From 41a4e9646229801624e38f7a1cc53033a0affdb1 Mon Sep 17 00:00:00 2001 From: Santosh Shilimkar Date: Sat, 19 Sep 2015 13:06:08 -0400 Subject: RDS: IB: use max_mr from HCA caps than max_fmr All HCA drivers seems to popullate max_mr caps and few of them do both max_mr and max_fmr. Hence update RDS code to make use of max_mr. Signed-off-by: Santosh Shilimkar Signed-off-by: Santosh Shilimkar --- net/rds/ib.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/rds/ib.c') diff --git a/net/rds/ib.c b/net/rds/ib.c index 2d3f2ab475df..883813a6cf44 100644 --- a/net/rds/ib.c +++ b/net/rds/ib.c @@ -148,8 +148,8 @@ static void rds_ib_add_one(struct ib_device *device) rds_ibdev->max_sge = min(dev_attr->max_sge, RDS_IB_MAX_SGE); rds_ibdev->fmr_max_remaps = dev_attr->max_map_per_fmr?: 32; - rds_ibdev->max_fmrs = dev_attr->max_fmr ? - min_t(unsigned int, dev_attr->max_fmr, fmr_pool_size) : + rds_ibdev->max_fmrs = dev_attr->max_mr ? + min_t(unsigned int, dev_attr->max_mr, fmr_pool_size) : fmr_pool_size; rds_ibdev->max_initiator_depth = dev_attr->max_qp_init_rd_atom; -- cgit v1.2.3 From 06766513232d1619ac84e87b1d839d3fcc23a540 Mon Sep 17 00:00:00 2001 From: Santosh Shilimkar Date: Thu, 10 Sep 2015 21:20:57 -0700 Subject: RDS: IB: split mr pool to improve 8K messages performance 8K message sizes are pretty important usecase for RDS current workloads so we make provison to have 8K mrs available from the pool. Based on number of SG's in the RDS message, we pick a pool to use. Also to make sure that we don't under utlise mrs when say 8k messages are dominating which could lead to 8k pull being exhausted, we fall-back to 1m pool till 8k pool recovers for use. This helps to at least push ~55 kB/s bidirectional data which is a nice improvement. Signed-off-by: Santosh Shilimkar Signed-off-by: Santosh Shilimkar --- net/rds/ib.c | 47 +++++++++++++++++-------- net/rds/ib.h | 43 ++++++++++++++++------- net/rds/ib_rdma.c | 101 +++++++++++++++++++++++++++++++++++++---------------- net/rds/ib_stats.c | 18 ++++++---- 4 files changed, 147 insertions(+), 62 deletions(-) (limited to 'net/rds/ib.c') diff --git a/net/rds/ib.c b/net/rds/ib.c index 883813a6cf44..a833ab7898fe 100644 --- a/net/rds/ib.c +++ b/net/rds/ib.c @@ -43,14 +43,14 @@ #include "rds.h" #include "ib.h" -static unsigned int fmr_pool_size = RDS_FMR_POOL_SIZE; -unsigned int fmr_message_size = RDS_FMR_SIZE + 1; /* +1 allows for unaligned MRs */ +unsigned int rds_ib_fmr_1m_pool_size = RDS_FMR_1M_POOL_SIZE; +unsigned int rds_ib_fmr_8k_pool_size = RDS_FMR_8K_POOL_SIZE; unsigned int rds_ib_retry_count = RDS_IB_DEFAULT_RETRY_COUNT; -module_param(fmr_pool_size, int, 0444); -MODULE_PARM_DESC(fmr_pool_size, " Max number of fmr per HCA"); -module_param(fmr_message_size, int, 0444); -MODULE_PARM_DESC(fmr_message_size, " Max size of a RDMA transfer"); +module_param(rds_ib_fmr_1m_pool_size, int, 0444); +MODULE_PARM_DESC(rds_ib_fmr_1m_pool_size, " Max number of 1M fmr per HCA"); +module_param(rds_ib_fmr_8k_pool_size, int, 0444); +MODULE_PARM_DESC(rds_ib_fmr_8k_pool_size, " Max number of 8K fmr per HCA"); module_param(rds_ib_retry_count, int, 0444); MODULE_PARM_DESC(rds_ib_retry_count, " Number of hw retries before reporting an error"); @@ -97,8 +97,10 @@ static void rds_ib_dev_free(struct work_struct *work) struct rds_ib_device *rds_ibdev = container_of(work, struct rds_ib_device, free_work); - if (rds_ibdev->mr_pool) - rds_ib_destroy_mr_pool(rds_ibdev->mr_pool); + if (rds_ibdev->mr_8k_pool) + rds_ib_destroy_mr_pool(rds_ibdev->mr_8k_pool); + if (rds_ibdev->mr_1m_pool) + rds_ib_destroy_mr_pool(rds_ibdev->mr_1m_pool); if (rds_ibdev->pd) ib_dealloc_pd(rds_ibdev->pd); @@ -148,9 +150,13 @@ static void rds_ib_add_one(struct ib_device *device) rds_ibdev->max_sge = min(dev_attr->max_sge, RDS_IB_MAX_SGE); rds_ibdev->fmr_max_remaps = dev_attr->max_map_per_fmr?: 32; - rds_ibdev->max_fmrs = dev_attr->max_mr ? - min_t(unsigned int, dev_attr->max_mr, fmr_pool_size) : - fmr_pool_size; + rds_ibdev->max_1m_fmrs = dev_attr->max_mr ? + min_t(unsigned int, (dev_attr->max_mr / 2), + rds_ib_fmr_1m_pool_size) : rds_ib_fmr_1m_pool_size; + + rds_ibdev->max_8k_fmrs = dev_attr->max_mr ? + min_t(unsigned int, ((dev_attr->max_mr / 2) * RDS_MR_8K_SCALE), + rds_ib_fmr_8k_pool_size) : rds_ib_fmr_8k_pool_size; rds_ibdev->max_initiator_depth = dev_attr->max_qp_init_rd_atom; rds_ibdev->max_responder_resources = dev_attr->max_qp_rd_atom; @@ -162,12 +168,25 @@ static void rds_ib_add_one(struct ib_device *device) goto put_dev; } - rds_ibdev->mr_pool = rds_ib_create_mr_pool(rds_ibdev); - if (IS_ERR(rds_ibdev->mr_pool)) { - rds_ibdev->mr_pool = NULL; + rds_ibdev->mr_1m_pool = + rds_ib_create_mr_pool(rds_ibdev, RDS_IB_MR_1M_POOL); + if (IS_ERR(rds_ibdev->mr_1m_pool)) { + rds_ibdev->mr_1m_pool = NULL; goto put_dev; } + rds_ibdev->mr_8k_pool = + rds_ib_create_mr_pool(rds_ibdev, RDS_IB_MR_8K_POOL); + if (IS_ERR(rds_ibdev->mr_8k_pool)) { + rds_ibdev->mr_8k_pool = NULL; + goto put_dev; + } + + rdsdebug("RDS/IB: max_mr = %d, max_wrs = %d, max_sge = %d, fmr_max_remaps = %d, max_1m_fmrs = %d, max_8k_fmrs = %d\n", + dev_attr->max_fmr, rds_ibdev->max_wrs, rds_ibdev->max_sge, + rds_ibdev->fmr_max_remaps, rds_ibdev->max_1m_fmrs, + rds_ibdev->max_8k_fmrs); + INIT_LIST_HEAD(&rds_ibdev->ipaddr_list); INIT_LIST_HEAD(&rds_ibdev->conn_list); diff --git a/net/rds/ib.h b/net/rds/ib.h index 3a8cd31d4048..f17d09567890 100644 --- a/net/rds/ib.h +++ b/net/rds/ib.h @@ -9,8 +9,11 @@ #include "rds.h" #include "rdma_transport.h" -#define RDS_FMR_SIZE 256 -#define RDS_FMR_POOL_SIZE 8192 +#define RDS_FMR_1M_POOL_SIZE (8192 / 2) +#define RDS_FMR_1M_MSG_SIZE 256 +#define RDS_FMR_8K_MSG_SIZE 2 +#define RDS_MR_8K_SCALE (256 / (RDS_FMR_8K_MSG_SIZE + 1)) +#define RDS_FMR_8K_POOL_SIZE (RDS_MR_8K_SCALE * (8192 / 2)) #define RDS_IB_MAX_SGE 8 #define RDS_IB_RECV_SGE 2 @@ -189,15 +192,23 @@ struct rds_ib_ipaddr { struct rcu_head rcu; }; +enum { + RDS_IB_MR_8K_POOL, + RDS_IB_MR_1M_POOL, +}; + struct rds_ib_device { struct list_head list; struct list_head ipaddr_list; struct list_head conn_list; struct ib_device *dev; struct ib_pd *pd; - struct rds_ib_mr_pool *mr_pool; - unsigned int fmr_max_remaps; unsigned int max_fmrs; + struct rds_ib_mr_pool *mr_1m_pool; + struct rds_ib_mr_pool *mr_8k_pool; + unsigned int fmr_max_remaps; + unsigned int max_8k_fmrs; + unsigned int max_1m_fmrs; int max_sge; unsigned int max_wrs; unsigned int max_initiator_depth; @@ -239,12 +250,18 @@ struct rds_ib_statistics { uint64_t s_ib_ack_send_delayed; uint64_t s_ib_ack_send_piggybacked; uint64_t s_ib_ack_received; - uint64_t s_ib_rdma_mr_alloc; - uint64_t s_ib_rdma_mr_free; - uint64_t s_ib_rdma_mr_used; - uint64_t s_ib_rdma_mr_pool_flush; - uint64_t s_ib_rdma_mr_pool_wait; - uint64_t s_ib_rdma_mr_pool_depleted; + uint64_t s_ib_rdma_mr_8k_alloc; + uint64_t s_ib_rdma_mr_8k_free; + uint64_t s_ib_rdma_mr_8k_used; + uint64_t s_ib_rdma_mr_8k_pool_flush; + uint64_t s_ib_rdma_mr_8k_pool_wait; + uint64_t s_ib_rdma_mr_8k_pool_depleted; + uint64_t s_ib_rdma_mr_1m_alloc; + uint64_t s_ib_rdma_mr_1m_free; + uint64_t s_ib_rdma_mr_1m_used; + uint64_t s_ib_rdma_mr_1m_pool_flush; + uint64_t s_ib_rdma_mr_1m_pool_wait; + uint64_t s_ib_rdma_mr_1m_pool_depleted; uint64_t s_ib_atomic_cswp; uint64_t s_ib_atomic_fadd; }; @@ -296,7 +313,8 @@ struct rds_ib_device *rds_ib_get_client_data(struct ib_device *device); void rds_ib_dev_put(struct rds_ib_device *rds_ibdev); extern struct ib_client rds_ib_client; -extern unsigned int fmr_message_size; +extern unsigned int rds_ib_fmr_1m_pool_size; +extern unsigned int rds_ib_fmr_8k_pool_size; extern unsigned int rds_ib_retry_count; extern spinlock_t ib_nodev_conns_lock; @@ -326,7 +344,8 @@ int rds_ib_update_ipaddr(struct rds_ib_device *rds_ibdev, __be32 ipaddr); void rds_ib_add_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn); void rds_ib_remove_conn(struct rds_ib_device *rds_ibdev, struct rds_connection *conn); void rds_ib_destroy_nodev_conns(void); -struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *); +struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_dev, + int npages); void rds_ib_get_mr_info(struct rds_ib_device *rds_ibdev, struct rds_info_rdma_connection *iinfo); void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *); void *rds_ib_get_mr(struct scatterlist *sg, unsigned long nents, diff --git a/net/rds/ib_rdma.c b/net/rds/ib_rdma.c index bb620246b3d6..a2340748ec86 100644 --- a/net/rds/ib_rdma.c +++ b/net/rds/ib_rdma.c @@ -65,6 +65,7 @@ struct rds_ib_mr { * Our own little FMR pool */ struct rds_ib_mr_pool { + unsigned int pool_type; struct mutex flush_lock; /* serialize fmr invalidate */ struct delayed_work flush_worker; /* flush worker */ @@ -234,7 +235,8 @@ void rds_ib_destroy_nodev_conns(void) rds_conn_destroy(ic->conn); } -struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev) +struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev, + int pool_type) { struct rds_ib_mr_pool *pool; @@ -242,6 +244,7 @@ struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev) if (!pool) return ERR_PTR(-ENOMEM); + pool->pool_type = pool_type; init_llist_head(&pool->free_list); init_llist_head(&pool->drop_list); init_llist_head(&pool->clean_list); @@ -249,28 +252,30 @@ struct rds_ib_mr_pool *rds_ib_create_mr_pool(struct rds_ib_device *rds_ibdev) init_waitqueue_head(&pool->flush_wait); INIT_DELAYED_WORK(&pool->flush_worker, rds_ib_mr_pool_flush_worker); - pool->fmr_attr.max_pages = fmr_message_size; + if (pool_type == RDS_IB_MR_1M_POOL) { + /* +1 allows for unaligned MRs */ + pool->fmr_attr.max_pages = RDS_FMR_1M_MSG_SIZE + 1; + pool->max_items = RDS_FMR_1M_POOL_SIZE; + } else { + /* pool_type == RDS_IB_MR_8K_POOL */ + pool->fmr_attr.max_pages = RDS_FMR_8K_MSG_SIZE + 1; + pool->max_items = RDS_FMR_8K_POOL_SIZE; + } + + pool->max_free_pinned = pool->max_items * pool->fmr_attr.max_pages / 4; pool->fmr_attr.max_maps = rds_ibdev->fmr_max_remaps; pool->fmr_attr.page_shift = PAGE_SHIFT; - pool->max_free_pinned = rds_ibdev->max_fmrs * fmr_message_size / 4; - - /* We never allow more than max_items MRs to be allocated. - * When we exceed more than max_items_soft, we start freeing - * items more aggressively. - * Make sure that max_items > max_items_soft > max_items / 2 - */ pool->max_items_soft = rds_ibdev->max_fmrs * 3 / 4; - pool->max_items = rds_ibdev->max_fmrs; return pool; } void rds_ib_get_mr_info(struct rds_ib_device *rds_ibdev, struct rds_info_rdma_connection *iinfo) { - struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool; + struct rds_ib_mr_pool *pool_1m = rds_ibdev->mr_1m_pool; - iinfo->rdma_mr_max = pool->max_items; - iinfo->rdma_mr_size = pool->fmr_attr.max_pages; + iinfo->rdma_mr_max = pool_1m->max_items; + iinfo->rdma_mr_size = pool_1m->fmr_attr.max_pages; } void rds_ib_destroy_mr_pool(struct rds_ib_mr_pool *pool) @@ -312,15 +317,29 @@ static inline void wait_clean_list_grace(void) } } -static struct rds_ib_mr *rds_ib_alloc_fmr(struct rds_ib_device *rds_ibdev) +static struct rds_ib_mr *rds_ib_alloc_fmr(struct rds_ib_device *rds_ibdev, + int npages) { - struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool; + struct rds_ib_mr_pool *pool; struct rds_ib_mr *ibmr = NULL; int err = 0, iter = 0; + if (npages <= RDS_FMR_8K_MSG_SIZE) + pool = rds_ibdev->mr_8k_pool; + else + pool = rds_ibdev->mr_1m_pool; + if (atomic_read(&pool->dirty_count) >= pool->max_items / 10) queue_delayed_work(rds_ib_fmr_wq, &pool->flush_worker, 10); + /* Switch pools if one of the pool is reaching upper limit */ + if (atomic_read(&pool->dirty_count) >= pool->max_items * 9 / 10) { + if (pool->pool_type == RDS_IB_MR_8K_POOL) + pool = rds_ibdev->mr_1m_pool; + else + pool = rds_ibdev->mr_8k_pool; + } + while (1) { ibmr = rds_ib_reuse_fmr(pool); if (ibmr) @@ -341,12 +360,18 @@ static struct rds_ib_mr *rds_ib_alloc_fmr(struct rds_ib_device *rds_ibdev) atomic_dec(&pool->item_count); if (++iter > 2) { - rds_ib_stats_inc(s_ib_rdma_mr_pool_depleted); + if (pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_pool_depleted); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_pool_depleted); return ERR_PTR(-EAGAIN); } /* We do have some empty MRs. Flush them out. */ - rds_ib_stats_inc(s_ib_rdma_mr_pool_wait); + if (pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_pool_wait); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_pool_wait); rds_ib_flush_mr_pool(pool, 0, &ibmr); if (ibmr) return ibmr; @@ -371,7 +396,12 @@ static struct rds_ib_mr *rds_ib_alloc_fmr(struct rds_ib_device *rds_ibdev) goto out_no_cigar; } - rds_ib_stats_inc(s_ib_rdma_mr_alloc); + ibmr->pool = pool; + if (pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_alloc); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_alloc); + return ibmr; out_no_cigar: @@ -427,7 +457,7 @@ static int rds_ib_map_fmr(struct rds_ib_device *rds_ibdev, struct rds_ib_mr *ibm } page_cnt += len >> PAGE_SHIFT; - if (page_cnt > fmr_message_size) + if (page_cnt > ibmr->pool->fmr_attr.max_pages) return -EINVAL; dma_pages = kmalloc_node(sizeof(u64) * page_cnt, GFP_ATOMIC, @@ -459,7 +489,10 @@ static int rds_ib_map_fmr(struct rds_ib_device *rds_ibdev, struct rds_ib_mr *ibm ibmr->sg_dma_len = sg_dma_len; ibmr->remap_count++; - rds_ib_stats_inc(s_ib_rdma_mr_used); + if (ibmr->pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_used); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_used); ret = 0; out: @@ -591,7 +624,7 @@ static void list_to_llist_nodes(struct rds_ib_mr_pool *pool, * to free as many MRs as needed to get back to this limit. */ static int rds_ib_flush_mr_pool(struct rds_ib_mr_pool *pool, - int free_all, struct rds_ib_mr **ibmr_ret) + int free_all, struct rds_ib_mr **ibmr_ret) { struct rds_ib_mr *ibmr, *next; struct llist_node *clean_nodes; @@ -602,11 +635,14 @@ static int rds_ib_flush_mr_pool(struct rds_ib_mr_pool *pool, unsigned int nfreed = 0, dirty_to_clean = 0, free_goal; int ret = 0; - rds_ib_stats_inc(s_ib_rdma_mr_pool_flush); + if (pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_pool_flush); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_pool_flush); if (ibmr_ret) { DEFINE_WAIT(wait); - while(!mutex_trylock(&pool->flush_lock)) { + while (!mutex_trylock(&pool->flush_lock)) { ibmr = rds_ib_reuse_fmr(pool); if (ibmr) { *ibmr_ret = ibmr; @@ -663,8 +699,12 @@ static int rds_ib_flush_mr_pool(struct rds_ib_mr_pool *pool, list_for_each_entry_safe(ibmr, next, &unmap_list, unmap_list) { unpinned += ibmr->sg_len; __rds_ib_teardown_mr(ibmr); - if (nfreed < free_goal || ibmr->remap_count >= pool->fmr_attr.max_maps) { - rds_ib_stats_inc(s_ib_rdma_mr_free); + if (nfreed < free_goal || + ibmr->remap_count >= pool->fmr_attr.max_maps) { + if (ibmr->pool->pool_type == RDS_IB_MR_8K_POOL) + rds_ib_stats_inc(s_ib_rdma_mr_8k_free); + else + rds_ib_stats_inc(s_ib_rdma_mr_1m_free); list_del(&ibmr->unmap_list); ib_dealloc_fmr(ibmr->fmr); kfree(ibmr); @@ -756,10 +796,11 @@ void rds_ib_flush_mrs(void) down_read(&rds_ib_devices_lock); list_for_each_entry(rds_ibdev, &rds_ib_devices, list) { - struct rds_ib_mr_pool *pool = rds_ibdev->mr_pool; + if (rds_ibdev->mr_8k_pool) + rds_ib_flush_mr_pool(rds_ibdev->mr_8k_pool, 0, NULL); - if (pool) - rds_ib_flush_mr_pool(pool, 0, NULL); + if (rds_ibdev->mr_1m_pool) + rds_ib_flush_mr_pool(rds_ibdev->mr_1m_pool, 0, NULL); } up_read(&rds_ib_devices_lock); } @@ -777,12 +818,12 @@ void *rds_ib_get_mr(struct scatterlist *sg, unsigned long nents, goto out; } - if (!rds_ibdev->mr_pool) { + if (!rds_ibdev->mr_8k_pool || !rds_ibdev->mr_1m_pool) { ret = -ENODEV; goto out; } - ibmr = rds_ib_alloc_fmr(rds_ibdev); + ibmr = rds_ib_alloc_fmr(rds_ibdev, nents); if (IS_ERR(ibmr)) { rds_ib_dev_put(rds_ibdev); return ibmr; diff --git a/net/rds/ib_stats.c b/net/rds/ib_stats.c index 8c8b84f7bfbb..d77e04473056 100644 --- a/net/rds/ib_stats.c +++ b/net/rds/ib_stats.c @@ -61,12 +61,18 @@ static const char *const rds_ib_stat_names[] = { "ib_ack_send_delayed", "ib_ack_send_piggybacked", "ib_ack_received", - "ib_rdma_mr_alloc", - "ib_rdma_mr_free", - "ib_rdma_mr_used", - "ib_rdma_mr_pool_flush", - "ib_rdma_mr_pool_wait", - "ib_rdma_mr_pool_depleted", + "ib_rdma_mr_8k_alloc", + "ib_rdma_mr_8k_free", + "ib_rdma_mr_8k_used", + "ib_rdma_mr_8k_pool_flush", + "ib_rdma_mr_8k_pool_wait", + "ib_rdma_mr_8k_pool_depleted", + "ib_rdma_mr_1m_alloc", + "ib_rdma_mr_1m_free", + "ib_rdma_mr_1m_used", + "ib_rdma_mr_1m_pool_flush", + "ib_rdma_mr_1m_pool_wait", + "ib_rdma_mr_1m_pool_depleted", "ib_atomic_cswp", "ib_atomic_fadd", }; -- cgit v1.2.3 From fa20105e09e97e81aadf02f722c31195e4a75c84 Mon Sep 17 00:00:00 2001 From: Guy Shapiro Date: Thu, 22 Oct 2015 15:20:10 +0300 Subject: IB/cma: Add support for network namespaces Add support for network namespaces in the ib_cma module. This is accomplished by: 1. Adding network namespace parameter for rdma_create_id. This parameter is used to populate the network namespace field in rdma_id_private. rdma_create_id keeps a reference on the network namespace. 2. Using the network namespace from the rdma_id instead of init_net inside of ib_cma, when listening on an ID and when looking for an ID for an incoming request. 3. Decrementing the reference count for the appropriate network namespace when calling rdma_destroy_id. In order to preserve the current behavior init_net is passed when calling from other modules. Signed-off-by: Guy Shapiro Signed-off-by: Haggai Eran Signed-off-by: Yotam Kenneth Signed-off-by: Shachar Raindel Signed-off-by: Doug Ledford --- drivers/infiniband/core/cma.c | 46 +++++++++++++--------- drivers/infiniband/core/ucma.c | 3 +- drivers/infiniband/ulp/iser/iser_verbs.c | 2 +- drivers/infiniband/ulp/isert/ib_isert.c | 2 +- .../staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h | 4 +- include/rdma/rdma_cm.h | 6 ++- net/9p/trans_rdma.c | 4 +- net/rds/ib.c | 2 +- net/rds/ib_cm.c | 2 +- net/rds/iw.c | 2 +- net/rds/iw_cm.c | 2 +- net/rds/rdma_transport.c | 4 +- net/sunrpc/xprtrdma/svc_rdma_transport.c | 4 +- net/sunrpc/xprtrdma/verbs.c | 3 +- 14 files changed, 52 insertions(+), 34 deletions(-) (limited to 'net/rds/ib.c') diff --git a/drivers/infiniband/core/cma.c b/drivers/infiniband/core/cma.c index ac03c32ca7f1..7e93eb1f33eb 100644 --- a/drivers/infiniband/core/cma.c +++ b/drivers/infiniband/core/cma.c @@ -600,7 +600,8 @@ static int cma_disable_callback(struct rdma_id_private *id_priv, return 0; } -struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler, +struct rdma_cm_id *rdma_create_id(struct net *net, + rdma_cm_event_handler event_handler, void *context, enum rdma_port_space ps, enum ib_qp_type qp_type) { @@ -624,7 +625,7 @@ struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler, INIT_LIST_HEAD(&id_priv->listen_list); INIT_LIST_HEAD(&id_priv->mc_list); get_random_bytes(&id_priv->seq_num, sizeof id_priv->seq_num); - id_priv->id.route.addr.dev_addr.net = &init_net; + id_priv->id.route.addr.dev_addr.net = get_net(net); return &id_priv->id; } @@ -1278,7 +1279,7 @@ static bool cma_match_net_dev(const struct rdma_id_private *id_priv, cma_protocol_roce(&id_priv->id); return !addr->dev_addr.bound_dev_if || - (net_eq(dev_net(net_dev), &init_net) && + (net_eq(dev_net(net_dev), addr->dev_addr.net) && addr->dev_addr.bound_dev_if == net_dev->ifindex); } @@ -1339,7 +1340,7 @@ static struct rdma_id_private *cma_id_from_event(struct ib_cm_id *cm_id, } } - bind_list = cma_ps_find(&init_net, + bind_list = cma_ps_find(*net_dev ? dev_net(*net_dev) : &init_net, rdma_ps_from_service_id(req.service_id), cma_port_from_service_id(req.service_id)); id_priv = cma_find_listener(bind_list, cm_id, ib_event, &req, *net_dev); @@ -1411,6 +1412,7 @@ static void cma_cancel_operation(struct rdma_id_private *id_priv, static void cma_release_port(struct rdma_id_private *id_priv) { struct rdma_bind_list *bind_list = id_priv->bind_list; + struct net *net = id_priv->id.route.addr.dev_addr.net; if (!bind_list) return; @@ -1418,7 +1420,7 @@ static void cma_release_port(struct rdma_id_private *id_priv) mutex_lock(&lock); hlist_del(&id_priv->node); if (hlist_empty(&bind_list->owners)) { - cma_ps_remove(&init_net, bind_list->ps, bind_list->port); + cma_ps_remove(net, bind_list->ps, bind_list->port); kfree(bind_list); } mutex_unlock(&lock); @@ -1477,6 +1479,7 @@ void rdma_destroy_id(struct rdma_cm_id *id) cma_deref_id(id_priv->id.context); kfree(id_priv->id.route.path_rec); + put_net(id_priv->id.route.addr.dev_addr.net); kfree(id_priv); } EXPORT_SYMBOL(rdma_destroy_id); @@ -1607,7 +1610,8 @@ static struct rdma_id_private *cma_new_conn_id(struct rdma_cm_id *listen_id, ib_event->param.req_rcvd.primary_path->service_id; int ret; - id = rdma_create_id(listen_id->event_handler, listen_id->context, + id = rdma_create_id(listen_id->route.addr.dev_addr.net, + listen_id->event_handler, listen_id->context, listen_id->ps, ib_event->param.req_rcvd.qp_type); if (IS_ERR(id)) return NULL; @@ -1662,9 +1666,10 @@ static struct rdma_id_private *cma_new_udp_id(struct rdma_cm_id *listen_id, struct rdma_id_private *id_priv; struct rdma_cm_id *id; const sa_family_t ss_family = listen_id->route.addr.src_addr.ss_family; + struct net *net = listen_id->route.addr.dev_addr.net; int ret; - id = rdma_create_id(listen_id->event_handler, listen_id->context, + id = rdma_create_id(net, listen_id->event_handler, listen_id->context, listen_id->ps, IB_QPT_UD); if (IS_ERR(id)) return NULL; @@ -1901,7 +1906,8 @@ static int iw_conn_req_handler(struct iw_cm_id *cm_id, return -ECONNABORTED; /* Create a new RDMA id for the new IW CM ID */ - new_cm_id = rdma_create_id(listen_id->id.event_handler, + new_cm_id = rdma_create_id(listen_id->id.route.addr.dev_addr.net, + listen_id->id.event_handler, listen_id->id.context, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(new_cm_id)) { @@ -2029,12 +2035,13 @@ static void cma_listen_on_dev(struct rdma_id_private *id_priv, { struct rdma_id_private *dev_id_priv; struct rdma_cm_id *id; + struct net *net = id_priv->id.route.addr.dev_addr.net; int ret; if (cma_family(id_priv) == AF_IB && !rdma_cap_ib_cm(cma_dev->device, 1)) return; - id = rdma_create_id(cma_listen_handler, id_priv, id_priv->id.ps, + id = rdma_create_id(net, cma_listen_handler, id_priv, id_priv->id.ps, id_priv->id.qp_type); if (IS_ERR(id)) return; @@ -2708,7 +2715,8 @@ static int cma_alloc_port(enum rdma_port_space ps, if (!bind_list) return -ENOMEM; - ret = cma_ps_alloc(&init_net, ps, bind_list, snum); + ret = cma_ps_alloc(id_priv->id.route.addr.dev_addr.net, ps, bind_list, + snum); if (ret < 0) goto err; @@ -2727,13 +2735,14 @@ static int cma_alloc_any_port(enum rdma_port_space ps, static unsigned int last_used_port; int low, high, remaining; unsigned int rover; + struct net *net = id_priv->id.route.addr.dev_addr.net; - inet_get_local_port_range(&init_net, &low, &high); + inet_get_local_port_range(net, &low, &high); remaining = (high - low) + 1; rover = prandom_u32() % remaining + low; retry: if (last_used_port != rover && - !cma_ps_find(&init_net, ps, (unsigned short)rover)) { + !cma_ps_find(net, ps, (unsigned short)rover)) { int ret = cma_alloc_port(ps, id_priv, rover); /* * Remember previously used port number in order to avoid @@ -2799,7 +2808,7 @@ static int cma_use_port(enum rdma_port_space ps, if (snum < PROT_SOCK && !capable(CAP_NET_BIND_SERVICE)) return -EACCES; - bind_list = cma_ps_find(&init_net, ps, snum); + bind_list = cma_ps_find(id_priv->id.route.addr.dev_addr.net, ps, snum); if (!bind_list) { ret = cma_alloc_port(ps, id_priv, snum); } else { @@ -2991,8 +3000,11 @@ int rdma_bind_addr(struct rdma_cm_id *id, struct sockaddr *addr) if (addr->sa_family == AF_INET) id_priv->afonly = 1; #if IS_ENABLED(CONFIG_IPV6) - else if (addr->sa_family == AF_INET6) - id_priv->afonly = init_net.ipv6.sysctl.bindv6only; + else if (addr->sa_family == AF_INET6) { + struct net *net = id_priv->id.route.addr.dev_addr.net; + + id_priv->afonly = net->ipv6.sysctl.bindv6only; + } #endif } ret = cma_get_port(id_priv); @@ -3797,6 +3809,7 @@ static int cma_netdev_change(struct net_device *ndev, struct rdma_id_private *id dev_addr = &id_priv->id.route.addr.dev_addr; if ((dev_addr->bound_dev_if == ndev->ifindex) && + (net_eq(dev_net(ndev), dev_addr->net)) && memcmp(dev_addr->src_dev_addr, ndev->dev_addr, ndev->addr_len)) { printk(KERN_INFO "RDMA CM addr change for ndev %s used by id %p\n", ndev->name, &id_priv->id); @@ -3822,9 +3835,6 @@ static int cma_netdev_callback(struct notifier_block *self, unsigned long event, struct rdma_id_private *id_priv; int ret = NOTIFY_DONE; - if (dev_net(ndev) != &init_net) - return NOTIFY_DONE; - if (event != NETDEV_BONDING_FAILOVER) return NOTIFY_DONE; diff --git a/drivers/infiniband/core/ucma.c b/drivers/infiniband/core/ucma.c index 3a342dfaac15..e80c107450ab 100644 --- a/drivers/infiniband/core/ucma.c +++ b/drivers/infiniband/core/ucma.c @@ -472,7 +472,8 @@ static ssize_t ucma_create_id(struct ucma_file *file, const char __user *inbuf, return -ENOMEM; ctx->uid = cmd.uid; - ctx->cm_id = rdma_create_id(ucma_event_handler, ctx, cmd.ps, qp_type); + ctx->cm_id = rdma_create_id(&init_net, ucma_event_handler, ctx, cmd.ps, + qp_type); if (IS_ERR(ctx->cm_id)) { ret = PTR_ERR(ctx->cm_id); goto err1; diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index 85132d867bc8..f3122372d49f 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -1017,7 +1017,7 @@ int iser_connect(struct iser_conn *iser_conn, ib_conn->beacon.wr_id = ISER_BEACON_WRID; ib_conn->beacon.opcode = IB_WR_SEND; - ib_conn->cma_id = rdma_create_id(iser_cma_handler, + ib_conn->cma_id = rdma_create_id(&init_net, iser_cma_handler, (void *)iser_conn, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(ib_conn->cma_id)) { diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c index b58ac25ce19d..7d41c9d6d6bf 100644 --- a/drivers/infiniband/ulp/isert/ib_isert.c +++ b/drivers/infiniband/ulp/isert/ib_isert.c @@ -3096,7 +3096,7 @@ isert_setup_id(struct isert_np *isert_np) sa = (struct sockaddr *)&np->np_sockaddr; isert_dbg("ksockaddr: %p, sa: %p\n", &np->np_sockaddr, sa); - id = rdma_create_id(isert_cma_handler, isert_np, + id = rdma_create_id(&init_net, isert_cma_handler, isert_np, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(id)) { isert_err("rdma_create_id() failed: %ld\n", PTR_ERR(id)); diff --git a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h index f4b6c33ac318..5015a9e830bd 100644 --- a/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h +++ b/drivers/staging/lustre/lnet/klnds/o2iblnd/o2iblnd.h @@ -128,7 +128,9 @@ extern kib_tunables_t kiblnd_tunables; IBLND_CREDIT_HIGHWATER_V1 : \ *kiblnd_tunables.kib_peercredits_hiw) /* when eagerly to return credits */ -#define kiblnd_rdma_create_id(cb, dev, ps, qpt) rdma_create_id(cb, dev, ps, qpt) +#define kiblnd_rdma_create_id(cb, dev, ps, qpt) rdma_create_id(&init_net, \ + cb, dev, \ + ps, qpt) static inline int kiblnd_concurrent_sends_v1(void) diff --git a/include/rdma/rdma_cm.h b/include/rdma/rdma_cm.h index c92522c192d2..463ec0ccbe51 100644 --- a/include/rdma/rdma_cm.h +++ b/include/rdma/rdma_cm.h @@ -160,13 +160,17 @@ struct rdma_cm_id { /** * rdma_create_id - Create an RDMA identifier. * + * @net: The network namespace in which to create the new id. * @event_handler: User callback invoked to report events associated with the * returned rdma_id. * @context: User specified context associated with the id. * @ps: RDMA port space. * @qp_type: type of queue pair associated with the id. + * + * The id holds a reference on the network namespace until it is destroyed. */ -struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler, +struct rdma_cm_id *rdma_create_id(struct net *net, + rdma_cm_event_handler event_handler, void *context, enum rdma_port_space ps, enum ib_qp_type qp_type); diff --git a/net/9p/trans_rdma.c b/net/9p/trans_rdma.c index ba1210253f5e..52b4a2f993f2 100644 --- a/net/9p/trans_rdma.c +++ b/net/9p/trans_rdma.c @@ -655,8 +655,8 @@ rdma_create_trans(struct p9_client *client, const char *addr, char *args) return -ENOMEM; /* Create the RDMA CM ID */ - rdma->cm_id = rdma_create_id(p9_cm_event_handler, client, RDMA_PS_TCP, - IB_QPT_RC); + rdma->cm_id = rdma_create_id(&init_net, p9_cm_event_handler, client, + RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(rdma->cm_id)) goto error; diff --git a/net/rds/ib.c b/net/rds/ib.c index 2d3f2ab475df..cd64ef9a4748 100644 --- a/net/rds/ib.c +++ b/net/rds/ib.c @@ -317,7 +317,7 @@ static int rds_ib_laddr_check(struct net *net, __be32 addr) /* Create a CMA ID and try to bind it. This catches both * IB and iWARP capable NICs. */ - cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC); + cm_id = rdma_create_id(&init_net, NULL, NULL, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(cm_id)) return PTR_ERR(cm_id); diff --git a/net/rds/ib_cm.c b/net/rds/ib_cm.c index 9043f5c04787..f5a98068faf0 100644 --- a/net/rds/ib_cm.c +++ b/net/rds/ib_cm.c @@ -565,7 +565,7 @@ int rds_ib_conn_connect(struct rds_connection *conn) /* XXX I wonder what affect the port space has */ /* delegate cm event handler to rdma_transport */ - ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn, + ic->i_cm_id = rdma_create_id(&init_net, rds_rdma_cm_event_handler, conn, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(ic->i_cm_id)) { ret = PTR_ERR(ic->i_cm_id); diff --git a/net/rds/iw.c b/net/rds/iw.c index 3df0295c6659..576f1825fc55 100644 --- a/net/rds/iw.c +++ b/net/rds/iw.c @@ -223,7 +223,7 @@ static int rds_iw_laddr_check(struct net *net, __be32 addr) /* Create a CMA ID and try to bind it. This catches both * IB and iWARP capable NICs. */ - cm_id = rdma_create_id(NULL, NULL, RDMA_PS_TCP, IB_QPT_RC); + cm_id = rdma_create_id(&init_net, NULL, NULL, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(cm_id)) return PTR_ERR(cm_id); diff --git a/net/rds/iw_cm.c b/net/rds/iw_cm.c index a6553a6fb2bc..aea4c911bc76 100644 --- a/net/rds/iw_cm.c +++ b/net/rds/iw_cm.c @@ -524,7 +524,7 @@ int rds_iw_conn_connect(struct rds_connection *conn) /* XXX I wonder what affect the port space has */ /* delegate cm event handler to rdma_transport */ - ic->i_cm_id = rdma_create_id(rds_rdma_cm_event_handler, conn, + ic->i_cm_id = rdma_create_id(&init_net, rds_rdma_cm_event_handler, conn, RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(ic->i_cm_id)) { ret = PTR_ERR(ic->i_cm_id); diff --git a/net/rds/rdma_transport.c b/net/rds/rdma_transport.c index b9b40af5345b..9c1fed81bf0f 100644 --- a/net/rds/rdma_transport.c +++ b/net/rds/rdma_transport.c @@ -142,8 +142,8 @@ static int rds_rdma_listen_init(void) struct rdma_cm_id *cm_id; int ret; - cm_id = rdma_create_id(rds_rdma_cm_event_handler, NULL, RDMA_PS_TCP, - IB_QPT_RC); + cm_id = rdma_create_id(&init_net, rds_rdma_cm_event_handler, NULL, + RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(cm_id)) { ret = PTR_ERR(cm_id); printk(KERN_ERR "RDS/RDMA: failed to setup listener, " diff --git a/net/sunrpc/xprtrdma/svc_rdma_transport.c b/net/sunrpc/xprtrdma/svc_rdma_transport.c index fcc3eb80c265..4a41122d586f 100644 --- a/net/sunrpc/xprtrdma/svc_rdma_transport.c +++ b/net/sunrpc/xprtrdma/svc_rdma_transport.c @@ -692,8 +692,8 @@ static struct svc_xprt *svc_rdma_create(struct svc_serv *serv, if (!cma_xprt) return ERR_PTR(-ENOMEM); - listen_id = rdma_create_id(rdma_listen_handler, cma_xprt, RDMA_PS_TCP, - IB_QPT_RC); + listen_id = rdma_create_id(&init_net, rdma_listen_handler, cma_xprt, + RDMA_PS_TCP, IB_QPT_RC); if (IS_ERR(listen_id)) { ret = PTR_ERR(listen_id); dprintk("svcrdma: rdma_create_id failed = %d\n", ret); diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index 81e8d31f3abc..6c06ba088fea 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -432,7 +432,8 @@ rpcrdma_create_id(struct rpcrdma_xprt *xprt, init_completion(&ia->ri_done); - id = rdma_create_id(rpcrdma_conn_upcall, xprt, RDMA_PS_TCP, IB_QPT_RC); + id = rdma_create_id(&init_net, rpcrdma_conn_upcall, xprt, RDMA_PS_TCP, + IB_QPT_RC); if (IS_ERR(id)) { rc = PTR_ERR(id); dprintk("RPC: %s: rdma_create_id() failed %i\n", -- cgit v1.2.3