Merge branches 'cxgb3', 'endian', 'ipath', 'ipoib', 'iser', 'mad', 'misc', 'mlx4', 'mthca', 'nes' and 'sysfs' into for-next
diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 7913b80..d1fba41 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -193,7 +193,7 @@
BUG_ON(device->reg_state != IB_DEV_UNREGISTERED);
- ib_device_unregister_sysfs(device);
+ kobject_put(&device->dev.kobj);
}
EXPORT_SYMBOL(ib_dealloc_device);
@@ -348,6 +348,8 @@
mutex_unlock(&device_mutex);
+ ib_device_unregister_sysfs(device);
+
spin_lock_irqsave(&device->client_data_lock, flags);
list_for_each_entry_safe(context, tmp, &device->client_data_list, list)
kfree(context);
diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c
index 5c54fc2..de922a0 100644
--- a/drivers/infiniband/core/mad.c
+++ b/drivers/infiniband/core/mad.c
@@ -301,6 +301,16 @@
mad_agent_priv->agent.context = context;
mad_agent_priv->agent.qp = port_priv->qp_info[qpn].qp;
mad_agent_priv->agent.port_num = port_num;
+ spin_lock_init(&mad_agent_priv->lock);
+ INIT_LIST_HEAD(&mad_agent_priv->send_list);
+ INIT_LIST_HEAD(&mad_agent_priv->wait_list);
+ INIT_LIST_HEAD(&mad_agent_priv->done_list);
+ INIT_LIST_HEAD(&mad_agent_priv->rmpp_list);
+ INIT_DELAYED_WORK(&mad_agent_priv->timed_work, timeout_sends);
+ INIT_LIST_HEAD(&mad_agent_priv->local_list);
+ INIT_WORK(&mad_agent_priv->local_work, local_completions);
+ atomic_set(&mad_agent_priv->refcount, 1);
+ init_completion(&mad_agent_priv->comp);
spin_lock_irqsave(&port_priv->reg_lock, flags);
mad_agent_priv->agent.hi_tid = ++ib_mad_client_id;
@@ -350,17 +360,6 @@
list_add_tail(&mad_agent_priv->agent_list, &port_priv->agent_list);
spin_unlock_irqrestore(&port_priv->reg_lock, flags);
- spin_lock_init(&mad_agent_priv->lock);
- INIT_LIST_HEAD(&mad_agent_priv->send_list);
- INIT_LIST_HEAD(&mad_agent_priv->wait_list);
- INIT_LIST_HEAD(&mad_agent_priv->done_list);
- INIT_LIST_HEAD(&mad_agent_priv->rmpp_list);
- INIT_DELAYED_WORK(&mad_agent_priv->timed_work, timeout_sends);
- INIT_LIST_HEAD(&mad_agent_priv->local_list);
- INIT_WORK(&mad_agent_priv->local_work, local_completions);
- atomic_set(&mad_agent_priv->refcount, 1);
- init_completion(&mad_agent_priv->comp);
-
return &mad_agent_priv->agent;
error4:
@@ -743,9 +742,7 @@
break;
case IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED:
kmem_cache_free(ib_mad_cache, mad_priv);
- kfree(local);
- ret = 1;
- goto out;
+ break;
case IB_MAD_RESULT_SUCCESS:
/* Treat like an incoming receive MAD */
port_priv = ib_get_mad_port(mad_agent_priv->agent.device,
@@ -756,10 +753,12 @@
&mad_priv->mad.mad);
}
if (!port_priv || !recv_mad_agent) {
+ /*
+ * No receiving agent so drop packet and
+ * generate send completion.
+ */
kmem_cache_free(ib_mad_cache, mad_priv);
- kfree(local);
- ret = 0;
- goto out;
+ break;
}
local->mad_priv = mad_priv;
local->recv_mad_agent = recv_mad_agent;
@@ -2356,7 +2355,7 @@
struct ib_mad_local_private *local;
struct ib_mad_agent_private *recv_mad_agent;
unsigned long flags;
- int recv = 0;
+ int free_mad;
struct ib_wc wc;
struct ib_mad_send_wc mad_send_wc;
@@ -2370,14 +2369,15 @@
completion_list);
list_del(&local->completion_list);
spin_unlock_irqrestore(&mad_agent_priv->lock, flags);
+ free_mad = 0;
if (local->mad_priv) {
recv_mad_agent = local->recv_mad_agent;
if (!recv_mad_agent) {
printk(KERN_ERR PFX "No receive MAD agent for local completion\n");
+ free_mad = 1;
goto local_send_completion;
}
- recv = 1;
/*
* Defined behavior is to complete response
* before request
@@ -2422,7 +2422,7 @@
spin_lock_irqsave(&mad_agent_priv->lock, flags);
atomic_dec(&mad_agent_priv->refcount);
- if (!recv)
+ if (free_mad)
kmem_cache_free(ib_mad_cache, local->mad_priv);
kfree(local);
}
diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c
index 7863a50..1865049 100644
--- a/drivers/infiniband/core/sa_query.c
+++ b/drivers/infiniband/core/sa_query.c
@@ -395,6 +395,8 @@
}
spin_lock_irq(&port->ah_lock);
+ if (port->sm_ah)
+ kref_put(&port->sm_ah->ref, free_sm_ah);
port->sm_ah = new_ah;
spin_unlock_irq(&port->ah_lock);
diff --git a/drivers/infiniband/core/sysfs.c b/drivers/infiniband/core/sysfs.c
index b43f7d368..5c04cfb 100644
--- a/drivers/infiniband/core/sysfs.c
+++ b/drivers/infiniband/core/sysfs.c
@@ -66,11 +66,6 @@
int index;
};
-static inline int ibdev_is_alive(const struct ib_device *dev)
-{
- return dev->reg_state == IB_DEV_REGISTERED;
-}
-
static ssize_t port_attr_show(struct kobject *kobj,
struct attribute *attr, char *buf)
{
@@ -80,8 +75,6 @@
if (!port_attr->show)
return -EIO;
- if (!ibdev_is_alive(p->ibdev))
- return -ENODEV;
return port_attr->show(p, port_attr, buf);
}
@@ -562,9 +555,6 @@
{
struct ib_device *dev = container_of(device, struct ib_device, dev);
- if (!ibdev_is_alive(dev))
- return -ENODEV;
-
switch (dev->node_type) {
case RDMA_NODE_IB_CA: return sprintf(buf, "%d: CA\n", dev->node_type);
case RDMA_NODE_RNIC: return sprintf(buf, "%d: RNIC\n", dev->node_type);
@@ -581,9 +571,6 @@
struct ib_device_attr attr;
ssize_t ret;
- if (!ibdev_is_alive(dev))
- return -ENODEV;
-
ret = ib_query_device(dev, &attr);
if (ret)
return ret;
@@ -600,9 +587,6 @@
{
struct ib_device *dev = container_of(device, struct ib_device, dev);
- if (!ibdev_is_alive(dev))
- return -ENODEV;
-
return sprintf(buf, "%04x:%04x:%04x:%04x\n",
be16_to_cpu(((__be16 *) &dev->node_guid)[0]),
be16_to_cpu(((__be16 *) &dev->node_guid)[1]),
@@ -848,6 +832,9 @@
struct kobject *p, *t;
struct ib_port *port;
+ /* Hold kobject until ib_dealloc_device() */
+ kobject_get(&device->dev.kobj);
+
list_for_each_entry_safe(p, t, &device->port_list, entry) {
list_del(&p->entry);
port = container_of(p, struct ib_port, kobj);
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.c b/drivers/infiniband/hw/cxgb3/cxio_hal.c
index 4dcf08b..d4d7204 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.c
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.c
@@ -450,7 +450,7 @@
if ((CQE_OPCODE(*cqe) == T3_READ_RESP) && SQ_TYPE(*cqe))
return 0;
- if ((CQE_OPCODE(*cqe) == T3_SEND) && RQ_TYPE(*cqe) &&
+ if (CQE_SEND_OPCODE(*cqe) && RQ_TYPE(*cqe) &&
Q_EMPTY(wq->rq_rptr, wq->rq_wptr))
return 0;
@@ -938,6 +938,23 @@
if (!rdev_p->t3cdev_p)
rdev_p->t3cdev_p = dev2t3cdev(netdev_p);
rdev_p->t3cdev_p->ulp = (void *) rdev_p;
+
+ err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, GET_EMBEDDED_INFO,
+ &(rdev_p->fw_info));
+ if (err) {
+ printk(KERN_ERR "%s t3cdev_p(%p)->ctl returned error %d.\n",
+ __func__, rdev_p->t3cdev_p, err);
+ goto err1;
+ }
+ if (G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers) != CXIO_FW_MAJ) {
+ printk(KERN_ERR MOD "fatal firmware version mismatch: "
+ "need version %u but adapter has version %u\n",
+ CXIO_FW_MAJ,
+ G_FW_VERSION_MAJOR(rdev_p->fw_info.fw_vers));
+ err = -EINVAL;
+ goto err1;
+ }
+
err = rdev_p->t3cdev_p->ctl(rdev_p->t3cdev_p, RDMA_GET_PARAMS,
&(rdev_p->rnic_info));
if (err) {
@@ -1204,11 +1221,12 @@
}
/* incoming SEND with no receive posted failures */
- if ((CQE_OPCODE(*hw_cqe) == T3_SEND) && RQ_TYPE(*hw_cqe) &&
+ if (CQE_SEND_OPCODE(*hw_cqe) && RQ_TYPE(*hw_cqe) &&
Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
ret = -1;
goto skip_cqe;
}
+ BUG_ON((*cqe_flushed == 0) && !SW_CQE(*hw_cqe));
goto proc_cqe;
}
@@ -1223,6 +1241,13 @@
* then we complete this with TPT_ERR_MSN and mark the wq in
* error.
*/
+
+ if (Q_EMPTY(wq->rq_rptr, wq->rq_wptr)) {
+ wq->error = 1;
+ ret = -1;
+ goto skip_cqe;
+ }
+
if (unlikely((CQE_WRID_MSN(*hw_cqe) != (wq->rq_rptr + 1)))) {
wq->error = 1;
hw_cqe->header |= htonl(V_CQE_STATUS(TPT_ERR_MSN));
@@ -1277,6 +1302,7 @@
cxio_hal_pblpool_free(wq->rdev,
wq->rq[Q_PTR2IDX(wq->rq_rptr,
wq->rq_size_log2)].pbl_addr, T3_STAG0_PBL_SIZE);
+ BUG_ON(Q_EMPTY(wq->rq_rptr, wq->rq_wptr));
wq->rq_rptr++;
}
diff --git a/drivers/infiniband/hw/cxgb3/cxio_hal.h b/drivers/infiniband/hw/cxgb3/cxio_hal.h
index 656fe47..e44dc22 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_hal.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_hal.h
@@ -61,6 +61,8 @@
#define T3_MAX_DEV_NAME_LEN 32
+#define CXIO_FW_MAJ 7
+
struct cxio_hal_ctrl_qp {
u32 wptr;
u32 rptr;
@@ -108,6 +110,7 @@
struct gen_pool *pbl_pool;
struct gen_pool *rqt_pool;
struct list_head entry;
+ struct ch_embedded_info fw_info;
};
static inline int cxio_num_stags(struct cxio_rdev *rdev_p)
diff --git a/drivers/infiniband/hw/cxgb3/cxio_wr.h b/drivers/infiniband/hw/cxgb3/cxio_wr.h
index 04618f7..ff9be1a 100644
--- a/drivers/infiniband/hw/cxgb3/cxio_wr.h
+++ b/drivers/infiniband/hw/cxgb3/cxio_wr.h
@@ -604,6 +604,12 @@
#define CQE_STATUS(x) (G_CQE_STATUS(be32_to_cpu((x).header)))
#define CQE_OPCODE(x) (G_CQE_OPCODE(be32_to_cpu((x).header)))
+#define CQE_SEND_OPCODE(x)( \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_INV) || \
+ (G_CQE_OPCODE(be32_to_cpu((x).header)) == T3_SEND_WITH_SE_INV))
+
#define CQE_LEN(x) (be32_to_cpu((x).len))
/* used for RQ completion processing */
diff --git a/drivers/infiniband/hw/cxgb3/iwch_cm.c b/drivers/infiniband/hw/cxgb3/iwch_cm.c
index 44e936e..8699947 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_cm.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_cm.c
@@ -1678,6 +1678,9 @@
{
struct iwch_ep *ep = ctx;
+ if (state_read(&ep->com) != FPDU_MODE)
+ return CPL_RET_BUF_DONE;
+
PDBG("%s ep %p\n", __func__, ep);
skb_pull(skb, sizeof(struct cpl_rdma_terminate));
PDBG("%s saving %d bytes of term msg\n", __func__, skb->len);
diff --git a/drivers/infiniband/hw/cxgb3/iwch_ev.c b/drivers/infiniband/hw/cxgb3/iwch_ev.c
index 7b67a67..743c5d8 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_ev.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_ev.c
@@ -179,11 +179,6 @@
case TPT_ERR_BOUND:
case TPT_ERR_INVALIDATE_SHARED_MR:
case TPT_ERR_INVALIDATE_MR_WITH_MW_BOUND:
- printk(KERN_ERR "%s - CQE Err qpid 0x%x opcode %d status 0x%x "
- "type %d wrid.hi 0x%x wrid.lo 0x%x \n", __func__,
- CQE_QPID(rsp_msg->cqe), CQE_OPCODE(rsp_msg->cqe),
- CQE_STATUS(rsp_msg->cqe), CQE_TYPE(rsp_msg->cqe),
- CQE_WRID_HI(rsp_msg->cqe), CQE_WRID_LOW(rsp_msg->cqe));
(*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context);
post_qp_event(rnicp, chp, rsp_msg, IB_EVENT_QP_ACCESS_ERR, 1);
break;
diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index 48e2b0b..c758fbd 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -195,15 +195,12 @@
return 0;
}
-/*
- * TBD: this is going to be moved to firmware. Missing pdid/qpid check for now.
- */
static int iwch_sgl2pbl_map(struct iwch_dev *rhp, struct ib_sge *sg_list,
u32 num_sgle, u32 * pbl_addr, u8 * page_size)
{
int i;
struct iwch_mr *mhp;
- u32 offset;
+ u64 offset;
for (i = 0; i < num_sgle; i++) {
mhp = get_mhp(rhp, (sg_list[i].lkey) >> 8);
@@ -235,8 +232,8 @@
return -EINVAL;
}
offset = sg_list[i].addr - mhp->attr.va_fbo;
- offset += ((u32) mhp->attr.va_fbo) %
- (1UL << (12 + mhp->attr.page_size));
+ offset += mhp->attr.va_fbo &
+ ((1UL << (12 + mhp->attr.page_size)) - 1);
pbl_addr[i] = ((mhp->attr.pbl_addr -
rhp->rdev.rnic_info.pbl_base) >> 3) +
(offset >> (12 + mhp->attr.page_size));
@@ -266,8 +263,8 @@
wqe->recv.sgl[i].len = cpu_to_be32(wr->sg_list[i].length);
/* to in the WQE == the offset into the page */
- wqe->recv.sgl[i].to = cpu_to_be64(((u32) wr->sg_list[i].addr) %
- (1UL << (12 + page_size[i])));
+ wqe->recv.sgl[i].to = cpu_to_be64(((u32)wr->sg_list[i].addr) &
+ ((1UL << (12 + page_size[i])) - 1));
/* pbl_addr is the adapters address in the PBL */
wqe->recv.pbl_addr[i] = cpu_to_be32(pbl_addr[i]);
diff --git a/drivers/infiniband/hw/ipath/ipath_init_chip.c b/drivers/infiniband/hw/ipath/ipath_init_chip.c
index 64aeefb..077879c 100644
--- a/drivers/infiniband/hw/ipath/ipath_init_chip.c
+++ b/drivers/infiniband/hw/ipath/ipath_init_chip.c
@@ -455,7 +455,7 @@
if (!addrs) {
ipath_dev_err(dd, "failed to allocate shadow dma handle "
"array, no expected sends!\n");
- vfree(dd->ipath_pageshadow);
+ vfree(pages);
dd->ipath_pageshadow = NULL;
return;
}
diff --git a/drivers/infiniband/hw/ipath/ipath_user_pages.c b/drivers/infiniband/hw/ipath/ipath_user_pages.c
index 0190edc..855911e 100644
--- a/drivers/infiniband/hw/ipath/ipath_user_pages.c
+++ b/drivers/infiniband/hw/ipath/ipath_user_pages.c
@@ -209,20 +209,20 @@
mm = get_task_mm(current);
if (!mm)
- goto bail;
+ return;
work = kmalloc(sizeof(*work), GFP_KERNEL);
if (!work)
goto bail_mm;
- goto bail;
-
INIT_WORK(&work->work, user_pages_account);
work->mm = mm;
work->num_pages = num_pages;
+ schedule_work(&work->work);
+ return;
+
bail_mm:
mmput(mm);
-bail:
return;
}
diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
index 606f1e2..19e68ab 100644
--- a/drivers/infiniband/hw/mlx4/mad.c
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -147,7 +147,8 @@
* Snoop SM MADs for port info and P_Key table sets, so we can
* synthesize LID change and P_Key change events.
*/
-static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad)
+static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad,
+ u16 prev_lid)
{
struct ib_event event;
@@ -157,6 +158,7 @@
if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) {
struct ib_port_info *pinfo =
(struct ib_port_info *) ((struct ib_smp *) mad)->data;
+ u16 lid = be16_to_cpu(pinfo->lid);
update_sm_ah(to_mdev(ibdev), port_num,
be16_to_cpu(pinfo->sm_lid),
@@ -165,12 +167,15 @@
event.device = ibdev;
event.element.port_num = port_num;
- if (pinfo->clientrereg_resv_subnetto & 0x80)
+ if (pinfo->clientrereg_resv_subnetto & 0x80) {
event.event = IB_EVENT_CLIENT_REREGISTER;
- else
- event.event = IB_EVENT_LID_CHANGE;
+ ib_dispatch_event(&event);
+ }
- ib_dispatch_event(&event);
+ if (prev_lid != lid) {
+ event.event = IB_EVENT_LID_CHANGE;
+ ib_dispatch_event(&event);
+ }
}
if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) {
@@ -228,8 +233,9 @@
struct ib_wc *in_wc, struct ib_grh *in_grh,
struct ib_mad *in_mad, struct ib_mad *out_mad)
{
- u16 slid;
+ u16 slid, prev_lid = 0;
int err;
+ struct ib_port_attr pattr;
slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
@@ -263,6 +269,13 @@
} else
return IB_MAD_RESULT_SUCCESS;
+ if ((in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+ in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+ in_mad->mad_hdr.method == IB_MGMT_METHOD_SET &&
+ in_mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO &&
+ !ib_query_port(ibdev, port_num, &pattr))
+ prev_lid = pattr.lid;
+
err = mlx4_MAD_IFC(to_mdev(ibdev),
mad_flags & IB_MAD_IGNORE_MKEY,
mad_flags & IB_MAD_IGNORE_BKEY,
@@ -271,7 +284,7 @@
return IB_MAD_RESULT_FAILURE;
if (!out_mad->mad_hdr.status) {
- smp_snoop(ibdev, port_num, in_mad);
+ smp_snoop(ibdev, port_num, in_mad, prev_lid);
node_desc_override(ibdev, out_mad);
}
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 61588bd2..2ccb9d3 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -699,11 +699,12 @@
struct mlx4_ib_dev *ibdev = ibdev_ptr;
int p;
+ mlx4_ib_mad_cleanup(ibdev);
+ ib_unregister_device(&ibdev->ib_dev);
+
for (p = 1; p <= ibdev->num_ports; ++p)
mlx4_CLOSE_PORT(dev, p);
- mlx4_ib_mad_cleanup(ibdev);
- ib_unregister_device(&ibdev->ib_dev);
iounmap(ibdev->uar_map);
mlx4_uar_free(dev, &ibdev->priv_uar);
mlx4_pd_free(dev, ibdev->priv_pdn);
diff --git a/drivers/infiniband/hw/mthca/mthca_mad.c b/drivers/infiniband/hw/mthca/mthca_mad.c
index 6404495..5648659 100644
--- a/drivers/infiniband/hw/mthca/mthca_mad.c
+++ b/drivers/infiniband/hw/mthca/mthca_mad.c
@@ -104,7 +104,8 @@
*/
static void smp_snoop(struct ib_device *ibdev,
u8 port_num,
- struct ib_mad *mad)
+ struct ib_mad *mad,
+ u16 prev_lid)
{
struct ib_event event;
@@ -114,6 +115,7 @@
if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) {
struct ib_port_info *pinfo =
(struct ib_port_info *) ((struct ib_smp *) mad)->data;
+ u16 lid = be16_to_cpu(pinfo->lid);
mthca_update_rate(to_mdev(ibdev), port_num);
update_sm_ah(to_mdev(ibdev), port_num,
@@ -123,12 +125,15 @@
event.device = ibdev;
event.element.port_num = port_num;
- if (pinfo->clientrereg_resv_subnetto & 0x80)
+ if (pinfo->clientrereg_resv_subnetto & 0x80) {
event.event = IB_EVENT_CLIENT_REREGISTER;
- else
- event.event = IB_EVENT_LID_CHANGE;
+ ib_dispatch_event(&event);
+ }
- ib_dispatch_event(&event);
+ if (prev_lid != lid) {
+ event.event = IB_EVENT_LID_CHANGE;
+ ib_dispatch_event(&event);
+ }
}
if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) {
@@ -196,6 +201,8 @@
int err;
u8 status;
u16 slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
+ u16 prev_lid = 0;
+ struct ib_port_attr pattr;
/* Forward locally generated traps to the SM */
if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP &&
@@ -233,6 +240,12 @@
return IB_MAD_RESULT_SUCCESS;
} else
return IB_MAD_RESULT_SUCCESS;
+ if ((in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED ||
+ in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) &&
+ in_mad->mad_hdr.method == IB_MGMT_METHOD_SET &&
+ in_mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO &&
+ !ib_query_port(ibdev, port_num, &pattr))
+ prev_lid = pattr.lid;
err = mthca_MAD_IFC(to_mdev(ibdev),
mad_flags & IB_MAD_IGNORE_MKEY,
@@ -252,7 +265,7 @@
}
if (!out_mad->mad_hdr.status) {
- smp_snoop(ibdev, port_num, in_mad);
+ smp_snoop(ibdev, port_num, in_mad, prev_lid);
node_desc_override(ibdev, out_mad);
}
diff --git a/drivers/infiniband/hw/nes/nes.c b/drivers/infiniband/hw/nes/nes.c
index b9611ad..ca59976 100644
--- a/drivers/infiniband/hw/nes/nes.c
+++ b/drivers/infiniband/hw/nes/nes.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
* Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
diff --git a/drivers/infiniband/hw/nes/nes.h b/drivers/infiniband/hw/nes/nes.h
index 13a5bb1..04b12ad 100644
--- a/drivers/infiniband/hw/nes/nes.h
+++ b/drivers/infiniband/hw/nes/nes.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
* Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
diff --git a/drivers/infiniband/hw/nes/nes_cm.c b/drivers/infiniband/hw/nes/nes_cm.c
index a01b448..5327f2b 100644
--- a/drivers/infiniband/hw/nes/nes_cm.c
+++ b/drivers/infiniband/hw/nes/nes_cm.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -103,6 +103,7 @@
static void nes_disconnect_worker(struct work_struct *work);
static int send_mpa_request(struct nes_cm_node *, struct sk_buff *);
+static int send_mpa_reject(struct nes_cm_node *);
static int send_syn(struct nes_cm_node *, u32, struct sk_buff *);
static int send_reset(struct nes_cm_node *, struct sk_buff *);
static int send_ack(struct nes_cm_node *cm_node, struct sk_buff *skb);
@@ -113,8 +114,7 @@
static void active_open_err(struct nes_cm_node *, struct sk_buff *, int);
static void passive_open_err(struct nes_cm_node *, struct sk_buff *, int);
static void cleanup_retrans_entry(struct nes_cm_node *);
-static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *,
- enum nes_cm_event_type);
+static void handle_rcv_mpa(struct nes_cm_node *, struct sk_buff *);
static void free_retrans_entry(struct nes_cm_node *cm_node);
static int handle_tcp_options(struct nes_cm_node *cm_node, struct tcphdr *tcph,
struct sk_buff *skb, int optionsize, int passive);
@@ -124,6 +124,8 @@
static void cm_event_connect_error(struct nes_cm_event *);
static void cm_event_reset(struct nes_cm_event *);
static void cm_event_mpa_req(struct nes_cm_event *);
+static void cm_event_mpa_reject(struct nes_cm_event *);
+static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node);
static void print_core(struct nes_cm_core *core);
@@ -196,7 +198,6 @@
*/
static int send_mpa_request(struct nes_cm_node *cm_node, struct sk_buff *skb)
{
- int ret;
if (!skb) {
nes_debug(NES_DBG_CM, "skb set to NULL\n");
return -1;
@@ -206,11 +207,27 @@
form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame,
cm_node->mpa_frame_size, SET_ACK);
- ret = schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
- if (ret < 0)
- return ret;
+ return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
+}
- return 0;
+
+
+static int send_mpa_reject(struct nes_cm_node *cm_node)
+{
+ struct sk_buff *skb = NULL;
+
+ skb = dev_alloc_skb(MAX_CM_BUFFER);
+ if (!skb) {
+ nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n");
+ return -ENOMEM;
+ }
+
+ /* send an MPA reject frame */
+ form_cm_frame(skb, cm_node, NULL, 0, &cm_node->mpa_frame,
+ cm_node->mpa_frame_size, SET_ACK | SET_FIN);
+
+ cm_node->state = NES_CM_STATE_FIN_WAIT1;
+ return schedule_nes_timer(cm_node, skb, NES_TIMER_TYPE_SEND, 1, 0);
}
@@ -218,14 +235,17 @@
* recv_mpa - process a received TCP pkt, we are expecting an
* IETF MPA frame
*/
-static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 len)
+static int parse_mpa(struct nes_cm_node *cm_node, u8 *buffer, u32 *type,
+ u32 len)
{
struct ietf_mpa_frame *mpa_frame;
+ *type = NES_MPA_REQUEST_ACCEPT;
+
/* assume req frame is in tcp data payload */
if (len < sizeof(struct ietf_mpa_frame)) {
nes_debug(NES_DBG_CM, "The received ietf buffer was too small (%x)\n", len);
- return -1;
+ return -EINVAL;
}
mpa_frame = (struct ietf_mpa_frame *)buffer;
@@ -234,14 +254,25 @@
if (cm_node->mpa_frame_size + sizeof(struct ietf_mpa_frame) != len) {
nes_debug(NES_DBG_CM, "The received ietf buffer was not right"
" complete (%x + %x != %x)\n",
- cm_node->mpa_frame_size, (u32)sizeof(struct ietf_mpa_frame), len);
- return -1;
+ cm_node->mpa_frame_size,
+ (u32)sizeof(struct ietf_mpa_frame), len);
+ return -EINVAL;
+ }
+ /* make sure it does not exceed the max size */
+ if (len > MAX_CM_BUFFER) {
+ nes_debug(NES_DBG_CM, "The received ietf buffer was too large"
+ " (%x + %x != %x)\n",
+ cm_node->mpa_frame_size,
+ (u32)sizeof(struct ietf_mpa_frame), len);
+ return -EINVAL;
}
/* copy entire MPA frame to our cm_node's frame */
memcpy(cm_node->mpa_frame_buf, buffer + sizeof(struct ietf_mpa_frame),
cm_node->mpa_frame_size);
+ if (mpa_frame->flags & IETF_MPA_FLAGS_REJECT)
+ *type = NES_MPA_REQUEST_REJECT;
return 0;
}
@@ -380,7 +411,7 @@
new_send = kzalloc(sizeof(*new_send), GFP_ATOMIC);
if (!new_send)
- return -1;
+ return -ENOMEM;
/* new_send->timetosend = currenttime */
new_send->retrycount = NES_DEFAULT_RETRYS;
@@ -394,9 +425,11 @@
if (type == NES_TIMER_TYPE_CLOSE) {
new_send->timetosend += (HZ/10);
- spin_lock_irqsave(&cm_node->recv_list_lock, flags);
- list_add_tail(&new_send->list, &cm_node->recv_list);
- spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
+ if (cm_node->recv_entry) {
+ WARN_ON(1);
+ return -EINVAL;
+ }
+ cm_node->recv_entry = new_send;
}
if (type == NES_TIMER_TYPE_SEND) {
@@ -435,24 +468,78 @@
return ret;
}
+static void nes_retrans_expired(struct nes_cm_node *cm_node)
+{
+ switch (cm_node->state) {
+ case NES_CM_STATE_SYN_RCVD:
+ case NES_CM_STATE_CLOSING:
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
+ break;
+ case NES_CM_STATE_LAST_ACK:
+ case NES_CM_STATE_FIN_WAIT1:
+ case NES_CM_STATE_MPAREJ_RCVD:
+ send_reset(cm_node, NULL);
+ break;
+ default:
+ create_event(cm_node, NES_CM_EVENT_ABORTED);
+ }
+}
+
+static void handle_recv_entry(struct nes_cm_node *cm_node, u32 rem_node)
+{
+ struct nes_timer_entry *recv_entry = cm_node->recv_entry;
+ struct iw_cm_id *cm_id = cm_node->cm_id;
+ struct nes_qp *nesqp;
+ unsigned long qplockflags;
+
+ if (!recv_entry)
+ return;
+ nesqp = (struct nes_qp *)recv_entry->skb;
+ if (nesqp) {
+ spin_lock_irqsave(&nesqp->lock, qplockflags);
+ if (nesqp->cm_id) {
+ nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
+ "refcount = %d: HIT A "
+ "NES_TIMER_TYPE_CLOSE with something "
+ "to do!!!\n", nesqp->hwqp.qp_id, cm_id,
+ atomic_read(&nesqp->refcount));
+ nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
+ nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
+ nesqp->ibqp_state = IB_QPS_ERR;
+ spin_unlock_irqrestore(&nesqp->lock, qplockflags);
+ nes_cm_disconn(nesqp);
+ } else {
+ spin_unlock_irqrestore(&nesqp->lock, qplockflags);
+ nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
+ "refcount = %d: HIT A "
+ "NES_TIMER_TYPE_CLOSE with nothing "
+ "to do!!!\n", nesqp->hwqp.qp_id, cm_id,
+ atomic_read(&nesqp->refcount));
+ }
+ } else if (rem_node) {
+ /* TIME_WAIT state */
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
+ }
+ if (cm_node->cm_id)
+ cm_id->rem_ref(cm_id);
+ kfree(recv_entry);
+ cm_node->recv_entry = NULL;
+}
/**
* nes_cm_timer_tick
*/
static void nes_cm_timer_tick(unsigned long pass)
{
- unsigned long flags, qplockflags;
+ unsigned long flags;
unsigned long nexttimeout = jiffies + NES_LONG_TIME;
- struct iw_cm_id *cm_id;
struct nes_cm_node *cm_node;
struct nes_timer_entry *send_entry, *recv_entry;
- struct list_head *list_core, *list_core_temp;
- struct list_head *list_node, *list_node_temp;
+ struct list_head *list_core_temp;
+ struct list_head *list_node;
struct nes_cm_core *cm_core = g_cm_core;
- struct nes_qp *nesqp;
u32 settimer = 0;
int ret = NETDEV_TX_OK;
- enum nes_cm_node_state last_state;
struct list_head timer_list;
INIT_LIST_HEAD(&timer_list);
@@ -461,7 +548,7 @@
list_for_each_safe(list_node, list_core_temp,
&cm_core->connected_nodes) {
cm_node = container_of(list_node, struct nes_cm_node, list);
- if (!list_empty(&cm_node->recv_list) || (cm_node->send_entry)) {
+ if ((cm_node->recv_entry) || (cm_node->send_entry)) {
add_ref_cm_node(cm_node);
list_add(&cm_node->timer_entry, &timer_list);
}
@@ -471,54 +558,18 @@
list_for_each_safe(list_node, list_core_temp, &timer_list) {
cm_node = container_of(list_node, struct nes_cm_node,
timer_entry);
- spin_lock_irqsave(&cm_node->recv_list_lock, flags);
- list_for_each_safe(list_core, list_node_temp,
- &cm_node->recv_list) {
- recv_entry = container_of(list_core,
- struct nes_timer_entry, list);
- if (!recv_entry)
- break;
+ recv_entry = cm_node->recv_entry;
+
+ if (recv_entry) {
if (time_after(recv_entry->timetosend, jiffies)) {
if (nexttimeout > recv_entry->timetosend ||
- !settimer) {
+ !settimer) {
nexttimeout = recv_entry->timetosend;
settimer = 1;
}
- continue;
- }
- list_del(&recv_entry->list);
- cm_id = cm_node->cm_id;
- spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
- nesqp = (struct nes_qp *)recv_entry->skb;
- spin_lock_irqsave(&nesqp->lock, qplockflags);
- if (nesqp->cm_id) {
- nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
- "refcount = %d: HIT A "
- "NES_TIMER_TYPE_CLOSE with something "
- "to do!!!\n", nesqp->hwqp.qp_id, cm_id,
- atomic_read(&nesqp->refcount));
- nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
- nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
- nesqp->ibqp_state = IB_QPS_ERR;
- spin_unlock_irqrestore(&nesqp->lock,
- qplockflags);
- nes_cm_disconn(nesqp);
- } else {
- spin_unlock_irqrestore(&nesqp->lock,
- qplockflags);
- nes_debug(NES_DBG_CM, "QP%u: cm_id = %p, "
- "refcount = %d: HIT A "
- "NES_TIMER_TYPE_CLOSE with nothing "
- "to do!!!\n", nesqp->hwqp.qp_id, cm_id,
- atomic_read(&nesqp->refcount));
- }
- if (cm_id)
- cm_id->rem_ref(cm_id);
-
- kfree(recv_entry);
- spin_lock_irqsave(&cm_node->recv_list_lock, flags);
+ } else
+ handle_recv_entry(cm_node, 1);
}
- spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
spin_lock_irqsave(&cm_node->retrans_list_lock, flags);
do {
@@ -533,12 +584,11 @@
nexttimeout =
send_entry->timetosend;
settimer = 1;
- break;
}
} else {
free_retrans_entry(cm_node);
- break;
}
+ break;
}
if ((cm_node->state == NES_CM_STATE_TSA) ||
@@ -550,16 +600,12 @@
if (!send_entry->retranscount ||
!send_entry->retrycount) {
cm_packets_dropped++;
- last_state = cm_node->state;
- cm_node->state = NES_CM_STATE_CLOSED;
free_retrans_entry(cm_node);
+
spin_unlock_irqrestore(
&cm_node->retrans_list_lock, flags);
- if (last_state == NES_CM_STATE_SYN_RCVD)
- rem_ref_cm_node(cm_core, cm_node);
- else
- create_event(cm_node,
- NES_CM_EVENT_ABORTED);
+ nes_retrans_expired(cm_node);
+ cm_node->state = NES_CM_STATE_CLOSED;
spin_lock_irqsave(&cm_node->retrans_list_lock,
flags);
break;
@@ -714,7 +760,7 @@
skb = dev_alloc_skb(MAX_CM_BUFFER);
if (!skb) {
nes_debug(NES_DBG_CM, "Failed to get a Free pkt\n");
- return -1;
+ return -ENOMEM;
}
form_cm_frame(skb, cm_node, NULL, 0, NULL, 0, flags);
@@ -778,14 +824,10 @@
unsigned long flags;
struct list_head *hte;
struct nes_cm_node *cm_node;
- __be32 tmp_addr = cpu_to_be32(loc_addr);
/* get a handle on the hte */
hte = &cm_core->connected_nodes;
- nes_debug(NES_DBG_CM, "Searching for an owner node: %pI4:%x from core %p->%p\n",
- &tmp_addr, loc_port, cm_core, hte);
-
/* walk list and find cm_node associated with this session ID */
spin_lock_irqsave(&cm_core->ht_lock, flags);
list_for_each_entry(cm_node, hte, list) {
@@ -875,7 +917,8 @@
static int mini_cm_dec_refcnt_listen(struct nes_cm_core *cm_core,
struct nes_cm_listener *listener, int free_hanging_nodes)
{
- int ret = 1;
+ int ret = -EINVAL;
+ int err = 0;
unsigned long flags;
struct list_head *list_pos = NULL;
struct list_head *list_temp = NULL;
@@ -904,10 +947,60 @@
list_for_each_safe(list_pos, list_temp, &reset_list) {
cm_node = container_of(list_pos, struct nes_cm_node,
- reset_entry);
- cleanup_retrans_entry(cm_node);
- send_reset(cm_node, NULL);
- rem_ref_cm_node(cm_node->cm_core, cm_node);
+ reset_entry);
+ {
+ struct nes_cm_node *loopback = cm_node->loopbackpartner;
+ if (NES_CM_STATE_FIN_WAIT1 <= cm_node->state) {
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
+ } else {
+ if (!loopback) {
+ cleanup_retrans_entry(cm_node);
+ err = send_reset(cm_node, NULL);
+ if (err) {
+ cm_node->state =
+ NES_CM_STATE_CLOSED;
+ WARN_ON(1);
+ } else {
+ cm_node->state =
+ NES_CM_STATE_CLOSED;
+ rem_ref_cm_node(
+ cm_node->cm_core,
+ cm_node);
+ }
+ } else {
+ struct nes_cm_event event;
+
+ event.cm_node = loopback;
+ event.cm_info.rem_addr =
+ loopback->rem_addr;
+ event.cm_info.loc_addr =
+ loopback->loc_addr;
+ event.cm_info.rem_port =
+ loopback->rem_port;
+ event.cm_info.loc_port =
+ loopback->loc_port;
+ event.cm_info.cm_id = loopback->cm_id;
+ cm_event_connect_error(&event);
+ loopback->state = NES_CM_STATE_CLOSED;
+
+ event.cm_node = cm_node;
+ event.cm_info.rem_addr =
+ cm_node->rem_addr;
+ event.cm_info.loc_addr =
+ cm_node->loc_addr;
+ event.cm_info.rem_port =
+ cm_node->rem_port;
+ event.cm_info.loc_port =
+ cm_node->loc_port;
+ event.cm_info.cm_id = cm_node->cm_id;
+ cm_event_reset(&event);
+
+ rem_ref_cm_node(cm_node->cm_core,
+ cm_node);
+
+ }
+ }
+ }
}
spin_lock_irqsave(&cm_core->listen_list_lock, flags);
@@ -968,6 +1061,7 @@
if (cm_node->accept_pend) {
BUG_ON(!cm_node->listener);
atomic_dec(&cm_node->listener->pend_accepts_cnt);
+ cm_node->accept_pend = 0;
BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0);
}
@@ -994,7 +1088,7 @@
memset(&fl, 0, sizeof fl);
fl.nl_u.ip4_u.daddr = htonl(dst_ip);
if (ip_route_output_key(&init_net, &rt, &fl)) {
- printk("%s: ip_route_output_key failed for 0x%08X\n",
+ printk(KERN_ERR "%s: ip_route_output_key failed for 0x%08X\n",
__func__, dst_ip);
return rc;
}
@@ -1057,8 +1151,6 @@
cm_node->cm_id);
spin_lock_init(&cm_node->retrans_list_lock);
- INIT_LIST_HEAD(&cm_node->recv_list);
- spin_lock_init(&cm_node->recv_list_lock);
cm_node->loopbackpartner = NULL;
atomic_set(&cm_node->ref_count, 1);
@@ -1126,10 +1218,7 @@
static int rem_ref_cm_node(struct nes_cm_core *cm_core,
struct nes_cm_node *cm_node)
{
- unsigned long flags, qplockflags;
- struct nes_timer_entry *recv_entry;
- struct iw_cm_id *cm_id;
- struct list_head *list_core, *list_node_temp;
+ unsigned long flags;
struct nes_qp *nesqp;
if (!cm_node)
@@ -1150,38 +1239,9 @@
atomic_dec(&cm_node->listener->pend_accepts_cnt);
BUG_ON(atomic_read(&cm_node->listener->pend_accepts_cnt) < 0);
}
- BUG_ON(cm_node->send_entry);
- spin_lock_irqsave(&cm_node->recv_list_lock, flags);
- list_for_each_safe(list_core, list_node_temp, &cm_node->recv_list) {
- recv_entry = container_of(list_core, struct nes_timer_entry,
- list);
- list_del(&recv_entry->list);
- cm_id = cm_node->cm_id;
- spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
- nesqp = (struct nes_qp *)recv_entry->skb;
- spin_lock_irqsave(&nesqp->lock, qplockflags);
- if (nesqp->cm_id) {
- nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A "
- "NES_TIMER_TYPE_CLOSE with something to do!\n",
- nesqp->hwqp.qp_id, cm_id);
- nesqp->hw_tcp_state = NES_AEQE_TCP_STATE_CLOSED;
- nesqp->last_aeq = NES_AEQE_AEID_RESET_SENT;
- nesqp->ibqp_state = IB_QPS_ERR;
- spin_unlock_irqrestore(&nesqp->lock, qplockflags);
- nes_cm_disconn(nesqp);
- } else {
- spin_unlock_irqrestore(&nesqp->lock, qplockflags);
- nes_debug(NES_DBG_CM, "QP%u: cm_id = %p: HIT A "
- "NES_TIMER_TYPE_CLOSE with nothing to do!\n",
- nesqp->hwqp.qp_id, cm_id);
- }
- cm_id->rem_ref(cm_id);
-
- kfree(recv_entry);
- spin_lock_irqsave(&cm_node->recv_list_lock, flags);
- }
- spin_unlock_irqrestore(&cm_node->recv_list_lock, flags);
-
+ WARN_ON(cm_node->send_entry);
+ if (cm_node->recv_entry)
+ handle_recv_entry(cm_node, 0);
if (cm_node->listener) {
mini_cm_dec_refcnt_listen(cm_core, cm_node->listener, 0);
} else {
@@ -1266,8 +1326,7 @@
dev_kfree_skb_any(skb);
}
-static void handle_fin_pkt(struct nes_cm_node *cm_node, struct sk_buff *skb,
- struct tcphdr *tcph)
+static void handle_fin_pkt(struct nes_cm_node *cm_node)
{
nes_debug(NES_DBG_CM, "Received FIN, cm_node = %p, state = %u. "
"refcnt=%d\n", cm_node, cm_node->state,
@@ -1279,23 +1338,30 @@
case NES_CM_STATE_SYN_SENT:
case NES_CM_STATE_ESTABLISHED:
case NES_CM_STATE_MPAREQ_SENT:
+ case NES_CM_STATE_MPAREJ_RCVD:
cm_node->state = NES_CM_STATE_LAST_ACK;
- send_fin(cm_node, skb);
+ send_fin(cm_node, NULL);
break;
case NES_CM_STATE_FIN_WAIT1:
cm_node->state = NES_CM_STATE_CLOSING;
- send_ack(cm_node, skb);
+ send_ack(cm_node, NULL);
+ /* Wait for ACK as this is simultanous close..
+ * After we receive ACK, do not send anything..
+ * Just rm the node.. Done.. */
break;
case NES_CM_STATE_FIN_WAIT2:
cm_node->state = NES_CM_STATE_TIME_WAIT;
- send_ack(cm_node, skb);
+ send_ack(cm_node, NULL);
+ schedule_nes_timer(cm_node, NULL, NES_TIMER_TYPE_CLOSE, 1, 0);
+ break;
+ case NES_CM_STATE_TIME_WAIT:
cm_node->state = NES_CM_STATE_CLOSED;
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
break;
case NES_CM_STATE_TSA:
default:
nes_debug(NES_DBG_CM, "Error Rcvd FIN for node-%p state = %d\n",
cm_node, cm_node->state);
- drop_packet(skb);
break;
}
}
@@ -1341,23 +1407,35 @@
cleanup_retrans_entry(cm_node);
drop_packet(skb);
break;
+ case NES_CM_STATE_TIME_WAIT:
+ cleanup_retrans_entry(cm_node);
+ cm_node->state = NES_CM_STATE_CLOSED;
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
+ drop_packet(skb);
+ break;
+ case NES_CM_STATE_FIN_WAIT1:
+ cleanup_retrans_entry(cm_node);
+ nes_debug(NES_DBG_CM, "Bad state %s[%u]\n", __func__, __LINE__);
default:
drop_packet(skb);
break;
}
}
-static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb,
- enum nes_cm_event_type type)
+
+static void handle_rcv_mpa(struct nes_cm_node *cm_node, struct sk_buff *skb)
{
- int ret;
+ int ret = 0;
int datasize = skb->len;
u8 *dataloc = skb->data;
- ret = parse_mpa(cm_node, dataloc, datasize);
- if (ret < 0) {
+
+ enum nes_cm_event_type type = NES_CM_EVENT_UNKNOWN;
+ u32 res_type;
+ ret = parse_mpa(cm_node, dataloc, &res_type, datasize);
+ if (ret) {
nes_debug(NES_DBG_CM, "didn't like MPA Request\n");
- if (type == NES_CM_EVENT_CONNECTED) {
+ if (cm_node->state == NES_CM_STATE_MPAREQ_SENT) {
nes_debug(NES_DBG_CM, "%s[%u] create abort for "
"cm_node=%p listener=%p state=%d\n", __func__,
__LINE__, cm_node, cm_node->listener,
@@ -1366,18 +1444,38 @@
} else {
passive_open_err(cm_node, skb, 1);
}
- } else {
- cleanup_retrans_entry(cm_node);
- dev_kfree_skb_any(skb);
- if (type == NES_CM_EVENT_CONNECTED)
- cm_node->state = NES_CM_STATE_TSA;
- else
- atomic_set(&cm_node->passive_state,
- NES_PASSIVE_STATE_INDICATED);
- create_event(cm_node, type);
-
+ return;
}
- return ;
+
+ switch (cm_node->state) {
+ case NES_CM_STATE_ESTABLISHED:
+ if (res_type == NES_MPA_REQUEST_REJECT) {
+ /*BIG problem as we are receiving the MPA.. So should
+ * not be REJECT.. This is Passive Open.. We can
+ * only receive it Reject for Active Open...*/
+ WARN_ON(1);
+ }
+ cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
+ type = NES_CM_EVENT_MPA_REQ;
+ atomic_set(&cm_node->passive_state,
+ NES_PASSIVE_STATE_INDICATED);
+ break;
+ case NES_CM_STATE_MPAREQ_SENT:
+ if (res_type == NES_MPA_REQUEST_REJECT) {
+ type = NES_CM_EVENT_MPA_REJECT;
+ cm_node->state = NES_CM_STATE_MPAREJ_RCVD;
+ } else {
+ type = NES_CM_EVENT_CONNECTED;
+ cm_node->state = NES_CM_STATE_TSA;
+ }
+
+ break;
+ default:
+ WARN_ON(1);
+ break;
+ }
+ dev_kfree_skb_any(skb);
+ create_event(cm_node, type);
}
static void indicate_pkt_err(struct nes_cm_node *cm_node, struct sk_buff *skb)
@@ -1465,8 +1563,6 @@
break;
case NES_CM_STATE_LISTENING:
/* Passive OPEN */
- cm_node->accept_pend = 1;
- atomic_inc(&cm_node->listener->pend_accepts_cnt);
if (atomic_read(&cm_node->listener->pend_accepts_cnt) >
cm_node->listener->backlog) {
nes_debug(NES_DBG_CM, "drop syn due to backlog "
@@ -1484,6 +1580,9 @@
}
cm_node->tcp_cntxt.rcv_nxt = inc_sequence + 1;
BUG_ON(cm_node->send_entry);
+ cm_node->accept_pend = 1;
+ atomic_inc(&cm_node->listener->pend_accepts_cnt);
+
cm_node->state = NES_CM_STATE_SYN_RCVD;
send_syn(cm_node, 1, skb);
break;
@@ -1518,6 +1617,7 @@
inc_sequence = ntohl(tcph->seq);
switch (cm_node->state) {
case NES_CM_STATE_SYN_SENT:
+ cleanup_retrans_entry(cm_node);
/* active open */
if (check_syn(cm_node, tcph, skb))
return;
@@ -1567,10 +1667,7 @@
u32 rem_seq;
int ret;
int optionsize;
- u32 temp_seq = cm_node->tcp_cntxt.loc_seq_num;
-
optionsize = (tcph->doff << 2) - sizeof(struct tcphdr);
- cm_node->tcp_cntxt.loc_seq_num = ntohl(tcph->ack_seq);
if (check_seq(cm_node, tcph, skb))
return;
@@ -1580,7 +1677,7 @@
rem_seq = ntohl(tcph->seq);
rem_seq_ack = ntohl(tcph->ack_seq);
datasize = skb->len;
-
+ cleanup_retrans_entry(cm_node);
switch (cm_node->state) {
case NES_CM_STATE_SYN_RCVD:
/* Passive OPEN */
@@ -1588,7 +1685,6 @@
if (ret)
break;
cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq);
- cm_node->tcp_cntxt.loc_seq_num = temp_seq;
if (cm_node->tcp_cntxt.rem_ack_num !=
cm_node->tcp_cntxt.loc_seq_num) {
nes_debug(NES_DBG_CM, "rem_ack_num != loc_seq_num\n");
@@ -1597,31 +1693,30 @@
return;
}
cm_node->state = NES_CM_STATE_ESTABLISHED;
+ cleanup_retrans_entry(cm_node);
if (datasize) {
cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
- cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
- handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_MPA_REQ);
- } else { /* rcvd ACK only */
+ handle_rcv_mpa(cm_node, skb);
+ } else { /* rcvd ACK only */
dev_kfree_skb_any(skb);
cleanup_retrans_entry(cm_node);
}
break;
case NES_CM_STATE_ESTABLISHED:
/* Passive OPEN */
- /* We expect mpa frame to be received only */
+ cleanup_retrans_entry(cm_node);
if (datasize) {
cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
- cm_node->state = NES_CM_STATE_MPAREQ_RCVD;
- handle_rcv_mpa(cm_node, skb,
- NES_CM_EVENT_MPA_REQ);
+ handle_rcv_mpa(cm_node, skb);
} else
drop_packet(skb);
break;
case NES_CM_STATE_MPAREQ_SENT:
+ cleanup_retrans_entry(cm_node);
cm_node->tcp_cntxt.rem_ack_num = ntohl(tcph->ack_seq);
if (datasize) {
cm_node->tcp_cntxt.rcv_nxt = inc_sequence + datasize;
- handle_rcv_mpa(cm_node, skb, NES_CM_EVENT_CONNECTED);
+ handle_rcv_mpa(cm_node, skb);
} else { /* Could be just an ack pkt.. */
cleanup_retrans_entry(cm_node);
dev_kfree_skb_any(skb);
@@ -1632,13 +1727,24 @@
cleanup_retrans_entry(cm_node);
send_reset(cm_node, skb);
break;
+ case NES_CM_STATE_LAST_ACK:
+ cleanup_retrans_entry(cm_node);
+ cm_node->state = NES_CM_STATE_CLOSED;
+ cm_node->cm_id->rem_ref(cm_node->cm_id);
+ case NES_CM_STATE_CLOSING:
+ cleanup_retrans_entry(cm_node);
+ rem_ref_cm_node(cm_node->cm_core, cm_node);
+ drop_packet(skb);
+ break;
case NES_CM_STATE_FIN_WAIT1:
+ cleanup_retrans_entry(cm_node);
+ drop_packet(skb);
+ cm_node->state = NES_CM_STATE_FIN_WAIT2;
+ break;
case NES_CM_STATE_SYN_SENT:
case NES_CM_STATE_FIN_WAIT2:
case NES_CM_STATE_TSA:
case NES_CM_STATE_MPAREQ_RCVD:
- case NES_CM_STATE_LAST_ACK:
- case NES_CM_STATE_CLOSING:
case NES_CM_STATE_UNKNOWN:
default:
drop_packet(skb);
@@ -1748,6 +1854,7 @@
{
enum nes_tcpip_pkt_type pkt_type = NES_PKT_TYPE_UNKNOWN;
struct tcphdr *tcph = tcp_hdr(skb);
+ u32 fin_set = 0;
skb_pull(skb, ip_hdr(skb)->ihl << 2);
nes_debug(NES_DBG_CM, "process_packet: cm_node=%p state =%d syn=%d "
@@ -1760,10 +1867,10 @@
pkt_type = NES_PKT_TYPE_SYN;
if (tcph->ack)
pkt_type = NES_PKT_TYPE_SYNACK;
- } else if (tcph->fin)
- pkt_type = NES_PKT_TYPE_FIN;
- else if (tcph->ack)
+ } else if (tcph->ack)
pkt_type = NES_PKT_TYPE_ACK;
+ if (tcph->fin)
+ fin_set = 1;
switch (pkt_type) {
case NES_PKT_TYPE_SYN:
@@ -1774,15 +1881,16 @@
break;
case NES_PKT_TYPE_ACK:
handle_ack_pkt(cm_node, skb, tcph);
+ if (fin_set)
+ handle_fin_pkt(cm_node);
break;
case NES_PKT_TYPE_RST:
handle_rst_pkt(cm_node, skb, tcph);
break;
- case NES_PKT_TYPE_FIN:
- handle_fin_pkt(cm_node, skb, tcph);
- break;
default:
drop_packet(skb);
+ if (fin_set)
+ handle_fin_pkt(cm_node);
break;
}
}
@@ -1925,7 +2033,7 @@
loopbackremotenode->tcp_cntxt.rcv_wscale;
loopbackremotenode->tcp_cntxt.snd_wscale =
cm_node->tcp_cntxt.rcv_wscale;
-
+ loopbackremotenode->state = NES_CM_STATE_MPAREQ_RCVD;
create_event(loopbackremotenode, NES_CM_EVENT_MPA_REQ);
}
return cm_node;
@@ -1980,7 +2088,11 @@
struct ietf_mpa_frame *mpa_frame, struct nes_cm_node *cm_node)
{
int ret = 0;
+ int err = 0;
int passive_state;
+ struct nes_cm_event event;
+ struct iw_cm_id *cm_id = cm_node->cm_id;
+ struct nes_cm_node *loopback = cm_node->loopbackpartner;
nes_debug(NES_DBG_CM, "%s cm_node=%p type=%d state=%d\n",
__func__, cm_node, cm_node->tcp_cntxt.client, cm_node->state);
@@ -1989,12 +2101,38 @@
return ret;
cleanup_retrans_entry(cm_node);
- passive_state = atomic_add_return(1, &cm_node->passive_state);
- cm_node->state = NES_CM_STATE_CLOSED;
- if (passive_state == NES_SEND_RESET_EVENT)
+ if (!loopback) {
+ passive_state = atomic_add_return(1, &cm_node->passive_state);
+ if (passive_state == NES_SEND_RESET_EVENT) {
+ cm_node->state = NES_CM_STATE_CLOSED;
+ rem_ref_cm_node(cm_core, cm_node);
+ } else {
+ ret = send_mpa_reject(cm_node);
+ if (ret) {
+ cm_node->state = NES_CM_STATE_CLOSED;
+ err = send_reset(cm_node, NULL);
+ if (err)
+ WARN_ON(1);
+ } else
+ cm_id->add_ref(cm_id);
+ }
+ } else {
+ cm_node->cm_id = NULL;
+ event.cm_node = loopback;
+ event.cm_info.rem_addr = loopback->rem_addr;
+ event.cm_info.loc_addr = loopback->loc_addr;
+ event.cm_info.rem_port = loopback->rem_port;
+ event.cm_info.loc_port = loopback->loc_port;
+ event.cm_info.cm_id = loopback->cm_id;
+ cm_event_mpa_reject(&event);
rem_ref_cm_node(cm_core, cm_node);
- else
- ret = send_reset(cm_node, NULL);
+ loopback->state = NES_CM_STATE_CLOSING;
+
+ cm_id = loopback->cm_id;
+ rem_ref_cm_node(cm_core, loopback);
+ cm_id->rem_ref(cm_id);
+ }
+
return ret;
}
@@ -2031,6 +2169,7 @@
case NES_CM_STATE_CLOSING:
ret = -1;
break;
+ case NES_CM_STATE_MPAREJ_RCVD:
case NES_CM_STATE_LISTENING:
case NES_CM_STATE_UNKNOWN:
case NES_CM_STATE_INITED:
@@ -2227,15 +2366,15 @@
int ret = 0;
switch (type) {
- case NES_CM_SET_PKT_SIZE:
- cm_core->mtu = value;
- break;
- case NES_CM_SET_FREE_PKT_Q_SIZE:
- cm_core->free_tx_pkt_max = value;
- break;
- default:
- /* unknown set option */
- ret = -EINVAL;
+ case NES_CM_SET_PKT_SIZE:
+ cm_core->mtu = value;
+ break;
+ case NES_CM_SET_FREE_PKT_Q_SIZE:
+ cm_core->free_tx_pkt_max = value;
+ break;
+ default:
+ /* unknown set option */
+ ret = -EINVAL;
}
return ret;
@@ -2625,9 +2764,7 @@
NES_QPCONTEXT_ORDIRD_WRPDU);
} else {
nesqp->nesqp_context->ird_ord_sizes |=
- cpu_to_le32((NES_QPCONTEXT_ORDIRD_LSMM_PRESENT |
- NES_QPCONTEXT_ORDIRD_WRPDU |
- NES_QPCONTEXT_ORDIRD_ALSMM));
+ cpu_to_le32(NES_QPCONTEXT_ORDIRD_WRPDU);
}
nesqp->skip_lsmm = 1;
@@ -2749,23 +2886,35 @@
int nes_reject(struct iw_cm_id *cm_id, const void *pdata, u8 pdata_len)
{
struct nes_cm_node *cm_node;
+ struct nes_cm_node *loopback;
+
struct nes_cm_core *cm_core;
atomic_inc(&cm_rejects);
cm_node = (struct nes_cm_node *) cm_id->provider_data;
+ loopback = cm_node->loopbackpartner;
cm_core = cm_node->cm_core;
+ cm_node->cm_id = cm_id;
cm_node->mpa_frame_size = sizeof(struct ietf_mpa_frame) + pdata_len;
- strcpy(&cm_node->mpa_frame.key[0], IEFT_MPA_KEY_REP);
- memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len);
+ if (cm_node->mpa_frame_size > MAX_CM_BUFFER)
+ return -EINVAL;
- cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len);
+ strcpy(&cm_node->mpa_frame.key[0], IEFT_MPA_KEY_REP);
+ if (loopback) {
+ memcpy(&loopback->mpa_frame.priv_data, pdata, pdata_len);
+ loopback->mpa_frame.priv_data_len = pdata_len;
+ loopback->mpa_frame_size = sizeof(struct ietf_mpa_frame) +
+ pdata_len;
+ } else {
+ memcpy(&cm_node->mpa_frame.priv_data, pdata, pdata_len);
+ cm_node->mpa_frame.priv_data_len = cpu_to_be16(pdata_len);
+ }
+
cm_node->mpa_frame.rev = mpa_version;
cm_node->mpa_frame.flags = IETF_MPA_FLAGS_CRC | IETF_MPA_FLAGS_REJECT;
- cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node);
-
- return 0;
+ return cm_core->api->reject(cm_core, &cm_node->mpa_frame, cm_node);
}
@@ -3274,13 +3423,56 @@
cm_event.remote_addr.sin_family = AF_INET;
cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port);
cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr);
-
- cm_event.private_data = cm_node->mpa_frame_buf;
- cm_event.private_data_len = (u8) cm_node->mpa_frame_size;
+ cm_event.private_data = cm_node->mpa_frame_buf;
+ cm_event.private_data_len = (u8) cm_node->mpa_frame_size;
ret = cm_id->event_handler(cm_id, &cm_event);
if (ret)
- printk("%s[%u] OFA CM event_handler returned, ret=%d\n",
+ printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n",
+ __func__, __LINE__, ret);
+ return;
+}
+
+
+static void cm_event_mpa_reject(struct nes_cm_event *event)
+{
+ struct iw_cm_id *cm_id;
+ struct iw_cm_event cm_event;
+ struct nes_cm_node *cm_node;
+ int ret;
+
+ cm_node = event->cm_node;
+ if (!cm_node)
+ return;
+ cm_id = cm_node->cm_id;
+
+ atomic_inc(&cm_connect_reqs);
+ nes_debug(NES_DBG_CM, "cm_node = %p - cm_id = %p, jiffies = %lu\n",
+ cm_node, cm_id, jiffies);
+
+ cm_event.event = IW_CM_EVENT_CONNECT_REPLY;
+ cm_event.status = -ECONNREFUSED;
+ cm_event.provider_data = cm_id->provider_data;
+
+ cm_event.local_addr.sin_family = AF_INET;
+ cm_event.local_addr.sin_port = htons(event->cm_info.loc_port);
+ cm_event.local_addr.sin_addr.s_addr = htonl(event->cm_info.loc_addr);
+
+ cm_event.remote_addr.sin_family = AF_INET;
+ cm_event.remote_addr.sin_port = htons(event->cm_info.rem_port);
+ cm_event.remote_addr.sin_addr.s_addr = htonl(event->cm_info.rem_addr);
+
+ cm_event.private_data = cm_node->mpa_frame_buf;
+ cm_event.private_data_len = (u8) cm_node->mpa_frame_size;
+
+ nes_debug(NES_DBG_CM, "call CM_EVENT_MPA_REJECTED, local_addr=%08x, "
+ "remove_addr=%08x\n",
+ cm_event.local_addr.sin_addr.s_addr,
+ cm_event.remote_addr.sin_addr.s_addr);
+
+ ret = cm_id->event_handler(cm_id, &cm_event);
+ if (ret)
+ printk(KERN_ERR "%s[%u] OFA CM event_handler returned, ret=%d\n",
__func__, __LINE__, ret);
return;
@@ -3345,6 +3537,14 @@
cm_event_connected(event);
nes_debug(NES_DBG_CM, "CM Event: CONNECTED\n");
break;
+ case NES_CM_EVENT_MPA_REJECT:
+ if ((!event->cm_node->cm_id) ||
+ (event->cm_node->state == NES_CM_STATE_TSA))
+ break;
+ cm_event_mpa_reject(event);
+ nes_debug(NES_DBG_CM, "CM Event: REJECT\n");
+ break;
+
case NES_CM_EVENT_ABORTED:
if ((!event->cm_node->cm_id) ||
(event->cm_node->state == NES_CM_STATE_TSA))
diff --git a/drivers/infiniband/hw/nes/nes_cm.h b/drivers/infiniband/hw/nes/nes_cm.h
index fafa350..d5f7782 100644
--- a/drivers/infiniband/hw/nes/nes_cm.h
+++ b/drivers/infiniband/hw/nes/nes_cm.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -39,6 +39,9 @@
#define NES_MANAGE_APBVT_DEL 0
#define NES_MANAGE_APBVT_ADD 1
+#define NES_MPA_REQUEST_ACCEPT 1
+#define NES_MPA_REQUEST_REJECT 2
+
/* IETF MPA -- defines, enums, structs */
#define IEFT_MPA_KEY_REQ "MPA ID Req Frame"
#define IEFT_MPA_KEY_REP "MPA ID Rep Frame"
@@ -186,6 +189,7 @@
NES_CM_STATE_ACCEPTING,
NES_CM_STATE_MPAREQ_SENT,
NES_CM_STATE_MPAREQ_RCVD,
+ NES_CM_STATE_MPAREJ_RCVD,
NES_CM_STATE_TSA,
NES_CM_STATE_FIN_WAIT1,
NES_CM_STATE_FIN_WAIT2,
@@ -278,13 +282,12 @@
struct nes_timer_entry *send_entry;
spinlock_t retrans_list_lock;
- struct list_head recv_list;
- spinlock_t recv_list_lock;
+ struct nes_timer_entry *recv_entry;
int send_write0;
union {
struct ietf_mpa_frame mpa_frame;
- u8 mpa_frame_buf[NES_CM_DEFAULT_MTU];
+ u8 mpa_frame_buf[MAX_CM_BUFFER];
};
u16 mpa_frame_size;
struct iw_cm_id *cm_id;
@@ -326,6 +329,7 @@
NES_CM_EVENT_MPA_REQ,
NES_CM_EVENT_MPA_CONNECT,
NES_CM_EVENT_MPA_ACCEPT,
+ NES_CM_EVENT_MPA_REJECT,
NES_CM_EVENT_MPA_ESTABLISHED,
NES_CM_EVENT_CONNECTED,
NES_CM_EVENT_CLOSED,
diff --git a/drivers/infiniband/hw/nes/nes_context.h b/drivers/infiniband/hw/nes/nes_context.h
index da9daba..0fb8d81 100644
--- a/drivers/infiniband/hw/nes/nes_context.h
+++ b/drivers/infiniband/hw/nes/nes_context.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c
index 5d139db..9a51f25 100644
--- a/drivers/infiniband/hw/nes/nes_hw.c
+++ b/drivers/infiniband/hw/nes/nes_hw.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -254,6 +254,7 @@
u32 adapter_size;
u32 arp_table_size;
u16 vendor_id;
+ u16 device_id;
u8 OneG_Mode;
u8 func_index;
@@ -356,6 +357,13 @@
return NULL;
}
+ nesadapter->vendor_id = (((u32) nesadapter->mac_addr_high) << 8) |
+ (nesadapter->mac_addr_low >> 24);
+
+ pci_bus_read_config_word(nesdev->pcidev->bus, nesdev->pcidev->devfn,
+ PCI_DEVICE_ID, &device_id);
+ nesadapter->vendor_part_id = device_id;
+
if (nes_init_serdes(nesdev, hw_rev, port_count, nesadapter,
OneG_Mode)) {
kfree(nesadapter);
@@ -1636,7 +1644,6 @@
nesvnic->post_cqp_request = nes_post_cqp_request;
nesvnic->mcrq_mcast_filter = NULL;
- spin_lock_init(&nesvnic->nic.sq_lock);
spin_lock_init(&nesvnic->nic.rq_lock);
/* setup the RQ */
@@ -2261,6 +2268,8 @@
if (++head >= aeq_size)
head = 0;
+
+ nes_write32(nesdev->regs + NES_AEQ_ALLOC, 1 << 16);
}
while (1);
aeq->aeq_head = head;
@@ -2622,9 +2631,9 @@
} else
break;
}
- if (skb)
- dev_kfree_skb_any(skb);
}
+ if (skb)
+ dev_kfree_skb_any(skb);
nesnic->sq_tail++;
nesnic->sq_tail &= nesnic->sq_size-1;
if (sq_cqes > 128) {
diff --git a/drivers/infiniband/hw/nes/nes_hw.h b/drivers/infiniband/hw/nes/nes_hw.h
index bc0b4de..f41a871 100644
--- a/drivers/infiniband/hw/nes/nes_hw.h
+++ b/drivers/infiniband/hw/nes/nes_hw.h
@@ -1,5 +1,5 @@
/*
-* Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+* Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -61,6 +61,7 @@
NES_CQ_ACK = 0x0034,
NES_WQE_ALLOC = 0x0040,
NES_CQE_ALLOC = 0x0044,
+ NES_AEQ_ALLOC = 0x0048
};
enum indexed_regs {
@@ -875,7 +876,6 @@
u8 replenishing_rq;
u8 reserved;
- spinlock_t sq_lock;
spinlock_t rq_lock;
};
@@ -1147,7 +1147,6 @@
struct nes_vnic {
struct nes_ib_device *nesibdev;
u64 sq_full;
- u64 sq_locked;
u64 tso_requests;
u64 segmented_tso_requests;
u64 linearized_skbs;
diff --git a/drivers/infiniband/hw/nes/nes_nic.c b/drivers/infiniband/hw/nes/nes_nic.c
index 57a47cf..025ed9f 100644
--- a/drivers/infiniband/hw/nes/nes_nic.c
+++ b/drivers/infiniband/hw/nes/nes_nic.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -400,8 +400,7 @@
if (skb_headlen(skb) == skb->len) {
if (skb_headlen(skb) <= NES_FIRST_FRAG_SIZE) {
nic_sqe->wqe_words[NES_NIC_SQ_WQE_LENGTH_2_1_IDX] = 0;
- nesnic->tx_skb[nesnic->sq_head] = NULL;
- dev_kfree_skb(skb);
+ nesnic->tx_skb[nesnic->sq_head] = skb;
}
} else {
/* Deal with Fragments */
@@ -453,7 +452,6 @@
u32 wqe_count=1;
u32 send_rc;
struct iphdr *iph;
- unsigned long flags;
__le16 *wqe_fragment_length;
u32 nr_frags;
u32 original_first_length;
@@ -480,13 +478,6 @@
if (netif_queue_stopped(netdev))
return NETDEV_TX_BUSY;
- local_irq_save(flags);
- if (!spin_trylock(&nesnic->sq_lock)) {
- local_irq_restore(flags);
- nesvnic->sq_locked++;
- return NETDEV_TX_LOCKED;
- }
-
/* Check if SQ is full */
if ((((nesnic->sq_tail+(nesnic->sq_size*2))-nesnic->sq_head) & (nesnic->sq_size - 1)) == 1) {
if (!netif_queue_stopped(netdev)) {
@@ -498,7 +489,6 @@
}
}
nesvnic->sq_full++;
- spin_unlock_irqrestore(&nesnic->sq_lock, flags);
return NETDEV_TX_BUSY;
}
@@ -531,7 +521,6 @@
}
}
nesvnic->sq_full++;
- spin_unlock_irqrestore(&nesnic->sq_lock, flags);
nes_debug(NES_DBG_NIC_TX, "%s: HNIC SQ full- TSO request has too many frags!\n",
netdev->name);
return NETDEV_TX_BUSY;
@@ -656,17 +645,13 @@
skb_set_transport_header(skb, hoffset);
skb_set_network_header(skb, nhoffset);
send_rc = nes_nic_send(skb, netdev);
- if (send_rc != NETDEV_TX_OK) {
- spin_unlock_irqrestore(&nesnic->sq_lock, flags);
+ if (send_rc != NETDEV_TX_OK)
return NETDEV_TX_OK;
- }
}
} else {
send_rc = nes_nic_send(skb, netdev);
- if (send_rc != NETDEV_TX_OK) {
- spin_unlock_irqrestore(&nesnic->sq_lock, flags);
+ if (send_rc != NETDEV_TX_OK)
return NETDEV_TX_OK;
- }
}
barrier();
@@ -676,7 +661,6 @@
(wqe_count << 24) | (1 << 23) | nesvnic->nic.qp_id);
netdev->trans_start = jiffies;
- spin_unlock_irqrestore(&nesnic->sq_lock, flags);
return NETDEV_TX_OK;
}
@@ -1012,7 +996,6 @@
"Pause Frames Received",
"Internal Routing Errors",
"SQ SW Dropped SKBs",
- "SQ Locked",
"SQ Full",
"Segmented TSO Requests",
"Rx Symbol Errors",
@@ -1129,16 +1112,17 @@
struct nes_device *nesdev = nesvnic->nesdev;
u32 nic_count;
u32 u32temp;
+ u32 index = 0;
target_ethtool_stats->n_stats = NES_ETHTOOL_STAT_COUNT;
- target_stat_values[0] = nesvnic->nesdev->link_status_interrupts;
- target_stat_values[1] = nesvnic->linearized_skbs;
- target_stat_values[2] = nesvnic->tso_requests;
+ target_stat_values[index] = nesvnic->nesdev->link_status_interrupts;
+ target_stat_values[++index] = nesvnic->linearized_skbs;
+ target_stat_values[++index] = nesvnic->tso_requests;
u32temp = nes_read_indexed(nesdev,
NES_IDX_MAC_TX_PAUSE_FRAMES + (nesvnic->nesdev->mac_index*0x200));
nesvnic->nesdev->mac_pause_frames_sent += u32temp;
- target_stat_values[3] = nesvnic->nesdev->mac_pause_frames_sent;
+ target_stat_values[++index] = nesvnic->nesdev->mac_pause_frames_sent;
u32temp = nes_read_indexed(nesdev,
NES_IDX_MAC_RX_PAUSE_FRAMES + (nesvnic->nesdev->mac_index*0x200));
@@ -1209,60 +1193,59 @@
nesvnic->endnode_ipv4_tcp_retransmits += u32temp;
}
- target_stat_values[4] = nesvnic->nesdev->mac_pause_frames_received;
- target_stat_values[5] = nesdev->nesadapter->nic_rx_eth_route_err;
- target_stat_values[6] = nesvnic->tx_sw_dropped;
- target_stat_values[7] = nesvnic->sq_locked;
- target_stat_values[8] = nesvnic->sq_full;
- target_stat_values[9] = nesvnic->segmented_tso_requests;
- target_stat_values[10] = nesvnic->nesdev->mac_rx_symbol_err_frames;
- target_stat_values[11] = nesvnic->nesdev->mac_rx_jabber_frames;
- target_stat_values[12] = nesvnic->nesdev->mac_rx_oversized_frames;
- target_stat_values[13] = nesvnic->nesdev->mac_rx_short_frames;
- target_stat_values[14] = nesvnic->endnode_nstat_rx_discard;
- target_stat_values[15] = nesvnic->endnode_nstat_rx_octets;
- target_stat_values[16] = nesvnic->endnode_nstat_rx_frames;
- target_stat_values[17] = nesvnic->endnode_nstat_tx_octets;
- target_stat_values[18] = nesvnic->endnode_nstat_tx_frames;
- target_stat_values[19] = mh_detected;
- target_stat_values[20] = mh_pauses_sent;
- target_stat_values[21] = nesvnic->endnode_ipv4_tcp_retransmits;
- target_stat_values[22] = atomic_read(&cm_connects);
- target_stat_values[23] = atomic_read(&cm_accepts);
- target_stat_values[24] = atomic_read(&cm_disconnects);
- target_stat_values[25] = atomic_read(&cm_connecteds);
- target_stat_values[26] = atomic_read(&cm_connect_reqs);
- target_stat_values[27] = atomic_read(&cm_rejects);
- target_stat_values[28] = atomic_read(&mod_qp_timouts);
- target_stat_values[29] = atomic_read(&qps_created);
- target_stat_values[30] = atomic_read(&sw_qps_destroyed);
- target_stat_values[31] = atomic_read(&qps_destroyed);
- target_stat_values[32] = atomic_read(&cm_closes);
- target_stat_values[33] = cm_packets_sent;
- target_stat_values[34] = cm_packets_bounced;
- target_stat_values[35] = cm_packets_created;
- target_stat_values[36] = cm_packets_received;
- target_stat_values[37] = cm_packets_dropped;
- target_stat_values[38] = cm_packets_retrans;
- target_stat_values[39] = cm_listens_created;
- target_stat_values[40] = cm_listens_destroyed;
- target_stat_values[41] = cm_backlog_drops;
- target_stat_values[42] = atomic_read(&cm_loopbacks);
- target_stat_values[43] = atomic_read(&cm_nodes_created);
- target_stat_values[44] = atomic_read(&cm_nodes_destroyed);
- target_stat_values[45] = atomic_read(&cm_accel_dropped_pkts);
- target_stat_values[46] = atomic_read(&cm_resets_recvd);
- target_stat_values[47] = int_mod_timer_init;
- target_stat_values[48] = int_mod_cq_depth_1;
- target_stat_values[49] = int_mod_cq_depth_4;
- target_stat_values[50] = int_mod_cq_depth_16;
- target_stat_values[51] = int_mod_cq_depth_24;
- target_stat_values[52] = int_mod_cq_depth_32;
- target_stat_values[53] = int_mod_cq_depth_128;
- target_stat_values[54] = int_mod_cq_depth_256;
- target_stat_values[55] = nesvnic->lro_mgr.stats.aggregated;
- target_stat_values[56] = nesvnic->lro_mgr.stats.flushed;
- target_stat_values[57] = nesvnic->lro_mgr.stats.no_desc;
+ target_stat_values[++index] = nesvnic->nesdev->mac_pause_frames_received;
+ target_stat_values[++index] = nesdev->nesadapter->nic_rx_eth_route_err;
+ target_stat_values[++index] = nesvnic->tx_sw_dropped;
+ target_stat_values[++index] = nesvnic->sq_full;
+ target_stat_values[++index] = nesvnic->segmented_tso_requests;
+ target_stat_values[++index] = nesvnic->nesdev->mac_rx_symbol_err_frames;
+ target_stat_values[++index] = nesvnic->nesdev->mac_rx_jabber_frames;
+ target_stat_values[++index] = nesvnic->nesdev->mac_rx_oversized_frames;
+ target_stat_values[++index] = nesvnic->nesdev->mac_rx_short_frames;
+ target_stat_values[++index] = nesvnic->endnode_nstat_rx_discard;
+ target_stat_values[++index] = nesvnic->endnode_nstat_rx_octets;
+ target_stat_values[++index] = nesvnic->endnode_nstat_rx_frames;
+ target_stat_values[++index] = nesvnic->endnode_nstat_tx_octets;
+ target_stat_values[++index] = nesvnic->endnode_nstat_tx_frames;
+ target_stat_values[++index] = mh_detected;
+ target_stat_values[++index] = mh_pauses_sent;
+ target_stat_values[++index] = nesvnic->endnode_ipv4_tcp_retransmits;
+ target_stat_values[++index] = atomic_read(&cm_connects);
+ target_stat_values[++index] = atomic_read(&cm_accepts);
+ target_stat_values[++index] = atomic_read(&cm_disconnects);
+ target_stat_values[++index] = atomic_read(&cm_connecteds);
+ target_stat_values[++index] = atomic_read(&cm_connect_reqs);
+ target_stat_values[++index] = atomic_read(&cm_rejects);
+ target_stat_values[++index] = atomic_read(&mod_qp_timouts);
+ target_stat_values[++index] = atomic_read(&qps_created);
+ target_stat_values[++index] = atomic_read(&sw_qps_destroyed);
+ target_stat_values[++index] = atomic_read(&qps_destroyed);
+ target_stat_values[++index] = atomic_read(&cm_closes);
+ target_stat_values[++index] = cm_packets_sent;
+ target_stat_values[++index] = cm_packets_bounced;
+ target_stat_values[++index] = cm_packets_created;
+ target_stat_values[++index] = cm_packets_received;
+ target_stat_values[++index] = cm_packets_dropped;
+ target_stat_values[++index] = cm_packets_retrans;
+ target_stat_values[++index] = cm_listens_created;
+ target_stat_values[++index] = cm_listens_destroyed;
+ target_stat_values[++index] = cm_backlog_drops;
+ target_stat_values[++index] = atomic_read(&cm_loopbacks);
+ target_stat_values[++index] = atomic_read(&cm_nodes_created);
+ target_stat_values[++index] = atomic_read(&cm_nodes_destroyed);
+ target_stat_values[++index] = atomic_read(&cm_accel_dropped_pkts);
+ target_stat_values[++index] = atomic_read(&cm_resets_recvd);
+ target_stat_values[++index] = int_mod_timer_init;
+ target_stat_values[++index] = int_mod_cq_depth_1;
+ target_stat_values[++index] = int_mod_cq_depth_4;
+ target_stat_values[++index] = int_mod_cq_depth_16;
+ target_stat_values[++index] = int_mod_cq_depth_24;
+ target_stat_values[++index] = int_mod_cq_depth_32;
+ target_stat_values[++index] = int_mod_cq_depth_128;
+ target_stat_values[++index] = int_mod_cq_depth_256;
+ target_stat_values[++index] = nesvnic->lro_mgr.stats.aggregated;
+ target_stat_values[++index] = nesvnic->lro_mgr.stats.flushed;
+ target_stat_values[++index] = nesvnic->lro_mgr.stats.no_desc;
}
@@ -1616,7 +1599,6 @@
nes_debug(NES_DBG_INIT, "Enabling VLAN Insert/Delete.\n");
netdev->features |= NETIF_F_HW_VLAN_TX | NETIF_F_HW_VLAN_RX;
netdev->vlan_rx_register = nes_netdev_vlan_rx_register;
- netdev->features |= NETIF_F_LLTX;
/* Fill in the port structure */
nesvnic->netdev = netdev;
diff --git a/drivers/infiniband/hw/nes/nes_user.h b/drivers/infiniband/hw/nes/nes_user.h
index e64306b..cc90c14 100644
--- a/drivers/infiniband/hw/nes/nes_user.h
+++ b/drivers/infiniband/hw/nes/nes_user.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
* Copyright (c) 2005 Topspin Communications. All rights reserved.
* Copyright (c) 2005 Cisco Systems. All rights reserved.
* Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
diff --git a/drivers/infiniband/hw/nes/nes_utils.c b/drivers/infiniband/hw/nes/nes_utils.c
index 6f3bc1b..a282031 100644
--- a/drivers/infiniband/hw/nes/nes_utils.c
+++ b/drivers/infiniband/hw/nes/nes_utils.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 4fdb7245..96d9535 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
@@ -551,6 +551,7 @@
struct nes_device *nesdev = nesvnic->nesdev;
struct nes_adapter *nesadapter = nesdev->nesadapter;
int i = 0;
+ int rc;
/* free the resources */
if (nesfmr->leaf_pbl_cnt == 0) {
@@ -572,7 +573,9 @@
nesmr->ibmw.rkey = ibfmr->rkey;
nesmr->ibmw.uobject = NULL;
- if (nesfmr->nesmr.pbls_used != 0) {
+ rc = nes_dealloc_mw(&nesmr->ibmw);
+
+ if ((rc == 0) && (nesfmr->nesmr.pbls_used != 0)) {
spin_lock_irqsave(&nesadapter->pbl_lock, flags);
if (nesfmr->nesmr.pbl_4k) {
nesadapter->free_4kpbl += nesfmr->nesmr.pbls_used;
@@ -584,7 +587,7 @@
spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
}
- return nes_dealloc_mw(&nesmr->ibmw);
+ return rc;
}
@@ -1884,21 +1887,75 @@
return ret;
}
+/**
+ * root_256
+ */
+static u32 root_256(struct nes_device *nesdev,
+ struct nes_root_vpbl *root_vpbl,
+ struct nes_root_vpbl *new_root,
+ u16 pbl_count_4k,
+ u16 pbl_count_256)
+{
+ u64 leaf_pbl;
+ int i, j, k;
+
+ if (pbl_count_4k == 1) {
+ new_root->pbl_vbase = pci_alloc_consistent(nesdev->pcidev,
+ 512, &new_root->pbl_pbase);
+
+ if (new_root->pbl_vbase == NULL)
+ return 0;
+
+ leaf_pbl = (u64)root_vpbl->pbl_pbase;
+ for (i = 0; i < 16; i++) {
+ new_root->pbl_vbase[i].pa_low =
+ cpu_to_le32((u32)leaf_pbl);
+ new_root->pbl_vbase[i].pa_high =
+ cpu_to_le32((u32)((((u64)leaf_pbl) >> 32)));
+ leaf_pbl += 256;
+ }
+ } else {
+ for (i = 3; i >= 0; i--) {
+ j = i * 16;
+ root_vpbl->pbl_vbase[j] = root_vpbl->pbl_vbase[i];
+ leaf_pbl = le32_to_cpu(root_vpbl->pbl_vbase[j].pa_low) +
+ (((u64)le32_to_cpu(root_vpbl->pbl_vbase[j].pa_high))
+ << 32);
+ for (k = 1; k < 16; k++) {
+ leaf_pbl += 256;
+ root_vpbl->pbl_vbase[j + k].pa_low =
+ cpu_to_le32((u32)leaf_pbl);
+ root_vpbl->pbl_vbase[j + k].pa_high =
+ cpu_to_le32((u32)((((u64)leaf_pbl) >> 32)));
+ }
+ }
+ }
+
+ return 1;
+}
+
/**
* nes_reg_mr
*/
static int nes_reg_mr(struct nes_device *nesdev, struct nes_pd *nespd,
u32 stag, u64 region_length, struct nes_root_vpbl *root_vpbl,
- dma_addr_t single_buffer, u16 pbl_count, u16 residual_page_count,
- int acc, u64 *iova_start)
+ dma_addr_t single_buffer, u16 pbl_count_4k,
+ u16 residual_page_count_4k, int acc, u64 *iova_start,
+ u16 *actual_pbl_cnt, u8 *used_4k_pbls)
{
struct nes_hw_cqp_wqe *cqp_wqe;
struct nes_cqp_request *cqp_request;
unsigned long flags;
int ret;
struct nes_adapter *nesadapter = nesdev->nesadapter;
- /* int count; */
+ uint pg_cnt = 0;
+ u16 pbl_count_256;
+ u16 pbl_count = 0;
+ u8 use_256_pbls = 0;
+ u8 use_4k_pbls = 0;
+ u16 use_two_level = (pbl_count_4k > 1) ? 1 : 0;
+ struct nes_root_vpbl new_root = {0, 0, 0};
u32 opcode = 0;
u16 major_code;
@@ -1911,41 +1968,70 @@
cqp_request->waiting = 1;
cqp_wqe = &cqp_request->cqp_wqe;
- spin_lock_irqsave(&nesadapter->pbl_lock, flags);
- /* track PBL resources */
- if (pbl_count != 0) {
- if (pbl_count > 1) {
- /* Two level PBL */
- if ((pbl_count+1) > nesadapter->free_4kpbl) {
- nes_debug(NES_DBG_MR, "Out of 4KB Pbls for two level request.\n");
- spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
- nes_free_cqp_request(nesdev, cqp_request);
- return -ENOMEM;
- } else {
- nesadapter->free_4kpbl -= pbl_count+1;
- }
- } else if (residual_page_count > 32) {
- if (pbl_count > nesadapter->free_4kpbl) {
- nes_debug(NES_DBG_MR, "Out of 4KB Pbls.\n");
- spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
- nes_free_cqp_request(nesdev, cqp_request);
- return -ENOMEM;
- } else {
- nesadapter->free_4kpbl -= pbl_count;
+ if (pbl_count_4k) {
+ spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+
+ pg_cnt = ((pbl_count_4k - 1) * 512) + residual_page_count_4k;
+ pbl_count_256 = (pg_cnt + 31) / 32;
+ if (pg_cnt <= 32) {
+ if (pbl_count_256 <= nesadapter->free_256pbl)
+ use_256_pbls = 1;
+ else if (pbl_count_4k <= nesadapter->free_4kpbl)
+ use_4k_pbls = 1;
+ } else if (pg_cnt <= 2048) {
+ if (((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) &&
+ (nesadapter->free_4kpbl > (nesadapter->max_4kpbl >> 1))) {
+ use_4k_pbls = 1;
+ } else if ((pbl_count_256 + 1) <= nesadapter->free_256pbl) {
+ use_256_pbls = 1;
+ use_two_level = 1;
+ } else if ((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) {
+ use_4k_pbls = 1;
}
} else {
- if (pbl_count > nesadapter->free_256pbl) {
- nes_debug(NES_DBG_MR, "Out of 256B Pbls.\n");
- spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
- nes_free_cqp_request(nesdev, cqp_request);
- return -ENOMEM;
- } else {
- nesadapter->free_256pbl -= pbl_count;
- }
+ if ((pbl_count_4k + 1) <= nesadapter->free_4kpbl)
+ use_4k_pbls = 1;
}
+
+ if (use_256_pbls) {
+ pbl_count = pbl_count_256;
+ nesadapter->free_256pbl -= pbl_count + use_two_level;
+ } else if (use_4k_pbls) {
+ pbl_count = pbl_count_4k;
+ nesadapter->free_4kpbl -= pbl_count + use_two_level;
+ } else {
+ spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+ nes_debug(NES_DBG_MR, "Out of Pbls\n");
+ nes_free_cqp_request(nesdev, cqp_request);
+ return -ENOMEM;
+ }
+
+ spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
}
- spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+ if (use_256_pbls && use_two_level) {
+ if (root_256(nesdev, root_vpbl, &new_root, pbl_count_4k, pbl_count_256) == 1) {
+ if (new_root.pbl_pbase != 0)
+ root_vpbl = &new_root;
+ } else {
+ spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+ nesadapter->free_256pbl += pbl_count_256 + use_two_level;
+ use_256_pbls = 0;
+
+ if (pbl_count_4k == 1)
+ use_two_level = 0;
+ pbl_count = pbl_count_4k;
+
+ if ((pbl_count_4k + use_two_level) <= nesadapter->free_4kpbl) {
+ nesadapter->free_4kpbl -= pbl_count + use_two_level;
+ use_4k_pbls = 1;
+ }
+ spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+
+ if (use_4k_pbls == 0)
+ return -ENOMEM;
+ }
+ }
opcode = NES_CQP_REGISTER_STAG | NES_CQP_STAG_RIGHTS_LOCAL_READ |
NES_CQP_STAG_VA_TO | NES_CQP_STAG_MR;
@@ -1974,10 +2060,9 @@
} else {
set_wqe_64bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PA_LOW_IDX, root_vpbl->pbl_pbase);
set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_BLK_COUNT_IDX, pbl_count);
- set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_LEN_IDX,
- (((pbl_count - 1) * 4096) + (residual_page_count*8)));
+ set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_STAG_WQE_PBL_LEN_IDX, (pg_cnt * 8));
- if ((pbl_count > 1) || (residual_page_count > 32))
+ if (use_4k_pbls)
cqp_wqe->wqe_words[NES_CQP_WQE_OPCODE_IDX] |= cpu_to_le32(NES_CQP_STAG_PBL_BLK_SIZE);
}
barrier();
@@ -1994,13 +2079,25 @@
major_code = cqp_request->major_code;
nes_put_cqp_request(nesdev, cqp_request);
+ if ((!ret || major_code) && pbl_count != 0) {
+ spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+ if (use_256_pbls)
+ nesadapter->free_256pbl += pbl_count + use_two_level;
+ else if (use_4k_pbls)
+ nesadapter->free_4kpbl += pbl_count + use_two_level;
+ spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+ }
+ if (new_root.pbl_pbase)
+ pci_free_consistent(nesdev->pcidev, 512, new_root.pbl_vbase,
+ new_root.pbl_pbase);
+
if (!ret)
return -ETIME;
else if (major_code)
return -EIO;
- else
- return 0;
+ *actual_pbl_cnt = pbl_count + use_two_level;
+ *used_4k_pbls = use_4k_pbls;
return 0;
}
@@ -2165,18 +2262,14 @@
pbl_count = root_pbl_index;
}
ret = nes_reg_mr(nesdev, nespd, stag, region_length, &root_vpbl,
- buffer_list[0].addr, pbl_count, (u16)cur_pbl_index, acc, iova_start);
+ buffer_list[0].addr, pbl_count, (u16)cur_pbl_index, acc, iova_start,
+ &nesmr->pbls_used, &nesmr->pbl_4k);
if (ret == 0) {
nesmr->ibmr.rkey = stag;
nesmr->ibmr.lkey = stag;
nesmr->mode = IWNES_MEMREG_TYPE_MEM;
ibmr = &nesmr->ibmr;
- nesmr->pbl_4k = ((pbl_count > 1) || (cur_pbl_index > 32)) ? 1 : 0;
- nesmr->pbls_used = pbl_count;
- if (pbl_count > 1) {
- nesmr->pbls_used++;
- }
} else {
kfree(nesmr);
ibmr = ERR_PTR(-ENOMEM);
@@ -2454,8 +2547,9 @@
stag, (unsigned int)iova_start,
(unsigned int)region_length, stag_index,
(unsigned long long)region->length, pbl_count);
- ret = nes_reg_mr( nesdev, nespd, stag, region->length, &root_vpbl,
- first_dma_addr, pbl_count, (u16)cur_pbl_index, acc, &iova_start);
+ ret = nes_reg_mr(nesdev, nespd, stag, region->length, &root_vpbl,
+ first_dma_addr, pbl_count, (u16)cur_pbl_index, acc,
+ &iova_start, &nesmr->pbls_used, &nesmr->pbl_4k);
nes_debug(NES_DBG_MR, "ret=%d\n", ret);
@@ -2464,11 +2558,6 @@
nesmr->ibmr.lkey = stag;
nesmr->mode = IWNES_MEMREG_TYPE_MEM;
ibmr = &nesmr->ibmr;
- nesmr->pbl_4k = ((pbl_count > 1) || (cur_pbl_index > 32)) ? 1 : 0;
- nesmr->pbls_used = pbl_count;
- if (pbl_count > 1) {
- nesmr->pbls_used++;
- }
} else {
ib_umem_release(region);
kfree(nesmr);
@@ -2607,24 +2696,6 @@
cqp_request->waiting = 1;
cqp_wqe = &cqp_request->cqp_wqe;
- spin_lock_irqsave(&nesadapter->pbl_lock, flags);
- if (nesmr->pbls_used != 0) {
- if (nesmr->pbl_4k) {
- nesadapter->free_4kpbl += nesmr->pbls_used;
- if (nesadapter->free_4kpbl > nesadapter->max_4kpbl) {
- printk(KERN_ERR PFX "free 4KB PBLs(%u) has exceeded the max(%u)\n",
- nesadapter->free_4kpbl, nesadapter->max_4kpbl);
- }
- } else {
- nesadapter->free_256pbl += nesmr->pbls_used;
- if (nesadapter->free_256pbl > nesadapter->max_256pbl) {
- printk(KERN_ERR PFX "free 256B PBLs(%u) has exceeded the max(%u)\n",
- nesadapter->free_256pbl, nesadapter->max_256pbl);
- }
- }
- }
-
- spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
nes_fill_init_cqp_wqe(cqp_wqe, nesdev);
set_wqe_32bit_value(cqp_wqe->wqe_words, NES_CQP_WQE_OPCODE_IDX,
NES_CQP_DEALLOCATE_STAG | NES_CQP_STAG_VA_TO |
@@ -2642,11 +2713,6 @@
" CQP Major:Minor codes = 0x%04X:0x%04X\n",
ib_mr->rkey, ret, cqp_request->major_code, cqp_request->minor_code);
- nes_free_resource(nesadapter, nesadapter->allocated_mrs,
- (ib_mr->rkey & 0x0fffff00) >> 8);
-
- kfree(nesmr);
-
major_code = cqp_request->major_code;
minor_code = cqp_request->minor_code;
@@ -2662,8 +2728,33 @@
" to destroy STag, ib_mr=%p, rkey = 0x%08X\n",
major_code, minor_code, ib_mr, ib_mr->rkey);
return -EIO;
- } else
- return 0;
+ }
+
+ if (nesmr->pbls_used != 0) {
+ spin_lock_irqsave(&nesadapter->pbl_lock, flags);
+ if (nesmr->pbl_4k) {
+ nesadapter->free_4kpbl += nesmr->pbls_used;
+ if (nesadapter->free_4kpbl > nesadapter->max_4kpbl)
+ printk(KERN_ERR PFX "free 4KB PBLs(%u) has "
+ "exceeded the max(%u)\n",
+ nesadapter->free_4kpbl,
+ nesadapter->max_4kpbl);
+ } else {
+ nesadapter->free_256pbl += nesmr->pbls_used;
+ if (nesadapter->free_256pbl > nesadapter->max_256pbl)
+ printk(KERN_ERR PFX "free 256B PBLs(%u) has "
+ "exceeded the max(%u)\n",
+ nesadapter->free_256pbl,
+ nesadapter->max_256pbl);
+ }
+ spin_unlock_irqrestore(&nesadapter->pbl_lock, flags);
+ }
+ nes_free_resource(nesadapter, nesadapter->allocated_mrs,
+ (ib_mr->rkey & 0x0fffff00) >> 8);
+
+ kfree(nesmr);
+
+ return 0;
}
diff --git a/drivers/infiniband/hw/nes/nes_verbs.h b/drivers/infiniband/hw/nes/nes_verbs.h
index 6c6b4da..da3c368 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.h
+++ b/drivers/infiniband/hw/nes/nes_verbs.h
@@ -1,5 +1,5 @@
/*
- * Copyright (c) 2006 - 2008 NetEffect, Inc. All rights reserved.
+ * Copyright (c) 2006 - 2009 Intel-NE, Inc. All rights reserved.
* Copyright (c) 2005 Open Grid Computing, Inc. All rights reserved.
*
* This software is available to you under a choice of one of two
diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c
index 0bd2a4f..353c13b 100644
--- a/drivers/infiniband/ulp/ipoib/ipoib_main.c
+++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c
@@ -660,8 +660,12 @@
path = __path_find(dev, phdr->hwaddr + 4);
if (!path || !path->valid) {
- if (!path)
+ int new_path = 0;
+
+ if (!path) {
path = path_rec_create(dev, phdr->hwaddr + 4);
+ new_path = 1;
+ }
if (path) {
/* put pseudoheader back on for next time */
skb_push(skb, sizeof *phdr);
@@ -669,7 +673,8 @@
if (!path->query && path_rec_start(dev, path)) {
spin_unlock_irqrestore(&priv->lock, flags);
- path_free(dev, path);
+ if (new_path)
+ path_free(dev, path);
return;
} else
__path_add(dev, path);
diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c
index 319b188..ea9e155 100644
--- a/drivers/infiniband/ulp/iser/iser_verbs.c
+++ b/drivers/infiniband/ulp/iser/iser_verbs.c
@@ -401,13 +401,6 @@
if (ret)
goto failure;
- iser_dbg("path.mtu is %d setting it to %d\n",
- cma_id->route.path_rec->mtu, IB_MTU_1024);
-
- /* we must set the MTU to 1024 as this is what the target is assuming */
- if (cma_id->route.path_rec->mtu > IB_MTU_1024)
- cma_id->route.path_rec->mtu = IB_MTU_1024;
-
memset(&conn_param, 0, sizeof conn_param);
conn_param.responder_resources = 4;
conn_param.initiator_depth = 1;
diff --git a/drivers/net/mlx4/Makefile b/drivers/net/mlx4/Makefile
index a7a97bf..21040a0 100644
--- a/drivers/net/mlx4/Makefile
+++ b/drivers/net/mlx4/Makefile
@@ -1,7 +1,7 @@
obj-$(CONFIG_MLX4_CORE) += mlx4_core.o
mlx4_core-y := alloc.o catas.o cmd.o cq.o eq.o fw.o icm.o intf.o main.o mcg.o \
- mr.o pd.o port.o profile.o qp.o reset.o srq.o
+ mr.o pd.o port.o profile.o qp.o reset.o sense.o srq.o
obj-$(CONFIG_MLX4_EN) += mlx4_en.o
diff --git a/drivers/net/mlx4/catas.c b/drivers/net/mlx4/catas.c
index f094ee0..aa9674b 100644
--- a/drivers/net/mlx4/catas.c
+++ b/drivers/net/mlx4/catas.c
@@ -42,7 +42,6 @@
static DEFINE_SPINLOCK(catas_lock);
static LIST_HEAD(catas_list);
-static struct workqueue_struct *catas_wq;
static struct work_struct catas_work;
static int internal_err_reset = 1;
@@ -77,7 +76,7 @@
list_add(&priv->catas_err.list, &catas_list);
spin_unlock(&catas_lock);
- queue_work(catas_wq, &catas_work);
+ queue_work(mlx4_wq, &catas_work);
}
} else
mod_timer(&priv->catas_err.timer,
@@ -146,18 +145,7 @@
spin_unlock_irq(&catas_lock);
}
-int __init mlx4_catas_init(void)
+void __init mlx4_catas_init(void)
{
INIT_WORK(&catas_work, catas_reset);
-
- catas_wq = create_singlethread_workqueue("mlx4_err");
- if (!catas_wq)
- return -ENOMEM;
-
- return 0;
-}
-
-void mlx4_catas_cleanup(void)
-{
- destroy_workqueue(catas_wq);
}
diff --git a/drivers/net/mlx4/eq.c b/drivers/net/mlx4/eq.c
index 2c19bff7c..8830dcb 100644
--- a/drivers/net/mlx4/eq.c
+++ b/drivers/net/mlx4/eq.c
@@ -163,6 +163,7 @@
int cqn;
int eqes_found = 0;
int set_ci = 0;
+ int port;
while ((eqe = next_eqe_sw(eq))) {
/*
@@ -203,11 +204,16 @@
break;
case MLX4_EVENT_TYPE_PORT_CHANGE:
- mlx4_dispatch_event(dev,
- eqe->subtype == MLX4_PORT_CHANGE_SUBTYPE_ACTIVE ?
- MLX4_DEV_EVENT_PORT_UP :
- MLX4_DEV_EVENT_PORT_DOWN,
- be32_to_cpu(eqe->event.port_change.port) >> 28);
+ port = be32_to_cpu(eqe->event.port_change.port) >> 28;
+ if (eqe->subtype == MLX4_PORT_CHANGE_SUBTYPE_DOWN) {
+ mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_DOWN,
+ port);
+ mlx4_priv(dev)->sense.do_sense_port[port] = 1;
+ } else {
+ mlx4_dispatch_event(dev, MLX4_DEV_EVENT_PORT_UP,
+ port);
+ mlx4_priv(dev)->sense.do_sense_port[port] = 0;
+ }
break;
case MLX4_EVENT_TYPE_CQ_ERROR:
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index 6ef2490..a66f5b2 100644
--- a/drivers/net/mlx4/main.c
+++ b/drivers/net/mlx4/main.c
@@ -51,6 +51,8 @@
MODULE_LICENSE("Dual BSD/GPL");
MODULE_VERSION(DRV_VERSION);
+struct workqueue_struct *mlx4_wq;
+
#ifdef CONFIG_MLX4_DEBUG
int mlx4_debug_level = 0;
@@ -98,24 +100,23 @@
MODULE_PARM_DESC(use_prio, "Enable steering by VLAN priority on ETH ports "
"(0/1, default 0)");
-static int mlx4_check_port_params(struct mlx4_dev *dev,
- enum mlx4_port_type *port_type)
+int mlx4_check_port_params(struct mlx4_dev *dev,
+ enum mlx4_port_type *port_type)
{
int i;
for (i = 0; i < dev->caps.num_ports - 1; i++) {
- if (port_type[i] != port_type[i+1] &&
- !(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
- mlx4_err(dev, "Only same port types supported "
- "on this HCA, aborting.\n");
- return -EINVAL;
+ if (port_type[i] != port_type[i + 1]) {
+ if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
+ mlx4_err(dev, "Only same port types supported "
+ "on this HCA, aborting.\n");
+ return -EINVAL;
+ }
+ if (port_type[i] == MLX4_PORT_TYPE_ETH &&
+ port_type[i + 1] == MLX4_PORT_TYPE_IB)
+ return -EINVAL;
}
}
- if ((port_type[0] == MLX4_PORT_TYPE_ETH) &&
- (port_type[1] == MLX4_PORT_TYPE_IB)) {
- mlx4_err(dev, "eth-ib configuration is not supported.\n");
- return -EINVAL;
- }
for (i = 0; i < dev->caps.num_ports; i++) {
if (!(port_type[i] & dev->caps.supported_type[i+1])) {
@@ -225,6 +226,9 @@
dev->caps.port_type[i] = MLX4_PORT_TYPE_IB;
else
dev->caps.port_type[i] = MLX4_PORT_TYPE_ETH;
+ dev->caps.possible_type[i] = dev->caps.port_type[i];
+ mlx4_priv(dev)->sense.sense_allowed[i] =
+ dev->caps.supported_type[i] == MLX4_PORT_TYPE_AUTO;
if (dev->caps.log_num_macs > dev_cap->log_max_macs[i]) {
dev->caps.log_num_macs = dev_cap->log_max_macs[i];
@@ -263,14 +267,16 @@
* Change the port configuration of the device.
* Every user of this function must hold the port mutex.
*/
-static int mlx4_change_port_types(struct mlx4_dev *dev,
- enum mlx4_port_type *port_types)
+int mlx4_change_port_types(struct mlx4_dev *dev,
+ enum mlx4_port_type *port_types)
{
int err = 0;
int change = 0;
int port;
for (port = 0; port < dev->caps.num_ports; port++) {
+ /* Change the port type only if the new type is different
+ * from the current, and not set to Auto */
if (port_types[port] != dev->caps.port_type[port + 1]) {
change = 1;
dev->caps.port_type[port + 1] = port_types[port];
@@ -302,10 +308,17 @@
struct mlx4_port_info *info = container_of(attr, struct mlx4_port_info,
port_attr);
struct mlx4_dev *mdev = info->dev;
+ char type[8];
- return sprintf(buf, "%s\n",
- mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB ?
- "ib" : "eth");
+ sprintf(type, "%s",
+ (mdev->caps.port_type[info->port] == MLX4_PORT_TYPE_IB) ?
+ "ib" : "eth");
+ if (mdev->caps.possible_type[info->port] == MLX4_PORT_TYPE_AUTO)
+ sprintf(buf, "auto (%s)\n", type);
+ else
+ sprintf(buf, "%s\n", type);
+
+ return strlen(buf);
}
static ssize_t set_port_type(struct device *dev,
@@ -317,6 +330,7 @@
struct mlx4_dev *mdev = info->dev;
struct mlx4_priv *priv = mlx4_priv(mdev);
enum mlx4_port_type types[MLX4_MAX_PORTS];
+ enum mlx4_port_type new_types[MLX4_MAX_PORTS];
int i;
int err = 0;
@@ -324,26 +338,56 @@
info->tmp_type = MLX4_PORT_TYPE_IB;
else if (!strcmp(buf, "eth\n"))
info->tmp_type = MLX4_PORT_TYPE_ETH;
+ else if (!strcmp(buf, "auto\n"))
+ info->tmp_type = MLX4_PORT_TYPE_AUTO;
else {
mlx4_err(mdev, "%s is not supported port type\n", buf);
return -EINVAL;
}
+ mlx4_stop_sense(mdev);
mutex_lock(&priv->port_mutex);
- for (i = 0; i < mdev->caps.num_ports; i++)
- types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
- mdev->caps.port_type[i+1];
+ /* Possible type is always the one that was delivered */
+ mdev->caps.possible_type[info->port] = info->tmp_type;
- err = mlx4_check_port_params(mdev, types);
+ for (i = 0; i < mdev->caps.num_ports; i++) {
+ types[i] = priv->port[i+1].tmp_type ? priv->port[i+1].tmp_type :
+ mdev->caps.possible_type[i+1];
+ if (types[i] == MLX4_PORT_TYPE_AUTO)
+ types[i] = mdev->caps.port_type[i+1];
+ }
+
+ if (!(mdev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP)) {
+ for (i = 1; i <= mdev->caps.num_ports; i++) {
+ if (mdev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) {
+ mdev->caps.possible_type[i] = mdev->caps.port_type[i];
+ err = -EINVAL;
+ }
+ }
+ }
+ if (err) {
+ mlx4_err(mdev, "Auto sensing is not supported on this HCA. "
+ "Set only 'eth' or 'ib' for both ports "
+ "(should be the same)\n");
+ goto out;
+ }
+
+ mlx4_do_sense_ports(mdev, new_types, types);
+
+ err = mlx4_check_port_params(mdev, new_types);
if (err)
goto out;
- for (i = 1; i <= mdev->caps.num_ports; i++)
- priv->port[i].tmp_type = 0;
+ /* We are about to apply the changes after the configuration
+ * was verified, no need to remember the temporary types
+ * any more */
+ for (i = 0; i < mdev->caps.num_ports; i++)
+ priv->port[i + 1].tmp_type = 0;
- err = mlx4_change_port_types(mdev, types);
+ err = mlx4_change_port_types(mdev, new_types);
out:
+ mlx4_start_sense(mdev);
mutex_unlock(&priv->port_mutex);
return err ? err : count;
}
@@ -1117,6 +1161,9 @@
if (err)
goto err_port;
+ mlx4_sense_init(dev);
+ mlx4_start_sense(dev);
+
pci_set_drvdata(pdev, dev);
return 0;
@@ -1182,6 +1229,7 @@
int p;
if (dev) {
+ mlx4_stop_sense(dev);
mlx4_unregister_device(dev);
for (p = 1; p <= dev->caps.num_ports; p++) {
@@ -1230,6 +1278,8 @@
{ PCI_VDEVICE(MELLANOX, 0x673c) }, /* MT25408 "Hermon" QDR PCIe gen2 */
{ PCI_VDEVICE(MELLANOX, 0x6368) }, /* MT25408 "Hermon" EN 10GigE */
{ PCI_VDEVICE(MELLANOX, 0x6750) }, /* MT25408 "Hermon" EN 10GigE PCIe gen2 */
+ { PCI_VDEVICE(MELLANOX, 0x6372) }, /* MT25458 ConnectX EN 10GBASE-T 10GigE */
+ { PCI_VDEVICE(MELLANOX, 0x675a) }, /* MT25458 ConnectX EN 10GBASE-T+Gen2 10GigE */
{ 0, }
};
@@ -1264,9 +1314,11 @@
if (mlx4_verify_params())
return -EINVAL;
- ret = mlx4_catas_init();
- if (ret)
- return ret;
+ mlx4_catas_init();
+
+ mlx4_wq = create_singlethread_workqueue("mlx4");
+ if (!mlx4_wq)
+ return -ENOMEM;
ret = pci_register_driver(&mlx4_driver);
return ret < 0 ? ret : 0;
@@ -1275,7 +1327,7 @@
static void __exit mlx4_cleanup(void)
{
pci_unregister_driver(&mlx4_driver);
- mlx4_catas_cleanup();
+ destroy_workqueue(mlx4_wq);
}
module_init(mlx4_init);
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
index e0213bad..5bd79c2 100644
--- a/drivers/net/mlx4/mlx4.h
+++ b/drivers/net/mlx4/mlx4.h
@@ -40,6 +40,7 @@
#include <linux/mutex.h>
#include <linux/radix-tree.h>
#include <linux/timer.h>
+#include <linux/workqueue.h>
#include <linux/mlx4/device.h>
#include <linux/mlx4/driver.h>
@@ -276,6 +277,13 @@
struct mlx4_vlan_table vlan_table;
};
+struct mlx4_sense {
+ struct mlx4_dev *dev;
+ u8 do_sense_port[MLX4_MAX_PORTS + 1];
+ u8 sense_allowed[MLX4_MAX_PORTS + 1];
+ struct delayed_work sense_poll;
+};
+
struct mlx4_priv {
struct mlx4_dev dev;
@@ -305,6 +313,7 @@
struct mlx4_uar driver_uar;
void __iomem *kar;
struct mlx4_port_info port[MLX4_MAX_PORTS + 1];
+ struct mlx4_sense sense;
struct mutex port_mutex;
};
@@ -313,6 +322,10 @@
return container_of(dev, struct mlx4_priv, dev);
}
+#define MLX4_SENSE_RANGE (HZ * 3)
+
+extern struct workqueue_struct *mlx4_wq;
+
u32 mlx4_bitmap_alloc(struct mlx4_bitmap *bitmap);
void mlx4_bitmap_free(struct mlx4_bitmap *bitmap, u32 obj);
u32 mlx4_bitmap_alloc_range(struct mlx4_bitmap *bitmap, int cnt, int align);
@@ -346,8 +359,7 @@
void mlx4_start_catas_poll(struct mlx4_dev *dev);
void mlx4_stop_catas_poll(struct mlx4_dev *dev);
-int mlx4_catas_init(void);
-void mlx4_catas_cleanup(void);
+void mlx4_catas_init(void);
int mlx4_restart_one(struct pci_dev *pdev);
int mlx4_register_device(struct mlx4_dev *dev);
void mlx4_unregister_device(struct mlx4_dev *dev);
@@ -379,6 +391,17 @@
void mlx4_handle_catas_err(struct mlx4_dev *dev);
+void mlx4_do_sense_ports(struct mlx4_dev *dev,
+ enum mlx4_port_type *stype,
+ enum mlx4_port_type *defaults);
+void mlx4_start_sense(struct mlx4_dev *dev);
+void mlx4_stop_sense(struct mlx4_dev *dev);
+void mlx4_sense_init(struct mlx4_dev *dev);
+int mlx4_check_port_params(struct mlx4_dev *dev,
+ enum mlx4_port_type *port_type);
+int mlx4_change_port_types(struct mlx4_dev *dev,
+ enum mlx4_port_type *port_types);
+
void mlx4_init_mac_table(struct mlx4_dev *dev, struct mlx4_mac_table *table);
void mlx4_init_vlan_table(struct mlx4_dev *dev, struct mlx4_vlan_table *table);
diff --git a/drivers/net/mlx4/port.c b/drivers/net/mlx4/port.c
index 0a057e5..7cce334 100644
--- a/drivers/net/mlx4/port.c
+++ b/drivers/net/mlx4/port.c
@@ -298,20 +298,17 @@
{
struct mlx4_cmd_mailbox *mailbox;
int err;
- u8 is_eth = dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH;
mailbox = mlx4_alloc_cmd_mailbox(dev);
if (IS_ERR(mailbox))
return PTR_ERR(mailbox);
memset(mailbox->buf, 0, 256);
- if (is_eth) {
- ((u8 *) mailbox->buf)[3] = 6;
- ((__be16 *) mailbox->buf)[4] = cpu_to_be16(1 << 15);
- ((__be16 *) mailbox->buf)[6] = cpu_to_be16(1 << 15);
- } else
- ((__be32 *) mailbox->buf)[1] = dev->caps.ib_port_def_cap[port];
- err = mlx4_cmd(dev, mailbox->dma, port, is_eth, MLX4_CMD_SET_PORT,
+ if (dev->caps.port_type[port] == MLX4_PORT_TYPE_ETH)
+ return 0;
+
+ ((__be32 *) mailbox->buf)[1] = dev->caps.ib_port_def_cap[port];
+ err = mlx4_cmd(dev, mailbox->dma, port, 0, MLX4_CMD_SET_PORT,
MLX4_CMD_TIME_CLASS_B);
mlx4_free_cmd_mailbox(dev, mailbox);
diff --git a/drivers/net/mlx4/sense.c b/drivers/net/mlx4/sense.c
new file mode 100644
index 0000000..6d5089e
--- /dev/null
+++ b/drivers/net/mlx4/sense.c
@@ -0,0 +1,156 @@
+/*
+ * Copyright (c) 2007 Mellanox Technologies. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses. You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the
+ * OpenIB.org BSD license below:
+ *
+ * Redistribution and use in source and binary forms, with or
+ * without modification, are permitted provided that the following
+ * conditions are met:
+ *
+ * - Redistributions of source code must retain the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer.
+ *
+ * - Redistributions in binary form must reproduce the above
+ * copyright notice, this list of conditions and the following
+ * disclaimer in the documentation and/or other materials
+ * provided with the distribution.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
+ * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
+ * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
+ * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+ * SOFTWARE.
+ *
+ */
+
+#include <linux/errno.h>
+#include <linux/if_ether.h>
+
+#include <linux/mlx4/cmd.h>
+
+#include "mlx4.h"
+
+static int mlx4_SENSE_PORT(struct mlx4_dev *dev, int port,
+ enum mlx4_port_type *type)
+{
+ u64 out_param;
+ int err = 0;
+
+ err = mlx4_cmd_imm(dev, 0, &out_param, port, 0,
+ MLX4_CMD_SENSE_PORT, MLX4_CMD_TIME_CLASS_B);
+ if (err) {
+ mlx4_err(dev, "Sense command failed for port: %d\n", port);
+ return err;
+ }
+
+ if (out_param > 2) {
+ mlx4_err(dev, "Sense returned illegal value: 0x%llx\n", out_param);
+ return EINVAL;
+ }
+
+ *type = out_param;
+ return 0;
+}
+
+void mlx4_do_sense_ports(struct mlx4_dev *dev,
+ enum mlx4_port_type *stype,
+ enum mlx4_port_type *defaults)
+{
+ struct mlx4_sense *sense = &mlx4_priv(dev)->sense;
+ int err;
+ int i;
+
+ for (i = 1; i <= dev->caps.num_ports; i++) {
+ stype[i - 1] = 0;
+ if (sense->do_sense_port[i] && sense->sense_allowed[i] &&
+ dev->caps.possible_type[i] == MLX4_PORT_TYPE_AUTO) {
+ err = mlx4_SENSE_PORT(dev, i, &stype[i - 1]);
+ if (err)
+ stype[i - 1] = defaults[i - 1];
+ } else
+ stype[i - 1] = defaults[i - 1];
+ }
+
+ /*
+ * Adjust port configuration:
+ * If port 1 sensed nothing and port 2 is IB, set both as IB
+ * If port 2 sensed nothing and port 1 is Eth, set both as Eth
+ */
+ if (stype[0] == MLX4_PORT_TYPE_ETH) {
+ for (i = 1; i < dev->caps.num_ports; i++)
+ stype[i] = stype[i] ? stype[i] : MLX4_PORT_TYPE_ETH;
+ }
+ if (stype[dev->caps.num_ports - 1] == MLX4_PORT_TYPE_IB) {
+ for (i = 0; i < dev->caps.num_ports - 1; i++)
+ stype[i] = stype[i] ? stype[i] : MLX4_PORT_TYPE_IB;
+ }
+
+ /*
+ * If sensed nothing, remain in current configuration.
+ */
+ for (i = 0; i < dev->caps.num_ports; i++)
+ stype[i] = stype[i] ? stype[i] : defaults[i];
+
+}
+
+static void mlx4_sense_port(struct work_struct *work)
+{
+ struct delayed_work *delay = container_of(work, struct delayed_work, work);
+ struct mlx4_sense *sense = container_of(delay, struct mlx4_sense,
+ sense_poll);
+ struct mlx4_dev *dev = sense->dev;
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ enum mlx4_port_type stype[MLX4_MAX_PORTS];
+
+ mutex_lock(&priv->port_mutex);
+ mlx4_do_sense_ports(dev, stype, &dev->caps.port_type[1]);
+
+ if (mlx4_check_port_params(dev, stype))
+ goto sense_again;
+
+ if (mlx4_change_port_types(dev, stype))
+ mlx4_err(dev, "Failed to change port_types\n");
+
+sense_again:
+ mutex_unlock(&priv->port_mutex);
+ queue_delayed_work(mlx4_wq , &sense->sense_poll,
+ round_jiffies_relative(MLX4_SENSE_RANGE));
+}
+
+void mlx4_start_sense(struct mlx4_dev *dev)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_sense *sense = &priv->sense;
+
+ if (!(dev->caps.flags & MLX4_DEV_CAP_FLAG_DPDP))
+ return;
+
+ queue_delayed_work(mlx4_wq , &sense->sense_poll,
+ round_jiffies_relative(MLX4_SENSE_RANGE));
+}
+
+void mlx4_stop_sense(struct mlx4_dev *dev)
+{
+ cancel_delayed_work_sync(&mlx4_priv(dev)->sense.sense_poll);
+}
+
+void mlx4_sense_init(struct mlx4_dev *dev)
+{
+ struct mlx4_priv *priv = mlx4_priv(dev);
+ struct mlx4_sense *sense = &priv->sense;
+ int port;
+
+ sense->dev = dev;
+ for (port = 1; port <= dev->caps.num_ports; port++)
+ sense->do_sense_port[port] = 1;
+
+ INIT_DELAYED_WORK_DEFERRABLE(&sense->sense_poll, mlx4_sense_port);
+}
diff --git a/include/linux/mlx4/cmd.h b/include/linux/mlx4/cmd.h
index cf9c679..0f82293 100644
--- a/include/linux/mlx4/cmd.h
+++ b/include/linux/mlx4/cmd.h
@@ -55,6 +55,7 @@
MLX4_CMD_CLOSE_PORT = 0xa,
MLX4_CMD_QUERY_HCA = 0xb,
MLX4_CMD_QUERY_PORT = 0x43,
+ MLX4_CMD_SENSE_PORT = 0x4d,
MLX4_CMD_SET_PORT = 0xc,
MLX4_CMD_ACCESS_DDR = 0x2e,
MLX4_CMD_MAP_ICM = 0xffa,
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 8f659cc..3aff8a6 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -155,8 +155,9 @@
};
enum mlx4_port_type {
- MLX4_PORT_TYPE_IB = 1 << 0,
- MLX4_PORT_TYPE_ETH = 1 << 1,
+ MLX4_PORT_TYPE_IB = 1,
+ MLX4_PORT_TYPE_ETH = 2,
+ MLX4_PORT_TYPE_AUTO = 3
};
enum mlx4_special_vlan_idx {
@@ -237,6 +238,7 @@
enum mlx4_port_type port_type[MLX4_MAX_PORTS + 1];
u8 supported_type[MLX4_MAX_PORTS + 1];
u32 port_mask;
+ enum mlx4_port_type possible_type[MLX4_MAX_PORTS + 1];
};
struct mlx4_buf_list {
diff --git a/include/rdma/ib_mad.h b/include/rdma/ib_mad.h
index 8cc71a1..d3b9401 100644
--- a/include/rdma/ib_mad.h
+++ b/include/rdma/ib_mad.h
@@ -290,7 +290,7 @@
*/
static inline void ib_set_rmpp_flags(struct ib_rmpp_hdr *rmpp_hdr, u8 flags)
{
- rmpp_hdr->rmpp_rtime_flags = (rmpp_hdr->rmpp_rtime_flags & 0xF1) |
+ rmpp_hdr->rmpp_rtime_flags = (rmpp_hdr->rmpp_rtime_flags & 0xF8) |
(flags & 0x7);
}