From b3d636b0d1b2eb870a55ae196b8f3838e1399554 Mon Sep 17 00:00:00 2001
From: Roland Dreier <rolandd@cisco.com>
Date: Wed, 16 Apr 2008 21:01:06 -0700
Subject: IB: Make struct ib_uobject.id a signed int

IDR IDs are signed, so struct ib_uobject.id should be signed.  This
avoids some sparse pointer signedness warnings.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 include/rdma/ib_verbs.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 701e7b40560..40ff51244d1 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -730,7 +730,7 @@ struct ib_uobject {
 	struct ib_ucontext     *context;	/* associated user context */
 	void		       *object;		/* containing object */
 	struct list_head	list;		/* link to context's list */
-	u32			id;		/* index into kernel idr */
+	int			id;		/* index into kernel idr */
 	struct kref		ref;
 	struct rw_semaphore	mutex;		/* protects .live */
 	int			live;
-- 
cgit v1.2.3-70-g09d2


From 37608eea86a358ac6a18df0af55d4f77d08a1f30 Mon Sep 17 00:00:00 2001
From: Roland Dreier <rolandd@cisco.com>
Date: Wed, 16 Apr 2008 21:01:08 -0700
Subject: mlx4_core: Fix confusion between mlx4_event and mlx4_dev_event enums

The struct mlx4_interface.event() method was supposed to get an enum
mlx4_dev_event, but the driver code was actually passing in the
hardware enum mlx4_event values.  Fix up the callers of
mlx4_dispatch_event() so that they pass in the right type of value,
and fix up the event method in mlx4_ib so that it can handle the enum
mlx4_dev_event values.

This eliminates the need for the subtype parameter to the event
method, so remove it.

This also fixes the sparse warning

    drivers/net/mlx4/intf.c:127:48: warning: mixing different enum types
    drivers/net/mlx4/intf.c:127:48:     int enum mlx4_event  versus
    drivers/net/mlx4/intf.c:127:48:     int enum mlx4_dev_event

Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/hw/mlx4/main.c | 14 ++++++++------
 drivers/net/mlx4/catas.c          |  2 +-
 drivers/net/mlx4/eq.c             |  5 ++++-
 drivers/net/mlx4/intf.c           |  8 ++------
 drivers/net/mlx4/mlx4.h           |  4 ++--
 include/linux/mlx4/driver.h       |  3 +--
 6 files changed, 18 insertions(+), 18 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 96a39b5c925..d5512011999 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -675,18 +675,20 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr)
 }
 
 static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr,
-			  enum mlx4_dev_event event, int subtype,
-			  int port)
+			  enum mlx4_dev_event event, int port)
 {
 	struct ib_event ibev;
 
 	switch (event) {
-	case MLX4_EVENT_TYPE_PORT_CHANGE:
-		ibev.event = subtype == MLX4_PORT_CHANGE_SUBTYPE_ACTIVE ?
-			IB_EVENT_PORT_ACTIVE : IB_EVENT_PORT_ERR;
+	case MLX4_DEV_EVENT_PORT_UP:
+		ibev.event = IB_EVENT_PORT_ACTIVE;
 		break;
 
-	case MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR:
+	case MLX4_DEV_EVENT_PORT_DOWN:
+		ibev.event = IB_EVENT_PORT_ERR;
+		break;
+
+	case MLX4_DEV_EVENT_CATASTROPHIC_ERROR:
 		ibev.event = IB_EVENT_DEVICE_FATAL;
 		break;
 
diff --git a/drivers/net/mlx4/catas.c b/drivers/net/mlx4/catas.c
index 6b32ec94b3a..aa952877904 100644
--- a/drivers/net/mlx4/catas.c
+++ b/drivers/net/mlx4/catas.c
@@ -69,7 +69,7 @@ static void poll_catas(unsigned long dev_ptr)
 	if (readl(priv->catas_err.map)) {
 		dump_err_buf(dev);
 
-		mlx4_dispatch_event(dev, MLX4_EVENT_TYPE_LOCAL_CATAS_ERROR, 0, 0);
+		mlx4_dispatch_event(dev, MLX4_DEV_EVENT_CATASTROPHIC_ERROR, 0);
 
 		if (internal_err_reset) {
 			spin_lock(&catas_lock);
diff --git a/drivers/net/mlx4/eq.c b/drivers/net/mlx4/eq.c
index 9c36c203403..e141a1513f0 100644
--- a/drivers/net/mlx4/eq.c
+++ b/drivers/net/mlx4/eq.c
@@ -202,7 +202,10 @@ static int mlx4_eq_int(struct mlx4_dev *dev, struct mlx4_eq *eq)
 			break;
 
 		case MLX4_EVENT_TYPE_PORT_CHANGE:
-			mlx4_dispatch_event(dev, eqe->type, eqe->subtype,
+			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);
 			break;
 
diff --git a/drivers/net/mlx4/intf.c b/drivers/net/mlx4/intf.c
index be5d9e90ccf..4a6c4d526f1 100644
--- a/drivers/net/mlx4/intf.c
+++ b/drivers/net/mlx4/intf.c
@@ -30,8 +30,6 @@
  * SOFTWARE.
  */
 
-#include <linux/mlx4/driver.h>
-
 #include "mlx4.h"
 
 struct mlx4_device_context {
@@ -113,8 +111,7 @@ void mlx4_unregister_interface(struct mlx4_interface *intf)
 }
 EXPORT_SYMBOL_GPL(mlx4_unregister_interface);
 
-void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_event type,
-			 int subtype, int port)
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_dev_event type, int port)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
 	struct mlx4_device_context *dev_ctx;
@@ -124,8 +121,7 @@ void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_event type,
 
 	list_for_each_entry(dev_ctx, &priv->ctx_list, list)
 		if (dev_ctx->intf->event)
-			dev_ctx->intf->event(dev, dev_ctx->context, type,
-					     subtype, port);
+			dev_ctx->intf->event(dev, dev_ctx->context, type, port);
 
 	spin_unlock_irqrestore(&priv->ctx_lock, flags);
 }
diff --git a/drivers/net/mlx4/mlx4.h b/drivers/net/mlx4/mlx4.h
index 53a1cdddfc1..73336810e65 100644
--- a/drivers/net/mlx4/mlx4.h
+++ b/drivers/net/mlx4/mlx4.h
@@ -42,6 +42,7 @@
 #include <linux/timer.h>
 
 #include <linux/mlx4/device.h>
+#include <linux/mlx4/driver.h>
 #include <linux/mlx4/doorbell.h>
 
 #define DRV_NAME	"mlx4_core"
@@ -313,8 +314,7 @@ void mlx4_catas_cleanup(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);
-void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_event type,
-			 int subtype, int port);
+void mlx4_dispatch_event(struct mlx4_dev *dev, enum mlx4_dev_event type, int port);
 
 struct mlx4_dev_cap;
 struct mlx4_init_hca_param;
diff --git a/include/linux/mlx4/driver.h b/include/linux/mlx4/driver.h
index 1b835ca49df..53c5fdb6eac 100644
--- a/include/linux/mlx4/driver.h
+++ b/include/linux/mlx4/driver.h
@@ -48,8 +48,7 @@ struct mlx4_interface {
 	void *			(*add)	 (struct mlx4_dev *dev);
 	void			(*remove)(struct mlx4_dev *dev, void *context);
 	void			(*event) (struct mlx4_dev *dev, void *context,
-					  enum mlx4_dev_event event, int subtype,
-					  int port);
+					  enum mlx4_dev_event event, int port);
 	struct list_head	list;
 };
 
-- 
cgit v1.2.3-70-g09d2


From 8ff095ec4bce7be943beff3b330562e2f0e42167 Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:01:10 -0700
Subject: IB/mlx4: Add IPoIB checksum offload support

ConnectX devices support checksum generation and verification of TCP
and UDP packets for UD IPoIB messages.  This patch checks if the HCA
supports this and sets the IB_DEVICE_UD_IP_CSUM capability flag if it
does.  It implements support for handling the IB_SEND_IP_CSUM send
flag and setting the csum_ok field in receive work completions.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Ali Ayub <ali@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/hw/mlx4/cq.c   | 16 ++++++++++++++++
 drivers/infiniband/hw/mlx4/main.c |  2 ++
 drivers/infiniband/hw/mlx4/qp.c   |  3 +++
 drivers/net/mlx4/fw.c             |  4 ++++
 include/linux/mlx4/cq.h           | 14 ++++++++++++--
 include/linux/mlx4/qp.h           | 10 ++++++----
 6 files changed, 43 insertions(+), 6 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index 7360bbafbe8..d2e32b03e2f 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -297,6 +297,20 @@ static void mlx4_ib_handle_error_cqe(struct mlx4_err_cqe *cqe,
 	wc->vendor_err = cqe->vendor_err_syndrome;
 }
 
+static int mlx4_ib_ipoib_csum_ok(__be32 status, __be16 checksum)
+{
+	return ((status & cpu_to_be32(MLX4_CQE_IPOIB_STATUS_IPV4	|
+				      MLX4_CQE_IPOIB_STATUS_IPV4F	|
+				      MLX4_CQE_IPOIB_STATUS_IPV4OPT	|
+				      MLX4_CQE_IPOIB_STATUS_IPV6	|
+				      MLX4_CQE_IPOIB_STATUS_IPOK)) ==
+		cpu_to_be32(MLX4_CQE_IPOIB_STATUS_IPV4	|
+			    MLX4_CQE_IPOIB_STATUS_IPOK))		&&
+		(status & cpu_to_be32(MLX4_CQE_IPOIB_STATUS_UDP	|
+				      MLX4_CQE_IPOIB_STATUS_TCP))	&&
+		checksum == cpu_to_be16(0xffff);
+}
+
 static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 			    struct mlx4_ib_qp **cur_qp,
 			    struct ib_wc *wc)
@@ -434,6 +448,8 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		wc->dlid_path_bits = (g_mlpath_rqpn >> 24) & 0x7f;
 		wc->wc_flags	  |= g_mlpath_rqpn & 0x80000000 ? IB_WC_GRH : 0;
 		wc->pkey_index     = be32_to_cpu(cqe->immed_rss_invalid) & 0x7f;
+		wc->csum_ok	   = mlx4_ib_ipoib_csum_ok(cqe->ipoib_status,
+							   cqe->checksum);
 	}
 
 	return 0;
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index d5512011999..6ea4746c2e9 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -99,6 +99,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
 		props->device_cap_flags |= IB_DEVICE_AUTO_PATH_MIG;
 	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_UD_AV_PORT)
 		props->device_cap_flags |= IB_DEVICE_UD_AV_PORT_ENFORCE;
+	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM)
+		props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM;
 
 	props->vendor_id	   = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
 		0xffffff;
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index ac965ab2884..31b2b5b230b 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -1436,6 +1436,9 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			 cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE) : 0) |
 			(wr->send_flags & IB_SEND_SOLICITED ?
 			 cpu_to_be32(MLX4_WQE_CTRL_SOLICITED) : 0) |
+			((wr->send_flags & IB_SEND_IP_CSUM) ?
+			 cpu_to_be32(MLX4_WQE_CTRL_IP_CSUM |
+				     MLX4_WQE_CTRL_TCP_UDP_CSUM) : 0) |
 			qp->sq_signal_bits;
 
 		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
diff --git a/drivers/net/mlx4/fw.c b/drivers/net/mlx4/fw.c
index 61dc4951d6b..f494c3e8bce 100644
--- a/drivers/net/mlx4/fw.c
+++ b/drivers/net/mlx4/fw.c
@@ -696,6 +696,10 @@ int mlx4_INIT_HCA(struct mlx4_dev *dev, struct mlx4_init_hca_param *param)
 	/* Check port for UD address vector: */
 	*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1);
 
+	/* Enable IPoIB checksumming if we can: */
+	if (dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM)
+		*(inbox + INIT_HCA_FLAGS_OFFSET / 4) |= cpu_to_be32(1 << 3);
+
 	/* QPC/EEC/CQC/EQC/RDMARC attributes */
 
 	MLX4_PUT(inbox, param->qpc_base,      INIT_HCA_QPC_BASE_OFFSET);
diff --git a/include/linux/mlx4/cq.h b/include/linux/mlx4/cq.h
index 0181e0a57cb..1243ebace56 100644
--- a/include/linux/mlx4/cq.h
+++ b/include/linux/mlx4/cq.h
@@ -45,11 +45,11 @@ struct mlx4_cqe {
 	u8			sl;
 	u8			reserved1;
 	__be16			rlid;
-	u32			reserved2;
+	__be32			ipoib_status;
 	__be32			byte_cnt;
 	__be16			wqe_index;
 	__be16			checksum;
-	u8			reserved3[3];
+	u8			reserved2[3];
 	u8			owner_sr_opcode;
 };
 
@@ -85,6 +85,16 @@ enum {
 	MLX4_CQE_SYNDROME_REMOTE_ABORTED_ERR		= 0x22,
 };
 
+enum {
+	MLX4_CQE_IPOIB_STATUS_IPV4			= 1 << 22,
+	MLX4_CQE_IPOIB_STATUS_IPV4F			= 1 << 23,
+	MLX4_CQE_IPOIB_STATUS_IPV6			= 1 << 24,
+	MLX4_CQE_IPOIB_STATUS_IPV4OPT			= 1 << 25,
+	MLX4_CQE_IPOIB_STATUS_TCP			= 1 << 26,
+	MLX4_CQE_IPOIB_STATUS_UDP			= 1 << 27,
+	MLX4_CQE_IPOIB_STATUS_IPOK			= 1 << 28,
+};
+
 static inline void mlx4_cq_arm(struct mlx4_cq *cq, u32 cmd,
 			       void __iomem *uar_page,
 			       spinlock_t *doorbell_lock)
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index 09a2230923f..31f9eb3ccbb 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -158,10 +158,12 @@ struct mlx4_qp_context {
 #define MLX4_FW_VER_WQE_CTRL_NEC mlx4_fw_ver(2, 2, 232)
 
 enum {
-	MLX4_WQE_CTRL_NEC	= 1 << 29,
-	MLX4_WQE_CTRL_FENCE	= 1 << 6,
-	MLX4_WQE_CTRL_CQ_UPDATE	= 3 << 2,
-	MLX4_WQE_CTRL_SOLICITED	= 1 << 1,
+	MLX4_WQE_CTRL_NEC		= 1 << 29,
+	MLX4_WQE_CTRL_FENCE		= 1 << 6,
+	MLX4_WQE_CTRL_CQ_UPDATE		= 3 << 2,
+	MLX4_WQE_CTRL_SOLICITED		= 1 << 1,
+	MLX4_WQE_CTRL_IP_CSUM		= 1 << 4,
+	MLX4_WQE_CTRL_TCP_UDP_CSUM	= 1 << 5,
 };
 
 struct mlx4_wqe_ctrl_seg {
-- 
cgit v1.2.3-70-g09d2


From b846f25aa2a353355aec5202fe4dbdc6674dfc64 Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:27 -0700
Subject: IB/core: Add creation flags to struct ib_qp_init_attr

Add a create_flags member to struct ib_qp_init_attr that will allow a
kernel verbs consumer to create a pass special flags when creating a QP.
Add a flag value for telling low-level drivers that a QP will be used
for IPoIB UD LSO.  The create_flags member will also be useful for XRC
and ehca low-latency QP support.

Since no create_flags handling is implemented yet, add code to all
low-level drivers to return -EINVAL if create_flags is non-zero.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/core/uverbs_cmd.c         | 1 +
 drivers/infiniband/hw/amso1100/c2_provider.c | 3 +++
 drivers/infiniband/hw/ehca/ehca_qp.c         | 3 +++
 drivers/infiniband/hw/ipath/ipath_qp.c       | 5 +++++
 drivers/infiniband/hw/mlx4/qp.c              | 3 +++
 drivers/infiniband/hw/mthca/mthca_provider.c | 3 +++
 drivers/infiniband/hw/nes/nes_verbs.c        | 3 +++
 include/rdma/ib_verbs.h                      | 5 +++++
 8 files changed, 26 insertions(+)

(limited to 'include')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 495c803fb11..9e98cec6230 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1065,6 +1065,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
 	attr.srq           = srq;
 	attr.sq_sig_type   = cmd.sq_sig_all ? IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR;
 	attr.qp_type       = cmd.qp_type;
+	attr.create_flags  = 0;
 
 	attr.cap.max_send_wr     = cmd.max_send_wr;
 	attr.cap.max_recv_wr     = cmd.max_recv_wr;
diff --git a/drivers/infiniband/hw/amso1100/c2_provider.c b/drivers/infiniband/hw/amso1100/c2_provider.c
index 21fb46d3931..e10d27a6e14 100644
--- a/drivers/infiniband/hw/amso1100/c2_provider.c
+++ b/drivers/infiniband/hw/amso1100/c2_provider.c
@@ -245,6 +245,9 @@ static struct ib_qp *c2_create_qp(struct ib_pd *pd,
 
 	pr_debug("%s:%u\n", __func__, __LINE__);
 
+	if (init_attr->create_flags)
+		return ERR_PTR(-EINVAL);
+
 	switch (init_attr->qp_type) {
 	case IB_QPT_RC:
 		qp = kzalloc(sizeof(*qp), GFP_KERNEL);
diff --git a/drivers/infiniband/hw/ehca/ehca_qp.c b/drivers/infiniband/hw/ehca/ehca_qp.c
index a9fd419855c..3eb14a52cbf 100644
--- a/drivers/infiniband/hw/ehca/ehca_qp.c
+++ b/drivers/infiniband/hw/ehca/ehca_qp.c
@@ -421,6 +421,9 @@ static struct ehca_qp *internal_create_qp(
 	u32 swqe_size = 0, rwqe_size = 0, ib_qp_num;
 	unsigned long flags;
 
+	if (init_attr->create_flags)
+		return ERR_PTR(-EINVAL);
+
 	memset(&parms, 0, sizeof(parms));
 	qp_type = init_attr->qp_type;
 
diff --git a/drivers/infiniband/hw/ipath/ipath_qp.c b/drivers/infiniband/hw/ipath/ipath_qp.c
index 6a4a5e3b78b..812b42c500e 100644
--- a/drivers/infiniband/hw/ipath/ipath_qp.c
+++ b/drivers/infiniband/hw/ipath/ipath_qp.c
@@ -747,6 +747,11 @@ struct ib_qp *ipath_create_qp(struct ib_pd *ibpd,
 	size_t sz;
 	struct ib_qp *ret;
 
+	if (init_attr->create_flags) {
+		ret = ERR_PTR(-EINVAL);
+		goto bail;
+	}
+
 	if (init_attr->cap.max_send_sge > ib_ipath_max_sges ||
 	    init_attr->cap.max_send_wr > ib_ipath_max_qp_wrs) {
 		ret = ERR_PTR(-EINVAL);
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index 31b2b5b230b..320c25fa74b 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -673,6 +673,9 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
 	struct mlx4_ib_qp *qp;
 	int err;
 
+	if (init_attr->create_flags)
+		return ERR_PTR(-EINVAL);
+
 	switch (init_attr->qp_type) {
 	case IB_QPT_RC:
 	case IB_QPT_UC:
diff --git a/drivers/infiniband/hw/mthca/mthca_provider.c b/drivers/infiniband/hw/mthca/mthca_provider.c
index ee9bc145622..81b257e18bb 100644
--- a/drivers/infiniband/hw/mthca/mthca_provider.c
+++ b/drivers/infiniband/hw/mthca/mthca_provider.c
@@ -540,6 +540,9 @@ static struct ib_qp *mthca_create_qp(struct ib_pd *pd,
 	struct mthca_qp *qp;
 	int err;
 
+	if (init_attr->create_flags)
+		return ERR_PTR(-EINVAL);
+
 	switch (init_attr->qp_type) {
 	case IB_QPT_RC:
 	case IB_QPT_UC:
diff --git a/drivers/infiniband/hw/nes/nes_verbs.c b/drivers/infiniband/hw/nes/nes_verbs.c
index 46b3b8ebcc1..7c27420c224 100644
--- a/drivers/infiniband/hw/nes/nes_verbs.c
+++ b/drivers/infiniband/hw/nes/nes_verbs.c
@@ -1252,6 +1252,9 @@ static struct ib_qp *nes_create_qp(struct ib_pd *ibpd,
 	u8 rq_encoded_size;
 	/* int counter; */
 
+	if (init_attr->create_flags)
+		return ERR_PTR(-EINVAL);
+
 	atomic_inc(&qps_created);
 	switch (init_attr->qp_type) {
 		case IB_QPT_RC:
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 40ff51244d1..c3299be41c6 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -495,6 +495,10 @@ enum ib_qp_type {
 	IB_QPT_RAW_ETY
 };
 
+enum ib_qp_create_flags {
+	IB_QP_CREATE_IPOIB_UD_LSO	= 1 << 0,
+};
+
 struct ib_qp_init_attr {
 	void                  (*event_handler)(struct ib_event *, void *);
 	void		       *qp_context;
@@ -504,6 +508,7 @@ struct ib_qp_init_attr {
 	struct ib_qp_cap	cap;
 	enum ib_sig_type	sq_sig_type;
 	enum ib_qp_type		qp_type;
+	enum ib_qp_create_flags	create_flags;
 	u8			port_num; /* special QP types only */
 };
 
-- 
cgit v1.2.3-70-g09d2


From c93570f23a98c633570397aedc6d1808f5d5846a Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:27 -0700
Subject: IB/core: Add IPoIB UD LSO support

LSO (large send offload) allows the networking stack to pass SKBs with
data size larger than the MTU to the IPoIB driver and have the HCA HW
fragment the data to multiple MSS-sized packets.  Add a device
capability flag IB_DEVICE_UD_TSO for devices that can perform TCP
segmentation offload, a new send work request opcode IB_WR_LSO,
header, hlen and mss fields for the work request structure, and a new
IB_WC_LSO completion type.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 include/rdma/ib_verbs.h | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

(limited to 'include')

diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index c3299be41c6..66928e9cab1 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -104,6 +104,7 @@ enum ib_device_cap_flags {
 	 * IPoIB driver may set NETIF_F_IP_CSUM for datagram mode.
 	 */
 	IB_DEVICE_UD_IP_CSUM		= (1<<18),
+	IB_DEVICE_UD_TSO		= (1<<19),
 };
 
 enum ib_atomic_cap {
@@ -411,6 +412,7 @@ enum ib_wc_opcode {
 	IB_WC_COMP_SWAP,
 	IB_WC_FETCH_ADD,
 	IB_WC_BIND_MW,
+	IB_WC_LSO,
 /*
  * Set value of IB_WC_RECV so consumers can test if a completion is a
  * receive by testing (opcode & IB_WC_RECV).
@@ -622,7 +624,8 @@ enum ib_wr_opcode {
 	IB_WR_SEND_WITH_IMM,
 	IB_WR_RDMA_READ,
 	IB_WR_ATOMIC_CMP_AND_SWP,
-	IB_WR_ATOMIC_FETCH_AND_ADD
+	IB_WR_ATOMIC_FETCH_AND_ADD,
+	IB_WR_LSO
 };
 
 enum ib_send_flags {
@@ -660,6 +663,9 @@ struct ib_send_wr {
 		} atomic;
 		struct {
 			struct ib_ah *ah;
+			void   *header;
+			int     hlen;
+			int     mss;
 			u32	remote_qpn;
 			u32	remote_qkey;
 			u16	pkey_index; /* valid for GSI only */
-- 
cgit v1.2.3-70-g09d2


From b832be1e4007f4a54954ec68bd865ff05d6babca Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:27 -0700
Subject: IB/mlx4: Add IPoIB LSO support

Add TSO support to the mlx4_ib driver.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/hw/mlx4/cq.c      |  3 ++
 drivers/infiniband/hw/mlx4/main.c    |  2 +
 drivers/infiniband/hw/mlx4/mlx4_ib.h |  5 +++
 drivers/infiniband/hw/mlx4/qp.c      | 72 +++++++++++++++++++++++++++++++-----
 drivers/net/mlx4/fw.c                |  9 +++++
 drivers/net/mlx4/fw.h                |  1 +
 drivers/net/mlx4/main.c              |  1 +
 include/linux/mlx4/device.h          |  1 +
 include/linux/mlx4/qp.h              |  5 +++
 9 files changed, 90 insertions(+), 9 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index d2e32b03e2f..7d70af7952b 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -420,6 +420,9 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		case MLX4_OPCODE_BIND_MW:
 			wc->opcode    = IB_WC_BIND_MW;
 			break;
+		case MLX4_OPCODE_LSO:
+			wc->opcode    = IB_WC_LSO;
+			break;
 		}
 	} else {
 		wc->byte_len = be32_to_cpu(cqe->byte_cnt);
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 6ea4746c2e9..e9330a0d6c0 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -101,6 +101,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
 		props->device_cap_flags |= IB_DEVICE_UD_AV_PORT_ENFORCE;
 	if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_IPOIB_CSUM)
 		props->device_cap_flags |= IB_DEVICE_UD_IP_CSUM;
+	if (dev->dev->caps.max_gso_sz)
+		props->device_cap_flags |= IB_DEVICE_UD_TSO;
 
 	props->vendor_id	   = be32_to_cpup((__be32 *) (out_mad->data + 36)) &
 		0xffffff;
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index 3726e451a32..3f8bd0a37b9 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -110,6 +110,10 @@ struct mlx4_ib_wq {
 	unsigned		tail;
 };
 
+enum mlx4_ib_qp_flags {
+	MLX4_IB_QP_LSO		= 1 << 0
+};
+
 struct mlx4_ib_qp {
 	struct ib_qp		ibqp;
 	struct mlx4_qp		mqp;
@@ -129,6 +133,7 @@ struct mlx4_ib_qp {
 	struct mlx4_mtt		mtt;
 	int			buf_size;
 	struct mutex		mutex;
+	u32			flags;
 	u8			port;
 	u8			alt_port;
 	u8			atomic_rd_en;
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index 320c25fa74b..2ba24308408 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -71,6 +71,7 @@ enum {
 
 static const __be32 mlx4_ib_opcode[] = {
 	[IB_WR_SEND]			= __constant_cpu_to_be32(MLX4_OPCODE_SEND),
+	[IB_WR_LSO]			= __constant_cpu_to_be32(MLX4_OPCODE_LSO),
 	[IB_WR_SEND_WITH_IMM]		= __constant_cpu_to_be32(MLX4_OPCODE_SEND_IMM),
 	[IB_WR_RDMA_WRITE]		= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE),
 	[IB_WR_RDMA_WRITE_WITH_IMM]	= __constant_cpu_to_be32(MLX4_OPCODE_RDMA_WRITE_IMM),
@@ -242,7 +243,7 @@ static void mlx4_ib_qp_event(struct mlx4_qp *qp, enum mlx4_event type)
 	}
 }
 
-static int send_wqe_overhead(enum ib_qp_type type)
+static int send_wqe_overhead(enum ib_qp_type type, u32 flags)
 {
 	/*
 	 * UD WQEs must have a datagram segment.
@@ -253,7 +254,8 @@ static int send_wqe_overhead(enum ib_qp_type type)
 	switch (type) {
 	case IB_QPT_UD:
 		return sizeof (struct mlx4_wqe_ctrl_seg) +
-			sizeof (struct mlx4_wqe_datagram_seg);
+			sizeof (struct mlx4_wqe_datagram_seg) +
+			((flags & MLX4_IB_QP_LSO) ? 64 : 0);
 	case IB_QPT_UC:
 		return sizeof (struct mlx4_wqe_ctrl_seg) +
 			sizeof (struct mlx4_wqe_raddr_seg);
@@ -315,7 +317,7 @@ static int set_kernel_sq_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap,
 	/* Sanity check SQ size before proceeding */
 	if (cap->max_send_wr	 > dev->dev->caps.max_wqes  ||
 	    cap->max_send_sge	 > dev->dev->caps.max_sq_sg ||
-	    cap->max_inline_data + send_wqe_overhead(type) +
+	    cap->max_inline_data + send_wqe_overhead(type, qp->flags) +
 	    sizeof (struct mlx4_wqe_inline_seg) > dev->dev->caps.max_sq_desc_sz)
 		return -EINVAL;
 
@@ -329,7 +331,7 @@ static int set_kernel_sq_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap,
 
 	s = max(cap->max_send_sge * sizeof (struct mlx4_wqe_data_seg),
 		cap->max_inline_data + sizeof (struct mlx4_wqe_inline_seg)) +
-		send_wqe_overhead(type);
+		send_wqe_overhead(type, qp->flags);
 
 	/*
 	 * Hermon supports shrinking WQEs, such that a single work
@@ -394,7 +396,8 @@ static int set_kernel_sq_size(struct mlx4_ib_dev *dev, struct ib_qp_cap *cap,
 	}
 
 	qp->sq.max_gs = ((qp->sq_max_wqes_per_wr << qp->sq.wqe_shift) -
-			 send_wqe_overhead(type)) / sizeof (struct mlx4_wqe_data_seg);
+			 send_wqe_overhead(type, qp->flags)) /
+		sizeof (struct mlx4_wqe_data_seg);
 
 	qp->buf_size = (qp->rq.wqe_cnt << qp->rq.wqe_shift) +
 		(qp->sq.wqe_cnt << qp->sq.wqe_shift);
@@ -503,6 +506,9 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
 	} else {
 		qp->sq_no_prefetch = 0;
 
+		if (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)
+			qp->flags |= MLX4_IB_QP_LSO;
+
 		err = set_kernel_sq_size(dev, &init_attr->cap, init_attr->qp_type, qp);
 		if (err)
 			goto err;
@@ -673,7 +679,11 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
 	struct mlx4_ib_qp *qp;
 	int err;
 
-	if (init_attr->create_flags)
+	/* We only support LSO, and only for kernel UD QPs. */
+	if (init_attr->create_flags & ~IB_QP_CREATE_IPOIB_UD_LSO)
+		return ERR_PTR(-EINVAL);
+	if (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO &&
+	    (pd->uobject || init_attr->qp_type != IB_QPT_UD))
 		return ERR_PTR(-EINVAL);
 
 	switch (init_attr->qp_type) {
@@ -879,10 +889,15 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
 		}
 	}
 
-	if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_SMI ||
-	    ibqp->qp_type == IB_QPT_UD)
+	if (ibqp->qp_type == IB_QPT_GSI || ibqp->qp_type == IB_QPT_SMI)
 		context->mtu_msgmax = (IB_MTU_4096 << 5) | 11;
-	else if (attr_mask & IB_QP_PATH_MTU) {
+	else if (ibqp->qp_type == IB_QPT_UD) {
+		if (qp->flags & MLX4_IB_QP_LSO)
+			context->mtu_msgmax = (IB_MTU_4096 << 5) |
+					      ilog2(dev->dev->caps.max_gso_sz);
+		else
+			context->mtu_msgmax = (IB_MTU_4096 << 5) | 11;
+	} else if (attr_mask & IB_QP_PATH_MTU) {
 		if (attr->path_mtu < IB_MTU_256 || attr->path_mtu > IB_MTU_4096) {
 			printk(KERN_ERR "path MTU (%u) is invalid\n",
 			       attr->path_mtu);
@@ -1399,6 +1414,34 @@ static void __set_data_seg(struct mlx4_wqe_data_seg *dseg, struct ib_sge *sg)
 	dseg->addr       = cpu_to_be64(sg->addr);
 }
 
+static int build_lso_seg(struct mlx4_lso_seg *wqe, struct ib_send_wr *wr,
+			 struct mlx4_ib_qp *qp, unsigned *lso_seg_len)
+{
+	unsigned halign = ALIGN(sizeof *wqe + wr->wr.ud.hlen, 16);
+
+	/*
+	 * This is a temporary limitation and will be removed in
+	 * a forthcoming FW release:
+	 */
+	if (unlikely(halign > 64))
+		return -EINVAL;
+
+	if (unlikely(!(qp->flags & MLX4_IB_QP_LSO) &&
+		     wr->num_sge > qp->sq.max_gs - (halign >> 4)))
+		return -EINVAL;
+
+	memcpy(wqe->header, wr->wr.ud.header, wr->wr.ud.hlen);
+
+	/* make sure LSO header is written before overwriting stamping */
+	wmb();
+
+	wqe->mss_hdr_size = cpu_to_be32((wr->wr.ud.mss - wr->wr.ud.hlen) << 16 |
+					wr->wr.ud.hlen);
+
+	*lso_seg_len = halign;
+	return 0;
+}
+
 int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 		      struct ib_send_wr **bad_wr)
 {
@@ -1412,6 +1455,7 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 	unsigned ind;
 	int uninitialized_var(stamp);
 	int uninitialized_var(size);
+	unsigned seglen;
 	int i;
 
 	spin_lock_irqsave(&qp->sq.lock, flags);
@@ -1490,6 +1534,16 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			set_datagram_seg(wqe, wr);
 			wqe  += sizeof (struct mlx4_wqe_datagram_seg);
 			size += sizeof (struct mlx4_wqe_datagram_seg) / 16;
+
+			if (wr->opcode == IB_WR_LSO) {
+				err = build_lso_seg(wqe, wr, qp, &seglen);
+				if (unlikely(err)) {
+					*bad_wr = wr;
+					goto out;
+				}
+				wqe  += seglen;
+				size += seglen / 16;
+			}
 			break;
 
 		case IB_QPT_SMI:
diff --git a/drivers/net/mlx4/fw.c b/drivers/net/mlx4/fw.c
index f494c3e8bce..d82f2751d2c 100644
--- a/drivers/net/mlx4/fw.c
+++ b/drivers/net/mlx4/fw.c
@@ -133,6 +133,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 #define QUERY_DEV_CAP_MAX_AV_OFFSET		0x27
 #define QUERY_DEV_CAP_MAX_REQ_QP_OFFSET		0x29
 #define QUERY_DEV_CAP_MAX_RES_QP_OFFSET		0x2b
+#define QUERY_DEV_CAP_MAX_GSO_OFFSET		0x2d
 #define QUERY_DEV_CAP_MAX_RDMA_OFFSET		0x2f
 #define QUERY_DEV_CAP_RSZ_SRQ_OFFSET		0x33
 #define QUERY_DEV_CAP_ACK_DELAY_OFFSET		0x35
@@ -215,6 +216,13 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 	dev_cap->max_requester_per_qp = 1 << (field & 0x3f);
 	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RES_QP_OFFSET);
 	dev_cap->max_responder_per_qp = 1 << (field & 0x3f);
+	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_GSO_OFFSET);
+	field &= 0x1f;
+	if (!field)
+		dev_cap->max_gso_sz = 0;
+	else
+		dev_cap->max_gso_sz = 1 << field;
+
 	MLX4_GET(field, outbox, QUERY_DEV_CAP_MAX_RDMA_OFFSET);
 	dev_cap->max_rdma_global = 1 << (field & 0x3f);
 	MLX4_GET(field, outbox, QUERY_DEV_CAP_ACK_DELAY_OFFSET);
@@ -377,6 +385,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 		 dev_cap->max_sq_desc_sz, dev_cap->max_sq_sg);
 	mlx4_dbg(dev, "Max RQ desc size: %d, max RQ S/G: %d\n",
 		 dev_cap->max_rq_desc_sz, dev_cap->max_rq_sg);
+	mlx4_dbg(dev, "Max GSO size: %d\n", dev_cap->max_gso_sz);
 
 	dump_dev_cap_flags(dev, dev_cap->flags);
 
diff --git a/drivers/net/mlx4/fw.h b/drivers/net/mlx4/fw.h
index e16dec89041..306cb9b0242 100644
--- a/drivers/net/mlx4/fw.h
+++ b/drivers/net/mlx4/fw.h
@@ -96,6 +96,7 @@ struct mlx4_dev_cap {
 	u8  bmme_flags;
 	u32 reserved_lkey;
 	u64 max_icm_sz;
+	int max_gso_sz;
 };
 
 struct mlx4_adapter {
diff --git a/drivers/net/mlx4/main.c b/drivers/net/mlx4/main.c
index 08bfc130a33..7cfbe75114d 100644
--- a/drivers/net/mlx4/main.c
+++ b/drivers/net/mlx4/main.c
@@ -159,6 +159,7 @@ static int mlx4_dev_cap(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap)
 	dev->caps.page_size_cap	     = ~(u32) (dev_cap->min_page_sz - 1);
 	dev->caps.flags		     = dev_cap->flags;
 	dev->caps.stat_rate_support  = dev_cap->stat_rate_support;
+	dev->caps.max_gso_sz	     = dev_cap->max_gso_sz;
 
 	return 0;
 }
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 6cdf813cd47..ff7df1a2222 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -186,6 +186,7 @@ struct mlx4_caps {
 	u32			flags;
 	u16			stat_rate_support;
 	u8			port_width_cap[MLX4_MAX_PORTS + 1];
+	int			max_gso_sz;
 };
 
 struct mlx4_buf_list {
diff --git a/include/linux/mlx4/qp.h b/include/linux/mlx4/qp.h
index 31f9eb3ccbb..a5e43febee4 100644
--- a/include/linux/mlx4/qp.h
+++ b/include/linux/mlx4/qp.h
@@ -219,6 +219,11 @@ struct mlx4_wqe_datagram_seg {
 	__be32			reservd[2];
 };
 
+struct mlx4_lso_seg {
+	__be32			mss_hdr_size;
+	__be32			header[0];
+};
+
 struct mlx4_wqe_bind_seg {
 	__be32			flags1;
 	__be32			flags2;
-- 
cgit v1.2.3-70-g09d2


From 0f39cf3d54e67a705773fd0ec56ca3dcd3e9272f Mon Sep 17 00:00:00 2001
From: Roland Dreier <rolandd@cisco.com>
Date: Wed, 16 Apr 2008 21:09:32 -0700
Subject: IB/core: Add support for "send with invalidate" work requests

Add a new IB_WR_SEND_WITH_INV send opcode that can be used to mark a
"send with invalidate" work request as defined in the iWARP verbs and
the InfiniBand base memory management extensions.  Also put "imm_data"
and a new "invalidate_rkey" member in a new "ex" union in struct
ib_send_wr. The invalidate_rkey member can be used to pass in an
R_Key/STag to be invalidated.  Add this new union to struct
ib_uverbs_send_wr.  Add code to copy the invalidate_rkey field in
ib_uverbs_post_send().

Fix up low-level drivers to deal with the change to struct ib_send_wr,
and just remove the imm_data initialization from net/sunrpc/xprtrdma/,
since that code never does any send with immediate operations.

Also, move the existing IB_DEVICE_SEND_W_INV flag to a new bit, since
the iWARP drivers currently in the tree set the bit.  The amso1100
driver at least will silently fail to honor the IB_SEND_INVALIDATE bit
if passed in as part of userspace send requests (since it does not
implement kernel bypass work request queueing).  Remove the flag from
all existing drivers that set it until we know which ones are OK.

The values chosen for the new flag is not consecutive to avoid clashing
with flags defined in the XRC patches, which are not merged yet but
which are already in use and are likely to be merged soon.

This resurrects a patch sent long ago by Mikkel Hagen <mhagen@iol.unh.edu>.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/core/uverbs_cmd.c        | 13 +++++++++++--
 drivers/infiniband/hw/amso1100/c2_rnic.c    |  2 +-
 drivers/infiniband/hw/cxgb3/iwch_provider.c |  3 +--
 drivers/infiniband/hw/cxgb3/iwch_qp.c       |  4 ++--
 drivers/infiniband/hw/ehca/ehca_reqs.c      |  2 +-
 drivers/infiniband/hw/ipath/ipath_rc.c      |  8 ++++----
 drivers/infiniband/hw/ipath/ipath_ruc.c     |  4 ++--
 drivers/infiniband/hw/ipath/ipath_uc.c      |  8 ++++----
 drivers/infiniband/hw/ipath/ipath_ud.c      |  4 ++--
 drivers/infiniband/hw/mlx4/qp.c             |  4 ++--
 drivers/infiniband/hw/mthca/mthca_qp.c      |  6 +++---
 drivers/infiniband/hw/nes/nes_hw.c          |  2 +-
 include/rdma/ib_user_verbs.h                |  5 ++++-
 include/rdma/ib_verbs.h                     | 11 ++++++++---
 net/sunrpc/xprtrdma/verbs.c                 |  1 -
 15 files changed, 46 insertions(+), 31 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c
index 9e98cec6230..2c3bff5fe86 100644
--- a/drivers/infiniband/core/uverbs_cmd.c
+++ b/drivers/infiniband/core/uverbs_cmd.c
@@ -1463,7 +1463,6 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file,
 		next->num_sge    = user_wr->num_sge;
 		next->opcode     = user_wr->opcode;
 		next->send_flags = user_wr->send_flags;
-		next->imm_data   = (__be32 __force) user_wr->imm_data;
 
 		if (is_ud) {
 			next->wr.ud.ah = idr_read_ah(user_wr->wr.ud.ah,
@@ -1476,14 +1475,24 @@ ssize_t ib_uverbs_post_send(struct ib_uverbs_file *file,
 			next->wr.ud.remote_qkey = user_wr->wr.ud.remote_qkey;
 		} else {
 			switch (next->opcode) {
-			case IB_WR_RDMA_WRITE:
 			case IB_WR_RDMA_WRITE_WITH_IMM:
+				next->ex.imm_data =
+					(__be32 __force) user_wr->ex.imm_data;
+			case IB_WR_RDMA_WRITE:
 			case IB_WR_RDMA_READ:
 				next->wr.rdma.remote_addr =
 					user_wr->wr.rdma.remote_addr;
 				next->wr.rdma.rkey        =
 					user_wr->wr.rdma.rkey;
 				break;
+			case IB_WR_SEND_WITH_IMM:
+				next->ex.imm_data =
+					(__be32 __force) user_wr->ex.imm_data;
+				break;
+			case IB_WR_SEND_WITH_INV:
+				next->ex.invalidate_rkey =
+					user_wr->ex.invalidate_rkey;
+				break;
 			case IB_WR_ATOMIC_CMP_AND_SWP:
 			case IB_WR_ATOMIC_FETCH_AND_ADD:
 				next->wr.atomic.remote_addr =
diff --git a/drivers/infiniband/hw/amso1100/c2_rnic.c b/drivers/infiniband/hw/amso1100/c2_rnic.c
index 7a625524e0c..b1441aeb60c 100644
--- a/drivers/infiniband/hw/amso1100/c2_rnic.c
+++ b/drivers/infiniband/hw/amso1100/c2_rnic.c
@@ -455,7 +455,7 @@ int __devinit c2_rnic_init(struct c2_dev *c2dev)
 	     IB_DEVICE_CURR_QP_STATE_MOD |
 	     IB_DEVICE_SYS_IMAGE_GUID |
 	     IB_DEVICE_ZERO_STAG |
-	     IB_DEVICE_SEND_W_INV | IB_DEVICE_MEM_WINDOW);
+	     IB_DEVICE_MEM_WINDOW);
 
 	/* Allocate the qptr_array */
 	c2dev->qptr_array = vmalloc(C2_MAX_CQS * sizeof(void *));
diff --git a/drivers/infiniband/hw/cxgb3/iwch_provider.c b/drivers/infiniband/hw/cxgb3/iwch_provider.c
index 50e1f2a16e0..ca7265443c0 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_provider.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_provider.c
@@ -1109,8 +1109,7 @@ int iwch_register_device(struct iwch_dev *dev)
 	memcpy(&dev->ibdev.node_guid, dev->rdev.t3cdev_p->lldev->dev_addr, 6);
 	dev->ibdev.owner = THIS_MODULE;
 	dev->device_cap_flags =
-	    (IB_DEVICE_ZERO_STAG |
-	     IB_DEVICE_SEND_W_INV | IB_DEVICE_MEM_WINDOW);
+	    (IB_DEVICE_ZERO_STAG | IB_DEVICE_MEM_WINDOW);
 
 	dev->ibdev.uverbs_cmd_mask =
 	    (1ull << IB_USER_VERBS_CMD_GET_CONTEXT) |
diff --git a/drivers/infiniband/hw/cxgb3/iwch_qp.c b/drivers/infiniband/hw/cxgb3/iwch_qp.c
index bc5d9b0813e..8891c3b0a3d 100644
--- a/drivers/infiniband/hw/cxgb3/iwch_qp.c
+++ b/drivers/infiniband/hw/cxgb3/iwch_qp.c
@@ -72,7 +72,7 @@ static int iwch_build_rdma_send(union t3_wr *wqe, struct ib_send_wr *wr,
 	wqe->send.reserved[2] = 0;
 	if (wr->opcode == IB_WR_SEND_WITH_IMM) {
 		plen = 4;
-		wqe->send.sgl[0].stag = wr->imm_data;
+		wqe->send.sgl[0].stag = wr->ex.imm_data;
 		wqe->send.sgl[0].len = __constant_cpu_to_be32(0);
 		wqe->send.num_sgle = __constant_cpu_to_be32(0);
 		*flit_cnt = 5;
@@ -112,7 +112,7 @@ static int iwch_build_rdma_write(union t3_wr *wqe, struct ib_send_wr *wr,
 
 	if (wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM) {
 		plen = 4;
-		wqe->write.sgl[0].stag = wr->imm_data;
+		wqe->write.sgl[0].stag = wr->ex.imm_data;
 		wqe->write.sgl[0].len = __constant_cpu_to_be32(0);
 		wqe->write.num_sgle = __constant_cpu_to_be32(0);
 		*flit_cnt = 6;
diff --git a/drivers/infiniband/hw/ehca/ehca_reqs.c b/drivers/infiniband/hw/ehca/ehca_reqs.c
index 2ce8cffb866..a20bbf46618 100644
--- a/drivers/infiniband/hw/ehca/ehca_reqs.c
+++ b/drivers/infiniband/hw/ehca/ehca_reqs.c
@@ -188,7 +188,7 @@ static inline int ehca_write_swqe(struct ehca_qp *qp,
 	if (send_wr->opcode == IB_WR_SEND_WITH_IMM ||
 	    send_wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM) {
 		/* this might not work as long as HW does not support it */
-		wqe_p->immediate_data = be32_to_cpu(send_wr->imm_data);
+		wqe_p->immediate_data = be32_to_cpu(send_wr->ex.imm_data);
 		wqe_p->wr_flag |= WQE_WRFLAG_IMM_DATA_PRESENT;
 	}
 
diff --git a/drivers/infiniband/hw/ipath/ipath_rc.c b/drivers/infiniband/hw/ipath/ipath_rc.c
index 467981905bb..c405dfba553 100644
--- a/drivers/infiniband/hw/ipath/ipath_rc.c
+++ b/drivers/infiniband/hw/ipath/ipath_rc.c
@@ -308,7 +308,7 @@ int ipath_make_rc_req(struct ipath_qp *qp)
 			else {
 				qp->s_state = OP(SEND_ONLY_WITH_IMMEDIATE);
 				/* Immediate data comes after the BTH */
-				ohdr->u.imm_data = wqe->wr.imm_data;
+				ohdr->u.imm_data = wqe->wr.ex.imm_data;
 				hwords += 1;
 			}
 			if (wqe->wr.send_flags & IB_SEND_SOLICITED)
@@ -346,7 +346,7 @@ int ipath_make_rc_req(struct ipath_qp *qp)
 				qp->s_state =
 					OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE);
 				/* Immediate data comes after RETH */
-				ohdr->u.rc.imm_data = wqe->wr.imm_data;
+				ohdr->u.rc.imm_data = wqe->wr.ex.imm_data;
 				hwords += 1;
 				if (wqe->wr.send_flags & IB_SEND_SOLICITED)
 					bth0 |= 1 << 23;
@@ -490,7 +490,7 @@ int ipath_make_rc_req(struct ipath_qp *qp)
 		else {
 			qp->s_state = OP(SEND_LAST_WITH_IMMEDIATE);
 			/* Immediate data comes after the BTH */
-			ohdr->u.imm_data = wqe->wr.imm_data;
+			ohdr->u.imm_data = wqe->wr.ex.imm_data;
 			hwords += 1;
 		}
 		if (wqe->wr.send_flags & IB_SEND_SOLICITED)
@@ -526,7 +526,7 @@ int ipath_make_rc_req(struct ipath_qp *qp)
 		else {
 			qp->s_state = OP(RDMA_WRITE_LAST_WITH_IMMEDIATE);
 			/* Immediate data comes after the BTH */
-			ohdr->u.imm_data = wqe->wr.imm_data;
+			ohdr->u.imm_data = wqe->wr.ex.imm_data;
 			hwords += 1;
 			if (wqe->wr.send_flags & IB_SEND_SOLICITED)
 				bth0 |= 1 << 23;
diff --git a/drivers/infiniband/hw/ipath/ipath_ruc.c b/drivers/infiniband/hw/ipath/ipath_ruc.c
index bcaa2914e34..8ac5c1d82cc 100644
--- a/drivers/infiniband/hw/ipath/ipath_ruc.c
+++ b/drivers/infiniband/hw/ipath/ipath_ruc.c
@@ -310,7 +310,7 @@ again:
 	switch (wqe->wr.opcode) {
 	case IB_WR_SEND_WITH_IMM:
 		wc.wc_flags = IB_WC_WITH_IMM;
-		wc.imm_data = wqe->wr.imm_data;
+		wc.imm_data = wqe->wr.ex.imm_data;
 		/* FALLTHROUGH */
 	case IB_WR_SEND:
 		if (!ipath_get_rwqe(qp, 0)) {
@@ -339,7 +339,7 @@ again:
 			goto err;
 		}
 		wc.wc_flags = IB_WC_WITH_IMM;
-		wc.imm_data = wqe->wr.imm_data;
+		wc.imm_data = wqe->wr.ex.imm_data;
 		if (!ipath_get_rwqe(qp, 1))
 			goto rnr_nak;
 		/* FALLTHROUGH */
diff --git a/drivers/infiniband/hw/ipath/ipath_uc.c b/drivers/infiniband/hw/ipath/ipath_uc.c
index 2dd8de20d22..bfe8926b551 100644
--- a/drivers/infiniband/hw/ipath/ipath_uc.c
+++ b/drivers/infiniband/hw/ipath/ipath_uc.c
@@ -94,7 +94,7 @@ int ipath_make_uc_req(struct ipath_qp *qp)
 				qp->s_state =
 					OP(SEND_ONLY_WITH_IMMEDIATE);
 				/* Immediate data comes after the BTH */
-				ohdr->u.imm_data = wqe->wr.imm_data;
+				ohdr->u.imm_data = wqe->wr.ex.imm_data;
 				hwords += 1;
 			}
 			if (wqe->wr.send_flags & IB_SEND_SOLICITED)
@@ -123,7 +123,7 @@ int ipath_make_uc_req(struct ipath_qp *qp)
 				qp->s_state =
 					OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE);
 				/* Immediate data comes after the RETH */
-				ohdr->u.rc.imm_data = wqe->wr.imm_data;
+				ohdr->u.rc.imm_data = wqe->wr.ex.imm_data;
 				hwords += 1;
 				if (wqe->wr.send_flags & IB_SEND_SOLICITED)
 					bth0 |= 1 << 23;
@@ -152,7 +152,7 @@ int ipath_make_uc_req(struct ipath_qp *qp)
 		else {
 			qp->s_state = OP(SEND_LAST_WITH_IMMEDIATE);
 			/* Immediate data comes after the BTH */
-			ohdr->u.imm_data = wqe->wr.imm_data;
+			ohdr->u.imm_data = wqe->wr.ex.imm_data;
 			hwords += 1;
 		}
 		if (wqe->wr.send_flags & IB_SEND_SOLICITED)
@@ -177,7 +177,7 @@ int ipath_make_uc_req(struct ipath_qp *qp)
 			qp->s_state =
 				OP(RDMA_WRITE_LAST_WITH_IMMEDIATE);
 			/* Immediate data comes after the BTH */
-			ohdr->u.imm_data = wqe->wr.imm_data;
+			ohdr->u.imm_data = wqe->wr.ex.imm_data;
 			hwords += 1;
 			if (wqe->wr.send_flags & IB_SEND_SOLICITED)
 				bth0 |= 1 << 23;
diff --git a/drivers/infiniband/hw/ipath/ipath_ud.c b/drivers/infiniband/hw/ipath/ipath_ud.c
index 918f5207065..8b6a261c89e 100644
--- a/drivers/infiniband/hw/ipath/ipath_ud.c
+++ b/drivers/infiniband/hw/ipath/ipath_ud.c
@@ -95,7 +95,7 @@ static void ipath_ud_loopback(struct ipath_qp *sqp, struct ipath_swqe *swqe)
 
 	if (swqe->wr.opcode == IB_WR_SEND_WITH_IMM) {
 		wc.wc_flags = IB_WC_WITH_IMM;
-		wc.imm_data = swqe->wr.imm_data;
+		wc.imm_data = swqe->wr.ex.imm_data;
 	} else {
 		wc.wc_flags = 0;
 		wc.imm_data = 0;
@@ -327,7 +327,7 @@ int ipath_make_ud_req(struct ipath_qp *qp)
 	}
 	if (wqe->wr.opcode == IB_WR_SEND_WITH_IMM) {
 		qp->s_hdrwords++;
-		ohdr->u.ud.imm_data = wqe->wr.imm_data;
+		ohdr->u.ud.imm_data = wqe->wr.ex.imm_data;
 		bth0 = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE << 24;
 	} else
 		bth0 = IB_OPCODE_UD_SEND_ONLY << 24;
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index f5210c17e31..38e651a6758 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -1249,7 +1249,7 @@ static int build_mlx_header(struct mlx4_ib_sqp *sqp, struct ib_send_wr *wr,
 	case IB_WR_SEND_WITH_IMM:
 		sqp->ud_header.bth.opcode	 = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;
 		sqp->ud_header.immediate_present = 1;
-		sqp->ud_header.immediate_data    = wr->imm_data;
+		sqp->ud_header.immediate_data    = wr->ex.imm_data;
 		break;
 	default:
 		return -EINVAL;
@@ -1492,7 +1492,7 @@ int mlx4_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 
 		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
 		    wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM)
-			ctrl->imm = wr->imm_data;
+			ctrl->imm = wr->ex.imm_data;
 		else
 			ctrl->imm = 0;
 
diff --git a/drivers/infiniband/hw/mthca/mthca_qp.c b/drivers/infiniband/hw/mthca/mthca_qp.c
index 8433897624b..b3fd6b05d79 100644
--- a/drivers/infiniband/hw/mthca/mthca_qp.c
+++ b/drivers/infiniband/hw/mthca/mthca_qp.c
@@ -1532,7 +1532,7 @@ static int build_mlx_header(struct mthca_dev *dev, struct mthca_sqp *sqp,
 	case IB_WR_SEND_WITH_IMM:
 		sqp->ud_header.bth.opcode = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;
 		sqp->ud_header.immediate_present = 1;
-		sqp->ud_header.immediate_data = wr->imm_data;
+		sqp->ud_header.immediate_data = wr->ex.imm_data;
 		break;
 	default:
 		return -EINVAL;
@@ -1679,7 +1679,7 @@ int mthca_tavor_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			cpu_to_be32(1);
 		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
 		    wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM)
-			((struct mthca_next_seg *) wqe)->imm = wr->imm_data;
+			((struct mthca_next_seg *) wqe)->imm = wr->ex.imm_data;
 
 		wqe += sizeof (struct mthca_next_seg);
 		size = sizeof (struct mthca_next_seg) / 16;
@@ -2020,7 +2020,7 @@ int mthca_arbel_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 			cpu_to_be32(1);
 		if (wr->opcode == IB_WR_SEND_WITH_IMM ||
 		    wr->opcode == IB_WR_RDMA_WRITE_WITH_IMM)
-			((struct mthca_next_seg *) wqe)->imm = wr->imm_data;
+			((struct mthca_next_seg *) wqe)->imm = wr->ex.imm_data;
 
 		wqe += sizeof (struct mthca_next_seg);
 		size = sizeof (struct mthca_next_seg) / 16;
diff --git a/drivers/infiniband/hw/nes/nes_hw.c b/drivers/infiniband/hw/nes/nes_hw.c
index 134189d77ed..aa53aab91bf 100644
--- a/drivers/infiniband/hw/nes/nes_hw.c
+++ b/drivers/infiniband/hw/nes/nes_hw.c
@@ -393,7 +393,7 @@ struct nes_adapter *nes_init_adapter(struct nes_device *nesdev, u8 hw_rev) {
 	nesadapter->base_pd = 1;
 
 	nesadapter->device_cap_flags =
-			IB_DEVICE_ZERO_STAG | IB_DEVICE_SEND_W_INV | IB_DEVICE_MEM_WINDOW;
+		IB_DEVICE_ZERO_STAG | IB_DEVICE_MEM_WINDOW;
 
 	nesadapter->allocated_qps = (unsigned long *)&(((unsigned char *)nesadapter)
 			[(sizeof(struct nes_adapter)+(sizeof(unsigned long)-1))&(~(sizeof(unsigned long)-1))]);
diff --git a/include/rdma/ib_user_verbs.h b/include/rdma/ib_user_verbs.h
index 64a721fcbc1..8d65bf0a625 100644
--- a/include/rdma/ib_user_verbs.h
+++ b/include/rdma/ib_user_verbs.h
@@ -533,7 +533,10 @@ struct ib_uverbs_send_wr {
 	__u32 num_sge;
 	__u32 opcode;
 	__u32 send_flags;
-	__u32 imm_data;
+	union {
+		__u32 imm_data;
+		__u32 invalidate_rkey;
+	} ex;
 	union {
 		struct {
 			__u64 remote_addr;
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 66928e9cab1..c48f6af5ef9 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -94,7 +94,7 @@ enum ib_device_cap_flags {
 	IB_DEVICE_SRQ_RESIZE		= (1<<13),
 	IB_DEVICE_N_NOTIFY_CQ		= (1<<14),
 	IB_DEVICE_ZERO_STAG		= (1<<15),
-	IB_DEVICE_SEND_W_INV		= (1<<16),
+	IB_DEVICE_RESERVED		= (1<<16), /* old SEND_W_INV */
 	IB_DEVICE_MEM_WINDOW		= (1<<17),
 	/*
 	 * Devices should set IB_DEVICE_UD_IP_SUM if they support
@@ -105,6 +105,7 @@ enum ib_device_cap_flags {
 	 */
 	IB_DEVICE_UD_IP_CSUM		= (1<<18),
 	IB_DEVICE_UD_TSO		= (1<<19),
+	IB_DEVICE_SEND_W_INV		= (1<<21),
 };
 
 enum ib_atomic_cap {
@@ -625,7 +626,8 @@ enum ib_wr_opcode {
 	IB_WR_RDMA_READ,
 	IB_WR_ATOMIC_CMP_AND_SWP,
 	IB_WR_ATOMIC_FETCH_AND_ADD,
-	IB_WR_LSO
+	IB_WR_LSO,
+	IB_WR_SEND_WITH_INV,
 };
 
 enum ib_send_flags {
@@ -649,7 +651,10 @@ struct ib_send_wr {
 	int			num_sge;
 	enum ib_wr_opcode	opcode;
 	int			send_flags;
-	__be32			imm_data;
+	union {
+		__be32		imm_data;
+		u32		invalidate_rkey;
+	} ex;
 	union {
 		struct {
 			u64	remote_addr;
diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c
index ffbf22a1d2c..8ea283ecc52 100644
--- a/net/sunrpc/xprtrdma/verbs.c
+++ b/net/sunrpc/xprtrdma/verbs.c
@@ -1573,7 +1573,6 @@ rpcrdma_ep_post(struct rpcrdma_ia *ia,
 	send_wr.sg_list = req->rl_send_iov;
 	send_wr.num_sge = req->rl_niovs;
 	send_wr.opcode = IB_WR_SEND;
-	send_wr.imm_data = 0;
 	if (send_wr.num_sge == 4)	/* no need to sync any pad (constant) */
 		ib_dma_sync_single_for_device(ia->ri_id->device,
 			req->rl_send_iov[3].addr, req->rl_send_iov[3].length,
-- 
cgit v1.2.3-70-g09d2


From 2dd5716227878d5950988514a2cbabf72f7fc888 Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:33 -0700
Subject: IB/core: Add support for modify CQ

Add support for modifying CQ parameters for controlling event
generation moderation.

Signed-off-by: Eli Cohen <eli@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/core/verbs.c |  7 +++++++
 include/rdma/ib_verbs.h         | 11 +++++++++++
 2 files changed, 18 insertions(+)

(limited to 'include')

diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 8ffb5f2c21f..05042089de6 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -630,6 +630,13 @@ struct ib_cq *ib_create_cq(struct ib_device *device,
 }
 EXPORT_SYMBOL(ib_create_cq);
 
+int ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
+{
+	return cq->device->modify_cq ?
+		cq->device->modify_cq(cq, cq_count, cq_period) : -ENOSYS;
+}
+EXPORT_SYMBOL(ib_modify_cq);
+
 int ib_destroy_cq(struct ib_cq *cq)
 {
 	if (atomic_read(&cq->usecnt))
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index c48f6af5ef9..95bf4bac44c 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -987,6 +987,8 @@ struct ib_device {
 						int comp_vector,
 						struct ib_ucontext *context,
 						struct ib_udata *udata);
+	int                        (*modify_cq)(struct ib_cq *cq, u16 cq_count,
+						u16 cq_period);
 	int                        (*destroy_cq)(struct ib_cq *cq);
 	int                        (*resize_cq)(struct ib_cq *cq, int cqe,
 						struct ib_udata *udata);
@@ -1391,6 +1393,15 @@ struct ib_cq *ib_create_cq(struct ib_device *device,
  */
 int ib_resize_cq(struct ib_cq *cq, int cqe);
 
+/**
+ * ib_modify_cq - Modifies moderation params of the CQ
+ * @cq: The CQ to modify.
+ * @cq_count: number of CQEs that will trigger an event
+ * @cq_period: max period of time in usec before triggering an event
+ *
+ */
+int ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period);
+
 /**
  * ib_destroy_cq - Destroys the specified CQ.
  * @cq: The CQ to destroy.
-- 
cgit v1.2.3-70-g09d2


From 3fdcb97f0b8d8a29117dc36acd0b15965d2a2160 Mon Sep 17 00:00:00 2001
From: Eli Cohen <eli@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:33 -0700
Subject: IB/mlx4: Add support for modifying CQ moderation parameters

Signed-off-by: Eli Cohen <eli@mellnaox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/hw/mlx4/cq.c      |  8 +++++++
 drivers/infiniband/hw/mlx4/main.c    |  1 +
 drivers/infiniband/hw/mlx4/mlx4_ib.h |  1 +
 drivers/net/mlx4/cq.c                | 44 ++++++++++++++++++++++++++++++------
 include/linux/mlx4/cmd.h             |  2 +-
 include/linux/mlx4/cq.h              |  3 +++
 6 files changed, 51 insertions(+), 8 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index 7d70af7952b..e4fb64b118e 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -85,6 +85,14 @@ static struct mlx4_cqe *next_cqe_sw(struct mlx4_ib_cq *cq)
 	return get_sw_cqe(cq, cq->mcq.cons_index);
 }
 
+int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
+{
+	struct mlx4_ib_cq *mcq = to_mcq(cq);
+	struct mlx4_ib_dev *dev = to_mdev(cq->device);
+
+	return mlx4_cq_modify(dev->dev, &mcq->mcq, cq_count, cq_period);
+}
+
 struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
 				struct ib_ucontext *context,
 				struct ib_udata *udata)
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index e9330a0d6c0..76dd45c764b 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -609,6 +609,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.post_send		= mlx4_ib_post_send;
 	ibdev->ib_dev.post_recv		= mlx4_ib_post_recv;
 	ibdev->ib_dev.create_cq		= mlx4_ib_create_cq;
+	ibdev->ib_dev.modify_cq		= mlx4_ib_modify_cq;
 	ibdev->ib_dev.destroy_cq	= mlx4_ib_destroy_cq;
 	ibdev->ib_dev.poll_cq		= mlx4_ib_poll_cq;
 	ibdev->ib_dev.req_notify_cq	= mlx4_ib_arm_cq;
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index 3f8bd0a37b9..ef8ad96e143 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -254,6 +254,7 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 				  struct ib_udata *udata);
 int mlx4_ib_dereg_mr(struct ib_mr *mr);
 
+int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period);
 struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
 				struct ib_ucontext *context,
 				struct ib_udata *udata);
diff --git a/drivers/net/mlx4/cq.c b/drivers/net/mlx4/cq.c
index d4441fee3d8..8c314341434 100644
--- a/drivers/net/mlx4/cq.c
+++ b/drivers/net/mlx4/cq.c
@@ -38,6 +38,7 @@
 #include <linux/hardirq.h>
 
 #include <linux/mlx4/cmd.h>
+#include <linux/mlx4/cq.h>
 
 #include "mlx4.h"
 #include "icm.h"
@@ -47,21 +48,19 @@ struct mlx4_cq_context {
 	u16			reserved1[3];
 	__be16			page_offset;
 	__be32			logsize_usrpage;
-	u8			reserved2;
-	u8			cq_period;
-	u8			reserved3;
-	u8			cq_max_count;
-	u8			reserved4[3];
+	__be16			cq_period;
+	__be16			cq_max_count;
+	u8			reserved2[3];
 	u8			comp_eqn;
 	u8			log_page_size;
-	u8			reserved5[2];
+	u8			reserved3[2];
 	u8			mtt_base_addr_h;
 	__be32			mtt_base_addr_l;
 	__be32			last_notified_index;
 	__be32			solicit_producer_index;
 	__be32			consumer_index;
 	__be32			producer_index;
-	u32			reserved6[2];
+	u32			reserved4[2];
 	__be64			db_rec_addr;
 };
 
@@ -121,6 +120,13 @@ static int mlx4_SW2HW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
 			MLX4_CMD_TIME_CLASS_A);
 }
 
+static int mlx4_MODIFY_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
+			 int cq_num, u32 opmod)
+{
+	return mlx4_cmd(dev, mailbox->dma, cq_num, opmod, MLX4_CMD_MODIFY_CQ,
+			MLX4_CMD_TIME_CLASS_A);
+}
+
 static int mlx4_HW2SW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
 			 int cq_num)
 {
@@ -129,6 +135,30 @@ static int mlx4_HW2SW_CQ(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox,
 			    MLX4_CMD_TIME_CLASS_A);
 }
 
+int mlx4_cq_modify(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   u16 count, u16 period)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	cq_context = mailbox->buf;
+	memset(cq_context, 0, sizeof *cq_context);
+
+	cq_context->cq_max_count = cpu_to_be16(count);
+	cq_context->cq_period    = cpu_to_be16(period);
+
+	err = mlx4_MODIFY_CQ(dev, mailbox, cq->cqn, 1);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_modify);
+
 int mlx4_cq_alloc(struct mlx4_dev *dev, int nent, struct mlx4_mtt *mtt,
 		  struct mlx4_uar *uar, u64 db_rec, struct mlx4_cq *cq)
 {
diff --git a/include/linux/mlx4/cmd.h b/include/linux/mlx4/cmd.h
index 7d1eaa97de1..77323a72dd3 100644
--- a/include/linux/mlx4/cmd.h
+++ b/include/linux/mlx4/cmd.h
@@ -81,7 +81,7 @@ enum {
 	MLX4_CMD_SW2HW_CQ	 = 0x16,
 	MLX4_CMD_HW2SW_CQ	 = 0x17,
 	MLX4_CMD_QUERY_CQ	 = 0x18,
-	MLX4_CMD_RESIZE_CQ	 = 0x2c,
+	MLX4_CMD_MODIFY_CQ	 = 0x2c,
 
 	/* SRQ commands */
 	MLX4_CMD_SW2HW_SRQ	 = 0x35,
diff --git a/include/linux/mlx4/cq.h b/include/linux/mlx4/cq.h
index 1243ebace56..f7c3511c594 100644
--- a/include/linux/mlx4/cq.h
+++ b/include/linux/mlx4/cq.h
@@ -130,4 +130,7 @@ enum {
 	MLX4_CQ_DB_REQ_NOT		= 2 << 24
 };
 
+int mlx4_cq_modify(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   u16 count, u16 period);
+
 #endif /* MLX4_CQ_H */
-- 
cgit v1.2.3-70-g09d2


From bbf8eed1a0f8949f7385146624f736f829992a70 Mon Sep 17 00:00:00 2001
From: Vladimir Sokolovsky <vlad@dev.mellanox.co.il>
Date: Wed, 16 Apr 2008 21:09:33 -0700
Subject: IB/mlx4: Add support for resizing CQs

Signed-off-by: Vladimir Sokolovsky <vlad@mellanox.co.il>
Signed-off-by: Roland Dreier <rolandd@cisco.com>
---
 drivers/infiniband/hw/mlx4/cq.c      | 292 +++++++++++++++++++++++++++++++----
 drivers/infiniband/hw/mlx4/main.c    |   2 +
 drivers/infiniband/hw/mlx4/mlx4_ib.h |   9 ++
 drivers/net/mlx4/cq.c                |  28 ++++
 include/linux/mlx4/cq.h              |   2 +
 5 files changed, 300 insertions(+), 33 deletions(-)

(limited to 'include')

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index e4fb64b118e..3557e7edc9b 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -93,6 +93,74 @@ int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period)
 	return mlx4_cq_modify(dev->dev, &mcq->mcq, cq_count, cq_period);
 }
 
+static int mlx4_ib_alloc_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int nent)
+{
+	int err;
+
+	err = mlx4_buf_alloc(dev->dev, nent * sizeof(struct mlx4_cqe),
+			     PAGE_SIZE * 2, &buf->buf);
+
+	if (err)
+		goto out;
+
+	err = mlx4_mtt_init(dev->dev, buf->buf.npages, buf->buf.page_shift,
+				    &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_buf_write_mtt(dev->dev, &buf->mtt, &buf->buf);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	mlx4_buf_free(dev->dev, nent * sizeof(struct mlx4_cqe),
+			      &buf->buf);
+
+out:
+	return err;
+}
+
+static void mlx4_ib_free_cq_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq_buf *buf, int cqe)
+{
+	mlx4_buf_free(dev->dev, (cqe + 1) * sizeof(struct mlx4_cqe), &buf->buf);
+}
+
+static int mlx4_ib_get_cq_umem(struct mlx4_ib_dev *dev, struct ib_ucontext *context,
+			       struct mlx4_ib_cq_buf *buf, struct ib_umem **umem,
+			       u64 buf_addr, int cqe)
+{
+	int err;
+
+	*umem = ib_umem_get(context, buf_addr, cqe * sizeof (struct mlx4_cqe),
+			    IB_ACCESS_LOCAL_WRITE);
+	if (IS_ERR(*umem))
+		return PTR_ERR(*umem);
+
+	err = mlx4_mtt_init(dev->dev, ib_umem_page_count(*umem),
+			    ilog2((*umem)->page_size), &buf->mtt);
+	if (err)
+		goto err_buf;
+
+	err = mlx4_ib_umem_write_mtt(dev, &buf->mtt, *umem);
+	if (err)
+		goto err_mtt;
+
+	return 0;
+
+err_mtt:
+	mlx4_mtt_cleanup(dev->dev, &buf->mtt);
+
+err_buf:
+	ib_umem_release(*umem);
+
+	return err;
+}
+
 struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
 				struct ib_ucontext *context,
 				struct ib_udata *udata)
@@ -100,7 +168,6 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 	struct mlx4_ib_dev *dev = to_mdev(ibdev);
 	struct mlx4_ib_cq *cq;
 	struct mlx4_uar *uar;
-	int buf_size;
 	int err;
 
 	if (entries < 1 || entries > dev->dev->caps.max_cqes)
@@ -112,8 +179,10 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 
 	entries      = roundup_pow_of_two(entries + 1);
 	cq->ibcq.cqe = entries - 1;
-	buf_size     = entries * sizeof (struct mlx4_cqe);
+	mutex_init(&cq->resize_mutex);
 	spin_lock_init(&cq->lock);
+	cq->resize_buf = NULL;
+	cq->resize_umem = NULL;
 
 	if (context) {
 		struct mlx4_ib_create_cq ucmd;
@@ -123,21 +192,10 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 			goto err_cq;
 		}
 
-		cq->umem = ib_umem_get(context, ucmd.buf_addr, buf_size,
-				       IB_ACCESS_LOCAL_WRITE);
-		if (IS_ERR(cq->umem)) {
-			err = PTR_ERR(cq->umem);
-			goto err_cq;
-		}
-
-		err = mlx4_mtt_init(dev->dev, ib_umem_page_count(cq->umem),
-				    ilog2(cq->umem->page_size), &cq->buf.mtt);
+		err = mlx4_ib_get_cq_umem(dev, context, &cq->buf, &cq->umem,
+					  ucmd.buf_addr, entries);
 		if (err)
-			goto err_buf;
-
-		err = mlx4_ib_umem_write_mtt(dev, &cq->buf.mtt, cq->umem);
-		if (err)
-			goto err_mtt;
+			goto err_cq;
 
 		err = mlx4_ib_db_map_user(to_mucontext(context), ucmd.db_addr,
 					  &cq->db);
@@ -155,19 +213,9 @@ struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector
 		*cq->mcq.set_ci_db = 0;
 		*cq->mcq.arm_db    = 0;
 
-		if (mlx4_buf_alloc(dev->dev, buf_size, PAGE_SIZE * 2, &cq->buf.buf)) {
-			err = -ENOMEM;
-			goto err_db;
-		}
-
-		err = mlx4_mtt_init(dev->dev, cq->buf.buf.npages, cq->buf.buf.page_shift,
-				    &cq->buf.mtt);
+		err = mlx4_ib_alloc_cq_buf(dev, &cq->buf, entries);
 		if (err)
-			goto err_buf;
-
-		err = mlx4_buf_write_mtt(dev->dev, &cq->buf.mtt, &cq->buf.buf);
-		if (err)
-			goto err_mtt;
+			goto err_db;
 
 		uar = &dev->priv_uar;
 	}
@@ -195,12 +243,10 @@ err_dbmap:
 err_mtt:
 	mlx4_mtt_cleanup(dev->dev, &cq->buf.mtt);
 
-err_buf:
 	if (context)
 		ib_umem_release(cq->umem);
 	else
-		mlx4_buf_free(dev->dev, entries * sizeof (struct mlx4_cqe),
-			      &cq->buf.buf);
+		mlx4_ib_free_cq_buf(dev, &cq->buf, entries);
 
 err_db:
 	if (!context)
@@ -212,6 +258,170 @@ err_cq:
 	return ERR_PTR(err);
 }
 
+static int mlx4_alloc_resize_buf(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				  int entries)
+{
+	int err;
+
+	if (cq->resize_buf)
+		return -EBUSY;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_alloc_cq_buf(dev, &cq->resize_buf->buf, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_alloc_resize_umem(struct mlx4_ib_dev *dev, struct mlx4_ib_cq *cq,
+				   int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_resize_cq ucmd;
+	int err;
+
+	if (cq->resize_umem)
+		return -EBUSY;
+
+	if (ib_copy_from_udata(&ucmd, udata, sizeof ucmd))
+		return -EFAULT;
+
+	cq->resize_buf = kmalloc(sizeof *cq->resize_buf, GFP_ATOMIC);
+	if (!cq->resize_buf)
+		return -ENOMEM;
+
+	err = mlx4_ib_get_cq_umem(dev, cq->umem->context, &cq->resize_buf->buf,
+				  &cq->resize_umem, ucmd.buf_addr, entries);
+	if (err) {
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		return err;
+	}
+
+	cq->resize_buf->cqe = entries - 1;
+
+	return 0;
+}
+
+static int mlx4_ib_get_outstanding_cqes(struct mlx4_ib_cq *cq)
+{
+	u32 i;
+
+	i = cq->mcq.cons_index;
+	while (get_sw_cqe(cq, i & cq->ibcq.cqe))
+		++i;
+
+	return i - cq->mcq.cons_index;
+}
+
+static void mlx4_ib_cq_resize_copy_cqes(struct mlx4_ib_cq *cq)
+{
+	struct mlx4_cqe *cqe;
+	int i;
+
+	i = cq->mcq.cons_index;
+	cqe = get_cqe(cq, i & cq->ibcq.cqe);
+	while ((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) != MLX4_CQE_OPCODE_RESIZE) {
+		memcpy(get_cqe_from_buf(&cq->resize_buf->buf,
+					(i + 1) & cq->resize_buf->cqe),
+			get_cqe(cq, i & cq->ibcq.cqe), sizeof(struct mlx4_cqe));
+		cqe = get_cqe(cq, ++i & cq->ibcq.cqe);
+	}
+	++cq->mcq.cons_index;
+}
+
+int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata)
+{
+	struct mlx4_ib_dev *dev = to_mdev(ibcq->device);
+	struct mlx4_ib_cq *cq = to_mcq(ibcq);
+	int outst_cqe;
+	int err;
+
+	mutex_lock(&cq->resize_mutex);
+
+	if (entries < 1 || entries > dev->dev->caps.max_cqes) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	entries = roundup_pow_of_two(entries + 1);
+	if (entries == ibcq->cqe + 1) {
+		err = 0;
+		goto out;
+	}
+
+	if (ibcq->uobject) {
+		err = mlx4_alloc_resize_umem(dev, cq, entries, udata);
+		if (err)
+			goto out;
+	} else {
+		/* Can't be smaller then the number of outstanding CQEs */
+		outst_cqe = mlx4_ib_get_outstanding_cqes(cq);
+		if (entries < outst_cqe + 1) {
+			err = 0;
+			goto out;
+		}
+
+		err = mlx4_alloc_resize_buf(dev, cq, entries);
+		if (err)
+			goto out;
+	}
+
+	err = mlx4_cq_resize(dev->dev, &cq->mcq, entries, &cq->resize_buf->buf.mtt);
+	if (err)
+		goto err_buf;
+
+	if (ibcq->uobject) {
+		cq->buf      = cq->resize_buf->buf;
+		cq->ibcq.cqe = cq->resize_buf->cqe;
+		ib_umem_release(cq->umem);
+		cq->umem     = cq->resize_umem;
+
+		kfree(cq->resize_buf);
+		cq->resize_buf = NULL;
+		cq->resize_umem = NULL;
+	} else {
+		spin_lock_irq(&cq->lock);
+		if (cq->resize_buf) {
+			mlx4_ib_cq_resize_copy_cqes(cq);
+			mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+		spin_unlock_irq(&cq->lock);
+	}
+
+	goto out;
+
+err_buf:
+	if (!ibcq->uobject)
+		mlx4_ib_free_cq_buf(dev, &cq->resize_buf->buf,
+				    cq->resize_buf->cqe);
+
+	kfree(cq->resize_buf);
+	cq->resize_buf = NULL;
+
+	if (cq->resize_umem) {
+		ib_umem_release(cq->resize_umem);
+		cq->resize_umem = NULL;
+	}
+
+out:
+	mutex_unlock(&cq->resize_mutex);
+	return err;
+}
+
 int mlx4_ib_destroy_cq(struct ib_cq *cq)
 {
 	struct mlx4_ib_dev *dev = to_mdev(cq->device);
@@ -224,8 +434,7 @@ int mlx4_ib_destroy_cq(struct ib_cq *cq)
 		mlx4_ib_db_unmap_user(to_mucontext(cq->uobject->context), &mcq->db);
 		ib_umem_release(mcq->umem);
 	} else {
-		mlx4_buf_free(dev->dev, (cq->cqe + 1) * sizeof (struct mlx4_cqe),
-			      &mcq->buf.buf);
+		mlx4_ib_free_cq_buf(dev, &mcq->buf, cq->cqe + 1);
 		mlx4_ib_db_free(dev, &mcq->db);
 	}
 
@@ -332,6 +541,7 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 	u32 g_mlpath_rqpn;
 	u16 wqe_ctr;
 
+repoll:
 	cqe = next_cqe_sw(cq);
 	if (!cqe)
 		return -EAGAIN;
@@ -354,6 +564,22 @@ static int mlx4_ib_poll_one(struct mlx4_ib_cq *cq,
 		return -EINVAL;
 	}
 
+	/* Resize CQ in progress */
+	if (unlikely((cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) == MLX4_CQE_OPCODE_RESIZE)) {
+		if (cq->resize_buf) {
+			struct mlx4_ib_dev *dev = to_mdev(cq->ibcq.device);
+
+			mlx4_ib_free_cq_buf(dev, &cq->buf, cq->ibcq.cqe);
+			cq->buf      = cq->resize_buf->buf;
+			cq->ibcq.cqe = cq->resize_buf->cqe;
+
+			kfree(cq->resize_buf);
+			cq->resize_buf = NULL;
+		}
+
+		goto repoll;
+	}
+
 	if (!*cur_qp ||
 	    (be32_to_cpu(cqe->my_qpn) & 0xffffff) != (*cur_qp)->mqp.qpn) {
 		/*
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 76dd45c764b..57885cd6193 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -571,6 +571,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 		(1ull << IB_USER_VERBS_CMD_DEREG_MR)		|
 		(1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL)	|
 		(1ull << IB_USER_VERBS_CMD_CREATE_CQ)		|
+		(1ull << IB_USER_VERBS_CMD_RESIZE_CQ)		|
 		(1ull << IB_USER_VERBS_CMD_DESTROY_CQ)		|
 		(1ull << IB_USER_VERBS_CMD_CREATE_QP)		|
 		(1ull << IB_USER_VERBS_CMD_MODIFY_QP)		|
@@ -610,6 +611,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev)
 	ibdev->ib_dev.post_recv		= mlx4_ib_post_recv;
 	ibdev->ib_dev.create_cq		= mlx4_ib_create_cq;
 	ibdev->ib_dev.modify_cq		= mlx4_ib_modify_cq;
+	ibdev->ib_dev.resize_cq		= mlx4_ib_resize_cq;
 	ibdev->ib_dev.destroy_cq	= mlx4_ib_destroy_cq;
 	ibdev->ib_dev.poll_cq		= mlx4_ib_poll_cq;
 	ibdev->ib_dev.req_notify_cq	= mlx4_ib_arm_cq;
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index ef8ad96e143..9e637323c15 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -78,13 +78,21 @@ struct mlx4_ib_cq_buf {
 	struct mlx4_mtt		mtt;
 };
 
+struct mlx4_ib_cq_resize {
+	struct mlx4_ib_cq_buf	buf;
+	int			cqe;
+};
+
 struct mlx4_ib_cq {
 	struct ib_cq		ibcq;
 	struct mlx4_cq		mcq;
 	struct mlx4_ib_cq_buf	buf;
+	struct mlx4_ib_cq_resize *resize_buf;
 	struct mlx4_ib_db	db;
 	spinlock_t		lock;
+	struct mutex		resize_mutex;
 	struct ib_umem	       *umem;
+	struct ib_umem	       *resize_umem;
 };
 
 struct mlx4_ib_mr {
@@ -255,6 +263,7 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length,
 int mlx4_ib_dereg_mr(struct ib_mr *mr);
 
 int mlx4_ib_modify_cq(struct ib_cq *cq, u16 cq_count, u16 cq_period);
+int mlx4_ib_resize_cq(struct ib_cq *ibcq, int entries, struct ib_udata *udata);
 struct ib_cq *mlx4_ib_create_cq(struct ib_device *ibdev, int entries, int vector,
 				struct ib_ucontext *context,
 				struct ib_udata *udata);
diff --git a/drivers/net/mlx4/cq.c b/drivers/net/mlx4/cq.c
index 8c314341434..caa5bcf54e3 100644
--- a/drivers/net/mlx4/cq.c
+++ b/drivers/net/mlx4/cq.c
@@ -159,6 +159,34 @@ int mlx4_cq_modify(struct mlx4_dev *dev, struct mlx4_cq *cq,
 }
 EXPORT_SYMBOL_GPL(mlx4_cq_modify);
 
+int mlx4_cq_resize(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   int entries, struct mlx4_mtt *mtt)
+{
+	struct mlx4_cmd_mailbox *mailbox;
+	struct mlx4_cq_context *cq_context;
+	u64 mtt_addr;
+	int err;
+
+	mailbox = mlx4_alloc_cmd_mailbox(dev);
+	if (IS_ERR(mailbox))
+		return PTR_ERR(mailbox);
+
+	cq_context = mailbox->buf;
+	memset(cq_context, 0, sizeof *cq_context);
+
+	cq_context->logsize_usrpage = cpu_to_be32(ilog2(entries) << 24);
+	cq_context->log_page_size   = mtt->page_shift - 12;
+	mtt_addr = mlx4_mtt_addr(dev, mtt);
+	cq_context->mtt_base_addr_h = mtt_addr >> 32;
+	cq_context->mtt_base_addr_l = cpu_to_be32(mtt_addr & 0xffffffff);
+
+	err = mlx4_MODIFY_CQ(dev, mailbox, cq->cqn, 1);
+
+	mlx4_free_cmd_mailbox(dev, mailbox);
+	return err;
+}
+EXPORT_SYMBOL_GPL(mlx4_cq_resize);
+
 int mlx4_cq_alloc(struct mlx4_dev *dev, int nent, struct mlx4_mtt *mtt,
 		  struct mlx4_uar *uar, u64 db_rec, struct mlx4_cq *cq)
 {
diff --git a/include/linux/mlx4/cq.h b/include/linux/mlx4/cq.h
index f7c3511c594..071cf96cf01 100644
--- a/include/linux/mlx4/cq.h
+++ b/include/linux/mlx4/cq.h
@@ -132,5 +132,7 @@ enum {
 
 int mlx4_cq_modify(struct mlx4_dev *dev, struct mlx4_cq *cq,
 		   u16 count, u16 period);
+int mlx4_cq_resize(struct mlx4_dev *dev, struct mlx4_cq *cq,
+		   int entries, struct mlx4_mtt *mtt);
 
 #endif /* MLX4_CQ_H */
-- 
cgit v1.2.3-70-g09d2