The branch main has been updated by jhb: URL: https://cgit.FreeBSD.org/src/commit/?id=85893a8399bce2dd66e8693037395e1979277967
commit 85893a8399bce2dd66e8693037395e1979277967 Author: John Baldwin <[email protected]> AuthorDate: 2026-06-24 19:30:55 +0000 Commit: John Baldwin <[email protected]> CommitDate: 2026-06-24 19:30:55 +0000 OFED: Various changes from Linux 4.15 This contains changes from the following Linux commits: c0b64f58e8d4 RDMA/cma: Avoid triggering undefined behavior c0348eb06968 IB: Let ib_core resolve destination mac address 5a3dc3237243 IB/cm: Fix memory corruption in handling CM request fec99ededf6b RDMA/umem: Avoid partial declaration of non-static function 19b57c6c4499 IB/core: Convert OPA AH to IB for Extended LIDs only e08ce2e82b2f RDMA/core: Make function rdma_copy_addr return void 4190b4e96954 RDMA/core: Rename kernel modify_cq to better describe its usage Tested by: Wafa Hamzah <[email protected]> (mlx5_ib) Tested by: John Baldwin <[email protected]> (iw_cxgbe) Sponsored by: Chelsio Communications --- sys/dev/bnxt/bnxt_re/bnxt_re.h | 14 --- sys/dev/bnxt/bnxt_re/ib_verbs.c | 70 +------------ sys/dev/bnxt/bnxt_re/ib_verbs.h | 1 - sys/dev/bnxt/bnxt_re/main.c | 22 ---- sys/dev/mlx4/mlx4_ib/mlx4_ib_ah.c | 8 +- sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c | 4 - sys/ofed/drivers/infiniband/core/ib_addr.c | 22 ++-- sys/ofed/drivers/infiniband/core/ib_cm.c | 11 +- sys/ofed/drivers/infiniband/core/ib_cma.c | 19 ++-- sys/ofed/drivers/infiniband/core/ib_umem_odp.c | 72 +++++++++++++ sys/ofed/drivers/infiniband/core/ib_umem_rbtree.c | 111 --------------------- sys/ofed/drivers/infiniband/core/ib_uverbs_cmd.c | 2 +- .../drivers/infiniband/core/ib_uverbs_marshall.c | 3 +- sys/ofed/drivers/infiniband/core/ib_verbs.c | 12 ++- .../drivers/infiniband/ulp/ipoib/ipoib_ethtool.c | 5 +- sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c | 2 +- sys/ofed/include/rdma/ib_addr.h | 5 +- sys/ofed/include/rdma/ib_umem_odp.h | 4 - sys/ofed/include/rdma/ib_verbs.h | 7 +- 19 files changed, 123 insertions(+), 271 deletions(-) diff --git a/sys/dev/bnxt/bnxt_re/bnxt_re.h b/sys/dev/bnxt/bnxt_re/bnxt_re.h index 8c78d78e742a..2e7f027380be 100644 --- a/sys/dev/bnxt/bnxt_re/bnxt_re.h +++ b/sys/dev/bnxt/bnxt_re/bnxt_re.h @@ -558,8 +558,6 @@ struct bnxt_re_dev { struct dentry *pdev_debug_dir; struct dentry *pdev_qpinfo_dir; struct bnxt_re_debug_entries *dbg_ent; - struct workqueue_struct *resolve_wq; - struct list_head mac_wq_list; struct workqueue_struct *dcb_wq; struct workqueue_struct *aer_wq; u32 event_bitmap[3]; @@ -608,16 +606,6 @@ struct bnxt_re_dev { unsigned char *dev_addr; /* For netdev->dev_addr */ }; -#define BNXT_RE_RESOLVE_RETRY_COUNT_US 5000000 /* 5 sec */ -struct bnxt_re_resolve_dmac_work{ - struct work_struct work; - struct list_head list; - struct bnxt_re_dev *rdev; - struct rdma_ah_attr *ah_attr; - struct bnxt_re_ah_info *ah_info; - atomic_t status_wait; -}; - static inline u8 bnxt_re_get_prio(u8 prio_map) { u8 prio = 0xFF; @@ -727,8 +715,6 @@ int bnxt_re_schedule_work(struct bnxt_re_dev *rdev, unsigned long event, void bnxt_re_get_link_speed(struct bnxt_re_dev *rdev); int _bnxt_re_ib_init(struct bnxt_re_dev *rdev); int _bnxt_re_ib_init2(struct bnxt_re_dev *rdev); -void bnxt_re_init_resolve_wq(struct bnxt_re_dev *rdev); -void bnxt_re_uninit_resolve_wq(struct bnxt_re_dev *rdev); /* The rdev ref_count is to protect immature removal of the device */ static inline void bnxt_re_hold(struct bnxt_re_dev *rdev) diff --git a/sys/dev/bnxt/bnxt_re/ib_verbs.c b/sys/dev/bnxt/bnxt_re/ib_verbs.c index 55d8d8b5b04a..eacd5c1abab5 100644 --- a/sys/dev/bnxt/bnxt_re/ib_verbs.c +++ b/sys/dev/bnxt/bnxt_re/ib_verbs.c @@ -49,23 +49,6 @@ static inline void bnxt_re_peer_mem_release(struct ib_umem *umem) ib_umem_release(umem); } -void bnxt_re_resolve_dmac_task(struct work_struct *work) -{ - int rc = -1; - struct bnxt_re_dev *rdev; - struct rdma_ah_attr *ah_attr; - struct bnxt_re_resolve_dmac_work *dmac_work = - container_of(work, struct bnxt_re_resolve_dmac_work, work); - - rdev = dmac_work->rdev; - ah_attr = dmac_work->ah_attr; - rc = ib_resolve_eth_dmac(&rdev->ibdev, ah_attr); - if (rc) - dev_err(rdev_to_dev(dmac_work->rdev), - "Failed to resolve dest mac rc = %d\n", rc); - atomic_set(&dmac_work->status_wait, rc << 8); -} - static int __from_ib_access_flags(int iflags) { int qflags = 0; @@ -999,52 +982,6 @@ static u8 _get_sgid_index(struct bnxt_re_dev *rdev, u8 gindx) return gindx; } -static int bnxt_re_init_dmac(struct bnxt_re_dev *rdev, struct rdma_ah_attr *ah_attr, - struct bnxt_re_ah_info *ah_info, bool is_user, - struct bnxt_re_ah *ah) -{ - const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr); - int rc = 0; - u8 *dmac; - - if (is_user && !rdma_is_multicast_addr((struct in6_addr *) - grh->dgid.raw) && - !rdma_link_local_addr((struct in6_addr *)grh->dgid.raw)) { - - u32 retry_count = BNXT_RE_RESOLVE_RETRY_COUNT_US; - struct bnxt_re_resolve_dmac_work *resolve_dmac_work; - - - resolve_dmac_work = kzalloc(sizeof(*resolve_dmac_work), GFP_ATOMIC); - - resolve_dmac_work->rdev = rdev; - resolve_dmac_work->ah_attr = ah_attr; - resolve_dmac_work->ah_info = ah_info; - - atomic_set(&resolve_dmac_work->status_wait, 1); - INIT_WORK(&resolve_dmac_work->work, bnxt_re_resolve_dmac_task); - queue_work(rdev->resolve_wq, &resolve_dmac_work->work); - - do { - rc = atomic_read(&resolve_dmac_work->status_wait) & 0xFF; - if (!rc) - break; - udelay(1); - } while (--retry_count); - if (atomic_read(&resolve_dmac_work->status_wait)) { - INIT_LIST_HEAD(&resolve_dmac_work->list); - list_add_tail(&resolve_dmac_work->list, - &rdev->mac_wq_list); - return -EFAULT; - } - kfree(resolve_dmac_work); - } - dmac = ROCE_DMAC(ah_attr); - if (dmac) - memcpy(ah->qplib_ah.dmac, dmac, ETH_ALEN); - return rc; -} - int bnxt_re_create_ah(struct ib_ah *ah_in, struct rdma_ah_attr *attr, u32 flags, struct ib_udata *udata) { @@ -1057,7 +994,6 @@ int bnxt_re_create_ah(struct ib_ah *ah_in, struct rdma_ah_attr *attr, const struct ib_global_route *grh = rdma_ah_read_grh(attr); struct bnxt_re_ah_info ah_info; u32 max_ah_count; - bool is_user; int rc; bool block = true; struct rdma_ah_attr *ah_attr = attr; @@ -1068,7 +1004,6 @@ int bnxt_re_create_ah(struct ib_ah *ah_in, struct rdma_ah_attr *attr, ah->rdev = rdev; ah->qplib_ah.pd = &pd->qplib_pd; - is_user = ib_pd->uobject ? true : false; /* Supply the configuration for the HW */ memcpy(ah->qplib_ah.dgid.data, grh->dgid.raw, @@ -1088,10 +1023,7 @@ int bnxt_re_create_ah(struct ib_ah *ah_in, struct rdma_ah_attr *attr, if (rc) goto fail; ah->qplib_ah.nw_type = ah_info.nw_type; - - rc = bnxt_re_init_dmac(rdev, ah_attr, &ah_info, is_user, ah); - if (rc) - goto fail; + memcpy(ah->qplib_ah.dmac, ROCE_DMAC(ah_attr), ETH_ALEN); rc = bnxt_qplib_create_ah(&rdev->qplib_res, &ah->qplib_ah, block); if (rc) { diff --git a/sys/dev/bnxt/bnxt_re/ib_verbs.h b/sys/dev/bnxt/bnxt_re/ib_verbs.h index 7dd8f6ece12f..332c88e07a3d 100644 --- a/sys/dev/bnxt/bnxt_re/ib_verbs.h +++ b/sys/dev/bnxt/bnxt_re/ib_verbs.h @@ -398,7 +398,6 @@ struct ib_umem *ib_umem_get_flags_compat(struct bnxt_re_dev *rdev, size_t size, int access, int dmasync); static inline size_t ib_umem_num_pages_compat(struct ib_umem *umem); static inline void bnxt_re_peer_mem_release(struct ib_umem *umem); -void bnxt_re_resolve_dmac_task(struct work_struct *work); static inline enum ib_qp_type __from_hw_to_ib_qp_type(u8 type) { diff --git a/sys/dev/bnxt/bnxt_re/main.c b/sys/dev/bnxt/bnxt_re/main.c index 5b1fe2ae0faa..991931bdd396 100644 --- a/sys/dev/bnxt/bnxt_re/main.c +++ b/sys/dev/bnxt/bnxt_re/main.c @@ -137,26 +137,6 @@ void bnxt_re_set_dma_device(struct ib_device *ibdev, struct bnxt_re_dev *rdev) ibdev->dma_device = &rdev->en_dev->pdev->dev; } -void bnxt_re_init_resolve_wq(struct bnxt_re_dev *rdev) -{ - rdev->resolve_wq = create_singlethread_workqueue("bnxt_re_resolve_wq"); - INIT_LIST_HEAD(&rdev->mac_wq_list); -} - -void bnxt_re_uninit_resolve_wq(struct bnxt_re_dev *rdev) -{ - struct bnxt_re_resolve_dmac_work *tmp_work = NULL, *tmp_st; - if (!rdev->resolve_wq) - return; - flush_workqueue(rdev->resolve_wq); - list_for_each_entry_safe(tmp_work, tmp_st, &rdev->mac_wq_list, list) { - list_del(&tmp_work->list); - kfree(tmp_work); - } - destroy_workqueue(rdev->resolve_wq); - rdev->resolve_wq = NULL; -} - u32 readl_fbsd(struct bnxt_softc *bp, u32 reg_off, u8 bar_idx) { @@ -3534,7 +3514,6 @@ static void bnxt_re_dev_uninit(struct bnxt_re_dev *rdev, u8 op_type) mutex_unlock(&bnxt_re_dev_lock); } - bnxt_re_uninit_resolve_wq(rdev); bnxt_re_uninit_dcb_wq(rdev); bnxt_re_uninit_aer_wq(rdev); @@ -3804,7 +3783,6 @@ static int bnxt_re_dev_init(struct bnxt_re_dev *rdev, u8 op_type) bnxt_re_init_dcb_wq(rdev); bnxt_re_init_aer_wq(rdev); - bnxt_re_init_resolve_wq(rdev); mutex_lock(&bnxt_re_dev_lock); list_add_tail_rcu(&rdev->list, &bnxt_re_dev_list); /* Added to the list, not in progress anymore */ diff --git a/sys/dev/mlx4/mlx4_ib/mlx4_ib_ah.c b/sys/dev/mlx4/mlx4_ib/mlx4_ib_ah.c index 6bb17975143f..e31c223c27e6 100644 --- a/sys/dev/mlx4/mlx4_ib/mlx4_ib_ah.c +++ b/sys/dev/mlx4/mlx4_ib/mlx4_ib_ah.c @@ -90,12 +90,10 @@ static int create_iboe_ah(struct ib_ah *ib_ah, struct rdma_ah_attr *ah_attr) int ret; memcpy(&in6, grh->dgid.raw, sizeof(in6)); - if (rdma_is_multicast_addr(&in6)) { + if (rdma_is_multicast_addr(&in6)) is_mcast = 1; - rdma_get_mcast_mac(&in6, ah->av.eth.mac); - } else { - memcpy(ah->av.eth.mac, ah_attr->roce.dmac, ETH_ALEN); - } + + memcpy(ah->av.eth.mac, ah_attr->roce.dmac, ETH_ALEN); ret = ib_get_cached_gid(pd->device, rdma_ah_get_port_num(ah_attr), grh->sgid_index, &sgid, &gid_attr); if (ret) diff --git a/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c b/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c index c42126241050..7439ec3fa4ea 100644 --- a/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c +++ b/sys/dev/mlx5/mlx5_ib/mlx5_ib_ah.c @@ -83,10 +83,6 @@ int mlx5_ib_create_ah(struct ib_ah *ibah, struct rdma_ah_attr *ah_attr, resp.response_length = min_resp_len; - err = ib_resolve_eth_dmac(&dev->ib_dev, ah_attr); - if (err) - return err; - memcpy(resp.dmac, ah_attr->roce.dmac, ETH_ALEN); err = ib_copy_to_udata(udata, &resp, resp.response_length); if (err) diff --git a/sys/ofed/drivers/infiniband/core/ib_addr.c b/sys/ofed/drivers/infiniband/core/ib_addr.c index c953ca604588..ecacfd6a43dd 100644 --- a/sys/ofed/drivers/infiniband/core/ib_addr.c +++ b/sys/ofed/drivers/infiniband/core/ib_addr.c @@ -139,8 +139,9 @@ rdma_copy_addr_sub(u8 *dst, const u8 *src, unsigned min, unsigned max) memset(dst + min, 0, max - min); } -int rdma_copy_addr(struct rdma_dev_addr *dev_addr, if_t dev, - const unsigned char *dst_dev_addr) +void rdma_copy_addr(struct rdma_dev_addr *dev_addr, + const if_t dev, + const unsigned char *dst_dev_addr) { int dev_type = if_gettype(dev); @@ -151,7 +152,7 @@ int rdma_copy_addr(struct rdma_dev_addr *dev_addr, if_t dev, memset(dev_addr->broadcast, 0, MAX_ADDR_LEN); memset(dev_addr->dst_dev_addr, 0, MAX_ADDR_LEN); dev_addr->bound_dev_if = if_getindex(dev); - return (0); + return; } else if (dev_type == IFT_INFINIBAND) dev_addr->dev_type = ARPHRD_INFINIBAND; else if (dev_type == IFT_ETHER || dev_type == IFT_L2VLAN) @@ -167,7 +168,6 @@ int rdma_copy_addr(struct rdma_dev_addr *dev_addr, if_t dev, if_getaddrlen(dev), MAX_ADDR_LEN); } dev_addr->bound_dev_if = if_getindex(dev); - return 0; } EXPORT_SYMBOL(rdma_copy_addr); @@ -175,7 +175,7 @@ int rdma_translate_ip(const struct sockaddr *addr, struct rdma_dev_addr *dev_addr) { if_t dev; - int ret; + int ret = 0; if (dev_addr->bound_dev_if) { dev = dev_get_by_index(dev_addr->net, dev_addr->bound_dev_if); @@ -202,7 +202,7 @@ int rdma_translate_ip(const struct sockaddr *addr, if (if_getflags(dev) & IFF_LOOPBACK) ret = -EINVAL; else - ret = rdma_copy_addr(dev_addr, dev, NULL); + rdma_copy_addr(dev_addr, dev, NULL); dev_put(dev); } else { ret = -ENODEV; @@ -655,10 +655,14 @@ static int addr_resolve_neigh(if_t dev, } /* If the device doesn't do ARP internally */ - if (!(if_getflags(dev) & IFF_NOARP)) - return rdma_copy_addr(addr, dev, edst); + if (!(if_getflags(dev) & IFF_NOARP)) { + rdma_copy_addr(addr, dev, edst); + return 0; + } - return rdma_copy_addr(addr, dev, NULL); + rdma_copy_addr(addr, dev, NULL); + + return 0; } static int addr_resolve(struct sockaddr *src_in, diff --git a/sys/ofed/drivers/infiniband/core/ib_cm.c b/sys/ofed/drivers/infiniband/core/ib_cm.c index 17d483f565ce..be29881c73c9 100644 --- a/sys/ofed/drivers/infiniband/core/ib_cm.c +++ b/sys/ofed/drivers/infiniband/core/ib_cm.c @@ -1601,7 +1601,7 @@ static void cm_format_req_event(struct cm_work *work, param->bth_pkey = cm_get_bth_pkey(work); param->port = cm_id_priv->av.port->port_num; param->primary_path = &work->path[0]; - if (req_msg->alt_local_lid) + if (cm_req_has_alt_path(req_msg)) param->alternate_path = &work->path[1]; else param->alternate_path = NULL; @@ -1884,7 +1884,8 @@ static int cm_req_handler(struct cm_work *work) cm_process_routed_req(req_msg, work->mad_recv_wc->wc); memset(&work->path[0], 0, sizeof(work->path[0])); - memset(&work->path[1], 0, sizeof(work->path[1])); + if (cm_req_has_alt_path(req_msg)) + memset(&work->path[1], 0, sizeof(work->path[1])); grh = rdma_ah_read_grh(&cm_id_priv->av.ah_attr); ret = ib_get_cached_gid(work->port->cm_dev->ib_device, work->port->port_num, @@ -3868,14 +3869,16 @@ static void cm_recv_handler(struct ib_mad_agent *mad_agent, struct cm_port *port = mad_agent->context; struct cm_work *work; enum ib_cm_event_type event; + bool alt_path = false; u16 attr_id; int paths = 0; int going_down = 0; switch (mad_recv_wc->recv_buf.mad->mad_hdr.attr_id) { case CM_REQ_ATTR_ID: - paths = 1 + (((struct cm_req_msg *) mad_recv_wc->recv_buf.mad)-> - alt_local_lid != 0); + alt_path = cm_req_has_alt_path((struct cm_req_msg *) + mad_recv_wc->recv_buf.mad); + paths = 1 + (alt_path != 0); event = IB_CM_REQ_RECEIVED; break; case CM_MRA_ATTR_ID: diff --git a/sys/ofed/drivers/infiniband/core/ib_cma.c b/sys/ofed/drivers/infiniband/core/ib_cma.c index ee7c47863aff..33a4189804e1 100644 --- a/sys/ofed/drivers/infiniband/core/ib_cma.c +++ b/sys/ofed/drivers/infiniband/core/ib_cma.c @@ -1823,7 +1823,7 @@ there_is_no_net_dev: return id_priv; } -static inline int cma_user_data_offset(struct rdma_id_private *id_priv) +static inline u8 cma_user_data_offset(struct rdma_id_private *id_priv) { if (cma_family(id_priv) == AF_IB) return 0; @@ -2144,9 +2144,7 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, rt->path_rec[1] = *ib_event->param.req_rcvd.alternate_path; if (net_dev) { - ret = rdma_copy_addr(&rt->addr.dev_addr, net_dev, NULL); - if (ret) - goto err; + rdma_copy_addr(&rt->addr.dev_addr, net_dev, NULL); } else { if (!cma_protocol_roce(listen_id) && cma_any_addr(vnet, cma_src_addr(id_priv))) { @@ -2192,9 +2190,7 @@ static struct rdma_id_private *cma_new_udp_id(struct rdma_cm_id *listen_id, goto err; if (net_dev) { - ret = rdma_copy_addr(&id->route.addr.dev_addr, net_dev, NULL); - if (ret) - goto err; + rdma_copy_addr(&id->route.addr.dev_addr, net_dev, NULL); } else { if (!cma_any_addr(vnet, cma_src_addr(id_priv))) { ret = cma_translate_addr(cma_src_addr(id_priv), @@ -2240,7 +2236,8 @@ static int cma_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *ib_event) struct rdma_id_private *listen_id, *conn_id = NULL; struct rdma_cm_event event; if_t net_dev; - int offset, ret; + u8 offset; + int ret; listen_id = cma_id_from_event(cm_id, ib_event, &net_dev); if (IS_ERR(listen_id)) @@ -3788,7 +3785,8 @@ static int cma_resolve_ib_udp(struct rdma_id_private *id_priv, struct ib_cm_sidr_req_param req; struct ib_cm_id *id; void *private_data; - int offset, ret; + u8 offset; + int ret; memset(&req, 0, sizeof req); offset = cma_user_data_offset(id_priv); @@ -3845,7 +3843,8 @@ static int cma_connect_ib(struct rdma_id_private *id_priv, struct rdma_route *route; void *private_data; struct ib_cm_id *id; - int offset, ret; + u8 offset; + int ret; memset(&req, 0, sizeof req); offset = cma_user_data_offset(id_priv); diff --git a/sys/ofed/drivers/infiniband/core/ib_umem_odp.c b/sys/ofed/drivers/infiniband/core/ib_umem_odp.c index 9e15a5bfa981..1b469f0a086e 100644 --- a/sys/ofed/drivers/infiniband/core/ib_umem_odp.c +++ b/sys/ofed/drivers/infiniband/core/ib_umem_odp.c @@ -38,11 +38,44 @@ #include <linux/slab.h> #include <linux/vmalloc.h> #include <linux/hugetlb.h> +#include <linux/interval_tree_generic.h> #include <rdma/ib_verbs.h> #include <rdma/ib_umem.h> #include <rdma/ib_umem_odp.h> +/* + * The ib_umem list keeps track of memory regions for which the HW + * device request to receive notification when the related memory + * mapping is changed. + * + * ib_umem_lock protects the list. + */ + +static u64 node_start(struct umem_odp_node *n) +{ + struct ib_umem_odp *umem_odp = + container_of(n, struct ib_umem_odp, interval_tree); + + return ib_umem_start(umem_odp->umem); +} + +/* Note that the representation of the intervals in the interval tree + * considers the ending point as contained in the interval, while the + * function ib_umem_end returns the first address which is not contained + * in the umem. + */ +static u64 node_last(struct umem_odp_node *n) +{ + struct ib_umem_odp *umem_odp = + container_of(n, struct ib_umem_odp, interval_tree); + + return ib_umem_end(umem_odp->umem) - 1; +} + +INTERVAL_TREE_DEFINE(struct umem_odp_node, rb, u64, __subtree_last, + node_start, node_last, static, rbt_ib_umem) + static void ib_umem_notifier_start_account(struct ib_umem *item) { mutex_lock(&item->odp_data->umem_mutex); @@ -747,3 +780,42 @@ void ib_umem_odp_unmap_dma_pages(struct ib_umem *umem, u64 virt, mutex_unlock(&umem->odp_data->umem_mutex); } EXPORT_SYMBOL(ib_umem_odp_unmap_dma_pages); + +/* @last is not a part of the interval. See comment for function + * node_last. + */ +int rbt_ib_umem_for_each_in_range(struct rb_root_cached *root, + u64 start, u64 last, + umem_call_back cb, + void *cookie) +{ + int ret_val = 0; + struct umem_odp_node *node, *next; + struct ib_umem_odp *umem; + + if (unlikely(start == last)) + return ret_val; + + for (node = rbt_ib_umem_iter_first(root, start, last - 1); + node; node = next) { + next = rbt_ib_umem_iter_next(node, start, last - 1); + umem = container_of(node, struct ib_umem_odp, interval_tree); + ret_val = cb(umem->umem, start, last, cookie) || ret_val; + } + + return ret_val; +} +EXPORT_SYMBOL(rbt_ib_umem_for_each_in_range); + +struct ib_umem_odp *rbt_ib_umem_lookup(struct rb_root_cached *root, + u64 addr, u64 length) +{ + struct umem_odp_node *node; + + node = rbt_ib_umem_iter_first(root, addr, addr + length - 1); + if (node) + return container_of(node, struct ib_umem_odp, interval_tree); + return NULL; + +} +EXPORT_SYMBOL(rbt_ib_umem_lookup); diff --git a/sys/ofed/drivers/infiniband/core/ib_umem_rbtree.c b/sys/ofed/drivers/infiniband/core/ib_umem_rbtree.c deleted file mode 100644 index 8025a14c2e4c..000000000000 --- a/sys/ofed/drivers/infiniband/core/ib_umem_rbtree.c +++ /dev/null @@ -1,111 +0,0 @@ -/*- - * SPDX-License-Identifier: BSD-2-Clause OR GPL-2.0 - * - * Copyright (c) 2014 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. - */ - -#include <sys/cdefs.h> -#include <linux/kernel.h> -#include <linux/module.h> -#include <linux/sched.h> -#include <linux/gfp.h> -#include <rdma/ib_umem_odp.h> - -/* - * The ib_umem list keeps track of memory regions for which the HW - * device request to receive notification when the related memory - * mapping is changed. - * - * ib_umem_lock protects the list. - */ - -static inline u64 node_start(struct umem_odp_node *n) -{ - struct ib_umem_odp *umem_odp = - container_of(n, struct ib_umem_odp, interval_tree); - - return ib_umem_start(umem_odp->umem); -} - -/* Note that the representation of the intervals in the interval tree - * considers the ending point as contained in the interval, while the - * function ib_umem_end returns the first address which is not contained - * in the umem. - */ -static inline u64 node_last(struct umem_odp_node *n) -{ - struct ib_umem_odp *umem_odp = - container_of(n, struct ib_umem_odp, interval_tree); - - return ib_umem_end(umem_odp->umem) - 1; -} - -INTERVAL_TREE_DEFINE(struct umem_odp_node, rb, u64, __subtree_last, - node_start, node_last, , rbt_ib_umem) - -/* @last is not a part of the interval. See comment for function - * node_last. - */ -int rbt_ib_umem_for_each_in_range(struct rb_root_cached *root, - u64 start, u64 last, - umem_call_back cb, - void *cookie) -{ - int ret_val = 0; - struct umem_odp_node *node, *next; - struct ib_umem_odp *umem; - - if (unlikely(start == last)) - return ret_val; - - for (node = rbt_ib_umem_iter_first(root, start, last - 1); - node; node = next) { - next = rbt_ib_umem_iter_next(node, start, last - 1); - umem = container_of(node, struct ib_umem_odp, interval_tree); - ret_val = cb(umem->umem, start, last, cookie) || ret_val; - } - - return ret_val; -} -EXPORT_SYMBOL(rbt_ib_umem_for_each_in_range); - -struct ib_umem_odp *rbt_ib_umem_lookup(struct rb_root_cached *root, - u64 addr, u64 length) -{ - struct umem_odp_node *node; - - node = rbt_ib_umem_iter_first(root, addr, addr + length - 1); - if (node) - return container_of(node, struct ib_umem_odp, interval_tree); - return NULL; - -} -EXPORT_SYMBOL(rbt_ib_umem_lookup); diff --git a/sys/ofed/drivers/infiniband/core/ib_uverbs_cmd.c b/sys/ofed/drivers/infiniband/core/ib_uverbs_cmd.c index 711c73c18115..a581d3c10b49 100644 --- a/sys/ofed/drivers/infiniband/core/ib_uverbs_cmd.c +++ b/sys/ofed/drivers/infiniband/core/ib_uverbs_cmd.c @@ -3714,7 +3714,7 @@ static int ib_uverbs_ex_modify_cq(struct uverbs_attr_bundle *attrs) if (!cq) return -EINVAL; - ret = ib_modify_cq(cq, cmd.attr.cq_count, cmd.attr.cq_period); + ret = rdma_set_cq_moderation(cq, cmd.attr.cq_count, cmd.attr.cq_period); rdma_lookup_put_uobject(&cq->uobject->uevent.uobject, UVERBS_LOOKUP_READ); diff --git a/sys/ofed/drivers/infiniband/core/ib_uverbs_marshall.c b/sys/ofed/drivers/infiniband/core/ib_uverbs_marshall.c index f7c4a221aa88..f74ecbf9cf96 100644 --- a/sys/ofed/drivers/infiniband/core/ib_uverbs_marshall.c +++ b/sys/ofed/drivers/infiniband/core/ib_uverbs_marshall.c @@ -71,8 +71,7 @@ void ib_copy_ah_attr_to_user(struct ib_device *device, memset(&dst->grh.reserved, 0, sizeof(dst->grh.reserved)); if ((ah_attr->type == RDMA_AH_ATTR_TYPE_OPA) && - (rdma_ah_get_dlid(ah_attr) >= - be16_to_cpu(IB_MULTICAST_LID_BASE)) && + (rdma_ah_get_dlid(ah_attr) > be16_to_cpu(IB_LID_PERMISSIVE)) && (!rdma_ah_conv_opa_to_ib(device, &conv_ah, ah_attr))) src = &conv_ah; diff --git a/sys/ofed/drivers/infiniband/core/ib_verbs.c b/sys/ofed/drivers/infiniband/core/ib_verbs.c index ef65aeefd8f4..b0cdf8a318d9 100644 --- a/sys/ofed/drivers/infiniband/core/ib_verbs.c +++ b/sys/ofed/drivers/infiniband/core/ib_verbs.c @@ -58,6 +58,9 @@ #include "core_priv.h" +static int ib_resolve_eth_dmac(struct ib_device *device, + struct rdma_ah_attr *ah_attr); + static const char * const ib_events[] = { [IB_EVENT_CQ_ERR] = "CQ error", [IB_EVENT_QP_FATAL] = "QP fatal error", @@ -1289,8 +1292,8 @@ bool ib_modify_qp_is_ok(enum ib_qp_state cur_state, enum ib_qp_state next_state, } EXPORT_SYMBOL(ib_modify_qp_is_ok); -int ib_resolve_eth_dmac(struct ib_device *device, - struct rdma_ah_attr *ah_attr) +static int ib_resolve_eth_dmac(struct ib_device *device, + struct rdma_ah_attr *ah_attr) { struct ib_gid_attr sgid_attr; union ib_gid sgid; @@ -1336,7 +1339,6 @@ int ib_resolve_eth_dmac(struct ib_device *device, grh->hop_limit = hop_limit; return ret; } -EXPORT_SYMBOL(ib_resolve_eth_dmac); static bool is_qp_type_connected(const struct ib_qp *qp) { @@ -1566,12 +1568,12 @@ struct ib_cq *__ib_create_cq(struct ib_device *device, } EXPORT_SYMBOL(__ib_create_cq); -int ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period) +int rdma_set_cq_moderation(struct ib_cq *cq, u16 cq_count, u16 cq_period) { return cq->device->modify_cq ? cq->device->modify_cq(cq, cq_count, cq_period) : -ENOSYS; } -EXPORT_SYMBOL(ib_modify_cq); +EXPORT_SYMBOL(rdma_set_cq_moderation); int ib_destroy_cq_user(struct ib_cq *cq, struct ib_udata *udata) { diff --git a/sys/ofed/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c b/sys/ofed/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c index 841326bd958d..10cc5be661b5 100644 --- a/sys/ofed/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c +++ b/sys/ofed/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c @@ -87,8 +87,9 @@ static int ipoib_set_coalesce(if_t dev, coal->rx_coalesce_usecs = 0xffff; } - ret = ib_modify_cq(priv->recv_cq, coal->rx_max_coalesced_frames, - coal->rx_coalesce_usecs); + ret = rdma_set_cq_moderation(priv->recv_cq, + coal->rx_max_coalesced_frames, + coal->rx_coalesce_usecs); if (ret && ret != -ENOSYS) { ipoib_warn(priv, "failed modifying CQ (%d)\n", ret); return ret; diff --git a/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c b/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c index cfc2390db02e..fd45e9961353 100644 --- a/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c +++ b/sys/ofed/drivers/infiniband/ulp/sdp/sdp_main.c @@ -1682,7 +1682,7 @@ sdp_set_default_moderation(struct sdp_sock *ssk) { if (sdp_mod_count <= 0 || sdp_mod_usec <= 0) return; - ib_modify_cq(ssk->rx_ring.cq, sdp_mod_count, sdp_mod_usec); + rdma_set_cq_moderation(ssk->rx_ring.cq, sdp_mod_count, sdp_mod_usec); } static void diff --git a/sys/ofed/include/rdma/ib_addr.h b/sys/ofed/include/rdma/ib_addr.h index fc938afabd05..b59d4a7bd572 100644 --- a/sys/ofed/include/rdma/ib_addr.h +++ b/sys/ofed/include/rdma/ib_addr.h @@ -140,8 +140,9 @@ int rdma_resolve_ip_route(struct sockaddr *src_addr, void rdma_addr_cancel(struct rdma_dev_addr *addr); -int rdma_copy_addr(struct rdma_dev_addr *dev_addr, if_t dev, - const unsigned char *dst_dev_addr); +void rdma_copy_addr(struct rdma_dev_addr *dev_addr, + const if_t dev, + const unsigned char *dst_dev_addr); int rdma_addr_size(struct sockaddr *addr); int rdma_addr_size_in6(struct sockaddr_in6 *addr); diff --git a/sys/ofed/include/rdma/ib_umem_odp.h b/sys/ofed/include/rdma/ib_umem_odp.h index 5affdc6427d9..99a6127a562e 100644 --- a/sys/ofed/include/rdma/ib_umem_odp.h +++ b/sys/ofed/include/rdma/ib_umem_odp.h @@ -114,10 +114,6 @@ int ib_umem_odp_map_dma_pages(struct ib_umem *umem, u64 start_offset, u64 bcnt, void ib_umem_odp_unmap_dma_pages(struct ib_umem *umem, u64 start_offset, u64 bound); -void rbt_ib_umem_insert(struct umem_odp_node *node, - struct rb_root_cached *root); -void rbt_ib_umem_remove(struct umem_odp_node *node, - struct rb_root_cached *root); typedef int (*umem_call_back)(struct ib_umem *item, u64 start, u64 end, void *cookie); /* diff --git a/sys/ofed/include/rdma/ib_verbs.h b/sys/ofed/include/rdma/ib_verbs.h index d526054a29c5..45c7d238de68 100644 --- a/sys/ofed/include/rdma/ib_verbs.h +++ b/sys/ofed/include/rdma/ib_verbs.h @@ -3420,13 +3420,13 @@ struct ib_cq *__ib_create_cq(struct ib_device *device, int ib_resize_cq(struct ib_cq *cq, int cqe); /** - * ib_modify_cq - Modifies moderation params of the CQ + * rdma_set_cq_moderation - Modifies moderation params of the CQ * @cq: The CQ to modify. * @cq_count: number of CQEs that will trigger an event * @cq_period: max period of time in usec before triggering an event * */ -int ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period); +int rdma_set_cq_moderation(struct ib_cq *cq, u16 cq_count, u16 cq_period); /** * ib_destroy_cq_user - Destroys the specified CQ. @@ -3943,9 +3943,6 @@ struct ib_ucontext *ib_uverbs_get_ucontext_file(struct ib_uverbs_file *ufile); int uverbs_destroy_def_handler(struct uverbs_attr_bundle *attrs); -int ib_resolve_eth_dmac(struct ib_device *device, - struct rdma_ah_attr *ah_attr); - static inline u8 *rdma_ah_retrieve_dmac(struct rdma_ah_attr *attr) { if (attr->type == RDMA_AH_ATTR_TYPE_ROCE)
