diff options
153 files changed, 6825 insertions, 4142 deletions
diff --git a/drivers/infiniband/core/cma_configfs.c b/drivers/infiniband/core/cma_configfs.c index 8c2dfb3e294e..3ec2c415bb70 100644 --- a/drivers/infiniband/core/cma_configfs.c +++ b/drivers/infiniband/core/cma_configfs.c @@ -33,7 +33,10 @@ #include <linux/module.h> #include <linux/configfs.h> #include <rdma/ib_verbs.h> +#include <rdma/rdma_cm.h> + #include "core_priv.h" +#include "cma_priv.h" struct cma_device; diff --git a/drivers/infiniband/core/cma_priv.h b/drivers/infiniband/core/cma_priv.h index 194cfe78c447..cf47c69436a7 100644 --- a/drivers/infiniband/core/cma_priv.h +++ b/drivers/infiniband/core/cma_priv.h @@ -94,4 +94,32 @@ struct rdma_id_private { */ struct rdma_restrack_entry res; }; + +#if IS_ENABLED(CONFIG_INFINIBAND_ADDR_TRANS_CONFIGFS) +int cma_configfs_init(void); +void cma_configfs_exit(void); +#else +static inline int cma_configfs_init(void) +{ + return 0; +} + +static inline void cma_configfs_exit(void) +{ +} +#endif + +void cma_ref_dev(struct cma_device *dev); +void cma_deref_dev(struct cma_device *dev); +typedef bool (*cma_device_filter)(struct ib_device *, void *); +struct cma_device *cma_enum_devices_by_ibdev(cma_device_filter filter, + void *cookie); +int cma_get_default_gid_type(struct cma_device *dev, unsigned int port); +int cma_set_default_gid_type(struct cma_device *dev, unsigned int port, + enum ib_gid_type default_gid_type); +int cma_get_default_roce_tos(struct cma_device *dev, unsigned int port); +int cma_set_default_roce_tos(struct cma_device *dev, unsigned int port, + u8 default_roce_tos); +struct ib_device *cma_get_ib_dev(struct cma_device *dev); + #endif /* _CMA_PRIV_H */ diff --git a/drivers/infiniband/core/core_priv.h b/drivers/infiniband/core/core_priv.h index bb9007a0cca7..cc7535c5e192 100644 --- a/drivers/infiniband/core/core_priv.h +++ b/drivers/infiniband/core/core_priv.h @@ -54,35 +54,6 @@ struct pkey_index_qp_list { struct list_head qp_list; }; -#if IS_ENABLED(CONFIG_INFINIBAND_ADDR_TRANS_CONFIGFS) -int cma_configfs_init(void); -void cma_configfs_exit(void); -#else -static inline int cma_configfs_init(void) -{ - return 0; -} - -static inline void cma_configfs_exit(void) -{ -} -#endif -struct cma_device; -void cma_ref_dev(struct cma_device *cma_dev); -void cma_deref_dev(struct cma_device *cma_dev); -typedef bool (*cma_device_filter)(struct ib_device *, void *); -struct cma_device *cma_enum_devices_by_ibdev(cma_device_filter filter, - void *cookie); -int cma_get_default_gid_type(struct cma_device *cma_dev, - unsigned int port); -int cma_set_default_gid_type(struct cma_device *cma_dev, - unsigned int port, - enum ib_gid_type default_gid_type); -int cma_get_default_roce_tos(struct cma_device *cma_dev, unsigned int port); -int cma_set_default_roce_tos(struct cma_device *a_dev, unsigned int port, - u8 default_roce_tos); -struct ib_device *cma_get_ib_dev(struct cma_device *cma_dev); - int ib_device_register_sysfs(struct ib_device *device, int (*port_callback)(struct ib_device *, u8, struct kobject *)); @@ -296,6 +267,7 @@ static inline int ib_mad_enforce_security(struct ib_mad_agent_private *map, #endif struct ib_device *ib_device_get_by_index(u32 ifindex); +void ib_device_put(struct ib_device *device); /* RDMA device netlink */ void nldev_init(void); void nldev_exit(void); diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c index 87eb4f2cdd7d..348a7fb1f945 100644 --- a/drivers/infiniband/core/device.c +++ b/drivers/infiniband/core/device.c @@ -145,7 +145,8 @@ static struct ib_device *__ib_device_get_by_index(u32 index) } /* - * Caller is responsible to return refrerence count by calling put_device() + * Caller must perform ib_device_put() to return the device reference count + * when ib_device_get_by_index() returns valid device pointer. */ struct ib_device *ib_device_get_by_index(u32 index) { @@ -153,13 +154,21 @@ struct ib_device *ib_device_get_by_index(u32 index) down_read(&lists_rwsem); device = __ib_device_get_by_index(index); - if (device) - get_device(&device->dev); - + if (device) { + /* Do not return a device if unregistration has started. */ + if (!refcount_inc_not_zero(&device->refcount)) + device = NULL; + } up_read(&lists_rwsem); return device; } +void ib_device_put(struct ib_device *device) +{ + if (refcount_dec_and_test(&device->refcount)) + complete(&device->unreg_completion); +} + static struct ib_device *__ib_device_get_by_name(const char *name) { struct ib_device *device; @@ -293,6 +302,8 @@ struct ib_device *ib_alloc_device(size_t size) rwlock_init(&device->client_data_lock); INIT_LIST_HEAD(&device->client_data_list); INIT_LIST_HEAD(&device->port_list); + refcount_set(&device->refcount, 1); + init_completion(&device->unreg_completion); return device; } @@ -641,6 +652,13 @@ void ib_unregister_device(struct ib_device *device) struct ib_client_data *context, *tmp; unsigned long flags; + /* + * Wait for all netlink command callers to finish working on the + * device. + */ + ib_device_put(device); + wait_for_completion(&device->unreg_completion); + mutex_lock(&device_mutex); down_write(&lists_rwsem); @@ -1024,6 +1042,9 @@ int ib_enum_all_devs(nldev_callback nldev_cb, struct sk_buff *skb, int ib_query_pkey(struct ib_device *device, u8 port_num, u16 index, u16 *pkey) { + if (!rdma_is_port_valid(device, port_num)) + return -EINVAL; + return device->query_pkey(device, port_num, index, pkey); } EXPORT_SYMBOL(ib_query_pkey); diff --git a/drivers/infiniband/core/fmr_pool.c b/drivers/infiniband/core/fmr_pool.c index 83ba0068e8bb..b00dfd2ad31e 100644 --- a/drivers/infiniband/core/fmr_pool.c +++ b/drivers/infiniband/core/fmr_pool.c @@ -474,7 +474,7 @@ EXPORT_SYMBOL(ib_fmr_pool_map_phys); * Unmap an FMR. The FMR mapping may remain valid until the FMR is * reused (or until ib_flush_fmr_pool() is called). */ -int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) +void ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) { struct ib_fmr_pool *pool; unsigned long flags; @@ -503,7 +503,5 @@ int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr) #endif spin_unlock_irqrestore(&pool->pool_lock, flags); - - return 0; } EXPORT_SYMBOL(ib_fmr_pool_unmap); diff --git a/drivers/infiniband/core/nldev.c b/drivers/infiniband/core/nldev.c index 573399e3ccc1..9abbadb9e366 100644 --- a/drivers/infiniband/core/nldev.c +++ b/drivers/infiniband/core/nldev.c @@ -308,6 +308,7 @@ static int fill_res_info(struct sk_buff *msg, struct ib_device *device) [RDMA_RESTRACK_QP] = "qp", [RDMA_RESTRACK_CM_ID] = "cm_id", [RDMA_RESTRACK_MR] = "mr", + [RDMA_RESTRACK_CTX] = "ctx", }; struct rdma_restrack_root *res = &device->res; @@ -636,13 +637,13 @@ static int nldev_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, nlmsg_end(msg, nlh); - put_device(&device->dev); + ib_device_put(device); return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); err: - put_device(&device->dev); + ib_device_put(device); return err; } @@ -672,7 +673,7 @@ static int nldev_set_doit(struct sk_buff *skb, struct nlmsghdr *nlh, err = ib_device_rename(device, name); } - put_device(&device->dev); + ib_device_put(device); return err; } @@ -756,14 +757,14 @@ static int nldev_port_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, goto err_free; nlmsg_end(msg, nlh); - put_device(&device->dev); + ib_device_put(device); return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); err: - put_device(&device->dev); + ib_device_put(device); return err; } @@ -820,7 +821,7 @@ static int nldev_port_get_dumpit(struct sk_buff *skb, } out: - put_device(&device->dev); + ib_device_put(device); cb->args[0] = idx; return skb->len; } @@ -859,13 +860,13 @@ static int nldev_res_get_doit(struct sk_buff *skb, struct nlmsghdr *nlh, goto err_free; nlmsg_end(msg, nlh); - put_device(&device->dev); + ib_device_put(device); return rdma_nl_unicast(msg, NETLINK_CB(skb).portid); err_free: nlmsg_free(msg); err: - put_device(&device->dev); + ib_device_put(device); return ret; } @@ -1058,7 +1059,7 @@ next: idx++; if (!filled) goto err; - put_device(&device->dev); + ib_device_put(device); return skb->len; res_err: @@ -1069,7 +1070,7 @@ err: nlmsg_cancel(skb, nlh); err_index: - put_device(&device->dev); + ib_device_put(device); return ret; } diff --git a/drivers/infiniband/core/rdma_core.c b/drivers/infiniband/core/rdma_core.c index 752a55c6bdce..7d2f1ef75025 100644 --- a/drivers/infiniband/core/rdma_core.c +++ b/drivers/infiniband/core/rdma_core.c @@ -224,12 +224,14 @@ out_unlock: * uverbs_put_destroy. */ struct ib_uobject *__uobj_get_destroy(const struct uverbs_api_object *obj, - u32 id, struct ib_uverbs_file *ufile) + u32 id, + const struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj; int ret; - uobj = rdma_lookup_get_uobject(obj, ufile, id, UVERBS_LOOKUP_DESTROY); + uobj = rdma_lookup_get_uobject(obj, attrs->ufile, id, + UVERBS_LOOKUP_DESTROY); if (IS_ERR(uobj)) return uobj; @@ -243,21 +245,20 @@ struct ib_uobject *__uobj_get_destroy(const struct uverbs_api_object *obj, } /* - * Does both uobj_get_destroy() and uobj_put_destroy(). Returns success_res - * on success (negative errno on failure). For use by callers that do not need - * the uobj. + * Does both uobj_get_destroy() and uobj_put_destroy(). Returns 0 on success + * (negative errno on failure). For use by callers that do not need the uobj. */ int __uobj_perform_destroy(const struct uverbs_api_object *obj, u32 id, - struct ib_uverbs_file *ufile, int success_res) + const struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj; - uobj = __uobj_get_destroy(obj, id, ufile); + uobj = __uobj_get_destroy(obj, id, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_WRITE); - return success_res; + return 0; } /* alloc_uobj must be undone by uverbs_destroy_uobject() */ @@ -267,7 +268,7 @@ static struct ib_uobject *alloc_uobj(struct ib_uverbs_file *ufile, struct ib_uobject *uobj; struct ib_ucontext *ucontext; - ucontext = ib_uverbs_get_ucontext(ufile); + ucontext = ib_uverbs_get_ucontext_file(ufile); if (IS_ERR(ucontext)) return ERR_CAST(ucontext); @@ -397,16 +398,23 @@ struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_api_object *obj, struct ib_uobject *uobj; int ret; - if (!obj) - return ERR_PTR(-EINVAL); + if (IS_ERR(obj) && PTR_ERR(obj) == -ENOMSG) { + /* must be UVERBS_IDR_ANY_OBJECT, see uapi_get_object() */ + uobj = lookup_get_idr_uobject(NULL, ufile, id, mode); + if (IS_ERR(uobj)) + return uobj; + } else { + if (IS_ERR(obj)) + return ERR_PTR(-EINVAL); - uobj = obj->type_class->lookup_get(obj, ufile, id, mode); - if (IS_ERR(uobj)) - return uobj; + uobj = obj->type_class->lookup_get(obj, ufile, id, mode); + if (IS_ERR(uobj)) + return uobj; - if (uobj->uapi_object != obj) { - ret = -EINVAL; - goto free; + if (uobj->uapi_object != obj) { + ret = -EINVAL; + goto free; + } } /* @@ -426,7 +434,7 @@ struct ib_uobject *rdma_lookup_get_uobject(const struct uverbs_api_object *obj, return uobj; free: - obj->type_class->lookup_put(uobj, mode); + uobj->uapi_object->type_class->lookup_put(uobj, mode); uverbs_uobject_put(uobj); return ERR_PTR(ret); } @@ -490,7 +498,7 @@ struct ib_uobject *rdma_alloc_begin_uobject(const struct uverbs_api_object *obj, { struct ib_uobject *ret; - if (!obj) + if (IS_ERR(obj)) return ERR_PTR(-EINVAL); /* @@ -819,6 +827,8 @@ static void ufile_destroy_ucontext(struct ib_uverbs_file *ufile, ib_rdmacg_uncharge(&ucontext->cg_obj, ib_dev, RDMACG_RESOURCE_HCA_HANDLE); + rdma_restrack_del(&ucontext->res); + /* * FIXME: Drivers are not permitted to fail dealloc_ucontext, remove * the error return. diff --git a/drivers/infiniband/core/rdma_core.h b/drivers/infiniband/core/rdma_core.h index 4886d2bba7c7..b3ca7457ac42 100644 --- a/drivers/infiniband/core/rdma_core.h +++ b/drivers/infiniband/core/rdma_core.h @@ -118,43 +118,67 @@ void release_ufile_idr_uobject(struct ib_uverbs_file *ufile); * Depending on ID the slot pointer in the radix tree points at one of these * structs. */ -struct uverbs_api_object { - const struct uverbs_obj_type *type_attrs; - const struct uverbs_obj_type_class *type_class; -}; struct uverbs_api_ioctl_method { - int (__rcu *handler)(struct ib_uverbs_file *ufile, - struct uverbs_attr_bundle *ctx); + int(__rcu *handler)(struct uverbs_attr_bundle *attrs); DECLARE_BITMAP(attr_mandatory, UVERBS_API_ATTR_BKEY_LEN); u16 bundle_size; u8 use_stack:1; u8 driver_method:1; + u8 disabled:1; + u8 has_udata:1; u8 key_bitmap_len; u8 destroy_bkey; }; +struct uverbs_api_write_method { + int (*handler)(struct uverbs_attr_bundle *attrs); + u8 disabled:1; + u8 is_ex:1; + u8 has_udata:1; + u8 has_resp:1; + u8 req_size; + u8 resp_size; +}; + struct uverbs_api_attr { struct uverbs_attr_spec spec; }; -struct uverbs_api_object; struct uverbs_api { /* radix tree contains struct uverbs_api_* pointers */ struct radix_tree_root radix; enum rdma_driver_id driver_id; + + unsigned int num_write; + unsigned int num_write_ex; + struct uverbs_api_write_method notsupp_method; + const struct uverbs_api_write_method **write_methods; + const struct uverbs_api_write_method **write_ex_methods; }; +/* + * Get an uverbs_api_object that corresponds to the given object_id. + * Note: + * -ENOMSG means that any object is allowed to match during lookup. + */ static inline const struct uverbs_api_object * uapi_get_object(struct uverbs_api *uapi, u16 object_id) { - return radix_tree_lookup(&uapi->radix, uapi_key_obj(object_id)); + const struct uverbs_api_object *res; + + if (object_id == UVERBS_IDR_ANY_OBJECT) + return ERR_PTR(-ENOMSG); + + res = radix_tree_lookup(&uapi->radix, uapi_key_obj(object_id)); + if (!res) + return ERR_PTR(-ENOENT); + + return res; } char *uapi_key_format(char *S, unsigned int key); -struct uverbs_api *uverbs_alloc_api( - const struct uverbs_object_tree_def *const *driver_specs, - enum rdma_driver_id driver_id); +struct uverbs_api *uverbs_alloc_api(struct ib_device *ibdev); void uverbs_disassociate_api_pre(struct ib_uverbs_device *uverbs_dev); void uverbs_disassociate_api(struct uverbs_api *uapi); void uverbs_destroy_api(struct uverbs_api *uapi); @@ -162,4 +186,32 @@ void uapi_compute_bundle_size(struct uverbs_api_ioctl_method *method_elm, unsigned int num_attrs); void uverbs_user_mmap_disassociate(struct ib_uverbs_file *ufile); +extern const struct uapi_definition uverbs_def_obj_counters[]; +extern const struct uapi_definition uverbs_def_obj_cq[]; +extern const struct uapi_definition uverbs_def_obj_dm[]; +extern const struct uapi_definition uverbs_def_obj_flow_action[]; +extern const struct uapi_definition uverbs_def_obj_intf[]; +extern const struct uapi_definition uverbs_def_obj_mr[]; +extern const struct uapi_definition uverbs_def_write_intf[]; + +static inline const struct uverbs_api_write_method * +uapi_get_method(const struct uverbs_api *uapi, u32 command) +{ + u32 cmd_idx = command & IB_USER_VERBS_CMD_COMMAND_MASK; + + if (command & ~(u32)(IB_USER_VERBS_CMD_FLAG_EXTENDED | + IB_USER_VERBS_CMD_COMMAND_MASK)) + return ERR_PTR(-EINVAL); + + if (command & IB_USER_VERBS_CMD_FLAG_EXTENDED) { + if (cmd_idx >= uapi->num_write_ex) + return ERR_PTR(-EOPNOTSUPP); + return uapi->write_ex_methods[cmd_idx]; + } + + if (cmd_idx >= uapi->num_write) + return ERR_PTR(-EOPNOTSUPP); + return uapi->write_methods[cmd_idx]; +} + #endif /* RDMA_CORE_H */ diff --git a/drivers/infiniband/core/restrack.c b/drivers/infiniband/core/restrack.c index 06d8657ce583..3dd316159f5f 100644 --- a/drivers/infiniband/core/restrack.c +++ b/drivers/infiniband/core/restrack.c @@ -32,6 +32,7 @@ static const char *type2str(enum rdma_restrack_type type) [RDMA_RESTRACK_QP] = "QP", [RDMA_RESTRACK_CM_ID] = "CM_ID", [RDMA_RESTRACK_MR] = "MR", + [RDMA_RESTRACK_CTX] = "CTX", }; return names[type]; @@ -130,6 +131,8 @@ static struct ib_device *res_to_dev(struct rdma_restrack_entry *res) res)->id.device; case RDMA_RESTRACK_MR: return container_of(res, struct ib_mr, res)->device; + case RDMA_RESTRACK_CTX: + return container_of(res, struct ib_ucontext, res)->device; default: WARN_ONCE(true, "Wrong resource tracking type %u\n", res->type); return NULL; @@ -149,6 +152,8 @@ static bool res_is_user(struct rdma_restrack_entry *res) return !res->kern_name; case RDMA_RESTRACK_MR: return container_of(res, struct ib_mr, res)->pd->uobject; + case RDMA_RESTRACK_CTX: + return true; default: WARN_ONCE(true, "Wrong resource tracking type %u\n", res->type); return false; diff --git a/drivers/infiniband/core/umem_odp.c b/drivers/infiniband/core/umem_odp.c index 676c1fd1119d..9608681224e6 100644 --- a/drivers/infiniband/core/umem_odp.c +++ b/drivers/infiniband/core/umem_odp.c @@ -647,8 +647,13 @@ int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 user_virt, flags, local_page_list, NULL, NULL); up_read(&owning_mm->mmap_sem); - if (npages < 0) + if (npages < 0) { + if (npages != -EAGAIN) + pr_warn("fail to get %zu user pages with error %d\n", gup_num_pages, npages); + else + pr_debug("fail to get %zu user pages with error %d\n", gup_num_pages, npages); break; + } bcnt -= min_t(size_t, npages << PAGE_SHIFT, bcnt); mutex_lock(&umem_odp->umem_mutex); @@ -666,8 +671,13 @@ int ib_umem_odp_map_dma_pages(struct ib_umem_odp *umem_odp, u64 user_virt, ret = ib_umem_odp_map_dma_single_page( umem_odp, k, local_page_list[j], access_mask, current_seq); - if (ret < 0) + if (ret < 0) { + if (ret != -EAGAIN) + pr_warn("ib_umem_odp_map_dma_single_page failed with error %d\n", ret); + else + pr_debug("ib_umem_odp_map_dma_single_page failed with error %d\n", ret); break; + } p = page_to_phys(local_page_list[j]); k++; diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index c97935a0c7c6..8b41c95300c6 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -161,9 +161,6 @@ struct ib_uverbs_file { struct mutex umap_lock; struct list_head umaps; - u64 uverbs_cmd_mask; - u64 uverbs_ex_cmd_mask; - struct idr idr; /* spinlock protects write access to idr */ spinlock_t idr_lock; @@ -297,63 +294,4 @@ extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_FLOW_ACTION); extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_DM); extern const struct uverbs_object_def UVERBS_OBJECT(UVERBS_OBJECT_COUNTERS); -#define IB_UVERBS_DECLARE_CMD(name) \ - ssize_t ib_uverbs_##name(struct ib_uverbs_file *file, \ - const char __user *buf, int in_len, \ - int out_len) - -IB_UVERBS_DECLARE_CMD(get_context); -IB_UVERBS_DECLARE_CMD(query_device); -IB_UVERBS_DECLARE_CMD(query_port); -IB_UVERBS_DECLARE_CMD(alloc_pd); -IB_UVERBS_DECLARE_CMD(dealloc_pd); -IB_UVERBS_DECLARE_CMD(reg_mr); -IB_UVERBS_DECLARE_CMD(rereg_mr); -IB_UVERBS_DECLARE_CMD(dereg_mr); -IB_UVERBS_DECLARE_CMD(alloc_mw); -IB_UVERBS_DECLARE_CMD(dealloc_mw); -IB_UVERBS_DECLARE_CMD(create_comp_channel); -IB_UVERBS_DECLARE_CMD(create_cq); -IB_UVERBS_DECLARE_CMD(resize_cq); -IB_UVERBS_DECLARE_CMD(poll_cq); -IB_UVERBS_DECLARE_CMD(req_notify_cq); -IB_UVERBS_DECLARE_CMD(destroy_cq); -IB_UVERBS_DECLARE_CMD(create_qp); -IB_UVERBS_DECLARE_CMD(open_qp); -IB_UVERBS_DECLARE_CMD(query_qp); -IB_UVERBS_DECLARE_CMD(modify_qp); -IB_UVERBS_DECLARE_CMD(destroy_qp); -IB_UVERBS_DECLARE_CMD(post_send); -IB_UVERBS_DECLARE_CMD(post_recv); -IB_UVERBS_DECLARE_CMD(post_srq_recv); -IB_UVERBS_DECLARE_CMD(create_ah); -IB_UVERBS_DECLARE_CMD(destroy_ah); -IB_UVERBS_DECLARE_CMD(attach_mcast); -IB_UVERBS_DECLARE_CMD(detach_mcast); -IB_UVERBS_DECLARE_CMD(create_srq); -IB_UVERBS_DECLARE_CMD(modify_srq); -IB_UVERBS_DECLARE_CMD(query_srq); -IB_UVERBS_DECLARE_CMD(destroy_srq); -IB_UVERBS_DECLARE_CMD(create_xsrq); -IB_UVERBS_DECLARE_CMD(open_xrcd); -IB_UVERBS_DECLARE_CMD(close_xrcd); - -#define IB_UVERBS_DECLARE_EX_CMD(name) \ - int ib_uverbs_ex_##name(struct ib_uverbs_file *file, \ - struct ib_udata *ucore, \ - struct ib_udata *uhw) - -IB_UVERBS_DECLARE_EX_CMD(create_flow); -IB_UVERBS_DECLARE_EX_CMD(destroy_flow); -IB_UVERBS_DECLARE_EX_CMD(query_device); -IB_UVERBS_DECLARE_EX_CMD(create_cq); -IB_UVERBS_DECLARE_EX_CMD(create_qp); -IB_UVERBS_DECLARE_EX_CMD(create_wq); -IB_UVERBS_DECLARE_EX_CMD(modify_wq); -IB_UVERBS_DECLARE_EX_CMD(destroy_wq); -IB_UVERBS_DECLARE_EX_CMD(create_rwq_ind_table); -IB_UVERBS_DECLARE_EX_CMD(destroy_rwq_ind_table); -IB_UVERBS_DECLARE_EX_CMD(modify_qp); -IB_UVERBS_DECLARE_EX_CMD(modify_cq); - #endif /* UVERBS_H */ diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index a93853770e3c..b70749542471 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -47,11 +47,132 @@ #include "uverbs.h" #include "core_priv.h" +/* + * Copy a response to userspace. If the provided 'resp' is larger than the + * user buffer it is silently truncated. If the user provided a larger buffer + * then the trailing portion is zero filled. + * + * These semantics are intended to support future extension of the output + * structures. + */ +static int uverbs_response(struct uverbs_attr_bundle *attrs, const void *resp, + size_t resp_len) +{ + u8 __user *cur = attrs->ucore.outbuf + resp_len; + u8 __user *end = attrs->ucore.outbuf + attrs->ucore.outlen; + int ret; + + if (copy_to_user(attrs->ucore.outbuf, resp, + min(attrs->ucore.outlen, resp_len))) + return -EFAULT; + + /* Zero fill any extra memory that user space might have provided */ + for (; cur < end; cur++) { + ret = put_user(0, cur); + if (ret) + return ret; + } + + return 0; +} + +/* + * Copy a request from userspace. If the provided 'req' is larger than the + * user buffer then the user buffer is zero extended into the 'req'. If 'req' + * is smaller than the user buffer then the uncopied bytes in the user buffer + * must be zero. + */ +static int uverbs_request(struct uverbs_attr_bundle *attrs, void *req, + size_t req_len) +{ + if (copy_from_user(req, attrs->ucore.inbuf, + min(attrs->ucore.inlen, req_len))) + return -EFAULT; + + if (attrs->ucore.inlen < req_len) { + memset(req + attrs->ucore.inlen, 0, + req_len - attrs->ucore.inlen); + } else if (attrs->ucore.inlen > req_len) { + if (!ib_is_buffer_cleared(attrs->ucore.inbuf + req_len, + attrs->ucore.inlen - req_len)) + return -EOPNOTSUPP; + } + return 0; +} + +/* + * Generate the value for the 'response_length' protocol used by write_ex. + * This is the number of bytes the kernel actually wrote. Userspace can use + * this to detect what structure members in the response the kernel + * understood. + */ +static u32 uverbs_response_length(struct uverbs_attr_bundle *attrs, + size_t resp_len) +{ + return min_t(size_t, attrs->ucore.outlen, resp_len); +} + +/* + * The iterator version of the request interface is for handlers that need to + * step over a flex array at the end of a command header. + */ +struct uverbs_req_iter { + const void __user *cur; + const void __user *end; +}; + +static int uverbs_request_start(struct uverbs_attr_bundle *attrs, + struct uverbs_req_iter *iter, + void *req, + size_t req_len) +{ + if (attrs->ucore.inlen < req_len) + return -ENOSPC; + + if (copy_from_user(req, attrs->ucore.inbuf, req_len)) + return -EFAULT; + + iter->cur = attrs->ucore.inbuf + req_len; + iter->end = attrs->ucore.inbuf + attrs->ucore.inlen; + return 0; +} + +static int uverbs_request_next(struct uverbs_req_iter *iter, void *val, + size_t len) +{ + if (iter->cur + len > iter->end) + return -ENOSPC; + + if (copy_from_user(val, iter->cur, len)) + return -EFAULT; + + iter->cur += len; + return 0; +} + +static const void __user *uverbs_request_next_ptr(struct uverbs_req_iter *iter, + size_t len) +{ + const void __user *res = iter->cur; + + if (iter->cur + len > iter->end) + return ERR_PTR(-ENOSPC); + iter->cur += len; + return res; +} + +static int uverbs_request_finish(struct uverbs_req_iter *iter) +{ + if (!ib_is_buffer_cleared(iter->cur, iter->end - iter->cur)) + return -EOPNOTSUPP; + return 0; +} + static struct ib_uverbs_completion_event_file * -_ib_uverbs_lookup_comp_file(s32 fd, struct ib_uverbs_file *ufile) +_ib_uverbs_lookup_comp_file(s32 fd, const struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = ufd_get_read(UVERBS_OBJECT_COMP_CHANNEL, - fd, ufile); + fd, attrs); if (IS_ERR(uobj)) return (void *)uobj; @@ -65,24 +186,20 @@ _ib_uverbs_lookup_comp_file(s32 fd, struct ib_uverbs_file *ufile) #define ib_uverbs_lookup_comp_file(_fd, _ufile) \ _ib_uverbs_lookup_comp_file((_fd)*typecheck(s32, _fd), _ufile) -ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_get_context(struct uverbs_attr_bundle *attrs) { + struct ib_uverbs_file *file = attrs->ufile; struct ib_uverbs_get_context cmd; struct ib_uverbs_get_context_resp resp; - struct ib_udata udata; struct ib_ucontext *ucontext; struct file *filp; struct ib_rdmacg_object cg_obj; struct ib_device *ib_dev; int ret; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; mutex_lock(&file->ucontext_lock); ib_dev = srcu_dereference(file->device->ib_dev, @@ -97,16 +214,11 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, goto err; } - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); - ret = ib_rdmacg_try_charge(&cg_obj, ib_dev, RDMACG_RESOURCE_HCA_HANDLE); if (ret) goto err; - ucontext = ib_dev->alloc_ucontext(ib_dev, &udata); + ucontext = ib_dev->alloc_ucontext(ib_dev, &attrs->driver_udata); if (IS_ERR(ucontext)) { ret = PTR_ERR(ucontext); goto err_alloc; @@ -141,13 +253,15 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, goto err_fd; } - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_file; - } fd_install(resp.async_fd, filp); + ucontext->res.type = RDMA_RESTRACK_CTX; + rdma_restrack_add(&ucontext->res); + /* * Make sure that ib_uverbs_get_ucontext() sees the pointer update * only after all writes to setup the ucontext have completed @@ -156,7 +270,7 @@ ssize_t ib_uverbs_get_context(struct ib_uverbs_file *file, mutex_unlock(&file->ucontext_lock); - return in_len; + return 0; err_file: ib_uverbs_free_async_event_file(file); @@ -224,31 +338,25 @@ static void copy_query_dev_fields(struct ib_ucontext *ucontext, resp->phys_port_cnt = ib_dev->phys_port_cnt; } -ssize_t ib_uverbs_query_device(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_query_device(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_query_device cmd; struct ib_uverbs_query_device_resp resp; struct ib_ucontext *ucontext; + int ret; - ucontext = ib_uverbs_get_ucontext(file); + ucontext = ib_uverbs_get_ucontext(attrs); if (IS_ERR(ucontext)) return PTR_ERR(ucontext); - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; memset(&resp, 0, sizeof resp); copy_query_dev_fields(ucontext, &resp, &ucontext->device->attrs); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } /* @@ -272,9 +380,7 @@ static u32 make_port_cap_flags(const struct ib_port_attr *attr) return res; } -ssize_t ib_uverbs_query_port(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_query_port(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_query_port cmd; struct ib_uverbs_query_port_resp resp; @@ -283,16 +389,14 @@ ssize_t ib_uverbs_query_port(struct ib_uverbs_file *file, struct ib_ucontext *ucontext; struct ib_device *ib_dev; - ucontext = ib_uverbs_get_ucontext(file); + ucontext = ib_uverbs_get_ucontext(attrs); if (IS_ERR(ucontext)) return PTR_ERR(ucontext); ib_dev = ucontext->device; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; ret = ib_query_port(ib_dev, cmd.port_num, &attr); if (ret) @@ -331,40 +435,27 @@ ssize_t ib_uverbs_query_port(struct ib_uverbs_file *file, resp.link_layer = rdma_port_get_link_layer(ib_dev, cmd.port_num); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } -ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_alloc_pd(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_alloc_pd cmd; struct ib_uverbs_alloc_pd_resp resp; - struct ib_udata udata; struct ib_uobject *uobj; struct ib_pd *pd; int ret; struct ib_device *ib_dev; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_alloc(UVERBS_OBJECT_PD, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_PD, attrs, &ib_dev); if (IS_ERR(uobj)) return PTR_ERR(uobj); - pd = ib_dev->alloc_pd(ib_dev, uobj->context, &udata); + pd = ib_dev->alloc_pd(ib_dev, uobj->context, &attrs->driver_udata); if (IS_ERR(pd)) { ret = PTR_ERR(pd); goto err; @@ -381,12 +472,11 @@ ssize_t ib_uverbs_alloc_pd(struct ib_uverbs_file *file, pd->res.type = RDMA_RESTRACK_PD; rdma_restrack_add(&pd->res); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } - return uobj_alloc_commit(uobj, in_len); + return uobj_alloc_commit(uobj); err_copy: ib_dealloc_pd(pd); @@ -396,17 +486,16 @@ err: return ret; } -ssize_t ib_uverbs_dealloc_pd(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_dealloc_pd(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_dealloc_pd cmd; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - return uobj_perform_destroy(UVERBS_OBJECT_PD, cmd.pd_handle, file, - in_len); + return uobj_perform_destroy(UVERBS_OBJECT_PD, cmd.pd_handle, attrs); } struct xrcd_table_entry { @@ -494,13 +583,11 @@ static void xrcd_table_delete(struct ib_uverbs_device *dev, } } -ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_open_xrcd(struct uverbs_attr_bundle *attrs) { + struct ib_uverbs_device *ibudev = attrs->ufile->device; struct ib_uverbs_open_xrcd cmd; struct ib_uverbs_open_xrcd_resp resp; - struct ib_udata udata; struct ib_uxrcd_object *obj; struct ib_xrcd *xrcd = NULL; struct fd f = {NULL, 0}; @@ -509,18 +596,11 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, int new_xrcd = 0; struct ib_device *ib_dev; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - mutex_lock(&file->device->xrcd_tree_mutex); + mutex_lock(&ibudev->xrcd_tree_mutex); if (cmd.fd != -1) { /* search for file descriptor */ @@ -531,7 +611,7 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, } inode = file_inode(f.file); - xrcd = find_xrcd(file->device, inode); + xrcd = find_xrcd(ibudev, inode); if (!xrcd && !(cmd.oflags & O_CREAT)) { /* no file descriptor. Need CREATE flag */ ret = -EAGAIN; @@ -544,7 +624,7 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, } } - obj = (struct ib_uxrcd_object *)uobj_alloc(UVERBS_OBJECT_XRCD, file, + obj = (struct ib_uxrcd_object *)uobj_alloc(UVERBS_OBJECT_XRCD, attrs, &ib_dev); if (IS_ERR(obj)) { ret = PTR_ERR(obj); @@ -552,7 +632,8 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, } if (!xrcd) { - xrcd = ib_dev->alloc_xrcd(ib_dev, obj->uobject.context, &udata); + xrcd = ib_dev->alloc_xrcd(ib_dev, obj->uobject.context, + &attrs->driver_udata); if (IS_ERR(xrcd)) { ret = PTR_ERR(xrcd); goto err; @@ -574,29 +655,28 @@ ssize_t ib_uverbs_open_xrcd(struct ib_uverbs_file *file, if (inode) { if (new_xrcd) { /* create new inode/xrcd table entry */ - ret = xrcd_table_insert(file->device, inode, xrcd); + ret = xrcd_table_insert(ibudev, inode, xrcd); if (ret) goto err_dealloc_xrcd; } atomic_inc(&xrcd->usecnt); } - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } if (f.file) fdput(f); - mutex_unlock(&file->device->xrcd_tree_mutex); + mutex_unlock(&ibudev->xrcd_tree_mutex); - return uobj_alloc_commit(&obj->uobject, in_len); + return uobj_alloc_commit(&obj->uobject); err_copy: if (inode) { if (new_xrcd) - xrcd_table_delete(file->device, inode); + xrcd_table_delete(ibudev, inode); atomic_dec(&xrcd->usecnt); } @@ -610,22 +690,21 @@ err_tree_mutex_unlock: if (f.file) fdput(f); - mutex_unlock(&file->device->xrcd_tree_mutex); + mutex_unlock(&ibudev->xrcd_tree_mutex); return ret; } -ssize_t ib_uverbs_close_xrcd(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_close_xrcd(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_close_xrcd cmd; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - return uobj_perform_destroy(UVERBS_OBJECT_XRCD, cmd.xrcd_handle, file, - in_len); + return uobj_perform_destroy(UVERBS_OBJECT_XRCD, cmd.xrcd_handle, attrs); } int ib_uverbs_dealloc_xrcd(struct ib_uobject *uobject, @@ -653,29 +732,19 @@ int ib_uverbs_dealloc_xrcd(struct ib_uobject *uobject, return ret; } -ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_reg_mr(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_reg_mr cmd; struct ib_uverbs_reg_mr_resp resp; - struct ib_udata udata; struct ib_uobject *uobj; struct ib_pd *pd; struct ib_mr *mr; int ret; struct ib_device *ib_dev; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; if ((cmd.start & ~PAGE_MASK) != (cmd.hca_va & ~PAGE_MASK)) return -EINVAL; @@ -684,11 +753,11 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file, if (ret) return ret; - uobj = uobj_alloc(UVERBS_OBJECT_MR, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_MR, attrs, &ib_dev); if (IS_ERR(uobj)) return PTR_ERR(uobj); - pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, file); + pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, attrs); if (!pd) { ret = -EINVAL; goto err_free; @@ -704,7 +773,7 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file, } mr = pd->device->reg_user_mr(pd, cmd.start, cmd.length, cmd.hca_va, - cmd.access_flags, &udata); + cmd.access_flags, &attrs->driver_udata); if (IS_ERR(mr)) { ret = PTR_ERR(mr); goto err_put; @@ -725,14 +794,13 @@ ssize_t ib_uverbs_reg_mr(struct ib_uverbs_file *file, resp.rkey = mr->rkey; resp.mr_handle = uobj->id; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } uobj_put_obj_read(pd); - return uobj_alloc_commit(uobj, in_len); + return uobj_alloc_commit(uobj); err_copy: ib_dereg_mr(mr); @@ -745,29 +813,19 @@ err_free: return ret; } -ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_rereg_mr(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_rereg_mr cmd; struct ib_uverbs_rereg_mr_resp resp; - struct ib_udata udata; struct ib_pd *pd = NULL; struct ib_mr *mr; struct ib_pd *old_pd; int ret; struct ib_uobject *uobj; - if (out_len < sizeof(resp)) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof(cmd))) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; if (cmd.flags & ~IB_MR_REREG_SUPPORTED || !cmd.flags) return -EINVAL; @@ -777,7 +835,7 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, (cmd.start & ~PAGE_MASK) != (cmd.hca_va & ~PAGE_MASK))) return -EINVAL; - uobj = uobj_get_write(UVERBS_OBJECT_MR, cmd.mr_handle, file); + uobj = uobj_get_write(UVERBS_OBJECT_MR, cmd.mr_handle, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -796,7 +854,7 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, if (cmd.flags & IB_MR_REREG_PD) { pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, - file); + attrs); if (!pd) { ret = -EINVAL; goto put_uobjs; @@ -804,9 +862,9 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, } old_pd = mr->pd; - ret = mr->device->rereg_user_mr(mr, cmd.flags, cmd.start, - cmd.length, cmd.hca_va, - cmd.access_flags, pd, &udata); + ret = mr->device->rereg_user_mr(mr, cmd.flags, cmd.start, cmd.length, + cmd.hca_va, cmd.access_flags, pd, + &attrs->driver_udata); if (!ret) { if (cmd.flags & IB_MR_REREG_PD) { atomic_inc(&pd->usecnt); @@ -821,10 +879,7 @@ ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, resp.lkey = mr->lkey; resp.rkey = mr->rkey; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof(resp))) - ret = -EFAULT; - else - ret = in_len; + ret = uverbs_response(attrs, &resp, sizeof(resp)); put_uobj_pd: if (cmd.flags & IB_MR_REREG_PD) @@ -836,54 +891,43 @@ put_uobjs: return ret; } -ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_dereg_mr(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_dereg_mr cmd; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - return uobj_perform_destroy(UVERBS_OBJECT_MR, cmd.mr_handle, file, - in_len); + return uobj_perform_destroy(UVERBS_OBJECT_MR, cmd.mr_handle, attrs); } -ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_alloc_mw(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_alloc_mw cmd; struct ib_uverbs_alloc_mw_resp resp; struct ib_uobject *uobj; struct ib_pd *pd; struct ib_mw *mw; - struct ib_udata udata; int ret; struct ib_device *ib_dev; - if (out_len < sizeof(resp)) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof(cmd))) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_alloc(UVERBS_OBJECT_MW, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_MW, attrs, &ib_dev); if (IS_ERR(uobj)) return PTR_ERR(uobj); - pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, file); + pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, attrs); if (!pd) { ret = -EINVAL; goto err_free; } - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); - - mw = pd->device->alloc_mw(pd, cmd.mw_type, &udata); + mw = pd->device->alloc_mw(pd, cmd.mw_type, &attrs->driver_udata); if (IS_ERR(mw)) { ret = PTR_ERR(mw); goto err_put; @@ -900,13 +944,12 @@ ssize_t ib_uverbs_alloc_mw(struct ib_uverbs_file *file, resp.rkey = mw->rkey; resp.mw_handle = uobj->id; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof(resp))) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } uobj_put_obj_read(pd); - return uobj_alloc_commit(uobj, in_len); + return uobj_alloc_commit(uobj); err_copy: uverbs_dealloc_mw(mw); @@ -917,36 +960,32 @@ err_free: return ret; } -ssize_t ib_uverbs_dealloc_mw(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_dealloc_mw(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_dealloc_mw cmd; + int ret; - if (copy_from_user(&cmd, buf, sizeof(cmd))) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - return uobj_perform_destroy(UVERBS_OBJECT_MW, cmd.mw_handle, file, - in_len); + return uobj_perform_destroy(UVERBS_OBJECT_MW, cmd.mw_handle, attrs); } -ssize_t ib_uverbs_create_comp_channel(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_create_comp_channel(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_comp_channel cmd; struct ib_uverbs_create_comp_channel_resp resp; struct ib_uobject *uobj; struct ib_uverbs_completion_event_file *ev_file; struct ib_device *ib_dev; + int ret; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_alloc(UVERBS_OBJECT_COMP_CHANNEL, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_COMP_CHANNEL, attrs, &ib_dev); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -956,25 +995,17 @@ ssize_t ib_uverbs_create_comp_channel(struct ib_uverbs_file *file, uobj); ib_uverbs_init_event_queue(&ev_file->ev_queue); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) { uobj_alloc_abort(uobj); - return -EFAULT; + return ret; } - return uobj_alloc_commit(uobj, in_len); + return uobj_alloc_commit(uobj); } -static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw, - struct ib_uverbs_ex_create_cq *cmd, - size_t cmd_sz, - int (*cb)(struct ib_uverbs_file *file, - struct ib_ucq_object *obj, - struct ib_uverbs_ex_create_cq_resp *resp, - struct ib_udata *udata, - void *context), - void *context) +static struct ib_ucq_object *create_cq(struct uverbs_attr_bundle *attrs, + struct ib_uverbs_ex_create_cq *cmd) { struct ib_ucq_object *obj; struct ib_uverbs_completion_event_file *ev_file = NULL; @@ -984,21 +1015,16 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file, struct ib_cq_init_attr attr = {}; struct ib_device *ib_dev; - if (cmd->comp_vector >= file->device->num_comp_vectors) + if (cmd->comp_vector >= attrs->ufile->device->num_comp_vectors) return ERR_PTR(-EINVAL); - obj = (struct ib_ucq_object *)uobj_alloc(UVERBS_OBJECT_CQ, file, + obj = (struct ib_ucq_object *)uobj_alloc(UVERBS_OBJECT_CQ, attrs, &ib_dev); if (IS_ERR(obj)) return obj; - if (!ib_dev->create_cq) { - ret = -EOPNOTSUPP; - goto err; - } - if (cmd->comp_channel >= 0) { - ev_file = ib_uverbs_lookup_comp_file(cmd->comp_channel, file); + ev_file = ib_uverbs_lookup_comp_file(cmd->comp_channel, attrs); if (IS_ERR(ev_file)) { ret = PTR_ERR(ev_file); goto err; @@ -1013,11 +1039,10 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file, attr.cqe = cmd->cqe; attr.comp_vector = cmd->comp_vector; + attr.flags = cmd->flags; - if (cmd_sz > offsetof(typeof(*cmd), flags) + sizeof(cmd->flags)) - attr.flags = cmd->flags; - - cq = ib_dev->create_cq(ib_dev, &attr, obj->uobject.context, uhw); + cq = ib_dev->create_cq(ib_dev, &attr, obj->uobject.context, + &attrs->driver_udata); if (IS_ERR(cq)) { ret = PTR_ERR(cq); goto err_file; @@ -1034,18 +1059,16 @@ static struct ib_ucq_object *create_cq(struct ib_uverbs_file *file, memset(&resp, 0, sizeof resp); resp.base.cq_handle = obj->uobject.id; resp.base.cqe = cq->cqe; - - resp.response_length = offsetof(typeof(resp), response_length) + - sizeof(resp.response_length); + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); cq->res.type = RDMA_RESTRACK_CQ; rdma_restrack_add(&cq->res); - ret = cb(file, obj, &resp, ucore, context); + ret = uverbs_response(attrs, &resp, sizeof(resp)); if (ret) goto err_cb; - ret = uobj_alloc_commit(&obj->uobject, 0); + ret = uobj_alloc_commit(&obj->uobject); if (ret) return ERR_PTR(ret); return obj; @@ -1055,7 +1078,7 @@ err_cb: err_file: if (ev_file) - ib_uverbs_release_ucq(file, ev_file, obj); + ib_uverbs_release_ucq(attrs->ufile, ev_file, obj); err: uobj_alloc_abort(&obj->uobject); @@ -1063,41 +1086,16 @@ err: return ERR_PTR(ret); } -static int ib_uverbs_create_cq_cb(struct ib_uverbs_file *file, - struct ib_ucq_object *obj, - struct ib_uverbs_ex_create_cq_resp *resp, - struct ib_udata *ucore, void *context) -{ - if (ib_copy_to_udata(ucore, &resp->base, sizeof(resp->base))) - return -EFAULT; - - return 0; -} - -ssize_t ib_uverbs_create_cq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_create_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_cq cmd; struct ib_uverbs_ex_create_cq cmd_ex; - struct ib_uverbs_create_cq_resp resp; - struct ib_udata ucore; - struct ib_udata uhw; struct ib_ucq_object *obj; + int ret; - if (out_len < sizeof(resp)) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof(cmd))) - return -EFAULT; - - ib_uverbs_init_udata(&ucore, buf, u64_to_user_ptr(cmd.response), - sizeof(cmd), sizeof(resp)); - - ib_uverbs_init_udata(&uhw, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; memset(&cmd_ex, 0, sizeof(cmd_ex)); cmd_ex.user_handle = cmd.user_handle; @@ -1105,43 +1103,19 @@ ssize_t ib_uverbs_create_cq(struct ib_uverbs_file *file, cmd_ex.comp_vector = cmd.comp_vector; cmd_ex.comp_channel = cmd.comp_channel; - obj = create_cq(file, &ucore, &uhw, &cmd_ex, - offsetof(typeof(cmd_ex), comp_channel) + - sizeof(cmd.comp_channel), ib_uverbs_create_cq_cb, - NULL); - - if (IS_ERR(obj)) - return PTR_ERR(obj); - - return in_len; -} - -static int ib_uverbs_ex_create_cq_cb(struct ib_uverbs_file *file, - struct ib_ucq_object *obj, - struct ib_uverbs_ex_create_cq_resp *resp, - struct ib_udata *ucore, void *context) -{ - if (ib_copy_to_udata(ucore, resp, resp->response_length)) - return -EFAULT; - - return 0; + obj = create_cq(attrs, &cmd_ex); + return PTR_ERR_OR_ZERO(obj); } -int ib_uverbs_ex_create_cq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_create_cq(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_create_cq_resp resp; struct ib_uverbs_ex_create_cq cmd; struct ib_ucq_object *obj; - int err; - - if (ucore->inlen < sizeof(cmd)) - return -EINVAL; + int ret; - err = ib_copy_from_udata(&cmd, ucore, sizeof(cmd)); - if (err) - return err; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; if (cmd.comp_mask) return -EINVAL; @@ -1149,52 +1123,36 @@ int ib_uverbs_ex_create_cq(struct ib_uverbs_file *file, if (cmd.reserved) return -EINVAL; - if (ucore->outlen < (offsetof(typeof(resp), response_length) + - sizeof(resp.response_length))) - return -ENOSPC; - - obj = create_cq(file, ucore, uhw, &cmd, - min(ucore->inlen, sizeof(cmd)), - ib_uverbs_ex_create_cq_cb, NULL); - + obj = create_cq(attrs, &cmd); return PTR_ERR_OR_ZERO(obj); } -ssize_t ib_uverbs_resize_cq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_resize_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_resize_cq cmd; struct ib_uverbs_resize_cq_resp resp = {}; - struct ib_udata udata; struct ib_cq *cq; int ret = -EINVAL; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, file); + cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (!cq) return -EINVAL; - ret = cq->device->resize_cq(cq, cmd.cqe, &udata); + ret = cq->device->resize_cq(cq, cmd.cqe, &attrs->driver_udata); if (ret) goto out; resp.cqe = cq->cqe; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp.cqe)) - ret = -EFAULT; - + ret = uverbs_response(attrs, &resp, sizeof(resp)); out: uobj_put_obj_read(cq); - return ret ? ret : in_len; + return ret; } static int copy_wc_to_user(struct ib_device *ib_dev, void __user *dest, @@ -1227,9 +1185,7 @@ static int copy_wc_to_user(struct ib_device *ib_dev, void __user *dest, return 0; } -ssize_t ib_uverbs_poll_cq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_poll_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_poll_cq cmd; struct ib_uverbs_poll_cq_resp resp; @@ -1239,15 +1195,16 @@ ssize_t ib_uverbs_poll_cq(struct ib_uverbs_file *file, struct ib_wc wc; int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, file); + cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (!cq) return -EINVAL; /* we copy a struct ib_uverbs_poll_cq_resp to user space */ - header_ptr = u64_to_user_ptr(cmd.response); + header_ptr = attrs->ucore.outbuf; data_ptr = header_ptr + sizeof resp; memset(&resp, 0, sizeof resp); @@ -1271,24 +1228,24 @@ ssize_t ib_uverbs_poll_cq(struct ib_uverbs_file *file, goto out_put; } - ret = in_len; + ret = 0; out_put: uobj_put_obj_read(cq); return ret; } -ssize_t ib_uverbs_req_notify_cq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_req_notify_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_req_notify_cq cmd; struct ib_cq *cq; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, file); + cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (!cq) return -EINVAL; @@ -1297,22 +1254,22 @@ ssize_t ib_uverbs_req_notify_cq(struct ib_uverbs_file *file, uobj_put_obj_read(cq); - return in_len; + return 0; } -ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_destroy_cq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_destroy_cq cmd; struct ib_uverbs_destroy_cq_resp resp; struct ib_uobject *uobj; struct ib_ucq_object *obj; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_get_destroy(UVERBS_OBJECT_CQ, cmd.cq_handle, file); + uobj = uobj_get_destroy(UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -1323,21 +1280,11 @@ ssize_t ib_uverbs_destroy_cq(struct ib_uverbs_file *file, uobj_put_destroy(uobj); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } -static int create_qp(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw, - struct ib_uverbs_ex_create_qp *cmd, - size_t cmd_sz, - int (*cb)(struct ib_uverbs_file *file, - struct ib_uverbs_ex_create_qp_resp *resp, - struct ib_udata *udata), - void *context) +static int create_qp(struct uverbs_attr_bundle *attrs, + struct ib_uverbs_ex_create_qp *cmd) { struct ib_uqp_object *obj; struct ib_device *device; @@ -1347,7 +1294,6 @@ static int create_qp(struct ib_uverbs_file *file, struct ib_cq *scq = NULL, *rcq = NULL; struct ib_srq *srq = NULL; struct ib_qp *qp; - char *buf; struct ib_qp_init_attr attr = {}; struct ib_uverbs_ex_create_qp_resp resp; int ret; @@ -1358,7 +1304,7 @@ static int create_qp(struct ib_uverbs_file *file, if (cmd->qp_type == IB_QPT_RAW_PACKET && !capable(CAP_NET_RAW)) return -EPERM; - obj = (struct ib_uqp_object *)uobj_alloc(UVERBS_OBJECT_QP, file, + obj = (struct ib_uqp_object *)uobj_alloc(UVERBS_OBJECT_QP, attrs, &ib_dev); if (IS_ERR(obj)) return PTR_ERR(obj); @@ -1366,12 +1312,10 @@ static int create_qp(struct ib_uverbs_file *file, obj->uevent.uobject.user_handle = cmd->user_handle; mutex_init(&obj->mcast_lock); - if (cmd_sz >= offsetof(typeof(*cmd), rwq_ind_tbl_handle) + - sizeof(cmd->rwq_ind_tbl_handle) && - (cmd->comp_mask & IB_UVERBS_CREATE_QP_MASK_IND_TABLE)) { + if (cmd->comp_mask & IB_UVERBS_CREATE_QP_MASK_IND_TABLE) { ind_tbl = uobj_get_obj_read(rwq_ind_table, UVERBS_OBJECT_RWQ_IND_TBL, - cmd->rwq_ind_tbl_handle, file); + cmd->rwq_ind_tbl_handle, attrs); if (!ind_tbl) { ret = -EINVAL; goto err_put; @@ -1380,13 +1324,6 @@ static int create_qp(struct ib_uverbs_file *file, attr.rwq_ind_tbl = ind_tbl; } - if (cmd_sz > sizeof(*cmd) && - !ib_is_udata_cleared(ucore, sizeof(*cmd), - cmd_sz - sizeof(*cmd))) { - ret = -EOPNOTSUPP; - goto err_put; - } - if (ind_tbl && (cmd->max_recv_wr || cmd->max_recv_sge || cmd->is_srq)) { ret = -EINVAL; goto err_put; @@ -1397,7 +1334,7 @@ static int create_qp(struct ib_uverbs_file *file, if (cmd->qp_type == IB_QPT_XRC_TGT) { xrcd_uobj = uobj_get_read(UVERBS_OBJECT_XRCD, cmd->pd_handle, - file); + attrs); if (IS_ERR(xrcd_uobj)) { ret = -EINVAL; @@ -1417,7 +1354,7 @@ static int create_qp(struct ib_uverbs_file *file, } else { if (cmd->is_srq) { srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, - cmd->srq_handle, file); + cmd->srq_handle, attrs); if (!srq || srq->srq_type == IB_SRQT_XRC) { ret = -EINVAL; goto err_put; @@ -1428,7 +1365,7 @@ static int create_qp(struct ib_uverbs_file *file, if (cmd->recv_cq_handle != cmd->send_cq_handle) { rcq = uobj_get_obj_read( cq, UVERBS_OBJECT_CQ, - cmd->recv_cq_handle, file); + cmd->recv_cq_handle, attrs); if (!rcq) { ret = -EINVAL; goto err_put; @@ -1439,11 +1376,11 @@ static int create_qp(struct ib_uverbs_file *file, if (has_sq) scq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, - cmd->send_cq_handle, file); + cmd->send_cq_handle, attrs); if (!ind_tbl) rcq = rcq ?: scq; pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd->pd_handle, - file); + attrs); if (!pd || (!scq && has_sq)) { ret = -EINVAL; goto err_put; @@ -1453,7 +1390,7 @@ static int create_qp(struct ib_uverbs_file *file, } attr.event_handler = ib_uverbs_qp_event_handler; - attr.qp_context = file; + attr.qp_context = attrs->ufile; attr.send_cq = scq; attr.recv_cq = rcq; attr.srq = srq; @@ -1473,10 +1410,7 @@ static int create_qp(struct ib_uverbs_file *file, INIT_LIST_HEAD(&obj->uevent.event_list); INIT_LIST_HEAD(&obj->mcast_list); - if (cmd_sz >= offsetof(typeof(*cmd), create_flags) + - sizeof(cmd->create_flags)) - attr.create_flags = cmd->create_flags; - + attr.create_flags = cmd->create_flags; if (attr.create_flags & ~(IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK | IB_QP_CREATE_CROSS_CHANNEL | IB_QP_CREATE_MANAGED_SEND | @@ -1498,18 +1432,10 @@ static int create_qp(struct ib_uverbs_file *file, attr.source_qpn = cmd->source_qpn; } - buf = (void *)cmd + sizeof(*cmd); - if (cmd_sz > sizeof(*cmd)) - if (!(buf[0] == 0 && !memcmp(buf, buf + 1, - cmd_sz - sizeof(*cmd) - 1))) { - ret = -EINVAL; - goto err_put; - } - if (cmd->qp_type == IB_QPT_XRC_TGT) qp = ib_create_qp(pd, &attr); else - qp = _ib_create_qp(device, pd, &attr, uhw, + qp = _ib_create_qp(device, pd, &attr, &attrs->driver_udata, &obj->uevent.uobject); if (IS_ERR(qp)) { @@ -1557,11 +1483,9 @@ static int create_qp(struct ib_uverbs_file *file, resp.base.max_recv_wr = attr.cap.max_recv_wr; resp.base.max_send_wr = attr.cap.max_send_wr; resp.base.max_inline_data = attr.cap.max_inline_data; + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); - resp.response_length = offsetof(typeof(resp), response_length) + - sizeof(resp.response_length); - - ret = cb(file, &resp, ucore); + ret = uverbs_response(attrs, &resp, sizeof(resp)); if (ret) goto err_cb; @@ -1583,7 +1507,7 @@ static int create_qp(struct ib_uverbs_file *file, if (ind_tbl) uobj_put_obj_read(ind_tbl); - return uobj_alloc_commit(&obj->uevent.uobject, 0); + return uobj_alloc_commit(&obj->uevent.uobject); err_cb: ib_destroy_qp(qp); @@ -1605,39 +1529,15 @@ err_put: return ret; } -static int ib_uverbs_create_qp_cb(struct ib_uverbs_file *file, - struct ib_uverbs_ex_create_qp_resp *resp, - struct ib_udata *ucore) -{ - if (ib_copy_to_udata(ucore, &resp->base, sizeof(resp->base))) - return -EFAULT; - - return 0; -} - -ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_create_qp(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_qp cmd; struct ib_uverbs_ex_create_qp cmd_ex; - struct ib_udata ucore; - struct ib_udata uhw; - ssize_t resp_size = sizeof(struct ib_uverbs_create_qp_resp); - int err; - - if (out_len < resp_size) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof(cmd))) - return -EFAULT; + int ret; - ib_uverbs_init_udata(&ucore, buf, u64_to_user_ptr(cmd.response), - sizeof(cmd), resp_size); - ib_uverbs_init_udata(&uhw, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + resp_size, - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - resp_size); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; memset(&cmd_ex, 0, sizeof(cmd_ex)); cmd_ex.user_handle = cmd.user_handle; @@ -1654,42 +1554,17 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file, cmd_ex.qp_type = cmd.qp_type; cmd_ex.is_srq = cmd.is_srq; - err = create_qp(file, &ucore, &uhw, &cmd_ex, - offsetof(typeof(cmd_ex), is_srq) + - sizeof(cmd.is_srq), ib_uverbs_create_qp_cb, - NULL); - - if (err) - return err; - - return in_len; + return create_qp(attrs, &cmd_ex); } -static int ib_uverbs_ex_create_qp_cb(struct ib_uverbs_file *file, - struct ib_uverbs_ex_create_qp_resp *resp, - struct ib_udata *ucore) +static int ib_uverbs_ex_create_qp(struct uverbs_attr_bundle *attrs) { - if (ib_copy_to_udata(ucore, resp, resp->response_length)) - return -EFAULT; - - return 0; -} - -int ib_uverbs_ex_create_qp(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) -{ - struct ib_uverbs_ex_create_qp_resp resp; - struct ib_uverbs_ex_create_qp cmd = {0}; - int err; - - if (ucore->inlen < (offsetof(typeof(cmd), comp_mask) + - sizeof(cmd.comp_mask))) - return -EINVAL; + struct ib_uverbs_ex_create_qp cmd; + int ret; - err = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); - if (err) - return err; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; if (cmd.comp_mask & ~IB_UVERBS_CREATE_QP_SUP_COMP_MASK) return -EINVAL; @@ -1697,26 +1572,13 @@ int ib_uverbs_ex_create_qp(struct ib_uverbs_file *file, if (cmd.reserved) return -EINVAL; - if (ucore->outlen < (offsetof(typeof(resp), response_length) + - sizeof(resp.response_length))) - return -ENOSPC; - - err = create_qp(file, ucore, uhw, &cmd, - min(ucore->inlen, sizeof(cmd)), - ib_uverbs_ex_create_qp_cb, NULL); - - if (err) - return err; - - return 0; + return create_qp(attrs, &cmd); } -ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file, - const char __user *buf, int in_len, int out_len) +static int ib_uverbs_open_qp(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_open_qp cmd; struct ib_uverbs_create_qp_resp resp; - struct ib_udata udata; struct ib_uqp_object *obj; struct ib_xrcd *xrcd; struct ib_uobject *uninitialized_var(xrcd_uobj); @@ -1725,23 +1587,16 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file, int ret; struct ib_device *ib_dev; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - obj = (struct ib_uqp_object *)uobj_alloc(UVERBS_OBJECT_QP, file, + obj = (struct ib_uqp_object *)uobj_alloc(UVERBS_OBJECT_QP, attrs, &ib_dev); if (IS_ERR(obj)) return PTR_ERR(obj); - xrcd_uobj = uobj_get_read(UVERBS_OBJECT_XRCD, cmd.pd_handle, file); + xrcd_uobj = uobj_get_read(UVERBS_OBJECT_XRCD, cmd.pd_handle, attrs); if (IS_ERR(xrcd_uobj)) { ret = -EINVAL; goto err_put; @@ -1754,7 +1609,7 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file, } attr.event_handler = ib_uverbs_qp_event_handler; - attr.qp_context = file; + attr.qp_context = attrs->ufile; attr.qp_num = cmd.qpn; attr.qp_type = cmd.qp_type; @@ -1775,17 +1630,16 @@ ssize_t ib_uverbs_open_qp(struct ib_uverbs_file *file, resp.qpn = qp->qp_num; resp.qp_handle = obj->uevent.uobject.id; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_destroy; - } obj->uxrcd = container_of(xrcd_uobj, struct ib_uxrcd_object, uobject); atomic_inc(&obj->uxrcd->refcnt); qp->uobject = &obj->uevent.uobject; uobj_put_read(xrcd_uobj); - return uobj_alloc_commit(&obj->uevent.uobject, in_len); + return uobj_alloc_commit(&obj->uevent.uobject); err_destroy: ib_destroy_qp(qp); @@ -1818,9 +1672,7 @@ static void copy_ah_attr_to_uverbs(struct ib_uverbs_qp_dest *uverb_attr, uverb_attr->port_num = rdma_ah_get_port_num(rdma_attr); } -ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_query_qp(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_query_qp cmd; struct ib_uverbs_query_qp_resp resp; @@ -1829,8 +1681,9 @@ ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file, struct ib_qp_init_attr *init_attr; int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; attr = kmalloc(sizeof *attr, GFP_KERNEL); init_attr = kmalloc(sizeof *init_attr, GFP_KERNEL); @@ -1839,7 +1692,7 @@ ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file, goto out; } - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (!qp) { ret = -EINVAL; goto out; @@ -1886,14 +1739,13 @@ ssize_t ib_uverbs_query_qp(struct ib_uverbs_file *file, resp.max_inline_data = init_attr->cap.max_inline_data; resp.sq_sig_all = init_attr->sq_sig_type == IB_SIGNAL_ALL_WR; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); out: kfree(attr); kfree(init_attr); - return ret ? ret : in_len; + return ret; } /* Remove ignored fields set in the attribute mask */ @@ -1933,8 +1785,8 @@ static void copy_ah_attr_from_uverbs(struct ib_device *dev, rdma_ah_set_make_grd(rdma_attr, false); } -static int modify_qp(struct ib_uverbs_file *file, - struct ib_uverbs_ex_modify_qp *cmd, struct ib_udata *udata) +static int modify_qp(struct uverbs_attr_bundle *attrs, + struct ib_uverbs_ex_modify_qp *cmd) { struct ib_qp_attr *attr; struct ib_qp *qp; @@ -1944,7 +1796,8 @@ static int modify_qp(struct ib_uverbs_file *file, if (!attr) return -ENOMEM; - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd->base.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd->base.qp_handle, + attrs); if (!qp) { ret = -EINVAL; goto out; @@ -2081,7 +1934,7 @@ static int modify_qp(struct ib_uverbs_file *file, ret = ib_modify_qp_with_udata(qp, attr, modify_qp_mask(qp->qp_type, cmd->base.attr_mask), - udata); + &attrs->driver_udata); release_qp: uobj_put_obj_read(qp); @@ -2091,80 +1944,64 @@ out: return ret; } -ssize_t ib_uverbs_modify_qp(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_modify_qp(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_modify_qp cmd = {}; - struct ib_udata udata; + struct ib_uverbs_ex_modify_qp cmd; int ret; - if (copy_from_user(&cmd.base, buf, sizeof(cmd.base))) - return -EFAULT; + ret = uverbs_request(attrs, &cmd.base, sizeof(cmd.base)); + if (ret) + return ret; if (cmd.base.attr_mask & ~((IB_USER_LEGACY_LAST_QP_ATTR_MASK << 1) - 1)) return -EOPNOTSUPP; - ib_uverbs_init_udata(&udata, buf + sizeof(cmd.base), NULL, - in_len - sizeof(cmd.base) - sizeof(struct ib_uverbs_cmd_hdr), - out_len); - - ret = modify_qp(file, &cmd, &udata); - if (ret) - return ret; - - return in_len; + return modify_qp(attrs, &cmd); } -int ib_uverbs_ex_modify_qp(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_modify_qp(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_modify_qp cmd = {}; + struct ib_uverbs_ex_modify_qp cmd; + struct ib_uverbs_ex_modify_qp_resp resp = { + .response_length = uverbs_response_length(attrs, sizeof(resp)) + }; int ret; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; + /* * Last bit is reserved for extending the attr_mask by * using another field. */ BUILD_BUG_ON(IB_USER_LAST_QP_ATTR_MASK == (1 << 31)); - if (ucore->inlen < sizeof(cmd.base)) - return -EINVAL; - - ret = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); - if (ret) - return ret; - if (cmd.base.attr_mask & ~((IB_USER_LAST_QP_ATTR_MASK << 1) - 1)) return -EOPNOTSUPP; - if (ucore->inlen > sizeof(cmd)) { - if (!ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; - } - - ret = modify_qp(file, &cmd, uhw); + ret = modify_qp(attrs, &cmd); + if (ret) + return ret; - return ret; + return uverbs_response(attrs, &resp, sizeof(resp)); } -ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_destroy_qp(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_destroy_qp cmd; struct ib_uverbs_destroy_qp_resp resp; struct ib_uobject *uobj; struct ib_uqp_object *obj; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_get_destroy(UVERBS_OBJECT_QP, cmd.qp_handle, file); + uobj = uobj_get_destroy(UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -2174,10 +2011,7 @@ ssize_t ib_uverbs_destroy_qp(struct ib_uverbs_file *file, uobj_put_destroy(uobj); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } static void *alloc_wr(size_t wr_size, __u32 num_sge) @@ -2190,9 +2024,7 @@ static void *alloc_wr(size_t wr_size, __u32 num_sge) num_sge * sizeof (struct ib_sge), GFP_KERNEL); } -ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_post_send(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_post_send cmd; struct ib_uverbs_post_send_resp resp; @@ -2202,24 +2034,31 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, struct ib_qp *qp; int i, sg_ind; int is_ud; - ssize_t ret = -EINVAL; + int ret, ret2; size_t next_size; + const struct ib_sge __user *sgls; + const void __user *wqes; + struct uverbs_req_iter iter; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - if (in_len < sizeof cmd + cmd.wqe_size * cmd.wr_count + - cmd.sge_count * sizeof (struct ib_uverbs_sge)) - return -EINVAL; - - if (cmd.wqe_size < sizeof (struct ib_uverbs_send_wr)) - return -EINVAL; + ret = uverbs_request_start(attrs, &iter, &cmd, sizeof(cmd)); + if (ret) + return ret; + wqes = uverbs_request_next_ptr(&iter, cmd.wqe_size * cmd.wr_count); + if (IS_ERR(wqes)) + return PTR_ERR(wqes); + sgls = uverbs_request_next_ptr( + &iter, cmd.sge_count * sizeof(struct ib_uverbs_sge)); + if (IS_ERR(sgls)) + return PTR_ERR(sgls); + ret = uverbs_request_finish(&iter); + if (ret) + return ret; user_wr = kmalloc(cmd.wqe_size, GFP_KERNEL); if (!user_wr) return -ENOMEM; - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (!qp) goto out; @@ -2227,8 +2066,7 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, sg_ind = 0; last = NULL; for (i = 0; i < cmd.wr_count; ++i) { - if (copy_from_user(user_wr, - buf + sizeof cmd + i * cmd.wqe_size, + if (copy_from_user(user_wr, wqes + i * cmd.wqe_size, cmd.wqe_size)) { ret = -EFAULT; goto out_put; @@ -2256,7 +2094,7 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, } ud->ah = uobj_get_obj_read(ah, UVERBS_OBJECT_AH, - user_wr->wr.ud.ah, file); + user_wr->wr.ud.ah, attrs); if (!ud->ah) { kfree(ud); ret = -EINVAL; @@ -2336,11 +2174,9 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, if (next->num_sge) { next->sg_list = (void *) next + ALIGN(next_size, sizeof(struct ib_sge)); - if (copy_from_user(next->sg_list, - buf + sizeof cmd + - cmd.wr_count * cmd.wqe_size + - sg_ind * sizeof (struct ib_sge), - next->num_sge * sizeof (struct ib_sge))) { + if (copy_from_user(next->sg_list, sgls + sg_ind, + next->num_sge * + sizeof(struct ib_sge))) { ret = -EFAULT; goto out_put; } @@ -2358,8 +2194,9 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file, break; } - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - ret = -EFAULT; + ret2 = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret2) + ret = ret2; out_put: uobj_put_obj_read(qp); @@ -2375,28 +2212,35 @@ out_put: out: kfree(user_wr); - return ret ? ret : in_len; + return ret; } -static struct ib_recv_wr *ib_uverbs_unmarshall_recv(const char __user *buf, - int in_len, - u32 wr_count, - u32 sge_count, - u32 wqe_size) +static struct ib_recv_wr * +ib_uverbs_unmarshall_recv(struct uverbs_req_iter *iter, u32 wr_count, + u32 wqe_size, u32 sge_count) { struct ib_uverbs_recv_wr *user_wr; struct ib_recv_wr *wr = NULL, *last, *next; int sg_ind; int i; int ret; - - if (in_len < wqe_size * wr_count + - sge_count * sizeof (struct ib_uverbs_sge)) - return ERR_PTR(-EINVAL); + const struct ib_sge __user *sgls; + const void __user *wqes; if (wqe_size < sizeof (struct ib_uverbs_recv_wr)) return ERR_PTR(-EINVAL); + wqes = uverbs_request_next_ptr(iter, wqe_size * wr_count); + if (IS_ERR(wqes)) + return ERR_CAST(wqes); + sgls = uverbs_request_next_ptr( + iter, sge_count * sizeof(struct ib_uverbs_sge)); + if (IS_ERR(sgls)) + return ERR_CAST(sgls); + ret = uverbs_request_finish(iter); + if (ret) + return ERR_PTR(ret); + user_wr = kmalloc(wqe_size, GFP_KERNEL); if (!user_wr) return ERR_PTR(-ENOMEM); @@ -2404,7 +2248,7 @@ static struct ib_recv_wr *ib_uverbs_unmarshall_recv(const char __user *buf, sg_ind = 0; last = NULL; for (i = 0; i < wr_count; ++i) { - if (copy_from_user(user_wr, buf + i * wqe_size, + if (copy_from_user(user_wr, wqes + i * wqe_size, wqe_size)) { ret = -EFAULT; goto err; @@ -2443,10 +2287,9 @@ static struct ib_recv_wr *ib_uverbs_unmarshall_recv(const char __user *buf, if (next->num_sge) { next->sg_list = (void *) next + ALIGN(sizeof *next, sizeof (struct ib_sge)); - if (copy_from_user(next->sg_list, - buf + wr_count * wqe_size + - sg_ind * sizeof (struct ib_sge), - next->num_sge * sizeof (struct ib_sge))) { + if (copy_from_user(next->sg_list, sgls + sg_ind, + next->num_sge * + sizeof(struct ib_sge))) { ret = -EFAULT; goto err; } @@ -2470,29 +2313,30 @@ err: return ERR_PTR(ret); } -ssize_t ib_uverbs_post_recv(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_post_recv(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_post_recv cmd; struct ib_uverbs_post_recv_resp resp; struct ib_recv_wr *wr, *next; const struct ib_recv_wr *bad_wr; struct ib_qp *qp; - ssize_t ret = -EINVAL; + int ret, ret2; + struct uverbs_req_iter iter; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request_start(attrs, &iter, &cmd, sizeof(cmd)); + if (ret) + return ret; - wr = ib_uverbs_unmarshall_recv(buf + sizeof cmd, - in_len - sizeof cmd, cmd.wr_count, - cmd.sge_count, cmd.wqe_size); + wr = ib_uverbs_unmarshall_recv(&iter, cmd.wr_count, cmd.wqe_size, + cmd.sge_count); if (IS_ERR(wr)) return PTR_ERR(wr); - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); - if (!qp) + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); + if (!qp) { + ret = -EINVAL; goto out; + } resp.bad_wr = 0; ret = qp->device->post_recv(qp->real_qp, wr, &bad_wr); @@ -2506,9 +2350,9 @@ ssize_t ib_uverbs_post_recv(struct ib_uverbs_file *file, } } - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - ret = -EFAULT; - + ret2 = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret2) + ret = ret2; out: while (wr) { next = wr->next; @@ -2516,36 +2360,36 @@ out: wr = next; } - return ret ? ret : in_len; + return ret; } -ssize_t ib_uverbs_post_srq_recv(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_post_srq_recv(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_post_srq_recv cmd; struct ib_uverbs_post_srq_recv_resp resp; struct ib_recv_wr *wr, *next; const struct ib_recv_wr *bad_wr; struct ib_srq *srq; - ssize_t ret = -EINVAL; + int ret, ret2; + struct uverbs_req_iter iter; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request_start(attrs, &iter, &cmd, sizeof(cmd)); + if (ret) + return ret; - wr = ib_uverbs_unmarshall_recv(buf + sizeof cmd, - in_len - sizeof cmd, cmd.wr_count, - cmd.sge_count, cmd.wqe_size); + wr = ib_uverbs_unmarshall_recv(&iter, cmd.wr_count, cmd.wqe_size, + cmd.sge_count); if (IS_ERR(wr)) return PTR_ERR(wr); - srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, file); - if (!srq) + srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, attrs); + if (!srq) { + ret = -EINVAL; goto out; + } resp.bad_wr = 0; - ret = srq->device->post_srq_recv ? - srq->device->post_srq_recv(srq, wr, &bad_wr) : -EOPNOTSUPP; + ret = srq->device->post_srq_recv(srq, wr, &bad_wr); uobj_put_obj_read(srq); @@ -2556,8 +2400,9 @@ ssize_t ib_uverbs_post_srq_recv(struct ib_uverbs_file *file, break; } - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - ret = -EFAULT; + ret2 = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret2) + ret = ret2; out: while (wr) { @@ -2566,12 +2411,10 @@ out: wr = next; } - return ret ? ret : in_len; + return ret; } -ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_create_ah(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_ah cmd; struct ib_uverbs_create_ah_resp resp; @@ -2580,21 +2423,13 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file, struct ib_ah *ah; struct rdma_ah_attr attr = {}; int ret; - struct ib_udata udata; struct ib_device *ib_dev; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_alloc(UVERBS_OBJECT_AH, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_AH, attrs, &ib_dev); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -2603,7 +2438,7 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file, goto err; } - pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, file); + pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, attrs); if (!pd) { ret = -EINVAL; goto err; @@ -2627,7 +2462,7 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file, rdma_ah_set_ah_flags(&attr, 0); } - ah = rdma_create_user_ah(pd, &attr, &udata); + ah = rdma_create_user_ah(pd, &attr, &attrs->driver_udata); if (IS_ERR(ah)) { ret = PTR_ERR(ah); goto err_put; @@ -2639,13 +2474,12 @@ ssize_t ib_uverbs_create_ah(struct ib_uverbs_file *file, resp.ah_handle = uobj->id; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } uobj_put_obj_read(pd); - return uobj_alloc_commit(uobj, in_len); + return uobj_alloc_commit(uobj); err_copy: rdma_destroy_ah(ah); @@ -2658,21 +2492,19 @@ err: return ret; } -ssize_t ib_uverbs_destroy_ah(struct ib_uverbs_file *file, - const char __user *buf, int in_len, int out_len) +static int ib_uverbs_destroy_ah(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_destroy_ah cmd; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - return uobj_perform_destroy(UVERBS_OBJECT_AH, cmd.ah_handle, file, - in_len); + return uobj_perform_destroy(UVERBS_OBJECT_AH, cmd.ah_handle, attrs); } -ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_attach_mcast(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_attach_mcast cmd; struct ib_qp *qp; @@ -2680,10 +2512,11 @@ ssize_t ib_uverbs_attach_mcast(struct ib_uverbs_file *file, struct ib_uverbs_mcast_entry *mcast; int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (!qp) return -EINVAL; @@ -2716,12 +2549,10 @@ out_put: mutex_unlock(&obj->mcast_lock); uobj_put_obj_read(qp); - return ret ? ret : in_len; + return ret; } -ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_detach_mcast(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_detach_mcast cmd; struct ib_uqp_object *obj; @@ -2730,10 +2561,11 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file, int ret = -EINVAL; bool found = false; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (!qp) return -EINVAL; @@ -2759,7 +2591,7 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file, out_put: mutex_unlock(&obj->mcast_lock); uobj_put_obj_read(qp); - return ret ? ret : in_len; + return ret; } struct ib_uflow_resources *flow_resources_alloc(size_t num_specs) @@ -2838,7 +2670,7 @@ void flow_resources_add(struct ib_uflow_resources *uflow_res, } EXPORT_SYMBOL(flow_resources_add); -static int kern_spec_to_ib_spec_action(struct ib_uverbs_file *ufile, +static int kern_spec_to_ib_spec_action(const struct uverbs_attr_bundle *attrs, struct ib_uverbs_flow_spec *kern_spec, union ib_flow_spec *ib_spec, struct ib_uflow_resources *uflow_res) @@ -2867,7 +2699,7 @@ static int kern_spec_to_ib_spec_action(struct ib_uverbs_file *ufile, ib_spec->action.act = uobj_get_obj_read(flow_action, UVERBS_OBJECT_FLOW_ACTION, kern_spec->action.handle, - ufile); + attrs); if (!ib_spec->action.act) return -EINVAL; ib_spec->action.size = @@ -2885,7 +2717,7 @@ static int kern_spec_to_ib_spec_action(struct ib_uverbs_file *ufile, uobj_get_obj_read(counters, UVERBS_OBJECT_COUNTERS, kern_spec->flow_count.handle, - ufile); + attrs); if (!ib_spec->flow_count.counters) return -EINVAL; ib_spec->flow_count.size = @@ -3066,7 +2898,7 @@ static int kern_spec_to_ib_spec_filter(struct ib_uverbs_flow_spec *kern_spec, kern_filter_sz, ib_spec); } -static int kern_spec_to_ib_spec(struct ib_uverbs_file *ufile, +static int kern_spec_to_ib_spec(struct uverbs_attr_bundle *attrs, struct ib_uverbs_flow_spec *kern_spec, union ib_flow_spec *ib_spec, struct ib_uflow_resources *uflow_res) @@ -3075,17 +2907,15 @@ static int kern_spec_to_ib_spec(struct ib_uverbs_file *ufile, return -EINVAL; if (kern_spec->type >= IB_FLOW_SPEC_ACTION_TAG) - return kern_spec_to_ib_spec_action(ufile, kern_spec, ib_spec, + return kern_spec_to_ib_spec_action(attrs, kern_spec, ib_spec, uflow_res); else return kern_spec_to_ib_spec_filter(kern_spec, ib_spec); } -int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_create_wq(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_create_wq cmd = {}; + struct ib_uverbs_ex_create_wq cmd; struct ib_uverbs_ex_create_wq_resp resp = {}; struct ib_uwq_object *obj; int err = 0; @@ -3093,43 +2923,27 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file, struct ib_pd *pd; struct ib_wq *wq; struct ib_wq_init_attr wq_init_attr = {}; - size_t required_cmd_sz; - size_t required_resp_len; struct ib_device *ib_dev; - required_cmd_sz = offsetof(typeof(cmd), max_sge) + sizeof(cmd.max_sge); - required_resp_len = offsetof(typeof(resp), wqn) + sizeof(resp.wqn); - - if (ucore->inlen < required_cmd_sz) - return -EINVAL; - - if (ucore->outlen < required_resp_len) - return -ENOSPC; - - if (ucore->inlen > sizeof(cmd) && - !ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; - - err = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); + err = uverbs_request(attrs, &cmd, sizeof(cmd)); if (err) return err; if (cmd.comp_mask) return -EOPNOTSUPP; - obj = (struct ib_uwq_object *)uobj_alloc(UVERBS_OBJECT_WQ, file, + obj = (struct ib_uwq_object *)uobj_alloc(UVERBS_OBJECT_WQ, attrs, &ib_dev); if (IS_ERR(obj)) return PTR_ERR(obj); - pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, file); + pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd.pd_handle, attrs); if (!pd) { err = -EINVAL; goto err_uobj; } - cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, file); + cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (!cq) { err = -EINVAL; goto err_put_pd; @@ -3138,20 +2952,14 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file, wq_init_attr.cq = cq; wq_init_attr.max_sge = cmd.max_sge; wq_init_attr.max_wr = cmd.max_wr; - wq_init_attr.wq_context = file; + wq_init_attr.wq_context = attrs->ufile; wq_init_attr.wq_type = cmd.wq_type; wq_init_attr.event_handler = ib_uverbs_wq_event_handler; - if (ucore->inlen >= (offsetof(typeof(cmd), create_flags) + - sizeof(cmd.create_flags))) - wq_init_attr.create_flags = cmd.create_flags; + wq_init_attr.create_flags = cmd.create_flags; obj->uevent.events_reported = 0; INIT_LIST_HEAD(&obj->uevent.event_list); - if (!pd->device->create_wq) { - err = -EOPNOTSUPP; - goto err_put_cq; - } - wq = pd->device->create_wq(pd, &wq_init_attr, uhw); + wq = pd->device->create_wq(pd, &wq_init_attr, &attrs->driver_udata); if (IS_ERR(wq)) { err = PTR_ERR(wq); goto err_put_cq; @@ -3175,15 +2983,14 @@ int ib_uverbs_ex_create_wq(struct ib_uverbs_file *file, resp.max_sge = wq_init_attr.max_sge; resp.max_wr = wq_init_attr.max_wr; resp.wqn = wq->wq_num; - resp.response_length = required_resp_len; - err = ib_copy_to_udata(ucore, - &resp, resp.response_length); + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + err = uverbs_response(attrs, &resp, sizeof(resp)); if (err) goto err_copy; uobj_put_obj_read(pd); uobj_put_obj_read(cq); - return uobj_alloc_commit(&obj->uevent.uobject, 0); + return uobj_alloc_commit(&obj->uevent.uobject); err_copy: ib_destroy_wq(wq); @@ -3197,41 +3004,23 @@ err_uobj: return err; } -int ib_uverbs_ex_destroy_wq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_destroy_wq(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_destroy_wq cmd = {}; + struct ib_uverbs_ex_destroy_wq cmd; struct ib_uverbs_ex_destroy_wq_resp resp = {}; struct ib_uobject *uobj; struct ib_uwq_object *obj; - size_t required_cmd_sz; - size_t required_resp_len; int ret; - required_cmd_sz = offsetof(typeof(cmd), wq_handle) + sizeof(cmd.wq_handle); - required_resp_len = offsetof(typeof(resp), reserved) + sizeof(resp.reserved); - - if (ucore->inlen < required_cmd_sz) - return -EINVAL; - - if (ucore->outlen < required_resp_len) - return -ENOSPC; - - if (ucore->inlen > sizeof(cmd) && - !ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; - - ret = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; if (cmd.comp_mask) return -EOPNOTSUPP; - resp.response_length = required_resp_len; - uobj = uobj_get_destroy(UVERBS_OBJECT_WQ, cmd.wq_handle, file); + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + uobj = uobj_get_destroy(UVERBS_OBJECT_WQ, cmd.wq_handle, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -3240,29 +3029,17 @@ int ib_uverbs_ex_destroy_wq(struct ib_uverbs_file *file, uobj_put_destroy(uobj); - return ib_copy_to_udata(ucore, &resp, resp.response_length); + return uverbs_response(attrs, &resp, sizeof(resp)); } -int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_modify_wq(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_modify_wq cmd = {}; + struct ib_uverbs_ex_modify_wq cmd; struct ib_wq *wq; struct ib_wq_attr wq_attr = {}; - size_t required_cmd_sz; int ret; - required_cmd_sz = offsetof(typeof(cmd), curr_wq_state) + sizeof(cmd.curr_wq_state); - if (ucore->inlen < required_cmd_sz) - return -EINVAL; - - if (ucore->inlen > sizeof(cmd) && - !ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; - - ret = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; @@ -3272,7 +3049,7 @@ int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file, if (cmd.attr_mask > (IB_WQ_STATE | IB_WQ_CUR_STATE | IB_WQ_FLAGS)) return -EINVAL; - wq = uobj_get_obj_read(wq, UVERBS_OBJECT_WQ, cmd.wq_handle, file); + wq = uobj_get_obj_read(wq, UVERBS_OBJECT_WQ, cmd.wq_handle, attrs); if (!wq) return -EINVAL; @@ -3282,24 +3059,18 @@ int ib_uverbs_ex_modify_wq(struct ib_uverbs_file *file, wq_attr.flags = cmd.flags; wq_attr.flags_mask = cmd.flags_mask; } - if (!wq->device->modify_wq) { - ret = -EOPNOTSUPP; - goto out; - } - ret = wq->device->modify_wq(wq, &wq_attr, cmd.attr_mask, uhw); -out: + ret = wq->device->modify_wq(wq, &wq_attr, cmd.attr_mask, + &attrs->driver_udata); uobj_put_obj_read(wq); return ret; } -int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_create_rwq_ind_table(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_create_rwq_ind_table cmd = {}; + struct ib_uverbs_ex_create_rwq_ind_table cmd; struct ib_uverbs_ex_create_rwq_ind_table_resp resp = {}; struct ib_uobject *uobj; - int err = 0; + int err; struct ib_rwq_ind_table_init_attr init_attr = {}; struct ib_rwq_ind_table *rwq_ind_tbl; struct ib_wq **wqs = NULL; @@ -3307,27 +3078,13 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, struct ib_wq *wq = NULL; int i, j, num_read_wqs; u32 num_wq_handles; - u32 expected_in_size; - size_t required_cmd_sz_header; - size_t required_resp_len; + struct uverbs_req_iter iter; struct ib_device *ib_dev; - required_cmd_sz_header = offsetof(typeof(cmd), log_ind_tbl_size) + sizeof(cmd.log_ind_tbl_size); - required_resp_len = offsetof(typeof(resp), ind_tbl_num) + sizeof(resp.ind_tbl_num); - - if (ucore->inlen < required_cmd_sz_header) - return -EINVAL; - - if (ucore->outlen < required_resp_len) - return -ENOSPC; - - err = ib_copy_from_udata(&cmd, ucore, required_cmd_sz_header); + err = uverbs_request_start(attrs, &iter, &cmd, sizeof(cmd)); if (err) return err; - ucore->inbuf += required_cmd_sz_header; - ucore->inlen -= required_cmd_sz_header; - if (cmd.comp_mask) return -EOPNOTSUPP; @@ -3335,26 +3092,17 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, return -EINVAL; num_wq_handles = 1 << cmd.log_ind_tbl_size; - expected_in_size = num_wq_handles * sizeof(__u32); - if (num_wq_handles == 1) - /* input size for wq handles is u64 aligned */ - expected_in_size += sizeof(__u32); - - if (ucore->inlen < expected_in_size) - return -EINVAL; - - if (ucore->inlen > expected_in_size && - !ib_is_udata_cleared(ucore, expected_in_size, - ucore->inlen - expected_in_size)) - return -EOPNOTSUPP; - wqs_handles = kcalloc(num_wq_handles, sizeof(*wqs_handles), GFP_KERNEL); if (!wqs_handles) return -ENOMEM; - err = ib_copy_from_udata(wqs_handles, ucore, - num_wq_handles * sizeof(__u32)); + err = uverbs_request_next(&iter, wqs_handles, + num_wq_handles * sizeof(__u32)); + if (err) + goto err_free; + + err = uverbs_request_finish(&iter); if (err) goto err_free; @@ -3367,7 +3115,7 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, for (num_read_wqs = 0; num_read_wqs < num_wq_handles; num_read_wqs++) { wq = uobj_get_obj_read(wq, UVERBS_OBJECT_WQ, - wqs_handles[num_read_wqs], file); + wqs_handles[num_read_wqs], attrs); if (!wq) { err = -EINVAL; goto put_wqs; @@ -3376,7 +3124,7 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, wqs[num_read_wqs] = wq; } - uobj = uobj_alloc(UVERBS_OBJECT_RWQ_IND_TBL, file, &ib_dev); + uobj = uobj_alloc(UVERBS_OBJECT_RWQ_IND_TBL, attrs, &ib_dev); if (IS_ERR(uobj)) { err = PTR_ERR(uobj); goto put_wqs; @@ -3385,11 +3133,8 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, init_attr.log_ind_tbl_size = cmd.log_ind_tbl_size; init_attr.ind_tbl = wqs; - if (!ib_dev->create_rwq_ind_table) { - err = -EOPNOTSUPP; - goto err_uobj; - } - rwq_ind_tbl = ib_dev->create_rwq_ind_table(ib_dev, &init_attr, uhw); + rwq_ind_tbl = ib_dev->create_rwq_ind_table(ib_dev, &init_attr, + &attrs->driver_udata); if (IS_ERR(rwq_ind_tbl)) { err = PTR_ERR(rwq_ind_tbl); @@ -3408,10 +3153,9 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, resp.ind_tbl_handle = uobj->id; resp.ind_tbl_num = rwq_ind_tbl->ind_tbl_num; - resp.response_length = required_resp_len; + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); - err = ib_copy_to_udata(ucore, - &resp, resp.response_length); + err = uverbs_response(attrs, &resp, sizeof(resp)); if (err) goto err_copy; @@ -3420,7 +3164,7 @@ int ib_uverbs_ex_create_rwq_ind_table(struct ib_uverbs_file *file, for (j = 0; j < num_read_wqs; j++) uobj_put_obj_read(wqs[j]); - return uobj_alloc_commit(uobj, 0); + return uobj_alloc_commit(uobj); err_copy: ib_destroy_rwq_ind_table(rwq_ind_tbl); @@ -3435,25 +3179,12 @@ err_free: return err; } -int ib_uverbs_ex_destroy_rwq_ind_table(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_destroy_rwq_ind_table(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_destroy_rwq_ind_table cmd = {}; - int ret; - size_t required_cmd_sz; - - required_cmd_sz = offsetof(typeof(cmd), ind_tbl_handle) + sizeof(cmd.ind_tbl_handle); - - if (ucore->inlen < required_cmd_sz) - return -EINVAL; - - if (ucore->inlen > sizeof(cmd) && - !ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; + struct ib_uverbs_ex_destroy_rwq_ind_table cmd; + int ret; - ret = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; @@ -3461,12 +3192,10 @@ int ib_uverbs_ex_destroy_rwq_ind_table(struct ib_uverbs_file *file, return -EOPNOTSUPP; return uobj_perform_destroy(UVERBS_OBJECT_RWQ_IND_TBL, - cmd.ind_tbl_handle, file, 0); + cmd.ind_tbl_handle, attrs); } -int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_create_flow(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_flow cmd; struct ib_uverbs_create_flow_resp resp; @@ -3477,24 +3206,16 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, struct ib_qp *qp; struct ib_uflow_resources *uflow_res; struct ib_uverbs_flow_spec_hdr *kern_spec; - int err = 0; + struct uverbs_req_iter iter; + int err; void *ib_spec; int i; struct ib_device *ib_dev; - if (ucore->inlen < sizeof(cmd)) - return -EINVAL; - - if (ucore->outlen < sizeof(resp)) - return -ENOSPC; - - err = ib_copy_from_udata(&cmd, ucore, sizeof(cmd)); + err = uverbs_request_start(attrs, &iter, &cmd, sizeof(cmd)); if (err) return err; - ucore->inbuf += sizeof(cmd); - ucore->inlen -= sizeof(cmd); - if (cmd.comp_mask) return -EINVAL; @@ -3512,8 +3233,7 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, if (cmd.flow_attr.num_of_specs > IB_FLOW_SPEC_SUPPORT_LAYERS) return -EINVAL; - if (cmd.flow_attr.size > ucore->inlen || - cmd.flow_attr.size > + if (cmd.flow_attr.size > (cmd.flow_attr.num_of_specs * sizeof(struct ib_uverbs_flow_spec))) return -EINVAL; @@ -3528,21 +3248,25 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, return -ENOMEM; *kern_flow_attr = cmd.flow_attr; - err = ib_copy_from_udata(&kern_flow_attr->flow_specs, ucore, - cmd.flow_attr.size); + err = uverbs_request_next(&iter, &kern_flow_attr->flow_specs, + cmd.flow_attr.size); if (err) goto err_free_attr; } else { kern_flow_attr = &cmd.flow_attr; } - uobj = uobj_alloc(UVERBS_OBJECT_FLOW, file, &ib_dev); + err = uverbs_request_finish(&iter); + if (err) + goto err_free_attr; + + uobj = uobj_alloc(UVERBS_OBJECT_FLOW, attrs, &ib_dev); if (IS_ERR(uobj)) { err = PTR_ERR(uobj); goto err_free_attr; } - qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, file); + qp = uobj_get_obj_read(qp, UVERBS_OBJECT_QP, cmd.qp_handle, attrs); if (!qp) { err = -EINVAL; goto err_uobj; @@ -3553,11 +3277,6 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, goto err_put; } - if (!qp->device->create_flow) { - err = -EOPNOTSUPP; - goto err_put; - } - flow_attr = kzalloc(struct_size(flow_attr, flows, cmd.flow_attr.num_of_specs), GFP_KERNEL); if (!flow_attr) { @@ -3584,7 +3303,7 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, cmd.flow_attr.size >= kern_spec->size; i++) { err = kern_spec_to_ib_spec( - file, (struct ib_uverbs_flow_spec *)kern_spec, + attrs, (struct ib_uverbs_flow_spec *)kern_spec, ib_spec, uflow_res); if (err) goto err_free; @@ -3602,8 +3321,8 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, goto err_free; } - flow_id = qp->device->create_flow(qp, flow_attr, - IB_FLOW_DOMAIN_USER, uhw); + flow_id = qp->device->create_flow(qp, flow_attr, IB_FLOW_DOMAIN_USER, + &attrs->driver_udata); if (IS_ERR(flow_id)) { err = PTR_ERR(flow_id); @@ -3615,8 +3334,7 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, memset(&resp, 0, sizeof(resp)); resp.flow_handle = uobj->id; - err = ib_copy_to_udata(ucore, - &resp, sizeof(resp)); + err = uverbs_response(attrs, &resp, sizeof(resp)); if (err) goto err_copy; @@ -3624,7 +3342,7 @@ int ib_uverbs_ex_create_flow(struct ib_uverbs_file *file, kfree(flow_attr); if (cmd.flow_attr.num_of_specs) kfree(kern_flow_attr); - return uobj_alloc_commit(uobj, 0); + return uobj_alloc_commit(uobj); err_copy: if (!qp->device->destroy_flow(flow_id)) atomic_dec(&qp->usecnt); @@ -3642,28 +3360,22 @@ err_free_attr: return err; } -int ib_uverbs_ex_destroy_flow(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_destroy_flow(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_destroy_flow cmd; int ret; - if (ucore->inlen < sizeof(cmd)) - return -EINVAL; - - ret = ib_copy_from_udata(&cmd, ucore, sizeof(cmd)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; if (cmd.comp_mask) return -EINVAL; - return uobj_perform_destroy(UVERBS_OBJECT_FLOW, cmd.flow_handle, file, - 0); + return uobj_perform_destroy(UVERBS_OBJECT_FLOW, cmd.flow_handle, attrs); } -static int __uverbs_create_xsrq(struct ib_uverbs_file *file, +static int __uverbs_create_xsrq(struct uverbs_attr_bundle *attrs, struct ib_uverbs_create_xsrq *cmd, struct ib_udata *udata) { @@ -3676,7 +3388,7 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file, int ret; struct ib_device *ib_dev; - obj = (struct ib_usrq_object *)uobj_alloc(UVERBS_OBJECT_SRQ, file, + obj = (struct ib_usrq_object *)uobj_alloc(UVERBS_OBJECT_SRQ, attrs, &ib_dev); if (IS_ERR(obj)) return PTR_ERR(obj); @@ -3686,7 +3398,7 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file, if (cmd->srq_type == IB_SRQT_XRC) { xrcd_uobj = uobj_get_read(UVERBS_OBJECT_XRCD, cmd->xrcd_handle, - file); + attrs); if (IS_ERR(xrcd_uobj)) { ret = -EINVAL; goto err; @@ -3704,21 +3416,21 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file, if (ib_srq_has_cq(cmd->srq_type)) { attr.ext.cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, - cmd->cq_handle, file); + cmd->cq_handle, attrs); if (!attr.ext.cq) { ret = -EINVAL; goto err_put_xrcd; } } - pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd->pd_handle, file); + pd = uobj_get_obj_read(pd, UVERBS_OBJECT_PD, cmd->pd_handle, attrs); if (!pd) { ret = -EINVAL; goto err_put_cq; } attr.event_handler = ib_uverbs_srq_event_handler; - attr.srq_context = file; + attr.srq_context = attrs->ufile; attr.srq_type = cmd->srq_type; attr.attr.max_wr = cmd->max_wr; attr.attr.max_sge = cmd->max_sge; @@ -3763,11 +3475,9 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file, if (cmd->srq_type == IB_SRQT_XRC) resp.srqn = srq->ext.xrc.srq_num; - if (copy_to_user(u64_to_user_ptr(cmd->response), - &resp, sizeof resp)) { - ret = -EFAULT; + ret = uverbs_response(attrs, &resp, sizeof(resp)); + if (ret) goto err_copy; - } if (cmd->srq_type == IB_SRQT_XRC) uobj_put_read(xrcd_uobj); @@ -3776,7 +3486,7 @@ static int __uverbs_create_xsrq(struct ib_uverbs_file *file, uobj_put_obj_read(attr.ext.cq); uobj_put_obj_read(pd); - return uobj_alloc_commit(&obj->uevent.uobject, 0); + return uobj_alloc_commit(&obj->uevent.uobject); err_copy: ib_destroy_srq(srq); @@ -3799,21 +3509,15 @@ err: return ret; } -ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_create_srq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_srq cmd; struct ib_uverbs_create_xsrq xcmd; - struct ib_uverbs_create_srq_resp resp; - struct ib_udata udata; int ret; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; memset(&xcmd, 0, sizeof(xcmd)); xcmd.response = cmd.response; @@ -3824,77 +3528,48 @@ ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file, xcmd.max_sge = cmd.max_sge; xcmd.srq_limit = cmd.srq_limit; - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); - - ret = __uverbs_create_xsrq(file, &xcmd, &udata); - if (ret) - return ret; - - return in_len; + return __uverbs_create_xsrq(attrs, &xcmd, &attrs->driver_udata); } -ssize_t ib_uverbs_create_xsrq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, int out_len) +static int ib_uverbs_create_xsrq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_create_xsrq cmd; - struct ib_uverbs_create_srq_resp resp; - struct ib_udata udata; int ret; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof(cmd), - u64_to_user_ptr(cmd.response) + sizeof(resp), - in_len - sizeof(cmd) - sizeof(struct ib_uverbs_cmd_hdr), - out_len - sizeof(resp)); - - ret = __uverbs_create_xsrq(file, &cmd, &udata); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; - return in_len; + return __uverbs_create_xsrq(attrs, &cmd, &attrs->driver_udata); } -ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_modify_srq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_modify_srq cmd; - struct ib_udata udata; struct ib_srq *srq; struct ib_srq_attr attr; int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; - - ib_uverbs_init_udata(&udata, buf + sizeof cmd, NULL, in_len - sizeof cmd, - out_len); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, file); + srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, attrs); if (!srq) return -EINVAL; attr.max_wr = cmd.max_wr; attr.srq_limit = cmd.srq_limit; - ret = srq->device->modify_srq(srq, &attr, cmd.attr_mask, &udata); + ret = srq->device->modify_srq(srq, &attr, cmd.attr_mask, + &attrs->driver_udata); uobj_put_obj_read(srq); - return ret ? ret : in_len; + return ret; } -ssize_t ib_uverbs_query_srq(struct ib_uverbs_file *file, - const char __user *buf, - int in_len, int out_len) +static int ib_uverbs_query_srq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_query_srq cmd; struct ib_uverbs_query_srq_resp resp; @@ -3902,13 +3577,11 @@ ssize_t ib_uverbs_query_srq(struct ib_uverbs_file *file, struct ib_srq *srq; int ret; - if (out_len < sizeof resp) - return -ENOSPC; - - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, file); + srq = uobj_get_obj_read(srq, UVERBS_OBJECT_SRQ, cmd.srq_handle, attrs); if (!srq) return -EINVAL; @@ -3925,25 +3598,22 @@ ssize_t ib_uverbs_query_srq(struct ib_uverbs_file *file, resp.max_sge = attr.max_sge; resp.srq_limit = attr.srq_limit; - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof resp)) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } -ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) +static int ib_uverbs_destroy_srq(struct uverbs_attr_bundle *attrs) { struct ib_uverbs_destroy_srq cmd; struct ib_uverbs_destroy_srq_resp resp; struct ib_uobject *uobj; struct ib_uevent_object *obj; + int ret; - if (copy_from_user(&cmd, buf, sizeof cmd)) - return -EFAULT; + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); + if (ret) + return ret; - uobj = uobj_get_destroy(UVERBS_OBJECT_SRQ, cmd.srq_handle, file); + uobj = uobj_get_destroy(UVERBS_OBJECT_SRQ, cmd.srq_handle, attrs); if (IS_ERR(uobj)) return PTR_ERR(uobj); @@ -3953,35 +3623,24 @@ ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file, uobj_put_destroy(uobj); - if (copy_to_user(u64_to_user_ptr(cmd.response), &resp, sizeof(resp))) - return -EFAULT; - - return in_len; + return uverbs_response(attrs, &resp, sizeof(resp)); } -int ib_uverbs_ex_query_device(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_query_device(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_query_device_resp resp = { {0} }; + struct ib_uverbs_ex_query_device_resp resp = {}; struct ib_uverbs_ex_query_device cmd; struct ib_device_attr attr = {0}; struct ib_ucontext *ucontext; struct ib_device *ib_dev; int err; - ucontext = ib_uverbs_get_ucontext(file); + ucontext = ib_uverbs_get_ucontext(attrs); if (IS_ERR(ucontext)) return PTR_ERR(ucontext); ib_dev = ucontext->device; - if (!ib_dev->query_device) - return -EOPNOTSUPP; - - if (ucore->inlen < sizeof(cmd)) - return -EINVAL; - - err = ib_copy_from_udata(&cmd, ucore, sizeof(cmd)); + err = uverbs_request(attrs, &cmd, sizeof(cmd)); if (err) return err; @@ -3991,20 +3650,12 @@ int ib_uverbs_ex_query_device(struct ib_uverbs_file *file, if (cmd.reserved) return -EINVAL; - resp.response_length = offsetof(typeof(resp), odp_caps); - - if (ucore->outlen < resp.response_length) - return -ENOSPC; - - err = ib_dev->query_device(ib_dev, &attr, uhw); + err = ib_dev->query_device(ib_dev, &attr, &attrs->driver_udata); if (err) return err; copy_query_dev_fields(ucontext, &resp.base, &attr); - if (ucore->outlen < resp.response_length + sizeof(resp.odp_caps)) - goto end; - #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING resp.odp_caps.general_caps = attr.odp_caps.general_caps; resp.odp_caps.per_transport_caps.rc_odp_caps = @@ -4014,99 +3665,39 @@ int ib_uverbs_ex_query_device(struct ib_uverbs_file *file, resp.odp_caps.per_transport_caps.ud_odp_caps = attr.odp_caps.per_transport_caps.ud_odp_caps; #endif - resp.response_length += sizeof(resp.odp_caps); - - if (ucore->outlen < resp.response_length + sizeof(resp.timestamp_mask)) - goto end; resp.timestamp_mask = attr.timestamp_mask; - resp.response_length += sizeof(resp.timestamp_mask); - - if (ucore->outlen < resp.response_length + sizeof(resp.hca_core_clock)) - goto end; - resp.hca_core_clock = attr.hca_core_clock; - resp.response_length += sizeof(resp.hca_core_clock); - - if (ucore->outlen < resp.response_length + sizeof(resp.device_cap_flags_ex)) - goto end; - resp.device_cap_flags_ex = attr.device_cap_flags; - resp.response_length += sizeof(resp.device_cap_flags_ex); - - if (ucore->outlen < resp.response_length + sizeof(resp.rss_caps)) - goto end; - resp.rss_caps.supported_qpts = attr.rss_caps.supported_qpts; resp.rss_caps.max_rwq_indirection_tables = attr.rss_caps.max_rwq_indirection_tables; resp.rss_caps.max_rwq_indirection_table_size = attr.rss_caps.max_rwq_indirection_table_size; - - resp.response_length += sizeof(resp.rss_caps); - - if (ucore->outlen < resp.response_length + sizeof(resp.max_wq_type_rq)) - goto end; - resp.max_wq_type_rq = attr.max_wq_type_rq; - resp.response_length += sizeof(resp.max_wq_type_rq); - - if (ucore->outlen < resp.response_length + sizeof(resp.raw_packet_caps)) - goto end; - resp.raw_packet_caps = attr.raw_packet_caps; - resp.response_length += sizeof(resp.raw_packet_caps); - - if (ucore->outlen < resp.response_length + sizeof(resp.tm_caps)) - goto end; - resp.tm_caps.max_rndv_hdr_size = attr.tm_caps.max_rndv_hdr_size; resp.tm_caps.max_num_tags = attr.tm_caps.max_num_tags; resp.tm_caps.max_ops = attr.tm_caps.max_ops; resp.tm_caps.max_sge = attr.tm_caps.max_sge; resp.tm_caps.flags = attr.tm_caps.flags; - resp.response_length += sizeof(resp.tm_caps); - - if (ucore->outlen < resp.response_length + sizeof(resp.cq_moderation_caps)) - goto end; - resp.cq_moderation_caps.max_cq_moderation_count = attr.cq_caps.max_cq_moderation_count; resp.cq_moderation_caps.max_cq_moderation_period = attr.cq_caps.max_cq_moderation_period; - resp.response_length += sizeof(resp.cq_moderation_caps); - - if (ucore->outlen < resp.response_length + sizeof(resp.max_dm_size)) - goto end; - resp.max_dm_size = attr.max_dm_size; - resp.response_length += sizeof(resp.max_dm_size); -end: - err = ib_copy_to_udata(ucore, &resp, resp.response_length); - return err; + resp.response_length = uverbs_response_length(attrs, sizeof(resp)); + + return uverbs_response(attrs, &resp, sizeof(resp)); } -int ib_uverbs_ex_modify_cq(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) +static int ib_uverbs_ex_modify_cq(struct uverbs_attr_bundle *attrs) { - struct ib_uverbs_ex_modify_cq cmd = {}; + struct ib_uverbs_ex_modify_cq cmd; struct ib_cq *cq; - size_t required_cmd_sz; int ret; - required_cmd_sz = offsetof(typeof(cmd), reserved) + - sizeof(cmd.reserved); - if (ucore->inlen < required_cmd_sz) - return -EINVAL; - - /* sanity checks */ - if (ucore->inlen > sizeof(cmd) && - !ib_is_udata_cleared(ucore, sizeof(cmd), - ucore->inlen - sizeof(cmd))) - return -EOPNOTSUPP; - - ret = ib_copy_from_udata(&cmd, ucore, min(sizeof(cmd), ucore->inlen)); + ret = uverbs_request(attrs, &cmd, sizeof(cmd)); if (ret) return ret; @@ -4116,7 +3707,7 @@ int ib_uverbs_ex_modify_cq(struct ib_uverbs_file *file, if (cmd.attr_mask > IB_CQ_MODERATE) return -EOPNOTSUPP; - cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, file); + cq = uobj_get_obj_read(cq, UVERBS_OBJECT_CQ, cmd.cq_handle, attrs); if (!cq) return -EINVAL; @@ -4126,3 +3717,381 @@ int ib_uverbs_ex_modify_cq(struct ib_uverbs_file *file, return ret; } + +/* + * Describe the input structs for write(). Some write methods have an input + * only struct, most have an input and output. If the struct has an output then + * the 'response' u64 must be the first field in the request structure. + * + * If udata is present then both the request and response structs have a + * trailing driver_data flex array. In this case the size of the base struct + * cannot be changed. + */ +#define offsetof_after(_struct, _member) \ + (offsetof(_struct, _member) + sizeof(((_struct *)NULL)->_member)) + +#define UAPI_DEF_WRITE_IO(req, resp) \ + .write.has_resp = 1 + \ + BUILD_BUG_ON_ZERO(offsetof(req, response) != 0) + \ + BUILD_BUG_ON_ZERO(sizeof(((req *)0)->response) != \ + sizeof(u64)), \ + .write.req_size = sizeof(req), .write.resp_size = sizeof(resp) + +#define UAPI_DEF_WRITE_I(req) .write.req_size = sizeof(req) + +#define UAPI_DEF_WRITE_UDATA_IO(req, resp) \ + UAPI_DEF_WRITE_IO(req, resp), \ + .write.has_udata = \ + 1 + \ + BUILD_BUG_ON_ZERO(offsetof(req, driver_data) != \ + sizeof(req)) + \ + BUILD_BUG_ON_ZERO(offsetof(resp, driver_data) != \ + sizeof(resp)) + +#define UAPI_DEF_WRITE_UDATA_I(req) \ + UAPI_DEF_WRITE_I(req), \ + .write.has_udata = \ + 1 + BUILD_BUG_ON_ZERO(offsetof(req, driver_data) != \ + sizeof(req)) + +/* + * The _EX versions are for use with WRITE_EX and allow the last struct member + * to be specified. Buffers that do not include that member will be rejected. + */ +#define UAPI_DEF_WRITE_IO_EX(req, req_last_member, resp, resp_last_member) \ + .write.has_resp = 1, \ + .write.req_size = offsetof_after(req, req_last_member), \ + .write.resp_size = offsetof_after(resp, resp_last_member) + +#define UAPI_DEF_WRITE_I_EX(req, req_last_member) \ + .write.req_size = offsetof_after(req, req_last_member) + +const struct uapi_definition uverbs_def_write_intf[] = { + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_AH, + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_CREATE_AH, + ib_uverbs_create_ah, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_create_ah, + struct ib_uverbs_create_ah_resp), + UAPI_DEF_METHOD_NEEDS_FN(create_ah)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DESTROY_AH, + ib_uverbs_destroy_ah, + UAPI_DEF_WRITE_I(struct ib_uverbs_destroy_ah), + UAPI_DEF_METHOD_NEEDS_FN(destroy_ah))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_COMP_CHANNEL, + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL, + ib_uverbs_create_comp_channel, + UAPI_DEF_WRITE_IO( + struct ib_uverbs_create_comp_channel, + struct ib_uverbs_create_comp_channel_resp))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_CQ, + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_CREATE_CQ, + ib_uverbs_create_cq, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_create_cq, + struct ib_uverbs_create_cq_resp), + UAPI_DEF_METHOD_NEEDS_FN(create_cq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DESTROY_CQ, + ib_uverbs_destroy_cq, + UAPI_DEF_WRITE_IO(struct ib_uverbs_destroy_cq, + struct ib_uverbs_destroy_cq_resp), + UAPI_DEF_METHOD_NEEDS_FN(destroy_cq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_POLL_CQ, + ib_uverbs_poll_cq, + UAPI_DEF_WRITE_IO(struct ib_uverbs_poll_cq, + struct ib_uverbs_poll_cq_resp), + UAPI_DEF_METHOD_NEEDS_FN(poll_cq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_REQ_NOTIFY_CQ, + ib_uverbs_req_notify_cq, + UAPI_DEF_WRITE_I(struct ib_uverbs_req_notify_cq), + UAPI_DEF_METHOD_NEEDS_FN(req_notify_cq)), + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_RESIZE_CQ, + ib_uverbs_resize_cq, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_resize_cq, + struct ib_uverbs_resize_cq_resp), + UAPI_DEF_METHOD_NEEDS_FN(resize_cq)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_CREATE_CQ, + ib_uverbs_ex_create_cq, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_ex_create_cq, + reserved, + struct ib_uverbs_ex_create_cq_resp, + response_length), + UAPI_DEF_METHOD_NEEDS_FN(create_cq)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_MODIFY_CQ, + ib_uverbs_ex_modify_cq, + UAPI_DEF_WRITE_I(struct ib_uverbs_ex_modify_cq), + UAPI_DEF_METHOD_NEEDS_FN(create_cq))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_DEVICE, + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_GET_CONTEXT, + ib_uverbs_get_context, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_get_context, + struct ib_uverbs_get_context_resp)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_QUERY_DEVICE, + ib_uverbs_query_device, + UAPI_DEF_WRITE_IO(struct ib_uverbs_query_device, + struct ib_uverbs_query_device_resp)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_QUERY_PORT, + ib_uverbs_query_port, + UAPI_DEF_WRITE_IO(struct ib_uverbs_query_port, + struct ib_uverbs_query_port_resp), + UAPI_DEF_METHOD_NEEDS_FN(query_port)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_QUERY_DEVICE, + ib_uverbs_ex_query_device, + UAPI_DEF_WRITE_IO_EX( + struct ib_uverbs_ex_query_device, + reserved, + struct ib_uverbs_ex_query_device_resp, + response_length), + UAPI_DEF_METHOD_NEEDS_FN(query_device)), + UAPI_DEF_OBJ_NEEDS_FN(alloc_ucontext), + UAPI_DEF_OBJ_NEEDS_FN(dealloc_ucontext)), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_FLOW, + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_CREATE_FLOW, + ib_uverbs_ex_create_flow, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_create_flow, + flow_attr, + struct ib_uverbs_create_flow_resp, + flow_handle), + UAPI_DEF_METHOD_NEEDS_FN(create_flow)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_DESTROY_FLOW, + ib_uverbs_ex_destroy_flow, + UAPI_DEF_WRITE_I(struct ib_uverbs_destroy_flow), + UAPI_DEF_METHOD_NEEDS_FN(destroy_flow))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_MR, + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_DEREG_MR, + ib_uverbs_dereg_mr, + UAPI_DEF_WRITE_I(struct ib_uverbs_dereg_mr), + UAPI_DEF_METHOD_NEEDS_FN(dereg_mr)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_REG_MR, + ib_uverbs_reg_mr, + UAPI_DEF_WRITE_UDATA_IO(struct ib_uverbs_reg_mr, + struct ib_uverbs_reg_mr_resp), + UAPI_DEF_METHOD_NEEDS_FN(reg_user_mr)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_REREG_MR, + ib_uverbs_rereg_mr, + UAPI_DEF_WRITE_UDATA_IO(struct ib_uverbs_rereg_mr, + struct ib_uverbs_rereg_mr_resp), + UAPI_DEF_METHOD_NEEDS_FN(rereg_user_mr))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_MW, + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_ALLOC_MW, + ib_uverbs_alloc_mw, + UAPI_DEF_WRITE_UDATA_IO(struct ib_uverbs_alloc_mw, + struct ib_uverbs_alloc_mw_resp), + UAPI_DEF_METHOD_NEEDS_FN(alloc_mw)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DEALLOC_MW, + ib_uverbs_dealloc_mw, + UAPI_DEF_WRITE_I(struct ib_uverbs_dealloc_mw), + UAPI_DEF_METHOD_NEEDS_FN(dealloc_mw))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_PD, + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_ALLOC_PD, + ib_uverbs_alloc_pd, + UAPI_DEF_WRITE_UDATA_IO(struct ib_uverbs_alloc_pd, + struct ib_uverbs_alloc_pd_resp), + UAPI_DEF_METHOD_NEEDS_FN(alloc_pd)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DEALLOC_PD, + ib_uverbs_dealloc_pd, + UAPI_DEF_WRITE_I(struct ib_uverbs_dealloc_pd), + UAPI_DEF_METHOD_NEEDS_FN(dealloc_pd))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_QP, + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_ATTACH_MCAST, + ib_uverbs_attach_mcast, + UAPI_DEF_WRITE_I(struct ib_uverbs_attach_mcast), + UAPI_DEF_METHOD_NEEDS_FN(attach_mcast), + UAPI_DEF_METHOD_NEEDS_FN(detach_mcast)), + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_CREATE_QP, + ib_uverbs_create_qp, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_create_qp, + struct ib_uverbs_create_qp_resp), + UAPI_DEF_METHOD_NEEDS_FN(create_qp)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DESTROY_QP, + ib_uverbs_destroy_qp, + UAPI_DEF_WRITE_IO(struct ib_uverbs_destroy_qp, + struct ib_uverbs_destroy_qp_resp), + UAPI_DEF_METHOD_NEEDS_FN(destroy_qp)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DETACH_MCAST, + ib_uverbs_detach_mcast, + UAPI_DEF_WRITE_I(struct ib_uverbs_detach_mcast), + UAPI_DEF_METHOD_NEEDS_FN(detach_mcast)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_MODIFY_QP, + ib_uverbs_modify_qp, + UAPI_DEF_WRITE_I(struct ib_uverbs_modify_qp), + UAPI_DEF_METHOD_NEEDS_FN(modify_qp)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_POST_RECV, + ib_uverbs_post_recv, + UAPI_DEF_WRITE_IO(struct ib_uverbs_post_recv, + struct ib_uverbs_post_recv_resp), + UAPI_DEF_METHOD_NEEDS_FN(post_recv)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_POST_SEND, + ib_uverbs_post_send, + UAPI_DEF_WRITE_IO(struct ib_uverbs_post_send, + struct ib_uverbs_post_send_resp), + UAPI_DEF_METHOD_NEEDS_FN(post_send)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_QUERY_QP, + ib_uverbs_query_qp, + UAPI_DEF_WRITE_IO(struct ib_uverbs_query_qp, + struct ib_uverbs_query_qp_resp), + UAPI_DEF_METHOD_NEEDS_FN(query_qp)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_CREATE_QP, + ib_uverbs_ex_create_qp, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_ex_create_qp, + comp_mask, + struct ib_uverbs_ex_create_qp_resp, + response_length), + UAPI_DEF_METHOD_NEEDS_FN(create_qp)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_MODIFY_QP, + ib_uverbs_ex_modify_qp, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_ex_modify_qp, + base, + struct ib_uverbs_ex_modify_qp_resp, + response_length), + UAPI_DEF_METHOD_NEEDS_FN(modify_qp))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_RWQ_IND_TBL, + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_CREATE_RWQ_IND_TBL, + ib_uverbs_ex_create_rwq_ind_table, + UAPI_DEF_WRITE_IO_EX( + struct ib_uverbs_ex_create_rwq_ind_table, + log_ind_tbl_size, + struct ib_uverbs_ex_create_rwq_ind_table_resp, + ind_tbl_num), + UAPI_DEF_METHOD_NEEDS_FN(create_rwq_ind_table)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_DESTROY_RWQ_IND_TBL, + ib_uverbs_ex_destroy_rwq_ind_table, + UAPI_DEF_WRITE_I( + struct ib_uverbs_ex_destroy_rwq_ind_table), + UAPI_DEF_METHOD_NEEDS_FN(destroy_rwq_ind_table))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_WQ, + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_CREATE_WQ, + ib_uverbs_ex_create_wq, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_ex_create_wq, + max_sge, + struct ib_uverbs_ex_create_wq_resp, + wqn), + UAPI_DEF_METHOD_NEEDS_FN(create_wq)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_DESTROY_WQ, + ib_uverbs_ex_destroy_wq, + UAPI_DEF_WRITE_IO_EX(struct ib_uverbs_ex_destroy_wq, + wq_handle, + struct ib_uverbs_ex_destroy_wq_resp, + reserved), + UAPI_DEF_METHOD_NEEDS_FN(destroy_wq)), + DECLARE_UVERBS_WRITE_EX( + IB_USER_VERBS_EX_CMD_MODIFY_WQ, + ib_uverbs_ex_modify_wq, + UAPI_DEF_WRITE_I_EX(struct ib_uverbs_ex_modify_wq, + curr_wq_state), + UAPI_DEF_METHOD_NEEDS_FN(modify_wq))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_SRQ, + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_CREATE_SRQ, + ib_uverbs_create_srq, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_create_srq, + struct ib_uverbs_create_srq_resp), + UAPI_DEF_METHOD_NEEDS_FN(create_srq)), + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_CREATE_XSRQ, + ib_uverbs_create_xsrq, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_create_xsrq, + struct ib_uverbs_create_srq_resp), + UAPI_DEF_METHOD_NEEDS_FN(create_srq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_DESTROY_SRQ, + ib_uverbs_destroy_srq, + UAPI_DEF_WRITE_IO(struct ib_uverbs_destroy_srq, + struct ib_uverbs_destroy_srq_resp), + UAPI_DEF_METHOD_NEEDS_FN(destroy_srq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_MODIFY_SRQ, + ib_uverbs_modify_srq, + UAPI_DEF_WRITE_UDATA_I(struct ib_uverbs_modify_srq), + UAPI_DEF_METHOD_NEEDS_FN(modify_srq)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_POST_SRQ_RECV, + ib_uverbs_post_srq_recv, + UAPI_DEF_WRITE_IO(struct ib_uverbs_post_srq_recv, + struct ib_uverbs_post_srq_recv_resp), + UAPI_DEF_METHOD_NEEDS_FN(post_srq_recv)), + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_QUERY_SRQ, + ib_uverbs_query_srq, + UAPI_DEF_WRITE_IO(struct ib_uverbs_query_srq, + struct ib_uverbs_query_srq_resp), + UAPI_DEF_METHOD_NEEDS_FN(query_srq))), + + DECLARE_UVERBS_OBJECT( + UVERBS_OBJECT_XRCD, + DECLARE_UVERBS_WRITE( + IB_USER_VERBS_CMD_CLOSE_XRCD, + ib_uverbs_close_xrcd, + UAPI_DEF_WRITE_I(struct ib_uverbs_close_xrcd), + UAPI_DEF_METHOD_NEEDS_FN(dealloc_xrcd)), + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_OPEN_QP, + ib_uverbs_open_qp, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_open_qp, + struct ib_uverbs_create_qp_resp)), + DECLARE_UVERBS_WRITE(IB_USER_VERBS_CMD_OPEN_XRCD, + ib_uverbs_open_xrcd, + UAPI_DEF_WRITE_UDATA_IO( + struct ib_uverbs_open_xrcd, + struct ib_uverbs_open_xrcd_resp), + UAPI_DEF_METHOD_NEEDS_FN(alloc_xrcd))), + + {}, +}; diff --git a/drivers/infiniband/core/uverbs_ioctl.c b/drivers/infiniband/core/uverbs_ioctl.c index b0e493e8d860..e643a43dce8d 100644 --- a/drivers/infiniband/core/uverbs_ioctl.c +++ b/drivers/infiniband/core/uverbs_ioctl.c @@ -404,8 +404,7 @@ static int uverbs_set_attr(struct bundle_priv *pbundle, static int ib_uverbs_run_method(struct bundle_priv *pbundle, unsigned int num_attrs) { - int (*handler)(struct ib_uverbs_file *ufile, - struct uverbs_attr_bundle *ctx); + int (*handler)(struct uverbs_attr_bundle *attrs); size_t uattrs_size = array_size(sizeof(*pbundle->uattrs), num_attrs); unsigned int destroy_bkey = pbundle->method_elm->destroy_bkey; unsigned int i; @@ -436,6 +435,9 @@ static int ib_uverbs_run_method(struct bundle_priv *pbundle, pbundle->method_elm->key_bitmap_len))) return -EINVAL; + if (pbundle->method_elm->has_udata) + create_udata(&pbundle->bundle, &pbundle->bundle.driver_udata); + if (destroy_bkey != UVERBS_API_ATTR_BKEY_LEN) { struct uverbs_obj_attr *destroy_attr = &pbundle->bundle.attrs[destroy_bkey].obj_attr; @@ -445,10 +447,10 @@ static int ib_uverbs_run_method(struct bundle_priv *pbundle, return ret; __clear_bit(destroy_bkey, pbundle->uobj_finalize); - ret = handler(pbundle->bundle.ufile, &pbundle->bundle); + ret = handler(&pbundle->bundle); uobj_put_destroy(destroy_attr->uobject); } else { - ret = handler(pbundle->bundle.ufile, &pbundle->bundle); + ret = handler(&pbundle->bundle); } /* diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 6d373f5515b7..96a5f89bbb75 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -74,64 +74,6 @@ static dev_t dynamic_uverbs_dev; static struct class *uverbs_class; static DEFINE_IDA(uverbs_ida); - -static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file *file, - const char __user *buf, int in_len, - int out_len) = { - [IB_USER_VERBS_CMD_GET_CONTEXT] = ib_uverbs_get_context, - [IB_USER_VERBS_CMD_QUERY_DEVICE] = ib_uverbs_query_device, - [IB_USER_VERBS_CMD_QUERY_PORT] = ib_uverbs_query_port, - [IB_USER_VERBS_CMD_ALLOC_PD] = ib_uverbs_alloc_pd, - [IB_USER_VERBS_CMD_DEALLOC_PD] = ib_uverbs_dealloc_pd, - [IB_USER_VERBS_CMD_REG_MR] = ib_uverbs_reg_mr, - [IB_USER_VERBS_CMD_REREG_MR] = ib_uverbs_rereg_mr, - [IB_USER_VERBS_CMD_DEREG_MR] = ib_uverbs_dereg_mr, - [IB_USER_VERBS_CMD_ALLOC_MW] = ib_uverbs_alloc_mw, - [IB_USER_VERBS_CMD_DEALLOC_MW] = ib_uverbs_dealloc_mw, - [IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL] = ib_uverbs_create_comp_channel, - [IB_USER_VERBS_CMD_CREATE_CQ] = ib_uverbs_create_cq, - [IB_USER_VERBS_CMD_RESIZE_CQ] = ib_uverbs_resize_cq, - [IB_USER_VERBS_CMD_POLL_CQ] = ib_uverbs_poll_cq, - [IB_USER_VERBS_CMD_REQ_NOTIFY_CQ] = ib_uverbs_req_notify_cq, - [IB_USER_VERBS_CMD_DESTROY_CQ] = ib_uverbs_destroy_cq, - [IB_USER_VERBS_CMD_CREATE_QP] = ib_uverbs_create_qp, - [IB_USER_VERBS_CMD_QUERY_QP] = ib_uverbs_query_qp, - [IB_USER_VERBS_CMD_MODIFY_QP] = ib_uverbs_modify_qp, - [IB_USER_VERBS_CMD_DESTROY_QP] = ib_uverbs_destroy_qp, - [IB_USER_VERBS_CMD_POST_SEND] = ib_uverbs_post_send, - [IB_USER_VERBS_CMD_POST_RECV] = ib_uverbs_post_recv, - [IB_USER_VERBS_CMD_POST_SRQ_RECV] = ib_uverbs_post_srq_recv, - [IB_USER_VERBS_CMD_CREATE_AH] = ib_uverbs_create_ah, - [IB_USER_VERBS_CMD_DESTROY_AH] = ib_uverbs_destroy_ah, - [IB_USER_VERBS_CMD_ATTACH_MCAST] = ib_uverbs_attach_mcast, - [IB_USER_VERBS_CMD_DETACH_MCAST] = ib_uverbs_detach_mcast, - [IB_USER_VERBS_CMD_CREATE_SRQ] = ib_uverbs_create_srq, - [IB_USER_VERBS_CMD_MODIFY_SRQ] = ib_uverbs_modify_srq, - [IB_USER_VERBS_CMD_QUERY_SRQ] = ib_uverbs_query_srq, - [IB_USER_VERBS_CMD_DESTROY_SRQ] = ib_uverbs_destroy_srq, - [IB_USER_VERBS_CMD_OPEN_XRCD] = ib_uverbs_open_xrcd, - [IB_USER_VERBS_CMD_CLOSE_XRCD] = ib_uverbs_close_xrcd, - [IB_USER_VERBS_CMD_CREATE_XSRQ] = ib_uverbs_create_xsrq, - [IB_USER_VERBS_CMD_OPEN_QP] = ib_uverbs_open_qp, -}; - -static int (*uverbs_ex_cmd_table[])(struct ib_uverbs_file *file, - struct ib_udata *ucore, - struct ib_udata *uhw) = { - [IB_USER_VERBS_EX_CMD_CREATE_FLOW] = ib_uverbs_ex_create_flow, - [IB_USER_VERBS_EX_CMD_DESTROY_FLOW] = ib_uverbs_ex_destroy_flow, - [IB_USER_VERBS_EX_CMD_QUERY_DEVICE] = ib_uverbs_ex_query_device, - [IB_USER_VERBS_EX_CMD_CREATE_CQ] = ib_uverbs_ex_create_cq, - [IB_USER_VERBS_EX_CMD_CREATE_QP] = ib_uverbs_ex_create_qp, - [IB_USER_VERBS_EX_CMD_CREATE_WQ] = ib_uverbs_ex_create_wq, - [IB_USER_VERBS_EX_CMD_MODIFY_WQ] = ib_uverbs_ex_modify_wq, - [IB_USER_VERBS_EX_CMD_DESTROY_WQ] = ib_uverbs_ex_destroy_wq, - [IB_USER_VERBS_EX_CMD_CREATE_RWQ_IND_TBL] = ib_uverbs_ex_create_rwq_ind_table, - [IB_USER_VERBS_EX_CMD_DESTROY_RWQ_IND_TBL] = ib_uverbs_ex_destroy_rwq_ind_table, - [IB_USER_VERBS_EX_CMD_MODIFY_QP] = ib_uverbs_ex_modify_qp, - [IB_USER_VERBS_EX_CMD_MODIFY_CQ] = ib_uverbs_ex_modify_cq, -}; - static void ib_uverbs_add_one(struct ib_device *device); static void ib_uverbs_remove_one(struct ib_device *device, void *client_data); @@ -139,7 +81,7 @@ static void ib_uverbs_remove_one(struct ib_device *device, void *client_data); * Must be called with the ufile->device->disassociate_srcu held, and the lock * must be held until use of the ucontext is finished. */ -struct ib_ucontext *ib_uverbs_get_ucontext(struct ib_uverbs_file *ufile) +struct ib_ucontext *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile) { /* * We do not hold the hw_destroy_rwsem lock for this flow, instead @@ -157,7 +99,7 @@ struct ib_ucontext *ib_uverbs_get_ucontext(struct ib_uverbs_file *ufile) return ucontext; } -EXPORT_SYMBOL(ib_uverbs_get_ucontext); +EXPORT_SYMBOL(ib_uverbs_get_ucontext_file); int uverbs_dealloc_mw(struct ib_mw *mw) { @@ -646,51 +588,19 @@ err_put_refs: return filp; } -static bool verify_command_mask(struct ib_uverbs_file *ufile, u32 command, - bool extended) -{ - if (!extended) - return ufile->uverbs_cmd_mask & BIT_ULL(command); - - return ufile->uverbs_ex_cmd_mask & BIT_ULL(command); -} - -static bool verify_command_idx(u32 command, bool extended) -{ - if (extended) - return command < ARRAY_SIZE(uverbs_ex_cmd_table) && - uverbs_ex_cmd_table[command]; - - return command < ARRAY_SIZE(uverbs_cmd_table) && - uverbs_cmd_table[command]; -} - -static ssize_t process_hdr(struct ib_uverbs_cmd_hdr *hdr, - u32 *command, bool *extended) -{ - if (hdr->command & ~(u32)(IB_USER_VERBS_CMD_FLAG_EXTENDED | - IB_USER_VERBS_CMD_COMMAND_MASK)) - return -EINVAL; - - *command = hdr->command & IB_USER_VERBS_CMD_COMMAND_MASK; - *extended = hdr->command & IB_USER_VERBS_CMD_FLAG_EXTENDED; - - if (!verify_command_idx(*command, *extended)) - return -EOPNOTSUPP; - - return 0; -} - static ssize_t verify_hdr(struct ib_uverbs_cmd_hdr *hdr, - struct ib_uverbs_ex_cmd_hdr *ex_hdr, - size_t count, bool extended) + struct ib_uverbs_ex_cmd_hdr *ex_hdr, size_t count, + const struct uverbs_api_write_method *method_elm) { - if (extended) { + if (method_elm->is_ex) { count -= sizeof(*hdr) + sizeof(*ex_hdr); if ((hdr->in_words + ex_hdr->provider_in_words) * 8 != count) return -EINVAL; + if (hdr->in_words * 8 < method_elm->req_size) + return -ENOSPC; + if (ex_hdr->cmd_hdr_reserved) return -EINVAL; @@ -698,6 +608,9 @@ static ssize_t verify_hdr(struct ib_uverbs_cmd_hdr *hdr, if (!hdr->out_words && !ex_hdr->provider_out_words) return -EINVAL; + if (hdr->out_words * 8 < method_elm->resp_size) + return -ENOSPC; + if (!access_ok(VERIFY_WRITE, u64_to_user_ptr(ex_hdr->response), (hdr->out_words + ex_hdr->provider_out_words) * 8)) @@ -714,6 +627,24 @@ static ssize_t verify_hdr(struct ib_uverbs_cmd_hdr *hdr, if (hdr->in_words * 4 != count) return -EINVAL; + if (count < method_elm->req_size + sizeof(hdr)) { + /* + * rdma-core v18 and v19 have a bug where they send DESTROY_CQ + * with a 16 byte write instead of 24. Old kernels didn't + * check the size so they allowed this. Now that the size is + * checked provide a compatibility work around to not break + * those userspaces. + */ + if (hdr->command == IB_USER_VERBS_CMD_DESTROY_CQ && + count == 16) { + hdr->in_words = 6; + return 0; + } + return -ENOSPC; + } + if (hdr->out_words * 4 < method_elm->resp_size) + return -ENOSPC; + return 0; } @@ -721,11 +652,12 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, size_t count, loff_t *pos) { struct ib_uverbs_file *file = filp->private_data; + const struct uverbs_api_write_method *method_elm; + struct uverbs_api *uapi = file->device->uapi; struct ib_uverbs_ex_cmd_hdr ex_hdr; struct ib_uverbs_cmd_hdr hdr; - bool extended; + struct uverbs_attr_bundle bundle; int srcu_key; - u32 command; ssize_t ret; if (!ib_safe_file_access(filp)) { @@ -740,57 +672,92 @@ static ssize_t ib_uverbs_write(struct file *filp, const char __user *buf, if (copy_from_user(&hdr, buf, sizeof(hdr))) return -EFAULT; - ret = process_hdr(&hdr, &command, &extended); - if (ret) - return ret; + method_elm = uapi_get_method(uapi, hdr.command); + if (IS_ERR(method_elm)) + return PTR_ERR(method_elm); - if (extended) { + if (method_elm->is_ex) { if (count < (sizeof(hdr) + sizeof(ex_hdr))) return -EINVAL; if (copy_from_user(&ex_hdr, buf + sizeof(hdr), sizeof(ex_hdr))) return -EFAULT; } - ret = verify_hdr(&hdr, &ex_hdr, count, extended); + ret = verify_hdr(&hdr, &ex_hdr, count, method_elm); if (ret) return ret; srcu_key = srcu_read_lock(&file->device->disassociate_srcu); - if (!verify_command_mask(file, command, extended)) { - ret = -EOPNOTSUPP; - goto out; - } - buf += sizeof(hdr); - if (!extended) { - ret = uverbs_cmd_table[command](file, buf, - hdr.in_words * 4, - hdr.out_words * 4); - } else { - struct ib_udata ucore; - struct ib_udata uhw; + bundle.ufile = file; + if (!method_elm->is_ex) { + size_t in_len = hdr.in_words * 4 - sizeof(hdr); + size_t out_len = hdr.out_words * 4; + u64 response = 0; + + if (method_elm->has_udata) { + bundle.driver_udata.inlen = + in_len - method_elm->req_size; + in_len = method_elm->req_size; + if (bundle.driver_udata.inlen) + bundle.driver_udata.inbuf = buf + in_len; + else + bundle.driver_udata.inbuf = NULL; + } else { + memset(&bundle.driver_udata, 0, + sizeof(bundle.driver_udata)); + } + + if (method_elm->has_resp) { + /* + * The macros check that if has_resp is set + * then the command request structure starts + * with a '__aligned u64 response' member. + */ + ret = get_user(response, (const u64 *)buf); + if (ret) + goto out_unlock; + + if (method_elm->has_udata) { + bundle.driver_udata.outlen = + out_len - method_elm->resp_size; + out_len = method_elm->resp_size; + if (bundle.driver_udata.outlen) + bundle.driver_udata.outbuf = + u64_to_user_ptr(response + + out_len); + else + bundle.driver_udata.outbuf = NULL; + } + } else { + bundle.driver_udata.outlen = 0; + bundle.driver_udata.outbuf = NULL; + } + ib_uverbs_init_udata_buf_or_null( + &bundle.ucore, buf, u64_to_user_ptr(response), + in_len, out_len); + } else { buf += sizeof(ex_hdr); - ib_uverbs_init_udata_buf_or_null(&ucore, buf, + ib_uverbs_init_udata_buf_or_null(&bundle.ucore, buf, u64_to_user_ptr(ex_hdr.response), hdr.in_words * 8, hdr.out_words * 8); - ib_uverbs_init_udata_buf_or_null(&uhw, - buf + ucore.inlen, - u64_to_user_ptr(ex_hdr.response) + ucore.outlen, - ex_hdr.provider_in_words * 8, - ex_hdr.provider_out_words * 8); + ib_uverbs_init_udata_buf_or_null( + &bundle.driver_udata, buf + bundle.ucore.inlen, + u64_to_user_ptr(ex_hdr.response) + bundle.ucore.outlen, + ex_hdr.provider_in_words * 8, + ex_hdr.provider_out_words * 8); - ret = uverbs_ex_cmd_table[command](file, &ucore, &uhw); - ret = (ret) ? : count; } -out: + ret = method_elm->handler(&bundle); +out_unlock: srcu_read_unlock(&file->device->disassociate_srcu, srcu_key); - return ret; + return (ret) ? : count; } static int ib_uverbs_mmap(struct file *filp, struct vm_area_struct *vma) @@ -801,7 +768,7 @@ static int ib_uverbs_mmap(struct file *filp, struct vm_area_struct *vma) int srcu_key; srcu_key = srcu_read_lock(&file->device->disassociate_srcu); - ucontext = ib_uverbs_get_ucontext(file); + ucontext = ib_uverbs_get_ucontext_file(file); if (IS_ERR(ucontext)) { ret = PTR_ERR(ucontext); goto out; @@ -1102,9 +1069,6 @@ static int ib_uverbs_open(struct inode *inode, struct file *filp) mutex_unlock(&dev->lists_mutex); srcu_read_unlock(&dev->disassociate_srcu, srcu_key); - file->uverbs_cmd_mask = ib_dev->uverbs_cmd_mask; - file->uverbs_ex_cmd_mask = ib_dev->uverbs_ex_cmd_mask; - setup_ufile_idr_uobject(file); return nonseekable_open(inode, filp); @@ -1224,7 +1188,7 @@ static int ib_uverbs_create_uapi(struct ib_device *device, { struct uverbs_api *uapi; - uapi = uverbs_alloc_api(device->driver_specs, device->driver_id); + uapi = uverbs_alloc_api(device); if (IS_ERR(uapi)) return PTR_ERR(uapi); diff --git a/drivers/infiniband/core/uverbs_std_types.c b/drivers/infiniband/core/uverbs_std_types.c index 203cc96ac6f5..063aff9e7a04 100644 --- a/drivers/infiniband/core/uverbs_std_types.c +++ b/drivers/infiniband/core/uverbs_std_types.c @@ -210,8 +210,7 @@ static int uverbs_hot_unplug_completion_event_file(struct ib_uobject *uobj, return 0; }; -int uverbs_destroy_def_handler(struct ib_uverbs_file *file, - struct uverbs_attr_bundle *attrs) +int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs) { return 0; } @@ -262,25 +261,28 @@ DECLARE_UVERBS_NAMED_OBJECT(UVERBS_OBJECT_PD, DECLARE_UVERBS_GLOBAL_METHODS(UVERBS_OBJECT_DEVICE); -DECLARE_UVERBS_OBJECT_TREE(uverbs_default_objects, - &UVERBS_OBJECT(UVERBS_OBJECT_DEVICE), - &UVERBS_OBJECT(UVERBS_OBJECT_PD), - &UVERBS_OBJECT(UVERBS_OBJECT_MR), - &UVERBS_OBJECT(UVERBS_OBJECT_COMP_CHANNEL), - &UVERBS_OBJECT(UVERBS_OBJECT_CQ), - &UVERBS_OBJECT(UVERBS_OBJECT_QP), - &UVERBS_OBJECT(UVERBS_OBJECT_AH), - &UVERBS_OBJECT(UVERBS_OBJECT_MW), - &UVERBS_OBJECT(UVERBS_OBJECT_SRQ), - &UVERBS_OBJECT(UVERBS_OBJECT_FLOW), - &UVERBS_OBJECT(UVERBS_OBJECT_WQ), - &UVERBS_OBJECT(UVERBS_OBJECT_RWQ_IND_TBL), - &UVERBS_OBJECT(UVERBS_OBJECT_XRCD), - &UVERBS_OBJECT(UVERBS_OBJECT_FLOW_ACTION), - &UVERBS_OBJECT(UVERBS_OBJECT_DM), - &UVERBS_OBJECT(UVERBS_OBJECT_COUNTERS)); - -const struct uverbs_object_tree_def *uverbs_default_get_objects(void) -{ - return &uverbs_default_objects; -} +const struct uapi_definition uverbs_def_obj_intf[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_DEVICE), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_PD, + UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_COMP_CHANNEL, + UAPI_DEF_OBJ_NEEDS_FN(dealloc_pd)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_QP, + UAPI_DEF_OBJ_NEEDS_FN(destroy_qp)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_AH, + UAPI_DEF_OBJ_NEEDS_FN(destroy_ah)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MW, + UAPI_DEF_OBJ_NEEDS_FN(dealloc_mw)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_SRQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_srq)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_FLOW, + UAPI_DEF_OBJ_NEEDS_FN(destroy_flow)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_WQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_wq)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + UVERBS_OBJECT_RWQ_IND_TBL, + UAPI_DEF_OBJ_NEEDS_FN(destroy_rwq_ind_table)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_XRCD, + UAPI_DEF_OBJ_NEEDS_FN(dealloc_xrcd)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_counters.c b/drivers/infiniband/core/uverbs_std_types_counters.c index a0ffdcf9a51c..8835bad5c6dd 100644 --- a/drivers/infiniband/core/uverbs_std_types_counters.c +++ b/drivers/infiniband/core/uverbs_std_types_counters.c @@ -48,7 +48,7 @@ static int uverbs_free_counters(struct ib_uobject *uobject, } static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_CREATE)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject( attrs, UVERBS_ATTR_CREATE_COUNTERS_HANDLE); @@ -82,7 +82,7 @@ err_create_counters: } static int UVERBS_HANDLER(UVERBS_METHOD_COUNTERS_READ)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_counters_read_attr read_attr = {}; const struct uverbs_attr *uattr; @@ -149,3 +149,9 @@ DECLARE_UVERBS_NAMED_OBJECT(UVERBS_OBJECT_COUNTERS, &UVERBS_METHOD(UVERBS_METHOD_COUNTERS_CREATE), &UVERBS_METHOD(UVERBS_METHOD_COUNTERS_DESTROY), &UVERBS_METHOD(UVERBS_METHOD_COUNTERS_READ)); + +const struct uapi_definition uverbs_def_obj_counters[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_COUNTERS, + UAPI_DEF_OBJ_NEEDS_FN(destroy_counters)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_cq.c b/drivers/infiniband/core/uverbs_std_types_cq.c index 5b5f2052cd52..859518eab583 100644 --- a/drivers/infiniband/core/uverbs_std_types_cq.c +++ b/drivers/infiniband/core/uverbs_std_types_cq.c @@ -58,13 +58,12 @@ static int uverbs_free_cq(struct ib_uobject *uobject, } static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_ucq_object *obj = container_of( uverbs_attr_get_uobject(attrs, UVERBS_ATTR_CREATE_CQ_HANDLE), typeof(*obj), uobject); struct ib_device *ib_dev = obj->uobject.context->device; - struct ib_udata uhw; int ret; u64 user_handle; struct ib_cq_init_attr attr = {}; @@ -101,7 +100,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( uverbs_uobject_get(ev_file_uobj); } - if (attr.comp_vector >= file->device->num_comp_vectors) { + if (attr.comp_vector >= attrs->ufile->device->num_comp_vectors) { ret = -EINVAL; goto err_event_file; } @@ -111,10 +110,8 @@ static int UVERBS_HANDLER(UVERBS_METHOD_CQ_CREATE)( INIT_LIST_HEAD(&obj->comp_list); INIT_LIST_HEAD(&obj->async_list); - /* Temporary, only until drivers get the new uverbs_attr_bundle */ - create_udata(attrs, &uhw); - - cq = ib_dev->create_cq(ib_dev, &attr, obj->uobject.context, &uhw); + cq = ib_dev->create_cq(ib_dev, &attr, obj->uobject.context, + &attrs->driver_udata); if (IS_ERR(cq)) { ret = PTR_ERR(cq); goto err_event_file; @@ -173,7 +170,7 @@ DECLARE_UVERBS_NAMED_METHOD( UVERBS_ATTR_UHW()); static int UVERBS_HANDLER(UVERBS_METHOD_CQ_DESTROY)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, UVERBS_ATTR_DESTROY_CQ_HANDLE); @@ -207,3 +204,9 @@ DECLARE_UVERBS_NAMED_OBJECT( &UVERBS_METHOD(UVERBS_METHOD_CQ_DESTROY) #endif ); + +const struct uapi_definition uverbs_def_obj_cq[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_CQ, + UAPI_DEF_OBJ_NEEDS_FN(destroy_cq)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_dm.c b/drivers/infiniband/core/uverbs_std_types_dm.c index edc3ff7733d4..658261b8f08e 100644 --- a/drivers/infiniband/core/uverbs_std_types_dm.c +++ b/drivers/infiniband/core/uverbs_std_types_dm.c @@ -46,9 +46,8 @@ static int uverbs_free_dm(struct ib_uobject *uobject, return dm->device->dealloc_dm(dm); } -static int -UVERBS_HANDLER(UVERBS_METHOD_DM_ALLOC)(struct ib_uverbs_file *file, - struct uverbs_attr_bundle *attrs) +static int UVERBS_HANDLER(UVERBS_METHOD_DM_ALLOC)( + struct uverbs_attr_bundle *attrs) { struct ib_dm_alloc_attr attr = {}; struct ib_uobject *uobj = @@ -109,3 +108,9 @@ DECLARE_UVERBS_NAMED_OBJECT(UVERBS_OBJECT_DM, UVERBS_TYPE_ALLOC_IDR(uverbs_free_dm), &UVERBS_METHOD(UVERBS_METHOD_DM_ALLOC), &UVERBS_METHOD(UVERBS_METHOD_DM_FREE)); + +const struct uapi_definition uverbs_def_obj_dm[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_DM, + UAPI_DEF_OBJ_NEEDS_FN(dealloc_dm)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_flow_action.c b/drivers/infiniband/core/uverbs_std_types_flow_action.c index cb9486ad5c67..e4d01fb5335d 100644 --- a/drivers/infiniband/core/uverbs_std_types_flow_action.c +++ b/drivers/infiniband/core/uverbs_std_types_flow_action.c @@ -223,7 +223,6 @@ struct ib_flow_action_esp_attr { #define ESP_LAST_SUPPORTED_FLAG IB_UVERBS_FLOW_ACTION_ESP_FLAGS_ESN_NEW_WINDOW static int parse_flow_action_esp(struct ib_device *ib_dev, - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs, struct ib_flow_action_esp_attr *esp_attr, bool is_modify) @@ -305,7 +304,7 @@ static int parse_flow_action_esp(struct ib_device *ib_dev, } static int UVERBS_HANDLER(UVERBS_METHOD_FLOW_ACTION_ESP_CREATE)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject( attrs, UVERBS_ATTR_CREATE_FLOW_ACTION_ESP_HANDLE); @@ -317,7 +316,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_FLOW_ACTION_ESP_CREATE)( if (!ib_dev->create_flow_action_esp) return -EOPNOTSUPP; - ret = parse_flow_action_esp(ib_dev, file, attrs, &esp_attr, false); + ret = parse_flow_action_esp(ib_dev, attrs, &esp_attr, false); if (ret) return ret; @@ -333,7 +332,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_FLOW_ACTION_ESP_CREATE)( } static int UVERBS_HANDLER(UVERBS_METHOD_FLOW_ACTION_ESP_MODIFY)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject( attrs, UVERBS_ATTR_MODIFY_FLOW_ACTION_ESP_HANDLE); @@ -344,8 +343,7 @@ static int UVERBS_HANDLER(UVERBS_METHOD_FLOW_ACTION_ESP_MODIFY)( if (!action->device->modify_flow_action_esp) return -EOPNOTSUPP; - ret = parse_flow_action_esp(action->device, file, attrs, &esp_attr, - true); + ret = parse_flow_action_esp(action->device, attrs, &esp_attr, true); if (ret) return ret; @@ -438,3 +436,10 @@ DECLARE_UVERBS_NAMED_OBJECT( &UVERBS_METHOD(UVERBS_METHOD_FLOW_ACTION_ESP_CREATE), &UVERBS_METHOD(UVERBS_METHOD_FLOW_ACTION_DESTROY), &UVERBS_METHOD(UVERBS_METHOD_FLOW_ACTION_ESP_MODIFY)); + +const struct uapi_definition uverbs_def_obj_flow_action[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + UVERBS_OBJECT_FLOW_ACTION, + UAPI_DEF_OBJ_NEEDS_FN(destroy_flow_action)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_std_types_mr.c b/drivers/infiniband/core/uverbs_std_types_mr.c index cf02e774303e..70ea48cfc047 100644 --- a/drivers/infiniband/core/uverbs_std_types_mr.c +++ b/drivers/infiniband/core/uverbs_std_types_mr.c @@ -40,7 +40,7 @@ static int uverbs_free_mr(struct ib_uobject *uobject, } static int UVERBS_HANDLER(UVERBS_METHOD_DM_MR_REG)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_dm_mr_attr attr = {}; struct ib_uobject *uobj = @@ -147,3 +147,9 @@ DECLARE_UVERBS_NAMED_OBJECT( UVERBS_OBJECT_MR, UVERBS_TYPE_ALLOC_IDR(uverbs_free_mr), &UVERBS_METHOD(UVERBS_METHOD_DM_MR_REG)); + +const struct uapi_definition uverbs_def_obj_mr[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(UVERBS_OBJECT_MR, + UAPI_DEF_OBJ_NEEDS_FN(dereg_mr)), + {} +}; diff --git a/drivers/infiniband/core/uverbs_uapi.c b/drivers/infiniband/core/uverbs_uapi.c index 86f3fc5e04b4..0136c1d78a0f 100644 --- a/drivers/infiniband/core/uverbs_uapi.c +++ b/drivers/infiniband/core/uverbs_uapi.c @@ -8,6 +8,11 @@ #include "rdma_core.h" #include "uverbs.h" +static int ib_uverbs_notsupp(struct uverbs_attr_bundle *attrs) +{ + return -EOPNOTSUPP; +} + static void *uapi_add_elm(struct uverbs_api *uapi, u32 key, size_t alloc_size) { void *elm; @@ -26,6 +31,70 @@ static void *uapi_add_elm(struct uverbs_api *uapi, u32 key, size_t alloc_size) return elm; } +static void *uapi_add_get_elm(struct uverbs_api *uapi, u32 key, + size_t alloc_size, bool *exists) +{ + void *elm; + + elm = uapi_add_elm(uapi, key, alloc_size); + if (!IS_ERR(elm)) { + *exists = false; + return elm; + } + + if (elm != ERR_PTR(-EEXIST)) + return elm; + + elm = radix_tree_lookup(&uapi->radix, key); + if (WARN_ON(!elm)) + return ERR_PTR(-EINVAL); + *exists = true; + return elm; +} + +static int uapi_create_write(struct uverbs_api *uapi, + struct ib_device *ibdev, + const struct uapi_definition *def, + u32 obj_key, + u32 *cur_method_key) +{ + struct uverbs_api_write_method *method_elm; + u32 method_key = obj_key; + bool exists; + + if (def->write.is_ex) + method_key |= uapi_key_write_ex_method(def->write.command_num); + else + method_key |= uapi_key_write_method(def->write.command_num); + + method_elm = uapi_add_get_elm(uapi, method_key, sizeof(*method_elm), + &exists); + if (IS_ERR(method_elm)) + return PTR_ERR(method_elm); + + if (WARN_ON(exists && (def->write.is_ex != method_elm->is_ex))) + return -EINVAL; + + method_elm->is_ex = def->write.is_ex; + method_elm->handler = def->func_write; + if (def->write.is_ex) + method_elm->disabled = !(ibdev->uverbs_ex_cmd_mask & + BIT_ULL(def->write.command_num)); + else + method_elm->disabled = !(ibdev->uverbs_cmd_mask & + BIT_ULL(def->write.command_num)); + + if (!def->write.is_ex && def->func_write) { + method_elm->has_udata = def->write.has_udata; + method_elm->has_resp = def->write.has_resp; + method_elm->req_size = def->write.req_size; + method_elm->resp_size = def->write.resp_size; + } + + *cur_method_key = method_key; + return 0; +} + static int uapi_merge_method(struct uverbs_api *uapi, struct uverbs_api_object *obj_elm, u32 obj_key, const struct uverbs_method_def *method, @@ -34,23 +103,21 @@ static int uapi_merge_method(struct uverbs_api *uapi, u32 method_key = obj_key | uapi_key_ioctl_method(method->id); struct uverbs_api_ioctl_method *method_elm; unsigned int i; + bool exists; if (!method->attrs) return 0; - method_elm = uapi_add_elm(uapi, method_key, sizeof(*method_elm)); - if (IS_ERR(method_elm)) { - if (method_elm != ERR_PTR(-EEXIST)) - return PTR_ERR(method_elm); - + method_elm = uapi_add_get_elm(uapi, method_key, sizeof(*method_elm), + &exists); + if (IS_ERR(method_elm)) + return PTR_ERR(method_elm); + if (exists) { /* * This occurs when a driver uses ADD_UVERBS_ATTRIBUTES_SIMPLE */ if (WARN_ON(method->handler)) return -EINVAL; - method_elm = radix_tree_lookup(&uapi->radix, method_key); - if (WARN_ON(!method_elm)) - return -EINVAL; } else { WARN_ON(!method->handler); rcu_assign_pointer(method_elm->handler, method->handler); @@ -98,72 +165,182 @@ static int uapi_merge_method(struct uverbs_api *uapi, return 0; } -static int uapi_merge_tree(struct uverbs_api *uapi, - const struct uverbs_object_tree_def *tree, - bool is_driver) +static int uapi_merge_obj_tree(struct uverbs_api *uapi, + const struct uverbs_object_def *obj, + bool is_driver) { - unsigned int i, j; + struct uverbs_api_object *obj_elm; + unsigned int i; + u32 obj_key; + bool exists; int rc; - if (!tree->objects) + obj_key = uapi_key_obj(obj->id); + obj_elm = uapi_add_get_elm(uapi, obj_key, sizeof(*obj_elm), &exists); + if (IS_ERR(obj_elm)) + return PTR_ERR(obj_elm); + + if (obj->type_attrs) { + if (WARN_ON(obj_elm->type_attrs)) + return -EINVAL; + + obj_elm->id = obj->id; + obj_elm->type_attrs = obj->type_attrs; + obj_elm->type_class = obj->type_attrs->type_class; + /* + * Today drivers are only permitted to use idr_class + * types. They cannot use FD types because we currently have + * no way to revoke the fops pointer after device + * disassociation. + */ + if (WARN_ON(is_driver && + obj->type_attrs->type_class != &uverbs_idr_class)) + return -EINVAL; + } + + if (!obj->methods) return 0; - for (i = 0; i != tree->num_objects; i++) { - const struct uverbs_object_def *obj = (*tree->objects)[i]; - struct uverbs_api_object *obj_elm; - u32 obj_key; + for (i = 0; i != obj->num_methods; i++) { + const struct uverbs_method_def *method = (*obj->methods)[i]; - if (!obj) + if (!method) continue; - obj_key = uapi_key_obj(obj->id); - obj_elm = uapi_add_elm(uapi, obj_key, sizeof(*obj_elm)); - if (IS_ERR(obj_elm)) { - if (obj_elm != ERR_PTR(-EEXIST)) - return PTR_ERR(obj_elm); + rc = uapi_merge_method(uapi, obj_elm, obj_key, method, + is_driver); + if (rc) + return rc; + } - /* This occurs when a driver uses ADD_UVERBS_METHODS */ - if (WARN_ON(obj->type_attrs)) - return -EINVAL; - obj_elm = radix_tree_lookup(&uapi->radix, obj_key); - if (WARN_ON(!obj_elm)) + return 0; +} + +static int uapi_disable_elm(struct uverbs_api *uapi, + const struct uapi_definition *def, + u32 obj_key, + u32 method_key) +{ + bool exists; + + if (def->scope == UAPI_SCOPE_OBJECT) { + struct uverbs_api_object *obj_elm; + + obj_elm = uapi_add_get_elm( + uapi, obj_key, sizeof(*obj_elm), &exists); + if (IS_ERR(obj_elm)) + return PTR_ERR(obj_elm); + obj_elm->disabled = 1; + return 0; + } + + if (def->scope == UAPI_SCOPE_METHOD && + uapi_key_is_ioctl_method(method_key)) { + struct uverbs_api_ioctl_method *method_elm; + + method_elm = uapi_add_get_elm(uapi, method_key, + sizeof(*method_elm), &exists); + if (IS_ERR(method_elm)) + return PTR_ERR(method_elm); + method_elm->disabled = 1; + return 0; + } + + if (def->scope == UAPI_SCOPE_METHOD && + (uapi_key_is_write_method(method_key) || + uapi_key_is_write_ex_method(method_key))) { + struct uverbs_api_write_method *write_elm; + + write_elm = uapi_add_get_elm(uapi, method_key, + sizeof(*write_elm), &exists); + if (IS_ERR(write_elm)) + return PTR_ERR(write_elm); + write_elm->disabled = 1; + return 0; + } + + WARN_ON(true); + return -EINVAL; +} + +static int uapi_merge_def(struct uverbs_api *uapi, struct ib_device *ibdev, + const struct uapi_definition *def_list, + bool is_driver) +{ + const struct uapi_definition *def = def_list; + u32 cur_obj_key = UVERBS_API_KEY_ERR; + u32 cur_method_key = UVERBS_API_KEY_ERR; + bool exists; + int rc; + + if (!def_list) + return 0; + + for (;; def++) { + switch ((enum uapi_definition_kind)def->kind) { + case UAPI_DEF_CHAIN: + rc = uapi_merge_def(uapi, ibdev, def->chain, is_driver); + if (rc) + return rc; + continue; + + case UAPI_DEF_CHAIN_OBJ_TREE: + if (WARN_ON(def->object_start.object_id != + def->chain_obj_tree->id)) return -EINVAL; - } else { - obj_elm->type_attrs = obj->type_attrs; - if (obj->type_attrs) { - obj_elm->type_class = - obj->type_attrs->type_class; - /* - * Today drivers are only permitted to use - * idr_class types. They cannot use FD types - * because we currently have no way to revoke - * the fops pointer after device - * disassociation. - */ - if (WARN_ON(is_driver && - obj->type_attrs->type_class != - &uverbs_idr_class)) - return -EINVAL; - } - } - if (!obj->methods) + cur_obj_key = uapi_key_obj(def->object_start.object_id); + rc = uapi_merge_obj_tree(uapi, def->chain_obj_tree, + is_driver); + if (rc) + return rc; continue; - for (j = 0; j != obj->num_methods; j++) { - const struct uverbs_method_def *method = - (*obj->methods)[j]; - if (!method) + case UAPI_DEF_END: + return 0; + + case UAPI_DEF_IS_SUPPORTED_DEV_FN: { + void **ibdev_fn = (void *)ibdev + def->needs_fn_offset; + + if (*ibdev_fn) continue; + rc = uapi_disable_elm( + uapi, def, cur_obj_key, cur_method_key); + if (rc) + return rc; + continue; + } - rc = uapi_merge_method(uapi, obj_elm, obj_key, method, - is_driver); + case UAPI_DEF_IS_SUPPORTED_FUNC: + if (def->func_is_supported(ibdev)) + continue; + rc = uapi_disable_elm( + uapi, def, cur_obj_key, cur_method_key); if (rc) return rc; + continue; + + case UAPI_DEF_OBJECT_START: { + struct uverbs_api_object *obj_elm; + + cur_obj_key = uapi_key_obj(def->object_start.object_id); + obj_elm = uapi_add_get_elm(uapi, cur_obj_key, + sizeof(*obj_elm), &exists); + if (IS_ERR(obj_elm)) + return PTR_ERR(obj_elm); + continue; } - } - return 0; + case UAPI_DEF_WRITE: + rc = uapi_create_write( + uapi, ibdev, def, cur_obj_key, &cur_method_key); + if (rc) + return rc; + continue; + } + WARN_ON(true); + return -EINVAL; + } } static int @@ -186,13 +363,16 @@ uapi_finalize_ioctl_method(struct uverbs_api *uapi, u32 attr_bkey = uapi_bkey_attr(attr_key); u8 type = elm->spec.type; - if (uapi_key_attr_to_method(iter.index) != - uapi_key_attr_to_method(method_key)) + if (uapi_key_attr_to_ioctl_method(iter.index) != + uapi_key_attr_to_ioctl_method(method_key)) break; if (elm->spec.mandatory) __set_bit(attr_bkey, method_elm->attr_mandatory); + if (elm->spec.is_udata) + method_elm->has_udata = true; + if (type == UVERBS_ATTR_TYPE_IDR || type == UVERBS_ATTR_TYPE_FD) { u8 access = elm->spec.u.obj.access; @@ -229,9 +409,13 @@ uapi_finalize_ioctl_method(struct uverbs_api *uapi, static int uapi_finalize(struct uverbs_api *uapi) { + const struct uverbs_api_write_method **data; + unsigned long max_write_ex = 0; + unsigned long max_write = 0; struct radix_tree_iter iter; void __rcu **slot; int rc; + int i; radix_tree_for_each_slot (slot, &uapi->radix, &iter, 0) { struct uverbs_api_ioctl_method *method_elm = @@ -243,29 +427,208 @@ static int uapi_finalize(struct uverbs_api *uapi) if (rc) return rc; } + + if (uapi_key_is_write_method(iter.index)) + max_write = max(max_write, + iter.index & UVERBS_API_ATTR_KEY_MASK); + if (uapi_key_is_write_ex_method(iter.index)) + max_write_ex = + max(max_write_ex, + iter.index & UVERBS_API_ATTR_KEY_MASK); + } + + uapi->notsupp_method.handler = ib_uverbs_notsupp; + uapi->num_write = max_write + 1; + uapi->num_write_ex = max_write_ex + 1; + data = kmalloc_array(uapi->num_write + uapi->num_write_ex, + sizeof(*uapi->write_methods), GFP_KERNEL); + for (i = 0; i != uapi->num_write + uapi->num_write_ex; i++) + data[i] = &uapi->notsupp_method; + uapi->write_methods = data; + uapi->write_ex_methods = data + uapi->num_write; + + radix_tree_for_each_slot (slot, &uapi->radix, &iter, 0) { + if (uapi_key_is_write_method(iter.index)) + uapi->write_methods[iter.index & + UVERBS_API_ATTR_KEY_MASK] = + rcu_dereference_protected(*slot, true); + if (uapi_key_is_write_ex_method(iter.index)) + uapi->write_ex_methods[iter.index & + UVERBS_API_ATTR_KEY_MASK] = + rcu_dereference_protected(*slot, true); } return 0; } -void uverbs_destroy_api(struct uverbs_api *uapi) +static void uapi_remove_range(struct uverbs_api *uapi, u32 start, u32 last) { struct radix_tree_iter iter; void __rcu **slot; - if (!uapi) - return; - - radix_tree_for_each_slot (slot, &uapi->radix, &iter, 0) { + radix_tree_for_each_slot (slot, &uapi->radix, &iter, start) { + if (iter.index > last) + return; kfree(rcu_dereference_protected(*slot, true)); radix_tree_iter_delete(&uapi->radix, &iter, slot); } +} + +static void uapi_remove_object(struct uverbs_api *uapi, u32 obj_key) +{ + uapi_remove_range(uapi, obj_key, + obj_key | UVERBS_API_METHOD_KEY_MASK | + UVERBS_API_ATTR_KEY_MASK); +} + +static void uapi_remove_method(struct uverbs_api *uapi, u32 method_key) +{ + uapi_remove_range(uapi, method_key, + method_key | UVERBS_API_ATTR_KEY_MASK); +} + + +static u32 uapi_get_obj_id(struct uverbs_attr_spec *spec) +{ + if (spec->type == UVERBS_ATTR_TYPE_IDR || + spec->type == UVERBS_ATTR_TYPE_FD) + return spec->u.obj.obj_type; + if (spec->type == UVERBS_ATTR_TYPE_IDRS_ARRAY) + return spec->u2.objs_arr.obj_type; + return UVERBS_API_KEY_ERR; +} + +static void uapi_key_okay(u32 key) +{ + unsigned int count = 0; + + if (uapi_key_is_object(key)) + count++; + if (uapi_key_is_ioctl_method(key)) + count++; + if (uapi_key_is_write_method(key)) + count++; + if (uapi_key_is_write_ex_method(key)) + count++; + if (uapi_key_is_attr(key)) + count++; + WARN(count != 1, "Bad count %d key=%x", count, key); +} + +static void uapi_finalize_disable(struct uverbs_api *uapi) +{ + struct radix_tree_iter iter; + u32 starting_key = 0; + bool scan_again = false; + void __rcu **slot; + +again: + radix_tree_for_each_slot (slot, &uapi->radix, &iter, starting_key) { + uapi_key_okay(iter.index); + + if (uapi_key_is_object(iter.index)) { + struct uverbs_api_object *obj_elm = + rcu_dereference_protected(*slot, true); + + if (obj_elm->disabled) { + /* Have to check all the attrs again */ + scan_again = true; + starting_key = iter.index; + uapi_remove_object(uapi, iter.index); + goto again; + } + continue; + } + + if (uapi_key_is_ioctl_method(iter.index)) { + struct uverbs_api_ioctl_method *method_elm = + rcu_dereference_protected(*slot, true); + + if (method_elm->disabled) { + starting_key = iter.index; + uapi_remove_method(uapi, iter.index); + goto again; + } + continue; + } + + if (uapi_key_is_write_method(iter.index) || + uapi_key_is_write_ex_method(iter.index)) { + struct uverbs_api_write_method *method_elm = + rcu_dereference_protected(*slot, true); + + if (method_elm->disabled) { + kfree(method_elm); + radix_tree_iter_delete(&uapi->radix, &iter, slot); + } + continue; + } + + if (uapi_key_is_attr(iter.index)) { + struct uverbs_api_attr *attr_elm = + rcu_dereference_protected(*slot, true); + const struct uverbs_api_object *tmp_obj; + u32 obj_key; + + /* + * If the method has a mandatory object handle + * attribute which relies on an object which is not + * present then the entire method is uncallable. + */ + if (!attr_elm->spec.mandatory) + continue; + obj_key = uapi_get_obj_id(&attr_elm->spec); + if (obj_key == UVERBS_API_KEY_ERR) + continue; + tmp_obj = uapi_get_object(uapi, obj_key); + if (IS_ERR(tmp_obj)) { + if (PTR_ERR(tmp_obj) == -ENOMSG) + continue; + } else { + if (!tmp_obj->disabled) + continue; + } + + starting_key = iter.index; + uapi_remove_method( + uapi, + iter.index & (UVERBS_API_OBJ_KEY_MASK | + UVERBS_API_METHOD_KEY_MASK)); + goto again; + } + + WARN_ON(false); + } + + if (!scan_again) + return; + scan_again = false; + starting_key = 0; + goto again; +} + +void uverbs_destroy_api(struct uverbs_api *uapi) +{ + if (!uapi) + return; + + uapi_remove_range(uapi, 0, U32_MAX); + kfree(uapi->write_methods); kfree(uapi); } -struct uverbs_api *uverbs_alloc_api( - const struct uverbs_object_tree_def *const *driver_specs, - enum rdma_driver_id driver_id) +static const struct uapi_definition uverbs_core_api[] = { + UAPI_DEF_CHAIN(uverbs_def_obj_counters), + UAPI_DEF_CHAIN(uverbs_def_obj_cq), + UAPI_DEF_CHAIN(uverbs_def_obj_dm), + UAPI_DEF_CHAIN(uverbs_def_obj_flow_action), + UAPI_DEF_CHAIN(uverbs_def_obj_intf), + UAPI_DEF_CHAIN(uverbs_def_obj_mr), + UAPI_DEF_CHAIN(uverbs_def_write_intf), + {}, +}; + +struct uverbs_api *uverbs_alloc_api(struct ib_device *ibdev) { struct uverbs_api *uapi; int rc; @@ -275,18 +638,16 @@ struct uverbs_api *uverbs_alloc_api( return ERR_PTR(-ENOMEM); INIT_RADIX_TREE(&uapi->radix, GFP_KERNEL); - uapi->driver_id = driver_id; + uapi->driver_id = ibdev->driver_id; - rc = uapi_merge_tree(uapi, uverbs_default_get_objects(), false); + rc = uapi_merge_def(uapi, ibdev, uverbs_core_api, false); + if (rc) + goto err; + rc = uapi_merge_def(uapi, ibdev, ibdev->driver_def, true); if (rc) goto err; - for (; driver_specs && *driver_specs; driver_specs++) { - rc = uapi_merge_tree(uapi, *driver_specs, true); - if (rc) - goto err; - } - + uapi_finalize_disable(uapi); rc = uapi_finalize(uapi); if (rc) goto err; @@ -294,8 +655,9 @@ struct uverbs_api *uverbs_alloc_api( return uapi; err: if (rc != -ENOMEM) - pr_err("Setup of uverbs_api failed, kernel parsing tree description is not valid (%d)??\n", - rc); + dev_err(&ibdev->dev, + "Setup of uverbs_api failed, kernel parsing tree description is not valid (%d)??\n", + rc); uverbs_destroy_api(uapi); return ERR_PTR(rc); diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c index 178899e3ce73..fb2fc0c7ecfb 100644 --- a/drivers/infiniband/core/verbs.c +++ b/drivers/infiniband/core/verbs.c @@ -141,6 +141,10 @@ __attribute_const__ int ib_rate_to_mult(enum ib_rate rate) case IB_RATE_100_GBPS: return 40; case IB_RATE_200_GBPS: return 80; case IB_RATE_300_GBPS: return 120; + case IB_RATE_28_GBPS: return 11; + case IB_RATE_50_GBPS: return 20; + case IB_RATE_400_GBPS: return 160; + case IB_RATE_600_GBPS: return 240; default: return -1; } } @@ -166,6 +170,10 @@ __attribute_const__ enum ib_rate mult_to_ib_rate(int mult) case 40: return IB_RATE_100_GBPS; case 80: return IB_RATE_200_GBPS; case 120: return IB_RATE_300_GBPS; + case 11: return IB_RATE_28_GBPS; + case 20: return IB_RATE_50_GBPS; + case 160: return IB_RATE_400_GBPS; + case 240: return IB_RATE_600_GBPS; default: return IB_RATE_PORT_CURRENT; } } @@ -191,6 +199,10 @@ __attribute_const__ int ib_rate_to_mbps(enum ib_rate rate) case IB_RATE_100_GBPS: return 103125; case IB_RATE_200_GBPS: return 206250; case IB_RATE_300_GBPS: return 309375; + case IB_RATE_28_GBPS: return 28125; + case IB_RATE_50_GBPS: return 53125; + case IB_RATE_400_GBPS: return 425000; + case IB_RATE_600_GBPS: return 637500; default: return -1; } } diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c index dcb4bba522ba..df4f7a3f043d 100644 --- a/drivers/infiniband/hw/cxgb3/cxio_hal.c +++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c @@ -291,13 +291,12 @@ int cxio_create_qp(struct cxio_rdev *rdev_p, u32 kernel_domain, if (!wq->sq) goto err3; - wq->queue = dma_alloc_coherent(&(rdev_p->rnic_info.pdev->dev), + wq->queue = dma_zalloc_coherent(&(rdev_p->rnic_info.pdev->dev), depth * sizeof(union t3_wr), &(wq->dma_addr), GFP_KERNEL); if (!wq->queue) goto err4; - memset(wq->queue, 0, depth * sizeof(union t3_wr)); dma_unmap_addr_set(wq, mapping, wq->dma_addr); wq->doorbell = (void __iomem *)rdev_p->rnic_info.kdb_addr; if (!kernel_domain) diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c index 615413bd3e8d..659175c9ae91 100644 --- a/drivers/infiniband/hw/cxgb4/cm.c +++ b/drivers/infiniband/hw/cxgb4/cm.c @@ -2795,7 +2795,8 @@ static int peer_abort(struct c4iw_dev *dev, struct sk_buff *skb) break; case MPA_REQ_SENT: (void)stop_ep_timer(ep); - if (mpa_rev == 1 || (mpa_rev == 2 && ep->tried_with_mpa_v1)) + if (status != CPL_ERR_CONN_RESET || mpa_rev == 1 || + (mpa_rev == 2 && ep->tried_with_mpa_v1)) connect_reply_upcall(ep, -ECONNRESET); else { /* diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c index 13478f3b7057..5a8030bd4208 100644 --- a/drivers/infiniband/hw/cxgb4/qp.c +++ b/drivers/infiniband/hw/cxgb4/qp.c @@ -2564,13 +2564,12 @@ static int alloc_srq_queue(struct c4iw_srq *srq, struct c4iw_dev_ucontext *uctx, wq->rqt_abs_idx = (wq->rqt_hwaddr - rdev->lldi.vr->rq.start) >> T4_RQT_ENTRY_SHIFT; - wq->queue = dma_alloc_coherent(&rdev->lldi.pdev->dev, + wq->queue = dma_zalloc_coherent(&rdev->lldi.pdev->dev, wq->memsize, &wq->dma_addr, GFP_KERNEL); if (!wq->queue) goto err_free_rqtpool; - memset(wq->queue, 0, wq->memsize); dma_unmap_addr_set(wq, mapping, wq->dma_addr); wq->bar2_va = c4iw_bar2_addrs(rdev, wq->qid, CXGB4_BAR2_QTYPE_EGRESS, diff --git a/drivers/infiniband/hw/hfi1/Makefile b/drivers/infiniband/hw/hfi1/Makefile index ff790390c91a..3ce9dc8c3463 100644 --- a/drivers/infiniband/hw/hfi1/Makefile +++ b/drivers/infiniband/hw/hfi1/Makefile @@ -34,6 +34,7 @@ hfi1-y := \ ruc.o \ sdma.o \ sysfs.o \ + tid_rdma.o \ trace.o \ uc.o \ ud.o \ diff --git a/drivers/infiniband/hw/hfi1/chip.c b/drivers/infiniband/hw/hfi1/chip.c index 9b20479dc710..385c33745c9f 100644 --- a/drivers/infiniband/hw/hfi1/chip.c +++ b/drivers/infiniband/hw/hfi1/chip.c @@ -1072,6 +1072,8 @@ static void log_state_transition(struct hfi1_pportdata *ppd, u32 state); static void log_physical_state(struct hfi1_pportdata *ppd, u32 state); static int wait_physical_linkstate(struct hfi1_pportdata *ppd, u32 state, int msecs); +static int wait_phys_link_out_of_offline(struct hfi1_pportdata *ppd, + int msecs); static void read_planned_down_reason_code(struct hfi1_devdata *dd, u8 *pdrrc); static void read_link_down_reason(struct hfi1_devdata *dd, u8 *ldr); static void handle_temp_err(struct hfi1_devdata *dd); @@ -10770,13 +10772,15 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state) break; ppd->port_error_action = 0; - ppd->host_link_state = HLS_DN_POLL; if (quick_linkup) { /* quick linkup does not go into polling */ ret = do_quick_linkup(dd); } else { ret1 = set_physical_link_state(dd, PLS_POLLING); + if (!ret1) + ret1 = wait_phys_link_out_of_offline(ppd, + 3000); if (ret1 != HCMD_SUCCESS) { dd_dev_err(dd, "Failed to transition to Polling link state, return 0x%x\n", @@ -10784,6 +10788,14 @@ int set_link_state(struct hfi1_pportdata *ppd, u32 state) ret = -EINVAL; } } + + /* + * Change the host link state after requesting DC8051 to + * change its physical state so that we can ignore any + * interrupt with stale LNI(XX) error, which will not be + * cleared until DC8051 transitions to Polling state. + */ + ppd->host_link_state = HLS_DN_POLL; ppd->offline_disabled_reason = HFI1_ODR_MASK(OPA_LINKDOWN_REASON_NONE); /* @@ -12927,6 +12939,39 @@ static int wait_phys_link_offline_substates(struct hfi1_pportdata *ppd, return read_state; } +/* + * wait_phys_link_out_of_offline - wait for any out of offline state + * @ppd: port device + * @msecs: the number of milliseconds to wait + * + * Wait up to msecs milliseconds for any out of offline physical link + * state change to occur. + * Returns 0 if at least one state is reached, otherwise -ETIMEDOUT. + */ +static int wait_phys_link_out_of_offline(struct hfi1_pportdata *ppd, + int msecs) +{ + u32 read_state; + unsigned long timeout; + + timeout = jiffies + msecs_to_jiffies(msecs); + while (1) { + read_state = read_physical_state(ppd->dd); + if ((read_state & 0xF0) != PLS_OFFLINE) + break; + if (time_after(jiffies, timeout)) { + dd_dev_err(ppd->dd, + "timeout waiting for phy link out of offline. Read state 0x%x, %dms\n", + read_state, msecs); + return -ETIMEDOUT; + } + usleep_range(1950, 2050); /* sleep 2ms-ish */ + } + + log_state_transition(ppd, read_state); + return read_state; +} + #define CLEAR_STATIC_RATE_CONTROL_SMASK(r) \ (r &= ~SEND_CTXT_CHECK_ENABLE_DISALLOW_PBC_STATIC_RATE_CONTROL_SMASK) diff --git a/drivers/infiniband/hw/hfi1/chip_registers.h b/drivers/infiniband/hw/hfi1/chip_registers.h index c6163a347e93..c0800ea5a3f8 100644 --- a/drivers/infiniband/hw/hfi1/chip_registers.h +++ b/drivers/infiniband/hw/hfi1/chip_registers.h @@ -935,6 +935,10 @@ #define SEND_CTXT_CREDIT_CTRL_THRESHOLD_MASK 0x7FFull #define SEND_CTXT_CREDIT_CTRL_THRESHOLD_SHIFT 0 #define SEND_CTXT_CREDIT_CTRL_THRESHOLD_SMASK 0x7FFull +#define SEND_CTXT_CREDIT_STATUS (TXE + 0x000000100018) +#define SEND_CTXT_CREDIT_STATUS_CURRENT_FREE_COUNTER_MASK 0x7FFull +#define SEND_CTXT_CREDIT_STATUS_CURRENT_FREE_COUNTER_SHIFT 32 +#define SEND_CTXT_CREDIT_STATUS_LAST_RETURNED_COUNTER_SMASK 0x7FFull #define SEND_CTXT_CREDIT_FORCE (TXE + 0x000000100028) #define SEND_CTXT_CREDIT_FORCE_FORCE_RETURN_SMASK 0x1ull #define SEND_CTXT_CREDIT_RETURN_ADDR (TXE + 0x000000100020) diff --git a/drivers/infiniband/hw/hfi1/common.h b/drivers/infiniband/hw/hfi1/common.h index 7108d4d92259..40d3cfb58bd1 100644 --- a/drivers/infiniband/hw/hfi1/common.h +++ b/drivers/infiniband/hw/hfi1/common.h @@ -1,5 +1,5 @@ /* - * Copyright(c) 2015, 2016 Intel Corporation. + * Copyright(c) 2015 - 2018 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -136,18 +136,21 @@ HFI1_CAP_ALLOW_PERM_JKEY | \ HFI1_CAP_STATIC_RATE_CTRL | \ HFI1_CAP_PRINT_UNIMPL | \ - HFI1_CAP_TID_UNMAP) + HFI1_CAP_TID_UNMAP | \ + HFI1_CAP_OPFN) /* * A set of capability bits that are "global" and are not allowed to be * set in the user bitmask. */ #define HFI1_CAP_RESERVED_MASK ((HFI1_CAP_SDMA | \ - HFI1_CAP_USE_SDMA_HEAD | \ - HFI1_CAP_EXTENDED_PSN | \ - HFI1_CAP_PRINT_UNIMPL | \ - HFI1_CAP_NO_INTEGRITY | \ - HFI1_CAP_PKEY_CHECK) << \ - HFI1_CAP_USER_SHIFT) + HFI1_CAP_USE_SDMA_HEAD | \ + HFI1_CAP_EXTENDED_PSN | \ + HFI1_CAP_PRINT_UNIMPL | \ + HFI1_CAP_NO_INTEGRITY | \ + HFI1_CAP_PKEY_CHECK | \ + HFI1_CAP_TID_RDMA | \ + HFI1_CAP_OPFN) << \ + HFI1_CAP_USER_SHIFT) /* * Set of capabilities that need to be enabled for kernel context in * order to be allowed for user contexts, as well. diff --git a/drivers/infiniband/hw/hfi1/debugfs.c b/drivers/infiniband/hw/hfi1/debugfs.c index 9f992ae36c89..0a557795563c 100644 --- a/drivers/infiniband/hw/hfi1/debugfs.c +++ b/drivers/infiniband/hw/hfi1/debugfs.c @@ -407,6 +407,54 @@ DEBUGFS_SEQ_FILE_OPS(rcds); DEBUGFS_SEQ_FILE_OPEN(rcds) DEBUGFS_FILE_OPS(rcds); +static void *_pios_seq_start(struct seq_file *s, loff_t *pos) +{ + struct hfi1_ibdev *ibd; + struct hfi1_devdata *dd; + + ibd = (struct hfi1_ibdev *)s->private; + dd = dd_from_dev(ibd); + if (!dd->send_contexts || *pos >= dd->num_send_contexts) + return NULL; + return pos; +} + +static void *_pios_seq_next(struct seq_file *s, void *v, loff_t *pos) +{ + struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private; + struct hfi1_devdata *dd = dd_from_dev(ibd); + + ++*pos; + if (!dd->send_contexts || *pos >= dd->num_send_contexts) + return NULL; + return pos; +} + +static void _pios_seq_stop(struct seq_file *s, void *v) +{ +} + +static int _pios_seq_show(struct seq_file *s, void *v) +{ + struct hfi1_ibdev *ibd = (struct hfi1_ibdev *)s->private; + struct hfi1_devdata *dd = dd_from_dev(ibd); + struct send_context_info *sci; + loff_t *spos = v; + loff_t i = *spos; + unsigned long flags; + + spin_lock_irqsave(&dd->sc_lock, flags); + sci = &dd->send_contexts[i]; + if (sci && sci->type != SC_USER && sci->allocated && sci->sc) + seqfile_dump_sci(s, i, sci); + spin_unlock_irqrestore(&dd->sc_lock, flags); + return 0; +} + +DEBUGFS_SEQ_FILE_OPS(pios); +DEBUGFS_SEQ_FILE_OPEN(pios) +DEBUGFS_FILE_OPS(pios); + /* read the per-device counters */ static ssize_t dev_counters_read(struct file *file, char __user *buf, size_t count, loff_t *ppos) @@ -1143,6 +1191,7 @@ void hfi1_dbg_ibdev_init(struct hfi1_ibdev *ibd) DEBUGFS_SEQ_FILE_CREATE(qp_stats, ibd->hfi1_ibdev_dbg, ibd); DEBUGFS_SEQ_FILE_CREATE(sdes, ibd->hfi1_ibdev_dbg, ibd); DEBUGFS_SEQ_FILE_CREATE(rcds, ibd->hfi1_ibdev_dbg, ibd); + DEBUGFS_SEQ_FILE_CREATE(pios, ibd->hfi1_ibdev_dbg, ibd); DEBUGFS_SEQ_FILE_CREATE(sdma_cpu_list, ibd->hfi1_ibdev_dbg, ibd); /* dev counter files */ for (i = 0; i < ARRAY_SIZE(cntr_ops); i++) diff --git a/drivers/infiniband/hw/hfi1/driver.c b/drivers/infiniband/hw/hfi1/driver.c index a41f85558312..a8ad70730203 100644 --- a/drivers/infiniband/hw/hfi1/driver.c +++ b/drivers/infiniband/hw/hfi1/driver.c @@ -430,40 +430,60 @@ static const hfi1_handle_cnp hfi1_handle_cnp_tbl[2] = { [HFI1_PKT_TYPE_16B] = &return_cnp_16B }; -void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt, - bool do_cnp) +/** + * hfi1_process_ecn_slowpath - Process FECN or BECN bits + * @qp: The packet's destination QP + * @pkt: The packet itself. + * @prescan: Is the caller the RXQ prescan + * + * Process the packet's FECN or BECN bits. By now, the packet + * has already been evaluated whether processing of those bit should + * be done. + * The significance of the @prescan argument is that if the caller + * is the RXQ prescan, a CNP will be send out instead of waiting for the + * normal packet processing to send an ACK with BECN set (or a CNP). + */ +bool hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt, + bool prescan) { struct hfi1_ibport *ibp = to_iport(qp->ibqp.device, qp->port_num); struct hfi1_pportdata *ppd = ppd_from_ibp(ibp); struct ib_other_headers *ohdr = pkt->ohdr; struct ib_grh *grh = pkt->grh; - u32 rqpn = 0, bth1; + u32 rqpn = 0; u16 pkey; u32 rlid, slid, dlid = 0; - u8 hdr_type, sc, svc_type; - bool is_mcast = false; + u8 hdr_type, sc, svc_type, opcode; + bool is_mcast = false, ignore_fecn = false, do_cnp = false, + fecn, becn; /* can be called from prescan */ if (pkt->etype == RHF_RCV_TYPE_BYPASS) { - is_mcast = hfi1_is_16B_mcast(dlid); pkey = hfi1_16B_get_pkey(pkt->hdr); sc = hfi1_16B_get_sc(pkt->hdr); dlid = hfi1_16B_get_dlid(pkt->hdr); slid = hfi1_16B_get_slid(pkt->hdr); + is_mcast = hfi1_is_16B_mcast(dlid); + opcode = ib_bth_get_opcode(ohdr); hdr_type = HFI1_PKT_TYPE_16B; + fecn = hfi1_16B_get_fecn(pkt->hdr); + becn = hfi1_16B_get_becn(pkt->hdr); } else { - is_mcast = (dlid > be16_to_cpu(IB_MULTICAST_LID_BASE)) && - (dlid != be16_to_cpu(IB_LID_PERMISSIVE)); pkey = ib_bth_get_pkey(ohdr); sc = hfi1_9B_get_sc5(pkt->hdr, pkt->rhf); - dlid = ib_get_dlid(pkt->hdr); + dlid = qp->ibqp.qp_type != IB_QPT_UD ? ib_get_dlid(pkt->hdr) : + ppd->lid; slid = ib_get_slid(pkt->hdr); + is_mcast = (dlid > be16_to_cpu(IB_MULTICAST_LID_BASE)) && + (dlid != be16_to_cpu(IB_LID_PERMISSIVE)); + opcode = ib_bth_get_opcode(ohdr); hdr_type = HFI1_PKT_TYPE_9B; + fecn = ib_bth_get_fecn(ohdr); + becn = ib_bth_get_becn(ohdr); } switch (qp->ibqp.qp_type) { case IB_QPT_UD: - dlid = ppd->lid; rlid = slid; rqpn = ib_get_sqpn(pkt->ohdr); svc_type = IB_CC_SVCTYPE_UD; @@ -485,22 +505,31 @@ void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt, svc_type = IB_CC_SVCTYPE_RC; break; default: - return; + return false; } - bth1 = be32_to_cpu(ohdr->bth[1]); + ignore_fecn = is_mcast || (opcode == IB_OPCODE_CNP) || + (opcode == IB_OPCODE_RC_ACKNOWLEDGE); + /* + * ACKNOWLEDGE packets do not get a CNP but this will be + * guarded by ignore_fecn above. + */ + do_cnp = prescan || + (opcode >= IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST && + opcode <= IB_OPCODE_RC_ATOMIC_ACKNOWLEDGE); + /* Call appropriate CNP handler */ - if (do_cnp && (bth1 & IB_FECN_SMASK)) + if (!ignore_fecn && do_cnp && fecn) hfi1_handle_cnp_tbl[hdr_type](ibp, qp, rqpn, pkey, dlid, rlid, sc, grh); - if (!is_mcast && (bth1 & IB_BECN_SMASK)) { - u32 lqpn = bth1 & RVT_QPN_MASK; + if (becn) { + u32 lqpn = be32_to_cpu(ohdr->bth[1]) & RVT_QPN_MASK; u8 sl = ibp->sc_to_sl[sc]; process_becn(ppd, sl, rlid, lqpn, rqpn, svc_type); } - + return !ignore_fecn && fecn; } struct ps_mdata { @@ -599,7 +628,6 @@ static void __prescan_rxq(struct hfi1_packet *packet) struct rvt_dev_info *rdi = &rcd->dd->verbs_dev.rdi; u64 rhf = rhf_to_cpu(rhf_addr); u32 etype = rhf_rcv_type(rhf), qpn, bth1; - int is_ecn = 0; u8 lnh; if (ps_done(&mdata, rhf, rcd)) @@ -625,12 +653,10 @@ static void __prescan_rxq(struct hfi1_packet *packet) goto next; /* just in case */ } - bth1 = be32_to_cpu(packet->ohdr->bth[1]); - is_ecn = !!(bth1 & (IB_FECN_SMASK | IB_BECN_SMASK)); - - if (!is_ecn) + if (!hfi1_may_ecn(packet)) goto next; + bth1 = be32_to_cpu(packet->ohdr->bth[1]); qpn = bth1 & RVT_QPN_MASK; rcu_read_lock(); qp = rvt_lookup_qpn(rdi, &ibp->rvp, qpn); @@ -640,7 +666,7 @@ static void __prescan_rxq(struct hfi1_packet *packet) goto next; } - process_ecn(qp, packet, true); + hfi1_process_ecn_slowpath(qp, packet, true); rcu_read_unlock(); /* turn off BECN, FECN */ @@ -1400,7 +1426,7 @@ static int hfi1_bypass_ingress_pkt_check(struct hfi1_packet *packet) if ((!(hfi1_is_16B_mcast(packet->dlid))) && (packet->dlid != opa_get_lid(be32_to_cpu(OPA_LID_PERMISSIVE), 16B))) { - if (packet->dlid != ppd->lid) + if ((packet->dlid & ~((1 << ppd->lmc) - 1)) != ppd->lid) return -EINVAL; } diff --git a/drivers/infiniband/hw/hfi1/hfi.h b/drivers/infiniband/hw/hfi1/hfi.h index 1401b6ea4a28..ca572ad6061c 100644 --- a/drivers/infiniband/hw/hfi1/hfi.h +++ b/drivers/infiniband/hw/hfi1/hfi.h @@ -1802,13 +1802,20 @@ static inline struct hfi1_ibport *rcd_to_iport(struct hfi1_ctxtdata *rcd) return &rcd->ppd->ibport_data; } -void hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt, - bool do_cnp); -static inline bool process_ecn(struct rvt_qp *qp, struct hfi1_packet *pkt, - bool do_cnp) +/** + * hfi1_may_ecn - Check whether FECN or BECN processing should be done + * @pkt: the packet to be evaluated + * + * Check whether the FECN or BECN bits in the packet's header are + * enabled, depending on packet type. + * + * This function only checks for FECN and BECN bits. Additional checks + * are done in the slowpath (hfi1_process_ecn_slowpath()) in order to + * ensure correct handling. + */ +static inline bool hfi1_may_ecn(struct hfi1_packet *pkt) { - bool becn; - bool fecn; + bool fecn, becn; if (pkt->etype == RHF_RCV_TYPE_BYPASS) { fecn = hfi1_16B_get_fecn(pkt->hdr); @@ -1817,10 +1824,18 @@ static inline bool process_ecn(struct rvt_qp *qp, struct hfi1_packet *pkt, fecn = ib_bth_get_fecn(pkt->ohdr); becn = ib_bth_get_becn(pkt->ohdr); } - if (unlikely(fecn || becn)) { - hfi1_process_ecn_slowpath(qp, pkt, do_cnp); - return fecn; - } + return fecn || becn; +} + +bool hfi1_process_ecn_slowpath(struct rvt_qp *qp, struct hfi1_packet *pkt, + bool prescan); +static inline bool process_ecn(struct rvt_qp *qp, struct hfi1_packet *pkt) +{ + bool do_work; + + do_work = hfi1_may_ecn(pkt); + if (unlikely(do_work)) + return hfi1_process_ecn_slowpath(qp, pkt, false); return false; } diff --git a/drivers/infiniband/hw/hfi1/pio.c b/drivers/infiniband/hw/hfi1/pio.c index 9ab50d2308dc..dd5a5c030066 100644 --- a/drivers/infiniband/hw/hfi1/pio.c +++ b/drivers/infiniband/hw/hfi1/pio.c @@ -742,6 +742,7 @@ struct send_context *sc_alloc(struct hfi1_devdata *dd, int type, spin_lock_init(&sc->alloc_lock); spin_lock_init(&sc->release_lock); spin_lock_init(&sc->credit_ctrl_lock); + seqlock_init(&sc->waitlock); INIT_LIST_HEAD(&sc->piowait); INIT_WORK(&sc->halt_work, sc_halted); init_waitqueue_head(&sc->halt_wait); @@ -1593,7 +1594,6 @@ void hfi1_sc_wantpiobuf_intr(struct send_context *sc, u32 needint) static void sc_piobufavail(struct send_context *sc) { struct hfi1_devdata *dd = sc->dd; - struct hfi1_ibdev *dev = &dd->verbs_dev; struct list_head *list; struct rvt_qp *qps[PIO_WAIT_BATCH_SIZE]; struct rvt_qp *qp; @@ -1612,7 +1612,7 @@ static void sc_piobufavail(struct send_context *sc) * could end up with QPs on the wait list with the interrupt * disabled. */ - write_seqlock_irqsave(&dev->iowait_lock, flags); + write_seqlock_irqsave(&sc->waitlock, flags); while (!list_empty(list)) { struct iowait *wait; @@ -1636,7 +1636,7 @@ static void sc_piobufavail(struct send_context *sc) if (!list_empty(list)) hfi1_sc_wantpiobuf_intr(sc, 1); } - write_sequnlock_irqrestore(&dev->iowait_lock, flags); + write_sequnlock_irqrestore(&sc->waitlock, flags); /* Wake up the most starved one first */ if (n) @@ -2137,3 +2137,28 @@ void free_credit_return(struct hfi1_devdata *dd) kfree(dd->cr_base); dd->cr_base = NULL; } + +void seqfile_dump_sci(struct seq_file *s, u32 i, + struct send_context_info *sci) +{ + struct send_context *sc = sci->sc; + u64 reg; + + seq_printf(s, "SCI %u: type %u base %u credits %u\n", + i, sci->type, sci->base, sci->credits); + seq_printf(s, " flags 0x%x sw_inx %u hw_ctxt %u grp %u\n", + sc->flags, sc->sw_index, sc->hw_context, sc->group); + seq_printf(s, " sr_size %u credits %u sr_head %u sr_tail %u\n", + sc->sr_size, sc->credits, sc->sr_head, sc->sr_tail); + seq_printf(s, " fill %lu free %lu fill_wrap %u alloc_free %lu\n", + sc->fill, sc->free, sc->fill_wrap, sc->alloc_free); + seq_printf(s, " credit_intr_count %u credit_ctrl 0x%llx\n", + sc->credit_intr_count, sc->credit_ctrl); + reg = read_kctxt_csr(sc->dd, sc->hw_context, SC(CREDIT_STATUS)); + seq_printf(s, " *hw_free %llu CurrentFree %llu LastReturned %llu\n", + (le64_to_cpu(*sc->hw_free) & CR_COUNTER_SMASK) >> + CR_COUNTER_SHIFT, + (reg >> SC(CREDIT_STATUS_CURRENT_FREE_COUNTER_SHIFT)) & + SC(CREDIT_STATUS_CURRENT_FREE_COUNTER_MASK), + reg & SC(CREDIT_STATUS_LAST_RETURNED_COUNTER_SMASK)); +} diff --git a/drivers/infiniband/hw/hfi1/pio.h b/drivers/infiniband/hw/hfi1/pio.h index aaf372c3e5d6..c9a58b642bdd 100644 --- a/drivers/infiniband/hw/hfi1/pio.h +++ b/drivers/infiniband/hw/hfi1/pio.h @@ -127,6 +127,8 @@ struct send_context { volatile __le64 *hw_free; /* HW free counter */ /* list for PIO waiters */ struct list_head piowait ____cacheline_aligned_in_smp; + seqlock_t waitlock; + spinlock_t credit_ctrl_lock ____cacheline_aligned_in_smp; u32 credit_intr_count; /* count of credit intr users */ u64 credit_ctrl; /* cache for credit control */ @@ -329,4 +331,7 @@ void seg_pio_copy_start(struct pio_buf *pbuf, u64 pbc, void seg_pio_copy_mid(struct pio_buf *pbuf, const void *from, size_t nbytes); void seg_pio_copy_end(struct pio_buf *pbuf); +void seqfile_dump_sci(struct seq_file *s, u32 i, + struct send_context_info *sci); + #endif /* _PIO_H */ diff --git a/drivers/infiniband/hw/hfi1/qp.c b/drivers/infiniband/hw/hfi1/qp.c index 6f3bc4dab858..e32fbfe029bc 100644 --- a/drivers/infiniband/hw/hfi1/qp.c +++ b/drivers/infiniband/hw/hfi1/qp.c @@ -368,20 +368,18 @@ bool _hfi1_schedule_send(struct rvt_qp *qp) static void qp_pio_drain(struct rvt_qp *qp) { - struct hfi1_ibdev *dev; struct hfi1_qp_priv *priv = qp->priv; if (!priv->s_sendcontext) return; - dev = to_idev(qp->ibqp.device); while (iowait_pio_pending(&priv->s_iowait)) { - write_seqlock_irq(&dev->iowait_lock); + write_seqlock_irq(&priv->s_sendcontext->waitlock); hfi1_sc_wantpiobuf_intr(priv->s_sendcontext, 1); - write_sequnlock_irq(&dev->iowait_lock); + write_sequnlock_irq(&priv->s_sendcontext->waitlock); iowait_pio_drain(&priv->s_iowait); - write_seqlock_irq(&dev->iowait_lock); + write_seqlock_irq(&priv->s_sendcontext->waitlock); hfi1_sc_wantpiobuf_intr(priv->s_sendcontext, 0); - write_sequnlock_irq(&dev->iowait_lock); + write_sequnlock_irq(&priv->s_sendcontext->waitlock); } } @@ -452,7 +450,6 @@ static int iowait_sleep( struct hfi1_qp_priv *priv; unsigned long flags; int ret = 0; - struct hfi1_ibdev *dev; qp = tx->qp; priv = qp->priv; @@ -465,9 +462,8 @@ static int iowait_sleep( * buffer and undoing the side effects of the copy. */ /* Make a common routine? */ - dev = &sde->dd->verbs_dev; list_add_tail(&stx->list, &wait->tx_head); - write_seqlock(&dev->iowait_lock); + write_seqlock(&sde->waitlock); if (sdma_progress(sde, seq, stx)) goto eagain; if (list_empty(&priv->s_iowait.list)) { @@ -478,11 +474,11 @@ static int iowait_sleep( qp->s_flags |= RVT_S_WAIT_DMA_DESC; iowait_queue(pkts_sent, &priv->s_iowait, &sde->dmawait); - priv->s_iowait.lock = &dev->iowait_lock; + priv->s_iowait.lock = &sde->waitlock; trace_hfi1_qpsleep(qp, RVT_S_WAIT_DMA_DESC); rvt_get_qp(qp); } - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sde->waitlock); hfi1_qp_unbusy(qp, wait); spin_unlock_irqrestore(&qp->s_lock, flags); ret = -EBUSY; @@ -492,7 +488,7 @@ static int iowait_sleep( } return ret; eagain: - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sde->waitlock); spin_unlock_irqrestore(&qp->s_lock, flags); list_del_init(&stx->list); return -EAGAIN; diff --git a/drivers/infiniband/hw/hfi1/rc.c b/drivers/infiniband/hw/hfi1/rc.c index 188aa4f686a0..be603f35d7e4 100644 --- a/drivers/infiniband/hw/hfi1/rc.c +++ b/drivers/infiniband/hw/hfi1/rc.c @@ -1157,6 +1157,7 @@ void hfi1_rc_send_complete(struct rvt_qp *qp, struct hfi1_opa_header *opah) if (cmp_psn(wqe->lpsn, qp->s_sending_psn) >= 0 && cmp_psn(qp->s_sending_psn, qp->s_sending_hpsn) <= 0) break; + rvt_qp_wqe_unreserve(qp, wqe); s_last = qp->s_last; trace_hfi1_qp_send_completion(qp, wqe, s_last); if (++s_last >= qp->s_size) @@ -1209,6 +1210,7 @@ static struct rvt_swqe *do_rc_completion(struct rvt_qp *qp, u32 s_last; rvt_put_swqe(wqe); + rvt_qp_wqe_unreserve(qp, wqe); s_last = qp->s_last; trace_hfi1_qp_send_completion(qp, wqe, s_last); if (++s_last >= qp->s_size) @@ -2049,8 +2051,7 @@ void hfi1_rc_rcv(struct hfi1_packet *packet) struct ib_reth *reth; unsigned long flags; int ret; - bool is_fecn = false; - bool copy_last = false; + bool copy_last = false, fecn; u32 rkey; u8 extra_bytes = pad + packet->extra_byte + (SIZE_OF_CRC << 2); @@ -2059,7 +2060,7 @@ void hfi1_rc_rcv(struct hfi1_packet *packet) if (hfi1_ruc_check_hdr(ibp, packet)) return; - is_fecn = process_ecn(qp, packet, false); + fecn = process_ecn(qp, packet); /* * Process responses (ACKs) before anything else. Note that the @@ -2070,8 +2071,6 @@ void hfi1_rc_rcv(struct hfi1_packet *packet) if (opcode >= OP(RDMA_READ_RESPONSE_FIRST) && opcode <= OP(ATOMIC_ACKNOWLEDGE)) { rc_rcv_resp(packet); - if (is_fecn) - goto send_ack; return; } @@ -2347,11 +2346,11 @@ send_last: /* Schedule the send engine. */ qp->s_flags |= RVT_S_RESP_PENDING; + if (fecn) + qp->s_flags |= RVT_S_ECN; hfi1_schedule_send(qp); spin_unlock_irqrestore(&qp->s_lock, flags); - if (is_fecn) - goto send_ack; return; } @@ -2413,11 +2412,11 @@ send_last: /* Schedule the send engine. */ qp->s_flags |= RVT_S_RESP_PENDING; + if (fecn) + qp->s_flags |= RVT_S_ECN; hfi1_schedule_send(qp); spin_unlock_irqrestore(&qp->s_lock, flags); - if (is_fecn) - goto send_ack; return; } @@ -2430,16 +2429,9 @@ send_last: qp->r_ack_psn = psn; qp->r_nak_state = 0; /* Send an ACK if requested or required. */ - if (psn & IB_BTH_REQ_ACK) { - if (packet->numpkt == 0) { - rc_cancel_ack(qp); - goto send_ack; - } - if (qp->r_adefered >= HFI1_PSN_CREDIT) { - rc_cancel_ack(qp); - goto send_ack; - } - if (unlikely(is_fecn)) { + if (psn & IB_BTH_REQ_ACK || fecn) { + if (packet->numpkt == 0 || fecn || + qp->r_adefered >= HFI1_PSN_CREDIT) { rc_cancel_ack(qp); goto send_ack; } @@ -2480,7 +2472,7 @@ nack_acc: qp->r_nak_state = IB_NAK_REMOTE_ACCESS_ERROR; qp->r_ack_psn = qp->r_psn; send_ack: - hfi1_send_rc_ack(packet, is_fecn); + hfi1_send_rc_ack(packet, fecn); } void hfi1_rc_hdrerr( diff --git a/drivers/infiniband/hw/hfi1/sdma.c b/drivers/infiniband/hw/hfi1/sdma.c index 891d2386d1ca..b84356e1a4c1 100644 --- a/drivers/infiniband/hw/hfi1/sdma.c +++ b/drivers/infiniband/hw/hfi1/sdma.c @@ -1424,6 +1424,7 @@ int sdma_init(struct hfi1_devdata *dd, u8 port) seqlock_init(&sde->head_lock); spin_lock_init(&sde->senddmactrl_lock); spin_lock_init(&sde->flushlist_lock); + seqlock_init(&sde->waitlock); /* insure there is always a zero bit */ sde->ahg_bits = 0xfffffffe00000000ULL; @@ -1758,7 +1759,6 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail) struct iowait *wait, *nw; struct iowait *waits[SDMA_WAIT_BATCH_SIZE]; uint i, n = 0, seq, max_idx = 0; - struct hfi1_ibdev *dev = &sde->dd->verbs_dev; u8 max_starved_cnt = 0; #ifdef CONFIG_SDMA_VERBOSITY @@ -1768,10 +1768,10 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail) #endif do { - seq = read_seqbegin(&dev->iowait_lock); + seq = read_seqbegin(&sde->waitlock); if (!list_empty(&sde->dmawait)) { /* at least one item */ - write_seqlock(&dev->iowait_lock); + write_seqlock(&sde->waitlock); /* Harvest waiters wanting DMA descriptors */ list_for_each_entry_safe( wait, @@ -1794,10 +1794,10 @@ static void sdma_desc_avail(struct sdma_engine *sde, uint avail) list_del_init(&wait->list); waits[n++] = wait; } - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sde->waitlock); break; } - } while (read_seqretry(&dev->iowait_lock, seq)); + } while (read_seqretry(&sde->waitlock, seq)); /* Schedule the most starved one first */ if (n) diff --git a/drivers/infiniband/hw/hfi1/sdma.h b/drivers/infiniband/hw/hfi1/sdma.h index 6dc63d7c5685..1e2e40f79cb2 100644 --- a/drivers/infiniband/hw/hfi1/sdma.h +++ b/drivers/infiniband/hw/hfi1/sdma.h @@ -382,6 +382,7 @@ struct sdma_engine { u64 progress_int_cnt; /* private: */ + seqlock_t waitlock; struct list_head dmawait; /* CONFIG SDMA for now, just blindly duplicate */ diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.c b/drivers/infiniband/hw/hfi1/tid_rdma.c new file mode 100644 index 000000000000..da1ecb68a928 --- /dev/null +++ b/drivers/infiniband/hw/hfi1/tid_rdma.c @@ -0,0 +1,48 @@ +// SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) +/* + * Copyright(c) 2018 Intel Corporation. + * + */ + +#include "hfi.h" +#include "verbs.h" +#include "tid_rdma.h" + +/** + * qp_to_rcd - determine the receive context used by a qp + * @qp - the qp + * + * This routine returns the receive context associated + * with a a qp's qpn. + * + * Returns the context. + */ +static struct hfi1_ctxtdata *qp_to_rcd(struct rvt_dev_info *rdi, + struct rvt_qp *qp) +{ + struct hfi1_ibdev *verbs_dev = container_of(rdi, + struct hfi1_ibdev, + rdi); + struct hfi1_devdata *dd = container_of(verbs_dev, + struct hfi1_devdata, + verbs_dev); + unsigned int ctxt; + + if (qp->ibqp.qp_num == 0) + ctxt = 0; + else + ctxt = ((qp->ibqp.qp_num >> dd->qos_shift) % + (dd->n_krcv_queues - 1)) + 1; + + return dd->rcd[ctxt]; +} + +int hfi1_qp_priv_init(struct rvt_dev_info *rdi, struct rvt_qp *qp, + struct ib_qp_init_attr *init_attr) +{ + struct hfi1_qp_priv *qpriv = qp->priv; + + qpriv->rcd = qp_to_rcd(rdi, qp); + + return 0; +} diff --git a/drivers/infiniband/hw/hfi1/tid_rdma.h b/drivers/infiniband/hw/hfi1/tid_rdma.h new file mode 100644 index 000000000000..6fcd3adcdcc3 --- /dev/null +++ b/drivers/infiniband/hw/hfi1/tid_rdma.h @@ -0,0 +1,13 @@ +/* SPDX-License-Identifier: (GPL-2.0 OR BSD-3-Clause) */ +/* + * Copyright(c) 2018 Intel Corporation. + * + */ +#ifndef HFI1_TID_RDMA_H +#define HFI1_TID_RDMA_H + +int hfi1_qp_priv_init(struct rvt_dev_info *rdi, struct rvt_qp *qp, + struct ib_qp_init_attr *init_attr); + +#endif /* HFI1_TID_RDMA_H */ + diff --git a/drivers/infiniband/hw/hfi1/uc.c b/drivers/infiniband/hw/hfi1/uc.c index 6aca0c5a7f97..6ba47037c424 100644 --- a/drivers/infiniband/hw/hfi1/uc.c +++ b/drivers/infiniband/hw/hfi1/uc.c @@ -321,7 +321,7 @@ void hfi1_uc_rcv(struct hfi1_packet *packet) if (hfi1_ruc_check_hdr(ibp, packet)) return; - process_ecn(qp, packet, true); + process_ecn(qp, packet); psn = ib_bth_get_psn(ohdr); /* Compare the PSN verses the expected PSN. */ diff --git a/drivers/infiniband/hw/hfi1/ud.c b/drivers/infiniband/hw/hfi1/ud.c index 4baa8f4d49de..88242fe95eaa 100644 --- a/drivers/infiniband/hw/hfi1/ud.c +++ b/drivers/infiniband/hw/hfi1/ud.c @@ -51,6 +51,7 @@ #include "hfi.h" #include "mad.h" #include "verbs_txreq.h" +#include "trace_ibhdrs.h" #include "qp.h" /* We support only two types - 9B and 16B for now */ @@ -656,18 +657,19 @@ void return_cnp_16B(struct hfi1_ibport *ibp, struct rvt_qp *qp, u32 bth0, plen, vl, hwords = 7; u16 len; u8 l4; - struct hfi1_16b_header hdr; + struct hfi1_opa_header hdr; struct ib_other_headers *ohdr; struct pio_buf *pbuf; struct send_context *ctxt = qp_to_send_context(qp, sc5); struct hfi1_pportdata *ppd = ppd_from_ibp(ibp); u32 nwords; + hdr.hdr_type = HFI1_PKT_TYPE_16B; /* Populate length */ nwords = ((hfi1_get_16b_padding(hwords << 2, 0) + SIZE_OF_LT) >> 2) + SIZE_OF_CRC; if (old_grh) { - struct ib_grh *grh = &hdr.u.l.grh; + struct ib_grh *grh = &hdr.opah.u.l.grh; grh->version_tclass_flow = old_grh->version_tclass_flow; grh->paylen = cpu_to_be16( @@ -675,11 +677,11 @@ void return_cnp_16B(struct hfi1_ibport *ibp, struct rvt_qp *qp, grh->hop_limit = 0xff; grh->sgid = old_grh->dgid; grh->dgid = old_grh->sgid; - ohdr = &hdr.u.l.oth; + ohdr = &hdr.opah.u.l.oth; l4 = OPA_16B_L4_IB_GLOBAL; hwords += sizeof(struct ib_grh) / sizeof(u32); } else { - ohdr = &hdr.u.oth; + ohdr = &hdr.opah.u.oth; l4 = OPA_16B_L4_IB_LOCAL; } @@ -693,7 +695,7 @@ void return_cnp_16B(struct hfi1_ibport *ibp, struct rvt_qp *qp, /* Convert dwords to flits */ len = (hwords + nwords) >> 1; - hfi1_make_16b_hdr(&hdr, slid, dlid, len, pkey, 1, 0, l4, sc5); + hfi1_make_16b_hdr(&hdr.opah, slid, dlid, len, pkey, 1, 0, l4, sc5); plen = 2 /* PBC */ + hwords + nwords; pbc_flags |= PBC_PACKET_BYPASS | PBC_INSERT_BYPASS_ICRC; @@ -701,9 +703,11 @@ void return_cnp_16B(struct hfi1_ibport *ibp, struct rvt_qp *qp, pbc = create_pbc(ppd, pbc_flags, qp->srate_mbps, vl, plen); if (ctxt) { pbuf = sc_buffer_alloc(ctxt, plen, NULL, NULL); - if (pbuf) + if (pbuf) { + trace_pio_output_ibhdr(ppd->dd, &hdr, sc5); ppd->dd->pio_inline_send(ppd->dd, pbuf, pbc, &hdr, hwords); + } } } @@ -715,14 +719,15 @@ void return_cnp(struct hfi1_ibport *ibp, struct rvt_qp *qp, u32 remote_qpn, u32 bth0, plen, vl, hwords = 5; u16 lrh0; u8 sl = ibp->sc_to_sl[sc5]; - struct ib_header hdr; + struct hfi1_opa_header hdr; struct ib_other_headers *ohdr; struct pio_buf *pbuf; struct send_context *ctxt = qp_to_send_context(qp, sc5); struct hfi1_pportdata *ppd = ppd_from_ibp(ibp); + hdr.hdr_type = HFI1_PKT_TYPE_9B; if (old_grh) { - struct ib_grh *grh = &hdr.u.l.grh; + struct ib_grh *grh = &hdr.ibh.u.l.grh; grh->version_tclass_flow = old_grh->version_tclass_flow; grh->paylen = cpu_to_be16( @@ -730,11 +735,11 @@ void return_cnp(struct hfi1_ibport *ibp, struct rvt_qp *qp, u32 remote_qpn, grh->hop_limit = 0xff; grh->sgid = old_grh->dgid; grh->dgid = old_grh->sgid; - ohdr = &hdr.u.l.oth; + ohdr = &hdr.ibh.u.l.oth; lrh0 = HFI1_LRH_GRH; hwords += sizeof(struct ib_grh) / sizeof(u32); } else { - ohdr = &hdr.u.oth; + ohdr = &hdr.ibh.u.oth; lrh0 = HFI1_LRH_BTH; } @@ -746,16 +751,18 @@ void return_cnp(struct hfi1_ibport *ibp, struct rvt_qp *qp, u32 remote_qpn, ohdr->bth[1] = cpu_to_be32(remote_qpn | (1 << IB_BECN_SHIFT)); ohdr->bth[2] = 0; /* PSN 0 */ - hfi1_make_ib_hdr(&hdr, lrh0, hwords + SIZE_OF_CRC, dlid, slid); + hfi1_make_ib_hdr(&hdr.ibh, lrh0, hwords + SIZE_OF_CRC, dlid, slid); plen = 2 /* PBC */ + hwords; pbc_flags |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT); vl = sc_to_vlt(ppd->dd, sc5); pbc = create_pbc(ppd, pbc_flags, qp->srate_mbps, vl, plen); if (ctxt) { pbuf = sc_buffer_alloc(ctxt, plen, NULL, NULL); - if (pbuf) + if (pbuf) { + trace_pio_output_ibhdr(ppd->dd, &hdr, sc5); ppd->dd->pio_inline_send(ppd->dd, pbuf, pbc, &hdr, hwords); + } } } @@ -912,7 +919,7 @@ void hfi1_ud_rcv(struct hfi1_packet *packet) src_qp = hfi1_16B_get_src_qpn(packet->mgmt); } - process_ecn(qp, packet, (opcode != IB_OPCODE_CNP)); + process_ecn(qp, packet); /* * Get the number of bytes the message was padded by * and drop incomplete packets. diff --git a/drivers/infiniband/hw/hfi1/user_sdma.c b/drivers/infiniband/hw/hfi1/user_sdma.c index 3f0aadccd9f6..e5e7fad09f32 100644 --- a/drivers/infiniband/hw/hfi1/user_sdma.c +++ b/drivers/infiniband/hw/hfi1/user_sdma.c @@ -130,7 +130,6 @@ static int defer_packet_queue( { struct hfi1_user_sdma_pkt_q *pq = container_of(wait->iow, struct hfi1_user_sdma_pkt_q, busy); - struct hfi1_ibdev *dev = &pq->dd->verbs_dev; struct user_sdma_txreq *tx = container_of(txreq, struct user_sdma_txreq, txreq); @@ -144,10 +143,10 @@ static int defer_packet_queue( * it is supposed to be enqueued. */ xchg(&pq->state, SDMA_PKT_Q_DEFERRED); - write_seqlock(&dev->iowait_lock); + write_seqlock(&sde->waitlock); if (list_empty(&pq->busy.list)) iowait_queue(pkts_sent, &pq->busy, &sde->dmawait); - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sde->waitlock); return -EBUSY; eagain: return -EAGAIN; diff --git a/drivers/infiniband/hw/hfi1/verbs.c b/drivers/infiniband/hw/hfi1/verbs.c index 48e11e510358..28f754d2a844 100644 --- a/drivers/infiniband/hw/hfi1/verbs.c +++ b/drivers/infiniband/hw/hfi1/verbs.c @@ -765,7 +765,6 @@ static int pio_wait(struct rvt_qp *qp, { struct hfi1_qp_priv *priv = qp->priv; struct hfi1_devdata *dd = sc->dd; - struct hfi1_ibdev *dev = &dd->verbs_dev; unsigned long flags; int ret = 0; @@ -777,7 +776,7 @@ static int pio_wait(struct rvt_qp *qp, */ spin_lock_irqsave(&qp->s_lock, flags); if (ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) { - write_seqlock(&dev->iowait_lock); + write_seqlock(&sc->waitlock); list_add_tail(&ps->s_txreq->txreq.list, &ps->wait->tx_head); if (list_empty(&priv->s_iowait.list)) { @@ -790,14 +789,14 @@ static int pio_wait(struct rvt_qp *qp, was_empty = list_empty(&sc->piowait); iowait_queue(ps->pkts_sent, &priv->s_iowait, &sc->piowait); - priv->s_iowait.lock = &dev->iowait_lock; + priv->s_iowait.lock = &sc->waitlock; trace_hfi1_qpsleep(qp, RVT_S_WAIT_PIO); rvt_get_qp(qp); /* counting: only call wantpiobuf_intr if first user */ if (was_empty) hfi1_sc_wantpiobuf_intr(sc, 1); } - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sc->waitlock); hfi1_qp_unbusy(qp, ps->wait); ret = -EBUSY; } @@ -919,6 +918,8 @@ int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps, if (slen > len) slen = len; + if (slen > ss->sge.sge_length) + slen = ss->sge.sge_length; rvt_update_sge(ss, slen, false); seg_pio_copy_mid(pbuf, addr, slen); len -= slen; @@ -1704,6 +1705,7 @@ int hfi1_register_ib_device(struct hfi1_devdata *dd) dd->verbs_dev.rdi.dparms.max_mad_size = OPA_MGMT_MAD_SIZE; dd->verbs_dev.rdi.driver_f.qp_priv_alloc = qp_priv_alloc; + dd->verbs_dev.rdi.driver_f.qp_priv_init = hfi1_qp_priv_init; dd->verbs_dev.rdi.driver_f.qp_priv_free = qp_priv_free; dd->verbs_dev.rdi.driver_f.free_all_qps = free_all_qps; dd->verbs_dev.rdi.driver_f.notify_qp_reset = notify_qp_reset; diff --git a/drivers/infiniband/hw/hfi1/verbs.h b/drivers/infiniband/hw/hfi1/verbs.h index 64c9054db5f3..1ad0b14bdb3c 100644 --- a/drivers/infiniband/hw/hfi1/verbs.h +++ b/drivers/infiniband/hw/hfi1/verbs.h @@ -71,6 +71,7 @@ struct hfi1_devdata; struct hfi1_packet; #include "iowait.h" +#include "tid_rdma.h" #define HFI1_MAX_RDMA_ATOMIC 16 @@ -156,6 +157,7 @@ struct hfi1_qp_priv { struct hfi1_ahg_info *s_ahg; /* ahg info for next header */ struct sdma_engine *s_sde; /* current sde */ struct send_context *s_sendcontext; /* current sendcontext */ + struct hfi1_ctxtdata *rcd; /* QP's receive context */ u8 s_sc; /* SC[0..4] for next packet */ struct iowait s_iowait; struct rvt_qp *owner; diff --git a/drivers/infiniband/hw/hfi1/vnic_main.c b/drivers/infiniband/hw/hfi1/vnic_main.c index c9876d9e3cb9..a922db58be14 100644 --- a/drivers/infiniband/hw/hfi1/vnic_main.c +++ b/drivers/infiniband/hw/hfi1/vnic_main.c @@ -816,14 +816,14 @@ struct net_device *hfi1_vnic_alloc_rn(struct ib_device *device, size = sizeof(struct opa_vnic_rdma_netdev) + sizeof(*vinfo); netdev = alloc_netdev_mqs(size, name, name_assign_type, setup, - chip_sdma_engines(dd), dd->num_vnic_contexts); + dd->num_sdma, dd->num_vnic_contexts); if (!netdev) return ERR_PTR(-ENOMEM); rn = netdev_priv(netdev); vinfo = opa_vnic_dev_priv(netdev); vinfo->dd = dd; - vinfo->num_tx_q = chip_sdma_engines(dd); + vinfo->num_tx_q = dd->num_sdma; vinfo->num_rx_q = dd->num_vnic_contexts; vinfo->netdev = netdev; rn->free_rdma_netdev = hfi1_vnic_free_rn; diff --git a/drivers/infiniband/hw/hfi1/vnic_sdma.c b/drivers/infiniband/hw/hfi1/vnic_sdma.c index 97bd940a056a..1f81c480e028 100644 --- a/drivers/infiniband/hw/hfi1/vnic_sdma.c +++ b/drivers/infiniband/hw/hfi1/vnic_sdma.c @@ -57,7 +57,6 @@ #define HFI1_VNIC_TXREQ_NAME_LEN 32 #define HFI1_VNIC_SDMA_DESC_WTRMRK 64 -#define HFI1_VNIC_SDMA_RETRY_COUNT 1 /* * struct vnic_txreq - VNIC transmit descriptor @@ -67,7 +66,6 @@ * @pad: pad buffer * @plen: pad length * @pbc_val: pbc value - * @retry_count: tx retry count */ struct vnic_txreq { struct sdma_txreq txreq; @@ -77,8 +75,6 @@ struct vnic_txreq { unsigned char pad[HFI1_VNIC_MAX_PAD]; u16 plen; __le64 pbc_val; - - u32 retry_count; }; static void vnic_sdma_complete(struct sdma_txreq *txreq, @@ -196,7 +192,6 @@ int hfi1_vnic_send_dma(struct hfi1_devdata *dd, u8 q_idx, ret = build_vnic_tx_desc(sde, tx, pbc); if (unlikely(ret)) goto free_desc; - tx->retry_count = 0; ret = sdma_send_txreq(sde, iowait_get_ib_work(&vnic_sdma->wait), &tx->txreq, vnic_sdma->pkts_sent); @@ -237,18 +232,17 @@ static int hfi1_vnic_sdma_sleep(struct sdma_engine *sde, { struct hfi1_vnic_sdma *vnic_sdma = container_of(wait->iow, struct hfi1_vnic_sdma, wait); - struct hfi1_ibdev *dev = &vnic_sdma->dd->verbs_dev; - struct vnic_txreq *tx = container_of(txreq, struct vnic_txreq, txreq); - if (sdma_progress(sde, seq, txreq)) - if (tx->retry_count++ < HFI1_VNIC_SDMA_RETRY_COUNT) - return -EAGAIN; + write_seqlock(&sde->waitlock); + if (sdma_progress(sde, seq, txreq)) { + write_sequnlock(&sde->waitlock); + return -EAGAIN; + } vnic_sdma->state = HFI1_VNIC_SDMA_Q_DEFERRED; - write_seqlock(&dev->iowait_lock); if (list_empty(&vnic_sdma->wait.list)) iowait_queue(pkts_sent, wait->iow, &sde->dmawait); - write_sequnlock(&dev->iowait_lock); + write_sequnlock(&sde->waitlock); return -EBUSY; } diff --git a/drivers/infiniband/hw/hns/Makefile b/drivers/infiniband/hw/hns/Makefile index cf03404b9d58..004c88b32e13 100644 --- a/drivers/infiniband/hw/hns/Makefile +++ b/drivers/infiniband/hw/hns/Makefile @@ -7,7 +7,7 @@ ccflags-y := -Idrivers/net/ethernet/hisilicon/hns3 obj-$(CONFIG_INFINIBAND_HNS) += hns-roce.o hns-roce-objs := hns_roce_main.o hns_roce_cmd.o hns_roce_pd.o \ hns_roce_ah.o hns_roce_hem.o hns_roce_mr.o hns_roce_qp.o \ - hns_roce_cq.o hns_roce_alloc.o hns_roce_db.o + hns_roce_cq.o hns_roce_alloc.o hns_roce_db.o hns_roce_srq.o obj-$(CONFIG_INFINIBAND_HNS_HIP06) += hns-roce-hw-v1.o hns-roce-hw-v1-objs := hns_roce_hw_v1.o obj-$(CONFIG_INFINIBAND_HNS_HIP08) += hns-roce-hw-v2.o diff --git a/drivers/infiniband/hw/hns/hns_roce_alloc.c b/drivers/infiniband/hw/hns/hns_roce_alloc.c index 46f65f9f59d0..6300033a448f 100644 --- a/drivers/infiniband/hw/hns/hns_roce_alloc.c +++ b/drivers/infiniband/hw/hns/hns_roce_alloc.c @@ -239,6 +239,8 @@ err_free: void hns_roce_cleanup_bitmap(struct hns_roce_dev *hr_dev) { + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) + hns_roce_cleanup_srq_table(hr_dev); hns_roce_cleanup_qp_table(hr_dev); hns_roce_cleanup_cq_table(hr_dev); hns_roce_cleanup_mr_table(hr_dev); diff --git a/drivers/infiniband/hw/hns/hns_roce_cmd.h b/drivers/infiniband/hw/hns/hns_roce_cmd.h index 9549ae51a0dd..927701df5eff 100644 --- a/drivers/infiniband/hw/hns/hns_roce_cmd.h +++ b/drivers/infiniband/hw/hns/hns_roce_cmd.h @@ -120,6 +120,10 @@ enum { HNS_ROCE_CMD_SQD2RTS_QP = 0x20, HNS_ROCE_CMD_2RST_QP = 0x21, HNS_ROCE_CMD_QUERY_QP = 0x22, + HNS_ROCE_CMD_SW2HW_SRQ = 0x70, + HNS_ROCE_CMD_MODIFY_SRQC = 0x72, + HNS_ROCE_CMD_QUERY_SRQC = 0x73, + HNS_ROCE_CMD_HW2SW_SRQ = 0x74, }; int hns_roce_cmd_mbox(struct hns_roce_dev *hr_dev, u64 in_param, u64 out_param, diff --git a/drivers/infiniband/hw/hns/hns_roce_common.h b/drivers/infiniband/hw/hns/hns_roce_common.h index 93d4b4ec002d..f4c92a7ac1ce 100644 --- a/drivers/infiniband/hw/hns/hns_roce_common.h +++ b/drivers/infiniband/hw/hns/hns_roce_common.h @@ -376,9 +376,6 @@ #define ROCEE_RX_CMQ_TAIL_REG 0x07024 #define ROCEE_RX_CMQ_HEAD_REG 0x07028 -#define ROCEE_VF_MB_CFG0_REG 0x40 -#define ROCEE_VF_MB_STATUS_REG 0x58 - #define ROCEE_VF_EQ_DB_CFG0_REG 0x238 #define ROCEE_VF_EQ_DB_CFG1_REG 0x23C diff --git a/drivers/infiniband/hw/hns/hns_roce_device.h b/drivers/infiniband/hw/hns/hns_roce_device.h index d39bdfdb5de9..779dd4c409cb 100644 --- a/drivers/infiniband/hw/hns/hns_roce_device.h +++ b/drivers/infiniband/hw/hns/hns_roce_device.h @@ -111,6 +111,9 @@ #define PAGES_SHIFT_24 24 #define PAGES_SHIFT_32 32 +#define HNS_ROCE_IDX_QUE_ENTRY_SZ 4 +#define SRQ_DB_REG 0x230 + enum { HNS_ROCE_SUPPORT_RQ_RECORD_DB = 1 << 0, HNS_ROCE_SUPPORT_SQ_RECORD_DB = 1 << 1, @@ -196,6 +199,7 @@ enum { HNS_ROCE_CAP_FLAG_RQ_INLINE = BIT(2), HNS_ROCE_CAP_FLAG_RECORD_DB = BIT(3), HNS_ROCE_CAP_FLAG_SQ_RECORD_DB = BIT(4), + HNS_ROCE_CAP_FLAG_SRQ = BIT(5), HNS_ROCE_CAP_FLAG_MW = BIT(7), HNS_ROCE_CAP_FLAG_FRMR = BIT(8), HNS_ROCE_CAP_FLAG_ATOMIC = BIT(10), @@ -204,6 +208,8 @@ enum { enum hns_roce_mtt_type { MTT_TYPE_WQE, MTT_TYPE_CQE, + MTT_TYPE_SRQWQE, + MTT_TYPE_IDX }; enum { @@ -339,6 +345,10 @@ struct hns_roce_mr_table { struct hns_roce_hem_table mtpt_table; struct hns_roce_buddy mtt_cqe_buddy; struct hns_roce_hem_table mtt_cqe_table; + struct hns_roce_buddy mtt_srqwqe_buddy; + struct hns_roce_hem_table mtt_srqwqe_table; + struct hns_roce_buddy mtt_idx_buddy; + struct hns_roce_hem_table mtt_idx_table; }; struct hns_roce_wq { @@ -429,9 +439,37 @@ struct hns_roce_cq { struct completion free; }; +struct hns_roce_idx_que { + struct hns_roce_buf idx_buf; + int entry_sz; + u32 buf_size; + struct ib_umem *umem; + struct hns_roce_mtt mtt; + u64 *bitmap; +}; + struct hns_roce_srq { struct ib_srq ibsrq; - int srqn; + void (*event)(struct hns_roce_srq *srq, enum hns_roce_event event); + unsigned long srqn; + int max; + int max_gs; + int wqe_shift; + void __iomem *db_reg_l; + + atomic_t refcount; + struct completion free; + + struct hns_roce_buf buf; + u64 *wrid; + struct ib_umem *umem; + struct hns_roce_mtt mtt; + struct hns_roce_idx_que idx_que; + spinlock_t lock; + int head; + int tail; + u16 wqe_ctr; + struct mutex mutex; }; struct hns_roce_uar_table { @@ -453,6 +491,12 @@ struct hns_roce_cq_table { struct hns_roce_hem_table table; }; +struct hns_roce_srq_table { + struct hns_roce_bitmap bitmap; + struct xarray xa; + struct hns_roce_hem_table table; +}; + struct hns_roce_raq_table { struct hns_roce_buf_list *e_raq_buf; }; @@ -603,6 +647,12 @@ struct hns_roce_aeqe { } qp_event; struct { + __le32 srq; + u32 rsv0; + u32 rsv1; + } srq_event; + + struct { __le32 cq; u32 rsv0; u32 rsv1; @@ -679,7 +729,12 @@ struct hns_roce_caps { u32 max_extend_sg; int num_qps; /* 256k */ int reserved_qps; + u32 max_srq_sg; + int num_srqs; u32 max_wqes; /* 16k */ + u32 max_srqs; + u32 max_srq_wrs; + u32 max_srq_sges; u32 max_sq_desc_sz; /* 64 */ u32 max_rq_desc_sz; /* 64 */ u32 max_srq_desc_sz; @@ -690,12 +745,16 @@ struct hns_roce_caps { int min_cqes; u32 min_wqes; int reserved_cqs; + int reserved_srqs; + u32 max_srqwqes; int num_aeq_vectors; /* 1 */ int num_comp_vectors; int num_other_vectors; int num_mtpts; u32 num_mtt_segs; u32 num_cqe_segs; + u32 num_srqwqe_segs; + u32 num_idx_segs; int reserved_mrws; int reserved_uars; int num_pds; @@ -709,6 +768,8 @@ struct hns_roce_caps { int irrl_entry_sz; int trrl_entry_sz; int cqc_entry_sz; + int srqc_entry_sz; + int idx_entry_sz; u32 pbl_ba_pg_sz; u32 pbl_buf_pg_sz; u32 pbl_hop_num; @@ -737,6 +798,12 @@ struct hns_roce_caps { u32 cqe_ba_pg_sz; u32 cqe_buf_pg_sz; u32 cqe_hop_num; + u32 srqwqe_ba_pg_sz; + u32 srqwqe_buf_pg_sz; + u32 srqwqe_hop_num; + u32 idx_ba_pg_sz; + u32 idx_buf_pg_sz; + u32 idx_hop_num; u32 eqe_ba_pg_sz; u32 eqe_buf_pg_sz; u32 eqe_hop_num; @@ -805,6 +872,17 @@ 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 (*modify_srq)(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr, + enum ib_srq_attr_mask srq_attr_mask, + struct ib_udata *udata); + int (*query_srq)(struct ib_srq *ibsrq, struct ib_srq_attr *attr); + int (*post_srq_recv)(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr); }; struct hns_roce_dev { @@ -839,6 +917,7 @@ struct hns_roce_dev { struct hns_roce_uar_table uar_table; struct hns_roce_mr_table mr_table; struct hns_roce_cq_table cq_table; + struct hns_roce_srq_table srq_table; struct hns_roce_qp_table qp_table; struct hns_roce_eq_table eq_table; @@ -951,12 +1030,14 @@ int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev); int hns_roce_init_eq_table(struct hns_roce_dev *hr_dev); int 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); void hns_roce_cleanup_pd_table(struct hns_roce_dev *hr_dev); void hns_roce_cleanup_mr_table(struct hns_roce_dev *hr_dev); void hns_roce_cleanup_eq_table(struct hns_roce_dev *hr_dev); void hns_roce_cleanup_cq_table(struct hns_roce_dev *hr_dev); void hns_roce_cleanup_qp_table(struct hns_roce_dev *hr_dev); +void hns_roce_cleanup_srq_table(struct hns_roce_dev *hr_dev); int hns_roce_bitmap_alloc(struct hns_roce_bitmap *bitmap, unsigned long *obj); void hns_roce_bitmap_free(struct hns_roce_bitmap *bitmap, unsigned long obj, @@ -1011,6 +1092,14 @@ int hns_roce_buf_alloc(struct hns_roce_dev *hr_dev, u32 size, u32 max_direct, int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev, struct hns_roce_mtt *mtt, struct ib_umem *umem); +struct ib_srq *hns_roce_create_srq(struct ib_pd *pd, + struct ib_srq_init_attr *srq_init_attr, + struct ib_udata *udata); +int hns_roce_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr, + enum ib_srq_attr_mask srq_attr_mask, + struct ib_udata *udata); +int hns_roce_destroy_srq(struct ib_srq *ibsrq); + struct ib_qp *hns_roce_create_qp(struct ib_pd *ib_pd, struct ib_qp_init_attr *init_attr, struct ib_udata *udata); @@ -1052,6 +1141,7 @@ void hns_roce_free_db(struct hns_roce_dev *hr_dev, struct hns_roce_db *db); void hns_roce_cq_completion(struct hns_roce_dev *hr_dev, u32 cqn); void hns_roce_cq_event(struct hns_roce_dev *hr_dev, u32 cqn, int event_type); void hns_roce_qp_event(struct hns_roce_dev *hr_dev, u32 qpn, int event_type); +void hns_roce_srq_event(struct hns_roce_dev *hr_dev, u32 srqn, int event_type); int hns_get_gid_index(struct hns_roce_dev *hr_dev, u8 port, int gid_index); int hns_roce_init(struct hns_roce_dev *hr_dev); void hns_roce_exit(struct hns_roce_dev *hr_dev); diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.c b/drivers/infiniband/hw/hns/hns_roce_hem.c index f6faefed96e8..4cdbcafa5915 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.c +++ b/drivers/infiniband/hw/hns/hns_roce_hem.c @@ -46,7 +46,9 @@ bool hns_roce_check_whether_mhop(struct hns_roce_dev *hr_dev, u32 type) (hr_dev->caps.cqc_hop_num && type == HEM_TYPE_CQC) || (hr_dev->caps.srqc_hop_num && type == HEM_TYPE_SRQC) || (hr_dev->caps.cqe_hop_num && type == HEM_TYPE_CQE) || - (hr_dev->caps.mtt_hop_num && type == HEM_TYPE_MTT)) + (hr_dev->caps.mtt_hop_num && type == HEM_TYPE_MTT) || + (hr_dev->caps.srqwqe_hop_num && type == HEM_TYPE_SRQWQE) || + (hr_dev->caps.idx_hop_num && type == HEM_TYPE_IDX)) return true; return false; @@ -147,6 +149,22 @@ int hns_roce_calc_hem_mhop(struct hns_roce_dev *hr_dev, mhop->ba_l0_num = mhop->bt_chunk_size / 8; mhop->hop_num = hr_dev->caps.cqe_hop_num; break; + case HEM_TYPE_SRQWQE: + mhop->buf_chunk_size = 1 << (hr_dev->caps.srqwqe_buf_pg_sz + + PAGE_SHIFT); + mhop->bt_chunk_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + + PAGE_SHIFT); + mhop->ba_l0_num = mhop->bt_chunk_size / 8; + mhop->hop_num = hr_dev->caps.srqwqe_hop_num; + break; + case HEM_TYPE_IDX: + mhop->buf_chunk_size = 1 << (hr_dev->caps.idx_buf_pg_sz + + PAGE_SHIFT); + mhop->bt_chunk_size = 1 << (hr_dev->caps.idx_ba_pg_sz + + PAGE_SHIFT); + mhop->ba_l0_num = mhop->bt_chunk_size / 8; + mhop->hop_num = hr_dev->caps.idx_hop_num; + break; default: dev_err(dev, "Table %d not support multi-hop addressing!\n", table->type); @@ -906,6 +924,18 @@ int hns_roce_init_hem_table(struct hns_roce_dev *hr_dev, bt_chunk_size = buf_chunk_size; hop_num = hr_dev->caps.cqe_hop_num; break; + case HEM_TYPE_SRQWQE: + buf_chunk_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + + PAGE_SHIFT); + bt_chunk_size = buf_chunk_size; + hop_num = hr_dev->caps.srqwqe_hop_num; + break; + case HEM_TYPE_IDX: + buf_chunk_size = 1 << (hr_dev->caps.idx_ba_pg_sz + + PAGE_SHIFT); + bt_chunk_size = buf_chunk_size; + hop_num = hr_dev->caps.idx_hop_num; + break; default: dev_err(dev, "Table %d not support to init hem table here!\n", @@ -1041,6 +1071,15 @@ void hns_roce_cleanup_hem_table(struct hns_roce_dev *hr_dev, void hns_roce_cleanup_hem(struct hns_roce_dev *hr_dev) { + if ((hr_dev->caps.num_idx_segs)) + hns_roce_cleanup_hem_table(hr_dev, + &hr_dev->mr_table.mtt_idx_table); + if (hr_dev->caps.num_srqwqe_segs) + hns_roce_cleanup_hem_table(hr_dev, + &hr_dev->mr_table.mtt_srqwqe_table); + if (hr_dev->caps.srqc_entry_sz) + hns_roce_cleanup_hem_table(hr_dev, + &hr_dev->srq_table.table); hns_roce_cleanup_hem_table(hr_dev, &hr_dev->cq_table.table); if (hr_dev->caps.trrl_entry_sz) hns_roce_cleanup_hem_table(hr_dev, diff --git a/drivers/infiniband/hw/hns/hns_roce_hem.h b/drivers/infiniband/hw/hns/hns_roce_hem.h index e8850d59e780..a650278c6fbd 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hem.h +++ b/drivers/infiniband/hw/hns/hns_roce_hem.h @@ -48,6 +48,8 @@ enum { /* UNMAP HEM */ HEM_TYPE_MTT, HEM_TYPE_CQE, + HEM_TYPE_SRQWQE, + HEM_TYPE_IDX, HEM_TYPE_IRRL, HEM_TYPE_TRRL, }; diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c index 3beb1523e17c..5007fe7af2f9 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.c +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.c @@ -1082,6 +1082,33 @@ static int hns_roce_query_pf_resource(struct hns_roce_dev *hr_dev) return 0; } +static int hns_roce_set_vf_switch_param(struct hns_roce_dev *hr_dev, + int vf_id) +{ + struct hns_roce_cmq_desc desc; + struct hns_roce_vf_switch *swt; + int ret; + + swt = (struct hns_roce_vf_switch *)desc.data; + hns_roce_cmq_setup_basic_desc(&desc, HNS_SWITCH_PARAMETER_CFG, true); + swt->rocee_sel |= cpu_to_le16(HNS_ICL_SWITCH_CMD_ROCEE_SEL); + roce_set_field(swt->fun_id, + VF_SWITCH_DATA_FUN_ID_VF_ID_M, + VF_SWITCH_DATA_FUN_ID_VF_ID_S, + vf_id); + ret = hns_roce_cmq_send(hr_dev, &desc, 1); + if (ret) + return ret; + desc.flag = + cpu_to_le16(HNS_ROCE_CMD_FLAG_NO_INTR | HNS_ROCE_CMD_FLAG_IN); + desc.flag &= cpu_to_le16(~HNS_ROCE_CMD_FLAG_WR); + roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_LPBK_S, 1); + roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_LCL_LPBK_S, 1); + roce_set_bit(swt->cfg, VF_SWITCH_DATA_CFG_ALW_DST_OVRD_S, 1); + + return hns_roce_cmq_send(hr_dev, &desc, 1); +} + static int hns_roce_alloc_vf_resource(struct hns_roce_dev *hr_dev) { struct hns_roce_cmq_desc desc[2]; @@ -1269,6 +1296,15 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) return ret; } + if (hr_dev->pci_dev->revision == 0x21) { + ret = hns_roce_set_vf_switch_param(hr_dev, 0); + if (ret) { + dev_err(hr_dev->dev, + "Set function switch param fail, ret = %d.\n", + ret); + return ret; + } + } hr_dev->vendor_part_id = hr_dev->pci_dev->device; hr_dev->sys_image_guid = be64_to_cpu(hr_dev->ib_dev.node_guid); @@ -1276,11 +1312,14 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->num_qps = HNS_ROCE_V2_MAX_QP_NUM; caps->max_wqes = HNS_ROCE_V2_MAX_WQE_NUM; caps->num_cqs = HNS_ROCE_V2_MAX_CQ_NUM; + caps->num_srqs = HNS_ROCE_V2_MAX_SRQ_NUM; caps->max_cqes = HNS_ROCE_V2_MAX_CQE_NUM; + caps->max_srqwqes = HNS_ROCE_V2_MAX_SRQWQE_NUM; caps->max_sq_sg = HNS_ROCE_V2_MAX_SQ_SGE_NUM; caps->max_extend_sg = HNS_ROCE_V2_MAX_EXTEND_SGE_NUM; caps->max_rq_sg = HNS_ROCE_V2_MAX_RQ_SGE_NUM; caps->max_sq_inline = HNS_ROCE_V2_MAX_SQ_INLINE; + caps->max_srq_sg = HNS_ROCE_V2_MAX_SRQ_SGE_NUM; caps->num_uars = HNS_ROCE_V2_UAR_NUM; caps->phy_num_uars = HNS_ROCE_V2_PHY_UAR_NUM; caps->num_aeq_vectors = HNS_ROCE_V2_AEQE_VEC_NUM; @@ -1289,6 +1328,8 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->num_mtpts = HNS_ROCE_V2_MAX_MTPT_NUM; caps->num_mtt_segs = HNS_ROCE_V2_MAX_MTT_SEGS; caps->num_cqe_segs = HNS_ROCE_V2_MAX_CQE_SEGS; + caps->num_srqwqe_segs = HNS_ROCE_V2_MAX_SRQWQE_SEGS; + caps->num_idx_segs = HNS_ROCE_V2_MAX_IDX_SEGS; caps->num_pds = HNS_ROCE_V2_MAX_PD_NUM; caps->max_qp_init_rdma = HNS_ROCE_V2_MAX_QP_INIT_RDMA; caps->max_qp_dest_rdma = HNS_ROCE_V2_MAX_QP_DEST_RDMA; @@ -1299,8 +1340,10 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->irrl_entry_sz = HNS_ROCE_V2_IRRL_ENTRY_SZ; caps->trrl_entry_sz = HNS_ROCE_V2_TRRL_ENTRY_SZ; caps->cqc_entry_sz = HNS_ROCE_V2_CQC_ENTRY_SZ; + caps->srqc_entry_sz = HNS_ROCE_V2_SRQC_ENTRY_SZ; caps->mtpt_entry_sz = HNS_ROCE_V2_MTPT_ENTRY_SZ; caps->mtt_entry_sz = HNS_ROCE_V2_MTT_ENTRY_SZ; + caps->idx_entry_sz = 4; caps->cq_entry_sz = HNS_ROCE_V2_CQE_ENTRY_SIZE; caps->page_size_cap = HNS_ROCE_V2_PAGE_SIZE_SUPPORTED; caps->reserved_lkey = 0; @@ -1308,6 +1351,7 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->reserved_mrws = 1; caps->reserved_uars = 0; caps->reserved_cqs = 0; + caps->reserved_srqs = 0; caps->reserved_qps = HNS_ROCE_V2_RSV_QPS; caps->qpc_ba_pg_sz = 0; @@ -1331,6 +1375,12 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->cqe_ba_pg_sz = 0; caps->cqe_buf_pg_sz = 0; caps->cqe_hop_num = HNS_ROCE_CQE_HOP_NUM; + caps->srqwqe_ba_pg_sz = 0; + caps->srqwqe_buf_pg_sz = 0; + caps->srqwqe_hop_num = HNS_ROCE_SRQWQE_HOP_NUM; + caps->idx_ba_pg_sz = 0; + caps->idx_buf_pg_sz = 0; + caps->idx_hop_num = HNS_ROCE_IDX_HOP_NUM; caps->eqe_ba_pg_sz = 0; caps->eqe_buf_pg_sz = 0; caps->eqe_hop_num = HNS_ROCE_EQE_HOP_NUM; @@ -1354,8 +1404,13 @@ static int hns_roce_v2_profile(struct hns_roce_dev *hr_dev) caps->local_ca_ack_delay = 0; caps->max_mtu = IB_MTU_4096; + caps->max_srqs = HNS_ROCE_V2_MAX_SRQ; + caps->max_srq_wrs = HNS_ROCE_V2_MAX_SRQ_WR; + caps->max_srq_sges = HNS_ROCE_V2_MAX_SRQ_SGE; + if (hr_dev->pci_dev->revision == 0x21) - caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC; + caps->flags |= HNS_ROCE_CAP_FLAG_ATOMIC | + HNS_ROCE_CAP_FLAG_SRQ; ret = hns_roce_v2_set_bt(hr_dev); if (ret) @@ -1587,30 +1642,62 @@ static void hns_roce_v2_exit(struct hns_roce_dev *hr_dev) hns_roce_free_link_table(hr_dev, &priv->tsq); } +static int hns_roce_query_mbox_status(struct hns_roce_dev *hr_dev) +{ + struct hns_roce_cmq_desc desc; + struct hns_roce_mbox_status *mb_st = + (struct hns_roce_mbox_status *)desc.data; + enum hns_roce_cmd_return_status status; + + hns_roce_cmq_setup_basic_desc(&desc, HNS_ROCE_OPC_QUERY_MB_ST, true); + + status = hns_roce_cmq_send(hr_dev, &desc, 1); + if (status) + return status; + + return cpu_to_le32(mb_st->mb_status_hw_run); +} + static int hns_roce_v2_cmd_pending(struct hns_roce_dev *hr_dev) { - u32 status = readl(hr_dev->reg_base + ROCEE_VF_MB_STATUS_REG); + u32 status = hns_roce_query_mbox_status(hr_dev); return status >> HNS_ROCE_HW_RUN_BIT_SHIFT; } static int hns_roce_v2_cmd_complete(struct hns_roce_dev *hr_dev) { - u32 status = readl(hr_dev->reg_base + ROCEE_VF_MB_STATUS_REG); + u32 status = hns_roce_query_mbox_status(hr_dev); return status & HNS_ROCE_HW_MB_STATUS_MASK; } +static int hns_roce_mbox_post(struct hns_roce_dev *hr_dev, u64 in_param, + u64 out_param, u32 in_modifier, u8 op_modifier, + u16 op, u16 token, int event) +{ + struct hns_roce_cmq_desc desc; + struct hns_roce_post_mbox *mb = (struct hns_roce_post_mbox *)desc.data; + + hns_roce_cmq_setup_basic_desc(&desc, HNS_ROCE_OPC_POST_MB, false); + + mb->in_param_l = cpu_to_le64(in_param); + mb->in_param_h = cpu_to_le64(in_param) >> 32; + mb->out_param_l = cpu_to_le64(out_param); + mb->out_param_h = cpu_to_le64(out_param) >> 32; + mb->cmd_tag = cpu_to_le32(in_modifier << 8 | op); + mb->token_event_en = cpu_to_le32(event << 16 | token); + + return hns_roce_cmq_send(hr_dev, &desc, 1); +} + static int hns_roce_v2_post_mbox(struct hns_roce_dev *hr_dev, u64 in_param, u64 out_param, u32 in_modifier, u8 op_modifier, u16 op, u16 token, int event) { struct device *dev = hr_dev->dev; - u32 __iomem *hcr = (u32 __iomem *)(hr_dev->reg_base + - ROCEE_VF_MB_CFG0_REG); unsigned long end; - u32 val0 = 0; - u32 val1 = 0; + int ret; end = msecs_to_jiffies(HNS_ROCE_V2_GO_BIT_TIMEOUT_MSECS) + jiffies; while (hns_roce_v2_cmd_pending(hr_dev)) { @@ -1622,27 +1709,12 @@ static int hns_roce_v2_post_mbox(struct hns_roce_dev *hr_dev, u64 in_param, cond_resched(); } - roce_set_field(val0, HNS_ROCE_VF_MB4_TAG_MASK, - HNS_ROCE_VF_MB4_TAG_SHIFT, in_modifier); - roce_set_field(val0, HNS_ROCE_VF_MB4_CMD_MASK, - HNS_ROCE_VF_MB4_CMD_SHIFT, op); - roce_set_field(val1, HNS_ROCE_VF_MB5_EVENT_MASK, - HNS_ROCE_VF_MB5_EVENT_SHIFT, event); - roce_set_field(val1, HNS_ROCE_VF_MB5_TOKEN_MASK, - HNS_ROCE_VF_MB5_TOKEN_SHIFT, token); - - writeq(in_param, hcr + 0); - writeq(out_param, hcr + 2); - - /* Memory barrier */ - wmb(); - - writel(val0, hcr + 4); - writel(val1, hcr + 5); - - mmiowb(); + ret = hns_roce_mbox_post(hr_dev, in_param, out_param, in_modifier, + op_modifier, op, token, event); + if (ret) + dev_err(dev, "Post mailbox fail(%d)\n", ret); - return 0; + return ret; } static int hns_roce_v2_chk_mbox(struct hns_roce_dev *hr_dev, @@ -2007,6 +2079,27 @@ static struct hns_roce_v2_cqe *next_cqe_sw_v2(struct hns_roce_cq *hr_cq) return get_sw_cqe_v2(hr_cq, hr_cq->cons_index); } +static void *get_srq_wqe(struct hns_roce_srq *srq, int n) +{ + return hns_roce_buf_offset(&srq->buf, n << srq->wqe_shift); +} + +static void hns_roce_free_srq_wqe(struct hns_roce_srq *srq, int wqe_index) +{ + u32 bitmap_num; + int bit_num; + + /* always called with interrupts disabled. */ + spin_lock(&srq->lock); + + bitmap_num = wqe_index / (sizeof(u64) * 8); + bit_num = wqe_index % (sizeof(u64) * 8); + srq->idx_que.bitmap[bitmap_num] |= (1ULL << bit_num); + srq->tail++; + + spin_unlock(&srq->lock); +} + static void hns_roce_v2_cq_set_ci(struct hns_roce_cq *hr_cq, u32 cons_index) { *hr_cq->set_ci_db = cons_index & 0xffffff; @@ -2018,6 +2111,7 @@ static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, struct hns_roce_v2_cqe *cqe, *dest; u32 prod_index; int nfreed = 0; + int wqe_index; u8 owner_bit; for (prod_index = hr_cq->cons_index; get_sw_cqe_v2(hr_cq, prod_index); @@ -2035,7 +2129,13 @@ static void __hns_roce_v2_cq_clean(struct hns_roce_cq *hr_cq, u32 qpn, if ((roce_get_field(cqe->byte_16, V2_CQE_BYTE_16_LCL_QPN_M, V2_CQE_BYTE_16_LCL_QPN_S) & HNS_ROCE_V2_CQE_QPN_MASK) == qpn) { - /* In v1 engine, not support SRQ */ + if (srq && + roce_get_bit(cqe->byte_4, V2_CQE_BYTE_4_S_R_S)) { + wqe_index = roce_get_field(cqe->byte_4, + V2_CQE_BYTE_4_WQE_INDX_M, + V2_CQE_BYTE_4_WQE_INDX_S); + hns_roce_free_srq_wqe(srq, wqe_index); + } ++nfreed; } else if (nfreed) { dest = get_cqe_v2(hr_cq, (prod_index + nfreed) & @@ -2212,6 +2312,7 @@ static int hns_roce_handle_recv_inl_wqe(struct hns_roce_v2_cqe *cqe, static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, struct hns_roce_qp **cur_qp, struct ib_wc *wc) { + struct hns_roce_srq *srq = NULL; struct hns_roce_dev *hr_dev; struct hns_roce_v2_cqe *cqe; struct hns_roce_qp *hr_qp; @@ -2254,6 +2355,37 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, wc->qp = &(*cur_qp)->ibqp; wc->vendor_err = 0; + if (is_send) { + wq = &(*cur_qp)->sq; + if ((*cur_qp)->sq_signal_bits) { + /* + * If sg_signal_bit is 1, + * firstly tail pointer updated to wqe + * which current cqe correspond to + */ + wqe_ctr = (u16)roce_get_field(cqe->byte_4, + V2_CQE_BYTE_4_WQE_INDX_M, + V2_CQE_BYTE_4_WQE_INDX_S); + wq->tail += (wqe_ctr - (u16)wq->tail) & + (wq->wqe_cnt - 1); + } + + wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; + ++wq->tail; + } else if ((*cur_qp)->ibqp.srq) { + srq = to_hr_srq((*cur_qp)->ibqp.srq); + wqe_ctr = le16_to_cpu(roce_get_field(cqe->byte_4, + V2_CQE_BYTE_4_WQE_INDX_M, + V2_CQE_BYTE_4_WQE_INDX_S)); + wc->wr_id = srq->wrid[wqe_ctr]; + hns_roce_free_srq_wqe(srq, wqe_ctr); + } else { + /* Update tail pointer, record wr_id */ + wq = &(*cur_qp)->rq; + wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; + ++wq->tail; + } + status = roce_get_field(cqe->byte_4, V2_CQE_BYTE_4_STATUS_M, V2_CQE_BYTE_4_STATUS_S); switch (status & HNS_ROCE_V2_CQE_STATUS_MASK) { @@ -2373,23 +2505,6 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, wc->status = IB_WC_GENERAL_ERR; break; } - - wq = &(*cur_qp)->sq; - if ((*cur_qp)->sq_signal_bits) { - /* - * If sg_signal_bit is 1, - * firstly tail pointer updated to wqe - * which current cqe correspond to - */ - wqe_ctr = (u16)roce_get_field(cqe->byte_4, - V2_CQE_BYTE_4_WQE_INDX_M, - V2_CQE_BYTE_4_WQE_INDX_S); - wq->tail += (wqe_ctr - (u16)wq->tail) & - (wq->wqe_cnt - 1); - } - - wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; - ++wq->tail; } else { /* RQ correspond to CQE */ wc->byte_len = le32_to_cpu(cqe->byte_cnt); @@ -2434,11 +2549,6 @@ static int hns_roce_v2_poll_one(struct hns_roce_cq *hr_cq, return -EAGAIN; } - /* Update tail pointer, record wr_id */ - wq = &(*cur_qp)->rq; - wc->wr_id = wq->wrid[wq->tail & (wq->wqe_cnt - 1)]; - ++wq->tail; - wc->sl = (u8)roce_get_field(cqe->byte_32, V2_CQE_BYTE_32_SL_M, V2_CQE_BYTE_32_SL_S); wc->src_qp = (u8)roce_get_field(cqe->byte_32, @@ -2747,6 +2857,8 @@ static void modify_qp_reset_to_init(struct ib_qp *ibqp, roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, + (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || + hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || ibqp->srq) ? 0 : ilog2((unsigned int)hr_qp->rq.wqe_cnt)); roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, 0); @@ -3088,6 +3200,8 @@ static void modify_qp_init_to_init(struct ib_qp *ibqp, roce_set_field(context->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, + (hr_qp->ibqp.qp_type == IB_QPT_XRC_INI || + hr_qp->ibqp.qp_type == IB_QPT_XRC_TGT || ibqp->srq) ? 0 : ilog2((unsigned int)hr_qp->rq.wqe_cnt)); roce_set_field(qpc_mask->byte_20_smac_sgid_idx, V2_QPC_BYTE_20_RQ_SHIFT_M, V2_QPC_BYTE_20_RQ_SHIFT_S, 0); @@ -3601,6 +3715,21 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp, return 0; } +static inline bool hns_roce_v2_check_qp_stat(enum ib_qp_state cur_state, + enum ib_qp_state new_state) +{ + + if ((cur_state != IB_QPS_RESET && + (new_state == IB_QPS_ERR || new_state == IB_QPS_RESET)) || + ((cur_state == IB_QPS_RTS || cur_state == IB_QPS_SQD) && + (new_state == IB_QPS_RTS || new_state == IB_QPS_SQD)) || + (cur_state == IB_QPS_SQE && new_state == IB_QPS_RTS)) + return true; + + return false; + +} + static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, const struct ib_qp_attr *attr, int attr_mask, enum ib_qp_state cur_state, @@ -3626,6 +3755,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, */ memset(qpc_mask, 0xff, sizeof(*qpc_mask)); if (cur_state == IB_QPS_RESET && new_state == IB_QPS_INIT) { + memset(qpc_mask, 0, sizeof(*qpc_mask)); modify_qp_reset_to_init(ibqp, attr, attr_mask, context, qpc_mask); } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_INIT) { @@ -3641,21 +3771,7 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, qpc_mask); if (ret) goto out; - } else if ((cur_state == IB_QPS_RTS && new_state == IB_QPS_RTS) || - (cur_state == IB_QPS_SQE && new_state == IB_QPS_RTS) || - (cur_state == IB_QPS_RTS && new_state == IB_QPS_SQD) || - (cur_state == IB_QPS_SQD && new_state == IB_QPS_SQD) || - (cur_state == IB_QPS_SQD && new_state == IB_QPS_RTS) || - (cur_state == IB_QPS_INIT && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_RTR && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_RTS && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_ERR && new_state == IB_QPS_RESET) || - (cur_state == IB_QPS_INIT && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_RTR && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_RTS && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_SQD && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_SQE && new_state == IB_QPS_ERR) || - (cur_state == IB_QPS_ERR && new_state == IB_QPS_ERR)) { + } else if (hns_roce_v2_check_qp_stat(cur_state, new_state)) { /* Nothing */ ; } else { @@ -3789,6 +3905,11 @@ static int hns_roce_v2_modify_qp(struct ib_qp *ibqp, if (attr_mask & (IB_QP_ACCESS_FLAGS | IB_QP_MAX_DEST_RD_ATOMIC)) set_access_flags(hr_qp, context, qpc_mask, attr, attr_mask); + roce_set_bit(context->byte_108_rx_reqepsn, V2_QPC_BYTE_108_INV_CREDIT_S, + ibqp->srq ? 1 : 0); + roce_set_bit(qpc_mask->byte_108_rx_reqepsn, + V2_QPC_BYTE_108_INV_CREDIT_S, 0); + /* Every status migrate must change state */ roce_set_field(context->byte_60_qpst_tempid, V2_QPC_BYTE_60_QP_ST_M, V2_QPC_BYTE_60_QP_ST_S, new_state); @@ -4074,7 +4195,8 @@ static int hns_roce_v2_destroy_qp_common(struct hns_roce_dev *hr_dev, hns_roce_free_db(hr_dev, &hr_qp->rdb); } - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) { + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && + hr_qp->rq.wqe_cnt) { kfree(hr_qp->rq_inl_buf.wqe_list[0].sg_list); kfree(hr_qp->rq_inl_buf.wqe_list); } @@ -4384,6 +4506,7 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, int aeqe_found = 0; int event_type; int sub_type; + u32 srqn; u32 qpn; u32 cqn; @@ -4406,6 +4529,9 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, cqn = roce_get_field(aeqe->event.cq_event.cq, HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_M, HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_S); + srqn = roce_get_field(aeqe->event.srq_event.srq, + HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_M, + HNS_ROCE_V2_AEQE_EVENT_QUEUE_NUM_S); switch (event_type) { case HNS_ROCE_EVENT_TYPE_PATH_MIG: @@ -4413,13 +4539,14 @@ static int hns_roce_v2_aeq_int(struct hns_roce_dev *hr_dev, case HNS_ROCE_EVENT_TYPE_COMM_EST: case HNS_ROCE_EVENT_TYPE_SQ_DRAINED: case HNS_ROCE_EVENT_TYPE_WQ_CATAS_ERROR: + case HNS_ROCE_EVENT_TYPE_SRQ_LAST_WQE_REACH: case HNS_ROCE_EVENT_TYPE_INV_REQ_LOCAL_WQ_ERROR: case HNS_ROCE_EVENT_TYPE_LOCAL_WQ_ACCESS_ERROR: hns_roce_qp_event(hr_dev, qpn, event_type); break; case HNS_ROCE_EVENT_TYPE_SRQ_LIMIT_REACH: - case HNS_ROCE_EVENT_TYPE_SRQ_LAST_WQE_REACH: case HNS_ROCE_EVENT_TYPE_SRQ_CATAS_ERROR: + hns_roce_srq_event(hr_dev, srqn, event_type); break; case HNS_ROCE_EVENT_TYPE_CQ_ACCESS_ERROR: case HNS_ROCE_EVENT_TYPE_CQ_OVERFLOW: @@ -4964,13 +5091,12 @@ static int hns_roce_mhop_alloc_eq(struct hns_roce_dev *hr_dev, eqe_alloc = i * (buf_chk_sz / eq->eqe_size); size = (eq->entries - eqe_alloc) * eq->eqe_size; } - eq->buf[i] = dma_alloc_coherent(dev, size, + eq->buf[i] = dma_zalloc_coherent(dev, size, &(eq->buf_dma[i]), GFP_KERNEL); if (!eq->buf[i]) goto err_dma_alloc_buf; - memset(eq->buf[i], 0, size); *(eq->bt_l0 + i) = eq->buf_dma[i]; eq_buf_cnt++; @@ -5000,13 +5126,12 @@ static int hns_roce_mhop_alloc_eq(struct hns_roce_dev *hr_dev, size = (eq->entries - eqe_alloc) * eq->eqe_size; } - eq->buf[idx] = dma_alloc_coherent(dev, size, + eq->buf[idx] = dma_zalloc_coherent(dev, size, &(eq->buf_dma[idx]), GFP_KERNEL); if (!eq->buf[idx]) goto err_dma_alloc_buf; - memset(eq->buf[idx], 0, size); *(eq->bt_l1[i] + j) = eq->buf_dma[idx]; eq_buf_cnt++; @@ -5116,7 +5241,7 @@ static int hns_roce_v2_create_eq(struct hns_roce_dev *hr_dev, goto free_cmd_mbox; } - eq->buf_list->buf = dma_alloc_coherent(dev, buf_chk_sz, + eq->buf_list->buf = dma_zalloc_coherent(dev, buf_chk_sz, &(eq->buf_list->map), GFP_KERNEL); if (!eq->buf_list->buf) { @@ -5124,7 +5249,6 @@ static int hns_roce_v2_create_eq(struct hns_roce_dev *hr_dev, goto err_alloc_buf; } - memset(eq->buf_list->buf, 0, buf_chk_sz); } else { ret = hns_roce_mhop_alloc_eq(hr_dev, eq); if (ret) { @@ -5332,6 +5456,284 @@ static void hns_roce_v2_cleanup_eq_table(struct hns_roce_dev *hr_dev) destroy_workqueue(hr_dev->irq_workq); } +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) +{ + struct hns_roce_srq_context *srq_context; + + 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, + (hr_dev->caps.srqwqe_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 : + hr_dev->caps.srqwqe_hop_num)); + roce_set_field(srq_context->byte_4_srqn_srqst, + SRQC_BYTE_4_SRQ_SHIFT_M, SRQC_BYTE_4_SRQ_SHIFT_S, + ilog2(srq->max)); + + 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, + cpu_to_le32(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 = (u32)(dma_handle_idx >> 3); + srq_context->idx_bt_ba = cpu_to_le32(srq_context->idx_bt_ba); + 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, + cpu_to_le32(dma_handle_idx >> 35)); + + srq_context->idx_cur_blk_addr = (u32)(mtts_idx[0] >> PAGE_ADDR_SHIFT); + srq_context->idx_cur_blk_addr = + cpu_to_le32(srq_context->idx_cur_blk_addr); + 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, + cpu_to_le32((mtts_idx[0]) >> (32 + PAGE_ADDR_SHIFT))); + 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, + hr_dev->caps.idx_hop_num == HNS_ROCE_HOP_NUM_0 ? 0 : + hr_dev->caps.idx_hop_num); + + 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, + hr_dev->caps.idx_ba_pg_sz); + 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, + hr_dev->caps.idx_buf_pg_sz); + + srq_context->idx_nxt_blk_addr = (u32)(mtts_idx[1] >> PAGE_ADDR_SHIFT); + srq_context->idx_nxt_blk_addr = + cpu_to_le32(srq_context->idx_nxt_blk_addr); + 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, + cpu_to_le32((mtts_idx[1]) >> (32 + PAGE_ADDR_SHIFT))); + 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, + hr_dev->caps.srqwqe_ba_pg_sz + PG_SHIFT_OFFSET); + 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, + hr_dev->caps.srqwqe_buf_pg_sz + PG_SHIFT_OFFSET); + + roce_set_bit(srq_context->db_record_addr_record_en, + SRQC_BYTE_60_SRQ_RECORD_EN_S, 0); +} + +static int hns_roce_v2_modify_srq(struct ib_srq *ibsrq, + struct ib_srq_attr *srq_attr, + enum ib_srq_attr_mask srq_attr_mask, + struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + struct hns_roce_srq_context *srq_context; + struct hns_roce_srq_context *srqc_mask; + struct hns_roce_cmd_mailbox *mailbox; + int ret; + + if (srq_attr_mask & IB_SRQ_LIMIT) { + if (srq_attr->srq_limit >= srq->max) + return -EINVAL; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + srq_context = mailbox->buf; + srqc_mask = (struct hns_roce_srq_context *)mailbox->buf + 1; + + memset(srqc_mask, 0xff, sizeof(*srqc_mask)); + + roce_set_field(srq_context->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S, srq_attr->srq_limit); + roce_set_field(srqc_mask->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S, 0); + + ret = hns_roce_cmd_mbox(hr_dev, mailbox->dma, 0, srq->srqn, 0, + HNS_ROCE_CMD_MODIFY_SRQC, + HNS_ROCE_CMD_TIMEOUT_MSECS); + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + if (ret) { + dev_err(hr_dev->dev, + "MODIFY SRQ Failed to cmd mailbox.\n"); + return ret; + } + } + + return 0; +} + +int hns_roce_v2_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + struct hns_roce_srq_context *srq_context; + struct hns_roce_cmd_mailbox *mailbox; + int limit_wl; + int ret; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) + return PTR_ERR(mailbox); + + srq_context = mailbox->buf; + ret = hns_roce_cmd_mbox(hr_dev, 0, mailbox->dma, srq->srqn, 0, + HNS_ROCE_CMD_QUERY_SRQC, + HNS_ROCE_CMD_TIMEOUT_MSECS); + if (ret) { + dev_err(hr_dev->dev, "QUERY SRQ cmd process error\n"); + goto out; + } + + limit_wl = roce_get_field(srq_context->byte_8_limit_wl, + SRQC_BYTE_8_SRQ_LIMIT_WL_M, + SRQC_BYTE_8_SRQ_LIMIT_WL_S); + + attr->srq_limit = limit_wl; + attr->max_wr = srq->max - 1; + attr->max_sge = srq->max_gs; + + memcpy(srq_context, mailbox->buf, sizeof(*srq_context)); + +out: + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + return ret; +} + +static int find_empty_entry(struct hns_roce_idx_que *idx_que) +{ + int bit_num; + int i; + + /* bitmap[i] is set zero if all bits are allocated */ + for (i = 0; idx_que->bitmap[i] == 0; ++i) + ; + bit_num = ffs(idx_que->bitmap[i]); + idx_que->bitmap[i] &= ~(1ULL << (bit_num - 1)); + + return i * sizeof(u64) * 8 + (bit_num - 1); +} + +static void fill_idx_queue(struct hns_roce_idx_que *idx_que, + int cur_idx, int wqe_idx) +{ + unsigned int *addr; + + addr = (unsigned int *)hns_roce_buf_offset(&idx_que->idx_buf, + cur_idx * idx_que->entry_sz); + *addr = wqe_idx; +} + +static int hns_roce_v2_post_srq_recv(struct ib_srq *ibsrq, + const struct ib_recv_wr *wr, + const struct ib_recv_wr **bad_wr) +{ + 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; + int ret = 0; + int wqe_idx; + void *wqe; + int nreq; + int ind; + int i; + + spin_lock_irqsave(&srq->lock, flags); + + ind = srq->head & (srq->max - 1); + + 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; + *bad_wr = wr; + break; + } + + wqe_idx = find_empty_entry(&srq->idx_que); + fill_idx_queue(&srq->idx_que, ind, wqe_idx); + 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 (i < srq->max_gs) { + dseg->len = 0; + dseg->lkey = cpu_to_le32(0x100); + dseg->addr = 0; + } + + srq->wrid[wqe_idx] = wr->wr_id; + ind = (ind + 1) & (srq->max - 1); + } + + if (likely(nreq)) { + srq->head += nreq; + + /* + * Make sure that descriptors are written before + * doorbell record. + */ + wmb(); + + srq_db.byte_4 = HNS_ROCE_V2_SRQ_DB << 24 | srq->srqn; + srq_db.parameter = srq->head; + + hns_roce_write64_k((__le32 *)&srq_db, srq->db_reg_l); + + } + + spin_unlock_irqrestore(&srq->lock, flags); + + return ret; +} + static const struct hns_roce_hw hns_roce_hw_v2 = { .cmq_init = hns_roce_v2_cmq_init, .cmq_exit = hns_roce_v2_cmq_exit, @@ -5359,6 +5761,10 @@ static const struct hns_roce_hw hns_roce_hw_v2 = { .poll_cq = hns_roce_v2_poll_cq, .init_eq = hns_roce_v2_init_eq_table, .cleanup_eq = hns_roce_v2_cleanup_eq_table, + .write_srqc = hns_roce_v2_write_srqc, + .modify_srq = hns_roce_v2_modify_srq, + .query_srq = hns_roce_v2_query_srq, + .post_srq_recv = hns_roce_v2_post_srq_recv, }; static const struct pci_device_id hns_roce_hw_v2_pci_tbl[] = { diff --git a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h index 8bc820635bbd..b72d0443c835 100644 --- a/drivers/infiniband/hw/hns/hns_roce_hw_v2.h +++ b/drivers/infiniband/hw/hns/hns_roce_hw_v2.h @@ -46,10 +46,16 @@ #define HNS_ROCE_V2_MAX_QP_NUM 0x2000 #define HNS_ROCE_V2_MAX_WQE_NUM 0x8000 +#define HNS_ROCE_V2_MAX_SRQ 0x100000 +#define HNS_ROCE_V2_MAX_SRQ_WR 0x8000 +#define HNS_ROCE_V2_MAX_SRQ_SGE 0x100 #define HNS_ROCE_V2_MAX_CQ_NUM 0x8000 +#define HNS_ROCE_V2_MAX_SRQ_NUM 0x100000 #define HNS_ROCE_V2_MAX_CQE_NUM 0x10000 +#define HNS_ROCE_V2_MAX_SRQWQE_NUM 0x8000 #define HNS_ROCE_V2_MAX_RQ_SGE_NUM 0x100 #define HNS_ROCE_V2_MAX_SQ_SGE_NUM 0xff +#define HNS_ROCE_V2_MAX_SRQ_SGE_NUM 0x100 #define HNS_ROCE_V2_MAX_EXTEND_SGE_NUM 0x200000 #define HNS_ROCE_V2_MAX_SQ_INLINE 0x20 #define HNS_ROCE_V2_UAR_NUM 256 @@ -61,6 +67,8 @@ #define HNS_ROCE_V2_MAX_MTPT_NUM 0x8000 #define HNS_ROCE_V2_MAX_MTT_SEGS 0x1000000 #define HNS_ROCE_V2_MAX_CQE_SEGS 0x1000000 +#define HNS_ROCE_V2_MAX_SRQWQE_SEGS 0x1000000 +#define HNS_ROCE_V2_MAX_IDX_SEGS 0x1000000 #define HNS_ROCE_V2_MAX_PD_NUM 0x1000000 #define HNS_ROCE_V2_MAX_QP_INIT_RDMA 128 #define HNS_ROCE_V2_MAX_QP_DEST_RDMA 128 @@ -71,6 +79,7 @@ #define HNS_ROCE_V2_IRRL_ENTRY_SZ 64 #define HNS_ROCE_V2_TRRL_ENTRY_SZ 48 #define HNS_ROCE_V2_CQC_ENTRY_SZ 64 +#define HNS_ROCE_V2_SRQC_ENTRY_SZ 64 #define HNS_ROCE_V2_MTPT_ENTRY_SZ 64 #define HNS_ROCE_V2_MTT_ENTRY_SZ 64 #define HNS_ROCE_V2_CQE_ENTRY_SIZE 32 @@ -84,8 +93,10 @@ #define HNS_ROCE_CONTEXT_HOP_NUM 1 #define HNS_ROCE_MTT_HOP_NUM 1 #define HNS_ROCE_CQE_HOP_NUM 1 +#define HNS_ROCE_SRQWQE_HOP_NUM 1 #define HNS_ROCE_PBL_HOP_NUM 2 #define HNS_ROCE_EQE_HOP_NUM 2 +#define HNS_ROCE_IDX_HOP_NUM 1 #define HNS_ROCE_V2_GID_INDEX_NUM 256 @@ -113,6 +124,8 @@ ((step_idx == 0 && hop_num == HNS_ROCE_HOP_NUM_0) || \ (step_idx == 1 && hop_num == 1) || \ (step_idx == 2 && hop_num == 2)) +#define HNS_ICL_SWITCH_CMD_ROCEE_SEL_SHIFT 0 +#define HNS_ICL_SWITCH_CMD_ROCEE_SEL BIT(HNS_ICL_SWITCH_CMD_ROCEE_SEL_SHIFT) #define CMD_CSQ_DESC_NUM 1024 #define CMD_CRQ_DESC_NUM 1024 @@ -213,7 +226,10 @@ enum hns_roce_opcode_type { HNS_ROCE_OPC_CFG_TMOUT_LLM = 0x8404, HNS_ROCE_OPC_CFG_SGID_TB = 0x8500, HNS_ROCE_OPC_CFG_SMAC_TB = 0x8501, + HNS_ROCE_OPC_POST_MB = 0x8504, + HNS_ROCE_OPC_QUERY_MB_ST = 0x8505, HNS_ROCE_OPC_CFG_BT_ATTR = 0x8506, + HNS_SWITCH_PARAMETER_CFG = 0x1033, }; enum { @@ -325,6 +341,90 @@ struct hns_roce_v2_cq_context { #define V2_CQC_BYTE_64_SE_CQE_IDX_S 0 #define V2_CQC_BYTE_64_SE_CQE_IDX_M GENMASK(23, 0) +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; +}; + +#define SRQC_BYTE_4_SRQ_ST_S 0 +#define SRQC_BYTE_4_SRQ_ST_M GENMASK(1, 0) + +#define SRQC_BYTE_4_SRQ_WQE_HOP_NUM_S 2 +#define SRQC_BYTE_4_SRQ_WQE_HOP_NUM_M GENMASK(3, 2) + +#define SRQC_BYTE_4_SRQ_SHIFT_S 4 +#define SRQC_BYTE_4_SRQ_SHIFT_M GENMASK(7, 4) + +#define SRQC_BYTE_4_SRQN_S 8 +#define SRQC_BYTE_4_SRQN_M GENMASK(31, 8) + +#define SRQC_BYTE_8_SRQ_LIMIT_WL_S 0 +#define SRQC_BYTE_8_SRQ_LIMIT_WL_M GENMASK(15, 0) + +#define SRQC_BYTE_12_SRQ_XRCD_S 0 +#define SRQC_BYTE_12_SRQ_XRCD_M GENMASK(23, 0) + +#define SRQC_BYTE_16_SRQ_PRODUCER_IDX_S 0 +#define SRQC_BYTE_16_SRQ_PRODUCER_IDX_M GENMASK(15, 0) + +#define SRQC_BYTE_16_SRQ_CONSUMER_IDX_S 0 +#define SRQC_BYTE_16_SRQ_CONSUMER_IDX_M GENMASK(31, 16) + +#define SRQC_BYTE_24_SRQ_WQE_BT_BA_S 0 +#define SRQC_BYTE_24_SRQ_WQE_BT_BA_M GENMASK(28, 0) + +#define SRQC_BYTE_28_PD_S 0 +#define SRQC_BYTE_28_PD_M GENMASK(23, 0) + +#define SRQC_BYTE_28_RQWS_S 24 +#define SRQC_BYTE_28_RQWS_M GENMASK(27, 24) + +#define SRQC_BYTE_36_SRQ_IDX_BT_BA_S 0 +#define SRQC_BYTE_36_SRQ_IDX_BT_BA_M GENMASK(28, 0) + +#define SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_S 0 +#define SRQC_BYTE_44_SRQ_IDX_CUR_BLK_ADDR_M GENMASK(19, 0) + +#define SRQC_BYTE_44_SRQ_IDX_HOP_NUM_S 22 +#define SRQC_BYTE_44_SRQ_IDX_HOP_NUM_M GENMASK(23, 22) + +#define SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_S 24 +#define SRQC_BYTE_44_SRQ_IDX_BA_PG_SZ_M GENMASK(27, 24) + +#define SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_S 28 +#define SRQC_BYTE_44_SRQ_IDX_BUF_PG_SZ_M GENMASK(31, 28) + +#define SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_S 0 +#define SRQC_BYTE_52_SRQ_IDX_NXT_BLK_ADDR_M GENMASK(19, 0) + +#define SRQC_BYTE_56_SRQ_XRC_CQN_S 0 +#define SRQC_BYTE_56_SRQ_XRC_CQN_M GENMASK(23, 0) + +#define SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_S 24 +#define SRQC_BYTE_56_SRQ_WQE_BA_PG_SZ_M GENMASK(27, 24) + +#define SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_S 28 +#define SRQC_BYTE_56_SRQ_WQE_BUF_PG_SZ_M GENMASK(31, 28) + +#define SRQC_BYTE_60_SRQ_RECORD_EN_S 0 + +#define SRQC_BYTE_60_SRQ_DB_RECORD_ADDR_S 1 +#define SRQC_BYTE_60_SRQ_DB_RECORD_ADDR_M GENMASK(31, 1) + enum{ V2_MPT_ST_VALID = 0x1, V2_MPT_ST_FREE = 0x2, @@ -1289,6 +1389,36 @@ struct hns_roce_vf_res_b { #define VF_RES_B_DATA_3_VF_SL_NUM_S 16 #define VF_RES_B_DATA_3_VF_SL_NUM_M GENMASK(19, 16) +struct hns_roce_vf_switch { + __le32 rocee_sel; + __le32 fun_id; + __le32 cfg; + __le32 resv1; + __le32 resv2; + __le32 resv3; +}; + +#define VF_SWITCH_DATA_FUN_ID_VF_ID_S 3 +#define VF_SWITCH_DATA_FUN_ID_VF_ID_M GENMASK(10, 3) + +#define VF_SWITCH_DATA_CFG_ALW_LPBK_S 1 +#define VF_SWITCH_DATA_CFG_ALW_LCL_LPBK_S 2 +#define VF_SWITCH_DATA_CFG_ALW_DST_OVRD_S 3 + +struct hns_roce_post_mbox { + __le32 in_param_l; + __le32 in_param_h; + __le32 out_param_l; + __le32 out_param_h; + __le32 cmd_tag; + __le32 token_event_en; +}; + +struct hns_roce_mbox_status { + __le32 mb_status_hw_run; + __le32 rsv[5]; +}; + struct hns_roce_cfg_bt_attr { __le32 vf_qpc_cfg; __le32 vf_srqc_cfg; @@ -1372,18 +1502,6 @@ struct hns_roce_cmq_desc { #define HNS_ROCE_HW_RUN_BIT_SHIFT 31 #define HNS_ROCE_HW_MB_STATUS_MASK 0xFF -#define HNS_ROCE_VF_MB4_TAG_MASK 0xFFFFFF00 -#define HNS_ROCE_VF_MB4_TAG_SHIFT 8 - -#define HNS_ROCE_VF_MB4_CMD_MASK 0xFF -#define HNS_ROCE_VF_MB4_CMD_SHIFT 0 - -#define HNS_ROCE_VF_MB5_EVENT_MASK 0x10000 -#define HNS_ROCE_VF_MB5_EVENT_SHIFT 16 - -#define HNS_ROCE_VF_MB5_TOKEN_MASK 0xFFFF -#define HNS_ROCE_VF_MB5_TOKEN_SHIFT 0 - struct hns_roce_v2_cmq_ring { dma_addr_t desc_dma_addr; struct hns_roce_cmq_desc *desc; diff --git a/drivers/infiniband/hw/hns/hns_roce_main.c b/drivers/infiniband/hw/hns/hns_roce_main.c index 1b3ee514f2ef..65ba43cee810 100644 --- a/drivers/infiniband/hw/hns/hns_roce_main.c +++ b/drivers/infiniband/hw/hns/hns_roce_main.c @@ -220,6 +220,11 @@ static int hns_roce_query_device(struct ib_device *ib_dev, IB_ATOMIC_HCA : IB_ATOMIC_NONE; props->max_pkeys = 1; props->local_ca_ack_delay = hr_dev->caps.local_ca_ack_delay; + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) { + props->max_srq = hr_dev->caps.max_srqs; + props->max_srq_wr = hr_dev->caps.max_srq_wrs; + props->max_srq_sge = hr_dev->caps.max_srq_sges; + } return 0; } @@ -541,6 +546,21 @@ static int hns_roce_register_device(struct hns_roce_dev *hr_dev) ib_dev->map_mr_sg = hns_roce_map_mr_sg; } + /* SRQ */ + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) { + ib_dev->create_srq = hns_roce_create_srq; + ib_dev->modify_srq = hr_dev->hw->modify_srq; + ib_dev->query_srq = hr_dev->hw->query_srq; + ib_dev->destroy_srq = hns_roce_destroy_srq; + ib_dev->post_srq_recv = hr_dev->hw->post_srq_recv; + ib_dev->uverbs_cmd_mask |= + (1ULL << IB_USER_VERBS_CMD_CREATE_SRQ) | + (1ULL << IB_USER_VERBS_CMD_MODIFY_SRQ) | + (1ULL << IB_USER_VERBS_CMD_QUERY_SRQ) | + (1ULL << IB_USER_VERBS_CMD_DESTROY_SRQ) | + (1ULL << IB_USER_VERBS_CMD_POST_SRQ_RECV); + } + /* OTHERS */ ib_dev->get_port_immutable = hns_roce_port_immutable; ib_dev->disassociate_ucontext = hns_roce_disassociate_ucontext; @@ -646,8 +666,58 @@ static int hns_roce_init_hem(struct hns_roce_dev *hr_dev) goto err_unmap_trrl; } + if (hr_dev->caps.srqc_entry_sz) { + ret = hns_roce_init_hem_table(hr_dev, &hr_dev->srq_table.table, + HEM_TYPE_SRQC, + hr_dev->caps.srqc_entry_sz, + hr_dev->caps.num_srqs, 1); + if (ret) { + dev_err(dev, + "Failed to init SRQ context memory, aborting.\n"); + goto err_unmap_cq; + } + } + + if (hr_dev->caps.num_srqwqe_segs) { + ret = hns_roce_init_hem_table(hr_dev, + &hr_dev->mr_table.mtt_srqwqe_table, + HEM_TYPE_SRQWQE, + hr_dev->caps.mtt_entry_sz, + hr_dev->caps.num_srqwqe_segs, 1); + if (ret) { + dev_err(dev, + "Failed to init MTT srqwqe memory, aborting.\n"); + goto err_unmap_srq; + } + } + + if (hr_dev->caps.num_idx_segs) { + ret = hns_roce_init_hem_table(hr_dev, + &hr_dev->mr_table.mtt_idx_table, + HEM_TYPE_IDX, + hr_dev->caps.idx_entry_sz, + hr_dev->caps.num_idx_segs, 1); + if (ret) { + dev_err(dev, + "Failed to init MTT idx memory, aborting.\n"); + goto err_unmap_srqwqe; + } + } + return 0; +err_unmap_srqwqe: + if (hr_dev->caps.num_srqwqe_segs) + hns_roce_cleanup_hem_table(hr_dev, + &hr_dev->mr_table.mtt_srqwqe_table); + +err_unmap_srq: + if (hr_dev->caps.srqc_entry_sz) + hns_roce_cleanup_hem_table(hr_dev, &hr_dev->srq_table.table); + +err_unmap_cq: + hns_roce_cleanup_hem_table(hr_dev, &hr_dev->cq_table.table); + err_unmap_trrl: if (hr_dev->caps.trrl_entry_sz) hns_roce_cleanup_hem_table(hr_dev, @@ -727,8 +797,21 @@ static int hns_roce_setup_hca(struct hns_roce_dev *hr_dev) goto err_cq_table_free; } + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) { + ret = hns_roce_init_srq_table(hr_dev); + if (ret) { + dev_err(dev, + "Failed to init share receive queue table.\n"); + goto err_qp_table_free; + } + } + return 0; +err_qp_table_free: + if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_SRQ) + hns_roce_cleanup_qp_table(hr_dev); + err_cq_table_free: hns_roce_cleanup_cq_table(hr_dev); diff --git a/drivers/infiniband/hw/hns/hns_roce_mr.c b/drivers/infiniband/hw/hns/hns_roce_mr.c index 521ad2aa3a4e..ee5991bd4171 100644 --- a/drivers/infiniband/hw/hns/hns_roce_mr.c +++ b/drivers/infiniband/hw/hns/hns_roce_mr.c @@ -184,12 +184,27 @@ static int hns_roce_alloc_mtt_range(struct hns_roce_dev *hr_dev, int order, struct hns_roce_buddy *buddy; int ret; - if (mtt_type == MTT_TYPE_WQE) { + switch (mtt_type) { + case MTT_TYPE_WQE: buddy = &mr_table->mtt_buddy; table = &mr_table->mtt_table; - } else { + break; + case MTT_TYPE_CQE: buddy = &mr_table->mtt_cqe_buddy; table = &mr_table->mtt_cqe_table; + break; + case MTT_TYPE_SRQWQE: + buddy = &mr_table->mtt_srqwqe_buddy; + table = &mr_table->mtt_srqwqe_table; + break; + case MTT_TYPE_IDX: + buddy = &mr_table->mtt_idx_buddy; + table = &mr_table->mtt_idx_table; + break; + default: + dev_err(hr_dev->dev, "Unsupport MTT table type: %d\n", + mtt_type); + return -EINVAL; } ret = hns_roce_buddy_alloc(buddy, order, seg); @@ -242,18 +257,40 @@ void hns_roce_mtt_cleanup(struct hns_roce_dev *hr_dev, struct hns_roce_mtt *mtt) if (mtt->order < 0) return; - if (mtt->mtt_type == MTT_TYPE_WQE) { + switch (mtt->mtt_type) { + case MTT_TYPE_WQE: hns_roce_buddy_free(&mr_table->mtt_buddy, mtt->first_seg, mtt->order); hns_roce_table_put_range(hr_dev, &mr_table->mtt_table, mtt->first_seg, mtt->first_seg + (1 << mtt->order) - 1); - } else { + break; + case MTT_TYPE_CQE: hns_roce_buddy_free(&mr_table->mtt_cqe_buddy, mtt->first_seg, mtt->order); hns_roce_table_put_range(hr_dev, &mr_table->mtt_cqe_table, mtt->first_seg, mtt->first_seg + (1 << mtt->order) - 1); + break; + case MTT_TYPE_SRQWQE: + hns_roce_buddy_free(&mr_table->mtt_srqwqe_buddy, mtt->first_seg, + mtt->order); + hns_roce_table_put_range(hr_dev, &mr_table->mtt_srqwqe_table, + mtt->first_seg, + mtt->first_seg + (1 << mtt->order) - 1); + break; + case MTT_TYPE_IDX: + hns_roce_buddy_free(&mr_table->mtt_idx_buddy, mtt->first_seg, + mtt->order); + hns_roce_table_put_range(hr_dev, &mr_table->mtt_idx_table, + mtt->first_seg, + mtt->first_seg + (1 << mtt->order) - 1); + break; + default: + dev_err(hr_dev->dev, + "Unsupport mtt type %d, clean mtt failed\n", + mtt->mtt_type); + break; } } EXPORT_SYMBOL_GPL(hns_roce_mtt_cleanup); @@ -713,10 +750,26 @@ static int hns_roce_write_mtt_chunk(struct hns_roce_dev *hr_dev, u32 bt_page_size; u32 i; - if (mtt->mtt_type == MTT_TYPE_WQE) + switch (mtt->mtt_type) { + case MTT_TYPE_WQE: + table = &hr_dev->mr_table.mtt_table; bt_page_size = 1 << (hr_dev->caps.mtt_ba_pg_sz + PAGE_SHIFT); - else + break; + case MTT_TYPE_CQE: + table = &hr_dev->mr_table.mtt_cqe_table; bt_page_size = 1 << (hr_dev->caps.cqe_ba_pg_sz + PAGE_SHIFT); + break; + case MTT_TYPE_SRQWQE: + table = &hr_dev->mr_table.mtt_srqwqe_table; + bt_page_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + PAGE_SHIFT); + break; + case MTT_TYPE_IDX: + table = &hr_dev->mr_table.mtt_idx_table; + bt_page_size = 1 << (hr_dev->caps.idx_ba_pg_sz + PAGE_SHIFT); + break; + default: + return -EINVAL; + } /* All MTTs must fit in the same page */ if (start_index / (bt_page_size / sizeof(u64)) != @@ -726,11 +779,6 @@ static int hns_roce_write_mtt_chunk(struct hns_roce_dev *hr_dev, if (start_index & (HNS_ROCE_MTT_ENTRY_PER_SEG - 1)) return -EINVAL; - if (mtt->mtt_type == MTT_TYPE_WQE) - table = &hr_dev->mr_table.mtt_table; - else - table = &hr_dev->mr_table.mtt_cqe_table; - mtts = hns_roce_table_find(hr_dev, table, mtt->first_seg + s / hr_dev->caps.mtt_entry_sz, &dma_handle); @@ -759,10 +807,25 @@ static int hns_roce_write_mtt(struct hns_roce_dev *hr_dev, if (mtt->order < 0) return -EINVAL; - if (mtt->mtt_type == MTT_TYPE_WQE) + switch (mtt->mtt_type) { + case MTT_TYPE_WQE: bt_page_size = 1 << (hr_dev->caps.mtt_ba_pg_sz + PAGE_SHIFT); - else + break; + case MTT_TYPE_CQE: bt_page_size = 1 << (hr_dev->caps.cqe_ba_pg_sz + PAGE_SHIFT); + break; + case MTT_TYPE_SRQWQE: + bt_page_size = 1 << (hr_dev->caps.srqwqe_ba_pg_sz + PAGE_SHIFT); + break; + case MTT_TYPE_IDX: + bt_page_size = 1 << (hr_dev->caps.idx_ba_pg_sz + PAGE_SHIFT); + break; + default: + dev_err(hr_dev->dev, + "Unsupport mtt type %d, write mtt failed\n", + mtt->mtt_type); + return -EINVAL; + } while (npages > 0) { chunk = min_t(int, bt_page_size / sizeof(u64), npages); @@ -828,8 +891,31 @@ int hns_roce_init_mr_table(struct hns_roce_dev *hr_dev) if (ret) goto err_buddy_cqe; } + + if (hr_dev->caps.num_srqwqe_segs) { + ret = hns_roce_buddy_init(&mr_table->mtt_srqwqe_buddy, + ilog2(hr_dev->caps.num_srqwqe_segs)); + if (ret) + goto err_buddy_srqwqe; + } + + if (hr_dev->caps.num_idx_segs) { + ret = hns_roce_buddy_init(&mr_table->mtt_idx_buddy, + ilog2(hr_dev->caps.num_idx_segs)); + if (ret) + goto err_buddy_idx; + } + return 0; +err_buddy_idx: + if (hr_dev->caps.num_srqwqe_segs) + hns_roce_buddy_cleanup(&mr_table->mtt_srqwqe_buddy); + +err_buddy_srqwqe: + if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) + hns_roce_buddy_cleanup(&mr_table->mtt_cqe_buddy); + err_buddy_cqe: hns_roce_buddy_cleanup(&mr_table->mtt_buddy); @@ -842,6 +928,10 @@ void hns_roce_cleanup_mr_table(struct hns_roce_dev *hr_dev) { struct hns_roce_mr_table *mr_table = &hr_dev->mr_table; + if (hr_dev->caps.num_idx_segs) + hns_roce_buddy_cleanup(&mr_table->mtt_idx_buddy); + if (hr_dev->caps.num_srqwqe_segs) + hns_roce_buddy_cleanup(&mr_table->mtt_srqwqe_buddy); hns_roce_buddy_cleanup(&mr_table->mtt_buddy); if (hns_roce_check_whether_mhop(hr_dev, HEM_TYPE_CQE)) hns_roce_buddy_cleanup(&mr_table->mtt_cqe_buddy); @@ -897,8 +987,25 @@ int hns_roce_ib_umem_write_mtt(struct hns_roce_dev *hr_dev, u32 bt_page_size; u32 n; - order = mtt->mtt_type == MTT_TYPE_WQE ? hr_dev->caps.mtt_ba_pg_sz : - hr_dev->caps.cqe_ba_pg_sz; + switch (mtt->mtt_type) { + case MTT_TYPE_WQE: + order = hr_dev->caps.mtt_ba_pg_sz; + break; + case MTT_TYPE_CQE: + order = hr_dev->caps.cqe_ba_pg_sz; + break; + case MTT_TYPE_SRQWQE: + order = hr_dev->caps.srqwqe_ba_pg_sz; + break; + case MTT_TYPE_IDX: + order = hr_dev->caps.idx_ba_pg_sz; + break; + default: + dev_err(dev, "Unsupport mtt type %d, write mtt failed\n", + mtt->mtt_type); + return -EINVAL; + } + bt_page_size = 1 << (order + PAGE_SHIFT); pages = (u64 *) __get_free_pages(GFP_KERNEL, order); @@ -1021,14 +1128,14 @@ struct ib_mr *hns_roce_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, goto err_umem; } } else { - int pbl_size = 1; + u64 pbl_size = 1; bt_size = (1 << (hr_dev->caps.pbl_ba_pg_sz + PAGE_SHIFT)) / 8; for (i = 0; i < hr_dev->caps.pbl_hop_num; i++) pbl_size *= bt_size; if (n > pbl_size) { dev_err(dev, - " MR len %lld err. MR page num is limited to %d!\n", + " MR len %lld err. MR page num is limited to %lld!\n", length, pbl_size); ret = -EINVAL; goto err_umem; diff --git a/drivers/infiniband/hw/hns/hns_roce_qp.c b/drivers/infiniband/hw/hns/hns_roce_qp.c index 5ebf481a39d9..52d2b299b3be 100644 --- a/drivers/infiniband/hw/hns/hns_roce_qp.c +++ b/drivers/infiniband/hw/hns/hns_roce_qp.c @@ -280,7 +280,7 @@ void hns_roce_release_range_qp(struct hns_roce_dev *hr_dev, int base_qpn, EXPORT_SYMBOL_GPL(hns_roce_release_range_qp); static int hns_roce_set_rq_size(struct hns_roce_dev *hr_dev, - struct ib_qp_cap *cap, int is_user, int has_srq, + struct ib_qp_cap *cap, int is_user, int has_rq, struct hns_roce_qp *hr_qp) { struct device *dev = hr_dev->dev; @@ -294,14 +294,12 @@ static int hns_roce_set_rq_size(struct hns_roce_dev *hr_dev, return -EINVAL; } - /* If srq exit, set zero for relative number of rq */ - if (has_srq) { - if (cap->max_recv_wr) { - dev_dbg(dev, "srq no need config max_recv_wr\n"); - return -EINVAL; - } - - hr_qp->rq.wqe_cnt = hr_qp->rq.max_gs = 0; + /* If srq exist, set zero for relative number of rq */ + if (!has_rq) { + hr_qp->rq.wqe_cnt = 0; + hr_qp->rq.max_gs = 0; + cap->max_recv_wr = 0; + cap->max_recv_sge = 0; } else { if (is_user && (!cap->max_recv_wr || !cap->max_recv_sge)) { dev_err(dev, "user space no need config max_recv_wr max_recv_sge\n"); @@ -563,13 +561,14 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev, hr_qp->sq_signal_bits = cpu_to_le32(IB_SIGNAL_REQ_WR); ret = hns_roce_set_rq_size(hr_dev, &init_attr->cap, !!ib_pd->uobject, - !!init_attr->srq, hr_qp); + hns_roce_qp_has_rq(init_attr), hr_qp); if (ret) { dev_err(dev, "hns_roce_set_rq_size failed\n"); goto err_out; } - if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) { + if ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_RQ_INLINE) && + hns_roce_qp_has_rq(init_attr)) { /* allocate recv inline buf */ hr_qp->rq_inl_buf.wqe_list = kcalloc(hr_qp->rq.wqe_cnt, sizeof(struct hns_roce_rinl_wqe), diff --git a/drivers/infiniband/hw/hns/hns_roce_srq.c b/drivers/infiniband/hw/hns/hns_roce_srq.c new file mode 100644 index 000000000000..463df60094e8 --- /dev/null +++ b/drivers/infiniband/hw/hns/hns_roce_srq.c @@ -0,0 +1,456 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +/* + * Copyright (c) 2018 Hisilicon Limited. + */ + +#include <rdma/ib_umem.h> +#include <rdma/hns-abi.h> +#include "hns_roce_device.h" +#include "hns_roce_cmd.h" +#include "hns_roce_hem.h" + +void hns_roce_srq_event(struct hns_roce_dev *hr_dev, u32 srqn, int event_type) +{ + struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; + struct hns_roce_srq *srq; + + xa_lock(&srq_table->xa); + srq = xa_load(&srq_table->xa, srqn & (hr_dev->caps.num_srqs - 1)); + if (srq) + atomic_inc(&srq->refcount); + xa_unlock(&srq_table->xa); + + if (!srq) { + dev_warn(hr_dev->dev, "Async event for bogus SRQ %08x\n", srqn); + return; + } + + srq->event(srq, event_type); + + if (atomic_dec_and_test(&srq->refcount)) + complete(&srq->free); +} +EXPORT_SYMBOL_GPL(hns_roce_srq_event); + +static void hns_roce_ib_srq_event(struct hns_roce_srq *srq, + enum hns_roce_event event_type) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(srq->ibsrq.device); + struct ib_srq *ibsrq = &srq->ibsrq; + struct ib_event event; + + if (ibsrq->event_handler) { + event.device = ibsrq->device; + event.element.srq = ibsrq; + switch (event_type) { + case HNS_ROCE_EVENT_TYPE_SRQ_LIMIT_REACH: + event.event = IB_EVENT_SRQ_LIMIT_REACHED; + break; + case HNS_ROCE_EVENT_TYPE_SRQ_CATAS_ERROR: + event.event = IB_EVENT_SRQ_ERR; + break; + default: + dev_err(hr_dev->dev, + "hns_roce:Unexpected event type 0x%x on SRQ %06lx\n", + event_type, srq->srqn); + return; + } + + ibsrq->event_handler(&event, ibsrq->srq_context); + } +} + +static int hns_roce_sw2hw_srq(struct hns_roce_dev *dev, + struct hns_roce_cmd_mailbox *mailbox, + unsigned long srq_num) +{ + return hns_roce_cmd_mbox(dev, mailbox->dma, 0, srq_num, 0, + HNS_ROCE_CMD_SW2HW_SRQ, + HNS_ROCE_CMD_TIMEOUT_MSECS); +} + +static int hns_roce_hw2sw_srq(struct hns_roce_dev *dev, + struct hns_roce_cmd_mailbox *mailbox, + unsigned long srq_num) +{ + return hns_roce_cmd_mbox(dev, 0, mailbox ? mailbox->dma : 0, srq_num, + mailbox ? 0 : 1, HNS_ROCE_CMD_HW2SW_SRQ, + HNS_ROCE_CMD_TIMEOUT_MSECS); +} + +int hns_roce_srq_alloc(struct hns_roce_dev *hr_dev, u32 pdn, u32 cqn, u16 xrcd, + struct hns_roce_mtt *hr_mtt, u64 db_rec_addr, + struct hns_roce_srq *srq) +{ + struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; + struct hns_roce_cmd_mailbox *mailbox; + dma_addr_t dma_handle_wqe; + dma_addr_t dma_handle_idx; + u64 *mtts_wqe; + u64 *mtts_idx; + int ret; + + /* Get the physical address of srq buf */ + mtts_wqe = hns_roce_table_find(hr_dev, + &hr_dev->mr_table.mtt_srqwqe_table, + srq->mtt.first_seg, + &dma_handle_wqe); + if (!mtts_wqe) { + dev_err(hr_dev->dev, + "SRQ alloc.Failed to find srq buf addr.\n"); + return -EINVAL; + } + + /* Get physical address of idx que buf */ + mtts_idx = hns_roce_table_find(hr_dev, &hr_dev->mr_table.mtt_idx_table, + srq->idx_que.mtt.first_seg, + &dma_handle_idx); + if (!mtts_idx) { + dev_err(hr_dev->dev, + "SRQ alloc.Failed to find idx que buf addr.\n"); + return -EINVAL; + } + + ret = hns_roce_bitmap_alloc(&srq_table->bitmap, &srq->srqn); + if (ret == -1) { + dev_err(hr_dev->dev, "SRQ alloc.Failed to alloc index.\n"); + return -ENOMEM; + } + + ret = hns_roce_table_get(hr_dev, &srq_table->table, srq->srqn); + if (ret) + goto err_out; + + ret = xa_err(xa_store(&srq_table->xa, srq->srqn, srq, GFP_KERNEL)); + if (ret) + goto err_put; + + mailbox = hns_roce_alloc_cmd_mailbox(hr_dev); + if (IS_ERR(mailbox)) { + ret = PTR_ERR(mailbox); + 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 = hns_roce_sw2hw_srq(hr_dev, mailbox, srq->srqn); + hns_roce_free_cmd_mailbox(hr_dev, mailbox); + if (ret) + goto err_xa; + + atomic_set(&srq->refcount, 1); + init_completion(&srq->free); + return ret; + +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; +} + +void hns_roce_srq_free(struct hns_roce_dev *hr_dev, struct hns_roce_srq *srq) +{ + struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; + int ret; + + ret = hns_roce_hw2sw_srq(hr_dev, NULL, srq->srqn); + if (ret) + dev_err(hr_dev->dev, "HW2SW_SRQ failed (%d) for CQN %06lx\n", + ret, srq->srqn); + + xa_erase(&srq_table->xa, srq->srqn); + + if (atomic_dec_and_test(&srq->refcount)) + complete(&srq->free); + wait_for_completion(&srq->free); + + hns_roce_table_put(hr_dev, &srq_table->table, srq->srqn); + hns_roce_bitmap_free(&srq_table->bitmap, srq->srqn, BITMAP_NO_RR); +} + +static int hns_roce_create_idx_que(struct ib_pd *pd, struct hns_roce_srq *srq, + u32 page_shift) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); + struct hns_roce_idx_que *idx_que = &srq->idx_que; + u32 bitmap_num; + int i; + + bitmap_num = HNS_ROCE_ALOGN_UP(srq->max, 8 * sizeof(u64)); + + idx_que->bitmap = kcalloc(1, bitmap_num / 8, GFP_KERNEL); + if (!idx_que->bitmap) + return -ENOMEM; + + bitmap_num = bitmap_num / (8 * sizeof(u64)); + + idx_que->buf_size = srq->idx_que.buf_size; + + if (hns_roce_buf_alloc(hr_dev, idx_que->buf_size, (1 << page_shift) * 2, + &idx_que->idx_buf, page_shift)) { + kfree(idx_que->bitmap); + return -ENOMEM; + } + + for (i = 0; i < bitmap_num; i++) + idx_que->bitmap[i] = ~(0UL); + + return 0; +} + +struct ib_srq *hns_roce_create_srq(struct ib_pd *pd, + struct ib_srq_init_attr *srq_init_attr, + struct ib_udata *udata) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(pd->device); + struct hns_roce_srq *srq; + int srq_desc_size; + int srq_buf_size; + u32 page_shift; + int ret = 0; + u32 npages; + u32 cqn; + + /* Check the actual SRQ wqe and SRQ sge num */ + if (srq_init_attr->attr.max_wr >= hr_dev->caps.max_srq_wrs || + srq_init_attr->attr.max_sge > hr_dev->caps.max_srq_sges) + return ERR_PTR(-EINVAL); + + srq = kzalloc(sizeof(*srq), GFP_KERNEL); + if (!srq) + return ERR_PTR(-ENOMEM); + + mutex_init(&srq->mutex); + spin_lock_init(&srq->lock); + + srq->max = roundup_pow_of_two(srq_init_attr->attr.max_wr + 1); + srq->max_gs = srq_init_attr->attr.max_sge; + + srq_desc_size = max(16, 16 * srq->max_gs); + + srq->wqe_shift = ilog2(srq_desc_size); + + srq_buf_size = srq->max * srq_desc_size; + + srq->idx_que.entry_sz = HNS_ROCE_IDX_QUE_ENTRY_SZ; + srq->idx_que.buf_size = srq->max * srq->idx_que.entry_sz; + srq->mtt.mtt_type = MTT_TYPE_SRQWQE; + srq->idx_que.mtt.mtt_type = MTT_TYPE_IDX; + + if (udata) { + struct hns_roce_ib_create_srq ucmd; + + if (ib_copy_from_udata(&ucmd, udata, sizeof(ucmd))) { + ret = -EFAULT; + goto err_srq; + } + + srq->umem = ib_umem_get(pd->uobject->context, ucmd.buf_addr, + srq_buf_size, 0, 0); + if (IS_ERR(srq->umem)) { + ret = PTR_ERR(srq->umem); + goto err_srq; + } + + if (hr_dev->caps.srqwqe_buf_pg_sz) { + npages = (ib_umem_page_count(srq->umem) + + (1 << hr_dev->caps.srqwqe_buf_pg_sz) - 1) / + (1 << hr_dev->caps.srqwqe_buf_pg_sz); + page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; + ret = hns_roce_mtt_init(hr_dev, npages, + page_shift, + &srq->mtt); + } else + ret = hns_roce_mtt_init(hr_dev, + ib_umem_page_count(srq->umem), + srq->umem->page_shift, + &srq->mtt); + if (ret) + goto err_buf; + + ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->mtt, srq->umem); + if (ret) + goto err_srq_mtt; + + /* config index queue BA */ + srq->idx_que.umem = ib_umem_get(pd->uobject->context, + ucmd.que_addr, + srq->idx_que.buf_size, 0, 0); + if (IS_ERR(srq->idx_que.umem)) { + dev_err(hr_dev->dev, + "ib_umem_get error for index queue\n"); + goto err_srq_mtt; + } + + if (hr_dev->caps.idx_buf_pg_sz) { + npages = (ib_umem_page_count(srq->idx_que.umem) + + (1 << hr_dev->caps.idx_buf_pg_sz) - 1) / + (1 << hr_dev->caps.idx_buf_pg_sz); + page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; + ret = hns_roce_mtt_init(hr_dev, npages, + page_shift, &srq->idx_que.mtt); + } else { + ret = hns_roce_mtt_init(hr_dev, + ib_umem_page_count(srq->idx_que.umem), + srq->idx_que.umem->page_shift, + &srq->idx_que.mtt); + } + + if (ret) { + dev_err(hr_dev->dev, + "hns_roce_mtt_init error for idx que\n"); + goto err_idx_mtt; + } + + ret = hns_roce_ib_umem_write_mtt(hr_dev, &srq->idx_que.mtt, + srq->idx_que.umem); + if (ret) { + dev_err(hr_dev->dev, + "hns_roce_ib_umem_write_mtt error for idx que\n"); + goto err_idx_buf; + } + } else { + page_shift = PAGE_SHIFT + hr_dev->caps.srqwqe_buf_pg_sz; + if (hns_roce_buf_alloc(hr_dev, srq_buf_size, + (1 << page_shift) * 2, + &srq->buf, page_shift)) { + ret = -ENOMEM; + goto err_srq; + } + + srq->head = 0; + srq->tail = srq->max - 1; + + ret = hns_roce_mtt_init(hr_dev, srq->buf.npages, + srq->buf.page_shift, &srq->mtt); + if (ret) + goto err_buf; + + ret = hns_roce_buf_write_mtt(hr_dev, &srq->mtt, &srq->buf); + if (ret) + goto err_srq_mtt; + + page_shift = PAGE_SHIFT + hr_dev->caps.idx_buf_pg_sz; + ret = hns_roce_create_idx_que(pd, srq, page_shift); + if (ret) { + dev_err(hr_dev->dev, "Create idx queue fail(%d)!\n", + ret); + goto err_srq_mtt; + } + + /* Init mtt table for idx_que */ + ret = hns_roce_mtt_init(hr_dev, srq->idx_que.idx_buf.npages, + srq->idx_que.idx_buf.page_shift, + &srq->idx_que.mtt); + if (ret) + goto err_create_idx; + + /* Write buffer address into the mtt table */ + ret = hns_roce_buf_write_mtt(hr_dev, &srq->idx_que.mtt, + &srq->idx_que.idx_buf); + if (ret) + goto err_idx_buf; + + srq->wrid = kvmalloc_array(srq->max, sizeof(u64), GFP_KERNEL); + if (!srq->wrid) { + ret = -ENOMEM; + goto err_idx_buf; + } + } + + cqn = ib_srq_has_cq(srq_init_attr->srq_type) ? + to_hr_cq(srq_init_attr->ext.cq)->cqn : 0; + + srq->db_reg_l = hr_dev->reg_base + SRQ_DB_REG; + + ret = hns_roce_srq_alloc(hr_dev, to_hr_pd(pd)->pdn, cqn, 0, + &srq->mtt, 0, srq); + if (ret) + goto err_wrid; + + srq->event = hns_roce_ib_srq_event; + srq->ibsrq.ext.xrc.srq_num = srq->srqn; + + if (pd->uobject) { + if (ib_copy_to_udata(udata, &srq->srqn, sizeof(__u32))) { + ret = -EFAULT; + goto err_wrid; + } + } + + return &srq->ibsrq; + +err_wrid: + kvfree(srq->wrid); + +err_idx_buf: + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + +err_idx_mtt: + if (udata) + ib_umem_release(srq->idx_que.umem); + +err_create_idx: + hns_roce_buf_free(hr_dev, srq->idx_que.buf_size, + &srq->idx_que.idx_buf); + kfree(srq->idx_que.bitmap); + +err_srq_mtt: + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + +err_buf: + if (udata) + ib_umem_release(srq->umem); + else + hns_roce_buf_free(hr_dev, srq_buf_size, &srq->buf); + +err_srq: + kfree(srq); + return ERR_PTR(ret); +} + +int hns_roce_destroy_srq(struct ib_srq *ibsrq) +{ + struct hns_roce_dev *hr_dev = to_hr_dev(ibsrq->device); + struct hns_roce_srq *srq = to_hr_srq(ibsrq); + + hns_roce_srq_free(hr_dev, srq); + hns_roce_mtt_cleanup(hr_dev, &srq->mtt); + + if (ibsrq->uobject) { + hns_roce_mtt_cleanup(hr_dev, &srq->idx_que.mtt); + ib_umem_release(srq->idx_que.umem); + ib_umem_release(srq->umem); + } else { + kvfree(srq->wrid); + hns_roce_buf_free(hr_dev, srq->max << srq->wqe_shift, + &srq->buf); + } + + kfree(srq); + + return 0; +} + +int hns_roce_init_srq_table(struct hns_roce_dev *hr_dev) +{ + struct hns_roce_srq_table *srq_table = &hr_dev->srq_table; + + xa_init(&srq_table->xa); + + return hns_roce_bitmap_init(&srq_table->bitmap, hr_dev->caps.num_srqs, + hr_dev->caps.num_srqs - 1, + hr_dev->caps.reserved_srqs, 0); +} + +void hns_roce_cleanup_srq_table(struct hns_roce_dev *hr_dev) +{ + hns_roce_bitmap_cleanup(&hr_dev->srq_table.bitmap); +} diff --git a/drivers/infiniband/hw/i40iw/i40iw_verbs.c b/drivers/infiniband/hw/i40iw/i40iw_verbs.c index 102875872bea..a773d1edf7fd 100644 --- a/drivers/infiniband/hw/i40iw/i40iw_verbs.c +++ b/drivers/infiniband/hw/i40iw/i40iw_verbs.c @@ -2722,25 +2722,6 @@ static int i40iw_query_pkey(struct ib_device *ibdev, } /** - * i40iw_get_vector_affinity - report IRQ affinity mask - * @ibdev: IB device - * @comp_vector: completion vector index - */ -static const struct cpumask *i40iw_get_vector_affinity(struct ib_device *ibdev, - int comp_vector) -{ - struct i40iw_device *iwdev = to_iwdev(ibdev); - struct i40iw_msix_vector *msix_vec; - - if (iwdev->msix_shared) - msix_vec = &iwdev->iw_msixtbl[comp_vector]; - else - msix_vec = &iwdev->iw_msixtbl[comp_vector + 1]; - - return irq_get_affinity_mask(msix_vec->irq); -} - -/** * i40iw_init_rdma_device - initialization of iwarp device * @iwdev: iwarp device */ @@ -2832,7 +2813,6 @@ static struct i40iw_ib_device *i40iw_init_rdma_device(struct i40iw_device *iwdev iwibdev->ibdev.req_notify_cq = i40iw_req_notify_cq; iwibdev->ibdev.post_send = i40iw_post_send; iwibdev->ibdev.post_recv = i40iw_post_recv; - iwibdev->ibdev.get_vector_affinity = i40iw_get_vector_affinity; return iwibdev; } diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 0def2323459c..b73b5fa1822a 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -2352,6 +2352,32 @@ static void mlx4_ib_scan_netdevs(struct mlx4_ib_dev *ibdev, event == NETDEV_UP || event == NETDEV_CHANGE)) update_qps_port = port; + if (dev == iboe->netdevs[port - 1] && + (event == NETDEV_UP || event == NETDEV_DOWN)) { + enum ib_port_state port_state; + struct ib_event ibev = { }; + + if (ib_get_cached_port_state(&ibdev->ib_dev, port, + &port_state)) + continue; + + if (event == NETDEV_UP && + (port_state != IB_PORT_ACTIVE || + iboe->last_port_state[port - 1] != IB_PORT_DOWN)) + continue; + if (event == NETDEV_DOWN && + (port_state != IB_PORT_DOWN || + iboe->last_port_state[port - 1] != IB_PORT_ACTIVE)) + continue; + iboe->last_port_state[port - 1] = port_state; + + ibev.device = &ibdev->ib_dev; + ibev.element.port_num = port; + ibev.event = event == NETDEV_UP ? IB_EVENT_PORT_ACTIVE : + IB_EVENT_PORT_ERR; + ib_dispatch_event(&ibev); + } + } spin_unlock_bh(&iboe->lock); @@ -2710,6 +2736,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) for (i = 0; i < ibdev->num_ports; ++i) { mutex_init(&ibdev->counters_table[i].mutex); INIT_LIST_HEAD(&ibdev->counters_table[i].counters_list); + iboe->last_port_state[i] = IB_PORT_DOWN; } num_req_counters = mlx4_is_bonded(dev) ? 1 : ibdev->num_ports; diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 8850dfc3826d..5cb52424912e 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -519,6 +519,7 @@ struct mlx4_ib_iboe { atomic64_t mac[MLX4_MAX_PORTS]; struct notifier_block nb; struct mlx4_port_gid_table gids[MLX4_MAX_PORTS]; + enum ib_port_state last_port_state[MLX4_MAX_PORTS]; }; struct pkey_mgt { diff --git a/drivers/infiniband/hw/mlx4/sysfs.c b/drivers/infiniband/hw/mlx4/sysfs.c index 752bdd536130..ea1f3a081b05 100644 --- a/drivers/infiniband/hw/mlx4/sysfs.c +++ b/drivers/infiniband/hw/mlx4/sysfs.c @@ -353,16 +353,12 @@ err: static void get_name(struct mlx4_ib_dev *dev, char *name, int i, int max) { - char base_name[9]; - - /* pci_name format is: bus:dev:func -> xxxx:yy:zz.n */ - strlcpy(name, pci_name(dev->dev->persist->pdev), max); - strncpy(base_name, name, 8); /*till xxxx:yy:*/ - base_name[8] = '\0'; - /* with no ARI only 3 last bits are used so when the fn is higher than 8 + /* pci_name format is: bus:dev:func -> xxxx:yy:zz.n + * with no ARI only 3 last bits are used so when the fn is higher than 8 * need to add it to the dev num, so count in the last number will be * modulo 8 */ - sprintf(name, "%s%.2d.%d", base_name, (i/8), (i%8)); + snprintf(name, max, "%.8s%.2d.%d", pci_name(dev->dev->persist->pdev), + i / 8, i % 8); } struct mlx4_port { diff --git a/drivers/infiniband/hw/mlx5/Makefile b/drivers/infiniband/hw/mlx5/Makefile index b8e4b15e2674..33f5adb14e4e 100644 --- a/drivers/infiniband/hw/mlx5/Makefile +++ b/drivers/infiniband/hw/mlx5/Makefile @@ -1,6 +1,8 @@ obj-$(CONFIG_MLX5_INFINIBAND) += mlx5_ib.o -mlx5_ib-y := main.o cq.o doorbell.o qp.o mem.o srq.o mr.o ah.o mad.o gsi.o ib_virt.o cmd.o cong.o +mlx5_ib-y := main.o cq.o doorbell.o qp.o mem.o srq_cmd.o \ + srq.o mr.o ah.o mad.o gsi.o ib_virt.o cmd.o \ + cong.o mlx5_ib-$(CONFIG_INFINIBAND_ON_DEMAND_PAGING) += odp.o mlx5_ib-$(CONFIG_MLX5_ESWITCH) += ib_rep.o mlx5_ib-$(CONFIG_INFINIBAND_USER_ACCESS) += devx.o diff --git a/drivers/infiniband/hw/mlx5/cq.c b/drivers/infiniband/hw/mlx5/cq.c index 7d769b5538b4..0b99f7d0630d 100644 --- a/drivers/infiniband/hw/mlx5/cq.c +++ b/drivers/infiniband/hw/mlx5/cq.c @@ -35,6 +35,7 @@ #include <rdma/ib_user_verbs.h> #include <rdma/ib_cache.h> #include "mlx5_ib.h" +#include "srq.h" static void mlx5_ib_cq_comp(struct mlx5_core_cq *cq) { @@ -177,8 +178,7 @@ static void handle_responder(struct ib_wc *wc, struct mlx5_cqe64 *cqe, struct mlx5_core_srq *msrq = NULL; if (qp->ibqp.xrcd) { - msrq = mlx5_core_get_srq(dev->mdev, - be32_to_cpu(cqe->srqn)); + msrq = mlx5_cmd_get_srq(dev, be32_to_cpu(cqe->srqn)); srq = to_mibsrq(msrq); } else { srq = to_msrq(qp->ibqp.srq); diff --git a/drivers/infiniband/hw/mlx5/devx.c b/drivers/infiniband/hw/mlx5/devx.c index 61aab7c0c513..5271469aad10 100644 --- a/drivers/infiniband/hw/mlx5/devx.c +++ b/drivers/infiniband/hw/mlx5/devx.c @@ -9,6 +9,7 @@ #include <rdma/uverbs_ioctl.h> #include <rdma/mlx5_user_ioctl_cmds.h> #include <rdma/ib_umem.h> +#include <rdma/uverbs_std_types.h> #include <linux/mlx5/driver.h> #include <linux/mlx5/fs.h> #include "mlx5_ib.h" @@ -40,29 +41,37 @@ struct devx_umem_reg_cmd { u32 out[MLX5_ST_SZ_DW(general_obj_out_cmd_hdr)]; }; -static struct mlx5_ib_ucontext *devx_ufile2uctx(struct ib_uverbs_file *file) +static struct mlx5_ib_ucontext * +devx_ufile2uctx(const struct uverbs_attr_bundle *attrs) { - return to_mucontext(ib_uverbs_get_ucontext(file)); + return to_mucontext(ib_uverbs_get_ucontext(attrs)); } -int mlx5_ib_devx_create(struct mlx5_ib_dev *dev) +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}; u64 general_obj_types; - void *hdr; + void *hdr, *uctx; int err; u16 uid; + u32 cap = 0; hdr = MLX5_ADDR_OF(create_uctx_in, in, hdr); + uctx = MLX5_ADDR_OF(create_uctx_in, in, uctx); general_obj_types = MLX5_CAP_GEN_64(dev->mdev, general_obj_types); if (!(general_obj_types & MLX5_GENERAL_OBJ_TYPES_CAP_UCTX) || !(general_obj_types & MLX5_GENERAL_OBJ_TYPES_CAP_UMEM)) return -EINVAL; + if (is_user && capable(CAP_NET_RAW) && + (MLX5_CAP_GEN(dev->mdev, uctx_cap) & MLX5_UCTX_CAP_RAW_TX)) + cap |= MLX5_UCTX_CAP_RAW_TX; + MLX5_SET(general_obj_in_cmd_hdr, hdr, opcode, MLX5_CMD_OP_CREATE_GENERAL_OBJECT); MLX5_SET(general_obj_in_cmd_hdr, hdr, obj_type, MLX5_OBJ_TYPE_UCTX); + MLX5_SET(uctx, uctx, cap, cap); err = mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); if (err) @@ -106,6 +115,21 @@ bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type) } } +bool mlx5_ib_devx_is_flow_counter(void *obj, u32 *counter_id) +{ + struct devx_obj *devx_obj = obj; + u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, devx_obj->dinbox, opcode); + + if (opcode == MLX5_CMD_OP_DEALLOC_FLOW_COUNTER) { + *counter_id = MLX5_GET(dealloc_flow_counter_in, + devx_obj->dinbox, + flow_counter_id); + return true; + } + + return false; +} + /* * As the obj_id in the firmware is not globally unique the object type * must be considered upon checking for a valid object id. @@ -116,7 +140,7 @@ static u64 get_enc_obj_id(u16 opcode, u32 obj_id) return ((u64)opcode << 32) | obj_id; } -static int devx_is_valid_obj_id(struct devx_obj *obj, const void *in) +static u64 devx_get_obj_id(const void *in) { u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode); u64 obj_id; @@ -290,6 +314,8 @@ static int devx_is_valid_obj_id(struct devx_obj *obj, const void *in) MLX5_GET(query_dct_in, in, dctn)); break; case MLX5_CMD_OP_QUERY_XRQ: + case MLX5_CMD_OP_QUERY_XRQ_DC_PARAMS_ENTRY: + case MLX5_CMD_OP_QUERY_XRQ_ERROR_PARAMS: obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_XRQ, MLX5_GET(query_xrq_in, in, xrqn)); break; @@ -316,17 +342,107 @@ static int devx_is_valid_obj_id(struct devx_obj *obj, const void *in) MLX5_GET(drain_dct_in, in, dctn)); break; case MLX5_CMD_OP_ARM_XRQ: + case MLX5_CMD_OP_SET_XRQ_DC_PARAMS_ENTRY: obj_id = get_enc_obj_id(MLX5_CMD_OP_CREATE_XRQ, MLX5_GET(arm_xrq_in, in, xrqn)); break; + case MLX5_CMD_OP_QUERY_PACKET_REFORMAT_CONTEXT: + obj_id = get_enc_obj_id + (MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT, + MLX5_GET(query_packet_reformat_context_in, + in, packet_reformat_id)); + break; default: + obj_id = 0; + } + + return obj_id; +} + +static bool devx_is_valid_obj_id(struct ib_uobject *uobj, const void *in) +{ + u64 obj_id = devx_get_obj_id(in); + + if (!obj_id) return false; + + switch (uobj_get_object_id(uobj)) { + case UVERBS_OBJECT_CQ: + return get_enc_obj_id(MLX5_CMD_OP_CREATE_CQ, + to_mcq(uobj->object)->mcq.cqn) == + obj_id; + + case UVERBS_OBJECT_SRQ: + { + struct mlx5_core_srq *srq = &(to_msrq(uobj->object)->msrq); + struct mlx5_ib_dev *dev = to_mdev(uobj->context->device); + u16 opcode; + + switch (srq->common.res) { + case MLX5_RES_XSRQ: + opcode = MLX5_CMD_OP_CREATE_XRC_SRQ; + break; + case MLX5_RES_XRQ: + opcode = MLX5_CMD_OP_CREATE_XRQ; + break; + default: + if (!dev->mdev->issi) + opcode = MLX5_CMD_OP_CREATE_SRQ; + else + opcode = MLX5_CMD_OP_CREATE_RMP; + } + + return get_enc_obj_id(opcode, + to_msrq(uobj->object)->msrq.srqn) == + obj_id; } - if (obj_id == obj->obj_id) - return true; + case UVERBS_OBJECT_QP: + { + struct mlx5_ib_qp *qp = to_mqp(uobj->object); + enum ib_qp_type qp_type = qp->ibqp.qp_type; + + if (qp_type == IB_QPT_RAW_PACKET || + (qp->flags & MLX5_IB_QP_UNDERLAY)) { + struct mlx5_ib_raw_packet_qp *raw_packet_qp = + &qp->raw_packet_qp; + struct mlx5_ib_rq *rq = &raw_packet_qp->rq; + struct mlx5_ib_sq *sq = &raw_packet_qp->sq; + + return (get_enc_obj_id(MLX5_CMD_OP_CREATE_RQ, + rq->base.mqp.qpn) == obj_id || + get_enc_obj_id(MLX5_CMD_OP_CREATE_SQ, + sq->base.mqp.qpn) == obj_id || + get_enc_obj_id(MLX5_CMD_OP_CREATE_TIR, + rq->tirn) == obj_id || + get_enc_obj_id(MLX5_CMD_OP_CREATE_TIS, + sq->tisn) == obj_id); + } + + if (qp_type == MLX5_IB_QPT_DCT) + return get_enc_obj_id(MLX5_CMD_OP_CREATE_DCT, + qp->dct.mdct.mqp.qpn) == obj_id; + + return get_enc_obj_id(MLX5_CMD_OP_CREATE_QP, + qp->ibqp.qp_num) == obj_id; + } - return false; + case UVERBS_OBJECT_WQ: + return get_enc_obj_id(MLX5_CMD_OP_CREATE_RQ, + to_mrwq(uobj->object)->core_qp.qpn) == + obj_id; + + case UVERBS_OBJECT_RWQ_IND_TBL: + return get_enc_obj_id(MLX5_CMD_OP_CREATE_RQT, + to_mrwq_ind_table(uobj->object)->rqtn) == + obj_id; + + case MLX5_IB_OBJECT_DEVX_OBJ: + return ((struct devx_obj *)uobj->object)->obj_id == obj_id; + + default: + return false; + } } static void devx_set_umem_valid(const void *in) @@ -494,6 +610,7 @@ static bool devx_is_obj_modify_cmd(const void *in) case MLX5_CMD_OP_DRAIN_DCT: case MLX5_CMD_OP_ARM_DCT_FOR_KEY_VIOLATION: case MLX5_CMD_OP_ARM_XRQ: + case MLX5_CMD_OP_SET_XRQ_DC_PARAMS_ENTRY: return true; case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY: { @@ -535,6 +652,9 @@ static bool devx_is_obj_query_cmd(const void *in) case MLX5_CMD_OP_QUERY_XRC_SRQ: case MLX5_CMD_OP_QUERY_DCT: case MLX5_CMD_OP_QUERY_XRQ: + case MLX5_CMD_OP_QUERY_XRQ_DC_PARAMS_ENTRY: + case MLX5_CMD_OP_QUERY_XRQ_ERROR_PARAMS: + case MLX5_CMD_OP_QUERY_PACKET_REFORMAT_CONTEXT: return true; default: return false; @@ -572,15 +692,16 @@ static int devx_get_uid(struct mlx5_ib_ucontext *c, void *cmd_in) if (!c->devx_uid) return -EINVAL; - if (!capable(CAP_NET_RAW)) - return -EPERM; - return c->devx_uid; } static bool devx_is_general_cmd(void *in) { u16 opcode = MLX5_GET(general_obj_in_cmd_hdr, in, opcode); + if (opcode >= MLX5_CMD_OP_GENERAL_START && + opcode < MLX5_CMD_OP_GENERAL_END) + return true; + switch (opcode) { case MLX5_CMD_OP_QUERY_HCA_CAP: case MLX5_CMD_OP_QUERY_HCA_VPORT_CONTEXT: @@ -603,7 +724,7 @@ static bool devx_is_general_cmd(void *in) } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_QUERY_EQN)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct mlx5_ib_ucontext *c; struct mlx5_ib_dev *dev; @@ -616,7 +737,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_QUERY_EQN)( MLX5_IB_ATTR_DEVX_QUERY_EQN_USER_VEC)) return -EFAULT; - c = devx_ufile2uctx(file); + c = devx_ufile2uctx(attrs); if (IS_ERR(c)) return PTR_ERR(c); dev = to_mdev(c->ibucontext.device); @@ -653,14 +774,14 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_QUERY_EQN)( * queue or arm its CQ for event generation), no further harm is expected. */ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_QUERY_UAR)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct mlx5_ib_ucontext *c; struct mlx5_ib_dev *dev; u32 user_idx; s32 dev_idx; - c = devx_ufile2uctx(file); + c = devx_ufile2uctx(attrs); if (IS_ERR(c)) return PTR_ERR(c); dev = to_mdev(c->ibucontext.device); @@ -681,7 +802,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_QUERY_UAR)( } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OTHER)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct mlx5_ib_ucontext *c; struct mlx5_ib_dev *dev; @@ -693,7 +814,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OTHER)( int err; int uid; - c = devx_ufile2uctx(file); + c = devx_ufile2uctx(attrs); if (IS_ERR(c)) return PTR_ERR(c); dev = to_mdev(c->ibucontext.device); @@ -908,7 +1029,7 @@ static int devx_obj_cleanup(struct ib_uobject *uobject, } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_CREATE)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { void *cmd_in = uverbs_attr_get_alloced_ptr(attrs, MLX5_IB_ATTR_DEVX_OBJ_CREATE_CMD_IN); int cmd_out_len = uverbs_attr_get_len(attrs, @@ -970,7 +1091,7 @@ obj_free: } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_MODIFY)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { void *cmd_in = uverbs_attr_get_alloced_ptr(attrs, MLX5_IB_ATTR_DEVX_OBJ_MODIFY_CMD_IN); int cmd_out_len = uverbs_attr_get_len(attrs, @@ -978,7 +1099,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_MODIFY)( struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_DEVX_OBJ_MODIFY_HANDLE); struct mlx5_ib_ucontext *c = to_mucontext(uobj->context); - struct devx_obj *obj = uobj->object; + struct mlx5_ib_dev *mdev = to_mdev(uobj->context->device); void *cmd_out; int err; int uid; @@ -990,7 +1111,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_MODIFY)( if (!devx_is_obj_modify_cmd(cmd_in)) return -EINVAL; - if (!devx_is_valid_obj_id(obj, cmd_in)) + if (!devx_is_valid_obj_id(uobj, cmd_in)) return -EINVAL; cmd_out = uverbs_zalloc(attrs, cmd_out_len); @@ -1000,7 +1121,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_MODIFY)( MLX5_SET(general_obj_in_cmd_hdr, cmd_in, uid, uid); devx_set_umem_valid(cmd_in); - err = mlx5_cmd_exec(obj->mdev, cmd_in, + err = mlx5_cmd_exec(mdev->mdev, cmd_in, uverbs_attr_get_len(attrs, MLX5_IB_ATTR_DEVX_OBJ_MODIFY_CMD_IN), cmd_out, cmd_out_len); if (err) @@ -1011,7 +1132,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_MODIFY)( } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_QUERY)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { void *cmd_in = uverbs_attr_get_alloced_ptr(attrs, MLX5_IB_ATTR_DEVX_OBJ_QUERY_CMD_IN); int cmd_out_len = uverbs_attr_get_len(attrs, @@ -1019,10 +1140,10 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_QUERY)( struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_DEVX_OBJ_QUERY_HANDLE); struct mlx5_ib_ucontext *c = to_mucontext(uobj->context); - struct devx_obj *obj = uobj->object; void *cmd_out; int err; int uid; + struct mlx5_ib_dev *mdev = to_mdev(uobj->context->device); uid = devx_get_uid(c, cmd_in); if (uid < 0) @@ -1031,7 +1152,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_QUERY)( if (!devx_is_obj_query_cmd(cmd_in)) return -EINVAL; - if (!devx_is_valid_obj_id(obj, cmd_in)) + if (!devx_is_valid_obj_id(uobj, cmd_in)) return -EINVAL; cmd_out = uverbs_zalloc(attrs, cmd_out_len); @@ -1039,7 +1160,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_OBJ_QUERY)( return PTR_ERR(cmd_out); MLX5_SET(general_obj_in_cmd_hdr, cmd_in, uid, uid); - err = mlx5_cmd_exec(obj->mdev, cmd_in, + err = mlx5_cmd_exec(mdev->mdev, cmd_in, uverbs_attr_get_len(attrs, MLX5_IB_ATTR_DEVX_OBJ_QUERY_CMD_IN), cmd_out, cmd_out_len); if (err) @@ -1125,7 +1246,7 @@ static void devx_umem_reg_cmd_build(struct mlx5_ib_dev *dev, } static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_UMEM_REG)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct devx_umem_reg_cmd cmd; struct devx_umem *obj; @@ -1139,9 +1260,6 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_DEVX_UMEM_REG)( if (!c->devx_uid) return -EINVAL; - if (!capable(CAP_NET_RAW)) - return -EPERM; - obj = kzalloc(sizeof(struct devx_umem), GFP_KERNEL); if (!obj) return -ENOMEM; @@ -1277,7 +1395,7 @@ DECLARE_UVERBS_NAMED_METHOD_DESTROY( DECLARE_UVERBS_NAMED_METHOD( MLX5_IB_METHOD_DEVX_OBJ_MODIFY, UVERBS_ATTR_IDR(MLX5_IB_ATTR_DEVX_OBJ_MODIFY_HANDLE, - MLX5_IB_OBJECT_DEVX_OBJ, + UVERBS_IDR_ANY_OBJECT, UVERBS_ACCESS_WRITE, UA_MANDATORY), UVERBS_ATTR_PTR_IN( @@ -1293,7 +1411,7 @@ DECLARE_UVERBS_NAMED_METHOD( DECLARE_UVERBS_NAMED_METHOD( MLX5_IB_METHOD_DEVX_OBJ_QUERY, UVERBS_ATTR_IDR(MLX5_IB_ATTR_DEVX_OBJ_QUERY_HANDLE, - MLX5_IB_OBJECT_DEVX_OBJ, + UVERBS_IDR_ANY_OBJECT, UVERBS_ACCESS_READ, UA_MANDATORY), UVERBS_ATTR_PTR_IN( @@ -1323,12 +1441,23 @@ DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_DEVX_UMEM, &UVERBS_METHOD(MLX5_IB_METHOD_DEVX_UMEM_REG), &UVERBS_METHOD(MLX5_IB_METHOD_DEVX_UMEM_DEREG)); -DECLARE_UVERBS_OBJECT_TREE(devx_objects, - &UVERBS_OBJECT(MLX5_IB_OBJECT_DEVX), - &UVERBS_OBJECT(MLX5_IB_OBJECT_DEVX_OBJ), - &UVERBS_OBJECT(MLX5_IB_OBJECT_DEVX_UMEM)); - -const struct uverbs_object_tree_def *mlx5_ib_get_devx_tree(void) +static bool devx_is_supported(struct ib_device *device) { - return &devx_objects; + struct mlx5_ib_dev *dev = to_mdev(device); + + return !dev->rep && MLX5_CAP_GEN_64(dev->mdev, general_obj_types) & + MLX5_GENERAL_OBJ_TYPES_CAP_UCTX; } + +const struct uapi_definition mlx5_ib_devx_defs[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + MLX5_IB_OBJECT_DEVX, + UAPI_DEF_IS_OBJ_SUPPORTED(devx_is_supported)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + MLX5_IB_OBJECT_DEVX_OBJ, + UAPI_DEF_IS_OBJ_SUPPORTED(devx_is_supported)), + UAPI_DEF_CHAIN_OBJ_TREE_NAMED( + MLX5_IB_OBJECT_DEVX_UMEM, + UAPI_DEF_IS_OBJ_SUPPORTED(devx_is_supported)), + {}, +}; diff --git a/drivers/infiniband/hw/mlx5/flow.c b/drivers/infiniband/hw/mlx5/flow.c index f86cdcafdafc..b73756bd38bd 100644 --- a/drivers/infiniband/hw/mlx5/flow.c +++ b/drivers/infiniband/hw/mlx5/flow.c @@ -60,7 +60,7 @@ static const struct uverbs_attr_spec mlx5_ib_flow_type[] = { #define MLX5_IB_CREATE_FLOW_MAX_FLOW_ACTIONS 2 static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct mlx5_flow_act flow_act = {.flow_tag = MLX5_FS_DEFAULT_FLOW_TAG}; struct mlx5_ib_flow_handler *flow_handler; @@ -77,6 +77,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( uverbs_attr_get_uobject(attrs, MLX5_IB_ATTR_CREATE_FLOW_HANDLE); struct mlx5_ib_dev *dev = to_mdev(uobj->context->device); int len, ret, i; + u32 counter_id = 0; if (!capable(CAP_NET_RAW)) return -EPERM; @@ -92,10 +93,6 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( ((dest_devx && dest_qp) || (!dest_devx && !dest_qp))) return -EINVAL; - if (fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS && - (dest_devx || dest_qp)) - return -EINVAL; - if (dest_devx) { devx_obj = uverbs_attr_get_obj( attrs, MLX5_IB_ATTR_CREATE_FLOW_DEST_DEVX); @@ -128,9 +125,22 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( dest_type = MLX5_FLOW_DESTINATION_TYPE_PORT; } + len = uverbs_attr_get_uobjs_arr(attrs, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, &arr_flow_actions); + if (len) { + devx_obj = arr_flow_actions[0]->object; + + if (!mlx5_ib_devx_is_flow_counter(devx_obj, &counter_id)) + return -EINVAL; + flow_act.action |= MLX5_FLOW_CONTEXT_ACTION_COUNT; + } if (dev->rep) return -ENOTSUPP; + if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR && + fs_matcher->ns_type == MLX5_FLOW_NAMESPACE_EGRESS) + return -EINVAL; + cmd_in = uverbs_attr_get_alloced_ptr( attrs, MLX5_IB_ATTR_CREATE_FLOW_MATCH_VALUE); inlen = uverbs_attr_get_len(attrs, @@ -164,6 +174,7 @@ static int UVERBS_HANDLER(MLX5_IB_METHOD_CREATE_FLOW)( } flow_handler = mlx5_ib_raw_fs_rule_add(dev, fs_matcher, &flow_act, + counter_id, cmd_in, inlen, dest_id, dest_type); if (IS_ERR(flow_handler)) { @@ -194,7 +205,7 @@ static int flow_matcher_cleanup(struct ib_uobject *uobject, } static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_MATCHER_CREATE)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) + struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject( attrs, MLX5_IB_ATTR_FLOW_MATCHER_CREATE_HANDLE); @@ -313,7 +324,6 @@ static bool mlx5_ib_modify_header_supported(struct mlx5_ib_dev *dev) } static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_MODIFY_HEADER)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject( @@ -435,7 +445,6 @@ static int mlx5_ib_flow_action_create_packet_reformat_ctx( } static int UVERBS_HANDLER(MLX5_IB_METHOD_FLOW_ACTION_CREATE_PACKET_REFORMAT)( - struct ib_uverbs_file *file, struct uverbs_attr_bundle *attrs) { struct ib_uobject *uobj = uverbs_attr_get_uobject(attrs, @@ -526,7 +535,11 @@ DECLARE_UVERBS_NAMED_METHOD( UA_OPTIONAL), UVERBS_ATTR_PTR_IN(MLX5_IB_ATTR_CREATE_FLOW_TAG, UVERBS_ATTR_TYPE(u32), - UA_OPTIONAL)); + UA_OPTIONAL), + UVERBS_ATTR_IDRS_ARR(MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, + MLX5_IB_OBJECT_DEVX_OBJ, + UVERBS_ACCESS_READ, 1, 1, + UA_OPTIONAL)); DECLARE_UVERBS_NAMED_METHOD_DESTROY( MLX5_IB_METHOD_DESTROY_FLOW, @@ -610,16 +623,10 @@ DECLARE_UVERBS_NAMED_OBJECT(MLX5_IB_OBJECT_FLOW_MATCHER, &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_CREATE), &UVERBS_METHOD(MLX5_IB_METHOD_FLOW_MATCHER_DESTROY)); -DECLARE_UVERBS_OBJECT_TREE(flow_objects, - &UVERBS_OBJECT(MLX5_IB_OBJECT_FLOW_MATCHER)); - -int mlx5_ib_get_flow_trees(const struct uverbs_object_tree_def **root) -{ - int i = 0; - - root[i++] = &flow_objects; - root[i++] = &mlx5_ib_fs; - root[i++] = &mlx5_ib_flow_actions; - - return i; -} +const struct uapi_definition mlx5_ib_flow_defs[] = { + UAPI_DEF_CHAIN_OBJ_TREE_NAMED(MLX5_IB_OBJECT_FLOW_MATCHER), + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW, &mlx5_ib_fs), + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION, + &mlx5_ib_flow_actions), + {}, +}; diff --git a/drivers/infiniband/hw/mlx5/ib_rep.c b/drivers/infiniband/hw/mlx5/ib_rep.c index 584ff2ea7810..8a682d86d634 100644 --- a/drivers/infiniband/hw/mlx5/ib_rep.c +++ b/drivers/infiniband/hw/mlx5/ib_rep.c @@ -4,6 +4,7 @@ */ #include "ib_rep.h" +#include "srq.h" static const struct mlx5_ib_profile rep_profile = { STAGE_CREATE(MLX5_IB_STAGE_INIT, @@ -21,6 +22,9 @@ static const struct mlx5_ib_profile rep_profile = { STAGE_CREATE(MLX5_IB_STAGE_ROCE, mlx5_ib_stage_rep_roce_init, mlx5_ib_stage_rep_roce_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_SRQ, + mlx5_init_srq_table, + mlx5_cleanup_srq_table), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES, mlx5_ib_stage_dev_res_init, mlx5_ib_stage_dev_res_cleanup), diff --git a/drivers/infiniband/hw/mlx5/mad.c b/drivers/infiniband/hw/mlx5/mad.c index 32a9e9228b13..a2735f246d5c 100644 --- a/drivers/infiniband/hw/mlx5/mad.c +++ b/drivers/infiniband/hw/mlx5/mad.c @@ -568,6 +568,10 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, props->max_vl_num = out_mad->data[37] >> 4; props->init_type_reply = out_mad->data[41] >> 4; + if (props->port_cap_flags & IB_PORT_CAP_MASK2_SUP) + props->port_cap_flags2 = + be16_to_cpup((__be16 *)(out_mad->data + 60)); + /* Check if extended speeds (EDR/FDR/...) are supported */ if (props->port_cap_flags & IB_PORT_EXTENDED_SPEEDS_SUP) { ext_active_speed = out_mad->data[62] >> 4; @@ -579,6 +583,11 @@ int mlx5_query_mad_ifc_port(struct ib_device *ibdev, u8 port, case 2: props->active_speed = 32; /* EDR */ break; + case 4: + if (props->port_cap_flags & IB_PORT_CAP_MASK2_SUP && + props->port_cap_flags2 & IB_PORT_LINK_SPEED_HDR_SUP) + props->active_speed = IB_SPEED_HDR; + break; } } diff --git a/drivers/infiniband/hw/mlx5/main.c b/drivers/infiniband/hw/mlx5/main.c index 3569fda07e07..5edd8d5f151f 100644 --- a/drivers/infiniband/hw/mlx5/main.c +++ b/drivers/infiniband/hw/mlx5/main.c @@ -60,6 +60,7 @@ #include "mlx5_ib.h" #include "ib_rep.h" #include "cmd.h" +#include "srq.h" #include <linux/mlx5/fs_helpers.h> #include <linux/mlx5/accel.h> #include <rdma/uverbs_std_types.h> @@ -82,10 +83,13 @@ static char mlx5_version[] = struct mlx5_ib_event_work { struct work_struct work; - struct mlx5_core_dev *dev; - void *context; - enum mlx5_dev_event event; - unsigned long param; + union { + struct mlx5_ib_dev *dev; + struct mlx5_ib_multiport_info *mpi; + }; + bool is_slave; + unsigned int event; + void *param; }; enum { @@ -1014,6 +1018,9 @@ static int mlx5_ib_query_device(struct ib_device *ibdev, if (MLX5_CAP_GEN(mdev, cqe_128_always)) resp.flags |= MLX5_IB_QUERY_DEV_RESP_FLAGS_CQE_128B_PAD; + if (MLX5_CAP_GEN(mdev, qp_packet_based)) + resp.flags |= + MLX5_IB_QUERY_DEV_RESP_PACKET_BASED_CREDIT_MODE; } if (field_avail(typeof(resp), sw_parsing_caps, @@ -1216,6 +1223,9 @@ static int mlx5_query_hca_port(struct ib_device *ibdev, u8 port, props->subnet_timeout = rep->subnet_timeout; props->init_type_reply = rep->init_type_reply; + if (props->port_cap_flags & IB_PORT_CAP_MASK2_SUP) + props->port_cap_flags2 = rep->cap_mask2; + err = mlx5_query_port_link_width_oper(mdev, &ib_link_width_oper, port); if (err) goto out; @@ -1752,7 +1762,7 @@ static struct ib_ucontext *mlx5_ib_alloc_ucontext(struct ib_device *ibdev, #endif if (req.flags & MLX5_IB_ALLOC_UCTX_DEVX) { - err = mlx5_ib_devx_create(dev); + err = mlx5_ib_devx_create(dev, true); if (err < 0) goto out_uars; context->devx_uid = err; @@ -3706,7 +3716,8 @@ _create_raw_flow_rule(struct mlx5_ib_dev *dev, struct mlx5_flow_destination *dst, struct mlx5_ib_flow_matcher *fs_matcher, struct mlx5_flow_act *flow_act, - void *cmd_in, int inlen) + void *cmd_in, int inlen, + int dst_num) { struct mlx5_ib_flow_handler *handler; struct mlx5_flow_spec *spec; @@ -3728,7 +3739,7 @@ _create_raw_flow_rule(struct mlx5_ib_dev *dev, spec->match_criteria_enable = fs_matcher->match_criteria_enable; handler->rule = mlx5_add_flow_rules(ft, spec, - flow_act, dst, 1); + flow_act, dst, dst_num); if (IS_ERR(handler->rule)) { err = PTR_ERR(handler->rule); @@ -3791,12 +3802,14 @@ struct mlx5_ib_flow_handler * mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher, struct mlx5_flow_act *flow_act, + u32 counter_id, void *cmd_in, int inlen, int dest_id, int dest_type) { struct mlx5_flow_destination *dst; struct mlx5_ib_flow_prio *ft_prio; struct mlx5_ib_flow_handler *handler; + int dst_num = 0; bool mcast; int err; @@ -3806,7 +3819,7 @@ mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev, if (fs_matcher->priority > MLX5_IB_FLOW_LAST_PRIO) return ERR_PTR(-ENOMEM); - dst = kzalloc(sizeof(*dst), GFP_KERNEL); + dst = kzalloc(sizeof(*dst) * 2, GFP_KERNEL); if (!dst) return ERR_PTR(-ENOMEM); @@ -3820,20 +3833,28 @@ mlx5_ib_raw_fs_rule_add(struct mlx5_ib_dev *dev, } if (dest_type == MLX5_FLOW_DESTINATION_TYPE_TIR) { - dst->type = dest_type; - dst->tir_num = dest_id; + dst[dst_num].type = dest_type; + dst[dst_num].tir_num = dest_id; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; } else if (dest_type == MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE) { - dst->type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM; - dst->ft_num = dest_id; + dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_FLOW_TABLE_NUM; + dst[dst_num].ft_num = dest_id; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_FWD_DEST; } else { - dst->type = MLX5_FLOW_DESTINATION_TYPE_PORT; + dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_PORT; flow_act->action |= MLX5_FLOW_CONTEXT_ACTION_ALLOW; } + dst_num++; + + if (flow_act->action & MLX5_FLOW_CONTEXT_ACTION_COUNT) { + dst[dst_num].type = MLX5_FLOW_DESTINATION_TYPE_COUNTER; + dst[dst_num].counter_id = counter_id; + dst_num++; + } + handler = _create_raw_flow_rule(dev, ft_prio, dst, fs_matcher, flow_act, - cmd_in, inlen); + cmd_in, inlen, dst_num); if (IS_ERR(handler)) { err = PTR_ERR(handler); @@ -4226,6 +4247,63 @@ static void delay_drop_handler(struct work_struct *work) mutex_unlock(&delay_drop->lock); } +static void handle_general_event(struct mlx5_ib_dev *ibdev, struct mlx5_eqe *eqe, + struct ib_event *ibev) +{ + switch (eqe->sub_type) { + case MLX5_GENERAL_SUBTYPE_DELAY_DROP_TIMEOUT: + schedule_work(&ibdev->delay_drop.delay_drop_work); + break; + default: /* do nothing */ + return; + } +} + +static int handle_port_change(struct mlx5_ib_dev *ibdev, struct mlx5_eqe *eqe, + struct ib_event *ibev) +{ + u8 port = (eqe->data.port.port >> 4) & 0xf; + + ibev->element.port_num = port; + + switch (eqe->sub_type) { + case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE: + case MLX5_PORT_CHANGE_SUBTYPE_DOWN: + case MLX5_PORT_CHANGE_SUBTYPE_INITIALIZED: + /* In RoCE, port up/down events are handled in + * mlx5_netdev_event(). + */ + if (mlx5_ib_port_link_layer(&ibdev->ib_dev, port) == + IB_LINK_LAYER_ETHERNET) + return -EINVAL; + + ibev->event = (eqe->sub_type == MLX5_PORT_CHANGE_SUBTYPE_ACTIVE) ? + IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR; + break; + + case MLX5_PORT_CHANGE_SUBTYPE_LID: + ibev->event = IB_EVENT_LID_CHANGE; + break; + + case MLX5_PORT_CHANGE_SUBTYPE_PKEY: + ibev->event = IB_EVENT_PKEY_CHANGE; + schedule_work(&ibdev->devr.ports[port - 1].pkey_change_work); + break; + + case MLX5_PORT_CHANGE_SUBTYPE_GUID: + ibev->event = IB_EVENT_GID_CHANGE; + break; + + case MLX5_PORT_CHANGE_SUBTYPE_CLIENT_REREG: + ibev->event = IB_EVENT_CLIENT_REREGISTER; + break; + default: + return -EINVAL; + } + + return 0; +} + static void mlx5_ib_handle_event(struct work_struct *_work) { struct mlx5_ib_event_work *work = @@ -4233,65 +4311,37 @@ static void mlx5_ib_handle_event(struct work_struct *_work) struct mlx5_ib_dev *ibdev; struct ib_event ibev; bool fatal = false; - u8 port = (u8)work->param; - if (mlx5_core_is_mp_slave(work->dev)) { - ibdev = mlx5_ib_get_ibdev_from_mpi(work->context); + if (work->is_slave) { + ibdev = mlx5_ib_get_ibdev_from_mpi(work->mpi); if (!ibdev) goto out; } else { - ibdev = work->context; + ibdev = work->dev; } switch (work->event) { case MLX5_DEV_EVENT_SYS_ERROR: ibev.event = IB_EVENT_DEVICE_FATAL; mlx5_ib_handle_internal_error(ibdev); + ibev.element.port_num = (u8)(unsigned long)work->param; fatal = true; break; - - case MLX5_DEV_EVENT_PORT_UP: - case MLX5_DEV_EVENT_PORT_DOWN: - case MLX5_DEV_EVENT_PORT_INITIALIZED: - /* In RoCE, port up/down events are handled in - * mlx5_netdev_event(). - */ - if (mlx5_ib_port_link_layer(&ibdev->ib_dev, port) == - IB_LINK_LAYER_ETHERNET) + case MLX5_EVENT_TYPE_PORT_CHANGE: + if (handle_port_change(ibdev, work->param, &ibev)) goto out; - - ibev.event = (work->event == MLX5_DEV_EVENT_PORT_UP) ? - IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR; - break; - - case MLX5_DEV_EVENT_LID_CHANGE: - ibev.event = IB_EVENT_LID_CHANGE; - break; - - case MLX5_DEV_EVENT_PKEY_CHANGE: - ibev.event = IB_EVENT_PKEY_CHANGE; - schedule_work(&ibdev->devr.ports[port - 1].pkey_change_work); - break; - - case MLX5_DEV_EVENT_GUID_CHANGE: - ibev.event = IB_EVENT_GID_CHANGE; - break; - - case MLX5_DEV_EVENT_CLIENT_REREG: - ibev.event = IB_EVENT_CLIENT_REREGISTER; break; - case MLX5_DEV_EVENT_DELAY_DROP_TIMEOUT: - schedule_work(&ibdev->delay_drop.delay_drop_work); - goto out; + case MLX5_EVENT_TYPE_GENERAL_EVENT: + handle_general_event(ibdev, work->param, &ibev); + /* fall through */ default: goto out; } - ibev.device = &ibdev->ib_dev; - ibev.element.port_num = port; + ibev.device = &ibdev->ib_dev; - if (!rdma_is_port_valid(&ibdev->ib_dev, port)) { - mlx5_ib_warn(ibdev, "warning: event on port %d\n", port); + if (!rdma_is_port_valid(&ibdev->ib_dev, ibev.element.port_num)) { + mlx5_ib_warn(ibdev, "warning: event on port %d\n", ibev.element.port_num); goto out; } @@ -4304,22 +4354,43 @@ out: kfree(work); } -static void mlx5_ib_event(struct mlx5_core_dev *dev, void *context, - enum mlx5_dev_event event, unsigned long param) +static int mlx5_ib_event(struct notifier_block *nb, + unsigned long event, void *param) { struct mlx5_ib_event_work *work; work = kmalloc(sizeof(*work), GFP_ATOMIC); if (!work) - return; + return NOTIFY_DONE; INIT_WORK(&work->work, mlx5_ib_handle_event); - work->dev = dev; + work->dev = container_of(nb, struct mlx5_ib_dev, mdev_events); + work->is_slave = false; work->param = param; - work->context = context; work->event = event; queue_work(mlx5_ib_event_wq, &work->work); + + return NOTIFY_OK; +} + +static int mlx5_ib_event_slave_port(struct notifier_block *nb, + unsigned long event, void *param) +{ + struct mlx5_ib_event_work *work; + + work = kmalloc(sizeof(*work), GFP_ATOMIC); + if (!work) + return NOTIFY_DONE; + + INIT_WORK(&work->work, mlx5_ib_handle_event); + work->mpi = container_of(nb, struct mlx5_ib_multiport_info, mdev_events); + work->is_slave = true; + work->param = param; + work->event = event; + queue_work(mlx5_ib_event_wq, &work->work); + + return NOTIFY_OK; } static int set_has_smi_cap(struct mlx5_ib_dev *dev) @@ -5325,14 +5396,6 @@ static void init_delay_drop(struct mlx5_ib_dev *dev) mlx5_ib_warn(dev, "Failed to init delay drop debugfs\n"); } -static const struct cpumask * -mlx5_ib_get_vector_affinity(struct ib_device *ibdev, int comp_vector) -{ - struct mlx5_ib_dev *dev = to_mdev(ibdev); - - return mlx5_get_vector_affinity_hint(dev->mdev, comp_vector); -} - /* The mlx5_ib_multiport_mutex should be held when calling this function */ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, struct mlx5_ib_multiport_info *mpi) @@ -5350,6 +5413,11 @@ static void mlx5_ib_unbind_slave_port(struct mlx5_ib_dev *ibdev, spin_unlock(&port->mp.mpi_lock); return; } + + if (mpi->mdev_events.notifier_call) + mlx5_notifier_unregister(mpi->mdev, &mpi->mdev_events); + mpi->mdev_events.notifier_call = NULL; + mpi->ibdev = NULL; spin_unlock(&port->mp.mpi_lock); @@ -5405,6 +5473,7 @@ static bool mlx5_ib_bind_slave_port(struct mlx5_ib_dev *ibdev, ibdev->port[port_num].mp.mpi = mpi; mpi->ibdev = ibdev; + mpi->mdev_events.notifier_call = NULL; spin_unlock(&ibdev->port[port_num].mp.mpi_lock); err = mlx5_nic_vport_affiliate_multiport(ibdev->mdev, mpi->mdev); @@ -5422,6 +5491,9 @@ static bool mlx5_ib_bind_slave_port(struct mlx5_ib_dev *ibdev, goto unbind; } + mpi->mdev_events.notifier_call = mlx5_ib_event_slave_port; + mlx5_notifier_register(mpi->mdev, &mpi->mdev_events); + err = mlx5_ib_init_cong_debugfs(ibdev, port_num); if (err) goto unbind; @@ -5551,30 +5623,17 @@ ADD_UVERBS_ATTRIBUTES_SIMPLE( UVERBS_ATTR_FLAGS_IN(MLX5_IB_ATTR_CREATE_FLOW_ACTION_FLAGS, enum mlx5_ib_uapi_flow_action_flags)); -static int populate_specs_root(struct mlx5_ib_dev *dev) -{ - const struct uverbs_object_tree_def **trees = dev->driver_trees; - size_t num_trees = 0; - - if (mlx5_accel_ipsec_device_caps(dev->mdev) & - MLX5_ACCEL_IPSEC_CAP_DEVICE) - trees[num_trees++] = &mlx5_ib_flow_action; - - if (MLX5_CAP_DEV_MEM(dev->mdev, memic)) - trees[num_trees++] = &mlx5_ib_dm; - - if (MLX5_CAP_GEN_64(dev->mdev, general_obj_types) & - MLX5_GENERAL_OBJ_TYPES_CAP_UCTX) - trees[num_trees++] = mlx5_ib_get_devx_tree(); - - num_trees += mlx5_ib_get_flow_trees(trees + num_trees); - - WARN_ON(num_trees >= ARRAY_SIZE(dev->driver_trees)); - trees[num_trees] = NULL; - dev->ib_dev.driver_specs = trees; +static const struct uapi_definition mlx5_ib_defs[] = { +#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) + UAPI_DEF_CHAIN(mlx5_ib_devx_defs), + UAPI_DEF_CHAIN(mlx5_ib_flow_defs), +#endif - return 0; -} + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_FLOW_ACTION, + &mlx5_ib_flow_action), + UAPI_DEF_CHAIN_OBJ_TREE(UVERBS_OBJECT_DM, &mlx5_ib_dm), + {} +}; static int mlx5_ib_read_counters(struct ib_counters *counters, struct ib_counters_read_attr *read_attr, @@ -5694,8 +5753,7 @@ int mlx5_ib_stage_init_init(struct mlx5_ib_dev *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 = - dev->mdev->priv.eq_table.num_comp_vectors; + dev->ib_dev.num_comp_vectors = mlx5_comp_vectors_count(mdev); dev->ib_dev.dev.parent = &mdev->pdev->dev; mutex_init(&dev->cap_mask_mutex); @@ -5838,7 +5896,6 @@ int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev) dev->ib_dev.map_mr_sg = mlx5_ib_map_mr_sg; dev->ib_dev.check_mr_status = mlx5_ib_check_mr_status; dev->ib_dev.get_dev_fw_str = get_dev_fw_str; - dev->ib_dev.get_vector_affinity = mlx5_ib_get_vector_affinity; if (MLX5_CAP_GEN(mdev, ipoib_enhanced_offloads) && IS_ENABLED(CONFIG_MLX5_CORE_IPOIB)) dev->ib_dev.rdma_netdev_get_params = mlx5_ib_rn_get_params; @@ -5881,14 +5938,22 @@ int mlx5_ib_stage_caps_init(struct mlx5_ib_dev *dev) dev->ib_dev.uverbs_ex_cmd_mask |= (1ull << IB_USER_VERBS_EX_CMD_CREATE_FLOW) | (1ull << IB_USER_VERBS_EX_CMD_DESTROY_FLOW); - dev->ib_dev.create_flow_action_esp = mlx5_ib_create_flow_action_esp; + if (mlx5_accel_ipsec_device_caps(dev->mdev) & + MLX5_ACCEL_IPSEC_CAP_DEVICE) { + dev->ib_dev.create_flow_action_esp = + mlx5_ib_create_flow_action_esp; + dev->ib_dev.modify_flow_action_esp = + mlx5_ib_modify_flow_action_esp; + } dev->ib_dev.destroy_flow_action = mlx5_ib_destroy_flow_action; - dev->ib_dev.modify_flow_action_esp = mlx5_ib_modify_flow_action_esp; dev->ib_dev.driver_id = RDMA_DRIVER_MLX5; dev->ib_dev.create_counters = mlx5_ib_create_counters; dev->ib_dev.destroy_counters = mlx5_ib_destroy_counters; dev->ib_dev.read_counters = mlx5_ib_read_counters; + if (IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS)) + dev->ib_dev.driver_def = mlx5_ib_defs; + err = init_node_data(dev); if (err) return err; @@ -6034,6 +6099,11 @@ static int mlx5_ib_stage_odp_init(struct mlx5_ib_dev *dev) return mlx5_ib_odp_init_one(dev); } +void mlx5_ib_stage_odp_cleanup(struct mlx5_ib_dev *dev) +{ + mlx5_ib_odp_cleanup_one(dev); +} + int mlx5_ib_stage_counters_init(struct mlx5_ib_dev *dev) { if (MLX5_CAP_GEN(dev->mdev, max_qp_cnt)) { @@ -6096,11 +6166,6 @@ void mlx5_ib_stage_bfrag_cleanup(struct mlx5_ib_dev *dev) mlx5_free_bfreg(dev->mdev, &dev->bfreg); } -static int mlx5_ib_stage_populate_specs(struct mlx5_ib_dev *dev) -{ - return populate_specs_root(dev); -} - int mlx5_ib_stage_ib_reg_init(struct mlx5_ib_dev *dev) { const char *name; @@ -6152,6 +6217,34 @@ static void mlx5_ib_stage_rep_reg_cleanup(struct mlx5_ib_dev *dev) mlx5_ib_unregister_vport_reps(dev); } +static int mlx5_ib_stage_dev_notifier_init(struct mlx5_ib_dev *dev) +{ + dev->mdev_events.notifier_call = mlx5_ib_event; + mlx5_notifier_register(dev->mdev, &dev->mdev_events); + return 0; +} + +static void mlx5_ib_stage_dev_notifier_cleanup(struct mlx5_ib_dev *dev) +{ + mlx5_notifier_unregister(dev->mdev, &dev->mdev_events); +} + +static int mlx5_ib_stage_devx_init(struct mlx5_ib_dev *dev) +{ + int uid; + + uid = mlx5_ib_devx_create(dev, false); + if (uid > 0) + dev->devx_whitelist_uid = uid; + + return 0; +} +static void mlx5_ib_stage_devx_cleanup(struct mlx5_ib_dev *dev) +{ + if (dev->devx_whitelist_uid) + mlx5_ib_devx_destroy(dev, dev->devx_whitelist_uid); +} + void __mlx5_ib_remove(struct mlx5_ib_dev *dev, const struct mlx5_ib_profile *profile, int stage) @@ -6163,8 +6256,6 @@ void __mlx5_ib_remove(struct mlx5_ib_dev *dev, profile->stage[stage].cleanup(dev); } - if (dev->devx_whitelist_uid) - mlx5_ib_devx_destroy(dev, dev->devx_whitelist_uid); ib_dealloc_device((struct ib_device *)dev); } @@ -6173,7 +6264,6 @@ void *__mlx5_ib_add(struct mlx5_ib_dev *dev, { int err; int i; - int uid; for (i = 0; i < MLX5_IB_STAGE_MAX; i++) { if (profile->stage[i].init) { @@ -6183,10 +6273,6 @@ void *__mlx5_ib_add(struct mlx5_ib_dev *dev, } } - uid = mlx5_ib_devx_create(dev); - if (uid > 0) - dev->devx_whitelist_uid = uid; - dev->profile = profile; dev->ib_active = true; @@ -6214,12 +6300,18 @@ static const struct mlx5_ib_profile pf_profile = { STAGE_CREATE(MLX5_IB_STAGE_ROCE, mlx5_ib_stage_roce_init, mlx5_ib_stage_roce_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_SRQ, + mlx5_init_srq_table, + mlx5_cleanup_srq_table), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES, mlx5_ib_stage_dev_res_init, mlx5_ib_stage_dev_res_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_DEVICE_NOTIFIER, + mlx5_ib_stage_dev_notifier_init, + mlx5_ib_stage_dev_notifier_cleanup), STAGE_CREATE(MLX5_IB_STAGE_ODP, mlx5_ib_stage_odp_init, - NULL), + mlx5_ib_stage_odp_cleanup), STAGE_CREATE(MLX5_IB_STAGE_COUNTERS, mlx5_ib_stage_counters_init, mlx5_ib_stage_counters_cleanup), @@ -6235,9 +6327,9 @@ static const struct mlx5_ib_profile pf_profile = { STAGE_CREATE(MLX5_IB_STAGE_PRE_IB_REG_UMR, NULL, mlx5_ib_stage_pre_ib_reg_umr_cleanup), - STAGE_CREATE(MLX5_IB_STAGE_SPECS, - mlx5_ib_stage_populate_specs, - NULL), + STAGE_CREATE(MLX5_IB_STAGE_WHITELIST_UID, + mlx5_ib_stage_devx_init, + mlx5_ib_stage_devx_cleanup), STAGE_CREATE(MLX5_IB_STAGE_IB_REG, mlx5_ib_stage_ib_reg_init, mlx5_ib_stage_ib_reg_cleanup), @@ -6265,9 +6357,15 @@ static const struct mlx5_ib_profile nic_rep_profile = { STAGE_CREATE(MLX5_IB_STAGE_ROCE, mlx5_ib_stage_rep_roce_init, mlx5_ib_stage_rep_roce_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_SRQ, + mlx5_init_srq_table, + mlx5_cleanup_srq_table), STAGE_CREATE(MLX5_IB_STAGE_DEVICE_RESOURCES, mlx5_ib_stage_dev_res_init, mlx5_ib_stage_dev_res_cleanup), + STAGE_CREATE(MLX5_IB_STAGE_DEVICE_NOTIFIER, + mlx5_ib_stage_dev_notifier_init, + mlx5_ib_stage_dev_notifier_cleanup), STAGE_CREATE(MLX5_IB_STAGE_COUNTERS, mlx5_ib_stage_counters_init, mlx5_ib_stage_counters_cleanup), @@ -6280,9 +6378,6 @@ static const struct mlx5_ib_profile nic_rep_profile = { STAGE_CREATE(MLX5_IB_STAGE_PRE_IB_REG_UMR, NULL, mlx5_ib_stage_pre_ib_reg_umr_cleanup), - STAGE_CREATE(MLX5_IB_STAGE_SPECS, - mlx5_ib_stage_populate_specs, - NULL), STAGE_CREATE(MLX5_IB_STAGE_IB_REG, mlx5_ib_stage_ib_reg_init, mlx5_ib_stage_ib_reg_cleanup), @@ -6388,10 +6483,6 @@ static void mlx5_ib_remove(struct mlx5_core_dev *mdev, void *context) static struct mlx5_interface mlx5_ib_interface = { .add = mlx5_ib_add, .remove = mlx5_ib_remove, - .event = mlx5_ib_event, -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - .pfault = mlx5_ib_pfault, -#endif .protocol = MLX5_INTERFACE_PROTOCOL_IB, }; diff --git a/drivers/infiniband/hw/mlx5/mlx5_ib.h b/drivers/infiniband/hw/mlx5/mlx5_ib.h index b651a7a6fde9..7145f512f948 100644 --- a/drivers/infiniband/hw/mlx5/mlx5_ib.h +++ b/drivers/infiniband/hw/mlx5/mlx5_ib.h @@ -41,7 +41,6 @@ #include <linux/mlx5/cq.h> #include <linux/mlx5/fs.h> #include <linux/mlx5/qp.h> -#include <linux/mlx5/srq.h> #include <linux/mlx5/fs.h> #include <linux/types.h> #include <linux/mlx5/transobj.h> @@ -50,6 +49,8 @@ #include <rdma/uverbs_ioctl.h> #include <rdma/mlx5_user_ioctl_cmds.h> +#include "srq.h" + #define mlx5_ib_dbg(_dev, format, arg...) \ dev_dbg(&(_dev)->ib_dev.dev, "%s:%d:(pid %d): " format, __func__, \ __LINE__, current->pid, ##arg) @@ -257,6 +258,7 @@ enum mlx5_ib_rq_flags { }; struct mlx5_ib_wq { + struct mlx5_frag_buf_ctrl fbc; u64 *wrid; u32 *wr_data; struct wr_list *w_list; @@ -275,7 +277,7 @@ struct mlx5_ib_wq { unsigned tail; u16 cur_post; u16 last_poll; - void *qend; + void *cur_edge; }; enum mlx5_ib_wq_flags { @@ -460,6 +462,7 @@ enum mlx5_ib_qp_flags { MLX5_IB_QP_UNDERLAY = 1 << 10, MLX5_IB_QP_PCI_WRITE_END_PADDING = 1 << 11, MLX5_IB_QP_TUNNEL_OFFLOAD = 1 << 12, + MLX5_IB_QP_PACKET_BASED_CREDIT = 1 << 13, }; struct mlx5_umr_wr { @@ -523,6 +526,7 @@ struct mlx5_ib_srq { struct mlx5_core_srq msrq; struct mlx5_frag_buf buf; struct mlx5_db db; + struct mlx5_frag_buf_ctrl fbc; u64 *wrid; /* protect SRQ hanlding */ @@ -540,7 +544,6 @@ struct mlx5_ib_srq { struct mlx5_ib_xrcd { struct ib_xrcd ibxrcd; u32 xrcdn; - u16 uid; }; enum mlx5_ib_mtt_access_flags { @@ -774,14 +777,16 @@ enum mlx5_ib_stages { MLX5_IB_STAGE_CAPS, MLX5_IB_STAGE_NON_DEFAULT_CB, MLX5_IB_STAGE_ROCE, + MLX5_IB_STAGE_SRQ, MLX5_IB_STAGE_DEVICE_RESOURCES, + MLX5_IB_STAGE_DEVICE_NOTIFIER, MLX5_IB_STAGE_ODP, MLX5_IB_STAGE_COUNTERS, MLX5_IB_STAGE_CONG_DEBUGFS, MLX5_IB_STAGE_UAR, MLX5_IB_STAGE_BFREG, MLX5_IB_STAGE_PRE_IB_REG_UMR, - MLX5_IB_STAGE_SPECS, + MLX5_IB_STAGE_WHITELIST_UID, MLX5_IB_STAGE_IB_REG, MLX5_IB_STAGE_POST_IB_REG_UMR, MLX5_IB_STAGE_DELAY_DROP, @@ -806,6 +811,7 @@ struct mlx5_ib_multiport_info { struct list_head list; struct mlx5_ib_dev *ibdev; struct mlx5_core_dev *mdev; + struct notifier_block mdev_events; struct completion unref_comp; u64 sys_image_guid; u32 mdev_refcnt; @@ -880,10 +886,19 @@ struct mlx5_ib_lb_state { bool enabled; }; +struct mlx5_ib_pf_eq { + struct mlx5_ib_dev *dev; + struct mlx5_eq *core; + struct work_struct work; + spinlock_t lock; /* Pagefaults spinlock */ + struct workqueue_struct *wq; + mempool_t *pool; +}; + struct mlx5_ib_dev { struct ib_device ib_dev; - const struct uverbs_object_tree_def *driver_trees[7]; struct mlx5_core_dev *mdev; + struct notifier_block mdev_events; struct mlx5_roce roce[MLX5_MAX_PORTS]; int num_ports; /* serialize update of capability mask @@ -902,6 +917,8 @@ struct mlx5_ib_dev { #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING struct ib_odp_caps odp_caps; 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. @@ -927,6 +944,7 @@ struct mlx5_ib_dev { u64 sys_image_guid; struct mlx5_memic memic; u16 devx_whitelist_uid; + struct mlx5_srq_table srq_table; }; static inline struct mlx5_ib_cq *to_mibcq(struct mlx5_core_cq *mcq) @@ -1158,9 +1176,8 @@ struct ib_mr *mlx5_ib_reg_dm_mr(struct ib_pd *pd, struct ib_dm *dm, #ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev); -void mlx5_ib_pfault(struct mlx5_core_dev *mdev, void *context, - struct mlx5_pagefault *pfault); int mlx5_ib_odp_init_one(struct mlx5_ib_dev *ibdev); +void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *ibdev); int __init mlx5_ib_odp_init(void); void mlx5_ib_odp_cleanup(void); void mlx5_ib_invalidate_range(struct ib_umem_odp *umem_odp, unsigned long start, @@ -1175,6 +1192,7 @@ static inline void mlx5_ib_internal_fill_odp_caps(struct mlx5_ib_dev *dev) } static inline int mlx5_ib_odp_init_one(struct mlx5_ib_dev *ibdev) { return 0; } +static inline void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *ibdev) {} static inline int mlx5_ib_odp_init(void) { return 0; } static inline void mlx5_ib_odp_cleanup(void) {} static inline void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) {} @@ -1250,32 +1268,29 @@ void mlx5_ib_put_native_port_mdev(struct mlx5_ib_dev *dev, u8 port_num); #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) -int mlx5_ib_devx_create(struct mlx5_ib_dev *dev); +int mlx5_ib_devx_create(struct mlx5_ib_dev *dev, bool is_user); void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid); const struct uverbs_object_tree_def *mlx5_ib_get_devx_tree(void); +extern const struct uapi_definition mlx5_ib_devx_defs[]; +extern const struct uapi_definition mlx5_ib_flow_defs[]; struct mlx5_ib_flow_handler *mlx5_ib_raw_fs_rule_add( struct mlx5_ib_dev *dev, struct mlx5_ib_flow_matcher *fs_matcher, - struct mlx5_flow_act *flow_act, void *cmd_in, int inlen, - int dest_id, int dest_type); + struct mlx5_flow_act *flow_act, u32 counter_id, + void *cmd_in, int inlen, int dest_id, int dest_type); bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type); +bool mlx5_ib_devx_is_flow_counter(void *obj, u32 *counter_id); int mlx5_ib_get_flow_trees(const struct uverbs_object_tree_def **root); void mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction); #else static inline int -mlx5_ib_devx_create(struct mlx5_ib_dev *dev) { return -EOPNOTSUPP; }; +mlx5_ib_devx_create(struct mlx5_ib_dev *dev, + bool is_user) { return -EOPNOTSUPP; } static inline void mlx5_ib_devx_destroy(struct mlx5_ib_dev *dev, u16 uid) {} -static inline const struct uverbs_object_tree_def * -mlx5_ib_get_devx_tree(void) { return NULL; } static inline bool mlx5_ib_devx_is_flow_dest(void *obj, int *dest_id, int *dest_type) { return false; } -static inline int -mlx5_ib_get_flow_trees(const struct uverbs_object_tree_def **root) -{ - return 0; -} static inline void mlx5_ib_destroy_flow_action_raw(struct mlx5_ib_flow_action *maction) { diff --git a/drivers/infiniband/hw/mlx5/odp.c b/drivers/infiniband/hw/mlx5/odp.c index 2cc3d69ab6f6..4ead8c0fff5a 100644 --- a/drivers/infiniband/hw/mlx5/odp.c +++ b/drivers/infiniband/hw/mlx5/odp.c @@ -37,6 +37,46 @@ #include "mlx5_ib.h" #include "cmd.h" +#include <linux/mlx5/eq.h> + +/* Contains the details of a pagefault. */ +struct mlx5_pagefault { + u32 bytes_committed; + u32 token; + u8 event_subtype; + u8 type; + union { + /* Initiator or send message responder pagefault details. */ + struct { + /* Received packet size, only valid for responders. */ + u32 packet_size; + /* + * Number of resource holding WQE, depends on type. + */ + u32 wq_num; + /* + * WQE index. Refers to either the send queue or + * receive queue, according to event_subtype. + */ + u16 wqe_index; + } wqe; + /* RDMA responder pagefault details */ + struct { + u32 r_key; + /* + * Received packet size, minimal size page fault + * resolution required for forward progress. + */ + u32 packet_size; + u32 rdma_op_len; + u64 rdma_va; + } rdma; + }; + + struct mlx5_ib_pf_eq *eq; + struct work_struct work; +}; + #define MAX_PREFETCH_LEN (4*1024*1024U) /* Timeout in ms to wait for an active mmu notifier to complete when handling @@ -304,14 +344,20 @@ static void mlx5_ib_page_fault_resume(struct mlx5_ib_dev *dev, { int wq_num = pfault->event_subtype == MLX5_PFAULT_SUBTYPE_WQE ? pfault->wqe.wq_num : pfault->token; - int ret = mlx5_core_page_fault_resume(dev->mdev, - pfault->token, - wq_num, - pfault->type, - error); - if (ret) - mlx5_ib_err(dev, "Failed to resolve the page fault on WQ 0x%x\n", - wq_num); + u32 out[MLX5_ST_SZ_DW(page_fault_resume_out)] = { }; + u32 in[MLX5_ST_SZ_DW(page_fault_resume_in)] = { }; + int err; + + MLX5_SET(page_fault_resume_in, in, opcode, MLX5_CMD_OP_PAGE_FAULT_RESUME); + MLX5_SET(page_fault_resume_in, in, page_fault_type, pfault->type); + MLX5_SET(page_fault_resume_in, in, token, pfault->token); + MLX5_SET(page_fault_resume_in, in, wq_number, wq_num); + MLX5_SET(page_fault_resume_in, in, error, !!error); + + err = mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); + if (err) + mlx5_ib_err(dev, "Failed to resolve the page fault on WQ 0x%x err %d\n", + wq_num, err); } static struct mlx5_ib_mr *implicit_mr_alloc(struct ib_pd *pd, @@ -607,8 +653,8 @@ out: if (!wait_for_completion_timeout( &odp->notifier_completion, timeout)) { - mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d\n", - current_seq, odp->notifiers_seq); + mlx5_ib_warn(dev, "timeout waiting for mmu notifier. seq %d against %d. notifiers_count=%d\n", + current_seq, odp->notifiers_seq, odp->notifiers_count); } } else { /* The MR is being killed, kill the QP as well. */ @@ -1026,16 +1072,31 @@ invalid_transport_or_opcode: return 0; } -static struct mlx5_ib_qp *mlx5_ib_odp_find_qp(struct mlx5_ib_dev *dev, - u32 wq_num) +static inline struct mlx5_core_rsc_common *odp_get_rsc(struct mlx5_ib_dev *dev, + u32 wq_num, int pf_type) { - struct mlx5_core_qp *mqp = __mlx5_qp_lookup(dev->mdev, wq_num); + enum mlx5_res_type res_type; - if (!mqp) { - mlx5_ib_err(dev, "QPN 0x%6x not found\n", wq_num); + switch (pf_type) { + case MLX5_WQE_PF_TYPE_RMP: + res_type = MLX5_RES_SRQ; + break; + case MLX5_WQE_PF_TYPE_REQ_SEND_OR_WRITE: + case MLX5_WQE_PF_TYPE_RESP: + case MLX5_WQE_PF_TYPE_REQ_READ_OR_ATOMIC: + res_type = MLX5_RES_QP; + break; + default: return NULL; } + return mlx5_core_res_hold(dev->mdev, wq_num, res_type); +} + +static inline struct mlx5_ib_qp *res_to_qp(struct mlx5_core_rsc_common *res) +{ + struct mlx5_core_qp *mqp = (struct mlx5_core_qp *)res; + return to_mibqp(mqp); } @@ -1049,18 +1110,30 @@ static void mlx5_ib_mr_wqe_pfault_handler(struct mlx5_ib_dev *dev, int resume_with_error = 1; u16 wqe_index = pfault->wqe.wqe_index; int requestor = pfault->type & MLX5_PFAULT_REQUESTOR; + struct mlx5_core_rsc_common *res; struct mlx5_ib_qp *qp; + res = odp_get_rsc(dev, pfault->wqe.wq_num, pfault->type); + if (!res) { + mlx5_ib_dbg(dev, "wqe page fault for missing resource %d\n", pfault->wqe.wq_num); + return; + } + + switch (res->res) { + case MLX5_RES_QP: + qp = res_to_qp(res); + break; + default: + mlx5_ib_err(dev, "wqe page fault for unsupported type %d\n", pfault->type); + goto resolve_page_fault; + } + buffer = (char *)__get_free_page(GFP_KERNEL); if (!buffer) { mlx5_ib_err(dev, "Error allocating memory for IO page fault handling.\n"); goto resolve_page_fault; } - qp = mlx5_ib_odp_find_qp(dev, pfault->wqe.wq_num); - if (!qp) - goto resolve_page_fault; - ret = mlx5_ib_read_user_wqe(qp, requestor, wqe_index, buffer, PAGE_SIZE, &qp->trans_qp.base); if (ret < 0) { @@ -1100,6 +1173,7 @@ resolve_page_fault: mlx5_ib_dbg(dev, "PAGE FAULT completed. QP 0x%x resume_with_error=%d, type: 0x%x\n", pfault->wqe.wq_num, resume_with_error, pfault->type); + mlx5_core_res_put(res); free_page((unsigned long)buffer); } @@ -1178,10 +1252,8 @@ static void mlx5_ib_mr_rdma_pfault_handler(struct mlx5_ib_dev *dev, } } -void mlx5_ib_pfault(struct mlx5_core_dev *mdev, void *context, - struct mlx5_pagefault *pfault) +static void mlx5_ib_pfault(struct mlx5_ib_dev *dev, struct mlx5_pagefault *pfault) { - struct mlx5_ib_dev *dev = context; u8 event_subtype = pfault->event_subtype; switch (event_subtype) { @@ -1198,6 +1270,203 @@ void mlx5_ib_pfault(struct mlx5_core_dev *mdev, void *context, } } +static void mlx5_ib_eqe_pf_action(struct work_struct *work) +{ + struct mlx5_pagefault *pfault = container_of(work, + struct mlx5_pagefault, + work); + struct mlx5_ib_pf_eq *eq = pfault->eq; + + mlx5_ib_pfault(eq->dev, pfault); + mempool_free(pfault, eq->pool); +} + +static void mlx5_ib_eq_pf_process(struct mlx5_ib_pf_eq *eq) +{ + struct mlx5_eqe_page_fault *pf_eqe; + struct mlx5_pagefault *pfault; + struct mlx5_eqe *eqe; + int cc = 0; + + while ((eqe = mlx5_eq_get_eqe(eq->core, cc))) { + pfault = mempool_alloc(eq->pool, GFP_ATOMIC); + if (!pfault) { + schedule_work(&eq->work); + break; + } + + pf_eqe = &eqe->data.page_fault; + pfault->event_subtype = eqe->sub_type; + pfault->bytes_committed = be32_to_cpu(pf_eqe->bytes_committed); + + mlx5_ib_dbg(eq->dev, + "PAGE_FAULT: subtype: 0x%02x, bytes_committed: 0x%06x\n", + eqe->sub_type, pfault->bytes_committed); + + switch (eqe->sub_type) { + case MLX5_PFAULT_SUBTYPE_RDMA: + /* RDMA based event */ + pfault->type = + be32_to_cpu(pf_eqe->rdma.pftype_token) >> 24; + pfault->token = + be32_to_cpu(pf_eqe->rdma.pftype_token) & + MLX5_24BIT_MASK; + pfault->rdma.r_key = + be32_to_cpu(pf_eqe->rdma.r_key); + pfault->rdma.packet_size = + be16_to_cpu(pf_eqe->rdma.packet_length); + pfault->rdma.rdma_op_len = + be32_to_cpu(pf_eqe->rdma.rdma_op_len); + pfault->rdma.rdma_va = + be64_to_cpu(pf_eqe->rdma.rdma_va); + mlx5_ib_dbg(eq->dev, + "PAGE_FAULT: type:0x%x, token: 0x%06x, r_key: 0x%08x\n", + pfault->type, pfault->token, + pfault->rdma.r_key); + mlx5_ib_dbg(eq->dev, + "PAGE_FAULT: rdma_op_len: 0x%08x, rdma_va: 0x%016llx\n", + pfault->rdma.rdma_op_len, + pfault->rdma.rdma_va); + break; + + case MLX5_PFAULT_SUBTYPE_WQE: + /* WQE based event */ + pfault->type = + (be32_to_cpu(pf_eqe->wqe.pftype_wq) >> 24) & 0x7; + pfault->token = + be32_to_cpu(pf_eqe->wqe.token); + pfault->wqe.wq_num = + be32_to_cpu(pf_eqe->wqe.pftype_wq) & + MLX5_24BIT_MASK; + pfault->wqe.wqe_index = + be16_to_cpu(pf_eqe->wqe.wqe_index); + pfault->wqe.packet_size = + be16_to_cpu(pf_eqe->wqe.packet_length); + mlx5_ib_dbg(eq->dev, + "PAGE_FAULT: type:0x%x, token: 0x%06x, wq_num: 0x%06x, wqe_index: 0x%04x\n", + pfault->type, pfault->token, + pfault->wqe.wq_num, + pfault->wqe.wqe_index); + break; + + default: + mlx5_ib_warn(eq->dev, + "Unsupported page fault event sub-type: 0x%02hhx\n", + eqe->sub_type); + /* Unsupported page faults should still be + * resolved by the page fault handler + */ + } + + pfault->eq = eq; + INIT_WORK(&pfault->work, mlx5_ib_eqe_pf_action); + queue_work(eq->wq, &pfault->work); + + cc = mlx5_eq_update_cc(eq->core, ++cc); + } + + mlx5_eq_update_ci(eq->core, cc, 1); +} + +static irqreturn_t mlx5_ib_eq_pf_int(int irq, void *eq_ptr) +{ + struct mlx5_ib_pf_eq *eq = eq_ptr; + unsigned long flags; + + if (spin_trylock_irqsave(&eq->lock, flags)) { + mlx5_ib_eq_pf_process(eq); + spin_unlock_irqrestore(&eq->lock, flags); + } else { + schedule_work(&eq->work); + } + + return IRQ_HANDLED; +} + +/* mempool_refill() was proposed but unfortunately wasn't accepted + * http://lkml.iu.edu/hypermail/linux/kernel/1512.1/05073.html + * Cheap workaround. + */ +static void mempool_refill(mempool_t *pool) +{ + while (pool->curr_nr < pool->min_nr) + mempool_free(mempool_alloc(pool, GFP_KERNEL), pool); +} + +static void mlx5_ib_eq_pf_action(struct work_struct *work) +{ + struct mlx5_ib_pf_eq *eq = + container_of(work, struct mlx5_ib_pf_eq, work); + + mempool_refill(eq->pool); + + spin_lock_irq(&eq->lock); + mlx5_ib_eq_pf_process(eq); + spin_unlock_irq(&eq->lock); +} + +enum { + MLX5_IB_NUM_PF_EQE = 0x1000, + MLX5_IB_NUM_PF_DRAIN = 64, +}; + +static int +mlx5_ib_create_pf_eq(struct mlx5_ib_dev *dev, struct mlx5_ib_pf_eq *eq) +{ + struct mlx5_eq_param param = {}; + int err; + + INIT_WORK(&eq->work, mlx5_ib_eq_pf_action); + spin_lock_init(&eq->lock); + eq->dev = dev; + + eq->pool = mempool_create_kmalloc_pool(MLX5_IB_NUM_PF_DRAIN, + sizeof(struct mlx5_pagefault)); + if (!eq->pool) + return -ENOMEM; + + eq->wq = alloc_workqueue("mlx5_ib_page_fault", + WQ_HIGHPRI | WQ_UNBOUND | WQ_MEM_RECLAIM, + MLX5_NUM_CMD_EQE); + if (!eq->wq) { + err = -ENOMEM; + goto err_mempool; + } + + param = (struct mlx5_eq_param) { + .index = MLX5_EQ_PFAULT_IDX, + .mask = 1 << MLX5_EVENT_TYPE_PAGE_FAULT, + .nent = MLX5_IB_NUM_PF_EQE, + .context = eq, + .handler = mlx5_ib_eq_pf_int + }; + eq->core = mlx5_eq_create_generic(dev->mdev, "mlx5_ib_page_fault_eq", ¶m); + if (IS_ERR(eq->core)) { + err = PTR_ERR(eq->core); + goto err_wq; + } + + return 0; +err_wq: + destroy_workqueue(eq->wq); +err_mempool: + mempool_destroy(eq->pool); + return err; +} + +static int +mlx5_ib_destroy_pf_eq(struct mlx5_ib_dev *dev, struct mlx5_ib_pf_eq *eq) +{ + int err; + + err = mlx5_eq_destroy_generic(dev->mdev, eq->core); + cancel_work_sync(&eq->work); + destroy_workqueue(eq->wq); + mempool_destroy(eq->pool); + + return err; +} + void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) { if (!(ent->dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT)) @@ -1226,7 +1495,7 @@ void mlx5_odp_init_mr_cache_entry(struct mlx5_cache_ent *ent) int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) { - int ret; + int ret = 0; if (dev->odp_caps.general_caps & IB_ODP_SUPPORT_IMPLICIT) { ret = mlx5_cmd_null_mkey(dev->mdev, &dev->null_mkey); @@ -1236,7 +1505,20 @@ int mlx5_ib_odp_init_one(struct mlx5_ib_dev *dev) } } - return 0; + if (!MLX5_CAP_GEN(dev->mdev, pg)) + return ret; + + ret = mlx5_ib_create_pf_eq(dev, &dev->odp_pf_eq); + + return ret; +} + +void mlx5_ib_odp_cleanup_one(struct mlx5_ib_dev *dev) +{ + if (!MLX5_CAP_GEN(dev->mdev, pg)) + return; + + mlx5_ib_destroy_pf_eq(dev, &dev->odp_pf_eq); } int mlx5_ib_odp_init(void) @@ -1246,4 +1528,3 @@ int mlx5_ib_odp_init(void) return 0; } - diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index 3747cc681b18..4a85748aeeea 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -108,21 +108,6 @@ static int is_sqp(enum ib_qp_type qp_type) return is_qp0(qp_type) || is_qp1(qp_type); } -static void *get_wqe(struct mlx5_ib_qp *qp, int offset) -{ - return mlx5_buf_offset(&qp->buf, offset); -} - -static void *get_recv_wqe(struct mlx5_ib_qp *qp, int n) -{ - return get_wqe(qp, qp->rq.offset + (n << qp->rq.wqe_shift)); -} - -void *mlx5_get_send_wqe(struct mlx5_ib_qp *qp, int n) -{ - return get_wqe(qp, qp->sq.offset + (n << MLX5_IB_SQ_STRIDE)); -} - /** * mlx5_ib_read_user_wqe() - Copy a user-space WQE to kernel space. * @@ -790,6 +775,7 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, __be64 *pas; void *qpc; int err; + u16 uid; err = ib_copy_from_udata(&ucmd, udata, sizeof(ucmd)); if (err) { @@ -851,7 +837,8 @@ static int create_user_qp(struct mlx5_ib_dev *dev, struct ib_pd *pd, goto err_umem; } - MLX5_SET(create_qp_in, *in, uid, to_mpd(pd)->uid); + uid = (attr->qp_type != IB_QPT_XRC_TGT) ? to_mpd(pd)->uid : 0; + MLX5_SET(create_qp_in, *in, uid, uid); pas = (__be64 *)MLX5_ADDR_OF(create_qp_in, *in, pas); if (ubuffer->umem) mlx5_ib_populate_pas(dev, ubuffer->umem, page_shift, pas, 0); @@ -917,6 +904,30 @@ static void destroy_qp_user(struct mlx5_ib_dev *dev, struct ib_pd *pd, mlx5_ib_free_bfreg(dev, &context->bfregi, qp->bfregn); } +/* get_sq_edge - Get the next nearby edge. + * + * An 'edge' is defined as the first following address after the end + * of the fragment or the SQ. Accordingly, during the WQE construction + * which repetitively increases the pointer to write the next data, it + * simply should check if it gets to an edge. + * + * @sq - SQ buffer. + * @idx - Stride index in the SQ buffer. + * + * Return: + * The new edge. + */ +static void *get_sq_edge(struct mlx5_ib_wq *sq, u32 idx) +{ + void *fragment_end; + + fragment_end = mlx5_frag_buf_get_wqe + (&sq->fbc, + mlx5_frag_buf_get_idx_last_contig_stride(&sq->fbc, idx)); + + return fragment_end + MLX5_SEND_WQE_BB; +} + static int create_kernel_qp(struct mlx5_ib_dev *dev, struct ib_qp_init_attr *init_attr, struct mlx5_ib_qp *qp, @@ -955,13 +966,29 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, qp->sq.offset = qp->rq.wqe_cnt << qp->rq.wqe_shift; base->ubuffer.buf_size = err + (qp->rq.wqe_cnt << qp->rq.wqe_shift); - err = mlx5_buf_alloc(dev->mdev, base->ubuffer.buf_size, &qp->buf); + err = mlx5_frag_buf_alloc_node(dev->mdev, base->ubuffer.buf_size, + &qp->buf, dev->mdev->priv.numa_node); if (err) { mlx5_ib_dbg(dev, "err %d\n", err); return err; } - qp->sq.qend = mlx5_get_send_wqe(qp, qp->sq.wqe_cnt); + if (qp->rq.wqe_cnt) + mlx5_init_fbc(qp->buf.frags, qp->rq.wqe_shift, + ilog2(qp->rq.wqe_cnt), &qp->rq.fbc); + + if (qp->sq.wqe_cnt) { + int sq_strides_offset = (qp->sq.offset & (PAGE_SIZE - 1)) / + MLX5_SEND_WQE_BB; + mlx5_init_fbc_offset(qp->buf.frags + + (qp->sq.offset / PAGE_SIZE), + ilog2(MLX5_SEND_WQE_BB), + ilog2(qp->sq.wqe_cnt), + sq_strides_offset, &qp->sq.fbc); + + qp->sq.cur_edge = get_sq_edge(&qp->sq, 0); + } + *inlen = MLX5_ST_SZ_BYTES(create_qp_in) + MLX5_FLD_SZ_BYTES(create_qp_in, pas[0]) * qp->buf.npages; *in = kvzalloc(*inlen, GFP_KERNEL); @@ -983,8 +1010,9 @@ static int create_kernel_qp(struct mlx5_ib_dev *dev, qp->flags |= MLX5_IB_QP_SQPN_QP1; } - mlx5_fill_page_array(&qp->buf, - (__be64 *)MLX5_ADDR_OF(create_qp_in, *in, pas)); + mlx5_fill_page_frag_array(&qp->buf, + (__be64 *)MLX5_ADDR_OF(create_qp_in, + *in, pas)); err = mlx5_db_alloc(dev->mdev, &qp->db); if (err) { @@ -1024,7 +1052,7 @@ err_free: kvfree(*in); err_buf: - mlx5_buf_free(dev->mdev, &qp->buf); + mlx5_frag_buf_free(dev->mdev, &qp->buf); return err; } @@ -1036,7 +1064,7 @@ static void destroy_qp_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_qp *qp) kvfree(qp->sq.wr_data); kvfree(qp->rq.wrid); mlx5_db_free(dev->mdev, &qp->db); - mlx5_buf_free(dev->mdev, &qp->buf); + mlx5_frag_buf_free(dev->mdev, &qp->buf); } static u32 get_rx_type(struct mlx5_ib_qp *qp, struct ib_qp_init_attr *attr) @@ -1889,7 +1917,8 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, MLX5_QP_FLAG_BFREG_INDEX | MLX5_QP_FLAG_TYPE_DCT | MLX5_QP_FLAG_TYPE_DCI | - MLX5_QP_FLAG_ALLOW_SCATTER_CQE)) + MLX5_QP_FLAG_ALLOW_SCATTER_CQE | + MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE)) return -EINVAL; err = get_qp_user_index(to_mucontext(pd->uobject->context), @@ -1925,6 +1954,15 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, qp->flags_en |= MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC; } + if (ucmd.flags & MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE) { + if (init_attr->qp_type != IB_QPT_RC || + !MLX5_CAP_GEN(dev->mdev, qp_packet_based)) { + mlx5_ib_dbg(dev, "packet based credit mode isn't supported\n"); + return -EOPNOTSUPP; + } + qp->flags |= MLX5_IB_QP_PACKET_BASED_CREDIT; + } + if (init_attr->create_flags & IB_QP_CREATE_SOURCE_QPN) { if (init_attr->qp_type != IB_QPT_UD || (MLX5_CAP_GEN(dev->mdev, port_type) != @@ -2021,7 +2059,8 @@ static int create_qp_common(struct mlx5_ib_dev *dev, struct ib_pd *pd, MLX5_SET(qpc, qpc, cd_slave_send, 1); if (qp->flags & MLX5_IB_QP_MANAGED_RECV) MLX5_SET(qpc, qpc, cd_slave_receive, 1); - + if (qp->flags & MLX5_IB_QP_PACKET_BASED_CREDIT) + MLX5_SET(qpc, qpc, req_e2e_credit_mode, 1); if (qp->scat_cqe && is_connected(init_attr->qp_type)) { configure_responder_scat_cqe(init_attr, qpc); configure_requester_scat_cqe(dev, init_attr, @@ -2663,7 +2702,7 @@ static int ib_rate_to_mlx5(struct mlx5_ib_dev *dev, u8 rate) if (rate == IB_RATE_PORT_CURRENT) return 0; - if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_300_GBPS) + if (rate < IB_RATE_2_5_GBPS || rate > IB_RATE_600_GBPS) return -EINVAL; while (rate != IB_RATE_PORT_CURRENT && @@ -3475,6 +3514,8 @@ static int __mlx5_ib_modify_qp(struct ib_qp *ibqp, qp->sq.head = 0; qp->sq.tail = 0; qp->sq.cur_post = 0; + if (qp->sq.wqe_cnt) + qp->sq.cur_edge = get_sq_edge(&qp->sq, 0); qp->sq.last_poll = 0; qp->db.db[MLX5_RCV_DBR] = 0; qp->db.db[MLX5_SND_DBR] = 0; @@ -3515,7 +3556,7 @@ static bool modify_dci_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state new return is_valid_mask(attr_mask, req, opt); } else if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR) { req |= IB_QP_PATH_MTU; - opt = IB_QP_PKEY_INDEX; + opt = IB_QP_PKEY_INDEX | IB_QP_AV; return is_valid_mask(attr_mask, req, opt); } else if (cur_state == IB_QPS_RTR && new_state == IB_QPS_RTS) { req |= IB_QP_TIMEOUT | IB_QP_RETRY_CNT | IB_QP_RNR_RETRY | @@ -3749,6 +3790,62 @@ out: return err; } +static void _handle_post_send_edge(struct mlx5_ib_wq *sq, void **seg, + u32 wqe_sz, void **cur_edge) +{ + u32 idx; + + idx = (sq->cur_post + (wqe_sz >> 2)) & (sq->wqe_cnt - 1); + *cur_edge = get_sq_edge(sq, idx); + + *seg = mlx5_frag_buf_get_wqe(&sq->fbc, idx); +} + +/* handle_post_send_edge - Check if we get to SQ edge. If yes, update to the + * next nearby edge and get new address translation for current WQE position. + * @sq - SQ buffer. + * @seg: Current WQE position (16B aligned). + * @wqe_sz: Total current WQE size [16B]. + * @cur_edge: Updated current edge. + */ +static inline void handle_post_send_edge(struct mlx5_ib_wq *sq, void **seg, + u32 wqe_sz, void **cur_edge) +{ + if (likely(*seg != *cur_edge)) + return; + + _handle_post_send_edge(sq, seg, wqe_sz, cur_edge); +} + +/* memcpy_send_wqe - copy data from src to WQE and update the relevant WQ's + * pointers. At the end @seg is aligned to 16B regardless the copied size. + * @sq - SQ buffer. + * @cur_edge: Updated current edge. + * @seg: Current WQE position (16B aligned). + * @wqe_sz: Total current WQE size [16B]. + * @src: Pointer to copy from. + * @n: Number of bytes to copy. + */ +static inline void memcpy_send_wqe(struct mlx5_ib_wq *sq, void **cur_edge, + void **seg, u32 *wqe_sz, const void *src, + size_t n) +{ + while (likely(n)) { + size_t leftlen = *cur_edge - *seg; + size_t copysz = min_t(size_t, leftlen, n); + size_t stride; + + memcpy(*seg, src, copysz); + + n -= copysz; + src += copysz; + stride = !n ? ALIGN(copysz, 16) : copysz; + *seg += stride; + *wqe_sz += stride >> 4; + handle_post_send_edge(sq, seg, *wqe_sz, cur_edge); + } +} + static int mlx5_wq_overflow(struct mlx5_ib_wq *wq, int nreq, struct ib_cq *ib_cq) { struct mlx5_ib_cq *cq; @@ -3774,11 +3871,10 @@ static __always_inline void set_raddr_seg(struct mlx5_wqe_raddr_seg *rseg, rseg->reserved = 0; } -static void *set_eth_seg(struct mlx5_wqe_eth_seg *eseg, - const struct ib_send_wr *wr, void *qend, - struct mlx5_ib_qp *qp, int *size) +static void set_eth_seg(const struct ib_send_wr *wr, struct mlx5_ib_qp *qp, + void **seg, int *size, void **cur_edge) { - void *seg = eseg; + struct mlx5_wqe_eth_seg *eseg = *seg; memset(eseg, 0, sizeof(struct mlx5_wqe_eth_seg)); @@ -3786,45 +3882,41 @@ static void *set_eth_seg(struct mlx5_wqe_eth_seg *eseg, eseg->cs_flags = MLX5_ETH_WQE_L3_CSUM | MLX5_ETH_WQE_L4_CSUM; - seg += sizeof(struct mlx5_wqe_eth_seg); - *size += sizeof(struct mlx5_wqe_eth_seg) / 16; - if (wr->opcode == IB_WR_LSO) { struct ib_ud_wr *ud_wr = container_of(wr, struct ib_ud_wr, wr); - int size_of_inl_hdr_start = sizeof(eseg->inline_hdr.start); - u64 left, leftlen, copysz; + size_t left, copysz; void *pdata = ud_wr->header; + size_t stride; left = ud_wr->hlen; eseg->mss = cpu_to_be16(ud_wr->mss); eseg->inline_hdr.sz = cpu_to_be16(left); - /* - * check if there is space till the end of queue, if yes, - * copy all in one shot, otherwise copy till the end of queue, - * rollback and than the copy the left + /* memcpy_send_wqe should get a 16B align address. Hence, we + * first copy up to the current edge and then, if needed, + * fall-through to memcpy_send_wqe. */ - leftlen = qend - (void *)eseg->inline_hdr.start; - copysz = min_t(u64, leftlen, left); - - memcpy(seg - size_of_inl_hdr_start, pdata, copysz); - - if (likely(copysz > size_of_inl_hdr_start)) { - seg += ALIGN(copysz - size_of_inl_hdr_start, 16); - *size += ALIGN(copysz - size_of_inl_hdr_start, 16) / 16; - } - - if (unlikely(copysz < left)) { /* the last wqe in the queue */ - seg = mlx5_get_send_wqe(qp, 0); + copysz = min_t(u64, *cur_edge - (void *)eseg->inline_hdr.start, + left); + memcpy(eseg->inline_hdr.start, pdata, copysz); + stride = ALIGN(sizeof(struct mlx5_wqe_eth_seg) - + sizeof(eseg->inline_hdr.start) + copysz, 16); + *size += stride / 16; + *seg += stride; + + if (copysz < left) { + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); left -= copysz; pdata += copysz; - memcpy(seg, pdata, left); - seg += ALIGN(left, 16); - *size += ALIGN(left, 16) / 16; + memcpy_send_wqe(&qp->sq, cur_edge, seg, size, pdata, + left); } + + return; } - return seg; + *seg += sizeof(struct mlx5_wqe_eth_seg); + *size += sizeof(struct mlx5_wqe_eth_seg) / 16; } static void set_datagram_seg(struct mlx5_wqe_datagram_seg *dseg, @@ -4083,24 +4175,6 @@ static void set_reg_data_seg(struct mlx5_wqe_data_seg *dseg, dseg->lkey = cpu_to_be32(pd->ibpd.local_dma_lkey); } -static void set_reg_umr_inline_seg(void *seg, struct mlx5_ib_qp *qp, - struct mlx5_ib_mr *mr, int mr_list_size) -{ - void *qend = qp->sq.qend; - void *addr = mr->descs; - int copy; - - if (unlikely(seg + mr_list_size > qend)) { - copy = qend - seg; - memcpy(seg, addr, copy); - addr += copy; - mr_list_size -= copy; - seg = mlx5_get_send_wqe(qp, 0); - } - memcpy(seg, addr, mr_list_size); - seg += mr_list_size; -} - static __be32 send_ieth(const struct ib_send_wr *wr) { switch (wr->opcode) { @@ -4134,40 +4208,48 @@ static u8 wq_sig(void *wqe) } static int set_data_inl_seg(struct mlx5_ib_qp *qp, const struct ib_send_wr *wr, - void *wqe, int *sz) + void **wqe, int *wqe_sz, void **cur_edge) { struct mlx5_wqe_inline_seg *seg; - void *qend = qp->sq.qend; - void *addr; + size_t offset; int inl = 0; - int copy; - int len; int i; - seg = wqe; - wqe += sizeof(*seg); + seg = *wqe; + *wqe += sizeof(*seg); + offset = sizeof(*seg); + for (i = 0; i < wr->num_sge; i++) { - addr = (void *)(unsigned long)(wr->sg_list[i].addr); - len = wr->sg_list[i].length; + size_t len = wr->sg_list[i].length; + void *addr = (void *)(unsigned long)(wr->sg_list[i].addr); + inl += len; if (unlikely(inl > qp->max_inline_data)) return -ENOMEM; - if (unlikely(wqe + len > qend)) { - copy = qend - wqe; - memcpy(wqe, addr, copy); - addr += copy; - len -= copy; - wqe = mlx5_get_send_wqe(qp, 0); + while (likely(len)) { + size_t leftlen; + size_t copysz; + + handle_post_send_edge(&qp->sq, wqe, + *wqe_sz + (offset >> 4), + cur_edge); + + leftlen = *cur_edge - *wqe; + copysz = min_t(size_t, leftlen, len); + + memcpy(*wqe, addr, copysz); + len -= copysz; + addr += copysz; + *wqe += copysz; + offset += copysz; } - memcpy(wqe, addr, len); - wqe += len; } seg->byte_count = cpu_to_be32(inl | MLX5_INLINE_SEG); - *sz = ALIGN(inl + sizeof(seg->byte_count), 16) / 16; + *wqe_sz += ALIGN(inl + sizeof(seg->byte_count), 16) / 16; return 0; } @@ -4280,7 +4362,8 @@ static int mlx5_set_bsf(struct ib_mr *sig_mr, } static int set_sig_data_segment(const struct ib_sig_handover_wr *wr, - struct mlx5_ib_qp *qp, void **seg, int *size) + struct mlx5_ib_qp *qp, void **seg, + int *size, void **cur_edge) { struct ib_sig_attrs *sig_attrs = wr->sig_attrs; struct ib_mr *sig_mr = wr->sig_mr; @@ -4364,8 +4447,7 @@ static int set_sig_data_segment(const struct ib_sig_handover_wr *wr, *seg += wqe_size; *size += wqe_size / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); bsf = *seg; ret = mlx5_set_bsf(sig_mr, sig_attrs, bsf, data_len); @@ -4374,8 +4456,7 @@ static int set_sig_data_segment(const struct ib_sig_handover_wr *wr, *seg += sizeof(*bsf); *size += sizeof(*bsf) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); return 0; } @@ -4413,7 +4494,8 @@ static void set_sig_umr_segment(struct mlx5_wqe_umr_ctrl_seg *umr, static int set_sig_umr_wr(const struct ib_send_wr *send_wr, - struct mlx5_ib_qp *qp, void **seg, int *size) + struct mlx5_ib_qp *qp, void **seg, int *size, + void **cur_edge) { const struct ib_sig_handover_wr *wr = sig_handover_wr(send_wr); struct mlx5_ib_mr *sig_mr = to_mmr(wr->sig_mr); @@ -4445,16 +4527,14 @@ static int set_sig_umr_wr(const struct ib_send_wr *send_wr, set_sig_umr_segment(*seg, xlt_size); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); set_sig_mkey_segment(*seg, wr, xlt_size, region_len, pdn); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); - ret = set_sig_data_segment(wr, qp, seg, size); + ret = set_sig_data_segment(wr, qp, seg, size, cur_edge); if (ret) return ret; @@ -4491,11 +4571,11 @@ static int set_psv_wr(struct ib_sig_domain *domain, static int set_reg_wr(struct mlx5_ib_qp *qp, const struct ib_reg_wr *wr, - void **seg, int *size) + void **seg, int *size, void **cur_edge) { struct mlx5_ib_mr *mr = to_mmr(wr->mr); struct mlx5_ib_pd *pd = to_mpd(qp->ibqp.pd); - int mr_list_size = mr->ndescs * mr->desc_size; + size_t mr_list_size = mr->ndescs * mr->desc_size; bool umr_inline = mr_list_size <= MLX5_IB_SQ_UMR_INLINE_THRESHOLD; if (unlikely(wr->wr.send_flags & IB_SEND_INLINE)) { @@ -4507,18 +4587,17 @@ static int set_reg_wr(struct mlx5_ib_qp *qp, set_reg_umr_seg(*seg, mr, umr_inline); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); set_reg_mkey_seg(*seg, mr, wr->key, wr->access); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); if (umr_inline) { - set_reg_umr_inline_seg(*seg, qp, mr, mr_list_size); - *size += get_xlt_octo(mr_list_size); + memcpy_send_wqe(&qp->sq, cur_edge, seg, size, mr->descs, + mr_list_size); + *size = ALIGN(*size, MLX5_SEND_WQE_BB >> 4); } else { set_reg_data_seg(*seg, mr, pd); *seg += sizeof(struct mlx5_wqe_data_seg); @@ -4527,32 +4606,31 @@ static int set_reg_wr(struct mlx5_ib_qp *qp, return 0; } -static void set_linv_wr(struct mlx5_ib_qp *qp, void **seg, int *size) +static void set_linv_wr(struct mlx5_ib_qp *qp, void **seg, int *size, + void **cur_edge) { set_linv_umr_seg(*seg); *seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); *size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); set_linv_mkey_seg(*seg); *seg += sizeof(struct mlx5_mkey_seg); *size += sizeof(struct mlx5_mkey_seg) / 16; - if (unlikely((*seg == qp->sq.qend))) - *seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, seg, *size, cur_edge); } -static void dump_wqe(struct mlx5_ib_qp *qp, int idx, int size_16) +static void dump_wqe(struct mlx5_ib_qp *qp, u32 idx, int size_16) { __be32 *p = NULL; - int tidx = idx; + u32 tidx = idx; int i, j; - pr_debug("dump wqe at %p\n", mlx5_get_send_wqe(qp, tidx)); + pr_debug("dump WQE index %u:\n", idx); for (i = 0, j = 0; i < size_16 * 4; i += 4, j += 4) { if ((i & 0xf) == 0) { - void *buf = mlx5_get_send_wqe(qp, tidx); tidx = (tidx + 1) & (qp->sq.wqe_cnt - 1); - p = buf; + p = mlx5_frag_buf_get_wqe(&qp->sq.fbc, tidx); + pr_debug("WQBB at %p:\n", (void *)p); j = 0; } pr_debug("%08x %08x %08x %08x\n", be32_to_cpu(p[j]), @@ -4562,15 +4640,16 @@ static void dump_wqe(struct mlx5_ib_qp *qp, int idx, int size_16) } static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, - struct mlx5_wqe_ctrl_seg **ctrl, - const struct ib_send_wr *wr, unsigned *idx, - int *size, int nreq, bool send_signaled, bool solicited) + struct mlx5_wqe_ctrl_seg **ctrl, + const struct ib_send_wr *wr, unsigned int *idx, + int *size, void **cur_edge, int nreq, + bool send_signaled, bool solicited) { if (unlikely(mlx5_wq_overflow(&qp->sq, nreq, qp->ibqp.send_cq))) return -ENOMEM; *idx = qp->sq.cur_post & (qp->sq.wqe_cnt - 1); - *seg = mlx5_get_send_wqe(qp, *idx); + *seg = mlx5_frag_buf_get_wqe(&qp->sq.fbc, *idx); *ctrl = *seg; *(uint32_t *)(*seg + 8) = 0; (*ctrl)->imm = send_ieth(wr); @@ -4580,6 +4659,7 @@ static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, *seg += sizeof(**ctrl); *size = sizeof(**ctrl) / 16; + *cur_edge = qp->sq.cur_edge; return 0; } @@ -4587,17 +4667,18 @@ static int __begin_wqe(struct mlx5_ib_qp *qp, void **seg, static int begin_wqe(struct mlx5_ib_qp *qp, void **seg, struct mlx5_wqe_ctrl_seg **ctrl, const struct ib_send_wr *wr, unsigned *idx, - int *size, int nreq) + int *size, void **cur_edge, int nreq) { - return __begin_wqe(qp, seg, ctrl, wr, idx, size, nreq, + return __begin_wqe(qp, seg, ctrl, wr, idx, size, cur_edge, nreq, wr->send_flags & IB_SEND_SIGNALED, wr->send_flags & IB_SEND_SOLICITED); } static void finish_wqe(struct mlx5_ib_qp *qp, struct mlx5_wqe_ctrl_seg *ctrl, - u8 size, unsigned idx, u64 wr_id, - int nreq, u8 fence, u32 mlx5_opcode) + void *seg, u8 size, void *cur_edge, + unsigned int idx, u64 wr_id, int nreq, u8 fence, + u32 mlx5_opcode) { u8 opmod = 0; @@ -4613,6 +4694,15 @@ static void finish_wqe(struct mlx5_ib_qp *qp, qp->sq.wqe_head[idx] = qp->sq.head + nreq; qp->sq.cur_post += DIV_ROUND_UP(size * 16, MLX5_SEND_WQE_BB); qp->sq.w_list[idx].next = qp->sq.cur_post; + + /* We save the edge which was possibly updated during the WQE + * construction, into SQ's cache. + */ + seg = PTR_ALIGN(seg, MLX5_SEND_WQE_BB); + qp->sq.cur_edge = (unlikely(seg == cur_edge)) ? + get_sq_edge(&qp->sq, qp->sq.cur_post & + (qp->sq.wqe_cnt - 1)) : + cur_edge; } static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, @@ -4623,11 +4713,10 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, struct mlx5_core_dev *mdev = dev->mdev; struct mlx5_ib_qp *qp; struct mlx5_ib_mr *mr; - struct mlx5_wqe_data_seg *dpseg; struct mlx5_wqe_xrc_seg *xrc; struct mlx5_bf *bf; + void *cur_edge; int uninitialized_var(size); - void *qend; unsigned long flags; unsigned idx; int err = 0; @@ -4649,7 +4738,6 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, qp = to_mqp(ibqp); bf = &qp->bf; - qend = qp->sq.qend; spin_lock_irqsave(&qp->sq.lock, flags); @@ -4669,7 +4757,8 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, goto out; } - err = begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, nreq); + err = begin_wqe(qp, &seg, &ctrl, wr, &idx, &size, &cur_edge, + nreq); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; @@ -4719,14 +4808,15 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, case IB_WR_LOCAL_INV: qp->sq.wr_data[idx] = IB_WR_LOCAL_INV; ctrl->imm = cpu_to_be32(wr->ex.invalidate_rkey); - set_linv_wr(qp, &seg, &size); + set_linv_wr(qp, &seg, &size, &cur_edge); num_sge = 0; break; case IB_WR_REG_MR: qp->sq.wr_data[idx] = IB_WR_REG_MR; ctrl->imm = cpu_to_be32(reg_wr(wr)->key); - err = set_reg_wr(qp, reg_wr(wr), &seg, &size); + err = set_reg_wr(qp, reg_wr(wr), &seg, &size, + &cur_edge); if (err) { *bad_wr = wr; goto out; @@ -4739,21 +4829,24 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, mr = to_mmr(sig_handover_wr(wr)->sig_mr); ctrl->imm = cpu_to_be32(mr->ibmr.rkey); - err = set_sig_umr_wr(wr, qp, &seg, &size); + err = set_sig_umr_wr(wr, qp, &seg, &size, + &cur_edge); if (err) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } - finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, - fence, MLX5_OPCODE_UMR); + finish_wqe(qp, ctrl, seg, size, cur_edge, idx, + wr->wr_id, nreq, fence, + MLX5_OPCODE_UMR); /* * SET_PSV WQEs are not signaled and solicited * on error */ err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, - &size, nreq, false, true); + &size, &cur_edge, nreq, false, + true); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; @@ -4770,10 +4863,12 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, goto out; } - finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, - fence, MLX5_OPCODE_SET_PSV); + finish_wqe(qp, ctrl, seg, size, cur_edge, idx, + wr->wr_id, nreq, fence, + MLX5_OPCODE_SET_PSV); err = __begin_wqe(qp, &seg, &ctrl, wr, &idx, - &size, nreq, false, true); + &size, &cur_edge, nreq, false, + true); if (err) { mlx5_ib_warn(dev, "\n"); err = -ENOMEM; @@ -4790,8 +4885,9 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, goto out; } - finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, - fence, MLX5_OPCODE_SET_PSV); + finish_wqe(qp, ctrl, seg, size, cur_edge, idx, + wr->wr_id, nreq, fence, + MLX5_OPCODE_SET_PSV); qp->next_fence = MLX5_FENCE_MODE_INITIATOR_SMALL; num_sge = 0; goto skip_psv; @@ -4828,16 +4924,14 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, set_datagram_seg(seg, wr); seg += sizeof(struct mlx5_wqe_datagram_seg); size += sizeof(struct mlx5_wqe_datagram_seg) / 16; - if (unlikely((seg == qend))) - seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); + break; case IB_QPT_UD: set_datagram_seg(seg, wr); seg += sizeof(struct mlx5_wqe_datagram_seg); size += sizeof(struct mlx5_wqe_datagram_seg) / 16; - - if (unlikely((seg == qend))) - seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); /* handle qp that supports ud offload */ if (qp->flags & IB_QP_CREATE_IPOIB_UD_LSO) { @@ -4847,11 +4941,9 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, memset(pad, 0, sizeof(struct mlx5_wqe_eth_pad)); seg += sizeof(struct mlx5_wqe_eth_pad); size += sizeof(struct mlx5_wqe_eth_pad) / 16; - - seg = set_eth_seg(seg, wr, qend, qp, &size); - - if (unlikely((seg == qend))) - seg = mlx5_get_send_wqe(qp, 0); + set_eth_seg(wr, qp, &seg, &size, &cur_edge); + handle_post_send_edge(&qp->sq, &seg, size, + &cur_edge); } break; case MLX5_IB_QPT_REG_UMR: @@ -4867,13 +4959,11 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, goto out; seg += sizeof(struct mlx5_wqe_umr_ctrl_seg); size += sizeof(struct mlx5_wqe_umr_ctrl_seg) / 16; - if (unlikely((seg == qend))) - seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); set_reg_mkey_segment(seg, wr); seg += sizeof(struct mlx5_mkey_seg); size += sizeof(struct mlx5_mkey_seg) / 16; - if (unlikely((seg == qend))) - seg = mlx5_get_send_wqe(qp, 0); + handle_post_send_edge(&qp->sq, &seg, size, &cur_edge); break; default: @@ -4881,33 +4971,29 @@ static int _mlx5_ib_post_send(struct ib_qp *ibqp, const struct ib_send_wr *wr, } if (wr->send_flags & IB_SEND_INLINE && num_sge) { - int uninitialized_var(sz); - - err = set_data_inl_seg(qp, wr, seg, &sz); + err = set_data_inl_seg(qp, wr, &seg, &size, &cur_edge); if (unlikely(err)) { mlx5_ib_warn(dev, "\n"); *bad_wr = wr; goto out; } - size += sz; } else { - dpseg = seg; for (i = 0; i < num_sge; i++) { - if (unlikely(dpseg == qend)) { - seg = mlx5_get_send_wqe(qp, 0); - dpseg = seg; - } + handle_post_send_edge(&qp->sq, &seg, size, + &cur_edge); if (likely(wr->sg_list[i].length)) { - set_data_ptr_seg(dpseg, wr->sg_list + i); + set_data_ptr_seg + ((struct mlx5_wqe_data_seg *)seg, + wr->sg_list + i); size += sizeof(struct mlx5_wqe_data_seg) / 16; - dpseg++; + seg += sizeof(struct mlx5_wqe_data_seg); } } } qp->next_fence = next_fence; - finish_wqe(qp, ctrl, size, idx, wr->wr_id, nreq, fence, - mlx5_ib_opcode[wr->opcode]); + finish_wqe(qp, ctrl, seg, size, cur_edge, idx, wr->wr_id, nreq, + fence, mlx5_ib_opcode[wr->opcode]); skip_psv: if (0) dump_wqe(qp, idx, size); @@ -4993,7 +5079,7 @@ static int _mlx5_ib_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, goto out; } - scat = get_recv_wqe(qp, ind); + scat = mlx5_frag_buf_get_wqe(&qp->rq.fbc, ind); if (qp->wq_sig) scat++; @@ -5441,7 +5527,6 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, struct mlx5_ib_dev *dev = to_mdev(ibdev); struct mlx5_ib_xrcd *xrcd; int err; - u16 uid; if (!MLX5_CAP_GEN(dev->mdev, xrc)) return ERR_PTR(-ENOSYS); @@ -5450,14 +5535,12 @@ struct ib_xrcd *mlx5_ib_alloc_xrcd(struct ib_device *ibdev, if (!xrcd) return ERR_PTR(-ENOMEM); - uid = context ? to_mucontext(context)->devx_uid : 0; - err = mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, uid); + err = mlx5_cmd_xrcd_alloc(dev->mdev, &xrcd->xrcdn, 0); if (err) { kfree(xrcd); return ERR_PTR(-ENOMEM); } - xrcd->uid = uid; return &xrcd->ibxrcd; } @@ -5465,10 +5548,9 @@ int mlx5_ib_dealloc_xrcd(struct ib_xrcd *xrcd) { struct mlx5_ib_dev *dev = to_mdev(xrcd->device); u32 xrcdn = to_mxrcd(xrcd)->xrcdn; - u16 uid = to_mxrcd(xrcd)->uid; int err; - err = mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, uid); + err = mlx5_cmd_xrcd_dealloc(dev->mdev, xrcdn, 0); if (err) mlx5_ib_warn(dev, "failed to dealloc xrcdn 0x%x\n", xrcdn); diff --git a/drivers/infiniband/hw/mlx5/srq.c b/drivers/infiniband/hw/mlx5/srq.c index d012e7dbcc38..0413b10dea71 100644 --- a/drivers/infiniband/hw/mlx5/srq.c +++ b/drivers/infiniband/hw/mlx5/srq.c @@ -1,50 +1,19 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright (c) 2013-2015, Mellanox Technologies. 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 - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright (c) 2013-2018, Mellanox Technologies inc. All rights reserved. */ #include <linux/module.h> #include <linux/mlx5/qp.h> -#include <linux/mlx5/srq.h> #include <linux/slab.h> #include <rdma/ib_umem.h> #include <rdma/ib_user_verbs.h> - #include "mlx5_ib.h" - -/* not supported currently */ -static int srq_signature; +#include "srq.h" static void *get_wqe(struct mlx5_ib_srq *srq, int n) { - return mlx5_buf_offset(&srq->buf, n << srq->msrq.wqe_shift); + return mlx5_frag_buf_get_wqe(&srq->fbc, n); } static void mlx5_ib_srq_event(struct mlx5_core_srq *srq, enum mlx5_event type) @@ -144,7 +113,7 @@ static int create_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq, in->log_page_size = page_shift - MLX5_ADAPTER_PAGE_SHIFT; in->page_offset = offset; - in->uid = to_mpd(pd)->uid; + in->uid = (in->type != IB_SRQT_XRC) ? to_mpd(pd)->uid : 0; if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && in->type != IB_SRQT_BASIC) in->user_index = uidx; @@ -173,12 +142,16 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, return err; } - if (mlx5_buf_alloc(dev->mdev, buf_size, &srq->buf)) { + if (mlx5_frag_buf_alloc_node(dev->mdev, buf_size, &srq->buf, + dev->mdev->priv.numa_node)) { mlx5_ib_dbg(dev, "buf alloc failed\n"); err = -ENOMEM; goto err_db; } + mlx5_init_fbc(srq->buf.frags, srq->msrq.wqe_shift, ilog2(srq->msrq.max), + &srq->fbc); + srq->head = 0; srq->tail = srq->msrq.max - 1; srq->wqe_ctr = 0; @@ -195,14 +168,14 @@ static int create_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq, err = -ENOMEM; goto err_buf; } - mlx5_fill_page_array(&srq->buf, in->pas); + mlx5_fill_page_frag_array(&srq->buf, in->pas); srq->wrid = kvmalloc_array(srq->msrq.max, sizeof(u64), GFP_KERNEL); if (!srq->wrid) { err = -ENOMEM; goto err_in; } - srq->wq_sig = !!srq_signature; + srq->wq_sig = 0; in->log_page_size = srq->buf.page_shift - MLX5_ADAPTER_PAGE_SHIFT; if (MLX5_CAP_GEN(dev->mdev, cqe_version) == MLX5_CQE_VERSION_V1 && @@ -215,7 +188,7 @@ err_in: kvfree(in->pas); err_buf: - mlx5_buf_free(dev->mdev, &srq->buf); + mlx5_frag_buf_free(dev->mdev, &srq->buf); err_db: mlx5_db_free(dev->mdev, &srq->db); @@ -232,7 +205,7 @@ static void destroy_srq_user(struct ib_pd *pd, struct mlx5_ib_srq *srq) static void destroy_srq_kernel(struct mlx5_ib_dev *dev, struct mlx5_ib_srq *srq) { kvfree(srq->wrid); - mlx5_buf_free(dev->mdev, &srq->buf); + mlx5_frag_buf_free(dev->mdev, &srq->buf); mlx5_db_free(dev->mdev, &srq->db); } @@ -327,7 +300,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, in.pd = to_mpd(pd)->pdn; in.db_record = srq->db.dma; - err = mlx5_core_create_srq(dev->mdev, &srq->msrq, &in); + err = mlx5_cmd_create_srq(dev, &srq->msrq, &in); kvfree(in.pas); if (err) { mlx5_ib_dbg(dev, "create SRQ failed, err %d\n", err); @@ -351,7 +324,7 @@ struct ib_srq *mlx5_ib_create_srq(struct ib_pd *pd, return &srq->ibsrq; err_core: - mlx5_core_destroy_srq(dev->mdev, &srq->msrq); + mlx5_cmd_destroy_srq(dev, &srq->msrq); err_usr_kern_srq: if (pd->uobject) @@ -381,7 +354,7 @@ int mlx5_ib_modify_srq(struct ib_srq *ibsrq, struct ib_srq_attr *attr, return -EINVAL; mutex_lock(&srq->mutex); - ret = mlx5_core_arm_srq(dev->mdev, &srq->msrq, attr->srq_limit, 1); + ret = mlx5_cmd_arm_srq(dev, &srq->msrq, attr->srq_limit, 1); mutex_unlock(&srq->mutex); if (ret) @@ -402,7 +375,7 @@ int mlx5_ib_query_srq(struct ib_srq *ibsrq, struct ib_srq_attr *srq_attr) if (!out) return -ENOMEM; - ret = mlx5_core_query_srq(dev->mdev, &srq->msrq, out); + ret = mlx5_cmd_query_srq(dev, &srq->msrq, out); if (ret) goto out_box; @@ -420,7 +393,7 @@ int mlx5_ib_destroy_srq(struct ib_srq *srq) struct mlx5_ib_dev *dev = to_mdev(srq->device); struct mlx5_ib_srq *msrq = to_msrq(srq); - mlx5_core_destroy_srq(dev->mdev, &msrq->msrq); + mlx5_cmd_destroy_srq(dev, &msrq->msrq); if (srq->uobject) { mlx5_ib_db_unmap_user(to_mucontext(srq->uobject->context), &msrq->db); diff --git a/drivers/infiniband/hw/mlx5/srq.h b/drivers/infiniband/hw/mlx5/srq.h new file mode 100644 index 000000000000..75eb5839ae95 --- /dev/null +++ b/drivers/infiniband/hw/mlx5/srq.h @@ -0,0 +1,73 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* + * Copyright (c) 2013-2018, Mellanox Technologies. All rights reserved. + */ + +#ifndef MLX5_IB_SRQ_H +#define MLX5_IB_SRQ_H + +enum { + MLX5_SRQ_FLAG_ERR = (1 << 0), + MLX5_SRQ_FLAG_WQ_SIG = (1 << 1), + MLX5_SRQ_FLAG_RNDV = (1 << 2), +}; + +struct mlx5_srq_attr { + u32 type; + u32 flags; + u32 log_size; + u32 wqe_shift; + u32 log_page_size; + u32 wqe_cnt; + u32 srqn; + u32 xrcd; + u32 page_offset; + u32 cqn; + u32 pd; + u32 lwm; + u32 user_index; + u64 db_record; + __be64 *pas; + u32 tm_log_list_size; + u32 tm_next_tag; + u32 tm_hw_phase_cnt; + u32 tm_sw_phase_cnt; + u16 uid; +}; + +struct mlx5_ib_dev; + +struct mlx5_core_srq { + struct mlx5_core_rsc_common common; /* must be first */ + u32 srqn; + int max; + size_t max_gs; + size_t max_avail_gather; + int wqe_shift; + void (*event)(struct mlx5_core_srq *srq, enum mlx5_event e); + + atomic_t refcount; + struct completion free; + u16 uid; +}; + +struct mlx5_srq_table { + struct notifier_block nb; + /* protect radix tree + */ + spinlock_t lock; + struct radix_tree_root tree; +}; + +int mlx5_cmd_create_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_srq_attr *in); +int mlx5_cmd_destroy_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq); +int mlx5_cmd_query_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_srq_attr *out); +int mlx5_cmd_arm_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + u16 lwm, int is_srq); +struct mlx5_core_srq *mlx5_cmd_get_srq(struct mlx5_ib_dev *dev, u32 srqn); + +int mlx5_init_srq_table(struct mlx5_ib_dev *dev); +void mlx5_cleanup_srq_table(struct mlx5_ib_dev *dev); +#endif /* MLX5_IB_SRQ_H */ diff --git a/drivers/net/ethernet/mellanox/mlx5/core/srq.c b/drivers/infiniband/hw/mlx5/srq_cmd.c index 6a6fc9be01e6..7aaaffbd4afa 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/srq.c +++ b/drivers/infiniband/hw/mlx5/srq_cmd.c @@ -1,67 +1,13 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB /* - * Copyright (c) 2013-2015, Mellanox Technologies. 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 - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. + * Copyright (c) 2013-2018, Mellanox Technologies inc. All rights reserved. */ #include <linux/kernel.h> -#include <linux/module.h> #include <linux/mlx5/driver.h> #include <linux/mlx5/cmd.h> -#include <linux/mlx5/srq.h> -#include <rdma/ib_verbs.h> -#include "mlx5_core.h" -#include <linux/mlx5/transobj.h> - -void mlx5_srq_event(struct mlx5_core_dev *dev, u32 srqn, int event_type) -{ - struct mlx5_srq_table *table = &dev->priv.srq_table; - struct mlx5_core_srq *srq; - - spin_lock(&table->lock); - - srq = radix_tree_lookup(&table->tree, srqn); - if (srq) - atomic_inc(&srq->refcount); - - spin_unlock(&table->lock); - - if (!srq) { - mlx5_core_warn(dev, "Async event for bogus SRQ 0x%08x\n", srqn); - return; - } - - srq->event(srq, event_type); - - if (atomic_dec_and_test(&srq->refcount)) - complete(&srq->free); -} +#include "mlx5_ib.h" +#include "srq.h" static int get_pas_size(struct mlx5_srq_attr *in) { @@ -132,9 +78,9 @@ static void get_srqc(void *srqc, struct mlx5_srq_attr *in) in->db_record = MLX5_GET64(srqc, srqc, dbr_addr); } -struct mlx5_core_srq *mlx5_core_get_srq(struct mlx5_core_dev *dev, u32 srqn) +struct mlx5_core_srq *mlx5_cmd_get_srq(struct mlx5_ib_dev *dev, u32 srqn) { - struct mlx5_srq_table *table = &dev->priv.srq_table; + struct mlx5_srq_table *table = &dev->srq_table; struct mlx5_core_srq *srq; spin_lock(&table->lock); @@ -147,9 +93,8 @@ struct mlx5_core_srq *mlx5_core_get_srq(struct mlx5_core_dev *dev, u32 srqn) return srq; } -EXPORT_SYMBOL(mlx5_core_get_srq); -static int create_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int create_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *in) { u32 create_out[MLX5_ST_SZ_DW(create_srq_out)] = {0}; @@ -176,7 +121,7 @@ static int create_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(create_srq_in, create_in, opcode, MLX5_CMD_OP_CREATE_SRQ); - err = mlx5_cmd_exec(dev, create_in, inlen, create_out, + err = mlx5_cmd_exec(dev->mdev, create_in, inlen, create_out, sizeof(create_out)); kvfree(create_in); if (!err) { @@ -187,8 +132,7 @@ static int create_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, return err; } -static int destroy_srq_cmd(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq) +static int destroy_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { u32 srq_in[MLX5_ST_SZ_DW(destroy_srq_in)] = {0}; u32 srq_out[MLX5_ST_SZ_DW(destroy_srq_out)] = {0}; @@ -198,11 +142,11 @@ static int destroy_srq_cmd(struct mlx5_core_dev *dev, MLX5_SET(destroy_srq_in, srq_in, srqn, srq->srqn); MLX5_SET(destroy_srq_in, srq_in, uid, srq->uid); - return mlx5_cmd_exec(dev, srq_in, sizeof(srq_in), - srq_out, sizeof(srq_out)); + return mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, + sizeof(srq_out)); } -static int arm_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int arm_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm, int is_srq) { u32 srq_in[MLX5_ST_SZ_DW(arm_rq_in)] = {0}; @@ -214,11 +158,11 @@ static int arm_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(arm_rq_in, srq_in, lwm, lwm); MLX5_SET(arm_rq_in, srq_in, uid, srq->uid); - return mlx5_cmd_exec(dev, srq_in, sizeof(srq_in), - srq_out, sizeof(srq_out)); + return mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, + sizeof(srq_out)); } -static int query_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int query_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { u32 srq_in[MLX5_ST_SZ_DW(query_srq_in)] = {0}; @@ -233,8 +177,8 @@ static int query_srq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(query_srq_in, srq_in, opcode, MLX5_CMD_OP_QUERY_SRQ); MLX5_SET(query_srq_in, srq_in, srqn, srq->srqn); - err = mlx5_cmd_exec(dev, srq_in, sizeof(srq_in), - srq_out, MLX5_ST_SZ_BYTES(query_srq_out)); + err = mlx5_cmd_exec(dev->mdev, srq_in, sizeof(srq_in), srq_out, + MLX5_ST_SZ_BYTES(query_srq_out)); if (err) goto out; @@ -247,7 +191,7 @@ out: return err; } -static int create_xrc_srq_cmd(struct mlx5_core_dev *dev, +static int create_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *in) { @@ -277,7 +221,7 @@ static int create_xrc_srq_cmd(struct mlx5_core_dev *dev, MLX5_CMD_OP_CREATE_XRC_SRQ); memset(create_out, 0, sizeof(create_out)); - err = mlx5_cmd_exec(dev, create_in, inlen, create_out, + err = mlx5_cmd_exec(dev->mdev, create_in, inlen, create_out, sizeof(create_out)); if (err) goto out; @@ -289,7 +233,7 @@ out: return err; } -static int destroy_xrc_srq_cmd(struct mlx5_core_dev *dev, +static int destroy_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { u32 xrcsrq_in[MLX5_ST_SZ_DW(destroy_xrc_srq_in)] = {0}; @@ -300,12 +244,12 @@ static int destroy_xrc_srq_cmd(struct mlx5_core_dev *dev, MLX5_SET(destroy_xrc_srq_in, xrcsrq_in, xrc_srqn, srq->srqn); MLX5_SET(destroy_xrc_srq_in, xrcsrq_in, uid, srq->uid); - return mlx5_cmd_exec(dev, xrcsrq_in, sizeof(xrcsrq_in), + return mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), xrcsrq_out, sizeof(xrcsrq_out)); } -static int arm_xrc_srq_cmd(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq, u16 lwm) +static int arm_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + u16 lwm) { u32 xrcsrq_in[MLX5_ST_SZ_DW(arm_xrc_srq_in)] = {0}; u32 xrcsrq_out[MLX5_ST_SZ_DW(arm_xrc_srq_out)] = {0}; @@ -316,11 +260,11 @@ static int arm_xrc_srq_cmd(struct mlx5_core_dev *dev, MLX5_SET(arm_xrc_srq_in, xrcsrq_in, lwm, lwm); MLX5_SET(arm_xrc_srq_in, xrcsrq_in, uid, srq->uid); - return mlx5_cmd_exec(dev, xrcsrq_in, sizeof(xrcsrq_in), + return mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), xrcsrq_out, sizeof(xrcsrq_out)); } -static int query_xrc_srq_cmd(struct mlx5_core_dev *dev, +static int query_xrc_srq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { @@ -338,8 +282,8 @@ static int query_xrc_srq_cmd(struct mlx5_core_dev *dev, MLX5_CMD_OP_QUERY_XRC_SRQ); MLX5_SET(query_xrc_srq_in, xrcsrq_in, xrc_srqn, srq->srqn); - err = mlx5_cmd_exec(dev, xrcsrq_in, sizeof(xrcsrq_in), xrcsrq_out, - MLX5_ST_SZ_BYTES(query_xrc_srq_out)); + err = mlx5_cmd_exec(dev->mdev, xrcsrq_in, sizeof(xrcsrq_in), + xrcsrq_out, MLX5_ST_SZ_BYTES(query_xrc_srq_out)); if (err) goto out; @@ -354,21 +298,27 @@ out: return err; } -static int create_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int create_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *in) { - void *create_in; + void *create_out = NULL; + void *create_in = NULL; void *rmpc; void *wq; int pas_size; + int outlen; int inlen; int err; pas_size = get_pas_size(in); inlen = MLX5_ST_SZ_BYTES(create_rmp_in) + pas_size; + outlen = MLX5_ST_SZ_BYTES(create_rmp_out); create_in = kvzalloc(inlen, GFP_KERNEL); - if (!create_in) - return -ENOMEM; + create_out = kvzalloc(outlen, GFP_KERNEL); + if (!create_in || !create_out) { + err = -ENOMEM; + goto out; + } rmpc = MLX5_ADDR_OF(create_rmp_in, create_in, ctx); wq = MLX5_ADDR_OF(rmpc, rmpc, wq); @@ -378,16 +328,20 @@ static int create_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, set_wq(wq, in); memcpy(MLX5_ADDR_OF(rmpc, rmpc, wq.pas), in->pas, pas_size); - err = mlx5_core_create_rmp(dev, create_in, inlen, &srq->srqn); - if (!err) + MLX5_SET(create_rmp_in, create_in, opcode, MLX5_CMD_OP_CREATE_RMP); + err = mlx5_cmd_exec(dev->mdev, create_in, inlen, create_out, outlen); + if (!err) { + srq->srqn = MLX5_GET(create_rmp_out, create_out, rmpn); srq->uid = in->uid; + } +out: kvfree(create_in); + kvfree(create_out); return err; } -static int destroy_rmp_cmd(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq) +static int destroy_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { u32 in[MLX5_ST_SZ_DW(destroy_rmp_in)] = {}; u32 out[MLX5_ST_SZ_DW(destroy_rmp_out)] = {}; @@ -395,22 +349,30 @@ static int destroy_rmp_cmd(struct mlx5_core_dev *dev, MLX5_SET(destroy_rmp_in, in, opcode, MLX5_CMD_OP_DESTROY_RMP); MLX5_SET(destroy_rmp_in, in, rmpn, srq->srqn); MLX5_SET(destroy_rmp_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); } -static int arm_rmp_cmd(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq, +static int arm_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm) { - void *in; + void *out = NULL; + void *in = NULL; void *rmpc; void *wq; void *bitmask; + int outlen; + int inlen; int err; - in = kvzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in), GFP_KERNEL); - if (!in) - return -ENOMEM; + inlen = MLX5_ST_SZ_BYTES(modify_rmp_in); + outlen = MLX5_ST_SZ_BYTES(modify_rmp_out); + + in = kvzalloc(inlen, GFP_KERNEL); + out = kvzalloc(outlen, GFP_KERNEL); + if (!in || !out) { + err = -ENOMEM; + goto out; + } rmpc = MLX5_ADDR_OF(modify_rmp_in, in, ctx); bitmask = MLX5_ADDR_OF(modify_rmp_in, in, bitmask); @@ -422,25 +384,39 @@ static int arm_rmp_cmd(struct mlx5_core_dev *dev, MLX5_SET(wq, wq, lwm, lwm); MLX5_SET(rmp_bitmask, bitmask, lwm, 1); MLX5_SET(rmpc, rmpc, state, MLX5_RMPC_STATE_RDY); + MLX5_SET(modify_rmp_in, in, opcode, MLX5_CMD_OP_MODIFY_RMP); - err = mlx5_core_modify_rmp(dev, in, MLX5_ST_SZ_BYTES(modify_rmp_in)); + err = mlx5_cmd_exec(dev->mdev, in, inlen, out, outlen); +out: kvfree(in); + kvfree(out); return err; } -static int query_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int query_rmp_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { - u32 *rmp_out; + u32 *rmp_out = NULL; + u32 *rmp_in = NULL; void *rmpc; + int outlen; + int inlen; int err; - rmp_out = kvzalloc(MLX5_ST_SZ_BYTES(query_rmp_out), GFP_KERNEL); - if (!rmp_out) - return -ENOMEM; + outlen = MLX5_ST_SZ_BYTES(query_rmp_out); + inlen = MLX5_ST_SZ_BYTES(query_rmp_in); - err = mlx5_core_query_rmp(dev, srq->srqn, rmp_out); + rmp_out = kvzalloc(outlen, GFP_KERNEL); + rmp_in = kvzalloc(inlen, GFP_KERNEL); + if (!rmp_out || !rmp_in) { + err = -ENOMEM; + goto out; + } + + MLX5_SET(query_rmp_in, rmp_in, opcode, MLX5_CMD_OP_QUERY_RMP); + MLX5_SET(query_rmp_in, rmp_in, rmpn, srq->srqn); + err = mlx5_cmd_exec(dev->mdev, rmp_in, inlen, rmp_out, outlen); if (err) goto out; @@ -451,10 +427,11 @@ static int query_rmp_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, out: kvfree(rmp_out); + kvfree(rmp_in); return err; } -static int create_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int create_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *in) { u32 create_out[MLX5_ST_SZ_DW(create_xrq_out)] = {0}; @@ -489,7 +466,7 @@ static int create_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(xrqc, xrqc, cqn, in->cqn); MLX5_SET(create_xrq_in, create_in, opcode, MLX5_CMD_OP_CREATE_XRQ); MLX5_SET(create_xrq_in, create_in, uid, in->uid); - err = mlx5_cmd_exec(dev, create_in, inlen, create_out, + err = mlx5_cmd_exec(dev->mdev, create_in, inlen, create_out, sizeof(create_out)); kvfree(create_in); if (!err) { @@ -500,7 +477,7 @@ static int create_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, return err; } -static int destroy_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq) +static int destroy_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { u32 in[MLX5_ST_SZ_DW(destroy_xrq_in)] = {0}; u32 out[MLX5_ST_SZ_DW(destroy_xrq_out)] = {0}; @@ -509,10 +486,10 @@ static int destroy_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq) MLX5_SET(destroy_xrq_in, in, xrqn, srq->srqn); MLX5_SET(destroy_xrq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); } -static int arm_xrq_cmd(struct mlx5_core_dev *dev, +static int arm_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, u16 lwm) { @@ -525,10 +502,10 @@ static int arm_xrq_cmd(struct mlx5_core_dev *dev, MLX5_SET(arm_rq_in, in, lwm, lwm); MLX5_SET(arm_rq_in, in, uid, srq->uid); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); + return mlx5_cmd_exec(dev->mdev, in, sizeof(in), out, sizeof(out)); } -static int query_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, +static int query_xrq_cmd(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *out) { u32 in[MLX5_ST_SZ_DW(query_xrq_in)] = {0}; @@ -544,7 +521,7 @@ static int query_xrq_cmd(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, MLX5_SET(query_xrq_in, in, opcode, MLX5_CMD_OP_QUERY_XRQ); MLX5_SET(query_xrq_in, in, xrqn, srq->srqn); - err = mlx5_cmd_exec(dev, in, sizeof(in), xrq_out, outlen); + err = mlx5_cmd_exec(dev->mdev, in, sizeof(in), xrq_out, outlen); if (err) goto out; @@ -567,11 +544,10 @@ out: return err; } -static int create_srq_split(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq, +static int create_srq_split(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, struct mlx5_srq_attr *in) { - if (!dev->issi) + if (!dev->mdev->issi) return create_srq_cmd(dev, srq, in); switch (srq->common.res) { case MLX5_RES_XSRQ: @@ -583,10 +559,9 @@ static int create_srq_split(struct mlx5_core_dev *dev, } } -static int destroy_srq_split(struct mlx5_core_dev *dev, - struct mlx5_core_srq *srq) +static int destroy_srq_split(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - if (!dev->issi) + if (!dev->mdev->issi) return destroy_srq_cmd(dev, srq); switch (srq->common.res) { case MLX5_RES_XSRQ: @@ -598,11 +573,11 @@ static int destroy_srq_split(struct mlx5_core_dev *dev, } } -int mlx5_core_create_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - struct mlx5_srq_attr *in) +int mlx5_cmd_create_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_srq_attr *in) { + struct mlx5_srq_table *table = &dev->srq_table; int err; - struct mlx5_srq_table *table = &dev->priv.srq_table; switch (in->type) { case IB_SRQT_XRC: @@ -625,10 +600,8 @@ int mlx5_core_create_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, spin_lock_irq(&table->lock); err = radix_tree_insert(&table->tree, srq->srqn, srq); spin_unlock_irq(&table->lock); - if (err) { - mlx5_core_warn(dev, "err %d, srqn 0x%x\n", err, srq->srqn); + if (err) goto err_destroy_srq_split; - } return 0; @@ -637,25 +610,18 @@ err_destroy_srq_split: return err; } -EXPORT_SYMBOL(mlx5_core_create_srq); -int mlx5_core_destroy_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq) +int mlx5_cmd_destroy_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq) { - struct mlx5_srq_table *table = &dev->priv.srq_table; + struct mlx5_srq_table *table = &dev->srq_table; struct mlx5_core_srq *tmp; int err; spin_lock_irq(&table->lock); tmp = radix_tree_delete(&table->tree, srq->srqn); spin_unlock_irq(&table->lock); - if (!tmp) { - mlx5_core_warn(dev, "srq 0x%x not found in tree\n", srq->srqn); - return -EINVAL; - } - if (tmp != srq) { - mlx5_core_warn(dev, "corruption on srqn 0x%x\n", srq->srqn); + if (!tmp || tmp != srq) return -EINVAL; - } err = destroy_srq_split(dev, srq); if (err) @@ -667,12 +633,11 @@ int mlx5_core_destroy_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq) return 0; } -EXPORT_SYMBOL(mlx5_core_destroy_srq); -int mlx5_core_query_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - struct mlx5_srq_attr *out) +int mlx5_cmd_query_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + struct mlx5_srq_attr *out) { - if (!dev->issi) + if (!dev->mdev->issi) return query_srq_cmd(dev, srq, out); switch (srq->common.res) { case MLX5_RES_XSRQ: @@ -683,12 +648,11 @@ int mlx5_core_query_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, return query_rmp_cmd(dev, srq, out); } } -EXPORT_SYMBOL(mlx5_core_query_srq); -int mlx5_core_arm_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - u16 lwm, int is_srq) +int mlx5_cmd_arm_srq(struct mlx5_ib_dev *dev, struct mlx5_core_srq *srq, + u16 lwm, int is_srq) { - if (!dev->issi) + if (!dev->mdev->issi) return arm_srq_cmd(dev, srq, lwm, is_srq); switch (srq->common.res) { case MLX5_RES_XSRQ: @@ -699,18 +663,60 @@ int mlx5_core_arm_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, return arm_rmp_cmd(dev, srq, lwm); } } -EXPORT_SYMBOL(mlx5_core_arm_srq); -void mlx5_init_srq_table(struct mlx5_core_dev *dev) +static int srq_event_notifier(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_srq_table *table; + struct mlx5_core_srq *srq; + struct mlx5_eqe *eqe; + u32 srqn; + + if (type != MLX5_EVENT_TYPE_SRQ_CATAS_ERROR && + type != MLX5_EVENT_TYPE_SRQ_RQ_LIMIT) + return NOTIFY_DONE; + + table = container_of(nb, struct mlx5_srq_table, nb); + + eqe = data; + srqn = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; + + spin_lock(&table->lock); + + srq = radix_tree_lookup(&table->tree, srqn); + if (srq) + atomic_inc(&srq->refcount); + + spin_unlock(&table->lock); + + if (!srq) + return NOTIFY_OK; + + srq->event(srq, eqe->type); + + if (atomic_dec_and_test(&srq->refcount)) + complete(&srq->free); + + return NOTIFY_OK; +} + +int mlx5_init_srq_table(struct mlx5_ib_dev *dev) { - struct mlx5_srq_table *table = &dev->priv.srq_table; + struct mlx5_srq_table *table = &dev->srq_table; memset(table, 0, sizeof(*table)); spin_lock_init(&table->lock); INIT_RADIX_TREE(&table->tree, GFP_ATOMIC); + + table->nb.notifier_call = srq_event_notifier; + mlx5_notifier_register(dev->mdev, &table->nb); + + return 0; } -void mlx5_cleanup_srq_table(struct mlx5_core_dev *dev) +void mlx5_cleanup_srq_table(struct mlx5_ib_dev *dev) { - /* nothing */ + struct mlx5_srq_table *table = &dev->srq_table; + + mlx5_notifier_unregister(dev->mdev, &table->nb); } diff --git a/drivers/infiniband/hw/qib/qib_iba6120.c b/drivers/infiniband/hw/qib/qib_iba6120.c index fb1ff59f40bd..cdbf707fa267 100644 --- a/drivers/infiniband/hw/qib/qib_iba6120.c +++ b/drivers/infiniband/hw/qib/qib_iba6120.c @@ -3237,7 +3237,6 @@ static int init_6120_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ ret = ib_mtu_enum_to_int(qib_ibmtu); dd->rcvegrbufsize = ret != -1 ? max(ret, 2048) : QIB_DEFAULT_MTU; - BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_6120_tidtemplate(dd); diff --git a/drivers/infiniband/hw/qib/qib_iba7220.c b/drivers/infiniband/hw/qib/qib_iba7220.c index 163a57a88742..9fde45538f6e 100644 --- a/drivers/infiniband/hw/qib/qib_iba7220.c +++ b/drivers/infiniband/hw/qib/qib_iba7220.c @@ -4043,7 +4043,6 @@ static int qib_init_7220_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ ret = ib_mtu_enum_to_int(qib_ibmtu); dd->rcvegrbufsize = ret != -1 ? max(ret, 2048) : QIB_DEFAULT_MTU; - BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_7220_tidtemplate(dd); @@ -4252,7 +4251,6 @@ static int init_sdma_7220_regs(struct qib_pportdata *ppd) unsigned word = i / 64; unsigned bit = i & 63; - BUG_ON(word >= 3); senddmabufmask[word] |= 1ULL << bit; } qib_write_kreg(dd, kr_senddmabufmask0, senddmabufmask[0]); diff --git a/drivers/infiniband/hw/qib/qib_iba7322.c b/drivers/infiniband/hw/qib/qib_iba7322.c index bf5e222eed8e..17d6b24b3473 100644 --- a/drivers/infiniband/hw/qib/qib_iba7322.c +++ b/drivers/infiniband/hw/qib/qib_iba7322.c @@ -1382,7 +1382,6 @@ static void err_decode(char *msg, size_t len, u64 errs, *msg++ = ','; len--; } - BUG_ON(!msp->sz); /* msp->sz counts the nul */ took = min_t(size_t, msp->sz - (size_t)1, len); memcpy(msg, msp->msg, took); @@ -6599,7 +6598,6 @@ static int qib_init_7322_variables(struct qib_devdata *dd) /* we always allocate at least 2048 bytes for eager buffers */ dd->rcvegrbufsize = max(mtu, 2048); - BUG_ON(!is_power_of_2(dd->rcvegrbufsize)); dd->rcvegrbufsize_shift = ilog2(dd->rcvegrbufsize); qib_7322_tidtemplate(dd); @@ -6904,7 +6902,6 @@ static int init_sdma_7322_regs(struct qib_pportdata *ppd) unsigned word = erstbuf / BITS_PER_LONG; unsigned bit = erstbuf & (BITS_PER_LONG - 1); - BUG_ON(word >= 3); senddmabufmask[word] |= 1ULL << bit; } qib_write_kreg_port(ppd, krp_senddmabufmask0, senddmabufmask[0]); diff --git a/drivers/infiniband/hw/qib/qib_init.c b/drivers/infiniband/hw/qib/qib_init.c index d7cdc77d6306..9fd69903ca57 100644 --- a/drivers/infiniband/hw/qib/qib_init.c +++ b/drivers/infiniband/hw/qib/qib_init.c @@ -209,7 +209,6 @@ struct qib_ctxtdata *qib_create_ctxtdata(struct qib_pportdata *ppd, u32 ctxt, rcd->rcvegrbuf_chunks = (rcd->rcvegrcnt + rcd->rcvegrbufs_perchunk - 1) / rcd->rcvegrbufs_perchunk; - BUG_ON(!is_power_of_2(rcd->rcvegrbufs_perchunk)); rcd->rcvegrbufs_perchunk_shift = ilog2(rcd->rcvegrbufs_perchunk); } diff --git a/drivers/infiniband/hw/qib/qib_pcie.c b/drivers/infiniband/hw/qib/qib_pcie.c index 30595b358d8f..864f2af171f7 100644 --- a/drivers/infiniband/hw/qib/qib_pcie.c +++ b/drivers/infiniband/hw/qib/qib_pcie.c @@ -387,7 +387,7 @@ void qib_pcie_reenable(struct qib_devdata *dd, u16 cmd, u8 iline, u8 cline) static int qib_pcie_coalesce; module_param_named(pcie_coalesce, qib_pcie_coalesce, int, S_IRUGO); -MODULE_PARM_DESC(pcie_coalesce, "tune PCIe colescing on some Intel chipsets"); +MODULE_PARM_DESC(pcie_coalesce, "tune PCIe coalescing on some Intel chipsets"); /* * Enable PCIe completion and data coalescing, on Intel 5x00 and 7300 diff --git a/drivers/infiniband/hw/qib/qib_sdma.c b/drivers/infiniband/hw/qib/qib_sdma.c index 757d4c9d713d..0b6ca424c11d 100644 --- a/drivers/infiniband/hw/qib/qib_sdma.c +++ b/drivers/infiniband/hw/qib/qib_sdma.c @@ -572,7 +572,6 @@ retry: len = sge->length; if (len > sge->sge_length) len = sge->sge_length; - BUG_ON(len == 0); dw = (len + 3) >> 2; addr = dma_map_single(&ppd->dd->pcidev->dev, sge->vaddr, dw << 2, DMA_TO_DEVICE); diff --git a/drivers/infiniband/hw/qib/qib_ud.c b/drivers/infiniband/hw/qib/qib_ud.c index 4d4c31ea4e2d..868da0ece7ba 100644 --- a/drivers/infiniband/hw/qib/qib_ud.c +++ b/drivers/infiniband/hw/qib/qib_ud.c @@ -178,7 +178,6 @@ static void qib_ud_loopback(struct rvt_qp *sqp, struct rvt_swqe *swqe) len = length; if (len > sge->sge_length) len = sge->sge_length; - BUG_ON(len == 0); rvt_copy_sge(qp, &qp->r_sge, sge->vaddr, len, true, false); sge->vaddr += len; sge->length -= len; diff --git a/drivers/infiniband/hw/qib/qib_user_sdma.c b/drivers/infiniband/hw/qib/qib_user_sdma.c index 926f3c8eba69..31c523b2a9f5 100644 --- a/drivers/infiniband/hw/qib/qib_user_sdma.c +++ b/drivers/infiniband/hw/qib/qib_user_sdma.c @@ -237,7 +237,6 @@ qib_user_sdma_queue_create(struct device *dev, int unit, int ctxt, int sctxt) ret = qib_user_sdma_rb_insert(&qib_user_sdma_rb_root, sdma_rb_node); - BUG_ON(ret == 0); } pq->sdma_rb_node = sdma_rb_node; diff --git a/drivers/infiniband/hw/qib/qib_verbs.c b/drivers/infiniband/hw/qib/qib_verbs.c index 4b0f5761a646..8914abdd7584 100644 --- a/drivers/infiniband/hw/qib/qib_verbs.c +++ b/drivers/infiniband/hw/qib/qib_verbs.c @@ -150,7 +150,6 @@ static u32 qib_count_sge(struct rvt_sge_state *ss, u32 length) len = length; if (len > sge.sge_length) len = sge.sge_length; - BUG_ON(len == 0); if (((long) sge.vaddr & (sizeof(u32) - 1)) || (len != length && (len & (sizeof(u32) - 1)))) { ndesc = 0; @@ -193,7 +192,6 @@ static void qib_copy_from_sge(void *data, struct rvt_sge_state *ss, u32 length) len = length; if (len > sge->sge_length) len = sge->sge_length; - BUG_ON(len == 0); memcpy(data, sge->vaddr, len); sge->vaddr += len; sge->length -= len; @@ -449,7 +447,6 @@ static void copy_io(u32 __iomem *piobuf, struct rvt_sge_state *ss, len = length; if (len > ss->sge.sge_length) len = ss->sge.sge_length; - BUG_ON(len == 0); /* If the source address is not aligned, try to align it. */ off = (unsigned long)ss->sge.vaddr & (sizeof(u32) - 1); if (off) { diff --git a/drivers/infiniband/hw/usnic/usnic_ib_main.c b/drivers/infiniband/hw/usnic/usnic_ib_main.c index 73bd00f8d2c8..413fa5732e2b 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_main.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_main.c @@ -649,7 +649,7 @@ static int __init usnic_ib_init(void) err = usnic_uiom_init(DRV_NAME); if (err) { - usnic_err("Unable to initalize umem with err %d\n", err); + usnic_err("Unable to initialize umem with err %d\n", err); return err; } diff --git a/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c b/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c index bf5136533d49..0cdb156e165e 100644 --- a/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c +++ b/drivers/infiniband/hw/usnic/usnic_ib_qp_grp.c @@ -681,7 +681,7 @@ usnic_ib_qp_grp_create(struct usnic_fwd_dev *ufdev, struct usnic_ib_vf *vf, err = usnic_vnic_res_spec_satisfied(&min_transport_spec[transport], res_spec); if (err) { - usnic_err("Spec does not meet miniumum req for transport %d\n", + usnic_err("Spec does not meet minimum req for transport %d\n", transport); log_spec(res_spec); return ERR_PTR(err); diff --git a/drivers/infiniband/sw/rdmavt/qp.c b/drivers/infiniband/sw/rdmavt/qp.c index 1735deb1a9d4..a1bd8cfc2c25 100644 --- a/drivers/infiniband/sw/rdmavt/qp.c +++ b/drivers/infiniband/sw/rdmavt/qp.c @@ -1,5 +1,5 @@ /* - * Copyright(c) 2016, 2017 Intel Corporation. + * Copyright(c) 2016 - 2018 Intel Corporation. * * This file is provided under a dual BSD/GPLv2 license. When using or * redistributing this file, you may do so under either license. @@ -1094,6 +1094,13 @@ struct ib_qp *rvt_create_qp(struct ib_pd *ibpd, qp->ibqp.qp_num = err; qp->port_num = init_attr->port_num; rvt_init_qp(rdi, qp, init_attr->qp_type); + if (rdi->driver_f.qp_priv_init) { + err = rdi->driver_f.qp_priv_init(rdi, qp, init_attr); + if (err) { + ret = ERR_PTR(err); + goto bail_rq_wq; + } + } break; default: diff --git a/drivers/infiniband/sw/rxe/rxe.h b/drivers/infiniband/sw/rxe/rxe.h index d9ec2de68738..8f79bd86d033 100644 --- a/drivers/infiniband/sw/rxe/rxe.h +++ b/drivers/infiniband/sw/rxe/rxe.h @@ -65,8 +65,9 @@ */ #define RXE_UVERBS_ABI_VERSION 2 -#define IB_PHYS_STATE_LINK_UP (5) -#define IB_PHYS_STATE_LINK_DOWN (3) +#define RDMA_LINK_PHYS_STATE_LINK_UP (5) +#define RDMA_LINK_PHYS_STATE_DISABLED (3) +#define RDMA_LINK_PHYS_STATE_POLLING (2) #define RXE_ROCE_V2_SPORT (0xc000) diff --git a/drivers/infiniband/sw/rxe/rxe_comp.c b/drivers/infiniband/sw/rxe/rxe_comp.c index ea089cb091ad..e996da67a851 100644 --- a/drivers/infiniband/sw/rxe/rxe_comp.c +++ b/drivers/infiniband/sw/rxe/rxe_comp.c @@ -439,6 +439,7 @@ static void make_send_cqe(struct rxe_qp *qp, struct rxe_send_wqe *wqe, */ static void do_complete(struct rxe_qp *qp, struct rxe_send_wqe *wqe) { + struct rxe_dev *rxe = to_rdev(qp->ibqp.device); struct rxe_cqe cqe; if ((qp->sq_sig_type == IB_SIGNAL_ALL_WR) || @@ -451,6 +452,11 @@ static void do_complete(struct rxe_qp *qp, struct rxe_send_wqe *wqe) advance_consumer(qp->sq.queue); } + if (wqe->wr.opcode == IB_WR_SEND || + wqe->wr.opcode == IB_WR_SEND_WITH_IMM || + wqe->wr.opcode == IB_WR_SEND_WITH_INV) + rxe_counter_inc(rxe, RXE_CNT_RDMA_SEND); + /* * we completed something so let req run again * if it is trying to fence diff --git a/drivers/infiniband/sw/rxe/rxe_hw_counters.c b/drivers/infiniband/sw/rxe/rxe_hw_counters.c index 6aeb7a165e46..4a24895846d3 100644 --- a/drivers/infiniband/sw/rxe/rxe_hw_counters.c +++ b/drivers/infiniband/sw/rxe/rxe_hw_counters.c @@ -37,15 +37,18 @@ static const char * const rxe_counter_name[] = { [RXE_CNT_SENT_PKTS] = "sent_pkts", [RXE_CNT_RCVD_PKTS] = "rcvd_pkts", [RXE_CNT_DUP_REQ] = "duplicate_request", - [RXE_CNT_OUT_OF_SEQ_REQ] = "out_of_sequence", + [RXE_CNT_OUT_OF_SEQ_REQ] = "out_of_seq_request", [RXE_CNT_RCV_RNR] = "rcvd_rnr_err", [RXE_CNT_SND_RNR] = "send_rnr_err", [RXE_CNT_RCV_SEQ_ERR] = "rcvd_seq_err", - [RXE_CNT_COMPLETER_SCHED] = "ack_deffered", + [RXE_CNT_COMPLETER_SCHED] = "ack_deferred", [RXE_CNT_RETRY_EXCEEDED] = "retry_exceeded_err", [RXE_CNT_RNR_RETRY_EXCEEDED] = "retry_rnr_exceeded_err", [RXE_CNT_COMP_RETRY] = "completer_retry_err", [RXE_CNT_SEND_ERR] = "send_err", + [RXE_CNT_LINK_DOWNED] = "link_downed", + [RXE_CNT_RDMA_SEND] = "rdma_sends", + [RXE_CNT_RDMA_RECV] = "rdma_recvs", }; int rxe_ib_get_hw_stats(struct ib_device *ibdev, diff --git a/drivers/infiniband/sw/rxe/rxe_hw_counters.h b/drivers/infiniband/sw/rxe/rxe_hw_counters.h index f44df1b76742..72c0d63c79e0 100644 --- a/drivers/infiniband/sw/rxe/rxe_hw_counters.h +++ b/drivers/infiniband/sw/rxe/rxe_hw_counters.h @@ -50,6 +50,9 @@ enum rxe_counters { RXE_CNT_RNR_RETRY_EXCEEDED, RXE_CNT_COMP_RETRY, RXE_CNT_SEND_ERR, + RXE_CNT_LINK_DOWNED, + RXE_CNT_RDMA_SEND, + RXE_CNT_RDMA_RECV, RXE_NUM_OF_COUNTERS }; diff --git a/drivers/infiniband/sw/rxe/rxe_loc.h b/drivers/infiniband/sw/rxe/rxe_loc.h index afd53f57a62b..a675c9f2b427 100644 --- a/drivers/infiniband/sw/rxe/rxe_loc.h +++ b/drivers/infiniband/sw/rxe/rxe_loc.h @@ -250,11 +250,12 @@ static inline unsigned int wr_opcode_mask(int opcode, struct rxe_qp *qp) return rxe_wr_opcode_info[opcode].mask[qp->ibqp.qp_type]; } -static inline int rxe_xmit_packet(struct rxe_dev *rxe, struct rxe_qp *qp, - struct rxe_pkt_info *pkt, struct sk_buff *skb) +static inline int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt, + struct sk_buff *skb) { int err; int is_request = pkt->mask & RXE_REQ_MASK; + struct rxe_dev *rxe = to_rdev(qp->ibqp.device); if ((is_request && (qp->req.state != QP_STATE_READY)) || (!is_request && (qp->resp.state != QP_STATE_READY))) { diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c index 40e82e0f6c2d..b26a8141f3ed 100644 --- a/drivers/infiniband/sw/rxe/rxe_net.c +++ b/drivers/infiniband/sw/rxe/rxe_net.c @@ -607,7 +607,6 @@ void rxe_port_up(struct rxe_dev *rxe) port = &rxe->port; port->attr.state = IB_PORT_ACTIVE; - port->attr.phys_state = IB_PHYS_STATE_LINK_UP; rxe_port_event(rxe, IB_EVENT_PORT_ACTIVE); dev_info(&rxe->ib_dev.dev, "set active\n"); @@ -620,9 +619,9 @@ void rxe_port_down(struct rxe_dev *rxe) port = &rxe->port; port->attr.state = IB_PORT_DOWN; - port->attr.phys_state = IB_PHYS_STATE_LINK_DOWN; rxe_port_event(rxe, IB_EVENT_PORT_ERR); + rxe_counter_inc(rxe, RXE_CNT_LINK_DOWNED); dev_info(&rxe->ib_dev.dev, "set down\n"); } diff --git a/drivers/infiniband/sw/rxe/rxe_pool.c b/drivers/infiniband/sw/rxe/rxe_pool.c index 36b53fb94a49..b5c91df22047 100644 --- a/drivers/infiniband/sw/rxe/rxe_pool.c +++ b/drivers/infiniband/sw/rxe/rxe_pool.c @@ -112,6 +112,18 @@ static inline struct kmem_cache *pool_cache(struct rxe_pool *pool) return rxe_type_info[pool->type].cache; } +static void rxe_cache_clean(size_t cnt) +{ + int i; + struct rxe_type_info *type; + + for (i = 0; i < cnt; i++) { + type = &rxe_type_info[i]; + kmem_cache_destroy(type->cache); + type->cache = NULL; + } +} + int rxe_cache_init(void) { int err; @@ -136,24 +148,14 @@ int rxe_cache_init(void) return 0; err1: - while (--i >= 0) { - kmem_cache_destroy(type->cache); - type->cache = NULL; - } + rxe_cache_clean(i); return err; } void rxe_cache_exit(void) { - int i; - struct rxe_type_info *type; - - for (i = 0; i < RXE_NUM_TYPES; i++) { - type = &rxe_type_info[i]; - kmem_cache_destroy(type->cache); - type->cache = NULL; - } + rxe_cache_clean(RXE_NUM_TYPES); } static int rxe_pool_init_index(struct rxe_pool *pool, u32 max, u32 min) @@ -241,7 +243,7 @@ static void rxe_pool_put(struct rxe_pool *pool) kref_put(&pool->ref_cnt, rxe_pool_release); } -int rxe_pool_cleanup(struct rxe_pool *pool) +void rxe_pool_cleanup(struct rxe_pool *pool) { unsigned long flags; @@ -253,8 +255,6 @@ int rxe_pool_cleanup(struct rxe_pool *pool) write_unlock_irqrestore(&pool->pool_lock, flags); rxe_pool_put(pool); - - return 0; } static u32 alloc_index(struct rxe_pool *pool) diff --git a/drivers/infiniband/sw/rxe/rxe_pool.h b/drivers/infiniband/sw/rxe/rxe_pool.h index aa4ba307097b..72968c29e01f 100644 --- a/drivers/infiniband/sw/rxe/rxe_pool.h +++ b/drivers/infiniband/sw/rxe/rxe_pool.h @@ -126,7 +126,7 @@ int rxe_pool_init(struct rxe_dev *rxe, struct rxe_pool *pool, enum rxe_elem_type type, u32 max_elem); /* free resources from object pool */ -int rxe_pool_cleanup(struct rxe_pool *pool); +void rxe_pool_cleanup(struct rxe_pool *pool); /* allocate an object from pool */ void *rxe_alloc(struct rxe_pool *pool); diff --git a/drivers/infiniband/sw/rxe/rxe_qp.c b/drivers/infiniband/sw/rxe/rxe_qp.c index b9710907dac2..2ca4ffe5015f 100644 --- a/drivers/infiniband/sw/rxe/rxe_qp.c +++ b/drivers/infiniband/sw/rxe/rxe_qp.c @@ -97,7 +97,7 @@ int rxe_qp_chk_init(struct rxe_dev *rxe, struct ib_qp_init_attr *init) goto err1; if (init->qp_type == IB_QPT_SMI || init->qp_type == IB_QPT_GSI) { - if (port_num != 1) { + if (!rdma_is_port_valid(&rxe->ib_dev, port_num)) { pr_warn("invalid port = %d\n", port_num); goto err1; } @@ -433,7 +433,7 @@ int rxe_qp_chk_attr(struct rxe_dev *rxe, struct rxe_qp *qp, } if (mask & IB_QP_PORT) { - if (attr->port_num != 1) { + if (!rdma_is_port_valid(&rxe->ib_dev, attr->port_num)) { pr_warn("invalid port %d\n", attr->port_num); goto err1; } @@ -448,7 +448,7 @@ int rxe_qp_chk_attr(struct rxe_dev *rxe, struct rxe_qp *qp, if (mask & IB_QP_ALT_PATH) { if (rxe_av_chk_attr(rxe, &attr->alt_ah_attr)) goto err1; - if (attr->alt_port_num != 1) { + if (!rdma_is_port_valid(&rxe->ib_dev, attr->alt_port_num)) { pr_warn("invalid alt port %d\n", attr->alt_port_num); goto err1; } diff --git a/drivers/infiniband/sw/rxe/rxe_req.c b/drivers/infiniband/sw/rxe/rxe_req.c index 6c361d70d7cd..c5d9b558fa90 100644 --- a/drivers/infiniband/sw/rxe/rxe_req.c +++ b/drivers/infiniband/sw/rxe/rxe_req.c @@ -643,6 +643,7 @@ next_wqe: rmr->access = wqe->wr.wr.reg.access; rmr->lkey = wqe->wr.wr.reg.key; rmr->rkey = wqe->wr.wr.reg.key; + rmr->iova = wqe->wr.wr.reg.mr->iova; wqe->state = wqe_state_done; wqe->status = IB_WC_SUCCESS; } else { @@ -728,7 +729,7 @@ next_wqe: save_state(wqe, qp, &rollback_wqe, &rollback_psn); update_wqe_state(qp, wqe, &pkt); update_wqe_psn(qp, wqe, &pkt, payload); - ret = rxe_xmit_packet(to_rdev(qp->ibqp.device), qp, &pkt, skb); + ret = rxe_xmit_packet(qp, &pkt, skb); if (ret) { qp->need_req_skb = 1; diff --git a/drivers/infiniband/sw/rxe/rxe_resp.c b/drivers/infiniband/sw/rxe/rxe_resp.c index c962160292f4..231528188250 100644 --- a/drivers/infiniband/sw/rxe/rxe_resp.c +++ b/drivers/infiniband/sw/rxe/rxe_resp.c @@ -124,12 +124,9 @@ static inline enum resp_states get_req(struct rxe_qp *qp, struct sk_buff *skb; if (qp->resp.state == QP_STATE_ERROR) { - skb = skb_dequeue(&qp->req_pkts); - if (skb) { - /* drain request packet queue */ + while ((skb = skb_dequeue(&qp->req_pkts))) { rxe_drop_ref(qp); kfree_skb(skb); - return RESPST_GET_REQ; } /* go drain recv wr queue */ @@ -660,7 +657,6 @@ static struct sk_buff *prepare_ack_packet(struct rxe_qp *qp, static enum resp_states read_reply(struct rxe_qp *qp, struct rxe_pkt_info *req_pkt) { - struct rxe_dev *rxe = to_rdev(qp->ibqp.device); struct rxe_pkt_info ack_pkt; struct sk_buff *skb; int mtu = qp->mtu; @@ -739,7 +735,7 @@ static enum resp_states read_reply(struct rxe_qp *qp, p = payload_addr(&ack_pkt) + payload + bth_pad(&ack_pkt); *p = ~icrc; - err = rxe_xmit_packet(rxe, qp, &ack_pkt, skb); + err = rxe_xmit_packet(qp, &ack_pkt, skb); if (err) { pr_err("Failed sending RDMA reply.\n"); return RESPST_ERR_RNR; @@ -838,18 +834,25 @@ static enum resp_states do_complete(struct rxe_qp *qp, struct ib_wc *wc = &cqe.ibwc; struct ib_uverbs_wc *uwc = &cqe.uibwc; struct rxe_recv_wqe *wqe = qp->resp.wqe; + struct rxe_dev *rxe = to_rdev(qp->ibqp.device); if (unlikely(!wqe)) return RESPST_CLEANUP; memset(&cqe, 0, sizeof(cqe)); - wc->wr_id = wqe->wr_id; - wc->status = qp->resp.status; - wc->qp = &qp->ibqp; + if (qp->rcq->is_user) { + uwc->status = qp->resp.status; + uwc->qp_num = qp->ibqp.qp_num; + uwc->wr_id = wqe->wr_id; + } else { + wc->status = qp->resp.status; + wc->qp = &qp->ibqp; + wc->wr_id = wqe->wr_id; + } - /* fields after status are not required for errors */ if (wc->status == IB_WC_SUCCESS) { + rxe_counter_inc(rxe, RXE_CNT_RDMA_RECV); wc->opcode = (pkt->mask & RXE_IMMDT_MASK && pkt->mask & RXE_WRITE_MASK) ? IB_WC_RECV_RDMA_WITH_IMM : IB_WC_RECV; @@ -898,7 +901,6 @@ static enum resp_states do_complete(struct rxe_qp *qp, } if (pkt->mask & RXE_IETH_MASK) { - struct rxe_dev *rxe = to_rdev(qp->ibqp.device); struct rxe_mem *rmr; wc->wc_flags |= IB_WC_WITH_INVALIDATE; @@ -950,7 +952,6 @@ static int send_ack(struct rxe_qp *qp, struct rxe_pkt_info *pkt, int err = 0; struct rxe_pkt_info ack_pkt; struct sk_buff *skb; - struct rxe_dev *rxe = to_rdev(qp->ibqp.device); skb = prepare_ack_packet(qp, pkt, &ack_pkt, IB_OPCODE_RC_ACKNOWLEDGE, 0, psn, syndrome, NULL); @@ -959,7 +960,7 @@ static int send_ack(struct rxe_qp *qp, struct rxe_pkt_info *pkt, goto err1; } - err = rxe_xmit_packet(rxe, qp, &ack_pkt, skb); + err = rxe_xmit_packet(qp, &ack_pkt, skb); if (err) pr_err_ratelimited("Failed sending ack\n"); @@ -973,7 +974,6 @@ static int send_atomic_ack(struct rxe_qp *qp, struct rxe_pkt_info *pkt, int rc = 0; struct rxe_pkt_info ack_pkt; struct sk_buff *skb; - struct rxe_dev *rxe = to_rdev(qp->ibqp.device); struct resp_res *res; skb = prepare_ack_packet(qp, pkt, &ack_pkt, @@ -1001,7 +1001,7 @@ static int send_atomic_ack(struct rxe_qp *qp, struct rxe_pkt_info *pkt, res->last_psn = ack_pkt.psn; res->cur_psn = ack_pkt.psn; - rc = rxe_xmit_packet(rxe, qp, &ack_pkt, skb); + rc = rxe_xmit_packet(qp, &ack_pkt, skb); if (rc) { pr_err_ratelimited("Failed sending ack\n"); rxe_drop_ref(qp); @@ -1131,8 +1131,7 @@ static enum resp_states duplicate_request(struct rxe_qp *qp, if (res) { skb_get(res->atomic.skb); /* Resend the result. */ - rc = rxe_xmit_packet(to_rdev(qp->ibqp.device), qp, - pkt, res->atomic.skb); + rc = rxe_xmit_packet(qp, pkt, res->atomic.skb); if (rc) { pr_err("Failed resending result. This flow is not handled - skb ignored\n"); rc = RESPST_CLEANUP; diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.c b/drivers/infiniband/sw/rxe/rxe_verbs.c index 9c19f2027511..30817c79ba96 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.c +++ b/drivers/infiniband/sw/rxe/rxe_verbs.c @@ -71,6 +71,14 @@ static int rxe_query_port(struct ib_device *dev, mutex_lock(&rxe->usdev_lock); rc = ib_get_eth_speed(dev, port_num, &attr->active_speed, &attr->active_width); + + if (attr->state == IB_PORT_ACTIVE) + attr->phys_state = RDMA_LINK_PHYS_STATE_LINK_UP; + else if (dev_get_flags(rxe->ndev) & IFF_UP) + attr->phys_state = RDMA_LINK_PHYS_STATE_POLLING; + else + attr->phys_state = RDMA_LINK_PHYS_STATE_DISABLED; + mutex_unlock(&rxe->usdev_lock); out: @@ -1279,11 +1287,9 @@ err1: return err; } -int rxe_unregister_device(struct rxe_dev *rxe) +void rxe_unregister_device(struct rxe_dev *rxe) { struct ib_device *dev = &rxe->ib_dev; ib_unregister_device(dev); - - return 0; } diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h index 82e670d6eeea..831381b7788d 100644 --- a/drivers/infiniband/sw/rxe/rxe_verbs.h +++ b/drivers/infiniband/sw/rxe/rxe_verbs.h @@ -467,7 +467,7 @@ static inline struct rxe_mem *to_rmw(struct ib_mw *mw) } int rxe_register_device(struct rxe_dev *rxe); -int rxe_unregister_device(struct rxe_dev *rxe); +void rxe_unregister_device(struct rxe_dev *rxe); void rxe_mc_cleanup(struct rxe_pool_entry *arg); diff --git a/drivers/infiniband/ulp/iser/iser_memory.c b/drivers/infiniband/ulp/iser/iser_memory.c index 009be8889d71..dbe97c02848c 100644 --- a/drivers/infiniband/ulp/iser/iser_memory.c +++ b/drivers/infiniband/ulp/iser/iser_memory.c @@ -277,16 +277,13 @@ void iser_unreg_mem_fmr(struct iscsi_iser_task *iser_task, enum iser_data_dir cmd_dir) { struct iser_mem_reg *reg = &iser_task->rdma_reg[cmd_dir]; - int ret; if (!reg->mem_h) return; iser_dbg("PHYSICAL Mem.Unregister mem_h %p\n", reg->mem_h); - ret = ib_fmr_pool_unmap((struct ib_pool_fmr *)reg->mem_h); - if (ret) - iser_err("ib_fmr_pool_unmap failed %d\n", ret); + ib_fmr_pool_unmap((struct ib_pool_fmr *)reg->mem_h); reg->mem_h = NULL; } diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index 2357aa727dcf..adc0e91d2bb5 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -3617,7 +3617,7 @@ static struct se_portal_group *srpt_make_tpg(struct se_wwn *wwn, const char *name) { struct srpt_port *sport = wwn->priv; - static struct se_portal_group *tpg; + struct se_portal_group *tpg; int res; WARN_ON_ONCE(wwn != &sport->port_guid_wwn && diff --git a/drivers/net/ethernet/mellanox/mlx5/core/Makefile b/drivers/net/ethernet/mellanox/mlx5/core/Makefile index d324a3884462..d499b3d00348 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/Makefile +++ b/drivers/net/ethernet/mellanox/mlx5/core/Makefile @@ -12,9 +12,9 @@ obj-$(CONFIG_MLX5_CORE) += mlx5_core.o # mlx5 core basic # mlx5_core-y := main.o cmd.o debugfs.o fw.o eq.o uar.o pagealloc.o \ - health.o mcg.o cq.o srq.o alloc.o qp.o port.o mr.o pd.o \ + health.o mcg.o cq.o alloc.o qp.o port.o mr.o pd.o \ mad.o transobj.o vport.o sriov.o fs_cmd.o fs_core.o \ - fs_counters.o rl.o lag.o dev.o wq.o lib/gid.o \ + fs_counters.o rl.o lag.o dev.o events.o wq.o lib/gid.o \ diag/fs_tracepoint.o diag/fw_tracer.o # diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c index a5a0823e5ada..8ab636d59edb 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c @@ -40,9 +40,11 @@ #include <linux/random.h> #include <linux/io-mapping.h> #include <linux/mlx5/driver.h> +#include <linux/mlx5/eq.h> #include <linux/debugfs.h> #include "mlx5_core.h" +#include "lib/eq.h" enum { CMD_IF_REV = 5, @@ -313,6 +315,7 @@ static int mlx5_internal_err_ret_value(struct mlx5_core_dev *dev, u16 op, case MLX5_CMD_OP_FPGA_DESTROY_QP: case MLX5_CMD_OP_DESTROY_GENERAL_OBJECT: case MLX5_CMD_OP_DEALLOC_MEMIC: + case MLX5_CMD_OP_PAGE_FAULT_RESUME: return MLX5_CMD_STAT_OK; case MLX5_CMD_OP_QUERY_HCA_CAP: @@ -326,7 +329,6 @@ static int mlx5_internal_err_ret_value(struct mlx5_core_dev *dev, u16 op, case MLX5_CMD_OP_CREATE_MKEY: case MLX5_CMD_OP_QUERY_MKEY: case MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS: - case MLX5_CMD_OP_PAGE_FAULT_RESUME: case MLX5_CMD_OP_CREATE_EQ: case MLX5_CMD_OP_QUERY_EQ: case MLX5_CMD_OP_GEN_EQE: @@ -805,6 +807,8 @@ static u16 msg_to_opcode(struct mlx5_cmd_msg *in) return MLX5_GET(mbox_in, in->first.data, opcode); } +static void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced); + static void cb_timeout_handler(struct work_struct *work) { struct delayed_work *dwork = container_of(work, struct delayed_work, @@ -1412,14 +1416,32 @@ static void mlx5_cmd_change_mod(struct mlx5_core_dev *dev, int mode) up(&cmd->sem); } +static int cmd_comp_notifier(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_core_dev *dev; + struct mlx5_cmd *cmd; + struct mlx5_eqe *eqe; + + cmd = mlx5_nb_cof(nb, struct mlx5_cmd, nb); + dev = container_of(cmd, struct mlx5_core_dev, cmd); + eqe = data; + + mlx5_cmd_comp_handler(dev, be32_to_cpu(eqe->data.cmd.vector), false); + + return NOTIFY_OK; +} void mlx5_cmd_use_events(struct mlx5_core_dev *dev) { + MLX5_NB_INIT(&dev->cmd.nb, cmd_comp_notifier, CMD); + mlx5_eq_notifier_register(dev, &dev->cmd.nb); mlx5_cmd_change_mod(dev, CMD_MODE_EVENTS); } void mlx5_cmd_use_polling(struct mlx5_core_dev *dev) { mlx5_cmd_change_mod(dev, CMD_MODE_POLLING); + mlx5_eq_notifier_unregister(dev, &dev->cmd.nb); } static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg) @@ -1435,7 +1457,7 @@ static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg) } } -void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced) +static void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced) { struct mlx5_cmd *cmd = &dev->cmd; struct mlx5_cmd_work_ent *ent; @@ -1533,7 +1555,29 @@ void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced) } } } -EXPORT_SYMBOL(mlx5_cmd_comp_handler); + +void mlx5_cmd_trigger_completions(struct mlx5_core_dev *dev) +{ + unsigned long flags; + u64 vector; + + /* wait for pending handlers to complete */ + mlx5_eq_synchronize_cmd_irq(dev); + spin_lock_irqsave(&dev->cmd.alloc_lock, flags); + vector = ~dev->cmd.bitmask & ((1ul << (1 << dev->cmd.log_sz)) - 1); + if (!vector) + goto no_trig; + + vector |= MLX5_TRIGGERED_CMD_COMP; + spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); + + mlx5_core_dbg(dev, "vector 0x%llx\n", vector); + mlx5_cmd_comp_handler(dev, vector, true); + return; + +no_trig: + spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); +} static int status_to_err(u8 status) { diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cq.c b/drivers/net/ethernet/mellanox/mlx5/core/cq.c index 4b85abb5c9f7..713a17ee3751 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cq.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cq.c @@ -38,6 +38,7 @@ #include <rdma/ib_verbs.h> #include <linux/mlx5/cq.h> #include "mlx5_core.h" +#include "lib/eq.h" #define TASKLET_MAX_TIME 2 #define TASKLET_MAX_TIME_JIFFIES msecs_to_jiffies(TASKLET_MAX_TIME) @@ -92,10 +93,10 @@ int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, u32 dout[MLX5_ST_SZ_DW(destroy_cq_out)]; u32 out[MLX5_ST_SZ_DW(create_cq_out)]; u32 din[MLX5_ST_SZ_DW(destroy_cq_in)]; - struct mlx5_eq *eq; + struct mlx5_eq_comp *eq; int err; - eq = mlx5_eqn2eq(dev, eqn); + eq = mlx5_eqn2comp_eq(dev, eqn); if (IS_ERR(eq)) return PTR_ERR(eq); @@ -119,12 +120,12 @@ int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, INIT_LIST_HEAD(&cq->tasklet_ctx.list); /* Add to comp EQ CQ tree to recv comp events */ - err = mlx5_eq_add_cq(eq, cq); + err = mlx5_eq_add_cq(&eq->core, cq); if (err) goto err_cmd; /* Add to async EQ CQ tree to recv async events */ - err = mlx5_eq_add_cq(&dev->priv.eq_table.async_eq, cq); + err = mlx5_eq_add_cq(mlx5_get_async_eq(dev), cq); if (err) goto err_cq_add; @@ -139,7 +140,7 @@ int mlx5_core_create_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq, return 0; err_cq_add: - mlx5_eq_del_cq(eq, cq); + mlx5_eq_del_cq(&eq->core, cq); err_cmd: memset(din, 0, sizeof(din)); memset(dout, 0, sizeof(dout)); @@ -157,11 +158,11 @@ int mlx5_core_destroy_cq(struct mlx5_core_dev *dev, struct mlx5_core_cq *cq) u32 in[MLX5_ST_SZ_DW(destroy_cq_in)] = {0}; int err; - err = mlx5_eq_del_cq(&dev->priv.eq_table.async_eq, cq); + err = mlx5_eq_del_cq(mlx5_get_async_eq(dev), cq); if (err) return err; - err = mlx5_eq_del_cq(cq->eq, cq); + err = mlx5_eq_del_cq(&cq->eq->core, cq); if (err) return err; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c index 90fabd612b6c..a11e22d0b0cc 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/debugfs.c @@ -36,6 +36,7 @@ #include <linux/mlx5/cq.h> #include <linux/mlx5/driver.h> #include "mlx5_core.h" +#include "lib/eq.h" enum { QP_PID, @@ -349,6 +350,16 @@ out: return param; } +static int mlx5_core_eq_query(struct mlx5_core_dev *dev, struct mlx5_eq *eq, + u32 *out, int outlen) +{ + u32 in[MLX5_ST_SZ_DW(query_eq_in)] = {}; + + MLX5_SET(query_eq_in, in, opcode, MLX5_CMD_OP_QUERY_EQ); + MLX5_SET(query_eq_in, in, eq_number, eq->eqn); + return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen); +} + static u64 eq_read_field(struct mlx5_core_dev *dev, struct mlx5_eq *eq, int index) { diff --git a/drivers/net/ethernet/mellanox/mlx5/core/dev.c b/drivers/net/ethernet/mellanox/mlx5/core/dev.c index 37ba7c78859d..d2ed14bc37c3 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/dev.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/dev.c @@ -45,75 +45,11 @@ struct mlx5_device_context { unsigned long state; }; -struct mlx5_delayed_event { - struct list_head list; - struct mlx5_core_dev *dev; - enum mlx5_dev_event event; - unsigned long param; -}; - enum { MLX5_INTERFACE_ADDED, MLX5_INTERFACE_ATTACHED, }; -static void add_delayed_event(struct mlx5_priv *priv, - struct mlx5_core_dev *dev, - enum mlx5_dev_event event, - unsigned long param) -{ - struct mlx5_delayed_event *delayed_event; - - delayed_event = kzalloc(sizeof(*delayed_event), GFP_ATOMIC); - if (!delayed_event) { - mlx5_core_err(dev, "event %d is missed\n", event); - return; - } - - mlx5_core_dbg(dev, "Accumulating event %d\n", event); - delayed_event->dev = dev; - delayed_event->event = event; - delayed_event->param = param; - list_add_tail(&delayed_event->list, &priv->waiting_events_list); -} - -static void delayed_event_release(struct mlx5_device_context *dev_ctx, - struct mlx5_priv *priv) -{ - struct mlx5_core_dev *dev = container_of(priv, struct mlx5_core_dev, priv); - struct mlx5_delayed_event *de; - struct mlx5_delayed_event *n; - struct list_head temp; - - INIT_LIST_HEAD(&temp); - - spin_lock_irq(&priv->ctx_lock); - - priv->is_accum_events = false; - list_splice_init(&priv->waiting_events_list, &temp); - if (!dev_ctx->context) - goto out; - list_for_each_entry_safe(de, n, &temp, list) - dev_ctx->intf->event(dev, dev_ctx->context, de->event, de->param); - -out: - spin_unlock_irq(&priv->ctx_lock); - - list_for_each_entry_safe(de, n, &temp, list) { - list_del(&de->list); - kfree(de); - } -} - -/* accumulating events that can come after mlx5_ib calls to - * ib_register_device, till adding that interface to the events list. - */ -static void delayed_event_start(struct mlx5_priv *priv) -{ - spin_lock_irq(&priv->ctx_lock); - priv->is_accum_events = true; - spin_unlock_irq(&priv->ctx_lock); -} void mlx5_add_device(struct mlx5_interface *intf, struct mlx5_priv *priv) { @@ -129,8 +65,6 @@ void mlx5_add_device(struct mlx5_interface *intf, struct mlx5_priv *priv) dev_ctx->intf = intf; - delayed_event_start(priv); - dev_ctx->context = intf->add(dev); if (dev_ctx->context) { set_bit(MLX5_INTERFACE_ADDED, &dev_ctx->state); @@ -139,22 +73,9 @@ void mlx5_add_device(struct mlx5_interface *intf, struct mlx5_priv *priv) spin_lock_irq(&priv->ctx_lock); list_add_tail(&dev_ctx->list, &priv->ctx_list); - -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (dev_ctx->intf->pfault) { - if (priv->pfault) { - mlx5_core_err(dev, "multiple page fault handlers not supported"); - } else { - priv->pfault_ctx = dev_ctx->context; - priv->pfault = dev_ctx->intf->pfault; - } - } -#endif spin_unlock_irq(&priv->ctx_lock); } - delayed_event_release(dev_ctx, priv); - if (!dev_ctx->context) kfree(dev_ctx); } @@ -179,15 +100,6 @@ void mlx5_remove_device(struct mlx5_interface *intf, struct mlx5_priv *priv) if (!dev_ctx) return; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - spin_lock_irq(&priv->ctx_lock); - if (priv->pfault == dev_ctx->intf->pfault) - priv->pfault = NULL; - spin_unlock_irq(&priv->ctx_lock); - - synchronize_srcu(&priv->pfault_srcu); -#endif - spin_lock_irq(&priv->ctx_lock); list_del(&dev_ctx->list); spin_unlock_irq(&priv->ctx_lock); @@ -207,26 +119,20 @@ static void mlx5_attach_interface(struct mlx5_interface *intf, struct mlx5_priv if (!dev_ctx) return; - delayed_event_start(priv); if (intf->attach) { if (test_bit(MLX5_INTERFACE_ATTACHED, &dev_ctx->state)) - goto out; + return; if (intf->attach(dev, dev_ctx->context)) - goto out; - + return; set_bit(MLX5_INTERFACE_ATTACHED, &dev_ctx->state); } else { if (test_bit(MLX5_INTERFACE_ADDED, &dev_ctx->state)) - goto out; + return; dev_ctx->context = intf->add(dev); if (!dev_ctx->context) - goto out; - + return; set_bit(MLX5_INTERFACE_ADDED, &dev_ctx->state); } - -out: - delayed_event_release(dev_ctx, priv); } void mlx5_attach_device(struct mlx5_core_dev *dev) @@ -422,44 +328,6 @@ struct mlx5_core_dev *mlx5_get_next_phys_dev(struct mlx5_core_dev *dev) return res; } -void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event, - unsigned long param) -{ - struct mlx5_priv *priv = &dev->priv; - struct mlx5_device_context *dev_ctx; - unsigned long flags; - - spin_lock_irqsave(&priv->ctx_lock, flags); - - if (priv->is_accum_events) - add_delayed_event(priv, dev, event, param); - - /* After mlx5_detach_device, the dev_ctx->intf is still set and dev_ctx is - * still in priv->ctx_list. In this case, only notify the dev_ctx if its - * ADDED or ATTACHED bit are set. - */ - list_for_each_entry(dev_ctx, &priv->ctx_list, list) - if (dev_ctx->intf->event && - (test_bit(MLX5_INTERFACE_ADDED, &dev_ctx->state) || - test_bit(MLX5_INTERFACE_ATTACHED, &dev_ctx->state))) - dev_ctx->intf->event(dev, dev_ctx->context, event, param); - - spin_unlock_irqrestore(&priv->ctx_lock, flags); -} - -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING -void mlx5_core_page_fault(struct mlx5_core_dev *dev, - struct mlx5_pagefault *pfault) -{ - struct mlx5_priv *priv = &dev->priv; - int srcu_idx; - - srcu_idx = srcu_read_lock(&priv->pfault_srcu); - if (priv->pfault) - priv->pfault(dev, priv->pfault_ctx, pfault); - srcu_read_unlock(&priv->pfault_srcu, srcu_idx); -} -#endif void mlx5_dev_list_lock(void) { diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c index d4ec93bde4de..6999f4486e9e 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.c @@ -30,6 +30,7 @@ * SOFTWARE. */ #define CREATE_TRACE_POINTS +#include "lib/eq.h" #include "fw_tracer.h" #include "fw_tracer_tracepoint.h" @@ -846,9 +847,9 @@ free_tracer: return ERR_PTR(err); } -/* Create HW resources + start tracer - * must be called before Async EQ is created - */ +static int fw_tracer_event(struct notifier_block *nb, unsigned long action, void *data); + +/* Create HW resources + start tracer */ int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer) { struct mlx5_core_dev *dev; @@ -874,6 +875,9 @@ int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer) goto err_dealloc_pd; } + MLX5_NB_INIT(&tracer->nb, fw_tracer_event, DEVICE_TRACER); + mlx5_eq_notifier_register(dev, &tracer->nb); + mlx5_fw_tracer_start(tracer); return 0; @@ -883,9 +887,7 @@ err_dealloc_pd: return err; } -/* Stop tracer + Cleanup HW resources - * must be called after Async EQ is destroyed - */ +/* Stop tracer + Cleanup HW resources */ void mlx5_fw_tracer_cleanup(struct mlx5_fw_tracer *tracer) { if (IS_ERR_OR_NULL(tracer)) @@ -893,7 +895,7 @@ void mlx5_fw_tracer_cleanup(struct mlx5_fw_tracer *tracer) mlx5_core_dbg(tracer->dev, "FWTracer: Cleanup, is owner ? (%d)\n", tracer->owner); - + mlx5_eq_notifier_unregister(tracer->dev, &tracer->nb); cancel_work_sync(&tracer->ownership_change_work); cancel_work_sync(&tracer->handle_traces_work); @@ -922,12 +924,11 @@ void mlx5_fw_tracer_destroy(struct mlx5_fw_tracer *tracer) kfree(tracer); } -void mlx5_fw_tracer_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe) +static int fw_tracer_event(struct notifier_block *nb, unsigned long action, void *data) { - struct mlx5_fw_tracer *tracer = dev->tracer; - - if (!tracer) - return; + struct mlx5_fw_tracer *tracer = mlx5_nb_cof(nb, struct mlx5_fw_tracer, nb); + struct mlx5_core_dev *dev = tracer->dev; + struct mlx5_eqe *eqe = data; switch (eqe->sub_type) { case MLX5_TRACER_SUBTYPE_OWNERSHIP_CHANGE: @@ -942,6 +943,8 @@ void mlx5_fw_tracer_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe) mlx5_core_dbg(dev, "FWTracer: Event with unrecognized subtype: sub_type %d\n", eqe->sub_type); } + + return NOTIFY_OK; } EXPORT_TRACEPOINT_SYMBOL(mlx5_fw); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h index 0347f2dd5cee..a8b8747f2b61 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/diag/fw_tracer.h @@ -55,6 +55,7 @@ struct mlx5_fw_tracer { struct mlx5_core_dev *dev; + struct mlx5_nb nb; bool owner; u8 trc_ver; struct workqueue_struct *work_queue; @@ -170,6 +171,5 @@ struct mlx5_fw_tracer *mlx5_fw_tracer_create(struct mlx5_core_dev *dev); int mlx5_fw_tracer_init(struct mlx5_fw_tracer *tracer); void mlx5_fw_tracer_cleanup(struct mlx5_fw_tracer *tracer); void mlx5_fw_tracer_destroy(struct mlx5_fw_tracer *tracer); -void mlx5_fw_tracer_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe); #endif diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en.h b/drivers/net/ethernet/mellanox/mlx5/core/en.h index 118324802926..7e0a0cf041d5 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/en.h @@ -178,8 +178,7 @@ static inline int mlx5e_get_max_num_channels(struct mlx5_core_dev *mdev) { return is_kdump_kernel() ? MLX5E_MIN_NUM_CHANNELS : - min_t(int, mdev->priv.eq_table.num_comp_vectors, - MLX5E_MAX_NUM_CHANNELS); + min_t(int, mlx5_comp_vectors_count(mdev), MLX5E_MAX_NUM_CHANNELS); } /* Use this function to get max num channels after netdev was created */ @@ -633,7 +632,6 @@ struct mlx5e_channel_stats { } ____cacheline_aligned_in_smp; enum { - MLX5E_STATE_ASYNC_EVENTS_ENABLED, MLX5E_STATE_OPENED, MLX5E_STATE_DESTROYING, }; @@ -692,6 +690,8 @@ struct mlx5e_priv { struct hwtstamp_config tstamp; u16 q_counter; u16 drop_rq_q_counter; + struct notifier_block events_nb; + #ifdef CONFIG_MLX5_CORE_EN_DCB struct mlx5e_dcbx dcbx; #endif diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c index 871313d6b34d..138c7679eebb 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_main.c @@ -49,6 +49,7 @@ #include "lib/clock.h" #include "en/port.h" #include "en/xdp.h" +#include "lib/eq.h" struct mlx5e_rq_param { u32 rqc[MLX5_ST_SZ_DW(rqc)]; @@ -293,33 +294,35 @@ void mlx5e_queue_update_stats(struct mlx5e_priv *priv) queue_work(priv->wq, &priv->update_stats_work); } -static void mlx5e_async_event(struct mlx5_core_dev *mdev, void *vpriv, - enum mlx5_dev_event event, unsigned long param) +static int async_event(struct notifier_block *nb, unsigned long event, void *data) { - struct mlx5e_priv *priv = vpriv; + struct mlx5e_priv *priv = container_of(nb, struct mlx5e_priv, events_nb); + struct mlx5_eqe *eqe = data; - if (!test_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLED, &priv->state)) - return; + if (event != MLX5_EVENT_TYPE_PORT_CHANGE) + return NOTIFY_DONE; - switch (event) { - case MLX5_DEV_EVENT_PORT_UP: - case MLX5_DEV_EVENT_PORT_DOWN: + switch (eqe->sub_type) { + case MLX5_PORT_CHANGE_SUBTYPE_DOWN: + case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE: queue_work(priv->wq, &priv->update_carrier_work); break; default: - break; + return NOTIFY_DONE; } + + return NOTIFY_OK; } static void mlx5e_enable_async_events(struct mlx5e_priv *priv) { - set_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLED, &priv->state); + priv->events_nb.notifier_call = async_event; + mlx5_notifier_register(priv->mdev, &priv->events_nb); } static void mlx5e_disable_async_events(struct mlx5e_priv *priv) { - clear_bit(MLX5E_STATE_ASYNC_EVENTS_ENABLED, &priv->state); - synchronize_irq(pci_irq_vector(priv->mdev->pdev, MLX5_EQ_VEC_ASYNC)); + mlx5_notifier_unregister(priv->mdev, &priv->events_nb); } static inline void mlx5e_build_umr_wqe(struct mlx5e_rq *rq, @@ -1763,11 +1766,6 @@ static void mlx5e_close_cq(struct mlx5e_cq *cq) mlx5e_free_cq(cq); } -static int mlx5e_get_cpu(struct mlx5e_priv *priv, int ix) -{ - return cpumask_first(priv->mdev->priv.irq_info[ix].mask); -} - static int mlx5e_open_tx_cqs(struct mlx5e_channel *c, struct mlx5e_params *params, struct mlx5e_channel_param *cparam) @@ -1918,9 +1916,9 @@ static int mlx5e_open_channel(struct mlx5e_priv *priv, int ix, struct mlx5e_channel_param *cparam, struct mlx5e_channel **cp) { + int cpu = cpumask_first(mlx5_comp_irq_get_affinity_mask(priv->mdev, ix)); struct net_dim_cq_moder icocq_moder = {0, 0}; struct net_device *netdev = priv->netdev; - int cpu = mlx5e_get_cpu(priv, ix); struct mlx5e_channel *c; unsigned int irq; int err; @@ -4134,17 +4132,17 @@ static netdev_features_t mlx5e_features_check(struct sk_buff *skb, static bool mlx5e_tx_timeout_eq_recover(struct net_device *dev, struct mlx5e_txqsq *sq) { - struct mlx5_eq *eq = sq->cq.mcq.eq; + struct mlx5_eq_comp *eq = sq->cq.mcq.eq; u32 eqe_count; netdev_err(dev, "EQ 0x%x: Cons = 0x%x, irqn = 0x%x\n", - eq->eqn, eq->cons_index, eq->irqn); + eq->core.eqn, eq->core.cons_index, eq->core.irqn); eqe_count = mlx5_eq_poll_irq_disabled(eq); if (!eqe_count) return false; - netdev_err(dev, "Recover %d eqes on EQ 0x%x\n", eqe_count, eq->eqn); + netdev_err(dev, "Recover %d eqes on EQ 0x%x\n", eqe_count, eq->core.eqn); sq->channel->stats->eq_rearm++; return true; } @@ -4975,7 +4973,7 @@ int mlx5e_netdev_init(struct net_device *netdev, netif_carrier_off(netdev); #ifdef CONFIG_MLX5_EN_ARFS - netdev->rx_cpu_rmap = mdev->rmap; + netdev->rx_cpu_rmap = mlx5_eq_table_get_rmap(mdev); #endif return 0; @@ -5199,7 +5197,6 @@ static struct mlx5_interface mlx5e_interface = { .remove = mlx5e_remove, .attach = mlx5e_attach, .detach = mlx5e_detach, - .event = mlx5e_async_event, .protocol = MLX5_INTERFACE_PROTOCOL_ETH, .get_dev = mlx5e_get_netdev, }; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c index 3e99d0728b2f..0b7f3ebb7b9d 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/en_stats.c @@ -30,6 +30,7 @@ * SOFTWARE. */ +#include "lib/mlx5.h" #include "en.h" #include "en_accel/ipsec.h" #include "en_accel/tls.h" @@ -1122,15 +1123,17 @@ static int mlx5e_grp_pme_fill_strings(struct mlx5e_priv *priv, u8 *data, static int mlx5e_grp_pme_fill_stats(struct mlx5e_priv *priv, u64 *data, int idx) { - struct mlx5_priv *mlx5_priv = &priv->mdev->priv; + struct mlx5_pme_stats pme_stats; int i; + mlx5_get_pme_stats(priv->mdev, &pme_stats); + for (i = 0; i < NUM_PME_STATUS_STATS; i++) - data[idx++] = MLX5E_READ_CTR64_CPU(mlx5_priv->pme_stats.status_counters, + data[idx++] = MLX5E_READ_CTR64_CPU(pme_stats.status_counters, mlx5e_pme_status_desc, i); for (i = 0; i < NUM_PME_ERR_STATS; i++) - data[idx++] = MLX5E_READ_CTR64_CPU(mlx5_priv->pme_stats.error_counters, + data[idx++] = MLX5E_READ_CTR64_CPU(pme_stats.error_counters, mlx5e_pme_error_desc, i); return idx; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c index c1e1a16a9b07..4aa39a1fe23f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c @@ -31,20 +31,22 @@ */ #include <linux/interrupt.h> +#include <linux/notifier.h> #include <linux/module.h> #include <linux/mlx5/driver.h> +#include <linux/mlx5/eq.h> #include <linux/mlx5/cmd.h> #ifdef CONFIG_RFS_ACCEL #include <linux/cpu_rmap.h> #endif #include "mlx5_core.h" +#include "lib/eq.h" #include "fpga/core.h" #include "eswitch.h" #include "lib/clock.h" #include "diag/fw_tracer.h" enum { - MLX5_EQE_SIZE = sizeof(struct mlx5_eqe), MLX5_EQE_OWNER_INIT_VAL = 0x1, }; @@ -55,14 +57,32 @@ enum { }; enum { - MLX5_NUM_SPARE_EQE = 0x80, - MLX5_NUM_ASYNC_EQE = 0x1000, - MLX5_NUM_CMD_EQE = 32, - MLX5_NUM_PF_DRAIN = 64, + MLX5_EQ_DOORBEL_OFFSET = 0x40, }; -enum { - MLX5_EQ_DOORBEL_OFFSET = 0x40, +struct mlx5_irq_info { + cpumask_var_t mask; + char name[MLX5_MAX_IRQ_NAME]; + void *context; /* dev_id provided to request_irq */ +}; + +struct mlx5_eq_table { + struct list_head comp_eqs_list; + struct mlx5_eq pages_eq; + struct mlx5_eq cmd_eq; + struct mlx5_eq async_eq; + + struct atomic_notifier_head nh[MLX5_EVENT_TYPE_MAX]; + + /* Since CQ DB is stored in async_eq */ + struct mlx5_nb cq_err_nb; + + struct mutex lock; /* sync async eqs creations */ + int num_comp_vectors; + struct mlx5_irq_info *irq_info; +#ifdef CONFIG_RFS_ACCEL + struct cpu_rmap *rmap; +#endif }; #define MLX5_ASYNC_EVENT_MASK ((1ull << MLX5_EVENT_TYPE_PATH_MIG) | \ @@ -78,17 +98,6 @@ enum { (1ull << MLX5_EVENT_TYPE_SRQ_LAST_WQE) | \ (1ull << MLX5_EVENT_TYPE_SRQ_RQ_LIMIT)) -struct map_eq_in { - u64 mask; - u32 reserved; - u32 unmap_eqn; -}; - -struct cre_des_eq { - u8 reserved[15]; - u8 eqn; -}; - static int mlx5_cmd_destroy_eq(struct mlx5_core_dev *dev, u8 eqn) { u32 out[MLX5_ST_SZ_DW(destroy_eq_out)] = {0}; @@ -99,213 +108,56 @@ static int mlx5_cmd_destroy_eq(struct mlx5_core_dev *dev, u8 eqn) return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); } -static struct mlx5_eqe *get_eqe(struct mlx5_eq *eq, u32 entry) -{ - return mlx5_buf_offset(&eq->buf, entry * MLX5_EQE_SIZE); -} - -static struct mlx5_eqe *next_eqe_sw(struct mlx5_eq *eq) -{ - struct mlx5_eqe *eqe = get_eqe(eq, eq->cons_index & (eq->nent - 1)); - - return ((eqe->owner & 1) ^ !!(eq->cons_index & eq->nent)) ? NULL : eqe; -} - -static const char *eqe_type_str(u8 type) -{ - switch (type) { - case MLX5_EVENT_TYPE_COMP: - return "MLX5_EVENT_TYPE_COMP"; - case MLX5_EVENT_TYPE_PATH_MIG: - return "MLX5_EVENT_TYPE_PATH_MIG"; - case MLX5_EVENT_TYPE_COMM_EST: - return "MLX5_EVENT_TYPE_COMM_EST"; - case MLX5_EVENT_TYPE_SQ_DRAINED: - return "MLX5_EVENT_TYPE_SQ_DRAINED"; - case MLX5_EVENT_TYPE_SRQ_LAST_WQE: - return "MLX5_EVENT_TYPE_SRQ_LAST_WQE"; - case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT: - return "MLX5_EVENT_TYPE_SRQ_RQ_LIMIT"; - case MLX5_EVENT_TYPE_CQ_ERROR: - return "MLX5_EVENT_TYPE_CQ_ERROR"; - case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: - return "MLX5_EVENT_TYPE_WQ_CATAS_ERROR"; - case MLX5_EVENT_TYPE_PATH_MIG_FAILED: - return "MLX5_EVENT_TYPE_PATH_MIG_FAILED"; - case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: - return "MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR"; - case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: - return "MLX5_EVENT_TYPE_WQ_ACCESS_ERROR"; - case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: - return "MLX5_EVENT_TYPE_SRQ_CATAS_ERROR"; - case MLX5_EVENT_TYPE_INTERNAL_ERROR: - return "MLX5_EVENT_TYPE_INTERNAL_ERROR"; - case MLX5_EVENT_TYPE_PORT_CHANGE: - return "MLX5_EVENT_TYPE_PORT_CHANGE"; - case MLX5_EVENT_TYPE_GPIO_EVENT: - return "MLX5_EVENT_TYPE_GPIO_EVENT"; - case MLX5_EVENT_TYPE_PORT_MODULE_EVENT: - return "MLX5_EVENT_TYPE_PORT_MODULE_EVENT"; - case MLX5_EVENT_TYPE_TEMP_WARN_EVENT: - return "MLX5_EVENT_TYPE_TEMP_WARN_EVENT"; - case MLX5_EVENT_TYPE_REMOTE_CONFIG: - return "MLX5_EVENT_TYPE_REMOTE_CONFIG"; - case MLX5_EVENT_TYPE_DB_BF_CONGESTION: - return "MLX5_EVENT_TYPE_DB_BF_CONGESTION"; - case MLX5_EVENT_TYPE_STALL_EVENT: - return "MLX5_EVENT_TYPE_STALL_EVENT"; - case MLX5_EVENT_TYPE_CMD: - return "MLX5_EVENT_TYPE_CMD"; - case MLX5_EVENT_TYPE_PAGE_REQUEST: - return "MLX5_EVENT_TYPE_PAGE_REQUEST"; - case MLX5_EVENT_TYPE_PAGE_FAULT: - return "MLX5_EVENT_TYPE_PAGE_FAULT"; - case MLX5_EVENT_TYPE_PPS_EVENT: - return "MLX5_EVENT_TYPE_PPS_EVENT"; - case MLX5_EVENT_TYPE_NIC_VPORT_CHANGE: - return "MLX5_EVENT_TYPE_NIC_VPORT_CHANGE"; - case MLX5_EVENT_TYPE_FPGA_ERROR: - return "MLX5_EVENT_TYPE_FPGA_ERROR"; - case MLX5_EVENT_TYPE_FPGA_QP_ERROR: - return "MLX5_EVENT_TYPE_FPGA_QP_ERROR"; - case MLX5_EVENT_TYPE_GENERAL_EVENT: - return "MLX5_EVENT_TYPE_GENERAL_EVENT"; - case MLX5_EVENT_TYPE_DEVICE_TRACER: - return "MLX5_EVENT_TYPE_DEVICE_TRACER"; - default: - return "Unrecognized event"; - } -} - -static enum mlx5_dev_event port_subtype_event(u8 subtype) -{ - switch (subtype) { - case MLX5_PORT_CHANGE_SUBTYPE_DOWN: - return MLX5_DEV_EVENT_PORT_DOWN; - case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE: - return MLX5_DEV_EVENT_PORT_UP; - case MLX5_PORT_CHANGE_SUBTYPE_INITIALIZED: - return MLX5_DEV_EVENT_PORT_INITIALIZED; - case MLX5_PORT_CHANGE_SUBTYPE_LID: - return MLX5_DEV_EVENT_LID_CHANGE; - case MLX5_PORT_CHANGE_SUBTYPE_PKEY: - return MLX5_DEV_EVENT_PKEY_CHANGE; - case MLX5_PORT_CHANGE_SUBTYPE_GUID: - return MLX5_DEV_EVENT_GUID_CHANGE; - case MLX5_PORT_CHANGE_SUBTYPE_CLIENT_REREG: - return MLX5_DEV_EVENT_CLIENT_REREG; - } - return -1; -} - -static void eq_update_ci(struct mlx5_eq *eq, int arm) +/* caller must eventually call mlx5_cq_put on the returned cq */ +static struct mlx5_core_cq *mlx5_eq_cq_get(struct mlx5_eq *eq, u32 cqn) { - __be32 __iomem *addr = eq->doorbell + (arm ? 0 : 2); - u32 val = (eq->cons_index & 0xffffff) | (eq->eqn << 24); - - __raw_writel((__force u32)cpu_to_be32(val), addr); - /* We still want ordering, just not swabbing, so add a barrier */ - mb(); -} + struct mlx5_cq_table *table = &eq->cq_table; + struct mlx5_core_cq *cq = NULL; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING -static void eqe_pf_action(struct work_struct *work) -{ - struct mlx5_pagefault *pfault = container_of(work, - struct mlx5_pagefault, - work); - struct mlx5_eq *eq = pfault->eq; + spin_lock(&table->lock); + cq = radix_tree_lookup(&table->tree, cqn); + if (likely(cq)) + mlx5_cq_hold(cq); + spin_unlock(&table->lock); - mlx5_core_page_fault(eq->dev, pfault); - mempool_free(pfault, eq->pf_ctx.pool); + return cq; } -static void eq_pf_process(struct mlx5_eq *eq) +static irqreturn_t mlx5_eq_comp_int(int irq, void *eq_ptr) { - struct mlx5_core_dev *dev = eq->dev; - struct mlx5_eqe_page_fault *pf_eqe; - struct mlx5_pagefault *pfault; + struct mlx5_eq_comp *eq_comp = eq_ptr; + struct mlx5_eq *eq = eq_ptr; struct mlx5_eqe *eqe; int set_ci = 0; + u32 cqn = -1; while ((eqe = next_eqe_sw(eq))) { - pfault = mempool_alloc(eq->pf_ctx.pool, GFP_ATOMIC); - if (!pfault) { - schedule_work(&eq->pf_ctx.work); - break; - } - + struct mlx5_core_cq *cq; + /* Make sure we read EQ entry contents after we've + * checked the ownership bit. + */ dma_rmb(); - pf_eqe = &eqe->data.page_fault; - pfault->event_subtype = eqe->sub_type; - pfault->bytes_committed = be32_to_cpu(pf_eqe->bytes_committed); - - mlx5_core_dbg(dev, - "PAGE_FAULT: subtype: 0x%02x, bytes_committed: 0x%06x\n", - eqe->sub_type, pfault->bytes_committed); - - switch (eqe->sub_type) { - case MLX5_PFAULT_SUBTYPE_RDMA: - /* RDMA based event */ - pfault->type = - be32_to_cpu(pf_eqe->rdma.pftype_token) >> 24; - pfault->token = - be32_to_cpu(pf_eqe->rdma.pftype_token) & - MLX5_24BIT_MASK; - pfault->rdma.r_key = - be32_to_cpu(pf_eqe->rdma.r_key); - pfault->rdma.packet_size = - be16_to_cpu(pf_eqe->rdma.packet_length); - pfault->rdma.rdma_op_len = - be32_to_cpu(pf_eqe->rdma.rdma_op_len); - pfault->rdma.rdma_va = - be64_to_cpu(pf_eqe->rdma.rdma_va); - mlx5_core_dbg(dev, - "PAGE_FAULT: type:0x%x, token: 0x%06x, r_key: 0x%08x\n", - pfault->type, pfault->token, - pfault->rdma.r_key); - mlx5_core_dbg(dev, - "PAGE_FAULT: rdma_op_len: 0x%08x, rdma_va: 0x%016llx\n", - pfault->rdma.rdma_op_len, - pfault->rdma.rdma_va); - break; - - case MLX5_PFAULT_SUBTYPE_WQE: - /* WQE based event */ - pfault->type = - (be32_to_cpu(pf_eqe->wqe.pftype_wq) >> 24) & 0x7; - pfault->token = - be32_to_cpu(pf_eqe->wqe.token); - pfault->wqe.wq_num = - be32_to_cpu(pf_eqe->wqe.pftype_wq) & - MLX5_24BIT_MASK; - pfault->wqe.wqe_index = - be16_to_cpu(pf_eqe->wqe.wqe_index); - pfault->wqe.packet_size = - be16_to_cpu(pf_eqe->wqe.packet_length); - mlx5_core_dbg(dev, - "PAGE_FAULT: type:0x%x, token: 0x%06x, wq_num: 0x%06x, wqe_index: 0x%04x\n", - pfault->type, pfault->token, - pfault->wqe.wq_num, - pfault->wqe.wqe_index); - break; - - default: - mlx5_core_warn(dev, - "Unsupported page fault event sub-type: 0x%02hhx\n", - eqe->sub_type); - /* Unsupported page faults should still be - * resolved by the page fault handler - */ + /* Assume (eqe->type) is always MLX5_EVENT_TYPE_COMP */ + cqn = be32_to_cpu(eqe->data.comp.cqn) & 0xffffff; + + cq = mlx5_eq_cq_get(eq, cqn); + if (likely(cq)) { + ++cq->arm_sn; + cq->comp(cq); + mlx5_cq_put(cq); + } else { + mlx5_core_warn(eq->dev, "Completion event for bogus CQ 0x%x\n", cqn); } - pfault->eq = eq; - INIT_WORK(&pfault->work, eqe_pf_action); - queue_work(eq->pf_ctx.wq, &pfault->work); - ++eq->cons_index; ++set_ci; + /* The HCA will think the queue has overflowed if we + * don't tell it we've been processing events. We + * create our EQs with MLX5_NUM_SPARE_EQE extra + * entries, so we must update our consumer index at + * least that often. + */ if (unlikely(set_ci >= MLX5_NUM_SPARE_EQE)) { eq_update_ci(eq, 0); set_ci = 0; @@ -313,165 +165,41 @@ static void eq_pf_process(struct mlx5_eq *eq) } eq_update_ci(eq, 1); -} -static irqreturn_t mlx5_eq_pf_int(int irq, void *eq_ptr) -{ - struct mlx5_eq *eq = eq_ptr; - unsigned long flags; - - if (spin_trylock_irqsave(&eq->pf_ctx.lock, flags)) { - eq_pf_process(eq); - spin_unlock_irqrestore(&eq->pf_ctx.lock, flags); - } else { - schedule_work(&eq->pf_ctx.work); - } + if (cqn != -1) + tasklet_schedule(&eq_comp->tasklet_ctx.task); return IRQ_HANDLED; } -/* mempool_refill() was proposed but unfortunately wasn't accepted - * http://lkml.iu.edu/hypermail/linux/kernel/1512.1/05073.html - * Chip workaround. +/* Some architectures don't latch interrupts when they are disabled, so using + * mlx5_eq_poll_irq_disabled could end up losing interrupts while trying to + * avoid losing them. It is not recommended to use it, unless this is the last + * resort. */ -static void mempool_refill(mempool_t *pool) -{ - while (pool->curr_nr < pool->min_nr) - mempool_free(mempool_alloc(pool, GFP_KERNEL), pool); -} - -static void eq_pf_action(struct work_struct *work) -{ - struct mlx5_eq *eq = container_of(work, struct mlx5_eq, pf_ctx.work); - - mempool_refill(eq->pf_ctx.pool); - - spin_lock_irq(&eq->pf_ctx.lock); - eq_pf_process(eq); - spin_unlock_irq(&eq->pf_ctx.lock); -} - -static int init_pf_ctx(struct mlx5_eq_pagefault *pf_ctx, const char *name) -{ - spin_lock_init(&pf_ctx->lock); - INIT_WORK(&pf_ctx->work, eq_pf_action); - - pf_ctx->wq = alloc_ordered_workqueue(name, - WQ_MEM_RECLAIM); - if (!pf_ctx->wq) - return -ENOMEM; - - pf_ctx->pool = mempool_create_kmalloc_pool - (MLX5_NUM_PF_DRAIN, sizeof(struct mlx5_pagefault)); - if (!pf_ctx->pool) - goto err_wq; - - return 0; -err_wq: - destroy_workqueue(pf_ctx->wq); - return -ENOMEM; -} - -int mlx5_core_page_fault_resume(struct mlx5_core_dev *dev, u32 token, - u32 wq_num, u8 type, int error) -{ - u32 out[MLX5_ST_SZ_DW(page_fault_resume_out)] = {0}; - u32 in[MLX5_ST_SZ_DW(page_fault_resume_in)] = {0}; - - MLX5_SET(page_fault_resume_in, in, opcode, - MLX5_CMD_OP_PAGE_FAULT_RESUME); - MLX5_SET(page_fault_resume_in, in, error, !!error); - MLX5_SET(page_fault_resume_in, in, page_fault_type, type); - MLX5_SET(page_fault_resume_in, in, wq_number, wq_num); - MLX5_SET(page_fault_resume_in, in, token, token); - - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); -} -EXPORT_SYMBOL_GPL(mlx5_core_page_fault_resume); -#endif - -static void general_event_handler(struct mlx5_core_dev *dev, - struct mlx5_eqe *eqe) -{ - switch (eqe->sub_type) { - case MLX5_GENERAL_SUBTYPE_DELAY_DROP_TIMEOUT: - if (dev->event) - dev->event(dev, MLX5_DEV_EVENT_DELAY_DROP_TIMEOUT, 0); - break; - default: - mlx5_core_dbg(dev, "General event with unrecognized subtype: sub_type %d\n", - eqe->sub_type); - } -} - -static void mlx5_temp_warning_event(struct mlx5_core_dev *dev, - struct mlx5_eqe *eqe) -{ - u64 value_lsb; - u64 value_msb; - - value_lsb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_lsb); - value_msb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_msb); - - mlx5_core_warn(dev, - "High temperature on sensors with bit set %llx %llx", - value_msb, value_lsb); -} - -/* caller must eventually call mlx5_cq_put on the returned cq */ -static struct mlx5_core_cq *mlx5_eq_cq_get(struct mlx5_eq *eq, u32 cqn) -{ - struct mlx5_cq_table *table = &eq->cq_table; - struct mlx5_core_cq *cq = NULL; - - spin_lock(&table->lock); - cq = radix_tree_lookup(&table->tree, cqn); - if (likely(cq)) - mlx5_cq_hold(cq); - spin_unlock(&table->lock); - - return cq; -} - -static void mlx5_eq_cq_completion(struct mlx5_eq *eq, u32 cqn) +u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq_comp *eq) { - struct mlx5_core_cq *cq = mlx5_eq_cq_get(eq, cqn); - - if (unlikely(!cq)) { - mlx5_core_warn(eq->dev, "Completion event for bogus CQ 0x%x\n", cqn); - return; - } - - ++cq->arm_sn; - - cq->comp(cq); - - mlx5_cq_put(cq); -} - -static void mlx5_eq_cq_event(struct mlx5_eq *eq, u32 cqn, int event_type) -{ - struct mlx5_core_cq *cq = mlx5_eq_cq_get(eq, cqn); - - if (unlikely(!cq)) { - mlx5_core_warn(eq->dev, "Async event for bogus CQ 0x%x\n", cqn); - return; - } + u32 count_eqe; - cq->event(cq, event_type); + disable_irq(eq->core.irqn); + count_eqe = eq->core.cons_index; + mlx5_eq_comp_int(eq->core.irqn, eq); + count_eqe = eq->core.cons_index - count_eqe; + enable_irq(eq->core.irqn); - mlx5_cq_put(cq); + return count_eqe; } -static irqreturn_t mlx5_eq_int(int irq, void *eq_ptr) +static irqreturn_t mlx5_eq_async_int(int irq, void *eq_ptr) { struct mlx5_eq *eq = eq_ptr; - struct mlx5_core_dev *dev = eq->dev; + struct mlx5_eq_table *eqt; + struct mlx5_core_dev *dev; struct mlx5_eqe *eqe; int set_ci = 0; - u32 cqn = -1; - u32 rsn; - u8 port; + + dev = eq->dev; + eqt = dev->priv.eq_table; while ((eqe = next_eqe_sw(eq))) { /* @@ -480,116 +208,12 @@ static irqreturn_t mlx5_eq_int(int irq, void *eq_ptr) */ dma_rmb(); - mlx5_core_dbg(eq->dev, "eqn %d, eqe type %s\n", - eq->eqn, eqe_type_str(eqe->type)); - switch (eqe->type) { - case MLX5_EVENT_TYPE_COMP: - cqn = be32_to_cpu(eqe->data.comp.cqn) & 0xffffff; - mlx5_eq_cq_completion(eq, cqn); - break; - case MLX5_EVENT_TYPE_DCT_DRAINED: - rsn = be32_to_cpu(eqe->data.dct.dctn) & 0xffffff; - rsn |= (MLX5_RES_DCT << MLX5_USER_INDEX_LEN); - mlx5_rsc_event(dev, rsn, eqe->type); - break; - case MLX5_EVENT_TYPE_PATH_MIG: - case MLX5_EVENT_TYPE_COMM_EST: - case MLX5_EVENT_TYPE_SQ_DRAINED: - case MLX5_EVENT_TYPE_SRQ_LAST_WQE: - case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: - case MLX5_EVENT_TYPE_PATH_MIG_FAILED: - case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: - case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: - rsn = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; - rsn |= (eqe->data.qp_srq.type << MLX5_USER_INDEX_LEN); - mlx5_core_dbg(dev, "event %s(%d) arrived on resource 0x%x\n", - eqe_type_str(eqe->type), eqe->type, rsn); - mlx5_rsc_event(dev, rsn, eqe->type); - break; - - case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT: - case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: - rsn = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; - mlx5_core_dbg(dev, "SRQ event %s(%d): srqn 0x%x\n", - eqe_type_str(eqe->type), eqe->type, rsn); - mlx5_srq_event(dev, rsn, eqe->type); - break; - - case MLX5_EVENT_TYPE_CMD: - mlx5_cmd_comp_handler(dev, be32_to_cpu(eqe->data.cmd.vector), false); - break; - - case MLX5_EVENT_TYPE_PORT_CHANGE: - port = (eqe->data.port.port >> 4) & 0xf; - switch (eqe->sub_type) { - case MLX5_PORT_CHANGE_SUBTYPE_DOWN: - case MLX5_PORT_CHANGE_SUBTYPE_ACTIVE: - case MLX5_PORT_CHANGE_SUBTYPE_LID: - case MLX5_PORT_CHANGE_SUBTYPE_PKEY: - case MLX5_PORT_CHANGE_SUBTYPE_GUID: - case MLX5_PORT_CHANGE_SUBTYPE_CLIENT_REREG: - case MLX5_PORT_CHANGE_SUBTYPE_INITIALIZED: - if (dev->event) - dev->event(dev, port_subtype_event(eqe->sub_type), - (unsigned long)port); - break; - default: - mlx5_core_warn(dev, "Port event with unrecognized subtype: port %d, sub_type %d\n", - port, eqe->sub_type); - } - break; - case MLX5_EVENT_TYPE_CQ_ERROR: - cqn = be32_to_cpu(eqe->data.cq_err.cqn) & 0xffffff; - mlx5_core_warn(dev, "CQ error on CQN 0x%x, syndrome 0x%x\n", - cqn, eqe->data.cq_err.syndrome); - mlx5_eq_cq_event(eq, cqn, eqe->type); - break; - - case MLX5_EVENT_TYPE_PAGE_REQUEST: - { - u16 func_id = be16_to_cpu(eqe->data.req_pages.func_id); - s32 npages = be32_to_cpu(eqe->data.req_pages.num_pages); + if (likely(eqe->type < MLX5_EVENT_TYPE_MAX)) + atomic_notifier_call_chain(&eqt->nh[eqe->type], eqe->type, eqe); + else + mlx5_core_warn_once(dev, "notifier_call_chain is not setup for eqe: %d\n", eqe->type); - mlx5_core_dbg(dev, "page request for func 0x%x, npages %d\n", - func_id, npages); - mlx5_core_req_pages_handler(dev, func_id, npages); - } - break; - - case MLX5_EVENT_TYPE_NIC_VPORT_CHANGE: - mlx5_eswitch_vport_event(dev->priv.eswitch, eqe); - break; - - case MLX5_EVENT_TYPE_PORT_MODULE_EVENT: - mlx5_port_module_event(dev, eqe); - break; - - case MLX5_EVENT_TYPE_PPS_EVENT: - mlx5_pps_event(dev, eqe); - break; - - case MLX5_EVENT_TYPE_FPGA_ERROR: - case MLX5_EVENT_TYPE_FPGA_QP_ERROR: - mlx5_fpga_event(dev, eqe->type, &eqe->data.raw); - break; - - case MLX5_EVENT_TYPE_TEMP_WARN_EVENT: - mlx5_temp_warning_event(dev, eqe); - break; - - case MLX5_EVENT_TYPE_GENERAL_EVENT: - general_event_handler(dev, eqe); - break; - - case MLX5_EVENT_TYPE_DEVICE_TRACER: - mlx5_fw_tracer_event(dev, eqe); - break; - - default: - mlx5_core_warn(dev, "Unhandled event 0x%x on EQ 0x%x\n", - eqe->type, eq->eqn); - break; - } + atomic_notifier_call_chain(&eqt->nh[MLX5_EVENT_TYPE_NOTIFY_ANY], eqe->type, eqe); ++eq->cons_index; ++set_ci; @@ -608,30 +232,9 @@ static irqreturn_t mlx5_eq_int(int irq, void *eq_ptr) eq_update_ci(eq, 1); - if (cqn != -1) - tasklet_schedule(&eq->tasklet_ctx.task); - return IRQ_HANDLED; } -/* Some architectures don't latch interrupts when they are disabled, so using - * mlx5_eq_poll_irq_disabled could end up losing interrupts while trying to - * avoid losing them. It is not recommended to use it, unless this is the last - * resort. - */ -u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq *eq) -{ - u32 count_eqe; - - disable_irq(eq->irqn); - count_eqe = eq->cons_index; - mlx5_eq_int(eq->irqn, eq); - count_eqe = eq->cons_index - count_eqe; - enable_irq(eq->irqn); - - return count_eqe; -} - static void init_eq_buf(struct mlx5_eq *eq) { struct mlx5_eqe *eqe; @@ -643,39 +246,35 @@ static void init_eq_buf(struct mlx5_eq *eq) } } -int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, - int nent, u64 mask, const char *name, - enum mlx5_eq_type type) +static int +create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, const char *name, + struct mlx5_eq_param *param) { + struct mlx5_eq_table *eq_table = dev->priv.eq_table; struct mlx5_cq_table *cq_table = &eq->cq_table; u32 out[MLX5_ST_SZ_DW(create_eq_out)] = {0}; struct mlx5_priv *priv = &dev->priv; - irq_handler_t handler; + u8 vecidx = param->index; __be64 *pas; void *eqc; int inlen; u32 *in; int err; + if (eq_table->irq_info[vecidx].context) + return -EEXIST; + /* Init CQ table */ memset(cq_table, 0, sizeof(*cq_table)); spin_lock_init(&cq_table->lock); INIT_RADIX_TREE(&cq_table->tree, GFP_ATOMIC); - eq->type = type; - eq->nent = roundup_pow_of_two(nent + MLX5_NUM_SPARE_EQE); + eq->nent = roundup_pow_of_two(param->nent + MLX5_NUM_SPARE_EQE); eq->cons_index = 0; err = mlx5_buf_alloc(dev, eq->nent * MLX5_EQE_SIZE, &eq->buf); if (err) return err; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (type == MLX5_EQ_TYPE_PF) - handler = mlx5_eq_pf_int; - else -#endif - handler = mlx5_eq_int; - init_eq_buf(eq); inlen = MLX5_ST_SZ_BYTES(create_eq_in) + @@ -691,7 +290,7 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, mlx5_fill_page_array(&eq->buf, pas); MLX5_SET(create_eq_in, in, opcode, MLX5_CMD_OP_CREATE_EQ); - MLX5_SET64(create_eq_in, in, event_bitmask, mask); + MLX5_SET64(create_eq_in, in, event_bitmask, param->mask); eqc = MLX5_ADDR_OF(create_eq_in, in, eq_context_entry); MLX5_SET(eqc, eqc, log_eq_size, ilog2(eq->nent)); @@ -704,15 +303,17 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, if (err) goto err_in; - snprintf(priv->irq_info[vecidx].name, MLX5_MAX_IRQ_NAME, "%s@pci:%s", + snprintf(eq_table->irq_info[vecidx].name, MLX5_MAX_IRQ_NAME, "%s@pci:%s", name, pci_name(dev->pdev)); + eq_table->irq_info[vecidx].context = param->context; + eq->vecidx = vecidx; eq->eqn = MLX5_GET(create_eq_out, out, eq_number); eq->irqn = pci_irq_vector(dev->pdev, vecidx); eq->dev = dev; eq->doorbell = priv->uar->map + MLX5_EQ_DOORBEL_OFFSET; - err = request_irq(eq->irqn, handler, 0, - priv->irq_info[vecidx].name, eq); + err = request_irq(eq->irqn, param->handler, 0, + eq_table->irq_info[vecidx].name, param->context); if (err) goto err_eq; @@ -720,21 +321,6 @@ int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, if (err) goto err_irq; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (type == MLX5_EQ_TYPE_PF) { - err = init_pf_ctx(&eq->pf_ctx, name); - if (err) - goto err_irq; - } else -#endif - { - INIT_LIST_HEAD(&eq->tasklet_ctx.list); - INIT_LIST_HEAD(&eq->tasklet_ctx.process_list); - spin_lock_init(&eq->tasklet_ctx.lock); - tasklet_init(&eq->tasklet_ctx.task, mlx5_cq_tasklet_cb, - (unsigned long)&eq->tasklet_ctx); - } - /* EQs are created in ARMED state */ eq_update_ci(eq, 1); @@ -756,27 +342,25 @@ err_buf: return err; } -int mlx5_destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq) +static int destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq) { + struct mlx5_eq_table *eq_table = dev->priv.eq_table; + struct mlx5_irq_info *irq_info; int err; + irq_info = &eq_table->irq_info[eq->vecidx]; + mlx5_debug_eq_remove(dev, eq); - free_irq(eq->irqn, eq); + + free_irq(eq->irqn, irq_info->context); + irq_info->context = NULL; + err = mlx5_cmd_destroy_eq(dev, eq->eqn); if (err) mlx5_core_warn(dev, "failed to destroy a previously created eq: eqn %d\n", eq->eqn); synchronize_irq(eq->irqn); - if (eq->type == MLX5_EQ_TYPE_COMP) { - tasklet_disable(&eq->tasklet_ctx.task); -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - } else if (eq->type == MLX5_EQ_TYPE_PF) { - cancel_work_sync(&eq->pf_ctx.work); - destroy_workqueue(eq->pf_ctx.wq); - mempool_destroy(eq->pf_ctx.pool); -#endif - } mlx5_buf_free(dev, &eq->buf); return err; @@ -816,28 +400,106 @@ int mlx5_eq_del_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq) return 0; } -int mlx5_eq_init(struct mlx5_core_dev *dev) +int mlx5_eq_table_init(struct mlx5_core_dev *dev) { - int err; + struct mlx5_eq_table *eq_table; + int i, err; - spin_lock_init(&dev->priv.eq_table.lock); + eq_table = kvzalloc(sizeof(*eq_table), GFP_KERNEL); + if (!eq_table) + return -ENOMEM; + + dev->priv.eq_table = eq_table; err = mlx5_eq_debugfs_init(dev); + if (err) + goto kvfree_eq_table; + mutex_init(&eq_table->lock); + for (i = 0; i < MLX5_EVENT_TYPE_MAX; i++) + ATOMIC_INIT_NOTIFIER_HEAD(&eq_table->nh[i]); + + return 0; + +kvfree_eq_table: + kvfree(eq_table); + dev->priv.eq_table = NULL; return err; } -void mlx5_eq_cleanup(struct mlx5_core_dev *dev) +void mlx5_eq_table_cleanup(struct mlx5_core_dev *dev) { mlx5_eq_debugfs_cleanup(dev); + kvfree(dev->priv.eq_table); } -int mlx5_start_eqs(struct mlx5_core_dev *dev) +/* Async EQs */ + +static int create_async_eq(struct mlx5_core_dev *dev, const char *name, + struct mlx5_eq *eq, struct mlx5_eq_param *param) { - struct mlx5_eq_table *table = &dev->priv.eq_table; - u64 async_event_mask = MLX5_ASYNC_EVENT_MASK; + struct mlx5_eq_table *eq_table = dev->priv.eq_table; + int err; + + mutex_lock(&eq_table->lock); + if (param->index >= MLX5_EQ_MAX_ASYNC_EQS) { + err = -ENOSPC; + goto unlock; + } + + err = create_map_eq(dev, eq, name, param); +unlock: + mutex_unlock(&eq_table->lock); + return err; +} + +static int destroy_async_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq) +{ + struct mlx5_eq_table *eq_table = dev->priv.eq_table; int err; + mutex_lock(&eq_table->lock); + err = destroy_unmap_eq(dev, eq); + mutex_unlock(&eq_table->lock); + return err; +} + +static int cq_err_event_notifier(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_eq_table *eqt; + struct mlx5_core_cq *cq; + struct mlx5_eqe *eqe; + struct mlx5_eq *eq; + u32 cqn; + + /* type == MLX5_EVENT_TYPE_CQ_ERROR */ + + eqt = mlx5_nb_cof(nb, struct mlx5_eq_table, cq_err_nb); + eq = &eqt->async_eq; + eqe = data; + + cqn = be32_to_cpu(eqe->data.cq_err.cqn) & 0xffffff; + mlx5_core_warn(eq->dev, "CQ error on CQN 0x%x, syndrome 0x%x\n", + cqn, eqe->data.cq_err.syndrome); + + cq = mlx5_eq_cq_get(eq, cqn); + if (unlikely(!cq)) { + mlx5_core_warn(eq->dev, "Async event for bogus CQ 0x%x\n", cqn); + return NOTIFY_OK; + } + + cq->event(cq, type); + + mlx5_cq_put(cq); + + return NOTIFY_OK; +} + +static u64 gather_async_events_mask(struct mlx5_core_dev *dev) +{ + u64 async_event_mask = MLX5_ASYNC_EVENT_MASK; + if (MLX5_VPORT_MANAGER(dev)) async_event_mask |= (1ull << MLX5_EVENT_TYPE_NIC_VPORT_CHANGE); @@ -865,127 +527,518 @@ int mlx5_start_eqs(struct mlx5_core_dev *dev) if (MLX5_CAP_MCAM_REG(dev, tracer_registers)) async_event_mask |= (1ull << MLX5_EVENT_TYPE_DEVICE_TRACER); - err = mlx5_create_map_eq(dev, &table->cmd_eq, MLX5_EQ_VEC_CMD, - MLX5_NUM_CMD_EQE, 1ull << MLX5_EVENT_TYPE_CMD, - "mlx5_cmd_eq", MLX5_EQ_TYPE_ASYNC); + return async_event_mask; +} + +static int create_async_eqs(struct mlx5_core_dev *dev) +{ + struct mlx5_eq_table *table = dev->priv.eq_table; + struct mlx5_eq_param param = {}; + int err; + + MLX5_NB_INIT(&table->cq_err_nb, cq_err_event_notifier, CQ_ERROR); + mlx5_eq_notifier_register(dev, &table->cq_err_nb); + + param = (struct mlx5_eq_param) { + .index = MLX5_EQ_CMD_IDX, + .mask = 1ull << MLX5_EVENT_TYPE_CMD, + .nent = MLX5_NUM_CMD_EQE, + .context = &table->cmd_eq, + .handler = mlx5_eq_async_int, + }; + err = create_async_eq(dev, "mlx5_cmd_eq", &table->cmd_eq, ¶m); if (err) { mlx5_core_warn(dev, "failed to create cmd EQ %d\n", err); - return err; + goto err0; } mlx5_cmd_use_events(dev); - err = mlx5_create_map_eq(dev, &table->async_eq, MLX5_EQ_VEC_ASYNC, - MLX5_NUM_ASYNC_EQE, async_event_mask, - "mlx5_async_eq", MLX5_EQ_TYPE_ASYNC); + param = (struct mlx5_eq_param) { + .index = MLX5_EQ_ASYNC_IDX, + .mask = gather_async_events_mask(dev), + .nent = MLX5_NUM_ASYNC_EQE, + .context = &table->async_eq, + .handler = mlx5_eq_async_int, + }; + err = create_async_eq(dev, "mlx5_async_eq", &table->async_eq, ¶m); if (err) { mlx5_core_warn(dev, "failed to create async EQ %d\n", err); goto err1; } - err = mlx5_create_map_eq(dev, &table->pages_eq, - MLX5_EQ_VEC_PAGES, - /* TODO: sriov max_vf + */ 1, - 1 << MLX5_EVENT_TYPE_PAGE_REQUEST, "mlx5_pages_eq", - MLX5_EQ_TYPE_ASYNC); + param = (struct mlx5_eq_param) { + .index = MLX5_EQ_PAGEREQ_IDX, + .mask = 1 << MLX5_EVENT_TYPE_PAGE_REQUEST, + .nent = /* TODO: sriov max_vf + */ 1, + .context = &table->pages_eq, + .handler = mlx5_eq_async_int, + }; + err = create_async_eq(dev, "mlx5_pages_eq", &table->pages_eq, ¶m); if (err) { mlx5_core_warn(dev, "failed to create pages EQ %d\n", err); goto err2; } -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (MLX5_CAP_GEN(dev, pg)) { - err = mlx5_create_map_eq(dev, &table->pfault_eq, - MLX5_EQ_VEC_PFAULT, - MLX5_NUM_ASYNC_EQE, - 1 << MLX5_EVENT_TYPE_PAGE_FAULT, - "mlx5_page_fault_eq", - MLX5_EQ_TYPE_PF); - if (err) { - mlx5_core_warn(dev, "failed to create page fault EQ %d\n", - err); - goto err3; - } - } - - return err; -err3: - mlx5_destroy_unmap_eq(dev, &table->pages_eq); -#else return err; -#endif err2: - mlx5_destroy_unmap_eq(dev, &table->async_eq); + destroy_async_eq(dev, &table->async_eq); err1: mlx5_cmd_use_polling(dev); - mlx5_destroy_unmap_eq(dev, &table->cmd_eq); + destroy_async_eq(dev, &table->cmd_eq); +err0: + mlx5_eq_notifier_unregister(dev, &table->cq_err_nb); return err; } -void mlx5_stop_eqs(struct mlx5_core_dev *dev) +static void destroy_async_eqs(struct mlx5_core_dev *dev) { - struct mlx5_eq_table *table = &dev->priv.eq_table; + struct mlx5_eq_table *table = dev->priv.eq_table; int err; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (MLX5_CAP_GEN(dev, pg)) { - err = mlx5_destroy_unmap_eq(dev, &table->pfault_eq); - if (err) - mlx5_core_err(dev, "failed to destroy page fault eq, err(%d)\n", - err); - } -#endif - - err = mlx5_destroy_unmap_eq(dev, &table->pages_eq); + err = destroy_async_eq(dev, &table->pages_eq); if (err) mlx5_core_err(dev, "failed to destroy pages eq, err(%d)\n", err); - err = mlx5_destroy_unmap_eq(dev, &table->async_eq); + err = destroy_async_eq(dev, &table->async_eq); if (err) mlx5_core_err(dev, "failed to destroy async eq, err(%d)\n", err); + mlx5_cmd_use_polling(dev); - err = mlx5_destroy_unmap_eq(dev, &table->cmd_eq); + err = destroy_async_eq(dev, &table->cmd_eq); if (err) mlx5_core_err(dev, "failed to destroy command eq, err(%d)\n", err); + + mlx5_eq_notifier_unregister(dev, &table->cq_err_nb); } -int mlx5_core_eq_query(struct mlx5_core_dev *dev, struct mlx5_eq *eq, - u32 *out, int outlen) +struct mlx5_eq *mlx5_get_async_eq(struct mlx5_core_dev *dev) { - u32 in[MLX5_ST_SZ_DW(query_eq_in)] = {0}; + return &dev->priv.eq_table->async_eq; +} - MLX5_SET(query_eq_in, in, opcode, MLX5_CMD_OP_QUERY_EQ); - MLX5_SET(query_eq_in, in, eq_number, eq->eqn); - return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen); +void mlx5_eq_synchronize_async_irq(struct mlx5_core_dev *dev) +{ + synchronize_irq(dev->priv.eq_table->async_eq.irqn); +} + +void mlx5_eq_synchronize_cmd_irq(struct mlx5_core_dev *dev) +{ + synchronize_irq(dev->priv.eq_table->cmd_eq.irqn); +} + +/* Generic EQ API for mlx5_core consumers + * Needed For RDMA ODP EQ for now + */ +struct mlx5_eq * +mlx5_eq_create_generic(struct mlx5_core_dev *dev, const char *name, + struct mlx5_eq_param *param) +{ + struct mlx5_eq *eq = kvzalloc(sizeof(*eq), GFP_KERNEL); + int err; + + if (!eq) + return ERR_PTR(-ENOMEM); + + err = create_async_eq(dev, name, eq, param); + if (err) { + kvfree(eq); + eq = ERR_PTR(err); + } + + return eq; +} +EXPORT_SYMBOL(mlx5_eq_create_generic); + +int mlx5_eq_destroy_generic(struct mlx5_core_dev *dev, struct mlx5_eq *eq) +{ + int err; + + if (IS_ERR(eq)) + return -EINVAL; + + err = destroy_async_eq(dev, eq); + if (err) + goto out; + + kvfree(eq); +out: + return err; +} +EXPORT_SYMBOL(mlx5_eq_destroy_generic); + +struct mlx5_eqe *mlx5_eq_get_eqe(struct mlx5_eq *eq, u32 cc) +{ + u32 ci = eq->cons_index + cc; + struct mlx5_eqe *eqe; + + eqe = get_eqe(eq, ci & (eq->nent - 1)); + eqe = ((eqe->owner & 1) ^ !!(ci & eq->nent)) ? NULL : eqe; + /* Make sure we read EQ entry contents after we've + * checked the ownership bit. + */ + if (eqe) + dma_rmb(); + + return eqe; +} +EXPORT_SYMBOL(mlx5_eq_get_eqe); + +void mlx5_eq_update_ci(struct mlx5_eq *eq, u32 cc, bool arm) +{ + __be32 __iomem *addr = eq->doorbell + (arm ? 0 : 2); + u32 val; + + eq->cons_index += cc; + val = (eq->cons_index & 0xffffff) | (eq->eqn << 24); + + __raw_writel((__force u32)cpu_to_be32(val), addr); + /* We still want ordering, just not swabbing, so add a barrier */ + mb(); +} +EXPORT_SYMBOL(mlx5_eq_update_ci); + +/* Completion EQs */ + +static int set_comp_irq_affinity_hint(struct mlx5_core_dev *mdev, int i) +{ + struct mlx5_priv *priv = &mdev->priv; + int vecidx = MLX5_EQ_VEC_COMP_BASE + i; + int irq = pci_irq_vector(mdev->pdev, vecidx); + struct mlx5_irq_info *irq_info = &priv->eq_table->irq_info[vecidx]; + + if (!zalloc_cpumask_var(&irq_info->mask, GFP_KERNEL)) { + mlx5_core_warn(mdev, "zalloc_cpumask_var failed"); + return -ENOMEM; + } + + cpumask_set_cpu(cpumask_local_spread(i, priv->numa_node), + irq_info->mask); + + if (IS_ENABLED(CONFIG_SMP) && + irq_set_affinity_hint(irq, irq_info->mask)) + mlx5_core_warn(mdev, "irq_set_affinity_hint failed, irq 0x%.4x", irq); + + return 0; +} + +static void clear_comp_irq_affinity_hint(struct mlx5_core_dev *mdev, int i) +{ + int vecidx = MLX5_EQ_VEC_COMP_BASE + i; + struct mlx5_priv *priv = &mdev->priv; + int irq = pci_irq_vector(mdev->pdev, vecidx); + struct mlx5_irq_info *irq_info = &priv->eq_table->irq_info[vecidx]; + + irq_set_affinity_hint(irq, NULL); + free_cpumask_var(irq_info->mask); +} + +static int set_comp_irq_affinity_hints(struct mlx5_core_dev *mdev) +{ + int err; + int i; + + for (i = 0; i < mdev->priv.eq_table->num_comp_vectors; i++) { + err = set_comp_irq_affinity_hint(mdev, i); + if (err) + goto err_out; + } + + return 0; + +err_out: + for (i--; i >= 0; i--) + clear_comp_irq_affinity_hint(mdev, i); + + return err; +} + +static void clear_comp_irqs_affinity_hints(struct mlx5_core_dev *mdev) +{ + int i; + + for (i = 0; i < mdev->priv.eq_table->num_comp_vectors; i++) + clear_comp_irq_affinity_hint(mdev, i); +} + +static void destroy_comp_eqs(struct mlx5_core_dev *dev) +{ + struct mlx5_eq_table *table = dev->priv.eq_table; + struct mlx5_eq_comp *eq, *n; + + clear_comp_irqs_affinity_hints(dev); + +#ifdef CONFIG_RFS_ACCEL + if (table->rmap) { + free_irq_cpu_rmap(table->rmap); + table->rmap = NULL; + } +#endif + list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) { + list_del(&eq->list); + if (destroy_unmap_eq(dev, &eq->core)) + mlx5_core_warn(dev, "failed to destroy comp EQ 0x%x\n", + eq->core.eqn); + tasklet_disable(&eq->tasklet_ctx.task); + kfree(eq); + } +} + +static int create_comp_eqs(struct mlx5_core_dev *dev) +{ + struct mlx5_eq_table *table = dev->priv.eq_table; + char name[MLX5_MAX_IRQ_NAME]; + struct mlx5_eq_comp *eq; + int ncomp_vec; + int nent; + int err; + int i; + + INIT_LIST_HEAD(&table->comp_eqs_list); + ncomp_vec = table->num_comp_vectors; + nent = MLX5_COMP_EQ_SIZE; +#ifdef CONFIG_RFS_ACCEL + table->rmap = alloc_irq_cpu_rmap(ncomp_vec); + if (!table->rmap) + return -ENOMEM; +#endif + for (i = 0; i < ncomp_vec; i++) { + int vecidx = i + MLX5_EQ_VEC_COMP_BASE; + struct mlx5_eq_param param = {}; + + eq = kzalloc(sizeof(*eq), GFP_KERNEL); + if (!eq) { + err = -ENOMEM; + goto clean; + } + + INIT_LIST_HEAD(&eq->tasklet_ctx.list); + INIT_LIST_HEAD(&eq->tasklet_ctx.process_list); + spin_lock_init(&eq->tasklet_ctx.lock); + tasklet_init(&eq->tasklet_ctx.task, mlx5_cq_tasklet_cb, + (unsigned long)&eq->tasklet_ctx); + +#ifdef CONFIG_RFS_ACCEL + irq_cpu_rmap_add(table->rmap, pci_irq_vector(dev->pdev, vecidx)); +#endif + snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_comp%d", i); + param = (struct mlx5_eq_param) { + .index = vecidx, + .mask = 0, + .nent = nent, + .context = &eq->core, + .handler = mlx5_eq_comp_int + }; + err = create_map_eq(dev, &eq->core, name, ¶m); + if (err) { + kfree(eq); + goto clean; + } + mlx5_core_dbg(dev, "allocated completion EQN %d\n", eq->core.eqn); + /* add tail, to keep the list ordered, for mlx5_vector2eqn to work */ + list_add_tail(&eq->list, &table->comp_eqs_list); + } + + err = set_comp_irq_affinity_hints(dev); + if (err) { + mlx5_core_err(dev, "Failed to alloc affinity hint cpumask\n"); + goto clean; + } + + return 0; + +clean: + destroy_comp_eqs(dev); + return err; +} + +int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn, + unsigned int *irqn) +{ + struct mlx5_eq_table *table = dev->priv.eq_table; + struct mlx5_eq_comp *eq, *n; + int err = -ENOENT; + int i = 0; + + list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) { + if (i++ == vector) { + *eqn = eq->core.eqn; + *irqn = eq->core.irqn; + err = 0; + break; + } + } + + return err; +} +EXPORT_SYMBOL(mlx5_vector2eqn); + +unsigned int mlx5_comp_vectors_count(struct mlx5_core_dev *dev) +{ + return dev->priv.eq_table->num_comp_vectors; +} +EXPORT_SYMBOL(mlx5_comp_vectors_count); + +struct cpumask * +mlx5_comp_irq_get_affinity_mask(struct mlx5_core_dev *dev, int vector) +{ + /* TODO: consider irq_get_affinity_mask(irq) */ + return dev->priv.eq_table->irq_info[vector + MLX5_EQ_VEC_COMP_BASE].mask; +} +EXPORT_SYMBOL(mlx5_comp_irq_get_affinity_mask); + +struct cpu_rmap *mlx5_eq_table_get_rmap(struct mlx5_core_dev *dev) +{ +#ifdef CONFIG_RFS_ACCEL + return dev->priv.eq_table->rmap; +#else + return NULL; +#endif +} + +struct mlx5_eq_comp *mlx5_eqn2comp_eq(struct mlx5_core_dev *dev, int eqn) +{ + struct mlx5_eq_table *table = dev->priv.eq_table; + struct mlx5_eq_comp *eq; + + list_for_each_entry(eq, &table->comp_eqs_list, list) { + if (eq->core.eqn == eqn) + return eq; + } + + return ERR_PTR(-ENOENT); } /* This function should only be called after mlx5_cmd_force_teardown_hca */ void mlx5_core_eq_free_irqs(struct mlx5_core_dev *dev) { - struct mlx5_eq_table *table = &dev->priv.eq_table; - struct mlx5_eq *eq; + struct mlx5_eq_table *table = dev->priv.eq_table; + int i, max_eqs; + + clear_comp_irqs_affinity_hints(dev); #ifdef CONFIG_RFS_ACCEL - if (dev->rmap) { - free_irq_cpu_rmap(dev->rmap); - dev->rmap = NULL; + if (table->rmap) { + free_irq_cpu_rmap(table->rmap); + table->rmap = NULL; } #endif - list_for_each_entry(eq, &table->comp_eqs_list, list) - free_irq(eq->irqn, eq); - - free_irq(table->pages_eq.irqn, &table->pages_eq); - free_irq(table->async_eq.irqn, &table->async_eq); - free_irq(table->cmd_eq.irqn, &table->cmd_eq); -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - if (MLX5_CAP_GEN(dev, pg)) - free_irq(table->pfault_eq.irqn, &table->pfault_eq); -#endif + + mutex_lock(&table->lock); /* sync with create/destroy_async_eq */ + max_eqs = table->num_comp_vectors + MLX5_EQ_VEC_COMP_BASE; + for (i = max_eqs - 1; i >= 0; i--) { + if (!table->irq_info[i].context) + continue; + free_irq(pci_irq_vector(dev->pdev, i), table->irq_info[i].context); + table->irq_info[i].context = NULL; + } + mutex_unlock(&table->lock); + pci_free_irq_vectors(dev->pdev); +} + +static int alloc_irq_vectors(struct mlx5_core_dev *dev) +{ + struct mlx5_priv *priv = &dev->priv; + struct mlx5_eq_table *table = priv->eq_table; + int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ? + MLX5_CAP_GEN(dev, max_num_eqs) : + 1 << MLX5_CAP_GEN(dev, log_max_eq); + int nvec; + int err; + + nvec = MLX5_CAP_GEN(dev, num_ports) * num_online_cpus() + + MLX5_EQ_VEC_COMP_BASE; + nvec = min_t(int, nvec, num_eqs); + if (nvec <= MLX5_EQ_VEC_COMP_BASE) + return -ENOMEM; + + table->irq_info = kcalloc(nvec, sizeof(*table->irq_info), GFP_KERNEL); + if (!table->irq_info) + return -ENOMEM; + + nvec = pci_alloc_irq_vectors(dev->pdev, MLX5_EQ_VEC_COMP_BASE + 1, + nvec, PCI_IRQ_MSIX); + if (nvec < 0) { + err = nvec; + goto err_free_irq_info; + } + + table->num_comp_vectors = nvec - MLX5_EQ_VEC_COMP_BASE; + + return 0; + +err_free_irq_info: + kfree(table->irq_info); + return err; +} + +static void free_irq_vectors(struct mlx5_core_dev *dev) +{ + struct mlx5_priv *priv = &dev->priv; + pci_free_irq_vectors(dev->pdev); + kfree(priv->eq_table->irq_info); +} + +int mlx5_eq_table_create(struct mlx5_core_dev *dev) +{ + int err; + + err = alloc_irq_vectors(dev); + if (err) { + mlx5_core_err(dev, "alloc irq vectors failed\n"); + return err; + } + + err = create_async_eqs(dev); + if (err) { + mlx5_core_err(dev, "Failed to create async EQs\n"); + goto err_async_eqs; + } + + err = create_comp_eqs(dev); + if (err) { + mlx5_core_err(dev, "Failed to create completion EQs\n"); + goto err_comp_eqs; + } + + return 0; +err_comp_eqs: + destroy_async_eqs(dev); +err_async_eqs: + free_irq_vectors(dev); + return err; +} + +void mlx5_eq_table_destroy(struct mlx5_core_dev *dev) +{ + destroy_comp_eqs(dev); + destroy_async_eqs(dev); + free_irq_vectors(dev); +} + +int mlx5_eq_notifier_register(struct mlx5_core_dev *dev, struct mlx5_nb *nb) +{ + struct mlx5_eq_table *eqt = dev->priv.eq_table; + + if (nb->event_type >= MLX5_EVENT_TYPE_MAX) + return -EINVAL; + + return atomic_notifier_chain_register(&eqt->nh[nb->event_type], &nb->nb); +} + +int mlx5_eq_notifier_unregister(struct mlx5_core_dev *dev, struct mlx5_nb *nb) +{ + struct mlx5_eq_table *eqt = dev->priv.eq_table; + + if (nb->event_type >= MLX5_EVENT_TYPE_MAX) + return -EINVAL; + + return atomic_notifier_chain_unregister(&eqt->nh[nb->event_type], &nb->nb); } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c index d004957328f9..e6a9b19d8626 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.c @@ -36,8 +36,10 @@ #include <linux/mlx5/vport.h> #include <linux/mlx5/fs.h> #include "mlx5_core.h" +#include "lib/eq.h" #include "eswitch.h" #include "fs_core.h" +#include "lib/eq.h" #define UPLINK_VPORT 0xFFFF @@ -1567,7 +1569,6 @@ static void esw_disable_vport(struct mlx5_eswitch *esw, int vport_num) /* Mark this vport as disabled to discard new events */ vport->enabled = false; - synchronize_irq(pci_irq_vector(esw->dev->pdev, MLX5_EQ_VEC_ASYNC)); /* Wait for current already scheduled events to complete */ flush_workqueue(esw->work_queue); /* Disable events from this vport */ @@ -1593,10 +1594,25 @@ static void esw_disable_vport(struct mlx5_eswitch *esw, int vport_num) mutex_unlock(&esw->state_lock); } +static int eswitch_vport_event(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_eswitch *esw = mlx5_nb_cof(nb, struct mlx5_eswitch, nb); + struct mlx5_eqe *eqe = data; + struct mlx5_vport *vport; + u16 vport_num; + + vport_num = be16_to_cpu(eqe->data.vport_change.vport_num); + vport = &esw->vports[vport_num]; + if (vport->enabled) + queue_work(esw->work_queue, &vport->vport_change_handler); + + return NOTIFY_OK; +} + /* Public E-Switch API */ #define ESW_ALLOWED(esw) ((esw) && MLX5_ESWITCH_MANAGER((esw)->dev)) - int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode) { int err; @@ -1640,6 +1656,11 @@ int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode) for (i = 0; i <= nvfs; i++) esw_enable_vport(esw, i, enabled_events); + if (mode == SRIOV_LEGACY) { + MLX5_NB_INIT(&esw->nb, eswitch_vport_event, NIC_VPORT_CHANGE); + mlx5_eq_notifier_register(esw->dev, &esw->nb); + } + esw_info(esw->dev, "SRIOV enabled: active vports(%d)\n", esw->enabled_vports); return 0; @@ -1669,6 +1690,9 @@ void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw) mc_promisc = &esw->mc_promisc; nvports = esw->enabled_vports; + if (esw->mode == SRIOV_LEGACY) + mlx5_eq_notifier_unregister(esw->dev, &esw->nb); + for (i = 0; i < esw->total_vports; i++) esw_disable_vport(esw, i); @@ -1777,23 +1801,6 @@ void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw) kfree(esw); } -void mlx5_eswitch_vport_event(struct mlx5_eswitch *esw, struct mlx5_eqe *eqe) -{ - struct mlx5_eqe_vport_change *vc_eqe = &eqe->data.vport_change; - u16 vport_num = be16_to_cpu(vc_eqe->vport_num); - struct mlx5_vport *vport; - - if (!esw) { - pr_warn("MLX5 E-Switch: vport %d got an event while eswitch is not initialized\n", - vport_num); - return; - } - - vport = &esw->vports[vport_num]; - if (vport->enabled) - queue_work(esw->work_queue, &vport->vport_change_handler); -} - /* Vport Administration */ #define LEGAL_VPORT(esw, vport) (vport >= 0 && vport < esw->total_vports) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h index aaafc9f17115..480ffa294867 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/eswitch.h @@ -181,6 +181,7 @@ struct esw_mc_addr { /* SRIOV only */ struct mlx5_eswitch { struct mlx5_core_dev *dev; + struct mlx5_nb nb; struct mlx5_eswitch_fdb fdb_table; struct hlist_head mc_table[MLX5_L2_ADDR_HASH_SIZE]; struct workqueue_struct *work_queue; @@ -211,7 +212,6 @@ int esw_offloads_init_reps(struct mlx5_eswitch *esw); /* E-Switch API */ int mlx5_eswitch_init(struct mlx5_core_dev *dev); void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw); -void mlx5_eswitch_vport_event(struct mlx5_eswitch *esw, struct mlx5_eqe *eqe); int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode); void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw); int mlx5_eswitch_set_vport_mac(struct mlx5_eswitch *esw, @@ -352,7 +352,6 @@ static inline bool mlx5_eswitch_vlan_actions_supported(struct mlx5_core_dev *dev /* eswitch API stubs */ static inline int mlx5_eswitch_init(struct mlx5_core_dev *dev) { return 0; } static inline void mlx5_eswitch_cleanup(struct mlx5_eswitch *esw) {} -static inline void mlx5_eswitch_vport_event(struct mlx5_eswitch *esw, struct mlx5_eqe *eqe) {} static inline int mlx5_eswitch_enable_sriov(struct mlx5_eswitch *esw, int nvfs, int mode) { return 0; } static inline void mlx5_eswitch_disable_sriov(struct mlx5_eswitch *esw) {} diff --git a/drivers/net/ethernet/mellanox/mlx5/core/events.c b/drivers/net/ethernet/mellanox/mlx5/core/events.c new file mode 100644 index 000000000000..e92df7020a26 --- /dev/null +++ b/drivers/net/ethernet/mellanox/mlx5/core/events.c @@ -0,0 +1,298 @@ +// SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB +// Copyright (c) 2018 Mellanox Technologies + +#include <linux/mlx5/driver.h> + +#include "mlx5_core.h" +#include "lib/eq.h" +#include "lib/mlx5.h" + +struct mlx5_event_nb { + struct mlx5_nb nb; + void *ctx; +}; + +/* General events handlers for the low level mlx5_core driver + * + * Other Major feature specific events such as + * clock/eswitch/fpga/FW trace and many others, are handled elsewhere, with + * separate notifiers callbacks, specifically by those mlx5 components. + */ +static int any_notifier(struct notifier_block *, unsigned long, void *); +static int temp_warn(struct notifier_block *, unsigned long, void *); +static int port_module(struct notifier_block *, unsigned long, void *); + +/* handler which forwards the event to events->nh, driver notifiers */ +static int forward_event(struct notifier_block *, unsigned long, void *); + +static struct mlx5_nb events_nbs_ref[] = { + /* Events to be proccessed by mlx5_core */ + {.nb.notifier_call = any_notifier, .event_type = MLX5_EVENT_TYPE_NOTIFY_ANY }, + {.nb.notifier_call = temp_warn, .event_type = MLX5_EVENT_TYPE_TEMP_WARN_EVENT }, + {.nb.notifier_call = port_module, .event_type = MLX5_EVENT_TYPE_PORT_MODULE_EVENT }, + + /* Events to be forwarded (as is) to mlx5 core interfaces (mlx5e/mlx5_ib) */ + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_PORT_CHANGE }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_GENERAL_EVENT }, + /* QP/WQ resource events to forward */ + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_DCT_DRAINED }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_PATH_MIG }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_COMM_EST }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_SQ_DRAINED }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_SRQ_LAST_WQE }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_WQ_CATAS_ERROR }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_PATH_MIG_FAILED }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_WQ_ACCESS_ERROR }, + /* SRQ events */ + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_SRQ_CATAS_ERROR }, + {.nb.notifier_call = forward_event, .event_type = MLX5_EVENT_TYPE_SRQ_RQ_LIMIT }, +}; + +struct mlx5_events { + struct mlx5_core_dev *dev; + struct mlx5_event_nb notifiers[ARRAY_SIZE(events_nbs_ref)]; + /* driver notifier chain */ + struct atomic_notifier_head nh; + /* port module events stats */ + struct mlx5_pme_stats pme_stats; +}; + +static const char *eqe_type_str(u8 type) +{ + switch (type) { + case MLX5_EVENT_TYPE_COMP: + return "MLX5_EVENT_TYPE_COMP"; + case MLX5_EVENT_TYPE_PATH_MIG: + return "MLX5_EVENT_TYPE_PATH_MIG"; + case MLX5_EVENT_TYPE_COMM_EST: + return "MLX5_EVENT_TYPE_COMM_EST"; + case MLX5_EVENT_TYPE_SQ_DRAINED: + return "MLX5_EVENT_TYPE_SQ_DRAINED"; + case MLX5_EVENT_TYPE_SRQ_LAST_WQE: + return "MLX5_EVENT_TYPE_SRQ_LAST_WQE"; + case MLX5_EVENT_TYPE_SRQ_RQ_LIMIT: + return "MLX5_EVENT_TYPE_SRQ_RQ_LIMIT"; + case MLX5_EVENT_TYPE_CQ_ERROR: + return "MLX5_EVENT_TYPE_CQ_ERROR"; + case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: + return "MLX5_EVENT_TYPE_WQ_CATAS_ERROR"; + case MLX5_EVENT_TYPE_PATH_MIG_FAILED: + return "MLX5_EVENT_TYPE_PATH_MIG_FAILED"; + case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: + return "MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR"; + case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: + return "MLX5_EVENT_TYPE_WQ_ACCESS_ERROR"; + case MLX5_EVENT_TYPE_SRQ_CATAS_ERROR: + return "MLX5_EVENT_TYPE_SRQ_CATAS_ERROR"; + case MLX5_EVENT_TYPE_INTERNAL_ERROR: + return "MLX5_EVENT_TYPE_INTERNAL_ERROR"; + case MLX5_EVENT_TYPE_PORT_CHANGE: + return "MLX5_EVENT_TYPE_PORT_CHANGE"; + case MLX5_EVENT_TYPE_GPIO_EVENT: + return "MLX5_EVENT_TYPE_GPIO_EVENT"; + case MLX5_EVENT_TYPE_PORT_MODULE_EVENT: + return "MLX5_EVENT_TYPE_PORT_MODULE_EVENT"; + case MLX5_EVENT_TYPE_TEMP_WARN_EVENT: + return "MLX5_EVENT_TYPE_TEMP_WARN_EVENT"; + case MLX5_EVENT_TYPE_REMOTE_CONFIG: + return "MLX5_EVENT_TYPE_REMOTE_CONFIG"; + case MLX5_EVENT_TYPE_DB_BF_CONGESTION: + return "MLX5_EVENT_TYPE_DB_BF_CONGESTION"; + case MLX5_EVENT_TYPE_STALL_EVENT: + return "MLX5_EVENT_TYPE_STALL_EVENT"; + case MLX5_EVENT_TYPE_CMD: + return "MLX5_EVENT_TYPE_CMD"; + case MLX5_EVENT_TYPE_PAGE_REQUEST: + return "MLX5_EVENT_TYPE_PAGE_REQUEST"; + case MLX5_EVENT_TYPE_PAGE_FAULT: + return "MLX5_EVENT_TYPE_PAGE_FAULT"; + case MLX5_EVENT_TYPE_PPS_EVENT: + return "MLX5_EVENT_TYPE_PPS_EVENT"; + case MLX5_EVENT_TYPE_NIC_VPORT_CHANGE: + return "MLX5_EVENT_TYPE_NIC_VPORT_CHANGE"; + case MLX5_EVENT_TYPE_FPGA_ERROR: + return "MLX5_EVENT_TYPE_FPGA_ERROR"; + case MLX5_EVENT_TYPE_FPGA_QP_ERROR: + return "MLX5_EVENT_TYPE_FPGA_QP_ERROR"; + case MLX5_EVENT_TYPE_GENERAL_EVENT: + return "MLX5_EVENT_TYPE_GENERAL_EVENT"; + case MLX5_EVENT_TYPE_DEVICE_TRACER: + return "MLX5_EVENT_TYPE_DEVICE_TRACER"; + default: + return "Unrecognized event"; + } +} + +/* handles all FW events, type == eqe->type */ +static int any_notifier(struct notifier_block *nb, + unsigned long type, void *data) +{ + struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb); + struct mlx5_events *events = event_nb->ctx; + struct mlx5_eqe *eqe = data; + + mlx5_core_dbg(events->dev, "Async eqe type %s, subtype (%d)\n", + eqe_type_str(eqe->type), eqe->sub_type); + return NOTIFY_OK; +} + +/* type == MLX5_EVENT_TYPE_TEMP_WARN_EVENT */ +static int temp_warn(struct notifier_block *nb, unsigned long type, void *data) +{ + struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb); + struct mlx5_events *events = event_nb->ctx; + struct mlx5_eqe *eqe = data; + u64 value_lsb; + u64 value_msb; + + value_lsb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_lsb); + value_msb = be64_to_cpu(eqe->data.temp_warning.sensor_warning_msb); + + mlx5_core_warn(events->dev, + "High temperature on sensors with bit set %llx %llx", + value_msb, value_lsb); + + return NOTIFY_OK; +} + +/* MLX5_EVENT_TYPE_PORT_MODULE_EVENT */ +static const char *mlx5_pme_status[MLX5_MODULE_STATUS_NUM] = { + "Cable plugged", /* MLX5_MODULE_STATUS_PLUGGED = 0x1 */ + "Cable unplugged", /* MLX5_MODULE_STATUS_UNPLUGGED = 0x2 */ + "Cable error", /* MLX5_MODULE_STATUS_ERROR = 0x3 */ +}; + +static const char *mlx5_pme_error[MLX5_MODULE_EVENT_ERROR_NUM] = { + "Power budget exceeded", + "Long Range for non MLNX cable", + "Bus stuck(I2C or data shorted)", + "No EEPROM/retry timeout", + "Enforce part number list", + "Unknown identifier", + "High Temperature", + "Bad or shorted cable/module", + "Unknown status", +}; + +/* type == MLX5_EVENT_TYPE_PORT_MODULE_EVENT */ +static int port_module(struct notifier_block *nb, unsigned long type, void *data) +{ + struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb); + struct mlx5_events *events = event_nb->ctx; + struct mlx5_eqe *eqe = data; + + enum port_module_event_status_type module_status; + enum port_module_event_error_type error_type; + struct mlx5_eqe_port_module *module_event_eqe; + u8 module_num; + + module_event_eqe = &eqe->data.port_module; + module_num = module_event_eqe->module; + module_status = module_event_eqe->module_status & + PORT_MODULE_EVENT_MODULE_STATUS_MASK; + error_type = module_event_eqe->error_type & + PORT_MODULE_EVENT_ERROR_TYPE_MASK; + if (module_status < MLX5_MODULE_STATUS_ERROR) { + events->pme_stats.status_counters[module_status - 1]++; + } else if (module_status == MLX5_MODULE_STATUS_ERROR) { + if (error_type >= MLX5_MODULE_EVENT_ERROR_UNKNOWN) + /* Unknown error type */ + error_type = MLX5_MODULE_EVENT_ERROR_UNKNOWN; + events->pme_stats.error_counters[error_type]++; + } + + if (!printk_ratelimit()) + return NOTIFY_OK; + + if (module_status < MLX5_MODULE_STATUS_ERROR) + mlx5_core_info(events->dev, + "Port module event: module %u, %s\n", + module_num, mlx5_pme_status[module_status - 1]); + + else if (module_status == MLX5_MODULE_STATUS_ERROR) + mlx5_core_info(events->dev, + "Port module event[error]: module %u, %s, %s\n", + module_num, mlx5_pme_status[module_status - 1], + mlx5_pme_error[error_type]); + + return NOTIFY_OK; +} + +void mlx5_get_pme_stats(struct mlx5_core_dev *dev, struct mlx5_pme_stats *stats) +{ + *stats = dev->priv.events->pme_stats; +} + +/* forward event as is to registered interfaces (mlx5e/mlx5_ib) */ +static int forward_event(struct notifier_block *nb, unsigned long event, void *data) +{ + struct mlx5_event_nb *event_nb = mlx5_nb_cof(nb, struct mlx5_event_nb, nb); + struct mlx5_events *events = event_nb->ctx; + struct mlx5_eqe *eqe = data; + + mlx5_core_dbg(events->dev, "Async eqe type %s, subtype (%d) forward to interfaces\n", + eqe_type_str(eqe->type), eqe->sub_type); + atomic_notifier_call_chain(&events->nh, event, data); + return NOTIFY_OK; +} + +int mlx5_events_init(struct mlx5_core_dev *dev) +{ + struct mlx5_events *events = kzalloc(sizeof(*events), GFP_KERNEL); + + if (!events) + return -ENOMEM; + + ATOMIC_INIT_NOTIFIER_HEAD(&events->nh); + events->dev = dev; + dev->priv.events = events; + return 0; +} + +void mlx5_events_cleanup(struct mlx5_core_dev *dev) +{ + kvfree(dev->priv.events); +} + +void mlx5_events_start(struct mlx5_core_dev *dev) +{ + struct mlx5_events *events = dev->priv.events; + int i; + + for (i = 0; i < ARRAY_SIZE(events_nbs_ref); i++) { + events->notifiers[i].nb = events_nbs_ref[i]; + events->notifiers[i].ctx = events; + mlx5_eq_notifier_register(dev, &events->notifiers[i].nb); + } +} + +void mlx5_events_stop(struct mlx5_core_dev *dev) +{ + struct mlx5_events *events = dev->priv.events; + int i; + + for (i = ARRAY_SIZE(events_nbs_ref) - 1; i >= 0 ; i--) + mlx5_eq_notifier_unregister(dev, &events->notifiers[i].nb); +} + +int mlx5_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb) +{ + struct mlx5_events *events = dev->priv.events; + + return atomic_notifier_chain_register(&events->nh, nb); +} +EXPORT_SYMBOL(mlx5_notifier_register); + +int mlx5_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb) +{ + struct mlx5_events *events = dev->priv.events; + + return atomic_notifier_chain_unregister(&events->nh, nb); +} +EXPORT_SYMBOL(mlx5_notifier_unregister); + +int mlx5_notifier_call_chain(struct mlx5_events *events, unsigned int event, void *data) +{ + return atomic_notifier_call_chain(&events->nh, event, data); +} diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c index 436a8136f26f..27c5f6c7d36a 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.c @@ -36,6 +36,7 @@ #include "mlx5_core.h" #include "lib/mlx5.h" +#include "lib/eq.h" #include "fpga/core.h" #include "fpga/conn.h" @@ -145,6 +146,22 @@ static int mlx5_fpga_device_brb(struct mlx5_fpga_device *fdev) return 0; } +static int mlx5_fpga_event(struct mlx5_fpga_device *, unsigned long, void *); + +static int fpga_err_event(struct notifier_block *nb, unsigned long event, void *eqe) +{ + struct mlx5_fpga_device *fdev = mlx5_nb_cof(nb, struct mlx5_fpga_device, fpga_err_nb); + + return mlx5_fpga_event(fdev, event, eqe); +} + +static int fpga_qp_err_event(struct notifier_block *nb, unsigned long event, void *eqe) +{ + struct mlx5_fpga_device *fdev = mlx5_nb_cof(nb, struct mlx5_fpga_device, fpga_qp_err_nb); + + return mlx5_fpga_event(fdev, event, eqe); +} + int mlx5_fpga_device_start(struct mlx5_core_dev *mdev) { struct mlx5_fpga_device *fdev = mdev->fpga; @@ -185,6 +202,11 @@ int mlx5_fpga_device_start(struct mlx5_core_dev *mdev) if (err) goto out; + MLX5_NB_INIT(&fdev->fpga_err_nb, fpga_err_event, FPGA_ERROR); + MLX5_NB_INIT(&fdev->fpga_qp_err_nb, fpga_qp_err_event, FPGA_QP_ERROR); + mlx5_eq_notifier_register(fdev->mdev, &fdev->fpga_err_nb); + mlx5_eq_notifier_register(fdev->mdev, &fdev->fpga_qp_err_nb); + err = mlx5_fpga_conn_device_init(fdev); if (err) goto err_rsvd_gid; @@ -201,6 +223,8 @@ err_conn_init: mlx5_fpga_conn_device_cleanup(fdev); err_rsvd_gid: + mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_err_nb); + mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_qp_err_nb); mlx5_core_unreserve_gids(mdev, max_num_qps); out: spin_lock_irqsave(&fdev->state_lock, flags); @@ -256,6 +280,9 @@ void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev) } mlx5_fpga_conn_device_cleanup(fdev); + mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_err_nb); + mlx5_eq_notifier_unregister(fdev->mdev, &fdev->fpga_qp_err_nb); + max_num_qps = MLX5_CAP_FPGA(mdev, shell_caps.max_num_qps); mlx5_core_unreserve_gids(mdev, max_num_qps); } @@ -283,9 +310,10 @@ static const char *mlx5_fpga_qp_syndrome_to_string(u8 syndrome) return "Unknown"; } -void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data) +static int mlx5_fpga_event(struct mlx5_fpga_device *fdev, + unsigned long event, void *eqe) { - struct mlx5_fpga_device *fdev = mdev->fpga; + void *data = ((struct mlx5_eqe *)eqe)->data.raw; const char *event_name; bool teardown = false; unsigned long flags; @@ -303,9 +331,7 @@ void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data) fpga_qpn = MLX5_GET(fpga_qp_error_event, data, fpga_qpn); break; default: - mlx5_fpga_warn_ratelimited(fdev, "Unexpected event %u\n", - event); - return; + return NOTIFY_DONE; } spin_lock_irqsave(&fdev->state_lock, flags); @@ -326,4 +352,6 @@ void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data) */ if (teardown) mlx5_trigger_health_work(fdev->mdev); + + return NOTIFY_OK; } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h index 3e2355c8df3f..7e2e871dbf83 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/fpga/core.h @@ -35,11 +35,16 @@ #ifdef CONFIG_MLX5_FPGA +#include <linux/mlx5/eq.h> + +#include "lib/eq.h" #include "fpga/cmd.h" /* Represents an Innova device */ struct mlx5_fpga_device { struct mlx5_core_dev *mdev; + struct mlx5_nb fpga_err_nb; + struct mlx5_nb fpga_qp_err_nb; spinlock_t state_lock; /* Protects state transitions */ enum mlx5_fpga_status state; enum mlx5_fpga_image last_admin_image; @@ -82,7 +87,6 @@ int mlx5_fpga_init(struct mlx5_core_dev *mdev); void mlx5_fpga_cleanup(struct mlx5_core_dev *mdev); int mlx5_fpga_device_start(struct mlx5_core_dev *mdev); void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev); -void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, void *data); #else @@ -104,11 +108,6 @@ static inline void mlx5_fpga_device_stop(struct mlx5_core_dev *mdev) { } -static inline void mlx5_fpga_event(struct mlx5_core_dev *mdev, u8 event, - void *data) -{ -} - #endif #endif /* __MLX5_FPGA_CORE_H__ */ diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c index 43118de8ee99..196c07383082 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/health.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c @@ -38,6 +38,8 @@ #include <linux/mlx5/driver.h> #include <linux/mlx5/cmd.h> #include "mlx5_core.h" +#include "lib/eq.h" +#include "lib/mlx5.h" enum { MLX5_HEALTH_POLL_INTERVAL = 2 * HZ, @@ -78,29 +80,6 @@ void mlx5_set_nic_state(struct mlx5_core_dev *dev, u8 state) &dev->iseg->cmdq_addr_l_sz); } -static void trigger_cmd_completions(struct mlx5_core_dev *dev) -{ - unsigned long flags; - u64 vector; - - /* wait for pending handlers to complete */ - synchronize_irq(pci_irq_vector(dev->pdev, MLX5_EQ_VEC_CMD)); - spin_lock_irqsave(&dev->cmd.alloc_lock, flags); - vector = ~dev->cmd.bitmask & ((1ul << (1 << dev->cmd.log_sz)) - 1); - if (!vector) - goto no_trig; - - vector |= MLX5_TRIGGERED_CMD_COMP; - spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); - - mlx5_core_dbg(dev, "vector 0x%llx\n", vector); - mlx5_cmd_comp_handler(dev, vector, true); - return; - -no_trig: - spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); -} - static int in_fatal(struct mlx5_core_dev *dev) { struct mlx5_core_health *health = &dev->priv.health; @@ -124,10 +103,10 @@ void mlx5_enter_error_state(struct mlx5_core_dev *dev, bool force) mlx5_core_err(dev, "start\n"); if (pci_channel_offline(dev->pdev) || in_fatal(dev) || force) { dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR; - trigger_cmd_completions(dev); + mlx5_cmd_trigger_completions(dev); } - mlx5_core_event(dev, MLX5_DEV_EVENT_SYS_ERROR, 1); + mlx5_notifier_call_chain(dev->priv.events, MLX5_DEV_EVENT_SYS_ERROR, (void *)1); mlx5_core_err(dev, "end\n"); unlock: diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c index 0d90b1b4a3d3..d27c239e7d6c 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.c @@ -33,6 +33,7 @@ #include <linux/clocksource.h> #include <linux/highmem.h> #include <rdma/mlx5-abi.h> +#include "lib/eq.h" #include "en.h" #include "clock.h" @@ -439,16 +440,17 @@ static void mlx5_get_pps_caps(struct mlx5_core_dev *mdev) clock->pps_info.pin_caps[7] = MLX5_GET(mtpps_reg, out, cap_pin_7_mode); } -void mlx5_pps_event(struct mlx5_core_dev *mdev, - struct mlx5_eqe *eqe) +static int mlx5_pps_event(struct notifier_block *nb, + unsigned long type, void *data) { - struct mlx5_clock *clock = &mdev->clock; + struct mlx5_clock *clock = mlx5_nb_cof(nb, struct mlx5_clock, pps_nb); + struct mlx5_core_dev *mdev = clock->mdev; struct ptp_clock_event ptp_event; - struct timespec64 ts; - u64 nsec_now, nsec_delta; u64 cycles_now, cycles_delta; + u64 nsec_now, nsec_delta, ns; + struct mlx5_eqe *eqe = data; int pin = eqe->data.pps.pin; - s64 ns; + struct timespec64 ts; unsigned long flags; switch (clock->ptp_info.pin_config[pin].func) { @@ -463,6 +465,7 @@ void mlx5_pps_event(struct mlx5_core_dev *mdev, } else { ptp_event.type = PTP_CLOCK_EXTTS; } + /* TODOL clock->ptp can be NULL if ptp_clock_register failes */ ptp_clock_event(clock->ptp, &ptp_event); break; case PTP_PF_PEROUT: @@ -481,8 +484,11 @@ void mlx5_pps_event(struct mlx5_core_dev *mdev, write_sequnlock_irqrestore(&clock->lock, flags); break; default: - mlx5_core_err(mdev, " Unhandled event\n"); + mlx5_core_err(mdev, " Unhandled clock PPS event, func %d\n", + clock->ptp_info.pin_config[pin].func); } + + return NOTIFY_OK; } void mlx5_init_clock(struct mlx5_core_dev *mdev) @@ -567,6 +573,9 @@ void mlx5_init_clock(struct mlx5_core_dev *mdev) PTR_ERR(clock->ptp)); clock->ptp = NULL; } + + MLX5_NB_INIT(&clock->pps_nb, mlx5_pps_event, PPS_EVENT); + mlx5_eq_notifier_register(mdev, &clock->pps_nb); } void mlx5_cleanup_clock(struct mlx5_core_dev *mdev) @@ -576,6 +585,7 @@ void mlx5_cleanup_clock(struct mlx5_core_dev *mdev) if (!MLX5_CAP_GEN(mdev, device_frequency_khz)) return; + mlx5_eq_notifier_unregister(mdev, &clock->pps_nb); if (clock->ptp) { ptp_clock_unregister(clock->ptp); clock->ptp = NULL; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h index 263cb6e2aeee..31600924bdc3 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/clock.h @@ -36,7 +36,6 @@ #if IS_ENABLED(CONFIG_PTP_1588_CLOCK) void mlx5_init_clock(struct mlx5_core_dev *mdev); void mlx5_cleanup_clock(struct mlx5_core_dev *mdev); -void mlx5_pps_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe); static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev) { @@ -60,8 +59,6 @@ static inline ktime_t mlx5_timecounter_cyc2time(struct mlx5_clock *clock, #else static inline void mlx5_init_clock(struct mlx5_core_dev *mdev) {} static inline void mlx5_cleanup_clock(struct mlx5_core_dev *mdev) {} -static inline void mlx5_pps_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe) {} - static inline int mlx5_clock_get_ptp_index(struct mlx5_core_dev *mdev) { return -1; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h new file mode 100644 index 000000000000..c0fb6d72b695 --- /dev/null +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/eq.h @@ -0,0 +1,98 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* Copyright (c) 2018 Mellanox Technologies */ + +#ifndef __LIB_MLX5_EQ_H__ +#define __LIB_MLX5_EQ_H__ +#include <linux/mlx5/driver.h> +#include <linux/mlx5/eq.h> +#include <linux/mlx5/cq.h> + +#define MLX5_MAX_IRQ_NAME (32) +#define MLX5_EQE_SIZE (sizeof(struct mlx5_eqe)) + +struct mlx5_eq_tasklet { + struct list_head list; + struct list_head process_list; + struct tasklet_struct task; + spinlock_t lock; /* lock completion tasklet list */ +}; + +struct mlx5_cq_table { + spinlock_t lock; /* protect radix tree */ + struct radix_tree_root tree; +}; + +struct mlx5_eq { + struct mlx5_core_dev *dev; + struct mlx5_cq_table cq_table; + __be32 __iomem *doorbell; + u32 cons_index; + struct mlx5_frag_buf buf; + int size; + unsigned int vecidx; + unsigned int irqn; + u8 eqn; + int nent; + struct mlx5_rsc_debug *dbg; +}; + +struct mlx5_eq_comp { + struct mlx5_eq core; /* Must be first */ + struct mlx5_eq_tasklet tasklet_ctx; + struct list_head list; +}; + +static inline struct mlx5_eqe *get_eqe(struct mlx5_eq *eq, u32 entry) +{ + return mlx5_buf_offset(&eq->buf, entry * MLX5_EQE_SIZE); +} + +static inline struct mlx5_eqe *next_eqe_sw(struct mlx5_eq *eq) +{ + struct mlx5_eqe *eqe = get_eqe(eq, eq->cons_index & (eq->nent - 1)); + + return ((eqe->owner & 1) ^ !!(eq->cons_index & eq->nent)) ? NULL : eqe; +} + +static inline void eq_update_ci(struct mlx5_eq *eq, int arm) +{ + __be32 __iomem *addr = eq->doorbell + (arm ? 0 : 2); + u32 val = (eq->cons_index & 0xffffff) | (eq->eqn << 24); + + __raw_writel((__force u32)cpu_to_be32(val), addr); + /* We still want ordering, just not swabbing, so add a barrier */ + mb(); +} + +int mlx5_eq_table_init(struct mlx5_core_dev *dev); +void mlx5_eq_table_cleanup(struct mlx5_core_dev *dev); +int mlx5_eq_table_create(struct mlx5_core_dev *dev); +void mlx5_eq_table_destroy(struct mlx5_core_dev *dev); + +int mlx5_eq_add_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq); +int mlx5_eq_del_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq); +struct mlx5_eq_comp *mlx5_eqn2comp_eq(struct mlx5_core_dev *dev, int eqn); +struct mlx5_eq *mlx5_get_async_eq(struct mlx5_core_dev *dev); +void mlx5_cq_tasklet_cb(unsigned long data); +struct cpumask *mlx5_eq_comp_cpumask(struct mlx5_core_dev *dev, int ix); + +u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq_comp *eq); +void mlx5_eq_synchronize_async_irq(struct mlx5_core_dev *dev); +void mlx5_eq_synchronize_cmd_irq(struct mlx5_core_dev *dev); + +int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq); +void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq); +int mlx5_eq_debugfs_init(struct mlx5_core_dev *dev); +void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev); + +/* This function should only be called after mlx5_cmd_force_teardown_hca */ +void mlx5_core_eq_free_irqs(struct mlx5_core_dev *dev); + +#ifdef CONFIG_RFS_ACCEL +struct cpu_rmap *mlx5_eq_table_get_rmap(struct mlx5_core_dev *dev); +#endif + +int mlx5_eq_notifier_register(struct mlx5_core_dev *dev, struct mlx5_nb *nb); +int mlx5_eq_notifier_unregister(struct mlx5_core_dev *dev, struct mlx5_nb *nb); + +#endif diff --git a/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h b/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h index 7550b1cc8c6a..4d78a459676e 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/lib/mlx5.h @@ -33,6 +33,8 @@ #ifndef __LIB_MLX5_H__ #define __LIB_MLX5_H__ +#include "mlx5_core.h" + void mlx5_init_reserved_gids(struct mlx5_core_dev *dev); void mlx5_cleanup_reserved_gids(struct mlx5_core_dev *dev); int mlx5_core_reserve_gids(struct mlx5_core_dev *dev, unsigned int count); @@ -40,4 +42,37 @@ void mlx5_core_unreserve_gids(struct mlx5_core_dev *dev, unsigned int count); int mlx5_core_reserved_gid_alloc(struct mlx5_core_dev *dev, int *gid_index); void mlx5_core_reserved_gid_free(struct mlx5_core_dev *dev, int gid_index); +/* TODO move to lib/events.h */ + +#define PORT_MODULE_EVENT_MODULE_STATUS_MASK 0xF +#define PORT_MODULE_EVENT_ERROR_TYPE_MASK 0xF + +enum port_module_event_status_type { + MLX5_MODULE_STATUS_PLUGGED = 0x1, + MLX5_MODULE_STATUS_UNPLUGGED = 0x2, + MLX5_MODULE_STATUS_ERROR = 0x3, + MLX5_MODULE_STATUS_NUM = 0x3, +}; + +enum port_module_event_error_type { + MLX5_MODULE_EVENT_ERROR_POWER_BUDGET_EXCEEDED, + MLX5_MODULE_EVENT_ERROR_LONG_RANGE_FOR_NON_MLNX_CABLE_MODULE, + MLX5_MODULE_EVENT_ERROR_BUS_STUCK, + MLX5_MODULE_EVENT_ERROR_NO_EEPROM_RETRY_TIMEOUT, + MLX5_MODULE_EVENT_ERROR_ENFORCE_PART_NUMBER_LIST, + MLX5_MODULE_EVENT_ERROR_UNKNOWN_IDENTIFIER, + MLX5_MODULE_EVENT_ERROR_HIGH_TEMPERATURE, + MLX5_MODULE_EVENT_ERROR_BAD_CABLE, + MLX5_MODULE_EVENT_ERROR_UNKNOWN, + MLX5_MODULE_EVENT_ERROR_NUM, +}; + +struct mlx5_pme_stats { + u64 status_counters[MLX5_MODULE_STATUS_NUM]; + u64 error_counters[MLX5_MODULE_EVENT_ERROR_NUM]; +}; + +void mlx5_get_pme_stats(struct mlx5_core_dev *dev, struct mlx5_pme_stats *stats); +int mlx5_notifier_call_chain(struct mlx5_events *events, unsigned int event, void *data); + #endif diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c index 28132c7dc05f..778995573812 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c @@ -43,7 +43,6 @@ #include <linux/mlx5/driver.h> #include <linux/mlx5/cq.h> #include <linux/mlx5/qp.h> -#include <linux/mlx5/srq.h> #include <linux/debugfs.h> #include <linux/kmod.h> #include <linux/mlx5/mlx5_ifc.h> @@ -53,6 +52,7 @@ #endif #include <net/devlink.h> #include "mlx5_core.h" +#include "lib/eq.h" #include "fs_core.h" #include "lib/mpfs.h" #include "eswitch.h" @@ -319,51 +319,6 @@ static void release_bar(struct pci_dev *pdev) pci_release_regions(pdev); } -static int mlx5_alloc_irq_vectors(struct mlx5_core_dev *dev) -{ - struct mlx5_priv *priv = &dev->priv; - struct mlx5_eq_table *table = &priv->eq_table; - int num_eqs = MLX5_CAP_GEN(dev, max_num_eqs) ? - MLX5_CAP_GEN(dev, max_num_eqs) : - 1 << MLX5_CAP_GEN(dev, log_max_eq); - int nvec; - int err; - - nvec = MLX5_CAP_GEN(dev, num_ports) * num_online_cpus() + - MLX5_EQ_VEC_COMP_BASE; - nvec = min_t(int, nvec, num_eqs); - if (nvec <= MLX5_EQ_VEC_COMP_BASE) - return -ENOMEM; - - priv->irq_info = kcalloc(nvec, sizeof(*priv->irq_info), GFP_KERNEL); - if (!priv->irq_info) - return -ENOMEM; - - nvec = pci_alloc_irq_vectors(dev->pdev, - MLX5_EQ_VEC_COMP_BASE + 1, nvec, - PCI_IRQ_MSIX); - if (nvec < 0) { - err = nvec; - goto err_free_irq_info; - } - - table->num_comp_vectors = nvec - MLX5_EQ_VEC_COMP_BASE; - - return 0; - -err_free_irq_info: - kfree(priv->irq_info); - return err; -} - -static void mlx5_free_irq_vectors(struct mlx5_core_dev *dev) -{ - struct mlx5_priv *priv = &dev->priv; - - pci_free_irq_vectors(dev->pdev); - kfree(priv->irq_info); -} - struct mlx5_reg_host_endianness { u8 he; u8 rsvd[15]; @@ -637,177 +592,6 @@ u64 mlx5_read_internal_timer(struct mlx5_core_dev *dev) return (u64)timer_l | (u64)timer_h1 << 32; } -static int mlx5_irq_set_affinity_hint(struct mlx5_core_dev *mdev, int i) -{ - struct mlx5_priv *priv = &mdev->priv; - int irq = pci_irq_vector(mdev->pdev, MLX5_EQ_VEC_COMP_BASE + i); - - if (!zalloc_cpumask_var(&priv->irq_info[i].mask, GFP_KERNEL)) { - mlx5_core_warn(mdev, "zalloc_cpumask_var failed"); - return -ENOMEM; - } - - cpumask_set_cpu(cpumask_local_spread(i, priv->numa_node), - priv->irq_info[i].mask); - - if (IS_ENABLED(CONFIG_SMP) && - irq_set_affinity_hint(irq, priv->irq_info[i].mask)) - mlx5_core_warn(mdev, "irq_set_affinity_hint failed, irq 0x%.4x", irq); - - return 0; -} - -static void mlx5_irq_clear_affinity_hint(struct mlx5_core_dev *mdev, int i) -{ - struct mlx5_priv *priv = &mdev->priv; - int irq = pci_irq_vector(mdev->pdev, MLX5_EQ_VEC_COMP_BASE + i); - - irq_set_affinity_hint(irq, NULL); - free_cpumask_var(priv->irq_info[i].mask); -} - -static int mlx5_irq_set_affinity_hints(struct mlx5_core_dev *mdev) -{ - int err; - int i; - - for (i = 0; i < mdev->priv.eq_table.num_comp_vectors; i++) { - err = mlx5_irq_set_affinity_hint(mdev, i); - if (err) - goto err_out; - } - - return 0; - -err_out: - for (i--; i >= 0; i--) - mlx5_irq_clear_affinity_hint(mdev, i); - - return err; -} - -static void mlx5_irq_clear_affinity_hints(struct mlx5_core_dev *mdev) -{ - int i; - - for (i = 0; i < mdev->priv.eq_table.num_comp_vectors; i++) - mlx5_irq_clear_affinity_hint(mdev, i); -} - -int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn, - unsigned int *irqn) -{ - struct mlx5_eq_table *table = &dev->priv.eq_table; - struct mlx5_eq *eq, *n; - int err = -ENOENT; - - spin_lock(&table->lock); - list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) { - if (eq->index == vector) { - *eqn = eq->eqn; - *irqn = eq->irqn; - err = 0; - break; - } - } - spin_unlock(&table->lock); - - return err; -} -EXPORT_SYMBOL(mlx5_vector2eqn); - -struct mlx5_eq *mlx5_eqn2eq(struct mlx5_core_dev *dev, int eqn) -{ - struct mlx5_eq_table *table = &dev->priv.eq_table; - struct mlx5_eq *eq; - - spin_lock(&table->lock); - list_for_each_entry(eq, &table->comp_eqs_list, list) - if (eq->eqn == eqn) { - spin_unlock(&table->lock); - return eq; - } - - spin_unlock(&table->lock); - - return ERR_PTR(-ENOENT); -} - -static void free_comp_eqs(struct mlx5_core_dev *dev) -{ - struct mlx5_eq_table *table = &dev->priv.eq_table; - struct mlx5_eq *eq, *n; - -#ifdef CONFIG_RFS_ACCEL - if (dev->rmap) { - free_irq_cpu_rmap(dev->rmap); - dev->rmap = NULL; - } -#endif - spin_lock(&table->lock); - list_for_each_entry_safe(eq, n, &table->comp_eqs_list, list) { - list_del(&eq->list); - spin_unlock(&table->lock); - if (mlx5_destroy_unmap_eq(dev, eq)) - mlx5_core_warn(dev, "failed to destroy EQ 0x%x\n", - eq->eqn); - kfree(eq); - spin_lock(&table->lock); - } - spin_unlock(&table->lock); -} - -static int alloc_comp_eqs(struct mlx5_core_dev *dev) -{ - struct mlx5_eq_table *table = &dev->priv.eq_table; - char name[MLX5_MAX_IRQ_NAME]; - struct mlx5_eq *eq; - int ncomp_vec; - int nent; - int err; - int i; - - INIT_LIST_HEAD(&table->comp_eqs_list); - ncomp_vec = table->num_comp_vectors; - nent = MLX5_COMP_EQ_SIZE; -#ifdef CONFIG_RFS_ACCEL - dev->rmap = alloc_irq_cpu_rmap(ncomp_vec); - if (!dev->rmap) - return -ENOMEM; -#endif - for (i = 0; i < ncomp_vec; i++) { - eq = kzalloc(sizeof(*eq), GFP_KERNEL); - if (!eq) { - err = -ENOMEM; - goto clean; - } - -#ifdef CONFIG_RFS_ACCEL - irq_cpu_rmap_add(dev->rmap, pci_irq_vector(dev->pdev, - MLX5_EQ_VEC_COMP_BASE + i)); -#endif - snprintf(name, MLX5_MAX_IRQ_NAME, "mlx5_comp%d", i); - err = mlx5_create_map_eq(dev, eq, - i + MLX5_EQ_VEC_COMP_BASE, nent, 0, - name, MLX5_EQ_TYPE_COMP); - if (err) { - kfree(eq); - goto clean; - } - mlx5_core_dbg(dev, "allocated completion EQN %d\n", eq->eqn); - eq->index = i; - spin_lock(&table->lock); - list_add_tail(&eq->list, &table->comp_eqs_list); - spin_unlock(&table->lock); - } - - return 0; - -clean: - free_comp_eqs(dev); - return err; -} - static int mlx5_core_set_issi(struct mlx5_core_dev *dev) { u32 query_in[MLX5_ST_SZ_DW(query_issi_in)] = {0}; @@ -944,22 +728,26 @@ static int mlx5_init_once(struct mlx5_core_dev *dev, struct mlx5_priv *priv) goto out; } - err = mlx5_eq_init(dev); + err = mlx5_eq_table_init(dev); if (err) { dev_err(&pdev->dev, "failed to initialize eq\n"); goto out; } + err = mlx5_events_init(dev); + if (err) { + dev_err(&pdev->dev, "failed to initialize events\n"); + goto err_eq_cleanup; + } + err = mlx5_cq_debugfs_init(dev); if (err) { dev_err(&pdev->dev, "failed to initialize cq debugfs\n"); - goto err_eq_cleanup; + goto err_events_cleanup; } mlx5_init_qp_table(dev); - mlx5_init_srq_table(dev); - mlx5_init_mkey_table(dev); mlx5_init_reserved_gids(dev); @@ -1013,12 +801,12 @@ err_rl_cleanup: err_tables_cleanup: mlx5_vxlan_destroy(dev->vxlan); mlx5_cleanup_mkey_table(dev); - mlx5_cleanup_srq_table(dev); mlx5_cleanup_qp_table(dev); mlx5_cq_debugfs_cleanup(dev); - +err_events_cleanup: + mlx5_events_cleanup(dev); err_eq_cleanup: - mlx5_eq_cleanup(dev); + mlx5_eq_table_cleanup(dev); out: return err; @@ -1036,10 +824,10 @@ static void mlx5_cleanup_once(struct mlx5_core_dev *dev) mlx5_cleanup_clock(dev); mlx5_cleanup_reserved_gids(dev); mlx5_cleanup_mkey_table(dev); - mlx5_cleanup_srq_table(dev); mlx5_cleanup_qp_table(dev); mlx5_cq_debugfs_cleanup(dev); - mlx5_eq_cleanup(dev); + mlx5_events_cleanup(dev); + mlx5_eq_table_cleanup(dev); } static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, @@ -1131,16 +919,10 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, goto reclaim_boot_pages; } - err = mlx5_pagealloc_start(dev); - if (err) { - dev_err(&pdev->dev, "mlx5_pagealloc_start failed\n"); - goto reclaim_boot_pages; - } - err = mlx5_cmd_init_hca(dev, sw_owner_id); if (err) { dev_err(&pdev->dev, "init hca failed\n"); - goto err_pagealloc_stop; + goto reclaim_boot_pages; } mlx5_set_driver_version(dev); @@ -1161,23 +943,20 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, } } - err = mlx5_alloc_irq_vectors(dev); - if (err) { - dev_err(&pdev->dev, "alloc irq vectors failed\n"); - goto err_cleanup_once; - } - dev->priv.uar = mlx5_get_uars_page(dev); if (IS_ERR(dev->priv.uar)) { dev_err(&pdev->dev, "Failed allocating uar, aborting\n"); err = PTR_ERR(dev->priv.uar); - goto err_disable_msix; + goto err_get_uars; } - err = mlx5_start_eqs(dev); + mlx5_events_start(dev); + mlx5_pagealloc_start(dev); + + err = mlx5_eq_table_create(dev); if (err) { - dev_err(&pdev->dev, "Failed to start pages and async EQs\n"); - goto err_put_uars; + dev_err(&pdev->dev, "Failed to create EQs\n"); + goto err_eq_table; } err = mlx5_fw_tracer_init(dev->tracer); @@ -1186,18 +965,6 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, goto err_fw_tracer; } - err = alloc_comp_eqs(dev); - if (err) { - dev_err(&pdev->dev, "Failed to alloc completion EQs\n"); - goto err_comp_eqs; - } - - err = mlx5_irq_set_affinity_hints(dev); - if (err) { - dev_err(&pdev->dev, "Failed to alloc affinity hint cpumask\n"); - goto err_affinity_hints; - } - err = mlx5_fpga_device_start(dev); if (err) { dev_err(&pdev->dev, "fpga device start failed %d\n", err); @@ -1266,24 +1033,17 @@ err_ipsec_start: mlx5_fpga_device_stop(dev); err_fpga_start: - mlx5_irq_clear_affinity_hints(dev); - -err_affinity_hints: - free_comp_eqs(dev); - -err_comp_eqs: mlx5_fw_tracer_cleanup(dev->tracer); err_fw_tracer: - mlx5_stop_eqs(dev); + mlx5_eq_table_destroy(dev); -err_put_uars: +err_eq_table: + mlx5_pagealloc_stop(dev); + mlx5_events_stop(dev); mlx5_put_uars_page(dev, priv->uar); -err_disable_msix: - mlx5_free_irq_vectors(dev); - -err_cleanup_once: +err_get_uars: if (boot) mlx5_cleanup_once(dev); @@ -1294,9 +1054,6 @@ err_stop_poll: goto out_err; } -err_pagealloc_stop: - mlx5_pagealloc_stop(dev); - reclaim_boot_pages: mlx5_reclaim_startup_pages(dev); @@ -1340,21 +1097,20 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, mlx5_accel_ipsec_cleanup(dev); mlx5_accel_tls_cleanup(dev); mlx5_fpga_device_stop(dev); - mlx5_irq_clear_affinity_hints(dev); - free_comp_eqs(dev); mlx5_fw_tracer_cleanup(dev->tracer); - mlx5_stop_eqs(dev); + mlx5_eq_table_destroy(dev); + mlx5_pagealloc_stop(dev); + mlx5_events_stop(dev); mlx5_put_uars_page(dev, priv->uar); - mlx5_free_irq_vectors(dev); if (cleanup) mlx5_cleanup_once(dev); mlx5_stop_health_poll(dev, cleanup); + err = mlx5_cmd_teardown_hca(dev); if (err) { dev_err(&dev->pdev->dev, "tear_down_hca failed, skip cleanup\n"); goto out; } - mlx5_pagealloc_stop(dev); mlx5_reclaim_startup_pages(dev); mlx5_core_disable_hca(dev, 0); mlx5_cmd_cleanup(dev); @@ -1364,12 +1120,6 @@ out: return err; } -struct mlx5_core_event_handler { - void (*event)(struct mlx5_core_dev *dev, - enum mlx5_dev_event event, - void *data); -}; - static const struct devlink_ops mlx5_devlink_ops = { #ifdef CONFIG_MLX5_ESWITCH .eswitch_mode_set = mlx5_devlink_eswitch_mode_set, @@ -1403,7 +1153,6 @@ static int init_one(struct pci_dev *pdev, pci_set_drvdata(pdev, dev); dev->pdev = pdev; - dev->event = mlx5_core_event; dev->profile = &profile[prof_sel]; INIT_LIST_HEAD(&priv->ctx_list); @@ -1411,17 +1160,6 @@ static int init_one(struct pci_dev *pdev, mutex_init(&dev->pci_status_mutex); mutex_init(&dev->intf_state_mutex); - INIT_LIST_HEAD(&priv->waiting_events_list); - priv->is_accum_events = false; - -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - err = init_srcu_struct(&priv->pfault_srcu); - if (err) { - dev_err(&pdev->dev, "init_srcu_struct failed with error code %d\n", - err); - goto clean_dev; - } -#endif mutex_init(&priv->bfregs.reg_head.lock); mutex_init(&priv->bfregs.wc_head.lock); INIT_LIST_HEAD(&priv->bfregs.reg_head.list); @@ -1430,7 +1168,7 @@ static int init_one(struct pci_dev *pdev, err = mlx5_pci_init(dev, priv); if (err) { dev_err(&pdev->dev, "mlx5_pci_init failed with error code %d\n", err); - goto clean_srcu; + goto clean_dev; } err = mlx5_health_init(dev); @@ -1439,12 +1177,14 @@ static int init_one(struct pci_dev *pdev, goto close_pci; } - mlx5_pagealloc_init(dev); + err = mlx5_pagealloc_init(dev); + if (err) + goto err_pagealloc_init; err = mlx5_load_one(dev, priv, true); if (err) { dev_err(&pdev->dev, "mlx5_load_one failed with error code %d\n", err); - goto clean_health; + goto err_load_one; } request_module_nowait(MLX5_IB_MOD); @@ -1458,16 +1198,13 @@ static int init_one(struct pci_dev *pdev, clean_load: mlx5_unload_one(dev, priv, true); -clean_health: +err_load_one: mlx5_pagealloc_cleanup(dev); +err_pagealloc_init: mlx5_health_cleanup(dev); close_pci: mlx5_pci_close(dev, priv); -clean_srcu: -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - cleanup_srcu_struct(&priv->pfault_srcu); clean_dev: -#endif devlink_free(devlink); return err; @@ -1491,9 +1228,6 @@ static void remove_one(struct pci_dev *pdev) mlx5_pagealloc_cleanup(dev); mlx5_health_cleanup(dev); mlx5_pci_close(dev, priv); -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - cleanup_srcu_struct(&priv->pfault_srcu); -#endif devlink_free(devlink); } @@ -1637,7 +1371,6 @@ succeed: * kexec. There is no need to cleanup the mlx5_core software * contexts. */ - mlx5_irq_clear_affinity_hints(dev); mlx5_core_eq_free_irqs(dev); return 0; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h index 0594d0961cb3..fd3141a4b3f1 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h @@ -78,6 +78,11 @@ do { \ __func__, __LINE__, current->pid, \ ##__VA_ARGS__) +#define mlx5_core_warn_once(__dev, format, ...) \ + dev_warn_once(&(__dev)->pdev->dev, "%s:%d:(pid %d): " format, \ + __func__, __LINE__, current->pid, \ + ##__VA_ARGS__) + #define mlx5_core_info(__dev, format, ...) \ dev_info(&(__dev)->pdev->dev, format, ##__VA_ARGS__) @@ -97,12 +102,6 @@ int mlx5_cmd_init_hca(struct mlx5_core_dev *dev, uint32_t *sw_owner_id); int mlx5_cmd_teardown_hca(struct mlx5_core_dev *dev); int mlx5_cmd_force_teardown_hca(struct mlx5_core_dev *dev); int mlx5_cmd_fast_teardown_hca(struct mlx5_core_dev *dev); - -void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event, - unsigned long param); -void mlx5_core_page_fault(struct mlx5_core_dev *dev, - struct mlx5_pagefault *pfault); -void mlx5_port_module_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe); void mlx5_enter_error_state(struct mlx5_core_dev *dev, bool force); void mlx5_disable_device(struct mlx5_core_dev *dev); void mlx5_recover_device(struct mlx5_core_dev *dev); @@ -124,28 +123,7 @@ int mlx5_destroy_scheduling_element_cmd(struct mlx5_core_dev *dev, u8 hierarchy, int mlx5_wait_for_vf_pages(struct mlx5_core_dev *dev); u64 mlx5_read_internal_timer(struct mlx5_core_dev *dev); -int mlx5_eq_init(struct mlx5_core_dev *dev); -void mlx5_eq_cleanup(struct mlx5_core_dev *dev); -int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx, - int nent, u64 mask, const char *name, - enum mlx5_eq_type type); -int mlx5_destroy_unmap_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq); -int mlx5_eq_add_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq); -int mlx5_eq_del_cq(struct mlx5_eq *eq, struct mlx5_core_cq *cq); -int mlx5_core_eq_query(struct mlx5_core_dev *dev, struct mlx5_eq *eq, - u32 *out, int outlen); -int mlx5_start_eqs(struct mlx5_core_dev *dev); -void mlx5_stop_eqs(struct mlx5_core_dev *dev); -/* This function should only be called after mlx5_cmd_force_teardown_hca */ -void mlx5_core_eq_free_irqs(struct mlx5_core_dev *dev); -struct mlx5_eq *mlx5_eqn2eq(struct mlx5_core_dev *dev, int eqn); -u32 mlx5_eq_poll_irq_disabled(struct mlx5_eq *eq); -void mlx5_cq_tasklet_cb(unsigned long data); -void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec, bool forced); -int mlx5_debug_eq_add(struct mlx5_core_dev *dev, struct mlx5_eq *eq); -void mlx5_debug_eq_remove(struct mlx5_core_dev *dev, struct mlx5_eq *eq); -int mlx5_eq_debugfs_init(struct mlx5_core_dev *dev); -void mlx5_eq_debugfs_cleanup(struct mlx5_core_dev *dev); +void mlx5_cmd_trigger_completions(struct mlx5_core_dev *dev); int mlx5_cq_debugfs_init(struct mlx5_core_dev *dev); void mlx5_cq_debugfs_cleanup(struct mlx5_core_dev *dev); @@ -159,6 +137,11 @@ int mlx5_query_qcam_reg(struct mlx5_core_dev *mdev, u32 *qcam, void mlx5_lag_add(struct mlx5_core_dev *dev, struct net_device *netdev); void mlx5_lag_remove(struct mlx5_core_dev *dev); +int mlx5_events_init(struct mlx5_core_dev *dev); +void mlx5_events_cleanup(struct mlx5_core_dev *dev); +void mlx5_events_start(struct mlx5_core_dev *dev); +void mlx5_events_stop(struct mlx5_core_dev *dev); + void mlx5_add_device(struct mlx5_interface *intf, struct mlx5_priv *priv); void mlx5_remove_device(struct mlx5_interface *intf, struct mlx5_priv *priv); void mlx5_attach_device(struct mlx5_core_dev *dev); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c index e36d3e3675f9..a83b517b0714 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c @@ -37,6 +37,7 @@ #include <linux/mlx5/driver.h> #include <linux/mlx5/cmd.h> #include "mlx5_core.h" +#include "lib/eq.h" enum { MLX5_PAGES_CANT_GIVE = 0, @@ -433,15 +434,28 @@ static void pages_work_handler(struct work_struct *work) kfree(req); } -void mlx5_core_req_pages_handler(struct mlx5_core_dev *dev, u16 func_id, - s32 npages) +static int req_pages_handler(struct notifier_block *nb, + unsigned long type, void *data) { struct mlx5_pages_req *req; - + struct mlx5_core_dev *dev; + struct mlx5_priv *priv; + struct mlx5_eqe *eqe; + u16 func_id; + s32 npages; + + priv = mlx5_nb_cof(nb, struct mlx5_priv, pg_nb); + dev = container_of(priv, struct mlx5_core_dev, priv); + eqe = data; + + func_id = be16_to_cpu(eqe->data.req_pages.func_id); + npages = be32_to_cpu(eqe->data.req_pages.num_pages); + mlx5_core_dbg(dev, "page request for func 0x%x, npages %d\n", + func_id, npages); req = kzalloc(sizeof(*req), GFP_ATOMIC); if (!req) { mlx5_core_warn(dev, "failed to allocate pages request\n"); - return; + return NOTIFY_DONE; } req->dev = dev; @@ -449,6 +463,7 @@ void mlx5_core_req_pages_handler(struct mlx5_core_dev *dev, u16 func_id, req->npages = npages; INIT_WORK(&req->work, pages_work_handler); queue_work(dev->priv.pg_wq, &req->work); + return NOTIFY_OK; } int mlx5_satisfy_startup_pages(struct mlx5_core_dev *dev, int boot) @@ -524,29 +539,32 @@ int mlx5_reclaim_startup_pages(struct mlx5_core_dev *dev) return 0; } -void mlx5_pagealloc_init(struct mlx5_core_dev *dev) +int mlx5_pagealloc_init(struct mlx5_core_dev *dev) { dev->priv.page_root = RB_ROOT; INIT_LIST_HEAD(&dev->priv.free_list); + dev->priv.pg_wq = create_singlethread_workqueue("mlx5_page_allocator"); + if (!dev->priv.pg_wq) + return -ENOMEM; + + return 0; } void mlx5_pagealloc_cleanup(struct mlx5_core_dev *dev) { - /* nothing */ + destroy_workqueue(dev->priv.pg_wq); } -int mlx5_pagealloc_start(struct mlx5_core_dev *dev) +void mlx5_pagealloc_start(struct mlx5_core_dev *dev) { - dev->priv.pg_wq = create_singlethread_workqueue("mlx5_page_allocator"); - if (!dev->priv.pg_wq) - return -ENOMEM; - - return 0; + MLX5_NB_INIT(&dev->priv.pg_nb, req_pages_handler, PAGE_REQUEST); + mlx5_eq_notifier_register(dev, &dev->priv.pg_nb); } void mlx5_pagealloc_stop(struct mlx5_core_dev *dev) { - destroy_workqueue(dev->priv.pg_wq); + mlx5_eq_notifier_unregister(dev, &dev->priv.pg_nb); + flush_workqueue(dev->priv.pg_wq); } int mlx5_wait_for_vf_pages(struct mlx5_core_dev *dev) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/port.c b/drivers/net/ethernet/mellanox/mlx5/core/port.c index 31a9cbd85689..2b82f35f4c35 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/port.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/port.c @@ -915,63 +915,6 @@ void mlx5_query_port_fcs(struct mlx5_core_dev *mdev, bool *supported, *enabled = !!(MLX5_GET(pcmr_reg, out, fcs_chk)); } -static const char *mlx5_pme_status[MLX5_MODULE_STATUS_NUM] = { - "Cable plugged", /* MLX5_MODULE_STATUS_PLUGGED = 0x1 */ - "Cable unplugged", /* MLX5_MODULE_STATUS_UNPLUGGED = 0x2 */ - "Cable error", /* MLX5_MODULE_STATUS_ERROR = 0x3 */ -}; - -static const char *mlx5_pme_error[MLX5_MODULE_EVENT_ERROR_NUM] = { - "Power budget exceeded", - "Long Range for non MLNX cable", - "Bus stuck(I2C or data shorted)", - "No EEPROM/retry timeout", - "Enforce part number list", - "Unknown identifier", - "High Temperature", - "Bad or shorted cable/module", - "Unknown status", -}; - -void mlx5_port_module_event(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe) -{ - enum port_module_event_status_type module_status; - enum port_module_event_error_type error_type; - struct mlx5_eqe_port_module *module_event_eqe; - struct mlx5_priv *priv = &dev->priv; - u8 module_num; - - module_event_eqe = &eqe->data.port_module; - module_num = module_event_eqe->module; - module_status = module_event_eqe->module_status & - PORT_MODULE_EVENT_MODULE_STATUS_MASK; - error_type = module_event_eqe->error_type & - PORT_MODULE_EVENT_ERROR_TYPE_MASK; - - if (module_status < MLX5_MODULE_STATUS_ERROR) { - priv->pme_stats.status_counters[module_status - 1]++; - } else if (module_status == MLX5_MODULE_STATUS_ERROR) { - if (error_type >= MLX5_MODULE_EVENT_ERROR_UNKNOWN) - /* Unknown error type */ - error_type = MLX5_MODULE_EVENT_ERROR_UNKNOWN; - priv->pme_stats.error_counters[error_type]++; - } - - if (!printk_ratelimit()) - return; - - if (module_status < MLX5_MODULE_STATUS_ERROR) - mlx5_core_info(dev, - "Port module event: module %u, %s\n", - module_num, mlx5_pme_status[module_status - 1]); - - else if (module_status == MLX5_MODULE_STATUS_ERROR) - mlx5_core_info(dev, - "Port module event[error]: module %u, %s, %s\n", - module_num, mlx5_pme_status[module_status - 1], - mlx5_pme_error[error_type]); -} - int mlx5_query_mtpps(struct mlx5_core_dev *mdev, u32 *mtpps, u32 mtpps_size) { u32 in[MLX5_ST_SZ_DW(mtpps_reg)] = {0}; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/qp.c b/drivers/net/ethernet/mellanox/mlx5/core/qp.c index 91b8139a388d..388f205a497f 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/qp.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/qp.c @@ -38,11 +38,11 @@ #include <linux/mlx5/transobj.h> #include "mlx5_core.h" +#include "lib/eq.h" -static struct mlx5_core_rsc_common *mlx5_get_rsc(struct mlx5_core_dev *dev, - u32 rsn) +static struct mlx5_core_rsc_common * +mlx5_get_rsc(struct mlx5_qp_table *table, u32 rsn) { - struct mlx5_qp_table *table = &dev->priv.qp_table; struct mlx5_core_rsc_common *common; spin_lock(&table->lock); @@ -53,11 +53,6 @@ static struct mlx5_core_rsc_common *mlx5_get_rsc(struct mlx5_core_dev *dev, spin_unlock(&table->lock); - if (!common) { - mlx5_core_warn(dev, "Async event for bogus resource 0x%x\n", - rsn); - return NULL; - } return common; } @@ -120,19 +115,57 @@ static bool is_event_type_allowed(int rsc_type, int event_type) } } -void mlx5_rsc_event(struct mlx5_core_dev *dev, u32 rsn, int event_type) +static int rsc_event_notifier(struct notifier_block *nb, + unsigned long type, void *data) { - struct mlx5_core_rsc_common *common = mlx5_get_rsc(dev, rsn); + struct mlx5_core_rsc_common *common; + struct mlx5_qp_table *table; + struct mlx5_core_dev *dev; struct mlx5_core_dct *dct; + u8 event_type = (u8)type; struct mlx5_core_qp *qp; + struct mlx5_priv *priv; + struct mlx5_eqe *eqe; + u32 rsn; + + switch (event_type) { + case MLX5_EVENT_TYPE_DCT_DRAINED: + eqe = data; + rsn = be32_to_cpu(eqe->data.dct.dctn) & 0xffffff; + rsn |= (MLX5_RES_DCT << MLX5_USER_INDEX_LEN); + break; + case MLX5_EVENT_TYPE_PATH_MIG: + case MLX5_EVENT_TYPE_COMM_EST: + case MLX5_EVENT_TYPE_SQ_DRAINED: + case MLX5_EVENT_TYPE_SRQ_LAST_WQE: + case MLX5_EVENT_TYPE_WQ_CATAS_ERROR: + case MLX5_EVENT_TYPE_PATH_MIG_FAILED: + case MLX5_EVENT_TYPE_WQ_INVAL_REQ_ERROR: + case MLX5_EVENT_TYPE_WQ_ACCESS_ERROR: + eqe = data; + rsn = be32_to_cpu(eqe->data.qp_srq.qp_srq_n) & 0xffffff; + rsn |= (eqe->data.qp_srq.type << MLX5_USER_INDEX_LEN); + break; + default: + return NOTIFY_DONE; + } + + table = container_of(nb, struct mlx5_qp_table, nb); + priv = container_of(table, struct mlx5_priv, qp_table); + dev = container_of(priv, struct mlx5_core_dev, priv); + + mlx5_core_dbg(dev, "event (%d) arrived on resource 0x%x\n", eqe->type, rsn); - if (!common) - return; + common = mlx5_get_rsc(table, rsn); + if (!common) { + mlx5_core_warn(dev, "Async event for bogus resource 0x%x\n", rsn); + return NOTIFY_OK; + } if (!is_event_type_allowed((rsn >> MLX5_USER_INDEX_LEN), event_type)) { mlx5_core_warn(dev, "event 0x%.2x is not allowed on resource 0x%.8x\n", event_type, rsn); - return; + goto out; } switch (common->res) { @@ -150,8 +183,10 @@ void mlx5_rsc_event(struct mlx5_core_dev *dev, u32 rsn, int event_type) default: mlx5_core_warn(dev, "invalid resource type for 0x%x\n", rsn); } - +out: mlx5_core_put_rsc(common); + + return NOTIFY_OK; } static int create_resource_common(struct mlx5_core_dev *dev, @@ -487,10 +522,16 @@ void mlx5_init_qp_table(struct mlx5_core_dev *dev) spin_lock_init(&table->lock); INIT_RADIX_TREE(&table->tree, GFP_ATOMIC); mlx5_qp_debugfs_init(dev); + + table->nb.notifier_call = rsc_event_notifier; + mlx5_notifier_register(dev, &table->nb); } void mlx5_cleanup_qp_table(struct mlx5_core_dev *dev) { + struct mlx5_qp_table *table = &dev->priv.qp_table; + + mlx5_notifier_unregister(dev, &table->nb); mlx5_qp_debugfs_cleanup(dev); } @@ -670,3 +711,20 @@ int mlx5_core_query_q_counter(struct mlx5_core_dev *dev, u16 counter_id, return mlx5_cmd_exec(dev, in, sizeof(in), out, out_size); } EXPORT_SYMBOL_GPL(mlx5_core_query_q_counter); + +struct mlx5_core_rsc_common *mlx5_core_res_hold(struct mlx5_core_dev *dev, + int res_num, + enum mlx5_res_type res_type) +{ + u32 rsn = res_num | (res_type << MLX5_USER_INDEX_LEN); + struct mlx5_qp_table *table = &dev->priv.qp_table; + + return mlx5_get_rsc(table, rsn); +} +EXPORT_SYMBOL_GPL(mlx5_core_res_hold); + +void mlx5_core_res_put(struct mlx5_core_rsc_common *res) +{ + mlx5_core_put_rsc(res); +} +EXPORT_SYMBOL_GPL(mlx5_core_res_put); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/transobj.c b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c index a1ee9a8a769e..c4d4b76096dc 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/transobj.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/transobj.c @@ -258,115 +258,6 @@ void mlx5_core_destroy_tis(struct mlx5_core_dev *dev, u32 tisn) } EXPORT_SYMBOL(mlx5_core_destroy_tis); -int mlx5_core_create_rmp(struct mlx5_core_dev *dev, u32 *in, int inlen, - u32 *rmpn) -{ - u32 out[MLX5_ST_SZ_DW(create_rmp_out)] = {0}; - int err; - - MLX5_SET(create_rmp_in, in, opcode, MLX5_CMD_OP_CREATE_RMP); - err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out)); - if (!err) - *rmpn = MLX5_GET(create_rmp_out, out, rmpn); - - return err; -} - -int mlx5_core_modify_rmp(struct mlx5_core_dev *dev, u32 *in, int inlen) -{ - u32 out[MLX5_ST_SZ_DW(modify_rmp_out)] = {0}; - - MLX5_SET(modify_rmp_in, in, opcode, MLX5_CMD_OP_MODIFY_RMP); - return mlx5_cmd_exec(dev, in, inlen, out, sizeof(out)); -} - -int mlx5_core_destroy_rmp(struct mlx5_core_dev *dev, u32 rmpn) -{ - u32 in[MLX5_ST_SZ_DW(destroy_rmp_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(destroy_rmp_out)] = {0}; - - MLX5_SET(destroy_rmp_in, in, opcode, MLX5_CMD_OP_DESTROY_RMP); - MLX5_SET(destroy_rmp_in, in, rmpn, rmpn); - return mlx5_cmd_exec(dev, in, sizeof(in), out, - sizeof(out)); -} - -int mlx5_core_query_rmp(struct mlx5_core_dev *dev, u32 rmpn, u32 *out) -{ - u32 in[MLX5_ST_SZ_DW(query_rmp_in)] = {0}; - int outlen = MLX5_ST_SZ_BYTES(query_rmp_out); - - MLX5_SET(query_rmp_in, in, opcode, MLX5_CMD_OP_QUERY_RMP); - MLX5_SET(query_rmp_in, in, rmpn, rmpn); - return mlx5_cmd_exec(dev, in, sizeof(in), out, outlen); -} - -int mlx5_core_arm_rmp(struct mlx5_core_dev *dev, u32 rmpn, u16 lwm) -{ - void *in; - void *rmpc; - void *wq; - void *bitmask; - int err; - - in = kvzalloc(MLX5_ST_SZ_BYTES(modify_rmp_in), GFP_KERNEL); - if (!in) - return -ENOMEM; - - rmpc = MLX5_ADDR_OF(modify_rmp_in, in, ctx); - bitmask = MLX5_ADDR_OF(modify_rmp_in, in, bitmask); - wq = MLX5_ADDR_OF(rmpc, rmpc, wq); - - MLX5_SET(modify_rmp_in, in, rmp_state, MLX5_RMPC_STATE_RDY); - MLX5_SET(modify_rmp_in, in, rmpn, rmpn); - MLX5_SET(wq, wq, lwm, lwm); - MLX5_SET(rmp_bitmask, bitmask, lwm, 1); - MLX5_SET(rmpc, rmpc, state, MLX5_RMPC_STATE_RDY); - - err = mlx5_core_modify_rmp(dev, in, MLX5_ST_SZ_BYTES(modify_rmp_in)); - - kvfree(in); - - return err; -} - -int mlx5_core_create_xsrq(struct mlx5_core_dev *dev, u32 *in, int inlen, - u32 *xsrqn) -{ - u32 out[MLX5_ST_SZ_DW(create_xrc_srq_out)] = {0}; - int err; - - MLX5_SET(create_xrc_srq_in, in, opcode, MLX5_CMD_OP_CREATE_XRC_SRQ); - err = mlx5_cmd_exec(dev, in, inlen, out, sizeof(out)); - if (!err) - *xsrqn = MLX5_GET(create_xrc_srq_out, out, xrc_srqn); - - return err; -} - -int mlx5_core_destroy_xsrq(struct mlx5_core_dev *dev, u32 xsrqn) -{ - u32 in[MLX5_ST_SZ_DW(destroy_xrc_srq_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(destroy_xrc_srq_out)] = {0}; - - MLX5_SET(destroy_xrc_srq_in, in, opcode, MLX5_CMD_OP_DESTROY_XRC_SRQ); - MLX5_SET(destroy_xrc_srq_in, in, xrc_srqn, xsrqn); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); -} - -int mlx5_core_arm_xsrq(struct mlx5_core_dev *dev, u32 xsrqn, u16 lwm) -{ - u32 in[MLX5_ST_SZ_DW(arm_xrc_srq_in)] = {0}; - u32 out[MLX5_ST_SZ_DW(arm_xrc_srq_out)] = {0}; - - MLX5_SET(arm_xrc_srq_in, in, opcode, MLX5_CMD_OP_ARM_XRC_SRQ); - MLX5_SET(arm_xrc_srq_in, in, xrc_srqn, xsrqn); - MLX5_SET(arm_xrc_srq_in, in, lwm, lwm); - MLX5_SET(arm_xrc_srq_in, in, op_mod, - MLX5_ARM_XRC_SRQ_IN_OP_MOD_XRC_SRQ); - return mlx5_cmd_exec(dev, in, sizeof(in), out, sizeof(out)); -} - int mlx5_core_create_rqt(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *rqtn) { diff --git a/include/linux/mlx5/cq.h b/include/linux/mlx5/cq.h index 31a750570c38..28b757a64029 100644 --- a/include/linux/mlx5/cq.h +++ b/include/linux/mlx5/cq.h @@ -60,7 +60,7 @@ struct mlx5_core_cq { } tasklet_ctx; int reset_notify_added; struct list_head reset_notify; - struct mlx5_eq *eq; + struct mlx5_eq_comp *eq; u16 uid; }; diff --git a/include/linux/mlx5/device.h b/include/linux/mlx5/device.h index b4c0457fbebd..f7c8bebfe472 100644 --- a/include/linux/mlx5/device.h +++ b/include/linux/mlx5/device.h @@ -212,6 +212,13 @@ enum { MLX5_PFAULT_SUBTYPE_RDMA = 1, }; +enum wqe_page_fault_type { + MLX5_WQE_PF_TYPE_RMP = 0, + MLX5_WQE_PF_TYPE_REQ_SEND_OR_WRITE = 1, + MLX5_WQE_PF_TYPE_RESP = 2, + MLX5_WQE_PF_TYPE_REQ_READ_OR_ATOMIC = 3, +}; + enum { MLX5_PERM_LOCAL_READ = 1 << 2, MLX5_PERM_LOCAL_WRITE = 1 << 3, @@ -294,9 +301,15 @@ enum { MLX5_EVENT_QUEUE_TYPE_DCT = 6, }; +/* mlx5 components can subscribe to any one of these events via + * mlx5_eq_notifier_register API. + */ enum mlx5_event { + /* Special value to subscribe to any event */ + MLX5_EVENT_TYPE_NOTIFY_ANY = 0x0, + /* HW events enum start: comp events are not subscribable */ MLX5_EVENT_TYPE_COMP = 0x0, - + /* HW Async events enum start: subscribable events */ MLX5_EVENT_TYPE_PATH_MIG = 0x01, MLX5_EVENT_TYPE_COMM_EST = 0x02, MLX5_EVENT_TYPE_SQ_DRAINED = 0x03, @@ -334,6 +347,8 @@ enum mlx5_event { MLX5_EVENT_TYPE_FPGA_QP_ERROR = 0x21, MLX5_EVENT_TYPE_DEVICE_TRACER = 0x26, + + MLX5_EVENT_TYPE_MAX = MLX5_EVENT_TYPE_DEVICE_TRACER + 1, }; enum { diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index aa5963b5d38e..b090a96f87df 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -46,10 +46,11 @@ #include <linux/mempool.h> #include <linux/interrupt.h> #include <linux/idr.h> +#include <linux/notifier.h> #include <linux/mlx5/device.h> #include <linux/mlx5/doorbell.h> -#include <linux/mlx5/srq.h> +#include <linux/mlx5/eq.h> #include <linux/timecounter.h> #include <linux/ptp_clock_kernel.h> @@ -85,18 +86,6 @@ enum { }; enum { - MLX5_EQ_VEC_PAGES = 0, - MLX5_EQ_VEC_CMD = 1, - MLX5_EQ_VEC_ASYNC = 2, - MLX5_EQ_VEC_PFAULT = 3, - MLX5_EQ_VEC_COMP_BASE, -}; - -enum { - MLX5_MAX_IRQ_NAME = 32 -}; - -enum { MLX5_ATOMIC_MODE_OFFSET = 16, MLX5_ATOMIC_MODE_IB_COMP = 1, MLX5_ATOMIC_MODE_CX = 2, @@ -205,16 +194,7 @@ struct mlx5_rsc_debug { }; enum mlx5_dev_event { - MLX5_DEV_EVENT_SYS_ERROR, - MLX5_DEV_EVENT_PORT_UP, - MLX5_DEV_EVENT_PORT_DOWN, - MLX5_DEV_EVENT_PORT_INITIALIZED, - MLX5_DEV_EVENT_LID_CHANGE, - MLX5_DEV_EVENT_PKEY_CHANGE, - MLX5_DEV_EVENT_GUID_CHANGE, - MLX5_DEV_EVENT_CLIENT_REREG, - MLX5_DEV_EVENT_PPS, - MLX5_DEV_EVENT_DELAY_DROP_TIMEOUT, + MLX5_DEV_EVENT_SYS_ERROR = 128, /* 0 - 127 are FW events */ }; enum mlx5_port_status { @@ -222,14 +202,6 @@ enum mlx5_port_status { MLX5_PORT_DOWN = 2, }; -enum mlx5_eq_type { - MLX5_EQ_TYPE_COMP, - MLX5_EQ_TYPE_ASYNC, -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - MLX5_EQ_TYPE_PF, -#endif -}; - struct mlx5_bfreg_info { u32 *sys_pages; int num_low_latency_bfregs; @@ -297,6 +269,8 @@ struct mlx5_cmd_stats { }; struct mlx5_cmd { + struct mlx5_nb nb; + void *cmd_alloc_buf; dma_addr_t alloc_dma; int alloc_size; @@ -366,51 +340,6 @@ struct mlx5_frag_buf_ctrl { u8 log_frag_strides; }; -struct mlx5_eq_tasklet { - struct list_head list; - struct list_head process_list; - struct tasklet_struct task; - /* lock on completion tasklet list */ - spinlock_t lock; -}; - -struct mlx5_eq_pagefault { - struct work_struct work; - /* Pagefaults lock */ - spinlock_t lock; - struct workqueue_struct *wq; - mempool_t *pool; -}; - -struct mlx5_cq_table { - /* protect radix tree */ - spinlock_t lock; - struct radix_tree_root tree; -}; - -struct mlx5_eq { - struct mlx5_core_dev *dev; - struct mlx5_cq_table cq_table; - __be32 __iomem *doorbell; - u32 cons_index; - struct mlx5_frag_buf buf; - int size; - unsigned int irqn; - u8 eqn; - int nent; - u64 mask; - struct list_head list; - int index; - struct mlx5_rsc_debug *dbg; - enum mlx5_eq_type type; - union { - struct mlx5_eq_tasklet tasklet_ctx; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - struct mlx5_eq_pagefault pf_ctx; -#endif - }; -}; - struct mlx5_core_psv { u32 psv_idx; struct psv_layout { @@ -463,36 +392,6 @@ struct mlx5_core_rsc_common { struct completion free; }; -struct mlx5_core_srq { - struct mlx5_core_rsc_common common; /* must be first */ - u32 srqn; - int max; - size_t max_gs; - size_t max_avail_gather; - int wqe_shift; - void (*event) (struct mlx5_core_srq *, enum mlx5_event); - - atomic_t refcount; - struct completion free; - u16 uid; -}; - -struct mlx5_eq_table { - void __iomem *update_ci; - void __iomem *update_arm_ci; - struct list_head comp_eqs_list; - struct mlx5_eq pages_eq; - struct mlx5_eq async_eq; - struct mlx5_eq cmd_eq; -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - struct mlx5_eq pfault_eq; -#endif - int num_comp_vectors; - /* protect EQs list - */ - spinlock_t lock; -}; - struct mlx5_uars_page { void __iomem *map; bool wc; @@ -542,13 +441,8 @@ struct mlx5_core_health { }; struct mlx5_qp_table { - /* protect radix tree - */ - spinlock_t lock; - struct radix_tree_root tree; -}; + struct notifier_block nb; -struct mlx5_srq_table { /* protect radix tree */ spinlock_t lock; @@ -575,11 +469,6 @@ struct mlx5_core_sriov { int enabled_vfs; }; -struct mlx5_irq_info { - cpumask_var_t mask; - char name[MLX5_MAX_IRQ_NAME]; -}; - struct mlx5_fc_stats { spinlock_t counters_idr_lock; /* protects counters_idr */ struct idr counters_idr; @@ -593,10 +482,11 @@ struct mlx5_fc_stats { unsigned long sampling_interval; /* jiffies */ }; +struct mlx5_events; struct mlx5_mpfs; struct mlx5_eswitch; struct mlx5_lag; -struct mlx5_pagefault; +struct mlx5_eq_table; struct mlx5_rate_limit { u32 rate; @@ -619,37 +509,12 @@ struct mlx5_rl_table { struct mlx5_rl_entry *rl_entry; }; -enum port_module_event_status_type { - MLX5_MODULE_STATUS_PLUGGED = 0x1, - MLX5_MODULE_STATUS_UNPLUGGED = 0x2, - MLX5_MODULE_STATUS_ERROR = 0x3, - MLX5_MODULE_STATUS_NUM = 0x3, -}; - -enum port_module_event_error_type { - MLX5_MODULE_EVENT_ERROR_POWER_BUDGET_EXCEEDED, - MLX5_MODULE_EVENT_ERROR_LONG_RANGE_FOR_NON_MLNX_CABLE_MODULE, - MLX5_MODULE_EVENT_ERROR_BUS_STUCK, - MLX5_MODULE_EVENT_ERROR_NO_EEPROM_RETRY_TIMEOUT, - MLX5_MODULE_EVENT_ERROR_ENFORCE_PART_NUMBER_LIST, - MLX5_MODULE_EVENT_ERROR_UNKNOWN_IDENTIFIER, - MLX5_MODULE_EVENT_ERROR_HIGH_TEMPERATURE, - MLX5_MODULE_EVENT_ERROR_BAD_CABLE, - MLX5_MODULE_EVENT_ERROR_UNKNOWN, - MLX5_MODULE_EVENT_ERROR_NUM, -}; - -struct mlx5_port_module_event_stats { - u64 status_counters[MLX5_MODULE_STATUS_NUM]; - u64 error_counters[MLX5_MODULE_EVENT_ERROR_NUM]; -}; - struct mlx5_priv { char name[MLX5_MAX_NAME_LEN]; - struct mlx5_eq_table eq_table; - struct mlx5_irq_info *irq_info; + struct mlx5_eq_table *eq_table; /* pages stuff */ + struct mlx5_nb pg_nb; struct workqueue_struct *pg_wq; struct rb_root page_root; int fw_pages; @@ -659,8 +524,6 @@ struct mlx5_priv { struct mlx5_core_health health; - struct mlx5_srq_table srq_table; - /* start: qp staff */ struct mlx5_qp_table qp_table; struct dentry *qp_debugfs; @@ -690,9 +553,7 @@ struct mlx5_priv { struct list_head dev_list; struct list_head ctx_list; spinlock_t ctx_lock; - - struct list_head waiting_events_list; - bool is_accum_events; + struct mlx5_events *events; struct mlx5_flow_steering *steering; struct mlx5_mpfs *mpfs; @@ -703,15 +564,6 @@ struct mlx5_priv { struct mlx5_fc_stats fc_stats; struct mlx5_rl_table rl_table; - struct mlx5_port_module_event_stats pme_stats; - -#ifdef CONFIG_INFINIBAND_ON_DEMAND_PAGING - void (*pfault)(struct mlx5_core_dev *dev, - void *context, - struct mlx5_pagefault *pfault); - void *pfault_ctx; - struct srcu_struct pfault_srcu; -#endif struct mlx5_bfreg_data bfregs; struct mlx5_uars_page *uar; }; @@ -736,44 +588,6 @@ enum mlx5_pagefault_type_flags { MLX5_PFAULT_RDMA = 1 << 2, }; -/* Contains the details of a pagefault. */ -struct mlx5_pagefault { - u32 bytes_committed; - u32 token; - u8 event_subtype; - u8 type; - union { - /* Initiator or send message responder pagefault details. */ - struct { - /* Received packet size, only valid for responders. */ - u32 packet_size; - /* - * Number of resource holding WQE, depends on type. - */ - u32 wq_num; - /* - * WQE index. Refers to either the send queue or - * receive queue, according to event_subtype. - */ - u16 wqe_index; - } wqe; - /* RDMA responder pagefault details */ - struct { - u32 r_key; - /* - * Received packet size, minimal size page fault - * resolution required for forward progress. - */ - u32 packet_size; - u32 rdma_op_len; - u64 rdma_va; - } rdma; - }; - - struct mlx5_eq *eq; - struct work_struct work; -}; - struct mlx5_td { struct list_head tirs_list; u32 tdn; @@ -803,6 +617,8 @@ struct mlx5_pps { }; struct mlx5_clock { + struct mlx5_core_dev *mdev; + struct mlx5_nb pps_nb; seqlock_t lock; struct cyclecounter cycles; struct timecounter tc; @@ -810,7 +626,6 @@ struct mlx5_clock { u32 nominal_c_mult; unsigned long overflow_period; struct delayed_work overflow_work; - struct mlx5_core_dev *mdev; struct ptp_clock *ptp; struct ptp_clock_info ptp_info; struct mlx5_pps pps_info; @@ -843,9 +658,6 @@ struct mlx5_core_dev { /* sync interface state */ struct mutex intf_state_mutex; unsigned long intf_state; - void (*event) (struct mlx5_core_dev *dev, - enum mlx5_dev_event event, - unsigned long param); struct mlx5_priv priv; struct mlx5_profile *profile; atomic_t num_qps; @@ -859,9 +671,6 @@ struct mlx5_core_dev { #ifdef CONFIG_MLX5_FPGA struct mlx5_fpga_device *fpga; #endif -#ifdef CONFIG_RFS_ACCEL - struct cpu_rmap *rmap; -#endif struct mlx5_clock clock; struct mlx5_ib_clock_info *clock_info; struct page *clock_info_page; @@ -940,8 +749,8 @@ struct mlx5_hca_vport_context { u64 node_guid; u32 cap_mask1; u32 cap_mask1_perm; - u32 cap_mask2; - u32 cap_mask2_perm; + u16 cap_mask2; + u16 cap_mask2_perm; u16 lid; u8 init_type_reply; /* bitmask: see ib spec 14.2.5.6 InitTypeReply */ u8 lmc; @@ -1070,13 +879,6 @@ struct mlx5_cmd_mailbox *mlx5_alloc_cmd_mailbox_chain(struct mlx5_core_dev *dev, gfp_t flags, int npages); void mlx5_free_cmd_mailbox_chain(struct mlx5_core_dev *dev, struct mlx5_cmd_mailbox *head); -int mlx5_core_create_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - struct mlx5_srq_attr *in); -int mlx5_core_destroy_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq); -int mlx5_core_query_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - struct mlx5_srq_attr *out); -int mlx5_core_arm_srq(struct mlx5_core_dev *dev, struct mlx5_core_srq *srq, - u16 lwm, int is_srq); void mlx5_init_mkey_table(struct mlx5_core_dev *dev); void mlx5_cleanup_mkey_table(struct mlx5_core_dev *dev); int mlx5_core_create_mkey_cb(struct mlx5_core_dev *dev, @@ -1095,9 +897,9 @@ int mlx5_core_alloc_pd(struct mlx5_core_dev *dev, u32 *pdn); int mlx5_core_dealloc_pd(struct mlx5_core_dev *dev, u32 pdn); int mlx5_core_mad_ifc(struct mlx5_core_dev *dev, const void *inb, void *outb, u16 opmod, u8 port); -void mlx5_pagealloc_init(struct mlx5_core_dev *dev); +int mlx5_pagealloc_init(struct mlx5_core_dev *dev); void mlx5_pagealloc_cleanup(struct mlx5_core_dev *dev); -int mlx5_pagealloc_start(struct mlx5_core_dev *dev); +void mlx5_pagealloc_start(struct mlx5_core_dev *dev); void mlx5_pagealloc_stop(struct mlx5_core_dev *dev); void mlx5_core_req_pages_handler(struct mlx5_core_dev *dev, u16 func_id, s32 npages); @@ -1108,9 +910,6 @@ void mlx5_unregister_debugfs(void); void mlx5_fill_page_array(struct mlx5_frag_buf *buf, __be64 *pas); void mlx5_fill_page_frag_array(struct mlx5_frag_buf *frag_buf, __be64 *pas); -void mlx5_rsc_event(struct mlx5_core_dev *dev, u32 rsn, int event_type); -void mlx5_srq_event(struct mlx5_core_dev *dev, u32 srqn, int event_type); -struct mlx5_core_srq *mlx5_core_get_srq(struct mlx5_core_dev *dev, u32 srqn); int mlx5_vector2eqn(struct mlx5_core_dev *dev, int vector, int *eqn, unsigned int *irqn); int mlx5_core_attach_mcg(struct mlx5_core_dev *dev, union ib_gid *mgid, u32 qpn); @@ -1155,6 +954,9 @@ int mlx5_alloc_bfreg(struct mlx5_core_dev *mdev, struct mlx5_sq_bfreg *bfreg, bool map_wc, bool fast_path); void mlx5_free_bfreg(struct mlx5_core_dev *mdev, struct mlx5_sq_bfreg *bfreg); +unsigned int mlx5_comp_vectors_count(struct mlx5_core_dev *dev); +struct cpumask * +mlx5_comp_irq_get_affinity_mask(struct mlx5_core_dev *dev, int vector); unsigned int mlx5_core_reserved_gids_count(struct mlx5_core_dev *dev); int mlx5_core_roce_gid_set(struct mlx5_core_dev *dev, unsigned int index, u8 roce_version, u8 roce_l3_type, const u8 *gid, @@ -1202,11 +1004,6 @@ struct mlx5_interface { void (*remove)(struct mlx5_core_dev *dev, void *context); int (*attach)(struct mlx5_core_dev *dev, void *context); void (*detach)(struct mlx5_core_dev *dev, void *context); - void (*event)(struct mlx5_core_dev *dev, void *context, - enum mlx5_dev_event event, unsigned long param); - void (*pfault)(struct mlx5_core_dev *dev, - void *context, - struct mlx5_pagefault *pfault); void * (*get_dev)(void *context); int protocol; struct list_head list; @@ -1215,6 +1012,9 @@ struct mlx5_interface { void *mlx5_get_protocol_dev(struct mlx5_core_dev *mdev, int protocol); int mlx5_register_interface(struct mlx5_interface *intf); void mlx5_unregister_interface(struct mlx5_interface *intf); +int mlx5_notifier_register(struct mlx5_core_dev *dev, struct notifier_block *nb); +int mlx5_notifier_unregister(struct mlx5_core_dev *dev, struct notifier_block *nb); + int mlx5_core_query_vendor_id(struct mlx5_core_dev *mdev, u32 *vendor_id); int mlx5_cmd_create_vport_lag(struct mlx5_core_dev *dev); @@ -1306,10 +1106,4 @@ enum { MLX5_TRIGGERED_CMD_COMP = (u64)1 << 32, }; -static inline const struct cpumask * -mlx5_get_vector_affinity_hint(struct mlx5_core_dev *dev, int vector) -{ - return dev->priv.irq_info[vector].mask; -} - #endif /* MLX5_DRIVER_H */ diff --git a/include/linux/mlx5/eq.h b/include/linux/mlx5/eq.h new file mode 100644 index 000000000000..00045cc4ea11 --- /dev/null +++ b/include/linux/mlx5/eq.h @@ -0,0 +1,72 @@ +/* SPDX-License-Identifier: GPL-2.0 OR Linux-OpenIB */ +/* Copyright (c) 2018 Mellanox Technologies. */ + +#ifndef MLX5_CORE_EQ_H +#define MLX5_CORE_EQ_H + +enum { + MLX5_EQ_PAGEREQ_IDX = 0, + MLX5_EQ_CMD_IDX = 1, + MLX5_EQ_ASYNC_IDX = 2, + /* reserved to be used by mlx5_core ulps (mlx5e/mlx5_ib) */ + MLX5_EQ_PFAULT_IDX = 3, + MLX5_EQ_MAX_ASYNC_EQS, + /* completion eqs vector indices start here */ + MLX5_EQ_VEC_COMP_BASE = MLX5_EQ_MAX_ASYNC_EQS, +}; + +#define MLX5_NUM_CMD_EQE (32) +#define MLX5_NUM_ASYNC_EQE (0x1000) +#define MLX5_NUM_SPARE_EQE (0x80) + +struct mlx5_eq; +struct mlx5_core_dev; + +struct mlx5_eq_param { + u8 index; + int nent; + u64 mask; + void *context; + irq_handler_t handler; +}; + +struct mlx5_eq * +mlx5_eq_create_generic(struct mlx5_core_dev *dev, const char *name, + struct mlx5_eq_param *param); +int +mlx5_eq_destroy_generic(struct mlx5_core_dev *dev, struct mlx5_eq *eq); + +struct mlx5_eqe *mlx5_eq_get_eqe(struct mlx5_eq *eq, u32 cc); +void mlx5_eq_update_ci(struct mlx5_eq *eq, u32 cc, bool arm); + +/* The HCA will think the queue has overflowed if we + * don't tell it we've been processing events. We + * create EQs with MLX5_NUM_SPARE_EQE extra entries, + * so we must update our consumer index at + * least that often. + * + * mlx5_eq_update_cc must be called on every EQE @EQ irq handler + */ +static inline u32 mlx5_eq_update_cc(struct mlx5_eq *eq, u32 cc) +{ + if (unlikely(cc >= MLX5_NUM_SPARE_EQE)) { + mlx5_eq_update_ci(eq, cc, 0); + cc = 0; + } + return cc; +} + +struct mlx5_nb { + struct notifier_block nb; + u8 event_type; +}; + +#define mlx5_nb_cof(ptr, type, member) \ + (container_of(container_of(ptr, struct mlx5_nb, nb), type, member)) + +#define MLX5_NB_INIT(name, handler, event) do { \ + (name)->nb.notifier_call = handler; \ + (name)->event_type = MLX5_EVENT_TYPE_##event; \ +} while (0) + +#endif /* MLX5_CORE_EQ_H */ diff --git a/include/linux/mlx5/mlx5_ifc.h b/include/linux/mlx5/mlx5_ifc.h index 34e17e6f8942..52393fbcf3b4 100644 --- a/include/linux/mlx5/mlx5_ifc.h +++ b/include/linux/mlx5/mlx5_ifc.h @@ -144,6 +144,9 @@ enum { MLX5_CMD_OP_DESTROY_XRQ = 0x718, MLX5_CMD_OP_QUERY_XRQ = 0x719, MLX5_CMD_OP_ARM_XRQ = 0x71a, + MLX5_CMD_OP_QUERY_XRQ_DC_PARAMS_ENTRY = 0x725, + MLX5_CMD_OP_SET_XRQ_DC_PARAMS_ENTRY = 0x726, + MLX5_CMD_OP_QUERY_XRQ_ERROR_PARAMS = 0x727, MLX5_CMD_OP_QUERY_VPORT_STATE = 0x750, MLX5_CMD_OP_MODIFY_VPORT_STATE = 0x751, MLX5_CMD_OP_QUERY_ESW_VPORT_CONTEXT = 0x752, @@ -245,6 +248,7 @@ enum { MLX5_CMD_OP_MODIFY_FLOW_TABLE = 0x93c, MLX5_CMD_OP_ALLOC_PACKET_REFORMAT_CONTEXT = 0x93d, MLX5_CMD_OP_DEALLOC_PACKET_REFORMAT_CONTEXT = 0x93e, + MLX5_CMD_OP_QUERY_PACKET_REFORMAT_CONTEXT = 0x93f, MLX5_CMD_OP_ALLOC_MODIFY_HEADER_CONTEXT = 0x940, MLX5_CMD_OP_DEALLOC_MODIFY_HEADER_CONTEXT = 0x941, MLX5_CMD_OP_QUERY_MODIFY_HEADER_CONTEXT = 0x942, @@ -260,6 +264,12 @@ enum { MLX5_CMD_OP_MAX }; +/* Valid range for general commands that don't work over an object */ +enum { + MLX5_CMD_OP_GENERAL_START = 0xb00, + MLX5_CMD_OP_GENERAL_END = 0xd00, +}; + struct mlx5_ifc_flow_table_fields_supported_bits { u8 outer_dmac[0x1]; u8 outer_smac[0x1]; @@ -349,7 +359,7 @@ struct mlx5_ifc_flow_table_prop_layout_bits { u8 reformat_l3_tunnel_to_l2[0x1]; u8 reformat_l2_to_l3_tunnel[0x1]; u8 reformat_and_modify_action[0x1]; - u8 reserved_at_14[0xb]; + u8 reserved_at_15[0xb]; u8 reserved_at_20[0x2]; u8 log_max_ft_size[0x6]; u8 log_max_modify_header_context[0x8]; @@ -586,7 +596,7 @@ struct mlx5_ifc_flow_table_eswitch_cap_bits { u8 fdb_multi_path_to_table[0x1]; u8 reserved_at_1d[0x1]; u8 multi_fdb_encap[0x1]; - u8 reserved_at_1e[0x1e1]; + u8 reserved_at_1f[0x1e1]; struct mlx5_ifc_flow_table_prop_layout_bits flow_table_properties_nic_esw_fdb; @@ -829,7 +839,7 @@ struct mlx5_ifc_vector_calc_cap_bits { struct mlx5_ifc_calc_op calc2; struct mlx5_ifc_calc_op calc3; - u8 reserved_at_e0[0x720]; + u8 reserved_at_c0[0x720]; }; enum { @@ -883,6 +893,10 @@ enum { MLX5_CAP_UMR_FENCE_NONE = 0x2, }; +enum { + MLX5_UCTX_CAP_RAW_TX = 1UL << 0, +}; + struct mlx5_ifc_cmd_hca_cap_bits { u8 reserved_at_0[0x30]; u8 vhca_id[0x10]; @@ -1043,7 +1057,8 @@ struct mlx5_ifc_cmd_hca_cap_bits { u8 vector_calc[0x1]; u8 umr_ptr_rlky[0x1]; u8 imaicl[0x1]; - u8 reserved_at_232[0x4]; + u8 qp_packet_based[0x1]; + u8 reserved_at_233[0x3]; u8 qkv[0x1]; u8 pkv[0x1]; u8 set_deth_sqpn[0x1]; @@ -1193,7 +1208,13 @@ struct mlx5_ifc_cmd_hca_cap_bits { u8 num_vhca_ports[0x8]; u8 reserved_at_618[0x6]; u8 sw_owner_id[0x1]; - u8 reserved_at_61f[0x1e1]; + u8 reserved_at_61f[0x1]; + + u8 reserved_at_620[0x80]; + + u8 uctx_cap[0x20]; + + u8 reserved_at_6c0[0x140]; }; enum mlx5_flow_destination_type { @@ -2249,7 +2270,8 @@ struct mlx5_ifc_qpc_bits { u8 st[0x8]; u8 reserved_at_10[0x3]; u8 pm_state[0x2]; - u8 reserved_at_15[0x3]; + u8 reserved_at_15[0x1]; + u8 req_e2e_credit_mode[0x2]; u8 offload_type[0x4]; u8 end_padding_mode[0x2]; u8 reserved_at_1e[0x2]; @@ -5567,7 +5589,7 @@ struct mlx5_ifc_modify_nic_vport_context_out_bits { struct mlx5_ifc_modify_nic_vport_field_select_bits { u8 reserved_at_0[0x12]; u8 affiliation[0x1]; - u8 reserved_at_e[0x1]; + u8 reserved_at_13[0x1]; u8 disable_uc_local_lb[0x1]; u8 disable_mc_local_lb[0x1]; u8 node_guid[0x1]; @@ -9028,7 +9050,7 @@ struct mlx5_ifc_dcbx_param_bits { u8 dcbx_cee_cap[0x1]; u8 dcbx_ieee_cap[0x1]; u8 dcbx_standby_cap[0x1]; - u8 reserved_at_0[0x5]; + u8 reserved_at_3[0x5]; u8 port_number[0x8]; u8 reserved_at_10[0xa]; u8 max_application_table_size[6]; @@ -9276,7 +9298,9 @@ struct mlx5_ifc_umem_bits { struct mlx5_ifc_uctx_bits { u8 modify_field_select[0x40]; - u8 reserved_at_40[0x1c0]; + u8 cap[0x20]; + + u8 reserved_at_60[0x1a0]; }; struct mlx5_ifc_create_umem_in_bits { diff --git a/include/linux/mlx5/port.h b/include/linux/mlx5/port.h index 34aed6032f86..bf4bc01ffb0c 100644 --- a/include/linux/mlx5/port.h +++ b/include/linux/mlx5/port.h @@ -107,9 +107,6 @@ enum mlx5e_connector_type { #define MLX5E_PROT_MASK(link_mode) (1 << link_mode) -#define PORT_MODULE_EVENT_MODULE_STATUS_MASK 0xF -#define PORT_MODULE_EVENT_ERROR_TYPE_MASK 0xF - int mlx5_set_port_caps(struct mlx5_core_dev *dev, u8 port_num, u32 caps); int mlx5_query_port_ptys(struct mlx5_core_dev *dev, u32 *ptys, int ptys_size, int proto_mask, u8 local_port); diff --git a/include/linux/mlx5/qp.h b/include/linux/mlx5/qp.h index fbe322c966bc..b26ea9077384 100644 --- a/include/linux/mlx5/qp.h +++ b/include/linux/mlx5/qp.h @@ -596,6 +596,11 @@ int mlx5_core_dealloc_q_counter(struct mlx5_core_dev *dev, u16 counter_id); int mlx5_core_query_q_counter(struct mlx5_core_dev *dev, u16 counter_id, int reset, void *out, int out_size); +struct mlx5_core_rsc_common *mlx5_core_res_hold(struct mlx5_core_dev *dev, + int res_num, + enum mlx5_res_type res_type); +void mlx5_core_res_put(struct mlx5_core_rsc_common *res); + static inline const char *mlx5_qp_type_str(int type) { switch (type) { diff --git a/include/linux/mlx5/srq.h b/include/linux/mlx5/srq.h deleted file mode 100644 index 1b1f3c20c6a3..000000000000 --- a/include/linux/mlx5/srq.h +++ /dev/null @@ -1,72 +0,0 @@ -/* - * Copyright (c) 2013-2015, Mellanox Technologies. 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 - * General Public License (GPL) Version 2, available from the file - * COPYING in the main directory of this source tree, or the - * OpenIB.org BSD license below: - * - * Redistribution and use in source and binary forms, with or - * without modification, are permitted provided that the following - * conditions are met: - * - * - Redistributions of source code must retain the above - * copyright notice, this list of conditions and the following - * disclaimer. - * - * - Redistributions in binary form must reproduce the above - * copyright notice, this list of conditions and the following - * disclaimer in the documentation and/or other materials - * provided with the distribution. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, - * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF - * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND - * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS - * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN - * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN - * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. - */ - -#ifndef MLX5_SRQ_H -#define MLX5_SRQ_H - -#include <linux/mlx5/driver.h> - -enum { - MLX5_SRQ_FLAG_ERR = (1 << 0), - MLX5_SRQ_FLAG_WQ_SIG = (1 << 1), - MLX5_SRQ_FLAG_RNDV = (1 << 2), -}; - -struct mlx5_srq_attr { - u32 type; - u32 flags; - u32 log_size; - u32 wqe_shift; - u32 log_page_size; - u32 wqe_cnt; - u32 srqn; - u32 xrcd; - u32 page_offset; - u32 cqn; - u32 pd; - u32 lwm; - u32 user_index; - u64 db_record; - __be64 *pas; - u32 tm_log_list_size; - u32 tm_next_tag; - u32 tm_hw_phase_cnt; - u32 tm_sw_phase_cnt; - u16 uid; -}; - -struct mlx5_core_dev; - -void mlx5_init_srq_table(struct mlx5_core_dev *dev); -void mlx5_cleanup_srq_table(struct mlx5_core_dev *dev); - -#endif /* MLX5_SRQ_H */ diff --git a/include/linux/mlx5/transobj.h b/include/linux/mlx5/transobj.h index 7f5ca2cd3a32..a261d5528ff7 100644 --- a/include/linux/mlx5/transobj.h +++ b/include/linux/mlx5/transobj.h @@ -58,17 +58,6 @@ int mlx5_core_create_tis(struct mlx5_core_dev *dev, u32 *in, int inlen, int mlx5_core_modify_tis(struct mlx5_core_dev *dev, u32 tisn, u32 *in, int inlen); void mlx5_core_destroy_tis(struct mlx5_core_dev *dev, u32 tisn); -int mlx5_core_create_rmp(struct mlx5_core_dev *dev, u32 *in, int inlen, - u32 *rmpn); -int mlx5_core_modify_rmp(struct mlx5_core_dev *dev, u32 *in, int inlen); -int mlx5_core_destroy_rmp(struct mlx5_core_dev *dev, u32 rmpn); -int mlx5_core_query_rmp(struct mlx5_core_dev *dev, u32 rmpn, u32 *out); -int mlx5_core_arm_rmp(struct mlx5_core_dev *dev, u32 rmpn, u16 lwm); -int mlx5_core_create_xsrq(struct mlx5_core_dev *dev, u32 *in, int inlen, - u32 *rmpn); -int mlx5_core_destroy_xsrq(struct mlx5_core_dev *dev, u32 rmpn); -int mlx5_core_arm_xsrq(struct mlx5_core_dev *dev, u32 rmpn, u16 lwm); - int mlx5_core_create_rqt(struct mlx5_core_dev *dev, u32 *in, int inlen, u32 *rqtn); int mlx5_core_modify_rqt(struct mlx5_core_dev *dev, u32 rqtn, u32 *in, diff --git a/include/rdma/ib_fmr_pool.h b/include/rdma/ib_fmr_pool.h index f62b842e6596..f8982e4e9702 100644 --- a/include/rdma/ib_fmr_pool.h +++ b/include/rdma/ib_fmr_pool.h @@ -88,6 +88,6 @@ struct ib_pool_fmr *ib_fmr_pool_map_phys(struct ib_fmr_pool *pool_handle, int list_len, u64 io_virtual_address); -int ib_fmr_pool_unmap(struct ib_pool_fmr *fmr); +void ib_fmr_pool_unmap(struct ib_pool_fmr *fmr); #endif /* IB_FMR_POOL_H */ diff --git a/include/rdma/ib_mad.h b/include/rdma/ib_mad.h index f6ba366051c7..fdef558e3a2d 100644 --- a/include/rdma/ib_mad.h +++ b/include/rdma/ib_mad.h @@ -277,6 +277,7 @@ enum ib_port_capability_mask_bits { IB_PORT_SYS_IMAGE_GUID_SUP = 1 << 11, IB_PORT_PKEY_SW_EXT_PORT_TRAP_SUP = 1 << 12, IB_PORT_EXTENDED_SPEEDS_SUP = 1 << 14, + IB_PORT_CAP_MASK2_SUP = 1 << 15, IB_PORT_CM_SUP = 1 << 16, IB_PORT_SNMP_TUNNEL_SUP = 1 << 17, IB_PORT_REINIT_SUP = 1 << 18, @@ -295,6 +296,15 @@ enum ib_port_capability_mask_bits { IB_PORT_HIERARCHY_INFO_SUP = 1ULL << 31, }; +enum ib_port_capability_mask2_bits { + IB_PORT_SET_NODE_DESC_SUP = 1 << 0, + IB_PORT_EX_PORT_INFO_EX_SUP = 1 << 1, + IB_PORT_VIRT_SUP = 1 << 2, + IB_PORT_SWITCH_PORT_STATE_TABLE_SUP = 1 << 3, + IB_PORT_LINK_WIDTH_2X_SUP = 1 << 4, + IB_PORT_LINK_SPEED_HDR_SUP = 1 << 5, +}; + #define OPA_CLASS_PORT_INFO_PR_SUPPORT BIT(26) struct opa_class_port_info { diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 9c0c2132a2d6..89eead636e68 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -41,14 +41,11 @@ #include <linux/types.h> #include <linux/device.h> -#include <linux/mm.h> #include <linux/dma-mapping.h> #include <linux/kref.h> #include <linux/list.h> #include <linux/rwsem.h> -#include <linux/scatterlist.h> #include <linux/workqueue.h> -#include <linux/socket.h> #include <linux/irq_poll.h> #include <uapi/linux/if_ether.h> #include <net/ipv6.h> @@ -56,7 +53,7 @@ #include <linux/string.h> #include <linux/slab.h> #include <linux/netdevice.h> - +#include <linux/refcount.h> #include <linux/if_link.h> #include <linux/atomic.h> #include <linux/mmu_notifier.h> @@ -437,6 +434,7 @@ enum ib_port_state { enum ib_port_width { IB_WIDTH_1X = 1, + IB_WIDTH_2X = 16, IB_WIDTH_4X = 2, IB_WIDTH_8X = 4, IB_WIDTH_12X = 8 @@ -446,6 +444,7 @@ static inline int ib_width_enum_to_int(enum ib_port_width width) { switch (width) { case IB_WIDTH_1X: return 1; + case IB_WIDTH_2X: return 2; case IB_WIDTH_4X: return 4; case IB_WIDTH_8X: return 8; case IB_WIDTH_12X: return 12; @@ -595,6 +594,7 @@ struct ib_port_attr { u8 active_width; u8 active_speed; u8 phys_state; + u16 port_cap_flags2; }; enum ib_device_modify_flags { @@ -732,7 +732,11 @@ enum ib_rate { IB_RATE_25_GBPS = 15, IB_RATE_100_GBPS = 16, IB_RATE_200_GBPS = 17, - IB_RATE_300_GBPS = 18 + IB_RATE_300_GBPS = 18, + IB_RATE_28_GBPS = 19, + IB_RATE_50_GBPS = 20, + IB_RATE_400_GBPS = 21, + IB_RATE_600_GBPS = 22, }; /** @@ -1508,6 +1512,10 @@ struct ib_ucontext { #endif struct ib_rdmacg_object cg_obj; + /* + * Implementation details of the RDMA core, don't use in drivers: + */ + struct rdma_restrack_entry res; }; struct ib_uobject { @@ -2301,7 +2309,7 @@ struct ib_device { * index - Updated the single counter pointed to by index * num_counters - Updated all counters (will reset the timestamp * and prevent further calls for lifespan milliseconds) - * Drivers are allowed to update all counters in leiu of just the + * Drivers are allowed to update all counters in lieu of just the * one given in index at their option */ int (*get_hw_stats)(struct ib_device *device, @@ -2603,8 +2611,14 @@ struct ib_device { const struct cpumask *(*get_vector_affinity)(struct ib_device *ibdev, int comp_vector); - const struct uverbs_object_tree_def *const *driver_specs; + const struct uapi_definition *driver_def; enum rdma_driver_id driver_id; + /* + * Provides synchronization between device unregistration and netlink + * commands on a device. To be used only by core. + */ + refcount_t refcount; + struct completion unreg_completion; }; struct ib_client { @@ -4204,10 +4218,10 @@ ib_get_vector_affinity(struct ib_device *device, int comp_vector) */ void rdma_roce_rescan_device(struct ib_device *ibdev); -struct ib_ucontext *ib_uverbs_get_ucontext(struct ib_uverbs_file *ufile); +struct ib_ucontext *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile); + -int uverbs_destroy_def_handler(struct ib_uverbs_file *file, - struct uverbs_attr_bundle *attrs); +int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs); struct net_device *rdma_alloc_netdev(struct ib_device *device, u8 port_num, enum rdma_netdev_t type, const char *name, diff --git a/include/rdma/rdma_vt.h b/include/rdma/rdma_vt.h index 3584d0816fcd..dd0ed8048bb4 100644 --- a/include/rdma/rdma_vt.h +++ b/include/rdma/rdma_vt.h @@ -269,6 +269,13 @@ struct rvt_driver_provided { void * (*qp_priv_alloc)(struct rvt_dev_info *rdi, struct rvt_qp *qp); /* + * Init a struture allocated with qp_priv_alloc(). This should be + * called after all qp fields have been initialized in rdmavt. + */ + int (*qp_priv_init)(struct rvt_dev_info *rdi, struct rvt_qp *qp, + struct ib_qp_init_attr *init_attr); + + /* * Free the driver's private qp structure. */ void (*qp_priv_free)(struct rvt_dev_info *rdi, struct rvt_qp *qp); diff --git a/include/rdma/restrack.h b/include/rdma/restrack.h index 2638fa7cd702..f34aa96e4518 100644 --- a/include/rdma/restrack.h +++ b/include/rdma/restrack.h @@ -39,6 +39,10 @@ enum rdma_restrack_type { */ RDMA_RESTRACK_MR, /** + * @RDMA_RESTRACK_CTX: Verbs contexts (CTX) + */ + RDMA_RESTRACK_CTX, + /** * @RDMA_RESTRACK_MAX: Last entry, used for array dclarations */ RDMA_RESTRACK_MAX diff --git a/include/rdma/uverbs_ioctl.h b/include/rdma/uverbs_ioctl.h index 84d3d15f1f38..2f56844fb7da 100644 --- a/include/rdma/uverbs_ioctl.h +++ b/include/rdma/uverbs_ioctl.h @@ -79,6 +79,8 @@ struct uverbs_attr_spec { */ u8 alloc_and_copy:1; u8 mandatory:1; + /* True if this is from UVERBS_ATTR_UHW */ + u8 is_udata:1; union { struct { @@ -140,6 +142,13 @@ struct uverbs_attr_spec { * * The tree encodes multiple types, and uses a scheme where OBJ_ID,0,0 returns * the object slot, and OBJ_ID,METH_ID,0 and returns the method slot. + * + * This also encodes the tables for the write() and write() extended commands + * using the coding + * OBJ_ID,UVERBS_API_METHOD_IS_WRITE,command # + * OBJ_ID,UVERBS_API_METHOD_IS_WRITE_EX,command_ex # + * ie the WRITE path is treated as a special method type in the ioctl + * framework. */ enum uapi_radix_data { UVERBS_API_NS_FLAG = 1U << UVERBS_ID_NS_SHIFT, @@ -147,12 +156,16 @@ enum uapi_radix_data { UVERBS_API_ATTR_KEY_BITS = 6, UVERBS_API_ATTR_KEY_MASK = GENMASK(UVERBS_API_ATTR_KEY_BITS - 1, 0), UVERBS_API_ATTR_BKEY_LEN = (1 << UVERBS_API_ATTR_KEY_BITS) - 1, + UVERBS_API_WRITE_KEY_NUM = 1 << UVERBS_API_ATTR_KEY_BITS, UVERBS_API_METHOD_KEY_BITS = 5, UVERBS_API_METHOD_KEY_SHIFT = UVERBS_API_ATTR_KEY_BITS, - UVERBS_API_METHOD_KEY_NUM_CORE = 24, - UVERBS_API_METHOD_KEY_NUM_DRIVER = (1 << UVERBS_API_METHOD_KEY_BITS) - - UVERBS_API_METHOD_KEY_NUM_CORE, + UVERBS_API_METHOD_KEY_NUM_CORE = 22, + UVERBS_API_METHOD_IS_WRITE = 30 << UVERBS_API_METHOD_KEY_SHIFT, + UVERBS_API_METHOD_IS_WRITE_EX = 31 << UVERBS_API_METHOD_KEY_SHIFT, + UVERBS_API_METHOD_KEY_NUM_DRIVER = + (UVERBS_API_METHOD_IS_WRITE >> UVERBS_API_METHOD_KEY_SHIFT) - + UVERBS_API_METHOD_KEY_NUM_CORE, UVERBS_API_METHOD_KEY_MASK = GENMASK( UVERBS_API_METHOD_KEY_BITS + UVERBS_API_METHOD_KEY_SHIFT - 1, UVERBS_API_METHOD_KEY_SHIFT), @@ -205,7 +218,22 @@ static inline __attribute_const__ u32 uapi_key_ioctl_method(u32 id) return id << UVERBS_API_METHOD_KEY_SHIFT; } -static inline __attribute_const__ u32 uapi_key_attr_to_method(u32 attr_key) +static inline __attribute_const__ u32 uapi_key_write_method(u32 id) +{ + if (id >= UVERBS_API_WRITE_KEY_NUM) + return UVERBS_API_KEY_ERR; + return UVERBS_API_METHOD_IS_WRITE | id; +} + +static inline __attribute_const__ u32 uapi_key_write_ex_method(u32 id) +{ + if (id >= UVERBS_API_WRITE_KEY_NUM) + return UVERBS_API_KEY_ERR; + return UVERBS_API_METHOD_IS_WRITE_EX | id; +} + +static inline __attribute_const__ u32 +uapi_key_attr_to_ioctl_method(u32 attr_key) { return attr_key & (UVERBS_API_OBJ_KEY_MASK | UVERBS_API_METHOD_KEY_MASK); @@ -213,10 +241,23 @@ static inline __attribute_const__ u32 uapi_key_attr_to_method(u32 attr_key) static inline __attribute_const__ bool uapi_key_is_ioctl_method(u32 key) { - return (key & UVERBS_API_METHOD_KEY_MASK) != 0 && + unsigned int method = key & UVERBS_API_METHOD_KEY_MASK; + + return method != 0 && method < UVERBS_API_METHOD_IS_WRITE && (key & UVERBS_API_ATTR_KEY_MASK) == 0; } +static inline __attribute_const__ bool uapi_key_is_write_method(u32 key) +{ + return (key & UVERBS_API_METHOD_KEY_MASK) == UVERBS_API_METHOD_IS_WRITE; +} + +static inline __attribute_const__ bool uapi_key_is_write_ex_method(u32 key) +{ + return (key & UVERBS_API_METHOD_KEY_MASK) == + UVERBS_API_METHOD_IS_WRITE_EX; +} + static inline __attribute_const__ u32 uapi_key_attrs_start(u32 ioctl_method_key) { /* 0 is the method slot itself */ @@ -246,9 +287,12 @@ static inline __attribute_const__ u32 uapi_key_attr(u32 id) return id; } +/* Only true for ioctl methods */ static inline __attribute_const__ bool uapi_key_is_attr(u32 key) { - return (key & UVERBS_API_METHOD_KEY_MASK) != 0 && + unsigned int method = key & UVERBS_API_METHOD_KEY_MASK; + + return method != 0 && method < UVERBS_API_METHOD_IS_WRITE && (key & UVERBS_API_ATTR_KEY_MASK) != 0; } @@ -285,8 +329,7 @@ struct uverbs_method_def { u32 flags; size_t num_attrs; const struct uverbs_attr_def * const (*attrs)[]; - int (*handler)(struct ib_uverbs_file *ufile, - struct uverbs_attr_bundle *ctx); + int (*handler)(struct uverbs_attr_bundle *attrs); }; struct uverbs_object_def { @@ -296,11 +339,131 @@ struct uverbs_object_def { const struct uverbs_method_def * const (*methods)[]; }; -struct uverbs_object_tree_def { - size_t num_objects; - const struct uverbs_object_def * const (*objects)[]; +enum uapi_definition_kind { + UAPI_DEF_END = 0, + UAPI_DEF_OBJECT_START, + UAPI_DEF_WRITE, + UAPI_DEF_CHAIN_OBJ_TREE, + UAPI_DEF_CHAIN, + UAPI_DEF_IS_SUPPORTED_FUNC, + UAPI_DEF_IS_SUPPORTED_DEV_FN, +}; + +enum uapi_definition_scope { + UAPI_SCOPE_OBJECT = 1, + UAPI_SCOPE_METHOD = 2, }; +struct uapi_definition { + u8 kind; + u8 scope; + union { + struct { + u16 object_id; + } object_start; + struct { + u16 command_num; + u8 is_ex:1; + u8 has_udata:1; + u8 has_resp:1; + u8 req_size; + u8 resp_size; + } write; + }; + + union { + bool (*func_is_supported)(struct ib_device *device); + int (*func_write)(struct uverbs_attr_bundle *attrs); + const struct uapi_definition *chain; + const struct uverbs_object_def *chain_obj_tree; + size_t needs_fn_offset; + }; +}; + +/* Define things connected to object_id */ +#define DECLARE_UVERBS_OBJECT(_object_id, ...) \ + { \ + .kind = UAPI_DEF_OBJECT_START, \ + .object_start = { .object_id = _object_id }, \ + }, \ + ##__VA_ARGS__ + +/* Use in a var_args of DECLARE_UVERBS_OBJECT */ +#define DECLARE_UVERBS_WRITE(_command_num, _func, _cmd_desc, ...) \ + { \ + .kind = UAPI_DEF_WRITE, \ + .scope = UAPI_SCOPE_OBJECT, \ + .write = { .is_ex = 0, .command_num = _command_num }, \ + .func_write = _func, \ + _cmd_desc, \ + }, \ + ##__VA_ARGS__ + +/* Use in a var_args of DECLARE_UVERBS_OBJECT */ +#define DECLARE_UVERBS_WRITE_EX(_command_num, _func, _cmd_desc, ...) \ + { \ + .kind = UAPI_DEF_WRITE, \ + .scope = UAPI_SCOPE_OBJECT, \ + .write = { .is_ex = 1, .command_num = _command_num }, \ + .func_write = _func, \ + _cmd_desc, \ + }, \ + ##__VA_ARGS__ + +/* + * Object is only supported if the function pointer named ibdev_fn in struct + * ib_device is not NULL. + */ +#define UAPI_DEF_OBJ_NEEDS_FN(ibdev_fn) \ + { \ + .kind = UAPI_DEF_IS_SUPPORTED_DEV_FN, \ + .scope = UAPI_SCOPE_OBJECT, \ + .needs_fn_offset = \ + offsetof(struct ib_device, ibdev_fn) + \ + BUILD_BUG_ON_ZERO( \ + sizeof(((struct ib_device *)0)->ibdev_fn) != \ + sizeof(void *)), \ + } + +/* + * Method is only supported if the function pointer named ibdev_fn in struct + * ib_device is not NULL. + */ +#define UAPI_DEF_METHOD_NEEDS_FN(ibdev_fn) \ + { \ + .kind = UAPI_DEF_IS_SUPPORTED_DEV_FN, \ + .scope = UAPI_SCOPE_METHOD, \ + .needs_fn_offset = \ + offsetof(struct ib_device, ibdev_fn) + \ + BUILD_BUG_ON_ZERO( \ + sizeof(((struct ib_device *)0)->ibdev_fn) != \ + sizeof(void *)), \ + } + +/* Call a function to determine if the entire object is supported or not */ +#define UAPI_DEF_IS_OBJ_SUPPORTED(_func) \ + { \ + .kind = UAPI_DEF_IS_SUPPORTED_FUNC, \ + .scope = UAPI_SCOPE_OBJECT, .func_is_supported = _func, \ + } + +/* Include another struct uapi_definition in this one */ +#define UAPI_DEF_CHAIN(_def_var) \ + { \ + .kind = UAPI_DEF_CHAIN, .chain = _def_var, \ + } + +/* Temporary until the tree base description is replaced */ +#define UAPI_DEF_CHAIN_OBJ_TREE(_object_enum, _object_ptr) \ + { \ + .kind = UAPI_DEF_CHAIN_OBJ_TREE, \ + .object_start = { .object_id = _object_enum }, \ + .chain_obj_tree = _object_ptr, \ + } +#define UAPI_DEF_CHAIN_OBJ_TREE_NAMED(_object_enum, ...) \ + UAPI_DEF_CHAIN_OBJ_TREE(_object_enum, &UVERBS_OBJECT(_object_enum)), \ + ##__VA_ARGS__ + /* * ======================================= * Attribute Specifications @@ -361,6 +524,12 @@ struct uverbs_object_tree_def { .u2.objs_arr.max_len = _max_len, \ __VA_ARGS__ } }) +/* + * Only for use with UVERBS_ATTR_IDR, allows any uobject type to be accepted, + * the user must validate the type of the uobject instead. + */ +#define UVERBS_IDR_ANY_OBJECT 0xFFFF + #define UVERBS_ATTR_IDR(_attr_id, _idr_type, _access, ...) \ (&(const struct uverbs_attr_def){ \ .id = _attr_id, \ @@ -433,25 +602,12 @@ struct uverbs_object_tree_def { #define UVERBS_ATTR_UHW() \ UVERBS_ATTR_PTR_IN(UVERBS_ATTR_UHW_IN, \ UVERBS_ATTR_MIN_SIZE(0), \ - UA_OPTIONAL), \ + UA_OPTIONAL, \ + .is_udata = 1), \ UVERBS_ATTR_PTR_OUT(UVERBS_ATTR_UHW_OUT, \ UVERBS_ATTR_MIN_SIZE(0), \ - UA_OPTIONAL) - -/* - * ======================================= - * Declaration helpers - * ======================================= - */ - -#define DECLARE_UVERBS_OBJECT_TREE(_name, ...) \ - static const struct uverbs_object_def *const _name##_ptr[] = { \ - __VA_ARGS__, \ - }; \ - static const struct uverbs_object_tree_def _name = { \ - .num_objects = ARRAY_SIZE(_name##_ptr), \ - .objects = &_name##_ptr, \ - } + UA_OPTIONAL, \ + .is_udata = 1) /* ================================================= * Parsing infrastructure @@ -492,6 +648,8 @@ struct uverbs_attr { }; struct uverbs_attr_bundle { + struct ib_udata driver_udata; + struct ib_udata ucore; struct ib_uverbs_file *ufile; DECLARE_BITMAP(attr_present, UVERBS_API_ATTR_BKEY_LEN); struct uverbs_attr attrs[]; @@ -660,6 +818,12 @@ static inline int _uverbs_copy_from_or_zero(void *to, #define uverbs_copy_from_or_zero(to, attrs_bundle, idx) \ _uverbs_copy_from_or_zero(to, attrs_bundle, idx, sizeof(*to)) +static inline struct ib_ucontext * +ib_uverbs_get_ucontext(const struct uverbs_attr_bundle *attrs) +{ + return ib_uverbs_get_ucontext_file(attrs->ufile); +} + #if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) int uverbs_get_flags64(u64 *to, const struct uverbs_attr_bundle *attrs_bundle, size_t idx, u64 allowed_bits); diff --git a/include/rdma/uverbs_named_ioctl.h b/include/rdma/uverbs_named_ioctl.h index b3b21733cc55..3447bfe356d6 100644 --- a/include/rdma/uverbs_named_ioctl.h +++ b/include/rdma/uverbs_named_ioctl.h @@ -43,7 +43,7 @@ #define _UVERBS_NAME(x, y) _UVERBS_PASTE(x, y) #define UVERBS_METHOD(id) _UVERBS_NAME(UVERBS_MODULE_NAME, _method_##id) #define UVERBS_HANDLER(id) _UVERBS_NAME(UVERBS_MODULE_NAME, _handler_##id) -#define UVERBS_OBJECT(id) _UVERBS_NAME(UVERBS_MOUDLE_NAME, _object_##id) +#define UVERBS_OBJECT(id) _UVERBS_NAME(UVERBS_MODULE_NAME, _object_##id) /* These are static so they do not need to be qualified */ #define UVERBS_METHOD_ATTRS(method_id) _method_attrs_##method_id @@ -102,18 +102,11 @@ #define ADD_UVERBS_METHODS(_name, _object_id, ...) \ static const struct uverbs_method_def *const UVERBS_OBJECT_METHODS( \ _object_id)[] = { __VA_ARGS__ }; \ - static const struct uverbs_object_def _name##_struct = { \ + static const struct uverbs_object_def _name = { \ .id = _object_id, \ .num_methods = ARRAY_SIZE(UVERBS_OBJECT_METHODS(_object_id)), \ .methods = &UVERBS_OBJECT_METHODS(_object_id) \ - }; \ - static const struct uverbs_object_def *const _name##_ptrs[] = { \ - &_name##_struct, \ - }; \ - static const struct uverbs_object_tree_def _name = { \ - .num_objects = 1, \ - .objects = &_name##_ptrs, \ - } + }; /* Used by drivers to declare a complete parsing tree for a single method that * differs only in having additional driver specific attributes. diff --git a/include/rdma/uverbs_std_types.h b/include/rdma/uverbs_std_types.h index 3db2802fbc68..883abcf6d36e 100644 --- a/include/rdma/uverbs_std_types.h +++ b/include/rdma/uverbs_std_types.h @@ -37,15 +37,6 @@ #include <rdma/uverbs_ioctl.h> #include <rdma/ib_user_ioctl_verbs.h> -#if IS_ENABLED(CONFIG_INFINIBAND_USER_ACCESS) -const struct uverbs_object_tree_def *uverbs_default_get_objects(void); -#else -static inline const struct uverbs_object_tree_def *uverbs_default_get_objects(void) -{ - return NULL; -} -#endif - /* Returns _id, or causes a compile error if _id is not a u32. * * The uobj APIs should only be used with the write based uAPI to access @@ -54,15 +45,15 @@ static inline const struct uverbs_object_tree_def *uverbs_default_get_objects(vo */ #define _uobj_check_id(_id) ((_id) * typecheck(u32, _id)) -#define uobj_get_type(_ufile, _object) \ - uapi_get_object((_ufile)->device->uapi, _object) +#define uobj_get_type(_attrs, _object) \ + uapi_get_object((_attrs)->ufile->device->uapi, _object) -#define uobj_get_read(_type, _id, _ufile) \ - rdma_lookup_get_uobject(uobj_get_type(_ufile, _type), _ufile, \ +#define uobj_get_read(_type, _id, _attrs) \ + rdma_lookup_get_uobject(uobj_get_type(_attrs, _type), (_attrs)->ufile, \ _uobj_check_id(_id), UVERBS_LOOKUP_READ) -#define ufd_get_read(_type, _fdnum, _ufile) \ - rdma_lookup_get_uobject(uobj_get_type(_ufile, _type), _ufile, \ +#define ufd_get_read(_type, _fdnum, _attrs) \ + rdma_lookup_get_uobject(uobj_get_type(_attrs, _type), (_attrs)->ufile, \ (_fdnum)*typecheck(s32, _fdnum), \ UVERBS_LOOKUP_READ) @@ -72,26 +63,27 @@ static inline void *_uobj_get_obj_read(struct ib_uobject *uobj) return NULL; return uobj->object; } -#define uobj_get_obj_read(_object, _type, _id, _ufile) \ +#define uobj_get_obj_read(_object, _type, _id, _attrs) \ ((struct ib_##_object *)_uobj_get_obj_read( \ - uobj_get_read(_type, _id, _ufile))) + uobj_get_read(_type, _id, _attrs))) -#define uobj_get_write(_type, _id, _ufile) \ - rdma_lookup_get_uobject(uobj_get_type(_ufile, _type), _ufile, \ +#define uobj_get_write(_type, _id, _attrs) \ + rdma_lookup_get_uobject(uobj_get_type(_attrs, _type), (_attrs)->ufile, \ _uobj_check_id(_id), UVERBS_LOOKUP_WRITE) int __uobj_perform_destroy(const struct uverbs_api_object *obj, u32 id, - struct ib_uverbs_file *ufile, int success_res); -#define uobj_perform_destroy(_type, _id, _ufile, _success_res) \ - __uobj_perform_destroy(uobj_get_type(_ufile, _type), \ - _uobj_check_id(_id), _ufile, _success_res) + const struct uverbs_attr_bundle *attrs); +#define uobj_perform_destroy(_type, _id, _attrs) \ + __uobj_perform_destroy(uobj_get_type(_attrs, _type), \ + _uobj_check_id(_id), _attrs) struct ib_uobject *__uobj_get_destroy(const struct uverbs_api_object *obj, - u32 id, struct ib_uverbs_file *ufile); + u32 id, + const struct uverbs_attr_bundle *attrs); -#define uobj_get_destroy(_type, _id, _ufile) \ - __uobj_get_destroy(uobj_get_type(_ufile, _type), _uobj_check_id(_id), \ - _ufile) +#define uobj_get_destroy(_type, _id, _attrs) \ + __uobj_get_destroy(uobj_get_type(_attrs, _type), _uobj_check_id(_id), \ + _attrs) static inline void uobj_put_destroy(struct ib_uobject *uobj) { @@ -111,14 +103,13 @@ static inline void uobj_put_write(struct ib_uobject *uobj) rdma_lookup_put_uobject(uobj, UVERBS_LOOKUP_WRITE); } -static inline int __must_check uobj_alloc_commit(struct ib_uobject *uobj, - int success_res) +static inline int __must_check uobj_alloc_commit(struct ib_uobject *uobj) { int ret = rdma_alloc_commit_uobject(uobj); if (ret) return ret; - return success_res; + return 0; } static inline void uobj_alloc_abort(struct ib_uobject *uobj) @@ -127,18 +118,18 @@ static inline void uobj_alloc_abort(struct ib_uobject *uobj) } static inline struct ib_uobject * -__uobj_alloc(const struct uverbs_api_object *obj, struct ib_uverbs_file *ufile, - struct ib_device **ib_dev) +__uobj_alloc(const struct uverbs_api_object *obj, + struct uverbs_attr_bundle *attrs, struct ib_device **ib_dev) { - struct ib_uobject *uobj = rdma_alloc_begin_uobject(obj, ufile); + struct ib_uobject *uobj = rdma_alloc_begin_uobject(obj, attrs->ufile); if (!IS_ERR(uobj)) *ib_dev = uobj->context->device; return uobj; } -#define uobj_alloc(_type, _ufile, _ib_dev) \ - __uobj_alloc(uobj_get_type(_ufile, _type), _ufile, _ib_dev) +#define uobj_alloc(_type, _attrs, _ib_dev) \ + __uobj_alloc(uobj_get_type(_attrs, _type), _attrs, _ib_dev) static inline void uverbs_flow_action_fill_action(struct ib_flow_action *action, struct ib_uobject *uobj, @@ -191,5 +182,17 @@ static inline void ib_set_flow(struct ib_uobject *uobj, struct ib_flow *ibflow, uflow->resources = uflow_res; } +struct uverbs_api_object { + const struct uverbs_obj_type *type_attrs; + const struct uverbs_obj_type_class *type_class; + u8 disabled:1; + u32 id; +}; + +static inline u32 uobj_get_object_id(struct ib_uobject *uobj) +{ + return uobj->uapi_object->id; +} + #endif diff --git a/include/uapi/rdma/hfi/hfi1_user.h b/include/uapi/rdma/hfi/hfi1_user.h index c6a984c0c881..01ac5853d9ac 100644 --- a/include/uapi/rdma/hfi/hfi1_user.h +++ b/include/uapi/rdma/hfi/hfi1_user.h @@ -6,7 +6,7 @@ * * GPL LICENSE SUMMARY * - * Copyright(c) 2015 Intel Corporation. + * Copyright(c) 2015 - 2018 Intel Corporation. * * This program is free software; you can redistribute it and/or modify * it under the terms of version 2 of the GNU General Public License as @@ -95,7 +95,7 @@ #define HFI1_CAP_SDMA_AHG (1UL << 2) /* Enable SDMA AHG support */ #define HFI1_CAP_EXTENDED_PSN (1UL << 3) /* Enable Extended PSN support */ #define HFI1_CAP_HDRSUPP (1UL << 4) /* Enable Header Suppression */ -/* 1UL << 5 unused */ +#define HFI1_CAP_TID_RDMA (1UL << 5) /* Enable TID RDMA operations */ #define HFI1_CAP_USE_SDMA_HEAD (1UL << 6) /* DMA Hdr Q tail vs. use CSR */ #define HFI1_CAP_MULTI_PKT_EGR (1UL << 7) /* Enable multi-packet Egr buffs*/ #define HFI1_CAP_NODROP_RHQ_FULL (1UL << 8) /* Don't drop on Hdr Q full */ @@ -106,7 +106,7 @@ #define HFI1_CAP_NO_INTEGRITY (1UL << 13) /* Enable ctxt integrity checks */ #define HFI1_CAP_PKEY_CHECK (1UL << 14) /* Enable ctxt PKey checking */ #define HFI1_CAP_STATIC_RATE_CTRL (1UL << 15) /* Allow PBC.StaticRateControl */ -/* 1UL << 16 unused */ +#define HFI1_CAP_OPFN (1UL << 16) /* Enable the OPFN protocol */ #define HFI1_CAP_SDMA_HEAD_CHECK (1UL << 17) /* SDMA head checking */ #define HFI1_CAP_EARLY_CREDIT_RETURN (1UL << 18) /* early credit return */ diff --git a/include/uapi/rdma/hns-abi.h b/include/uapi/rdma/hns-abi.h index c1f87735514f..ef3c7ec793a7 100644 --- a/include/uapi/rdma/hns-abi.h +++ b/include/uapi/rdma/hns-abi.h @@ -46,6 +46,12 @@ struct hns_roce_ib_create_cq_resp { __aligned_u64 cap_flags; }; +struct hns_roce_ib_create_srq { + __aligned_u64 buf_addr; + __aligned_u64 db_addr; + __aligned_u64 que_addr; +}; + struct hns_roce_ib_create_qp { __aligned_u64 buf_addr; __aligned_u64 db_addr; diff --git a/include/uapi/rdma/ib_user_verbs.h b/include/uapi/rdma/ib_user_verbs.h index 1254b51a551a..c586fc43739c 100644 --- a/include/uapi/rdma/ib_user_verbs.h +++ b/include/uapi/rdma/ib_user_verbs.h @@ -164,6 +164,7 @@ struct ib_uverbs_get_context { struct ib_uverbs_get_context_resp { __u32 async_fd; __u32 num_comp_vectors; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_query_device { @@ -310,6 +311,7 @@ struct ib_uverbs_alloc_pd { struct ib_uverbs_alloc_pd_resp { __u32 pd_handle; + __u32 driver_data[0]; }; struct ib_uverbs_dealloc_pd { @@ -325,6 +327,7 @@ struct ib_uverbs_open_xrcd { struct ib_uverbs_open_xrcd_resp { __u32 xrcd_handle; + __u32 driver_data[0]; }; struct ib_uverbs_close_xrcd { @@ -345,6 +348,7 @@ struct ib_uverbs_reg_mr_resp { __u32 mr_handle; __u32 lkey; __u32 rkey; + __u32 driver_data[0]; }; struct ib_uverbs_rereg_mr { @@ -356,11 +360,13 @@ struct ib_uverbs_rereg_mr { __aligned_u64 hca_va; __u32 pd_handle; __u32 access_flags; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_rereg_mr_resp { __u32 lkey; __u32 rkey; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_dereg_mr { @@ -372,11 +378,13 @@ struct ib_uverbs_alloc_mw { __u32 pd_handle; __u8 mw_type; __u8 reserved[3]; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_alloc_mw_resp { __u32 mw_handle; __u32 rkey; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_dealloc_mw { @@ -419,6 +427,7 @@ struct ib_uverbs_ex_create_cq { struct ib_uverbs_create_cq_resp { __u32 cq_handle; __u32 cqe; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_ex_create_cq_resp { @@ -629,6 +638,7 @@ struct ib_uverbs_create_qp_resp { __u32 max_recv_sge; __u32 max_inline_data; __u32 reserved; + __u32 driver_data[0]; }; struct ib_uverbs_ex_create_qp_resp { @@ -733,9 +743,6 @@ struct ib_uverbs_ex_modify_qp { __u32 reserved; }; -struct ib_uverbs_modify_qp_resp { -}; - struct ib_uverbs_ex_modify_qp_resp { __u32 comp_mask; __u32 response_length; @@ -863,10 +870,12 @@ struct ib_uverbs_create_ah { __u32 pd_handle; __u32 reserved; struct ib_uverbs_ah_attr attr; + __aligned_u64 driver_data[0]; }; struct ib_uverbs_create_ah_resp { __u32 ah_handle; + __u32 driver_data[0]; }; struct ib_uverbs_destroy_ah { @@ -1175,6 +1184,7 @@ struct ib_uverbs_create_srq_resp { __u32 max_wr; __u32 max_sge; __u32 srqn; + __u32 driver_data[0]; }; struct ib_uverbs_modify_srq { diff --git a/include/uapi/rdma/mlx5-abi.h b/include/uapi/rdma/mlx5-abi.h index 8fa9f90e2bb1..87b3198f4b5d 100644 --- a/include/uapi/rdma/mlx5-abi.h +++ b/include/uapi/rdma/mlx5-abi.h @@ -48,6 +48,7 @@ enum { MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_UC = 1 << 6, MLX5_QP_FLAG_TIR_ALLOW_SELF_LB_MC = 1 << 7, MLX5_QP_FLAG_ALLOW_SCATTER_CQE = 1 << 8, + MLX5_QP_FLAG_PACKET_BASED_CREDIT_MODE = 1 << 9, }; enum { @@ -236,6 +237,7 @@ enum mlx5_ib_query_dev_resp_flags { /* Support 128B CQE compression */ MLX5_IB_QUERY_DEV_RESP_FLAGS_CQE_128B_COMP = 1 << 0, MLX5_IB_QUERY_DEV_RESP_FLAGS_CQE_128B_PAD = 1 << 1, + MLX5_IB_QUERY_DEV_RESP_PACKET_BASED_CREDIT_MODE = 1 << 2, }; enum mlx5_ib_tunnel_offloads { diff --git a/include/uapi/rdma/mlx5_user_ioctl_cmds.h b/include/uapi/rdma/mlx5_user_ioctl_cmds.h index 408e220034de..b8d121d457f1 100644 --- a/include/uapi/rdma/mlx5_user_ioctl_cmds.h +++ b/include/uapi/rdma/mlx5_user_ioctl_cmds.h @@ -158,6 +158,7 @@ enum mlx5_ib_create_flow_attrs { MLX5_IB_ATTR_CREATE_FLOW_MATCHER, MLX5_IB_ATTR_CREATE_FLOW_ARR_FLOW_ACTIONS, MLX5_IB_ATTR_CREATE_FLOW_TAG, + MLX5_IB_ATTR_CREATE_FLOW_ARR_COUNTERS_DEVX, }; enum mlx5_ib_destoy_flow_attrs { |