summaryrefslogtreecommitdiff
path: root/drivers/infiniband/ulp
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/infiniband/ulp')
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib.h4
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_cm.c23
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_ib.c45
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_main.c65
-rw-r--r--drivers/infiniband/ulp/ipoib/ipoib_verbs.c5
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.c25
-rw-r--r--drivers/infiniband/ulp/iser/iscsi_iser.h7
-rw-r--r--drivers/infiniband/ulp/iser/iser_initiator.c7
-rw-r--r--drivers/infiniband/ulp/iser/iser_verbs.c38
-rw-r--r--drivers/infiniband/ulp/isert/ib_isert.c713
-rw-r--r--drivers/infiniband/ulp/isert/ib_isert.h72
-rw-r--r--drivers/infiniband/ulp/srp/ib_srp.c54
-rw-r--r--drivers/infiniband/ulp/srpt/ib_srpt.c878
-rw-r--r--drivers/infiniband/ulp/srpt/ib_srpt.h31
14 files changed, 843 insertions, 1124 deletions
diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h
index a6f3eab0f..caec8e9c4 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib.h
+++ b/drivers/infiniband/ulp/ipoib/ipoib.h
@@ -244,6 +244,7 @@ struct ipoib_cm_tx {
unsigned tx_tail;
unsigned long flags;
u32 mtu;
+ unsigned max_send_sge;
};
struct ipoib_cm_rx_buf {
@@ -387,9 +388,10 @@ struct ipoib_dev_priv {
struct dentry *mcg_dentry;
struct dentry *path_dentry;
#endif
- int hca_caps;
+ u64 hca_caps;
struct ipoib_ethtool_st ethtool;
struct timer_list poll_timer;
+ unsigned max_send_sge;
};
struct ipoib_ah {
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_cm.c b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
index 917e46ea3..c8ed53562 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_cm.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_cm.c
@@ -710,6 +710,7 @@ void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_
struct ipoib_dev_priv *priv = netdev_priv(dev);
struct ipoib_tx_buf *tx_req;
int rc;
+ unsigned usable_sge = tx->max_send_sge - !!skb_headlen(skb);
if (unlikely(skb->len > tx->mtu)) {
ipoib_warn(priv, "packet len %d (> %d) too long to send, dropping\n",
@@ -719,7 +720,23 @@ void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_
ipoib_cm_skb_too_long(dev, skb, tx->mtu - IPOIB_ENCAP_LEN);
return;
}
-
+ if (skb_shinfo(skb)->nr_frags > usable_sge) {
+ if (skb_linearize(skb) < 0) {
+ ipoib_warn(priv, "skb could not be linearized\n");
+ ++dev->stats.tx_dropped;
+ ++dev->stats.tx_errors;
+ dev_kfree_skb_any(skb);
+ return;
+ }
+ /* Does skb_linearize return ok without reducing nr_frags? */
+ if (skb_shinfo(skb)->nr_frags > usable_sge) {
+ ipoib_warn(priv, "too many frags after skb linearize\n");
+ ++dev->stats.tx_dropped;
+ ++dev->stats.tx_errors;
+ dev_kfree_skb_any(skb);
+ return;
+ }
+ }
ipoib_dbg_data(priv, "sending packet: head 0x%x length %d connection 0x%x\n",
tx->tx_head, skb->len, tx->qp->qp_num);
@@ -1031,7 +1048,8 @@ static struct ib_qp *ipoib_cm_create_tx_qp(struct net_device *dev, struct ipoib_
struct ib_qp *tx_qp;
if (dev->features & NETIF_F_SG)
- attr.cap.max_send_sge = MAX_SKB_FRAGS + 1;
+ attr.cap.max_send_sge =
+ min_t(u32, priv->ca->attrs.max_sge, MAX_SKB_FRAGS + 1);
tx_qp = ib_create_qp(priv->pd, &attr);
if (PTR_ERR(tx_qp) == -EINVAL) {
@@ -1040,6 +1058,7 @@ static struct ib_qp *ipoib_cm_create_tx_qp(struct net_device *dev, struct ipoib_
attr.create_flags &= ~IB_QP_CREATE_USE_GFP_NOIO;
tx_qp = ib_create_qp(priv->pd, &attr);
}
+ tx->max_send_sge = attr.cap.max_send_sge;
return tx_qp;
}
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
index fa9c42ff1..f0e55e47e 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c
@@ -180,6 +180,7 @@ static void ipoib_ib_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
struct sk_buff *skb;
u64 mapping[IPOIB_UD_RX_SG];
union ib_gid *dgid;
+ union ib_gid *sgid;
ipoib_dbg_data(priv, "recv completion: id %d, status: %d\n",
wr_id, wc->status);
@@ -203,13 +204,6 @@ static void ipoib_ib_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
return;
}
- /*
- * Drop packets that this interface sent, ie multicast packets
- * that the HCA has replicated.
- */
- if (wc->slid == priv->local_lid && wc->src_qp == priv->qp->qp_num)
- goto repost;
-
memcpy(mapping, priv->rx_ring[wr_id].mapping,
IPOIB_UD_RX_SG * sizeof *mapping);
@@ -239,6 +233,25 @@ static void ipoib_ib_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
else
skb->pkt_type = PACKET_MULTICAST;
+ sgid = &((struct ib_grh *)skb->data)->sgid;
+
+ /*
+ * Drop packets that this interface sent, ie multicast packets
+ * that the HCA has replicated.
+ */
+ if (wc->slid == priv->local_lid && wc->src_qp == priv->qp->qp_num) {
+ int need_repost = 1;
+
+ if ((wc->wc_flags & IB_WC_GRH) &&
+ sgid->global.interface_id != priv->local_gid.global.interface_id)
+ need_repost = 0;
+
+ if (need_repost) {
+ dev_kfree_skb_any(skb);
+ goto repost;
+ }
+ }
+
skb_pull(skb, IB_GRH_BYTES);
skb->protocol = ((struct ipoib_header *) skb->data)->proto;
@@ -538,6 +551,7 @@ void ipoib_send(struct net_device *dev, struct sk_buff *skb,
struct ipoib_tx_buf *tx_req;
int hlen, rc;
void *phead;
+ unsigned usable_sge = priv->max_send_sge - !!skb_headlen(skb);
if (skb_is_gso(skb)) {
hlen = skb_transport_offset(skb) + tcp_hdrlen(skb);
@@ -561,6 +575,23 @@ void ipoib_send(struct net_device *dev, struct sk_buff *skb,
phead = NULL;
hlen = 0;
}
+ if (skb_shinfo(skb)->nr_frags > usable_sge) {
+ if (skb_linearize(skb) < 0) {
+ ipoib_warn(priv, "skb could not be linearized\n");
+ ++dev->stats.tx_dropped;
+ ++dev->stats.tx_errors;
+ dev_kfree_skb_any(skb);
+ return;
+ }
+ /* Does skb_linearize return ok without reducing nr_frags? */
+ if (skb_shinfo(skb)->nr_frags > usable_sge) {
+ ipoib_warn(priv, "too many frags after skb linearize\n");
+ ++dev->stats.tx_dropped;
+ ++dev->stats.tx_errors;
+ dev_kfree_skb_any(skb);
+ return;
+ }
+ }
ipoib_dbg_data(priv, "sending packet, length=%d address=%p qpn=0x%06x\n",
skb->len, address, qpn);
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 25509bbd4..80807d6e5 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -51,6 +51,7 @@
#include <net/addrconf.h>
#include <linux/inetdevice.h>
#include <rdma/ib_cache.h>
+#include <linux/pci.h>
#define DRV_VERSION "1.0.0"
@@ -1590,11 +1591,67 @@ void ipoib_dev_cleanup(struct net_device *dev)
priv->tx_ring = NULL;
}
+static int ipoib_set_vf_link_state(struct net_device *dev, int vf, int link_state)
+{
+ struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+ return ib_set_vf_link_state(priv->ca, vf, priv->port, link_state);
+}
+
+static int ipoib_get_vf_config(struct net_device *dev, int vf,
+ struct ifla_vf_info *ivf)
+{
+ struct ipoib_dev_priv *priv = netdev_priv(dev);
+ int err;
+
+ err = ib_get_vf_config(priv->ca, vf, priv->port, ivf);
+ if (err)
+ return err;
+
+ ivf->vf = vf;
+
+ return 0;
+}
+
+static int ipoib_set_vf_guid(struct net_device *dev, int vf, u64 guid, int type)
+{
+ struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+ if (type != IFLA_VF_IB_NODE_GUID && type != IFLA_VF_IB_PORT_GUID)
+ return -EINVAL;
+
+ return ib_set_vf_guid(priv->ca, vf, priv->port, guid, type);
+}
+
+static int ipoib_get_vf_stats(struct net_device *dev, int vf,
+ struct ifla_vf_stats *vf_stats)
+{
+ struct ipoib_dev_priv *priv = netdev_priv(dev);
+
+ return ib_get_vf_stats(priv->ca, vf, priv->port, vf_stats);
+}
+
static const struct header_ops ipoib_header_ops = {
.create = ipoib_hard_header,
};
-static const struct net_device_ops ipoib_netdev_ops = {
+static const struct net_device_ops ipoib_netdev_ops_pf = {
+ .ndo_uninit = ipoib_uninit,
+ .ndo_open = ipoib_open,
+ .ndo_stop = ipoib_stop,
+ .ndo_change_mtu = ipoib_change_mtu,
+ .ndo_fix_features = ipoib_fix_features,
+ .ndo_start_xmit = ipoib_start_xmit,
+ .ndo_tx_timeout = ipoib_timeout,
+ .ndo_set_rx_mode = ipoib_set_mcast_list,
+ .ndo_get_iflink = ipoib_get_iflink,
+ .ndo_set_vf_link_state = ipoib_set_vf_link_state,
+ .ndo_get_vf_config = ipoib_get_vf_config,
+ .ndo_get_vf_stats = ipoib_get_vf_stats,
+ .ndo_set_vf_guid = ipoib_set_vf_guid,
+};
+
+static const struct net_device_ops ipoib_netdev_ops_vf = {
.ndo_uninit = ipoib_uninit,
.ndo_open = ipoib_open,
.ndo_stop = ipoib_stop,
@@ -1610,7 +1667,11 @@ void ipoib_setup(struct net_device *dev)
{
struct ipoib_dev_priv *priv = netdev_priv(dev);
- dev->netdev_ops = &ipoib_netdev_ops;
+ if (priv->hca_caps & IB_DEVICE_VIRTUAL_FUNCTION)
+ dev->netdev_ops = &ipoib_netdev_ops_vf;
+ else
+ dev->netdev_ops = &ipoib_netdev_ops_pf;
+
dev->header_ops = &ipoib_header_ops;
ipoib_set_ethtool_ops(dev);
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
index d48c5bae7..b809c373e 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_verbs.c
@@ -206,7 +206,8 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca)
init_attr.create_flags |= IB_QP_CREATE_NETIF_QP;
if (dev->features & NETIF_F_SG)
- init_attr.cap.max_send_sge = MAX_SKB_FRAGS + 1;
+ init_attr.cap.max_send_sge =
+ min_t(u32, priv->ca->attrs.max_sge, MAX_SKB_FRAGS + 1);
priv->qp = ib_create_qp(priv->pd, &init_attr);
if (IS_ERR(priv->qp)) {
@@ -233,6 +234,8 @@ int ipoib_transport_dev_init(struct net_device *dev, struct ib_device *ca)
priv->rx_wr.next = NULL;
priv->rx_wr.sg_list = priv->rx_sge;
+ priv->max_send_sge = init_attr.cap.max_send_sge;
+
return 0;
out_free_send_cq:
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c
index c827c93f4..64b3d11dc 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.c
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.c
@@ -612,6 +612,7 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
struct Scsi_Host *shost;
struct iser_conn *iser_conn = NULL;
struct ib_conn *ib_conn;
+ u32 max_fr_sectors;
u16 max_cmds;
shost = iscsi_host_alloc(&iscsi_iser_sht, 0, 0);
@@ -632,7 +633,6 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
iser_conn = ep->dd_data;
max_cmds = iser_conn->max_cmds;
shost->sg_tablesize = iser_conn->scsi_sg_tablesize;
- shost->max_sectors = iser_conn->scsi_max_sectors;
mutex_lock(&iser_conn->state_mutex);
if (iser_conn->state != ISER_CONN_UP) {
@@ -657,8 +657,6 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
*/
shost->sg_tablesize = min_t(unsigned short, shost->sg_tablesize,
ib_conn->device->ib_device->attrs.max_fast_reg_page_list_len);
- shost->max_sectors = min_t(unsigned int,
- 1024, (shost->sg_tablesize * PAGE_SIZE) >> 9);
if (iscsi_host_add(shost,
ib_conn->device->ib_device->dma_device)) {
@@ -672,6 +670,15 @@ iscsi_iser_session_create(struct iscsi_endpoint *ep,
goto free_host;
}
+ /*
+ * FRs or FMRs can only map up to a (device) page per entry, but if the
+ * first entry is misaligned we'll end up using using two entries
+ * (head and tail) for a single page worth data, so we have to drop
+ * one segment from the calculation.
+ */
+ max_fr_sectors = ((shost->sg_tablesize - 1) * PAGE_SIZE) >> 9;
+ shost->max_sectors = min(iser_max_sectors, max_fr_sectors);
+
if (cmds_max > max_cmds) {
iser_info("cmds_max changed from %u to %u\n",
cmds_max, max_cmds);
@@ -969,7 +976,16 @@ static umode_t iser_attr_is_visible(int param_type, int param)
static int iscsi_iser_slave_alloc(struct scsi_device *sdev)
{
- blk_queue_virt_boundary(sdev->request_queue, ~MASK_4K);
+ struct iscsi_session *session;
+ struct iser_conn *iser_conn;
+ struct ib_device *ib_dev;
+
+ session = starget_to_session(scsi_target(sdev))->dd_data;
+ iser_conn = session->leadconn->dd_data;
+ ib_dev = iser_conn->ib_conn.device->ib_device;
+
+ if (!(ib_dev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG))
+ blk_queue_virt_boundary(sdev->request_queue, ~MASK_4K);
return 0;
}
@@ -980,7 +996,6 @@ static struct scsi_host_template iscsi_iser_sht = {
.queuecommand = iscsi_queuecommand,
.change_queue_depth = scsi_change_queue_depth,
.sg_tablesize = ISCSI_ISER_DEF_SG_TABLESIZE,
- .max_sectors = ISER_DEF_MAX_SECTORS,
.cmd_per_lun = ISER_DEF_CMD_PER_LUN,
.eh_abort_handler = iscsi_eh_abort,
.eh_device_reset_handler= iscsi_eh_device_reset,
diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h
index 95f0a64e0..035105978 100644
--- a/drivers/infiniband/ulp/iser/iscsi_iser.h
+++ b/drivers/infiniband/ulp/iser/iscsi_iser.h
@@ -458,9 +458,6 @@ struct iser_fr_pool {
* @comp: iser completion context
* @fr_pool: connection fast registration poool
* @pi_support: Indicate device T10-PI support
- * @last: last send wr to signal all flush errors were drained
- * @last_cqe: cqe handler for last wr
- * @last_comp: completes when all connection completions consumed
*/
struct ib_conn {
struct rdma_cm_id *cma_id;
@@ -472,10 +469,7 @@ struct ib_conn {
struct iser_comp *comp;
struct iser_fr_pool fr_pool;
bool pi_support;
- struct ib_send_wr last;
- struct ib_cqe last_cqe;
struct ib_cqe reg_cqe;
- struct completion last_comp;
};
/**
@@ -617,7 +611,6 @@ void iser_cmd_comp(struct ib_cq *cq, struct ib_wc *wc);
void iser_ctrl_comp(struct ib_cq *cq, struct ib_wc *wc);
void iser_dataout_comp(struct ib_cq *cq, struct ib_wc *wc);
void iser_reg_comp(struct ib_cq *cq, struct ib_wc *wc);
-void iser_last_comp(struct ib_cq *cq, struct ib_wc *wc);
void iser_task_rdma_init(struct iscsi_iser_task *task);
diff --git a/drivers/infiniband/ulp/iser/iser_initiator.c b/drivers/infiniband/ulp/iser/iser_initiator.c
index ed54b388e..81ae2e30d 100644
--- a/drivers/infiniband/ulp/iser/iser_initiator.c
+++ b/drivers/infiniband/ulp/iser/iser_initiator.c
@@ -729,13 +729,6 @@ void iser_dataout_comp(struct ib_cq *cq, struct ib_wc *wc)
kmem_cache_free(ig.desc_cache, desc);
}
-void iser_last_comp(struct ib_cq *cq, struct ib_wc *wc)
-{
- struct ib_conn *ib_conn = wc->qp->qp_context;
-
- complete(&ib_conn->last_comp);
-}
-
void iser_task_rdma_init(struct iscsi_iser_task *iser_task)
{
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 40c0f4978..1b4945367 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -252,14 +252,21 @@ void iser_free_fmr_pool(struct ib_conn *ib_conn)
}
static int
-iser_alloc_reg_res(struct ib_device *ib_device,
+iser_alloc_reg_res(struct iser_device *device,
struct ib_pd *pd,
struct iser_reg_resources *res,
unsigned int size)
{
+ struct ib_device *ib_dev = device->ib_device;
+ enum ib_mr_type mr_type;
int ret;
- res->mr = ib_alloc_mr(pd, IB_MR_TYPE_MEM_REG, size);
+ if (ib_dev->attrs.device_cap_flags & IB_DEVICE_SG_GAPS_REG)
+ mr_type = IB_MR_TYPE_SG_GAPS;
+ else
+ mr_type = IB_MR_TYPE_MEM_REG;
+
+ res->mr = ib_alloc_mr(pd, mr_type, size);
if (IS_ERR(res->mr)) {
ret = PTR_ERR(res->mr);
iser_err("Failed to allocate ib_fast_reg_mr err=%d\n", ret);
@@ -277,7 +284,7 @@ iser_free_reg_res(struct iser_reg_resources *rsc)
}
static int
-iser_alloc_pi_ctx(struct ib_device *ib_device,
+iser_alloc_pi_ctx(struct iser_device *device,
struct ib_pd *pd,
struct iser_fr_desc *desc,
unsigned int size)
@@ -291,7 +298,7 @@ iser_alloc_pi_ctx(struct ib_device *ib_device,
pi_ctx = desc->pi_ctx;
- ret = iser_alloc_reg_res(ib_device, pd, &pi_ctx->rsc, size);
+ ret = iser_alloc_reg_res(device, pd, &pi_ctx->rsc, size);
if (ret) {
iser_err("failed to allocate reg_resources\n");
goto alloc_reg_res_err;
@@ -324,7 +331,7 @@ iser_free_pi_ctx(struct iser_pi_context *pi_ctx)
}
static struct iser_fr_desc *
-iser_create_fastreg_desc(struct ib_device *ib_device,
+iser_create_fastreg_desc(struct iser_device *device,
struct ib_pd *pd,
bool pi_enable,
unsigned int size)
@@ -336,12 +343,12 @@ iser_create_fastreg_desc(struct ib_device *ib_device,
if (!desc)
return ERR_PTR(-ENOMEM);
- ret = iser_alloc_reg_res(ib_device, pd, &desc->rsc, size);
+ ret = iser_alloc_reg_res(device, pd, &desc->rsc, size);
if (ret)
goto reg_res_alloc_failure;
if (pi_enable) {
- ret = iser_alloc_pi_ctx(ib_device, pd, desc, size);
+ ret = iser_alloc_pi_ctx(device, pd, desc, size);
if (ret)
goto pi_ctx_alloc_failure;
}
@@ -374,7 +381,7 @@ int iser_alloc_fastreg_pool(struct ib_conn *ib_conn,
spin_lock_init(&fr_pool->lock);
fr_pool->size = 0;
for (i = 0; i < cmds_max; i++) {
- desc = iser_create_fastreg_desc(device->ib_device, device->pd,
+ desc = iser_create_fastreg_desc(device, device->pd,
ib_conn->pi_support, size);
if (IS_ERR(desc)) {
ret = PTR_ERR(desc);
@@ -663,7 +670,6 @@ void iser_conn_release(struct iser_conn *iser_conn)
int iser_conn_terminate(struct iser_conn *iser_conn)
{
struct ib_conn *ib_conn = &iser_conn->ib_conn;
- struct ib_send_wr *bad_wr;
int err = 0;
/* terminate the iser conn only if the conn state is UP */
@@ -688,14 +694,8 @@ int iser_conn_terminate(struct iser_conn *iser_conn)
iser_err("Failed to disconnect, conn: 0x%p err %d\n",
iser_conn, err);
- /* post an indication that all flush errors were consumed */
- err = ib_post_send(ib_conn->qp, &ib_conn->last, &bad_wr);
- if (err) {
- iser_err("conn %p failed to post last wr", ib_conn);
- return 1;
- }
-
- wait_for_completion(&ib_conn->last_comp);
+ /* block until all flush errors are consumed */
+ ib_drain_sq(ib_conn->qp);
}
return 1;
@@ -954,10 +954,6 @@ void iser_conn_init(struct iser_conn *iser_conn)
ib_conn->post_recv_buf_count = 0;
ib_conn->reg_cqe.done = iser_reg_comp;
- ib_conn->last_cqe.done = iser_last_comp;
- ib_conn->last.wr_cqe = &ib_conn->last_cqe;
- ib_conn->last.opcode = IB_WR_SEND;
- init_completion(&ib_conn->last_comp);
}
/**
diff --git a/drivers/infiniband/ulp/isert/ib_isert.c b/drivers/infiniband/ulp/isert/ib_isert.c
index 0e1a802c3..411e4464c 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.c
+++ b/drivers/infiniband/ulp/isert/ib_isert.c
@@ -49,23 +49,24 @@ static struct workqueue_struct *isert_release_wq;
static void
isert_unmap_cmd(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn);
static int
-isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
- struct isert_rdma_wr *wr);
+isert_map_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn);
static void
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);
+isert_reg_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn);
static int
isert_put_response(struct iscsi_conn *conn, struct iscsi_cmd *cmd);
static int
-isert_rdma_post_recvl(struct isert_conn *isert_conn);
+isert_login_post_recv(struct isert_conn *isert_conn);
static int
isert_rdma_accept(struct isert_conn *isert_conn);
struct rdma_cm_id *isert_setup_id(struct isert_np *isert_np);
static void isert_release_work(struct work_struct *work);
-static void isert_wait4flush(struct isert_conn *isert_conn);
+static void isert_recv_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_send_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc);
+static void isert_login_send_done(struct ib_cq *cq, struct ib_wc *wc);
static inline bool
isert_prot_cmd(struct isert_conn *conn, struct se_cmd *cmd)
@@ -139,7 +140,7 @@ isert_create_qp(struct isert_conn *isert_conn,
attr.qp_context = isert_conn;
attr.send_cq = comp->cq;
attr.recv_cq = comp->cq;
- attr.cap.max_send_wr = ISERT_QP_MAX_REQ_DTOS;
+ attr.cap.max_send_wr = ISERT_QP_MAX_REQ_DTOS + 1;
attr.cap.max_recv_wr = ISERT_QP_MAX_RECV_DTOS + 1;
attr.cap.max_send_sge = device->ib_device->attrs.max_sge;
isert_conn->max_sge = min(device->ib_device->attrs.max_sge,
@@ -178,12 +179,6 @@ err:
return ret;
}
-static void
-isert_cq_event_callback(struct ib_event *e, void *context)
-{
- isert_dbg("event: %d\n", e->event);
-}
-
static int
isert_alloc_rx_descriptors(struct isert_conn *isert_conn)
{
@@ -213,6 +208,7 @@ isert_alloc_rx_descriptors(struct isert_conn *isert_conn)
rx_sg->addr = rx_desc->dma_addr;
rx_sg->length = ISER_RX_PAYLOAD_SIZE;
rx_sg->lkey = device->pd->local_dma_lkey;
+ rx_desc->rx_cqe.done = isert_recv_done;
}
return 0;
@@ -251,9 +247,6 @@ isert_free_rx_descriptors(struct isert_conn *isert_conn)
isert_conn->rx_descs = NULL;
}
-static void isert_cq_work(struct work_struct *);
-static void isert_cq_callback(struct ib_cq *, void *);
-
static void
isert_free_comps(struct isert_device *device)
{
@@ -262,10 +255,8 @@ isert_free_comps(struct isert_device *device)
for (i = 0; i < device->comps_used; i++) {
struct isert_comp *comp = &device->comps[i];
- if (comp->cq) {
- cancel_work_sync(&comp->work);
- ib_destroy_cq(comp->cq);
- }
+ if (comp->cq)
+ ib_free_cq(comp->cq);
}
kfree(device->comps);
}
@@ -294,28 +285,17 @@ isert_alloc_comps(struct isert_device *device)
max_cqe = min(ISER_MAX_CQ_LEN, device->ib_device->attrs.max_cqe);
for (i = 0; i < device->comps_used; i++) {
- struct ib_cq_init_attr cq_attr = {};
struct isert_comp *comp = &device->comps[i];
comp->device = device;
- INIT_WORK(&comp->work, isert_cq_work);
- cq_attr.cqe = max_cqe;
- cq_attr.comp_vector = i;
- comp->cq = ib_create_cq(device->ib_device,
- isert_cq_callback,
- isert_cq_event_callback,
- (void *)comp,
- &cq_attr);
+ comp->cq = ib_alloc_cq(device->ib_device, comp, max_cqe, i,
+ IB_POLL_WORKQUEUE);
if (IS_ERR(comp->cq)) {
isert_err("Unable to allocate cq\n");
ret = PTR_ERR(comp->cq);
comp->cq = NULL;
goto out_cq;
}
-
- ret = ib_req_notify_cq(comp->cq, IB_CQ_NEXT_COMP);
- if (ret)
- goto out_cq;
}
return 0;
@@ -583,7 +563,6 @@ isert_init_conn(struct isert_conn *isert_conn)
INIT_LIST_HEAD(&isert_conn->node);
init_completion(&isert_conn->login_comp);
init_completion(&isert_conn->login_req_comp);
- init_completion(&isert_conn->wait);
kref_init(&isert_conn->kref);
mutex_init(&isert_conn->mutex);
spin_lock_init(&isert_conn->pool_lock);
@@ -597,11 +576,13 @@ isert_free_login_buf(struct isert_conn *isert_conn)
struct ib_device *ib_dev = isert_conn->device->ib_device;
ib_dma_unmap_single(ib_dev, isert_conn->login_rsp_dma,
- ISER_RX_LOGIN_SIZE, DMA_TO_DEVICE);
+ ISER_RX_PAYLOAD_SIZE, DMA_TO_DEVICE);
+ kfree(isert_conn->login_rsp_buf);
+
ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma,
- ISCSI_DEF_MAX_RECV_SEG_LEN,
+ ISER_RX_PAYLOAD_SIZE,
DMA_FROM_DEVICE);
- kfree(isert_conn->login_buf);
+ kfree(isert_conn->login_req_buf);
}
static int
@@ -610,50 +591,48 @@ isert_alloc_login_buf(struct isert_conn *isert_conn,
{
int ret;
- isert_conn->login_buf = kzalloc(ISCSI_DEF_MAX_RECV_SEG_LEN +
- ISER_RX_LOGIN_SIZE, GFP_KERNEL);
- if (!isert_conn->login_buf) {
+ isert_conn->login_req_buf = kzalloc(sizeof(*isert_conn->login_req_buf),
+ GFP_KERNEL);
+ if (!isert_conn->login_req_buf) {
isert_err("Unable to allocate isert_conn->login_buf\n");
return -ENOMEM;
}
- isert_conn->login_req_buf = isert_conn->login_buf;
- isert_conn->login_rsp_buf = isert_conn->login_buf +
- ISCSI_DEF_MAX_RECV_SEG_LEN;
-
- isert_dbg("Set login_buf: %p login_req_buf: %p login_rsp_buf: %p\n",
- isert_conn->login_buf, isert_conn->login_req_buf,
- isert_conn->login_rsp_buf);
-
isert_conn->login_req_dma = ib_dma_map_single(ib_dev,
- (void *)isert_conn->login_req_buf,
- ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_FROM_DEVICE);
-
+ isert_conn->login_req_buf,
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
ret = ib_dma_mapping_error(ib_dev, isert_conn->login_req_dma);
if (ret) {
isert_err("login_req_dma mapping error: %d\n", ret);
isert_conn->login_req_dma = 0;
- goto out_login_buf;
+ goto out_free_login_req_buf;
}
- isert_conn->login_rsp_dma = ib_dma_map_single(ib_dev,
- (void *)isert_conn->login_rsp_buf,
- ISER_RX_LOGIN_SIZE, DMA_TO_DEVICE);
+ isert_conn->login_rsp_buf = kzalloc(ISER_RX_PAYLOAD_SIZE, GFP_KERNEL);
+ if (!isert_conn->login_rsp_buf) {
+ isert_err("Unable to allocate isert_conn->login_rspbuf\n");
+ goto out_unmap_login_req_buf;
+ }
+ isert_conn->login_rsp_dma = ib_dma_map_single(ib_dev,
+ isert_conn->login_rsp_buf,
+ ISER_RX_PAYLOAD_SIZE, DMA_TO_DEVICE);
ret = ib_dma_mapping_error(ib_dev, isert_conn->login_rsp_dma);
if (ret) {
isert_err("login_rsp_dma mapping error: %d\n", ret);
isert_conn->login_rsp_dma = 0;
- goto out_req_dma_map;
+ goto out_free_login_rsp_buf;
}
return 0;
-out_req_dma_map:
+out_free_login_rsp_buf:
+ kfree(isert_conn->login_rsp_buf);
+out_unmap_login_req_buf:
ib_dma_unmap_single(ib_dev, isert_conn->login_req_dma,
- ISCSI_DEF_MAX_RECV_SEG_LEN, DMA_FROM_DEVICE);
-out_login_buf:
- kfree(isert_conn->login_buf);
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
+out_free_login_req_buf:
+ kfree(isert_conn->login_req_buf);
return ret;
}
@@ -727,7 +706,7 @@ isert_connect_request(struct rdma_cm_id *cma_id, struct rdma_cm_event *event)
if (ret)
goto out_conn_dev;
- ret = isert_rdma_post_recvl(isert_conn);
+ ret = isert_login_post_recv(isert_conn);
if (ret)
goto out_conn_dev;
@@ -774,7 +753,7 @@ isert_connect_release(struct isert_conn *isert_conn)
ib_destroy_qp(isert_conn->qp);
}
- if (isert_conn->login_buf)
+ if (isert_conn->login_req_buf)
isert_free_login_buf(isert_conn);
isert_device_put(device);
@@ -834,7 +813,6 @@ isert_handle_unbound_conn(struct isert_conn *isert_conn)
*/
list_del_init(&isert_conn->node);
isert_put_conn(isert_conn);
- complete(&isert_conn->wait);
queue_work(isert_release_wq, &isert_conn->release_work);
}
mutex_unlock(&isert_np->mutex);
@@ -867,9 +845,6 @@ isert_conn_terminate(struct isert_conn *isert_conn)
if (err)
isert_warn("Failed rdma_disconnect isert_conn %p\n",
isert_conn);
-
- isert_info("conn %p completing wait\n", isert_conn);
- complete(&isert_conn->wait);
}
static int
@@ -911,7 +886,7 @@ isert_disconnected_handler(struct rdma_cm_id *cma_id,
break;
case ISER_CONN_UP:
isert_conn_terminate(isert_conn);
- isert_wait4flush(isert_conn);
+ ib_drain_qp(isert_conn->qp);
isert_handle_unbound_conn(isert_conn);
break;
case ISER_CONN_BOUND:
@@ -989,7 +964,8 @@ isert_post_recvm(struct isert_conn *isert_conn, u32 count)
for (rx_wr = isert_conn->rx_wr, i = 0; i < count; i++, rx_wr++) {
rx_desc = &isert_conn->rx_descs[i];
- rx_wr->wr_id = (uintptr_t)rx_desc;
+
+ rx_wr->wr_cqe = &rx_desc->rx_cqe;
rx_wr->sg_list = &rx_desc->rx_sg;
rx_wr->num_sge = 1;
rx_wr->next = rx_wr + 1;
@@ -1011,7 +987,7 @@ isert_post_recv(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc)
struct ib_recv_wr *rx_wr_failed, rx_wr;
int ret;
- rx_wr.wr_id = (uintptr_t)rx_desc;
+ rx_wr.wr_cqe = &rx_desc->rx_cqe;
rx_wr.sg_list = &rx_desc->rx_sg;
rx_wr.num_sge = 1;
rx_wr.next = NULL;
@@ -1024,7 +1000,7 @@ isert_post_recv(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc)
}
static int
-isert_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
+isert_login_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
{
struct ib_device *ib_dev = isert_conn->cm_id->device;
struct ib_send_wr send_wr, *send_wr_failed;
@@ -1033,8 +1009,10 @@ isert_post_send(struct isert_conn *isert_conn, struct iser_tx_desc *tx_desc)
ib_dma_sync_single_for_device(ib_dev, tx_desc->dma_addr,
ISER_HEADERS_LEN, DMA_TO_DEVICE);
+ tx_desc->tx_cqe.done = isert_login_send_done;
+
send_wr.next = NULL;
- send_wr.wr_id = (uintptr_t)tx_desc;
+ send_wr.wr_cqe = &tx_desc->tx_cqe;
send_wr.sg_list = tx_desc->tx_sg;
send_wr.num_sge = tx_desc->num_sge;
send_wr.opcode = IB_WR_SEND;
@@ -1062,7 +1040,6 @@ isert_create_send_desc(struct isert_conn *isert_conn,
tx_desc->iser_header.flags = ISCSI_CTRL;
tx_desc->num_sge = 1;
- tx_desc->isert_cmd = isert_cmd;
if (tx_desc->tx_sg[0].lkey != device->pd->local_dma_lkey) {
tx_desc->tx_sg[0].lkey = device->pd->local_dma_lkey;
@@ -1103,8 +1080,9 @@ isert_init_send_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
{
struct iser_tx_desc *tx_desc = &isert_cmd->tx_desc;
- isert_cmd->rdma_wr.iser_ib_op = ISER_IB_SEND;
- send_wr->wr_id = (uintptr_t)&isert_cmd->tx_desc;
+ isert_cmd->iser_ib_op = ISER_IB_SEND;
+ tx_desc->tx_cqe.done = isert_send_done;
+ send_wr->wr_cqe = &tx_desc->tx_cqe;
if (isert_conn->snd_w_inv && isert_cmd->inv_rkey) {
send_wr->opcode = IB_WR_SEND_WITH_INV;
@@ -1119,7 +1097,7 @@ isert_init_send_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
}
static int
-isert_rdma_post_recvl(struct isert_conn *isert_conn)
+isert_login_post_recv(struct isert_conn *isert_conn)
{
struct ib_recv_wr rx_wr, *rx_wr_fail;
struct ib_sge sge;
@@ -1127,14 +1105,16 @@ isert_rdma_post_recvl(struct isert_conn *isert_conn)
memset(&sge, 0, sizeof(struct ib_sge));
sge.addr = isert_conn->login_req_dma;
- sge.length = ISER_RX_LOGIN_SIZE;
+ sge.length = ISER_RX_PAYLOAD_SIZE;
sge.lkey = isert_conn->device->pd->local_dma_lkey;
isert_dbg("Setup sge: addr: %llx length: %d 0x%08x\n",
sge.addr, sge.length, sge.lkey);
+ isert_conn->login_req_buf->rx_cqe.done = isert_login_recv_done;
+
memset(&rx_wr, 0, sizeof(struct ib_recv_wr));
- rx_wr.wr_id = (uintptr_t)isert_conn->login_req_buf;
+ rx_wr.wr_cqe = &isert_conn->login_req_buf->rx_cqe;
rx_wr.sg_list = &sge;
rx_wr.num_sge = 1;
@@ -1206,12 +1186,12 @@ isert_put_login_tx(struct iscsi_conn *conn, struct iscsi_login *login,
goto post_send;
}
- ret = isert_rdma_post_recvl(isert_conn);
+ ret = isert_login_post_recv(isert_conn);
if (ret)
return ret;
}
post_send:
- ret = isert_post_send(isert_conn, tx_desc);
+ ret = isert_login_post_send(isert_conn, tx_desc);
if (ret)
return ret;
@@ -1221,7 +1201,7 @@ post_send:
static void
isert_rx_login_req(struct isert_conn *isert_conn)
{
- struct iser_rx_desc *rx_desc = (void *)isert_conn->login_req_buf;
+ struct iser_rx_desc *rx_desc = isert_conn->login_req_buf;
int rx_buflen = isert_conn->login_req_len;
struct iscsi_conn *conn = isert_conn->conn;
struct iscsi_login *login = conn->conn_login;
@@ -1554,12 +1534,42 @@ isert_rx_opcode(struct isert_conn *isert_conn, struct iser_rx_desc *rx_desc,
}
static void
-isert_rx_do_work(struct iser_rx_desc *rx_desc, struct isert_conn *isert_conn)
+isert_print_wc(struct ib_wc *wc, const char *type)
{
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ isert_err("%s failure: %s (%d) vend_err %x\n", type,
+ ib_wc_status_msg(wc->status), wc->status,
+ wc->vendor_err);
+ else
+ isert_dbg("%s failure: %s (%d)\n", type,
+ ib_wc_status_msg(wc->status), wc->status);
+}
+
+static void
+isert_recv_done(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct isert_conn *isert_conn = wc->qp->qp_context;
+ struct ib_device *ib_dev = isert_conn->cm_id->device;
+ struct iser_rx_desc *rx_desc = cqe_to_rx_desc(wc->wr_cqe);
+ struct iscsi_hdr *hdr = &rx_desc->iscsi_header;
struct iser_ctrl *iser_ctrl = &rx_desc->iser_header;
uint64_t read_va = 0, write_va = 0;
uint32_t read_stag = 0, write_stag = 0;
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "recv");
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+ return;
+ }
+
+ ib_dma_sync_single_for_cpu(ib_dev, rx_desc->dma_addr,
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
+
+ isert_dbg("DMA: 0x%llx, iSCSI opcode: 0x%02x, ITT: 0x%08x, flags: 0x%02x dlen: %d\n",
+ rx_desc->dma_addr, hdr->opcode, hdr->itt, hdr->flags,
+ (int)(wc->byte_len - ISER_HEADERS_LEN));
+
switch (iser_ctrl->flags & 0xF0) {
case ISCSI_CTRL:
if (iser_ctrl->flags & ISER_RSV) {
@@ -1587,55 +1597,40 @@ isert_rx_do_work(struct iser_rx_desc *rx_desc, struct isert_conn *isert_conn)
isert_rx_opcode(isert_conn, rx_desc,
read_stag, read_va, write_stag, write_va);
+
+ ib_dma_sync_single_for_device(ib_dev, rx_desc->dma_addr,
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
}
static void
-isert_rcv_completion(struct iser_rx_desc *desc,
- struct isert_conn *isert_conn,
- u32 xfer_len)
+isert_login_recv_done(struct ib_cq *cq, struct ib_wc *wc)
{
+ struct isert_conn *isert_conn = wc->qp->qp_context;
struct ib_device *ib_dev = isert_conn->cm_id->device;
- struct iscsi_hdr *hdr;
- u64 rx_dma;
- int rx_buflen;
-
- if ((char *)desc == isert_conn->login_req_buf) {
- rx_dma = isert_conn->login_req_dma;
- rx_buflen = ISER_RX_LOGIN_SIZE;
- isert_dbg("login_buf: Using rx_dma: 0x%llx, rx_buflen: %d\n",
- rx_dma, rx_buflen);
- } else {
- rx_dma = desc->dma_addr;
- rx_buflen = ISER_RX_PAYLOAD_SIZE;
- isert_dbg("req_buf: Using rx_dma: 0x%llx, rx_buflen: %d\n",
- rx_dma, rx_buflen);
+
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "login recv");
+ return;
}
- ib_dma_sync_single_for_cpu(ib_dev, rx_dma, rx_buflen, DMA_FROM_DEVICE);
+ ib_dma_sync_single_for_cpu(ib_dev, isert_conn->login_req_dma,
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
- hdr = &desc->iscsi_header;
- isert_dbg("iSCSI opcode: 0x%02x, ITT: 0x%08x, flags: 0x%02x dlen: %d\n",
- hdr->opcode, hdr->itt, hdr->flags,
- (int)(xfer_len - ISER_HEADERS_LEN));
+ isert_conn->login_req_len = wc->byte_len - ISER_HEADERS_LEN;
- if ((char *)desc == isert_conn->login_req_buf) {
- isert_conn->login_req_len = xfer_len - ISER_HEADERS_LEN;
- if (isert_conn->conn) {
- struct iscsi_login *login = isert_conn->conn->conn_login;
+ if (isert_conn->conn) {
+ struct iscsi_login *login = isert_conn->conn->conn_login;
- if (login && !login->first_request)
- isert_rx_login_req(isert_conn);
- }
- mutex_lock(&isert_conn->mutex);
- complete(&isert_conn->login_req_comp);
- mutex_unlock(&isert_conn->mutex);
- } else {
- isert_rx_do_work(desc, isert_conn);
+ if (login && !login->first_request)
+ isert_rx_login_req(isert_conn);
}
- ib_dma_sync_single_for_device(ib_dev, rx_dma, rx_buflen,
- DMA_FROM_DEVICE);
+ mutex_lock(&isert_conn->mutex);
+ complete(&isert_conn->login_req_comp);
+ mutex_unlock(&isert_conn->mutex);
+ ib_dma_sync_single_for_device(ib_dev, isert_conn->login_req_dma,
+ ISER_RX_PAYLOAD_SIZE, DMA_FROM_DEVICE);
}
static int
@@ -1685,54 +1680,50 @@ isert_unmap_data_buf(struct isert_conn *isert_conn, struct isert_data_buf *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;
-
isert_dbg("Cmd %p\n", isert_cmd);
- if (wr->data.sg) {
+ if (isert_cmd->data.sg) {
isert_dbg("Cmd %p unmap_sg op\n", isert_cmd);
- isert_unmap_data_buf(isert_conn, &wr->data);
+ isert_unmap_data_buf(isert_conn, &isert_cmd->data);
}
- if (wr->rdma_wr) {
+ if (isert_cmd->rdma_wr) {
isert_dbg("Cmd %p free send_wr\n", isert_cmd);
- kfree(wr->rdma_wr);
- wr->rdma_wr = NULL;
+ kfree(isert_cmd->rdma_wr);
+ isert_cmd->rdma_wr = NULL;
}
- if (wr->ib_sge) {
+ if (isert_cmd->ib_sge) {
isert_dbg("Cmd %p free ib_sge\n", isert_cmd);
- kfree(wr->ib_sge);
- wr->ib_sge = NULL;
+ kfree(isert_cmd->ib_sge);
+ isert_cmd->ib_sge = NULL;
}
}
static void
isert_unreg_rdma(struct isert_cmd *isert_cmd, struct isert_conn *isert_conn)
{
- struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
-
isert_dbg("Cmd %p\n", isert_cmd);
- if (wr->fr_desc) {
- isert_dbg("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;
+ if (isert_cmd->fr_desc) {
+ isert_dbg("Cmd %p free fr_desc %p\n", isert_cmd, isert_cmd->fr_desc);
+ if (isert_cmd->fr_desc->ind & ISERT_PROTECTED) {
+ isert_unmap_data_buf(isert_conn, &isert_cmd->prot);
+ isert_cmd->fr_desc->ind &= ~ISERT_PROTECTED;
}
spin_lock_bh(&isert_conn->pool_lock);
- list_add_tail(&wr->fr_desc->list, &isert_conn->fr_pool);
+ list_add_tail(&isert_cmd->fr_desc->list, &isert_conn->fr_pool);
spin_unlock_bh(&isert_conn->pool_lock);
- wr->fr_desc = NULL;
+ isert_cmd->fr_desc = NULL;
}
- if (wr->data.sg) {
+ if (isert_cmd->data.sg) {
isert_dbg("Cmd %p unmap_sg op\n", isert_cmd);
- isert_unmap_data_buf(isert_conn, &wr->data);
+ isert_unmap_data_buf(isert_conn, &isert_cmd->data);
}
- wr->ib_sge = NULL;
- wr->rdma_wr = NULL;
+ isert_cmd->ib_sge = NULL;
+ isert_cmd->rdma_wr = NULL;
}
static void
@@ -1884,52 +1875,70 @@ fail_mr_status:
}
static void
-isert_completion_rdma_write(struct iser_tx_desc *tx_desc,
- struct isert_cmd *isert_cmd)
+isert_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
{
- 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_conn *isert_conn = wc->qp->qp_context;
struct isert_device *device = isert_conn->device;
+ struct iser_tx_desc *desc = cqe_to_tx_desc(wc->wr_cqe);
+ struct isert_cmd *isert_cmd = tx_desc_to_cmd(desc);
+ struct se_cmd *cmd = &isert_cmd->iscsi_cmd->se_cmd;
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;
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "rdma write");
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+ isert_completion_put(desc, isert_cmd, device->ib_device, true);
+ return;
+ }
+
+ isert_dbg("Cmd %p\n", isert_cmd);
+
+ if (isert_cmd->fr_desc && isert_cmd->fr_desc->ind & ISERT_PROTECTED) {
+ ret = isert_check_pi_status(cmd,
+ isert_cmd->fr_desc->pi_ctx->sig_mr);
+ isert_cmd->fr_desc->ind &= ~ISERT_PROTECTED;
}
device->unreg_rdma_mem(isert_cmd, isert_conn);
- wr->rdma_wr_num = 0;
+ isert_cmd->rdma_wr_num = 0;
if (ret)
- transport_send_check_condition_and_sense(se_cmd,
- se_cmd->pi_err, 0);
+ transport_send_check_condition_and_sense(cmd, cmd->pi_err, 0);
else
- isert_put_response(isert_conn->conn, cmd);
+ isert_put_response(isert_conn->conn, isert_cmd->iscsi_cmd);
}
static void
-isert_completion_rdma_read(struct iser_tx_desc *tx_desc,
- struct isert_cmd *isert_cmd)
+isert_rdma_read_done(struct ib_cq *cq, struct ib_wc *wc)
{
- struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
+ struct isert_conn *isert_conn = wc->qp->qp_context;
+ struct isert_device *device = isert_conn->device;
+ struct iser_tx_desc *desc = cqe_to_tx_desc(wc->wr_cqe);
+ struct isert_cmd *isert_cmd = tx_desc_to_cmd(desc);
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->device;
int ret = 0;
- if (wr->fr_desc && wr->fr_desc->ind & ISERT_PROTECTED) {
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "rdma read");
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+ isert_completion_put(desc, isert_cmd, device->ib_device, true);
+ return;
+ }
+
+ isert_dbg("Cmd %p\n", isert_cmd);
+
+ if (isert_cmd->fr_desc && isert_cmd->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;
+ isert_cmd->fr_desc->pi_ctx->sig_mr);
+ isert_cmd->fr_desc->ind &= ~ISERT_PROTECTED;
}
iscsit_stop_dataout_timer(cmd);
device->unreg_rdma_mem(isert_cmd, isert_conn);
- cmd->write_data_done = wr->data.len;
- wr->rdma_wr_num = 0;
+ cmd->write_data_done = isert_cmd->data.len;
+ isert_cmd->rdma_wr_num = 0;
isert_dbg("Cmd: %p RDMA_READ comp calling execute_cmd\n", isert_cmd);
spin_lock_bh(&cmd->istate_lock);
@@ -1977,167 +1986,56 @@ isert_do_control_comp(struct work_struct *work)
}
static void
-isert_response_completion(struct iser_tx_desc *tx_desc,
- struct isert_cmd *isert_cmd,
- struct isert_conn *isert_conn,
- struct ib_device *ib_dev)
+isert_login_send_done(struct ib_cq *cq, struct ib_wc *wc)
{
- struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
-
- if (cmd->i_state == ISTATE_SEND_TASKMGTRSP ||
- cmd->i_state == ISTATE_SEND_LOGOUTRSP ||
- cmd->i_state == ISTATE_SEND_REJECT ||
- cmd->i_state == ISTATE_SEND_TEXTRSP) {
- isert_unmap_tx_desc(tx_desc, ib_dev);
+ struct isert_conn *isert_conn = wc->qp->qp_context;
+ struct ib_device *ib_dev = isert_conn->cm_id->device;
+ struct iser_tx_desc *tx_desc = cqe_to_tx_desc(wc->wr_cqe);
- INIT_WORK(&isert_cmd->comp_work, isert_do_control_comp);
- queue_work(isert_comp_wq, &isert_cmd->comp_work);
- return;
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "login send");
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
}
- cmd->i_state = ISTATE_SENT_STATUS;
- isert_completion_put(tx_desc, isert_cmd, ib_dev, false);
+ isert_unmap_tx_desc(tx_desc, ib_dev);
}
static void
-isert_snd_completion(struct iser_tx_desc *tx_desc,
- struct isert_conn *isert_conn)
+isert_send_done(struct ib_cq *cq, struct ib_wc *wc)
{
+ struct isert_conn *isert_conn = wc->qp->qp_context;
struct ib_device *ib_dev = isert_conn->cm_id->device;
- struct isert_cmd *isert_cmd = tx_desc->isert_cmd;
- struct isert_rdma_wr *wr;
+ struct iser_tx_desc *tx_desc = cqe_to_tx_desc(wc->wr_cqe);
+ struct isert_cmd *isert_cmd = tx_desc_to_cmd(tx_desc);
- if (!isert_cmd) {
- isert_unmap_tx_desc(tx_desc, ib_dev);
+ if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ isert_print_wc(wc, "send");
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ iscsit_cause_connection_reinstatement(isert_conn->conn, 0);
+ isert_completion_put(tx_desc, isert_cmd, ib_dev, true);
return;
}
- wr = &isert_cmd->rdma_wr;
- isert_dbg("Cmd %p iser_ib_op %d\n", isert_cmd, wr->iser_ib_op);
+ isert_dbg("Cmd %p\n", isert_cmd);
- switch (wr->iser_ib_op) {
- case ISER_IB_SEND:
- isert_response_completion(tx_desc, isert_cmd,
- isert_conn, ib_dev);
- break;
- case ISER_IB_RDMA_WRITE:
- isert_completion_rdma_write(tx_desc, isert_cmd);
- break;
- case ISER_IB_RDMA_READ:
- isert_completion_rdma_read(tx_desc, isert_cmd);
- break;
+ switch (isert_cmd->iscsi_cmd->i_state) {
+ case ISTATE_SEND_TASKMGTRSP:
+ case ISTATE_SEND_LOGOUTRSP:
+ case ISTATE_SEND_REJECT:
+ case ISTATE_SEND_TEXTRSP:
+ isert_unmap_tx_desc(tx_desc, ib_dev);
+
+ INIT_WORK(&isert_cmd->comp_work, isert_do_control_comp);
+ queue_work(isert_comp_wq, &isert_cmd->comp_work);
+ return;
default:
- isert_err("Unknown wr->iser_ib_op: 0x%x\n", wr->iser_ib_op);
- dump_stack();
+ isert_cmd->iscsi_cmd->i_state = ISTATE_SENT_STATUS;
+ isert_completion_put(tx_desc, isert_cmd, ib_dev, false);
break;
}
}
-/**
- * is_isert_tx_desc() - Indicate if the completion wr_id
- * is a TX descriptor or not.
- * @isert_conn: iser connection
- * @wr_id: completion WR identifier
- *
- * Since we cannot rely on wc opcode in FLUSH errors
- * we must work around it by checking if the wr_id address
- * falls in the iser connection rx_descs buffer. If so
- * it is an RX descriptor, otherwize it is a TX.
- */
-static inline bool
-is_isert_tx_desc(struct isert_conn *isert_conn, void *wr_id)
-{
- void *start = isert_conn->rx_descs;
- int len = ISERT_QP_MAX_RECV_DTOS * sizeof(*isert_conn->rx_descs);
-
- if ((wr_id >= start && wr_id < start + len) ||
- (wr_id == isert_conn->login_req_buf))
- return false;
-
- return true;
-}
-
-static void
-isert_cq_comp_err(struct isert_conn *isert_conn, struct ib_wc *wc)
-{
- if (wc->wr_id == ISER_BEACON_WRID) {
- isert_info("conn %p completing wait_comp_err\n",
- isert_conn);
- complete(&isert_conn->wait_comp_err);
- } else if (is_isert_tx_desc(isert_conn, (void *)(uintptr_t)wc->wr_id)) {
- struct ib_device *ib_dev = isert_conn->cm_id->device;
- struct isert_cmd *isert_cmd;
- struct iser_tx_desc *desc;
-
- desc = (struct iser_tx_desc *)(uintptr_t)wc->wr_id;
- isert_cmd = desc->isert_cmd;
- if (!isert_cmd)
- isert_unmap_tx_desc(desc, ib_dev);
- else
- isert_completion_put(desc, isert_cmd, ib_dev, true);
- }
-}
-
-static void
-isert_handle_wc(struct ib_wc *wc)
-{
- struct isert_conn *isert_conn;
- struct iser_tx_desc *tx_desc;
- struct iser_rx_desc *rx_desc;
-
- isert_conn = wc->qp->qp_context;
- if (likely(wc->status == IB_WC_SUCCESS)) {
- if (wc->opcode == IB_WC_RECV) {
- rx_desc = (struct iser_rx_desc *)(uintptr_t)wc->wr_id;
- isert_rcv_completion(rx_desc, isert_conn, wc->byte_len);
- } else {
- tx_desc = (struct iser_tx_desc *)(uintptr_t)wc->wr_id;
- isert_snd_completion(tx_desc, isert_conn);
- }
- } else {
- if (wc->status != IB_WC_WR_FLUSH_ERR)
- isert_err("%s (%d): wr id %llx vend_err %x\n",
- ib_wc_status_msg(wc->status), wc->status,
- wc->wr_id, wc->vendor_err);
- else
- isert_dbg("%s (%d): wr id %llx\n",
- ib_wc_status_msg(wc->status), wc->status,
- wc->wr_id);
-
- if (wc->wr_id != ISER_FASTREG_LI_WRID)
- isert_cq_comp_err(isert_conn, wc);
- }
-}
-
-static void
-isert_cq_work(struct work_struct *work)
-{
- enum { isert_poll_budget = 65536 };
- struct isert_comp *comp = container_of(work, struct isert_comp,
- work);
- struct ib_wc *const wcs = comp->wcs;
- int i, n, completed = 0;
-
- while ((n = ib_poll_cq(comp->cq, ARRAY_SIZE(comp->wcs), wcs)) > 0) {
- for (i = 0; i < n; i++)
- isert_handle_wc(&wcs[i]);
-
- completed += n;
- if (completed >= isert_poll_budget)
- break;
- }
-
- ib_req_notify_cq(comp->cq, IB_CQ_NEXT_COMP);
-}
-
-static void
-isert_cq_callback(struct ib_cq *cq, void *context)
-{
- struct isert_comp *comp = context;
-
- queue_work(isert_comp_wq, &comp->work);
-}
-
static int
isert_post_response(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd)
{
@@ -2394,7 +2292,8 @@ isert_build_rdma_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
page_off = offset % PAGE_SIZE;
rdma_wr->wr.sg_list = ib_sge;
- rdma_wr->wr.wr_id = (uintptr_t)&isert_cmd->tx_desc;
+ rdma_wr->wr.wr_cqe = &isert_cmd->tx_desc.tx_cqe;
+
/*
* Perform mapping of TCM scatterlist memory ib_sge dma_addr.
*/
@@ -2427,24 +2326,23 @@ isert_build_rdma_wr(struct isert_conn *isert_conn, struct isert_cmd *isert_cmd,
}
static int
-isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
- struct isert_rdma_wr *wr)
+isert_map_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn)
{
+ struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
struct se_cmd *se_cmd = &cmd->se_cmd;
- struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
struct isert_conn *isert_conn = conn->context;
- struct isert_data_buf *data = &wr->data;
+ struct isert_data_buf *data = &isert_cmd->data;
struct ib_rdma_wr *rdma_wr;
struct ib_sge *ib_sge;
u32 offset, data_len, data_left, rdma_write_max, va_offset = 0;
int ret = 0, i, ib_sge_cnt;
- isert_cmd->tx_desc.isert_cmd = isert_cmd;
-
- offset = wr->iser_ib_op == ISER_IB_RDMA_READ ? cmd->write_data_done : 0;
+ offset = isert_cmd->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);
+ offset, isert_cmd->iser_ib_op,
+ &isert_cmd->data);
if (ret)
return ret;
@@ -2457,41 +2355,44 @@ isert_map_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
ret = -ENOMEM;
goto unmap_cmd;
}
- wr->ib_sge = ib_sge;
+ isert_cmd->ib_sge = ib_sge;
- wr->rdma_wr_num = DIV_ROUND_UP(data->nents, isert_conn->max_sge);
- wr->rdma_wr = kzalloc(sizeof(struct ib_rdma_wr) * wr->rdma_wr_num,
- GFP_KERNEL);
- if (!wr->rdma_wr) {
- isert_dbg("Unable to allocate wr->rdma_wr\n");
+ isert_cmd->rdma_wr_num = DIV_ROUND_UP(data->nents, isert_conn->max_sge);
+ isert_cmd->rdma_wr = kzalloc(sizeof(struct ib_rdma_wr) *
+ isert_cmd->rdma_wr_num, GFP_KERNEL);
+ if (!isert_cmd->rdma_wr) {
+ isert_dbg("Unable to allocate isert_cmd->rdma_wr\n");
ret = -ENOMEM;
goto unmap_cmd;
}
- wr->isert_cmd = isert_cmd;
rdma_write_max = isert_conn->max_sge * PAGE_SIZE;
- for (i = 0; i < wr->rdma_wr_num; i++) {
- rdma_wr = &isert_cmd->rdma_wr.rdma_wr[i];
+ for (i = 0; i < isert_cmd->rdma_wr_num; i++) {
+ rdma_wr = &isert_cmd->rdma_wr[i];
data_len = min(data_left, rdma_write_max);
rdma_wr->wr.send_flags = 0;
- if (wr->iser_ib_op == ISER_IB_RDMA_WRITE) {
+ if (isert_cmd->iser_ib_op == ISER_IB_RDMA_WRITE) {
+ isert_cmd->tx_desc.tx_cqe.done = isert_rdma_write_done;
+
rdma_wr->wr.opcode = IB_WR_RDMA_WRITE;
rdma_wr->remote_addr = isert_cmd->read_va + offset;
rdma_wr->rkey = isert_cmd->read_stag;
- if (i + 1 == wr->rdma_wr_num)
+ if (i + 1 == isert_cmd->rdma_wr_num)
rdma_wr->wr.next = &isert_cmd->tx_desc.send_wr;
else
- rdma_wr->wr.next = &wr->rdma_wr[i + 1].wr;
+ rdma_wr->wr.next = &isert_cmd->rdma_wr[i + 1].wr;
} else {
+ isert_cmd->tx_desc.tx_cqe.done = isert_rdma_read_done;
+
rdma_wr->wr.opcode = IB_WR_RDMA_READ;
rdma_wr->remote_addr = isert_cmd->write_va + va_offset;
rdma_wr->rkey = isert_cmd->write_stag;
- if (i + 1 == wr->rdma_wr_num)
+ if (i + 1 == isert_cmd->rdma_wr_num)
rdma_wr->wr.send_flags = IB_SEND_SIGNALED;
else
- rdma_wr->wr.next = &wr->rdma_wr[i + 1].wr;
+ rdma_wr->wr.next = &isert_cmd->rdma_wr[i + 1].wr;
}
ib_sge_cnt = isert_build_rdma_wr(isert_conn, isert_cmd, ib_sge,
@@ -2516,7 +2417,7 @@ isert_inv_rkey(struct ib_send_wr *inv_wr, struct ib_mr *mr)
u32 rkey;
memset(inv_wr, 0, sizeof(*inv_wr));
- inv_wr->wr_id = ISER_FASTREG_LI_WRID;
+ inv_wr->wr_cqe = NULL;
inv_wr->opcode = IB_WR_LOCAL_INV;
inv_wr->ex.invalidate_rkey = mr->rkey;
@@ -2572,7 +2473,7 @@ isert_fast_reg_mr(struct isert_conn *isert_conn,
reg_wr.wr.next = NULL;
reg_wr.wr.opcode = IB_WR_REG_MR;
- reg_wr.wr.wr_id = ISER_FASTREG_LI_WRID;
+ reg_wr.wr.wr_cqe = NULL;
reg_wr.wr.send_flags = 0;
reg_wr.wr.num_sge = 0;
reg_wr.mr = mr;
@@ -2659,10 +2560,10 @@ isert_set_prot_checks(u8 prot_checks)
static int
isert_reg_sig_mr(struct isert_conn *isert_conn,
- struct se_cmd *se_cmd,
- struct isert_rdma_wr *rdma_wr,
+ struct isert_cmd *isert_cmd,
struct fast_reg_descriptor *fr_desc)
{
+ struct se_cmd *se_cmd = &isert_cmd->iscsi_cmd->se_cmd;
struct ib_sig_handover_wr sig_wr;
struct ib_send_wr inv_wr, *bad_wr, *wr = NULL;
struct pi_context *pi_ctx = fr_desc->pi_ctx;
@@ -2683,14 +2584,14 @@ isert_reg_sig_mr(struct isert_conn *isert_conn,
memset(&sig_wr, 0, sizeof(sig_wr));
sig_wr.wr.opcode = IB_WR_REG_SIG_MR;
- sig_wr.wr.wr_id = ISER_FASTREG_LI_WRID;
- sig_wr.wr.sg_list = &rdma_wr->ib_sg[DATA];
+ sig_wr.wr.wr_cqe = NULL;
+ sig_wr.wr.sg_list = &isert_cmd->ib_sg[DATA];
sig_wr.wr.num_sge = 1;
sig_wr.access_flags = IB_ACCESS_LOCAL_WRITE;
sig_wr.sig_attrs = &sig_attrs;
sig_wr.sig_mr = pi_ctx->sig_mr;
if (se_cmd->t_prot_sg)
- sig_wr.prot = &rdma_wr->ib_sg[PROT];
+ sig_wr.prot = &isert_cmd->ib_sg[PROT];
if (!wr)
wr = &sig_wr.wr;
@@ -2704,35 +2605,34 @@ isert_reg_sig_mr(struct isert_conn *isert_conn,
}
fr_desc->ind &= ~ISERT_SIG_KEY_VALID;
- rdma_wr->ib_sg[SIG].lkey = pi_ctx->sig_mr->lkey;
- rdma_wr->ib_sg[SIG].addr = 0;
- rdma_wr->ib_sg[SIG].length = se_cmd->data_length;
+ isert_cmd->ib_sg[SIG].lkey = pi_ctx->sig_mr->lkey;
+ isert_cmd->ib_sg[SIG].addr = 0;
+ isert_cmd->ib_sg[SIG].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
*/
- rdma_wr->ib_sg[SIG].length += se_cmd->prot_length;
+ isert_cmd->ib_sg[SIG].length += se_cmd->prot_length;
isert_dbg("sig_sge: addr: 0x%llx length: %u lkey: %x\n",
- rdma_wr->ib_sg[SIG].addr, rdma_wr->ib_sg[SIG].length,
- rdma_wr->ib_sg[SIG].lkey);
+ isert_cmd->ib_sg[SIG].addr, isert_cmd->ib_sg[SIG].length,
+ isert_cmd->ib_sg[SIG].lkey);
err:
return ret;
}
static int
isert_handle_prot_cmd(struct isert_conn *isert_conn,
- struct isert_cmd *isert_cmd,
- struct isert_rdma_wr *wr)
+ struct isert_cmd *isert_cmd)
{
struct isert_device *device = isert_conn->device;
struct se_cmd *se_cmd = &isert_cmd->iscsi_cmd->se_cmd;
int ret;
- if (!wr->fr_desc->pi_ctx) {
- ret = isert_create_pi_ctx(wr->fr_desc,
+ if (!isert_cmd->fr_desc->pi_ctx) {
+ ret = isert_create_pi_ctx(isert_cmd->fr_desc,
device->ib_device,
device->pd);
if (ret) {
@@ -2747,16 +2647,20 @@ isert_handle_prot_cmd(struct isert_conn *isert_conn,
se_cmd->t_prot_sg,
se_cmd->t_prot_nents,
se_cmd->prot_length,
- 0, wr->iser_ib_op, &wr->prot);
+ 0,
+ isert_cmd->iser_ib_op,
+ &isert_cmd->prot);
if (ret) {
isert_err("conn %p failed to map protection buffer\n",
isert_conn);
return ret;
}
- memset(&wr->ib_sg[PROT], 0, sizeof(wr->ib_sg[PROT]));
- ret = isert_fast_reg_mr(isert_conn, wr->fr_desc, &wr->prot,
- ISERT_PROT_KEY_VALID, &wr->ib_sg[PROT]);
+ memset(&isert_cmd->ib_sg[PROT], 0, sizeof(isert_cmd->ib_sg[PROT]));
+ ret = isert_fast_reg_mr(isert_conn, isert_cmd->fr_desc,
+ &isert_cmd->prot,
+ ISERT_PROT_KEY_VALID,
+ &isert_cmd->ib_sg[PROT]);
if (ret) {
isert_err("conn %p failed to fast reg mr\n",
isert_conn);
@@ -2764,29 +2668,28 @@ isert_handle_prot_cmd(struct isert_conn *isert_conn,
}
}
- ret = isert_reg_sig_mr(isert_conn, se_cmd, wr, wr->fr_desc);
+ ret = isert_reg_sig_mr(isert_conn, isert_cmd, isert_cmd->fr_desc);
if (ret) {
isert_err("conn %p failed to fast reg mr\n",
isert_conn);
goto unmap_prot_cmd;
}
- wr->fr_desc->ind |= ISERT_PROTECTED;
+ isert_cmd->fr_desc->ind |= ISERT_PROTECTED;
return 0;
unmap_prot_cmd:
if (se_cmd->t_prot_sg)
- isert_unmap_data_buf(isert_conn, &wr->prot);
+ isert_unmap_data_buf(isert_conn, &isert_cmd->prot);
return ret;
}
static int
-isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
- struct isert_rdma_wr *wr)
+isert_reg_rdma(struct isert_cmd *isert_cmd, struct iscsi_conn *conn)
{
+ struct iscsi_cmd *cmd = isert_cmd->iscsi_cmd;
struct se_cmd *se_cmd = &cmd->se_cmd;
- struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
struct isert_conn *isert_conn = conn->context;
struct fast_reg_descriptor *fr_desc = NULL;
struct ib_rdma_wr *rdma_wr;
@@ -2795,57 +2698,61 @@ isert_reg_rdma(struct iscsi_conn *conn, struct iscsi_cmd *cmd,
int ret = 0;
unsigned long flags;
- isert_cmd->tx_desc.isert_cmd = isert_cmd;
-
- offset = wr->iser_ib_op == ISER_IB_RDMA_READ ? cmd->write_data_done : 0;
+ offset = isert_cmd->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);
+ offset, isert_cmd->iser_ib_op,
+ &isert_cmd->data);
if (ret)
return ret;
- if (wr->data.dma_nents != 1 || isert_prot_cmd(isert_conn, se_cmd)) {
+ if (isert_cmd->data.dma_nents != 1 ||
+ isert_prot_cmd(isert_conn, se_cmd)) {
spin_lock_irqsave(&isert_conn->pool_lock, flags);
fr_desc = list_first_entry(&isert_conn->fr_pool,
struct fast_reg_descriptor, list);
list_del(&fr_desc->list);
spin_unlock_irqrestore(&isert_conn->pool_lock, flags);
- wr->fr_desc = fr_desc;
+ isert_cmd->fr_desc = fr_desc;
}
- ret = isert_fast_reg_mr(isert_conn, fr_desc, &wr->data,
- ISERT_DATA_KEY_VALID, &wr->ib_sg[DATA]);
+ ret = isert_fast_reg_mr(isert_conn, fr_desc, &isert_cmd->data,
+ ISERT_DATA_KEY_VALID, &isert_cmd->ib_sg[DATA]);
if (ret)
goto unmap_cmd;
if (isert_prot_cmd(isert_conn, se_cmd)) {
- ret = isert_handle_prot_cmd(isert_conn, isert_cmd, wr);
+ ret = isert_handle_prot_cmd(isert_conn, isert_cmd);
if (ret)
goto unmap_cmd;
- ib_sg = &wr->ib_sg[SIG];
+ ib_sg = &isert_cmd->ib_sg[SIG];
} else {
- ib_sg = &wr->ib_sg[DATA];
+ ib_sg = &isert_cmd->ib_sg[DATA];
}
- memcpy(&wr->s_ib_sge, ib_sg, sizeof(*ib_sg));
- wr->ib_sge = &wr->s_ib_sge;
- wr->rdma_wr_num = 1;
- memset(&wr->s_rdma_wr, 0, sizeof(wr->s_rdma_wr));
- wr->rdma_wr = &wr->s_rdma_wr;
- wr->isert_cmd = isert_cmd;
+ memcpy(&isert_cmd->s_ib_sge, ib_sg, sizeof(*ib_sg));
+ isert_cmd->ib_sge = &isert_cmd->s_ib_sge;
+ isert_cmd->rdma_wr_num = 1;
+ memset(&isert_cmd->s_rdma_wr, 0, sizeof(isert_cmd->s_rdma_wr));
+ isert_cmd->rdma_wr = &isert_cmd->s_rdma_wr;
- rdma_wr = &isert_cmd->rdma_wr.s_rdma_wr;
- rdma_wr->wr.sg_list = &wr->s_ib_sge;
+ rdma_wr = &isert_cmd->s_rdma_wr;
+ rdma_wr->wr.sg_list = &isert_cmd->s_ib_sge;
rdma_wr->wr.num_sge = 1;
- rdma_wr->wr.wr_id = (uintptr_t)&isert_cmd->tx_desc;
- if (wr->iser_ib_op == ISER_IB_RDMA_WRITE) {
+ rdma_wr->wr.wr_cqe = &isert_cmd->tx_desc.tx_cqe;
+ if (isert_cmd->iser_ib_op == ISER_IB_RDMA_WRITE) {
+ isert_cmd->tx_desc.tx_cqe.done = isert_rdma_write_done;
+
rdma_wr->wr.opcode = IB_WR_RDMA_WRITE;
rdma_wr->remote_addr = isert_cmd->read_va;
rdma_wr->rkey = isert_cmd->read_stag;
rdma_wr->wr.send_flags = !isert_prot_cmd(isert_conn, se_cmd) ?
0 : IB_SEND_SIGNALED;
} else {
+ isert_cmd->tx_desc.tx_cqe.done = isert_rdma_read_done;
+
rdma_wr->wr.opcode = IB_WR_RDMA_READ;
rdma_wr->remote_addr = isert_cmd->write_va;
rdma_wr->rkey = isert_cmd->write_stag;
@@ -2860,7 +2767,7 @@ unmap_cmd:
list_add_tail(&fr_desc->list, &isert_conn->fr_pool);
spin_unlock_irqrestore(&isert_conn->pool_lock, flags);
}
- isert_unmap_data_buf(isert_conn, &wr->data);
+ isert_unmap_data_buf(isert_conn, &isert_cmd->data);
return ret;
}
@@ -2870,7 +2777,6 @@ isert_put_datain(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_rdma_wr *wr = &isert_cmd->rdma_wr;
struct isert_conn *isert_conn = conn->context;
struct isert_device *device = isert_conn->device;
struct ib_send_wr *wr_failed;
@@ -2879,8 +2785,8 @@ isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
isert_dbg("Cmd: %p RDMA_WRITE data_length: %u\n",
isert_cmd, se_cmd->data_length);
- wr->iser_ib_op = ISER_IB_RDMA_WRITE;
- rc = device->reg_rdma_mem(conn, cmd, wr);
+ isert_cmd->iser_ib_op = ISER_IB_RDMA_WRITE;
+ rc = device->reg_rdma_mem(isert_cmd, conn);
if (rc) {
isert_err("Cmd: %p failed to prepare RDMA res\n", isert_cmd);
return rc;
@@ -2897,8 +2803,8 @@ isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
isert_init_tx_hdrs(isert_conn, &isert_cmd->tx_desc);
isert_init_send_wr(isert_conn, isert_cmd,
&isert_cmd->tx_desc.send_wr);
- isert_cmd->rdma_wr.s_rdma_wr.wr.next = &isert_cmd->tx_desc.send_wr;
- wr->rdma_wr_num += 1;
+ isert_cmd->s_rdma_wr.wr.next = &isert_cmd->tx_desc.send_wr;
+ isert_cmd->rdma_wr_num += 1;
rc = isert_post_recv(isert_conn, isert_cmd->rx_desc);
if (rc) {
@@ -2907,7 +2813,7 @@ isert_put_datain(struct iscsi_conn *conn, struct iscsi_cmd *cmd)
}
}
- rc = ib_post_send(isert_conn->qp, &wr->rdma_wr->wr, &wr_failed);
+ rc = ib_post_send(isert_conn->qp, &isert_cmd->rdma_wr->wr, &wr_failed);
if (rc)
isert_warn("ib_post_send() failed for IB_WR_RDMA_WRITE\n");
@@ -2926,7 +2832,6 @@ isert_get_dataout(struct iscsi_conn *conn, struct iscsi_cmd *cmd, bool recovery)
{
struct se_cmd *se_cmd = &cmd->se_cmd;
struct isert_cmd *isert_cmd = iscsit_priv_cmd(cmd);
- struct isert_rdma_wr *wr = &isert_cmd->rdma_wr;
struct isert_conn *isert_conn = conn->context;
struct isert_device *device = isert_conn->device;
struct ib_send_wr *wr_failed;
@@ -2934,14 +2839,14 @@ isert_get_dataout(struct iscsi_conn *conn, struct iscsi_cmd *cmd, bool recovery)
isert_dbg("Cmd: %p RDMA_READ data_length: %u write_data_done: %u\n",
isert_cmd, se_cmd->data_length, cmd->write_data_done);
- wr->iser_ib_op = ISER_IB_RDMA_READ;
- rc = device->reg_rdma_mem(conn, cmd, wr);
+ isert_cmd->iser_ib_op = ISER_IB_RDMA_READ;
+ rc = device->reg_rdma_mem(isert_cmd, conn);
if (rc) {
isert_err("Cmd: %p failed to prepare RDMA res\n", isert_cmd);
return rc;
}
- rc = ib_post_send(isert_conn->qp, &wr->rdma_wr->wr, &wr_failed);
+ rc = ib_post_send(isert_conn->qp, &isert_cmd->rdma_wr->wr, &wr_failed);
if (rc)
isert_warn("ib_post_send() failed for IB_WR_RDMA_READ\n");
@@ -3274,8 +3179,6 @@ static void isert_release_work(struct work_struct *work)
isert_info("Starting release conn %p\n", isert_conn);
- wait_for_completion(&isert_conn->wait);
-
mutex_lock(&isert_conn->mutex);
isert_conn->state = ISER_CONN_DOWN;
mutex_unlock(&isert_conn->mutex);
@@ -3309,24 +3212,6 @@ isert_wait4cmds(struct iscsi_conn *conn)
}
}
-static void
-isert_wait4flush(struct isert_conn *isert_conn)
-{
- struct ib_recv_wr *bad_wr;
-
- isert_info("conn %p\n", isert_conn);
-
- init_completion(&isert_conn->wait_comp_err);
- isert_conn->beacon.wr_id = ISER_BEACON_WRID;
- /* post an indication that all flush errors were consumed */
- if (ib_post_recv(isert_conn->qp, &isert_conn->beacon, &bad_wr)) {
- isert_err("conn %p failed to post beacon", isert_conn);
- return;
- }
-
- wait_for_completion(&isert_conn->wait_comp_err);
-}
-
/**
* isert_put_unsol_pending_cmds() - Drop commands waiting for
* unsolicitate dataout
@@ -3369,18 +3254,10 @@ static void isert_wait_conn(struct iscsi_conn *conn)
isert_info("Starting conn %p\n", isert_conn);
mutex_lock(&isert_conn->mutex);
- /*
- * Only wait for wait_comp_err if the isert_conn made it
- * into full feature phase..
- */
- if (isert_conn->state == ISER_CONN_INIT) {
- mutex_unlock(&isert_conn->mutex);
- return;
- }
isert_conn_terminate(isert_conn);
mutex_unlock(&isert_conn->mutex);
- isert_wait4flush(isert_conn);
+ ib_drain_qp(isert_conn->qp);
isert_put_unsol_pending_cmds(conn);
isert_wait4cmds(conn);
isert_wait4logout(isert_conn);
@@ -3392,7 +3269,7 @@ static void isert_free_conn(struct iscsi_conn *conn)
{
struct isert_conn *isert_conn = conn->context;
- isert_wait4flush(isert_conn);
+ ib_drain_qp(isert_conn->qp);
isert_put_conn(isert_conn);
}
diff --git a/drivers/infiniband/ulp/isert/ib_isert.h b/drivers/infiniband/ulp/isert/ib_isert.h
index 1aa019ab9..147900cbb 100644
--- a/drivers/infiniband/ulp/isert/ib_isert.h
+++ b/drivers/infiniband/ulp/isert/ib_isert.h
@@ -36,9 +36,7 @@
/* Constant PDU lengths calculations */
#define ISER_HEADERS_LEN (sizeof(struct iser_ctrl) + \
sizeof(struct iscsi_hdr))
-#define ISER_RECV_DATA_SEG_LEN 8192
-#define ISER_RX_PAYLOAD_SIZE (ISER_HEADERS_LEN + ISER_RECV_DATA_SEG_LEN)
-#define ISER_RX_LOGIN_SIZE (ISER_HEADERS_LEN + ISCSI_DEF_MAX_RECV_SEG_LEN)
+#define ISER_RX_PAYLOAD_SIZE (ISER_HEADERS_LEN + ISCSI_DEF_MAX_RECV_SEG_LEN)
/* QP settings */
/* Maximal bounds on received asynchronous PDUs */
@@ -62,12 +60,11 @@
ISERT_MAX_TX_MISC_PDUS + \
ISERT_MAX_RX_MISC_PDUS)
-#define ISER_RX_PAD_SIZE (ISER_RECV_DATA_SEG_LEN + 4096 - \
- (ISER_RX_PAYLOAD_SIZE + sizeof(u64) + sizeof(struct ib_sge)))
+#define ISER_RX_PAD_SIZE (ISCSI_DEF_MAX_RECV_SEG_LEN + 4096 - \
+ (ISER_RX_PAYLOAD_SIZE + sizeof(u64) + sizeof(struct ib_sge) + \
+ sizeof(struct ib_cqe)))
#define ISCSI_ISER_SG_TABLESIZE 256
-#define ISER_FASTREG_LI_WRID 0xffffffffffffffffULL
-#define ISER_BEACON_WRID 0xfffffffffffffffeULL
enum isert_desc_type {
ISCSI_TX_CONTROL,
@@ -93,23 +90,35 @@ enum iser_conn_state {
struct iser_rx_desc {
struct iser_ctrl iser_header;
struct iscsi_hdr iscsi_header;
- char data[ISER_RECV_DATA_SEG_LEN];
+ char data[ISCSI_DEF_MAX_RECV_SEG_LEN];
u64 dma_addr;
struct ib_sge rx_sg;
+ struct ib_cqe rx_cqe;
char pad[ISER_RX_PAD_SIZE];
} __packed;
+static inline struct iser_rx_desc *cqe_to_rx_desc(struct ib_cqe *cqe)
+{
+ return container_of(cqe, struct iser_rx_desc, rx_cqe);
+}
+
struct iser_tx_desc {
struct iser_ctrl iser_header;
struct iscsi_hdr iscsi_header;
enum isert_desc_type type;
u64 dma_addr;
struct ib_sge tx_sg[2];
+ struct ib_cqe tx_cqe;
int num_sge;
- struct isert_cmd *isert_cmd;
struct ib_send_wr send_wr;
} __packed;
+static inline struct iser_tx_desc *cqe_to_tx_desc(struct ib_cqe *cqe)
+{
+ return container_of(cqe, struct iser_tx_desc, tx_cqe);
+}
+
+
enum isert_indicator {
ISERT_PROTECTED = 1 << 0,
ISERT_DATA_KEY_VALID = 1 << 1,
@@ -145,20 +154,6 @@ enum {
SIG = 2,
};
-struct isert_rdma_wr {
- struct isert_cmd *isert_cmd;
- enum iser_ib_op_code iser_ib_op;
- struct ib_sge *ib_sge;
- struct ib_sge s_ib_sge;
- int rdma_wr_num;
- struct ib_rdma_wr *rdma_wr;
- struct ib_rdma_wr s_rdma_wr;
- struct ib_sge ib_sg[3];
- struct isert_data_buf data;
- struct isert_data_buf prot;
- struct fast_reg_descriptor *fr_desc;
-};
-
struct isert_cmd {
uint32_t read_stag;
uint32_t write_stag;
@@ -171,11 +166,25 @@ struct isert_cmd {
struct iscsi_cmd *iscsi_cmd;
struct iser_tx_desc tx_desc;
struct iser_rx_desc *rx_desc;
- struct isert_rdma_wr rdma_wr;
+ enum iser_ib_op_code iser_ib_op;
+ struct ib_sge *ib_sge;
+ struct ib_sge s_ib_sge;
+ int rdma_wr_num;
+ struct ib_rdma_wr *rdma_wr;
+ struct ib_rdma_wr s_rdma_wr;
+ struct ib_sge ib_sg[3];
+ struct isert_data_buf data;
+ struct isert_data_buf prot;
+ struct fast_reg_descriptor *fr_desc;
struct work_struct comp_work;
struct scatterlist sg;
};
+static inline struct isert_cmd *tx_desc_to_cmd(struct iser_tx_desc *desc)
+{
+ return container_of(desc, struct isert_cmd, tx_desc);
+}
+
struct isert_device;
struct isert_conn {
@@ -184,8 +193,7 @@ struct isert_conn {
u32 initiator_depth;
bool pi_support;
u32 max_sge;
- char *login_buf;
- char *login_req_buf;
+ struct iser_rx_desc *login_req_buf;
char *login_rsp_buf;
u64 login_req_dma;
int login_req_len;
@@ -201,15 +209,12 @@ struct isert_conn {
struct ib_qp *qp;
struct isert_device *device;
struct mutex mutex;
- struct completion wait;
- struct completion wait_comp_err;
struct kref kref;
struct list_head fr_pool;
int fr_pool_size;
/* lock to protect fastreg pool */
spinlock_t pool_lock;
struct work_struct release_work;
- struct ib_recv_wr beacon;
bool logout_posted;
bool snd_w_inv;
};
@@ -221,17 +226,13 @@ struct isert_conn {
*
* @device: pointer to device handle
* @cq: completion queue
- * @wcs: work completion array
* @active_qps: Number of active QPs attached
* to completion context
- * @work: completion work handle
*/
struct isert_comp {
struct isert_device *device;
struct ib_cq *cq;
- struct ib_wc wcs[16];
int active_qps;
- struct work_struct work;
};
struct isert_device {
@@ -243,9 +244,8 @@ struct isert_device {
struct isert_comp *comps;
int comps_used;
struct list_head dev_node;
- int (*reg_rdma_mem)(struct iscsi_conn *conn,
- struct iscsi_cmd *cmd,
- struct isert_rdma_wr *wr);
+ int (*reg_rdma_mem)(struct isert_cmd *isert_cmd,
+ struct iscsi_conn *conn);
void (*unreg_rdma_mem)(struct isert_cmd *isert_cmd,
struct isert_conn *isert_conn);
};
diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c
index 03022f642..845ce90c2 100644
--- a/drivers/infiniband/ulp/srp/ib_srp.c
+++ b/drivers/infiniband/ulp/srp/ib_srp.c
@@ -446,50 +446,18 @@ static struct srp_fr_pool *srp_alloc_fr_pool(struct srp_target_port *target)
dev->max_pages_per_mr);
}
-static void srp_drain_done(struct ib_cq *cq, struct ib_wc *wc)
-{
- struct srp_rdma_ch *ch = cq->cq_context;
-
- complete(&ch->done);
-}
-
-static struct ib_cqe srp_drain_cqe = {
- .done = srp_drain_done,
-};
-
/**
* srp_destroy_qp() - destroy an RDMA queue pair
- * @ch: SRP RDMA channel.
+ * @qp: RDMA queue pair.
*
- * Change a queue pair into the error state and wait until all receive
- * completions have been processed before destroying it. This avoids that
- * the receive completion handler can access the queue pair while it is
+ * Drain the qp before destroying it. This avoids that the receive
+ * completion handler can access the queue pair while it is
* being destroyed.
*/
-static void srp_destroy_qp(struct srp_rdma_ch *ch)
+static void srp_destroy_qp(struct ib_qp *qp)
{
- static struct ib_qp_attr attr = { .qp_state = IB_QPS_ERR };
- static struct ib_recv_wr wr = { 0 };
- struct ib_recv_wr *bad_wr;
- int ret;
-
- wr.wr_cqe = &srp_drain_cqe;
- /* Destroying a QP and reusing ch->done is only safe if not connected */
- WARN_ON_ONCE(ch->connected);
-
- ret = ib_modify_qp(ch->qp, &attr, IB_QP_STATE);
- WARN_ONCE(ret, "ib_cm_init_qp_attr() returned %d\n", ret);
- if (ret)
- goto out;
-
- init_completion(&ch->done);
- ret = ib_post_recv(ch->qp, &wr, &bad_wr);
- WARN_ONCE(ret, "ib_post_recv() returned %d\n", ret);
- if (ret == 0)
- wait_for_completion(&ch->done);
-
-out:
- ib_destroy_qp(ch->qp);
+ ib_drain_rq(qp);
+ ib_destroy_qp(qp);
}
static int srp_create_ch_ib(struct srp_rdma_ch *ch)
@@ -508,7 +476,7 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch)
if (!init_attr)
return -ENOMEM;
- /* queue_size + 1 for ib_drain_qp */
+ /* queue_size + 1 for ib_drain_rq() */
recv_cq = ib_alloc_cq(dev->dev, ch, target->queue_size + 1,
ch->comp_vector, IB_POLL_SOFTIRQ);
if (IS_ERR(recv_cq)) {
@@ -562,7 +530,7 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch)
}
if (ch->qp)
- srp_destroy_qp(ch);
+ srp_destroy_qp(ch->qp);
if (ch->recv_cq)
ib_free_cq(ch->recv_cq);
if (ch->send_cq)
@@ -586,7 +554,7 @@ static int srp_create_ch_ib(struct srp_rdma_ch *ch)
return 0;
err_qp:
- srp_destroy_qp(ch);
+ srp_destroy_qp(qp);
err_send_cq:
ib_free_cq(send_cq);
@@ -629,7 +597,7 @@ static void srp_free_ch_ib(struct srp_target_port *target,
ib_destroy_fmr_pool(ch->fmr_pool);
}
- srp_destroy_qp(ch);
+ srp_destroy_qp(ch->qp);
ib_free_cq(ch->send_cq);
ib_free_cq(ch->recv_cq);
@@ -1541,7 +1509,7 @@ static int srp_map_idb(struct srp_rdma_ch *ch, struct srp_request *req,
if (dev->use_fast_reg) {
state.sg = idb_sg;
- sg_set_buf(idb_sg, req->indirect_desc, idb_len);
+ sg_init_one(idb_sg, req->indirect_desc, idb_len);
idb_sg->dma_address = req->indirect_dma_addr; /* hack! */
#ifdef CONFIG_NEED_SG_DMA_LENGTH
idb_sg->dma_length = idb_sg->length; /* hack^2 */
diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c
index 4328679a6..8b42401d4 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.c
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.c
@@ -91,76 +91,32 @@ MODULE_PARM_DESC(srpt_service_guid,
" instead of using the node_guid of the first HCA.");
static struct ib_client srpt_client;
-static void srpt_release_channel(struct srpt_rdma_ch *ch);
+static void srpt_release_cmd(struct se_cmd *se_cmd);
+static void srpt_free_ch(struct kref *kref);
static int srpt_queue_status(struct se_cmd *cmd);
static void srpt_recv_done(struct ib_cq *cq, struct ib_wc *wc);
static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc);
+static void srpt_process_wait_list(struct srpt_rdma_ch *ch);
-/**
- * opposite_dma_dir() - Swap DMA_TO_DEVICE and DMA_FROM_DEVICE.
- */
-static inline
-enum dma_data_direction opposite_dma_dir(enum dma_data_direction dir)
-{
- switch (dir) {
- case DMA_TO_DEVICE: return DMA_FROM_DEVICE;
- case DMA_FROM_DEVICE: return DMA_TO_DEVICE;
- default: return dir;
- }
-}
-
-/**
- * srpt_sdev_name() - Return the name associated with the HCA.
- *
- * Examples are ib0, ib1, ...
- */
-static inline const char *srpt_sdev_name(struct srpt_device *sdev)
-{
- return sdev->device->name;
-}
-
-static enum rdma_ch_state srpt_get_ch_state(struct srpt_rdma_ch *ch)
-{
- unsigned long flags;
- enum rdma_ch_state state;
-
- spin_lock_irqsave(&ch->spinlock, flags);
- state = ch->state;
- spin_unlock_irqrestore(&ch->spinlock, flags);
- return state;
-}
-
-static enum rdma_ch_state
-srpt_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state new_state)
-{
- unsigned long flags;
- enum rdma_ch_state prev;
-
- spin_lock_irqsave(&ch->spinlock, flags);
- prev = ch->state;
- ch->state = new_state;
- spin_unlock_irqrestore(&ch->spinlock, flags);
- return prev;
-}
-
-/**
- * srpt_test_and_set_ch_state() - Test and set the channel state.
- *
- * Returns true if and only if the channel state has been set to the new state.
+/*
+ * The only allowed channel state changes are those that change the channel
+ * state into a state with a higher numerical value. Hence the new > prev test.
*/
-static bool
-srpt_test_and_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state old,
- enum rdma_ch_state new)
+static bool srpt_set_ch_state(struct srpt_rdma_ch *ch, enum rdma_ch_state new)
{
unsigned long flags;
enum rdma_ch_state prev;
+ bool changed = false;
spin_lock_irqsave(&ch->spinlock, flags);
prev = ch->state;
- if (prev == old)
+ if (new > prev) {
ch->state = new;
+ changed = true;
+ }
spin_unlock_irqrestore(&ch->spinlock, flags);
- return prev == old;
+
+ return changed;
}
/**
@@ -182,7 +138,7 @@ static void srpt_event_handler(struct ib_event_handler *handler,
return;
pr_debug("ASYNC event= %d on device= %s\n", event->event,
- srpt_sdev_name(sdev));
+ sdev->device->name);
switch (event->event) {
case IB_EVENT_PORT_ERR:
@@ -220,25 +176,39 @@ static void srpt_srq_event(struct ib_event *event, void *ctx)
pr_info("SRQ event %d\n", event->event);
}
+static const char *get_ch_state_name(enum rdma_ch_state s)
+{
+ switch (s) {
+ case CH_CONNECTING:
+ return "connecting";
+ case CH_LIVE:
+ return "live";
+ case CH_DISCONNECTING:
+ return "disconnecting";
+ case CH_DRAINING:
+ return "draining";
+ case CH_DISCONNECTED:
+ return "disconnected";
+ }
+ return "???";
+}
+
/**
* srpt_qp_event() - QP event callback function.
*/
static void srpt_qp_event(struct ib_event *event, struct srpt_rdma_ch *ch)
{
pr_debug("QP event %d on cm_id=%p sess_name=%s state=%d\n",
- event->event, ch->cm_id, ch->sess_name, srpt_get_ch_state(ch));
+ event->event, ch->cm_id, ch->sess_name, ch->state);
switch (event->event) {
case IB_EVENT_COMM_EST:
ib_cm_notify(ch->cm_id, event->event);
break;
case IB_EVENT_QP_LAST_WQE_REACHED:
- if (srpt_test_and_set_ch_state(ch, CH_DRAINING,
- CH_RELEASING))
- srpt_release_channel(ch);
- else
- pr_debug("%s: state %d - ignored LAST_WQE.\n",
- ch->sess_name, srpt_get_ch_state(ch));
+ pr_debug("%s-%d, state %s: received Last WQE event.\n",
+ ch->sess_name, ch->qp->qp_num,
+ get_ch_state_name(ch->state));
break;
default:
pr_err("received unrecognized IB QP event %d\n", event->event);
@@ -281,7 +251,7 @@ static void srpt_get_class_port_info(struct ib_dm_mad *mad)
struct ib_class_port_info *cif;
cif = (struct ib_class_port_info *)mad->data;
- memset(cif, 0, sizeof *cif);
+ memset(cif, 0, sizeof(*cif));
cif->base_version = 1;
cif->class_version = 1;
cif->resp_time_value = 20;
@@ -340,7 +310,7 @@ static void srpt_get_ioc(struct srpt_port *sport, u32 slot,
return;
}
- memset(iocp, 0, sizeof *iocp);
+ memset(iocp, 0, sizeof(*iocp));
strcpy(iocp->id_string, SRPT_ID_STRING);
iocp->guid = cpu_to_be64(srpt_service_guid);
iocp->vendor_id = cpu_to_be32(sdev->device->attrs.vendor_id);
@@ -390,7 +360,7 @@ static void srpt_get_svc_entries(u64 ioc_guid,
}
svc_entries = (struct ib_dm_svc_entries *)mad->data;
- memset(svc_entries, 0, sizeof *svc_entries);
+ memset(svc_entries, 0, sizeof(*svc_entries));
svc_entries->service_entries[0].id = cpu_to_be64(ioc_guid);
snprintf(svc_entries->service_entries[0].name,
sizeof(svc_entries->service_entries[0].name),
@@ -484,7 +454,7 @@ static void srpt_mad_recv_handler(struct ib_mad_agent *mad_agent,
rsp->ah = ah;
dm_mad = rsp->mad;
- memcpy(dm_mad, mad_wc->recv_buf.mad, sizeof *dm_mad);
+ memcpy(dm_mad, mad_wc->recv_buf.mad, sizeof(*dm_mad));
dm_mad->mad_hdr.method = IB_MGMT_METHOD_GET_RESP;
dm_mad->mad_hdr.status = 0;
@@ -532,7 +502,7 @@ static int srpt_refresh_port(struct srpt_port *sport)
struct ib_port_attr port_attr;
int ret;
- memset(&port_modify, 0, sizeof port_modify);
+ memset(&port_modify, 0, sizeof(port_modify));
port_modify.set_port_cap_mask = IB_PORT_DEVICE_MGMT_SUP;
port_modify.clr_port_cap_mask = 0;
@@ -553,7 +523,7 @@ static int srpt_refresh_port(struct srpt_port *sport)
goto err_query_port;
if (!sport->mad_agent) {
- memset(&reg_req, 0, sizeof reg_req);
+ memset(&reg_req, 0, sizeof(reg_req));
reg_req.mgmt_class = IB_MGMT_CLASS_DEVICE_MGMT;
reg_req.mgmt_class_version = IB_MGMT_BASE_VERSION;
set_bit(IB_MGMT_METHOD_GET, reg_req.method_mask);
@@ -841,6 +811,39 @@ out:
}
/**
+ * srpt_zerolength_write() - Perform a zero-length RDMA write.
+ *
+ * A quote from the InfiniBand specification: C9-88: For an HCA responder
+ * using Reliable Connection service, for each zero-length RDMA READ or WRITE
+ * request, the R_Key shall not be validated, even if the request includes
+ * Immediate data.
+ */
+static int srpt_zerolength_write(struct srpt_rdma_ch *ch)
+{
+ struct ib_send_wr wr, *bad_wr;
+
+ memset(&wr, 0, sizeof(wr));
+ wr.opcode = IB_WR_RDMA_WRITE;
+ wr.wr_cqe = &ch->zw_cqe;
+ wr.send_flags = IB_SEND_SIGNALED;
+ return ib_post_send(ch->qp, &wr, &bad_wr);
+}
+
+static void srpt_zerolength_write_done(struct ib_cq *cq, struct ib_wc *wc)
+{
+ struct srpt_rdma_ch *ch = cq->cq_context;
+
+ if (wc->status == IB_WC_SUCCESS) {
+ srpt_process_wait_list(ch);
+ } else {
+ if (srpt_set_ch_state(ch, CH_DISCONNECTED))
+ schedule_work(&ch->release_work);
+ else
+ WARN_ONCE(1, "%s-%d\n", ch->sess_name, ch->qp->qp_num);
+ }
+}
+
+/**
* srpt_get_desc_tbl() - Parse the data descriptors of an SRP_CMD request.
* @ioctx: Pointer to the I/O context associated with the request.
* @srp_cmd: Pointer to the SRP_CMD request data.
@@ -903,14 +906,14 @@ static int srpt_get_desc_tbl(struct srpt_send_ioctx *ioctx,
db = (struct srp_direct_buf *)(srp_cmd->add_data
+ add_cdb_offset);
- memcpy(ioctx->rbufs, db, sizeof *db);
+ memcpy(ioctx->rbufs, db, sizeof(*db));
*data_len = be32_to_cpu(db->len);
} else if (((srp_cmd->buf_fmt & 0xf) == SRP_DATA_DESC_INDIRECT) ||
((srp_cmd->buf_fmt >> 4) == SRP_DATA_DESC_INDIRECT)) {
idb = (struct srp_indirect_buf *)(srp_cmd->add_data
+ add_cdb_offset);
- ioctx->n_rbuf = be32_to_cpu(idb->table_desc.len) / sizeof *db;
+ ioctx->n_rbuf = be32_to_cpu(idb->table_desc.len) / sizeof(*db);
if (ioctx->n_rbuf >
(srp_cmd->data_out_desc_cnt + srp_cmd->data_in_desc_cnt)) {
@@ -929,7 +932,7 @@ static int srpt_get_desc_tbl(struct srpt_send_ioctx *ioctx,
ioctx->rbufs = &ioctx->single_rbuf;
else {
ioctx->rbufs =
- kmalloc(ioctx->n_rbuf * sizeof *db, GFP_ATOMIC);
+ kmalloc(ioctx->n_rbuf * sizeof(*db), GFP_ATOMIC);
if (!ioctx->rbufs) {
ioctx->n_rbuf = 0;
ret = -ENOMEM;
@@ -938,7 +941,7 @@ static int srpt_get_desc_tbl(struct srpt_send_ioctx *ioctx,
}
db = idb->desc_list;
- memcpy(ioctx->rbufs, db, ioctx->n_rbuf * sizeof *db);
+ memcpy(ioctx->rbufs, db, ioctx->n_rbuf * sizeof(*db));
*data_len = be32_to_cpu(idb->len);
}
out:
@@ -956,7 +959,7 @@ static int srpt_init_ch_qp(struct srpt_rdma_ch *ch, struct ib_qp *qp)
struct ib_qp_attr *attr;
int ret;
- attr = kzalloc(sizeof *attr, GFP_KERNEL);
+ attr = kzalloc(sizeof(*attr), GFP_KERNEL);
if (!attr)
return -ENOMEM;
@@ -1070,7 +1073,7 @@ static void srpt_unmap_sg_to_ib_sge(struct srpt_rdma_ch *ch,
dir = ioctx->cmd.data_direction;
BUG_ON(dir == DMA_NONE);
ib_dma_unmap_sg(ch->sport->sdev->device, sg, ioctx->sg_cnt,
- opposite_dma_dir(dir));
+ target_reverse_dma_direction(&ioctx->cmd));
ioctx->mapped_sg_count = 0;
}
}
@@ -1107,7 +1110,7 @@ static int srpt_map_sg_to_ib_sge(struct srpt_rdma_ch *ch,
ioctx->sg_cnt = sg_cnt = cmd->t_data_nents;
count = ib_dma_map_sg(ch->sport->sdev->device, sg, sg_cnt,
- opposite_dma_dir(dir));
+ target_reverse_dma_direction(cmd));
if (unlikely(!count))
return -EAGAIN;
@@ -1313,10 +1316,7 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
/*
* If the command is in a state where the target core is waiting for
- * the ib_srpt driver, change the state to the next state. Changing
- * the state of the command from SRPT_STATE_NEED_DATA to
- * SRPT_STATE_DATA_IN ensures that srpt_xmit_response() will call this
- * function a second time.
+ * the ib_srpt driver, change the state to the next state.
*/
spin_lock_irqsave(&ioctx->spinlock, flags);
@@ -1325,25 +1325,17 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
case SRPT_STATE_NEED_DATA:
ioctx->state = SRPT_STATE_DATA_IN;
break;
- case SRPT_STATE_DATA_IN:
case SRPT_STATE_CMD_RSP_SENT:
case SRPT_STATE_MGMT_RSP_SENT:
ioctx->state = SRPT_STATE_DONE;
break;
default:
+ WARN_ONCE(true, "%s: unexpected I/O context state %d\n",
+ __func__, state);
break;
}
spin_unlock_irqrestore(&ioctx->spinlock, flags);
- if (state == SRPT_STATE_DONE) {
- struct srpt_rdma_ch *ch = ioctx->ch;
-
- BUG_ON(ch->sess == NULL);
-
- target_put_sess_cmd(&ioctx->cmd);
- goto out;
- }
-
pr_debug("Aborting cmd with state %d and tag %lld\n", state,
ioctx->cmd.tag);
@@ -1351,19 +1343,16 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
case SRPT_STATE_NEW:
case SRPT_STATE_DATA_IN:
case SRPT_STATE_MGMT:
+ case SRPT_STATE_DONE:
/*
* Do nothing - defer abort processing until
* srpt_queue_response() is invoked.
*/
- WARN_ON(!transport_check_aborted_status(&ioctx->cmd, false));
break;
case SRPT_STATE_NEED_DATA:
- /* DMA_TO_DEVICE (write) - RDMA read error. */
-
- /* XXX(hch): this is a horrible layering violation.. */
- spin_lock_irqsave(&ioctx->cmd.t_state_lock, flags);
- ioctx->cmd.transport_state &= ~CMD_T_ACTIVE;
- spin_unlock_irqrestore(&ioctx->cmd.t_state_lock, flags);
+ pr_debug("tag %#llx: RDMA read error\n", ioctx->cmd.tag);
+ transport_generic_request_failure(&ioctx->cmd,
+ TCM_CHECK_CONDITION_ABORT_CMD);
break;
case SRPT_STATE_CMD_RSP_SENT:
/*
@@ -1371,18 +1360,16 @@ static int srpt_abort_cmd(struct srpt_send_ioctx *ioctx)
* not been received in time.
*/
srpt_unmap_sg_to_ib_sge(ioctx->ch, ioctx);
- target_put_sess_cmd(&ioctx->cmd);
+ transport_generic_free_cmd(&ioctx->cmd, 0);
break;
case SRPT_STATE_MGMT_RSP_SENT:
- srpt_set_cmd_state(ioctx, SRPT_STATE_DONE);
- target_put_sess_cmd(&ioctx->cmd);
+ transport_generic_free_cmd(&ioctx->cmd, 0);
break;
default:
WARN(1, "Unexpected command state (%d)", state);
break;
}
-out:
return state;
}
@@ -1422,9 +1409,14 @@ static void srpt_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
container_of(wc->wr_cqe, struct srpt_send_ioctx, rdma_cqe);
if (unlikely(wc->status != IB_WC_SUCCESS)) {
+ /*
+ * Note: if an RDMA write error completion is received that
+ * means that a SEND also has been posted. Defer further
+ * processing of the associated command until the send error
+ * completion has been received.
+ */
pr_info("RDMA_WRITE for ioctx 0x%p failed with status %d\n",
ioctx, wc->status);
- srpt_abort_cmd(ioctx);
}
}
@@ -1464,7 +1456,7 @@ static int srpt_build_cmd_rsp(struct srpt_rdma_ch *ch,
sense_data_len = ioctx->cmd.scsi_sense_length;
WARN_ON(sense_data_len > sizeof(ioctx->sense_data));
- memset(srp_rsp, 0, sizeof *srp_rsp);
+ memset(srp_rsp, 0, sizeof(*srp_rsp));
srp_rsp->opcode = SRP_RSP;
srp_rsp->req_lim_delta =
cpu_to_be32(1 + atomic_xchg(&ch->req_lim_delta, 0));
@@ -1514,7 +1506,7 @@ static int srpt_build_tskmgmt_rsp(struct srpt_rdma_ch *ch,
srp_rsp = ioctx->ioctx.buf;
BUG_ON(!srp_rsp);
- memset(srp_rsp, 0, sizeof *srp_rsp);
+ memset(srp_rsp, 0, sizeof(*srp_rsp));
srp_rsp->opcode = SRP_RSP;
srp_rsp->req_lim_delta =
@@ -1528,80 +1520,6 @@ static int srpt_build_tskmgmt_rsp(struct srpt_rdma_ch *ch,
return resp_len;
}
-#define NO_SUCH_LUN ((uint64_t)-1LL)
-
-/*
- * SCSI LUN addressing method. See also SAM-2 and the section about
- * eight byte LUNs.
- */
-enum scsi_lun_addr_method {
- SCSI_LUN_ADDR_METHOD_PERIPHERAL = 0,
- SCSI_LUN_ADDR_METHOD_FLAT = 1,
- SCSI_LUN_ADDR_METHOD_LUN = 2,
- SCSI_LUN_ADDR_METHOD_EXTENDED_LUN = 3,
-};
-
-/*
- * srpt_unpack_lun() - Convert from network LUN to linear LUN.
- *
- * Convert an 2-byte, 4-byte, 6-byte or 8-byte LUN structure in network byte
- * order (big endian) to a linear LUN. Supports three LUN addressing methods:
- * peripheral, flat and logical unit. See also SAM-2, section 4.9.4 (page 40).
- */
-static uint64_t srpt_unpack_lun(const uint8_t *lun, int len)
-{
- uint64_t res = NO_SUCH_LUN;
- int addressing_method;
-
- if (unlikely(len < 2)) {
- pr_err("Illegal LUN length %d, expected 2 bytes or more\n",
- len);
- goto out;
- }
-
- switch (len) {
- case 8:
- if ((*((__be64 *)lun) &
- cpu_to_be64(0x0000FFFFFFFFFFFFLL)) != 0)
- goto out_err;
- break;
- case 4:
- if (*((__be16 *)&lun[2]) != 0)
- goto out_err;
- break;
- case 6:
- if (*((__be32 *)&lun[2]) != 0)
- goto out_err;
- break;
- case 2:
- break;
- default:
- goto out_err;
- }
-
- addressing_method = (*lun) >> 6; /* highest two bits of byte 0 */
- switch (addressing_method) {
- case SCSI_LUN_ADDR_METHOD_PERIPHERAL:
- case SCSI_LUN_ADDR_METHOD_FLAT:
- case SCSI_LUN_ADDR_METHOD_LUN:
- res = *(lun + 1) | (((*lun) & 0x3f) << 8);
- break;
-
- case SCSI_LUN_ADDR_METHOD_EXTENDED_LUN:
- default:
- pr_err("Unimplemented LUN addressing method %u\n",
- addressing_method);
- break;
- }
-
-out:
- return res;
-
-out_err:
- pr_err("Support for multi-level LUNs has not yet been implemented\n");
- goto out;
-}
-
static int srpt_check_stop_free(struct se_cmd *cmd)
{
struct srpt_send_ioctx *ioctx = container_of(cmd,
@@ -1613,16 +1531,14 @@ static int srpt_check_stop_free(struct se_cmd *cmd)
/**
* srpt_handle_cmd() - Process SRP_CMD.
*/
-static int srpt_handle_cmd(struct srpt_rdma_ch *ch,
- struct srpt_recv_ioctx *recv_ioctx,
- struct srpt_send_ioctx *send_ioctx)
+static void srpt_handle_cmd(struct srpt_rdma_ch *ch,
+ struct srpt_recv_ioctx *recv_ioctx,
+ struct srpt_send_ioctx *send_ioctx)
{
struct se_cmd *cmd;
struct srp_cmd *srp_cmd;
- uint64_t unpacked_lun;
u64 data_len;
enum dma_data_direction dir;
- sense_reason_t ret;
int rc;
BUG_ON(!send_ioctx);
@@ -1650,24 +1566,23 @@ static int srpt_handle_cmd(struct srpt_rdma_ch *ch,
if (srpt_get_desc_tbl(send_ioctx, srp_cmd, &dir, &data_len)) {
pr_err("0x%llx: parsing SRP descriptor table failed.\n",
srp_cmd->tag);
- ret = TCM_INVALID_CDB_FIELD;
- goto send_sense;
+ goto release_ioctx;
}
- unpacked_lun = srpt_unpack_lun((uint8_t *)&srp_cmd->lun,
- sizeof(srp_cmd->lun));
rc = target_submit_cmd(cmd, ch->sess, srp_cmd->cdb,
- &send_ioctx->sense_data[0], unpacked_lun, data_len,
- TCM_SIMPLE_TAG, dir, TARGET_SCF_ACK_KREF);
+ &send_ioctx->sense_data[0],
+ scsilun_to_int(&srp_cmd->lun), data_len,
+ TCM_SIMPLE_TAG, dir, TARGET_SCF_ACK_KREF);
if (rc != 0) {
- ret = TCM_LOGICAL_UNIT_COMMUNICATION_FAILURE;
- goto send_sense;
+ pr_debug("target_submit_cmd() returned %d for tag %#llx\n", rc,
+ srp_cmd->tag);
+ goto release_ioctx;
}
- return 0;
+ return;
-send_sense:
- transport_send_check_condition_and_sense(cmd, ret, 0);
- return -1;
+release_ioctx:
+ send_ioctx->state = SRPT_STATE_DONE;
+ srpt_release_cmd(cmd);
}
static int srp_tmr_to_tcm(int fn)
@@ -1703,7 +1618,6 @@ static void srpt_handle_tsk_mgmt(struct srpt_rdma_ch *ch,
struct srp_tsk_mgmt *srp_tsk;
struct se_cmd *cmd;
struct se_session *sess = ch->sess;
- uint64_t unpacked_lun;
int tcm_tmr;
int rc;
@@ -1719,11 +1633,10 @@ static void srpt_handle_tsk_mgmt(struct srpt_rdma_ch *ch,
srpt_set_cmd_state(send_ioctx, SRPT_STATE_MGMT);
send_ioctx->cmd.tag = srp_tsk->tag;
tcm_tmr = srp_tmr_to_tcm(srp_tsk->tsk_mgmt_func);
- unpacked_lun = srpt_unpack_lun((uint8_t *)&srp_tsk->lun,
- sizeof(srp_tsk->lun));
- rc = target_submit_tmr(&send_ioctx->cmd, sess, NULL, unpacked_lun,
- srp_tsk, tcm_tmr, GFP_KERNEL, srp_tsk->task_tag,
- TARGET_SCF_ACK_KREF);
+ rc = target_submit_tmr(&send_ioctx->cmd, sess, NULL,
+ scsilun_to_int(&srp_tsk->lun), srp_tsk, tcm_tmr,
+ GFP_KERNEL, srp_tsk->task_tag,
+ TARGET_SCF_ACK_KREF);
if (rc != 0) {
send_ioctx->cmd.se_tmr_req->response = TMR_FUNCTION_REJECTED;
goto fail;
@@ -1743,7 +1656,6 @@ static void srpt_handle_new_iu(struct srpt_rdma_ch *ch,
struct srpt_send_ioctx *send_ioctx)
{
struct srp_cmd *srp_cmd;
- enum rdma_ch_state ch_state;
BUG_ON(!ch);
BUG_ON(!recv_ioctx);
@@ -1752,13 +1664,12 @@ static void srpt_handle_new_iu(struct srpt_rdma_ch *ch,
recv_ioctx->ioctx.dma, srp_max_req_size,
DMA_FROM_DEVICE);
- ch_state = srpt_get_ch_state(ch);
- if (unlikely(ch_state == CH_CONNECTING)) {
+ if (unlikely(ch->state == CH_CONNECTING)) {
list_add_tail(&recv_ioctx->wait_list, &ch->cmd_wait_list);
goto out;
}
- if (unlikely(ch_state != CH_LIVE))
+ if (unlikely(ch->state != CH_LIVE))
goto out;
srp_cmd = recv_ioctx->ioctx.buf;
@@ -1821,6 +1732,28 @@ static void srpt_recv_done(struct ib_cq *cq, struct ib_wc *wc)
}
}
+/*
+ * This function must be called from the context in which RDMA completions are
+ * processed because it accesses the wait list without protection against
+ * access from other threads.
+ */
+static void srpt_process_wait_list(struct srpt_rdma_ch *ch)
+{
+ struct srpt_send_ioctx *ioctx;
+
+ while (!list_empty(&ch->cmd_wait_list) &&
+ ch->state >= CH_LIVE &&
+ (ioctx = srpt_get_send_ioctx(ch)) != NULL) {
+ struct srpt_recv_ioctx *recv_ioctx;
+
+ recv_ioctx = list_first_entry(&ch->cmd_wait_list,
+ struct srpt_recv_ioctx,
+ wait_list);
+ list_del(&recv_ioctx->wait_list);
+ srpt_handle_new_iu(ch, recv_ioctx, ioctx);
+ }
+}
+
/**
* Note: Although this has not yet been observed during tests, at least in
* theory it is possible that the srpt_get_send_ioctx() call invoked by
@@ -1848,15 +1781,10 @@ static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc)
atomic_inc(&ch->sq_wr_avail);
- if (wc->status != IB_WC_SUCCESS) {
+ if (wc->status != IB_WC_SUCCESS)
pr_info("sending response for ioctx 0x%p failed"
" with status %d\n", ioctx, wc->status);
- atomic_dec(&ch->req_lim);
- srpt_abort_cmd(ioctx);
- goto out;
- }
-
if (state != SRPT_STATE_DONE) {
srpt_unmap_sg_to_ib_sge(ch, ioctx);
transport_generic_free_cmd(&ioctx->cmd, 0);
@@ -1865,18 +1793,7 @@ static void srpt_send_done(struct ib_cq *cq, struct ib_wc *wc)
" wr_id = %u.\n", ioctx->ioctx.index);
}
-out:
- while (!list_empty(&ch->cmd_wait_list) &&
- srpt_get_ch_state(ch) == CH_LIVE &&
- (ioctx = srpt_get_send_ioctx(ch)) != NULL) {
- struct srpt_recv_ioctx *recv_ioctx;
-
- recv_ioctx = list_first_entry(&ch->cmd_wait_list,
- struct srpt_recv_ioctx,
- wait_list);
- list_del(&recv_ioctx->wait_list);
- srpt_handle_new_iu(ch, recv_ioctx, ioctx);
- }
+ srpt_process_wait_list(ch);
}
/**
@@ -1893,7 +1810,7 @@ static int srpt_create_ch_ib(struct srpt_rdma_ch *ch)
WARN_ON(ch->rq_size < 1);
ret = -ENOMEM;
- qp_init = kzalloc(sizeof *qp_init, GFP_KERNEL);
+ qp_init = kzalloc(sizeof(*qp_init), GFP_KERNEL);
if (!qp_init)
goto out;
@@ -1960,168 +1877,102 @@ static void srpt_destroy_ch_ib(struct srpt_rdma_ch *ch)
}
/**
- * __srpt_close_ch() - Close an RDMA channel by setting the QP error state.
+ * srpt_close_ch() - Close an RDMA channel.
*
- * Reset the QP and make sure all resources associated with the channel will
- * be deallocated at an appropriate time.
+ * Make sure all resources associated with the channel will be deallocated at
+ * an appropriate time.
*
- * Note: The caller must hold ch->sport->sdev->spinlock.
+ * Returns true if and only if the channel state has been modified into
+ * CH_DRAINING.
*/
-static void __srpt_close_ch(struct srpt_rdma_ch *ch)
+static bool srpt_close_ch(struct srpt_rdma_ch *ch)
{
- enum rdma_ch_state prev_state;
- unsigned long flags;
-
- spin_lock_irqsave(&ch->spinlock, flags);
- prev_state = ch->state;
- switch (prev_state) {
- case CH_CONNECTING:
- case CH_LIVE:
- ch->state = CH_DISCONNECTING;
- break;
- default:
- break;
- }
- spin_unlock_irqrestore(&ch->spinlock, flags);
+ int ret;
- switch (prev_state) {
- case CH_CONNECTING:
- ib_send_cm_rej(ch->cm_id, IB_CM_REJ_NO_RESOURCES, NULL, 0,
- NULL, 0);
- /* fall through */
- case CH_LIVE:
- if (ib_send_cm_dreq(ch->cm_id, NULL, 0) < 0)
- pr_err("sending CM DREQ failed.\n");
- break;
- case CH_DISCONNECTING:
- break;
- case CH_DRAINING:
- case CH_RELEASING:
- break;
+ if (!srpt_set_ch_state(ch, CH_DRAINING)) {
+ pr_debug("%s-%d: already closed\n", ch->sess_name,
+ ch->qp->qp_num);
+ return false;
}
-}
-
-/**
- * srpt_close_ch() - Close an RDMA channel.
- */
-static void srpt_close_ch(struct srpt_rdma_ch *ch)
-{
- struct srpt_device *sdev;
- sdev = ch->sport->sdev;
- spin_lock_irq(&sdev->spinlock);
- __srpt_close_ch(ch);
- spin_unlock_irq(&sdev->spinlock);
-}
+ kref_get(&ch->kref);
-/**
- * srpt_shutdown_session() - Whether or not a session may be shut down.
- */
-static int srpt_shutdown_session(struct se_session *se_sess)
-{
- struct srpt_rdma_ch *ch = se_sess->fabric_sess_ptr;
- unsigned long flags;
+ ret = srpt_ch_qp_err(ch);
+ if (ret < 0)
+ pr_err("%s-%d: changing queue pair into error state failed: %d\n",
+ ch->sess_name, ch->qp->qp_num, ret);
- spin_lock_irqsave(&ch->spinlock, flags);
- if (ch->in_shutdown) {
- spin_unlock_irqrestore(&ch->spinlock, flags);
- return true;
+ pr_debug("%s-%d: queued zerolength write\n", ch->sess_name,
+ ch->qp->qp_num);
+ ret = srpt_zerolength_write(ch);
+ if (ret < 0) {
+ pr_err("%s-%d: queuing zero-length write failed: %d\n",
+ ch->sess_name, ch->qp->qp_num, ret);
+ if (srpt_set_ch_state(ch, CH_DISCONNECTED))
+ schedule_work(&ch->release_work);
+ else
+ WARN_ON_ONCE(true);
}
- ch->in_shutdown = true;
- target_sess_cmd_list_set_waiting(se_sess);
- spin_unlock_irqrestore(&ch->spinlock, flags);
+ kref_put(&ch->kref, srpt_free_ch);
return true;
}
-/**
- * srpt_drain_channel() - Drain a channel by resetting the IB queue pair.
- * @cm_id: Pointer to the CM ID of the channel to be drained.
- *
- * Note: Must be called from inside srpt_cm_handler to avoid a race between
- * accessing sdev->spinlock and the call to kfree(sdev) in srpt_remove_one()
- * (the caller of srpt_cm_handler holds the cm_id spinlock; srpt_remove_one()
- * waits until all target sessions for the associated IB device have been
- * unregistered and target session registration involves a call to
- * ib_destroy_cm_id(), which locks the cm_id spinlock and hence waits until
- * this function has finished).
+/*
+ * Change the channel state into CH_DISCONNECTING. If a channel has not yet
+ * reached the connected state, close it. If a channel is in the connected
+ * state, send a DREQ. If a DREQ has been received, send a DREP. Note: it is
+ * the responsibility of the caller to ensure that this function is not
+ * invoked concurrently with the code that accepts a connection. This means
+ * that this function must either be invoked from inside a CM callback
+ * function or that it must be invoked with the srpt_port.mutex held.
*/
-static void srpt_drain_channel(struct ib_cm_id *cm_id)
+static int srpt_disconnect_ch(struct srpt_rdma_ch *ch)
{
- struct srpt_device *sdev;
- struct srpt_rdma_ch *ch;
int ret;
- bool do_reset = false;
- WARN_ON_ONCE(irqs_disabled());
+ if (!srpt_set_ch_state(ch, CH_DISCONNECTING))
+ return -ENOTCONN;
- sdev = cm_id->context;
- BUG_ON(!sdev);
- spin_lock_irq(&sdev->spinlock);
- list_for_each_entry(ch, &sdev->rch_list, list) {
- if (ch->cm_id == cm_id) {
- do_reset = srpt_test_and_set_ch_state(ch,
- CH_CONNECTING, CH_DRAINING) ||
- srpt_test_and_set_ch_state(ch,
- CH_LIVE, CH_DRAINING) ||
- srpt_test_and_set_ch_state(ch,
- CH_DISCONNECTING, CH_DRAINING);
- break;
- }
- }
- spin_unlock_irq(&sdev->spinlock);
+ ret = ib_send_cm_dreq(ch->cm_id, NULL, 0);
+ if (ret < 0)
+ ret = ib_send_cm_drep(ch->cm_id, NULL, 0);
- if (do_reset) {
- if (ch->sess)
- srpt_shutdown_session(ch->sess);
+ if (ret < 0 && srpt_close_ch(ch))
+ ret = 0;
- ret = srpt_ch_qp_err(ch);
- if (ret < 0)
- pr_err("Setting queue pair in error state"
- " failed: %d\n", ret);
- }
+ return ret;
}
-/**
- * srpt_find_channel() - Look up an RDMA channel.
- * @cm_id: Pointer to the CM ID of the channel to be looked up.
- *
- * Return NULL if no matching RDMA channel has been found.
- */
-static struct srpt_rdma_ch *srpt_find_channel(struct srpt_device *sdev,
- struct ib_cm_id *cm_id)
+static void __srpt_close_all_ch(struct srpt_device *sdev)
{
struct srpt_rdma_ch *ch;
- bool found;
- WARN_ON_ONCE(irqs_disabled());
- BUG_ON(!sdev);
+ lockdep_assert_held(&sdev->mutex);
- found = false;
- spin_lock_irq(&sdev->spinlock);
list_for_each_entry(ch, &sdev->rch_list, list) {
- if (ch->cm_id == cm_id) {
- found = true;
- break;
- }
+ if (srpt_disconnect_ch(ch) >= 0)
+ pr_info("Closing channel %s-%d because target %s has been disabled\n",
+ ch->sess_name, ch->qp->qp_num,
+ sdev->device->name);
+ srpt_close_ch(ch);
}
- spin_unlock_irq(&sdev->spinlock);
-
- return found ? ch : NULL;
}
/**
- * srpt_release_channel() - Release channel resources.
- *
- * Schedules the actual release because:
- * - Calling the ib_destroy_cm_id() call from inside an IB CM callback would
- * trigger a deadlock.
- * - It is not safe to call TCM transport_* functions from interrupt context.
+ * srpt_shutdown_session() - Whether or not a session may be shut down.
*/
-static void srpt_release_channel(struct srpt_rdma_ch *ch)
+static int srpt_shutdown_session(struct se_session *se_sess)
{
- schedule_work(&ch->release_work);
+ return 1;
+}
+
+static void srpt_free_ch(struct kref *kref)
+{
+ struct srpt_rdma_ch *ch = container_of(kref, struct srpt_rdma_ch, kref);
+
+ kfree(ch);
}
static void srpt_release_channel_work(struct work_struct *w)
@@ -2131,8 +1982,8 @@ static void srpt_release_channel_work(struct work_struct *w)
struct se_session *se_sess;
ch = container_of(w, struct srpt_rdma_ch, release_work);
- pr_debug("ch = %p; ch->sess = %p; release_done = %p\n", ch, ch->sess,
- ch->release_done);
+ pr_debug("%s: %s-%d; release_done = %p\n", __func__, ch->sess_name,
+ ch->qp->qp_num, ch->release_done);
sdev = ch->sport->sdev;
BUG_ON(!sdev);
@@ -2140,6 +1991,7 @@ static void srpt_release_channel_work(struct work_struct *w)
se_sess = ch->sess;
BUG_ON(!se_sess);
+ target_sess_cmd_list_set_waiting(se_sess);
target_wait_for_sess_cmds(se_sess);
transport_deregister_session_configfs(se_sess);
@@ -2154,16 +2006,15 @@ static void srpt_release_channel_work(struct work_struct *w)
ch->sport->sdev, ch->rq_size,
ch->rsp_size, DMA_TO_DEVICE);
- spin_lock_irq(&sdev->spinlock);
- list_del(&ch->list);
- spin_unlock_irq(&sdev->spinlock);
-
+ mutex_lock(&sdev->mutex);
+ list_del_init(&ch->list);
if (ch->release_done)
complete(ch->release_done);
+ mutex_unlock(&sdev->mutex);
wake_up(&sdev->ch_releaseQ);
- kfree(ch);
+ kref_put(&ch->kref, srpt_free_ch);
}
/**
@@ -2183,7 +2034,6 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
struct srp_login_rej *rej;
struct ib_cm_rep_param *rep_param;
struct srpt_rdma_ch *ch, *tmp_ch;
- struct se_node_acl *se_acl;
u32 it_iu_len;
int i, ret = 0;
unsigned char *p;
@@ -2209,9 +2059,9 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
be64_to_cpu(*(__be64 *)&sdev->port[param->port - 1].gid.raw[0]),
be64_to_cpu(*(__be64 *)&sdev->port[param->port - 1].gid.raw[8]));
- rsp = kzalloc(sizeof *rsp, GFP_KERNEL);
- rej = kzalloc(sizeof *rej, GFP_KERNEL);
- rep_param = kzalloc(sizeof *rep_param, GFP_KERNEL);
+ rsp = kzalloc(sizeof(*rsp), GFP_KERNEL);
+ rej = kzalloc(sizeof(*rej), GFP_KERNEL);
+ rep_param = kzalloc(sizeof(*rep_param), GFP_KERNEL);
if (!rsp || !rej || !rep_param) {
ret = -ENOMEM;
@@ -2240,7 +2090,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
if ((req->req_flags & SRP_MTCH_ACTION) == SRP_MULTICHAN_SINGLE) {
rsp->rsp_flags = SRP_LOGIN_RSP_MULTICHAN_NO_CHAN;
- spin_lock_irq(&sdev->spinlock);
+ mutex_lock(&sdev->mutex);
list_for_each_entry_safe(ch, tmp_ch, &sdev->rch_list, list) {
if (!memcmp(ch->i_port_id, req->initiator_port_id, 16)
@@ -2248,26 +2098,16 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
&& param->port == ch->sport->port
&& param->listen_id == ch->sport->sdev->cm_id
&& ch->cm_id) {
- enum rdma_ch_state ch_state;
-
- ch_state = srpt_get_ch_state(ch);
- if (ch_state != CH_CONNECTING
- && ch_state != CH_LIVE)
+ if (srpt_disconnect_ch(ch) < 0)
continue;
-
- /* found an existing channel */
- pr_debug("Found existing channel %s"
- " cm_id= %p state= %d\n",
- ch->sess_name, ch->cm_id, ch_state);
-
- __srpt_close_ch(ch);
-
+ pr_info("Relogin - closed existing channel %s\n",
+ ch->sess_name);
rsp->rsp_flags =
SRP_LOGIN_RSP_MULTICHAN_TERMINATED;
}
}
- spin_unlock_irq(&sdev->spinlock);
+ mutex_unlock(&sdev->mutex);
} else
rsp->rsp_flags = SRP_LOGIN_RSP_MULTICHAN_MAINTAINED;
@@ -2283,7 +2123,7 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
goto reject;
}
- ch = kzalloc(sizeof *ch, GFP_KERNEL);
+ ch = kzalloc(sizeof(*ch), GFP_KERNEL);
if (!ch) {
rej->reason = cpu_to_be32(
SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
@@ -2292,11 +2132,14 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
goto reject;
}
+ kref_init(&ch->kref);
+ ch->zw_cqe.done = srpt_zerolength_write_done;
INIT_WORK(&ch->release_work, srpt_release_channel_work);
memcpy(ch->i_port_id, req->initiator_port_id, 16);
memcpy(ch->t_port_id, req->target_port_id, 16);
ch->sport = &sdev->port[param->port - 1];
ch->cm_id = cm_id;
+ cm_id->context = ch;
/*
* Avoid QUEUE_FULL conditions by limiting the number of buffers used
* for the SRP protocol to the command queue size.
@@ -2349,19 +2192,12 @@ static int srpt_cm_req_recv(struct ib_cm_id *cm_id,
pr_debug("registering session %s\n", ch->sess_name);
p = &ch->sess_name[0];
- ch->sess = transport_init_session(TARGET_PROT_NORMAL);
- if (IS_ERR(ch->sess)) {
- rej->reason = cpu_to_be32(
- SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES);
- pr_debug("Failed to create session\n");
- goto destroy_ib;
- }
-
try_again:
- se_acl = core_tpg_get_initiator_node_acl(&sport->port_tpg_1, p);
- if (!se_acl) {
+ ch->sess = target_alloc_session(&sport->port_tpg_1, 0, 0,
+ TARGET_PROT_NORMAL, p, ch, NULL);
+ if (IS_ERR(ch->sess)) {
pr_info("Rejected login because no ACL has been"
- " configured yet for initiator %s.\n", ch->sess_name);
+ " configured yet for initiator %s.\n", p);
/*
* XXX: Hack to retry of ch->i_port_id without leading '0x'
*/
@@ -2369,14 +2205,11 @@ try_again:
p += 2;
goto try_again;
}
- rej->reason = cpu_to_be32(
+ rej->reason = cpu_to_be32((PTR_ERR(ch->sess) == -ENOMEM) ?
+ SRP_LOGIN_REJ_INSUFFICIENT_RESOURCES :
SRP_LOGIN_REJ_CHANNEL_LIMIT_REACHED);
- transport_free_session(ch->sess);
goto destroy_ib;
}
- ch->sess->se_node_acl = se_acl;
-
- transport_register_session(&sport->port_tpg_1, se_acl, ch->sess, ch);
pr_debug("Establish connection sess=%p name=%s cm_id=%p\n", ch->sess,
ch->sess_name, ch->cm_id);
@@ -2396,7 +2229,7 @@ try_again:
/* create cm reply */
rep_param->qp_num = ch->qp->qp_num;
rep_param->private_data = (void *)rsp;
- rep_param->private_data_len = sizeof *rsp;
+ rep_param->private_data_len = sizeof(*rsp);
rep_param->rnr_retry_count = 7;
rep_param->flow_control = 1;
rep_param->failover_accepted = 0;
@@ -2411,14 +2244,14 @@ try_again:
goto release_channel;
}
- spin_lock_irq(&sdev->spinlock);
+ mutex_lock(&sdev->mutex);
list_add_tail(&ch->list, &sdev->rch_list);
- spin_unlock_irq(&sdev->spinlock);
+ mutex_unlock(&sdev->mutex);
goto out;
release_channel:
- srpt_set_ch_state(ch, CH_RELEASING);
+ srpt_disconnect_ch(ch);
transport_deregister_session_configfs(ch->sess);
transport_deregister_session(ch->sess);
ch->sess = NULL;
@@ -2440,7 +2273,7 @@ reject:
| SRP_BUF_FORMAT_INDIRECT);
ib_send_cm_rej(cm_id, IB_CM_REJ_CONSUMER_DEFINED, NULL, 0,
- (void *)rej, sizeof *rej);
+ (void *)rej, sizeof(*rej));
out:
kfree(rep_param);
@@ -2450,10 +2283,23 @@ out:
return ret;
}
-static void srpt_cm_rej_recv(struct ib_cm_id *cm_id)
+static void srpt_cm_rej_recv(struct srpt_rdma_ch *ch,
+ enum ib_cm_rej_reason reason,
+ const u8 *private_data,
+ u8 private_data_len)
{
- pr_info("Received IB REJ for cm_id %p.\n", cm_id);
- srpt_drain_channel(cm_id);
+ char *priv = NULL;
+ int i;
+
+ if (private_data_len && (priv = kmalloc(private_data_len * 3 + 1,
+ GFP_KERNEL))) {
+ for (i = 0; i < private_data_len; i++)
+ sprintf(priv + 3 * i, " %02x", private_data[i]);
+ }
+ pr_info("Received CM REJ for ch %s-%d; reason %d%s%s.\n",
+ ch->sess_name, ch->qp->qp_num, reason, private_data_len ?
+ "; private data" : "", priv ? priv : " (?)");
+ kfree(priv);
}
/**
@@ -2462,87 +2308,23 @@ static void srpt_cm_rej_recv(struct ib_cm_id *cm_id)
* An IB_CM_RTU_RECEIVED message indicates that the connection is established
* and that the recipient may begin transmitting (RTU = ready to use).
*/
-static void srpt_cm_rtu_recv(struct ib_cm_id *cm_id)
+static void srpt_cm_rtu_recv(struct srpt_rdma_ch *ch)
{
- struct srpt_rdma_ch *ch;
int ret;
- ch = srpt_find_channel(cm_id->context, cm_id);
- BUG_ON(!ch);
-
- if (srpt_test_and_set_ch_state(ch, CH_CONNECTING, CH_LIVE)) {
- struct srpt_recv_ioctx *ioctx, *ioctx_tmp;
-
+ if (srpt_set_ch_state(ch, CH_LIVE)) {
ret = srpt_ch_qp_rts(ch, ch->qp);
- list_for_each_entry_safe(ioctx, ioctx_tmp, &ch->cmd_wait_list,
- wait_list) {
- list_del(&ioctx->wait_list);
- srpt_handle_new_iu(ch, ioctx, NULL);
- }
- if (ret)
+ if (ret == 0) {
+ /* Trigger wait list processing. */
+ ret = srpt_zerolength_write(ch);
+ WARN_ONCE(ret < 0, "%d\n", ret);
+ } else {
srpt_close_ch(ch);
+ }
}
}
-static void srpt_cm_timewait_exit(struct ib_cm_id *cm_id)
-{
- pr_info("Received IB TimeWait exit for cm_id %p.\n", cm_id);
- srpt_drain_channel(cm_id);
-}
-
-static void srpt_cm_rep_error(struct ib_cm_id *cm_id)
-{
- pr_info("Received IB REP error for cm_id %p.\n", cm_id);
- srpt_drain_channel(cm_id);
-}
-
-/**
- * srpt_cm_dreq_recv() - Process reception of a DREQ message.
- */
-static void srpt_cm_dreq_recv(struct ib_cm_id *cm_id)
-{
- struct srpt_rdma_ch *ch;
- unsigned long flags;
- bool send_drep = false;
-
- ch = srpt_find_channel(cm_id->context, cm_id);
- BUG_ON(!ch);
-
- pr_debug("cm_id= %p ch->state= %d\n", cm_id, srpt_get_ch_state(ch));
-
- spin_lock_irqsave(&ch->spinlock, flags);
- switch (ch->state) {
- case CH_CONNECTING:
- case CH_LIVE:
- send_drep = true;
- ch->state = CH_DISCONNECTING;
- break;
- case CH_DISCONNECTING:
- case CH_DRAINING:
- case CH_RELEASING:
- WARN(true, "unexpected channel state %d\n", ch->state);
- break;
- }
- spin_unlock_irqrestore(&ch->spinlock, flags);
-
- if (send_drep) {
- if (ib_send_cm_drep(ch->cm_id, NULL, 0) < 0)
- pr_err("Sending IB DREP failed.\n");
- pr_info("Received DREQ and sent DREP for session %s.\n",
- ch->sess_name);
- }
-}
-
-/**
- * srpt_cm_drep_recv() - Process reception of a DREP message.
- */
-static void srpt_cm_drep_recv(struct ib_cm_id *cm_id)
-{
- pr_info("Received InfiniBand DREP message for cm_id %p.\n", cm_id);
- srpt_drain_channel(cm_id);
-}
-
/**
* srpt_cm_handler() - IB connection manager callback function.
*
@@ -2555,6 +2337,7 @@ static void srpt_cm_drep_recv(struct ib_cm_id *cm_id)
*/
static int srpt_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
{
+ struct srpt_rdma_ch *ch = cm_id->context;
int ret;
ret = 0;
@@ -2564,32 +2347,39 @@ static int srpt_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
event->private_data);
break;
case IB_CM_REJ_RECEIVED:
- srpt_cm_rej_recv(cm_id);
+ srpt_cm_rej_recv(ch, event->param.rej_rcvd.reason,
+ event->private_data,
+ IB_CM_REJ_PRIVATE_DATA_SIZE);
break;
case IB_CM_RTU_RECEIVED:
case IB_CM_USER_ESTABLISHED:
- srpt_cm_rtu_recv(cm_id);
+ srpt_cm_rtu_recv(ch);
break;
case IB_CM_DREQ_RECEIVED:
- srpt_cm_dreq_recv(cm_id);
+ srpt_disconnect_ch(ch);
break;
case IB_CM_DREP_RECEIVED:
- srpt_cm_drep_recv(cm_id);
+ pr_info("Received CM DREP message for ch %s-%d.\n",
+ ch->sess_name, ch->qp->qp_num);
+ srpt_close_ch(ch);
break;
case IB_CM_TIMEWAIT_EXIT:
- srpt_cm_timewait_exit(cm_id);
+ pr_info("Received CM TimeWait exit for ch %s-%d.\n",
+ ch->sess_name, ch->qp->qp_num);
+ srpt_close_ch(ch);
break;
case IB_CM_REP_ERROR:
- srpt_cm_rep_error(cm_id);
+ pr_info("Received CM REP error for ch %s-%d.\n", ch->sess_name,
+ ch->qp->qp_num);
break;
case IB_CM_DREQ_ERROR:
- pr_info("Received IB DREQ ERROR event.\n");
+ pr_info("Received CM DREQ ERROR event.\n");
break;
case IB_CM_MRA_RECEIVED:
- pr_info("Received IB MRA event\n");
+ pr_info("Received CM MRA event\n");
break;
default:
- pr_err("received unrecognized IB CM event %d\n", event->event);
+ pr_err("received unrecognized CM event %d\n", event->event);
break;
}
@@ -2698,41 +2488,14 @@ static int srpt_write_pending_status(struct se_cmd *se_cmd)
*/
static int srpt_write_pending(struct se_cmd *se_cmd)
{
- struct srpt_rdma_ch *ch;
- struct srpt_send_ioctx *ioctx;
+ struct srpt_send_ioctx *ioctx =
+ container_of(se_cmd, struct srpt_send_ioctx, cmd);
+ struct srpt_rdma_ch *ch = ioctx->ch;
enum srpt_command_state new_state;
- enum rdma_ch_state ch_state;
- int ret;
-
- ioctx = container_of(se_cmd, struct srpt_send_ioctx, cmd);
new_state = srpt_set_cmd_state(ioctx, SRPT_STATE_NEED_DATA);
WARN_ON(new_state == SRPT_STATE_DONE);
-
- ch = ioctx->ch;
- BUG_ON(!ch);
-
- ch_state = srpt_get_ch_state(ch);
- switch (ch_state) {
- case CH_CONNECTING:
- WARN(true, "unexpected channel state %d\n", ch_state);
- ret = -EINVAL;
- goto out;
- case CH_LIVE:
- break;
- case CH_DISCONNECTING:
- case CH_DRAINING:
- case CH_RELEASING:
- pr_debug("cmd with tag %lld: channel disconnecting\n",
- ioctx->cmd.tag);
- srpt_set_cmd_state(ioctx, SRPT_STATE_DATA_IN);
- ret = -EINVAL;
- goto out;
- }
- ret = srpt_xfer_data(ch, ioctx);
-
-out:
- return ret;
+ return srpt_xfer_data(ch, ioctx);
}
static u8 tcm_to_srp_tsk_mgmt_status(const int tcm_mgmt_status)
@@ -2863,36 +2626,25 @@ static void srpt_refresh_port_work(struct work_struct *work)
srpt_refresh_port(sport);
}
-static int srpt_ch_list_empty(struct srpt_device *sdev)
-{
- int res;
-
- spin_lock_irq(&sdev->spinlock);
- res = list_empty(&sdev->rch_list);
- spin_unlock_irq(&sdev->spinlock);
-
- return res;
-}
-
/**
* srpt_release_sdev() - Free the channel resources associated with a target.
*/
static int srpt_release_sdev(struct srpt_device *sdev)
{
- struct srpt_rdma_ch *ch, *tmp_ch;
- int res;
+ int i, res;
WARN_ON_ONCE(irqs_disabled());
BUG_ON(!sdev);
- spin_lock_irq(&sdev->spinlock);
- list_for_each_entry_safe(ch, tmp_ch, &sdev->rch_list, list)
- __srpt_close_ch(ch);
- spin_unlock_irq(&sdev->spinlock);
+ mutex_lock(&sdev->mutex);
+ for (i = 0; i < ARRAY_SIZE(sdev->port); i++)
+ sdev->port[i].enabled = false;
+ __srpt_close_all_ch(sdev);
+ mutex_unlock(&sdev->mutex);
res = wait_event_interruptible(sdev->ch_releaseQ,
- srpt_ch_list_empty(sdev));
+ list_empty_careful(&sdev->rch_list));
if (res)
pr_err("%s: interrupted.\n", __func__);
@@ -2946,14 +2698,14 @@ static void srpt_add_one(struct ib_device *device)
pr_debug("device = %p, device->dma_ops = %p\n", device,
device->dma_ops);
- sdev = kzalloc(sizeof *sdev, GFP_KERNEL);
+ sdev = kzalloc(sizeof(*sdev), GFP_KERNEL);
if (!sdev)
goto err;
sdev->device = device;
INIT_LIST_HEAD(&sdev->rch_list);
init_waitqueue_head(&sdev->ch_releaseQ);
- spin_lock_init(&sdev->spinlock);
+ mutex_init(&sdev->mutex);
sdev->pd = ib_alloc_pd(device);
if (IS_ERR(sdev->pd))
@@ -3025,7 +2777,7 @@ static void srpt_add_one(struct ib_device *device)
if (srpt_refresh_port(sport)) {
pr_err("MAD registration failed for %s-%d.\n",
- srpt_sdev_name(sdev), i);
+ sdev->device->name, i);
goto err_ring;
}
snprintf(sport->port_guid, sizeof(sport->port_guid),
@@ -3174,24 +2926,26 @@ static void srpt_release_cmd(struct se_cmd *se_cmd)
static void srpt_close_session(struct se_session *se_sess)
{
DECLARE_COMPLETION_ONSTACK(release_done);
- struct srpt_rdma_ch *ch;
- struct srpt_device *sdev;
- unsigned long res;
-
- ch = se_sess->fabric_sess_ptr;
- WARN_ON(ch->sess != se_sess);
+ struct srpt_rdma_ch *ch = se_sess->fabric_sess_ptr;
+ struct srpt_device *sdev = ch->sport->sdev;
+ bool wait;
- pr_debug("ch %p state %d\n", ch, srpt_get_ch_state(ch));
+ pr_debug("ch %s-%d state %d\n", ch->sess_name, ch->qp->qp_num,
+ ch->state);
- sdev = ch->sport->sdev;
- spin_lock_irq(&sdev->spinlock);
+ mutex_lock(&sdev->mutex);
BUG_ON(ch->release_done);
ch->release_done = &release_done;
- __srpt_close_ch(ch);
- spin_unlock_irq(&sdev->spinlock);
+ wait = !list_empty(&ch->list);
+ srpt_disconnect_ch(ch);
+ mutex_unlock(&sdev->mutex);
- res = wait_for_completion_timeout(&release_done, 60 * HZ);
- WARN_ON(res == 0);
+ if (!wait)
+ return;
+
+ while (wait_for_completion_timeout(&release_done, 180 * HZ) == 0)
+ pr_info("%s(%s-%d state %d): still waiting ...\n", __func__,
+ ch->sess_name, ch->qp->qp_num, ch->state);
}
/**
@@ -3399,6 +3153,8 @@ static ssize_t srpt_tpg_enable_store(struct config_item *item,
{
struct se_portal_group *se_tpg = to_tpg(item);
struct srpt_port *sport = container_of(se_tpg, struct srpt_port, port_tpg_1);
+ struct srpt_device *sdev = sport->sdev;
+ struct srpt_rdma_ch *ch;
unsigned long tmp;
int ret;
@@ -3412,11 +3168,24 @@ static ssize_t srpt_tpg_enable_store(struct config_item *item,
pr_err("Illegal value for srpt_tpg_store_enable: %lu\n", tmp);
return -EINVAL;
}
- if (tmp == 1)
- sport->enabled = true;
- else
- sport->enabled = false;
+ if (sport->enabled == tmp)
+ goto out;
+ sport->enabled = tmp;
+ if (sport->enabled)
+ goto out;
+
+ mutex_lock(&sdev->mutex);
+ list_for_each_entry(ch, &sdev->rch_list, list) {
+ if (ch->sport == sport) {
+ pr_debug("%s: ch %p %s-%d\n", __func__, ch,
+ ch->sess_name, ch->qp->qp_num);
+ srpt_disconnect_ch(ch);
+ srpt_close_ch(ch);
+ }
+ }
+ mutex_unlock(&sdev->mutex);
+out:
return count;
}
@@ -3508,7 +3277,6 @@ static struct configfs_attribute *srpt_wwn_attrs[] = {
static const struct target_core_fabric_ops srpt_template = {
.module = THIS_MODULE,
.name = "srpt",
- .node_acl_size = sizeof(struct srpt_node_acl),
.get_fabric_name = srpt_get_fabric_name,
.tpg_get_wwn = srpt_get_fabric_wwn,
.tpg_get_tag = srpt_get_tag,
diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.h b/drivers/infiniband/ulp/srpt/ib_srpt.h
index 09037f2b0..af9b8b527 100644
--- a/drivers/infiniband/ulp/srpt/ib_srpt.h
+++ b/drivers/infiniband/ulp/srpt/ib_srpt.h
@@ -218,20 +218,20 @@ struct srpt_send_ioctx {
/**
* enum rdma_ch_state - SRP channel state.
- * @CH_CONNECTING: QP is in RTR state; waiting for RTU.
- * @CH_LIVE: QP is in RTS state.
- * @CH_DISCONNECTING: DREQ has been received; waiting for DREP
- * or DREQ has been send and waiting for DREP
- * or .
- * @CH_DRAINING: QP is in ERR state; waiting for last WQE event.
- * @CH_RELEASING: Last WQE event has been received; releasing resources.
+ * @CH_CONNECTING: QP is in RTR state; waiting for RTU.
+ * @CH_LIVE: QP is in RTS state.
+ * @CH_DISCONNECTING: DREQ has been sent and waiting for DREP or DREQ has
+ * been received.
+ * @CH_DRAINING: DREP has been received or waiting for DREP timed out
+ * and last work request has been queued.
+ * @CH_DISCONNECTED: Last completion has been received.
*/
enum rdma_ch_state {
CH_CONNECTING,
CH_LIVE,
CH_DISCONNECTING,
CH_DRAINING,
- CH_RELEASING
+ CH_DISCONNECTED,
};
/**
@@ -267,6 +267,8 @@ struct srpt_rdma_ch {
struct ib_cm_id *cm_id;
struct ib_qp *qp;
struct ib_cq *cq;
+ struct ib_cqe zw_cqe;
+ struct kref kref;
int rq_size;
u32 rsp_size;
atomic_t sq_wr_avail;
@@ -286,7 +288,6 @@ struct srpt_rdma_ch {
u8 sess_name[36];
struct work_struct release_work;
struct completion *release_done;
- bool in_shutdown;
};
/**
@@ -343,7 +344,7 @@ struct srpt_port {
* @ioctx_ring: Per-HCA SRQ.
* @rch_list: Per-device channel list -- see also srpt_rdma_ch.list.
* @ch_releaseQ: Enables waiting for removal from rch_list.
- * @spinlock: Protects rch_list and tpg.
+ * @mutex: Protects rch_list.
* @port: Information about the ports owned by this HCA.
* @event_handler: Per-HCA asynchronous IB event handler.
* @list: Node in srpt_dev_list.
@@ -357,18 +358,10 @@ struct srpt_device {
struct srpt_recv_ioctx **ioctx_ring;
struct list_head rch_list;
wait_queue_head_t ch_releaseQ;
- spinlock_t spinlock;
+ struct mutex mutex;
struct srpt_port port[2];
struct ib_event_handler event_handler;
struct list_head list;
};
-/**
- * struct srpt_node_acl - Per-initiator ACL data (managed via configfs).
- * @nacl: Target core node ACL information.
- */
-struct srpt_node_acl {
- struct se_node_acl nacl;
-};
-
#endif /* IB_SRPT_H */