[PATCH 0/8] Misc fixes for 3.17 rdma stack

[PATCH 0/8] Misc fixes for 3.17 rdma stack

[Doug Ledford]

This is a collection of miscellaneous fixes that people have posted
previously for inclusion in the 3.17 kernel.  The one from Bart was
(apparently) skipped when you picked up the rest of the set.  I'm
requesting that you reconsider skipping this patch as the pkey cache
is a perfectly valid optimization to use and saves considerable time
on pkey lookups (my measurements were up to 5ms per lookup that goes
to the card instead of hitting the cache, so that's a 5ms of totally
unneeded latency).

I have three in here, one that I've posted before and two that are new.

The remainder are very much trivial patches, but they weren't in your
tree so I included them here.  These have all been built into my test
kernel for my next patch set, so they've at least been compile tested
by me, and they've also been reviewed and I've added my sign off.

Bart Van Assche (1):
  IB/srp: Use P_Key cache for P_Key lookups

Dan Carpenter (1):
  RDMA/amso1100: integer overflow in c2_alloc_cq_buf()

Doug Ledford (3):
  ib/srpt: Enhance printk output
  ib/srpt: Handle GID change events
  uapi/rdma_user_cm.h: include socket.h

Fabian Frederick (3):
  drivers/infiniband/ulp/ipoib/ipoib_fs.c: remove unnecessary null test
    before debugfs_remove
  IB/mlx4: use ARRAY_SIZE instead of sizeof/sizeof[0]
  IB/mlx5: use ARRAY_SIZE instead of sizeof/sizeof[0]

 drivers/infiniband/hw/amso1100/c2_cq.c  |   7 +-
 drivers/infiniband/hw/mlx4/main.c       |   6 +-
 drivers/infiniband/hw/mlx5/qp.c         |   2 +-
 drivers/infiniband/ulp/ipoib/ipoib_fs.c |   6 +-
 drivers/infiniband/ulp/srp/ib_srp.c     |   9 +-
 drivers/infiniband/ulp/srpt/ib_srpt.c   | 202 ++++++++++++++++----------------
 include/uapi/rdma/rdma_user_cm.h        |   1 +
 7 files changed, 118 insertions(+), 115 deletions(-)

-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1/8] RDMA/amso1100: integer overflow in c2_alloc_cq_buf()

[Doug Ledford]

From: Dan Carpenter <dan.carpenter-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>

This is a static checker fix.  The static checker says that q_size comes
from the user and can be any 32 bit value.  The call tree is:
  --> ib_uverbs_create_cq()
      --> c2_create_cq()
          --> c2_init_cq()

Signed-off-by: Dan Carpenter <dan.carpenter-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>
Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/amso1100/c2_cq.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/amso1100/c2_cq.c b/drivers/infiniband/hw/amso1100/c2_cq.c
index 49e0e8533f7..1b63185b4ad 100644
--- a/drivers/infiniband/hw/amso1100/c2_cq.c
+++ b/drivers/infiniband/hw/amso1100/c2_cq.c
@@ -260,11 +260,14 @@ static void c2_free_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq)
 			  mq->msg_pool.host, dma_unmap_addr(mq, mapping));
 }
 
-static int c2_alloc_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq, int q_size,
-			   int msg_size)
+static int c2_alloc_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq,
+			   size_t q_size, size_t msg_size)
 {
 	u8 *pool_start;
 
+	if (q_size > SIZE_MAX / msg_size)
+		return -EINVAL;
+
 	pool_start = dma_alloc_coherent(&c2dev->pcidev->dev, q_size * msg_size,
 					&mq->host_dma, GFP_KERNEL);
 	if (!pool_start)
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/8] IB/srp: Use P_Key cache for P_Key lookups

[Doug Ledford]

From: Bart Van Assche <bvanassche-HInyCGIudOg@public.gmane.org>

This change slightly reduces the time needed to log in.

Signed-off-by: Bart Van Assche <bvanassche-HInyCGIudOg@public.gmane.org>
Reviewed-by: Sagi Grimberg <sagig-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Reviewed-by: David Dillow <dave-i1Mk8JYDVaaSihdK6806/g@public.gmane.org>
Cc: Sebastian Parschauer <sebastian.riemer-EIkl63zCoXaH+58JC4qpiA@public.gmane.org>
Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/ulp/srp/ib_srp.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c
index 7f5ee7fc02a..bcec45a2c95 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -40,6 +40,7 @@
 #include <linux/parser.h>
 #include <linux/random.h>
 #include <linux/jiffies.h>
+#include <rdma/ib_cache.h>
 
 #include <linux/atomic.h>
 
@@ -260,10 +261,10 @@ static int srp_init_qp(struct srp_target_port *target,
 	if (!attr)
 		return -ENOMEM;
 
-	ret = ib_find_pkey(target->srp_host->srp_dev->dev,
-			   target->srp_host->port,
-			   be16_to_cpu(target->path.pkey),
-			   &attr->pkey_index);
+	ret = ib_find_cached_pkey(target->srp_host->srp_dev->dev,
+				  target->srp_host->port,
+				  be16_to_cpu(target->path.pkey),
+				  &attr->pkey_index);
 	if (ret)
 		goto out;
 
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 3/8] drivers/infiniband/ulp/ipoib/ipoib_fs.c: remove unnecessary null test before debugfs_remove

[Doug Ledford]

From: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>

Fix checkpatch warning:
WARNING: debugfs_remove(NULL) is safe this check is probably not required

Cc: Roland Dreier <roland-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
Cc: Sean Hefty <sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Signed-off-by: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>
Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/ulp/ipoib/ipoib_fs.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/ulp/ipoib/ipoib_fs.c b/drivers/infiniband/ulp/ipoib/ipoib_fs.c
index 50061854616..6bd5740e269 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_fs.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_fs.c
@@ -281,10 +281,8 @@ void ipoib_delete_debug_files(struct net_device *dev)
 {
 	struct ipoib_dev_priv *priv = netdev_priv(dev);
 
-	if (priv->mcg_dentry)
-		debugfs_remove(priv->mcg_dentry);
-	if (priv->path_dentry)
-		debugfs_remove(priv->path_dentry);
+	debugfs_remove(priv->mcg_dentry);
+	debugfs_remove(priv->path_dentry);
 }
 
 int ipoib_register_debugfs(void)
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 4/8] IB/mlx4: use ARRAY_SIZE instead of sizeof/sizeof[0]

[Doug Ledford]

From: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>

use macro definition

Cc: Roland Dreier <roland-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
Cc: Sean Hefty <sean.hefty-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Signed-off-by: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>
Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/mlx4/main.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 828a37b2481..e1e558a3d69 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -910,8 +910,7 @@ static int __mlx4_ib_default_rules_match(struct ib_qp *qp,
 	const struct default_rules *pdefault_rules = default_table;
 	u8 link_layer = rdma_port_get_link_layer(qp->device, flow_attr->port);
 
-	for (i = 0; i < sizeof(default_table)/sizeof(default_table[0]); i++,
-	     pdefault_rules++) {
+	for (i = 0; i < ARRAY_SIZE(default_table); i++, pdefault_rules++) {
 		__u32 field_types[IB_FLOW_SPEC_SUPPORT_LAYERS];
 		memset(&field_types, 0, sizeof(field_types));
 
@@ -965,8 +964,7 @@ static int __mlx4_ib_create_default_rules(
 	int size = 0;
 	int i;
 
-	for (i = 0; i < sizeof(pdefault_rules->rules_create_list)/
-			sizeof(pdefault_rules->rules_create_list[0]); i++) {
+	for (i = 0; i < ARRAY_SIZE(pdefault_rules->rules_create_list); i++) {
 		int ret;
 		union ib_flow_spec ib_spec;
 		switch (pdefault_rules->rules_create_list[i]) {
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 5/8] IB/mlx5: use ARRAY_SIZE instead of sizeof/sizeof[0]

[Doug Ledford]

From: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>

use macro definition

Cc: Eli Cohen <eli-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Cc: Roland Dreier <roland-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org>
Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Acked-by: Eli Cohen <eli-VPRAkNaXOzVWk0Htik3J/w@public.gmane.org>
Signed-off-by: Fabian Frederick <fabf-AgBVmzD5pcezQB+pC5nmwQ@public.gmane.org>
Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/hw/mlx5/qp.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c
index bbbcf389272..416cb724422 100644
--- a/drivers/infiniband/hw/mlx5/qp.c
+++ b/drivers/infiniband/hw/mlx5/qp.c
@@ -2501,7 +2501,7 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 	spin_lock_irqsave(&qp->sq.lock, flags);
 
 	for (nreq = 0; wr; nreq++, wr = wr->next) {
-		if (unlikely(wr->opcode >= sizeof(mlx5_ib_opcode) / sizeof(mlx5_ib_opcode[0]))) {
+		if (unlikely(wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) {
 			mlx5_ib_warn(dev, "\n");
 			err = -EINVAL;
 			*bad_wr = wr;
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 6/8] ib/srpt: Enhance printk output

[Doug Ledford]

The ib_srpt module has a metric ton of printks for debugging and
warnings and such.  And it never prefaces a single one of them with
information to allow an admin to know what errant module is spewing
garbage all through their dmesg output.  Fix that.

Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/ulp/srpt/ib_srpt.c | 201 +++++++++++++++++-----------------
 1 file changed, 101 insertions(+), 100 deletions(-)

diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 8a8311e35cb..238dfe2d8fb 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -182,7 +182,7 @@ static void srpt_event_handler(struct ib_event_handler *handler,
 	if (!sdev || sdev->device != event->device)
 		return;
 
-	pr_debug("ASYNC event= %d on device= %s\n", event->event,
+	pr_debug("ib_srpt: ASYNC event= %d on device= %s\n", event->event,
 		 srpt_sdev_name(sdev));
 
 	switch (event->event) {
@@ -206,7 +206,7 @@ static void srpt_event_handler(struct ib_event_handler *handler,
 		}
 		break;
 	default:
-		printk(KERN_ERR "received unrecognized IB event %d\n",
+		printk(KERN_ERR "ib_srpt: received unrecognized IB event %d\n",
 		       event->event);
 		break;
 	}
@@ -217,7 +217,7 @@ static void srpt_event_handler(struct ib_event_handler *handler,
  */
 static void srpt_srq_event(struct ib_event *event, void *ctx)
 {
-	printk(KERN_INFO "SRQ event %d\n", event->event);
+	printk(KERN_INFO "ib_srpt: SRQ event %d\n", event->event);
 }
 
 /**
@@ -225,7 +225,7 @@ static void srpt_srq_event(struct ib_event *event, void *ctx)
  */
 static void srpt_qp_event(struct ib_event *event, struct srpt_rdma_ch *ch)
 {
-	pr_debug("QP event %d on cm_id=%p sess_name=%s state=%d\n",
+	pr_debug("ib_srpt: QP event %d on cm_id=%p sess_name=%s state=%d\n",
 		 event->event, ch->cm_id, ch->sess_name, srpt_get_ch_state(ch));
 
 	switch (event->event) {
@@ -237,11 +237,11 @@ static void srpt_qp_event(struct ib_event *event, struct srpt_rdma_ch *ch)
 					       CH_RELEASING))
 			srpt_release_channel(ch);
 		else
-			pr_debug("%s: state %d - ignored LAST_WQE.\n",
+			pr_debug("ib_srpt: %s: state %d - ignored LAST_WQE.\n",
 				 ch->sess_name, srpt_get_ch_state(ch));
 		break;
 	default:
-		printk(KERN_ERR "received unrecognized IB QP event %d\n",
+		printk(KERN_ERR "ib_srpt: received unrecognized IB QP event %d\n",
 		       event->event);
 		break;
 	}
@@ -601,7 +601,7 @@ static void srpt_unregister_mad_agent(struct srpt_device *sdev)
 		sport = &sdev->port[i - 1];
 		WARN_ON(sport->port != i);
 		if (ib_modify_port(sdev->device, i, 0, &port_modify) < 0)
-			printk(KERN_ERR "disabling MAD processing failed.\n");
+			printk(KERN_ERR "ib_srpt: disabling MAD processing failed.\n");
 		if (sport->mad_agent) {
 			ib_unregister_mad_agent(sport->mad_agent);
 			sport->mad_agent = NULL;
@@ -809,7 +809,7 @@ static int srpt_post_send(struct srpt_rdma_ch *ch,
 
 	ret = -ENOMEM;
 	if (unlikely(atomic_dec_return(&ch->sq_wr_avail) < 0)) {
-		printk(KERN_WARNING "IB send queue full (needed 1)\n");
+		printk(KERN_WARNING "ib_srpt: IB send queue full (needed 1)\n");
 		goto out;
 	}
 
@@ -911,7 +911,7 @@ static int srpt_get_desc_tbl(struct srpt_send_ioctx *ioctx,
 
 		if (ioctx->n_rbuf >
 		    (srp_cmd->data_out_desc_cnt + srp_cmd->data_in_desc_cnt)) {
-			printk(KERN_ERR "received unsupported SRP_CMD request"
+			printk(KERN_ERR "ib_srpt: received unsupported SRP_CMD request"
 			       " type (%u out + %u in != %u / %zu)\n",
 			       srp_cmd->data_out_desc_cnt,
 			       srp_cmd->data_in_desc_cnt,
@@ -1338,7 +1338,7 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
 		goto out;
 	}
 
-	pr_debug("Aborting cmd with state %d and tag %lld\n", state,
+	pr_debug("ib_srpt: Aborting cmd with state %d and tag %lld\n", state,
 		 ioctx->tag);
 
 	switch (state) {
@@ -1372,7 +1372,7 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
 		target_put_sess_cmd(ioctx->ch->sess, &ioctx->cmd);
 		break;
 	default:
-		WARN(1, "Unexpected command state (%d)", state);
+		WARN(1, "ib_srpt: Unexpected command state (%d)", state);
 		break;
 	}
 
@@ -1425,14 +1425,15 @@ static void srpt_handle_send_comp(struct srpt_rdma_ch *ch,
 	if (WARN_ON(state != SRPT_STATE_CMD_RSP_SENT
 		    && state != SRPT_STATE_MGMT_RSP_SENT
 		    && state != SRPT_STATE_DONE))
-		pr_debug("state = %d\n", state);
+		pr_debug("ib_srpt: bad state in srpt_handle_send_comp, "
+			 "state = %d\n", state);
 
 	if (state != SRPT_STATE_DONE) {
 		srpt_unmap_sg_to_ib_sge(ch, ioctx);
 		transport_generic_free_cmd(&ioctx->cmd, 0);
 	} else {
-		printk(KERN_ERR "IB completion has been received too late for"
-		       " wr_id = %u.\n", ioctx->ioctx.index);
+		printk(KERN_ERR "ib_srpt: IB completion has been received too "
+		       "late for wr_id = %u.\n", ioctx->ioctx.index);
 	}
 }
 
@@ -1461,7 +1462,7 @@ static void srpt_handle_rdma_comp(struct srpt_rdma_ch *ch,
 	} else if (opcode == SRPT_RDMA_ABORT) {
 		ioctx->rdma_aborted = true;
 	} else {
-		WARN(true, "unexpected opcode %d\n", opcode);
+		WARN(true, "ib_srpt: unexpected opcode %d\n", opcode);
 	}
 }
 
@@ -1480,7 +1481,7 @@ static void srpt_handle_rdma_err_comp(struct srpt_rdma_ch *ch,
 	switch (opcode) {
 	case SRPT_RDMA_READ_LAST:
 		if (ioctx->n_rdma <= 0) {
-			printk(KERN_ERR "Received invalid RDMA read"
+			printk(KERN_ERR "ib_srpt: Received invalid RDMA read"
 			       " error completion with idx %d\n",
 			       ioctx->ioctx.index);
 			break;
@@ -1548,7 +1549,7 @@ static int srpt_build_cmd_rsp(struct srpt_rdma_ch *ch,
 		BUILD_BUG_ON(MIN_MAX_RSP_SIZE <= sizeof(*srp_rsp));
 		max_sense_len = ch->max_ti_iu_len - sizeof(*srp_rsp);
 		if (sense_data_len > max_sense_len) {
-			printk(KERN_WARNING "truncated sense data from %d to %d"
+			printk(KERN_WARNING "ib_srpt: truncated sense data from %d to %d"
 			       " bytes\n", sense_data_len, max_sense_len);
 			sense_data_len = max_sense_len;
 		}
@@ -1627,7 +1628,7 @@ static uint64_t srpt_unpack_lun(const uint8_t *lun, int len)
 	int addressing_method;
 
 	if (unlikely(len < 2)) {
-		printk(KERN_ERR "Illegal LUN length %d, expected 2 bytes or "
+		printk(KERN_ERR "ib_srpt: Illegal LUN length %d, expected 2 bytes or "
 		       "more", len);
 		goto out;
 	}
@@ -1662,7 +1663,7 @@ static uint64_t srpt_unpack_lun(const uint8_t *lun, int len)
 
 	case SCSI_LUN_ADDR_METHOD_EXTENDED_LUN:
 	default:
-		printk(KERN_ERR "Unimplemented LUN addressing method %u",
+		printk(KERN_ERR "ib_srpt: Unimplemented LUN addressing method %u",
 		       addressing_method);
 		break;
 	}
@@ -1671,7 +1672,7 @@ out:
 	return res;
 
 out_err:
-	printk(KERN_ERR "Support for multi-level LUNs has not yet been"
+	printk(KERN_ERR "ib_srpt: Support for multi-level LUNs has not yet been"
 	       " implemented");
 	goto out;
 }
@@ -1722,7 +1723,7 @@ static int srpt_handle_cmd(struct srpt_rdma_ch *ch,
 	}
 
 	if (srpt_get_desc_tbl(send_ioctx, srp_cmd, &dir, &data_len)) {
-		printk(KERN_ERR "0x%llx: parsing SRP descriptor table failed.\n",
+		printk(KERN_ERR "ib_srpt: 0x%llx: parsing SRP descriptor table failed.\n",
 		       srp_cmd->tag);
 		ret = TCM_INVALID_CDB_FIELD;
 		goto send_sense;
@@ -1828,7 +1829,7 @@ static void srpt_handle_tsk_mgmt(struct srpt_rdma_ch *ch,
 	srp_tsk = recv_ioctx->ioctx.buf;
 	cmd = &send_ioctx->cmd;
 
-	pr_debug("recv tsk_mgmt fn %d for task_tag %lld and cmd tag %lld"
+	pr_debug("ib_srpt: recv tsk_mgmt fn %d for task_tag %lld and cmd tag %lld"
 		 " cm_id %p sess %p\n", srp_tsk->tsk_mgmt_func,
 		 srp_tsk->task_tag, srp_tsk->tag, ch->cm_id, ch->sess);
 
@@ -1911,19 +1912,19 @@ static void srpt_handle_new_iu(struct srpt_rdma_ch *ch,
 		srpt_handle_tsk_mgmt(ch, recv_ioctx, send_ioctx);
 		break;
 	case SRP_I_LOGOUT:
-		printk(KERN_ERR "Not yet implemented: SRP_I_LOGOUT\n");
+		printk(KERN_ERR "ib_srpt: Not yet implemented: SRP_I_LOGOUT\n");
 		break;
 	case SRP_CRED_RSP:
-		pr_debug("received SRP_CRED_RSP\n");
+		pr_debug("ib_srpt: received SRP_CRED_RSP\n");
 		break;
 	case SRP_AER_RSP:
-		pr_debug("received SRP_AER_RSP\n");
+		pr_debug("ib_srpt: received SRP_AER_RSP\n");
 		break;
 	case SRP_RSP:
-		printk(KERN_ERR "Received SRP_RSP\n");
+		printk(KERN_ERR "ib_srpt: Received SRP_RSP\n");
 		break;
 	default:
-		printk(KERN_ERR "received IU with unknown opcode 0x%x\n",
+		printk(KERN_ERR "ib_srpt: received IU with unknown opcode 0x%x\n",
 		       srp_cmd->opcode);
 		break;
 	}
@@ -1947,11 +1948,11 @@ static void srpt_process_rcv_completion(struct ib_cq *cq,
 
 		req_lim = atomic_dec_return(&ch->req_lim);
 		if (unlikely(req_lim < 0))
-			printk(KERN_ERR "req_lim = %d < 0\n", req_lim);
+			printk(KERN_ERR "ib_srpt: req_lim = %d < 0\n", req_lim);
 		ioctx = sdev->ioctx_ring[index];
 		srpt_handle_new_iu(ch, ioctx, NULL);
 	} else {
-		printk(KERN_INFO "receiving failed for idx %u with status %d\n",
+		printk(KERN_INFO "ib_srpt: receiving failed for idx %u with status %d\n",
 		       index, wc->status);
 	}
 }
@@ -1992,11 +1993,11 @@ static void srpt_process_send_completion(struct ib_cq *cq,
 		}
 	} else {
 		if (opcode == SRPT_SEND) {
-			printk(KERN_INFO "sending response for idx %u failed"
+			printk(KERN_INFO "ib_srpt: sending response for idx %u failed"
 			       " with status %d\n", index, wc->status);
 			srpt_handle_send_err_comp(ch, wc->wr_id);
 		} else if (opcode != SRPT_RDMA_MID) {
-			printk(KERN_INFO "RDMA t %d for idx %u failed with"
+			printk(KERN_INFO "ib_srpt: RDMA t %d for idx %u failed with"
 				" status %d", opcode, index, wc->status);
 			srpt_handle_rdma_err_comp(ch, send_ioctx, opcode);
 		}
@@ -2061,14 +2062,14 @@ static int srpt_compl_thread(void *arg)
 
 	ch = arg;
 	BUG_ON(!ch);
-	printk(KERN_INFO "Session %s: kernel thread %s (PID %d) started\n",
+	printk(KERN_INFO "ib_srpt: Session %s: kernel thread %s (PID %d) started\n",
 	       ch->sess_name, ch->thread->comm, current->pid);
 	while (!kthread_should_stop()) {
 		wait_event_interruptible(ch->wait_queue,
 			(srpt_process_completion(ch->cq, ch),
 			 kthread_should_stop()));
 	}
-	printk(KERN_INFO "Session %s: kernel thread %s (PID %d) stopped\n",
+	printk(KERN_INFO "ib_srpt: Session %s: kernel thread %s (PID %d) stopped\n",
 	       ch->sess_name, ch->thread->comm, current->pid);
 	return 0;
 }
@@ -2095,7 +2096,7 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch)
 			      ch->rq_size + srp_sq_size, 0);
 	if (IS_ERR(ch->cq)) {
 		ret = PTR_ERR(ch->cq);
-		printk(KERN_ERR "failed to create CQ cqe= %d ret= %d\n",
+		printk(KERN_ERR "ib_srpt: failed to create CQ cqe= %d ret= %d\n",
 		       ch->rq_size + srp_sq_size, ret);
 		goto out;
 	}
@@ -2114,13 +2115,13 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch)
 	ch->qp = ib_create_qp(sdev->pd, qp_init);
 	if (IS_ERR(ch->qp)) {
 		ret = PTR_ERR(ch->qp);
-		printk(KERN_ERR "failed to create_qp ret= %d\n", ret);
+		printk(KERN_ERR "ib_srpt: failed to create_qp ret= %d\n", ret);
 		goto err_destroy_cq;
 	}
 
 	atomic_set(&ch->sq_wr_avail, qp_init->cap.max_send_wr);
 
-	pr_debug("%s: max_cqe= %d max_sge= %d sq_size = %d cm_id= %p\n",
+	pr_debug("ib_srpt: %s: max_cqe= %d max_sge= %d sq_size = %d cm_id= %p\n",
 		 __func__, ch->cq->cqe, qp_init->cap.max_send_sge,
 		 qp_init->cap.max_send_wr, ch->cm_id);
 
@@ -2130,11 +2131,11 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch)
 
 	init_waitqueue_head(&ch->wait_queue);
 
-	pr_debug("creating thread for session %s\n", ch->sess_name);
+	pr_debug("ib_srpt: creating thread for session %s\n", ch->sess_name);
 
 	ch->thread = kthread_run(srpt_compl_thread, ch, "ib_srpt_compl");
 	if (IS_ERR(ch->thread)) {
-		printk(KERN_ERR "failed to create kernel thread %ld\n",
+		printk(KERN_ERR "ib_srpt: failed to create kernel thread %ld\n",
 		       PTR_ERR(ch->thread));
 		ch->thread = NULL;
 		goto err_destroy_qp;
@@ -2195,7 +2196,7 @@ static void __srpt_close_ch(struct srpt_rdma_ch *ch)
 		/* fall through */
 	case CH_LIVE:
 		if (ib_send_cm_dreq(ch->cm_id, NULL, 0) < 0)
-			printk(KERN_ERR "sending CM DREQ failed.\n");
+			printk(KERN_ERR "ib_srpt: sending CM DREQ failed.\n");
 		break;
 	case CH_DISCONNECTING:
 		break;
@@ -2282,7 +2283,7 @@ static void srpt_drain_channel(struct ib_cm_id *cm_id)
 
 		ret = srpt_ch_qp_err(ch);
 		if (ret < 0)
-			printk(KERN_ERR "Setting queue pair in error state"
+			printk(KERN_ERR "ib_srpt: Setting queue pair in error state"
 			       " failed: %d\n", ret);
 	}
 }
@@ -2335,7 +2336,7 @@ static void srpt_release_channel_work(struct work_struct *w)
 	struct se_session *se_sess;
 
 	ch = container_of(w, struct srpt_rdma_ch, release_work);
-	pr_debug("ch = %p; ch->sess = %p; release_done = %p\n", ch, ch->sess,
+	pr_debug("ib_srpt: ch = %p; ch->sess = %p; release_done = %p\n", ch, ch->sess,
 		 ch->release_done);
 
 	sdev = ch->sport->sdev;
@@ -2426,7 +2427,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 
 	it_iu_len = be32_to_cpu(req->req_it_iu_len);
 
-	printk(KERN_INFO "Received SRP_LOGIN_REQ with i_port_id 0x%llx:0x%llx,"
+	printk(KERN_INFO "ib_srpt: Received SRP_LOGIN_REQ with i_port_id 0x%llx:0x%llx,"
 	       " t_port_id 0x%llx:0x%llx and it_iu_len %d on port %d"
 	       " (guid=0x%llx:0x%llx)\n",
 	       be64_to_cpu(*(__be64 *)&req->initiator_port_id[0]),
@@ -2451,7 +2452,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_REQ_IT_IU_LENGTH_TOO_LARGE);
 		ret = -EINVAL;
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because its"
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because its"
 		       " length (%d bytes) is out of range (%d .. %d)\n",
 		       it_iu_len, 64, srp_max_req_size);
 		goto reject;
@@ -2461,7 +2462,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 		rej->reason = __constant_cpu_to_be32(
 			     SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
 		ret = -EINVAL;
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because the target port"
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because the target port"
 		       " has not yet been enabled\n");
 		goto reject;
 	}
@@ -2485,7 +2486,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 					continue;
 
 				/* found an existing channel */
-				pr_debug("Found existing channel %s"
+				pr_debug("ib_srpt: Found existing channel %s"
 					 " cm_id= %p state= %d\n",
 					 ch->sess_name, ch->cm_id, ch_state);
 
@@ -2507,7 +2508,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_UNABLE_ASSOCIATE_CHANNEL);
 		ret = -ENOMEM;
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because it"
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because it"
 		       " has an invalid target port identifier.\n");
 		goto reject;
 	}
@@ -2516,7 +2517,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 	if (!ch) {
 		rej->reason = __constant_cpu_to_be32(
 					SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because no memory.\n");
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because no memory.\n");
 		ret = -ENOMEM;
 		goto reject;
 	}
@@ -2553,7 +2554,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 	if (ret) {
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because creating"
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because creating"
 		       " a new RDMA channel failed.\n");
 		goto free_ring;
 	}
@@ -2562,7 +2563,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 	if (ret) {
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
-		printk(KERN_ERR "rejected SRP_LOGIN_REQ because enabling"
+		printk(KERN_ERR "ib_srpt: rejected SRP_LOGIN_REQ because enabling"
 		       " RTR failed (error code = %d)\n", ret);
 		goto destroy_ib;
 	}
@@ -2573,11 +2574,11 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 			be64_to_cpu(*(__be64 *)ch->i_port_id),
 			be64_to_cpu(*(__be64 *)(ch->i_port_id + 8)));
 
-	pr_debug("registering session %s\n", ch->sess_name);
+	pr_debug("ib_srpt: registering session %s\n", ch->sess_name);
 
 	nacl = srpt_lookup_acl(sport, ch->i_port_id);
 	if (!nacl) {
-		printk(KERN_INFO "Rejected login because no ACL has been"
+		printk(KERN_INFO "ib_srpt: Rejected login because no ACL has been"
 		       " configured yet for initiator %s.\n", ch->sess_name);
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_CHANNEL_LIMIT_REACHED);
@@ -2588,13 +2589,13 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 	if (IS_ERR(ch->sess)) {
 		rej->reason = __constant_cpu_to_be32(
 				SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
-		pr_debug("Failed to create session\n");
+		pr_debug("ib_srpt: Failed to create session\n");
 		goto deregister_session;
 	}
 	ch->sess->se_node_acl = &nacl->nacl;
 	transport_register_session(&sport->port_tpg_1, &nacl->nacl, ch->sess, ch);
 
-	pr_debug("Establish connection sess=%p name=%s cm_id=%p\n", ch->sess,
+	pr_debug("ib_srpt: Establish connection sess=%p name=%s cm_id=%p\n", ch->sess,
 		 ch->sess_name, ch->cm_id);
 
 	/* create srp_login_response */
@@ -2622,7 +2623,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
 
 	ret = ib_send_cm_rep(cm_id, rep_param);
 	if (ret) {
-		printk(KERN_ERR "sending SRP_LOGIN_REQ response failed"
+		printk(KERN_ERR "ib_srpt: sending SRP_LOGIN_REQ response failed"
 		       " (error code = %d)\n", ret);
 		goto release_channel;
 	}
@@ -2670,7 +2671,7 @@ out:
 
 static void srpt_cm_rej_recv(struct ib_cm_id *cm_id)
 {
-	printk(KERN_INFO "Received IB REJ for cm_id %p.\n", cm_id);
+	printk(KERN_INFO "ib_srpt: Received IB REJ for cm_id %p.\n", cm_id);
 	srpt_drain_channel(cm_id);
 }
 
@@ -2705,13 +2706,13 @@ static void srpt_cm_rtu_recv(struct ib_cm_id *cm_id)
 
 static void srpt_cm_timewait_exit(struct ib_cm_id *cm_id)
 {
-	printk(KERN_INFO "Received IB TimeWait exit for cm_id %p.\n", cm_id);
+	printk(KERN_INFO "ib_srpt: Received IB TimeWait exit for cm_id %p.\n", cm_id);
 	srpt_drain_channel(cm_id);
 }
 
 static void srpt_cm_rep_error(struct ib_cm_id *cm_id)
 {
-	printk(KERN_INFO "Received IB REP error for cm_id %p.\n", cm_id);
+	printk(KERN_INFO "ib_srpt: Received IB REP error for cm_id %p.\n", cm_id);
 	srpt_drain_channel(cm_id);
 }
 
@@ -2727,7 +2728,7 @@ static void srpt_cm_dreq_recv(struct ib_cm_id *cm_id)
 	ch = srpt_find_channel(cm_id->context, cm_id);
 	BUG_ON(!ch);
 
-	pr_debug("cm_id= %p ch->state= %d\n", cm_id, srpt_get_ch_state(ch));
+	pr_debug("ib_srpt: cm_id= %p ch->state= %d\n", cm_id, srpt_get_ch_state(ch));
 
 	spin_lock_irqsave(&ch->spinlock, flags);
 	switch (ch->state) {
@@ -2739,15 +2740,15 @@ static void srpt_cm_dreq_recv(struct ib_cm_id *cm_id)
 	case CH_DISCONNECTING:
 	case CH_DRAINING:
 	case CH_RELEASING:
-		WARN(true, "unexpected channel state %d\n", ch->state);
+		WARN(true, "ib_srpt: unexpected channel state %d\n", ch->state);
 		break;
 	}
 	spin_unlock_irqrestore(&ch->spinlock, flags);
 
 	if (send_drep) {
 		if (ib_send_cm_drep(ch->cm_id, NULL, 0) < 0)
-			printk(KERN_ERR "Sending IB DREP failed.\n");
-		printk(KERN_INFO "Received DREQ and sent DREP for session %s.\n",
+			printk(KERN_ERR "ib_srpt: Sending IB DREP failed.\n");
+		printk(KERN_INFO "ib_srpt: Received DREQ and sent DREP for session %s.\n",
 		       ch->sess_name);
 	}
 }
@@ -2757,7 +2758,7 @@ static void srpt_cm_dreq_recv(struct ib_cm_id *cm_id)
  */
 static void srpt_cm_drep_recv(struct ib_cm_id *cm_id)
 {
-	printk(KERN_INFO "Received InfiniBand DREP message for cm_id %p.\n",
+	printk(KERN_INFO "ib_srpt: Received InfiniBand DREP message for cm_id %p.\n",
 	       cm_id);
 	srpt_drain_channel(cm_id);
 }
@@ -2802,13 +2803,13 @@ static int srpt_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
 		srpt_cm_rep_error(cm_id);
 		break;
 	case IB_CM_DREQ_ERROR:
-		printk(KERN_INFO "Received IB DREQ ERROR event.\n");
+		printk(KERN_INFO "ib_srpt: Received IB DREQ ERROR event.\n");
 		break;
 	case IB_CM_MRA_RECEIVED:
-		printk(KERN_INFO "Received IB MRA event\n");
+		printk(KERN_INFO "ib_srpt: Received IB MRA event\n");
 		break;
 	default:
-		printk(KERN_ERR "received unrecognized IB CM event %d\n",
+		printk(KERN_ERR "ib_srpt: received unrecognized IB CM event %d\n",
 		       event->event);
 		break;
 	}
@@ -2839,7 +2840,7 @@ static int srpt_perform_rdmas(struct srpt_rdma_ch *ch,
 		ret = -ENOMEM;
 		sq_wr_avail = atomic_sub_return(n_rdma, &ch->sq_wr_avail);
 		if (sq_wr_avail < 0) {
-			printk(KERN_WARNING "IB send queue full (needed %d)\n",
+			printk(KERN_WARNING "ib_srpt: IB send queue full (needed %d)\n",
 			       n_rdma);
 			goto out;
 		}
@@ -2888,12 +2889,12 @@ static int srpt_perform_rdmas(struct srpt_rdma_ch *ch,
 		wr.send_flags = IB_SEND_SIGNALED;
 		while (ch->state == CH_LIVE &&
 			ib_post_send(ch->qp, &wr, &bad_wr) != 0) {
-			printk(KERN_INFO "Trying to abort failed RDMA transfer [%d]",
+			printk(KERN_INFO "ib_srpt: Trying to abort failed RDMA transfer [%d]",
 				ioctx->ioctx.index);
 			msleep(1000);
 		}
 		while (ch->state != CH_RELEASING && !ioctx->rdma_aborted) {
-			printk(KERN_INFO "Waiting until RDMA abort finished [%d]",
+			printk(KERN_INFO "ib_srpt: Waiting until RDMA abort finished [%d]",
 				ioctx->ioctx.index);
 			msleep(1000);
 		}
@@ -2966,7 +2967,7 @@ static int srpt_write_pending(struct se_cmd *se_cmd)
 	ch_state = srpt_get_ch_state(ch);
 	switch (ch_state) {
 	case CH_CONNECTING:
-		WARN(true, "unexpected channel state %d\n", ch_state);
+		WARN(true, "ib_srpt: unexpected channel state %d\n", ch_state);
 		ret = -EINVAL;
 		goto out;
 	case CH_LIVE:
@@ -2974,7 +2975,7 @@ static int srpt_write_pending(struct se_cmd *se_cmd)
 	case CH_DISCONNECTING:
 	case CH_DRAINING:
 	case CH_RELEASING:
-		pr_debug("cmd with tag %lld: channel disconnecting\n",
+		pr_debug("ib_srpt: cmd with tag %lld: channel disconnecting\n",
 			 ioctx->tag);
 		srpt_set_cmd_state(ioctx, SRPT_STATE_DATA_IN);
 		ret = -EINVAL;
@@ -3029,7 +3030,7 @@ static void srpt_queue_response(struct se_cmd *cmd)
 		ioctx->state = SRPT_STATE_MGMT_RSP_SENT;
 		break;
 	default:
-		WARN(true, "ch %p; cmd %d: unexpected command state %d\n",
+		WARN(true, "ib_srpt: ch %p; cmd %d: unexpected command state %d\n",
 			ch, ioctx->ioctx.index, ioctx->state);
 		break;
 	}
@@ -3049,7 +3050,7 @@ static void srpt_queue_response(struct se_cmd *cmd)
 	    !ioctx->queue_status_only) {
 		ret = srpt_xfer_data(ch, ioctx);
 		if (ret) {
-			printk(KERN_ERR "xfer_data failed for tag %llu\n",
+			printk(KERN_ERR "ib_srpt: xfer_data failed for tag %llu\n",
 			       ioctx->tag);
 			return;
 		}
@@ -3066,7 +3067,7 @@ static void srpt_queue_response(struct se_cmd *cmd)
 	}
 	ret = srpt_post_send(ch, ioctx, resp_len);
 	if (ret) {
-		printk(KERN_ERR "sending cmd response failed for tag %llu\n",
+		printk(KERN_ERR "ib_srpt: sending cmd response failed for tag %llu\n",
 		       ioctx->tag);
 		srpt_unmap_sg_to_ib_sge(ch, ioctx);
 		srpt_set_cmd_state(ioctx, SRPT_STATE_DONE);
@@ -3194,7 +3195,7 @@ static void srpt_add_one(struct ib_device *device)
 	struct ib_srq_init_attr srq_attr;
 	int i;
 
-	pr_debug("device = %p, device->dma_ops = %p\n", device,
+	pr_debug("ib_srpt: device = %p, device->dma_ops = %p\n", device,
 		 device->dma_ops);
 
 	sdev = kzalloc(sizeof *sdev, GFP_KERNEL);
@@ -3230,7 +3231,7 @@ static void srpt_add_one(struct ib_device *device)
 	if (IS_ERR(sdev->srq))
 		goto err_mr;
 
-	pr_debug("%s: create SRQ #wr= %d max_allow=%d dev= %s\n",
+	pr_debug("ib_srpt: %s: create SRQ #wr= %d max_allow=%d dev= %s\n",
 		 __func__, sdev->srq_size, sdev->dev_attr.max_srq_wr,
 		 device->name);
 
@@ -3242,7 +3243,7 @@ static void srpt_add_one(struct ib_device *device)
 		goto err_srq;
 
 	/* print out target login information */
-	pr_debug("Target login info: id_ext=%016llx,ioc_guid=%016llx,"
+	pr_debug("ib_srpt: Target login info: id_ext=%016llx,ioc_guid=%016llx,"
 		 "pkey=ffff,service_id=%016llx\n", srpt_service_guid,
 		 srpt_service_guid, srpt_service_guid);
 
@@ -3284,7 +3285,7 @@ static void srpt_add_one(struct ib_device *device)
 		spin_lock_init(&sport->port_acl_lock);
 
 		if (srpt_refresh_port(sport)) {
-			printk(KERN_ERR "MAD registration failed for %s-%d.\n",
+			printk(KERN_ERR "ib_srpt: MAD registration failed for %s-%d.\n",
 			       srpt_sdev_name(sdev), i);
 			goto err_ring;
 		}
@@ -3300,7 +3301,7 @@ static void srpt_add_one(struct ib_device *device)
 
 out:
 	ib_set_client_data(device, &srpt_client, sdev);
-	pr_debug("added %s.\n", device->name);
+	pr_debug("ib_srpt: added %s.\n", device->name);
 	return;
 
 err_ring:
@@ -3455,7 +3456,7 @@ static struct se_node_acl *srpt_alloc_fabric_acl(struct se_portal_group *se_tpg)
 
 	nacl = kzalloc(sizeof(struct srpt_node_acl), GFP_KERNEL);
 	if (!nacl) {
-		printk(KERN_ERR "Unable to allocate struct srpt_node_acl\n");
+		printk(KERN_ERR "ib_srpt: Unable to allocate struct srpt_node_acl\n");
 		return NULL;
 	}
 
@@ -3514,7 +3515,7 @@ static void srpt_close_session(struct se_session *se_sess)
 	ch = se_sess->fabric_sess_ptr;
 	WARN_ON(ch->sess != se_sess);
 
-	pr_debug("ch %p state %d\n", ch, srpt_get_ch_state(ch));
+	pr_debug("ib_srpt: ch %p state %d\n", ch, srpt_get_ch_state(ch));
 
 	sdev = ch->sport->sdev;
 	spin_lock_irq(&sdev->spinlock);
@@ -3584,7 +3585,7 @@ static int srpt_parse_i_port_id(u8 i_port_id[16], const char *name)
 	memset(i_port_id, 0, leading_zero_bytes);
 	rc = hex2bin(i_port_id + leading_zero_bytes, p, count);
 	if (rc < 0)
-		pr_debug("hex2bin failed for srpt_parse_i_port_id: %d\n", rc);
+		pr_debug("ib_srpt: hex2bin failed for srpt_parse_i_port_id: %d\n", rc);
 	ret = 0;
 out:
 	return ret;
@@ -3606,7 +3607,7 @@ static struct se_node_acl *srpt_make_nodeacl(struct se_portal_group *tpg,
 	u8 i_port_id[16];
 
 	if (srpt_parse_i_port_id(i_port_id, name) < 0) {
-		printk(KERN_ERR "invalid initiator port ID %s\n", name);
+		printk(KERN_ERR "ib_srpt: invalid initiator port ID %s\n", name);
 		ret = -EINVAL;
 		goto err;
 	}
@@ -3680,16 +3681,16 @@ static ssize_t srpt_tpg_attrib_store_srp_max_rdma_size(
 
 	ret = kstrtoul(page, 0, &val);
 	if (ret < 0) {
-		pr_err("kstrtoul() failed with ret: %d\n", ret);
+		pr_err("ib_srpt: kstrtoul() failed with ret: %d\n", ret);
 		return -EINVAL;
 	}
 	if (val > MAX_SRPT_RDMA_SIZE) {
-		pr_err("val: %lu exceeds MAX_SRPT_RDMA_SIZE: %d\n", val,
+		pr_err("ib_srpt: val: %lu exceeds MAX_SRPT_RDMA_SIZE: %d\n", val,
 			MAX_SRPT_RDMA_SIZE);
 		return -EINVAL;
 	}
 	if (val < DEFAULT_MAX_RDMA_SIZE) {
-		pr_err("val: %lu smaller than DEFAULT_MAX_RDMA_SIZE: %d\n",
+		pr_err("ib_srpt: val: %lu smaller than DEFAULT_MAX_RDMA_SIZE: %d\n",
 			val, DEFAULT_MAX_RDMA_SIZE);
 		return -EINVAL;
 	}
@@ -3720,16 +3721,16 @@ static ssize_t srpt_tpg_attrib_store_srp_max_rsp_size(
 
 	ret = kstrtoul(page, 0, &val);
 	if (ret < 0) {
-		pr_err("kstrtoul() failed with ret: %d\n", ret);
+		pr_err("ib_srpt: kstrtoul() failed with ret: %d\n", ret);
 		return -EINVAL;
 	}
 	if (val > MAX_SRPT_RSP_SIZE) {
-		pr_err("val: %lu exceeds MAX_SRPT_RSP_SIZE: %d\n", val,
+		pr_err("ib_srpt: val: %lu exceeds MAX_SRPT_RSP_SIZE: %d\n", val,
 			MAX_SRPT_RSP_SIZE);
 		return -EINVAL;
 	}
 	if (val < MIN_MAX_RSP_SIZE) {
-		pr_err("val: %lu smaller than MIN_MAX_RSP_SIZE: %d\n", val,
+		pr_err("ib_srpt: val: %lu smaller than MIN_MAX_RSP_SIZE: %d\n", val,
 			MIN_MAX_RSP_SIZE);
 		return -EINVAL;
 	}
@@ -3760,16 +3761,16 @@ static ssize_t srpt_tpg_attrib_store_srp_sq_size(
 
 	ret = kstrtoul(page, 0, &val);
 	if (ret < 0) {
-		pr_err("kstrtoul() failed with ret: %d\n", ret);
+		pr_err("ib_srpt: kstrtoul() failed with ret: %d\n", ret);
 		return -EINVAL;
 	}
 	if (val > MAX_SRPT_SRQ_SIZE) {
-		pr_err("val: %lu exceeds MAX_SRPT_SRQ_SIZE: %d\n", val,
+		pr_err("ib_srpt: val: %lu exceeds MAX_SRPT_SRQ_SIZE: %d\n", val,
 			MAX_SRPT_SRQ_SIZE);
 		return -EINVAL;
 	}
 	if (val < MIN_SRPT_SRQ_SIZE) {
-		pr_err("val: %lu smaller than MIN_SRPT_SRQ_SIZE: %d\n", val,
+		pr_err("ib_srpt: val: %lu smaller than MIN_SRPT_SRQ_SIZE: %d\n", val,
 			MIN_SRPT_SRQ_SIZE);
 		return -EINVAL;
 	}
@@ -3807,12 +3808,12 @@ static ssize_t srpt_tpg_store_enable(
 
 	ret = kstrtoul(page, 0, &tmp);
 	if (ret < 0) {
-		printk(KERN_ERR "Unable to extract srpt_tpg_store_enable\n");
+		printk(KERN_ERR "ib_srpt: Unable to extract srpt_tpg_store_enable\n");
 		return -EINVAL;
 	}
 
 	if ((tmp != 0) && (tmp != 1)) {
-		printk(KERN_ERR "Illegal value for srpt_tpg_store_enable: %lu\n", tmp);
+		printk(KERN_ERR "ib_srpt: Illegal value for srpt_tpg_store_enable: %lu\n", tmp);
 		return -EINVAL;
 	}
 	if (tmp == 1)
@@ -3875,7 +3876,7 @@ static struct se_wwn *srpt_make_tport(struct target_fabric_configfs *tf,
 	int ret;
 
 	sport = srpt_lookup_port(name);
-	pr_debug("make_tport(%s)\n", name);
+	pr_debug("ib_srpt: make_tport(%s)\n", name);
 	ret = -EINVAL;
 	if (!sport)
 		goto err;
@@ -3894,7 +3895,7 @@ static void srpt_drop_tport(struct se_wwn *wwn)
 {
 	struct srpt_port *sport = container_of(wwn, struct srpt_port, port_wwn);
 
-	pr_debug("drop_tport(%s\n", config_item_name(&sport->port_wwn.wwn_group.cg_item));
+	pr_debug("ib_srpt: drop_tport(%s\n", config_item_name(&sport->port_wwn.wwn_group.cg_item));
 }
 
 static ssize_t srpt_wwn_show_attr_version(struct target_fabric_configfs *tf,
@@ -3971,7 +3972,7 @@ static int __init srpt_init_module(void)
 
 	ret = -EINVAL;
 	if (srp_max_req_size < MIN_MAX_REQ_SIZE) {
-		printk(KERN_ERR "invalid value %d for kernel module parameter"
+		printk(KERN_ERR "ib_srpt: invalid value %d for kernel module parameter"
 		       " srp_max_req_size -- must be at least %d.\n",
 		       srp_max_req_size, MIN_MAX_REQ_SIZE);
 		goto out;
@@ -3979,7 +3980,7 @@ static int __init srpt_init_module(void)
 
 	if (srpt_srq_size < MIN_SRPT_SRQ_SIZE
 	    || srpt_srq_size > MAX_SRPT_SRQ_SIZE) {
-		printk(KERN_ERR "invalid value %d for kernel module parameter"
+		printk(KERN_ERR "ib_srpt: invalid value %d for kernel module parameter"
 		       " srpt_srq_size -- must be in the range [%d..%d].\n",
 		       srpt_srq_size, MIN_SRPT_SRQ_SIZE, MAX_SRPT_SRQ_SIZE);
 		goto out;
@@ -3987,7 +3988,7 @@ static int __init srpt_init_module(void)
 
 	srpt_target = target_fabric_configfs_init(THIS_MODULE, "srpt");
 	if (IS_ERR(srpt_target)) {
-		printk(KERN_ERR "couldn't register\n");
+		printk(KERN_ERR "ib_srpt: couldn't register\n");
 		ret = PTR_ERR(srpt_target);
 		goto out;
 	}
@@ -4009,13 +4010,13 @@ static int __init srpt_init_module(void)
 
 	ret = target_fabric_configfs_register(srpt_target);
 	if (ret < 0) {
-		printk(KERN_ERR "couldn't register\n");
+		printk(KERN_ERR "ib_srpt: couldn't register\n");
 		goto out_free_target;
 	}
 
 	ret = ib_register_client(&srpt_client);
 	if (ret) {
-		printk(KERN_ERR "couldn't register IB client\n");
+		printk(KERN_ERR "ib_srpt: couldn't register IB client\n");
 		goto out_unregister_target;
 	}
 
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 7/8] ib/srpt: Handle GID change events

[Doug Ledford]

GID change events need a refresh just like LID change events and several
others.  Handle this the same as the others.

Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 drivers/infiniband/ulp/srpt/ib_srpt.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 238dfe2d8fb..1d00666cd0d 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -198,6 +198,7 @@ static void srpt_event_handler(struct ib_event_handler *handler,
 	case IB_EVENT_PKEY_CHANGE:
 	case IB_EVENT_SM_CHANGE:
 	case IB_EVENT_CLIENT_REREGISTER:
+	case IB_EVENT_GID_CHANGE:
 		/* Refresh port data asynchronously. */
 		if (event->element.port_num <= sdev->device->phys_port_cnt) {
 			sport = &sdev->port[event->element.port_num - 1];
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 8/8] uapi/rdma_user_cm.h: include socket.h

[Doug Ledford]

Commit ee7aed4528f (RDMA/ucma: Support querying for AF_IB addresses)
added struct sockaddr_storage to rdma_user_cm.h without also adding an
include for linux/socket.h to make sure it is defined.  As systemtap
needs the header files to build standalone and not rely on the code
files to pre-include other headers, add linux/socket.h to the list of
includes in this file.

This patch resolves Red Hat bugzilla #1072645.

Signed-off-by: Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
---
 include/uapi/rdma/rdma_user_cm.h | 1 +
 1 file changed, 1 insertion(+)

diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h
index 99b80abf360..3066718eb12 100644
--- a/include/uapi/rdma/rdma_user_cm.h
+++ b/include/uapi/rdma/rdma_user_cm.h
@@ -34,6 +34,7 @@
 #define RDMA_USER_CM_H
 
 #include <linux/types.h>
+#include <linux/socket.h>
 #include <linux/in6.h>
 #include <rdma/ib_user_verbs.h>
 #include <rdma/ib_user_sa.h>
-- 
1.9.3

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 6/8] ib/srpt: Enhance printk output

[Roland Dreier]

On Tue, Aug 12, 2014 at 4:20 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> The ib_srpt module has a metric ton of printks for debugging and
> warnings and such.  And it never prefaces a single one of them with
> information to allow an admin to know what errant module is spewing
> garbage all through their dmesg output.  Fix that.

Target patches should probably go through Nic's tree, although for
srpt stuff like this I guess I can take it.

However I also think it would be a lot cleaner to define pr_fmt() at
the top of the file like

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

and then convert the printks to pr_info() or pr_warn() instead of
having to type "ib_srpt:" everywhere.

 - R.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] IB/srp: Use P_Key cache for P_Key lookups

[Roland Dreier]

On Tue, Aug 12, 2014 at 4:20 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> -       ret = ib_find_pkey(target->srp_host->srp_dev->dev,
> -                          target->srp_host->port,
> -                          be16_to_cpu(target->path.pkey),
> -                          &attr->pkey_index);
> +       ret = ib_find_cached_pkey(target->srp_host->srp_dev->dev,
> +                                 target->srp_host->port,
> +                                 be16_to_cpu(target->path.pkey),
> +                                 &attr->pkey_index);

I consider the "cached" operations to be a deprecated interface, and
I'd rather remove uses than add new ones.  I have a vague dream of
fixing everything up so the normal interfaces (ib_find_pkey() etc)
don't sleep and are usable everywhere, although that dream will
probably never come true.

But is there any practical benefit to this change?  I don't think
saving a few milliseconds at log in time counts -- I would think it's
lost in the noise of how long it takes to actually log in, scan for
LUNs, etc.

 - R.
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/8] IB/srp: Use P_Key cache for P_Key lookups

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 4061 bytes --]

On Tue, 2014-08-12 at 21:56 -0700, Roland Dreier wrote:
> On Tue, Aug 12, 2014 at 4:20 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> > -       ret = ib_find_pkey(target->srp_host->srp_dev->dev,
> > -                          target->srp_host->port,
> > -                          be16_to_cpu(target->path.pkey),
> > -                          &attr->pkey_index);
> > +       ret = ib_find_cached_pkey(target->srp_host->srp_dev->dev,
> > +                                 target->srp_host->port,
> > +                                 be16_to_cpu(target->path.pkey),
> > +                                 &attr->pkey_index);
> 
> I consider the "cached" operations to be a deprecated interface,

I disagree.

>  and
> I'd rather remove uses than add new ones.

This isn't practical.  For the ib_get_* cache operations, you could
theoretically do away with the cache and not suffer too bad.  But this
is one of the ib_find_* operations.  These functions can not be done
away with.

>   I have a vague dream of
> fixing everything up so the normal interfaces (ib_find_pkey() etc)
> don't sleep and are usable everywhere,

Sleeping isn't the problem.  And that's probably part of why you are
still dreaming about this instead of accepting reality.  The problem is
the latency required to query the firmware on a card.  Some IB adapters
might keep their tables in kernel memory and can access them fast, but
some cards use firmware and the tables are there and when you query the
driver you are actually doing a query to the card via a firmware and a
mailbox and you have to wait for a return.  *That's* the problem.  And
specifically as it relates to the ib_find_* functions as this function
is, when the P_Key you're searching for isn't one of the first found,
the difference between the cache and direct to the card is very
significant.

>  although that dream will
> probably never come true.

No offense Roland, but not if I have anything to say about it ;-)  Not
because I don't want you to have your dreams of course, but this
particular dream is more nightmare for the rest of us than the unicorns
and rainbows raining skittles on the world that you envision it being.

> But is there any practical benefit to this change? 

Absolutely yes!

>  I don't think
> saving a few milliseconds at log in time counts

You can't think that way Roland....it breaks real world usage in very
unexpected ways.  Ways that took me, 3 support engineers from two
companies, and top level admins at a joint customer's sight over a month
to finally figure out.  So, let me be clear, this particular dream of
yours has wasted well over 4 man months of support/engineering time.
Email me off list if you want more details of how this nightmare dream
of yours did it.

>  -- I would think it's
> lost in the noise of how long it takes to actually log in, scan for
> LUNs, etc.

No, it's really not.  The simple case of finding the right element in
the first lookup or two is OK.  But this is the ib_find_* function, and
if it doesn't find the right element at first, it searches all available
elements until it does.  There are 128 slots (like there are for GID
indexes), so it's a few ms times 128 in the worst case scenario.  That's
not noise.  It's so far from noise that it kills real world
applications.  Things start timing out and resetting when you have more
than just a few items search the table for an element that isn't there.
It really can't be thought of in the terms you are thinking of it, which
is the common/best case scenario.  You *have* to consider worst case
scenario, and in that scenario, the cache functions are absolutely
essential to scalability.  Especially the ib_find_* variants.  And since
they are essential, they *must* be maintained and kept working properly,
and so there is no reason not to use them when they can be used.

-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 819 bytes --]