq); + int ret; + + mutex_init(&srq->mutex); + spin_lock_init(&srq->lock); + + ret = set_srq_param(srq, init_attr, udata); + if (ret) + return ret; + + ret = alloc_srq_buf(hr_dev, srq, udata); + if (ret) + return ret; + + ret = alloc_srqc(hr_dev, srq); + if (ret) + goto err_srq_buf; if (udata) { - ret = ib_copy_to_udata(udata, &resp, - min(udata->outlen, sizeof(resp))); - if (ret) - goto err_srqc_alloc; + resp.srqn = srq->srqn; + if (ib_copy_to_udata(udata, &resp, + min(udata->outlen, sizeof(resp)))) { + ret = -EFAULT; + goto err_srqc; + } } + srq->db_reg_l = hr_dev->reg_base + SRQ_DB_REG; + srq->event = hns_roce_ib_srq_event; + atomic_set(&srq->refcount, 1); + init_completion(&srq->free); + return 0; -err_srqc_alloc: +err_srqc: free_srqc(hr_dev, srq); -err_wrid_alloc: - free_srq_wrid(srq); -err_idx_alloc: - free_srq_idx(hr_dev, srq); -err_buf_alloc: +err_srq_buf: free_srq_buf(hr_dev, srq); + return ret; } @@ -378,8 +433,6 @@ int hns_roce_destroy_srq(struct ib_srq *ibsrq, struct ib_udata *udata) struct hns_roce_srq *srq = to_hr_srq(ibsrq); free_srqc(hr_dev, srq); - free_srq_idx(hr_dev, srq); - free_srq_wrid(srq); free_srq_buf(hr_dev, srq); return 0; } diff --git a/drivers/infiniband/hw/i40iw/i40iw_cm.c b/drivers/infiniband/hw/i40iw/i40iw_cm.c index 9acc0ecc9a43..ac65c8237b2e 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_cm.c +++ b/drivers/infiniband/hw/i40iw/i40iw_cm.c @@ -70,7 +70,7 @@ static void i40iw_disconnect_worker(struct work_struct *work); /** * i40iw_free_sqbuf - put back puda buffer if refcount = 0 * @vsi: pointer to vsi structure - * @buf: puda buffer to free + * @bufp: puda buffer to free */ void i40iw_free_sqbuf(struct i40iw_sc_vsi *vsi, void *bufp) { @@ -729,6 +729,7 @@ static int i40iw_handle_tcp_options(struct i40iw_cm_node *cm_node, /** * i40iw_build_mpa_v1 - build a MPA V1 frame * @cm_node: connection's node + * @start_addr: MPA frame start address * @mpa_key: to do read0 or write0 */ static void i40iw_build_mpa_v1(struct i40iw_cm_node *cm_node, @@ -1040,7 +1041,7 @@ negotiate_done: /** * i40iw_schedule_cm_timer - * @@cm_node: connection's node + * @cm_node: connection's node * @sqbuf: buffer to send * @type: if it is send or close * @send_retrans: if rexmits to be done @@ -1205,7 +1206,7 @@ static void i40iw_build_timer_list(struct list_head *timer_list, /** * i40iw_cm_timer_tick - system's timer expired callback - * @pass: Pointing to cm_core + * @t: Timer instance to fetch the cm_core pointer from */ static void i40iw_cm_timer_tick(struct timer_list *t) { @@ -1463,6 +1464,7 @@ struct i40iw_cm_node *i40iw_find_node(struct i40iw_cm_core *cm_core, * @cm_core: cm's core * @dst_port: listener tcp port num * @dst_addr: listener ip addr + * @vlan_id: vlan id for the given address * @listener_state: state to match with listen node's */ static struct i40iw_cm_listener *i40iw_find_listener( @@ -1521,7 +1523,7 @@ static void i40iw_add_hte_node(struct i40iw_cm_core *cm_core, /** * i40iw_find_port - find port that matches reference port * @hte: ptr to accelerated or non-accelerated list - * @accelerated_list: flag for accelerated vs non-accelerated list + * @port: port number to locate */ static bool i40iw_find_port(struct list_head *hte, u16 port) { @@ -1834,6 +1836,7 @@ exit: /** * i40iw_dec_refcnt_listen - delete listener and associated cm nodes * @cm_core: cm's core + * @listener: passive connection's listener * @free_hanging_nodes: to free associated cm_nodes * @apbvt_del: flag to delete the apbvt */ @@ -2029,7 +2032,7 @@ static int i40iw_addr_resolve_neigh(struct i40iw_device *iwdev, return rc; } -/** +/* * i40iw_get_dst_ipv6 */ static struct dst_entry *i40iw_get_dst_ipv6(struct sockaddr_in6 *src_addr, @@ -2051,7 +2054,8 @@ static struct dst_entry *i40iw_get_dst_ipv6(struct sockaddr_in6 *src_addr, /** * i40iw_addr_resolve_neigh_ipv6 - resolve neighbor ipv6 address * @iwdev: iwarp device structure - * @dst_ip: remote ip address + * @src: source ip address + * @dest: remote ip address * @arpindex: if there is an arp entry */ static int i40iw_addr_resolve_neigh_ipv6(struct i40iw_device *iwdev, @@ -3004,7 +3008,7 @@ static struct i40iw_cm_node *i40iw_create_cm_node( /** * i40iw_cm_reject - reject and teardown a connection * @cm_node: connection's node - * @pdate: ptr to private data for reject + * @pdata: ptr to private data for reject * @plen: size of private data */ static int i40iw_cm_reject(struct i40iw_cm_node *cm_node, const void *pdata, u8 plen) @@ -4302,7 +4306,7 @@ set_qhash: * i40iw_cm_teardown_connections - teardown QPs * @iwdev: device pointer * @ipaddr: Pointer to IPv4 or IPv6 address - * @ipv4: flag indicating IPv4 when true + * @nfo: cm info node * @disconnect_all: flag indicating disconnect all QPs * teardown QPs where source or destination addr matches ip addr */ @@ -4358,6 +4362,7 @@ void i40iw_cm_teardown_connections(struct i40iw_device *iwdev, u32 *ipaddr, /** * i40iw_ifdown_notify - process an ifdown on an interface * @iwdev: device pointer + * @netdev: network interface device structure * @ipaddr: Pointer to IPv4 or IPv6 address * @ipv4: flag indicating IPv4 when true * @ifup: flag indicating interface up when true diff --git a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c index c943d491b72b..eaea5d545eb8 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_ctrl.c +++ b/drivers/infiniband/hw/i40iw/i40iw_ctrl.c @@ -181,7 +181,7 @@ static enum i40iw_status_code i40iw_sc_parse_fpm_commit_buf( * i40iw_sc_decode_fpm_query() - Decode a 64 bit value into max count and size * @buf: ptr to fpm query buffer * @buf_idx: index into buf - * @info: ptr to i40iw_hmc_obj_info struct + * @obj_info: ptr to i40iw_hmc_obj_info struct * @rsrc_idx: resource index into info * * Decode a 64 bit value from fpm query buffer into max count and size @@ -205,7 +205,7 @@ static u64 i40iw_sc_decode_fpm_query(u64 *buf, /** * i40iw_sc_parse_fpm_query_buf() - parses fpm query buffer * @buf: ptr to fpm query buffer - * @info: ptr to i40iw_hmc_obj_info struct + * @hmc_info: ptr to i40iw_hmc_obj_info struct * @hmc_fpm_misc: ptr to fpm data * * parses fpm query buffer and copy max_cnt and @@ -775,7 +775,7 @@ static enum i40iw_status_code i40iw_sc_ccq_get_cqe_info( * i40iw_sc_poll_for_cqp_op_done - Waits for last write to complete in CQP SQ * @cqp: struct for cqp hw * @op_code: cqp opcode for completion - * @info: completion q entry to return + * @compl_info: completion q entry to return */ static enum i40iw_status_code i40iw_sc_poll_for_cqp_op_done( struct i40iw_sc_cqp *cqp, @@ -933,7 +933,7 @@ static enum i40iw_status_code i40iw_sc_commit_fpm_values_done(struct i40iw_sc_cq * @cqp: struct for cqp hw * @scratch: u64 saved to be used during cqp completion * @hmc_fn_id: hmc function id - * @commit_fpm_mem; Memory for fpm values + * @commit_fpm_mem: Memory for fpm values * @post_sq: flag for cqp db to ring * @wait_type: poll ccq or cqp registers for cqp completion */ @@ -1026,7 +1026,7 @@ i40iw_sc_query_rdma_features(struct i40iw_sc_cqp *cqp, /** * i40iw_get_rdma_features - get RDMA features - * @dev - sc device struct + * @dev: sc device struct */ enum i40iw_status_code i40iw_get_rdma_features(struct i40iw_sc_dev *dev) { @@ -1456,7 +1456,7 @@ static enum i40iw_status_code i40iw_sc_add_local_mac_ipaddr_entry( * @cqp: struct for cqp hw * @scratch: u64 saved to be used during cqp completion * @entry_idx: index of mac entry - * @ ignore_ref_count: to force mac adde delete + * @ignore_ref_count: to force mac adde delete * @post_sq: flag for cqp db to ring */ static enum i40iw_status_code i40iw_sc_del_local_mac_ipaddr_entry( @@ -2304,7 +2304,7 @@ static enum i40iw_status_code i40iw_sc_cq_destroy(struct i40iw_sc_cq *cq, * i40iw_sc_cq_modify - modify a Completion Queue * @cq: cq struct * @info: modification info struct - * @scratch: + * @scratch: u64 saved to be used during cqp completion * @post_sq: flag to post to sq */ static enum i40iw_status_code i40iw_sc_cq_modify(struct i40iw_sc_cq *cq, @@ -3673,7 +3673,7 @@ static enum i40iw_status_code i40iw_sc_configure_iw_fpm(struct i40iw_sc_dev *dev /** * cqp_sds_wqe_fill - fill cqp wqe doe sd * @cqp: struct for cqp hw - * @info; sd info for wqe + * @info: sd info for wqe * @scratch: u64 saved to be used during cqp completion */ static enum i40iw_status_code cqp_sds_wqe_fill(struct i40iw_sc_cqp *cqp, @@ -4884,7 +4884,7 @@ void i40iw_hw_stats_init(struct i40iw_vsi_pestat *stats, u8 fcn_idx, bool is_pf) /** * i40iw_hw_stats_read_32 - Read 32-bit HW stats counters and accommodates for roll-overs. - * @stat: pestat struct + * @stats: pestat struct * @index: index in HW stats table which contains offset reg-addr * @value: hw stats value */ diff --git a/drivers/infiniband/hw/i40iw/i40iw_hmc.c b/drivers/infiniband/hw/i40iw/i40iw_hmc.c index 5484cbf55f0f..8bd72af9e099 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_hmc.c +++ b/drivers/infiniband/hw/i40iw/i40iw_hmc.c @@ -46,7 +46,7 @@ * i40iw_find_sd_index_limit - finds segment descriptor index limit * @hmc_info: pointer to the HMC configuration information structure * @type: type of HMC resources we're searching - * @index: starting index for the object + * @idx: starting index for the object * @cnt: number of objects we're trying to create * @sd_idx: pointer to return index of the segment descriptor in question * @sd_limit: pointer to return the maximum number of segment descriptors @@ -78,7 +78,7 @@ static inline void i40iw_find_sd_index_limit(struct i40iw_hmc_info *hmc_info, * @type: HMC resource type we're examining * @idx: starting index for the object * @cnt: number of objects we're trying to create - * @pd_index: pointer to return page descriptor index + * @pd_idx: pointer to return page descriptor index * @pd_limit: pointer to return page descriptor index limit * * Calculates the page descriptor index and index limit for the resource diff --git a/drivers/infiniband/hw/i40iw/i40iw_hw.c b/drivers/infiniband/hw/i40iw/i40iw_hw.c index 56fdc161f6f8..d167ac10c751 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_hw.c +++ b/drivers/infiniband/hw/i40iw/i40iw_hw.c @@ -165,7 +165,7 @@ static void i40iw_cqp_ce_handler(struct i40iw_device *iwdev, struct i40iw_sc_cq /** * i40iw_iwarp_ce_handler - handle iwarp completions * @iwdev: iwarp device - * @iwcp: iwarp cq receiving event + * @iwcq: iwarp cq receiving event */ static void i40iw_iwarp_ce_handler(struct i40iw_device *iwdev, struct i40iw_sc_cq *iwcq) @@ -519,6 +519,7 @@ enum i40iw_status_code i40iw_manage_apbvt(struct i40iw_device *iwdev, * @iwdev: iwarp device * @mac_addr: mac address ptr * @ip_addr: ip addr for arp cache + * @ipv4: flag indicating IPv4 when true * @action: add, delete or modify */ void i40iw_manage_arp_cache(struct i40iw_device *iwdev, @@ -581,7 +582,6 @@ static void i40iw_send_syn_cqp_callback(struct i40iw_cqp_request *cqp_request, u * @mtype: type of qhash * @cmnode: cmnode associated with connection * @wait: wait for completion - * @user_pri:user pri of the connection */ enum i40iw_status_code i40iw_manage_qhash(struct i40iw_device *iwdev, struct i40iw_cm_info *cminfo, diff --git a/drivers/infiniband/hw/i40iw/i40iw_main.c b/drivers/infiniband/hw/i40iw/i40iw_main.c index 584932d3cc44..ab4cb11950dc 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_main.c +++ b/drivers/infiniband/hw/i40iw/i40iw_main.c @@ -186,7 +186,7 @@ static void i40iw_enable_intr(struct i40iw_sc_dev *dev, u32 msix_id) /** * i40iw_dpc - tasklet for aeq and ceq 0 - * @data: iwarp device + * @t: Timer context to fetch pointer to iwarp device */ static void i40iw_dpc(struct tasklet_struct *t) { @@ -200,7 +200,7 @@ static void i40iw_dpc(struct tasklet_struct *t) /** * i40iw_ceq_dpc - dpc handler for CEQ - * @data: data points to CEQ + * @t: Timer context to fetch pointer to CEQ data */ static void i40iw_ceq_dpc(struct tasklet_struct *t) { @@ -227,7 +227,7 @@ static irqreturn_t i40iw_irq_handler(int irq, void *data) /** * i40iw_destroy_cqp - destroy control qp * @iwdev: iwarp device - * @create_done: 1 if cqp create poll was success + * @free_hwcqp: 1 if CQP should be destroyed * * Issue destroy cqp request and * free the resources associated with the cqp @@ -253,7 +253,7 @@ static void i40iw_destroy_cqp(struct i40iw_device *iwdev, bool free_hwcqp) /** * i40iw_disable_irqs - disable device interrupts * @dev: hardware control device structure - * @msic_vec: msix vector to disable irq + * @msix_vec: msix vector to disable irq * @dev_id: parameter to pass to free_irq (used during irq setup) * * The function is called when destroying aeq/ceq @@ -394,8 +394,9 @@ static enum i40iw_hmc_rsrc_type iw_hmc_obj_types[] = { /** * i40iw_close_hmc_objects_type - delete hmc objects of a given type - * @iwdev: iwarp device + * @dev: iwarp device * @obj_type: the hmc object type to be deleted + * @hmc_info: pointer to the HMC configuration information * @is_pf: true if the function is PF otherwise false * @reset: true if called before reset */ @@ -437,6 +438,7 @@ static void i40iw_del_hmc_objects(struct i40iw_sc_dev *dev, /** * i40iw_ceq_handler - interrupt handler for ceq + * @irq: interrupt request number * @data: ceq pointer */ static irqreturn_t i40iw_ceq_handler(int irq, void *data) @@ -1777,6 +1779,7 @@ static void i40iw_l2param_change(struct i40e_info *ldev, struct i40e_client *cli /** * i40iw_close - client interface operation close for iwarp/uda device * @ldev: lan device information + * @reset: true if called before reset * @client: client to close * * Called by the lan driver during the processing of client unregister diff --git a/drivers/infiniband/hw/i40iw/i40iw_pble.c b/drivers/infiniband/hw/i40iw/i40iw_pble.c index 5f97643e22e5..53e5cd1a2bd6 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_pble.c +++ b/drivers/infiniband/hw/i40iw/i40iw_pble.c @@ -54,6 +54,7 @@ static void i40iw_free_vmalloc_mem(struct i40iw_hw *hw, struct i40iw_chunk *chun /** * i40iw_destroy_pble_pool - destroy pool during module unload + * @dev: i40iw_sc_dev struct * @pble_rsrc: pble resources */ void i40iw_destroy_pble_pool(struct i40iw_sc_dev *dev, struct i40iw_hmc_pble_rsrc *pble_rsrc) @@ -112,8 +113,8 @@ enum i40iw_status_code i40iw_hmc_init_pble(struct i40iw_sc_dev *dev, /** * get_sd_pd_idx - Returns sd index, pd index and rel_pd_idx from fpm address - * @ pble_rsrc: structure containing fpm address - * @ idx: where to return indexes + * @pble_rsrc: structure containing fpm address + * @idx: where to return indexes */ static inline void get_sd_pd_idx(struct i40iw_hmc_pble_rsrc *pble_rsrc, struct sd_pd_idx *idx) diff --git a/drivers/infiniband/hw/i40iw/i40iw_puda.c b/drivers/infiniband/hw/i40iw/i40iw_puda.c index 924be4b03c9a..d1c8cc0a6236 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_puda.c +++ b/drivers/infiniband/hw/i40iw/i40iw_puda.c @@ -511,7 +511,8 @@ static void i40iw_puda_qp_setctx(struct i40iw_puda_rsrc *rsrc) /** * i40iw_puda_qp_wqe - setup wqe for qp create - * @rsrc: resource for qp + * @dev: iwarp device + * @qp: resource for qp */ static enum i40iw_status_code i40iw_puda_qp_wqe(struct i40iw_sc_dev *dev, struct i40iw_sc_qp *qp) { @@ -623,7 +624,8 @@ static enum i40iw_status_code i40iw_puda_qp_create(struct i40iw_puda_rsrc *rsrc) /** * i40iw_puda_cq_wqe - setup wqe for cq create - * @rsrc: resource for cq + * @dev: iwarp device + * @cq: cq to setup */ static enum i40iw_status_code i40iw_puda_cq_wqe(struct i40iw_sc_dev *dev, struct i40iw_sc_cq *cq) { @@ -782,7 +784,7 @@ static void i40iw_puda_free_cq(struct i40iw_puda_rsrc *rsrc) /** * i40iw_puda_dele_resources - delete all resources during close - * @dev: iwarp device + * @vsi: pointer to vsi structure * @type: type of resource to dele * @reset: true if reset chip */ @@ -876,7 +878,7 @@ static enum i40iw_status_code i40iw_puda_allocbufs(struct i40iw_puda_rsrc *rsrc, /** * i40iw_puda_create_rsrc - create resouce (ilq or ieq) - * @dev: iwarp device + * @vsi: pointer to vsi structure * @info: resource information */ enum i40iw_status_code i40iw_puda_create_rsrc(struct i40iw_sc_vsi *vsi, @@ -1121,6 +1123,7 @@ static void i40iw_ieq_compl_pfpdu(struct i40iw_puda_rsrc *ieq, /** * i40iw_ieq_create_pbufl - create buffer list for single fpdu + * @pfpdu: partial management per user qp * @rxlist: resource list for receive ieq buffes * @pbufl: temp. list for buffers for fpddu * @buf: first receive buffer @@ -1434,7 +1437,7 @@ static void i40iw_ieq_handle_exception(struct i40iw_puda_rsrc *ieq, /** * i40iw_ieq_receive - received exception buffer - * @dev: iwarp device + * @vsi: pointer to vsi structure * @buf: exception buffer received */ static void i40iw_ieq_receive(struct i40iw_sc_vsi *vsi, diff --git a/drivers/infiniband/hw/i40iw/i40iw_uk.c b/drivers/infiniband/hw/i40iw/i40iw_uk.c index c3633c9944db..f521be16bf31 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_uk.c +++ b/drivers/infiniband/hw/i40iw/i40iw_uk.c @@ -119,6 +119,8 @@ void i40iw_qp_post_wr(struct i40iw_qp_uk *qp) * @qp: hw qp ptr * @wqe_idx: return wqe index * @wqe_size: size of sq wqe + * @total_size: work request length + * @wr_id: work request id */ u64 *i40iw_qp_get_next_send_wqe(struct i40iw_qp_uk *qp, u32 *wqe_idx, @@ -717,7 +719,6 @@ static enum i40iw_status_code i40iw_cq_post_entries(struct i40iw_cq_uk *cq, * i40iw_cq_poll_completion - get cq completion info * @cq: hw cq * @info: cq poll information returned - * @post_cq: update cq tail */ static enum i40iw_status_code i40iw_cq_poll_completion(struct i40iw_cq_uk *cq, struct i40iw_cq_poll_info *info) @@ -1051,7 +1052,7 @@ void i40iw_device_init_uk(struct i40iw_dev_uk *dev) /** * i40iw_clean_cq - clean cq entries - * @ queue completion context + * @queue: completion context * @cq: cq to clean */ void i40iw_clean_cq(void *queue, struct i40iw_cq_uk *cq) diff --git a/drivers/infiniband/hw/i40iw/i40iw_utils.c b/drivers/infiniband/hw/i40iw/i40iw_utils.c index 644f8c641aa0..76f052b12c14 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_utils.c +++ b/drivers/infiniband/hw/i40iw/i40iw_utils.c @@ -55,6 +55,7 @@ * i40iw_arp_table - manage arp table * @iwdev: iwarp device * @ip_addr: ip address for device + * @ipv4: flag indicating IPv4 when true * @mac_addr: mac address ptr * @action: modify, delete or add */ @@ -138,7 +139,7 @@ inline u32 i40iw_rd32(struct i40iw_hw *hw, u32 reg) /** * i40iw_inetaddr_event - system notifier for ipv4 addr events - * @notfier: not used + * @notifier: not used * @event: event for notifier * @ptr: if address */ @@ -214,7 +215,7 @@ int i40iw_inetaddr_event(struct notifier_block *notifier, /** * i40iw_inet6addr_event - system notifier for ipv6 addr events - * @notfier: not used + * @notifier: not used * @event: event for notifier * @ptr: if address */ @@ -265,7 +266,7 @@ int i40iw_inet6addr_event(struct notifier_block *notifier, /** * i40iw_net_event - system notifier for netevents - * @notfier: not used + * @notifier: not used * @event: event for notifier * @ptr: neighbor */ @@ -310,7 +311,7 @@ int i40iw_net_event(struct notifier_block *notifier, unsigned long event, void * /** * i40iw_netdevice_event - system notifier for netdev events - * @notfier: not used + * @notifier: not used * @event: event for notifier * @ptr: netdev */ @@ -652,6 +653,7 @@ struct ib_qp *i40iw_get_qp(struct ib_device *device, int qpn) * i40iw_debug_buf - print debug msg and buffer is mask set * @dev: hardware control device structure * @mask: mask to compare if to print debug buffer + * @desc: identifying string * @buf: points buffer addr * @size: saize of buffer to print */ @@ -784,7 +786,7 @@ enum i40iw_status_code i40iw_free_virt_mem(struct i40iw_hw *hw, /** * i40iw_cqp_sds_cmd - create cqp command for sd * @dev: hardware control device structure - * @sd_info: information for sd cqp + * @sdinfo: information for sd cqp * */ enum i40iw_status_code i40iw_cqp_sds_cmd(struct i40iw_sc_dev *dev, @@ -889,7 +891,7 @@ void i40iw_terminate_done(struct i40iw_sc_qp *qp, int timeout_occurred) /** * i40iw_terminate_imeout - timeout happened - * @context: points to iwarp qp + * @t: points to iwarp qp */ static void i40iw_terminate_timeout(struct timer_list *t) { @@ -943,7 +945,7 @@ static void i40iw_cqp_generic_worker(struct work_struct *work) /** * i40iw_cqp_spawn_worker - spawn worket thread - * @iwdev: device struct pointer + * @dev: device struct pointer * @work_info: work request info * @iw_vf_idx: virtual function index */ @@ -1048,7 +1050,7 @@ enum i40iw_status_code i40iw_cqp_manage_hmc_fcn_cmd(struct i40iw_sc_dev *dev, /** * i40iw_cqp_query_fpm_values_cmd - send cqp command for fpm - * @iwdev: function device struct + * @dev: function device struct * @values_mem: buffer for fpm * @hmc_fn_id: function id for fpm */ @@ -1114,7 +1116,7 @@ enum i40iw_status_code i40iw_cqp_commit_fpm_values_cmd(struct i40iw_sc_dev *dev, /** * i40iw_vf_wait_vchnl_resp - wait for channel msg - * @iwdev: function's device struct + * @dev: function's device struct */ enum i40iw_status_code i40iw_vf_wait_vchnl_resp(struct i40iw_sc_dev *dev) { @@ -1461,7 +1463,7 @@ enum i40iw_status_code i40iw_puda_get_tcpip_info(struct i40iw_puda_completion_in /** * i40iw_hw_stats_timeout - Stats timer-handler which updates all HW stats - * @vsi: pointer to the vsi structure + * @t: Timer context containing pointer to the vsi structure */ static void i40iw_hw_stats_timeout(struct timer_list *t) { diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index 65aedfe57e77..f18d146a6079 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -265,9 +265,7 @@ static struct i40iw_pbl *i40iw_get_pbl(unsigned long va, /** * i40iw_free_qp_resources - free up memory resources for qp - * @iwdev: iwarp device * @iwqp: qp ptr (user or kernel) - * @qp_num: qp number assigned */ void i40iw_free_qp_resources(struct i40iw_qp *iwqp) { @@ -302,6 +300,7 @@ static void i40iw_clean_cqes(struct i40iw_qp *iwqp, struct i40iw_cq *iwcq) /** * i40iw_destroy_qp - destroy qp * @ibqp: qp's ib pointer also to get to device's qp address + * @udata: user data */ static int i40iw_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) { @@ -338,8 +337,8 @@ static int i40iw_destroy_qp(struct ib_qp *ibqp, struct ib_udata *udata) /** * i40iw_setup_virt_qp - setup for allocation of virtual qp - * @dev: iwarp device - * @qp: qp ptr + * @iwdev: iwarp device + * @iwqp: qp ptr * @init_info: initialize info to return */ static int i40iw_setup_virt_qp(struct i40iw_device *iwdev, @@ -1241,7 +1240,7 @@ static void i40iw_copy_user_pgaddrs(struct i40iw_mr *iwmr, * i40iw_check_mem_contiguous - check if pbls stored in arr are contiguous * @arr: lvl1 pbl array * @npages: page count - * pg_size: page size + * @pg_size: page size * */ static bool i40iw_check_mem_contiguous(u64 *arr, u32 npages, u32 pg_size) @@ -1258,7 +1257,7 @@ static bool i40iw_check_mem_contiguous(u64 *arr, u32 npages, u32 pg_size) /** * i40iw_check_mr_contiguous - check if MR is physically contiguous * @palloc: pbl allocation struct - * pg_size: page size + * @pg_size: page size */ static bool i40iw_check_mr_contiguous(struct i40iw_pble_alloc *palloc, u32 pg_size) { @@ -1533,6 +1532,7 @@ static int i40iw_set_page(struct ib_mr *ibmr, u64 addr) * @ibmr: ib mem to access iwarp mr pointer * @sg: scatter gather list for fmr * @sg_nents: number of sg pages + * @sg_offset: scatter gather offset */ static int i40iw_map_mr_sg(struct ib_mr *ibmr, struct scatterlist *sg, int sg_nents, unsigned int *sg_offset) @@ -1881,6 +1881,7 @@ static void i40iw_del_memlist(struct i40iw_mr *iwmr, /** * i40iw_dereg_mr - deregister mr * @ib_mr: mr ptr for dereg + * @udata: user data */ static int i40iw_dereg_mr(struct ib_mr *ib_mr, struct ib_udata *udata) { @@ -1945,7 +1946,7 @@ static int i40iw_dereg_mr(struct ib_mr *ib_mr, struct ib_udata *udata) return 0; } -/** +/* * hw_rev_show */ static ssize_t hw_rev_show(struct device *dev, @@ -1959,7 +1960,7 @@ static ssize_t hw_rev_show(struct device *dev, } static DEVICE_ATTR_RO(hw_rev); -/** +/* * hca_type_show */ static ssize_t hca_type_show(struct device *dev, @@ -1969,7 +1970,7 @@ static ssize_t hca_type_show(struct device *dev, } static DEVICE_ATTR_RO(hca_type); -/** +/* * board_id_show */ static ssize_t board_id_show(struct device *dev, diff --git a/drivers/infiniband/hw/i40iw/i40iw_virtchnl.c b/drivers/infiniband/hw/i40iw/i40iw_virtchnl.c index 48fd327f876b..aca9061688ae 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_virtchnl.c +++ b/drivers/infiniband/hw/i40iw/i40iw_virtchnl.c @@ -119,7 +119,7 @@ static enum i40iw_status_code vchnl_vf_send_get_pe_stats_req(struct i40iw_sc_dev return ret_code; } -/** +/* * vchnl_vf_send_add_hmc_objs_req - Add HMC objects * @dev: IWARP device pointer * @vchnl_req: Virtual channel message request pointer @@ -158,9 +158,9 @@ static enum i40iw_status_code vchnl_vf_send_add_hmc_objs_req(struct i40iw_sc_dev * vchnl_vf_send_del_hmc_objs_req - del HMC objects * @dev: IWARP device pointer * @vchnl_req: Virtual channel message request pointer - * @ rsrc_type - resource type to delete - * @ start_index - starting index for resource - * @ rsrc_count - number of resource type to delete + * @rsrc_type: resource type to delete + * @start_index: starting index for resource + * @rsrc_count: number of resource type to delete */ static enum i40iw_status_code vchnl_vf_send_del_hmc_objs_req(struct i40iw_sc_dev *dev, struct i40iw_virtchnl_req *vchnl_req, @@ -222,6 +222,7 @@ static void vchnl_pf_send_get_ver_resp(struct i40iw_sc_dev *dev, * @dev: IWARP device pointer * @vf_id: Virtual function ID associated with the message * @vchnl_msg: Virtual channel message buffer pointer + * @hmc_fcn: HMC function index pointer */ static void vchnl_pf_send_get_hmc_fcn_resp(struct i40iw_sc_dev *dev, u32 vf_id, @@ -276,6 +277,7 @@ static void vchnl_pf_send_get_pe_stats_resp(struct i40iw_sc_dev *dev, * @dev: IWARP device pointer * @vf_id: Virtual function ID associated with the message * @vchnl_msg: Virtual channel message buffer pointer + * @op_ret_code: I40IW_ERR_* status code */ static void vchnl_pf_send_error_resp(struct i40iw_sc_dev *dev, u32 vf_id, struct i40iw_virtchnl_op_buf *vchnl_msg, @@ -297,8 +299,9 @@ static void vchnl_pf_send_error_resp(struct i40iw_sc_dev *dev, u32 vf_id, /** * pf_cqp_get_hmc_fcn_callback - Callback for Get HMC Fcn - * @cqp_req_param: CQP Request param value - * @not_used: unused CQP callback parameter + * @dev: IWARP device pointer + * @callback_param: unused CQP callback parameter + * @cqe_info: CQE information pointer */ static void pf_cqp_get_hmc_fcn_callback(struct i40iw_sc_dev *dev, void *callback_param, struct i40iw_ccq_cqe_info *cqe_info) @@ -331,7 +334,7 @@ static void pf_cqp_get_hmc_fcn_callback(struct i40iw_sc_dev *dev, void *callback /** * pf_add_hmc_obj - Callback for Add HMC Object - * @vf_dev: pointer to the VF Device + * @work_vf_dev: pointer to the VF Device */ static void pf_add_hmc_obj_callback(void *work_vf_dev) { @@ -404,7 +407,7 @@ del_out: /** * i40iw_vf_init_pestat - Initialize stats for VF - * @devL pointer to the VF Device + * @dev: pointer to the VF Device * @stats: Statistics structure pointer * @index: Stats index */ diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index e3cd402c079a..f26a0d920842 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -1699,7 +1699,7 @@ static struct ib_flow *mlx4_ib_create_flow(struct ib_qp *qp, struct mlx4_dev *dev = (to_mdev(qp->device))->dev; int is_bonded = mlx4_is_bonded(dev); - if (flow_attr->port < 1 || flow_attr->port > qp->device->phys_port_cnt) + if (!rdma_is_port_valid(qp->device, flow_attr->port)) return ERR_PTR(-EINVAL); if (flow_attr->flags & ~IB_FLOW_ATTR_FLAGS_DONT_TRAP) diff --git a/drivers/infiniband/hw/mlx4/sysfs.c b/drivers/infiniband/hw/mlx4/sysfs.c index 1b5891130aab..24ee79aa2122 100644 --- a/drivers/infiniband/hw/mlx4/sysfs.c +++ b/drivers/infiniband/hw/mlx4/sysfs.c @@ -798,7 +798,7 @@ static void unregister_pkey_tree(struct mlx4_ib_dev *device) int mlx4_ib_device_register_sysfs(struct mlx4_ib_dev *dev) { - int i; + unsigned int i; int ret = 0; if (!mlx4_is_master(dev->dev)) @@ -817,7 +817,7 @@ int mlx4_ib_device_register_sysfs(struct mlx4_ib_dev *dev) goto err_ports; } - for (i = 1; i <= dev->ib_dev.phys_port_cnt; ++i) { + rdma_for_each_port(&dev->ib_dev, i) { ret = add_port_entries(dev, i); if (ret) goto err_add_entries; diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index 819c142857d6..ebc2a4355fa5 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -94,13 +94,13 @@ struct devx_umem { struct mlx5_core_dev *mdev; struct ib_umem *umem; u32 dinlen; - u32 dinbox[MLX5_ST_SZ_DW(general_obj_in_cmd_hdr)]; + u32 dinbox[MLX5_ST_SZ_DW(destroy_umem_in)]; }; struct devx_umem_reg_cmd { void *in; u32 inlen; - u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)]; + u32 out[MLX5_ST_SZ_DW(create_umem_out)]; }; static struct mlx5_ib_ucontext * @@ -111,8 +111,8 @@ devx_ufile2uctx(const struct uverbs_attr_bundle *attrs) int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user) { - u32 in[MLX5_ST_SZ_DW(create_uctx_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {0}; + u32 in[MLX5_ST_SZ_DW(create_uctx_in)] = {}; + u32 out[MLX5_ST_SZ_DW(create_uctx_out)] = {}; void *uctx; int err; u16 uid; @@ -138,14 +138,14 @@ int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user) if (err) return err; - uid = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id); + uid = MLX5_GET(create_uctx_out, out, uid); return uid; } void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid) { - u32 in[MLX5_ST_SZ_DW(destroy_uctx_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)] = {0}; + u32 in[MLX5_ST_SZ_DW(destroy_uctx_in)] = {}; + u32 out[MLX5_ST_SZ_DW(destroy_uctx_out)] = {}; MLX5_SET(destroy_uctx_in, in, opcode, MLX5_CMD_OP_DESTROY_UCTX); MLX5_SET(destroy_uctx_in, in, uid, uid); @@ -288,6 +288,80 @@ static u64 get_enc_obj_id(u32 opcode, u32 obj_id) return ((u64)opcode << 32) | obj_id; } +static u32 devx_get_created_obj_id(const void *in, const void *out, u16 opcode) +{ + switch (opcode) { + case MLX5_CMD_OP_CREATE_GENERAL_OBJECT: + return MLX5_GET(general_obj_out_cmd_hdr, out, obj_id); + case MLX5_CMD_OP_CREATE_UMEM: + return MLX5_GET(create_umem_out, out, umem_id); + case MLX5_CMD_OP_CREATE_MKEY: + return MLX5_GET(create_mkey_out, out, mkey_index); + case MLX5_CMD_OP_CREATE_CQ: + return MLX5_GET(create_cq_out, out, cqn); + case MLX5_CMD_OP_ALLOC_PD: + return MLX5_GET(alloc_pd_out, out, pd); + case MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN: + return MLX5_GET(alloc_transport_domain_out, out, + transport_domain); + case MLX5_CMD_OP_CREATE_RMP: + return MLX5_GET(create_rmp_out, out, rmpn); + case MLX5_CMD_OP_CREATE_SQ: + return MLX5_GET(create_sq_out, out, sqn); + case MLX5_CMD_OP_CREATE_RQ: + return MLX5_GET(create_rq_out, out, rqn); + case MLX5_CMD_OP_CREATE_RQT: + return MLX5_GET(create_rqt_out, out, rqtn); + case MLX5_CMD_OP_CREATE_TIR: + return MLX5_GET(create_tir_out, out, tirn); + case MLX5_CMD_OP_CREATE_TIS: + return MLX5_GET(create_tis_out, out, tisn); + case MLX5_CMD_OP_ALLOC_Q_COUNTER: + return MLX5_GET(alloc_q_counter_out, out, counter_set_id); + case MLX5_CMD_OP_CREATE_FLOW_TABLE: + return MLX5_GET(create_flow_table_out, out, table_id); + case MLX5_CMD_OP_CREATE_FLOW_GROUP: + return MLX5_GET(create_flow_group_out, out, group_id); + case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY: + return MLX5_GET(set_fte_in, in, flow_index); + case MLX5_CMD_OP_ALLOC_FLOW_COUNTER: + return MLX5_GET(alloc_flow_counter_out, out, flow_counter_id); + case MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT: + return MLX5_GET(alloc_packet_reformat_context_out, out, + packet_reformat_id); + case MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT: + return MLX5_GET(alloc_modify_header_context_out, out, + modify_header_id); + case MLX5_CMD_OP_CREATE_SCHEDULING_ELEMENT: + return MLX5_GET(create_scheduling_element_out, out, + scheduling_element_id); + case MLX5_CMD_OP_ADD_VXLAN_UDP_DPORT: + return MLX5_GET(add_vxlan_udp_dport_in, in, vxlan_udp_port); + case MLX5_CMD_OP_SET_L2_TABLE_ENTRY: + return MLX5_GET(set_l2_table_entry_in, in, table_index); + case MLX5_CMD_OP_CREATE_QP: + return MLX5_GET(create_qp_out, out, qpn); + case MLX5_CMD_OP_CREATE_SRQ: + return MLX5_GET(create_srq_out, out, srqn); + case MLX5_CMD_OP_CREATE_XRC_SRQ: + return MLX5_GET(create_xrc_srq_out, out, xrc_srqn); + case MLX5_CMD_OP_CREATE_DCT: + return MLX5_GET(create_dct_out, out, dctn); + case MLX5_CMD_OP_CREATE_XRQ: + return MLX5_GET(create_xrq_out, out, xrqn); + case MLX5_CMD_OP_ATTACH_TO_MCG: + return MLX5_GET(attach_to_mcg_in, in, qpn); + case MLX5_CMD_OP_ALLOC_XRCD: + return MLX5_GET(alloc_xrcd_out, out, xrcd); + case MLX5_CMD_OP_CREATE_PSV: + return MLX5_GET(create_psv_out, out, psv0_index); + default: + /* The entry must match to one of the devx_is_obj_create_cmd */ + WARN_ON(true); + return 0; + } +} + static u64 devx_get_obj_id(const void *in) { u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode); @@ -399,8 +473,8 @@ static u64 devx_get_obj_id(const void *in) break; case MLX5_CMD_OP_QUERY_MODIFY_HEADER_CONTEXT: obj_id = get_enc_obj_id(MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT, - MLX5_GET(general_obj_in_cmd_hdr, in, - obj_id)); + MLX5_GET(query_modify_header_context_in, + in, modify_header_id)); break; case MLX5_CMD_OP_QUERY_SCHEDULING_ELEMENT: obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_SCHEDULING_ELEMENT, @@ -1019,63 +1093,76 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, u32 *dinlen, u32 *obj_id) { - u16 obj_type = MLX5_GET(general_obj_in_cmd_hdr, in, obj_type); + u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode); u16 uid = MLX5_GET(general_obj_in_cmd_hdr, in, uid); - *obj_id = MLX5_GET(general_obj_out_cmd_hdr, out, obj_id); + *obj_id = devx_get_created_obj_id(in, out, opcode); *dinlen = MLX5_ST_SZ_BYTES(general_obj_in_cmd_hdr); - - MLX5_SET(general_obj_in_cmd_hdr, din, obj_id, *obj_id); MLX5_SET(general_obj_in_cmd_hdr, din, uid, uid); - switch (MLX5_GET(general_obj_in_cmd_hdr, in, opcode)) { + switch (opcode) { case MLX5_CMD_OP_CREATE_GENERAL_OBJECT: MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_GENERAL_OBJECT); - MLX5_SET(general_obj_in_cmd_hdr, din, obj_type, obj_type); + MLX5_SET(general_obj_in_cmd_hdr, din, obj_id, *obj_id); + MLX5_SET(general_obj_in_cmd_hdr, din, obj_type, + MLX5_GET(general_obj_in_cmd_hdr, in, obj_type)); break; case MLX5_CMD_OP_CREATE_UMEM: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_umem_in, din, opcode, MLX5_CMD_OP_DESTROY_UMEM); + MLX5_SET(destroy_umem_in, din, umem_id, *obj_id); break; case MLX5_CMD_OP_CREATE_MKEY: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_MKEY); + MLX5_SET(destroy_mkey_in, din, opcode, + MLX5_CMD_OP_DESTROY_MKEY); + MLX5_SET(destroy_mkey_in, in, mkey_index, *obj_id); break; case MLX5_CMD_OP_CREATE_CQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_CQ); + MLX5_SET(destroy_cq_in, din, opcode, MLX5_CMD_OP_DESTROY_CQ); + MLX5_SET(destroy_cq_in, din, cqn, *obj_id); break; case MLX5_CMD_OP_ALLOC_PD: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DEALLOC_PD); + MLX5_SET(dealloc_pd_in, din, opcode, MLX5_CMD_OP_DEALLOC_PD); + MLX5_SET(dealloc_pd_in, din, pd, *obj_id); break; case MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(dealloc_transport_domain_in, din, opcode, MLX5_CMD_OP_DEALLOC_TRANSPORT_DOMAIN); + MLX5_SET(dealloc_transport_domain_in, din, transport_domain, + *obj_id); break; case MLX5_CMD_OP_CREATE_RMP: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_RMP); + MLX5_SET(destroy_rmp_in, din, opcode, MLX5_CMD_OP_DESTROY_RMP); + MLX5_SET(destroy_rmp_in, din, rmpn, *obj_id); break; case MLX5_CMD_OP_CREATE_SQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_SQ); + MLX5_SET(destroy_sq_in, din, opcode, MLX5_CMD_OP_DESTROY_SQ); + MLX5_SET(destroy_sq_in, din, sqn, *obj_id); break; case MLX5_CMD_OP_CREATE_RQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_RQ); + MLX5_SET(destroy_rq_in, din, opcode, MLX5_CMD_OP_DESTROY_RQ); + MLX5_SET(destroy_rq_in, din, rqn, *obj_id); break; case MLX5_CMD_OP_CREATE_RQT: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_RQT); + MLX5_SET(destroy_rqt_in, din, opcode, MLX5_CMD_OP_DESTROY_RQT); + MLX5_SET(destroy_rqt_in, din, rqtn, *obj_id); break; case MLX5_CMD_OP_CREATE_TIR: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_TIR); + MLX5_SET(destroy_tir_in, din, opcode, MLX5_CMD_OP_DESTROY_TIR); + MLX5_SET(destroy_tir_in, din, tirn, *obj_id); break; case MLX5_CMD_OP_CREATE_TIS: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_TIS); + MLX5_SET(destroy_tis_in, din, opcode, MLX5_CMD_OP_DESTROY_TIS); + MLX5_SET(destroy_tis_in, din, tisn, *obj_id); break; case MLX5_CMD_OP_ALLOC_Q_COUNTER: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(dealloc_q_counter_in, din, opcode, MLX5_CMD_OP_DEALLOC_Q_COUNTER); + MLX5_SET(dealloc_q_counter_in, din, counter_set_id, *obj_id); break; case MLX5_CMD_OP_CREATE_FLOW_TABLE: *dinlen = MLX5_ST_SZ_BYTES(destroy_flow_table_in); - *obj_id = MLX5_GET(create_flow_table_out, out, table_id); MLX5_SET(destroy_flow_table_in, din, other_vport, MLX5_GET(create_flow_table_in, in, other_vport)); MLX5_SET(destroy_flow_table_in, din, vport_number, @@ -1083,12 +1170,11 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, MLX5_SET(destroy_flow_table_in, din, table_type, MLX5_GET(create_flow_table_in, in, table_type)); MLX5_SET(destroy_flow_table_in, din, table_id, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_flow_table_in, din, opcode, MLX5_CMD_OP_DESTROY_FLOW_TABLE); break; case MLX5_CMD_OP_CREATE_FLOW_GROUP: *dinlen = MLX5_ST_SZ_BYTES(destroy_flow_group_in); - *obj_id = MLX5_GET(create_flow_group_out, out, group_id); MLX5_SET(destroy_flow_group_in, din, other_vport, MLX5_GET(create_flow_group_in, in, other_vport)); MLX5_SET(destroy_flow_group_in, din, vport_number, @@ -1098,12 +1184,11 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, MLX5_SET(destroy_flow_group_in, din, table_id, MLX5_GET(create_flow_group_in, in, table_id)); MLX5_SET(destroy_flow_group_in, din, group_id, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_flow_group_in, din, opcode, MLX5_CMD_OP_DESTROY_FLOW_GROUP); break; case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY: *dinlen = MLX5_ST_SZ_BYTES(delete_fte_in); - *obj_id = MLX5_GET(set_fte_in, in, flow_index); MLX5_SET(delete_fte_in, din, other_vport, MLX5_GET(set_fte_in, in, other_vport)); MLX5_SET(delete_fte_in, din, vport_number, @@ -1113,63 +1198,70 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, MLX5_SET(delete_fte_in, din, table_id, MLX5_GET(set_fte_in, in, table_id)); MLX5_SET(delete_fte_in, din, flow_index, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(delete_fte_in, din, opcode, MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY); break; case MLX5_CMD_OP_ALLOC_FLOW_COUNTER: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(dealloc_flow_counter_in, din, opcode, MLX5_CMD_OP_DEALLOC_FLOW_COUNTER); + MLX5_SET(dealloc_flow_counter_in, din, flow_counter_id, + *obj_id); break; case MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(dealloc_packet_reformat_context_in, din, opcode, MLX5_CMD_OP_DEALLOC_PACKET_REFORMAT_CONTEXT); + MLX5_SET(dealloc_packet_reformat_context_in, din, + packet_reformat_id, *obj_id); break; case MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(dealloc_modify_header_context_in, din, opcode, MLX5_CMD_OP_DEALLOC_MODIFY_HEADER_CONTEXT); + MLX5_SET(dealloc_modify_header_context_in, din, + modify_header_id, *obj_id); break; case MLX5_CMD_OP_CREATE_SCHEDULING_ELEMENT: *dinlen = MLX5_ST_SZ_BYTES(destroy_scheduling_element_in); - *obj_id = MLX5_GET(create_scheduling_element_out, out, - scheduling_element_id); MLX5_SET(destroy_scheduling_element_in, din, scheduling_hierarchy, MLX5_GET(create_scheduling_element_in, in, scheduling_hierarchy)); MLX5_SET(destroy_scheduling_element_in, din, scheduling_element_id, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_scheduling_element_in, din, opcode, MLX5_CMD_OP_DESTROY_SCHEDULING_ELEMENT); break; case MLX5_CMD_OP_ADD_VXLAN_UDP_DPORT: *dinlen = MLX5_ST_SZ_BYTES(delete_vxlan_udp_dport_in); - *obj_id = MLX5_GET(add_vxlan_udp_dport_in, in, vxlan_udp_port); MLX5_SET(delete_vxlan_udp_dport_in, din, vxlan_udp_port, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(delete_vxlan_udp_dport_in, din, opcode, MLX5_CMD_OP_DELETE_VXLAN_UDP_DPORT); break; case MLX5_CMD_OP_SET_L2_TABLE_ENTRY: *dinlen = MLX5_ST_SZ_BYTES(delete_l2_table_entry_in); - *obj_id = MLX5_GET(set_l2_table_entry_in, in, table_index); MLX5_SET(delete_l2_table_entry_in, din, table_index, *obj_id); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(delete_l2_table_entry_in, din, opcode, MLX5_CMD_OP_DELETE_L2_TABLE_ENTRY); break; case MLX5_CMD_OP_CREATE_QP: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_QP); + MLX5_SET(destroy_qp_in, din, opcode, MLX5_CMD_OP_DESTROY_QP); + MLX5_SET(destroy_qp_in, din, qpn, *obj_id); break; case MLX5_CMD_OP_CREATE_SRQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_SRQ); + MLX5_SET(destroy_srq_in, din, opcode, MLX5_CMD_OP_DESTROY_SRQ); + MLX5_SET(destroy_srq_in, din, srqn, *obj_id); break; case MLX5_CMD_OP_CREATE_XRC_SRQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_xrc_srq_in, din, opcode, MLX5_CMD_OP_DESTROY_XRC_SRQ); + MLX5_SET(destroy_xrc_srq_in, din, xrc_srqn, *obj_id); break; case MLX5_CMD_OP_CREATE_DCT: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_DCT); + MLX5_SET(destroy_dct_in, din, opcode, MLX5_CMD_OP_DESTROY_DCT); + MLX5_SET(destroy_dct_in, din, dctn, *obj_id); break; case MLX5_CMD_OP_CREATE_XRQ: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DESTROY_XRQ); + MLX5_SET(destroy_xrq_in, din, opcode, MLX5_CMD_OP_DESTROY_XRQ); + MLX5_SET(destroy_xrq_in, din, xrqn, *obj_id); break; case MLX5_CMD_OP_ATTACH_TO_MCG: *dinlen = MLX5_ST_SZ_BYTES(detach_from_mcg_in); @@ -1178,16 +1270,19 @@ static void devx_obj_build_destroy_cmd(void *in, void *out, void *din, memcpy(MLX5_ADDR_OF(detach_from_mcg_in, din, multicast_gid), MLX5_ADDR_OF(attach_to_mcg_in, in, multicast_gid), MLX5_FLD_SZ_BYTES(attach_to_mcg_in, multicast_gid)); - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DETACH_FROM_MCG); + MLX5_SET(detach_from_mcg_in, din, opcode, + MLX5_CMD_OP_DETACH_FROM_MCG); + MLX5_SET(detach_from_mcg_in, din, qpn, *obj_id); break; case MLX5_CMD_OP_ALLOC_XRCD: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, MLX5_CMD_OP_DEALLOC_XRCD); + MLX5_SET(dealloc_xrcd_in, din, opcode, + MLX5_CMD_OP_DEALLOC_XRCD); + MLX5_SET(dealloc_xrcd_in, din, xrcd, *obj_id); break; case MLX5_CMD_OP_CREATE_PSV: - MLX5_SET(general_obj_in_cmd_hdr, din, opcode, + MLX5_SET(destroy_psv_in, din, opcode, MLX5_CMD_OP_DESTROY_PSV); - MLX5_SET(destroy_psv_in, din, psvn, - MLX5_GET(create_psv_out, out, psv0_index)); + MLX5_SET(destroy_psv_in, din, psvn, *obj_id); break; default: /* The entry must match to one of the devx_is_obj_create_cmd */ @@ -1215,9 +1310,9 @@ static int devx_handle_mkey_indirect(struct devx_obj *obj, mkey->size = MLX5_GET64(mkc, mkc, len); mkey->pd = MLX5_GET(mkc, mkc, pd); devx_mr->ndescs = MLX5_GET(mkc, mkc, translations_octword_size); + init_waitqueue_head(&mkey->wait); - return xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(mkey->key), mkey, - GFP_KERNEL)); + return mlx5r_store_odp_mkey(dev, mkey); } static int devx_handle_mkey_create(struct mlx5_ib_dev *dev, @@ -1290,16 +1385,15 @@ static int devx_obj_cleanup(struct ib_uobject *uobject, int ret; dev = mlx5_udata_to_mdev(&attrs->driver_udata); - if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY) { + if (obj->flags & DEVX_OBJ_FLAGS_INDIRECT_MKEY && + xa_erase(&obj->ib_dev->odp_mkeys, + mlx5_base_mkey(obj->devx_mr.mmkey.key))) /* * The pagefault_single_data_segment() does commands against * the mmkey, we must wait for that to stop before freeing the * mkey, as another allocation could get the same mkey #. */ - xa_erase(&obj->ib_dev->odp_mkeys, - mlx5_base_mkey(obj->devx_mr.mmkey.key)); - synchronize_srcu(&dev->odp_srcu); - } + mlx5r_deref_wait_odp_mkey(&obj->devx_mr.mmkey); if (obj->flags & DEVX_OBJ_FLAGS_DCT) ret = mlx5_core_destroy_dct(obj->ib_dev, &obj->core_dct); @@ -1345,6 +1439,16 @@ out: rcu_read_unlock(); } +static bool is_apu_thread_cq(struct mlx5_ib_dev *dev, const void *in) +{ + if (!MLX5_CAP_GEN(dev->mdev, apu) || + !MLX5_GET(cqc, MLX5_ADDR_OF(create_cq_in, in, cq_context), + apu_thread_cq)) + return false; + + return true; +} + static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( struct uverbs_attr_bundle *attrs) { @@ -1398,7 +1502,8 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( obj->flags |= DEVX_OBJ_FLAGS_DCT; err = mlx5_core_create_dct(dev, &obj->core_dct, cmd_in, cmd_in_len, cmd_out, cmd_out_len); - } else if (opcode == MLX5_CMD_OP_CREATE_CQ) { + } else if (opcode == MLX5_CMD_OP_CREATE_CQ && + !is_apu_thread_cq(dev, cmd_in)) { obj->flags |= DEVX_OBJ_FLAGS_CQ; obj->core_cq.comp = devx_cq_comp; err = mlx5_core_create_cq(dev->mdev, &obj->core_cq, diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c index 9bb9bb058932..652c6ccf1881 100644 --- a/drivers/infiniband/hw/mlx5/mad.c +++ b/drivers/infiniband/hw/mlx5/mad.c @@ -48,7 +48,7 @@ static bool can_do_mad_ifc(struct mlx5_ib_dev *dev, u8 port_num, if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_LID_ROUTED && in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) return true; - return dev->mdev->port_caps[port_num - 1].has_smi; + return dev->port_caps[port_num - 1].has_smi; } static int mlx5_MAD_IFC(struct mlx5_ib_dev *dev, int ignore_mkey, @@ -279,7 +279,7 @@ int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY; } -int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port) +int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, unsigned int port) { struct ib_smp *in_mad = NULL; struct ib_smp *out_mad = NULL; @@ -299,7 +299,7 @@ int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port) packet_error = be16_to_cpu(out_mad->status); - dev->mdev->port_caps[port - 1].ext_port_cap = (!err && !packet_error) ? + dev->port_caps[port - 1].ext_port_cap = (!err && !packet_error) ? MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO : 0; out: @@ -308,8 +308,8 @@ out: return err; } -int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, - struct ib_smp *out_mad) +static int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, + struct ib_smp *out_mad) { struct ib_smp *in_mad = NULL; int err = -ENOMEM; @@ -549,7 +549,7 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, props->port_cap_flags = be32_to_cpup((__be32 *)(out_mad->data + 20)); props->gid_tbl_len = out_mad->data[50]; props->max_msg_sz = 1 << MLX5_CAP_GEN(mdev, log_max_msg); - props->pkey_tbl_len = mdev->port_caps[port - 1].pkey_table_len; + props->pkey_tbl_len = dev->pkey_table_len; props->bad_pkey_cntr = be16_to_cpup((__be16 *)(out_mad->data + 46)); props->qkey_viol_cntr = be16_to_cpup((__be16 *)(out_mad->data + 48)); props->active_width = out_mad->data[31] & 0xf; @@ -589,7 +589,7 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, /* If reported active speed is QDR, check if is FDR-10 */ if (props->active_speed == 4) { - if (mdev->port_caps[port - 1].ext_port_cap & + if (dev->port_caps[port - 1].ext_port_cap & MLX_EXT_PORT_CAP_FLAG_EXTENDED_PORT_INFO) { init_query_mad(in_mad); in_mad->attr_id = MLX5_ATTR_EXTENDED_PORT_INFO; diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index aabdc07e4753..0d69a697d75f 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -1,6 +1,7 @@ // SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + * Copyright (c) 2020, Intel Corporation. All rights reserved. */ #include <linux/debugfs.h> @@ -461,7 +462,6 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, struct net_device *ndev, *upper; enum ib_mtu ndev_ib_mtu; bool put_mdev = true; - u16 qkey_viol_cntr; u32 eth_prot_oper; u8 mdev_port_num; bool ext; @@ -499,20 +499,22 @@ static int mlx5_query_port_roce(struct ib_device *device, u8 port_num, translate_eth_proto_oper(eth_prot_oper, &props->active_speed, &props->active_width, ext); - props->port_cap_flags |= IB_PORT_CM_SUP; - props->ip_gids = true; + if (!dev->is_rep && mlx5_is_roce_enabled(mdev)) { + u16 qkey_viol_cntr; - props->gid_tbl_len = MLX5_CAP_ROCE(dev->mdev, - roce_address_table_size); + props->port_cap_flags |= IB_PORT_CM_SUP; + props->ip_gids = true; + props->gid_tbl_len = MLX5_CAP_ROCE(dev->mdev, + roce_address_table_size); + mlx5_query_nic_vport_qkey_viol_cntr(mdev, &qkey_viol_cntr); + props->qkey_viol_cntr = qkey_viol_cntr; + } props->max_mtu = IB_MTU_4096; 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 = IB_PORT_PHYS_STATE_DISABLED; - mlx5_query_nic_vport_qkey_viol_cntr(mdev, &qkey_viol_cntr); - props->qkey_viol_cntr = qkey_viol_cntr; - /* If this is a stub query for an unaffiliated port stop here */ if (!put_mdev) goto out; @@ -815,9 +817,7 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, if (err) return err; - err = mlx5_query_max_pkeys(ibdev, &props->max_pkeys); - if (err) - return err; + props->max_pkeys = dev->pkey_table_len; err = mlx5_query_vendor_id(ibdev, &props->vendor_id); if (err) @@ -1384,19 +1384,17 @@ int mlx5_ib_query_port(struct ib_device *ibdev, u8 port, static int mlx5_ib_rep_query_port(struct ib_device *ibdev, u8 port, struct ib_port_attr *props) { - int ret; + return mlx5_query_port_roce(ibdev, port, props); +} - /* Only link layer == ethernet is valid for representors - * and we always use port 1 +static int mlx5_ib_rep_query_pkey(struct ib_device *ibdev, u8 port, u16 index, + u16 *pkey) +{ + /* Default special Pkey for representor device port as per the + * IB specification 1.3 section 10.9.1.2. */ - ret = mlx5_query_port_roce(ibdev, port, props); - if (ret || !props) - return ret; - - /* We don't support GIDS */ - props->gid_tbl_len = 0; - - return ret; + *pkey = 0xffff; + return 0; } static int mlx5_ib_query_gid(struct ib_device *ibdev, u8 port, int index, @@ -2935,8 +2933,8 @@ static int set_has_smi_cap(struct mlx5_ib_dev *dev) int err; int port; - for (port = 1; port <= ARRAY_SIZE(dev->mdev->port_caps); port++) { - dev->mdev->port_caps[port - 1].has_smi = false; + for (port = 1; port <= ARRAY_SIZE(dev->port_caps); port++) { + dev->port_caps[port - 1].has_smi = false; if (MLX5_CAP_GEN(dev->mdev, port_type) == MLX5_CAP_PORT_TYPE_IB) { if (MLX5_CAP_GEN(dev->mdev, ib_virt)) { @@ -2948,10 +2946,10 @@ static int set_has_smi_cap(struct mlx5_ib_dev *dev) port, err); return err; } - dev->mdev->port_caps[port - 1].has_smi = + dev->port_caps[port - 1].has_smi = vport_ctx.has_smi; } else { - dev->mdev->port_caps[port - 1].has_smi = true; + dev->port_caps[port - 1].has_smi = true; } } } @@ -2960,63 +2958,12 @@ static int set_has_smi_cap(struct mlx5_ib_dev *dev) static void get_ext_port_caps(struct mlx5_ib_dev *dev) { - int port; + unsigned int port; - for (port = 1; port <= dev->num_ports; port++) + rdma_for_each_port (&dev->ib_dev, port) mlx5_query_ext_port_caps(dev, port); } -static int __get_port_caps(struct mlx5_ib_dev *dev, u8 port) -{ - struct ib_device_attr *dprops = NULL; - struct ib_port_attr *pprops = NULL; - int err = -ENOMEM; - - pprops = kzalloc(sizeof(*pprops), GFP_KERNEL); - if (!pprops) - goto out; - - dprops = kmalloc(sizeof(*dprops), GFP_KERNEL); - if (!dprops) - goto out; - - err = mlx5_ib_query_device(&dev->ib_dev, dprops, NULL); - if (err) { - mlx5_ib_warn(dev, "query_device failed %d\n", err); - goto out; - } - - err = mlx5_ib_query_port(&dev->ib_dev, port, pprops); - if (err) { - mlx5_ib_warn(dev, "query_port %d failed %d\n", - port, err); - goto out; - } - - dev->mdev->port_caps[port - 1].pkey_table_len = - dprops->max_pkeys; - dev->mdev->port_caps[port - 1].gid_table_len = - pprops->gid_tbl_len; - mlx5_ib_dbg(dev, "port %d: pkey_table_len %d, gid_table_len %d\n", - port, dprops->max_pkeys, pprops->gid_tbl_len); - -out: - kfree(pprops); - kfree(dprops); - - return err; -} - -static int get_port_caps(struct mlx5_ib_dev *dev, u8 port) -{ - /* For representors use port 1, is this is the only native - * port - */ - if (dev->is_rep) - return __get_port_caps(dev, 1); - return __get_port_caps(dev, port); -} - static u8 mlx5_get_umr_fence(u8 umr_fence_cap) { switch (umr_fence_cap) { @@ -3488,10 +3435,6 @@ static bool mlx5_ib_bind_slave_port(struct mlx5_ib_dev *ibdev, if (err) goto unbind; - err = get_port_caps(ibdev, mlx5_core_native_port_num(mpi->mdev)); - if (err) - goto unbind; - err = mlx5_add_netdev_notifier(ibdev, port_num); if (err) { mlx5_ib_err(ibdev, "failed adding netdev notifier for port %u\n", @@ -3569,11 +3512,9 @@ static int mlx5_ib_init_multiport_master(struct mlx5_ib_dev *dev) break; } } - if (!bound) { - get_port_caps(dev, i + 1); + if (!bound) mlx5_ib_dbg(dev, "no free port found for port %d\n", i + 1); - } } list_add_tail(&dev->ib_dev_list, &mlx5_ib_dev_list); @@ -3926,8 +3867,7 @@ static void mlx5_ib_stage_init_cleanup(struct mlx5_ib_dev *dev) { mlx5_ib_cleanup_multiport_master(dev); WARN_ON(!xa_empty(&dev->odp_mkeys)); - cleanup_srcu_struct(&dev->odp_srcu); - + mutex_destroy(&dev->cap_mask_mutex); WARN_ON(!xa_empty(&dev->sig_mrs)); WARN_ON(!bitmap_empty(dev->dm.memic_alloc_pages, MLX5_MAX_MEMIC_PAGES)); } @@ -3938,6 +3878,12 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev) int err; int i; + dev->ib_dev.node_type = RDMA_NODE_IB_CA; + dev->ib_dev.local_dma_lkey = 0 /* not supported for now */; + dev->ib_dev.phys_port_cnt = dev->num_ports; + dev->ib_dev.dev.parent = mdev->device; + dev->ib_dev.lag_flags = RDMA_LAG_FLAGS_HASH_ALL_SLAVES; + for (i = 0; i < dev->num_ports; i++) { spin_lock_init(&dev->port[i].mp.mpi_lock); rwlock_init(&dev->port[i].roce.netdev_lock); @@ -3956,27 +3902,14 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev) if (err) goto err_mp; - if (!mlx5_core_mp_enabled(mdev)) { - for (i = 1; i <= dev->num_ports; i++) { - err = get_port_caps(dev, i); - if (err) - break; - } - } else { - err = get_port_caps(dev, mlx5_core_native_port_num(mdev)); - } + err = mlx5_query_max_pkeys(&dev->ib_dev, &dev->pkey_table_len); if (err) goto err_mp; if (mlx5_use_mad_ifc(dev)) get_ext_port_caps(dev); - dev->ib_dev.node_type = RDMA_NODE_IB_CA; - dev->ib_dev.local_dma_lkey = 0 /* not supported for now */; - dev->ib_dev.phys_port_cnt = dev->num_ports; dev->ib_dev.num_comp_vectors = mlx5_comp_vectors_count(mdev); - dev->ib_dev.dev.parent = mdev->device; - dev->ib_dev.lag_flags = RDMA_LAG_FLAGS_HASH_ALL_SLAVES; mutex_init(&dev->cap_mask_mutex); INIT_LIST_HEAD(&dev->qp_list); @@ -3987,17 +3920,11 @@ static int mlx5_ib_stage_init_init(struct mlx5_ib_dev *dev) spin_lock_init(&dev->dm.lock); dev->dm.dev = mdev; - - err = init_srcu_struct(&dev->odp_srcu); - if (err) - goto err_mp; - return 0; err_mp: mlx5_ib_cleanup_multiport_master(dev); - - return -ENOMEM; + return err; } static int mlx5_ib_enable_driver(struct ib_device *dev) @@ -4067,6 +3994,7 @@ static const struct ib_device_ops mlx5_ib_dev_ops = { .query_srq = mlx5_ib_query_srq, .query_ucontext = mlx5_ib_query_ucontext, .reg_user_mr = mlx5_ib_reg_user_mr, + .reg_user_mr_dmabuf = mlx5_ib_reg_user_mr_dmabuf, .req_notify_cq = mlx5_ib_arm_cq, .rereg_user_mr = mlx5_ib_rereg_user_mr, .resize_cq = mlx5_ib_resize_cq, @@ -4207,6 +4135,7 @@ static int mlx5_ib_stage_non_default_cb(struct mlx5_ib_dev *dev) static const struct ib_device_ops mlx5_ib_dev_port_rep_ops = { .get_port_immutable = mlx5_port_rep_immutable, .query_port = mlx5_ib_rep_query_port, + .query_pkey = mlx5_ib_rep_query_pkey, }; static int mlx5_ib_stage_raw_eth_non_default_cb(struct mlx5_ib_dev *dev) diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index b0fdc1b08e06..88cc26e008fc 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -1,6 +1,7 @@ /* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ /* * Copyright (c) 2013-2020, Mellanox Technologies inc. All rights reserved. + * Copyright (c) 2020, Intel Corporation. All rights reserved. */ #ifndef MLX5_IB_H @@ -683,11 +684,8 @@ struct mlx5_ib_mr { u64 pi_iova; /* For ODP and implicit */ - atomic_t num_deferred_work; - wait_queue_head_t q_deferred_work; struct xarray implicit_children; union { - struct rcu_head rcu; struct list_head elm; struct work_struct work; } odp_destroy; @@ -703,6 +701,12 @@ static inline bool is_odp_mr(struct mlx5_ib_mr *mr) mr->umem->is_odp; } +static inline bool is_dmabuf_mr(struct mlx5_ib_mr *mr) +{ + return IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) && mr->umem && + mr->umem->is_dmabuf; +} + struct mlx5_ib_mw { struct ib_mw ibmw; struct mlx5_core_mkey mmkey; @@ -1029,6 +1033,11 @@ struct mlx5_var_table { u64 num_var_hw_entries; }; +struct mlx5_port_caps { + bool has_smi; + u8 ext_port_cap; +}; + struct mlx5_ib_dev { struct ib_device ib_dev; struct mlx5_core_dev *mdev; @@ -1056,11 +1065,6 @@ struct mlx5_ib_dev { u64 odp_max_size; struct mlx5_ib_pf_eq odp_pf_eq; - /* - * Sleepable RCU that prevents destruction of MRs while they are still - * being used by a page fault handler. - */ - struct srcu_struct odp_srcu; struct xarray odp_mkeys; u32 null_mkey; @@ -1089,6 +1093,8 @@ struct mlx5_ib_dev { struct mlx5_var_table var_table; struct xarray sig_mrs; + struct mlx5_port_caps port_caps[MLX5_MAX_PORTS]; + u16 pkey_table_len; }; static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq) @@ -1243,6 +1249,10 @@ struct ib_mr *mlx5_ib_get_dma_mr(struct ib_pd *pd, int acc); struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, u64 virt_addr, int access_flags, struct ib_udata *udata); +struct ib_mr *mlx5_ib_reg_user_mr_dmabuf(struct ib_pd *pd, u64 start, + u64 length, u64 virt_addr, + int fd, int access_flags, + struct ib_udata *udata); int mlx5_ib_advise_mr(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, u32 flags, @@ -1253,11 +1263,13 @@ int mlx5_ib_alloc_mw(struct ib_mw *mw, struct ib_udata *udata); int mlx5_ib_dealloc_mw(struct ib_mw *mw); int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages, int page_shift, int flags); +int mlx5_ib_update_mr_pas(struct mlx5_ib_mr *mr, unsigned int flags); struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd, struct ib_udata *udata, int access_flags); void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *mr); void mlx5_ib_fence_odp_mr(struct mlx5_ib_mr *mr); +void mlx5_ib_fence_dmabuf_mr(struct mlx5_ib_mr *mr); struct ib_mr *mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start, u64 length, u64 virt_addr, int access_flags, struct ib_pd *pd, struct ib_udata *udata); @@ -1279,9 +1291,7 @@ int mlx5_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, size_t *out_mad_size, u16 *out_mad_pkey_index); int mlx5_ib_alloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata); int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd, struct ib_udata *udata); -int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, u8 port); -int mlx5_query_mad_ifc_smp_attr_node_info(struct ib_device *ibdev, - struct ib_smp *out_mad); +int mlx5_query_ext_port_caps(struct mlx5_ib_dev *dev, unsigned int port); int mlx5_query_mad_ifc_system_image_guid(struct ib_device *ibdev, __be64 *sys_image_guid); int mlx5_query_mad_ifc_max_pkeys(struct ib_device *ibdev, @@ -1345,6 +1355,7 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, u32 flags, struct ib_sge *sg_list, u32 num_sge); int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr); +int mlx5_ib_init_dmabuf_mr(struct mlx5_ib_mr *mr); #else /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */ static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) { @@ -1370,6 +1381,10 @@ static inline int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr) { return -EOPNOTSUPP; } +static inline int mlx5_ib_init_dmabuf_mr(struct mlx5_ib_mr *mr) +{ + return -EOPNOTSUPP; +} #endif /* CONFIG_INFINIBAND_ON_DEMAND_PAGING */ extern const struct mmu_interval_notifier_ops mlx5_mn_ops; @@ -1576,6 +1591,29 @@ static inline bool mlx5_ib_can_reconfig_with_umr(struct mlx5_ib_dev *dev, return true; } +static inline int mlx5r_store_odp_mkey(struct mlx5_ib_dev *dev, + struct mlx5_core_mkey *mmkey) +{ + refcount_set(&mmkey->usecount, 1); + + return xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(mmkey->key), + mmkey, GFP_KERNEL)); +} + +/* deref an mkey that can participate in ODP flow */ +static inline void mlx5r_deref_odp_mkey(struct mlx5_core_mkey *mmkey) +{ + if (refcount_dec_and_test(&mmkey->usecount)) + wake_up(&mmkey->wait); +} + +/* deref an mkey that can participate in ODP flow and wait for relese */ +static inline void mlx5r_deref_wait_odp_mkey(struct mlx5_core_mkey *mmkey) +{ + mlx5r_deref_odp_mkey(mmkey); + wait_event(mmkey->wait, refcount_read(&mmkey->usecount) == 0); +} + int mlx5_ib_test_wc(struct mlx5_ib_dev *dev); static inline bool mlx5_ib_lag_should_assign_affinity(struct mlx5_ib_dev *dev) diff --git a/drivers/infiniband/hw/mlx5/mr.c b/drivers/infiniband/hw/mlx5/mr.c index 24f8d59a42ea..db05b0e0a8d7 100644 --- a/drivers/infiniband/hw/mlx5/mr.c +++ b/drivers/infiniband/hw/mlx5/mr.c @@ -1,5 +1,6 @@ /* * Copyright (c) 2013-2015, Mellanox Technologies. All rights reserved. + * Copyright (c) 2020, Intel Corporation. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -36,6 +37,8 @@ #include <linux/debugfs.h> #include <linux/export.h> #include <linux/delay.h> +#include <linux/dma-buf.h> +#include <linux/dma-resv.h> #include <rdma/ib_umem.h> #include <rdma/ib_umem_odp.h> #include <rdma/ib_verbs.h> @@ -155,6 +158,7 @@ static void create_mkey_callback(int status, struct mlx5_async_work *context) mr->mmkey.type = MLX5_MKEY_MR; mr->mmkey.key |= mlx5_idx_to_mkey( MLX5_GET(create_mkey_out, mr->out, mkey_index)); + init_waitqueue_head(&mr->mmkey.wait); WRITE_ONCE(dev->cache.last_add, jiffies); @@ -935,6 +939,17 @@ static void set_mr_fields(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr, mr->access_flags = access_flags; } +static unsigned int mlx5_umem_dmabuf_default_pgsz(struct ib_umem *umem, + u64 iova) +{ + /* + * The alignment of iova has already been checked upon entering + * UVERBS_METHOD_REG_DMABUF_MR + */ + umem->iova = iova; + return PAGE_SIZE; +} + static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd, struct ib_umem *umem, u64 iova, int access_flags) @@ -944,7 +959,11 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd, struct mlx5_ib_mr *mr; unsigned int page_size; - page_size = mlx5_umem_find_best_pgsz(umem, mkc, log_page_size, 0, iova); + if (umem->is_dmabuf) + page_size = mlx5_umem_dmabuf_default_pgsz(umem, iova); + else + page_size = mlx5_umem_find_best_pgsz(umem, mkc, log_page_size, + 0, iova); if (WARN_ON(!page_size)) return ERR_PTR(-EINVAL); ent = mr_cache_ent_from_order( @@ -980,7 +999,6 @@ static struct mlx5_ib_mr *alloc_cacheable_mr(struct ib_pd *pd, mr->mmkey.size = umem->length; mr->mmkey.pd = to_mpd(pd)->pdn; mr->page_shift = order_base_2(page_size); - mr->umem = umem; set_mr_fields(dev, mr, umem->length, access_flags); return mr; @@ -1201,8 +1219,10 @@ int mlx5_ib_update_xlt(struct mlx5_ib_mr *mr, u64 idx, int npages, /* * Send the DMA list to the HW for a normal MR using UMR. + * Dmabuf MR is handled in a similar way, except that the MLX5_IB_UPD_XLT_ZAP + * flag may be used. */ -static int mlx5_ib_update_mr_pas(struct mlx5_ib_mr *mr, unsigned int flags) +int mlx5_ib_update_mr_pas(struct mlx5_ib_mr *mr, unsigned int flags) { struct mlx5_ib_dev *dev = mr_to_mdev(mr); struct device *ddev = &dev->mdev->pdev->dev; @@ -1244,6 +1264,10 @@ static int mlx5_ib_update_mr_pas(struct mlx5_ib_mr *mr, unsigned int flags) cur_mtt->ptag = cpu_to_be64(rdma_block_iter_dma_address(&biter) | MLX5_IB_MTT_PRESENT); + + if (mr->umem->is_dmabuf && (flags & MLX5_IB_UPD_XLT_ZAP)) + cur_mtt->ptag = 0; + cur_mtt++; } @@ -1528,10 +1552,7 @@ static struct ib_mr *create_user_odp_mr(struct ib_pd *pd, u64 start, u64 length, } odp->private = mr; - init_waitqueue_head(&mr->q_deferred_work); - atomic_set(&mr->num_deferred_work, 0); - err = xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(mr->mmkey.key), - &mr->mmkey, GFP_KERNEL)); + err = mlx5r_store_odp_mkey(dev, &mr->mmkey); if (err) goto err_dereg_mr; @@ -1567,6 +1588,81 @@ struct ib_mr *mlx5_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, return create_real_mr(pd, umem, iova, access_flags); } +static void mlx5_ib_dmabuf_invalidate_cb(struct dma_buf_attachment *attach) +{ + struct ib_umem_dmabuf *umem_dmabuf = attach->importer_priv; + struct mlx5_ib_mr *mr = umem_dmabuf->private; + + dma_resv_assert_held(umem_dmabuf->attach->dmabuf->resv); + + if (!umem_dmabuf->sgt) + return; + + mlx5_ib_update_mr_pas(mr, MLX5_IB_UPD_XLT_ZAP); + ib_umem_dmabuf_unmap_pages(umem_dmabuf); +} + +static struct dma_buf_attach_ops mlx5_ib_dmabuf_attach_ops = { + .allow_peer2peer = 1, + .move_notify = mlx5_ib_dmabuf_invalidate_cb, +}; + +struct ib_mr *mlx5_ib_reg_user_mr_dmabuf(struct ib_pd *pd, u64 offset, + u64 length, u64 virt_addr, + int fd, int access_flags, + struct ib_udata *udata) +{ + struct mlx5_ib_dev *dev = to_mdev(pd->device); + struct mlx5_ib_mr *mr = NULL; + struct ib_umem_dmabuf *umem_dmabuf; + int err; + + if (!IS_ENABLED(CONFIG_INFINIBAND_USER_MEM) || + !IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) + return ERR_PTR(-EOPNOTSUPP); + + mlx5_ib_dbg(dev, + "offset 0x%llx, virt_addr 0x%llx, length 0x%llx, fd %d, access_flags 0x%x\n", + offset, virt_addr, length, fd, access_flags); + + /* dmabuf requires xlt update via umr to work. */ + if (!mlx5_ib_can_load_pas_with_umr(dev, length)) + return ERR_PTR(-EINVAL); + + umem_dmabuf = ib_umem_dmabuf_get(&dev->ib_dev, offset, length, fd, + access_flags, + &mlx5_ib_dmabuf_attach_ops); + if (IS_ERR(umem_dmabuf)) { + mlx5_ib_dbg(dev, "umem_dmabuf get failed (%ld)\n", + PTR_ERR(umem_dmabuf)); + return ERR_CAST(umem_dmabuf); + } + + mr = alloc_cacheable_mr(pd, &umem_dmabuf->umem, virt_addr, + access_flags); + if (IS_ERR(mr)) { + ib_umem_release(&umem_dmabuf->umem); + return ERR_CAST(mr); + } + + mlx5_ib_dbg(dev, "mkey 0x%x\n", mr->mmkey.key); + + atomic_add(ib_umem_num_pages(mr->umem), &dev->mdev->priv.reg_pages); + umem_dmabuf->private = mr; + err = mlx5r_store_odp_mkey(dev, &mr->mmkey); + if (err) + goto err_dereg_mr; + + err = mlx5_ib_init_dmabuf_mr(mr); + if (err) + goto err_dereg_mr; + return &mr->ibmr; + +err_dereg_mr: + dereg_mr(dev, mr); + return ERR_PTR(err); +} + /** * mlx5_mr_cache_invalidate - Fence all DMA on the MR * @mr: The MR to fence @@ -1740,8 +1836,8 @@ struct ib_mr *mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start, return ERR_PTR(err); return NULL; } - /* DM or ODP MR's don't have a umem so we can't re-use it */ - if (!mr->umem || is_odp_mr(mr)) + /* DM or ODP MR's don't have a normal umem so we can't re-use it */ + if (!mr->umem || is_odp_mr(mr) || is_dmabuf_mr(mr)) goto recreate; /* @@ -1760,10 +1856,10 @@ struct ib_mr *mlx5_ib_rereg_user_mr(struct ib_mr *ib_mr, int flags, u64 start, } /* - * DM doesn't have a PAS list so we can't re-use it, odp does but the - * logic around releasing the umem is different + * DM doesn't have a PAS list so we can't re-use it, odp/dmabuf does + * but the logic around releasing the umem is different */ - if (!mr->umem || is_odp_mr(mr)) + if (!mr->umem || is_odp_mr(mr) || is_dmabuf_mr(mr)) goto recreate; if (!(new_access_flags & IB_ACCESS_ON_DEMAND) && @@ -1876,6 +1972,8 @@ static void dereg_mr(struct mlx5_ib_dev *dev, struct mlx5_ib_mr *mr) /* Stop all DMA */ if (is_odp_mr(mr)) mlx5_ib_fence_odp_mr(mr); + else if (is_dmabuf_mr(mr)) + mlx5_ib_fence_dmabuf_mr(mr); else clean_mr(dev, mr); @@ -2227,9 +2325,7 @@ int mlx5_ib_alloc_mw(struct ib_mw *ibmw, struct ib_udata *udata) } if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) { - err = xa_err(xa_store(&dev->odp_mkeys, - mlx5_base_mkey(mw->mmkey.key), &mw->mmkey, - GFP_KERNEL)); + err = mlx5r_store_odp_mkey(dev, &mw->mmkey); if (err) goto free_mkey; } @@ -2249,14 +2345,13 @@ int mlx5_ib_dealloc_mw(struct ib_mw *mw) struct mlx5_ib_dev *dev = to_mdev(mw->device); struct mlx5_ib_mw *mmw = to_mmw(mw); - if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING)) { - xa_erase(&dev->odp_mkeys, mlx5_base_mkey(mmw->mmkey.key)); + if (IS_ENABLED(CONFIG_INFINIBAND_ON_DEMAND_PAGING) && + xa_erase(&dev->odp_mkeys, mlx5_base_mkey(mmw->mmkey.key))) /* - * pagefault_single_data_segment() may be accessing mmw under - * SRCU if the user bound an ODP MR to this MW. + * pagefault_single_data_segment() may be accessing mmw + * if the user bound an ODP MR to this MW. */ - synchronize_srcu(&dev->odp_srcu); - } + mlx5r_deref_wait_odp_mkey(&mmw->mmkey); return mlx5_core_destroy_mkey(dev->mdev, &mmw->mmkey); } diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index aa2413b50adc..374698186662 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -33,6 +33,8 @@ #include <rdma/ib_umem.h> #include <rdma/ib_umem_odp.h> #include <linux/kernel.h> +#include <linux/dma-buf.h> +#include <linux/dma-resv.h> #include "mlx5_ib.h" #include "cmd.h" @@ -113,7 +115,6 @@ static void populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries, * xarray would be protected by the umem_mutex, however that is not * possible. Instead this uses a weaker update-then-lock pattern: * - * srcu_read_lock() * xa_store() * mutex_lock(umem_mutex) * mlx5_ib_update_xlt() @@ -124,12 +125,9 @@ static void populate_klm(struct mlx5_klm *pklm, size_t idx, size_t nentries, * before destroying. * * The umem_mutex provides the acquire/release semantic needed to make - * the xa_store() visible to a racing thread. While SRCU is not - * technically required, using it gives consistent use of the SRCU - * locking around the xarray. + * the xa_store() visible to a racing thread. */ lockdep_assert_held(&to_ib_umem_odp(imr->umem)->umem_mutex); - lockdep_assert_held(&mr_to_mdev(imr)->odp_srcu); for (; pklm != end; pklm++, idx++) { struct mlx5_ib_mr *mtt = xa_load(&imr->implicit_children, idx); @@ -205,8 +203,8 @@ static void dma_fence_odp_mr(struct mlx5_ib_mr *mr) } /* - * This must be called after the mr has been removed from implicit_children - * and the SRCU synchronized. NOTE: The MR does not necessarily have to be + * This must be called after the mr has been removed from implicit_children. + * NOTE: The MR does not necessarily have to be * empty here, parallel page faults could have raced with the free process and * added pages to it. */ @@ -216,19 +214,15 @@ static void free_implicit_child_mr(struct mlx5_ib_mr *mr, bool need_imr_xlt) struct ib_umem_odp *odp_imr = to_ib_umem_odp(imr->umem); struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem); unsigned long idx = ib_umem_start(odp) >> MLX5_IMR_MTT_SHIFT; - int srcu_key; - /* implicit_child_mr's are not allowed to have deferred work */ - WARN_ON(atomic_read(&mr->num_deferred_work)); + mlx5r_deref_wait_odp_mkey(&mr->mmkey); if (need_imr_xlt) { - srcu_key = srcu_read_lock(&mr_to_mdev(mr)->odp_srcu); mutex_lock(&odp_imr->umem_mutex); mlx5_ib_update_xlt(mr->parent, idx, 1, 0, MLX5_IB_UPD_XLT_INDIRECT | MLX5_IB_UPD_XLT_ATOMIC); mutex_unlock(&odp_imr->umem_mutex); - srcu_read_unlock(&mr_to_mdev(mr)->odp_srcu, srcu_key); } dma_fence_odp_mr(mr); @@ -236,26 +230,16 @@ static void free_implicit_child_mr(struct mlx5_ib_mr *mr, bool need_imr_xlt) mr->parent = NULL; mlx5_mr_cache_free(mr_to_mdev(mr), mr); ib_umem_odp_release(odp); - if (atomic_dec_and_test(&imr->num_deferred_work)) - wake_up(&imr->q_deferred_work); } static void free_implicit_child_mr_work(struct work_struct *work) { struct mlx5_ib_mr *mr = container_of(work, struct mlx5_ib_mr, odp_destroy.work); + struct mlx5_ib_mr *imr = mr->parent; free_implicit_child_mr(mr, true); -} - -static void free_implicit_child_mr_rcu(struct rcu_head *head) -{ - struct mlx5_ib_mr *mr = - container_of(head, struct mlx5_ib_mr, odp_destroy.rcu); - - /* Freeing a MR is a sleeping operation, so bounce to a work queue */ - INIT_WORK(&mr->odp_destroy.work, free_implicit_child_mr_work); - queue_work(system_unbound_wq, &mr->odp_destroy.work); + mlx5r_deref_odp_mkey(&imr->mmkey); } static void destroy_unused_implicit_child_mr(struct mlx5_ib_mr *mr) @@ -264,21 +248,14 @@ static void destroy_unused_implicit_child_mr(struct mlx5_ib_mr *mr) unsigned long idx = ib_umem_start(odp) >> MLX5_IMR_MTT_SHIFT; struct mlx5_ib_mr *imr = mr->parent; - xa_lock(&imr->implicit_children); - /* - * This can race with mlx5_ib_free_implicit_mr(), the first one to - * reach the xa lock wins the race and destroys the MR. - */ - if (__xa_cmpxchg(&imr->implicit_children, idx, mr, NULL, GFP_ATOMIC) != - mr) - goto out_unlock; + if (!refcount_inc_not_zero(&imr->mmkey.usecount)) + return; - atomic_inc(&imr->num_deferred_work); - call_srcu(&mr_to_mdev(mr)->odp_srcu, &mr->odp_destroy.rcu, - free_implicit_child_mr_rcu); + xa_erase(&imr->implicit_children, idx); -out_unlock: - xa_unlock(&imr->implicit_children); + /* Freeing a MR is a sleeping operation, so bounce to a work queue */ + INIT_WORK(&mr->odp_destroy.work, free_implicit_child_mr_work); + queue_work(system_unbound_wq, &mr->odp_destroy.work); } static bool mlx5_ib_invalidate_range(struct mmu_interval_notifier *mni, @@ -490,6 +467,12 @@ static struct mlx5_ib_mr *implicit_get_child_mr(struct mlx5_ib_mr *imr, mr->parent = imr; odp->private = mr; + /* + * First refcount is owned by the xarray and second refconut + * is returned to the caller. + */ + refcount_set(&mr->mmkey.usecount, 2); + err = mlx5_ib_update_xlt(mr, 0, MLX5_IMR_MTT_ENTRIES, PAGE_SHIFT, @@ -500,27 +483,28 @@ static struct mlx5_ib_mr *implicit_get_child_mr(struct mlx5_ib_mr *imr, goto out_mr; } - /* - * Once the store to either xarray completes any error unwind has to - * use synchronize_srcu(). Avoid this with xa_reserve() - */ - ret = xa_cmpxchg(&imr->implicit_children, idx, NULL, mr, - GFP_KERNEL); + xa_lock(&imr->implicit_children); + ret = __xa_cmpxchg(&imr->implicit_children, idx, NULL, mr, + GFP_KERNEL); if (unlikely(ret)) { if (xa_is_err(ret)) { ret = ERR_PTR(xa_err(ret)); - goto out_mr; + goto out_lock; } /* * Another thread beat us to creating the child mr, use * theirs. */ - goto out_mr; + refcount_inc(&ret->mmkey.usecount); + goto out_lock; } + xa_unlock(&imr->implicit_children); mlx5_ib_dbg(mr_to_mdev(imr), "key %x mr %p\n", mr->mmkey.key, mr); return mr; +out_lock: + xa_unlock(&imr->implicit_children); out_mr: mlx5_mr_cache_free(mr_to_mdev(imr), mr); out_umem: @@ -559,8 +543,6 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd, imr->ibmr.device = &dev->ib_dev; imr->umem = &umem_odp->umem; imr->is_odp_implicit = true; - atomic_set(&imr->num_deferred_work, 0); - init_waitqueue_head(&imr->q_deferred_work); xa_init(&imr->implicit_children); err = mlx5_ib_update_xlt(imr, 0, @@ -572,8 +554,7 @@ struct mlx5_ib_mr *mlx5_ib_alloc_implicit_mr(struct mlx5_ib_pd *pd, if (err) goto out_mr; - err = xa_err(xa_store(&dev->odp_mkeys, mlx5_base_mkey(imr->mmkey.key), - &imr->mmkey, GFP_KERNEL)); + err = mlx5r_store_odp_mkey(dev, &imr->mmkey); if (err) goto out_mr; @@ -591,60 +572,35 @@ void mlx5_ib_free_implicit_mr(struct mlx5_ib_mr *imr) { struct ib_umem_odp *odp_imr = to_ib_umem_odp(imr->umem); struct mlx5_ib_dev *dev = mr_to_mdev(imr); - struct list_head destroy_list; struct mlx5_ib_mr *mtt; - struct mlx5_ib_mr *tmp; unsigned long idx; - INIT_LIST_HEAD(&destroy_list); - xa_erase(&dev->odp_mkeys, mlx5_base_mkey(imr->mmkey.key)); /* - * This stops the SRCU protected page fault path from touching either - * the imr or any children. The page fault path can only reach the - * children xarray via the imr. - */ - synchronize_srcu(&dev->odp_srcu); - - /* * All work on the prefetch list must be completed, xa_erase() prevented * new work from being created. */ - wait_event(imr->q_deferred_work, !atomic_read(&imr->num_deferred_work)); - + mlx5r_deref_wait_odp_mkey(&imr->mmkey); /* * At this point it is forbidden for any other thread to enter * pagefault_mr() on this imr. It is already forbidden to call * pagefault_mr() on an implicit child. Due to this additions to * implicit_children are prevented. + * In addition, any new call to destroy_unused_implicit_child_mr() + * may return immediately. */ /* - * Block destroy_unused_implicit_child_mr() from incrementing - * num_deferred_work. - */ - xa_lock(&imr->implicit_children); - xa_for_each (&imr->implicit_children, idx, mtt) { - __xa_erase(&imr->implicit_children, idx); - list_add(&mtt->odp_destroy.elm, &destroy_list); - } - xa_unlock(&imr->implicit_children); - - /* - * Wait for any concurrent destroy_unused_implicit_child_mr() to - * complete. - */ - wait_event(imr->q_deferred_work, !atomic_read(&imr->num_deferred_work)); - - /* * Fence the imr before we destroy the children. This allows us to * skip updating the XLT of the imr during destroy of the child mkey * the imr points to. */ mlx5_mr_cache_invalidate(imr); - list_for_each_entry_safe (mtt, tmp, &destroy_list, odp_destroy.elm) + xa_for_each(&imr->implicit_children, idx, mtt) { + xa_erase(&imr->implicit_children, idx); free_implicit_child_mr(mtt, false); + } mlx5_mr_cache_free(dev, imr); ib_umem_odp_release(odp_imr); @@ -663,13 +619,39 @@ void mlx5_ib_fence_odp_mr(struct mlx5_ib_mr *mr) xa_erase(&mr_to_mdev(mr)->odp_mkeys, mlx5_base_mkey(mr->mmkey.key)); /* Wait for all running page-fault handlers to finish. */ - synchronize_srcu(&mr_to_mdev(mr)->odp_srcu); - - wait_event(mr->q_deferred_work, !atomic_read(&mr->num_deferred_work)); + mlx5r_deref_wait_odp_mkey(&mr->mmkey); dma_fence_odp_mr(mr); } +/** + * mlx5_ib_fence_dmabuf_mr - Stop all access to the dmabuf MR + * @mr: to fence + * + * On return no parallel threads will be touching this MR and no DMA will be + * active. + */ +void mlx5_ib_fence_dmabuf_mr(struct mlx5_ib_mr *mr) +{ + struct ib_umem_dmabuf *umem_dmabuf = to_ib_umem_dmabuf(mr->umem); + + /* Prevent new page faults and prefetch requests from succeeding */ + xa_erase(&mr_to_mdev(mr)->odp_mkeys, mlx5_base_mkey(mr->mmkey.key)); + + mlx5r_deref_wait_odp_mkey(&mr->mmkey); + + dma_resv_lock(umem_dmabuf->attach->dmabuf->resv, NULL); + mlx5_mr_cache_invalidate(mr); + umem_dmabuf->private = NULL; + ib_umem_dmabuf_unmap_pages(umem_dmabuf); + dma_resv_unlock(umem_dmabuf->attach->dmabuf->resv); + + if (!mr->cache_ent) { + mlx5_core_destroy_mkey(mr_to_mdev(mr)->mdev, &mr->mmkey); + WARN_ON(mr->descs); + } +} + #define MLX5_PF_FLAGS_DOWNGRADE BIT(1) #define MLX5_PF_FLAGS_SNAPSHOT BIT(2) #define MLX5_PF_FLAGS_ENABLE BIT(3) @@ -747,8 +729,10 @@ static int pagefault_implicit_mr(struct mlx5_ib_mr *imr, struct mlx5_ib_mr *mtt; u64 len; + xa_lock(&imr->implicit_children); mtt = xa_load(&imr->implicit_children, idx); if (unlikely(!mtt)) { + xa_unlock(&imr->implicit_children); mtt = implicit_get_child_mr(imr, idx); if (IS_ERR(mtt)) { ret = PTR_ERR(mtt); @@ -756,6 +740,9 @@ static int pagefault_implicit_mr(struct mlx5_ib_mr *imr, } upd_start_idx = min(upd_start_idx, idx); upd_len = idx - upd_start_idx + 1; + } else { + refcount_inc(&mtt->mmkey.usecount); + xa_unlock(&imr->implicit_children); } umem_odp = to_ib_umem_odp(mtt->umem); @@ -764,6 +751,9 @@ static int pagefault_implicit_mr(struct mlx5_ib_mr *imr, ret = pagefault_real_mr(mtt, umem_odp, user_va, len, bytes_mapped, flags); + + mlx5r_deref_odp_mkey(&mtt->mmkey); + if (ret < 0) goto out; user_va += len; @@ -803,6 +793,44 @@ out: return ret; } +static int pagefault_dmabuf_mr(struct mlx5_ib_mr *mr, size_t bcnt, + u32 *bytes_mapped, u32 flags) +{ + struct ib_umem_dmabuf *umem_dmabuf = to_ib_umem_dmabuf(mr->umem); + u32 xlt_flags = 0; + int err; + unsigned int page_size; + + if (flags & MLX5_PF_FLAGS_ENABLE) + xlt_flags |= MLX5_IB_UPD_XLT_ENABLE; + + dma_resv_lock(umem_dmabuf->attach->dmabuf->resv, NULL); + err = ib_umem_dmabuf_map_pages(umem_dmabuf); + if (err) { + dma_resv_unlock(umem_dmabuf->attach->dmabuf->resv); + return err; + } + + page_size = mlx5_umem_find_best_pgsz(&umem_dmabuf->umem, mkc, + log_page_size, 0, + umem_dmabuf->umem.iova); + if (unlikely(page_size < PAGE_SIZE)) { + ib_umem_dmabuf_unmap_pages(umem_dmabuf); + err = -EINVAL; + } else { + err = mlx5_ib_update_mr_pas(mr, xlt_flags); + } + dma_resv_unlock(umem_dmabuf->attach->dmabuf->resv); + + if (err) + return err; + + if (bytes_mapped) + *bytes_mapped += bcnt; + + return ib_umem_num_pages(mr->umem); +} + /* * Returns: * -EFAULT: The io_virt->bcnt is not within the MR, it covers pages that are @@ -817,10 +845,12 @@ static int pagefault_mr(struct mlx5_ib_mr *mr, u64 io_virt, size_t bcnt, { struct ib_umem_odp *odp = to_ib_umem_odp(mr->umem); - lockdep_assert_held(&mr_to_mdev(mr)->odp_srcu); if (unlikely(io_virt < mr->mmkey.iova)) return -EFAULT; + if (mr->umem->is_dmabuf) + return pagefault_dmabuf_mr(mr, bcnt, bytes_mapped, flags); + if (!odp->is_implicit_odp) { u64 user_va; @@ -847,6 +877,16 @@ int mlx5_ib_init_odp_mr(struct mlx5_ib_mr *mr) return ret >= 0 ? 0 : ret; } +int mlx5_ib_init_dmabuf_mr(struct mlx5_ib_mr *mr) +{ + int ret; + + ret = pagefault_dmabuf_mr(mr, mr->umem->length, NULL, + MLX5_PF_FLAGS_ENABLE); + + return ret >= 0 ? 0 : ret; +} + struct pf_frame { struct pf_frame *next; u32 key; @@ -896,7 +936,7 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *dev, u32 *bytes_committed, u32 *bytes_mapped) { - int npages = 0, srcu_key, ret, i, outlen, cur_outlen = 0, depth = 0; + int npages = 0, ret, i, outlen, cur_outlen = 0, depth = 0; struct pf_frame *head = NULL, *frame; struct mlx5_core_mkey *mmkey; struct mlx5_ib_mr *mr; @@ -905,14 +945,14 @@ static int pagefault_single_data_segment(struct mlx5_ib_dev *dev, size_t offset; int ndescs; - srcu_key = srcu_read_lock(&dev->odp_srcu); - io_virt += *bytes_committed; bcnt -= *bytes_committed; next_mr: + xa_lock(&dev->odp_mkeys); mmkey = xa_load(&dev->odp_mkeys, mlx5_base_mkey(key)); if (!mmkey) { + xa_unlock(&dev->odp_mkeys); mlx5_ib_dbg( dev, "skipping non ODP MR (lkey=0x%06x) in page fault handler.\n", @@ -925,12 +965,15 @@ next_mr: * faulted. */ ret = 0; - goto srcu_unlock; + goto end; } + refcount_inc(&mmkey->usecount); + xa_unlock(&dev->odp_mkeys); + if (!mkey_is_eq(mmkey, key)) { mlx5_ib_dbg(dev, "failed to find mkey %x\n", key); ret = -EFAULT; - goto srcu_unlock; + goto end; } switch (mmkey->type) { @@ -939,7 +982,7 @@ next_mr: ret = pagefault_mr(mr, io_virt, bcnt, bytes_mapped, 0); if (ret < 0) - goto srcu_unlock; + goto end; mlx5_update_odp_stats(mr, faults, ret); @@ -954,7 +997,7 @@ next_mr: if (depth >= MLX5_CAP_GEN(dev->mdev, max_indirection)) { mlx5_ib_dbg(dev, "indirection level exceeded\n"); ret = -EFAULT; - goto srcu_unlock; + goto end; } outlen = MLX5_ST_SZ_BYTES(query_mkey_out) + @@ -965,7 +1008,7 @@ next_mr: out = kzalloc(outlen, GFP_KERNEL); if (!out) { ret = -ENOMEM; - goto srcu_unlock; + goto end; } cur_outlen = outlen; } @@ -975,7 +1018,7 @@ next_mr: ret = mlx5_core_query_mkey(dev->mdev, mmkey, out, outlen); if (ret) - goto srcu_unlock; + goto end; offset = io_virt - MLX5_GET64(query_mkey_out, out, memory_key_mkey_entry.start_addr); @@ -989,7 +1032,7 @@ next_mr: frame = kzalloc(sizeof(*frame), GFP_KERNEL); if (!frame) { ret = -ENOMEM; - goto srcu_unlock; + goto end; } frame->key = be32_to_cpu(pklm->key); @@ -1008,7 +1051,7 @@ next_mr: default: mlx5_ib_dbg(dev, "wrong mkey type %d\n", mmkey->type); ret = -EFAULT; - goto srcu_unlock; + goto end; } if (head) { @@ -1021,10 +1064,13 @@ next_mr: depth = frame->depth; kfree(frame); + mlx5r_deref_odp_mkey(mmkey); goto next_mr; } -srcu_unlock: +end: + if (mmkey) + mlx5r_deref_odp_mkey(mmkey); while (head) { frame = head; head = frame->next; @@ -1032,7 +1078,6 @@ srcu_unlock: } kfree(out); - srcu_read_unlock(&dev->odp_srcu, srcu_key); *bytes_committed = 0; return ret ? ret : npages; } @@ -1040,16 +1085,18 @@ srcu_unlock: /** * Parse a series of data segments for page fault handling. * - * @pfault contains page fault information. - * @wqe points at the first data segment in the WQE. - * @wqe_end points after the end of the WQE. - * @bytes_mapped receives the number of bytes that the function was able to - * map. This allows the caller to decide intelligently whether - * enough memory was mapped to resolve the page fault - * successfully (e.g. enough for the next MTU, or the entire - * WQE). - * @total_wqe_bytes receives the total data size of this WQE in bytes (minus - * the committed bytes). + * @dev: Pointer to mlx5 IB device + * @pfault: contains page fault information. + * @wqe: points at the first data segment in the WQE. + * @wqe_end: points after the end of the WQE. + * @bytes_mapped: receives the number of bytes that the function was able to + * map. This allows the caller to decide intelligently whether + * enough memory was mapped to resolve the page fault + * successfully (e.g. enough for the next MTU, or the entire + * WQE). + * @total_wqe_bytes: receives the total data size of this WQE in bytes (minus + * the committed bytes). + * @receive_queue: receive WQE end of sg list * * Returns the number of pages loaded if positive, zero for an empty WQE, or a * negative error code. @@ -1738,8 +1785,8 @@ static void destroy_prefetch_work(struct prefetch_mr_work *work) u32 i; for (i = 0; i < work->num_sge; ++i) - if (atomic_dec_and_test(&work->frags[i].mr->num_deferred_work)) - wake_up(&work->frags[i].mr->q_deferred_work); + mlx5r_deref_odp_mkey(&work->frags[i].mr->mmkey); + kvfree(work); } @@ -1749,27 +1796,30 @@ get_prefetchable_mr(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, { struct mlx5_ib_dev *dev = to_mdev(pd->device); struct mlx5_core_mkey *mmkey; - struct ib_umem_odp *odp; - struct mlx5_ib_mr *mr; - - lockdep_assert_held(&dev->odp_srcu); + struct mlx5_ib_mr *mr = NULL; + xa_lock(&dev->odp_mkeys); mmkey = xa_load(&dev->odp_mkeys, mlx5_base_mkey(lkey)); if (!mmkey || mmkey->key != lkey || mmkey->type != MLX5_MKEY_MR) - return NULL; + goto end; mr = container_of(mmkey, struct mlx5_ib_mr, mmkey); - if (mr->ibmr.pd != pd) - return NULL; - - odp = to_ib_umem_odp(mr->umem); + if (mr->ibmr.pd != pd) { + mr = NULL; + goto end; + } /* prefetch with write-access must be supported by the MR */ if (advice == IB_UVERBS_ADVISE_MR_ADVICE_PREFETCH_WRITE && - !odp->umem.writable) - return NULL; + !mr->umem->writable) { + mr = NULL; + goto end; + } + refcount_inc(&mmkey->usecount); +end: + xa_unlock(&dev->odp_mkeys); return mr; } @@ -1777,17 +1827,12 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *w) { struct prefetch_mr_work *work = container_of(w, struct prefetch_mr_work, work); - struct mlx5_ib_dev *dev; u32 bytes_mapped = 0; - int srcu_key; int ret; u32 i; /* We rely on IB/core that work is executed if we have num_sge != 0 only. */ WARN_ON(!work->num_sge); - dev = mr_to_mdev(work->frags[0].mr); - /* SRCU should be held when calling to mlx5_odp_populate_xlt() */ - srcu_key = srcu_read_lock(&dev->odp_srcu); for (i = 0; i < work->num_sge; ++i) { ret = pagefault_mr(work->frags[i].mr, work->frags[i].io_virt, work->frags[i].length, &bytes_mapped, @@ -1796,7 +1841,6 @@ static void mlx5_ib_prefetch_mr_work(struct work_struct *w) continue; mlx5_update_odp_stats(work->frags[i].mr, prefetch, ret); } - srcu_read_unlock(&dev->odp_srcu, srcu_key); destroy_prefetch_work(work); } @@ -1820,9 +1864,6 @@ static bool init_prefetch_work(struct ib_pd *pd, work->num_sge = i; return false; } - - /* Keep the MR pointer will valid outside the SRCU */ - atomic_inc(&work->frags[i].mr->num_deferred_work); } work->num_sge = num_sge; return true; @@ -1833,42 +1874,35 @@ static int mlx5_ib_prefetch_sg_list(struct ib_pd *pd, u32 pf_flags, struct ib_sge *sg_list, u32 num_sge) { - struct mlx5_ib_dev *dev = to_mdev(pd->device); u32 bytes_mapped = 0; - int srcu_key; int ret = 0; u32 i; - srcu_key = srcu_read_lock(&dev->odp_srcu); for (i = 0; i < num_sge; ++i) { struct mlx5_ib_mr *mr; mr = get_prefetchable_mr(pd, advice, sg_list[i].lkey); - if (!mr) { - ret = -ENOENT; - goto out; - } + if (!mr) + return -ENOENT; ret = pagefault_mr(mr, sg_list[i].addr, sg_list[i].length, &bytes_mapped, pf_flags); - if (ret < 0) - goto out; + if (ret < 0) { + mlx5r_deref_odp_mkey(&mr->mmkey); + return ret; + } mlx5_update_odp_stats(mr, prefetch, ret); + mlx5r_deref_odp_mkey(&mr->mmkey); } - ret = 0; -out: - srcu_read_unlock(&dev->odp_srcu, srcu_key); - return ret; + return 0; } int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, enum ib_uverbs_advise_mr_advice advice, u32 flags, struct ib_sge *sg_list, u32 num_sge) { - struct mlx5_ib_dev *dev = to_mdev(pd->device); u32 pf_flags = 0; struct prefetch_mr_work *work; - int srcu_key; if (advice == IB_UVERBS_ADVISE_MR_ADVICE_PREFETCH) pf_flags |= MLX5_PF_FLAGS_DOWNGRADE; @@ -1884,13 +1918,10 @@ int mlx5_ib_advise_mr_prefetch(struct ib_pd *pd, if (!work) return -ENOMEM; - srcu_key = srcu_read_lock(&dev->odp_srcu); if (!init_prefetch_work(pd, advice, pf_flags, work, sg_list, num_sge)) { - srcu_read_unlock(&dev->odp_srcu, srcu_key); destroy_prefetch_work(work); return -EINVAL; } queue_work(system_unbound_wq, &work->work); - srcu_read_unlock(&dev->odp_srcu, srcu_key); return 0; } diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 0cb7cc642d87..ec4b3f6a8222 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -1078,6 +1078,7 @@ static int _create_kernel_qp(struct mlx5_ib_dev *dev, qpc = MLX5_ADDR_OF(create_qp_in, *in, qpc); MLX5_SET(qpc, qpc, uar_page, uar_index); + MLX5_SET(qpc, qpc, ts_format, MLX5_QPC_TIMESTAMP_FORMAT_DEFAULT); MLX5_SET(qpc, qpc, log_page_size, qp->buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT); /* Set "fast registration enabled" for all kernel QPs */ @@ -1172,10 +1173,72 @@ static void destroy_flow_rule_vport_sq(struct mlx5_ib_sq *sq) sq->flow_rule = NULL; } +static int get_rq_ts_format(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *send_cq) +{ + bool fr_supported = + MLX5_CAP_GEN(dev->mdev, rq_ts_format) == + MLX5_RQ_TIMESTAMP_FORMAT_CAP_FREE_RUNNING || + MLX5_CAP_GEN(dev->mdev, rq_ts_format) == + MLX5_RQ_TIMESTAMP_FORMAT_CAP_FREE_RUNNING_AND_REAL_TIME; + + if (send_cq->create_flags & IB_UVERBS_CQ_FLAGS_TIMESTAMP_COMPLETION) { + if (!fr_supported) { + mlx5_ib_dbg(dev, "Free running TS format is not supported\n"); + return -EOPNOTSUPP; + } + return MLX5_RQC_TIMESTAMP_FORMAT_FREE_RUNNING; + } + return MLX5_RQC_TIMESTAMP_FORMAT_DEFAULT; +} + +static int get_sq_ts_format(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *send_cq) +{ + bool fr_supported = + MLX5_CAP_GEN(dev->mdev, sq_ts_format) == + MLX5_SQ_TIMESTAMP_FORMAT_CAP_FREE_RUNNING || + MLX5_CAP_GEN(dev->mdev, sq_ts_format) == + MLX5_SQ_TIMESTAMP_FORMAT_CAP_FREE_RUNNING_AND_REAL_TIME; + + if (send_cq->create_flags & IB_UVERBS_CQ_FLAGS_TIMESTAMP_COMPLETION) { + if (!fr_supported) { + mlx5_ib_dbg(dev, "Free running TS format is not supported\n"); + return -EOPNOTSUPP; + } + return MLX5_SQC_TIMESTAMP_FORMAT_FREE_RUNNING; + } + return MLX5_SQC_TIMESTAMP_FORMAT_DEFAULT; +} + +static int get_qp_ts_format(struct mlx5_ib_dev *dev, struct mlx5_ib_cq *send_cq, + struct mlx5_ib_cq *recv_cq) +{ + bool fr_supported = + MLX5_CAP_ROCE(dev->mdev, qp_ts_format) == + MLX5_QP_TIMESTAMP_FORMAT_CAP_FREE_RUNNING || + MLX5_CAP_ROCE(dev->mdev, qp_ts_format) == + MLX5_QP_TIMESTAMP_FORMAT_CAP_FREE_RUNNING_AND_REAL_TIME; + int ts_format = MLX5_QPC_TIMESTAMP_FORMAT_DEFAULT; + + if (recv_cq && + recv_cq->create_flags & IB_UVERBS_CQ_FLAGS_TIMESTAMP_COMPLETION) + ts_format = MLX5_QPC_TIMESTAMP_FORMAT_FREE_RUNNING; + + if (send_cq && + send_cq->create_flags & IB_UVERBS_CQ_FLAGS_TIMESTAMP_COMPLETION) + ts_format = MLX5_QPC_TIMESTAMP_FORMAT_FREE_RUNNING; + + if (ts_format == MLX5_QPC_TIMESTAMP_FORMAT_FREE_RUNNING && + !fr_supported) { + mlx5_ib_dbg(dev, "Free running TS format is not supported\n"); + return -EOPNOTSUPP; + } + return ts_format; +} + static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev, struct ib_udata *udata, struct mlx5_ib_sq *sq, void *qpin, - struct ib_pd *pd) + struct ib_pd *pd, struct mlx5_ib_cq *cq) { struct mlx5_ib_ubuffer *ubuffer = &sq->ubuffer; __be64 *pas; @@ -1187,6 +1250,11 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev, int err; unsigned int page_offset_quantized; unsigned long page_size; + int ts_format; + + ts_format = get_sq_ts_format(dev, cq); + if (ts_format < 0) + return ts_format; sq->ubuffer.umem = ib_umem_get(&dev->ib_dev, ubuffer->buf_addr, ubuffer->buf_size, 0); @@ -1215,6 +1283,7 @@ static int create_raw_packet_qp_sq(struct mlx5_ib_dev *dev, if (MLX5_CAP_ETH(dev->mdev, multi_pkt_send_wqe)) MLX5_SET(sqc, sqc, allow_multi_pkt_send_wqe, 1); MLX5_SET(sqc, sqc, state, MLX5_SQC_STATE_RST); + MLX5_SET(sqc, sqc, ts_format, ts_format); MLX5_SET(sqc, sqc, user_index, MLX5_GET(qpc, qpc, user_index)); MLX5_SET(sqc, sqc, cqn, MLX5_GET(qpc, qpc, cqn_snd)); MLX5_SET(sqc, sqc, tis_lst_sz, 1); @@ -1263,7 +1332,7 @@ static void destroy_raw_packet_qp_sq(struct mlx5_ib_dev *dev, static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct mlx5_ib_rq *rq, void *qpin, - struct ib_pd *pd) + struct ib_pd *pd, struct mlx5_ib_cq *cq) { struct mlx5_ib_qp *mqp = rq->base.container_mibqp; __be64 *pas; @@ -1274,9 +1343,14 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, struct ib_umem *umem = rq->base.ubuffer.umem; unsigned int page_offset_quantized; unsigned long page_size = 0; + int ts_format; size_t inlen; int err; + ts_format = get_rq_ts_format(dev, cq); + if (ts_format < 0) + return ts_format; + page_size = mlx5_umem_find_best_quantized_pgoff(umem, wq, log_wq_pg_sz, MLX5_ADAPTER_PAGE_SHIFT, page_offset, 64, @@ -1296,6 +1370,7 @@ static int create_raw_packet_qp_rq(struct mlx5_ib_dev *dev, MLX5_SET(rqc, rqc, vsd, 1); MLX5_SET(rqc, rqc, mem_rq_type, MLX5_RQC_MEM_RQ_TYPE_MEMORY_RQ_INLINE); MLX5_SET(rqc, rqc, state, MLX5_RQC_STATE_RST); + MLX5_SET(rqc, rqc, ts_format, ts_format); MLX5_SET(rqc, rqc, flush_in_error_en, 1); MLX5_SET(rqc, rqc, user_index, MLX5_GET(qpc, qpc, user_index)); MLX5_SET(rqc, rqc, cqn, MLX5_GET(qpc, qpc, cqn_rcv)); @@ -1393,10 +1468,10 @@ static int create_raw_packet_qp_tir(struct mlx5_ib_dev *dev, } static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, - u32 *in, size_t inlen, - struct ib_pd *pd, + u32 *in, size_t inlen, struct ib_pd *pd, struct ib_udata *udata, - struct mlx5_ib_create_qp_resp *resp) + struct mlx5_ib_create_qp_resp *resp, + struct ib_qp_init_attr *init_attr) { struct mlx5_ib_raw_packet_qp *raw_packet_qp = &qp->raw_packet_qp; struct mlx5_ib_sq *sq = &raw_packet_qp->sq; @@ -1415,7 +1490,8 @@ static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, if (err) return err; - err = create_raw_packet_qp_sq(dev, udata, sq, in, pd); + err = create_raw_packet_qp_sq(dev, udata, sq, in, pd, + to_mcq(init_attr->send_cq)); if (err) goto err_destroy_tis; @@ -1437,7 +1513,8 @@ static int create_raw_packet_qp(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, rq->flags |= MLX5_IB_RQ_CVLAN_STRIPPING; if (qp->flags & IB_QP_CREATE_PCI_WRITE_END_PADDING) rq->flags |= MLX5_IB_RQ_PCI_WRITE_END_PADDING; - err = create_raw_packet_qp_rq(dev, rq, in, pd); + err = create_raw_packet_qp_rq(dev, rq, in, pd, + to_mcq(init_attr->recv_cq)); if (err) goto err_destroy_sq; @@ -1907,6 +1984,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, struct mlx5_ib_cq *recv_cq; unsigned long flags; struct mlx5_ib_qp_base *base; + int ts_format; int mlx5_st; void *qpc; u32 *in; @@ -1944,6 +2022,13 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, if (ucmd->sq_wqe_count > (1 << MLX5_CAP_GEN(mdev, log_max_qp_sz))) return -EINVAL; + if (init_attr->qp_type != IB_QPT_RAW_PACKET) { + ts_format = get_qp_ts_format(dev, to_mcq(init_attr->send_cq), + to_mcq(init_attr->recv_cq)); + if (ts_format < 0) + return ts_format; + } + err = _create_user_qp(dev, pd, qp, udata, init_attr, &in, ¶ms->resp, &inlen, base, ucmd); if (err) @@ -1992,6 +2077,9 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, MLX5_SET(qpc, qpc, log_rq_size, ilog2(qp->rq.wqe_cnt)); } + if (init_attr->qp_type != IB_QPT_RAW_PACKET) + MLX5_SET(qpc, qpc, ts_format, ts_format); + MLX5_SET(qpc, qpc, rq_type, get_rx_type(qp, init_attr)); if (qp->sq.wqe_cnt) { @@ -2046,7 +2134,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, qp->raw_packet_qp.sq.ubuffer.buf_addr = ucmd->sq_buf_addr; raw_packet_qp_copy_info(qp, &qp->raw_packet_qp); err = create_raw_packet_qp(dev, qp, in, inlen, pd, udata, - ¶ms->resp); + ¶ms->resp, init_attr); } else err = mlx5_qpc_create_qp(dev, &base->mqp, in, inlen, out); @@ -2432,9 +2520,6 @@ static int check_qp_type(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *attr, case MLX5_IB_QPT_HW_GSI: case IB_QPT_DRIVER: case IB_QPT_GSI: - if (dev->profile == &raw_eth_profile) - goto out; - fallthrough; case IB_QPT_RAW_PACKET: case IB_QPT_UD: case MLX5_IB_QPT_REG_UMR: @@ -2629,10 +2714,6 @@ static int process_create_flags(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, int create_flags = attr->create_flags; bool cond; - if (qp->type == IB_QPT_UD && dev->profile == &raw_eth_profile) - if (create_flags & ~MLX5_IB_QP_CREATE_WC_TEST) - return -EINVAL; - if (qp_type == MLX5_IB_QPT_DCT) return (create_flags) ? -EINVAL : 0; @@ -3076,6 +3157,8 @@ static int ib_to_mlx5_rate_map(u8 rate) return 4; case IB_RATE_50_GBPS: return 5; + case IB_RATE_400_GBPS: + return 6; default: return rate + MLX5_STAT_RATE_OFFSET; } @@ -3183,11 +3266,13 @@ static int mlx5_set_path(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp, alt ? attr->alt_pkey_index : attr->pkey_index); if (ah_flags & IB_AH_GRH) { - if (grh->sgid_index >= - dev->mdev->port_caps[port - 1].gid_table_len) { + const struct ib_port_immutable *immutable; + + immutable = ib_port_immutable_read(&dev->ib_dev, port); + if (grh->sgid_index >= immutable->gid_tbl_len) { pr_err("sgid_index (%u) too large. max is %d\n", grh->sgid_index, - dev->mdev->port_caps[port - 1].gid_table_len); + immutable->gid_tbl_len); return -EINVAL; } } @@ -4211,6 +4296,23 @@ static int mlx5_ib_modify_dct(struct ib_qp *ibqp, struct ib_qp_attr *attr, return 0; } +static bool mlx5_ib_modify_qp_allowed(struct mlx5_ib_dev *dev, + struct mlx5_ib_qp *qp, + enum ib_qp_type qp_type) +{ + if (dev->profile != &raw_eth_profile) + return true; + + if (qp_type == IB_QPT_RAW_PACKET || qp_type == MLX5_IB_QPT_REG_UMR) + return true; + + /* Internal QP used for wc testing, with NOPs in wq */ + if (qp->flags & MLX5_IB_QP_CREATE_WC_TEST) + return true; + + return false; +} + int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, int attr_mask, struct ib_udata *udata) { @@ -4221,7 +4323,9 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, enum ib_qp_type qp_type; enum ib_qp_state cur_state, new_state; int err = -EINVAL; - int port; + + if (!mlx5_ib_modify_qp_allowed(dev, qp, ibqp->qp_type)) + return -EOPNOTSUPP; if (attr_mask & ~(IB_QP_ATTR_STANDARD_BITS | IB_QP_RATE_LIMIT)) return -EOPNOTSUPP; @@ -4263,10 +4367,6 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state; new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state; - if (!(cur_state == new_state && cur_state == IB_QPS_RESET)) { - port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; - } - if (qp->flags & IB_QP_CREATE_SOURCE_QPN) { if (attr_mask & ~(IB_QP_STATE | IB_QP_CUR_STATE)) { mlx5_ib_dbg(dev, "invalid attr_mask 0x%x when underlay QP is used\n", @@ -4295,14 +4395,10 @@ int mlx5_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, goto out; } - if (attr_mask & IB_QP_PKEY_INDEX) { - port = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; - if (attr->pkey_index >= - dev->mdev->port_caps[port - 1].pkey_table_len) { - mlx5_ib_dbg(dev, "invalid pkey index %d\n", - attr->pkey_index); - goto out; - } + if ((attr_mask & IB_QP_PKEY_INDEX) && + attr->pkey_index >= dev->pkey_table_len) { + mlx5_ib_dbg(dev, "invalid pkey index %d\n", attr->pkey_index); + goto out; } if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && @@ -5376,7 +5472,7 @@ void mlx5_ib_drain_rq(struct ib_qp *qp) handle_drain_completion(cq, &rdrain, dev); } -/** +/* * Bind a qp to a counter. If @counter is NULL then bind the qp to * the default counter */ diff --git a/drivers/infiniband/hw/mlx5/wr.c b/drivers/infiniband/hw/mlx5/wr.c index d6038fb6c50c..cf2852cba45c 100644 --- a/drivers/infiniband/hw/mlx5/wr.c +++ b/drivers/infiniband/hw/mlx5/wr.c @@ -1369,7 +1369,7 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, handle_qpt_uc(wr, &seg, &size); break; case IB_QPT_SMI: - if (unlikely(!mdev->port_caps[qp->port - 1].has_smi)) { + if (unlikely(!dev->port_caps[qp->port - 1].has_smi)) { mlx5_ib_warn(dev, "Send SMP MADs is not allowed\n"); err = -EPERM; *bad_wr = wr; diff --git a/drivers/infiniband/hw/qedr/qedr.h b/drivers/infiniband/hw/qedr/qedr.h index 9dde70373a55..3cb4febaad0f 100644 --- a/drivers/infiniband/hw/qedr/qedr.h +++ b/drivers/infiniband/hw/qedr/qedr.h @@ -617,18 +617,18 @@ static inline bool qedr_qp_has_srq(struct qedr_qp *qp) static inline bool qedr_qp_has_sq(struct qedr_qp *qp) { if (qp->qp_type == IB_QPT_GSI || qp->qp_type == IB_QPT_XRC_TGT) - return 0; + return false; - return 1; + return true; } static inline bool qedr_qp_has_rq(struct qedr_qp *qp) { if (qp->qp_type == IB_QPT_GSI || qp->qp_type == IB_QPT_XRC_INI || qp->qp_type == IB_QPT_XRC_TGT || qedr_qp_has_srq(qp)) - return 0; + return false; - return 1; + return true; } static inline struct qedr_user_mmap_entry * diff --git a/drivers/infiniband/hw/qedr/qedr_roce_cm.c b/drivers/infiniband/hw/qedr/qedr_roce_cm.c index f5542d703ef9..13e5e6bbec99 100644 --- a/drivers/infiniband/hw/qedr/qedr_roce_cm.c +++ b/drivers/infiniband/hw/qedr/qedr_roce_cm.c @@ -586,8 +586,8 @@ int qedr_gsi_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, qp->wqe_wr_id[qp->sq.prod].wr_id = wr->wr_id; qedr_inc_sw_prod(&qp->sq); DP_DEBUG(qp->dev, QEDR_MSG_GSI, - "gsi post send: opcode=%d, in_irq=%ld, irqs_disabled=%d, wr_id=%llx\n", - wr->opcode, in_irq(), irqs_disabled(), wr->wr_id); + "gsi post send: opcode=%d, wr_id=%llx\n", wr->opcode, + wr->wr_id); } else { DP_ERR(dev, "gsi post send: failed to transmit (rc=%d)\n", rc); rc = -EAGAIN; diff --git a/drivers/infiniband/hw/qib/qib_driver.c b/drivers/infiniband/hw/qib/qib_driver.c index 92eeea5679e2..84fc4dcc5399 100644 --- a/drivers/infiniband/hw/qib/qib_driver.c +++ b/drivers/infiniband/hw/qib/qib_driver.c @@ -151,7 +151,7 @@ int qib_count_units(int *npresentp, int *nupp) /** * qib_wait_linkstate - wait for an IB link state change to occur - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @state: the state to wait for * @msecs: the number of milliseconds to wait * diff --git a/drivers/infiniband/hw/qib/qib_eeprom.c b/drivers/infiniband/hw/qib/qib_eeprom.c index 5838b3bf34b9..bf660c001b6d 100644 --- a/drivers/infiniband/hw/qib/qib_eeprom.c +++ b/drivers/infiniband/hw/qib/qib_eeprom.c @@ -47,7 +47,7 @@ * qib_eeprom_read - receives bytes from the eeprom via I2C * @dd: the qlogic_ib device * @eeprom_offset: address to read from - * @buffer: where to store result + * @buff: where to store result * @len: number of bytes to receive */ int qib_eeprom_read(struct qib_devdata *dd, u8 eeprom_offset, @@ -94,7 +94,7 @@ static int eeprom_write_with_enable(struct qib_devdata *dd, u8 offset, * qib_eeprom_write - writes data to the eeprom via I2C * @dd: the qlogic_ib device * @eeprom_offset: where to place data - * @buffer: data to write + * @buff: data to write * @len: number of bytes to write */ int qib_eeprom_write(struct qib_devdata *dd, u8 eeprom_offset, diff --git a/drivers/infiniband/hw/qib/qib_iba6120.c b/drivers/infiniband/hw/qib/qib_iba6120.c index 44150be215bf..b35e1174be22 100644 --- a/drivers/infiniband/hw/qib/qib_iba6120.c +++ b/drivers/infiniband/hw/qib/qib_iba6120.c @@ -1223,7 +1223,7 @@ static void qib_set_ib_6120_lstate(struct qib_pportdata *ppd, u16 linkcmd, /** * qib_6120_bringup_serdes - bring up the serdes - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device */ static int qib_6120_bringup_serdes(struct qib_pportdata *ppd) { @@ -1412,7 +1412,7 @@ static void qib_6120_quiet_serdes(struct qib_pportdata *ppd) /** * qib_6120_setup_setextled - set the state of the two external LEDs - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @on: whether the link is up or not * * The exact combo of LEDs if on is true is determined by looking @@ -1823,7 +1823,7 @@ bail: * qib_6120_put_tid - write a TID in chip * @dd: the qlogic_ib device * @tidptr: pointer to the expected TID (in chip) to update - * @tidtype: RCVHQ_RCV_TYPE_EAGER (1) for eager, RCVHQ_RCV_TYPE_EXPECTED (0) + * @type: RCVHQ_RCV_TYPE_EAGER (1) for eager, RCVHQ_RCV_TYPE_EXPECTED (0) * for expected * @pa: physical address of in memory buffer; tidinvalid if freeing * @@ -1890,7 +1890,7 @@ static void qib_6120_put_tid(struct qib_devdata *dd, u64 __iomem *tidptr, * qib_6120_put_tid_2 - write a TID in chip, Revision 2 or higher * @dd: the qlogic_ib device * @tidptr: pointer to the expected TID (in chip) to update - * @tidtype: RCVHQ_RCV_TYPE_EAGER (1) for eager, RCVHQ_RCV_TYPE_EXPECTED (0) + * @type: RCVHQ_RCV_TYPE_EAGER (1) for eager, RCVHQ_RCV_TYPE_EXPECTED (0) * for expected * @pa: physical address of in memory buffer; tidinvalid if freeing * @@ -1932,7 +1932,7 @@ static void qib_6120_put_tid_2(struct qib_devdata *dd, u64 __iomem *tidptr, /** * qib_6120_clear_tids - clear all TID entries for a context, expected and eager * @dd: the qlogic_ib device - * @ctxt: the context + * @rcd: the context * * clear all TID entries for a context, expected and eager. * Used from qib_close(). On this chip, TIDs are only 32 bits, @@ -2008,7 +2008,7 @@ int __attribute__((weak)) qib_unordered_wc(void) /** * qib_6120_get_base_info - set chip-specific flags for user code * @rcd: the qlogic_ib ctxt - * @kbase: qib_base_info pointer + * @kinfo: qib_base_info pointer * * We set the PCIE flag because the lower bandwidth on PCIe vs * HyperTransport can affect some user packet algorithms. @@ -2270,8 +2270,8 @@ static void sendctrl_6120_mod(struct qib_pportdata *ppd, u32 op) /** * qib_portcntr_6120 - read a per-port counter - * @dd: the qlogic_ib device - * @creg: the counter to snapshot + * @ppd: the qlogic_ib device + * @reg: the counter to snapshot */ static u64 qib_portcntr_6120(struct qib_pportdata *ppd, u32 reg) { @@ -2610,7 +2610,7 @@ static void qib_chk_6120_errormask(struct qib_devdata *dd) /** * qib_get_faststats - get word counters from chip before they overflow - * @opaque - contains a pointer to the qlogic_ib device qib_devdata + * @t: contains a pointer to the qlogic_ib device qib_devdata * * This needs more work; in particular, decision on whether we really * need traffic_wds done the way it is diff --git a/drivers/infiniband/hw/qib/qib_iba7220.c b/drivers/infiniband/hw/qib/qib_iba7220.c index 0a6f26d4cb31..229dcd6ead95 100644 --- a/drivers/infiniband/hw/qib/qib_iba7220.c +++ b/drivers/infiniband/hw/qib/qib_iba7220.c @@ -1701,7 +1701,7 @@ static void qib_7220_quiet_serdes(struct qib_pportdata *ppd) /** * qib_setup_7220_setextled - set the state of the two external LEDs - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @on: whether the link is up or not * * The exact combo of LEDs if on is true is determined by looking @@ -2146,7 +2146,7 @@ bail: * qib_7220_put_tid - write a TID to the chip * @dd: the qlogic_ib device * @tidptr: pointer to the expected TID (in chip) to update - * @tidtype: 0 for eager, 1 for expected + * @type: 0 for eager, 1 for expected * @pa: physical address of in memory buffer; tidinvalid if freeing */ static void qib_7220_put_tid(struct qib_devdata *dd, u64 __iomem *tidptr, @@ -2180,7 +2180,7 @@ static void qib_7220_put_tid(struct qib_devdata *dd, u64 __iomem *tidptr, /** * qib_7220_clear_tids - clear all TID entries for a ctxt, expected and eager * @dd: the qlogic_ib device - * @ctxt: the ctxt + * @rcd: the ctxt * * clear all TID entries for a ctxt, expected and eager. * Used from qib_close(). On this chip, TIDs are only 32 bits, @@ -2238,7 +2238,7 @@ static void qib_7220_tidtemplate(struct qib_devdata *dd) /** * qib_init_7220_get_base_info - set chip-specific flags for user code * @rcd: the qlogic_ib ctxt - * @kbase: qib_base_info pointer + * @kinfo: qib_base_info pointer * * We set the PCIE flag because the lower bandwidth on PCIe vs * HyperTransport can affect some user packet algorithims. @@ -2896,8 +2896,8 @@ static void sendctrl_7220_mod(struct qib_pportdata *ppd, u32 op) /** * qib_portcntr_7220 - read a per-port counter - * @dd: the qlogic_ib device - * @creg: the counter to snapshot + * @ppd: the qlogic_ib device + * @reg: the counter to snapshot */ static u64 qib_portcntr_7220(struct qib_pportdata *ppd, u32 reg) { @@ -3232,7 +3232,7 @@ done: /** * qib_get_7220_faststats - get word counters from chip before they overflow - * @opaque - contains a pointer to the qlogic_ib device qib_devdata + * @t: contains a pointer to the qlogic_ib device qib_devdata * * This needs more work; in particular, decision on whether we really * need traffic_wds done the way it is @@ -4468,7 +4468,7 @@ static int qib_7220_eeprom_wen(struct qib_devdata *dd, int wen) /** * qib_init_iba7220_funcs - set up the chip-specific function pointers - * @dev: the pci_dev for qlogic_ib device + * @pdev: the pci_dev for qlogic_ib device * @ent: pci_device_id struct for this dev * * This is global, and is called directly at init to set up the diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c index 189a0ce6056a..9fe6ea75b45e 100644 --- a/drivers/infiniband/hw/qib/qib_iba7322.c +++ b/drivers/infiniband/hw/qib/qib_iba7322.c @@ -2514,7 +2514,7 @@ static int qib_7322_bringup_serdes(struct qib_pportdata *ppd) /** * qib_7322_quiet_serdes - set serdes to txidle - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * Called when driver is being unloaded */ static void qib_7322_mini_quiet_serdes(struct qib_pportdata *ppd) @@ -3760,7 +3760,7 @@ bail: * qib_7322_put_tid - write a TID to the chip * @dd: the qlogic_ib device * @tidptr: pointer to the expected TID (in chip) to update - * @tidtype: 0 for eager, 1 for expected + * @type: 0 for eager, 1 for expected * @pa: physical address of in memory buffer; tidinvalid if freeing */ static void qib_7322_put_tid(struct qib_devdata *dd, u64 __iomem *tidptr, @@ -3796,7 +3796,7 @@ static void qib_7322_put_tid(struct qib_devdata *dd, u64 __iomem *tidptr, /** * qib_7322_clear_tids - clear all TID entries for a ctxt, expected and eager * @dd: the qlogic_ib device - * @ctxt: the ctxt + * @rcd: the ctxt * * clear all TID entries for a ctxt, expected and eager. * Used from qib_close(). @@ -3861,7 +3861,7 @@ static void qib_7322_tidtemplate(struct qib_devdata *dd) /** * qib_init_7322_get_base_info - set chip-specific flags for user code * @rcd: the qlogic_ib ctxt - * @kbase: qib_base_info pointer + * @kinfo: qib_base_info pointer * * We set the PCIE flag because the lower bandwidth on PCIe vs * HyperTransport can affect some user packet algorithims. @@ -4724,7 +4724,7 @@ static void sendctrl_7322_mod(struct qib_pportdata *ppd, u32 op) /** * qib_portcntr_7322 - read a per-port chip counter * @ppd: the qlogic_ib pport - * @creg: the counter to read (not a chip offset) + * @reg: the counter to read (not a chip offset) */ static u64 qib_portcntr_7322(struct qib_pportdata *ppd, u32 reg) { @@ -5096,7 +5096,7 @@ done: /** * qib_get_7322_faststats - get word counters from chip before they overflow - * @opaque - contains a pointer to the qlogic_ib device qib_devdata + * @t: contains a pointer to the qlogic_ib device qib_devdata * * VESTIGIAL IBA7322 has no "small fast counters", so the only * real purpose of this function is to maintain the notion of @@ -7175,7 +7175,7 @@ static int qib_7322_tempsense_rd(struct qib_devdata *dd, int regnum) /** * qib_init_iba7322_funcs - set up the chip-specific function pointers - * @dev: the pci_dev for qlogic_ib device + * @pdev: the pci_dev for qlogic_ib device * @ent: pci_device_id struct for this dev * * Also allocates, inits, and returns the devdata struct for this diff --git a/drivers/infiniband/hw/qib/qib_intr.c b/drivers/infiniband/hw/qib/qib_intr.c index 65c3b964ad1b..85c3187d796d 100644 --- a/drivers/infiniband/hw/qib/qib_intr.c +++ b/drivers/infiniband/hw/qib/qib_intr.c @@ -40,9 +40,9 @@ /** * qib_format_hwmsg - format a single hwerror message - * @msg message buffer - * @msgl length of message buffer - * @hwmsg message to add to message buffer + * @msg: message buffer + * @msgl: length of message buffer + * @hwmsg: message to add to message buffer */ static void qib_format_hwmsg(char *msg, size_t msgl, const char *hwmsg) { @@ -53,11 +53,11 @@ static void qib_format_hwmsg(char *msg, size_t msgl, const char *hwmsg) /** * qib_format_hwerrors - format hardware error messages for display - * @hwerrs hardware errors bit vector - * @hwerrmsgs hardware error descriptions - * @nhwerrmsgs number of hwerrmsgs - * @msg message buffer - * @msgl message buffer length + * @hwerrs: hardware errors bit vector + * @hwerrmsgs: hardware error descriptions + * @nhwerrmsgs: number of hwerrmsgs + * @msg: message buffer + * @msgl: message buffer length */ void qib_format_hwerrors(u64 hwerrs, const struct qib_hwerror_msgs *hwerrmsgs, size_t nhwerrmsgs, char *msg, size_t msgl) diff --git a/drivers/infiniband/hw/qib/qib_mad.c b/drivers/infiniband/hw/qib/qib_mad.c index f83e331977f8..44e2f813024a 100644 --- a/drivers/infiniband/hw/qib/qib_mad.c +++ b/drivers/infiniband/hw/qib/qib_mad.c @@ -886,7 +886,7 @@ done: /** * rm_pkey - decrecment the reference count for the given PKEY - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @key: the PKEY index * * Return true if this was the last reference and the hardware table entry @@ -916,7 +916,7 @@ bail: /** * add_pkey - add the given PKEY to the hardware table - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @key: the PKEY * * Return an error code if unable to add the entry, zero if no change, @@ -2346,8 +2346,10 @@ static int process_cc(struct ib_device *ibdev, int mad_flags, * @port: the port number this packet came in on * @in_wc: the work completion entry for this packet * @in_grh: the global route header for this packet - * @in_mad: the incoming MAD - * @out_mad: any outgoing MAD reply + * @in: the incoming MAD + * @out: any outgoing MAD reply + * @out_mad_size: size of the outgoing MAD reply + * @out_mad_pkey_index: unused * * Returns IB_MAD_RESULT_SUCCESS if this is a MAD that we are not * interested in processing. diff --git a/drivers/infiniband/hw/qib/qib_pcie.c b/drivers/infiniband/hw/qib/qib_pcie.c index 2e07b3749b88..cb2a02d671e2 100644 --- a/drivers/infiniband/hw/qib/qib_pcie.c +++ b/drivers/infiniband/hw/qib/qib_pcie.c @@ -181,7 +181,7 @@ void qib_pcie_ddcleanup(struct qib_devdata *dd) pci_set_drvdata(dd->pcidev, NULL); } -/** +/* * We save the msi lo and hi values, so we can restore them after * chip reset (the kernel PCI infrastructure doesn't yet handle that * correctly. diff --git a/drivers/infiniband/hw/qib/qib_qp.c b/drivers/infiniband/hw/qib/qib_qp.c index 8d0563ef5be1..ca39a029e4af 100644 --- a/drivers/infiniband/hw/qib/qib_qp.c +++ b/drivers/infiniband/hw/qib/qib_qp.c @@ -207,7 +207,7 @@ bail: return ret; } -/** +/* * qib_free_all_qps - check for QPs still in use */ unsigned qib_free_all_qps(struct rvt_dev_info *rdi) @@ -376,9 +376,9 @@ void qib_flush_qp_waiters(struct rvt_qp *qp) /** * qib_check_send_wqe - validate wr/wqe - * @qp - The qp - * @wqe - The built wqe - * @call_send - Determine if the send should be posted or scheduled + * @qp: The qp + * @wqe: The built wqe + * @call_send: Determine if the send should be posted or scheduled * * Returns 0 on success, -EINVAL on failure */ @@ -418,8 +418,8 @@ static const char * const qp_type_str[] = { /** * qib_qp_iter_print - print information to seq_file - * @s - the seq_file - * @iter - the iterator + * @s: the seq_file + * @iter: the iterator */ void qib_qp_iter_print(struct seq_file *s, struct rvt_qp_iter *iter) { diff --git a/drivers/infiniband/hw/qib/qib_rc.c b/drivers/infiniband/hw/qib/qib_rc.c index 3915e5b4a9bc..a1c20ffb4490 100644 --- a/drivers/infiniband/hw/qib/qib_rc.c +++ b/drivers/infiniband/hw/qib/qib_rc.c @@ -207,6 +207,7 @@ bail: /** * qib_make_rc_req - construct a request packet (SEND, RDMA r/w, ATOMIC) * @qp: a pointer to the QP + * @flags: unused * * Assumes the s_lock is held. * @@ -992,7 +993,7 @@ static struct rvt_swqe *do_rc_completion(struct rvt_qp *qp, return wqe; } -/** +/* * do_rc_ack - process an incoming RC ACK * @qp: the QP the ACK came in on * @psn: the packet sequence number of the ACK @@ -1259,6 +1260,7 @@ static void rdma_seq_err(struct rvt_qp *qp, struct qib_ibport *ibp, u32 psn, * @psn: the packet sequence number for this packet * @hdrsize: the header length * @pmtu: the path MTU + * @rcd: the context pointer * * This is called from qib_rc_rcv() to process an incoming RC response * packet for the given QP. @@ -1480,6 +1482,7 @@ bail: * @opcode: the opcode for this packet * @psn: the packet sequence number for this packet * @diff: the difference between the PSN and the expected PSN + * @rcd: the context pointer * * This is called from qib_rc_rcv() to process an unexpected * incoming RC packet for the given QP. diff --git a/drivers/infiniband/hw/qib/qib_twsi.c b/drivers/infiniband/hw/qib/qib_twsi.c index f5698664419b..97b8a2bf5c69 100644 --- a/drivers/infiniband/hw/qib/qib_twsi.c +++ b/drivers/infiniband/hw/qib/qib_twsi.c @@ -168,6 +168,7 @@ static void stop_cmd(struct qib_devdata *dd); /** * rd_byte - read a byte, sending STOP on last, else ACK * @dd: the qlogic_ib device + * @last: identifies the last read * * Returns byte shifted out of device */ diff --git a/drivers/infiniband/hw/qib/qib_tx.c b/drivers/infiniband/hw/qib/qib_tx.c index 29785eb84646..6a8148851f21 100644 --- a/drivers/infiniband/hw/qib/qib_tx.c +++ b/drivers/infiniband/hw/qib/qib_tx.c @@ -377,6 +377,7 @@ void qib_sendbuf_done(struct qib_devdata *dd, unsigned n) * @start: the starting send buffer number * @len: the number of send buffers * @avail: true if the buffers are available for kernel use, false otherwise + * @rcd: the context pointer */ void qib_chg_pioavailkernel(struct qib_devdata *dd, unsigned start, unsigned len, u32 avail, struct qib_ctxtdata *rcd) diff --git a/drivers/infiniband/hw/qib/qib_uc.c b/drivers/infiniband/hw/qib/qib_uc.c index 554af4273a13..8e2bda77d8b9 100644 --- a/drivers/infiniband/hw/qib/qib_uc.c +++ b/drivers/infiniband/hw/qib/qib_uc.c @@ -40,6 +40,7 @@ /** * qib_make_uc_req - construct a request packet (SEND, RDMA write) * @qp: a pointer to the QP + * @flags: unused * * Assumes the s_lock is held. * diff --git a/drivers/infiniband/hw/qib/qib_ud.c b/drivers/infiniband/hw/qib/qib_ud.c index 93ca21347959..81eda94bd279 100644 --- a/drivers/infiniband/hw/qib/qib_ud.c +++ b/drivers/infiniband/hw/qib/qib_ud.c @@ -222,6 +222,7 @@ drop: /** * qib_make_ud_req - construct a UD request packet * @qp: the QP + * @flags: flags to modify and pass back to caller * * Assumes the s_lock is held. * diff --git a/drivers/infiniband/hw/qib/qib_user_pages.c b/drivers/infiniband/hw/qib/qib_user_pages.c index 4c24e83f3175..5d6cf7427431 100644 --- a/drivers/infiniband/hw/qib/qib_user_pages.c +++ b/drivers/infiniband/hw/qib/qib_user_pages.c @@ -43,7 +43,7 @@ static void __qib_release_user_pages(struct page **p, size_t num_pages, unpin_user_pages_dirty_lock(p, num_pages, dirty); } -/** +/* * qib_map_page - a safety wrapper around pci_map_page() * * A dma_addr of all 0's is interpreted by the chip as "disabled". diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c index f6c01bad5a74..8e0de265ad57 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.c +++ b/drivers/infiniband/hw/qib/qib_verbs.c @@ -1067,7 +1067,7 @@ bail: /** * qib_get_counters - get various chip counters - * @dd: the qlogic_ib device + * @ppd: the qlogic_ib device * @cntrs: counters are placed here * * Return the counters needed by recv_pma_get_portcounters(). @@ -1675,7 +1675,7 @@ void qib_unregister_ib_device(struct qib_devdata *dd) /** * _qib_schedule_send - schedule progress - * @qp - the qp + * @qp: the qp * * This schedules progress w/o regard to the s_flags. * @@ -1694,7 +1694,7 @@ bool _qib_schedule_send(struct rvt_qp *qp) /** * qib_schedule_send - schedule progress - * @qp - the qp + * @qp: the qp * * This schedules qp progress. The s_lock * should be held. diff --git a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c index 00a330909bb3..4b6019e7de67 100644 --- a/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c +++ b/drivers/infiniband/hw/vmw_pvrdma/pvrdma_main.c @@ -474,7 +474,6 @@ static irqreturn_t pvrdma_intrx_handler(int irq, void *dev_id) int ring_slots = (dev->dsr->cq_ring_pages.num_pages - 1) * PAGE_SIZE / sizeof(struct pvrdma_cqne); unsigned int head; - unsigned long flags; dev_dbg(&dev->pdev->dev, "interrupt x (completion) handler\n"); @@ -483,11 +482,11 @@ static irqreturn_t pvrdma_intrx_handler(int irq, void *dev_id) struct pvrdma_cq *cq; cqne = get_cqne(dev, head); - spin_lock_irqsave(&dev->cq_tbl_lock, flags); + spin_lock(&dev->cq_tbl_lock); cq = dev->cq_tbl[cqne->info % dev->dsr->caps.max_cq]; if (cq) refcount_inc(&cq->refcnt); - spin_unlock_irqrestore(&dev->cq_tbl_lock, flags); + spin_unlock(&dev->cq_tbl_lock); if (cq && cq->ibcq.comp_handler) cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context); diff --git a/drivers/infiniband/sw/rdmavt/cq.c b/drivers/infiniband/sw/rdmavt/cq.c index 20cc0799ac4b..5138afca067f 100644 --- a/drivers/infiniband/sw/rdmavt/cq.c +++ b/drivers/infiniband/sw/rdmavt/cq.c @@ -371,7 +371,7 @@ int rvt_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags notify_flags) return ret; } -/** +/* * rvt_resize_cq - change the size of the CQ * @ibcq: the completion queue * diff --git a/drivers/infiniband/sw/rdmavt/mad.c b/drivers/infiniband/sw/rdmavt/mad.c index 108c71e3ac23..fa5be13a4394 100644 --- a/drivers/infiniband/sw/rdmavt/mad.c +++ b/drivers/infiniband/sw/rdmavt/mad.c @@ -56,8 +56,11 @@ * @port_num: the port number this packet came in on, 1 based from ib core * @in_wc: the work completion entry for this packet * @in_grh: the global route header for this packet - * @in_mad: the incoming MAD - * @out_mad: any outgoing MAD reply + * @in: the incoming MAD + * @in_mad_size: size of the incoming MAD reply + * @out: any outgoing MAD reply + * @out_mad_size: size of the outgoing MAD reply + * @out_mad_pkey_index: unused * * Note that the verbs framework has already done the MAD sanity checks, * and hop count/pointer updating for IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE diff --git a/drivers/infiniband/sw/rdmavt/mcast.c b/drivers/infiniband/sw/rdmavt/mcast.c index 5233a63d99a6..951abac13dbb 100644 --- a/drivers/infiniband/sw/rdmavt/mcast.c +++ b/drivers/infiniband/sw/rdmavt/mcast.c @@ -180,7 +180,7 @@ struct rvt_mcast *rvt_mcast_find(struct rvt_ibport *ibp, union ib_gid *mgid, } EXPORT_SYMBOL(rvt_mcast_find); -/** +/* * rvt_mcast_add - insert mcast GID into table and attach QP struct * @mcast: the mcast GID table * @mqp: the QP to attach diff --git a/drivers/infiniband/sw/rdmavt/mr.c b/drivers/infiniband/sw/rdmavt/mr.c index 90fc234f489a..601d18dda1f5 100644 --- a/drivers/infiniband/sw/rdmavt/mr.c +++ b/drivers/infiniband/sw/rdmavt/mr.c @@ -369,6 +369,7 @@ bail: * @pd: protection domain for this memory region * @start: starting userspace address * @length: length of region to register + * @virt_addr: associated virtual address * @mr_access_flags: access flags for this memory region * @udata: unused by the driver * @@ -438,8 +439,8 @@ bail_umem: /** * rvt_dereg_clean_qp_cb - callback from iterator - * @qp - the qp - * @v - the mregion (as u64) + * @qp: the qp + * @v: the mregion (as u64) * * This routine fields the callback for all QPs and * for QPs in the same PD as the MR will call the @@ -457,7 +458,7 @@ static void rvt_dereg_clean_qp_cb(struct rvt_qp *qp, u64 v) /** * rvt_dereg_clean_qps - find QPs for reference cleanup - * @mr - the MR that is being deregistered + * @mr: the MR that is being deregistered * * This routine iterates RC QPs looking for references * to the lkey noted in mr. @@ -471,8 +472,8 @@ static void rvt_dereg_clean_qps(struct rvt_mregion *mr) /** * rvt_check_refs - check references - * @mr - the megion - * @t - the caller identification + * @mr: the megion + * @t: the caller identification * * This routine checks MRs holding a reference during * when being de-registered. @@ -506,8 +507,8 @@ static int rvt_check_refs(struct rvt_mregion *mr, const char *t) /** * rvt_mr_has_lkey - is MR - * @mr - the mregion - * @lkey - the lkey + * @mr: the mregion + * @lkey: the lkey */ bool rvt_mr_has_lkey(struct rvt_mregion *mr, u32 lkey) { @@ -516,8 +517,8 @@ bool rvt_mr_has_lkey(struct rvt_mregion *mr, u32 lkey) /** * rvt_ss_has_lkey - is mr in sge tests - * @ss - the sge state - * @lkey + * @ss: the sge state + * @lkey: the lkey * * This code tests for an MR in the indicated * sge state. @@ -540,7 +541,7 @@ bool rvt_ss_has_lkey(struct rvt_sge_state *ss, u32 lkey) /** * rvt_dereg_mr - unregister and free a memory region * @ibmr: the memory region to free - * + * @udata: unused by the driver * * Note that this is called to free MRs created by rvt_get_dma_mr() * or rvt_reg_user_mr(). diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c index 22fa9bde5419..9d13db68283c 100644 --- a/drivers/infiniband/sw/rdmavt/qp.c +++ b/drivers/infiniband/sw/rdmavt/qp.c @@ -156,7 +156,7 @@ void rvt_wss_exit(struct rvt_dev_info *rdi) rdi->wss = NULL; } -/** +/* * rvt_wss_init - Init wss data structures * * Return: 0 on success @@ -323,6 +323,7 @@ static void get_map_page(struct rvt_qpn_table *qpt, /** * init_qpn_table - initialize the QP number table for a device + * @rdi: rvt dev struct * @qpt: the QPN table */ static int init_qpn_table(struct rvt_dev_info *rdi, struct rvt_qpn_table *qpt) @@ -524,6 +525,7 @@ static inline unsigned mk_qpn(struct rvt_qpn_table *qpt, * IB_QPT_SMI/IB_QPT_GSI * @rdi: rvt device info structure * @qpt: queue pair number table pointer + * @type: the QP type * @port_num: IB port number, 1 based, comes from core * @exclude_prefix: prefix of special queue pair number being allocated * @@ -655,8 +657,8 @@ static void rvt_clear_mr_refs(struct rvt_qp *qp, int clr_sends) /** * rvt_swqe_has_lkey - return true if lkey is used by swqe - * @wqe - the send wqe - * @lkey - the lkey + * @wqe: the send wqe + * @lkey: the lkey * * Test the swqe for using lkey */ @@ -675,8 +677,8 @@ static bool rvt_swqe_has_lkey(struct rvt_swqe *wqe, u32 lkey) /** * rvt_qp_sends_has_lkey - return true is qp sends use lkey - * @qp - the rvt_qp - * @lkey - the lkey + * @qp: the rvt_qp + * @lkey: the lkey */ static bool rvt_qp_sends_has_lkey(struct rvt_qp *qp, u32 lkey) { @@ -699,8 +701,8 @@ static bool rvt_qp_sends_has_lkey(struct rvt_qp *qp, u32 lkey) /** * rvt_qp_acks_has_lkey - return true if acks have lkey - * @qp - the qp - * @lkey - the lkey + * @qp: the qp + * @lkey: the lkey */ static bool rvt_qp_acks_has_lkey(struct rvt_qp *qp, u32 lkey) { @@ -716,10 +718,10 @@ static bool rvt_qp_acks_has_lkey(struct rvt_qp *qp, u32 lkey) return false; } -/* +/** * rvt_qp_mr_clean - clean up remote ops for lkey - * @qp - the qp - * @lkey - the lkey that is being de-registered + * @qp: the qp + * @lkey: the lkey that is being de-registered * * This routine checks if the lkey is being used by * the qp. @@ -853,6 +855,7 @@ bail: /** * rvt_init_qp - initialize the QP state to the reset state + * @rdi: rvt dev struct * @qp: the QP to init or reinit * @type: the QP type * @@ -907,6 +910,7 @@ static void rvt_init_qp(struct rvt_dev_info *rdi, struct rvt_qp *qp, /** * _rvt_reset_qp - initialize the QP state to the reset state + * @rdi: rvt dev struct * @qp: the QP to reset * @type: the QP type * @@ -1726,6 +1730,7 @@ inval: /** * rvt_destroy_qp - destroy a queue pair * @ibqp: the queue pair to destroy + * @udata: unused by the driver * * Note that this can be called while the QP is actively sending or * receiving! @@ -1901,9 +1906,9 @@ int rvt_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, /** * rvt_qp_valid_operation - validate post send wr request - * @qp - the qp - * @post-parms - the post send table for the driver - * @wr - the work request + * @qp: the qp + * @post_parms: the post send table for the driver + * @wr: the work request * * The routine validates the operation based on the * validation table an returns the length of the operation @@ -2013,6 +2018,7 @@ static inline int rvt_qp_is_avail( * rvt_post_one_wr - post one RC, UC, or UD send work request * @qp: the QP to post on * @wr: the work request to send + * @call_send: kick the send engine into gear */ static int rvt_post_one_wr(struct rvt_qp *qp, const struct ib_send_wr *wr, @@ -2612,7 +2618,7 @@ EXPORT_SYMBOL(rvt_stop_rc_timers); /** * rvt_stop_rnr_timer - stop an rnr timer - * @qp - the QP + * @qp: the QP * * stop an rnr timer and return if the timer * had been pending. diff --git a/drivers/infiniband/sw/rdmavt/srq.c b/drivers/infiniband/sw/rdmavt/srq.c index 64d98bf238ab..2a7c2f12d372 100644 --- a/drivers/infiniband/sw/rdmavt/srq.c +++ b/drivers/infiniband/sw/rdmavt/srq.c @@ -67,7 +67,7 @@ void rvt_driver_srq_init(struct rvt_dev_info *rdi) /** * rvt_create_srq - create a shared receive queue - * @ibpd: the protection domain of the SRQ to create + * @ibsrq: the protection domain of the SRQ to create * @srq_init_attr: the attributes of the SRQ * @udata: data from libibverbs when creating a user SRQ * @@ -311,7 +311,8 @@ bail_free: return ret; } -/** rvt_query_srq - query srq data +/** + * rvt_query_srq - query srq data * @ibsrq: srq to query * @attr: return info in attr * @@ -330,7 +331,7 @@ int rvt_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr) /** * rvt_destroy_srq - destory an srq * @ibsrq: srq object to destroy - * + * @udata: user data for libibverbs.so */ int rvt_destroy_srq(struct ib_srq *ibsrq, struct ib_udata *udata) { diff --git a/drivers/infiniband/sw/rdmavt/vt.c b/drivers/infiniband/sw/rdmavt/vt.c index 49cec85a372a..8fd0128a9336 100644 --- a/drivers/infiniband/sw/rdmavt/vt.c +++ b/drivers/infiniband/sw/rdmavt/vt.c @@ -294,7 +294,7 @@ static int rvt_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) /** * rvt_dealloc_ucontext - Free a user context - * @context - Free this + * @context: Unused */ static void rvt_dealloc_ucontext(struct ib_ucontext *context) { diff --git a/drivers/infiniband/sw/rxe/rxe_comp.c b/drivers/infiniband/sw/rxe/rxe_comp.c index 0a1e6393250b..a8ac791a1bb9 100644 --- a/drivers/infiniband/sw/rxe/rxe_comp.c +++ b/drivers/infiniband/sw/rxe/rxe_comp.c @@ -515,6 +515,7 @@ static void rxe_drain_resp_pkts(struct rxe_qp *qp, bool notify) while ((skb = skb_dequeue(&qp->resp_pkts))) { rxe_drop_ref(qp); kfree_skb(skb); + ib_device_put(qp->ibqp.device); } while ((wqe = queue_head(qp->sq.queue))) { @@ -527,6 +528,17 @@ static void rxe_drain_resp_pkts(struct rxe_qp *qp, bool notify) } } +static void free_pkt(struct rxe_pkt_info *pkt) +{ + struct sk_buff *skb = PKT_TO_SKB(pkt); + struct rxe_qp *qp = pkt->qp; + struct ib_device *dev = qp->ibqp.device; + + kfree_skb(skb); + rxe_drop_ref(qp); + ib_device_put(dev); +} + int rxe_completer(void *arg) { struct rxe_qp *qp = (struct rxe_qp *)arg; @@ -624,11 +636,8 @@ int rxe_completer(void *arg) break; case COMPST_DONE: - if (pkt) { - rxe_drop_ref(pkt->qp); - kfree_skb(skb); - skb = NULL; - } + if (pkt) + free_pkt(pkt); goto done; case COMPST_EXIT: @@ -671,12 +680,8 @@ int rxe_completer(void *arg) */ if (qp->comp.started_retry && !qp->comp.timeout_retry) { - if (pkt) { - rxe_drop_ref(pkt->qp); - kfree_skb(skb); - skb = NULL; - } - + if (pkt) + free_pkt(pkt); goto done; } @@ -699,13 +704,8 @@ int rxe_completer(void *arg) qp->comp.started_retry = 1; rxe_run_task(&qp->req.task, 0); } - - if (pkt) { - rxe_drop_ref(pkt->qp); - kfree_skb(skb); - skb = NULL; - } - + if (pkt) + free_pkt(pkt); goto done; } else { @@ -726,9 +726,7 @@ int rxe_completer(void *arg) mod_timer(&qp->rnr_nak_timer, jiffies + rnrnak_jiffies(aeth_syn(pkt) & ~AETH_TYPE_MASK)); - rxe_drop_ref(pkt->qp); - kfree_skb(skb); - skb = NULL; + free_pkt(pkt); goto exit; } else { rxe_counter_inc(rxe, @@ -742,13 +740,8 @@ int rxe_completer(void *arg) WARN_ON_ONCE(wqe->status == IB_WC_SUCCESS); do_complete(qp, wqe); rxe_qp_error(qp); - - if (pkt) { - rxe_drop_ref(pkt->qp); - kfree_skb(skb); - skb = NULL; - } - + if (pkt) + free_pkt(pkt); goto exit; } } diff --git a/drivers/infiniband/sw/rxe/rxe_hdr.h b/drivers/infiniband/sw/rxe/rxe_hdr.h index 3b483b75dfe3..e432f9e37795 100644 --- a/drivers/infiniband/sw/rxe/rxe_hdr.h +++ b/drivers/infiniband/sw/rxe/rxe_hdr.h @@ -22,7 +22,6 @@ struct rxe_pkt_info { u16 paylen; /* length of bth - icrc */ u8 port_num; /* port pkt received on */ u8 opcode; /* bth opcode of packet */ - u8 offset; /* bth offset from pkt->hdr */ }; /* Macros should be used only for received skb */ @@ -280,134 +279,134 @@ static inline void __bth_set_psn(void *arg, u32 psn) static inline u8 bth_opcode(struct rxe_pkt_info *pkt) { - return __bth_opcode(pkt->hdr + pkt->offset); + return __bth_opcode(pkt->hdr); } static inline void bth_set_opcode(struct rxe_pkt_info *pkt, u8 opcode) { - __bth_set_opcode(pkt->hdr + pkt->offset, opcode); + __bth_set_opcode(pkt->hdr, opcode); } static inline u8 bth_se(struct rxe_pkt_info *pkt) { - return __bth_se(pkt->hdr + pkt->offset); + return __bth_se(pkt->hdr); } static inline void bth_set_se(struct rxe_pkt_info *pkt, int se) { - __bth_set_se(pkt->hdr + pkt->offset, se); + __bth_set_se(pkt->hdr, se); } static inline u8 bth_mig(struct rxe_pkt_info *pkt) { - return __bth_mig(pkt->hdr + pkt->offset); + return __bth_mig(pkt->hdr); } static inline void bth_set_mig(struct rxe_pkt_info *pkt, u8 mig) { - __bth_set_mig(pkt->hdr + pkt->offset, mig); + __bth_set_mig(pkt->hdr, mig); } static inline u8 bth_pad(struct rxe_pkt_info *pkt) { - return __bth_pad(pkt->hdr + pkt->offset); + return __bth_pad(pkt->hdr); } static inline void bth_set_pad(struct rxe_pkt_info *pkt, u8 pad) { - __bth_set_pad(pkt->hdr + pkt->offset, pad); + __bth_set_pad(pkt->hdr, pad); } static inline u8 bth_tver(struct rxe_pkt_info *pkt) { - return __bth_tver(pkt->hdr + pkt->offset); + return __bth_tver(pkt->hdr); } static inline void bth_set_tver(struct rxe_pkt_info *pkt, u8 tver) { - __bth_set_tver(pkt->hdr + pkt->offset, tver); + __bth_set_tver(pkt->hdr, tver); } static inline u16 bth_pkey(struct rxe_pkt_info *pkt) { - return __bth_pkey(pkt->hdr + pkt->offset); + return __bth_pkey(pkt->hdr); } static inline void bth_set_pkey(struct rxe_pkt_info *pkt, u16 pkey) { - __bth_set_pkey(pkt->hdr + pkt->offset, pkey); + __bth_set_pkey(pkt->hdr, pkey); } static inline u32 bth_qpn(struct rxe_pkt_info *pkt) { - return __bth_qpn(pkt->hdr + pkt->offset); + return __bth_qpn(pkt->hdr); } static inline void bth_set_qpn(struct rxe_pkt_info *pkt, u32 qpn) { - __bth_set_qpn(pkt->hdr + pkt->offset, qpn); + __bth_set_qpn(pkt->hdr, qpn); } static inline int bth_fecn(struct rxe_pkt_info *pkt) { - return __bth_fecn(pkt->hdr + pkt->offset); + return __bth_fecn(pkt->hdr); } static inline void bth_set_fecn(struct rxe_pkt_info *pkt, int fecn) { - __bth_set_fecn(pkt->hdr + pkt->offset, fecn); + __bth_set_fecn(pkt->hdr, fecn); } static inline int bth_becn(struct rxe_pkt_info *pkt) { - return __bth_becn(pkt->hdr + pkt->offset); + return __bth_becn(pkt->hdr); } static inline void bth_set_becn(struct rxe_pkt_info *pkt, int becn) { - __bth_set_becn(pkt->hdr + pkt->offset, becn); + __bth_set_becn(pkt->hdr, becn); } static inline u8 bth_resv6a(struct rxe_pkt_info *pkt) { - return __bth_resv6a(pkt->hdr + pkt->offset); + return __bth_resv6a(pkt->hdr); } static inline void bth_set_resv6a(struct rxe_pkt_info *pkt) { - __bth_set_resv6a(pkt->hdr + pkt->offset); + __bth_set_resv6a(pkt->hdr); } static inline int bth_ack(struct rxe_pkt_info *pkt) { - return __bth_ack(pkt->hdr + pkt->offset); + return __bth_ack(pkt->hdr); } static inline void bth_set_ack(struct rxe_pkt_info *pkt, int ack) { - __bth_set_ack(pkt->hdr + pkt->offset, ack); + __bth_set_ack(pkt->hdr, ack); } static inline void bth_set_resv7(struct rxe_pkt_info *pkt) { - __bth_set_resv7(pkt->hdr + pkt->offset); + __bth_set_resv7(pkt->hdr); } static inline u32 bth_psn(struct rxe_pkt_info *pkt) { - return __bth_psn(pkt->hdr + pkt->offset); + return __bth_psn(pkt->hdr); } static inline void bth_set_psn(struct rxe_pkt_info *pkt, u32 psn) { - __bth_set_psn(pkt->hdr + pkt->offset, psn); + __bth_set_psn(pkt->hdr, psn); } static inline void bth_init(struct rxe_pkt_info *pkt, u8 opcode, int se, int mig, int pad, u16 pkey, u32 qpn, int ack_req, u32 psn) { - struct rxe_bth *bth = (struct rxe_bth *)(pkt->hdr + pkt->offset); + struct rxe_bth *bth = (struct rxe_bth *)(pkt->hdr); bth->opcode = opcode; bth->flags = (pad << 4) & BTH_PAD_MASK; @@ -448,14 +447,14 @@ static inline void __rdeth_set_een(void *arg, u32 een) static inline u8 rdeth_een(struct rxe_pkt_info *pkt) { - return __rdeth_een(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RDETH]); + return __rdeth_een(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RDETH]); } static inline void rdeth_set_een(struct rxe_pkt_info *pkt, u32 een) { - __rdeth_set_een(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RDETH], een); + __rdeth_set_een(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RDETH], een); } /****************************************************************************** @@ -499,26 +498,26 @@ static inline void __deth_set_sqp(void *arg, u32 sqp) static inline u32 deth_qkey(struct rxe_pkt_info *pkt) { - return __deth_qkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_DETH]); + return __deth_qkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_DETH]); } static inline void deth_set_qkey(struct rxe_pkt_info *pkt, u32 qkey) { - __deth_set_qkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_DETH], qkey); + __deth_set_qkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_DETH], qkey); } static inline u32 deth_sqp(struct rxe_pkt_info *pkt) { - return __deth_sqp(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_DETH]); + return __deth_sqp(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_DETH]); } static inline void deth_set_sqp(struct rxe_pkt_info *pkt, u32 sqp) { - __deth_set_sqp(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_DETH], sqp); + __deth_set_sqp(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_DETH], sqp); } /****************************************************************************** @@ -574,38 +573,38 @@ static inline void __reth_set_len(void *arg, u32 len) static inline u64 reth_va(struct rxe_pkt_info *pkt) { - return __reth_va(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH]); + return __reth_va(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH]); } static inline void reth_set_va(struct rxe_pkt_info *pkt, u64 va) { - __reth_set_va(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH], va); + __reth_set_va(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH], va); } static inline u32 reth_rkey(struct rxe_pkt_info *pkt) { - return __reth_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH]); + return __reth_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH]); } static inline void reth_set_rkey(struct rxe_pkt_info *pkt, u32 rkey) { - __reth_set_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH], rkey); + __reth_set_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH], rkey); } static inline u32 reth_len(struct rxe_pkt_info *pkt) { - return __reth_len(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH]); + return __reth_len(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH]); } static inline void reth_set_len(struct rxe_pkt_info *pkt, u32 len) { - __reth_set_len(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_RETH], len); + __reth_set_len(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_RETH], len); } /****************************************************************************** @@ -676,50 +675,50 @@ static inline void __atmeth_set_comp(void *arg, u64 comp) static inline u64 atmeth_va(struct rxe_pkt_info *pkt) { - return __atmeth_va(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); + return __atmeth_va(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); } static inline void atmeth_set_va(struct rxe_pkt_info *pkt, u64 va) { - __atmeth_set_va(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], va); + __atmeth_set_va(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], va); } static inline u32 atmeth_rkey(struct rxe_pkt_info *pkt) { - return __atmeth_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); + return __atmeth_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); } static inline void atmeth_set_rkey(struct rxe_pkt_info *pkt, u32 rkey) { - __atmeth_set_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], rkey); + __atmeth_set_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], rkey); } static inline u64 atmeth_swap_add(struct rxe_pkt_info *pkt) { - return __atmeth_swap_add(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); + return __atmeth_swap_add(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); } static inline void atmeth_set_swap_add(struct rxe_pkt_info *pkt, u64 swap_add) { - __atmeth_set_swap_add(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], swap_add); + __atmeth_set_swap_add(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], swap_add); } static inline u64 atmeth_comp(struct rxe_pkt_info *pkt) { - return __atmeth_comp(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); + return __atmeth_comp(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH]); } static inline void atmeth_set_comp(struct rxe_pkt_info *pkt, u64 comp) { - __atmeth_set_comp(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], comp); + __atmeth_set_comp(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMETH], comp); } /****************************************************************************** @@ -780,26 +779,26 @@ static inline void __aeth_set_msn(void *arg, u32 msn) static inline u8 aeth_syn(struct rxe_pkt_info *pkt) { - return __aeth_syn(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_AETH]); + return __aeth_syn(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_AETH]); } static inline void aeth_set_syn(struct rxe_pkt_info *pkt, u8 syn) { - __aeth_set_syn(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_AETH], syn); + __aeth_set_syn(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_AETH], syn); } static inline u32 aeth_msn(struct rxe_pkt_info *pkt) { - return __aeth_msn(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_AETH]); + return __aeth_msn(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_AETH]); } static inline void aeth_set_msn(struct rxe_pkt_info *pkt, u32 msn) { - __aeth_set_msn(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_AETH], msn); + __aeth_set_msn(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_AETH], msn); } /****************************************************************************** @@ -825,14 +824,14 @@ static inline void __atmack_set_orig(void *arg, u64 orig) static inline u64 atmack_orig(struct rxe_pkt_info *pkt) { - return __atmack_orig(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMACK]); + return __atmack_orig(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMACK]); } static inline void atmack_set_orig(struct rxe_pkt_info *pkt, u64 orig) { - __atmack_set_orig(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_ATMACK], orig); + __atmack_set_orig(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_ATMACK], orig); } /****************************************************************************** @@ -858,14 +857,14 @@ static inline void __immdt_set_imm(void *arg, __be32 imm) static inline __be32 immdt_imm(struct rxe_pkt_info *pkt) { - return __immdt_imm(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_IMMDT]); + return __immdt_imm(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_IMMDT]); } static inline void immdt_set_imm(struct rxe_pkt_info *pkt, __be32 imm) { - __immdt_set_imm(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_IMMDT], imm); + __immdt_set_imm(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_IMMDT], imm); } /****************************************************************************** @@ -891,14 +890,14 @@ static inline void __ieth_set_rkey(void *arg, u32 rkey) static inline u32 ieth_rkey(struct rxe_pkt_info *pkt) { - return __ieth_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_IETH]); + return __ieth_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_IETH]); } static inline void ieth_set_rkey(struct rxe_pkt_info *pkt, u32 rkey) { - __ieth_set_rkey(pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_IETH], rkey); + __ieth_set_rkey(pkt->hdr + + rxe_opcode[pkt->opcode].offset[RXE_IETH], rkey); } enum rxe_hdr_length { @@ -915,13 +914,12 @@ enum rxe_hdr_length { static inline size_t header_size(struct rxe_pkt_info *pkt) { - return pkt->offset + rxe_opcode[pkt->opcode].length; + return rxe_opcode[pkt->opcode].length; } static inline void *payload_addr(struct rxe_pkt_info *pkt) { - return pkt->hdr + pkt->offset - + rxe_opcode[pkt->opcode].offset[RXE_PAYLOAD]; + return pkt->hdr + rxe_opcode[pkt->opcode].offset[RXE_PAYLOAD]; } static inline size_t payload_size(struct rxe_pkt_info *pkt) diff --git a/drivers/infiniband/sw/rxe/rxe_mcast.c b/drivers/infiniband/sw/rxe/rxe_mcast.c index c02315aed8d1..0ea9a5aa4ec0 100644 --- a/drivers/infiniband/sw/rxe/rxe_mcast.c +++ b/drivers/infiniband/sw/rxe/rxe_mcast.c @@ -7,45 +7,61 @@ #include "rxe.h" #include "rxe_loc.h" +/* caller should hold mc_grp_pool->pool_lock */ +static struct rxe_mc_grp *create_grp(struct rxe_dev *rxe, + struct rxe_pool *pool, + union ib_gid *mgid) +{ + int err; + struct rxe_mc_grp *grp; + + grp = rxe_alloc_locked(&rxe->mc_grp_pool); + if (!grp) + return ERR_PTR(-ENOMEM); + + INIT_LIST_HEAD(&grp->qp_list); + spin_lock_init(&grp->mcg_lock); + grp->rxe = rxe; + rxe_add_key_locked(grp, mgid); + + err = rxe_mcast_add(rxe, mgid); + if (unlikely(err)) { + rxe_drop_key_locked(grp); + rxe_drop_ref(grp); + return ERR_PTR(err); + } + + return grp; +} + int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid, struct rxe_mc_grp **grp_p) { int err; struct rxe_mc_grp *grp; + struct rxe_pool *pool = &rxe->mc_grp_pool; + unsigned long flags; - if (rxe->attr.max_mcast_qp_attach == 0) { - err = -EINVAL; - goto err1; - } + if (rxe->attr.max_mcast_qp_attach == 0) + return -EINVAL; - grp = rxe_pool_get_key(&rxe->mc_grp_pool, mgid); + write_lock_irqsave(&pool->pool_lock, flags); + + grp = rxe_pool_get_key_locked(pool, mgid); if (grp) goto done; - grp = rxe_alloc(&rxe->mc_grp_pool); - if (!grp) { - err = -ENOMEM; - goto err1; + grp = create_grp(rxe, pool, mgid); + if (IS_ERR(grp)) { + write_unlock_irqrestore(&pool->pool_lock, flags); + err = PTR_ERR(grp); + return err; } - INIT_LIST_HEAD(&grp->qp_list); - spin_lock_init(&grp->mcg_lock); - grp->rxe = rxe; - - rxe_add_key(grp, mgid); - - err = rxe_mcast_add(rxe, mgid); - if (err) - goto err2; - done: + write_unlock_irqrestore(&pool->pool_lock, flags); *grp_p = grp; return 0; - -err2: - rxe_drop_ref(grp); -err1: - return err; } int rxe_mcast_add_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp, diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c index 943914c2a50c..0701bd1ffd1a 100644 --- a/drivers/infiniband/sw/rxe/rxe_net.c +++ b/drivers/infiniband/sw/rxe/rxe_net.c @@ -153,15 +153,16 @@ static struct dst_entry *rxe_find_route(struct net_device *ndev, static int rxe_udp_encap_recv(struct sock *sk, struct sk_buff *skb) { struct udphdr *udph; + struct rxe_dev *rxe; struct net_device *ndev = skb->dev; - struct net_device *rdev = ndev; - struct rxe_dev *rxe = rxe_get_dev_from_net(ndev); struct rxe_pkt_info *pkt = SKB_TO_PKT(skb); - if (!rxe && is_vlan_dev(rdev)) { - rdev = vlan_dev_real_dev(ndev); - rxe = rxe_get_dev_from_net(rdev); - } + /* takes a reference on rxe->ib_dev + * drop when skb is freed + */ + rxe = rxe_get_dev_from_net(ndev); + if (!rxe && is_vlan_dev(ndev)) + rxe = rxe_get_dev_from_net(vlan_dev_real_dev(ndev)); if (!rxe) goto drop; @@ -180,12 +181,6 @@ static int rxe_udp_encap_recv(struct sock *sk, struct sk_buff *skb) rxe_rcv(skb); - /* - * FIXME: this is in the wrong place, it needs to be done when pkt is - * destroyed - */ - ib_device_put(&rxe->ib_dev); - return 0; drop: kfree_skb(skb); @@ -414,6 +409,11 @@ int rxe_send(struct rxe_pkt_info *pkt, struct sk_buff *skb) void rxe_loopback(struct sk_buff *skb) { + if (skb->protocol == htons(ETH_P_IP)) + skb_pull(skb, sizeof(struct iphdr)); + else + skb_pull(skb, sizeof(struct ipv6hdr)); + rxe_rcv(skb); } diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c index b374eb53e2fe..307d8986e7c9 100644 --- a/drivers/infiniband/sw/rxe/rxe_pool.c +++ b/drivers/infiniband/sw/rxe/rxe_pool.c @@ -15,21 +15,25 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_UC] = { .name = "rxe-uc", .size = sizeof(struct rxe_ucontext), + .elem_offset = offsetof(struct rxe_ucontext, pelem), .flags = RXE_POOL_NO_ALLOC, }, [RXE_TYPE_PD] = { .name = "rxe-pd", .size = sizeof(struct rxe_pd), + .elem_offset = offsetof(struct rxe_pd, pelem), .flags = RXE_POOL_NO_ALLOC, }, [RXE_TYPE_AH] = { .name = "rxe-ah", .size = sizeof(struct rxe_ah), - .flags = RXE_POOL_ATOMIC | RXE_POOL_NO_ALLOC, + .elem_offset = offsetof(struct rxe_ah, pelem), + .flags = RXE_POOL_NO_ALLOC, }, [RXE_TYPE_SRQ] = { .name = "rxe-srq", .size = sizeof(struct rxe_srq), + .elem_offset = offsetof(struct rxe_srq, pelem), .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_SRQ_INDEX, .max_index = RXE_MAX_SRQ_INDEX, @@ -37,6 +41,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_QP] = { .name = "rxe-qp", .size = sizeof(struct rxe_qp), + .elem_offset = offsetof(struct rxe_qp, pelem), .cleanup = rxe_qp_cleanup, .flags = RXE_POOL_INDEX, .min_index = RXE_MIN_QP_INDEX, @@ -45,12 +50,14 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_CQ] = { .name = "rxe-cq", .size = sizeof(struct rxe_cq), + .elem_offset = offsetof(struct rxe_cq, pelem), .flags = RXE_POOL_NO_ALLOC, .cleanup = rxe_cq_cleanup, }, [RXE_TYPE_MR] = { .name = "rxe-mr", .size = sizeof(struct rxe_mem), + .elem_offset = offsetof(struct rxe_mem, pelem), .cleanup = rxe_mem_cleanup, .flags = RXE_POOL_INDEX, .max_index = RXE_MAX_MR_INDEX, @@ -59,6 +66,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_MW] = { .name = "rxe-mw", .size = sizeof(struct rxe_mem), + .elem_offset = offsetof(struct rxe_mem, pelem), .flags = RXE_POOL_INDEX, .max_index = RXE_MAX_MW_INDEX, .min_index = RXE_MIN_MW_INDEX, @@ -66,6 +74,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_MC_GRP] = { .name = "rxe-mc_grp", .size = sizeof(struct rxe_mc_grp), + .elem_offset = offsetof(struct rxe_mc_grp, pelem), .cleanup = rxe_mc_cleanup, .flags = RXE_POOL_KEY, .key_offset = offsetof(struct rxe_mc_grp, mgid), @@ -74,7 +83,7 @@ struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { [RXE_TYPE_MC_ELEM] = { .name = "rxe-mc_elem", .size = sizeof(struct rxe_mc_elem), - .flags = RXE_POOL_ATOMIC, + .elem_offset = offsetof(struct rxe_mc_elem, pelem), }, }; @@ -94,18 +103,18 @@ static int rxe_pool_init_index(struct rxe_pool *pool, u32 max, u32 min) goto out; } - pool->max_index = max; - pool->min_index = min; + pool->index.max_index = max; + pool->index.min_index = min; size = BITS_TO_LONGS(max - min + 1) * sizeof(long); - pool->table = kmalloc(size, GFP_KERNEL); - if (!pool->table) { + pool->index.table = kmalloc(size, GFP_KERNEL); + if (!pool->index.table) { err = -ENOMEM; goto out; } - pool->table_size = size; - bitmap_zero(pool->table, max - min + 1); + pool->index.table_size = size; + bitmap_zero(pool->index.table, max - min + 1); out: return err; @@ -127,13 +136,12 @@ int rxe_pool_init( pool->max_elem = max_elem; pool->elem_size = ALIGN(size, RXE_POOL_ALIGN); pool->flags = rxe_type_info[type].flags; - pool->tree = RB_ROOT; + pool->index.tree = RB_ROOT; + pool->key.tree = RB_ROOT; pool->cleanup = rxe_type_info[type].cleanup; atomic_set(&pool->num_elem, 0); - kref_init(&pool->ref_cnt); - rwlock_init(&pool->pool_lock); if (rxe_type_info[type].flags & RXE_POOL_INDEX) { @@ -145,67 +153,47 @@ int rxe_pool_init( } if (rxe_type_info[type].flags & RXE_POOL_KEY) { - pool->key_offset = rxe_type_info[type].key_offset; - pool->key_size = rxe_type_info[type].key_size; + pool->key.key_offset = rxe_type_info[type].key_offset; + pool->key.key_size = rxe_type_info[type].key_size; } - pool->state = RXE_POOL_STATE_VALID; - out: return err; } -static void rxe_pool_release(struct kref *kref) -{ - struct rxe_pool *pool = container_of(kref, struct rxe_pool, ref_cnt); - - pool->state = RXE_POOL_STATE_INVALID; - kfree(pool->table); -} - -static void rxe_pool_put(struct rxe_pool *pool) -{ - kref_put(&pool->ref_cnt, rxe_pool_release); -} - void rxe_pool_cleanup(struct rxe_pool *pool) { - unsigned long flags; - - write_lock_irqsave(&pool->pool_lock, flags); - pool->state = RXE_POOL_STATE_INVALID; if (atomic_read(&pool->num_elem) > 0) pr_warn("%s pool destroyed with unfree'd elem\n", pool_name(pool)); - write_unlock_irqrestore(&pool->pool_lock, flags); - rxe_pool_put(pool); + kfree(pool->index.table); } static u32 alloc_index(struct rxe_pool *pool) { u32 index; - u32 range = pool->max_index - pool->min_index + 1; + u32 range = pool->index.max_index - pool->index.min_index + 1; - index = find_next_zero_bit(pool->table, range, pool->last); + index = find_next_zero_bit(pool->index.table, range, pool->index.last); if (index >= range) - index = find_first_zero_bit(pool->table, range); + index = find_first_zero_bit(pool->index.table, range); WARN_ON_ONCE(index >= range); - set_bit(index, pool->table); - pool->last = index; - return index + pool->min_index; + set_bit(index, pool->index.table); + pool->index.last = index; + return index + pool->index.min_index; } static void insert_index(struct rxe_pool *pool, struct rxe_pool_entry *new) { - struct rb_node **link = &pool->tree.rb_node; + struct rb_node **link = &pool->index.tree.rb_node; struct rb_node *parent = NULL; struct rxe_pool_entry *elem; while (*link) { parent = *link; - elem = rb_entry(parent, struct rxe_pool_entry, node); + elem = rb_entry(parent, struct rxe_pool_entry, index_node); if (elem->index == new->index) { pr_warn("element already exists!\n"); @@ -218,25 +206,25 @@ static void insert_index(struct rxe_pool *pool, struct rxe_pool_entry *new) link = &(*link)->rb_right; } - rb_link_node(&new->node, parent, link); - rb_insert_color(&new->node, &pool->tree); + rb_link_node(&new->index_node, parent, link); + rb_insert_color(&new->index_node, &pool->index.tree); out: return; } static void insert_key(struct rxe_pool *pool, struct rxe_pool_entry *new) { - struct rb_node **link = &pool->tree.rb_node; + struct rb_node **link = &pool->key.tree.rb_node; struct rb_node *parent = NULL; struct rxe_pool_entry *elem; int cmp; while (*link) { parent = *link; - elem = rb_entry(parent, struct rxe_pool_entry, node); + elem = rb_entry(parent, struct rxe_pool_entry, key_node); - cmp = memcmp((u8 *)elem + pool->key_offset, - (u8 *)new + pool->key_offset, pool->key_size); + cmp = memcmp((u8 *)elem + pool->key.key_offset, + (u8 *)new + pool->key.key_offset, pool->key.key_size); if (cmp == 0) { pr_warn("key already exists!\n"); @@ -249,116 +237,135 @@ static void insert_key(struct rxe_pool *pool, struct rxe_pool_entry *new) link = &(*link)->rb_right; } - rb_link_node(&new->node, parent, link); - rb_insert_color(&new->node, &pool->tree); + rb_link_node(&new->key_node, parent, link); + rb_insert_color(&new->key_node, &pool->key.tree); out: return; } -void rxe_add_key(void *arg, void *key) +void __rxe_add_key_locked(struct rxe_pool_entry *elem, void *key) { - struct rxe_pool_entry *elem = arg; struct rxe_pool *pool = elem->pool; - unsigned long flags; - write_lock_irqsave(&pool->pool_lock, flags); - memcpy((u8 *)elem + pool->key_offset, key, pool->key_size); + memcpy((u8 *)elem + pool->key.key_offset, key, pool->key.key_size); insert_key(pool, elem); - write_unlock_irqrestore(&pool->pool_lock, flags); } -void rxe_drop_key(void *arg) +void __rxe_add_key(struct rxe_pool_entry *elem, void *key) { - struct rxe_pool_entry *elem = arg; struct rxe_pool *pool = elem->pool; unsigned long flags; write_lock_irqsave(&pool->pool_lock, flags); - rb_erase(&elem->node, &pool->tree); + __rxe_add_key_locked(elem, key); write_unlock_irqrestore(&pool->pool_lock, flags); } -void rxe_add_index(void *arg) +void __rxe_drop_key_locked(struct rxe_pool_entry *elem) +{ + struct rxe_pool *pool = elem->pool; + + rb_erase(&elem->key_node, &pool->key.tree); +} + +void __rxe_drop_key(struct rxe_pool_entry *elem) { - struct rxe_pool_entry *elem = arg; struct rxe_pool *pool = elem->pool; unsigned long flags; write_lock_irqsave(&pool->pool_lock, flags); + __rxe_drop_key_locked(elem); + write_unlock_irqrestore(&pool->pool_lock, flags); +} + +void __rxe_add_index_locked(struct rxe_pool_entry *elem) +{ + struct rxe_pool *pool = elem->pool; + elem->index = alloc_index(pool); insert_index(pool, elem); - write_unlock_irqrestore(&pool->pool_lock, flags); } -void rxe_drop_index(void *arg) +void __rxe_add_index(struct rxe_pool_entry *elem) { - struct rxe_pool_entry *elem = arg; struct rxe_pool *pool = elem->pool; unsigned long flags; write_lock_irqsave(&pool->pool_lock, flags); - clear_bit(elem->index - pool->min_index, pool->table); - rb_erase(&elem->node, &pool->tree); + __rxe_add_index_locked(elem); write_unlock_irqrestore(&pool->pool_lock, flags); } -void *rxe_alloc(struct rxe_pool *pool) +void __rxe_drop_index_locked(struct rxe_pool_entry *elem) { - struct rxe_pool_entry *elem; - unsigned long flags; + struct rxe_pool *pool = elem->pool; - might_sleep_if(!(pool->flags & RXE_POOL_ATOMIC)); + clear_bit(elem->index - pool->index.min_index, pool->index.table); + rb_erase(&elem->index_node, &pool->index.tree); +} - read_lock_irqsave(&pool->pool_lock, flags); - if (pool->state != RXE_POOL_STATE_VALID) { - read_unlock_irqrestore(&pool->pool_lock, flags); - return NULL; - } - kref_get(&pool->ref_cnt); - read_unlock_irqrestore(&pool->pool_lock, flags); +void __rxe_drop_index(struct rxe_pool_entry *elem) +{ + struct rxe_pool *pool = elem->pool; + unsigned long flags; + + write_lock_irqsave(&pool->pool_lock, flags); + __rxe_drop_index_locked(elem); + write_unlock_irqrestore(&pool->pool_lock, flags); +} - if (!ib_device_try_get(&pool->rxe->ib_dev)) - goto out_put_pool; +void *rxe_alloc_locked(struct rxe_pool *pool) +{ + struct rxe_type_info *info = &rxe_type_info[pool->type]; + struct rxe_pool_entry *elem; + u8 *obj; if (atomic_inc_return(&pool->num_elem) > pool->max_elem) goto out_cnt; - elem = kzalloc(rxe_type_info[pool->type].size, - (pool->flags & RXE_POOL_ATOMIC) ? - GFP_ATOMIC : GFP_KERNEL); - if (!elem) + obj = kzalloc(info->size, GFP_ATOMIC); + if (!obj) goto out_cnt; + elem = (struct rxe_pool_entry *)(obj + info->elem_offset); + elem->pool = pool; kref_init(&elem->ref_cnt); - return elem; + return obj; out_cnt: atomic_dec(&pool->num_elem); - ib_device_put(&pool->rxe->ib_dev); -out_put_pool: - rxe_pool_put(pool); return NULL; } -int rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem) +void *rxe_alloc(struct rxe_pool *pool) { - unsigned long flags; + struct rxe_type_info *info = &rxe_type_info[pool->type]; + struct rxe_pool_entry *elem; + u8 *obj; + + if (atomic_inc_return(&pool->num_elem) > pool->max_elem) + goto out_cnt; - might_sleep_if(!(pool->flags & RXE_POOL_ATOMIC)); + obj = kzalloc(info->size, GFP_KERNEL); + if (!obj) + goto out_cnt; - read_lock_irqsave(&pool->pool_lock, flags); - if (pool->state != RXE_POOL_STATE_VALID) { - read_unlock_irqrestore(&pool->pool_lock, flags); - return -EINVAL; - } - kref_get(&pool->ref_cnt); - read_unlock_irqrestore(&pool->pool_lock, flags); + elem = (struct rxe_pool_entry *)(obj + info->elem_offset); - if (!ib_device_try_get(&pool->rxe->ib_dev)) - goto out_put_pool; + elem->pool = pool; + kref_init(&elem->ref_cnt); + + return obj; + +out_cnt: + atomic_dec(&pool->num_elem); + return NULL; +} +int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem) +{ if (atomic_inc_return(&pool->num_elem) > pool->max_elem) goto out_cnt; @@ -369,9 +376,6 @@ int rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem) out_cnt: atomic_dec(&pool->num_elem); - ib_device_put(&pool->rxe->ib_dev); -out_put_pool: - rxe_pool_put(pool); return -EINVAL; } @@ -380,67 +384,77 @@ void rxe_elem_release(struct kref *kref) struct rxe_pool_entry *elem = container_of(kref, struct rxe_pool_entry, ref_cnt); struct rxe_pool *pool = elem->pool; + struct rxe_type_info *info = &rxe_type_info[pool->type]; + u8 *obj; if (pool->cleanup) pool->cleanup(elem); - if (!(pool->flags & RXE_POOL_NO_ALLOC)) - kfree(elem); + if (!(pool->flags & RXE_POOL_NO_ALLOC)) { + obj = (u8 *)elem - info->elem_offset; + kfree(obj); + } + atomic_dec(&pool->num_elem); - ib_device_put(&pool->rxe->ib_dev); - rxe_pool_put(pool); } -void *rxe_pool_get_index(struct rxe_pool *pool, u32 index) +void *rxe_pool_get_index_locked(struct rxe_pool *pool, u32 index) { - struct rb_node *node = NULL; - struct rxe_pool_entry *elem = NULL; - unsigned long flags; - - read_lock_irqsave(&pool->pool_lock, flags); - - if (pool->state != RXE_POOL_STATE_VALID) - goto out; + struct rxe_type_info *info = &rxe_type_info[pool->type]; + struct rb_node *node; + struct rxe_pool_entry *elem; + u8 *obj; - node = pool->tree.rb_node; + node = pool->index.tree.rb_node; while (node) { - elem = rb_entry(node, struct rxe_pool_entry, node); + elem = rb_entry(node, struct rxe_pool_entry, index_node); if (elem->index > index) node = node->rb_left; else if (elem->index < index) node = node->rb_right; - else { - kref_get(&elem->ref_cnt); + else break; - } } -out: - read_unlock_irqrestore(&pool->pool_lock, flags); - return node ? elem : NULL; + if (node) { + kref_get(&elem->ref_cnt); + obj = (u8 *)elem - info->elem_offset; + } else { + obj = NULL; + } + + return obj; } -void *rxe_pool_get_key(struct rxe_pool *pool, void *key) +void *rxe_pool_get_index(struct rxe_pool *pool, u32 index) { - struct rb_node *node = NULL; - struct rxe_pool_entry *elem = NULL; - int cmp; + u8 *obj; unsigned long flags; read_lock_irqsave(&pool->pool_lock, flags); + obj = rxe_pool_get_index_locked(pool, index); + read_unlock_irqrestore(&pool->pool_lock, flags); - if (pool->state != RXE_POOL_STATE_VALID) - goto out; + return obj; +} - node = pool->tree.rb_node; +void *rxe_pool_get_key_locked(struct rxe_pool *pool, void *key) +{ + struct rxe_type_info *info = &rxe_type_info[pool->type]; + struct rb_node *node; + struct rxe_pool_entry *elem; + u8 *obj; + int cmp; + + node = pool->key.tree.rb_node; while (node) { - elem = rb_entry(node, struct rxe_pool_entry, node); + elem = rb_entry(node, struct rxe_pool_entry, key_node); - cmp = memcmp((u8 *)elem + pool->key_offset, - key, pool->key_size); + cmp = memcmp((u8 *)elem + pool->key.key_offset, + key, pool->key.key_size); if (cmp > 0) node = node->rb_left; @@ -450,10 +464,24 @@ void *rxe_pool_get_key(struct rxe_pool *pool, void *key) break; } - if (node) + if (node) { kref_get(&elem->ref_cnt); + obj = (u8 *)elem - info->elem_offset; + } else { + obj = NULL; + } -out: + return obj; +} + +void *rxe_pool_get_key(struct rxe_pool *pool, void *key) +{ + u8 *obj; + unsigned long flags; + + read_lock_irqsave(&pool->pool_lock, flags); + obj = rxe_pool_get_key_locked(pool, key); read_unlock_irqrestore(&pool->pool_lock, flags); - return node ? elem : NULL; + + return obj; } diff --git a/drivers/infiniband/sw/rxe/rxe_pool.h b/drivers/infiniband/sw/rxe/rxe_pool.h index 432745ffc8d4..61210b300a78 100644 --- a/drivers/infiniband/sw/rxe/rxe_pool.h +++ b/drivers/infiniband/sw/rxe/rxe_pool.h @@ -11,7 +11,6 @@ #define RXE_POOL_CACHE_FLAGS (0) enum rxe_pool_flags { - RXE_POOL_ATOMIC = BIT(0), RXE_POOL_INDEX = BIT(1), RXE_POOL_KEY = BIT(2), RXE_POOL_NO_ALLOC = BIT(4), @@ -36,6 +35,7 @@ struct rxe_pool_entry; struct rxe_type_info { const char *name; size_t size; + size_t elem_offset; void (*cleanup)(struct rxe_pool_entry *obj); enum rxe_pool_flags flags; u32 max_index; @@ -46,18 +46,16 @@ struct rxe_type_info { extern struct rxe_type_info rxe_type_info[]; -enum rxe_pool_state { - RXE_POOL_STATE_INVALID, - RXE_POOL_STATE_VALID, -}; - struct rxe_pool_entry { struct rxe_pool *pool; struct kref ref_cnt; struct list_head list; - /* only used if indexed or keyed */ - struct rb_node node; + /* only used if keyed */ + struct rb_node key_node; + + /* only used if indexed */ + struct rb_node index_node; u32 index; }; @@ -65,24 +63,29 @@ struct rxe_pool { struct rxe_dev *rxe; rwlock_t pool_lock; /* protects pool add/del/search */ size_t elem_size; - struct kref ref_cnt; void (*cleanup)(struct rxe_pool_entry *obj); - enum rxe_pool_state state; enum rxe_pool_flags flags; enum rxe_elem_type type; unsigned int max_elem; atomic_t num_elem; - /* only used if indexed or keyed */ - struct rb_root tree; - unsigned long *table; - size_t table_size; - u32 max_index; - u32 min_index; - u32 last; - size_t key_offset; - size_t key_size; + /* only used if indexed */ + struct { + struct rb_root tree; + unsigned long *table; + size_t table_size; + u32 last; + u32 max_index; + u32 min_index; + } index; + + /* only used if keyed */ + struct { + struct rb_root tree; + size_t key_offset; + size_t key_size; + } key; }; /* initialize a pool of objects with given limit on @@ -95,32 +98,70 @@ int rxe_pool_init(struct rxe_dev *rxe, struct rxe_pool *pool, /* free resources from object pool */ void rxe_pool_cleanup(struct rxe_pool *pool); -/* allocate an object from pool */ +/* allocate an object from pool holding and not holding the pool lock */ +void *rxe_alloc_locked(struct rxe_pool *pool); + void *rxe_alloc(struct rxe_pool *pool); /* connect already allocated object to pool */ -int rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem); +int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_entry *elem); + +#define rxe_add_to_pool(pool, obj) __rxe_add_to_pool(pool, &(obj)->pelem) /* assign an index to an indexed object and insert object into - * pool's rb tree + * pool's rb tree holding and not holding the pool_lock */ -void rxe_add_index(void *elem); +void __rxe_add_index_locked(struct rxe_pool_entry *elem); + +#define rxe_add_index_locked(obj) __rxe_add_index_locked(&(obj)->pelem) -/* drop an index and remove object from rb tree */ -void rxe_drop_index(void *elem); +void __rxe_add_index(struct rxe_pool_entry *elem); + +#define rxe_add_index(obj) __rxe_add_index(&(obj)->pelem) + +/* drop an index and remove object from rb tree + * holding and not holding the pool_lock + */ +void __rxe_drop_index_locked(struct rxe_pool_entry *elem); + +#define rxe_drop_index_locked(obj) __rxe_drop_index_locked(&(obj)->pelem) + +void __rxe_drop_index(struct rxe_pool_entry *elem); + +#define rxe_drop_index(obj) __rxe_drop_index(&(obj)->pelem) /* assign a key to a keyed object and insert object into - * pool's rb tree + * pool's rb tree holding and not holding pool_lock */ -void rxe_add_key(void *elem, void *key); +void __rxe_add_key_locked(struct rxe_pool_entry *elem, void *key); + +#define rxe_add_key_locked(obj, key) __rxe_add_key_locked(&(obj)->pelem, key) + +void __rxe_add_key(struct rxe_pool_entry *elem, void *key); + +#define rxe_add_key(obj, key) __rxe_add_key(&(obj)->pelem, key) + +/* remove elem from rb tree holding and not holding the pool_lock */ +void __rxe_drop_key_locked(struct rxe_pool_entry *elem); -/* remove elem from rb tree */ -void rxe_drop_key(void *elem); +#define rxe_drop_key_locked(obj) __rxe_drop_key_locked(&(obj)->pelem) + +void __rxe_drop_key(struct rxe_pool_entry *elem); + +#define rxe_drop_key(obj) __rxe_drop_key(&(obj)->pelem) + +/* lookup an indexed object from index holding and not holding the pool_lock. + * takes a reference on object + */ +void *rxe_pool_get_index_locked(struct rxe_pool *pool, u32 index); -/* lookup an indexed object from index. takes a reference on object */ void *rxe_pool_get_index(struct rxe_pool *pool, u32 index); -/* lookup keyed object from key. takes a reference on the object */ +/* lookup keyed object from key holding and not holding the pool_lock. + * takes a reference on the objecti + */ +void *rxe_pool_get_key_locked(struct rxe_pool *pool, void *key); + void *rxe_pool_get_key(struct rxe_pool *pool, void *key); /* cleanup an object when all references are dropped */ diff --git a/drivers/infiniband/sw/rxe/rxe_qp.c b/drivers/infiniband/sw/rxe/rxe_qp.c index 656a5b4be847..34ae957a315c 100644 --- a/drivers/infiniband/sw/rxe/rxe_qp.c +++ b/drivers/infiniband/sw/rxe/rxe_qp.c @@ -62,6 +62,17 @@ int rxe_qp_chk_init(struct rxe_dev *rxe, struct ib_qp_init_attr *init) struct rxe_port *port; int port_num = init->port_num; + switch (init->qp_type) { + case IB_QPT_SMI: + case IB_QPT_GSI: + case IB_QPT_RC: + case IB_QPT_UC: + case IB_QPT_UD: + break; + default: + return -EOPNOTSUPP; + } + if (!init->recv_cq || !init->send_cq) { pr_warn("missing cq\n"); goto err1; diff --git a/drivers/infiniband/sw/rxe/rxe_recv.c b/drivers/infiniband/sw/rxe/rxe_recv.c index c9984a28eecc..45d2f711bce2 100644 --- a/drivers/infiniband/sw/rxe/rxe_recv.c +++ b/drivers/infiniband/sw/rxe/rxe_recv.c @@ -9,21 +9,26 @@ #include "rxe.h" #include "rxe_loc.h" +/* check that QP matches packet opcode type and is in a valid state */ static int check_type_state(struct rxe_dev *rxe, struct rxe_pkt_info *pkt, struct rxe_qp *qp) { + unsigned int pkt_type; + if (unlikely(!qp->valid)) goto err1; + pkt_type = pkt->opcode & 0xe0; + switch (qp_type(qp)) { case IB_QPT_RC: - if (unlikely((pkt->opcode & IB_OPCODE_RC) != 0)) { + if (unlikely(pkt_type != IB_OPCODE_RC)) { pr_warn_ratelimited("bad qp type\n"); goto err1; } break; case IB_QPT_UC: - if (unlikely(!(pkt->opcode & IB_OPCODE_UC))) { + if (unlikely(pkt_type != IB_OPCODE_UC)) { pr_warn_ratelimited("bad qp type\n"); goto err1; } @@ -31,7 +36,7 @@ static int check_type_state(struct rxe_dev *rxe, struct rxe_pkt_info *pkt, case IB_QPT_UD: case IB_QPT_SMI: case IB_QPT_GSI: - if (unlikely(!(pkt->opcode & IB_OPCODE_UD))) { + if (unlikely(pkt_type != IB_OPCODE_UD)) { pr_warn_ratelimited("bad qp type\n"); goto err1; } @@ -85,8 +90,7 @@ static int check_keys(struct rxe_dev *rxe, struct rxe_pkt_info *pkt, goto err1; } - if ((qp_type(qp) == IB_QPT_UD || qp_type(qp) == IB_QPT_GSI) && - pkt->mask) { + if (qp_type(qp) == IB_QPT_UD || qp_type(qp) == IB_QPT_GSI) { u32 qkey = (qpn == 1) ? GSI_QKEY : qp->attr.qkey; if (unlikely(deth_qkey(pkt) != qkey)) { @@ -252,7 +256,6 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb) list_for_each_entry(mce, &mcg->qp_list, qp_list) { qp = mce->qp; - pkt = SKB_TO_PKT(skb); /* validate qp for incoming packet */ err = check_type_state(rxe, pkt, qp); @@ -264,12 +267,22 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb) continue; /* for all but the last qp create a new clone of the - * skb and pass to the qp. + * skb and pass to the qp. If an error occurs in the + * checks for the last qp in the list we need to + * free the skb since it hasn't been passed on to + * rxe_rcv_pkt() which would free it later. */ - if (mce->qp_list.next != &mcg->qp_list) + if (mce->qp_list.next != &mcg->qp_list) { per_qp_skb = skb_clone(skb, GFP_ATOMIC); - else + if (WARN_ON(!ib_device_try_get(&rxe->ib_dev))) { + kfree_skb(per_qp_skb); + continue; + } + } else { per_qp_skb = skb; + /* show we have consumed the skb */ + skb = NULL; + } if (unlikely(!per_qp_skb)) continue; @@ -284,10 +297,10 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb) rxe_drop_ref(mcg); /* drop ref from rxe_pool_get_key. */ - return; - err1: + /* free skb if not consumed */ kfree_skb(skb); + ib_device_put(&rxe->ib_dev); } /** @@ -340,9 +353,7 @@ void rxe_rcv(struct sk_buff *skb) __be32 *icrcp; u32 calc_icrc, pack_icrc; - pkt->offset = 0; - - if (unlikely(skb->len < pkt->offset + RXE_BTH_BYTES)) + if (unlikely(skb->len < RXE_BTH_BYTES)) goto drop; if (rxe_chk_dgid(rxe, skb) < 0) { @@ -397,4 +408,5 @@ drop: rxe_drop_ref(pkt->qp); kfree_skb(skb); + ib_device_put(&rxe->ib_dev); } diff --git a/drivers/infiniband/sw/rxe/rxe_req.c b/drivers/infiniband/sw/rxe/rxe_req.c index d4917646641a..889290793d75 100644 --- a/drivers/infiniband/sw/rxe/rxe_req.c +++ b/drivers/infiniband/sw/rxe/rxe_req.c @@ -375,7 +375,6 @@ static struct sk_buff *init_req_packet(struct rxe_qp *qp, pkt->psn = qp->req.psn; pkt->mask = rxe_opcode[opcode].mask; pkt->paylen = paylen; - pkt->offset = 0; pkt->wqe = wqe; /* init skb */ diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c index c7e3b6a4af38..142f3d8014d8 100644 --- a/drivers/infiniband/sw/rxe/rxe_resp.c +++ b/drivers/infiniband/sw/rxe/rxe_resp.c @@ -99,6 +99,7 @@ static inline enum resp_states get_req(struct rxe_qp *qp, while ((skb = skb_dequeue(&qp->req_pkts))) { rxe_drop_ref(qp); kfree_skb(skb); + ib_device_put(qp->ibqp.device); } /* go drain recv wr queue */ @@ -585,11 +586,10 @@ static struct sk_buff *prepare_ack_packet(struct rxe_qp *qp, ack->qp = qp; ack->opcode = opcode; ack->mask = rxe_opcode[opcode].mask; - ack->offset = pkt->offset; ack->paylen = paylen; /* fill in bth using the request packet headers */ - memcpy(ack->hdr, pkt->hdr, pkt->offset + RXE_BTH_BYTES); + memcpy(ack->hdr, pkt->hdr, RXE_BTH_BYTES); bth_set_opcode(ack, opcode); bth_set_qpn(ack, qp->attr.dest_qp_num); @@ -1017,6 +1017,7 @@ static enum resp_states cleanup(struct rxe_qp *qp, skb = skb_dequeue(&qp->req_pkts); rxe_drop_ref(qp); kfree_skb(skb); + ib_device_put(qp->ibqp.device); } if (qp->resp.mr) { @@ -1181,6 +1182,7 @@ static void rxe_drain_req_pkts(struct rxe_qp *qp, bool notify) while ((skb = skb_dequeue(&qp->req_pkts))) { rxe_drop_ref(qp); kfree_skb(skb); + ib_device_put(qp->ibqp.device); } if (notify) diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c index a031514e2f41..dee5e0e919d2 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.c +++ b/drivers/infiniband/sw/rxe/rxe_verbs.c @@ -106,12 +106,12 @@ static enum rdma_link_layer rxe_get_link_layer(struct ib_device *dev, return IB_LINK_LAYER_ETHERNET; } -static int rxe_alloc_ucontext(struct ib_ucontext *uctx, struct ib_udata *udata) +static int rxe_alloc_ucontext(struct ib_ucontext *ibuc, struct ib_udata *udata) { - struct rxe_dev *rxe = to_rdev(uctx->device); - struct rxe_ucontext *uc = to_ruc(uctx); + struct rxe_dev *rxe = to_rdev(ibuc->device); + struct rxe_ucontext *uc = to_ruc(ibuc); - return rxe_add_to_pool(&rxe->uc_pool, &uc->pelem); + return rxe_add_to_pool(&rxe->uc_pool, uc); } static void rxe_dealloc_ucontext(struct ib_ucontext *ibuc) @@ -145,7 +145,7 @@ static int rxe_alloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) struct rxe_dev *rxe = to_rdev(ibpd->device); struct rxe_pd *pd = to_rpd(ibpd); - return rxe_add_to_pool(&rxe->pd_pool, &pd->pelem); + return rxe_add_to_pool(&rxe->pd_pool, pd); } static int rxe_dealloc_pd(struct ib_pd *ibpd, struct ib_udata *udata) @@ -169,7 +169,7 @@ static int rxe_create_ah(struct ib_ah *ibah, if (err) return err; - err = rxe_add_to_pool(&rxe->ah_pool, &ah->pelem); + err = rxe_add_to_pool(&rxe->ah_pool, ah); if (err) return err; @@ -273,7 +273,7 @@ static int rxe_create_srq(struct ib_srq *ibsrq, struct ib_srq_init_attr *init, if (err) goto err1; - err = rxe_add_to_pool(&rxe->srq_pool, &srq->pelem); + err = rxe_add_to_pool(&rxe->srq_pool, srq); if (err) goto err1; @@ -555,37 +555,42 @@ static void init_send_wr(struct rxe_qp *qp, struct rxe_send_wr *wr, } } -static int init_send_wqe(struct rxe_qp *qp, const struct ib_send_wr *ibwr, +static void copy_inline_data_to_wqe(struct rxe_send_wqe *wqe, + const struct ib_send_wr *ibwr) +{ + struct ib_sge *sge = ibwr->sg_list; + u8 *p = wqe->dma.inline_data; + int i; + + for (i = 0; i < ibwr->num_sge; i++, sge++) { + memcpy(p, (void *)(uintptr_t)sge->addr, sge->length); + p += sge->length; + } +} + +static void init_send_wqe(struct rxe_qp *qp, const struct ib_send_wr *ibwr, unsigned int mask, unsigned int length, struct rxe_send_wqe *wqe) { int num_sge = ibwr->num_sge; - struct ib_sge *sge; - int i; - u8 *p; init_send_wr(qp, &wqe->wr, ibwr); + /* local operation */ + if (unlikely(mask & WR_REG_MASK)) { + wqe->mask = mask; + wqe->state = wqe_state_posted; + return; + } + if (qp_type(qp) == IB_QPT_UD || qp_type(qp) == IB_QPT_SMI || qp_type(qp) == IB_QPT_GSI) memcpy(&wqe->av, &to_rah(ud_wr(ibwr)->ah)->av, sizeof(wqe->av)); - if (unlikely(ibwr->send_flags & IB_SEND_INLINE)) { - p = wqe->dma.inline_data; - - sge = ibwr->sg_list; - for (i = 0; i < num_sge; i++, sge++) { - memcpy(p, (void *)(uintptr_t)sge->addr, - sge->length); - - p += sge->length; - } - } else if (mask & WR_REG_MASK) { - wqe->mask = mask; - wqe->state = wqe_state_posted; - return 0; - } else + if (unlikely(ibwr->send_flags & IB_SEND_INLINE)) + copy_inline_data_to_wqe(wqe, ibwr); + else memcpy(wqe->dma.sge, ibwr->sg_list, num_sge * sizeof(struct ib_sge)); @@ -599,8 +604,6 @@ static int init_send_wqe(struct rxe_qp *qp, const struct ib_send_wr *ibwr, wqe->dma.sge_offset = 0; wqe->state = wqe_state_posted; wqe->ssn = atomic_add_return(1, &qp->ssn); - - return 0; } static int post_one_send(struct rxe_qp *qp, const struct ib_send_wr *ibwr, @@ -623,10 +626,7 @@ static int post_one_send(struct rxe_qp *qp, const struct ib_send_wr *ibwr, } send_wqe = producer_addr(sq->queue); - - err = init_send_wqe(qp, ibwr, mask, length, send_wqe); - if (unlikely(err)) - goto err1; + init_send_wqe(qp, ibwr, mask, length, send_wqe); advance_producer(sq->queue); spin_unlock_irqrestore(&qp->sq.sq_lock, flags); @@ -774,7 +774,7 @@ static int rxe_create_cq(struct ib_cq *ibcq, const struct ib_cq_init_attr *attr, if (err) return err; - return rxe_add_to_pool(&rxe->cq_pool, &cq->pelem); + return rxe_add_to_pool(&rxe->cq_pool, cq); } static int rxe_destroy_cq(struct ib_cq *ibcq, struct ib_udata *udata) @@ -1118,7 +1118,7 @@ int rxe_register_device(struct rxe_dev *rxe, const char *ibdev_name) struct ib_device *dev = &rxe->ib_dev; struct crypto_shash *tfm; - strlcpy(dev->node_desc, "rxe", sizeof(dev->node_desc)); + strscpy(dev->node_desc, "rxe", sizeof(dev->node_desc)); dev->node_type = RDMA_NODE_IB_CA; dev->phys_port_cnt = 1; diff --git a/drivers/infiniband/sw/siw/siw.h b/drivers/infiniband/sw/siw/siw.h index adda78996219..368959ae9a8c 100644 --- a/drivers/infiniband/sw/siw/siw.h +++ b/drivers/infiniband/sw/siw/siw.h @@ -653,7 +653,7 @@ static inline struct siw_sqe *orq_get_free(struct siw_qp *qp) { struct siw_sqe *orq_e = orq_get_tail(qp); - if (orq_e && READ_ONCE(orq_e->flags) == 0) + if (READ_ONCE(orq_e->flags) == 0) return orq_e; return NULL; diff --git a/drivers/infiniband/sw/siw/siw_main.c b/drivers/infiniband/sw/siw/siw_main.c index ee95cf29179d..cf55326f2ab4 100644 --- a/drivers/infiniband/sw/siw/siw_main.c +++ b/drivers/infiniband/sw/siw/siw_main.c @@ -135,7 +135,7 @@ static struct { static int siw_init_cpulist(void) { - int i, num_nodes = num_possible_nodes(); + int i, num_nodes = nr_node_ids; memset(siw_tx_thread, 0, sizeof(siw_tx_thread)); @@ -357,7 +357,7 @@ static struct siw_device *siw_device_create(struct net_device *netdev) sizeof(base_dev->iw_ifname)); /* Disable TCP port mapping */ - base_dev->iw_driver_flags = IW_F_NO_PORT_MAP, + base_dev->iw_driver_flags = IW_F_NO_PORT_MAP; sdev->attrs.max_qp = SIW_MAX_QP; sdev->attrs.max_qp_wr = SIW_MAX_QP_WR; diff --git a/drivers/infiniband/sw/siw/siw_qp.c b/drivers/infiniband/sw/siw/siw_qp.c index 875d36d4b1c6..ddb2e66f9f13 100644 --- a/drivers/infiniband/sw/siw/siw_qp.c +++ b/drivers/infiniband/sw/siw/siw_qp.c @@ -199,26 +199,26 @@ void siw_qp_llp_write_space(struct sock *sk) static int siw_qp_readq_init(struct siw_qp *qp, int irq_size, int orq_size) { - irq_size = roundup_pow_of_two(irq_size); - orq_size = roundup_pow_of_two(orq_size); - - qp->attrs.irq_size = irq_size; - qp->attrs.orq_size = orq_size; - - qp->irq = vzalloc(irq_size * sizeof(struct siw_sqe)); - if (!qp->irq) { - siw_dbg_qp(qp, "irq malloc for %d failed\n", irq_size); - qp->attrs.irq_size = 0; - return -ENOMEM; + if (irq_size) { + irq_size = roundup_pow_of_two(irq_size); + qp->irq = vzalloc(irq_size * sizeof(struct siw_sqe)); + if (!qp->irq) { + qp->attrs.irq_size = 0; + return -ENOMEM; + } } - qp->orq = vzalloc(orq_size * sizeof(struct siw_sqe)); - if (!qp->orq) { - siw_dbg_qp(qp, "orq malloc for %d failed\n", orq_size); - qp->attrs.orq_size = 0; - qp->attrs.irq_size = 0; - vfree(qp->irq); - return -ENOMEM; + if (orq_size) { + orq_size = roundup_pow_of_two(orq_size); + qp->orq = vzalloc(orq_size * sizeof(struct siw_sqe)); + if (!qp->orq) { + qp->attrs.orq_size = 0; + qp->attrs.irq_size = 0; + vfree(qp->irq); + return -ENOMEM; + } } + qp->attrs.irq_size = irq_size; + qp->attrs.orq_size = orq_size; siw_dbg_qp(qp, "ORD %d, IRD %d\n", orq_size, irq_size); return 0; } @@ -288,13 +288,14 @@ int siw_qp_mpa_rts(struct siw_qp *qp, enum mpa_v2_ctrl ctrl) if (ctrl & MPA_V2_RDMA_WRITE_RTR) wqe->sqe.opcode = SIW_OP_WRITE; else if (ctrl & MPA_V2_RDMA_READ_RTR) { - struct siw_sqe *rreq; + struct siw_sqe *rreq = NULL; wqe->sqe.opcode = SIW_OP_READ; spin_lock(&qp->orq_lock); - rreq = orq_get_free(qp); + if (qp->attrs.orq_size) + rreq = orq_get_free(qp); if (rreq) { siw_read_to_orq(rreq, &wqe->sqe); qp->orq_put++; @@ -877,135 +878,88 @@ void siw_read_to_orq(struct siw_sqe *rreq, struct siw_sqe *sqe) rreq->num_sge = 1; } -/* - * Must be called with SQ locked. - * To avoid complete SQ starvation by constant inbound READ requests, - * the active IRQ will not be served after qp->irq_burst, if the - * SQ has pending work. - */ -int siw_activate_tx(struct siw_qp *qp) +static int siw_activate_tx_from_sq(struct siw_qp *qp) { - struct siw_sqe *irqe, *sqe; + struct siw_sqe *sqe; struct siw_wqe *wqe = tx_wqe(qp); int rv = 1; - irqe = &qp->irq[qp->irq_get % qp->attrs.irq_size]; - - if (irqe->flags & SIW_WQE_VALID) { - sqe = sq_get_next(qp); - - /* - * Avoid local WQE processing starvation in case - * of constant inbound READ request stream - */ - if (sqe && ++qp->irq_burst >= SIW_IRQ_MAXBURST_SQ_ACTIVE) { - qp->irq_burst = 0; - goto skip_irq; - } - memset(wqe->mem, 0, sizeof(*wqe->mem) * SIW_MAX_SGE); - wqe->wr_status = SIW_WR_QUEUED; - - /* start READ RESPONSE */ - wqe->sqe.opcode = SIW_OP_READ_RESPONSE; - wqe->sqe.flags = 0; - if (irqe->num_sge) { - wqe->sqe.num_sge = 1; - wqe->sqe.sge[0].length = irqe->sge[0].length; - wqe->sqe.sge[0].laddr = irqe->sge[0].laddr; - wqe->sqe.sge[0].lkey = irqe->sge[0].lkey; - } else { - wqe->sqe.num_sge = 0; - } - - /* Retain original RREQ's message sequence number for - * potential error reporting cases. - */ - wqe->sqe.sge[1].length = irqe->sge[1].length; - - wqe->sqe.rkey = irqe->rkey; - wqe->sqe.raddr = irqe->raddr; + sqe = sq_get_next(qp); + if (!sqe) + return 0; - wqe->processed = 0; - qp->irq_get++; + memset(wqe->mem, 0, sizeof(*wqe->mem) * SIW_MAX_SGE); + wqe->wr_status = SIW_WR_QUEUED; - /* mark current IRQ entry free */ - smp_store_mb(irqe->flags, 0); + /* First copy SQE to kernel private memory */ + memcpy(&wqe->sqe, sqe, sizeof(*sqe)); + if (wqe->sqe.opcode >= SIW_NUM_OPCODES) { + rv = -EINVAL; goto out; } - sqe = sq_get_next(qp); - if (sqe) { -skip_irq: - memset(wqe->mem, 0, sizeof(*wqe->mem) * SIW_MAX_SGE); - wqe->wr_status = SIW_WR_QUEUED; - - /* First copy SQE to kernel private memory */ - memcpy(&wqe->sqe, sqe, sizeof(*sqe)); - - if (wqe->sqe.opcode >= SIW_NUM_OPCODES) { + if (wqe->sqe.flags & SIW_WQE_INLINE) { + if (wqe->sqe.opcode != SIW_OP_SEND && + wqe->sqe.opcode != SIW_OP_WRITE) { rv = -EINVAL; goto out; } - if (wqe->sqe.flags & SIW_WQE_INLINE) { - if (wqe->sqe.opcode != SIW_OP_SEND && - wqe->sqe.opcode != SIW_OP_WRITE) { - rv = -EINVAL; - goto out; - } - if (wqe->sqe.sge[0].length > SIW_MAX_INLINE) { - rv = -EINVAL; - goto out; - } - wqe->sqe.sge[0].laddr = (uintptr_t)&wqe->sqe.sge[1]; - wqe->sqe.sge[0].lkey = 0; - wqe->sqe.num_sge = 1; + if (wqe->sqe.sge[0].length > SIW_MAX_INLINE) { + rv = -EINVAL; + goto out; } - if (wqe->sqe.flags & SIW_WQE_READ_FENCE) { - /* A READ cannot be fenced */ - if (unlikely(wqe->sqe.opcode == SIW_OP_READ || - wqe->sqe.opcode == - SIW_OP_READ_LOCAL_INV)) { - siw_dbg_qp(qp, "cannot fence read\n"); - rv = -EINVAL; - goto out; - } - spin_lock(&qp->orq_lock); + wqe->sqe.sge[0].laddr = (uintptr_t)&wqe->sqe.sge[1]; + wqe->sqe.sge[0].lkey = 0; + wqe->sqe.num_sge = 1; + } + if (wqe->sqe.flags & SIW_WQE_READ_FENCE) { + /* A READ cannot be fenced */ + if (unlikely(wqe->sqe.opcode == SIW_OP_READ || + wqe->sqe.opcode == + SIW_OP_READ_LOCAL_INV)) { + siw_dbg_qp(qp, "cannot fence read\n"); + rv = -EINVAL; + goto out; + } + spin_lock(&qp->orq_lock); - if (!siw_orq_empty(qp)) { - qp->tx_ctx.orq_fence = 1; - rv = 0; - } - spin_unlock(&qp->orq_lock); + if (qp->attrs.orq_size && !siw_orq_empty(qp)) { + qp->tx_ctx.orq_fence = 1; + rv = 0; + } + spin_unlock(&qp->orq_lock); - } else if (wqe->sqe.opcode == SIW_OP_READ || - wqe->sqe.opcode == SIW_OP_READ_LOCAL_INV) { - struct siw_sqe *rreq; + } else if (wqe->sqe.opcode == SIW_OP_READ || + wqe->sqe.opcode == SIW_OP_READ_LOCAL_INV) { + struct siw_sqe *rreq; - wqe->sqe.num_sge = 1; + if (unlikely(!qp->attrs.orq_size)) { + /* We negotiated not to send READ req's */ + rv = -EINVAL; + goto out; + } + wqe->sqe.num_sge = 1; - spin_lock(&qp->orq_lock); + spin_lock(&qp->orq_lock); - rreq = orq_get_free(qp); - if (rreq) { - /* - * Make an immediate copy in ORQ to be ready - * to process loopback READ reply - */ - siw_read_to_orq(rreq, &wqe->sqe); - qp->orq_put++; - } else { - qp->tx_ctx.orq_fence = 1; - rv = 0; - } - spin_unlock(&qp->orq_lock); + rreq = orq_get_free(qp); + if (rreq) { + /* + * Make an immediate copy in ORQ to be ready + * to process loopback READ reply + */ + siw_read_to_orq(rreq, &wqe->sqe); + qp->orq_put++; + } else { + qp->tx_ctx.orq_fence = 1; + rv = 0; } - - /* Clear SQE, can be re-used by application */ - smp_store_mb(sqe->flags, 0); - qp->sq_get++; - } else { - rv = 0; + spin_unlock(&qp->orq_lock); } + + /* Clear SQE, can be re-used by application */ + smp_store_mb(sqe->flags, 0); + qp->sq_get++; out: if (unlikely(rv < 0)) { siw_dbg_qp(qp, "error %d\n", rv); @@ -1015,6 +969,65 @@ out: } /* + * Must be called with SQ locked. + * To avoid complete SQ starvation by constant inbound READ requests, + * the active IRQ will not be served after qp->irq_burst, if the + * SQ has pending work. + */ +int siw_activate_tx(struct siw_qp *qp) +{ + struct siw_sqe *irqe; + struct siw_wqe *wqe = tx_wqe(qp); + + if (!qp->attrs.irq_size) + return siw_activate_tx_from_sq(qp); + + irqe = &qp->irq[qp->irq_get % qp->attrs.irq_size]; + + if (!(irqe->flags & SIW_WQE_VALID)) + return siw_activate_tx_from_sq(qp); + + /* + * Avoid local WQE processing starvation in case + * of constant inbound READ request stream + */ + if (sq_get_next(qp) && ++qp->irq_burst >= SIW_IRQ_MAXBURST_SQ_ACTIVE) { + qp->irq_burst = 0; + return siw_activate_tx_from_sq(qp); + } + memset(wqe->mem, 0, sizeof(*wqe->mem) * SIW_MAX_SGE); + wqe->wr_status = SIW_WR_QUEUED; + + /* start READ RESPONSE */ + wqe->sqe.opcode = SIW_OP_READ_RESPONSE; + wqe->sqe.flags = 0; + if (irqe->num_sge) { + wqe->sqe.num_sge = 1; + wqe->sqe.sge[0].length = irqe->sge[0].length; + wqe->sqe.sge[0].laddr = irqe->sge[0].laddr; + wqe->sqe.sge[0].lkey = irqe->sge[0].lkey; + } else { + wqe->sqe.num_sge = 0; + } + + /* Retain original RREQ's message sequence number for + * potential error reporting cases. + */ + wqe->sqe.sge[1].length = irqe->sge[1].length; + + wqe->sqe.rkey = irqe->rkey; + wqe->sqe.raddr = irqe->raddr; + + wqe->processed = 0; + qp->irq_get++; + + /* mark current IRQ entry free */ + smp_store_mb(irqe->flags, 0); + + return 1; +} + +/* * Check if current CQ state qualifies for calling CQ completion * handler. Must be called with CQ lock held. */ diff --git a/drivers/infiniband/sw/siw/siw_qp_rx.c b/drivers/infiniband/sw/siw/siw_qp_rx.c index 4bd1f1f84057..60116f20653c 100644 --- a/drivers/infiniband/sw/siw/siw_qp_rx.c +++ b/drivers/infiniband/sw/siw/siw_qp_rx.c @@ -680,6 +680,10 @@ static int siw_init_rresp(struct siw_qp *qp, struct siw_rx_stream *srx) } spin_lock_irqsave(&qp->sq_lock, flags); + if (unlikely(!qp->attrs.irq_size)) { + run_sq = 0; + goto error_irq; + } if (tx_work->wr_status == SIW_WR_IDLE) { /* * immediately schedule READ response w/o @@ -712,8 +716,9 @@ static int siw_init_rresp(struct siw_qp *qp, struct siw_rx_stream *srx) /* RRESP now valid as current TX wqe or placed into IRQ */ smp_store_mb(resp->flags, SIW_WQE_VALID); } else { - pr_warn("siw: [QP %u]: irq %d exceeded %d\n", qp_id(qp), - qp->irq_put % qp->attrs.irq_size, qp->attrs.irq_size); +error_irq: + pr_warn("siw: [QP %u]: IRQ exceeded or null, size %d\n", + qp_id(qp), qp->attrs.irq_size); siw_init_terminate(qp, TERM_ERROR_LAYER_RDMAP, RDMAP_ETYPE_REMOTE_OPERATION, @@ -740,6 +745,9 @@ static int siw_orqe_start_rx(struct siw_qp *qp) struct siw_sqe *orqe; struct siw_wqe *wqe = NULL; + if (unlikely(!qp->attrs.orq_size)) + return -EPROTO; + /* make sure ORQ indices are current */ smp_mb(); @@ -796,8 +804,8 @@ int siw_proc_rresp(struct siw_qp *qp) */ rv = siw_orqe_start_rx(qp); if (rv) { - pr_warn("siw: [QP %u]: ORQ empty at idx %d\n", - qp_id(qp), qp->orq_get % qp->attrs.orq_size); + pr_warn("siw: [QP %u]: ORQ empty, size %d\n", + qp_id(qp), qp->attrs.orq_size); goto error_term; } rv = siw_rresp_check_ntoh(srx, frx); @@ -1290,11 +1298,13 @@ static int siw_rdmap_complete(struct siw_qp *qp, int error) wc_status); siw_wqe_put_mem(wqe, SIW_OP_READ); - if (!error) + if (!error) { rv = siw_check_tx_fence(qp); - else - /* Disable current ORQ eleement */ - WRITE_ONCE(orq_get_current(qp)->flags, 0); + } else { + /* Disable current ORQ element */ + if (qp->attrs.orq_size) + WRITE_ONCE(orq_get_current(qp)->flags, 0); + } break; case RDMAP_RDMA_READ_REQ: diff --git a/drivers/infiniband/sw/siw/siw_qp_tx.c b/drivers/infiniband/sw/siw/siw_qp_tx.c index d19d8325588b..7989c4043db4 100644 --- a/drivers/infiniband/sw/siw/siw_qp_tx.c +++ b/drivers/infiniband/sw/siw/siw_qp_tx.c @@ -1107,8 +1107,8 @@ next_wqe: /* * RREQ may have already been completed by inbound RRESP! */ - if (tx_type == SIW_OP_READ || - tx_type == SIW_OP_READ_LOCAL_INV) { + if ((tx_type == SIW_OP_READ || + tx_type == SIW_OP_READ_LOCAL_INV) && qp->attrs.orq_size) { /* Cleanup pending entry in ORQ */ qp->orq_put--; qp->orq[qp->orq_put % qp->attrs.orq_size].flags = 0; diff --git a/drivers/infiniband/sw/siw/siw_verbs.c b/drivers/infiniband/sw/siw/siw_verbs.c index 68fd053fc774..e389d44e5591 100644 --- a/drivers/infiniband/sw/siw/siw_verbs.c +++ b/drivers/infiniband/sw/siw/siw_verbs.c @@ -365,13 +365,23 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd, if (rv) goto err_out; + num_sqe = attrs->cap.max_send_wr; + num_rqe = attrs->cap.max_recv_wr; + /* All queue indices are derived from modulo operations * on a free running 'get' (consumer) and 'put' (producer) * unsigned counter. Having queue sizes at power of two * avoids handling counter wrap around. */ - num_sqe = roundup_pow_of_two(attrs->cap.max_send_wr); - num_rqe = roundup_pow_of_two(attrs->cap.max_recv_wr); + if (num_sqe) + num_sqe = roundup_pow_of_two(num_sqe); + else { + /* Zero sized SQ is not supported */ + rv = -EINVAL; + goto err_out; + } + if (num_rqe) + num_rqe = roundup_pow_of_two(num_rqe); if (udata) qp->sendq = vmalloc_user(num_sqe * sizeof(struct siw_sqe)); @@ -379,7 +389,6 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd, qp->sendq = vzalloc(num_sqe * sizeof(struct siw_sqe)); if (qp->sendq == NULL) { - siw_dbg(base_dev, "SQ size %d alloc failed\n", num_sqe); rv = -ENOMEM; goto err_out_xa; } @@ -413,7 +422,6 @@ struct ib_qp *siw_create_qp(struct ib_pd *pd, qp->recvq = vzalloc(num_rqe * sizeof(struct siw_rqe)); if (qp->recvq == NULL) { - siw_dbg(base_dev, "RQ size %d alloc failed\n", num_rqe); rv = -ENOMEM; goto err_out_xa; } @@ -966,9 +974,9 @@ int siw_post_receive(struct ib_qp *base_qp, const struct ib_recv_wr *wr, unsigned long flags; int rv = 0; - if (qp->srq) { + if (qp->srq || qp->attrs.rq_size == 0) { *bad_wr = wr; - return -EOPNOTSUPP; /* what else from errno.h? */ + return -EINVAL; } if (!rdma_is_kernel_res(&qp->base_qp.res)) { siw_dbg_qp(qp, "no kernel post_recv for user mapped rq\n"); diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index 3440dc48d02c..179ff1d068e5 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -413,7 +413,6 @@ struct ipoib_dev_priv { u64 hca_caps; struct ipoib_ethtool_st ethtool; unsigned int max_send_sge; - bool sm_fullmember_sendonly_support; const struct net_device_ops *rn_ops; }; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index a6f413491321..e16b40c09f82 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -141,8 +141,6 @@ int ipoib_open(struct net_device *dev) set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags); - priv->sm_fullmember_sendonly_support = false; - if (ipoib_ib_dev_open(dev)) { if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) return 0; diff --git a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c index 86e4ed64e4e2..5b3154503bf4 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_multicast.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_multicast.c @@ -275,7 +275,7 @@ static int ipoib_mcast_join_finish(struct ipoib_mcast *mcast, memset(&av, 0, sizeof(av)); av.type = rdma_ah_find_type(priv->ca, priv->port); - rdma_ah_set_dlid(&av, be16_to_cpu(mcast->mcmember.mlid)), + rdma_ah_set_dlid(&av, be16_to_cpu(mcast->mcmember.mlid)); rdma_ah_set_port_num(&av, priv->port); rdma_ah_set_sl(&av, mcast->mcmember.sl); rdma_ah_set_static_rate(&av, mcast->mcmember.rate); @@ -334,15 +334,6 @@ void ipoib_mcast_carrier_on_task(struct work_struct *work) return; } /* - * Check if can send sendonly MCG's with sendonly-fullmember join state. - * It done here after the successfully join to the broadcast group, - * because the broadcast group must always be joined first and is always - * re-joined if the SM changes substantially. - */ - priv->sm_fullmember_sendonly_support = - ib_sa_sendonly_fullmem_support(&ipoib_sa_client, - priv->ca, priv->port); - /* * Take rtnl_lock to avoid racing with ipoib_stop() and * turning the carrier back on while a device is being * removed. However, ipoib_stop() will attempt to flush @@ -537,9 +528,7 @@ static int ipoib_mcast_join(struct net_device *dev, struct ipoib_mcast *mcast) * most closely emulates the behavior, from a user space * application perspective, of Ethernet multicast operation. */ - if (test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags) && - priv->sm_fullmember_sendonly_support) - /* SM supports sendonly-fullmember, otherwise fallback to full-member */ + if (test_bit(IPOIB_MCAST_FLAG_SENDONLY, &mcast->flags)) rec.join_state = SENDONLY_FULLMEMBER_JOIN; } spin_unlock_irq(&priv->lock); diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c index 4792b9bf400f..8fcaa1136f2c 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.c +++ b/drivers/infiniband/ulp/iser/iscsi_iser.c @@ -89,13 +89,20 @@ int iser_debug_level = 0; module_param_named(debug_level, iser_debug_level, int, S_IRUGO | S_IWUSR); MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0 (default:disabled)"); +static int iscsi_iser_set(const char *val, const struct kernel_param *kp); +static const struct kernel_param_ops iscsi_iser_size_ops = { + .set = iscsi_iser_set, + .get = param_get_uint, +}; + static unsigned int iscsi_max_lun = 512; -module_param_named(max_lun, iscsi_max_lun, uint, S_IRUGO); -MODULE_PARM_DESC(max_lun, "Max LUNs to allow per session (default:512"); +module_param_cb(max_lun, &iscsi_iser_size_ops, &iscsi_max_lun, S_IRUGO); +MODULE_PARM_DESC(max_lun, "Max LUNs to allow per session, should > 0 (default:512)"); unsigned int iser_max_sectors = ISER_DEF_MAX_SECTORS; -module_param_named(max_sectors, iser_max_sectors, uint, S_IRUGO | S_IWUSR); -MODULE_PARM_DESC(max_sectors, "Max number of sectors in a single scsi command (default:1024"); +module_param_cb(max_sectors, &iscsi_iser_size_ops, &iser_max_sectors, + S_IRUGO | S_IWUSR); +MODULE_PARM_DESC(max_sectors, "Max number of sectors in a single scsi command, should > 0 (default:1024)"); bool iser_always_reg = true; module_param_named(always_register, iser_always_reg, bool, S_IRUGO); @@ -110,6 +117,18 @@ int iser_pi_guard; module_param_named(pi_guard, iser_pi_guard, int, S_IRUGO); MODULE_PARM_DESC(pi_guard, "T10-PI guard_type [deprecated]"); +static int iscsi_iser_set(const char *val, const struct kernel_param *kp) +{ + int ret; + unsigned int n = 0; + + ret = kstrtouint(val, 10, &n); + if (ret != 0 || n == 0) + return -EINVAL; + + return param_set_uint(val, kp); +} + /* * iscsi_iser_recv() - Process a successful recv completion * @conn: iscsi connection @@ -571,13 +590,20 @@ iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session) static inline unsigned int iser_dif_prot_caps(int prot_caps) { - return ((prot_caps & IB_PROT_T10DIF_TYPE_1) ? - SHOST_DIF_TYPE1_PROTECTION | SHOST_DIX_TYPE0_PROTECTION | - SHOST_DIX_TYPE1_PROTECTION : 0) | - ((prot_caps & IB_PROT_T10DIF_TYPE_2) ? - SHOST_DIF_TYPE2_PROTECTION | SHOST_DIX_TYPE2_PROTECTION : 0) | - ((prot_caps & IB_PROT_T10DIF_TYPE_3) ? - SHOST_DIF_TYPE3_PROTECTION | SHOST_DIX_TYPE3_PROTECTION : 0); + int ret = 0; + + if (prot_caps & IB_PROT_T10DIF_TYPE_1) + ret |= SHOST_DIF_TYPE1_PROTECTION | + SHOST_DIX_TYPE0_PROTECTION | + SHOST_DIX_TYPE1_PROTECTION; + if (prot_caps & IB_PROT_T10DIF_TYPE_2) + ret |= SHOST_DIF_TYPE2_PROTECTION | + SHOST_DIX_TYPE2_PROTECTION; + if (prot_caps & IB_PROT_T10DIF_TYPE_3) + ret |= SHOST_DIF_TYPE3_PROTECTION | + SHOST_DIX_TYPE3_PROTECTION; + + return ret; } /** @@ -1009,11 +1035,6 @@ static int __init iser_init(void) iser_dbg("Starting iSER datamover...\n"); - if (iscsi_max_lun < 1) { - iser_err("Invalid max_lun value of %u\n", iscsi_max_lun); - return -EINVAL; - } - memset(&ig, 0, sizeof(struct iser_global)); ig.desc_cache = kmem_cache_create("iser_descriptors", diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c index d4e057fac219..afec40da9b58 100644 --- a/drivers/infiniband/ulp/iser/iser_memory.c +++ b/drivers/infiniband/ulp/iser/iser_memory.c @@ -169,7 +169,7 @@ iser_set_dif_domain(struct scsi_cmnd *sc, struct ib_sig_domain *domain) domain->sig.dif.ref_escape = true; if (sc->prot_flags & SCSI_PROT_REF_INCREMENT) domain->sig.dif.ref_remap = true; -}; +} static int iser_set_sig_attrs(struct scsi_cmnd *sc, struct ib_sig_attrs *sig_attrs) @@ -390,4 +390,3 @@ err_reg: return err; } - diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index 2bd18b006893..136f6c4492e0 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -685,7 +685,7 @@ static void iser_cleanup_handler(struct rdma_cm_id *cma_id, iser_disconnected_handler(cma_id); iser_free_ib_conn_res(iser_conn, destroy); complete(&iser_conn->ib_completion); -}; +} static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) { diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index 2ba27221ea85..7305ed8976c2 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -71,7 +71,6 @@ static int isert_sg_tablesize_set(const char *val, const struct kernel_param *kp return param_set_int(val, kp); } - static inline bool isert_prot_cmd(struct isert_conn *conn, struct se_cmd *cmd) { @@ -79,7 +78,6 @@ isert_prot_cmd(struct isert_conn *conn, struct se_cmd *cmd) cmd->prot_op != TARGET_PROT_NORMAL); } - static void isert_qp_event_callback(struct ib_event *e, void *context) { @@ -232,8 +230,10 @@ isert_create_device_ib_res(struct isert_device *device) } /* Check signature cap */ - device->pi_capable = ib_dev->attrs.device_cap_flags & - IB_DEVICE_INTEGRITY_HANDOVER ? true : false; + if (ib_dev->attrs.device_cap_flags & IB_DEVICE_INTEGRITY_HANDOVER) + device->pi_capable = true; + else + device->pi_capable = false; return 0; } @@ -1993,7 +1993,7 @@ isert_set_dif_domain(struct se_cmd *se_cmd, struct ib_sig_domain *domain) if (se_cmd->prot_type == TARGET_DIF_TYPE1_PROT || se_cmd->prot_type == TARGET_DIF_TYPE2_PROT) domain->sig.dif.ref_remap = true; -}; +} static int isert_set_sig_attrs(struct se_cmd *se_cmd, struct ib_sig_attrs *sig_attrs) diff --git a/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c b/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c index 4933085a864a..cecf0f7cadf9 100644 --- a/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c +++ b/drivers/infiniband/ulp/opa_vnic/opa_vnic_vema.c @@ -233,7 +233,7 @@ static void vema_get_class_port_info(struct opa_vnic_vema_port *port, port_info = (struct opa_class_port_info *)rsp_mad->data; memcpy(port_info, &port->class_port_info, sizeof(*port_info)); - port_info->base_version = OPA_MGMT_BASE_VERSION, + port_info->base_version = OPA_MGMT_BASE_VERSION; port_info->class_version = OPA_EMA_CLASS_VERSION; /* diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c index ba00f0de14ca..b6a0abf40589 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt-sysfs.c @@ -408,6 +408,7 @@ int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess) "%s", str); if (err) { pr_err("kobject_init_and_add: %d\n", err); + kobject_put(&sess->kobj); return err; } err = sysfs_create_group(&sess->kobj, &rtrs_clt_sess_attr_group); @@ -419,6 +420,7 @@ int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess) &sess->kobj, "stats"); if (err) { pr_err("kobject_init_and_add: %d\n", err); + kobject_put(&sess->stats->kobj_stats); goto remove_group; } @@ -469,15 +471,12 @@ int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt) return sysfs_create_group(&clt->dev.kobj, &rtrs_clt_attr_group); } -void rtrs_clt_destroy_sysfs_root_folders(struct rtrs_clt *clt) +void rtrs_clt_destroy_sysfs_root(struct rtrs_clt *clt) { + sysfs_remove_group(&clt->dev.kobj, &rtrs_clt_attr_group); + if (clt->kobj_paths) { kobject_del(clt->kobj_paths); kobject_put(clt->kobj_paths); } } - -void rtrs_clt_destroy_sysfs_root_files(struct rtrs_clt *clt) -{ - sysfs_remove_group(&clt->dev.kobj, &rtrs_clt_attr_group); -} diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.c b/drivers/infiniband/ulp/rtrs/rtrs-clt.c index 67f86c405a26..0a08b4b742a3 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.c @@ -31,6 +31,8 @@ */ #define RTRS_RECONNECT_SEED 8 +#define FIRST_CONN 0x01 + MODULE_DESCRIPTION("RDMA Transport Client"); MODULE_LICENSE("GPL"); @@ -178,18 +180,18 @@ struct rtrs_clt_con *rtrs_permit_to_clt_con(struct rtrs_clt_sess *sess, } /** - * __rtrs_clt_change_state() - change the session state through session state + * rtrs_clt_change_state() - change the session state through session state * machine. * * @sess: client session to change the state of. * @new_state: state to change to. * - * returns true if successful, false if the requested state can not be set. + * returns true if sess's state is changed to new state, otherwise return false. * * Locks: * state_wq lock must be hold. */ -static bool __rtrs_clt_change_state(struct rtrs_clt_sess *sess, +static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess, enum rtrs_clt_state new_state) { enum rtrs_clt_state old_state; @@ -286,7 +288,7 @@ static bool rtrs_clt_change_state_from_to(struct rtrs_clt_sess *sess, spin_lock_irq(&sess->state_wq.lock); if (sess->state == old_state) - changed = __rtrs_clt_change_state(sess, new_state); + changed = rtrs_clt_change_state(sess, new_state); spin_unlock_irq(&sess->state_wq.lock); return changed; @@ -494,7 +496,7 @@ static void rtrs_clt_recv_done(struct rtrs_clt_con *con, struct ib_wc *wc) int err; struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - WARN_ON(sess->flags != RTRS_MSG_NEW_RKEY_F); + WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0); iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); err = rtrs_iu_post_recv(&con->c, iu); @@ -514,7 +516,7 @@ static void rtrs_clt_rkey_rsp_done(struct rtrs_clt_con *con, struct ib_wc *wc) u32 buf_id; int err; - WARN_ON(sess->flags != RTRS_MSG_NEW_RKEY_F); + WARN_ON((sess->flags & RTRS_MSG_NEW_RKEY_F) == 0); iu = container_of(wc->wr_cqe, struct rtrs_iu, cqe); @@ -621,12 +623,12 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) } else if (imm_type == RTRS_HB_MSG_IMM) { WARN_ON(con->c.cid); rtrs_send_hb_ack(&sess->s); - if (sess->flags == RTRS_MSG_NEW_RKEY_F) + if (sess->flags & RTRS_MSG_NEW_RKEY_F) return rtrs_clt_recv_done(con, wc); } else if (imm_type == RTRS_HB_ACK_IMM) { WARN_ON(con->c.cid); sess->s.hb_missed_cnt = 0; - if (sess->flags == RTRS_MSG_NEW_RKEY_F) + if (sess->flags & RTRS_MSG_NEW_RKEY_F) return rtrs_clt_recv_done(con, wc); } else { rtrs_wrn(con->c.sess, "Unknown IMM type %u\n", @@ -654,7 +656,7 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) WARN_ON(!(wc->wc_flags & IB_WC_WITH_INVALIDATE || wc->wc_flags & IB_WC_WITH_IMM)); WARN_ON(wc->wr_cqe->done != rtrs_clt_rdma_done); - if (sess->flags == RTRS_MSG_NEW_RKEY_F) { + if (sess->flags & RTRS_MSG_NEW_RKEY_F) { if (wc->wc_flags & IB_WC_WITH_INVALIDATE) return rtrs_clt_recv_done(con, wc); @@ -664,7 +666,6 @@ static void rtrs_clt_rdma_done(struct ib_cq *cq, struct ib_wc *wc) case IB_WC_RDMA_WRITE: /* * post_send() RDMA write completions of IO reqs (read/write) - * and hb */ break; @@ -680,7 +681,7 @@ static int post_recv_io(struct rtrs_clt_con *con, size_t q_size) struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); for (i = 0; i < q_size; i++) { - if (sess->flags == RTRS_MSG_NEW_RKEY_F) { + if (sess->flags & RTRS_MSG_NEW_RKEY_F) { struct rtrs_iu *iu = &con->rsp_ius[i]; err = rtrs_iu_post_recv(&con->c, iu); @@ -1318,6 +1319,12 @@ out_err: static void free_permits(struct rtrs_clt *clt) { + if (clt->permits_map) { + size_t sz = clt->queue_depth; + + wait_event(clt->permits_wait, + find_first_bit(clt->permits_map, sz) >= sz); + } kfree(clt->permits_map); clt->permits_map = NULL; kfree(clt->permits); @@ -1353,21 +1360,14 @@ static bool rtrs_clt_change_state_get_old(struct rtrs_clt_sess *sess, bool changed; spin_lock_irq(&sess->state_wq.lock); - *old_state = sess->state; - changed = __rtrs_clt_change_state(sess, new_state); + if (old_state) + *old_state = sess->state; + changed = rtrs_clt_change_state(sess, new_state); spin_unlock_irq(&sess->state_wq.lock); return changed; } -static bool rtrs_clt_change_state(struct rtrs_clt_sess *sess, - enum rtrs_clt_state new_state) -{ - enum rtrs_clt_state old_state; - - return rtrs_clt_change_state_get_old(sess, new_state, &old_state); -} - static void rtrs_clt_hb_err_handler(struct rtrs_con *c) { struct rtrs_clt_con *con = container_of(c, typeof(*con), c); @@ -1511,7 +1511,7 @@ static void destroy_con(struct rtrs_clt_con *con) static int create_con_cq_qp(struct rtrs_clt_con *con) { struct rtrs_clt_sess *sess = to_clt_sess(con->c.sess); - u16 wr_queue_size; + u32 max_send_wr, max_recv_wr, cq_size; int err, cq_vector; struct rtrs_msg_rkey_rsp *rsp; @@ -1523,7 +1523,8 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) * + 2 for drain and heartbeat * in case qp gets into error state */ - wr_queue_size = SERVICE_CON_QUEUE_DEPTH * 3 + 2; + max_send_wr = SERVICE_CON_QUEUE_DEPTH * 2 + 2; + max_recv_wr = SERVICE_CON_QUEUE_DEPTH * 2 + 2; /* We must be the first here */ if (WARN_ON(sess->s.dev)) return -EINVAL; @@ -1555,25 +1556,29 @@ static int create_con_cq_qp(struct rtrs_clt_con *con) /* Shared between connections */ sess->s.dev_ref++; - wr_queue_size = + max_send_wr = min_t(int, sess->s.dev->ib_dev->attrs.max_qp_wr, /* QD * (REQ + RSP + FR REGS or INVS) + drain */ sess->queue_depth * 3 + 1); + max_recv_wr = + min_t(int, sess->s.dev->ib_dev->attrs.max_qp_wr, + sess->queue_depth * 3 + 1); } /* alloc iu to recv new rkey reply when server reports flags set */ - if (sess->flags == RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) { - con->rsp_ius = rtrs_iu_alloc(wr_queue_size, sizeof(*rsp), + if (sess->flags & RTRS_MSG_NEW_RKEY_F || con->c.cid == 0) { + con->rsp_ius = rtrs_iu_alloc(max_recv_wr, sizeof(*rsp), GFP_KERNEL, sess->s.dev->ib_dev, DMA_FROM_DEVICE, rtrs_clt_rdma_done); if (!con->rsp_ius) return -ENOMEM; - con->queue_size = wr_queue_size; + con->queue_size = max_recv_wr; } + cq_size = max_send_wr + max_recv_wr; cq_vector = con->cpu % sess->s.dev->ib_dev->num_comp_vectors; err = rtrs_cq_qp_create(&sess->s, &con->c, sess->max_send_sge, - cq_vector, wr_queue_size, wr_queue_size, - IB_POLL_SOFTIRQ); + cq_vector, cq_size, max_send_wr, + max_recv_wr, IB_POLL_SOFTIRQ); /* * In case of error we do not bother to clean previous allocations, * since destroy_con_cq_qp() must be called. @@ -1657,6 +1662,7 @@ static int rtrs_rdma_route_resolved(struct rtrs_clt_con *con) .cid_num = cpu_to_le16(sess->s.con_num), .recon_cnt = cpu_to_le16(sess->s.recon_cnt), }; + msg.first_conn = sess->for_new_clt ? FIRST_CONN : 0; uuid_copy(&msg.sess_uuid, &sess->s.uuid); uuid_copy(&msg.paths_uuid, &clt->paths_uuid); @@ -1742,6 +1748,8 @@ static int rtrs_rdma_conn_established(struct rtrs_clt_con *con, scnprintf(sess->hca_name, sizeof(sess->hca_name), sess->s.dev->ib_dev->name); sess->s.src_addr = con->c.cm_id->route.addr.src_addr; + /* set for_new_clt, to allow future reconnect on any path */ + sess->for_new_clt = 1; } return 0; @@ -1788,7 +1796,7 @@ static int rtrs_rdma_conn_rejected(struct rtrs_clt_con *con, static void rtrs_clt_close_conns(struct rtrs_clt_sess *sess, bool wait) { - if (rtrs_clt_change_state(sess, RTRS_CLT_CLOSING)) + if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_CLOSING, NULL)) queue_work(rtrs_wq, &sess->close_work); if (wait) flush_work(&sess->close_work); @@ -2174,7 +2182,7 @@ static void rtrs_clt_close_work(struct work_struct *work) cancel_delayed_work_sync(&sess->reconnect_dwork); rtrs_clt_stop_and_destroy_conns(sess); - rtrs_clt_change_state(sess, RTRS_CLT_CLOSED); + rtrs_clt_change_state_get_old(sess, RTRS_CLT_CLOSED, NULL); } static int init_conns(struct rtrs_clt_sess *sess) @@ -2226,7 +2234,7 @@ destroy: * doing rdma_resolve_addr(), switch to CONNECTION_ERR state * manually to keep reconnecting. */ - rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); return err; } @@ -2243,7 +2251,7 @@ static void rtrs_clt_info_req_done(struct ib_cq *cq, struct ib_wc *wc) if (unlikely(wc->status != IB_WC_SUCCESS)) { rtrs_err(sess->clt, "Sess info request send failed: %s\n", ib_wc_status_msg(wc->status)); - rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); return; } @@ -2367,7 +2375,7 @@ static void rtrs_clt_info_rsp_done(struct ib_cq *cq, struct ib_wc *wc) out: rtrs_clt_update_wc_stats(con); rtrs_iu_free(iu, sess->s.dev->ib_dev, 1); - rtrs_clt_change_state(sess, state); + rtrs_clt_change_state_get_old(sess, state, NULL); } static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) @@ -2423,7 +2431,6 @@ static int rtrs_send_sess_info(struct rtrs_clt_sess *sess) err = -ECONNRESET; else err = -ETIMEDOUT; - goto out; } out: @@ -2433,7 +2440,7 @@ out: rtrs_iu_free(rx_iu, sess->s.dev->ib_dev, 1); if (unlikely(err)) /* If we've never taken async path because of malloc problems */ - rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING_ERR); + rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING_ERR, NULL); return err; } @@ -2490,7 +2497,7 @@ static void rtrs_clt_reconnect_work(struct work_struct *work) /* Stop everything */ rtrs_clt_stop_and_destroy_conns(sess); msleep(RTRS_RECONNECT_BACKOFF); - if (rtrs_clt_change_state(sess, RTRS_CLT_CONNECTING)) { + if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_CONNECTING, NULL)) { err = init_sess(sess); if (err) goto reconnect_again; @@ -2499,7 +2506,7 @@ static void rtrs_clt_reconnect_work(struct work_struct *work) return; reconnect_again: - if (rtrs_clt_change_state(sess, RTRS_CLT_RECONNECTING)) { + if (rtrs_clt_change_state_get_old(sess, RTRS_CLT_RECONNECTING, NULL)) { sess->stats->reconnects.fail_cnt++; delay_ms = clt->reconnect_delay_sec * 1000; queue_delayed_work(rtrs_wq, &sess->reconnect_dwork, @@ -2565,11 +2572,8 @@ static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, clt->dev.class = rtrs_clt_dev_class; clt->dev.release = rtrs_clt_dev_release; err = dev_set_name(&clt->dev, "%s", sessname); - if (err) { - free_percpu(clt->pcpu_path); - kfree(clt); - return ERR_PTR(err); - } + if (err) + goto err; /* * Suppress user space notification until * sysfs files are created @@ -2577,44 +2581,35 @@ static struct rtrs_clt *alloc_clt(const char *sessname, size_t paths_num, dev_set_uevent_suppress(&clt->dev, true); err = device_register(&clt->dev); if (err) { - free_percpu(clt->pcpu_path); put_device(&clt->dev); - return ERR_PTR(err); + goto err; } clt->kobj_paths = kobject_create_and_add("paths", &clt->dev.kobj); if (!clt->kobj_paths) { - free_percpu(clt->pcpu_path); - device_unregister(&clt->dev); - return NULL; + err = -ENOMEM; + goto err_dev; } err = rtrs_clt_create_sysfs_root_files(clt); if (err) { - free_percpu(clt->pcpu_path); kobject_del(clt->kobj_paths); kobject_put(clt->kobj_paths); - device_unregister(&clt->dev); - return ERR_PTR(err); + goto err_dev; } dev_set_uevent_suppress(&clt->dev, false); kobject_uevent(&clt->dev.kobj, KOBJ_ADD); return clt; -} - -static void wait_for_inflight_permits(struct rtrs_clt *clt) -{ - if (clt->permits_map) { - size_t sz = clt->queue_depth; - - wait_event(clt->permits_wait, - find_first_bit(clt->permits_map, sz) >= sz); - } +err_dev: + device_unregister(&clt->dev); +err: + free_percpu(clt->pcpu_path); + kfree(clt); + return ERR_PTR(err); } static void free_clt(struct rtrs_clt *clt) { - wait_for_inflight_permits(clt); free_permits(clt); free_percpu(clt->pcpu_path); mutex_destroy(&clt->paths_ev_mutex); @@ -2672,6 +2667,8 @@ struct rtrs_clt *rtrs_clt_open(struct rtrs_clt_ops *ops, err = PTR_ERR(sess); goto close_all_sess; } + if (!i) + sess->for_new_clt = 1; list_add_tail_rcu(&sess->s.entry, &clt->paths_list); err = init_sess(sess); @@ -2702,8 +2699,7 @@ close_all_sess: rtrs_clt_close_conns(sess, true); kobject_put(&sess->kobj); } - rtrs_clt_destroy_sysfs_root_files(clt); - rtrs_clt_destroy_sysfs_root_folders(clt); + rtrs_clt_destroy_sysfs_root(clt); free_clt(clt); out: @@ -2720,8 +2716,7 @@ void rtrs_clt_close(struct rtrs_clt *clt) struct rtrs_clt_sess *sess, *tmp; /* Firstly forbid sysfs access */ - rtrs_clt_destroy_sysfs_root_files(clt); - rtrs_clt_destroy_sysfs_root_folders(clt); + rtrs_clt_destroy_sysfs_root(clt); /* Now it is safe to iterate over all paths without locks */ list_for_each_entry_safe(sess, tmp, &clt->paths_list, s.entry) { diff --git a/drivers/infiniband/ulp/rtrs/rtrs-clt.h b/drivers/infiniband/ulp/rtrs/rtrs-clt.h index b8dbd701b3cb..692bc83e1f09 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-clt.h +++ b/drivers/infiniband/ulp/rtrs/rtrs-clt.h @@ -143,6 +143,7 @@ struct rtrs_clt_sess { int max_send_sge; u32 flags; struct kobject kobj; + u8 for_new_clt; struct rtrs_clt_stats *stats; /* cache hca_port and hca_name to display in sysfs */ u8 hca_port; @@ -243,8 +244,7 @@ ssize_t rtrs_clt_reset_all_help(struct rtrs_clt_stats *stats, /* rtrs-clt-sysfs.c */ int rtrs_clt_create_sysfs_root_files(struct rtrs_clt *clt); -void rtrs_clt_destroy_sysfs_root_folders(struct rtrs_clt *clt); -void rtrs_clt_destroy_sysfs_root_files(struct rtrs_clt *clt); +void rtrs_clt_destroy_sysfs_root(struct rtrs_clt *clt); int rtrs_clt_create_sess_files(struct rtrs_clt_sess *sess); void rtrs_clt_destroy_sess_files(struct rtrs_clt_sess *sess, diff --git a/drivers/infiniband/ulp/rtrs/rtrs-pri.h b/drivers/infiniband/ulp/rtrs/rtrs-pri.h index 3f2918671dbe..8caad0a2322b 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-pri.h +++ b/drivers/infiniband/ulp/rtrs/rtrs-pri.h @@ -188,7 +188,9 @@ struct rtrs_msg_conn_req { __le16 recon_cnt; uuid_t sess_uuid; uuid_t paths_uuid; - u8 reserved[12]; + u8 first_conn : 1; + u8 reserved_bits : 7; + u8 reserved[11]; }; /** @@ -303,8 +305,9 @@ int rtrs_post_rdma_write_imm_empty(struct rtrs_con *con, struct ib_cqe *cqe, struct ib_send_wr *head); int rtrs_cq_qp_create(struct rtrs_sess *rtrs_sess, struct rtrs_con *con, - u32 max_send_sge, int cq_vector, u16 cq_size, - u16 wr_queue_size, enum ib_poll_context poll_ctx); + u32 max_send_sge, int cq_vector, int cq_size, + u32 max_send_wr, u32 max_recv_wr, + enum ib_poll_context poll_ctx); void rtrs_cq_qp_destroy(struct rtrs_con *con); void rtrs_init_hb(struct rtrs_sess *sess, struct ib_cqe *cqe, diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c index d2edff3b8f0d..126a96e75c62 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv-sysfs.c @@ -51,6 +51,8 @@ static ssize_t rtrs_srv_disconnect_store(struct kobject *kobj, sockaddr_to_str((struct sockaddr *)&sess->s.dst_addr, str, sizeof(str)); rtrs_info(s, "disconnect for path %s requested\n", str); + /* first remove sysfs itself to avoid deadlock */ + sysfs_remove_file_self(&sess->kobj, &attr->attr); close_sess(sess); return count; @@ -181,6 +183,7 @@ static int rtrs_srv_create_once_sysfs_root_folders(struct rtrs_srv_sess *sess) err = -ENOMEM; pr_err("kobject_create_and_add(): %d\n", err); device_del(&srv->dev); + put_device(&srv->dev); goto unlock; } dev_set_uevent_suppress(&srv->dev, false); @@ -206,6 +209,7 @@ rtrs_srv_destroy_once_sysfs_root_folders(struct rtrs_srv_sess *sess) kobject_put(srv->kobj_paths); mutex_unlock(&srv->paths_mutex); device_del(&srv->dev); + put_device(&srv->dev); } else { mutex_unlock(&srv->paths_mutex); } @@ -234,6 +238,7 @@ static int rtrs_srv_create_stats_files(struct rtrs_srv_sess *sess) &sess->kobj, "stats"); if (err) { rtrs_err(s, "kobject_init_and_add(): %d\n", err); + kobject_put(&sess->stats->kobj_stats); return err; } err = sysfs_create_group(&sess->stats->kobj_stats, @@ -290,8 +295,8 @@ remove_group: sysfs_remove_group(&sess->kobj, &rtrs_srv_sess_attr_group); put_kobj: kobject_del(&sess->kobj); - kobject_put(&sess->kobj); destroy_root: + kobject_put(&sess->kobj); rtrs_srv_destroy_once_sysfs_root_folders(sess); return err; @@ -302,7 +307,7 @@ void rtrs_srv_destroy_sess_files(struct rtrs_srv_sess *sess) if (sess->kobj.state_in_sysfs) { kobject_del(&sess->stats->kobj_stats); kobject_put(&sess->stats->kobj_stats); - kobject_del(&sess->kobj); + sysfs_remove_group(&sess->kobj, &rtrs_srv_sess_attr_group); kobject_put(&sess->kobj); rtrs_srv_destroy_once_sysfs_root_folders(sess); diff --git a/drivers/infiniband/ulp/rtrs/rtrs-srv.c b/drivers/infiniband/ulp/rtrs/rtrs-srv.c index c42fd470c4eb..d071809e3ed2 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs-srv.c +++ b/drivers/infiniband/ulp/rtrs/rtrs-srv.c @@ -222,7 +222,8 @@ static int rdma_write_sg(struct rtrs_srv_op *id) dma_addr_t dma_addr = sess->dma_addr[id->msg_id]; struct rtrs_srv_mr *srv_mr; struct rtrs_srv *srv = sess->srv; - struct ib_send_wr inv_wr, imm_wr; + struct ib_send_wr inv_wr; + struct ib_rdma_wr imm_wr; struct ib_rdma_wr *wr = NULL; enum ib_send_flags flags; size_t sg_cnt; @@ -267,21 +268,22 @@ static int rdma_write_sg(struct rtrs_srv_op *id) WARN_ON_ONCE(rkey != wr->rkey); wr->wr.opcode = IB_WR_RDMA_WRITE; + wr->wr.wr_cqe = &io_comp_cqe; wr->wr.ex.imm_data = 0; wr->wr.send_flags = 0; if (need_inval && always_invalidate) { wr->wr.next = &rwr.wr; rwr.wr.next = &inv_wr; - inv_wr.next = &imm_wr; + inv_wr.next = &imm_wr.wr; } else if (always_invalidate) { wr->wr.next = &rwr.wr; - rwr.wr.next = &imm_wr; + rwr.wr.next = &imm_wr.wr; } else if (need_inval) { wr->wr.next = &inv_wr; - inv_wr.next = &imm_wr; + inv_wr.next = &imm_wr.wr; } else { - wr->wr.next = &imm_wr; + wr->wr.next = &imm_wr.wr; } /* * From time to time we have to post signaled sends, @@ -294,16 +296,18 @@ static int rdma_write_sg(struct rtrs_srv_op *id) inv_wr.sg_list = NULL; inv_wr.num_sge = 0; inv_wr.opcode = IB_WR_SEND_WITH_INV; + inv_wr.wr_cqe = &io_comp_cqe; inv_wr.send_flags = 0; inv_wr.ex.invalidate_rkey = rkey; } - imm_wr.next = NULL; + imm_wr.wr.next = NULL; if (always_invalidate) { struct rtrs_msg_rkey_rsp *msg; srv_mr = &sess->mrs[id->msg_id]; rwr.wr.opcode = IB_WR_REG_MR; + rwr.wr.wr_cqe = &local_reg_cqe; rwr.wr.num_sge = 0; rwr.mr = srv_mr->mr; rwr.wr.send_flags = 0; @@ -318,22 +322,22 @@ static int rdma_write_sg(struct rtrs_srv_op *id) list.addr = srv_mr->iu->dma_addr; list.length = sizeof(*msg); list.lkey = sess->s.dev->ib_pd->local_dma_lkey; - imm_wr.sg_list = &list; - imm_wr.num_sge = 1; - imm_wr.opcode = IB_WR_SEND_WITH_IMM; + imm_wr.wr.sg_list = &list; + imm_wr.wr.num_sge = 1; + imm_wr.wr.opcode = IB_WR_SEND_WITH_IMM; ib_dma_sync_single_for_device(sess->s.dev->ib_dev, srv_mr->iu->dma_addr, srv_mr->iu->size, DMA_TO_DEVICE); } else { - imm_wr.sg_list = NULL; - imm_wr.num_sge = 0; - imm_wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; + imm_wr.wr.sg_list = NULL; + imm_wr.wr.num_sge = 0; + imm_wr.wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; } - imm_wr.send_flags = flags; - imm_wr.ex.imm_data = cpu_to_be32(rtrs_to_io_rsp_imm(id->msg_id, + imm_wr.wr.send_flags = flags; + imm_wr.wr.ex.imm_data = cpu_to_be32(rtrs_to_io_rsp_imm(id->msg_id, 0, need_inval)); - imm_wr.wr_cqe = &io_comp_cqe; + imm_wr.wr.wr_cqe = &io_comp_cqe; ib_dma_sync_single_for_device(sess->s.dev->ib_dev, dma_addr, offset, DMA_BIDIRECTIONAL); @@ -360,7 +364,8 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, { struct rtrs_sess *s = con->c.sess; struct rtrs_srv_sess *sess = to_srv_sess(s); - struct ib_send_wr inv_wr, imm_wr, *wr = NULL; + struct ib_send_wr inv_wr, *wr = NULL; + struct ib_rdma_wr imm_wr; struct ib_reg_wr rwr; struct rtrs_srv *srv = sess->srv; struct rtrs_srv_mr *srv_mr; @@ -379,6 +384,7 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, if (need_inval) { if (likely(sg_cnt)) { + inv_wr.wr_cqe = &io_comp_cqe; inv_wr.sg_list = NULL; inv_wr.num_sge = 0; inv_wr.opcode = IB_WR_SEND_WITH_INV; @@ -396,15 +402,15 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, if (need_inval && always_invalidate) { wr = &inv_wr; inv_wr.next = &rwr.wr; - rwr.wr.next = &imm_wr; + rwr.wr.next = &imm_wr.wr; } else if (always_invalidate) { wr = &rwr.wr; - rwr.wr.next = &imm_wr; + rwr.wr.next = &imm_wr.wr; } else if (need_inval) { wr = &inv_wr; - inv_wr.next = &imm_wr; + inv_wr.next = &imm_wr.wr; } else { - wr = &imm_wr; + wr = &imm_wr.wr; } /* * From time to time we have to post signalled sends, @@ -413,14 +419,15 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, flags = (atomic_inc_return(&con->wr_cnt) % srv->queue_depth) ? 0 : IB_SEND_SIGNALED; imm = rtrs_to_io_rsp_imm(id->msg_id, errno, need_inval); - imm_wr.next = NULL; + imm_wr.wr.next = NULL; if (always_invalidate) { struct ib_sge list; struct rtrs_msg_rkey_rsp *msg; srv_mr = &sess->mrs[id->msg_id]; - rwr.wr.next = &imm_wr; + rwr.wr.next = &imm_wr.wr; rwr.wr.opcode = IB_WR_REG_MR; + rwr.wr.wr_cqe = &local_reg_cqe; rwr.wr.num_sge = 0; rwr.wr.send_flags = 0; rwr.mr = srv_mr->mr; @@ -435,21 +442,21 @@ static int send_io_resp_imm(struct rtrs_srv_con *con, struct rtrs_srv_op *id, list.addr = srv_mr->iu->dma_addr; list.length = sizeof(*msg); list.lkey = sess->s.dev->ib_pd->local_dma_lkey; - imm_wr.sg_list = &list; - imm_wr.num_sge = 1; - imm_wr.opcode = IB_WR_SEND_WITH_IMM; + imm_wr.wr.sg_list = &list; + imm_wr.wr.num_sge = 1; + imm_wr.wr.opcode = IB_WR_SEND_WITH_IMM; ib_dma_sync_single_for_device(sess->s.dev->ib_dev, srv_mr->iu->dma_addr, srv_mr->iu->size, DMA_TO_DEVICE); } else { - imm_wr.sg_list = NULL; - imm_wr.num_sge = 0; - imm_wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; + imm_wr.wr.sg_list = NULL; + imm_wr.wr.num_sge = 0; + imm_wr.wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM; } - imm_wr.send_flags = flags; - imm_wr.wr_cqe = &io_comp_cqe; + imm_wr.wr.send_flags = flags; + imm_wr.wr.wr_cqe = &io_comp_cqe; - imm_wr.ex.imm_data = cpu_to_be32(imm); + imm_wr.wr.ex.imm_data = cpu_to_be32(imm); err = ib_post_send(id->con->c.qp, wr, NULL); if (unlikely(err)) @@ -651,7 +658,7 @@ static int map_cont_bufs(struct rtrs_srv_sess *sess) if (!srv_mr->iu) { err = -ENOMEM; rtrs_err(ss, "rtrs_iu_alloc(), err: %d\n", err); - goto free_iu; + goto dereg_mr; } } /* Eventually dma addr for each chunk can be cached */ @@ -667,7 +674,6 @@ err: srv_mr = &sess->mrs[mri]; sgt = &srv_mr->sgt; mr = srv_mr->mr; -free_iu: rtrs_iu_free(srv_mr->iu, sess->s.dev->ib_dev, 1); dereg_mr: ib_dereg_mr(mr); @@ -814,7 +820,7 @@ static int process_info_req(struct rtrs_srv_con *con, rwr[mri].wr.opcode = IB_WR_REG_MR; rwr[mri].wr.wr_cqe = &local_reg_cqe; rwr[mri].wr.num_sge = 0; - rwr[mri].wr.send_flags = mri ? 0 : IB_SEND_SIGNALED; + rwr[mri].wr.send_flags = 0; rwr[mri].mr = mr; rwr[mri].key = mr->rkey; rwr[mri].access = (IB_ACCESS_LOCAL_WRITE | @@ -1238,7 +1244,6 @@ static void rtrs_srv_rdma_done(struct ib_cq *cq, struct ib_wc *wc) case IB_WC_SEND: /* * post_send() RDMA write completions of IO reqs (read/write) - * and hb */ atomic_add(srv->queue_depth, &con->sq_wr_avail); @@ -1328,7 +1333,8 @@ static void free_srv(struct rtrs_srv *srv) } static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, - const uuid_t *paths_uuid) + const uuid_t *paths_uuid, + bool first_conn) { struct rtrs_srv *srv; int i; @@ -1341,13 +1347,18 @@ static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, return srv; } } + mutex_unlock(&ctx->srv_mutex); + /* + * If this request is not the first connection request from the + * client for this session then fail and return error. + */ + if (!first_conn) + return ERR_PTR(-ENXIO); /* need to allocate a new srv */ srv = kzalloc(sizeof(*srv), GFP_KERNEL); - if (!srv) { - mutex_unlock(&ctx->srv_mutex); - return NULL; - } + if (!srv) + return ERR_PTR(-ENOMEM); INIT_LIST_HEAD(&srv->paths_list); mutex_init(&srv->paths_mutex); @@ -1357,8 +1368,6 @@ static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, srv->ctx = ctx; device_initialize(&srv->dev); srv->dev.release = rtrs_srv_dev_release; - list_add(&srv->ctx_list, &ctx->srv_list); - mutex_unlock(&ctx->srv_mutex); srv->chunks = kcalloc(srv->queue_depth, sizeof(*srv->chunks), GFP_KERNEL); @@ -1371,6 +1380,9 @@ static struct rtrs_srv *get_or_create_srv(struct rtrs_srv_ctx *ctx, goto err_free_chunks; } refcount_set(&srv->refcount, 1); + mutex_lock(&ctx->srv_mutex); + list_add(&srv->ctx_list, &ctx->srv_list); + mutex_unlock(&ctx->srv_mutex); return srv; @@ -1381,7 +1393,7 @@ err_free_chunks: err_free_srv: kfree(srv); - return NULL; + return ERR_PTR(-ENOMEM); } static void put_srv(struct rtrs_srv *srv) @@ -1461,10 +1473,12 @@ static bool __is_path_w_addr_exists(struct rtrs_srv *srv, static void free_sess(struct rtrs_srv_sess *sess) { - if (sess->kobj.state_in_sysfs) + if (sess->kobj.state_in_sysfs) { + kobject_del(&sess->kobj); kobject_put(&sess->kobj); - else + } else { kfree(sess); + } } static void rtrs_srv_close_work(struct work_struct *work) @@ -1586,7 +1600,7 @@ static int create_con(struct rtrs_srv_sess *sess, struct rtrs_sess *s = &sess->s; struct rtrs_srv_con *con; - u16 cq_size, wr_queue_size; + u32 cq_size, wr_queue_size; int err, cq_vector; con = kzalloc(sizeof(*con), GFP_KERNEL); @@ -1600,7 +1614,7 @@ static int create_con(struct rtrs_srv_sess *sess, con->c.cm_id = cm_id; con->c.sess = &sess->s; con->c.cid = cid; - atomic_set(&con->wr_cnt, 0); + atomic_set(&con->wr_cnt, 1); if (con->c.cid == 0) { /* @@ -1630,7 +1644,8 @@ static int create_con(struct rtrs_srv_sess *sess, /* TODO: SOFTIRQ can be faster, but be careful with softirq context */ err = rtrs_cq_qp_create(&sess->s, &con->c, 1, cq_vector, cq_size, - wr_queue_size, IB_POLL_WORKQUEUE); + wr_queue_size, wr_queue_size, + IB_POLL_WORKQUEUE); if (err) { rtrs_err(s, "rtrs_cq_qp_create(), err: %d\n", err); goto free_con; @@ -1781,13 +1796,9 @@ static int rtrs_rdma_connect(struct rdma_cm_id *cm_id, goto reject_w_econnreset; } recon_cnt = le16_to_cpu(msg->recon_cnt); - srv = get_or_create_srv(ctx, &msg->paths_uuid); - /* - * "refcount == 0" happens if a previous thread calls get_or_create_srv - * allocate srv, but chunks of srv are not allocated yet. - */ - if (!srv || refcount_read(&srv->refcount) == 0) { - err = -ENOMEM; + srv = get_or_create_srv(ctx, &msg->paths_uuid, msg->first_conn); + if (IS_ERR(srv)) { + err = PTR_ERR(srv); goto reject_w_err; } mutex_lock(&srv->paths_mutex); @@ -1862,8 +1873,8 @@ reject_w_econnreset: return rtrs_rdma_do_reject(cm_id, -ECONNRESET); close_and_return_err: - close_sess(sess); mutex_unlock(&srv->paths_mutex); + close_sess(sess); return err; } diff --git a/drivers/infiniband/ulp/rtrs/rtrs.c b/drivers/infiniband/ulp/rtrs/rtrs.c index 2e3a849e0a77..d13aff0aa816 100644 --- a/drivers/infiniband/ulp/rtrs/rtrs.c +++ b/drivers/infiniband/ulp/rtrs/rtrs.c @@ -182,16 +182,16 @@ int rtrs_post_rdma_write_imm_empty(struct rtrs_con *con, struct ib_cqe *cqe, u32 imm_data, enum ib_send_flags flags, struct ib_send_wr *head) { - struct ib_send_wr wr; + struct ib_rdma_wr wr; - wr = (struct ib_send_wr) { - .wr_cqe = cqe, - .send_flags = flags, - .opcode = IB_WR_RDMA_WRITE_WITH_IMM, - .ex.imm_data = cpu_to_be32(imm_data), + wr = (struct ib_rdma_wr) { + .wr.wr_cqe = cqe, + .wr.send_flags = flags, + .wr.opcode = IB_WR_RDMA_WRITE_WITH_IMM, + .wr.ex.imm_data = cpu_to_be32(imm_data), }; - return rtrs_post_send(con->qp, head, &wr); + return rtrs_post_send(con->qp, head, &wr.wr); } EXPORT_SYMBOL_GPL(rtrs_post_rdma_write_imm_empty); @@ -231,14 +231,14 @@ static int create_cq(struct rtrs_con *con, int cq_vector, u16 cq_size, } static int create_qp(struct rtrs_con *con, struct ib_pd *pd, - u16 wr_queue_size, u32 max_sge) + u32 max_send_wr, u32 max_recv_wr, u32 max_sge) { struct ib_qp_init_attr init_attr = {NULL}; struct rdma_cm_id *cm_id = con->cm_id; int ret; - init_attr.cap.max_send_wr = wr_queue_size; - init_attr.cap.max_recv_wr = wr_queue_size; + init_attr.cap.max_send_wr = max_send_wr; + init_attr.cap.max_recv_wr = max_recv_wr; init_attr.cap.max_recv_sge = 1; init_attr.event_handler = qp_event_handler; init_attr.qp_context = con; @@ -260,8 +260,9 @@ static int create_qp(struct rtrs_con *con, struct ib_pd *pd, } int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, - u32 max_send_sge, int cq_vector, u16 cq_size, - u16 wr_queue_size, enum ib_poll_context poll_ctx) + u32 max_send_sge, int cq_vector, int cq_size, + u32 max_send_wr, u32 max_recv_wr, + enum ib_poll_context poll_ctx) { int err; @@ -269,7 +270,8 @@ int rtrs_cq_qp_create(struct rtrs_sess *sess, struct rtrs_con *con, if (err) return err; - err = create_qp(con, sess->dev->ib_pd, wr_queue_size, max_send_sge); + err = create_qp(con, sess->dev->ib_pd, max_send_wr, max_recv_wr, + max_send_sge); if (err) { ib_free_cq(con->cq); con->cq = NULL; @@ -308,7 +310,7 @@ void rtrs_send_hb_ack(struct rtrs_sess *sess) imm = rtrs_to_imm(RTRS_HB_ACK_IMM, 0); err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, - IB_SEND_SIGNALED, NULL); + 0, NULL); if (err) { sess->hb_err_handler(usr_con); return; @@ -337,7 +339,7 @@ static void hb_work(struct work_struct *work) } imm = rtrs_to_imm(RTRS_HB_MSG_IMM, 0); err = rtrs_post_rdma_write_imm_empty(usr_con, sess->hb_cqe, imm, - IB_SEND_SIGNALED, NULL); + 0, NULL); if (err) { sess->hb_err_handler(usr_con); return; diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index 5492b66a8153..31f8aa2c40ed 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -3628,7 +3628,7 @@ static ssize_t srp_create_target(struct device *dev, struct srp_rdma_ch *ch; struct srp_device *srp_dev = host->srp_dev; struct ib_device *ibdev = srp_dev->dev; - int ret, node_idx, node, cpu, i; + int ret, i, ch_idx; unsigned int max_sectors_per_mr, mr_per_cmd = 0; bool multich = false; uint32_t max_iu_len; @@ -3753,81 +3753,61 @@ static ssize_t srp_create_target(struct device *dev, goto out; ret = -ENOMEM; - if (target->ch_count == 0) + if (target->ch_count == 0) { target->ch_count = - max_t(unsigned int, num_online_nodes(), - min(ch_count ?: - min(4 * num_online_nodes(), - ibdev->num_comp_vectors), - num_online_cpus())); + min(ch_count ?: + max(4 * num_online_nodes(), + ibdev->num_comp_vectors), + num_online_cpus()); + } + target->ch = kcalloc(target->ch_count, sizeof(*target->ch), GFP_KERNEL); if (!target->ch) goto out; - node_idx = 0; - for_each_online_node(node) { - const int ch_start = (node_idx * target->ch_count / - num_online_nodes()); - const int ch_end = ((node_idx + 1) * target->ch_count / - num_online_nodes()); - const int cv_start = node_idx * ibdev->num_comp_vectors / - num_online_nodes(); - const int cv_end = (node_idx + 1) * ibdev->num_comp_vectors / - num_online_nodes(); - int cpu_idx = 0; - - for_each_online_cpu(cpu) { - if (cpu_to_node(cpu) != node) - continue; - if (ch_start + cpu_idx >= ch_end) - continue; - ch = &target->ch[ch_start + cpu_idx]; - ch->target = target; - ch->comp_vector = cv_start == cv_end ? cv_start : - cv_start + cpu_idx % (cv_end - cv_start); - spin_lock_init(&ch->lock); - INIT_LIST_HEAD(&ch->free_tx); - ret = srp_new_cm_id(ch); - if (ret) - goto err_disconnect; + for (ch_idx = 0; ch_idx < target->ch_count; ++ch_idx) { + ch = &target->ch[ch_idx]; + ch->target = target; + ch->comp_vector = ch_idx % ibdev->num_comp_vectors; + spin_lock_init(&ch->lock); + INIT_LIST_HEAD(&ch->free_tx); + ret = srp_new_cm_id(ch); + if (ret) + goto err_disconnect; - ret = srp_create_ch_ib(ch); - if (ret) - goto err_disconnect; + ret = srp_create_ch_ib(ch); + if (ret) + goto err_disconnect; - ret = srp_alloc_req_data(ch); - if (ret) - goto err_disconnect; + ret = srp_alloc_req_data(ch); + if (ret) + goto err_disconnect; - ret = srp_connect_ch(ch, max_iu_len, multich); - if (ret) { - char dst[64]; - - if (target->using_rdma_cm) - snprintf(dst, sizeof(dst), "%pIS", - &target->rdma_cm.dst); - else - snprintf(dst, sizeof(dst), "%pI6", - target->ib_cm.orig_dgid.raw); - shost_printk(KERN_ERR, target->scsi_host, - PFX "Connection %d/%d to %s failed\n", - ch_start + cpu_idx, - target->ch_count, dst); - if (node_idx == 0 && cpu_idx == 0) { - goto free_ch; - } else { - srp_free_ch_ib(target, ch); - srp_free_req_data(target, ch); - target->ch_count = ch - target->ch; - goto connected; - } - } + ret = srp_connect_ch(ch, max_iu_len, multich); + if (ret) { + char dst[64]; - multich = true; - cpu_idx++; + if (target->using_rdma_cm) + snprintf(dst, sizeof(dst), "%pIS", + &target->rdma_cm.dst); + else + snprintf(dst, sizeof(dst), "%pI6", + target->ib_cm.orig_dgid.raw); + shost_printk(KERN_ERR, target->scsi_host, + PFX "Connection %d/%d to %s failed\n", + ch_idx, + target->ch_count, dst); + if (ch_idx == 0) { + goto free_ch; + } else { + srp_free_ch_ib(target, ch); + srp_free_req_data(target, ch); + target->ch_count = ch - target->ch; + goto connected; + } } - node_idx++; + multich = true; } connected: diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mr.c b/drivers/net/ethernet/mellanox/mlx5/core/mr.c index 9eb51f06d3ae..50af84e76fb6 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/mr.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/mr.c @@ -56,6 +56,7 @@ int mlx5_core_create_mkey(struct mlx5_core_dev *dev, mkey->size = MLX5_GET64(mkc, mkc, len); mkey->key |= mlx5_idx_to_mkey(mkey_index); mkey->pd = MLX5_GET(mkc, mkc, pd); + init_waitqueue_head(&mkey->wait); mlx5_core_dbg(dev, "out 0x%x, mkey 0x%x\n", mkey_index, mkey->key); return 0; diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index 6ea8d67e3cb8..53b89631a1d9 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -307,13 +307,6 @@ struct mlx5_cmd { struct mlx5_cmd_stats *stats; }; -struct mlx5_port_caps { - int gid_table_len; - int pkey_table_len; - u8 ext_port_cap; - bool has_smi; -}; - struct mlx5_cmd_mailbox { void *buf; dma_addr_t dma; @@ -375,6 +368,8 @@ struct mlx5_core_mkey { u32 key; u32 pd; u32 type; + struct wait_queue_head wait; + refcount_t usecount; }; #define MLX5_24BIT_MASK ((1 << 24) - 1) @@ -713,7 +708,6 @@ struct mlx5_core_dev { u8 rev_id; char board_id[MLX5_BOARD_ID_LEN]; struct mlx5_cmd cmd; - struct mlx5_port_caps port_caps[MLX5_MAX_PORTS]; struct { u32 hca_cur[MLX5_CAP_NUM][MLX5_UN_SZ_DW(hca_cap_union)]; u32 hca_max[MLX5_CAP_NUM][MLX5_UN_SZ_DW(hca_cap_union)]; diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 6f0b866fb495..df5d91c8b2d4 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -1661,7 +1661,8 @@ struct mlx5_ifc_cmd_hca_cap_bits { u8 sf_set_partition[0x1]; u8 reserved_at_682[0x1]; u8 log_max_sf[0x5]; - u8 reserved_at_688[0x8]; + u8 apu[0x1]; + u8 reserved_at_689[0x7]; u8 log_min_sf_size[0x8]; u8 max_num_sf_partitions[0x8]; @@ -3868,7 +3869,7 @@ struct mlx5_ifc_cqc_bits { u8 status[0x4]; u8 reserved_at_4[0x2]; u8 dbr_umem_valid[0x1]; - u8 reserved_at_7[0x1]; + u8 apu_thread_cq[0x1]; u8 cqe_sz[0x3]; u8 cc[0x1]; u8 reserved_at_c[0x1]; diff --git a/include/rdma/ib_sa.h b/include/rdma/ib_sa.h index 693285e76f13..4c52c2fd22a1 100644 --- a/include/rdma/ib_sa.h +++ b/include/rdma/ib_sa.h @@ -547,10 +547,6 @@ int ib_sa_guid_info_rec_query(struct ib_sa_client *client, void *context), void *context, struct ib_sa_query **sa_query); -bool ib_sa_sendonly_fullmem_support(struct ib_sa_client *client, - struct ib_device *device, - u8 port_num); - static inline bool sa_path_is_roce(struct sa_path_rec *rec) { return ((rec->rec_type == SA_PATH_REC_TYPE_ROCE_V1) || diff --git a/include/rdma/ib_umem.h b/include/rdma/ib_umem.h index 7752211c9638..676c57f5ca80 100644 --- a/include/rdma/ib_umem.h +++ b/include/rdma/ib_umem.h @@ -1,6 +1,7 @@ /* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ /* * Copyright (c) 2007 Cisco Systems. All rights reserved. + * Copyright (c) 2020 Intel Corporation. All rights reserved. */ #ifndef IB_UMEM_H @@ -13,6 +14,7 @@ struct ib_ucontext; struct ib_umem_odp; +struct dma_buf_attach_ops; struct ib_umem { struct ib_device *ibdev; @@ -22,12 +24,29 @@ struct ib_umem { unsigned long address; u32 writable : 1; u32 is_odp : 1; + u32 is_dmabuf : 1; struct work_struct work; struct sg_table sg_head; int nmap; unsigned int sg_nents; }; +struct ib_umem_dmabuf { + struct ib_umem umem; + struct dma_buf_attachment *attach; + struct sg_table *sgt; + struct scatterlist *first_sg; + struct scatterlist *last_sg; + unsigned long first_sg_offset; + unsigned long last_sg_trim; + void *private; +}; + +static inline struct ib_umem_dmabuf *to_ib_umem_dmabuf(struct ib_umem *umem) +{ + return container_of(umem, struct ib_umem_dmabuf, umem); +} + /* Returns the offset of the umem start relative to the first page. */ static inline int ib_umem_offset(struct ib_umem *umem) { @@ -86,6 +105,7 @@ int ib_umem_copy_from(void *dst, struct ib_umem *umem, size_t offset, unsigned long ib_umem_find_best_pgsz(struct ib_umem *umem, unsigned long pgsz_bitmap, unsigned long virt); + /** * ib_umem_find_best_pgoff - Find best HW page size * @@ -116,6 +136,14 @@ static inline unsigned long ib_umem_find_best_pgoff(struct ib_umem *umem, dma_addr & pgoff_bitmask); } +struct ib_umem_dmabuf *ib_umem_dmabuf_get(struct ib_device *device, + unsigned long offset, size_t size, + int fd, int access, + const struct dma_buf_attach_ops *ops); +int ib_umem_dmabuf_map_pages(struct ib_umem_dmabuf *umem_dmabuf); +void ib_umem_dmabuf_unmap_pages(struct ib_umem_dmabuf *umem_dmabuf); +void ib_umem_dmabuf_release(struct ib_umem_dmabuf *umem_dmabuf); + #else /* CONFIG_INFINIBAND_USER_MEM */ #include <linux/err.h> @@ -124,12 +152,12 @@ static inline struct ib_umem *ib_umem_get(struct ib_device *device, unsigned long addr, size_t size, int access) { - return ERR_PTR(-EINVAL); + return ERR_PTR(-EOPNOTSUPP); } static inline void ib_umem_release(struct ib_umem *umem) { } static inline int ib_umem_copy_from(void *dst, struct ib_umem *umem, size_t offset, size_t length) { - return -EINVAL; + return -EOPNOTSUPP; } static inline unsigned long ib_umem_find_best_pgsz(struct ib_umem *umem, unsigned long pgsz_bitmap, @@ -143,7 +171,21 @@ static inline unsigned long ib_umem_find_best_pgoff(struct ib_umem *umem, { return 0; } +static inline +struct ib_umem_dmabuf *ib_umem_dmabuf_get(struct ib_device *device, + unsigned long offset, + size_t size, int fd, + int access, + struct dma_buf_attach_ops *ops) +{ + return ERR_PTR(-EOPNOTSUPP); +} +static inline int ib_umem_dmabuf_map_pages(struct ib_umem_dmabuf *umem_dmabuf) +{ + return -EOPNOTSUPP; +} +static inline void ib_umem_dmabuf_unmap_pages(struct ib_umem_dmabuf *umem_dmabuf) { } +static inline void ib_umem_dmabuf_release(struct ib_umem_dmabuf *umem_dmabuf) { } #endif /* CONFIG_INFINIBAND_USER_MEM */ - #endif /* IB_UMEM_H */ diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 9fed65bf9279..ca28fca5736b 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -2,7 +2,7 @@ /* * Copyright (c) 2004 Mellanox Technologies Ltd. All rights reserved. * Copyright (c) 2004 Infinicon Corporation. All rights reserved. - * Copyright (c) 2004 Intel Corporation. All rights reserved. + * Copyright (c) 2004, 2020 Intel Corporation. All rights reserved. * Copyright (c) 2004 Topspin Corporation. All rights reserved. * Copyright (c) 2004 Voltaire Corporation. All rights reserved. * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved. @@ -2434,6 +2434,10 @@ struct ib_device_ops { struct ib_mr *(*reg_user_mr)(struct ib_pd *pd, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_udata *udata); + struct ib_mr *(*reg_user_mr_dmabuf)(struct ib_pd *pd, u64 offset, + u64 length, u64 virt_addr, int fd, + int mr_access_flags, + struct ib_udata *udata); struct ib_mr *(*rereg_user_mr)(struct ib_mr *mr, int flags, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_pd *pd, @@ -4670,4 +4674,7 @@ static inline u32 rdma_calc_flow_label(u32 lqpn, u32 rqpn) return (u32)(v & IB_GRH_FLOWLABEL_MASK); } + +const struct ib_port_immutable* +ib_port_immutable_read(struct ib_device *dev, unsigned int port); #endif /* IB_VERBS_H */ diff --git a/include/rdma/rdma_counter.h b/include/rdma/rdma_counter.h index eb99856e8b30..e75cf9742e04 100644 --- a/include/rdma/rdma_counter.h +++ b/include/rdma/rdma_counter.h @@ -46,7 +46,8 @@ struct rdma_counter { void rdma_counter_init(struct ib_device *dev); void rdma_counter_release(struct ib_device *dev); int rdma_counter_set_auto_mode(struct ib_device *dev, u8 port, - bool on, enum rdma_nl_counter_mask mask); + enum rdma_nl_counter_mask mask, + struct netlink_ext_ack *extack); int rdma_counter_bind_qp_auto(struct ib_qp *qp, u8 port); int rdma_counter_unbind_qp(struct ib_qp *qp, bool force); diff --git a/include/uapi/rdma/ib_user_ioctl_cmds.h b/include/uapi/rdma/ib_user_ioctl_cmds.h index 7968a1845355..dafc7ebe545b 100644 --- a/include/uapi/rdma/ib_user_ioctl_cmds.h +++ b/include/uapi/rdma/ib_user_ioctl_cmds.h @@ -1,5 +1,6 @@ /* * Copyright (c) 2018, Mellanox Technologies inc. All rights reserved. + * Copyright (c) 2020, Intel Corporation. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -251,6 +252,7 @@ enum uverbs_methods_mr { UVERBS_METHOD_MR_DESTROY, UVERBS_METHOD_ADVISE_MR, UVERBS_METHOD_QUERY_MR, + UVERBS_METHOD_REG_DMABUF_MR, }; enum uverbs_attrs_mr_destroy_ids { @@ -272,6 +274,18 @@ enum uverbs_attrs_query_mr_cmd_attr_ids { UVERBS_ATTR_QUERY_MR_RESP_IOVA, }; +enum uverbs_attrs_reg_dmabuf_mr_cmd_attr_ids { + UVERBS_ATTR_REG_DMABUF_MR_HANDLE, + UVERBS_ATTR_REG_DMABUF_MR_PD_HANDLE, + UVERBS_ATTR_REG_DMABUF_MR_OFFSET, + UVERBS_ATTR_REG_DMABUF_MR_LENGTH, + UVERBS_ATTR_REG_DMABUF_MR_IOVA, + UVERBS_ATTR_REG_DMABUF_MR_FD, + UVERBS_ATTR_REG_DMABUF_MR_ACCESS_FLAGS, + UVERBS_ATTR_REG_DMABUF_MR_RESP_LKEY, + UVERBS_ATTR_REG_DMABUF_MR_RESP_RKEY, +}; + enum uverbs_attrs_create_counters_cmd_attr_ids { UVERBS_ATTR_CREATE_COUNTERS_HANDLE, }; diff --git a/tools/testing/scatterlist/main.c b/tools/testing/scatterlist/main.c index 71c960dcd8a4..652254754b4c 100644 --- a/tools/testing/scatterlist/main.c +++ b/tools/testing/scatterlist/main.c @@ -55,7 +55,6 @@ int main(void) struct test *test, tests[] = { { -EINVAL, 1, pfn(0), NULL, PAGE_SIZE, 0, 1 }, { 0, 1, pfn(0), NULL, PAGE_SIZE, PAGE_SIZE + 1, 1 }, - { 0, 1, pfn(0), NULL, PAGE_SIZE, sgmax + 1, 1 }, { 0, 1, pfn(0), NULL, PAGE_SIZE, sgmax, 1 }, { 0, 1, pfn(0), NULL, 1, sgmax, 1 }, { 0, 2, pfn(0, 1), NULL, 2 * PAGE_SIZE, sgmax, 1 }, |