Merge tag 'rdma-for-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband

Pull InfiniBand/RDMA changes from Roland Dreier:
 - Add ocrdma hardware driver for Emulex IB-over-Ethernet adapters
 - Add generic and mlx4 support for "raw" QPs: allow suitably privileged
   applications to send and receive arbitrary packets directly to/from
   the hardware
 - Add "doorbell drop" handling to the cxgb4 driver
 - A fairly large batch of qib hardware driver changes
 - A few fixes for lockdep-detected issues
 - A few other miscellaneous fixes and cleanups

Fix up trivial conflict in drivers/net/ethernet/emulex/benet/be.h.

* tag 'rdma-for-3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband: (53 commits)
  RDMA/cxgb4: Include vmalloc.h for vmalloc and vfree
  IB/mlx4: Fix mlx4_ib_add() error flow
  IB/core: Fix IB_SA_COMP_MASK macro
  IB/iser: Fix error flow in iser ep connection establishment
  IB/mlx4: Increase the number of vectors (EQs) available for ULPs
  RDMA/cxgb4: Add query_qp support
  RDMA/cxgb4: Remove kfifo usage
  RDMA/cxgb4: Use vmalloc() for debugfs QP dump
  RDMA/cxgb4: DB Drop Recovery for RDMA and LLD queues
  RDMA/cxgb4: Disable interrupts in c4iw_ev_dispatch()
  RDMA/cxgb4: Add DB Overflow Avoidance
  RDMA/cxgb4: Add debugfs RDMA memory stats
  cxgb4: DB Drop Recovery for RDMA and LLD queues
  cxgb4: Common platform specific changes for DB Drop Recovery
  cxgb4: Detect DB FULL events and notify RDMA ULD
  RDMA/cxgb4: Drop peer_abort when no endpoint found
  RDMA/cxgb4: Always wake up waiters in c4iw_peer_abort_intr()
  mlx4_core: Change bitmap allocator to work in round-robin fashion
  RDMA/nes: Don't call event handler if pointer is NULL
  RDMA/nes: Fix for the ORD value of the connecting peer
  ...

12 files changed, 821 insertions, 162 deletions

diff --git a/drivers/infiniband/hw/cxgb4/Makefile b/drivers/infiniband/hw/cxgb4/Makefile
index 46b878ca2c3b..e11cf7299945 100644
--- a/drivers/infiniband/hw/cxgb4/Makefile
+++ b/drivers/infiniband/hw/cxgb4/Makefile
@@ -2,4 +2,4 @@ ccflags-y := -Idrivers/net/ethernet/chelsio/cxgb4
 
 obj-$(CONFIG_INFINIBAND_CXGB4) += iw_cxgb4.o
 
-iw_cxgb4-y :=  device.o cm.o provider.o mem.o cq.o qp.o resource.o ev.o
+iw_cxgb4-y :=  device.o cm.o provider.o mem.o cq.o qp.o resource.o ev.o id_table.o
diff --git a/drivers/infiniband/hw/cxgb4/cm.c b/drivers/infiniband/hw/cxgb4/cm.c
index 92b4c2b0308b..55ab284e22f2 100644
--- a/drivers/infiniband/hw/cxgb4/cm.c
+++ b/drivers/infiniband/hw/cxgb4/cm.c
@@ -1362,7 +1362,10 @@ static int abort_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
 
 	ep = lookup_tid(t, tid);
 	PDBG("%s ep %p tid %u\n", __func__, ep, ep->hwtid);
-	BUG_ON(!ep);
+	if (!ep) {
+		printk(KERN_WARNING MOD "Abort rpl to freed endpoint\n");
+		return 0;
+	}
 	mutex_lock(&ep->com.mutex);
 	switch (ep->com.state) {
 	case ABORTING:
@@ -1410,6 +1413,24 @@ static int act_open_rpl(struct c4iw_dev *dev, struct sk_buff *skb)
 		return 0;
 	}
 
+	/*
+	 * Log interesting failures.
+	 */
+	switch (status) {
+	case CPL_ERR_CONN_RESET:
+	case CPL_ERR_CONN_TIMEDOUT:
+		break;
+	default:
+		printk(KERN_INFO MOD "Active open failure - "
+		       "atid %u status %u errno %d %pI4:%u->%pI4:%u\n",
+		       atid, status, status2errno(status),
+		       &ep->com.local_addr.sin_addr.s_addr,
+		       ntohs(ep->com.local_addr.sin_port),
+		       &ep->com.remote_addr.sin_addr.s_addr,
+		       ntohs(ep->com.remote_addr.sin_port));
+		break;
+	}
+
 	connect_reply_upcall(ep, status2errno(status));
 	state_set(&ep->com, DEAD);
 
@@ -1593,7 +1614,7 @@ static int import_ep(struct c4iw_ep *ep, __be32 peer_ip, struct dst_entry *dst,
 					n, n->dev, 0);
 		if (!ep->l2t)
 			goto out;
-		ep->mtu = dst_mtu(ep->dst);
+		ep->mtu = dst_mtu(dst);
 		ep->tx_chan = cxgb4_port_chan(n->dev);
 		ep->smac_idx = (cxgb4_port_viid(n->dev) & 0x7F) << 1;
 		step = cdev->rdev.lldi.ntxq /
@@ -2656,6 +2677,12 @@ static int peer_abort_intr(struct c4iw_dev *dev, struct sk_buff *skb)
 	unsigned int tid = GET_TID(req);
 
 	ep = lookup_tid(t, tid);
+	if (!ep) {
+		printk(KERN_WARNING MOD
+		       "Abort on non-existent endpoint, tid %d\n", tid);
+		kfree_skb(skb);
+		return 0;
+	}
 	if (is_neg_adv_abort(req->status)) {
 		PDBG("%s neg_adv_abort ep %p tid %u\n", __func__, ep,
 		     ep->hwtid);
@@ -2667,11 +2694,8 @@ static int peer_abort_intr(struct c4iw_dev *dev, struct sk_buff *skb)
 
 	/*
 	 * Wake up any threads in rdma_init() or rdma_fini().
-	 * However, this is not needed if com state is just
-	 * MPA_REQ_SENT
 	 */
-	if (ep->com.state != MPA_REQ_SENT)
-		c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
+	c4iw_wake_up(&ep->com.wr_wait, -ECONNRESET);
 	sched(dev, skb);
 	return 0;
 }
diff --git a/drivers/infiniband/hw/cxgb4/device.c b/drivers/infiniband/hw/cxgb4/device.c
index 6d0df6ec161b..cb4ecd783700 100644
--- a/drivers/infiniband/hw/cxgb4/device.c
+++ b/drivers/infiniband/hw/cxgb4/device.c
@@ -32,6 +32,7 @@
 #include <linux/module.h>
 #include <linux/moduleparam.h>
 #include <linux/debugfs.h>
+#include <linux/vmalloc.h>
 
 #include <rdma/ib_verbs.h>
 
@@ -44,6 +45,12 @@ MODULE_DESCRIPTION("Chelsio T4 RDMA Driver");
 MODULE_LICENSE("Dual BSD/GPL");
 MODULE_VERSION(DRV_VERSION);
 
+struct uld_ctx {
+	struct list_head entry;
+	struct cxgb4_lld_info lldi;
+	struct c4iw_dev *dev;
+};
+
 static LIST_HEAD(uld_ctx_list);
 static DEFINE_MUTEX(dev_mutex);
 
@@ -115,7 +122,7 @@ static int qp_release(struct inode *inode, struct file *file)
 		printk(KERN_INFO "%s null qpd?\n", __func__);
 		return 0;
 	}
-	kfree(qpd->buf);
+	vfree(qpd->buf);
 	kfree(qpd);
 	return 0;
 }
@@ -139,7 +146,7 @@ static int qp_open(struct inode *inode, struct file *file)
 	spin_unlock_irq(&qpd->devp->lock);
 
 	qpd->bufsize = count * 128;
-	qpd->buf = kmalloc(qpd->bufsize, GFP_KERNEL);
+	qpd->buf = vmalloc(qpd->bufsize);
 	if (!qpd->buf) {
 		ret = -ENOMEM;
 		goto err1;
@@ -240,6 +247,81 @@ static const struct file_operations stag_debugfs_fops = {
 	.llseek  = default_llseek,
 };
 
+static char *db_state_str[] = {"NORMAL", "FLOW_CONTROL", "RECOVERY"};
+
+static int stats_show(struct seq_file *seq, void *v)
+{
+	struct c4iw_dev *dev = seq->private;
+
+	seq_printf(seq, "   Object: %10s %10s %10s %10s\n", "Total", "Current",
+		   "Max", "Fail");
+	seq_printf(seq, "     PDID: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.pd.total, dev->rdev.stats.pd.cur,
+			dev->rdev.stats.pd.max, dev->rdev.stats.pd.fail);
+	seq_printf(seq, "      QID: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.qid.total, dev->rdev.stats.qid.cur,
+			dev->rdev.stats.qid.max, dev->rdev.stats.qid.fail);
+	seq_printf(seq, "   TPTMEM: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.stag.total, dev->rdev.stats.stag.cur,
+			dev->rdev.stats.stag.max, dev->rdev.stats.stag.fail);
+	seq_printf(seq, "   PBLMEM: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.pbl.total, dev->rdev.stats.pbl.cur,
+			dev->rdev.stats.pbl.max, dev->rdev.stats.pbl.fail);
+	seq_printf(seq, "   RQTMEM: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.rqt.total, dev->rdev.stats.rqt.cur,
+			dev->rdev.stats.rqt.max, dev->rdev.stats.rqt.fail);
+	seq_printf(seq, "  OCQPMEM: %10llu %10llu %10llu %10llu\n",
+			dev->rdev.stats.ocqp.total, dev->rdev.stats.ocqp.cur,
+			dev->rdev.stats.ocqp.max, dev->rdev.stats.ocqp.fail);
+	seq_printf(seq, "  DB FULL: %10llu\n", dev->rdev.stats.db_full);
+	seq_printf(seq, " DB EMPTY: %10llu\n", dev->rdev.stats.db_empty);
+	seq_printf(seq, "  DB DROP: %10llu\n", dev->rdev.stats.db_drop);
+	seq_printf(seq, " DB State: %s Transitions %llu\n",
+		   db_state_str[dev->db_state],
+		   dev->rdev.stats.db_state_transitions);
+	return 0;
+}
+
+static int stats_open(struct inode *inode, struct file *file)
+{
+	return single_open(file, stats_show, inode->i_private);
+}
+
+static ssize_t stats_clear(struct file *file, const char __user *buf,
+		size_t count, loff_t *pos)
+{
+	struct c4iw_dev *dev = ((struct seq_file *)file->private_data)->private;
+
+	mutex_lock(&dev->rdev.stats.lock);
+	dev->rdev.stats.pd.max = 0;
+	dev->rdev.stats.pd.fail = 0;
+	dev->rdev.stats.qid.max = 0;
+	dev->rdev.stats.qid.fail = 0;
+	dev->rdev.stats.stag.max = 0;
+	dev->rdev.stats.stag.fail = 0;
+	dev->rdev.stats.pbl.max = 0;
+	dev->rdev.stats.pbl.fail = 0;
+	dev->rdev.stats.rqt.max = 0;
+	dev->rdev.stats.rqt.fail = 0;
+	dev->rdev.stats.ocqp.max = 0;
+	dev->rdev.stats.ocqp.fail = 0;
+	dev->rdev.stats.db_full = 0;
+	dev->rdev.stats.db_empty = 0;
+	dev->rdev.stats.db_drop = 0;
+	dev->rdev.stats.db_state_transitions = 0;
+	mutex_unlock(&dev->rdev.stats.lock);
+	return count;
+}
+
+static const struct file_operations stats_debugfs_fops = {
+	.owner   = THIS_MODULE,
+	.open    = stats_open,
+	.release = single_release,
+	.read    = seq_read,
+	.llseek  = seq_lseek,
+	.write   = stats_clear,
+};
+
 static int setup_debugfs(struct c4iw_dev *devp)
 {
 	struct dentry *de;
@@ -256,6 +338,12 @@ static int setup_debugfs(struct c4iw_dev *devp)
 				 (void *)devp, &stag_debugfs_fops);
 	if (de && de->d_inode)
 		de->d_inode->i_size = 4096;
+
+	de = debugfs_create_file("stats", S_IWUSR, devp->debugfs_root,
+			(void *)devp, &stats_debugfs_fops);
+	if (de && de->d_inode)
+		de->d_inode->i_size = 4096;
+
 	return 0;
 }
 
@@ -269,9 +357,13 @@ void c4iw_release_dev_ucontext(struct c4iw_rdev *rdev,
 	list_for_each_safe(pos, nxt, &uctx->qpids) {
 		entry = list_entry(pos, struct c4iw_qid_list, entry);
 		list_del_init(&entry->entry);
-		if (!(entry->qid & rdev->qpmask))
-			c4iw_put_resource(&rdev->resource.qid_fifo, entry->qid,
-					  &rdev->resource.qid_fifo_lock);
+		if (!(entry->qid & rdev->qpmask)) {
+			c4iw_put_resource(&rdev->resource.qid_table,
+					  entry->qid);
+			mutex_lock(&rdev->stats.lock);
+			rdev->stats.qid.cur -= rdev->qpmask + 1;
+			mutex_unlock(&rdev->stats.lock);
+		}
 		kfree(entry);
 	}
 
@@ -332,6 +424,13 @@ static int c4iw_rdev_open(struct c4iw_rdev *rdev)
 		goto err1;
 	}
 
+	rdev->stats.pd.total = T4_MAX_NUM_PD;
+	rdev->stats.stag.total = rdev->lldi.vr->stag.size;
+	rdev->stats.pbl.total = rdev->lldi.vr->pbl.size;
+	rdev->stats.rqt.total = rdev->lldi.vr->rq.size;
+	rdev->stats.ocqp.total = rdev->lldi.vr->ocq.size;
+	rdev->stats.qid.total = rdev->lldi.vr->qp.size;
+
 	err = c4iw_init_resource(rdev, c4iw_num_stags(rdev), T4_MAX_NUM_PD);
 	if (err) {
 		printk(KERN_ERR MOD "error %d initializing resources\n", err);
@@ -370,12 +469,6 @@ static void c4iw_rdev_close(struct c4iw_rdev *rdev)
 	c4iw_destroy_resource(&rdev->resource);
 }
 
-struct uld_ctx {
-	struct list_head entry;
-	struct cxgb4_lld_info lldi;
-	struct c4iw_dev *dev;
-};
-
 static void c4iw_dealloc(struct uld_ctx *ctx)
 {
 	c4iw_rdev_close(&ctx->dev->rdev);
@@ -440,6 +533,8 @@ static struct c4iw_dev *c4iw_alloc(const struct cxgb4_lld_info *infop)
 	idr_init(&devp->qpidr);
 	idr_init(&devp->mmidr);
 	spin_lock_init(&devp->lock);
+	mutex_init(&devp->rdev.stats.lock);
+	mutex_init(&devp->db_mutex);
 
 	if (c4iw_debugfs_root) {
 		devp->debugfs_root = debugfs_create_dir(
@@ -585,11 +680,234 @@ static int c4iw_uld_state_change(void *handle, enum cxgb4_state new_state)
 	return 0;
 }
 
+static int disable_qp_db(int id, void *p, void *data)
+{
+	struct c4iw_qp *qp = p;
+
+	t4_disable_wq_db(&qp->wq);
+	return 0;
+}
+
+static void stop_queues(struct uld_ctx *ctx)
+{
+	spin_lock_irq(&ctx->dev->lock);
+	if (ctx->dev->db_state == NORMAL) {
+		ctx->dev->rdev.stats.db_state_transitions++;
+		ctx->dev->db_state = FLOW_CONTROL;
+		idr_for_each(&ctx->dev->qpidr, disable_qp_db, NULL);
+	}
+	spin_unlock_irq(&ctx->dev->lock);
+}
+
+static int enable_qp_db(int id, void *p, void *data)
+{
+	struct c4iw_qp *qp = p;
+
+	t4_enable_wq_db(&qp->wq);
+	return 0;
+}
+
+static void resume_queues(struct uld_ctx *ctx)
+{
+	spin_lock_irq(&ctx->dev->lock);
+	if (ctx->dev->qpcnt <= db_fc_threshold &&
+	    ctx->dev->db_state == FLOW_CONTROL) {
+		ctx->dev->db_state = NORMAL;
+		ctx->dev->rdev.stats.db_state_transitions++;
+		idr_for_each(&ctx->dev->qpidr, enable_qp_db, NULL);
+	}
+	spin_unlock_irq(&ctx->dev->lock);
+}
+
+struct qp_list {
+	unsigned idx;
+	struct c4iw_qp **qps;
+};
+
+static int add_and_ref_qp(int id, void *p, void *data)
+{
+	struct qp_list *qp_listp = data;
+	struct c4iw_qp *qp = p;
+
+	c4iw_qp_add_ref(&qp->ibqp);
+	qp_listp->qps[qp_listp->idx++] = qp;
+	return 0;
+}
+
+static int count_qps(int id, void *p, void *data)
+{
+	unsigned *countp = data;
+	(*countp)++;
+	return 0;
+}
+
+static void deref_qps(struct qp_list qp_list)
+{
+	int idx;
+
+	for (idx = 0; idx < qp_list.idx; idx++)
+		c4iw_qp_rem_ref(&qp_list.qps[idx]->ibqp);
+}
+
+static void recover_lost_dbs(struct uld_ctx *ctx, struct qp_list *qp_list)
+{
+	int idx;
+	int ret;
+
+	for (idx = 0; idx < qp_list->idx; idx++) {
+		struct c4iw_qp *qp = qp_list->qps[idx];
+
+		ret = cxgb4_sync_txq_pidx(qp->rhp->rdev.lldi.ports[0],
+					  qp->wq.sq.qid,
+					  t4_sq_host_wq_pidx(&qp->wq),
+					  t4_sq_wq_size(&qp->wq));
+		if (ret) {
+			printk(KERN_ERR MOD "%s: Fatal error - "
+			       "DB overflow recovery failed - "
+			       "error syncing SQ qid %u\n",
+			       pci_name(ctx->lldi.pdev), qp->wq.sq.qid);
+			return;
+		}
+
+		ret = cxgb4_sync_txq_pidx(qp->rhp->rdev.lldi.ports[0],
+					  qp->wq.rq.qid,
+					  t4_rq_host_wq_pidx(&qp->wq),
+					  t4_rq_wq_size(&qp->wq));
+
+		if (ret) {
+			printk(KERN_ERR MOD "%s: Fatal error - "
+			       "DB overflow recovery failed - "
+			       "error syncing RQ qid %u\n",
+			       pci_name(ctx->lldi.pdev), qp->wq.rq.qid);
+			return;
+		}
+
+		/* Wait for the dbfifo to drain */
+		while (cxgb4_dbfifo_count(qp->rhp->rdev.lldi.ports[0], 1) > 0) {
+			set_current_state(TASK_UNINTERRUPTIBLE);
+			schedule_timeout(usecs_to_jiffies(10));
+		}
+	}
+}
+
+static void recover_queues(struct uld_ctx *ctx)
+{
+	int count = 0;
+	struct qp_list qp_list;
+	int ret;
+
+	/* lock out kernel db ringers */
+	mutex_lock(&ctx->dev->db_mutex);
+
+	/* put all queues in to recovery mode */
+	spin_lock_irq(&ctx->dev->lock);
+	ctx->dev->db_state = RECOVERY;
+	ctx->dev->rdev.stats.db_state_transitions++;
+	idr_for_each(&ctx->dev->qpidr, disable_qp_db, NULL);
+	spin_unlock_irq(&ctx->dev->lock);
+
+	/* slow everybody down */
+	set_current_state(TASK_UNINTERRUPTIBLE);
+	schedule_timeout(usecs_to_jiffies(1000));
+
+	/* Wait for the dbfifo to completely drain. */
+	while (cxgb4_dbfifo_count(ctx->dev->rdev.lldi.ports[0], 1) > 0) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		schedule_timeout(usecs_to_jiffies(10));
+	}
+
+	/* flush the SGE contexts */
+	ret = cxgb4_flush_eq_cache(ctx->dev->rdev.lldi.ports[0]);
+	if (ret) {
+		printk(KERN_ERR MOD "%s: Fatal error - DB overflow recovery failed\n",
+		       pci_name(ctx->lldi.pdev));
+		goto out;
+	}
+
+	/* Count active queues so we can build a list of queues to recover */
+	spin_lock_irq(&ctx->dev->lock);
+	idr_for_each(&ctx->dev->qpidr, count_qps, &count);
+
+	qp_list.qps = kzalloc(count * sizeof *qp_list.qps, GFP_ATOMIC);
+	if (!qp_list.qps) {
+		printk(KERN_ERR MOD "%s: Fatal error - DB overflow recovery failed\n",
+		       pci_name(ctx->lldi.pdev));
+		spin_unlock_irq(&ctx->dev->lock);
+		goto out;
+	}
+	qp_list.idx = 0;
+
+	/* add and ref each qp so it doesn't get freed */
+	idr_for_each(&ctx->dev->qpidr, add_and_ref_qp, &qp_list);
+
+	spin_unlock_irq(&ctx->dev->lock);
+
+	/* now traverse the list in a safe context to recover the db state*/
+	recover_lost_dbs(ctx, &qp_list);
+
+	/* we're almost done!  deref the qps and clean up */
+	deref_qps(qp_list);
+	kfree(qp_list.qps);
+
+	/* Wait for the dbfifo to completely drain again */
+	while (cxgb4_dbfifo_count(ctx->dev->rdev.lldi.ports[0], 1) > 0) {
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		schedule_timeout(usecs_to_jiffies(10));
+	}
+
+	/* resume the queues */
+	spin_lock_irq(&ctx->dev->lock);
+	if (ctx->dev->qpcnt > db_fc_threshold)
+		ctx->dev->db_state = FLOW_CONTROL;
+	else {
+		ctx->dev->db_state = NORMAL;
+		idr_for_each(&ctx->dev->qpidr, enable_qp_db, NULL);
+	}
+	ctx->dev->rdev.stats.db_state_transitions++;
+	spin_unlock_irq(&ctx->dev->lock);
+
+out:
+	/* start up kernel db ringers again */
+	mutex_unlock(&ctx->dev->db_mutex);
+}
+
+static int c4iw_uld_control(void *handle, enum cxgb4_control control, ...)
+{
+	struct uld_ctx *ctx = handle;
+
+	switch (control) {
+	case CXGB4_CONTROL_DB_FULL:
+		stop_queues(ctx);
+		mutex_lock(&ctx->dev->rdev.stats.lock);
+		ctx->dev->rdev.stats.db_full++;
+		mutex_unlock(&ctx->dev->rdev.stats.lock);
+		break;
+	case CXGB4_CONTROL_DB_EMPTY:
+		resume_queues(ctx);
+		mutex_lock(&ctx->dev->rdev.stats.lock);
+		ctx->dev->rdev.stats.db_empty++;
+		mutex_unlock(&ctx->dev->rdev.stats.lock);
+		break;
+	case CXGB4_CONTROL_DB_DROP:
+		recover_queues(ctx);
+		mutex_lock(&ctx->dev->rdev.stats.lock);
+		ctx->dev->rdev.stats.db_drop++;
+		mutex_unlock(&ctx->dev->rdev.stats.lock);
+		break;
+	default:
+		printk(KERN_WARNING MOD "%s: unknown control cmd %u\n",
+		       pci_name(ctx->lldi.pdev), control);
+		break;
+	}
+	return 0;
+}
+
 static struct cxgb4_uld_info c4iw_uld_info = {
 	.name = DRV_NAME,
 	.add = c4iw_uld_add,
 	.rx_handler = c4iw_uld_rx_handler,
 	.state_change = c4iw_uld_state_change,
+	.control = c4iw_uld_control,
 };
 
 static int __init c4iw_init_module(void)
diff --git a/drivers/infiniband/hw/cxgb4/ev.c b/drivers/infiniband/hw/cxgb4/ev.c
index 397cb36cf103..cf2f6b47617a 100644
--- a/drivers/infiniband/hw/cxgb4/ev.c
+++ b/drivers/infiniband/hw/cxgb4/ev.c
@@ -84,7 +84,7 @@ void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe)
 	struct c4iw_qp *qhp;
 	u32 cqid;
 
-	spin_lock(&dev->lock);
+	spin_lock_irq(&dev->lock);
 	qhp = get_qhp(dev, CQE_QPID(err_cqe));
 	if (!qhp) {
 		printk(KERN_ERR MOD "BAD AE qpid 0x%x opcode %d "
@@ -93,7 +93,7 @@ void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe)
 		       CQE_OPCODE(err_cqe), CQE_STATUS(err_cqe),
 		       CQE_TYPE(err_cqe), CQE_WRID_HI(err_cqe),
 		       CQE_WRID_LOW(err_cqe));
-		spin_unlock(&dev->lock);
+		spin_unlock_irq(&dev->lock);
 		goto out;
 	}
 
@@ -109,13 +109,13 @@ void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe)
 		       CQE_OPCODE(err_cqe), CQE_STATUS(err_cqe),
 		       CQE_TYPE(err_cqe), CQE_WRID_HI(err_cqe),
 		       CQE_WRID_LOW(err_cqe));
-		spin_unlock(&dev->lock);
+		spin_unlock_irq(&dev->lock);
 		goto out;
 	}
 
 	c4iw_qp_add_ref(&qhp->ibqp);
 	atomic_inc(&chp->refcnt);
-	spin_unlock(&dev->lock);
+	spin_unlock_irq(&dev->lock);
 
 	/* Bad incoming write */
 	if (RQ_TYPE(err_cqe) &&
diff --git a/drivers/infiniband/hw/cxgb4/id_table.c b/drivers/infiniband/hw/cxgb4/id_table.c
new file mode 100644
index 000000000000..f95e5df30db2
--- /dev/null
+++ b/drivers/infiniband/hw/cxgb4/id_table.c
@@ -0,0 +1,112 @@
+/*
+ * Copyright (c) 2011 Chelsio Communications.  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 <linux/kernel.h>
+#include <linux/random.h>
+#include "iw_cxgb4.h"
+
+#define RANDOM_SKIP 16
+
+/*
+ * Trivial bitmap-based allocator. If the random flag is set, the
+ * allocator is designed to:
+ * - pseudo-randomize the id returned such that it is not trivially predictable.
+ * - avoid reuse of recently used id (at the expense of predictability)
+ */
+u32 c4iw_id_alloc(struct c4iw_id_table *alloc)
+{
+	unsigned long flags;
+	u32 obj;
+
+	spin_lock_irqsave(&alloc->lock, flags);
+
+	obj = find_next_zero_bit(alloc->table, alloc->max, alloc->last);
+	if (obj >= alloc->max)
+		obj = find_first_zero_bit(alloc->table, alloc->max);
+
+	if (obj < alloc->max) {
+		if (alloc->flags & C4IW_ID_TABLE_F_RANDOM)
+			alloc->last += random32() % RANDOM_SKIP;
+		else
+			alloc->last = obj + 1;
+		if (alloc->last >= alloc->max)
+			alloc->last = 0;
+		set_bit(obj, alloc->table);
+		obj += alloc->start;
+	} else
+		obj = -1;
+
+	spin_unlock_irqrestore(&alloc->lock, flags);
+	return obj;
+}
+
+void c4iw_id_free(struct c4iw_id_table *alloc, u32 obj)
+{
+	unsigned long flags;
+
+	obj -= alloc->start;
+	BUG_ON((int)obj < 0);
+
+	spin_lock_irqsave(&alloc->lock, flags);
+	clear_bit(obj, alloc->table);
+	spin_unlock_irqrestore(&alloc->lock, flags);
+}
+
+int c4iw_id_table_alloc(struct c4iw_id_table *alloc, u32 start, u32 num,
+			u32 reserved, u32 flags)
+{
+	int i;
+
+	alloc->start = start;
+	alloc->flags = flags;
+	if (flags & C4IW_ID_TABLE_F_RANDOM)
+		alloc->last = random32() % RANDOM_SKIP;
+	else
+		alloc->last = 0;
+	alloc->max  = num;
+	spin_lock_init(&alloc->lock);
+	alloc->table = kmalloc(BITS_TO_LONGS(num) * sizeof(long),
+				GFP_KERNEL);
+	if (!alloc->table)
+		return -ENOMEM;
+
+	bitmap_zero(alloc->table, num);
+	if (!(alloc->flags & C4IW_ID_TABLE_F_EMPTY))
+		for (i = 0; i < reserved; ++i)
+			set_bit(i, alloc->table);
+
+	return 0;
+}
+
+void c4iw_id_table_free(struct c4iw_id_table *alloc)
+{
+	kfree(alloc->table);
+}
diff --git a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
index 1357c5bf209b..9beb3a9f0336 100644
--- a/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
+++ b/drivers/infiniband/hw/cxgb4/iw_cxgb4.h
@@ -45,7 +45,6 @@
 #include <linux/kref.h>
 #include <linux/timer.h>
 #include <linux/io.h>
-#include <linux/kfifo.h>
 
 #include <asm/byteorder.h>
 
@@ -79,13 +78,22 @@ static inline void *cplhdr(struct sk_buff *skb)
 	return skb->data;
 }
 
+#define C4IW_ID_TABLE_F_RANDOM 1       /* Pseudo-randomize the id's returned */
+#define C4IW_ID_TABLE_F_EMPTY  2       /* Table is initially empty */
+
+struct c4iw_id_table {
+	u32 flags;
+	u32 start;              /* logical minimal id */
+	u32 last;               /* hint for find */
+	u32 max;
+	spinlock_t lock;
+	unsigned long *table;
+};
+
 struct c4iw_resource {
-	struct kfifo tpt_fifo;
-	spinlock_t tpt_fifo_lock;
-	struct kfifo qid_fifo;
-	spinlock_t qid_fifo_lock;
-	struct kfifo pdid_fifo;
-	spinlock_t pdid_fifo_lock;
+	struct c4iw_id_table tpt_table;
+	struct c4iw_id_table qid_table;
+	struct c4iw_id_table pdid_table;
 };
 
 struct c4iw_qid_list {
@@ -103,6 +111,27 @@ enum c4iw_rdev_flags {
 	T4_FATAL_ERROR = (1<<0),
 };
 
+struct c4iw_stat {
+	u64 total;
+	u64 cur;
+	u64 max;
+	u64 fail;
+};
+
+struct c4iw_stats {
+	struct mutex lock;
+	struct c4iw_stat qid;
+	struct c4iw_stat pd;
+	struct c4iw_stat stag;
+	struct c4iw_stat pbl;
+	struct c4iw_stat rqt;
+	struct c4iw_stat ocqp;
+	u64  db_full;
+	u64  db_empty;
+	u64  db_drop;
+	u64  db_state_transitions;
+};
+
 struct c4iw_rdev {
 	struct c4iw_resource resource;
 	unsigned long qpshift;
@@ -117,6 +146,7 @@ struct c4iw_rdev {
 	struct cxgb4_lld_info lldi;
 	unsigned long oc_mw_pa;
 	void __iomem *oc_mw_kva;
+	struct c4iw_stats stats;
 };
 
 static inline int c4iw_fatal_error(struct c4iw_rdev *rdev)
@@ -175,6 +205,12 @@ static inline int c4iw_wait_for_reply(struct c4iw_rdev *rdev,
 	return wr_waitp->ret;
 }
 
+enum db_state {
+	NORMAL = 0,
+	FLOW_CONTROL = 1,
+	RECOVERY = 2
+};
+
 struct c4iw_dev {
 	struct ib_device ibdev;
 	struct c4iw_rdev rdev;
@@ -183,7 +219,10 @@ struct c4iw_dev {
 	struct idr qpidr;
 	struct idr mmidr;
 	spinlock_t lock;
+	struct mutex db_mutex;
 	struct dentry *debugfs_root;
+	enum db_state db_state;
+	int qpcnt;
 };
 
 static inline struct c4iw_dev *to_c4iw_dev(struct ib_device *ibdev)
@@ -211,29 +250,57 @@ static inline struct c4iw_mr *get_mhp(struct c4iw_dev *rhp, u32 mmid)
 	return idr_find(&rhp->mmidr, mmid);
 }
 
-static inline int insert_handle(struct c4iw_dev *rhp, struct idr *idr,
-				void *handle, u32 id)
+static inline int _insert_handle(struct c4iw_dev *rhp, struct idr *idr,
+				 void *handle, u32 id, int lock)
 {
 	int ret;
 	int newid;
 
 	do {
-		if (!idr_pre_get(idr, GFP_KERNEL))
+		if (!idr_pre_get(idr, lock ? GFP_KERNEL : GFP_ATOMIC))
 			return -ENOMEM;
-		spin_lock_irq(&rhp->lock);
+		if (lock)
+			spin_lock_irq(&rhp->lock);
 		ret = idr_get_new_above(idr, handle, id, &newid);
-		BUG_ON(newid != id);
-		spin_unlock_irq(&rhp->lock);
+		BUG_ON(!ret && newid != id);
+		if (lock)
+			spin_unlock_irq(&rhp->lock);
 	} while (ret == -EAGAIN);
 
 	return ret;
 }
 
-static inline void remove_handle(struct c4iw_dev *rhp, struct idr *idr, u32 id)
+static inline int insert_handle(struct c4iw_dev *rhp, struct idr *idr,
+				void *handle, u32 id)
+{
+	return _insert_handle(rhp, idr, handle, id, 1);
+}
+
+static inline int insert_handle_nolock(struct c4iw_dev *rhp, struct idr *idr,
+				       void *handle, u32 id)
+{
+	return _insert_handle(rhp, idr, handle, id, 0);
+}
+
+static inline void _remove_handle(struct c4iw_dev *rhp, struct idr *idr,
+				   u32 id, int lock)
 {
-	spin_lock_irq(&rhp->lock);
+	if (lock)
+		spin_lock_irq(&rhp->lock);
 	idr_remove(idr, id);
-	spin_unlock_irq(&rhp->lock);
+	if (lock)
+		spin_unlock_irq(&rhp->lock);
+}
+
+static inline void remove_handle(struct c4iw_dev *rhp, struct idr *idr, u32 id)
+{
+	_remove_handle(rhp, idr, id, 1);
+}
+
+static inline void remove_handle_nolock(struct c4iw_dev *rhp,
+					 struct idr *idr, u32 id)
+{
+	_remove_handle(rhp, idr, id, 0);
 }
 
 struct c4iw_pd {
@@ -353,6 +420,8 @@ struct c4iw_qp_attributes {
 	struct c4iw_ep *llp_stream_handle;
 	u8 layer_etype;
 	u8 ecode;
+	u16 sq_db_inc;
+	u16 rq_db_inc;
 };
 
 struct c4iw_qp {
@@ -427,6 +496,8 @@ static inline void insert_mmap(struct c4iw_ucontext *ucontext,
 
 enum c4iw_qp_attr_mask {
 	C4IW_QP_ATTR_NEXT_STATE = 1 << 0,
+	C4IW_QP_ATTR_SQ_DB = 1<<1,
+	C4IW_QP_ATTR_RQ_DB = 1<<2,
 	C4IW_QP_ATTR_ENABLE_RDMA_READ = 1 << 7,
 	C4IW_QP_ATTR_ENABLE_RDMA_WRITE = 1 << 8,
 	C4IW_QP_ATTR_ENABLE_RDMA_BIND = 1 << 9,
@@ -480,6 +551,23 @@ static inline int c4iw_convert_state(enum ib_qp_state ib_state)
 	}
 }
 
+static inline int to_ib_qp_state(int c4iw_qp_state)
+{
+	switch (c4iw_qp_state) {
+	case C4IW_QP_STATE_IDLE:
+		return IB_QPS_INIT;
+	case C4IW_QP_STATE_RTS:
+		return IB_QPS_RTS;
+	case C4IW_QP_STATE_CLOSING:
+		return IB_QPS_SQD;
+	case C4IW_QP_STATE_TERMINATE:
+		return IB_QPS_SQE;
+	case C4IW_QP_STATE_ERROR:
+		return IB_QPS_ERR;
+	}
+	return IB_QPS_ERR;
+}
+
 static inline u32 c4iw_ib_to_tpt_access(int a)
 {
 	return (a & IB_ACCESS_REMOTE_WRITE ? FW_RI_MEM_ACCESS_REM_WRITE : 0) |
@@ -693,14 +781,20 @@ static inline int compute_wscale(int win)
 	return wscale;
 }
 
+u32 c4iw_id_alloc(struct c4iw_id_table *alloc);
+void c4iw_id_free(struct c4iw_id_table *alloc, u32 obj);
+int c4iw_id_table_alloc(struct c4iw_id_table *alloc, u32 start, u32 num,
+			u32 reserved, u32 flags);
+void c4iw_id_table_free(struct c4iw_id_table *alloc);
+
 typedef int (*c4iw_handler_func)(struct c4iw_dev *dev, struct sk_buff *skb);
 
 int c4iw_ep_redirect(void *ctx, struct dst_entry *old, struct dst_entry *new,
 		     struct l2t_entry *l2t);
 void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qpid,
 		   struct c4iw_dev_ucontext *uctx);
-u32 c4iw_get_resource(struct kfifo *fifo, spinlock_t *lock);
-void c4iw_put_resource(struct kfifo *fifo, u32 entry, spinlock_t *lock);
+u32 c4iw_get_resource(struct c4iw_id_table *id_table);
+void c4iw_put_resource(struct c4iw_id_table *id_table, u32 entry);
 int c4iw_init_resource(struct c4iw_rdev *rdev, u32 nr_tpt, u32 nr_pdid);
 int c4iw_init_ctrl_qp(struct c4iw_rdev *rdev);
 int c4iw_pblpool_create(struct c4iw_rdev *rdev);
@@ -769,6 +863,8 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd,
 			     struct ib_udata *udata);
 int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 				 int attr_mask, struct ib_udata *udata);
+int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		     int attr_mask, struct ib_qp_init_attr *init_attr);
 struct ib_qp *c4iw_get_qp(struct ib_device *dev, int qpn);
 u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size);
 void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size);
@@ -797,5 +893,7 @@ void c4iw_ev_dispatch(struct c4iw_dev *dev, struct t4_cqe *err_cqe);
 extern struct cxgb4_client t4c_client;
 extern c4iw_handler_func c4iw_handlers[NUM_CPL_CMDS];
 extern int c4iw_max_read_depth;
+extern int db_fc_threshold;
+
 
 #endif
diff --git a/drivers/infiniband/hw/cxgb4/mem.c b/drivers/infiniband/hw/cxgb4/mem.c
index 40c835309e49..57e07c61ace2 100644
--- a/drivers/infiniband/hw/cxgb4/mem.c
+++ b/drivers/infiniband/hw/cxgb4/mem.c
@@ -131,10 +131,14 @@ static int write_tpt_entry(struct c4iw_rdev *rdev, u32 reset_tpt_entry,
 	stag_idx = (*stag) >> 8;
 
 	if ((!reset_tpt_entry) && (*stag == T4_STAG_UNSET)) {
-		stag_idx = c4iw_get_resource(&rdev->resource.tpt_fifo,
-					     &rdev->resource.tpt_fifo_lock);
+		stag_idx = c4iw_get_resource(&rdev->resource.tpt_table);
 		if (!stag_idx)
 			return -ENOMEM;
+		mutex_lock(&rdev->stats.lock);
+		rdev->stats.stag.cur += 32;
+		if (rdev->stats.stag.cur > rdev->stats.stag.max)
+			rdev->stats.stag.max = rdev->stats.stag.cur;
+		mutex_unlock(&rdev->stats.lock);
 		*stag = (stag_idx << 8) | (atomic_inc_return(&key) & 0xff);
 	}
 	PDBG("%s stag_state 0x%0x type 0x%0x pdid 0x%0x, stag_idx 0x%x\n",
@@ -165,9 +169,12 @@ static int write_tpt_entry(struct c4iw_rdev *rdev, u32 reset_tpt_entry,
 				(rdev->lldi.vr->stag.start >> 5),
 				sizeof(tpt), &tpt);
 
-	if (reset_tpt_entry)
-		c4iw_put_resource(&rdev->resource.tpt_fifo, stag_idx,
-				  &rdev->resource.tpt_fifo_lock);
+	if (reset_tpt_entry) {
+		c4iw_put_resource(&rdev->resource.tpt_table, stag_idx);
+		mutex_lock(&rdev->stats.lock);
+		rdev->stats.stag.cur -= 32;
+		mutex_unlock(&rdev->stats.lock);
+	}
 	return err;
 }
 
@@ -686,8 +693,8 @@ int c4iw_dealloc_mw(struct ib_mw *mw)
 	mhp = to_c4iw_mw(mw);
 	rhp = mhp->rhp;
 	mmid = (mw->rkey) >> 8;
-	deallocate_window(&rhp->rdev, mhp->attr.stag);
 	remove_handle(rhp, &rhp->mmidr, mmid);
+	deallocate_window(&rhp->rdev, mhp->attr.stag);
 	kfree(mhp);
 	PDBG("%s ib_mw %p mmid 0x%x ptr %p\n", __func__, mw, mmid, mhp);
 	return 0;
@@ -789,12 +796,12 @@ int c4iw_dereg_mr(struct ib_mr *ib_mr)
 	mhp = to_c4iw_mr(ib_mr);
 	rhp = mhp->rhp;
 	mmid = mhp->attr.stag >> 8;
+	remove_handle(rhp, &rhp->mmidr, mmid);
 	dereg_mem(&rhp->rdev, mhp->attr.stag, mhp->attr.pbl_size,
 		       mhp->attr.pbl_addr);
 	if (mhp->attr.pbl_size)
 		c4iw_pblpool_free(&mhp->rhp->rdev, mhp->attr.pbl_addr,
 				  mhp->attr.pbl_size << 3);
-	remove_handle(rhp, &rhp->mmidr, mmid);
 	if (mhp->kva)
 		kfree((void *) (unsigned long) mhp->kva);
 	if (mhp->umem)
diff --git a/drivers/infiniband/hw/cxgb4/provider.c b/drivers/infiniband/hw/cxgb4/provider.c
index be1c18f44400..e084fdc6da7f 100644
--- a/drivers/infiniband/hw/cxgb4/provider.c
+++ b/drivers/infiniband/hw/cxgb4/provider.c
@@ -188,8 +188,10 @@ static int c4iw_deallocate_pd(struct ib_pd *pd)
 	php = to_c4iw_pd(pd);
 	rhp = php->rhp;
 	PDBG("%s ibpd %p pdid 0x%x\n", __func__, pd, php->pdid);
-	c4iw_put_resource(&rhp->rdev.resource.pdid_fifo, php->pdid,
-			  &rhp->rdev.resource.pdid_fifo_lock);
+	c4iw_put_resource(&rhp->rdev.resource.pdid_table, php->pdid);
+	mutex_lock(&rhp->rdev.stats.lock);
+	rhp->rdev.stats.pd.cur--;
+	mutex_unlock(&rhp->rdev.stats.lock);
 	kfree(php);
 	return 0;
 }
@@ -204,14 +206,12 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
 
 	PDBG("%s ibdev %p\n", __func__, ibdev);
 	rhp = (struct c4iw_dev *) ibdev;
-	pdid =  c4iw_get_resource(&rhp->rdev.resource.pdid_fifo,
-				  &rhp->rdev.resource.pdid_fifo_lock);
+	pdid =  c4iw_get_resource(&rhp->rdev.resource.pdid_table);
 	if (!pdid)
 		return ERR_PTR(-EINVAL);
 	php = kzalloc(sizeof(*php), GFP_KERNEL);
 	if (!php) {
-		c4iw_put_resource(&rhp->rdev.resource.pdid_fifo, pdid,
-				  &rhp->rdev.resource.pdid_fifo_lock);
+		c4iw_put_resource(&rhp->rdev.resource.pdid_table, pdid);
 		return ERR_PTR(-ENOMEM);
 	}
 	php->pdid = pdid;
@@ -222,6 +222,11 @@ static struct ib_pd *c4iw_allocate_pd(struct ib_device *ibdev,
 			return ERR_PTR(-EFAULT);
 		}
 	}
+	mutex_lock(&rhp->rdev.stats.lock);
+	rhp->rdev.stats.pd.cur++;
+	if (rhp->rdev.stats.pd.cur > rhp->rdev.stats.pd.max)
+		rhp->rdev.stats.pd.max = rhp->rdev.stats.pd.cur;
+	mutex_unlock(&rhp->rdev.stats.lock);
 	PDBG("%s pdid 0x%0x ptr 0x%p\n", __func__, pdid, php);
 	return &php->ibpd;
 }
@@ -438,6 +443,7 @@ int c4iw_register_device(struct c4iw_dev *dev)
 	    (1ull << IB_USER_VERBS_CMD_REQ_NOTIFY_CQ) |
 	    (1ull << IB_USER_VERBS_CMD_CREATE_QP) |
 	    (1ull << IB_USER_VERBS_CMD_MODIFY_QP) |
+	    (1ull << IB_USER_VERBS_CMD_QUERY_QP) |
 	    (1ull << IB_USER_VERBS_CMD_POLL_CQ) |
 	    (1ull << IB_USER_VERBS_CMD_DESTROY_QP) |
 	    (1ull << IB_USER_VERBS_CMD_POST_SEND) |
@@ -460,6 +466,7 @@ int c4iw_register_device(struct c4iw_dev *dev)
 	dev->ibdev.destroy_ah = c4iw_ah_destroy;
 	dev->ibdev.create_qp = c4iw_create_qp;
 	dev->ibdev.modify_qp = c4iw_ib_modify_qp;
+	dev->ibdev.query_qp = c4iw_ib_query_qp;
 	dev->ibdev.destroy_qp = c4iw_destroy_qp;
 	dev->ibdev.create_cq = c4iw_create_cq;
 	dev->ibdev.destroy_cq = c4iw_destroy_cq;
diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c
index 5f940aeaab1e..45aedf1d9338 100644
--- a/drivers/infiniband/hw/cxgb4/qp.c
+++ b/drivers/infiniband/hw/cxgb4/qp.c
@@ -34,10 +34,19 @@
 
 #include "iw_cxgb4.h"
 
+static int db_delay_usecs = 1;
+module_param(db_delay_usecs, int, 0644);
+MODULE_PARM_DESC(db_delay_usecs, "Usecs to delay awaiting db fifo to drain");
+
 static int ocqp_support = 1;
 module_param(ocqp_support, int, 0644);
 MODULE_PARM_DESC(ocqp_support, "Support on-chip SQs (default=1)");
 
+int db_fc_threshold = 2000;
+module_param(db_fc_threshold, int, 0644);
+MODULE_PARM_DESC(db_fc_threshold, "QP count/threshold that triggers automatic "
+		 "db flow control mode (default = 2000)");
+
 static void set_state(struct c4iw_qp *qhp, enum c4iw_qp_state state)
 {
 	unsigned long flag;
@@ -1128,6 +1137,35 @@ out:
 	return ret;
 }
 
+/*
+ * Called by the library when the qp has user dbs disabled due to
+ * a DB_FULL condition.  This function will single-thread all user
+ * DB rings to avoid overflowing the hw db-fifo.
+ */
+static int ring_kernel_db(struct c4iw_qp *qhp, u32 qid, u16 inc)
+{
+	int delay = db_delay_usecs;
+
+	mutex_lock(&qhp->rhp->db_mutex);
+	do {
+
+		/*
+		 * The interrupt threshold is dbfifo_int_thresh << 6. So
+		 * make sure we don't cross that and generate an interrupt.
+		 */
+		if (cxgb4_dbfifo_count(qhp->rhp->rdev.lldi.ports[0], 1) <
+		    (qhp->rhp->rdev.lldi.dbfifo_int_thresh << 5)) {
+			writel(V_QID(qid) | V_PIDX(inc), qhp->wq.db);
+			break;
+		}
+		set_current_state(TASK_UNINTERRUPTIBLE);
+		schedule_timeout(usecs_to_jiffies(delay));
+		delay = min(delay << 1, 2000);
+	} while (1);
+	mutex_unlock(&qhp->rhp->db_mutex);
+	return 0;
+}
+
 int c4iw_modify_qp(struct c4iw_dev *rhp, struct c4iw_qp *qhp,
 		   enum c4iw_qp_attr_mask mask,
 		   struct c4iw_qp_attributes *attrs,
@@ -1176,6 +1214,15 @@ int c4iw_modify_qp(struct c4iw_dev *rhp, struct c4iw_qp *qhp,
 		qhp->attr = newattr;
 	}
 
+	if (mask & C4IW_QP_ATTR_SQ_DB) {
+		ret = ring_kernel_db(qhp, qhp->wq.sq.qid, attrs->sq_db_inc);
+		goto out;
+	}
+	if (mask & C4IW_QP_ATTR_RQ_DB) {
+		ret = ring_kernel_db(qhp, qhp->wq.rq.qid, attrs->rq_db_inc);
+		goto out;
+	}
+
 	if (!(mask & C4IW_QP_ATTR_NEXT_STATE))
 		goto out;
 	if (qhp->attr.state == attrs->next_state)
@@ -1352,6 +1399,14 @@ out:
 	return ret;
 }
 
+static int enable_qp_db(int id, void *p, void *data)
+{
+	struct c4iw_qp *qp = p;
+
+	t4_enable_wq_db(&qp->wq);
+	return 0;
+}
+
 int c4iw_destroy_qp(struct ib_qp *ib_qp)
 {
 	struct c4iw_dev *rhp;
@@ -1369,7 +1424,16 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp)
 		c4iw_modify_qp(rhp, qhp, C4IW_QP_ATTR_NEXT_STATE, &attrs, 0);
 	wait_event(qhp->wait, !qhp->ep);
 
-	remove_handle(rhp, &rhp->qpidr, qhp->wq.sq.qid);
+	spin_lock_irq(&rhp->lock);
+	remove_handle_nolock(rhp, &rhp->qpidr, qhp->wq.sq.qid);
+	rhp->qpcnt--;
+	BUG_ON(rhp->qpcnt < 0);
+	if (rhp->qpcnt <= db_fc_threshold && rhp->db_state == FLOW_CONTROL) {
+		rhp->rdev.stats.db_state_transitions++;
+		rhp->db_state = NORMAL;
+		idr_for_each(&rhp->qpidr, enable_qp_db, NULL);
+	}
+	spin_unlock_irq(&rhp->lock);
 	atomic_dec(&qhp->refcnt);
 	wait_event(qhp->wait, !atomic_read(&qhp->refcnt));
 
@@ -1383,6 +1447,14 @@ int c4iw_destroy_qp(struct ib_qp *ib_qp)
 	return 0;
 }
 
+static int disable_qp_db(int id, void *p, void *data)
+{
+	struct c4iw_qp *qp = p;
+
+	t4_disable_wq_db(&qp->wq);
+	return 0;
+}
+
 struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 			     struct ib_udata *udata)
 {
@@ -1469,7 +1541,16 @@ struct ib_qp *c4iw_create_qp(struct ib_pd *pd, struct ib_qp_init_attr *attrs,
 	init_waitqueue_head(&qhp->wait);
 	atomic_set(&qhp->refcnt, 1);
 
-	ret = insert_handle(rhp, &rhp->qpidr, qhp, qhp->wq.sq.qid);
+	spin_lock_irq(&rhp->lock);
+	if (rhp->db_state != NORMAL)
+		t4_disable_wq_db(&qhp->wq);
+	if (++rhp->qpcnt > db_fc_threshold && rhp->db_state == NORMAL) {
+		rhp->rdev.stats.db_state_transitions++;
+		rhp->db_state = FLOW_CONTROL;
+		idr_for_each(&rhp->qpidr, disable_qp_db, NULL);
+	}
+	ret = insert_handle_nolock(rhp, &rhp->qpidr, qhp, qhp->wq.sq.qid);
+	spin_unlock_irq(&rhp->lock);
 	if (ret)
 		goto err2;
 
@@ -1613,6 +1694,15 @@ int c4iw_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
 			 C4IW_QP_ATTR_ENABLE_RDMA_WRITE |
 			 C4IW_QP_ATTR_ENABLE_RDMA_BIND) : 0;
 
+	/*
+	 * Use SQ_PSN and RQ_PSN to pass in IDX_INC values for
+	 * ringing the queue db when we're in DB_FULL mode.
+	 */
+	attrs.sq_db_inc = attr->sq_psn;
+	attrs.rq_db_inc = attr->rq_psn;
+	mask |= (attr_mask & IB_QP_SQ_PSN) ? C4IW_QP_ATTR_SQ_DB : 0;
+	mask |= (attr_mask & IB_QP_RQ_PSN) ? C4IW_QP_ATTR_RQ_DB : 0;
+
 	return c4iw_modify_qp(rhp, qhp, mask, &attrs, 0);
 }
 
@@ -1621,3 +1711,14 @@ struct ib_qp *c4iw_get_qp(struct ib_device *dev, int qpn)
 	PDBG("%s ib_dev %p qpn 0x%x\n", __func__, dev, qpn);
 	return (struct ib_qp *)get_qhp(to_c4iw_dev(dev), qpn);
 }
+
+int c4iw_ib_query_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
+		     int attr_mask, struct ib_qp_init_attr *init_attr)
+{
+	struct c4iw_qp *qhp = to_c4iw_qp(ibqp);
+
+	memset(attr, 0, sizeof *attr);
+	memset(init_attr, 0, sizeof *init_attr);
+	attr->qp_state = to_ib_qp_state(qhp->attr.state);
+	return 0;
+}
diff --git a/drivers/infiniband/hw/cxgb4/resource.c b/drivers/infiniband/hw/cxgb4/resource.c
index 407ff3924150..cdef4d7fb6d8 100644
--- a/drivers/infiniband/hw/cxgb4/resource.c
+++ b/drivers/infiniband/hw/cxgb4/resource.c
@@ -30,96 +30,25 @@
  * SOFTWARE.
  */
 /* Crude resource management */
-#include <linux/kernel.h>
-#include <linux/random.h>
-#include <linux/slab.h>
-#include <linux/kfifo.h>
 #include <linux/spinlock.h>
-#include <linux/errno.h>
 #include <linux/genalloc.h>
 #include <linux/ratelimit.h>
 #include "iw_cxgb4.h"
 
-#define RANDOM_SIZE 16
-
-static int __c4iw_init_resource_fifo(struct kfifo *fifo,
-				   spinlock_t *fifo_lock,
-				   u32 nr, u32 skip_low,
-				   u32 skip_high,
-				   int random)
-{
-	u32 i, j, entry = 0, idx;
-	u32 random_bytes;
-	u32 rarray[16];
-	spin_lock_init(fifo_lock);
-
-	if (kfifo_alloc(fifo, nr * sizeof(u32), GFP_KERNEL))
-		return -ENOMEM;
-
-	for (i = 0; i < skip_low + skip_high; i++)
-		kfifo_in(fifo, (unsigned char *) &entry, sizeof(u32));
-	if (random) {
-		j = 0;
-		random_bytes = random32();
-		for (i = 0; i < RANDOM_SIZE; i++)
-			rarray[i] = i + skip_low;
-		for (i = skip_low + RANDOM_SIZE; i < nr - skip_high; i++) {
-			if (j >= RANDOM_SIZE) {
-				j = 0;
-				random_bytes = random32();
-			}
-			idx = (random_bytes >> (j * 2)) & 0xF;
-			kfifo_in(fifo,
-				(unsigned char *) &rarray[idx],
-				sizeof(u32));
-			rarray[idx] = i;
-			j++;
-		}
-		for (i = 0; i < RANDOM_SIZE; i++)
-			kfifo_in(fifo,
-				(unsigned char *) &rarray[i],
-				sizeof(u32));
-	} else
-		for (i = skip_low; i < nr - skip_high; i++)
-			kfifo_in(fifo, (unsigned char *) &i, sizeof(u32));
-
-	for (i = 0; i < skip_low + skip_high; i++)
-		if (kfifo_out_locked(fifo, (unsigned char *) &entry,
-				     sizeof(u32), fifo_lock))
-			break;
-	return 0;
-}
-
-static int c4iw_init_resource_fifo(struct kfifo *fifo, spinlock_t * fifo_lock,
-				   u32 nr, u32 skip_low, u32 skip_high)
-{
-	return __c4iw_init_resource_fifo(fifo, fifo_lock, nr, skip_low,
-					  skip_high, 0);
-}
-
-static int c4iw_init_resource_fifo_random(struct kfifo *fifo,
-				   spinlock_t *fifo_lock,
-				   u32 nr, u32 skip_low, u32 skip_high)
-{
-	return __c4iw_init_resource_fifo(fifo, fifo_lock, nr, skip_low,
-					  skip_high, 1);
-}
-
-static int c4iw_init_qid_fifo(struct c4iw_rdev *rdev)
+static int c4iw_init_qid_table(struct c4iw_rdev *rdev)
 {
 	u32 i;
 
-	spin_lock_init(&rdev->resource.qid_fifo_lock);
-
-	if (kfifo_alloc(&rdev->resource.qid_fifo, rdev->lldi.vr->qp.size *
-			sizeof(u32), GFP_KERNEL))
+	if (c4iw_id_table_alloc(&rdev->resource.qid_table,
+				rdev->lldi.vr->qp.start,
+				rdev->lldi.vr->qp.size,
+				rdev->lldi.vr->qp.size, 0))
 		return -ENOMEM;
 
 	for (i = rdev->lldi.vr->qp.start;
-	     i < rdev->lldi.vr->qp.start + rdev->lldi.vr->qp.size; i++)
+		i < rdev->lldi.vr->qp.start + rdev->lldi.vr->qp.size; i++)
 		if (!(i & rdev->qpmask))
-			kfifo_in(&rdev->resource.qid_fifo,
-				    (unsigned char *) &i, sizeof(u32));
+			c4iw_id_free(&rdev->resource.qid_table, i);
 	return 0;
 }
 
@@ -127,44 +56,42 @@ static int c4iw_init_qid_fifo(struct c4iw_rdev *rdev)
 int c4iw_init_resource(struct c4iw_rdev *rdev, u32 nr_tpt, u32 nr_pdid)
 {
 	int err = 0;
-	err = c4iw_init_resource_fifo_random(&rdev->resource.tpt_fifo,
-					     &rdev->resource.tpt_fifo_lock,
-					     nr_tpt, 1, 0);
+	err = c4iw_id_table_alloc(&rdev->resource.tpt_table, 0, nr_tpt, 1,
+					C4IW_ID_TABLE_F_RANDOM);
 	if (err)
 		goto tpt_err;
-	err = c4iw_init_qid_fifo(rdev);
+	err = c4iw_init_qid_table(rdev);
 	if (err)
 		goto qid_err;
-	err = c4iw_init_resource_fifo(&rdev->resource.pdid_fifo,
-				      &rdev->resource.pdid_fifo_lock,
-				      nr_pdid, 1, 0);
+	err = c4iw_id_table_alloc(&rdev->resource.pdid_table, 0,
+					nr_pdid, 1, 0);
 	if (err)
 		goto pdid_err;
 	return 0;
-pdid_err:
-	kfifo_free(&rdev->resource.qid_fifo);
-qid_err:
-	kfifo_free(&rdev->resource.tpt_fifo);
-tpt_err:
+ pdid_err:
+	c4iw_id_table_free(&rdev->resource.qid_table);
+ qid_err:
+	c4iw_id_table_free(&rdev->resource.tpt_table);
+ tpt_err:
 	return -ENOMEM;
 }
 
 /*
  * returns 0 if no resource available
  */
-u32 c4iw_get_resource(struct kfifo *fifo, spinlock_t *lock)
+u32 c4iw_get_resource(struct c4iw_id_table *id_table)
 {
 	u32 entry;
-	if (kfifo_out_locked(fifo, (unsigned char *) &entry, sizeof(u32), lock))
-		return entry;
-	else
+	entry = c4iw_id_alloc(id_table);
+	if (entry == (u32)(-1))
 		return 0;
+	return entry;
 }
 
-void c4iw_put_resource(struct kfifo *fifo, u32 entry, spinlock_t *lock)
+void c4iw_put_resource(struct c4iw_id_table *id_table, u32 entry)
 {
 	PDBG("%s entry 0x%x\n", __func__, entry);
-	kfifo_in_locked(fifo, (unsigned char *) &entry, sizeof(u32), lock);
+	c4iw_id_free(id_table, entry);
 }
 
 u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
@@ -181,10 +108,12 @@ u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		qid = entry->qid;
 		kfree(entry);
 	} else {
-		qid = c4iw_get_resource(&rdev->resource.qid_fifo,
-					&rdev->resource.qid_fifo_lock);
+		qid = c4iw_get_resource(&rdev->resource.qid_table);
 		if (!qid)
 			goto out;
+		mutex_lock(&rdev->stats.lock);
+		rdev->stats.qid.cur += rdev->qpmask + 1;
+		mutex_unlock(&rdev->stats.lock);
 		for (i = qid+1; i & rdev->qpmask; i++) {
 			entry = kmalloc(sizeof *entry, GFP_KERNEL);
 			if (!entry)
@@ -213,6 +142,10 @@ u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 out:
 	mutex_unlock(&uctx->lock);
 	PDBG("%s qid 0x%x\n", __func__, qid);
+	mutex_lock(&rdev->stats.lock);
+	if (rdev->stats.qid.cur > rdev->stats.qid.max)
+		rdev->stats.qid.max = rdev->stats.qid.cur;
+	mutex_unlock(&rdev->stats.lock);
 	return qid;
 }
 
@@ -245,10 +178,12 @@ u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 		qid = entry->qid;
 		kfree(entry);
 	} else {
-		qid = c4iw_get_resource(&rdev->resource.qid_fifo,
-					&rdev->resource.qid_fifo_lock);
+		qid = c4iw_get_resource(&rdev->resource.qid_table);
 		if (!qid)
 			goto out;
+		mutex_lock(&rdev->stats.lock);
+		rdev->stats.qid.cur += rdev->qpmask + 1;
+		mutex_unlock(&rdev->stats.lock);
 		for (i = qid+1; i & rdev->qpmask; i++) {
 			entry = kmalloc(sizeof *entry, GFP_KERNEL);
 			if (!entry)
@@ -277,6 +212,10 @@ u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
 out:
 	mutex_unlock(&uctx->lock);
 	PDBG("%s qid 0x%x\n", __func__, qid);
+	mutex_lock(&rdev->stats.lock);
+	if (rdev->stats.qid.cur > rdev->stats.qid.max)
+		rdev->stats.qid.max = rdev->stats.qid.cur;
+	mutex_unlock(&rdev->stats.lock);
 	return qid;
 }
 
@@ -297,9 +236,9 @@ void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qid,
 
 void c4iw_destroy_resource(struct c4iw_resource *rscp)
 {
-	kfifo_free(&rscp->tpt_fifo);
-	kfifo_free(&rscp->qid_fifo);
-	kfifo_free(&rscp->pdid_fifo);
+	c4iw_id_table_free(&rscp->tpt_table);
+	c4iw_id_table_free(&rscp->qid_table);
+	c4iw_id_table_free(&rscp->pdid_table);
 }
 
 /*
@@ -312,15 +251,23 @@ u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size)
 {
 	unsigned long addr = gen_pool_alloc(rdev->pbl_pool, size);
 	PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size);
-	if (!addr)
-		printk_ratelimited(KERN_WARNING MOD "%s: Out of PBL memory\n",
-		       pci_name(rdev->lldi.pdev));
+	mutex_lock(&rdev->stats.lock);
+	if (addr) {
+		rdev->stats.pbl.cur += roundup(size, 1 << MIN_PBL_SHIFT);
+		if (rdev->stats.pbl.cur > rdev->stats.pbl.max)
+			rdev->stats.pbl.max = rdev->stats.pbl.cur;
+	} else
+		rdev->stats.pbl.fail++;
+	mutex_unlock(&rdev->stats.lock);
 	return (u32)addr;
 }
 
 void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 {
 	PDBG("%s addr 0x%x size %d\n", __func__, addr, size);
+	mutex_lock(&rdev->stats.lock);
+	rdev->stats.pbl.cur -= roundup(size, 1 << MIN_PBL_SHIFT);
+	mutex_unlock(&rdev->stats.lock);
 	gen_pool_free(rdev->pbl_pool, (unsigned long)addr, size);
 }
 
@@ -377,12 +324,23 @@ u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size)
 	if (!addr)
 		printk_ratelimited(KERN_WARNING MOD "%s: Out of RQT memory\n",
 		       pci_name(rdev->lldi.pdev));
+	mutex_lock(&rdev->stats.lock);
+	if (addr) {
+		rdev->stats.rqt.cur += roundup(size << 6, 1 << MIN_RQT_SHIFT);
+		if (rdev->stats.rqt.cur > rdev->stats.rqt.max)
+			rdev->stats.rqt.max = rdev->stats.rqt.cur;
+	} else
+		rdev->stats.rqt.fail++;
+	mutex_unlock(&rdev->stats.lock);
 	return (u32)addr;
 }
 
 void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 {
 	PDBG("%s addr 0x%x size %d\n", __func__, addr, size << 6);
+	mutex_lock(&rdev->stats.lock);
+	rdev->stats.rqt.cur -= roundup(size << 6, 1 << MIN_RQT_SHIFT);
+	mutex_unlock(&rdev->stats.lock);
 	gen_pool_free(rdev->rqt_pool, (unsigned long)addr, size << 6);
 }
 
@@ -433,12 +391,22 @@ u32 c4iw_ocqp_pool_alloc(struct c4iw_rdev *rdev, int size)
 {
 	unsigned long addr = gen_pool_alloc(rdev->ocqp_pool, size);
 	PDBG("%s addr 0x%x size %d\n", __func__, (u32)addr, size);
+	if (addr) {
+		mutex_lock(&rdev->stats.lock);
+		rdev->stats.ocqp.cur += roundup(size, 1 << MIN_OCQP_SHIFT);
+		if (rdev->stats.ocqp.cur > rdev->stats.ocqp.max)
+			rdev->stats.ocqp.max = rdev->stats.ocqp.cur;
+		mutex_unlock(&rdev->stats.lock);
+	}
 	return (u32)addr;
 }
 
 void c4iw_ocqp_pool_free(struct c4iw_rdev *rdev, u32 addr, int size)
 {
 	PDBG("%s addr 0x%x size %d\n", __func__, addr, size);
+	mutex_lock(&rdev->stats.lock);
+	rdev->stats.ocqp.cur -= roundup(size, 1 << MIN_OCQP_SHIFT);
+	mutex_unlock(&rdev->stats.lock);
 	gen_pool_free(rdev->ocqp_pool, (unsigned long)addr, size);
 }
 
diff --git a/drivers/infiniband/hw/cxgb4/t4.h b/drivers/infiniband/hw/cxgb4/t4.h
index c0221eec8817..16f26ab29302 100644
--- a/drivers/infiniband/hw/cxgb4/t4.h
+++ b/drivers/infiniband/hw/cxgb4/t4.h
@@ -62,6 +62,10 @@ struct t4_status_page {
 	__be16 pidx;
 	u8 qp_err;	/* flit 1 - sw owns */
 	u8 db_off;
+	u8 pad;
+	u16 host_wq_pidx;
+	u16 host_cidx;
+	u16 host_pidx;
 };
 
 #define T4_EQ_ENTRY_SIZE 64
@@ -375,6 +379,16 @@ static inline void t4_rq_consume(struct t4_wq *wq)
 		wq->rq.cidx = 0;
 }
 
+static inline u16 t4_rq_host_wq_pidx(struct t4_wq *wq)
+{
+	return wq->rq.queue[wq->rq.size].status.host_wq_pidx;
+}
+
+static inline u16 t4_rq_wq_size(struct t4_wq *wq)
+{
+		return wq->rq.size * T4_RQ_NUM_SLOTS;
+}
+
 static inline int t4_sq_onchip(struct t4_sq *sq)
 {
 	return sq->flags & T4_SQ_ONCHIP;
@@ -412,6 +426,16 @@ static inline void t4_sq_consume(struct t4_wq *wq)
 		wq->sq.cidx = 0;
 }
 
+static inline u16 t4_sq_host_wq_pidx(struct t4_wq *wq)
+{
+	return wq->sq.queue[wq->sq.size].status.host_wq_pidx;
+}
+
+static inline u16 t4_sq_wq_size(struct t4_wq *wq)
+{
+		return wq->sq.size * T4_SQ_NUM_SLOTS;
+}
+
 static inline void t4_ring_sq_db(struct t4_wq *wq, u16 inc)
 {
 	wmb();
diff --git a/drivers/infiniband/hw/cxgb4/user.h b/drivers/infiniband/hw/cxgb4/user.h
index e6669d54770e..32b754c35ab7 100644
--- a/drivers/infiniband/hw/cxgb4/user.h
+++ b/drivers/infiniband/hw/cxgb4/user.h
@@ -32,7 +32,7 @@
 #ifndef __C4IW_USER_H__
 #define __C4IW_USER_H__
 
-#define C4IW_UVERBS_ABI_VERSION	1
+#define C4IW_UVERBS_ABI_VERSION	2
 
 /*
  * Make sure that all structs defined in this file remain laid out so