Replace red-black tree by xarray
Subsystems (2)
cover Replace red-black tree by xarray
This series of patches starts with some minor cleanups in rxe pools and then replaces the red-black trees used to lookup rxe objects from indices with xarrays. The change over is made in small steps starting by adding xarrays and then convrting red-black trees to xarrays and finally deleting the original index code and renaming the xarray code to take its place.
These patches apply cleanly to current for-next branch after the "Correct race conditions in rdma_rxe" patch series
commit ed0d7445ed4651f4171c38d9446dc43c823ad32b applied to commit ac0fffa0859b8e1e991939663b3ebdd80bf979e6 (origin/for-next)
Bob Pearson (7):
RDMA/rxe: Replace irqsave locks with bh locks
RDMA/rxe: Cleanup rxe_pool_entry
RDMA/rxe: Add xarray support to rxe_pool.c
RDMA/rxe: Replace pool_lock by xa_lock
RDMA/rxe: Convert remaining pools to xarrays
RDMA/rxe: Remove old index code from rxe_pool.c
RDMA/rxe: Rename XARRAY as INDEX
drivers/infiniband/sw/rxe/rxe.c | 86 ++------- drivers/infiniband/sw/rxe/rxe_comp.c | 8 +- drivers/infiniband/sw/rxe/rxe_cq.c | 23 +-- drivers/infiniband/sw/rxe/rxe_loc.h | 10 +- drivers/infiniband/sw/rxe/rxe_mcast.c | 10 +- drivers/infiniband/sw/rxe/rxe_mr.c | 6 +- drivers/infiniband/sw/rxe/rxe_mw.c | 19 +- drivers/infiniband/sw/rxe/rxe_pool.c | 266 +++++++------------------- drivers/infiniband/sw/rxe/rxe_pool.h | 53 +++-- drivers/infiniband/sw/rxe/rxe_qp.c | 6 +- drivers/infiniband/sw/rxe/rxe_queue.c | 9 +- drivers/infiniband/sw/rxe/rxe_req.c | 11 +- drivers/infiniband/sw/rxe/rxe_srq.c | 6 +- drivers/infiniband/sw/rxe/rxe_task.c | 18 +- drivers/infiniband/sw/rxe/rxe_verbs.c | 29 ++- drivers/infiniband/sw/rxe/rxe_verbs.h | 22 +-- 16 files changed, 185 insertions(+), 397 deletions(-)
Signature · 1 line
2.30.2
PATCH 1/7 RDMA/rxe: Replace irqsave locks with bh locks
Most of the locks in the rxe driver are _irqsave/restore locks but in fact there are no interrupt threads that run rxe code or share data with rxe. There are softirq threads and data sharing so the appropriate lock type is _bh. This patch replaces all irqsave type locks with bh type locks.
--- drivers/infiniband/sw/rxe/rxe_comp.c | 8 ++--- drivers/infiniband/sw/rxe/rxe_cq.c | 19 +++++------- drivers/infiniband/sw/rxe/rxe_mcast.c | 10 +++---- drivers/infiniband/sw/rxe/rxe_mw.c | 15 ++++------ drivers/infiniband/sw/rxe/rxe_pool.c | 42 +++++++++++---------------- drivers/infiniband/sw/rxe/rxe_queue.c | 9 +++--- drivers/infiniband/sw/rxe/rxe_req.c | 11 +++---- drivers/infiniband/sw/rxe/rxe_task.c | 18 +++++------- drivers/infiniband/sw/rxe/rxe_verbs.c | 27 +++++++---------- 9 files changed, 65 insertions(+), 94 deletions(-)
drivers/infiniband/sw/rxe/rxe_comp.c +3 -5
@@ -458,8 +458,6 @@ static inline enum comp_state complete_ack(struct rxe_qp *qp, struct rxe_pkt_info *pkt, struct rxe_send_wqe *wqe) { - unsigned long flags; -if (wqe->has_rd_atomic) { wqe->has_rd_atomic =0 ;atomic_inc (&qp->req.rd_atomic);
@@ -472,11 +470,11 @@ static inline enum comp_state complete_ack(struct rxe_qp *qp,if (unlikely (qp->req.state ==QP_STATE_DRAIN )) {/* state_lock used by requester & completer */ -spin_lock_irqsave (&qp->state_lock, flags); +spin_lock_bh (&qp->state_lock);if ((qp->req.state ==QP_STATE_DRAIN ) && (qp->comp.psn == qp->req.psn)) { qp->req.state =QP_STATE_DRAINED ; -spin_unlock_irqrestore (&qp->state_lock, flags); +spin_unlock_bh (&qp->state_lock);if (qp->ibqp.event_handler) {struct ib_event ev;
@@ -488,7 +486,7 @@ static inline enum comp_state complete_ack(struct rxe_qp *qp, qp->ibqp.qp_context); } }else { -spin_unlock_irqrestore (&qp->state_lock, flags); +spin_unlock_bh (&qp->state_lock); } }
drivers/infiniband/sw/rxe/rxe_cq.c +8 -11
@@ -42,14 +42,13 @@ int rxe_cq_chk_attr(struct rxe_dev *rxe, struct rxe_cq *cq,static void rxe_send_complete (struct tasklet_struct *t) {struct rxe_cq *cq =from_tasklet (cq, t, comp_task); -unsigned long flags; -spin_lock_irqsave (&cq->cq_lock, flags); +spin_lock_bh (&cq->cq_lock);if (cq->is_dying) { -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock);return ; } -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock); cq->ibcq.comp_handler (&cq->ibcq, cq->ibcq.cq_context); }
@@ -106,15 +105,14 @@ int rxe_cq_resize_queue(struct rxe_cq *cq, int cqe,int rxe_cq_post (struct rxe_cq *cq,struct rxe_cqe *cqe,int solicited) {struct ib_event ev; -unsigned long flags;int full;void *addr; -spin_lock_irqsave (&cq->cq_lock, flags); +spin_lock_bh (&cq->cq_lock); full =queue_full (cq->queue,QUEUE_TYPE_TO_CLIENT );if (unlikely (full)) { -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock);if (cq->ibcq.event_handler) { ev.device = cq->ibcq.device; ev.element.cq = &cq->ibcq;
@@ -130,7 +128,7 @@ int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)queue_advance_producer (cq->queue,QUEUE_TYPE_TO_CLIENT ); -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock);if ((cq->notify ==IB_CQ_NEXT_COMP ) || (cq->notify ==IB_CQ_SOLICITED && solicited)) {
@@ -144,12 +142,11 @@ int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)void rxe_cq_cleanup (struct rxe_pool_entry *arg) {struct rxe_cq *cq =container_of (arg,typeof (*cq), pelem); -unsigned long flags;/* TODO get rid of this */ -spin_lock_irqsave (&cq->cq_lock, flags); +spin_lock_bh (&cq->cq_lock); cq->is_dying = true; -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock);if (cq->queue)rxe_queue_cleanup (cq->queue);
drivers/infiniband/sw/rxe/rxe_mcast.c +4 -6
@@ -33,14 +33,13 @@ static int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid, {struct rxe_pool *pool = &rxe->mc_grp_pool;struct rxe_mc_grp *grp; -unsigned long flags;int err =0 ;/* Perform this while holding the mc_grp_pool lock * to prevent races where two coincident calls fail to lookup the * same group and then both create the same group. */ -write_lock_irqsave (&pool->pool_lock, flags); +write_lock_bh (&pool->pool_lock); grp =rxe_pool_get_key_locked (pool, mgid);if (grp) goto done;
@@ -66,7 +65,7 @@ static int rxe_mcast_get_grp(struct rxe_dev *rxe, union ib_gid *mgid,rxe_add_ref_locked (grp); done: *grp_p = grp; -write_unlock_irqrestore (&pool->pool_lock, flags); +write_unlock_bh (&pool->pool_lock);return err; }
@@ -75,9 +74,8 @@ static void rxe_mcast_put_grp(struct rxe_mc_grp *grp) {struct rxe_dev *rxe = grp->rxe;struct rxe_pool *pool = &rxe->mc_grp_pool; -unsigned long flags; -write_lock_irqsave (&pool->pool_lock, flags); +write_lock_bh (&pool->pool_lock);rxe_drop_ref_locked (grp);
@@ -86,7 +84,7 @@ static void rxe_mcast_put_grp(struct rxe_mc_grp *grp)rxe_fini_ref_locked (grp); } -write_unlock_irqrestore (&pool->pool_lock, flags); +write_unlock_bh (&pool->pool_lock); }/**
drivers/infiniband/sw/rxe/rxe_mw.c +6 -9
@@ -55,12 +55,11 @@ int rxe_dealloc_mw(struct ib_mw *ibmw) {struct rxe_mw *mw =to_rmw (ibmw);struct rxe_pd *pd =to_rpd (ibmw->pd); -unsigned long flags;int err; -spin_lock_irqsave (&mw->lock, flags); +spin_lock_bh (&mw->lock);rxe_do_dealloc_mw (mw); -spin_unlock_irqrestore (&mw->lock, flags); +spin_unlock_bh (&mw->lock); err =rxe_fini_ref (mw);if (err)
@@ -199,7 +198,6 @@ int rxe_bind_mw(struct rxe_qp *qp, struct rxe_send_wqe *wqe)struct rxe_dev *rxe =to_rdev (qp->ibqp.device);u32 mw_rkey = wqe->wr.wr.mw.mw_rkey;u32 mr_lkey = wqe->wr.wr.mw.mr_lkey; -unsigned long flags; mw =rxe_pool_get_index (&rxe->mw_pool, mw_rkey >>8 );if (unlikely (!mw)) {
@@ -226,7 +224,7 @@ int rxe_bind_mw(struct rxe_qp *qp, struct rxe_send_wqe *wqe) } } -spin_lock_irqsave (&mw->lock, flags); +spin_lock_bh (&mw->lock); ret =rxe_check_bind_mw (qp, wqe, mw, mr);if (ret) {if (mr)
@@ -236,7 +234,7 @@ int rxe_bind_mw(struct rxe_qp *qp, struct rxe_send_wqe *wqe)rxe_do_bind_mw (qp, wqe, mw, mr); err_unlock: -spin_unlock_irqrestore (&mw->lock, flags); +spin_unlock_bh (&mw->lock); err_drop_mw:rxe_drop_ref (mw); err:
@@ -280,7 +278,6 @@ static void rxe_do_invalidate_mw(struct rxe_mw *mw)int rxe_invalidate_mw (struct rxe_qp *qp,u32 rkey) {struct rxe_dev *rxe =to_rdev (qp->ibqp.device); -unsigned long flags;struct rxe_mw *mw;int ret;
@@ -295,7 +292,7 @@ int rxe_invalidate_mw(struct rxe_qp *qp, u32 rkey) goto err_drop_ref; } -spin_lock_irqsave (&mw->lock, flags); +spin_lock_bh (&mw->lock); ret =rxe_check_invalidate_mw (qp, mw);if (ret)
@@ -303,7 +300,7 @@ int rxe_invalidate_mw(struct rxe_qp *qp, u32 rkey)rxe_do_invalidate_mw (mw); err_unlock: -spin_unlock_irqrestore (&mw->lock, flags); +spin_unlock_bh (&mw->lock); err_drop_ref:rxe_drop_ref (mw); err:
drivers/infiniband/sw/rxe/rxe_pool.c +17 -25
122 lines not on this page · raw message
drivers/infiniband/sw/rxe/rxe_queue.c +4 -5
@@ -151,7 +151,6 @@ int rxe_queue_resize(struct rxe_queue *q, unsigned int *num_elem_p,struct rxe_queue *new_q;unsigned int num_elem = *num_elem_p;int err; -unsigned long flags =0 , flags1; new_q =rxe_queue_init (q->rxe, &num_elem, elem_size, q->type);if (!new_q)
@@ -165,17 +164,17 @@ int rxe_queue_resize(struct rxe_queue *q, unsigned int *num_elem_p, goto err1; } -spin_lock_irqsave (consumer_lock, flags1); +spin_lock_bh (consumer_lock);if (producer_lock) { -spin_lock_irqsave (producer_lock, flags); +spin_lock_bh (producer_lock); err =resize_finish (q, new_q, num_elem); -spin_unlock_irqrestore (producer_lock, flags); +spin_unlock_bh (producer_lock); }else { err =resize_finish (q, new_q, num_elem); } -spin_unlock_irqrestore (consumer_lock, flags1); +spin_unlock_bh (consumer_lock);rxe_queue_cleanup (new_q);/* new/old dep on err */ if (err)
drivers/infiniband/sw/rxe/rxe_req.c +4 -7
@@ -110,7 +110,6 @@ void rnr_nak_timer(struct timer_list *t)static struct rxe_send_wqe *req_next_wqe (struct rxe_qp *qp) {struct rxe_send_wqe *wqe; -unsigned long flags;struct rxe_queue *q = qp->sq.queue;unsigned int index = qp->req.wqe_index;unsigned int cons;
@@ -124,25 +123,23 @@ static struct rxe_send_wqe *req_next_wqe(struct rxe_qp *qp)/* check to see if we are drained; * state_lock used by requester and completer */ -spin_lock_irqsave (&qp->state_lock, flags); +spin_lock_bh (&qp->state_lock); do { if (qp->req.state != QP_STATE_DRAIN) { /* comp just finished */ - spin_unlock_irqrestore(&qp->state_lock, - flags); + spin_unlock_bh(&qp->state_lock); break; } if (wqe && ((index != cons) || (wqe->state != wqe_state_posted))) { /* comp not done yet */ - spin_unlock_irqrestore(&qp->state_lock, - flags); + spin_unlock_bh(&qp->state_lock); break; } qp->req.state = QP_STATE_DRAINED; - spin_unlock_irqrestore(&qp->state_lock, flags); + spin_unlock_bh(&qp->state_lock); if (qp->ibqp.event_handler) { struct ib_event ev;
drivers/infiniband/sw/rxe/rxe_task.c +8 -10
@@ -32,25 +32,24 @@ void rxe_do_task(struct tasklet_struct *t) {int cont;int ret; -unsigned long flags;struct rxe_task *task =from_tasklet (task, t, tasklet); -spin_lock_irqsave (&task->state_lock, flags); +spin_lock_bh (&task->state_lock);switch (task->state) {case TASK_STATE_START : task->state =TASK_STATE_BUSY ; -spin_unlock_irqrestore (&task->state_lock, flags); +spin_unlock_bh (&task->state_lock);break ;case TASK_STATE_BUSY : task->state =TASK_STATE_ARMED ; fallthrough;case TASK_STATE_ARMED : -spin_unlock_irqrestore (&task->state_lock, flags); +spin_unlock_bh (&task->state_lock);return ;default : -spin_unlock_irqrestore (&task->state_lock, flags); +spin_unlock_bh (&task->state_lock);pr_warn ("%s failed with bad state %d\n" , __func__, task->state);return ; }
@@ -59,7 +58,7 @@ void rxe_do_task(struct tasklet_struct *t) cont =0 ; ret = task->func (task->arg); -spin_lock_irqsave (&task->state_lock, flags); +spin_lock_bh (&task->state_lock);switch (task->state) {case TASK_STATE_BUSY :if (ret)
@@ -81,7 +80,7 @@ void rxe_do_task(struct tasklet_struct *t)pr_warn ("%s failed with bad state %d\n" , __func__, task->state); } -spin_unlock_irqrestore (&task->state_lock, flags); +spin_unlock_bh (&task->state_lock); }while (cont); task->ret = ret;
@@ -106,7 +105,6 @@ int rxe_init_task(void *obj, struct rxe_task *task,void rxe_cleanup_task (struct rxe_task *task) { -unsigned long flags;bool idle;/*
@@ -116,9 +114,9 @@ void rxe_cleanup_task(struct rxe_task *task) task->destroyed = true;do { -spin_lock_irqsave (&task->state_lock, flags); +spin_lock_bh (&task->state_lock); idle = (task->state ==TASK_STATE_START ); -spin_unlock_irqrestore (&task->state_lock, flags); +spin_unlock_bh (&task->state_lock); }while (!idle);tasklet_kill (&task->tasklet);
drivers/infiniband/sw/rxe/rxe_verbs.c +11 -16
@@ -381,10 +381,9 @@ static int rxe_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, const struct ib_recv_wr **bad_wr) {int err =0 ; -unsigned long flags;struct rxe_srq *srq =to_rsrq (ibsrq); -spin_lock_irqsave (&srq->rq.producer_lock, flags); +spin_lock_bh (&srq->rq.producer_lock);while (wr) { err =post_one_recv (&srq->rq, wr);
@@ -393,7 +392,7 @@ static int rxe_post_srq_recv(struct ib_srq *ibsrq, const struct ib_recv_wr *wr, wr = wr->next; } -spin_unlock_irqrestore (&srq->rq.producer_lock, flags); +spin_unlock_bh (&srq->rq.producer_lock);if (err) *bad_wr = wr;
@@ -627,19 +626,18 @@ static int post_one_send(struct rxe_qp *qp, const struct ib_send_wr *ibwr,int err;struct rxe_sq *sq = &qp->sq;struct rxe_send_wqe *send_wqe; -unsigned long flags;int full; err =validate_send_wr (qp, ibwr, mask, length);if (err)return err; -spin_lock_irqsave (&qp->sq.sq_lock, flags); +spin_lock_bh (&qp->sq.sq_lock); full =queue_full (sq->queue,QUEUE_TYPE_TO_DRIVER );if (unlikely (full)) { -spin_unlock_irqrestore (&qp->sq.sq_lock, flags); +spin_unlock_bh (&qp->sq.sq_lock);return -ENOMEM ; }
@@ -648,7 +646,7 @@ static int post_one_send(struct rxe_qp *qp, const struct ib_send_wr *ibwr,queue_advance_producer (sq->queue,QUEUE_TYPE_TO_DRIVER ); -spin_unlock_irqrestore (&qp->sq.sq_lock, flags); +spin_unlock_bh (&qp->sq.sq_lock);return 0 ; }
@@ -728,7 +726,6 @@ static int rxe_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr,int err =0 ;struct rxe_qp *qp =to_rqp (ibqp);struct rxe_rq *rq = &qp->rq; -unsigned long flags;if (unlikely ((qp_state (qp) <IB_QPS_INIT ) || !qp->valid)) { *bad_wr = wr;
@@ -742,7 +739,7 @@ static int rxe_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, goto err1; } -spin_lock_irqsave (&rq->producer_lock, flags); +spin_lock_bh (&rq->producer_lock);while (wr) { err =post_one_recv (rq, wr);
@@ -753,7 +750,7 @@ static int rxe_post_recv(struct ib_qp *ibqp, const struct ib_recv_wr *wr, wr = wr->next; } -spin_unlock_irqrestore (&rq->producer_lock, flags); +spin_unlock_bh (&rq->producer_lock);if (qp->resp.state ==QP_STATE_ERROR )rxe_run_task (&qp->resp.task,1 );
@@ -831,9 +828,8 @@ static int rxe_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)int i;struct rxe_cq *cq =to_rcq (ibcq);struct rxe_cqe *cqe; -unsigned long flags; -spin_lock_irqsave (&cq->cq_lock, flags); +spin_lock_bh (&cq->cq_lock);for (i =0 ; i < num_entries; i++) { cqe =queue_head (cq->queue,QUEUE_TYPE_FROM_DRIVER );if (!cqe)
@@ -842,7 +838,7 @@ static int rxe_poll_cq(struct ib_cq *ibcq, int num_entries, struct ib_wc *wc)memcpy (wc++, &cqe->ibwc,sizeof (*wc));queue_advance_consumer (cq->queue,QUEUE_TYPE_FROM_DRIVER ); } -spin_unlock_irqrestore (&cq->cq_lock, flags); +spin_unlock_bh (&cq->cq_lock);return i; }
@@ -860,11 +856,10 @@ static int rxe_peek_cq(struct ib_cq *ibcq, int wc_cnt)static int rxe_req_notify_cq (struct ib_cq *ibcq,enum ib_cq_notify_flags flags) {struct rxe_cq *cq =to_rcq (ibcq); -unsigned long irq_flags;int ret =0 ;int empty; -spin_lock_irqsave (&cq->cq_lock, irq_flags); +spin_lock_bh (&cq->cq_lock);if (cq->notify !=IB_CQ_NEXT_COMP ) cq->notify = flags &IB_CQ_SOLICITED_MASK ;
@@ -873,7 +868,7 @@ static int rxe_req_notify_cq(struct ib_cq *ibcq, enum ib_cq_notify_flags flags)if ((flags &IB_CQ_REPORT_MISSED_EVENTS ) && !empty) ret =1 ; -spin_unlock_irqrestore (&cq->cq_lock, irq_flags); +spin_unlock_bh (&cq->cq_lock);return ret; }
Signature · 1 line
2.30.2
PATCH 2/7 RDMA/rxe: Cleanup rxe_pool_entry
In reply to Bob PearsonCurrently three different names are used to describe rxe pool elements. They are referred to as entries, elems or pelems. This patch chooses one 'elem' and changes the other ones.
--- drivers/infiniband/sw/rxe/rxe_cq.c | 4 +- drivers/infiniband/sw/rxe/rxe_loc.h | 10 ++-- drivers/infiniband/sw/rxe/rxe_mr.c | 6 +-- drivers/infiniband/sw/rxe/rxe_mw.c | 4 +- drivers/infiniband/sw/rxe/rxe_pool.c | 70 +++++++++++++-------------- drivers/infiniband/sw/rxe/rxe_pool.h | 38 +++++++-------- drivers/infiniband/sw/rxe/rxe_qp.c | 6 +-- drivers/infiniband/sw/rxe/rxe_srq.c | 6 +-- drivers/infiniband/sw/rxe/rxe_verbs.c | 2 +- drivers/infiniband/sw/rxe/rxe_verbs.h | 22 ++++----- 10 files changed, 84 insertions(+), 84 deletions(-)
drivers/infiniband/sw/rxe/rxe_cq.c +2 -2
@@ -139,9 +139,9 @@ int rxe_cq_post(struct rxe_cq *cq, struct rxe_cqe *cqe, int solicited)return 0 ; } -void rxe_cq_cleanup (struct rxe_pool_entry *arg) +void rxe_cq_cleanup (struct rxe_pool_elem *arg) { -struct rxe_cq *cq =container_of (arg,typeof (*cq), pelem); +struct rxe_cq *cq =container_of (arg,typeof (*cq), elem);/* TODO get rid of this */ spin_lock_bh (&cq->cq_lock);
drivers/infiniband/sw/rxe/rxe_loc.h +5 -5
@@ -35,7 +35,7 @@ int rxe_cq_resize_queue(struct rxe_cq *cq, int new_cqe,int rxe_cq_post (struct rxe_cq *cq,struct rxe_cqe *cqe,int solicited); -void rxe_cq_cleanup (struct rxe_pool_entry *arg); +void rxe_cq_cleanup (struct rxe_pool_elem *arg);/* rxe_mcast.c */ int rxe_mcast_add_grp_elem(struct rxe_dev *rxe, struct rxe_qp *qp,
@@ -80,7 +80,7 @@ int rxe_invalidate_mr(struct rxe_qp *qp, u32 rkey);int rxe_reg_fast_mr (struct rxe_qp *qp,struct rxe_send_wqe *wqe);int rxe_mr_set_page (struct ib_mr *ibmr,u64 addr);int rxe_dereg_mr (struct ib_mr *ibmr,struct ib_udata *udata); -void rxe_mr_cleanup (struct rxe_pool_entry *arg); +void rxe_mr_cleanup (struct rxe_pool_elem *arg);/* rxe_mw.c */ int rxe_alloc_mw (struct ib_mw *ibmw,struct ib_udata *udata);
@@ -88,7 +88,7 @@ int rxe_dealloc_mw(struct ib_mw *ibmw);int rxe_bind_mw (struct rxe_qp *qp,struct rxe_send_wqe *wqe);int rxe_invalidate_mw (struct rxe_qp *qp,u32 rkey);struct rxe_mw *rxe_lookup_mw (struct rxe_qp *qp,int access,u32 rkey); -void rxe_mw_cleanup (struct rxe_pool_entry *arg); +void rxe_mw_cleanup (struct rxe_pool_elem *arg);/* rxe_net.c */ struct sk_buff *rxe_init_packet(struct rxe_dev *rxe, struct rxe_av *av,
@@ -121,7 +121,7 @@ void rxe_qp_error(struct rxe_qp *qp);void rxe_qp_destroy (struct rxe_qp *qp); -void rxe_qp_cleanup (struct rxe_pool_entry *arg); +void rxe_qp_cleanup (struct rxe_pool_elem *arg);static inline int qp_num (struct rxe_qp *qp) {
@@ -177,7 +177,7 @@ int rxe_srq_from_init(struct rxe_dev *rxe, struct rxe_srq *srq,int rxe_srq_from_attr (struct rxe_dev *rxe,struct rxe_srq *srq,struct ib_srq_attr *attr,enum ib_srq_attr_mask mask,struct rxe_modify_srq_cmd *ucmd,struct ib_udata *udata); -void rxe_srq_cleanup (struct rxe_pool_entry *arg); +void rxe_srq_cleanup (struct rxe_pool_elem *arg);void rxe_dealloc (struct ib_device *ib_dev);
drivers/infiniband/sw/rxe/rxe_mr.c +3 -3
@@ -50,7 +50,7 @@ int mr_check_range(struct rxe_mr *mr, u64 iova, size_t length)static void rxe_mr_init (int access,struct rxe_mr *mr) { -u32 lkey = mr->pelem.index <<8 |rxe_get_next_key (-1 ); +u32 lkey = mr->elem.index <<8 |rxe_get_next_key (-1 );u32 rkey = (access &IB_ACCESS_REMOTE ) ? lkey :0 ;/* set ibmr->l/rkey and also copy into private l/rkey
@@ -704,9 +704,9 @@ int rxe_dereg_mr(struct ib_mr *ibmr, struct ib_udata *udata)return 0 ; } -void rxe_mr_cleanup (struct rxe_pool_entry *arg) +void rxe_mr_cleanup (struct rxe_pool_elem *arg) { -struct rxe_mr *mr =container_of (arg,typeof (*mr), pelem); +struct rxe_mr *mr =container_of (arg,typeof (*mr), elem);ib_umem_release (mr->umem);
drivers/infiniband/sw/rxe/rxe_mw.c +2 -2
@@ -20,7 +20,7 @@ int rxe_alloc_mw(struct ib_mw *ibmw, struct ib_udata *udata)return ret; } - mw->rkey = ibmw->rkey = (mw->pelem.index <<8 ) |rxe_get_next_key (-1 ); + mw->rkey = ibmw->rkey = (mw->elem.index <<8 ) |rxe_get_next_key (-1 ); mw->state = (mw->ibmw.type ==IB_MW_TYPE_2 ) ?RXE_MW_STATE_FREE :RXE_MW_STATE_VALID ;spin_lock_init (&mw->lock);
@@ -330,7 +330,7 @@ struct rxe_mw *rxe_lookup_mw(struct rxe_qp *qp, int access, u32 rkey)return mw; } -void rxe_mw_cleanup (struct rxe_pool_entry *elem) +void rxe_mw_cleanup (struct rxe_pool_elem *elem) {/* nothing to do currently */ }
drivers/infiniband/sw/rxe/rxe_pool.c +35 -35
266 lines not on this page · raw message
drivers/infiniband/sw/rxe/rxe_pool.h +19 -19
@@ -31,13 +31,13 @@ enum rxe_elem_type { RXE_NUM_TYPES, /* keep me last */ }; -struct rxe_pool_entry ; +struct rxe_pool_elem ;struct rxe_type_info {const char *name;size_t size;size_t elem_offset; -void (*cleanup)(struct rxe_pool_entry *obj); +void (*cleanup)(struct rxe_pool_elem *obj);enum rxe_pool_flags flags;u32 max_index;u32 min_index;
@@ -45,7 +45,7 @@ struct rxe_type_info {size_t key_size; }; -struct rxe_pool_entry { +struct rxe_pool_elem {struct rxe_pool *pool;void *obj;refcount_t refcnt;
@@ -63,7 +63,7 @@ struct rxe_pool {struct rxe_dev *rxe;const char *name;rwlock_t pool_lock;/* protects pool add/del/search */ -void (*cleanup)(struct rxe_pool_entry *obj); +void (*cleanup)(struct rxe_pool_elem *obj);enum rxe_pool_flags flags;enum rxe_elem_type type;
@@ -110,9 +110,9 @@ void *rxe_alloc_with_key_locked(struct rxe_pool *pool, void *key);void *rxe_alloc_with_key (struct rxe_pool *pool,void *key);/* connect already allocated object to pool */ -int __rxe_add_to_pool (struct rxe_pool *pool,struct rxe_pool_entry *elem); +int __rxe_add_to_pool (struct rxe_pool *pool,struct rxe_pool_elem *elem); -#define rxe_add_to_pool (pool, obj) __rxe_add_to_pool(pool, &(obj)->pelem) +#define rxe_add_to_pool (pool, obj) __rxe_add_to_pool(pool, &(obj)->elem)/* lookup an indexed object from index holding and not holding the pool_lock. * takes a reference on object
@@ -129,32 +129,32 @@ void *rxe_pool_get_key_locked(struct rxe_pool *pool, void *key);void *rxe_pool_get_key (struct rxe_pool *pool,void *key);/* take a reference on an object */ -int __rxe_add_ref_locked (struct rxe_pool_entry *elem); +int __rxe_add_ref_locked (struct rxe_pool_elem *elem); -#define rxe_add_ref_locked (obj) __rxe_add_ref_locked(&(obj)->pelem) +#define rxe_add_ref_locked (obj) __rxe_add_ref_locked(&(obj)->elem) -int __rxe_add_ref (struct rxe_pool_entry *elem); +int __rxe_add_ref (struct rxe_pool_elem *elem); -#define rxe_add_ref (obj) __rxe_add_ref(&(obj)->pelem) +#define rxe_add_ref (obj) __rxe_add_ref(&(obj)->elem)/* drop a reference on an object */ -int __rxe_drop_ref_locked (struct rxe_pool_entry *elem); +int __rxe_drop_ref_locked (struct rxe_pool_elem *elem); -#define rxe_drop_ref_locked (obj) __rxe_drop_ref_locked(&(obj)->pelem) +#define rxe_drop_ref_locked (obj) __rxe_drop_ref_locked(&(obj)->elem) -int __rxe_drop_ref (struct rxe_pool_entry *elem); +int __rxe_drop_ref (struct rxe_pool_elem *elem); -#define rxe_drop_ref (obj) __rxe_drop_ref(&(obj)->pelem) +#define rxe_drop_ref (obj) __rxe_drop_ref(&(obj)->elem)/* drop last reference on an object */ -int __rxe_fini_ref_locked (struct rxe_pool_entry *elem); +int __rxe_fini_ref_locked (struct rxe_pool_elem *elem); -#define rxe_fini_ref_locked (obj) __rxe_fini_ref_locked(&(obj)->pelem) +#define rxe_fini_ref_locked (obj) __rxe_fini_ref_locked(&(obj)->elem) -int __rxe_fini_ref (struct rxe_pool_entry *elem); +int __rxe_fini_ref (struct rxe_pool_elem *elem); -#define rxe_fini_ref (obj) __rxe_fini_ref(&(obj)->pelem) +#define rxe_fini_ref (obj) __rxe_fini_ref(&(obj)->elem) -#define rxe_read_ref (obj) refcount_read(&(obj)->pelem.refcnt) +#define rxe_read_ref (obj) refcount_read(&(obj)->elem.refcnt)#endif /* RXE_POOL_H */
drivers/infiniband/sw/rxe/rxe_qp.c +3 -3
@@ -163,7 +163,7 @@ static void rxe_qp_init_misc(struct rxe_dev *rxe, struct rxe_qp *qp, qp->attr.path_mtu =1 ; qp->mtu =ib_mtu_enum_to_int (qp->attr.path_mtu); - qpn = qp->pelem.index; + qpn = qp->elem.index; port = &rxe->port;switch (init->qp_type) {
@@ -825,9 +825,9 @@ static void rxe_qp_do_cleanup(struct work_struct *work) }/* called when the last reference to the qp is dropped */ -void rxe_qp_cleanup (struct rxe_pool_entry *arg) +void rxe_qp_cleanup (struct rxe_pool_elem *arg) { -struct rxe_qp *qp =container_of (arg,typeof (*qp), pelem); +struct rxe_qp *qp =container_of (arg,typeof (*qp), elem);rxe_qp_destroy (qp);execute_in_process_context (rxe_qp_do_cleanup, &qp->cleanup_work);
drivers/infiniband/sw/rxe/rxe_srq.c +3 -3
@@ -83,7 +83,7 @@ int rxe_srq_from_init(struct rxe_dev *rxe, struct rxe_srq *srq, srq->ibsrq.event_handler = init->event_handler; srq->ibsrq.srq_context = init->srq_context; srq->limit = init->attr.srq_limit; - srq->srq_num = srq->pelem.index; + srq->srq_num = srq->elem.index; srq->rq.max_wr = init->attr.max_wr; srq->rq.max_sge = init->attr.max_sge;
@@ -155,9 +155,9 @@ int rxe_srq_from_attr(struct rxe_dev *rxe, struct rxe_srq *srq,return err; } -void rxe_srq_cleanup (struct rxe_pool_entry *arg) +void rxe_srq_cleanup (struct rxe_pool_elem *arg) { -struct rxe_srq *srq =container_of (arg,typeof (*srq), pelem); +struct rxe_srq *srq =container_of (arg,typeof (*srq), elem);if (srq->rq.queue)rxe_queue_cleanup (srq->rq.queue);
drivers/infiniband/sw/rxe/rxe_verbs.c +1 -1
@@ -180,7 +180,7 @@ static int rxe_create_ah(struct ib_ah *ibah,return err;/* create index > 0 */ - ah->ah_num = ah->pelem.index; + ah->ah_num = ah->elem.index;if (uresp) {/* only if new user provider */
drivers/infiniband/sw/rxe/rxe_verbs.h +11 -11
@@ -35,17 +35,17 @@ static inline int psn_compare(u32 psn_a, u32 psn_b)struct rxe_ucontext {struct ib_ucontext ibuc; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem; };struct rxe_pd {struct ib_pd ibpd; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem; };struct rxe_ah {struct ib_ah ibah; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct rxe_av av;bool is_user;int ah_num;
@@ -60,7 +60,7 @@ struct rxe_cqe {struct rxe_cq {struct ib_cq ibcq; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct rxe_queue *queue;spinlock_t cq_lock;u8 notify;
@@ -95,7 +95,7 @@ struct rxe_rq {struct rxe_srq {struct ib_srq ibsrq; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct rxe_pd *pd;struct rxe_rq rq;u32 srq_num;
@@ -208,7 +208,7 @@ struct rxe_resp_info {struct rxe_qp {struct ib_qp ibqp; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct ib_qp_attr attr;unsigned int valid;unsigned int mtu;
@@ -308,7 +308,7 @@ static inline int rkey_is_mw(u32 rkey) }struct rxe_mr { -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct ib_mr ibmr;struct ib_umem *umem;
@@ -341,7 +341,7 @@ enum rxe_mw_state {struct rxe_mw {struct ib_mw ibmw; -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;spinlock_t lock;enum rxe_mw_state state;struct rxe_qp *qp;/* Type 2 only */
@@ -353,7 +353,7 @@ struct rxe_mw { };struct rxe_mc_grp { -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;spinlock_t mcg_lock;/* guard group */ struct rxe_dev *rxe;struct list_head qp_list;
@@ -364,7 +364,7 @@ struct rxe_mc_grp { };struct rxe_mc_elem { -struct rxe_pool_entry pelem; +struct rxe_pool_elem elem;struct list_head qp_list;struct list_head grp_list;struct rxe_qp *qp;
@@ -484,6 +484,6 @@ static inline struct rxe_pd *rxe_mw_pd(struct rxe_mw *mw)int rxe_register_device (struct rxe_dev *rxe,const char *ibdev_name); -void rxe_mc_cleanup (struct rxe_pool_entry *arg); +void rxe_mc_cleanup (struct rxe_pool_elem *arg);#endif /* RXE_VERBS_H */
Signature · 1 line
2.30.2
PATCH 3/7 RDMA/rxe: Add xarray support to rxe_pool.c
In reply to Bob PearsonCurrently the rxe driver uses red-black trees to add indices and keys to the rxe object pool. Linux xarrays provide a better way to implement the same functionality for indices but not keys. This patch adds a second alternative to adding indices based on cyclic allocating xarrays. The AH pool is modified to hold either xarrays or red-black trees. The code is tested for both options.
--- drivers/infiniband/sw/rxe/rxe_pool.c | 100 ++++++++++++++++++++++----- drivers/infiniband/sw/rxe/rxe_pool.h | 9 +++ 2 files changed, 92 insertions(+), 17 deletions(-)
drivers/infiniband/sw/rxe/rxe_pool.c +83 -17
@@ -25,7 +25,8 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .name = "rxe-ah", .size = sizeof(struct rxe_ah), .elem_offset = offsetof(struct rxe_ah, elem), - .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, +//.flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_AH_INDEX, .max_index = RXE_MAX_AH_INDEX, },
@@ -128,15 +129,20 @@ int rxe_pool_init( pool->elem_size =ALIGN (info->size,RXE_POOL_ALIGN ); pool->elem_offset = info->elem_offset; pool->flags = info->flags; - pool->index.tree =RB_ROOT ; - pool->key.tree =RB_ROOT ; pool->cleanup = info->cleanup;atomic_set (&pool->num_elem,0 );rwlock_init (&pool->pool_lock); +if (info->flags &RXE_POOL_XARRAY ) { +xa_init_flags (&pool->xarray.xa,XA_FLAGS_ALLOC ); + pool->xarray.limit.max = info->max_index; + pool->xarray.limit.min = info->min_index; + } +if (info->flags &RXE_POOL_INDEX ) { + pool->index.tree =RB_ROOT ; err =rxe_pool_init_index (pool, info->max_index, info->min_index);if (err)
@@ -144,6 +150,7 @@ int rxe_pool_init( }if (info->flags &RXE_POOL_KEY ) { + pool->key.tree =RB_ROOT ; pool->key.key_offset = info->key_offset; pool->key.key_size = info->key_size; }
@@ -158,7 +165,8 @@ void rxe_pool_cleanup(struct rxe_pool *pool)pr_warn ("%s pool destroyed with unfree'd elem\n" , pool->name); -kfree (pool->index.table); +if (pool->flags &RXE_POOL_INDEX ) +kfree (pool->index.table); }/* should never fail because there are at least as many indices as
@@ -272,28 +280,35 @@ static void *__rxe_alloc_locked(struct rxe_pool *pool)int err;if (atomic_inc_return (&pool->num_elem) > pool->max_elem) - goto out_cnt; + goto err; obj =kzalloc (pool->elem_size,GFP_ATOMIC );if (!obj) - goto out_cnt; + goto err; elem = (struct rxe_pool_elem *)((u8 *)obj + pool->elem_offset); elem->pool = pool; elem->obj = obj; +if (pool->flags &RXE_POOL_XARRAY ) { + err =xa_alloc_cyclic_bh (&pool->xarray.xa, &elem->index, elem, + pool->xarray.limit, + &pool->xarray.next,GFP_KERNEL ); +if (err) + goto err; + } +if (pool->flags &RXE_POOL_INDEX ) { err =rxe_add_index (elem); -if (err) { -kfree (obj); - goto out_cnt; - } +if (err) + goto err; }return obj; -out_cnt: +err: +kfree (obj);atomic_dec (&pool->num_elem);return NULL ; }
@@ -368,15 +383,23 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)write_lock_bh (&pool->pool_lock);if (atomic_inc_return (&pool->num_elem) > pool->max_elem) - goto out_cnt; + goto err; elem->pool = pool; elem->obj = (u8 *)elem - pool->elem_offset; +if (pool->flags &RXE_POOL_XARRAY ) { + err =xa_alloc_cyclic_bh (&pool->xarray.xa, &elem->index, elem, + pool->xarray.limit, + &pool->xarray.next,GFP_KERNEL ); +if (err) + goto err; + } +if (pool->flags &RXE_POOL_INDEX ) { err =rxe_add_index (elem);if (err) - goto out_cnt; + goto err; }refcount_set (&elem->refcnt,1 );
@@ -384,13 +407,13 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)return 0 ; -out_cnt: +err:atomic_dec (&pool->num_elem);write_unlock_bh (&pool->pool_lock);return -EINVAL ; } -void *rxe_pool_get_index_locked (struct rxe_pool *pool,u32 index) +static void *__rxe_get_index_locked (struct rxe_pool *pool,u32 index) {struct rb_node *node;struct rxe_pool_elem *elem;
@@ -415,17 +438,58 @@ void *rxe_pool_get_index_locked(struct rxe_pool *pool, u32 index)return obj; } -void *rxe_pool_get_index (struct rxe_pool *pool,u32 index) +static void *__rxe_get_index (struct rxe_pool *pool,u32 index) +{ +void *obj; + +read_lock_bh (&pool->pool_lock); + obj =__rxe_get_index_locked (pool, index); +read_unlock_bh (&pool->pool_lock); + +return obj; +} + +static void *__rxe_get_xarray_locked (struct rxe_pool *pool,u32 index) +{ +struct rxe_pool_elem *elem; +void *obj =NULL ; + + elem =xa_load (&pool->xarray.xa, index); +if (elem &&refcount_inc_not_zero (&elem->refcnt)) + obj = elem->obj; + +return obj; +} + +static void *__rxe_get_xarray (struct rxe_pool *pool,u32 index) {void *obj;read_lock_bh (&pool->pool_lock); - obj =rxe_pool_get_index_locked (pool, index); + obj =__rxe_get_xarray_locked (pool, index);read_unlock_bh (&pool->pool_lock);return obj; } +void *rxe_pool_get_index_locked (struct rxe_pool *pool,u32 index) +{ +if (pool->flags &RXE_POOL_XARRAY ) +return __rxe_get_xarray_locked (pool, index); +if (pool->flags &RXE_POOL_INDEX ) +return __rxe_get_index_locked (pool, index); +return NULL ; +} + +void *rxe_pool_get_index (struct rxe_pool *pool,u32 index) +{ +if (pool->flags &RXE_POOL_XARRAY ) +return __rxe_get_xarray (pool, index); +if (pool->flags &RXE_POOL_INDEX ) +return __rxe_get_index (pool, index); +return NULL ; +} +void *rxe_pool_get_key_locked (struct rxe_pool *pool,void *key) {struct rb_node *node;
@@ -519,6 +583,8 @@ static int __rxe_fini(struct rxe_pool_elem *elem) done =refcount_dec_if_one (&elem->refcnt);if (done) { +if (pool->flags &RXE_POOL_XARRAY ) +xa_erase (&pool->xarray.xa, elem->index);if (pool->flags &RXE_POOL_INDEX )rxe_drop_index (elem);if (pool->flags &RXE_POOL_KEY )
drivers/infiniband/sw/rxe/rxe_pool.h +9 -0
@@ -8,6 +8,7 @@#define RXE_POOL_H #include <linux/refcount.h> +#include <linux/xarray.h> #define RXE_POOL_ALIGN (16)#define RXE_POOL_CACHE_FLAGS (0)
@@ -15,6 +16,7 @@enum rxe_pool_flags {RXE_POOL_INDEX =BIT (1 ),RXE_POOL_KEY =BIT (2 ), +RXE_POOL_XARRAY =BIT (3 ),RXE_POOL_NO_ALLOC =BIT (4 ), };
@@ -72,6 +74,13 @@ struct rxe_pool {size_t elem_size;size_t elem_offset; +/* only used if xarray */ +struct { +struct xarray xa; +struct xa_limit limit; +u32 next; + } xarray; +/* only used if indexed */ struct {struct rb_root tree;
Signature · 1 line
2.30.2
Re: [PATCH for-next 3/7] RDMA/rxe: Add xarray support to rxe_pool.c
drivers/infiniband/sw/rxe/rxe_pool.c static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { · line 29
Quoted 3 lines · Bob Pearson
drivers/infiniband/sw/rxe/rxe_pool.c +2
-1- .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, +//.flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC,
Assume you meant to remove that comment line.
-Denny
Re: [PATCH for-next 3/7] RDMA/rxe: Add xarray support to rxe_pool.c
Quoted 9 lines · Dennis Dalessandro
drivers/infiniband/sw/rxe/rxe_pool.c static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { · line 29
Quoted 3 lines · Bob Pearson
drivers/infiniband/sw/rxe/rxe_pool.c +2
-1- .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, +//.flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC,Assume you meant to remove that comment line.
-Denny
This was an intentionally intermediate step and was a hint that you can run with either choice. In the end of the series all the old stuff goes away including the comment.
PATCH 4/7 RDMA/rxe: Replace pool_lock by xa_lock
In reply to Bob PearsonIn rxe_pool.c xa_alloc_bh and xa_erase_bh and variants already include
spin_lock_bh() __xa_alloc() spin_unlock_bh()
So we are double locking. Replacing pool_lock by xa_lock and using xa_lock in all the places that were previously locked by pool_lock but dropping the double locks is a performance improvement.
--- drivers/infiniband/sw/rxe/rxe_pool.c | 54 ++++++++++++++-------------- 1 file changed, 26 insertions(+), 28 deletions(-)
drivers/infiniband/sw/rxe/rxe_pool.c +26 -28
@@ -133,8 +133,6 @@ int rxe_pool_init(atomic_set (&pool->num_elem,0 ); -rwlock_init (&pool->pool_lock); -if (info->flags &RXE_POOL_XARRAY ) {xa_init_flags (&pool->xarray.xa,XA_FLAGS_ALLOC ); pool->xarray.limit.max = info->max_index;
@@ -292,9 +290,9 @@ static void *__rxe_alloc_locked(struct rxe_pool *pool) elem->obj = obj;if (pool->flags &RXE_POOL_XARRAY ) { - err =xa_alloc_cyclic_bh (&pool->xarray.xa, &elem->index, elem, - pool->xarray.limit, - &pool->xarray.next,GFP_KERNEL ); + err =__xa_alloc_cyclic (&pool->xarray.xa, &elem->index, elem, + pool->xarray.limit, + &pool->xarray.next,GFP_KERNEL );if (err) goto err; }
@@ -359,9 +357,9 @@ void *rxe_alloc(struct rxe_pool *pool) {void *obj; -write_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); obj =rxe_alloc_locked (pool); -write_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return obj; }
@@ -370,9 +368,9 @@ void *rxe_alloc_with_key(struct rxe_pool *pool, void *key) {void *obj; -write_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); obj =rxe_alloc_with_key_locked (pool, key); -write_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return obj; }
@@ -381,7 +379,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem) {int err; -write_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa);if (atomic_inc_return (&pool->num_elem) > pool->max_elem) goto err;
@@ -389,9 +387,9 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem) elem->obj = (u8 *)elem - pool->elem_offset;if (pool->flags &RXE_POOL_XARRAY ) { - err =xa_alloc_cyclic_bh (&pool->xarray.xa, &elem->index, elem, - pool->xarray.limit, - &pool->xarray.next,GFP_KERNEL ); + err =__xa_alloc_cyclic (&pool->xarray.xa, &elem->index, elem, + pool->xarray.limit, + &pool->xarray.next,GFP_KERNEL );if (err) goto err; }
@@ -403,13 +401,13 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem) }refcount_set (&elem->refcnt,1 ); -write_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return 0 ; err:atomic_dec (&pool->num_elem); -write_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return -EINVAL ; }
@@ -442,9 +440,9 @@ static void *__rxe_get_index(struct rxe_pool *pool, u32 index) {void *obj; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); obj =__rxe_get_index_locked (pool, index); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return obj; }
@@ -465,9 +463,9 @@ static void *__rxe_get_xarray(struct rxe_pool *pool, u32 index) {void *obj; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); obj =__rxe_get_xarray_locked (pool, index); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return obj; }
@@ -523,9 +521,9 @@ void *rxe_pool_get_key(struct rxe_pool *pool, void *key) {void *obj; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); obj =rxe_pool_get_key_locked (pool, key); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return obj; }
@@ -546,9 +544,9 @@ int __rxe_add_ref(struct rxe_pool_elem *elem)struct rxe_pool *pool = elem->pool;int ret; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); ret =__rxe_add_ref_locked (elem); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return ret; }
@@ -569,9 +567,9 @@ int __rxe_drop_ref(struct rxe_pool_elem *elem)struct rxe_pool *pool = elem->pool;int ret; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); ret =__rxe_drop_ref_locked (elem); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);return ret; }
@@ -584,7 +582,7 @@ static int __rxe_fini(struct rxe_pool_elem *elem) done =refcount_dec_if_one (&elem->refcnt);if (done) {if (pool->flags &RXE_POOL_XARRAY ) -xa_erase (&pool->xarray.xa, elem->index); +__xa_erase (&pool->xarray.xa, elem->index);if (pool->flags &RXE_POOL_INDEX )rxe_drop_index (elem);if (pool->flags &RXE_POOL_KEY )
@@ -621,9 +619,9 @@ int __rxe_fini_ref(struct rxe_pool_elem *elem)struct rxe_pool *pool = elem->pool;int ret; -read_lock_bh (&pool->pool_lock); +xa_lock_bh (&pool->xarray.xa); ret =__rxe_fini (elem); -read_unlock_bh (&pool->pool_lock); +xa_unlock_bh (&pool->xarray.xa);if (!ret) {if (pool->cleanup)
Signature · 1 line
2.30.2
PATCH 5/7 RDMA/rxe: Convert remaining pools to xarrays
In reply to Bob PearsonThis patch converts the remaining pools with RXE_POOL_INDEX set to RXE_POOL_XARRAY.
--- drivers/infiniband/sw/rxe/rxe_pool.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)
drivers/infiniband/sw/rxe/rxe_pool.c +4 -5
@@ -25,7 +25,6 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .name = "rxe-ah", .size = sizeof(struct rxe_ah), .elem_offset = offsetof(struct rxe_ah, elem), -//.flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_AH_INDEX, .max_index = RXE_MAX_AH_INDEX,
@@ -35,7 +34,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_srq), .elem_offset = offsetof(struct rxe_srq, elem), .cleanup = rxe_srq_cleanup, - .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_SRQ_INDEX, .max_index = RXE_MAX_SRQ_INDEX, },
@@ -44,7 +43,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_qp), .elem_offset = offsetof(struct rxe_qp, elem), .cleanup = rxe_qp_cleanup, - .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_QP_INDEX, .max_index = RXE_MAX_QP_INDEX, },
@@ -60,7 +59,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_mr), .elem_offset = offsetof(struct rxe_mr, elem), .cleanup = rxe_mr_cleanup, - .flags = RXE_POOL_INDEX, + .flags = RXE_POOL_XARRAY, .max_index = RXE_MAX_MR_INDEX, .min_index = RXE_MIN_MR_INDEX, },
@@ -69,7 +68,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_mw), .elem_offset = offsetof(struct rxe_mw, elem), .cleanup = rxe_mw_cleanup, - .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, .max_index = RXE_MAX_MW_INDEX, .min_index = RXE_MIN_MW_INDEX, },
Signature · 1 line
2.30.2
PATCH 6/7 RDMA/rxe: Remove old index code from rxe_pool.c
In reply to Bob PearsonRemove all red-black tree based index code from rxe_pool.c. Change some functions from int to void as errors are no longer returned.
--- drivers/infiniband/sw/rxe/rxe.c | 86 ++------------ drivers/infiniband/sw/rxe/rxe_pool.c | 171 +-------------------------- drivers/infiniband/sw/rxe/rxe_pool.h | 14 +-- 3 files changed, 15 insertions(+), 256 deletions(-)
drivers/infiniband/sw/rxe/rxe.c +13 -73
@@ -115,90 +115,37 @@ static void rxe_init_ports(struct rxe_dev *rxe) }/* init pools of managed objects */ -static int rxe_init_pools (struct rxe_dev *rxe) +static void rxe_init_pools (struct rxe_dev *rxe) { -int err; - - err =rxe_pool_init (rxe, &rxe->uc_pool,RXE_TYPE_UC , +rxe_pool_init (rxe, &rxe->uc_pool,RXE_TYPE_UC , rxe->max_ucontext); -if (err) - goto err1; - - err =rxe_pool_init (rxe, &rxe->pd_pool,RXE_TYPE_PD , +rxe_pool_init (rxe, &rxe->pd_pool,RXE_TYPE_PD , rxe->attr.max_pd); -if (err) - goto err2; - - err =rxe_pool_init (rxe, &rxe->ah_pool,RXE_TYPE_AH , +rxe_pool_init (rxe, &rxe->ah_pool,RXE_TYPE_AH , rxe->attr.max_ah); -if (err) - goto err3; - - err =rxe_pool_init (rxe, &rxe->srq_pool,RXE_TYPE_SRQ , +rxe_pool_init (rxe, &rxe->srq_pool,RXE_TYPE_SRQ , rxe->attr.max_srq); -if (err) - goto err4; - - err =rxe_pool_init (rxe, &rxe->qp_pool,RXE_TYPE_QP , +rxe_pool_init (rxe, &rxe->qp_pool,RXE_TYPE_QP , rxe->attr.max_qp); -if (err) - goto err5; - - err =rxe_pool_init (rxe, &rxe->cq_pool,RXE_TYPE_CQ , +rxe_pool_init (rxe, &rxe->cq_pool,RXE_TYPE_CQ , rxe->attr.max_cq); -if (err) - goto err6; - - err =rxe_pool_init (rxe, &rxe->mr_pool,RXE_TYPE_MR , +rxe_pool_init (rxe, &rxe->mr_pool,RXE_TYPE_MR , rxe->attr.max_mr); -if (err) - goto err7; - - err =rxe_pool_init (rxe, &rxe->mw_pool,RXE_TYPE_MW , +rxe_pool_init (rxe, &rxe->mw_pool,RXE_TYPE_MW , rxe->attr.max_mw); -if (err) - goto err8; - - err =rxe_pool_init (rxe, &rxe->mc_grp_pool,RXE_TYPE_MC_GRP , +rxe_pool_init (rxe, &rxe->mc_grp_pool,RXE_TYPE_MC_GRP , rxe->attr.max_mcast_grp); -if (err) - goto err9; - -return 0 ; - -err9: -rxe_pool_cleanup (&rxe->mw_pool); -err8: -rxe_pool_cleanup (&rxe->mr_pool); -err7: -rxe_pool_cleanup (&rxe->cq_pool); -err6: -rxe_pool_cleanup (&rxe->qp_pool); -err5: -rxe_pool_cleanup (&rxe->srq_pool); -err4: -rxe_pool_cleanup (&rxe->ah_pool); -err3: -rxe_pool_cleanup (&rxe->pd_pool); -err2: -rxe_pool_cleanup (&rxe->uc_pool); -err1: -return err; }/* initialize rxe device state */ -static int rxe_init (struct rxe_dev *rxe) +static void rxe_init (struct rxe_dev *rxe) { -int err; -/* init default device parameters */ rxe_init_device_param (rxe);rxe_init_ports (rxe); - err =rxe_init_pools (rxe); -if (err) -return err; +rxe_init_pools (rxe);/* init pending mmap list */ spin_lock_init (&rxe->mmap_offset_lock);
@@ -206,8 +153,6 @@ static int rxe_init(struct rxe_dev *rxe)INIT_LIST_HEAD (&rxe->pending_mmaps);mutex_init (&rxe->usdev_lock); - -return 0 ; }void rxe_set_mtu (struct rxe_dev *rxe,unsigned int ndev_mtu)
@@ -229,12 +174,7 @@ void rxe_set_mtu(struct rxe_dev *rxe, unsigned int ndev_mtu)*/ int rxe_add (struct rxe_dev *rxe,unsigned int mtu,const char *ibdev_name) { -int err; - - err =rxe_init (rxe); -if (err) -return err; - +rxe_init (rxe);rxe_set_mtu (rxe, mtu);return rxe_register_device (rxe, ibdev_name);
drivers/infiniband/sw/rxe/rxe_pool.c +1 -170
242 lines not on this page · raw message
drivers/infiniband/sw/rxe/rxe_pool.h +1 -13
@@ -14,7 +14,6 @@#define RXE_POOL_CACHE_FLAGS (0)enum rxe_pool_flags { -RXE_POOL_INDEX =BIT (1 ),RXE_POOL_KEY =BIT (2 ),RXE_POOL_XARRAY =BIT (3 ),RXE_POOL_NO_ALLOC =BIT (4 ),
@@ -57,7 +56,6 @@ struct rxe_pool_elem {struct rb_node key_node;/* only used if indexed */ -struct rb_node index_node;u32 index; };
@@ -81,16 +79,6 @@ struct rxe_pool {u32 next; } xarray; -/* only used if indexed */ -struct { -struct rb_root tree; -unsigned long *table; -size_t table_size; -u32 last; -u32 max_index; -u32 min_index; - } index; -/* only used if keyed */ struct {struct rb_root tree;
@@ -103,7 +91,7 @@ struct rxe_pool {* number of elements. gets parameters from rxe_type_info * pool elements will be allocated out of a slab cache */ -int rxe_pool_init (struct rxe_dev *rxe,struct rxe_pool *pool, +void rxe_pool_init (struct rxe_dev *rxe,struct rxe_pool *pool,enum rxe_elem_type type,u32 max_elem);/* free resources from object pool */
Signature · 1 line
2.30.2
PATCH 7/7 RDMA/rxe: Rename XARRAY as INDEX
In reply to Bob PearsonRename RXE_POOL_XARRAY as RXE_POOL_INDEX and change several function names .._index_... from .._xarray_.. which completes the process of replacing red-black trees by xarrays.
--- drivers/infiniband/sw/rxe/rxe_pool.c | 38 +++++++++------------------- drivers/infiniband/sw/rxe/rxe_pool.h | 4 +-- 2 files changed, 14 insertions(+), 28 deletions(-)
drivers/infiniband/sw/rxe/rxe_pool.c +12 -26
@@ -25,7 +25,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .name = "rxe-ah", .size = sizeof(struct rxe_ah), .elem_offset = offsetof(struct rxe_ah, elem), - .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_AH_INDEX, .max_index = RXE_MAX_AH_INDEX, },
@@ -34,7 +34,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_srq), .elem_offset = offsetof(struct rxe_srq, elem), .cleanup = rxe_srq_cleanup, - .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_SRQ_INDEX, .max_index = RXE_MAX_SRQ_INDEX, },
@@ -43,7 +43,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_qp), .elem_offset = offsetof(struct rxe_qp, elem), .cleanup = rxe_qp_cleanup, - .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .min_index = RXE_MIN_QP_INDEX, .max_index = RXE_MAX_QP_INDEX, },
@@ -59,7 +59,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_mr), .elem_offset = offsetof(struct rxe_mr, elem), .cleanup = rxe_mr_cleanup, - .flags = RXE_POOL_XARRAY, + .flags = RXE_POOL_INDEX, .max_index = RXE_MAX_MR_INDEX, .min_index = RXE_MIN_MR_INDEX, },
@@ -68,7 +68,7 @@ static const struct rxe_type_info rxe_type_info[RXE_NUM_TYPES] = { .size = sizeof(struct rxe_mw), .elem_offset = offsetof(struct rxe_mw, elem), .cleanup = rxe_mw_cleanup, - .flags = RXE_POOL_XARRAY | RXE_POOL_NO_ALLOC, + .flags = RXE_POOL_INDEX | RXE_POOL_NO_ALLOC, .max_index = RXE_MAX_MW_INDEX, .min_index = RXE_MIN_MW_INDEX, },
@@ -103,7 +103,7 @@ void rxe_pool_init(atomic_set (&pool->num_elem,0 ); -if (info->flags &RXE_POOL_XARRAY ) { +if (info->flags &RXE_POOL_INDEX ) {xa_init_flags (&pool->xarray.xa,XA_FLAGS_ALLOC ); pool->xarray.limit.max = info->max_index; pool->xarray.limit.min = info->min_index;
@@ -173,7 +173,7 @@ static void *__rxe_alloc_locked(struct rxe_pool *pool) elem->pool = pool; elem->obj = obj; -if (pool->flags &RXE_POOL_XARRAY ) { +if (pool->flags &RXE_POOL_INDEX ) { err =__xa_alloc_cyclic (&pool->xarray.xa, &elem->index, elem, pool->xarray.limit, &pool->xarray.next,GFP_KERNEL );
@@ -264,7 +264,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem) elem->pool = pool; elem->obj = (u8 *)elem - pool->elem_offset; -if (pool->flags &RXE_POOL_XARRAY ) { +if (pool->flags &RXE_POOL_INDEX ) { err =__xa_alloc_cyclic (&pool->xarray.xa, &elem->index, elem, pool->xarray.limit, &pool->xarray.next,GFP_KERNEL );
@@ -283,7 +283,7 @@ int __rxe_add_to_pool(struct rxe_pool *pool, struct rxe_pool_elem *elem)return -EINVAL ; } -static void *__rxe_get_xarray_locked (struct rxe_pool *pool,u32 index) +void *rxe_pool_get_index_locked (struct rxe_pool *pool,u32 index) {struct rxe_pool_elem *elem;void *obj =NULL ;
@@ -295,31 +295,17 @@ static void *__rxe_get_xarray_locked(struct rxe_pool *pool, u32 index)return obj; } -static void *__rxe_get_xarray (struct rxe_pool *pool,u32 index) +void *rxe_pool_get_index (struct rxe_pool *pool,u32 index) {void *obj;xa_lock_bh (&pool->xarray.xa); - obj =__rxe_get_xarray_locked (pool, index); + obj =rxe_pool_get_index_locked (pool, index);xa_unlock_bh (&pool->xarray.xa);return obj; } -void *rxe_pool_get_index_locked (struct rxe_pool *pool,u32 index) -{ -if (pool->flags &RXE_POOL_XARRAY ) -return __rxe_get_xarray_locked (pool, index); -return NULL ; -} - -void *rxe_pool_get_index (struct rxe_pool *pool,u32 index) -{ -if (pool->flags &RXE_POOL_XARRAY ) -return __rxe_get_xarray (pool, index); -return NULL ; -} -void *rxe_pool_get_key_locked (struct rxe_pool *pool,void *key) {struct rb_node *node;
@@ -413,7 +399,7 @@ static int __rxe_fini(struct rxe_pool_elem *elem) done =refcount_dec_if_one (&elem->refcnt);if (done) { -if (pool->flags &RXE_POOL_XARRAY ) +if (pool->flags &RXE_POOL_INDEX )__xa_erase (&pool->xarray.xa, elem->index);if (pool->flags &RXE_POOL_KEY )rb_erase (&elem->key_node, &pool->key.tree);
drivers/infiniband/sw/rxe/rxe_pool.h +2 -2
@@ -14,8 +14,8 @@#define RXE_POOL_CACHE_FLAGS (0)enum rxe_pool_flags { +RXE_POOL_INDEX =BIT (1 ),RXE_POOL_KEY =BIT (2 ), -RXE_POOL_XARRAY =BIT (3 ),RXE_POOL_NO_ALLOC =BIT (4 ), };
@@ -72,7 +72,7 @@ struct rxe_pool {size_t elem_size;size_t elem_offset; -/* only used if xarray */ +/* only used if indexed */ struct {struct xarray xa;struct xa_limit limit;
Signature · 1 line
2.30.2