aboutsummaryrefslogtreecommitdiff
path: root/drivers/infiniband/ulp
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/ulp')
-rw-r--r--drivers/infiniband/ulp/Makefile5
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_cm.c18
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_ethtool.c2
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.c105
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.h10
-rw-r--r--drivers/infiniband/ulp/iser/iser_initiator.c34
-rw-r--r--drivers/infiniband/ulp/iser/iser_verbs.c89
-rw-r--r--drivers/infiniband/ulp/isert/ib_isert.c928
-rw-r--r--drivers/infiniband/ulp/isert/ib_isert.h42
-rw-r--r--drivers/infiniband/ulp/srp/ib_srp.c672
-rw-r--r--drivers/infiniband/ulp/srp/ib_srp.h94
-rw-r--r--drivers/infiniband/ulp/srpt/ib_srpt.c27
12 files changed, 1469 insertions, 557 deletions
diff --git a/drivers/infiniband/ulp/Makefile b/drivers/infiniband/ulp/Makefile
new file mode 100644
index 00000000000..f3c7dcf0309
--- /dev/null
+++ b/drivers/infiniband/ulp/Makefile
@@ -0,0 +1,5 @@
+obj-$(CONFIG_INFINIBAND_IPOIB) += ipoib/
+obj-$(CONFIG_INFINIBAND_SRP) += srp/
+obj-$(CONFIG_INFINIBAND_SRPT) += srpt/
+obj-$(CONFIG_INFINIBAND_ISER) += iser/
+obj-$(CONFIG_INFINIBAND_ISERT) += isert/
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
index 1377f85911c..933efcea0d0 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
@@ -1030,10 +1030,20 @@ static struct ib_qp *ipoib_cm_create_tx_qp(struct net_device *dev, struct ipoib_
.cap.max_send_sge = 1,
.sq_sig_type = IB_SIGNAL_ALL_WR,
.qp_type = IB_QPT_RC,
- .qp_context = tx
+ .qp_context = tx,
+ .create_flags = IB_QP_CREATE_USE_GFP_NOIO
};
- return ib_create_qp(priv->pd, &attr);
+ struct ib_qp *tx_qp;
+
+ tx_qp = ib_create_qp(priv->pd, &attr);
+ if (PTR_ERR(tx_qp) == -EINVAL) {
+ ipoib_warn(priv, "can't use GFP_NOIO for QPs on device %s, using GFP_KERNEL\n",
+ priv->ca->name);
+ attr.create_flags &= ~IB_QP_CREATE_USE_GFP_NOIO;
+ tx_qp = ib_create_qp(priv->pd, &attr);
+ }
+ return tx_qp;
}
static int ipoib_cm_send_req(struct net_device *dev,
@@ -1104,12 +1114,14 @@ static int ipoib_cm_tx_init(struct ipoib_cm_tx *p, u32 qpn,
struct ipoib_dev_priv *priv = netdev_priv(p->dev);
int ret;
- p->tx_ring = vzalloc(ipoib_sendq_size * sizeof *p->tx_ring);
+ p->tx_ring = __vmalloc(ipoib_sendq_size * sizeof *p->tx_ring,
+ GFP_NOIO, PAGE_KERNEL);
if (!p->tx_ring) {
ipoib_warn(priv, "failed to allocate tx ring\n");
ret = -ENOMEM;
goto err_tx;
}
+ memset(p->tx_ring, 0, ipoib_sendq_size * sizeof *p->tx_ring);
p->qp = ipoib_cm_create_tx_qp(p->dev, p);
if (IS_ERR(p->qp)) {
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c b/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c
index c4b3940845e..078cadd6c79 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ethtool.c
@@ -105,5 +105,5 @@ static const struct ethtool_ops ipoib_ethtool_ops = {
void ipoib_set_ethtool_ops(struct net_device *dev)
{
- SET_ETHTOOL_OPS(dev, &ipoib_ethtool_ops);
+ dev->ethtool_ops = &ipoib_ethtool_ops;
}
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c
index 25f195ef44b..eb7973957a6 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.c
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.c
@@ -99,6 +99,7 @@ MODULE_PARM_DESC(pi_enable, "Enable T10-PI offload support (default:disabled)");
module_param_named(pi_guard, iser_pi_guard, int, 0644);
MODULE_PARM_DESC(pi_guard, "T10-PI guard_type, 0:CRC|1:IP_CSUM (default:CRC)");
+static struct workqueue_struct *release_wq;
struct iser_global ig;
void
@@ -337,24 +338,6 @@ iscsi_iser_conn_create(struct iscsi_cls_session *cls_session, uint32_t conn_idx)
return cls_conn;
}
-static void
-iscsi_iser_conn_destroy(struct iscsi_cls_conn *cls_conn)
-{
- struct iscsi_conn *conn = cls_conn->dd_data;
- struct iser_conn *ib_conn = conn->dd_data;
-
- iscsi_conn_teardown(cls_conn);
- /*
- * Userspace will normally call the stop callback and
- * already have freed the ib_conn, but if it goofed up then
- * we free it here.
- */
- if (ib_conn) {
- ib_conn->iscsi_conn = NULL;
- iser_conn_put(ib_conn, 1); /* deref iscsi/ib conn unbinding */
- }
-}
-
static int
iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session,
struct iscsi_cls_conn *cls_conn, uint64_t transport_eph,
@@ -392,29 +375,39 @@ iscsi_iser_conn_bind(struct iscsi_cls_session *cls_session,
conn->dd_data = ib_conn;
ib_conn->iscsi_conn = conn;
- iser_conn_get(ib_conn); /* ref iscsi/ib conn binding */
return 0;
}
+static int
+iscsi_iser_conn_start(struct iscsi_cls_conn *cls_conn)
+{
+ struct iscsi_conn *iscsi_conn;
+ struct iser_conn *ib_conn;
+
+ iscsi_conn = cls_conn->dd_data;
+ ib_conn = iscsi_conn->dd_data;
+ reinit_completion(&ib_conn->stop_completion);
+
+ return iscsi_conn_start(cls_conn);
+}
+
static void
iscsi_iser_conn_stop(struct iscsi_cls_conn *cls_conn, int flag)
{
struct iscsi_conn *conn = cls_conn->dd_data;
struct iser_conn *ib_conn = conn->dd_data;
+ iser_dbg("stopping iscsi_conn: %p, ib_conn: %p\n", conn, ib_conn);
+ iscsi_conn_stop(cls_conn, flag);
+
/*
* Userspace may have goofed up and not bound the connection or
* might have only partially setup the connection.
*/
if (ib_conn) {
- iscsi_conn_stop(cls_conn, flag);
- /*
- * There is no unbind event so the stop callback
- * must release the ref from the bind.
- */
- iser_conn_put(ib_conn, 1); /* deref iscsi/ib conn unbinding */
+ conn->dd_data = NULL;
+ complete(&ib_conn->stop_completion);
}
- conn->dd_data = NULL;
}
static void iscsi_iser_session_destroy(struct iscsi_cls_session *cls_session)
@@ -515,28 +508,28 @@ iscsi_iser_set_param(struct iscsi_cls_conn *cls_conn,
case ISCSI_PARAM_HDRDGST_EN:
sscanf(buf, "%d", &value);
if (value) {
- iser_err("DataDigest wasn't negotiated to None");
+ iser_err("DataDigest wasn't negotiated to None\n");
return -EPROTO;
}
break;
case ISCSI_PARAM_DATADGST_EN:
sscanf(buf, "%d", &value);
if (value) {
- iser_err("DataDigest wasn't negotiated to None");
+ iser_err("DataDigest wasn't negotiated to None\n");
return -EPROTO;
}
break;
case ISCSI_PARAM_IFMARKER_EN:
sscanf(buf, "%d", &value);
if (value) {
- iser_err("IFMarker wasn't negotiated to No");
+ iser_err("IFMarker wasn't negotiated to No\n");
return -EPROTO;
}
break;
case ISCSI_PARAM_OFMARKER_EN:
sscanf(buf, "%d", &value);
if (value) {
- iser_err("OFMarker wasn't negotiated to No");
+ iser_err("OFMarker wasn't negotiated to No\n");
return -EPROTO;
}
break;
@@ -652,19 +645,20 @@ iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep)
struct iser_conn *ib_conn;
ib_conn = ep->dd_data;
- if (ib_conn->iscsi_conn)
- /*
- * Must suspend xmit path if the ep is bound to the
- * iscsi_conn, so we know we are not accessing the ib_conn
- * when we free it.
- *
- * This may not be bound if the ep poll failed.
- */
- iscsi_suspend_tx(ib_conn->iscsi_conn);
-
-
- iser_info("ib conn %p state %d\n", ib_conn, ib_conn->state);
+ iser_info("ep %p ib conn %p state %d\n", ep, ib_conn, ib_conn->state);
iser_conn_terminate(ib_conn);
+
+ /*
+ * if iser_conn and iscsi_conn are bound, we must wait iscsi_conn_stop
+ * call and ISER_CONN_DOWN state before freeing the iser resources.
+ * otherwise we are safe to free resources immediately.
+ */
+ if (ib_conn->iscsi_conn) {
+ INIT_WORK(&ib_conn->release_work, iser_release_work);
+ queue_work(release_wq, &ib_conn->release_work);
+ } else {
+ iser_conn_release(ib_conn);
+ }
}
static umode_t iser_attr_is_visible(int param_type, int param)
@@ -748,13 +742,13 @@ static struct iscsi_transport iscsi_iser_transport = {
/* connection management */
.create_conn = iscsi_iser_conn_create,
.bind_conn = iscsi_iser_conn_bind,
- .destroy_conn = iscsi_iser_conn_destroy,
+ .destroy_conn = iscsi_conn_teardown,
.attr_is_visible = iser_attr_is_visible,
.set_param = iscsi_iser_set_param,
.get_conn_param = iscsi_conn_get_param,
.get_ep_param = iscsi_iser_get_ep_param,
.get_session_param = iscsi_session_get_param,
- .start_conn = iscsi_conn_start,
+ .start_conn = iscsi_iser_conn_start,
.stop_conn = iscsi_iser_conn_stop,
/* iscsi host params */
.get_host_param = iscsi_host_get_param,
@@ -801,6 +795,12 @@ static int __init iser_init(void)
mutex_init(&ig.connlist_mutex);
INIT_LIST_HEAD(&ig.connlist);
+ release_wq = alloc_workqueue("release workqueue", 0, 0);
+ if (!release_wq) {
+ iser_err("failed to allocate release workqueue\n");
+ return -ENOMEM;
+ }
+
iscsi_iser_scsi_transport = iscsi_register_transport(
&iscsi_iser_transport);
if (!iscsi_iser_scsi_transport) {
@@ -819,7 +819,24 @@ register_transport_failure:
static void __exit iser_exit(void)
{
+ struct iser_conn *ib_conn, *n;
+ int connlist_empty;
+
iser_dbg("Removing iSER datamover...\n");
+ destroy_workqueue(release_wq);
+
+ mutex_lock(&ig.connlist_mutex);
+ connlist_empty = list_empty(&ig.connlist);
+ mutex_unlock(&ig.connlist_mutex);
+
+ if (!connlist_empty) {
+ iser_err("Error cleanup stage completed but we still have iser "
+ "connections, destroying them anyway.\n");
+ list_for_each_entry_safe(ib_conn, n, &ig.connlist, conn_list) {
+ iser_conn_release(ib_conn);
+ }
+ }
+
iscsi_unregister_transport(&iscsi_iser_transport);
kmem_cache_destroy(ig.desc_cache);
}
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h
index 324129f80d4..97cd385bf7f 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.h
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.h
@@ -69,7 +69,7 @@
#define DRV_NAME "iser"
#define PFX DRV_NAME ": "
-#define DRV_VER "1.3"
+#define DRV_VER "1.4"
#define iser_dbg(fmt, arg...) \
do { \
@@ -333,6 +333,8 @@ struct iser_conn {
int post_recv_buf_count; /* posted rx count */
atomic_t post_send_buf_count; /* posted tx count */
char name[ISER_OBJECT_NAME_SIZE];
+ struct work_struct release_work;
+ struct completion stop_completion;
struct list_head conn_list; /* entry in ig conn list */
char *login_buf;
@@ -417,12 +419,12 @@ void iscsi_iser_recv(struct iscsi_conn *conn,
void iser_conn_init(struct iser_conn *ib_conn);
-void iser_conn_get(struct iser_conn *ib_conn);
-
-int iser_conn_put(struct iser_conn *ib_conn, int destroy_cma_id_allowed);
+void iser_conn_release(struct iser_conn *ib_conn);
void iser_conn_terminate(struct iser_conn *ib_conn);
+void iser_release_work(struct work_struct *work);
+
void iser_rcv_completion(struct iser_rx_desc *desc,
unsigned long dto_xfer_len,
struct iser_conn *ib_conn);
diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index 2e2d903db83..8d44a406063 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -41,11 +41,11 @@
#include "iscsi_iser.h"
/* Register user buffer memory and initialize passive rdma
- * dto descriptor. Total data size is stored in
- * iser_task->data[ISER_DIR_IN].data_len
+ * dto descriptor. Data size is stored in
+ * task->data[ISER_DIR_IN].data_len, Protection size
+ * os stored in task->prot[ISER_DIR_IN].data_len
*/
-static int iser_prepare_read_cmd(struct iscsi_task *task,
- unsigned int edtl)
+static int iser_prepare_read_cmd(struct iscsi_task *task)
{
struct iscsi_iser_task *iser_task = task->dd_data;
@@ -73,14 +73,6 @@ static int iser_prepare_read_cmd(struct iscsi_task *task,
return err;
}
- if (edtl > iser_task->data[ISER_DIR_IN].data_len) {
- iser_err("Total data length: %ld, less than EDTL: "
- "%d, in READ cmd BHS itt: %d, conn: 0x%p\n",
- iser_task->data[ISER_DIR_IN].data_len, edtl,
- task->itt, iser_task->ib_conn);
- return -EINVAL;
- }
-
err = device->iser_reg_rdma_mem(iser_task, ISER_DIR_IN);
if (err) {
iser_err("Failed to set up Data-IN RDMA\n");
@@ -100,8 +92,9 @@ static int iser_prepare_read_cmd(struct iscsi_task *task,
}
/* Register user buffer memory and initialize passive rdma
- * dto descriptor. Total data size is stored in
- * task->data[ISER_DIR_OUT].data_len
+ * dto descriptor. Data size is stored in
+ * task->data[ISER_DIR_OUT].data_len, Protection size
+ * is stored at task->prot[ISER_DIR_OUT].data_len
*/
static int
iser_prepare_write_cmd(struct iscsi_task *task,
@@ -135,14 +128,6 @@ iser_prepare_write_cmd(struct iscsi_task *task,
return err;
}
- if (edtl > iser_task->data[ISER_DIR_OUT].data_len) {
- iser_err("Total data length: %ld, less than EDTL: %d, "
- "in WRITE cmd BHS itt: %d, conn: 0x%p\n",
- iser_task->data[ISER_DIR_OUT].data_len,
- edtl, task->itt, task->conn);
- return -EINVAL;
- }
-
err = device->iser_reg_rdma_mem(iser_task, ISER_DIR_OUT);
if (err != 0) {
iser_err("Failed to register write cmd RDMA mem\n");
@@ -417,11 +402,12 @@ int iser_send_command(struct iscsi_conn *conn,
if (scsi_prot_sg_count(sc)) {
prot_buf->buf = scsi_prot_sglist(sc);
prot_buf->size = scsi_prot_sg_count(sc);
- prot_buf->data_len = sc->prot_sdb->length;
+ prot_buf->data_len = data_buf->data_len >>
+ ilog2(sc->device->sector_size) * 8;
}
if (hdr->flags & ISCSI_FLAG_CMD_READ) {
- err = iser_prepare_read_cmd(task, edtl);
+ err = iser_prepare_read_cmd(task);
if (err)
goto send_command_error;
}
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 32849f2becd..ea01075f9f9 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -581,14 +581,30 @@ static int iser_conn_state_comp_exch(struct iser_conn *ib_conn,
return ret;
}
+void iser_release_work(struct work_struct *work)
+{
+ struct iser_conn *ib_conn;
+
+ ib_conn = container_of(work, struct iser_conn, release_work);
+
+ /* wait for .conn_stop callback */
+ wait_for_completion(&ib_conn->stop_completion);
+
+ /* wait for the qp`s post send and post receive buffers to empty */
+ wait_event_interruptible(ib_conn->wait,
+ ib_conn->state == ISER_CONN_DOWN);
+
+ iser_conn_release(ib_conn);
+}
+
/**
* Frees all conn objects and deallocs conn descriptor
*/
-static void iser_conn_release(struct iser_conn *ib_conn, int can_destroy_id)
+void iser_conn_release(struct iser_conn *ib_conn)
{
struct iser_device *device = ib_conn->device;
- BUG_ON(ib_conn->state != ISER_CONN_DOWN);
+ BUG_ON(ib_conn->state == ISER_CONN_UP);
mutex_lock(&ig.connlist_mutex);
list_del(&ib_conn->conn_list);
@@ -600,27 +616,13 @@ static void iser_conn_release(struct iser_conn *ib_conn, int can_destroy_id)
if (device != NULL)
iser_device_try_release(device);
/* if cma handler context, the caller actually destroy the id */
- if (ib_conn->cma_id != NULL && can_destroy_id) {
+ if (ib_conn->cma_id != NULL) {
rdma_destroy_id(ib_conn->cma_id);
ib_conn->cma_id = NULL;
}
iscsi_destroy_endpoint(ib_conn->ep);
}
-void iser_conn_get(struct iser_conn *ib_conn)
-{
- atomic_inc(&ib_conn->refcount);
-}
-
-int iser_conn_put(struct iser_conn *ib_conn, int can_destroy_id)
-{
- if (atomic_dec_and_test(&ib_conn->refcount)) {
- iser_conn_release(ib_conn, can_destroy_id);
- return 1;
- }
- return 0;
-}
-
/**
* triggers start of the disconnect procedures and wait for them to be done
*/
@@ -638,24 +640,19 @@ void iser_conn_terminate(struct iser_conn *ib_conn)
if (err)
iser_err("Failed to disconnect, conn: 0x%p err %d\n",
ib_conn,err);
-
- wait_event_interruptible(ib_conn->wait,
- ib_conn->state == ISER_CONN_DOWN);
-
- iser_conn_put(ib_conn, 1); /* deref ib conn deallocate */
}
-static int iser_connect_error(struct rdma_cm_id *cma_id)
+static void iser_connect_error(struct rdma_cm_id *cma_id)
{
struct iser_conn *ib_conn;
+
ib_conn = (struct iser_conn *)cma_id->context;
ib_conn->state = ISER_CONN_DOWN;
wake_up_interruptible(&ib_conn->wait);
- return iser_conn_put(ib_conn, 0); /* deref ib conn's cma id */
}
-static int iser_addr_handler(struct rdma_cm_id *cma_id)
+static void iser_addr_handler(struct rdma_cm_id *cma_id)
{
struct iser_device *device;
struct iser_conn *ib_conn;
@@ -664,7 +661,8 @@ static int iser_addr_handler(struct rdma_cm_id *cma_id)
device = iser_device_find_by_ib_device(cma_id);
if (!device) {
iser_err("device lookup/creation failed\n");
- return iser_connect_error(cma_id);
+ iser_connect_error(cma_id);
+ return;
}
ib_conn = (struct iser_conn *)cma_id->context;
@@ -686,13 +684,12 @@ static int iser_addr_handler(struct rdma_cm_id *cma_id)
ret = rdma_resolve_route(cma_id, 1000);
if (ret) {
iser_err("resolve route failed: %d\n", ret);
- return iser_connect_error(cma_id);
+ iser_connect_error(cma_id);
+ return;
}
-
- return 0;
}
-static int iser_route_handler(struct rdma_cm_id *cma_id)
+static void iser_route_handler(struct rdma_cm_id *cma_id)
{
struct rdma_conn_param conn_param;
int ret;
@@ -720,9 +717,9 @@ static int iser_route_handler(struct rdma_cm_id *cma_id)
goto failure;
}
- return 0;
+ return;
failure:
- return iser_connect_error(cma_id);
+ iser_connect_error(cma_id);
}
static void iser_connected_handler(struct rdma_cm_id *cma_id)
@@ -735,14 +732,13 @@ static void iser_connected_handler(struct rdma_cm_id *cma_id)
iser_info("remote qpn:%x my qpn:%x\n", attr.dest_qp_num, cma_id->qp->qp_num);
ib_conn = (struct iser_conn *)cma_id->context;
- ib_conn->state = ISER_CONN_UP;
- wake_up_interruptible(&ib_conn->wait);
+ if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_PENDING, ISER_CONN_UP))
+ wake_up_interruptible(&ib_conn->wait);
}
-static int iser_disconnected_handler(struct rdma_cm_id *cma_id)
+static void iser_disconnected_handler(struct rdma_cm_id *cma_id)
{
struct iser_conn *ib_conn;
- int ret;
ib_conn = (struct iser_conn *)cma_id->context;
@@ -762,24 +758,19 @@ static int iser_disconnected_handler(struct rdma_cm_id *cma_id)
ib_conn->state = ISER_CONN_DOWN;
wake_up_interruptible(&ib_conn->wait);
}
-
- ret = iser_conn_put(ib_conn, 0); /* deref ib conn's cma id */
- return ret;
}
static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
{
- int ret = 0;
-
iser_info("event %d status %d conn %p id %p\n",
event->event, event->status, cma_id->context, cma_id);
switch (event->event) {
case RDMA_CM_EVENT_ADDR_RESOLVED:
- ret = iser_addr_handler(cma_id);
+ iser_addr_handler(cma_id);
break;
case RDMA_CM_EVENT_ROUTE_RESOLVED:
- ret = iser_route_handler(cma_id);
+ iser_route_handler(cma_id);
break;
case RDMA_CM_EVENT_ESTABLISHED:
iser_connected_handler(cma_id);
@@ -789,18 +780,18 @@ static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *eve
case RDMA_CM_EVENT_CONNECT_ERROR:
case RDMA_CM_EVENT_UNREACHABLE:
case RDMA_CM_EVENT_REJECTED:
- ret = iser_connect_error(cma_id);
+ iser_connect_error(cma_id);
break;
case RDMA_CM_EVENT_DISCONNECTED:
case RDMA_CM_EVENT_DEVICE_REMOVAL:
case RDMA_CM_EVENT_ADDR_CHANGE:
- ret = iser_disconnected_handler(cma_id);
+ iser_disconnected_handler(cma_id);
break;
default:
iser_err("Unexpected RDMA CM event (%d)\n", event->event);
break;
}
- return ret;
+ return 0;
}
void iser_conn_init(struct iser_conn *ib_conn)
@@ -809,7 +800,7 @@ void iser_conn_init(struct iser_conn *ib_conn)
init_waitqueue_head(&ib_conn->wait);
ib_conn->post_recv_buf_count = 0;
atomic_set(&ib_conn->post_send_buf_count, 0);
- atomic_set(&ib_conn->refcount, 1); /* ref ib conn allocation */
+ init_completion(&ib_conn->stop_completion);
INIT_LIST_HEAD(&ib_conn->conn_list);
spin_lock_init(&ib_conn->lock);
}
@@ -837,7 +828,6 @@ int iser_connect(struct iser_conn *ib_conn,
ib_conn->state = ISER_CONN_PENDING;
- iser_conn_get(ib_conn); /* ref ib conn's cma id */
ib_conn->cma_id = rdma_create_id(iser_cma_handler,
(void *)ib_conn,
RDMA_PS_TCP, IB_QPT_RC);
@@ -874,9 +864,8 @@ id_failure:
ib_conn->cma_id = NULL;
addr_failure:
ib_conn->state = ISER_CONN_DOWN;
- iser_conn_put(ib_conn, 1); /* deref ib conn's cma id */
connect_failure:
- iser_conn_put(ib_conn, 1); /* deref ib conn deallocate */
+ iser_conn_release(ib_conn);
return err;
}
diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
index 8ee228e9ab5..d4c7928a0f3 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -28,6 +28,7 @@
#include <target/target_core_base.h>
#include <target/target_core_fabric.h>
#include <target/iscsi/iscsi_transport.h>
+#include <linux/semaphore.h>
#include "isert_proto.h"
#include "ib_isert.h"
@@ -51,6 +52,8 @@ isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
static int
isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
struct isert_rdma_wr *wr);
+static int
+isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd);
static void
isert_qp_event_callback(struct ib_event *e, void *context)
@@ -87,7 +90,8 @@ isert_query_device(struct ib_device *ib_dev, struct ib_device_attr *devattr)
}
static int
-isert_conn_setup_qp(struct isert_conn *isert_conn, struct rdma_cm_id *cma_id)
+isert_conn_setup_qp(struct isert_conn *isert_conn, struct rdma_cm_id *cma_id,
+ u8 protection)
{
struct isert_device *device = isert_conn->conn_device;
struct ib_qp_init_attr attr;
@@ -119,6 +123,8 @@ isert_conn_setup_qp(struct isert_conn *isert_conn, struct rdma_cm_id *cma_id)
attr.cap.max_recv_sge = 1;
attr.sq_sig_type = IB_SIGNAL_REQ_WR;
attr.qp_type = IB_QPT_RC;
+ if (protection)
+ attr.create_flags |= IB_QP_CREATE_SIGNATURE_EN;
pr_debug("isert_conn_setup_qp cma_id->device: %p\n",
cma_id->device);
@@ -226,7 +232,8 @@ isert_create_device_ib_res(struct isert_device *device)
return ret;
/* asign function handlers */
- if (dev_attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS) {
+ if (dev_attr->device_cap_flags & IB_DEVICE_MEM_MGT_EXTENSIONS &&
+ dev_attr->device_cap_flags & IB_DEVICE_SIGNATURE_HANDOVER) {
device->use_fastreg = 1;
device->reg_rdma_mem = isert_reg_rdma;
device->unreg_rdma_mem = isert_unreg_rdma;
@@ -236,13 +243,18 @@ isert_create_device_ib_res(struct isert_device *device)
device->unreg_rdma_mem = isert_unmap_cmd;
}
+ /* Check signature cap */
+ device->pi_capable = dev_attr->device_cap_flags &
+ IB_DEVICE_SIGNATURE_HANDOVER ? true : false;
+
device->cqs_used = min_t(int, num_online_cpus(),
device->ib_device->num_comp_vectors);
device->cqs_used = min(ISERT_MAX_CQ, device->cqs_used);
pr_debug("Using %d CQs, device %s supports %d vectors support "
- "Fast registration %d\n",
+ "Fast registration %d pi_capable %d\n",
device->cqs_used, device->ib_device->name,
- device->ib_device->num_comp_vectors, device->use_fastreg);
+ device->ib_device->num_comp_vectors, device->use_fastreg,
+ device->pi_capable);
device->cq_desc = kzalloc(sizeof(struct isert_cq_desc) *
device->cqs_used, GFP_KERNEL);
if (!device->cq_desc) {
@@ -395,6 +407,12 @@ isert_conn_free_fastreg_pool(struct isert_conn *isert_conn)
list_del(&fr_desc->list);
ib_free_fast_reg_page_list(fr_desc->data_frpl);
ib_dereg_mr(fr_desc->data_mr);
+ if (fr_desc->pi_ctx) {
+ ib_free_fast_reg_page_list(fr_desc->pi_ctx->prot_frpl);
+ ib_dereg_mr(fr_desc->pi_ctx->prot_mr);
+ ib_destroy_mr(fr_desc->pi_ctx->sig_mr);
+ kfree(fr_desc->pi_ctx);
+ }
kfree(fr_desc);
++i;
}
@@ -406,8 +424,10 @@ isert_conn_free_fastreg_pool(struct isert_conn *isert_conn)
static int
isert_create_fr_desc(struct ib_device *ib_device, struct ib_pd *pd,
- struct fast_reg_descriptor *fr_desc)
+ struct fast_reg_descriptor *fr_desc, u8 protection)
{
+ int ret;
+
fr_desc->data_frpl = ib_alloc_fast_reg_page_list(ib_device,
ISCSI_ISER_SG_TABLESIZE);
if (IS_ERR(fr_desc->data_frpl)) {
@@ -420,27 +440,88 @@ isert_create_fr_desc(struct ib_device *ib_device, struct ib_pd *pd,
if (IS_ERR(fr_desc->data_mr)) {
pr_err("Failed to allocate data frmr err=%ld\n",
PTR_ERR(fr_desc->data_mr));
- ib_free_fast_reg_page_list(fr_desc->data_frpl);
- return PTR_ERR(fr_desc->data_mr);
+ ret = PTR_ERR(fr_desc->data_mr);
+ goto err_data_frpl;
}
pr_debug("Create fr_desc %p page_list %p\n",
fr_desc, fr_desc->data_frpl->page_list);
+ fr_desc->ind |= ISERT_DATA_KEY_VALID;
+
+ if (protection) {
+ struct ib_mr_init_attr mr_init_attr = {0};
+ struct pi_context *pi_ctx;
+
+ fr_desc->pi_ctx = kzalloc(sizeof(*fr_desc->pi_ctx), GFP_KERNEL);
+ if (!fr_desc->pi_ctx) {
+ pr_err("Failed to allocate pi context\n");
+ ret = -ENOMEM;
+ goto err_data_mr;
+ }
+ pi_ctx = fr_desc->pi_ctx;
+
+ pi_ctx->prot_frpl = ib_alloc_fast_reg_page_list(ib_device,
+ ISCSI_ISER_SG_TABLESIZE);
+ if (IS_ERR(pi_ctx->prot_frpl)) {
+ pr_err("Failed to allocate prot frpl err=%ld\n",
+ PTR_ERR(pi_ctx->prot_frpl));
+ ret = PTR_ERR(pi_ctx->prot_frpl);
+ goto err_pi_ctx;
+ }
- fr_desc->valid = true;
+ pi_ctx->prot_mr = ib_alloc_fast_reg_mr(pd, ISCSI_ISER_SG_TABLESIZE);
+ if (IS_ERR(pi_ctx->prot_mr)) {
+ pr_err("Failed to allocate prot frmr err=%ld\n",
+ PTR_ERR(pi_ctx->prot_mr));
+ ret = PTR_ERR(pi_ctx->prot_mr);
+ goto err_prot_frpl;
+ }
+ fr_desc->ind |= ISERT_PROT_KEY_VALID;
+
+ mr_init_attr.max_reg_descriptors = 2;
+ mr_init_attr.flags |= IB_MR_SIGNATURE_EN;
+ pi_ctx->sig_mr = ib_create_mr(pd, &mr_init_attr);
+ if (IS_ERR(pi_ctx->sig_mr)) {
+ pr_err("Failed to allocate signature enabled mr err=%ld\n",
+ PTR_ERR(pi_ctx->sig_mr));
+ ret = PTR_ERR(pi_ctx->sig_mr);
+ goto err_prot_mr;
+ }
+ fr_desc->ind |= ISERT_SIG_KEY_VALID;
+ }
+ fr_desc->ind &= ~ISERT_PROTECTED;
return 0;
+err_prot_mr:
+ ib_dereg_mr(fr_desc->pi_ctx->prot_mr);
+err_prot_frpl:
+ ib_free_fast_reg_page_list(fr_desc->pi_ctx->prot_frpl);
+err_pi_ctx:
+ kfree(fr_desc->pi_ctx);
+err_data_mr:
+ ib_dereg_mr(fr_desc->data_mr);
+err_data_frpl:
+ ib_free_fast_reg_page_list(fr_desc->data_frpl);
+
+ return ret;
}
static int
-isert_conn_create_fastreg_pool(struct isert_conn *isert_conn)
+isert_conn_create_fastreg_pool(struct isert_conn *isert_conn, u8 pi_support)
{
struct fast_reg_descriptor *fr_desc;
struct isert_device *device = isert_conn->conn_device;
- int i, ret;
+ struct se_session *se_sess = isert_conn->conn->sess->se_sess;
+ struct se_node_acl *se_nacl = se_sess->se_node_acl;
+ int i, ret, tag_num;
+ /*
+ * Setup the number of FRMRs based upon the number of tags
+ * available to session in iscsi_target_locate_portal().
+ */
+ tag_num = max_t(u32, ISCSIT_MIN_TAGS, se_nacl->queue_depth);
+ tag_num = (tag_num * 2) + ISCSIT_EXTRA_TAGS;
- INIT_LIST_HEAD(&isert_conn->conn_fr_pool);
isert_conn->conn_fr_pool_size = 0;
- for (i = 0; i < ISCSI_DEF_XMIT_CMDS_MAX; i++) {
+ for (i = 0; i < tag_num; i++) {
fr_desc = kzalloc(sizeof(*fr_desc), GFP_KERNEL);
if (!fr_desc) {
pr_err("Failed to allocate fast_reg descriptor\n");
@@ -449,7 +530,8 @@ isert_conn_create_fastreg_pool(struct isert_conn *isert_conn)
}
ret = isert_create_fr_desc(device->ib_device,
- isert_conn->conn_pd, fr_desc);
+ isert_conn->conn_pd, fr_desc,
+ pi_support);
if (ret) {
pr_err("Failed to create fastreg descriptor err=%d\n",
ret);
@@ -480,6 +562,15 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
struct isert_device *device;
struct ib_device *ib_dev = cma_id->device;
int ret = 0;
+ u8 pi_support;
+
+ spin_lock_bh(&np->np_thread_lock);
+ if (!np->enabled) {
+ spin_unlock_bh(&np->np_thread_lock);
+ pr_debug("iscsi_np is not enabled, reject connect request\n");
+ return rdma_reject(cma_id, NULL, 0);
+ }
+ spin_unlock_bh(&np->np_thread_lock);
pr_debug("Entering isert_connect_request cma_id: %p, context: %p\n",
cma_id, cma_id->context);
@@ -498,6 +589,7 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
kref_get(&isert_conn->conn_kref);
mutex_init(&isert_conn->conn_mutex);
spin_lock_init(&isert_conn->conn_lock);
+ INIT_LIST_HEAD(&isert_conn->conn_fr_pool);
cma_id->context = isert_conn;
isert_conn->conn_cm_id = cma_id;
@@ -569,31 +661,27 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
goto out_mr;
}
- if (device->use_fastreg) {
- ret = isert_conn_create_fastreg_pool(isert_conn);
- if (ret) {
- pr_err("Conn: %p failed to create fastreg pool\n",
- isert_conn);
- goto out_fastreg;
- }
+ pi_support = np->tpg_np->tpg->tpg_attrib.t10_pi;
+ if (pi_support && !device->pi_capable) {
+ pr_err("Protection information requested but not supported, "
+ "rejecting connect request\n");
+ ret = rdma_reject(cma_id, NULL, 0);
+ goto out_mr;
}
- ret = isert_conn_setup_qp(isert_conn, cma_id);
+ ret = isert_conn_setup_qp(isert_conn, cma_id, pi_support);
if (ret)
goto out_conn_dev;
mutex_lock(&isert_np->np_accept_mutex);
- list_add_tail(&isert_np->np_accept_list, &isert_conn->conn_accept_node);
+ list_add_tail(&isert_conn->conn_accept_node, &isert_np->np_accept_list);
mutex_unlock(&isert_np->np_accept_mutex);
- pr_debug("isert_connect_request() waking up np_accept_wq: %p\n", np);
- wake_up(&isert_np->np_accept_wq);
+ pr_debug("isert_connect_request() up np_sem np: %p\n", np);
+ up(&isert_np->np_sem);
return 0;
out_conn_dev:
- if (device->use_fastreg)
- isert_conn_free_fastreg_pool(isert_conn);
-out_fastreg:
ib_dereg_mr(isert_conn->conn_mr);
out_mr:
ib_dealloc_pd(isert_conn->conn_pd);
@@ -700,14 +788,12 @@ isert_disconnect_work(struct work_struct *work)
isert_put_conn(isert_conn);
return;
}
- if (!isert_conn->logout_posted) {
- pr_debug("Calling rdma_disconnect for !logout_posted from"
- " isert_disconnect_work\n");
+
+ if (isert_conn->disconnect) {
+ /* Send DREQ/DREP towards our initiator */
rdma_disconnect(isert_conn->conn_cm_id);
- mutex_unlock(&isert_conn->conn_mutex);
- iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
- goto wake_up;
}
+
mutex_unlock(&isert_conn->conn_mutex);
wake_up:
@@ -716,10 +802,11 @@ wake_up:
}
static void
-isert_disconnected_handler(struct rdma_cm_id *cma_id)
+isert_disconnected_handler(struct rdma_cm_id *cma_id, bool disconnect)
{
struct isert_conn *isert_conn = (struct isert_conn *)cma_id->context;
+ isert_conn->disconnect = disconnect;
INIT_WORK(&isert_conn->conn_logout_work, isert_disconnect_work);
schedule_work(&isert_conn->conn_logout_work);
}
@@ -728,29 +815,28 @@ static int
isert_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
{
int ret = 0;
+ bool disconnect = false;
pr_debug("isert_cma_handler: event %d status %d conn %p id %p\n",
event->event, event->status, cma_id->context, cma_id);
switch (event->event) {
case RDMA_CM_EVENT_CONNECT_REQUEST:
- pr_debug("RDMA_CM_EVENT_CONNECT_REQUEST: >>>>>>>>>>>>>>>\n");
ret = isert_connect_request(cma_id, event);
break;
case RDMA_CM_EVENT_ESTABLISHED:
- pr_debug("RDMA_CM_EVENT_ESTABLISHED >>>>>>>>>>>>>>\n");
isert_connected_handler(cma_id);
break;
- case RDMA_CM_EVENT_DISCONNECTED:
- pr_debug("RDMA_CM_EVENT_DISCONNECTED: >>>>>>>>>>>>>>\n");
- isert_disconnected_handler(cma_id);
- break;
- case RDMA_CM_EVENT_DEVICE_REMOVAL:
- case RDMA_CM_EVENT_ADDR_CHANGE:
+ case RDMA_CM_EVENT_ADDR_CHANGE: /* FALLTHRU */
+ case RDMA_CM_EVENT_DISCONNECTED: /* FALLTHRU */
+ case RDMA_CM_EVENT_DEVICE_REMOVAL: /* FALLTHRU */
+ disconnect = true;
+ case RDMA_CM_EVENT_TIMEWAIT_EXIT: /* FALLTHRU */
+ isert_disconnected_handler(cma_id, disconnect);
break;
case RDMA_CM_EVENT_CONNECT_ERROR:
default:
- pr_err("Unknown RDMA CMA event: %d\n", event->event);
+ pr_err("Unhandled RDMA CMA event: %d\n", event->event);
break;
}
@@ -967,6 +1053,20 @@ isert_put_login_tx(struct iscsi_conn *conn, struct iscsi_login *login,
}
if (!login->login_failed) {
if (login->login_complete) {
+ if (!conn->sess->sess_ops->SessionType &&
+ isert_conn->conn_device->use_fastreg) {
+ /* Normal Session and fastreg is used */
+ u8 pi_support = login->np->tpg_np->tpg->tpg_attrib.t10_pi;
+
+ ret = isert_conn_create_fastreg_pool(isert_conn,
+ pi_support);
+ if (ret) {
+ pr_err("Conn: %p failed to create"
+ " fastreg pool\n", isert_conn);
+ return ret;
+ }
+ }
+
ret = isert_alloc_rx_descriptors(isert_conn);
if (ret)
return ret;
@@ -1111,6 +1211,8 @@ sequence_cmd:
if (!rc && dump_payload == false && unsol_data)
iscsit_set_unsoliticed_dataout(cmd);
+ else if (dump_payload && imm_data)
+ target_put_sess_cmd(conn->sess->se_sess, &cmd->se_cmd);
return 0;
}
@@ -1392,19 +1494,60 @@ isert_rx_completion(struct iser_rx_desc *desc, struct isert_conn *isert_conn,
}
}
+static int
+isert_map_data_buf(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
+ struct scatterlist *sg, u32 nents, u32 length, u32 offset,
+ enum iser_ib_op_code op, struct isert_data_buf *data)
+{
+ struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
+
+ data->dma_dir = op == ISER_IB_RDMA_WRITE ?
+ DMA_TO_DEVICE : DMA_FROM_DEVICE;
+
+ data->len = length - offset;
+ data->offset = offset;
+ data->sg_off = data->offset / PAGE_SIZE;
+
+ data->sg = &sg[data->sg_off];
+ data->nents = min_t(unsigned int, nents - data->sg_off,
+ ISCSI_ISER_SG_TABLESIZE);
+ data->len = min_t(unsigned int, data->len, ISCSI_ISER_SG_TABLESIZE *
+ PAGE_SIZE);
+
+ data->dma_nents = ib_dma_map_sg(ib_dev, data->sg, data->nents,
+ data->dma_dir);
+ if (unlikely(!data->dma_nents)) {
+ pr_err("Cmd: unable to dma map SGs %p\n", sg);
+ return -EINVAL;
+ }
+
+ pr_debug("Mapped cmd: %p count: %u sg: %p sg_nents: %u rdma_len %d\n",
+ isert_cmd, data->dma_nents, data->sg, data->nents, data->len);
+
+ return 0;
+}
+
+static void
+isert_unmap_data_buf(struct isert_conn *isert_conn, struct isert_data_buf *data)
+{
+ struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
+
+ ib_dma_unmap_sg(ib_dev, data->sg, data->nents, data->dma_dir);
+ memset(data, 0, sizeof(*data));
+}
+
+
+
static void
isert_unmap_cmd(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
{
struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
- struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
pr_debug("isert_unmap_cmd: %p\n", isert_cmd);
- if (wr->sge) {
+
+ if (wr->data.sg) {
pr_debug("isert_unmap_cmd: %p unmap_sg op\n", isert_cmd);
- ib_dma_unmap_sg(ib_dev, wr->sge, wr->num_sge,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
- wr->sge = NULL;
+ isert_unmap_data_buf(isert_conn, &wr->data);
}
if (wr->send_wr) {
@@ -1424,7 +1567,6 @@ static void
isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
{
struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
- struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
LIST_HEAD(unmap_list);
pr_debug("unreg_fastreg_cmd: %p\n", isert_cmd);
@@ -1432,18 +1574,19 @@ isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
if (wr->fr_desc) {
pr_debug("unreg_fastreg_cmd: %p free fr_desc %p\n",
isert_cmd, wr->fr_desc);
+ if (wr->fr_desc->ind & ISERT_PROTECTED) {
+ isert_unmap_data_buf(isert_conn, &wr->prot);
+ wr->fr_desc->ind &= ~ISERT_PROTECTED;
+ }
spin_lock_bh(&isert_conn->conn_lock);
list_add_tail(&wr->fr_desc->list, &isert_conn->conn_fr_pool);
spin_unlock_bh(&isert_conn->conn_lock);
wr->fr_desc = NULL;
}
- if (wr->sge) {
+ if (wr->data.sg) {
pr_debug("unreg_fastreg_cmd: %p unmap_sg op\n", isert_cmd);
- ib_dma_unmap_sg(ib_dev, wr->sge, wr->num_sge,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
- wr->sge = NULL;
+ isert_unmap_data_buf(isert_conn, &wr->data);
}
wr->ib_sge = NULL;
@@ -1451,7 +1594,7 @@ isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
}
static void
-isert_put_cmd(struct isert_cmd *isert_cmd)
+isert_put_cmd(struct isert_cmd *isert_cmd, bool comp_err)
{
struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
struct isert_conn *isert_conn = isert_cmd->conn;
@@ -1467,8 +1610,21 @@ isert_put_cmd(struct isert_cmd *isert_cmd)
list_del_init(&cmd->i_conn_node);
spin_unlock_bh(&conn->cmd_lock);
- if (cmd->data_direction == DMA_TO_DEVICE)
+ if (cmd->data_direction == DMA_TO_DEVICE) {
iscsit_stop_dataout_timer(cmd);
+ /*
+ * Check for special case during comp_err where
+ * WRITE_PENDING has been handed off from core,
+ * but requires an extra target_put_sess_cmd()
+ * before transport_generic_free_cmd() below.
+ */
+ if (comp_err &&
+ cmd->se_cmd.t_state == TRANSPORT_WRITE_PENDING) {
+ struct se_cmd *se_cmd = &cmd->se_cmd;
+
+ target_put_sess_cmd(se_cmd->se_sess, se_cmd);
+ }
+ }
device->unreg_rdma_mem(isert_cmd, isert_conn);
transport_generic_free_cmd(&cmd->se_cmd, 0);
@@ -1523,7 +1679,7 @@ isert_unmap_tx_desc(struct iser_tx_desc *tx_desc, struct ib_device *ib_dev)
static void
isert_completion_put(struct iser_tx_desc *tx_desc, struct isert_cmd *isert_cmd,
- struct ib_device *ib_dev)
+ struct ib_device *ib_dev, bool comp_err)
{
if (isert_cmd->pdu_buf_dma != 0) {
pr_debug("Calling ib_dma_unmap_single for isert_cmd->pdu_buf_dma\n");
@@ -1533,7 +1689,77 @@ isert_completion_put(struct iser_tx_desc *tx_desc, struct isert_cmd *isert_cmd,
}
isert_unmap_tx_desc(tx_desc, ib_dev);
- isert_put_cmd(isert_cmd);
+ isert_put_cmd(isert_cmd, comp_err);
+}
+
+static int
+isert_check_pi_status(struct se_cmd *se_cmd, struct ib_mr *sig_mr)
+{
+ struct ib_mr_status mr_status;
+ int ret;
+
+ ret = ib_check_mr_status(sig_mr, IB_MR_CHECK_SIG_STATUS, &mr_status);
+ if (ret) {
+ pr_err("ib_check_mr_status failed, ret %d\n", ret);
+ goto fail_mr_status;
+ }
+
+ if (mr_status.fail_status & IB_MR_CHECK_SIG_STATUS) {
+ u64 sec_offset_err;
+ u32 block_size = se_cmd->se_dev->dev_attrib.block_size + 8;
+
+ switch (mr_status.sig_err.err_type) {
+ case IB_SIG_BAD_GUARD:
+ se_cmd->pi_err = TCM_LOGICAL_BLOCK_GUARD_CHECK_FAILED;
+ break;
+ case IB_SIG_BAD_REFTAG:
+ se_cmd->pi_err = TCM_LOGICAL_BLOCK_REF_TAG_CHECK_FAILED;
+ break;
+ case IB_SIG_BAD_APPTAG:
+ se_cmd->pi_err = TCM_LOGICAL_BLOCK_APP_TAG_CHECK_FAILED;
+ break;
+ }
+ sec_offset_err = mr_status.sig_err.sig_err_offset;
+ do_div(sec_offset_err, block_size);
+ se_cmd->bad_sector = sec_offset_err + se_cmd->t_task_lba;
+
+ pr_err("isert: PI error found type %d at sector 0x%llx "
+ "expected 0x%x vs actual 0x%x\n",
+ mr_status.sig_err.err_type,
+ (unsigned long long)se_cmd->bad_sector,
+ mr_status.sig_err.expected,
+ mr_status.sig_err.actual);
+ ret = 1;
+ }
+
+fail_mr_status:
+ return ret;
+}
+
+static void
+isert_completion_rdma_write(struct iser_tx_desc *tx_desc,
+ struct isert_cmd *isert_cmd)
+{
+ struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
+ struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
+ struct se_cmd *se_cmd = &cmd->se_cmd;
+ struct isert_conn *isert_conn = isert_cmd->conn;
+ struct isert_device *device = isert_conn->conn_device;
+ int ret = 0;
+
+ if (wr->fr_desc && wr->fr_desc->ind & ISERT_PROTECTED) {
+ ret = isert_check_pi_status(se_cmd,
+ wr->fr_desc->pi_ctx->sig_mr);
+ wr->fr_desc->ind &= ~ISERT_PROTECTED;
+ }
+
+ device->unreg_rdma_mem(isert_cmd, isert_conn);
+ wr->send_wr_num = 0;
+ if (ret)
+ transport_send_check_condition_and_sense(se_cmd,
+ se_cmd->pi_err, 0);
+ else
+ isert_put_response(isert_conn->conn, cmd);
}
static void
@@ -1545,10 +1771,17 @@ isert_completion_rdma_read(struct iser_tx_desc *tx_desc,
struct se_cmd *se_cmd = &cmd->se_cmd;
struct isert_conn *isert_conn = isert_cmd->conn;
struct isert_device *device = isert_conn->conn_device;
+ int ret = 0;
+
+ if (wr->fr_desc && wr->fr_desc->ind & ISERT_PROTECTED) {
+ ret = isert_check_pi_status(se_cmd,
+ wr->fr_desc->pi_ctx->sig_mr);
+ wr->fr_desc->ind &= ~ISERT_PROTECTED;
+ }
iscsit_stop_dataout_timer(cmd);
device->unreg_rdma_mem(isert_cmd, isert_conn);
- cmd->write_data_done = wr->cur_rdma_length;
+ cmd->write_data_done = wr->data.len;
wr->send_wr_num = 0;
pr_debug("Cmd: %p RDMA_READ comp calling execute_cmd\n", isert_cmd);
@@ -1557,7 +1790,11 @@ isert_completion_rdma_read(struct iser_tx_desc *tx_desc,
cmd->i_state = ISTATE_RECEIVED_LAST_DATAOUT;
spin_unlock_bh(&cmd->istate_lock);
- target_execute_cmd(se_cmd);
+ if (ret)
+ transport_send_check_condition_and_sense(se_cmd,
+ se_cmd->pi_err, 0);
+ else
+ target_execute_cmd(se_cmd);
}
static void
@@ -1577,28 +1814,25 @@ isert_do_control_comp(struct work_struct *work)
iscsit_tmr_post_handler(cmd, cmd->conn);
cmd->i_state = ISTATE_SENT_STATUS;
- isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev);
+ isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev, false);
break;
case ISTATE_SEND_REJECT:
pr_debug("Got isert_do_control_comp ISTATE_SEND_REJECT: >>>\n");
atomic_dec(&isert_conn->post_send_buf_count);
cmd->i_state = ISTATE_SENT_STATUS;
- isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev);
+ isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev, false);
break;
case ISTATE_SEND_LOGOUTRSP:
pr_debug("Calling iscsit_logout_post_handler >>>>>>>>>>>>>>\n");
- /*
- * Call atomic_dec(&isert_conn->post_send_buf_count)
- * from isert_wait_conn()
- */
- isert_conn->logout_posted = true;
+
+ atomic_dec(&isert_conn->post_send_buf_count);
iscsit_logout_post_handler(cmd, cmd->conn);
break;
case ISTATE_SEND_TEXTRSP:
atomic_dec(&isert_conn->post_send_buf_count);
cmd->i_state = ISTATE_SENT_STATUS;
- isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev);
+ isert_completion_put(&isert_cmd->tx_desc, isert_cmd, ib_dev, false);
break;
default:
pr_err("Unknown do_control_comp i_state %d\n", cmd->i_state);
@@ -1626,10 +1860,21 @@ isert_response_completion(struct iser_tx_desc *tx_desc,
queue_work(isert_comp_wq, &isert_cmd->comp_work);
return;
}
- atomic_sub(wr->send_wr_num + 1, &isert_conn->post_send_buf_count);
+
+ /**
+ * If send_wr_num is 0 this means that we got
+ * RDMA completion and we cleared it and we should
+ * simply decrement the response post. else the
+ * response is incorporated in send_wr_num, just
+ * sub it.
+ **/
+ if (wr->send_wr_num)
+ atomic_sub(wr->send_wr_num, &isert_conn->post_send_buf_count);
+ else
+ atomic_dec(&isert_conn->post_send_buf_count);
cmd->i_state = ISTATE_SENT_STATUS;
- isert_completion_put(tx_desc, isert_cmd, ib_dev);
+ isert_completion_put(tx_desc, isert_cmd, ib_dev, false);
}
static void
@@ -1658,8 +1903,9 @@ __isert_send_completion(struct iser_tx_desc *tx_desc,
isert_conn, ib_dev);
break;
case ISER_IB_RDMA_WRITE:
- pr_err("isert_send_completion: Got ISER_IB_RDMA_WRITE\n");
- dump_stack();
+ pr_debug("isert_send_completion: Got ISER_IB_RDMA_WRITE\n");
+ atomic_sub(wr->send_wr_num, &isert_conn->post_send_buf_count);
+ isert_completion_rdma_write(tx_desc, isert_cmd);
break;
case ISER_IB_RDMA_READ:
pr_debug("isert_send_completion: Got ISER_IB_RDMA_READ:\n");
@@ -1709,8 +1955,20 @@ isert_cq_drain_comp_llist(struct isert_conn *isert_conn, struct ib_device *ib_de
llnode = llist_next(llnode);
wr = &t->isert_cmd->rdma_wr;
- atomic_sub(wr->send_wr_num + 1, &isert_conn->post_send_buf_count);
- isert_completion_put(t, t->isert_cmd, ib_dev);
+ /**
+ * If send_wr_num is 0 this means that we got
+ * RDMA completion and we cleared it and we should
+ * simply decrement the response post. else the
+ * response is incorporated in send_wr_num, just
+ * sub it.
+ **/
+ if (wr->send_wr_num)
+ atomic_sub(wr->send_wr_num,
+ &isert_conn->post_send_buf_count);
+ else
+ atomic_dec(&isert_conn->post_send_buf_count);
+
+ isert_completion_put(t, t->isert_cmd, ib_dev, true);
}
}
@@ -1728,15 +1986,27 @@ isert_cq_tx_comp_err(struct iser_tx_desc *tx_desc, struct isert_conn *isert_conn
llnode = llist_next(llnode);
wr = &t->isert_cmd->rdma_wr;
- atomic_sub(wr->send_wr_num + 1, &isert_conn->post_send_buf_count);
- isert_completion_put(t, t->isert_cmd, ib_dev);
+ /**
+ * If send_wr_num is 0 this means that we got
+ * RDMA completion and we cleared it and we should
+ * simply decrement the response post. else the
+ * response is incorporated in send_wr_num, just
+ * sub it.
+ **/
+ if (wr->send_wr_num)
+ atomic_sub(wr->send_wr_num,
+ &isert_conn->post_send_buf_count);
+ else
+ atomic_dec(&isert_conn->post_send_buf_count);
+
+ isert_completion_put(t, t->isert_cmd, ib_dev, true);
}
tx_desc->comp_llnode_batch = NULL;
if (!isert_cmd)
isert_unmap_tx_desc(tx_desc, ib_dev);
else
- isert_completion_put(tx_desc, isert_cmd, ib_dev);
+ isert_completion_put(tx_desc, isert_cmd, ib_dev, true);
}
static void
@@ -1762,6 +2032,8 @@ isert_cq_rx_comp_err(struct isert_conn *isert_conn)
isert_conn->state = ISER_CONN_DOWN;
mutex_unlock(&isert_conn->conn_mutex);
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+
complete(&isert_conn->conn_wait_comp_err);
}
@@ -1918,6 +2190,36 @@ isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
return isert_post_response(isert_conn, isert_cmd);
}
+static void
+isert_aborted_task(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
+{
+ struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
+ struct isert_conn *isert_conn = (struct isert_conn *)conn->context;
+ struct isert_device *device = isert_conn->conn_device;
+
+ spin_lock_bh(&conn->cmd_lock);
+ if (!list_empty(&cmd->i_conn_node))
+ list_del_init(&cmd->i_conn_node);
+ spin_unlock_bh(&conn->cmd_lock);
+
+ if (cmd->data_direction == DMA_TO_DEVICE)
+ iscsit_stop_dataout_timer(cmd);
+
+ device->unreg_rdma_mem(isert_cmd, isert_conn);
+}
+
+static enum target_prot_op
+isert_get_sup_prot_ops(struct iscsi_conn *conn)
+{
+ struct isert_conn *isert_conn = (struct isert_conn *)conn->context;
+ struct isert_device *device = isert_conn->conn_device;
+
+ if (device->pi_capable)
+ return TARGET_PROT_ALL;
+
+ return TARGET_PROT_NORMAL;
+}
+
static int
isert_put_nopin(struct iscsi_cmd *cmd, struct iscsi_conn *conn,
bool nopout_response)
@@ -2018,7 +2320,7 @@ isert_put_text_rsp(struct iscsi_cmd *cmd, struct iscsi_conn *conn)
int rc;
isert_create_send_desc(isert_conn, isert_cmd, &isert_cmd->tx_desc);
- rc = iscsit_build_text_rsp(cmd, conn, hdr);
+ rc = iscsit_build_text_rsp(cmd, conn, hdr, ISCSI_INFINIBAND);
if (rc < 0)
return rc;
@@ -2099,54 +2401,39 @@ isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
struct se_cmd *se_cmd = &cmd->se_cmd;
struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
struct isert_conn *isert_conn = (struct isert_conn *)conn->context;
- struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
+ struct isert_data_buf *data = &wr->data;
struct ib_send_wr *send_wr;
struct ib_sge *ib_sge;
- struct scatterlist *sg_start;
- u32 sg_off = 0, sg_nents;
- u32 offset = 0, data_len, data_left, rdma_write_max, va_offset = 0;
- int ret = 0, count, i, ib_sge_cnt;
+ u32 offset, data_len, data_left, rdma_write_max, va_offset = 0;
+ int ret = 0, i, ib_sge_cnt;
- if (wr->iser_ib_op == ISER_IB_RDMA_WRITE) {
- data_left = se_cmd->data_length;
- } else {
- sg_off = cmd->write_data_done / PAGE_SIZE;
- data_left = se_cmd->data_length - cmd->write_data_done;
- offset = cmd->write_data_done;
- isert_cmd->tx_desc.isert_cmd = isert_cmd;
- }
+ isert_cmd->tx_desc.isert_cmd = isert_cmd;
- sg_start = &cmd->se_cmd.t_data_sg[sg_off];
- sg_nents = se_cmd->t_data_nents - sg_off;
+ offset = wr->iser_ib_op == ISER_IB_RDMA_READ ? cmd->write_data_done : 0;
+ ret = isert_map_data_buf(isert_conn, isert_cmd, se_cmd->t_data_sg,
+ se_cmd->t_data_nents, se_cmd->data_length,
+ offset, wr->iser_ib_op, &wr->data);
+ if (ret)
+ return ret;
- count = ib_dma_map_sg(ib_dev, sg_start, sg_nents,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
- if (unlikely(!count)) {
- pr_err("Cmd: %p unrable to map SGs\n", isert_cmd);
- return -EINVAL;
- }
- wr->sge = sg_start;
- wr->num_sge = sg_nents;
- wr->cur_rdma_length = data_left;
- pr_debug("Mapped cmd: %p count: %u sg: %p sg_nents: %u rdma_len %d\n",
- isert_cmd, count, sg_start, sg_nents, data_left);
+ data_left = data->len;
+ offset = data->offset;
- ib_sge = kzalloc(sizeof(struct ib_sge) * sg_nents, GFP_KERNEL);
+ ib_sge = kzalloc(sizeof(struct ib_sge) * data->nents, GFP_KERNEL);
if (!ib_sge) {
pr_warn("Unable to allocate ib_sge\n");
ret = -ENOMEM;
- goto unmap_sg;
+ goto unmap_cmd;
}
wr->ib_sge = ib_sge;
- wr->send_wr_num = DIV_ROUND_UP(sg_nents, isert_conn->max_sge);
+ wr->send_wr_num = DIV_ROUND_UP(data->nents, isert_conn->max_sge);
wr->send_wr = kzalloc(sizeof(struct ib_send_wr) * wr->send_wr_num,
GFP_KERNEL);
if (!wr->send_wr) {
pr_debug("Unable to allocate wr->send_wr\n");
ret = -ENOMEM;
- goto unmap_sg;
+ goto unmap_cmd;
}
wr->isert_cmd = isert_cmd;
@@ -2185,10 +2472,9 @@ isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
}
return 0;
-unmap_sg:
- ib_dma_unmap_sg(ib_dev, sg_start, sg_nents,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
+unmap_cmd:
+ isert_unmap_data_buf(isert_conn, data);
+
return ret;
}
@@ -2232,49 +2518,70 @@ isert_map_fr_pagelist(struct ib_device *ib_dev,
}
static int
-isert_fast_reg_mr(struct fast_reg_descriptor *fr_desc,
- struct isert_conn *isert_conn, struct scatterlist *sg_start,
- struct ib_sge *ib_sge, u32 sg_nents, u32 offset,
- unsigned int data_len)
+isert_fast_reg_mr(struct isert_conn *isert_conn,
+ struct fast_reg_descriptor *fr_desc,
+ struct isert_data_buf *mem,
+ enum isert_indicator ind,
+ struct ib_sge *sge)
{
struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
+ struct ib_mr *mr;
+ struct ib_fast_reg_page_list *frpl;
struct ib_send_wr fr_wr, inv_wr;
struct ib_send_wr *bad_wr, *wr = NULL;
int ret, pagelist_len;
u32 page_off;
u8 key;
- sg_nents = min_t(unsigned int, sg_nents, ISCSI_ISER_SG_TABLESIZE);
- page_off = offset % PAGE_SIZE;
+ if (mem->dma_nents == 1) {
+ sge->lkey = isert_conn->conn_mr->lkey;
+ sge->addr = ib_sg_dma_address(ib_dev, &mem->sg[0]);
+ sge->length = ib_sg_dma_len(ib_dev, &mem->sg[0]);
+ pr_debug("%s:%d sge: addr: 0x%llx length: %u lkey: %x\n",
+ __func__, __LINE__, sge->addr, sge->length,
+ sge->lkey);
+ return 0;
+ }
+
+ if (ind == ISERT_DATA_KEY_VALID) {
+ /* Registering data buffer */
+ mr = fr_desc->data_mr;
+ frpl = fr_desc->data_frpl;
+ } else {
+ /* Registering protection buffer */
+ mr = fr_desc->pi_ctx->prot_mr;
+ frpl = fr_desc->pi_ctx->prot_frpl;
+ }
+
+ page_off = mem->offset % PAGE_SIZE;
pr_debug("Use fr_desc %p sg_nents %d offset %u\n",
- fr_desc, sg_nents, offset);
+ fr_desc, mem->nents, mem->offset);
- pagelist_len = isert_map_fr_pagelist(ib_dev, sg_start, sg_nents,
- &fr_desc->data_frpl->page_list[0]);
+ pagelist_len = isert_map_fr_pagelist(ib_dev, mem->sg, mem->nents,
+ &frpl->page_list[0]);
- if (!fr_desc->valid) {
+ if (!(fr_desc->ind & ISERT_DATA_KEY_VALID)) {
memset(&inv_wr, 0, sizeof(inv_wr));
inv_wr.wr_id = ISER_FASTREG_LI_WRID;
inv_wr.opcode = IB_WR_LOCAL_INV;
- inv_wr.ex.invalidate_rkey = fr_desc->data_mr->rkey;
+ inv_wr.ex.invalidate_rkey = mr->rkey;
wr = &inv_wr;
/* Bump the key */
- key = (u8)(fr_desc->data_mr->rkey & 0x000000FF);
- ib_update_fast_reg_key(fr_desc->data_mr, ++key);
+ key = (u8)(mr->rkey & 0x000000FF);
+ ib_update_fast_reg_key(mr, ++key);
}
/* Prepare FASTREG WR */
memset(&fr_wr, 0, sizeof(fr_wr));
fr_wr.wr_id = ISER_FASTREG_LI_WRID;
fr_wr.opcode = IB_WR_FAST_REG_MR;
- fr_wr.wr.fast_reg.iova_start =
- fr_desc->data_frpl->page_list[0] + page_off;
- fr_wr.wr.fast_reg.page_list = fr_desc->data_frpl;
+ fr_wr.wr.fast_reg.iova_start = frpl->page_list[0] + page_off;
+ fr_wr.wr.fast_reg.page_list = frpl;
fr_wr.wr.fast_reg.page_list_len = pagelist_len;
fr_wr.wr.fast_reg.page_shift = PAGE_SHIFT;
- fr_wr.wr.fast_reg.length = data_len;
- fr_wr.wr.fast_reg.rkey = fr_desc->data_mr->rkey;
+ fr_wr.wr.fast_reg.length = mem->len;
+ fr_wr.wr.fast_reg.rkey = mr->rkey;
fr_wr.wr.fast_reg.access_flags = IB_ACCESS_LOCAL_WRITE;
if (!wr)
@@ -2287,15 +2594,157 @@ isert_fast_reg_mr(struct fast_reg_descriptor *fr_desc,
pr_err("fast registration failed, ret:%d\n", ret);
return ret;
}
- fr_desc->valid = false;
+ fr_desc->ind &= ~ind;
+
+ sge->lkey = mr->lkey;
+ sge->addr = frpl->page_list[0] + page_off;
+ sge->length = mem->len;
+
+ pr_debug("%s:%d sge: addr: 0x%llx length: %u lkey: %x\n",
+ __func__, __LINE__, sge->addr, sge->length,
+ sge->lkey);
+
+ return ret;
+}
+
+static inline enum ib_t10_dif_type
+se2ib_prot_type(enum target_prot_type prot_type)
+{
+ switch (prot_type) {
+ case TARGET_DIF_TYPE0_PROT:
+ return IB_T10DIF_NONE;
+ case TARGET_DIF_TYPE1_PROT:
+ return IB_T10DIF_TYPE1;
+ case TARGET_DIF_TYPE2_PROT:
+ return IB_T10DIF_TYPE2;
+ case TARGET_DIF_TYPE3_PROT:
+ return IB_T10DIF_TYPE3;
+ default:
+ return IB_T10DIF_NONE;
+ }
+}
+
+static int
+isert_set_sig_attrs(struct se_cmd *se_cmd, struct ib_sig_attrs *sig_attrs)
+{
+ enum ib_t10_dif_type ib_prot_type = se2ib_prot_type(se_cmd->prot_type);
+
+ sig_attrs->mem.sig_type = IB_SIG_TYPE_T10_DIF;
+ sig_attrs->wire.sig_type = IB_SIG_TYPE_T10_DIF;
+ sig_attrs->mem.sig.dif.pi_interval =
+ se_cmd->se_dev->dev_attrib.block_size;
+ sig_attrs->wire.sig.dif.pi_interval =
+ se_cmd->se_dev->dev_attrib.block_size;
+
+ switch (se_cmd->prot_op) {
+ case TARGET_PROT_DIN_INSERT:
+ case TARGET_PROT_DOUT_STRIP:
+ sig_attrs->mem.sig.dif.type = IB_T10DIF_NONE;
+ sig_attrs->wire.sig.dif.type = ib_prot_type;
+ sig_attrs->wire.sig.dif.bg_type = IB_T10DIF_CRC;
+ sig_attrs->wire.sig.dif.ref_tag = se_cmd->reftag_seed;
+ break;
+ case TARGET_PROT_DOUT_INSERT:
+ case TARGET_PROT_DIN_STRIP:
+ sig_attrs->mem.sig.dif.type = ib_prot_type;
+ sig_attrs->mem.sig.dif.bg_type = IB_T10DIF_CRC;
+ sig_attrs->mem.sig.dif.ref_tag = se_cmd->reftag_seed;
+ sig_attrs->wire.sig.dif.type = IB_T10DIF_NONE;
+ break;
+ case TARGET_PROT_DIN_PASS:
+ case TARGET_PROT_DOUT_PASS:
+ sig_attrs->mem.sig.dif.type = ib_prot_type;
+ sig_attrs->mem.sig.dif.bg_type = IB_T10DIF_CRC;
+ sig_attrs->mem.sig.dif.ref_tag = se_cmd->reftag_seed;
+ sig_attrs->wire.sig.dif.type = ib_prot_type;
+ sig_attrs->wire.sig.dif.bg_type = IB_T10DIF_CRC;
+ sig_attrs->wire.sig.dif.ref_tag = se_cmd->reftag_seed;
+ break;
+ default:
+ pr_err("Unsupported PI operation %d\n", se_cmd->prot_op);
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static inline u8
+isert_set_prot_checks(u8 prot_checks)
+{
+ return (prot_checks & TARGET_DIF_CHECK_GUARD ? 0xc0 : 0) |
+ (prot_checks & TARGET_DIF_CHECK_REFTAG ? 0x30 : 0) |
+ (prot_checks & TARGET_DIF_CHECK_REFTAG ? 0x0f : 0);
+}
+
+static int
+isert_reg_sig_mr(struct isert_conn *isert_conn, struct se_cmd *se_cmd,
+ struct fast_reg_descriptor *fr_desc,
+ struct ib_sge *data_sge, struct ib_sge *prot_sge,
+ struct ib_sge *sig_sge)
+{
+ struct ib_send_wr sig_wr, inv_wr;
+ struct ib_send_wr *bad_wr, *wr = NULL;
+ struct pi_context *pi_ctx = fr_desc->pi_ctx;
+ struct ib_sig_attrs sig_attrs;
+ int ret;
+ u32 key;
+
+ memset(&sig_attrs, 0, sizeof(sig_attrs));
+ ret = isert_set_sig_attrs(se_cmd, &sig_attrs);
+ if (ret)
+ goto err;
+
+ sig_attrs.check_mask = isert_set_prot_checks(se_cmd->prot_checks);
+
+ if (!(fr_desc->ind & ISERT_SIG_KEY_VALID)) {
+ memset(&inv_wr, 0, sizeof(inv_wr));
+ inv_wr.opcode = IB_WR_LOCAL_INV;
+ inv_wr.wr_id = ISER_FASTREG_LI_WRID;
+ inv_wr.ex.invalidate_rkey = pi_ctx->sig_mr->rkey;
+ wr = &inv_wr;
+ /* Bump the key */
+ key = (u8)(pi_ctx->sig_mr->rkey & 0x000000FF);
+ ib_update_fast_reg_key(pi_ctx->sig_mr, ++key);
+ }
- ib_sge->lkey = fr_desc->data_mr->lkey;
- ib_sge->addr = fr_desc->data_frpl->page_list[0] + page_off;
- ib_sge->length = data_len;
+ memset(&sig_wr, 0, sizeof(sig_wr));
+ sig_wr.opcode = IB_WR_REG_SIG_MR;
+ sig_wr.wr_id = ISER_FASTREG_LI_WRID;
+ sig_wr.sg_list = data_sge;
+ sig_wr.num_sge = 1;
+ sig_wr.wr.sig_handover.access_flags = IB_ACCESS_LOCAL_WRITE;
+ sig_wr.wr.sig_handover.sig_attrs = &sig_attrs;
+ sig_wr.wr.sig_handover.sig_mr = pi_ctx->sig_mr;
+ if (se_cmd->t_prot_sg)
+ sig_wr.wr.sig_handover.prot = prot_sge;
- pr_debug("RDMA ib_sge: addr: 0x%16llx length: %u lkey: %08x\n",
- ib_sge->addr, ib_sge->length, ib_sge->lkey);
+ if (!wr)
+ wr = &sig_wr;
+ else
+ wr->next = &sig_wr;
+ ret = ib_post_send(isert_conn->conn_qp, wr, &bad_wr);
+ if (ret) {
+ pr_err("fast registration failed, ret:%d\n", ret);
+ goto err;
+ }
+ fr_desc->ind &= ~ISERT_SIG_KEY_VALID;
+
+ sig_sge->lkey = pi_ctx->sig_mr->lkey;
+ sig_sge->addr = 0;
+ sig_sge->length = se_cmd->data_length;
+ if (se_cmd->prot_op != TARGET_PROT_DIN_STRIP &&
+ se_cmd->prot_op != TARGET_PROT_DOUT_INSERT)
+ /*
+ * We have protection guards on the wire
+ * so we need to set a larget transfer
+ */
+ sig_sge->length += se_cmd->prot_length;
+
+ pr_debug("sig_sge: addr: 0x%llx length: %u lkey: %x\n",
+ sig_sge->addr, sig_sge->length,
+ sig_sge->lkey);
+err:
return ret;
}
@@ -2305,62 +2754,82 @@ isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
{
struct se_cmd *se_cmd = &cmd->se_cmd;
struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
- struct isert_conn *isert_conn = (struct isert_conn *)conn->context;
- struct ib_device *ib_dev = isert_conn->conn_cm_id->device;
+ struct isert_conn *isert_conn = conn->context;
+ struct ib_sge data_sge;
struct ib_send_wr *send_wr;
- struct ib_sge *ib_sge;
- struct scatterlist *sg_start;
- struct fast_reg_descriptor *fr_desc;
- u32 sg_off = 0, sg_nents;
- u32 offset = 0, data_len, data_left, rdma_write_max;
- int ret = 0, count;
+ struct fast_reg_descriptor *fr_desc = NULL;
+ u32 offset;
+ int ret = 0;
unsigned long flags;
- if (wr->iser_ib_op == ISER_IB_RDMA_WRITE) {
- data_left = se_cmd->data_length;
- } else {
- offset = cmd->write_data_done;
- sg_off = offset / PAGE_SIZE;
- data_left = se_cmd->data_length - cmd->write_data_done;
- isert_cmd->tx_desc.isert_cmd = isert_cmd;
- }
+ isert_cmd->tx_desc.isert_cmd = isert_cmd;
- sg_start = &cmd->se_cmd.t_data_sg[sg_off];
- sg_nents = se_cmd->t_data_nents - sg_off;
+ offset = wr->iser_ib_op == ISER_IB_RDMA_READ ? cmd->write_data_done : 0;
+ ret = isert_map_data_buf(isert_conn, isert_cmd, se_cmd->t_data_sg,
+ se_cmd->t_data_nents, se_cmd->data_length,
+ offset, wr->iser_ib_op, &wr->data);
+ if (ret)
+ return ret;
- count = ib_dma_map_sg(ib_dev, sg_start, sg_nents,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
- if (unlikely(!count)) {
- pr_err("Cmd: %p unrable to map SGs\n", isert_cmd);
- return -EINVAL;
+ if (wr->data.dma_nents != 1 ||
+ se_cmd->prot_op != TARGET_PROT_NORMAL) {
+ spin_lock_irqsave(&isert_conn->conn_lock, flags);
+ fr_desc = list_first_entry(&isert_conn->conn_fr_pool,
+ struct fast_reg_descriptor, list);
+ list_del(&fr_desc->list);
+ spin_unlock_irqrestore(&isert_conn->conn_lock, flags);
+ wr->fr_desc = fr_desc;
}
- wr->sge = sg_start;
- wr->num_sge = sg_nents;
- pr_debug("Mapped cmd: %p count: %u sg: %p sg_nents: %u rdma_len %d\n",
- isert_cmd, count, sg_start, sg_nents, data_left);
- memset(&wr->s_ib_sge, 0, sizeof(*ib_sge));
- ib_sge = &wr->s_ib_sge;
- wr->ib_sge = ib_sge;
+ ret = isert_fast_reg_mr(isert_conn, fr_desc, &wr->data,
+ ISERT_DATA_KEY_VALID, &data_sge);
+ if (ret)
+ goto unmap_cmd;
+ if (se_cmd->prot_op != TARGET_PROT_NORMAL) {
+ struct ib_sge prot_sge, sig_sge;
+
+ if (se_cmd->t_prot_sg) {
+ ret = isert_map_data_buf(isert_conn, isert_cmd,
+ se_cmd->t_prot_sg,
+ se_cmd->t_prot_nents,
+ se_cmd->prot_length,
+ 0, wr->iser_ib_op, &wr->prot);
+ if (ret)
+ goto unmap_cmd;
+
+ ret = isert_fast_reg_mr(isert_conn, fr_desc, &wr->prot,
+ ISERT_PROT_KEY_VALID, &prot_sge);
+ if (ret)
+ goto unmap_prot_cmd;
+ }
+
+ ret = isert_reg_sig_mr(isert_conn, se_cmd, fr_desc,
+ &data_sge, &prot_sge, &sig_sge);
+ if (ret)
+ goto unmap_prot_cmd;
+
+ fr_desc->ind |= ISERT_PROTECTED;
+ memcpy(&wr->s_ib_sge, &sig_sge, sizeof(sig_sge));
+ } else
+ memcpy(&wr->s_ib_sge, &data_sge, sizeof(data_sge));
+
+ wr->ib_sge = &wr->s_ib_sge;
wr->send_wr_num = 1;
memset(&wr->s_send_wr, 0, sizeof(*send_wr));
wr->send_wr = &wr->s_send_wr;
-
wr->isert_cmd = isert_cmd;
- rdma_write_max = ISCSI_ISER_SG_TABLESIZE * PAGE_SIZE;
send_wr = &isert_cmd->rdma_wr.s_send_wr;
- send_wr->sg_list = ib_sge;
+ send_wr->sg_list = &wr->s_ib_sge;
send_wr->num_sge = 1;
send_wr->wr_id = (unsigned long)&isert_cmd->tx_desc;
if (wr->iser_ib_op == ISER_IB_RDMA_WRITE) {
send_wr->opcode = IB_WR_RDMA_WRITE;
send_wr->wr.rdma.remote_addr = isert_cmd->read_va;
send_wr->wr.rdma.rkey = isert_cmd->read_stag;
- send_wr->send_flags = 0;
- send_wr->next = &isert_cmd->tx_desc.send_wr;
+ send_wr->send_flags = se_cmd->prot_op == TARGET_PROT_NORMAL ?
+ 0 : IB_SEND_SIGNALED;
} else {
send_wr->opcode = IB_WR_RDMA_READ;
send_wr->wr.rdma.remote_addr = isert_cmd->write_va;
@@ -2368,37 +2837,18 @@ isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
send_wr->send_flags = IB_SEND_SIGNALED;
}
- data_len = min(data_left, rdma_write_max);
- wr->cur_rdma_length = data_len;
-
- /* if there is a single dma entry, dma mr is sufficient */
- if (count == 1) {
- ib_sge->addr = ib_sg_dma_address(ib_dev, &sg_start[0]);
- ib_sge->length = ib_sg_dma_len(ib_dev, &sg_start[0]);
- ib_sge->lkey = isert_conn->conn_mr->lkey;
- wr->fr_desc = NULL;
- } else {
+ return 0;
+unmap_prot_cmd:
+ if (se_cmd->t_prot_sg)
+ isert_unmap_data_buf(isert_conn, &wr->prot);
+unmap_cmd:
+ if (fr_desc) {
spin_lock_irqsave(&isert_conn->conn_lock, flags);
- fr_desc = list_first_entry(&isert_conn->conn_fr_pool,
- struct fast_reg_descriptor, list);
- list_del(&fr_desc->list);
+ list_add_tail(&fr_desc->list, &isert_conn->conn_fr_pool);
spin_unlock_irqrestore(&isert_conn->conn_lock, flags);
- wr->fr_desc = fr_desc;
-
- ret = isert_fast_reg_mr(fr_desc, isert_conn, sg_start,
- ib_sge, sg_nents, offset, data_len);
- if (ret) {
- list_add_tail(&fr_desc->list, &isert_conn->conn_fr_pool);
- goto unmap_sg;
- }
}
+ isert_unmap_data_buf(isert_conn, &wr->data);
- return 0;
-
-unmap_sg:
- ib_dma_unmap_sg(ib_dev, sg_start, sg_nents,
- (wr->iser_ib_op == ISER_IB_RDMA_WRITE) ?
- DMA_TO_DEVICE : DMA_FROM_DEVICE);
return ret;
}
@@ -2422,25 +2872,35 @@ isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
return rc;
}
- /*
- * Build isert_conn->tx_desc for iSCSI response PDU and attach
- */
- isert_create_send_desc(isert_conn, isert_cmd, &isert_cmd->tx_desc);
- iscsit_build_rsp_pdu(cmd, conn, true, (struct iscsi_scsi_rsp *)
- &isert_cmd->tx_desc.iscsi_header);
- isert_init_tx_hdrs(isert_conn, &isert_cmd->tx_desc);
- isert_init_send_wr(isert_conn, isert_cmd,
- &isert_cmd->tx_desc.send_wr, true);
+ if (se_cmd->prot_op == TARGET_PROT_NORMAL) {
+ /*
+ * Build isert_conn->tx_desc for iSCSI response PDU and attach
+ */
+ isert_create_send_desc(isert_conn, isert_cmd,
+ &isert_cmd->tx_desc);
+ iscsit_build_rsp_pdu(cmd, conn, true, (struct iscsi_scsi_rsp *)
+ &isert_cmd->tx_desc.iscsi_header);
+ isert_init_tx_hdrs(isert_conn, &isert_cmd->tx_desc);
+ isert_init_send_wr(isert_conn, isert_cmd,
+ &isert_cmd->tx_desc.send_wr, true);
+ isert_cmd->rdma_wr.s_send_wr.next = &isert_cmd->tx_desc.send_wr;
+ wr->send_wr_num += 1;
+ }
- atomic_add(wr->send_wr_num + 1, &isert_conn->post_send_buf_count);
+ atomic_add(wr->send_wr_num, &isert_conn->post_send_buf_count);
rc = ib_post_send(isert_conn->conn_qp, wr->send_wr, &wr_failed);
if (rc) {
pr_warn("ib_post_send() failed for IB_WR_RDMA_WRITE\n");
- atomic_sub(wr->send_wr_num + 1, &isert_conn->post_send_buf_count);
+ atomic_sub(wr->send_wr_num, &isert_conn->post_send_buf_count);
}
- pr_debug("Cmd: %p posted RDMA_WRITE + Response for iSER Data READ\n",
- isert_cmd);
+
+ if (se_cmd->prot_op == TARGET_PROT_NORMAL)
+ pr_debug("Cmd: %p posted RDMA_WRITE + Response for iSER Data "
+ "READ\n", isert_cmd);
+ else
+ pr_debug("Cmd: %p posted RDMA_WRITE for iSER Data READ\n",
+ isert_cmd);
return 1;
}
@@ -2551,7 +3011,7 @@ isert_setup_np(struct iscsi_np *np,
pr_err("Unable to allocate struct isert_np\n");
return -ENOMEM;
}
- init_waitqueue_head(&isert_np->np_accept_wq);
+ sema_init(&isert_np->np_sem, 0);
mutex_init(&isert_np->np_accept_mutex);
INIT_LIST_HEAD(&isert_np->np_accept_list);
init_completion(&isert_np->np_login_comp);
@@ -2600,18 +3060,6 @@ out:
}
static int
-isert_check_accept_queue(struct isert_np *isert_np)
-{
- int empty;
-
- mutex_lock(&isert_np->np_accept_mutex);
- empty = list_empty(&isert_np->np_accept_list);
- mutex_unlock(&isert_np->np_accept_mutex);
-
- return empty;
-}
-
-static int
isert_rdma_accept(struct isert_conn *isert_conn)
{
struct rdma_cm_id *cm_id = isert_conn->conn_cm_id;
@@ -2703,16 +3151,19 @@ isert_accept_np(struct iscsi_np *np, struct iscsi_conn *conn)
int max_accept = 0, ret;
accept_wait:
- ret = wait_event_interruptible(isert_np->np_accept_wq,
- !isert_check_accept_queue(isert_np) ||
- np->np_thread_state == ISCSI_NP_THREAD_RESET);
+ ret = down_interruptible(&isert_np->np_sem);
if (max_accept > 5)
return -ENODEV;
spin_lock_bh(&np->np_thread_lock);
- if (np->np_thread_state == ISCSI_NP_THREAD_RESET) {
+ if (np->np_thread_state >= ISCSI_NP_THREAD_RESET) {
spin_unlock_bh(&np->np_thread_lock);
- pr_err("ISCSI_NP_THREAD_RESET for isert_accept_np\n");
+ pr_debug("np_thread_state %d for isert_accept_np\n",
+ np->np_thread_state);
+ /**
+ * No point in stalling here when np_thread
+ * is in state RESET/SHUTDOWN/EXIT - bail
+ **/
return -ENODEV;
}
spin_unlock_bh(&np->np_thread_lock);
@@ -2762,15 +3213,9 @@ static void isert_wait_conn(struct iscsi_conn *conn)
struct isert_conn *isert_conn = conn->context;
pr_debug("isert_wait_conn: Starting \n");
- /*
- * Decrement post_send_buf_count for special case when called
- * from isert_do_control_comp() -> iscsit_logout_post_handler()
- */
- mutex_lock(&isert_conn->conn_mutex);
- if (isert_conn->logout_posted)
- atomic_dec(&isert_conn->post_send_buf_count);
- if (isert_conn->conn_cm_id && isert_conn->state != ISER_CONN_DOWN) {
+ mutex_lock(&isert_conn->conn_mutex);
+ if (isert_conn->conn_cm_id) {
pr_debug("Calling rdma_disconnect from isert_wait_conn\n");
rdma_disconnect(isert_conn->conn_cm_id);
}
@@ -2815,6 +3260,8 @@ static struct iscsit_transport iser_target_transport = {
.iscsit_get_dataout = isert_get_dataout,
.iscsit_queue_data_in = isert_put_datain,
.iscsit_queue_status = isert_put_response,
+ .iscsit_aborted_task = isert_aborted_task,
+ .iscsit_get_sup_prot_ops = isert_get_sup_prot_ops,
};
static int __init isert_init(void)
@@ -2845,6 +3292,7 @@ destroy_rx_wq:
static void __exit isert_exit(void)
{
+ flush_scheduled_work();
destroy_workqueue(isert_comp_wq);
destroy_workqueue(isert_rx_wq);
iscsit_unregister_transport(&iser_target_transport);
diff --git a/drivers/infiniband/ulp/isert/ib_isert.h b/drivers/infiniband/ulp/isert/ib_isert.h
index f6ae7f5dd40..04f51f7bf61 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.h
+++ b/drivers/infiniband/ulp/isert/ib_isert.h
@@ -50,11 +50,35 @@ struct iser_tx_desc {
struct ib_send_wr send_wr;
} __packed;
+enum isert_indicator {
+ ISERT_PROTECTED = 1 << 0,
+ ISERT_DATA_KEY_VALID = 1 << 1,
+ ISERT_PROT_KEY_VALID = 1 << 2,
+ ISERT_SIG_KEY_VALID = 1 << 3,
+};
+
+struct pi_context {
+ struct ib_mr *prot_mr;
+ struct ib_fast_reg_page_list *prot_frpl;
+ struct ib_mr *sig_mr;
+};
+
struct fast_reg_descriptor {
- struct list_head list;
- struct ib_mr *data_mr;
- struct ib_fast_reg_page_list *data_frpl;
- bool valid;
+ struct list_head list;
+ struct ib_mr *data_mr;
+ struct ib_fast_reg_page_list *data_frpl;
+ u8 ind;
+ struct pi_context *pi_ctx;
+};
+
+struct isert_data_buf {
+ struct scatterlist *sg;
+ int nents;
+ u32 sg_off;
+ u32 len; /* cur_rdma_length */
+ u32 offset;
+ unsigned int dma_nents;
+ enum dma_data_direction dma_dir;
};
struct isert_rdma_wr {
@@ -63,12 +87,11 @@ struct isert_rdma_wr {
enum iser_ib_op_code iser_ib_op;
struct ib_sge *ib_sge;
struct ib_sge s_ib_sge;
- int num_sge;
- struct scatterlist *sge;
int send_wr_num;
struct ib_send_wr *send_wr;
struct ib_send_wr s_send_wr;
- u32 cur_rdma_length;
+ struct isert_data_buf data;
+ struct isert_data_buf prot;
struct fast_reg_descriptor *fr_desc;
};
@@ -93,7 +116,6 @@ struct isert_device;
struct isert_conn {
enum iser_conn_state state;
- bool logout_posted;
int post_recv_buf_count;
atomic_t post_send_buf_count;
u32 responder_resources;
@@ -128,6 +150,7 @@ struct isert_conn {
#define ISERT_COMP_BATCH_COUNT 8
int conn_comp_batch;
struct llist_head conn_comp_llist;
+ bool disconnect;
};
#define ISERT_MAX_CQ 64
@@ -141,6 +164,7 @@ struct isert_cq_desc {
struct isert_device {
int use_fastreg;
+ bool pi_capable;
int cqs_used;
int refcount;
int cq_active_qps[ISERT_MAX_CQ];
@@ -158,7 +182,7 @@ struct isert_device {
};
struct isert_np {
- wait_queue_head_t np_accept_wq;
+ struct semaphore np_sem;
struct rdma_cm_id *np_cm_id;
struct mutex np_accept_mutex;
struct list_head np_accept_list;
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c
index 66a908bf3fb..e3c2c5b4297 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -30,7 +30,7 @@
* SOFTWARE.
*/
-#define pr_fmt(fmt) PFX fmt
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
#include <linux/module.h>
#include <linux/init.h>
@@ -66,6 +66,8 @@ static unsigned int srp_sg_tablesize;
static unsigned int cmd_sg_entries;
static unsigned int indirect_sg_entries;
static bool allow_ext_sg;
+static bool prefer_fr;
+static bool register_always;
static int topspin_workarounds = 1;
module_param(srp_sg_tablesize, uint, 0444);
@@ -87,6 +89,14 @@ module_param(topspin_workarounds, int, 0444);
MODULE_PARM_DESC(topspin_workarounds,
"Enable workarounds for Topspin/Cisco SRP target bugs if != 0");
+module_param(prefer_fr, bool, 0444);
+MODULE_PARM_DESC(prefer_fr,
+"Whether to use fast registration if both FMR and fast registration are supported");
+
+module_param(register_always, bool, 0444);
+MODULE_PARM_DESC(register_always,
+ "Use memory registration even for contiguous memory regions");
+
static struct kernel_param_ops srp_tmo_ops;
static int srp_reconnect_delay = 10;
@@ -288,28 +298,174 @@ static int srp_new_cm_id(struct srp_target_port *target)
return 0;
}
+static struct ib_fmr_pool *srp_alloc_fmr_pool(struct srp_target_port *target)
+{
+ struct srp_device *dev = target->srp_host->srp_dev;
+ struct ib_fmr_pool_param fmr_param;
+
+ memset(&fmr_param, 0, sizeof(fmr_param));
+ fmr_param.pool_size = target->scsi_host->can_queue;
+ fmr_param.dirty_watermark = fmr_param.pool_size / 4;
+ fmr_param.cache = 1;
+ fmr_param.max_pages_per_fmr = dev->max_pages_per_mr;
+ fmr_param.page_shift = ilog2(dev->mr_page_size);
+ fmr_param.access = (IB_ACCESS_LOCAL_WRITE |
+ IB_ACCESS_REMOTE_WRITE |
+ IB_ACCESS_REMOTE_READ);
+
+ return ib_create_fmr_pool(dev->pd, &fmr_param);
+}
+
+/**
+ * srp_destroy_fr_pool() - free the resources owned by a pool
+ * @pool: Fast registration pool to be destroyed.
+ */
+static void srp_destroy_fr_pool(struct srp_fr_pool *pool)
+{
+ int i;
+ struct srp_fr_desc *d;
+
+ if (!pool)
+ return;
+
+ for (i = 0, d = &pool->desc[0]; i < pool->size; i++, d++) {
+ if (d->frpl)
+ ib_free_fast_reg_page_list(d->frpl);
+ if (d->mr)
+ ib_dereg_mr(d->mr);
+ }
+ kfree(pool);
+}
+
+/**
+ * srp_create_fr_pool() - allocate and initialize a pool for fast registration
+ * @device: IB device to allocate fast registration descriptors for.
+ * @pd: Protection domain associated with the FR descriptors.
+ * @pool_size: Number of descriptors to allocate.
+ * @max_page_list_len: Maximum fast registration work request page list length.
+ */
+static struct srp_fr_pool *srp_create_fr_pool(struct ib_device *device,
+ struct ib_pd *pd, int pool_size,
+ int max_page_list_len)
+{
+ struct srp_fr_pool *pool;
+ struct srp_fr_desc *d;
+ struct ib_mr *mr;
+ struct ib_fast_reg_page_list *frpl;
+ int i, ret = -EINVAL;
+
+ if (pool_size <= 0)
+ goto err;
+ ret = -ENOMEM;
+ pool = kzalloc(sizeof(struct srp_fr_pool) +
+ pool_size * sizeof(struct srp_fr_desc), GFP_KERNEL);
+ if (!pool)
+ goto err;
+ pool->size = pool_size;
+ pool->max_page_list_len = max_page_list_len;
+ spin_lock_init(&pool->lock);
+ INIT_LIST_HEAD(&pool->free_list);
+
+ for (i = 0, d = &pool->desc[0]; i < pool->size; i++, d++) {
+ mr = ib_alloc_fast_reg_mr(pd, max_page_list_len);
+ if (IS_ERR(mr)) {
+ ret = PTR_ERR(mr);
+ goto destroy_pool;
+ }
+ d->mr = mr;
+ frpl = ib_alloc_fast_reg_page_list(device, max_page_list_len);
+ if (IS_ERR(frpl)) {
+ ret = PTR_ERR(frpl);
+ goto destroy_pool;
+ }
+ d->frpl = frpl;
+ list_add_tail(&d->entry, &pool->free_list);
+ }
+
+out:
+ return pool;
+
+destroy_pool:
+ srp_destroy_fr_pool(pool);
+
+err:
+ pool = ERR_PTR(ret);
+ goto out;
+}
+
+/**
+ * srp_fr_pool_get() - obtain a descriptor suitable for fast registration
+ * @pool: Pool to obtain descriptor from.
+ */
+static struct srp_fr_desc *srp_fr_pool_get(struct srp_fr_pool *pool)
+{
+ struct srp_fr_desc *d = NULL;
+ unsigned long flags;
+
+ spin_lock_irqsave(&pool->lock, flags);
+ if (!list_empty(&pool->free_list)) {
+ d = list_first_entry(&pool->free_list, typeof(*d), entry);
+ list_del(&d->entry);
+ }
+ spin_unlock_irqrestore(&pool->lock, flags);
+
+ return d;
+}
+
+/**
+ * srp_fr_pool_put() - put an FR descriptor back in the free list
+ * @pool: Pool the descriptor was allocated from.
+ * @desc: Pointer to an array of fast registration descriptor pointers.
+ * @n: Number of descriptors to put back.
+ *
+ * Note: The caller must already have queued an invalidation request for
+ * desc->mr->rkey before calling this function.
+ */
+static void srp_fr_pool_put(struct srp_fr_pool *pool, struct srp_fr_desc **desc,
+ int n)
+{
+ unsigned long flags;
+ int i;
+
+ spin_lock_irqsave(&pool->lock, flags);
+ for (i = 0; i < n; i++)
+ list_add(&desc[i]->entry, &pool->free_list);
+ spin_unlock_irqrestore(&pool->lock, flags);
+}
+
+static struct srp_fr_pool *srp_alloc_fr_pool(struct srp_target_port *target)
+{
+ struct srp_device *dev = target->srp_host->srp_dev;
+
+ return srp_create_fr_pool(dev->dev, dev->pd,
+ target->scsi_host->can_queue,
+ dev->max_pages_per_mr);
+}
+
static int srp_create_target_ib(struct srp_target_port *target)
{
+ struct srp_device *dev = target->srp_host->srp_dev;
struct ib_qp_init_attr *init_attr;
struct ib_cq *recv_cq, *send_cq;
struct ib_qp *qp;
+ struct ib_fmr_pool *fmr_pool = NULL;
+ struct srp_fr_pool *fr_pool = NULL;
+ const int m = 1 + dev->use_fast_reg;
int ret;
init_attr = kzalloc(sizeof *init_attr, GFP_KERNEL);
if (!init_attr)
return -ENOMEM;
- recv_cq = ib_create_cq(target->srp_host->srp_dev->dev,
- srp_recv_completion, NULL, target,
+ recv_cq = ib_create_cq(dev->dev, srp_recv_completion, NULL, target,
target->queue_size, target->comp_vector);
if (IS_ERR(recv_cq)) {
ret = PTR_ERR(recv_cq);
goto err;
}
- send_cq = ib_create_cq(target->srp_host->srp_dev->dev,
- srp_send_completion, NULL, target,
- target->queue_size, target->comp_vector);
+ send_cq = ib_create_cq(dev->dev, srp_send_completion, NULL, target,
+ m * target->queue_size, target->comp_vector);
if (IS_ERR(send_cq)) {
ret = PTR_ERR(send_cq);
goto err_recv_cq;
@@ -318,16 +474,16 @@ static int srp_create_target_ib(struct srp_target_port *target)
ib_req_notify_cq(recv_cq, IB_CQ_NEXT_COMP);
init_attr->event_handler = srp_qp_event;
- init_attr->cap.max_send_wr = target->queue_size;
+ init_attr->cap.max_send_wr = m * target->queue_size;
init_attr->cap.max_recv_wr = target->queue_size;
init_attr->cap.max_recv_sge = 1;
init_attr->cap.max_send_sge = 1;
- init_attr->sq_sig_type = IB_SIGNAL_ALL_WR;
+ init_attr->sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr->qp_type = IB_QPT_RC;
init_attr->send_cq = send_cq;
init_attr->recv_cq = recv_cq;
- qp = ib_create_qp(target->srp_host->srp_dev->pd, init_attr);
+ qp = ib_create_qp(dev->pd, init_attr);
if (IS_ERR(qp)) {
ret = PTR_ERR(qp);
goto err_send_cq;
@@ -337,6 +493,30 @@ static int srp_create_target_ib(struct srp_target_port *target)
if (ret)
goto err_qp;
+ if (dev->use_fast_reg && dev->has_fr) {
+ fr_pool = srp_alloc_fr_pool(target);
+ if (IS_ERR(fr_pool)) {
+ ret = PTR_ERR(fr_pool);
+ shost_printk(KERN_WARNING, target->scsi_host, PFX
+ "FR pool allocation failed (%d)\n", ret);
+ goto err_qp;
+ }
+ if (target->fr_pool)
+ srp_destroy_fr_pool(target->fr_pool);
+ target->fr_pool = fr_pool;
+ } else if (!dev->use_fast_reg && dev->has_fmr) {
+ fmr_pool = srp_alloc_fmr_pool(target);
+ if (IS_ERR(fmr_pool)) {
+ ret = PTR_ERR(fmr_pool);
+ shost_printk(KERN_WARNING, target->scsi_host, PFX
+ "FMR pool allocation failed (%d)\n", ret);
+ goto err_qp;
+ }
+ if (target->fmr_pool)
+ ib_destroy_fmr_pool(target->fmr_pool);
+ target->fmr_pool = fmr_pool;
+ }
+
if (target->qp)
ib_destroy_qp(target->qp);
if (target->recv_cq)
@@ -371,8 +551,16 @@ err:
*/
static void srp_free_target_ib(struct srp_target_port *target)
{
+ struct srp_device *dev = target->srp_host->srp_dev;
int i;
+ if (dev->use_fast_reg) {
+ if (target->fr_pool)
+ srp_destroy_fr_pool(target->fr_pool);
+ } else {
+ if (target->fmr_pool)
+ ib_destroy_fmr_pool(target->fmr_pool);
+ }
ib_destroy_qp(target->qp);
ib_destroy_cq(target->send_cq);
ib_destroy_cq(target->recv_cq);
@@ -577,7 +765,8 @@ static void srp_disconnect_target(struct srp_target_port *target)
static void srp_free_req_data(struct srp_target_port *target)
{
- struct ib_device *ibdev = target->srp_host->srp_dev->dev;
+ struct srp_device *dev = target->srp_host->srp_dev;
+ struct ib_device *ibdev = dev->dev;
struct srp_request *req;
int i;
@@ -586,7 +775,10 @@ static void srp_free_req_data(struct srp_target_port *target)
for (i = 0; i < target->req_ring_size; ++i) {
req = &target->req_ring[i];
- kfree(req->fmr_list);
+ if (dev->use_fast_reg)
+ kfree(req->fr_list);
+ else
+ kfree(req->fmr_list);
kfree(req->map_page);
if (req->indirect_dma_addr) {
ib_dma_unmap_single(ibdev, req->indirect_dma_addr,
@@ -605,6 +797,7 @@ static int srp_alloc_req_data(struct srp_target_port *target)
struct srp_device *srp_dev = target->srp_host->srp_dev;
struct ib_device *ibdev = srp_dev->dev;
struct srp_request *req;
+ void *mr_list;
dma_addr_t dma_addr;
int i, ret = -ENOMEM;
@@ -617,12 +810,20 @@ static int srp_alloc_req_data(struct srp_target_port *target)
for (i = 0; i < target->req_ring_size; ++i) {
req = &target->req_ring[i];
- req->fmr_list = kmalloc(target->cmd_sg_cnt * sizeof(void *),
- GFP_KERNEL);
- req->map_page = kmalloc(SRP_FMR_SIZE * sizeof(void *),
- GFP_KERNEL);
+ mr_list = kmalloc(target->cmd_sg_cnt * sizeof(void *),
+ GFP_KERNEL);
+ if (!mr_list)
+ goto out;
+ if (srp_dev->use_fast_reg)
+ req->fr_list = mr_list;
+ else
+ req->fmr_list = mr_list;
+ req->map_page = kmalloc(srp_dev->max_pages_per_mr *
+ sizeof(void *), GFP_KERNEL);
+ if (!req->map_page)
+ goto out;
req->indirect_desc = kmalloc(target->indirect_size, GFP_KERNEL);
- if (!req->fmr_list || !req->map_page || !req->indirect_desc)
+ if (!req->indirect_desc)
goto out;
dma_addr = ib_dma_map_single(ibdev, req->indirect_desc,
@@ -759,21 +960,56 @@ static int srp_connect_target(struct srp_target_port *target)
}
}
+static int srp_inv_rkey(struct srp_target_port *target, u32 rkey)
+{
+ struct ib_send_wr *bad_wr;
+ struct ib_send_wr wr = {
+ .opcode = IB_WR_LOCAL_INV,
+ .wr_id = LOCAL_INV_WR_ID_MASK,
+ .next = NULL,
+ .num_sge = 0,
+ .send_flags = 0,
+ .ex.invalidate_rkey = rkey,
+ };
+
+ return ib_post_send(target->qp, &wr, &bad_wr);
+}
+
static void srp_unmap_data(struct scsi_cmnd *scmnd,
struct srp_target_port *target,
struct srp_request *req)
{
- struct ib_device *ibdev = target->srp_host->srp_dev->dev;
- struct ib_pool_fmr **pfmr;
+ struct srp_device *dev = target->srp_host->srp_dev;
+ struct ib_device *ibdev = dev->dev;
+ int i, res;
if (!scsi_sglist(scmnd) ||
(scmnd->sc_data_direction != DMA_TO_DEVICE &&
scmnd->sc_data_direction != DMA_FROM_DEVICE))
return;
- pfmr = req->fmr_list;
- while (req->nfmr--)
- ib_fmr_pool_unmap(*pfmr++);
+ if (dev->use_fast_reg) {
+ struct srp_fr_desc **pfr;
+
+ for (i = req->nmdesc, pfr = req->fr_list; i > 0; i--, pfr++) {
+ res = srp_inv_rkey(target, (*pfr)->mr->rkey);
+ if (res < 0) {
+ shost_printk(KERN_ERR, target->scsi_host, PFX
+ "Queueing INV WR for rkey %#x failed (%d)\n",
+ (*pfr)->mr->rkey, res);
+ queue_work(system_long_wq,
+ &target->tl_err_work);
+ }
+ }
+ if (req->nmdesc)
+ srp_fr_pool_put(target->fr_pool, req->fr_list,
+ req->nmdesc);
+ } else {
+ struct ib_pool_fmr **pfmr;
+
+ for (i = req->nmdesc, pfmr = req->fmr_list; i > 0; i--, pfmr++)
+ ib_fmr_pool_unmap(*pfmr);
+ }
ib_dma_unmap_sg(ibdev, scsi_sglist(scmnd), scsi_sg_count(scmnd),
scmnd->sc_data_direction);
@@ -813,6 +1049,10 @@ static struct scsi_cmnd *srp_claim_req(struct srp_target_port *target,
/**
* srp_free_req() - Unmap data and add request to the free request list.
+ * @target: SRP target port.
+ * @req: Request to be freed.
+ * @scmnd: SCSI command associated with @req.
+ * @req_lim_delta: Amount to be added to @target->req_lim.
*/
static void srp_free_req(struct srp_target_port *target,
struct srp_request *req, struct scsi_cmnd *scmnd,
@@ -882,21 +1122,19 @@ static int srp_rport_reconnect(struct srp_rport *rport)
* callbacks will have finished before a new QP is allocated.
*/
ret = srp_new_cm_id(target);
- /*
- * Whether or not creating a new CM ID succeeded, create a new
- * QP. This guarantees that all completion callback function
- * invocations have finished before request resetting starts.
- */
- if (ret == 0)
- ret = srp_create_target_ib(target);
- else
- srp_create_target_ib(target);
for (i = 0; i < target->req_ring_size; ++i) {
struct srp_request *req = &target->req_ring[i];
srp_finish_req(target, req, NULL, DID_RESET << 16);
}
+ /*
+ * Whether or not creating a new CM ID succeeded, create a new
+ * QP. This guarantees that all callback functions for the old QP have
+ * finished before any send requests are posted on the new QP.
+ */
+ ret += srp_create_target_ib(target);
+
INIT_LIST_HEAD(&target->free_tx);
for (i = 0; i < target->queue_size; ++i)
list_add(&target->tx_ring[i]->list, &target->free_tx);
@@ -928,33 +1166,87 @@ static void srp_map_desc(struct srp_map_state *state, dma_addr_t dma_addr,
static int srp_map_finish_fmr(struct srp_map_state *state,
struct srp_target_port *target)
{
- struct srp_device *dev = target->srp_host->srp_dev;
struct ib_pool_fmr *fmr;
u64 io_addr = 0;
- if (!state->npages)
- return 0;
-
- if (state->npages == 1) {
- srp_map_desc(state, state->base_dma_addr, state->fmr_len,
- target->rkey);
- state->npages = state->fmr_len = 0;
- return 0;
- }
-
- fmr = ib_fmr_pool_map_phys(dev->fmr_pool, state->pages,
+ fmr = ib_fmr_pool_map_phys(target->fmr_pool, state->pages,
state->npages, io_addr);
if (IS_ERR(fmr))
return PTR_ERR(fmr);
*state->next_fmr++ = fmr;
- state->nfmr++;
+ state->nmdesc++;
+
+ srp_map_desc(state, 0, state->dma_len, fmr->fmr->rkey);
- srp_map_desc(state, 0, state->fmr_len, fmr->fmr->rkey);
- state->npages = state->fmr_len = 0;
return 0;
}
+static int srp_map_finish_fr(struct srp_map_state *state,
+ struct srp_target_port *target)
+{
+ struct srp_device *dev = target->srp_host->srp_dev;
+ struct ib_send_wr *bad_wr;
+ struct ib_send_wr wr;
+ struct srp_fr_desc *desc;
+ u32 rkey;
+
+ desc = srp_fr_pool_get(target->fr_pool);
+ if (!desc)
+ return -ENOMEM;
+
+ rkey = ib_inc_rkey(desc->mr->rkey);
+ ib_update_fast_reg_key(desc->mr, rkey);
+
+ memcpy(desc->frpl->page_list, state->pages,
+ sizeof(state->pages[0]) * state->npages);
+
+ memset(&wr, 0, sizeof(wr));
+ wr.opcode = IB_WR_FAST_REG_MR;
+ wr.wr_id = FAST_REG_WR_ID_MASK;
+ wr.wr.fast_reg.iova_start = state->base_dma_addr;
+ wr.wr.fast_reg.page_list = desc->frpl;
+ wr.wr.fast_reg.page_list_len = state->npages;
+ wr.wr.fast_reg.page_shift = ilog2(dev->mr_page_size);
+ wr.wr.fast_reg.length = state->dma_len;
+ wr.wr.fast_reg.access_flags = (IB_ACCESS_LOCAL_WRITE |
+ IB_ACCESS_REMOTE_READ |
+ IB_ACCESS_REMOTE_WRITE);
+ wr.wr.fast_reg.rkey = desc->mr->lkey;
+
+ *state->next_fr++ = desc;
+ state->nmdesc++;
+
+ srp_map_desc(state, state->base_dma_addr, state->dma_len,
+ desc->mr->rkey);
+
+ return ib_post_send(target->qp, &wr, &bad_wr);
+}
+
+static int srp_finish_mapping(struct srp_map_state *state,
+ struct srp_target_port *target)
+{
+ int ret = 0;
+
+ if (state->npages == 0)
+ return 0;
+
+ if (state->npages == 1 && !register_always)
+ srp_map_desc(state, state->base_dma_addr, state->dma_len,
+ target->rkey);
+ else
+ ret = target->srp_host->srp_dev->use_fast_reg ?
+ srp_map_finish_fr(state, target) :
+ srp_map_finish_fmr(state, target);
+
+ if (ret == 0) {
+ state->npages = 0;
+ state->dma_len = 0;
+ }
+
+ return ret;
+}
+
static void srp_map_update_start(struct srp_map_state *state,
struct scatterlist *sg, int sg_index,
dma_addr_t dma_addr)
@@ -967,7 +1259,7 @@ static void srp_map_update_start(struct srp_map_state *state,
static int srp_map_sg_entry(struct srp_map_state *state,
struct srp_target_port *target,
struct scatterlist *sg, int sg_index,
- int use_fmr)
+ bool use_mr)
{
struct srp_device *dev = target->srp_host->srp_dev;
struct ib_device *ibdev = dev->dev;
@@ -979,23 +1271,25 @@ static int srp_map_sg_entry(struct srp_map_state *state,
if (!dma_len)
return 0;
- if (use_fmr == SRP_MAP_NO_FMR) {
- /* Once we're in direct map mode for a request, we don't
- * go back to FMR mode, so no need to update anything
+ if (!use_mr) {
+ /*
+ * Once we're in direct map mode for a request, we don't
+ * go back to FMR or FR mode, so no need to update anything
* other than the descriptor.
*/
srp_map_desc(state, dma_addr, dma_len, target->rkey);
return 0;
}
- /* If we start at an offset into the FMR page, don't merge into
- * the current FMR. Finish it out, and use the kernel's MR for this
- * sg entry. This is to avoid potential bugs on some SRP targets
- * that were never quite defined, but went away when the initiator
- * avoided using FMR on such page fragments.
+ /*
+ * Since not all RDMA HW drivers support non-zero page offsets for
+ * FMR, if we start at an offset into a page, don't merge into the
+ * current FMR mapping. Finish it out, and use the kernel's MR for
+ * this sg entry.
*/
- if (dma_addr & ~dev->fmr_page_mask || dma_len > dev->fmr_max_size) {
- ret = srp_map_finish_fmr(state, target);
+ if ((!dev->use_fast_reg && dma_addr & ~dev->mr_page_mask) ||
+ dma_len > dev->mr_max_size) {
+ ret = srp_finish_mapping(state, target);
if (ret)
return ret;
@@ -1004,52 +1298,106 @@ static int srp_map_sg_entry(struct srp_map_state *state,
return 0;
}
- /* If this is the first sg to go into the FMR, save our position.
- * We need to know the first unmapped entry, its index, and the
- * first unmapped address within that entry to be able to restart
- * mapping after an error.
+ /*
+ * If this is the first sg that will be mapped via FMR or via FR, save
+ * our position. We need to know the first unmapped entry, its index,
+ * and the first unmapped address within that entry to be able to
+ * restart mapping after an error.
*/
if (!state->unmapped_sg)
srp_map_update_start(state, sg, sg_index, dma_addr);
while (dma_len) {
- if (state->npages == SRP_FMR_SIZE) {
- ret = srp_map_finish_fmr(state, target);
+ unsigned offset = dma_addr & ~dev->mr_page_mask;
+ if (state->npages == dev->max_pages_per_mr || offset != 0) {
+ ret = srp_finish_mapping(state, target);
if (ret)
return ret;
srp_map_update_start(state, sg, sg_index, dma_addr);
}
- len = min_t(unsigned int, dma_len, dev->fmr_page_size);
+ len = min_t(unsigned int, dma_len, dev->mr_page_size - offset);
if (!state->npages)
state->base_dma_addr = dma_addr;
- state->pages[state->npages++] = dma_addr;
- state->fmr_len += len;
+ state->pages[state->npages++] = dma_addr & dev->mr_page_mask;
+ state->dma_len += len;
dma_addr += len;
dma_len -= len;
}
- /* If the last entry of the FMR wasn't a full page, then we need to
+ /*
+ * If the last entry of the MR wasn't a full page, then we need to
* close it out and start a new one -- we can only merge at page
* boundries.
*/
ret = 0;
- if (len != dev->fmr_page_size) {
- ret = srp_map_finish_fmr(state, target);
+ if (len != dev->mr_page_size) {
+ ret = srp_finish_mapping(state, target);
if (!ret)
srp_map_update_start(state, NULL, 0, 0);
}
return ret;
}
+static int srp_map_sg(struct srp_map_state *state,
+ struct srp_target_port *target, struct srp_request *req,
+ struct scatterlist *scat, int count)
+{
+ struct srp_device *dev = target->srp_host->srp_dev;
+ struct ib_device *ibdev = dev->dev;
+ struct scatterlist *sg;
+ int i;
+ bool use_mr;
+
+ state->desc = req->indirect_desc;
+ state->pages = req->map_page;
+ if (dev->use_fast_reg) {
+ state->next_fr = req->fr_list;
+ use_mr = !!target->fr_pool;
+ } else {
+ state->next_fmr = req->fmr_list;
+ use_mr = !!target->fmr_pool;
+ }
+
+ for_each_sg(scat, sg, count, i) {
+ if (srp_map_sg_entry(state, target, sg, i, use_mr)) {
+ /*
+ * Memory registration failed, so backtrack to the
+ * first unmapped entry and continue on without using
+ * memory registration.
+ */
+ dma_addr_t dma_addr;
+ unsigned int dma_len;
+
+backtrack:
+ sg = state->unmapped_sg;
+ i = state->unmapped_index;
+
+ dma_addr = ib_sg_dma_address(ibdev, sg);
+ dma_len = ib_sg_dma_len(ibdev, sg);
+ dma_len -= (state->unmapped_addr - dma_addr);
+ dma_addr = state->unmapped_addr;
+ use_mr = false;
+ srp_map_desc(state, dma_addr, dma_len, target->rkey);
+ }
+ }
+
+ if (use_mr && srp_finish_mapping(state, target))
+ goto backtrack;
+
+ req->nmdesc = state->nmdesc;
+
+ return 0;
+}
+
static int srp_map_data(struct scsi_cmnd *scmnd, struct srp_target_port *target,
struct srp_request *req)
{
- struct scatterlist *scat, *sg;
+ struct scatterlist *scat;
struct srp_cmd *cmd = req->cmd->buf;
- int i, len, nents, count, use_fmr;
+ int len, nents, count;
struct srp_device *dev;
struct ib_device *ibdev;
struct srp_map_state state;
@@ -1081,7 +1429,7 @@ static int srp_map_data(struct scsi_cmnd *scmnd, struct srp_target_port *target,
fmt = SRP_DATA_DESC_DIRECT;
len = sizeof (struct srp_cmd) + sizeof (struct srp_direct_buf);
- if (count == 1) {
+ if (count == 1 && !register_always) {
/*
* The midlayer only generated a single gather/scatter
* entry, or DMA mapping coalesced everything to a
@@ -1094,13 +1442,13 @@ static int srp_map_data(struct scsi_cmnd *scmnd, struct srp_target_port *target,
buf->key = cpu_to_be32(target->rkey);
buf->len = cpu_to_be32(ib_sg_dma_len(ibdev, scat));
- req->nfmr = 0;
+ req->nmdesc = 0;
goto map_complete;
}
- /* We have more than one scatter/gather entry, so build our indirect
- * descriptor table, trying to merge as many entries with FMR as we
- * can.
+ /*
+ * We have more than one scatter/gather entry, so build our indirect
+ * descriptor table, trying to merge as many entries as we can.
*/
indirect_hdr = (void *) cmd->add_data;
@@ -1108,35 +1456,7 @@ static int srp_map_data(struct scsi_cmnd *scmnd, struct srp_target_port *target,
target->indirect_size, DMA_TO_DEVICE);
memset(&state, 0, sizeof(state));
- state.desc = req->indirect_desc;
- state.pages = req->map_page;
- state.next_fmr = req->fmr_list;
-
- use_fmr = dev->fmr_pool ? SRP_MAP_ALLOW_FMR : SRP_MAP_NO_FMR;
-
- for_each_sg(scat, sg, count, i) {
- if (srp_map_sg_entry(&state, target, sg, i, use_fmr)) {
- /* FMR mapping failed, so backtrack to the first
- * unmapped entry and continue on without using FMR.
- */
- dma_addr_t dma_addr;
- unsigned int dma_len;
-
-backtrack:
- sg = state.unmapped_sg;
- i = state.unmapped_index;
-
- dma_addr = ib_sg_dma_address(ibdev, sg);
- dma_len = ib_sg_dma_len(ibdev, sg);
- dma_len -= (state.unmapped_addr - dma_addr);
- dma_addr = state.unmapped_addr;
- use_fmr = SRP_MAP_NO_FMR;
- srp_map_desc(&state, dma_addr, dma_len, target->rkey);
- }
- }
-
- if (use_fmr == SRP_MAP_ALLOW_FMR && srp_map_finish_fmr(&state, target))
- goto backtrack;
+ srp_map_sg(&state, target, req, scat, count);
/* We've mapped the request, now pull as much of the indirect
* descriptor table as we can into the command buffer. If this
@@ -1144,9 +1464,9 @@ backtrack:
* guaranteed to fit into the command, as the SCSI layer won't
* give us more S/G entries than we allow.
*/
- req->nfmr = state.nfmr;
if (state.ndesc == 1) {
- /* FMR mapping was able to collapse this to one entry,
+ /*
+ * Memory registration collapsed the sg-list into one entry,
* so use a direct descriptor.
*/
struct srp_direct_buf *buf = (void *) cmd->add_data;
@@ -1455,6 +1775,7 @@ static void srp_handle_recv(struct srp_target_port *target, struct ib_wc *wc)
/**
* srp_tl_err_work() - handle a transport layer error
+ * @work: Work structure embedded in an SRP target port.
*
* Note: This function may get invoked before the rport has been created,
* hence the target->rport test.
@@ -1468,14 +1789,24 @@ static void srp_tl_err_work(struct work_struct *work)
srp_start_tl_fail_timers(target->rport);
}
-static void srp_handle_qp_err(enum ib_wc_status wc_status, bool send_err,
- struct srp_target_port *target)
+static void srp_handle_qp_err(u64 wr_id, enum ib_wc_status wc_status,
+ bool send_err, struct srp_target_port *target)
{
if (target->connected && !target->qp_in_error) {
- shost_printk(KERN_ERR, target->scsi_host,
- PFX "failed %s status %d\n",
- send_err ? "send" : "receive",
- wc_status);
+ if (wr_id & LOCAL_INV_WR_ID_MASK) {
+ shost_printk(KERN_ERR, target->scsi_host, PFX
+ "LOCAL_INV failed with status %d\n",
+ wc_status);
+ } else if (wr_id & FAST_REG_WR_ID_MASK) {
+ shost_printk(KERN_ERR, target->scsi_host, PFX
+ "FAST_REG_MR failed status %d\n",
+ wc_status);
+ } else {
+ shost_printk(KERN_ERR, target->scsi_host,
+ PFX "failed %s status %d for iu %p\n",
+ send_err ? "send" : "receive",
+ wc_status, (void *)(uintptr_t)wr_id);
+ }
queue_work(system_long_wq, &target->tl_err_work);
}
target->qp_in_error = true;
@@ -1491,7 +1822,7 @@ static void srp_recv_completion(struct ib_cq *cq, void *target_ptr)
if (likely(wc.status == IB_WC_SUCCESS)) {
srp_handle_recv(target, &wc);
} else {
- srp_handle_qp_err(wc.status, false, target);
+ srp_handle_qp_err(wc.wr_id, wc.status, false, target);
}
}
}
@@ -1507,7 +1838,7 @@ static void srp_send_completion(struct ib_cq *cq, void *target_ptr)
iu = (struct srp_iu *) (uintptr_t) wc.wr_id;
list_add(&iu->list, &target->free_tx);
} else {
- srp_handle_qp_err(wc.status, true, target);
+ srp_handle_qp_err(wc.wr_id, wc.status, true, target);
}
}
}
@@ -1521,7 +1852,7 @@ static int srp_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *scmnd)
struct srp_cmd *cmd;
struct ib_device *dev;
unsigned long flags;
- int len, result;
+ int len, ret;
const bool in_scsi_eh = !in_interrupt() && current == shost->ehandler;
/*
@@ -1533,12 +1864,9 @@ static int srp_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *scmnd)
if (in_scsi_eh)
mutex_lock(&rport->mutex);
- result = srp_chkready(target->rport);
- if (unlikely(result)) {
- scmnd->result = result;
- scmnd->scsi_done(scmnd);
- goto unlock_rport;
- }
+ scmnd->result = srp_chkready(target->rport);
+ if (unlikely(scmnd->result))
+ goto err;
spin_lock_irqsave(&target->lock, flags);
iu = __srp_get_tx_iu(target, SRP_IU_CMD);
@@ -1553,7 +1881,6 @@ static int srp_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *scmnd)
ib_dma_sync_single_for_cpu(dev, iu->dma, target->max_iu_len,
DMA_TO_DEVICE);
- scmnd->result = 0;
scmnd->host_scribble = (void *) req;
cmd = iu->buf;
@@ -1570,7 +1897,15 @@ static int srp_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *scmnd)
len = srp_map_data(scmnd, target, req);
if (len < 0) {
shost_printk(KERN_ERR, target->scsi_host,
- PFX "Failed to map data\n");
+ PFX "Failed to map data (%d)\n", len);
+ /*
+ * If we ran out of memory descriptors (-ENOMEM) because an
+ * application is queuing many requests with more than
+ * max_pages_per_mr sg-list elements, tell the SCSI mid-layer
+ * to reduce queue depth temporarily.
+ */
+ scmnd->result = len == -ENOMEM ?
+ DID_OK << 16 | QUEUE_FULL << 1 : DID_ERROR << 16;
goto err_iu;
}
@@ -1582,11 +1917,13 @@ static int srp_queuecommand(struct Scsi_Host *shost, struct scsi_cmnd *scmnd)
goto err_unmap;
}
+ ret = 0;
+
unlock_rport:
if (in_scsi_eh)
mutex_unlock(&rport->mutex);
- return 0;
+ return ret;
err_unmap:
srp_unmap_data(scmnd, target, req);
@@ -1594,16 +1931,27 @@ err_unmap:
err_iu:
srp_put_tx_iu(target, iu, SRP_IU_CMD);
+ /*
+ * Avoid that the loops that iterate over the request ring can
+ * encounter a dangling SCSI command pointer.
+ */
+ req->scmnd = NULL;
+
spin_lock_irqsave(&target->lock, flags);
list_add(&req->list, &target->free_reqs);
err_unlock:
spin_unlock_irqrestore(&target->lock, flags);
- if (in_scsi_eh)
- mutex_unlock(&rport->mutex);
+err:
+ if (scmnd->result) {
+ scmnd->scsi_done(scmnd);
+ ret = 0;
+ } else {
+ ret = SCSI_MLQUEUE_HOST_BUSY;
+ }
- return SCSI_MLQUEUE_HOST_BUSY;
+ goto unlock_rport;
}
/*
@@ -2310,6 +2658,8 @@ static struct class srp_class = {
/**
* srp_conn_unique() - check whether the connection to a target is unique
+ * @host: SRP host.
+ * @target: SRP target port.
*/
static bool srp_conn_unique(struct srp_host *host,
struct srp_target_port *target)
@@ -2605,7 +2955,8 @@ static ssize_t srp_create_target(struct device *dev,
container_of(dev, struct srp_host, dev);
struct Scsi_Host *target_host;
struct srp_target_port *target;
- struct ib_device *ibdev = host->srp_dev->dev;
+ struct srp_device *srp_dev = host->srp_dev;
+ struct ib_device *ibdev = srp_dev->dev;
int ret;
target_host = scsi_host_alloc(&srp_template,
@@ -2650,9 +3001,9 @@ static ssize_t srp_create_target(struct device *dev,
goto err;
}
- if (!host->srp_dev->fmr_pool && !target->allow_ext_sg &&
- target->cmd_sg_cnt < target->sg_tablesize) {
- pr_warn("No FMR pool and no external indirect descriptors, limiting sg_tablesize to cmd_sg_cnt\n");
+ if (!srp_dev->has_fmr && !srp_dev->has_fr && !target->allow_ext_sg &&
+ target->cmd_sg_cnt < target->sg_tablesize) {
+ pr_warn("No MR pool and no external indirect descriptors, limiting sg_tablesize to cmd_sg_cnt\n");
target->sg_tablesize = target->cmd_sg_cnt;
}
@@ -2790,9 +3141,9 @@ static void srp_add_one(struct ib_device *device)
{
struct srp_device *srp_dev;
struct ib_device_attr *dev_attr;
- struct ib_fmr_pool_param fmr_param;
struct srp_host *host;
- int max_pages_per_fmr, fmr_page_shift, s, e, p;
+ int mr_page_shift, s, e, p;
+ u64 max_pages_per_mr;
dev_attr = kmalloc(sizeof *dev_attr, GFP_KERNEL);
if (!dev_attr)
@@ -2807,15 +3158,39 @@ static void srp_add_one(struct ib_device *device)
if (!srp_dev)
goto free_attr;
+ srp_dev->has_fmr = (device->alloc_fmr && device->dealloc_fmr &&
+ device->map_phys_fmr && device->unmap_fmr);
+ srp_dev->has_fr = (dev_attr->device_cap_flags &
+ IB_DEVICE_MEM_MGT_EXTENSIONS);
+ if (!srp_dev->has_fmr && !srp_dev->has_fr)
+ dev_warn(&device->dev, "neither FMR nor FR is supported\n");
+
+ srp_dev->use_fast_reg = (srp_dev->has_fr &&
+ (!srp_dev->has_fmr || prefer_fr));
+
/*
* Use the smallest page size supported by the HCA, down to a
* minimum of 4096 bytes. We're unlikely to build large sglists
* out of smaller entries.
*/
- fmr_page_shift = max(12, ffs(dev_attr->page_size_cap) - 1);
- srp_dev->fmr_page_size = 1 << fmr_page_shift;
- srp_dev->fmr_page_mask = ~((u64) srp_dev->fmr_page_size - 1);
- srp_dev->fmr_max_size = srp_dev->fmr_page_size * SRP_FMR_SIZE;
+ mr_page_shift = max(12, ffs(dev_attr->page_size_cap) - 1);
+ srp_dev->mr_page_size = 1 << mr_page_shift;
+ srp_dev->mr_page_mask = ~((u64) srp_dev->mr_page_size - 1);
+ max_pages_per_mr = dev_attr->max_mr_size;
+ do_div(max_pages_per_mr, srp_dev->mr_page_size);
+ srp_dev->max_pages_per_mr = min_t(u64, SRP_MAX_PAGES_PER_MR,
+ max_pages_per_mr);
+ if (srp_dev->use_fast_reg) {
+ srp_dev->max_pages_per_mr =
+ min_t(u32, srp_dev->max_pages_per_mr,
+ dev_attr->max_fast_reg_page_list_len);
+ }
+ srp_dev->mr_max_size = srp_dev->mr_page_size *
+ srp_dev->max_pages_per_mr;
+ pr_debug("%s: mr_page_shift = %d, dev_attr->max_mr_size = %#llx, dev_attr->max_fast_reg_page_list_len = %u, max_pages_per_mr = %d, mr_max_size = %#x\n",
+ device->name, mr_page_shift, dev_attr->max_mr_size,
+ dev_attr->max_fast_reg_page_list_len,
+ srp_dev->max_pages_per_mr, srp_dev->mr_max_size);
INIT_LIST_HEAD(&srp_dev->dev_list);
@@ -2831,27 +3206,6 @@ static void srp_add_one(struct ib_device *device)
if (IS_ERR(srp_dev->mr))
goto err_pd;
- for (max_pages_per_fmr = SRP_FMR_SIZE;
- max_pages_per_fmr >= SRP_FMR_MIN_SIZE;
- max_pages_per_fmr /= 2, srp_dev->fmr_max_size /= 2) {
- memset(&fmr_param, 0, sizeof fmr_param);
- fmr_param.pool_size = SRP_FMR_POOL_SIZE;
- fmr_param.dirty_watermark = SRP_FMR_DIRTY_SIZE;
- fmr_param.cache = 1;
- fmr_param.max_pages_per_fmr = max_pages_per_fmr;
- fmr_param.page_shift = fmr_page_shift;
- fmr_param.access = (IB_ACCESS_LOCAL_WRITE |
- IB_ACCESS_REMOTE_WRITE |
- IB_ACCESS_REMOTE_READ);
-
- srp_dev->fmr_pool = ib_create_fmr_pool(srp_dev->pd, &fmr_param);
- if (!IS_ERR(srp_dev->fmr_pool))
- break;
- }
-
- if (IS_ERR(srp_dev->fmr_pool))
- srp_dev->fmr_pool = NULL;
-
if (device->node_type == RDMA_NODE_IB_SWITCH) {
s = 0;
e = 0;
@@ -2914,8 +3268,6 @@ static void srp_remove_one(struct ib_device *device)
kfree(host);
}
- if (srp_dev->fmr_pool)
- ib_destroy_fmr_pool(srp_dev->fmr_pool);
ib_dereg_mr(srp_dev->mr);
ib_dealloc_pd(srp_dev->pd);
diff --git a/drivers/infiniband/ulp/srp/ib_srp.h b/drivers/infiniband/ulp/srp/ib_srp.h
index aad27b7b4a4..e46ecb15aa0 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.h
+++ b/drivers/infiniband/ulp/srp/ib_srp.h
@@ -66,13 +66,10 @@ enum {
SRP_TAG_NO_REQ = ~0U,
SRP_TAG_TSK_MGMT = 1U << 31,
- SRP_FMR_SIZE = 512,
- SRP_FMR_MIN_SIZE = 128,
- SRP_FMR_POOL_SIZE = 1024,
- SRP_FMR_DIRTY_SIZE = SRP_FMR_POOL_SIZE / 4,
+ SRP_MAX_PAGES_PER_MR = 512,
- SRP_MAP_ALLOW_FMR = 0,
- SRP_MAP_NO_FMR = 1,
+ LOCAL_INV_WR_ID_MASK = 1,
+ FAST_REG_WR_ID_MASK = 2,
};
enum srp_target_state {
@@ -86,15 +83,24 @@ enum srp_iu_type {
SRP_IU_RSP,
};
+/*
+ * @mr_page_mask: HCA memory registration page mask.
+ * @mr_page_size: HCA memory registration page size.
+ * @mr_max_size: Maximum size in bytes of a single FMR / FR registration
+ * request.
+ */
struct srp_device {
struct list_head dev_list;
struct ib_device *dev;
struct ib_pd *pd;
struct ib_mr *mr;
- struct ib_fmr_pool *fmr_pool;
- u64 fmr_page_mask;
- int fmr_page_size;
- int fmr_max_size;
+ u64 mr_page_mask;
+ int mr_page_size;
+ int mr_max_size;
+ int max_pages_per_mr;
+ bool has_fmr;
+ bool has_fr;
+ bool use_fast_reg;
};
struct srp_host {
@@ -112,11 +118,14 @@ struct srp_request {
struct list_head list;
struct scsi_cmnd *scmnd;
struct srp_iu *cmd;
- struct ib_pool_fmr **fmr_list;
+ union {
+ struct ib_pool_fmr **fmr_list;
+ struct srp_fr_desc **fr_list;
+ };
u64 *map_page;
struct srp_direct_buf *indirect_desc;
dma_addr_t indirect_dma_addr;
- short nfmr;
+ short nmdesc;
short index;
};
@@ -131,6 +140,10 @@ struct srp_target_port {
struct ib_cq *send_cq ____cacheline_aligned_in_smp;
struct ib_cq *recv_cq;
struct ib_qp *qp;
+ union {
+ struct ib_fmr_pool *fmr_pool;
+ struct srp_fr_pool *fr_pool;
+ };
u32 lkey;
u32 rkey;
enum srp_target_state state;
@@ -197,15 +210,66 @@ struct srp_iu {
enum dma_data_direction direction;
};
+/**
+ * struct srp_fr_desc - fast registration work request arguments
+ * @entry: Entry in srp_fr_pool.free_list.
+ * @mr: Memory region.
+ * @frpl: Fast registration page list.
+ */
+struct srp_fr_desc {
+ struct list_head entry;
+ struct ib_mr *mr;
+ struct ib_fast_reg_page_list *frpl;
+};
+
+/**
+ * struct srp_fr_pool - pool of fast registration descriptors
+ *
+ * An entry is available for allocation if and only if it occurs in @free_list.
+ *
+ * @size: Number of descriptors in this pool.
+ * @max_page_list_len: Maximum fast registration work request page list length.
+ * @lock: Protects free_list.
+ * @free_list: List of free descriptors.
+ * @desc: Fast registration descriptor pool.
+ */
+struct srp_fr_pool {
+ int size;
+ int max_page_list_len;
+ spinlock_t lock;
+ struct list_head free_list;
+ struct srp_fr_desc desc[0];
+};
+
+/**
+ * struct srp_map_state - per-request DMA memory mapping state
+ * @desc: Pointer to the element of the SRP buffer descriptor array
+ * that is being filled in.
+ * @pages: Array with DMA addresses of pages being considered for
+ * memory registration.
+ * @base_dma_addr: DMA address of the first page that has not yet been mapped.
+ * @dma_len: Number of bytes that will be registered with the next
+ * FMR or FR memory registration call.
+ * @total_len: Total number of bytes in the sg-list being mapped.
+ * @npages: Number of page addresses in the pages[] array.
+ * @nmdesc: Number of FMR or FR memory descriptors used for mapping.
+ * @ndesc: Number of SRP buffer descriptors that have been filled in.
+ * @unmapped_sg: First element of the sg-list that is mapped via FMR or FR.
+ * @unmapped_index: Index of the first element mapped via FMR or FR.
+ * @unmapped_addr: DMA address of the first element mapped via FMR or FR.
+ */
struct srp_map_state {
- struct ib_pool_fmr **next_fmr;
+ union {
+ struct ib_pool_fmr **next_fmr;
+ struct srp_fr_desc **next_fr;
+ };
struct srp_direct_buf *desc;
u64 *pages;
dma_addr_t base_dma_addr;
- u32 fmr_len;
+ u32 dma_len;
u32 total_len;
unsigned int npages;
- unsigned int nfmr;
+ unsigned int nmdesc;
unsigned int ndesc;
struct scatterlist *unmapped_sg;
int unmapped_index;
diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 0e537d8d0e4..fe09f2788b1 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -1078,6 +1078,7 @@ static void srpt_unmap_sg_to_ib_sge(struct srpt_rdma_ch *ch,
static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
struct srpt_send_ioctx *ioctx)
{
+ struct ib_device *dev = ch->sport->sdev->device;
struct se_cmd *cmd;
struct scatterlist *sg, *sg_orig;
int sg_cnt;
@@ -1124,7 +1125,7 @@ static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
db = ioctx->rbufs;
tsize = cmd->data_length;
- dma_len = sg_dma_len(&sg[0]);
+ dma_len = ib_sg_dma_len(dev, &sg[0]);
riu = ioctx->rdma_ius;
/*
@@ -1155,7 +1156,8 @@ static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
++j;
if (j < count) {
sg = sg_next(sg);
- dma_len = sg_dma_len(sg);
+ dma_len = ib_sg_dma_len(
+ dev, sg);
}
}
} else {
@@ -1192,8 +1194,8 @@ static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
tsize = cmd->data_length;
riu = ioctx->rdma_ius;
sg = sg_orig;
- dma_len = sg_dma_len(&sg[0]);
- dma_addr = sg_dma_address(&sg[0]);
+ dma_len = ib_sg_dma_len(dev, &sg[0]);
+ dma_addr = ib_sg_dma_address(dev, &sg[0]);
/* this second loop is really mapped sg_addres to rdma_iu->ib_sge */
for (i = 0, j = 0;
@@ -1216,8 +1218,10 @@ static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
++j;
if (j < count) {
sg = sg_next(sg);
- dma_len = sg_dma_len(sg);
- dma_addr = sg_dma_address(sg);
+ dma_len = ib_sg_dma_len(
+ dev, sg);
+ dma_addr = ib_sg_dma_address(
+ dev, sg);
}
}
} else {
@@ -2580,7 +2584,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
goto destroy_ib;
}
- ch->sess = transport_init_session();
+ ch->sess = transport_init_session(TARGET_PROT_NORMAL);
if (IS_ERR(ch->sess)) {
rej->reason = __constant_cpu_to_be32(
SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
@@ -3081,6 +3085,14 @@ static void srpt_queue_tm_rsp(struct se_cmd *cmd)
srpt_queue_response(cmd);
}
+static void srpt_aborted_task(struct se_cmd *cmd)
+{
+ struct srpt_send_ioctx *ioctx = container_of(cmd,
+ struct srpt_send_ioctx, cmd);
+
+ srpt_unmap_sg_to_ib_sge(ioctx->ch, ioctx);
+}
+
static int srpt_queue_status(struct se_cmd *cmd)
{
struct srpt_send_ioctx *ioctx;
@@ -3928,6 +3940,7 @@ static struct target_core_fabric_ops srpt_template = {
.queue_data_in = srpt_queue_data_in,
.queue_status = srpt_queue_status,
.queue_tm_rsp = srpt_queue_tm_rsp,
+ .aborted_task = srpt_aborted_task,
/*
* Setup function pointers for generic logic in
* target_core_fabric_configfs.c
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-12 04:06:34 +0000