Hi Eli, are there known IBoE fixes which are in ofed but missed 2.6.37-rc1?
Also, can the below and/or any other enhancements you've placed in ofed be
pushed upstream? it would be great to have perf counters operating fine for IBoE
Or.
>From 72c316b60f62401e031520fe3f55ec6879bbc42b Mon Sep 17 00:00:00 2001
From: Eli Cohen <e...@mellanox.co.il>
Date: Wed, 6 Jan 2010 14:09:38 +0200
Subject: [PATCH 12/12] mlx4: add support for reading performance counters
This patch uses basic or extended counters which can be read by a command
interface, to report counters for all the QPs that work on an rdmaoe port. This
effectively allows to implement performance counter ala IB.
Signed-off-by: Eli Cohen <e...@mellanox.co.il>
---
drivers/infiniband/hw/mlx4/mad.c | 86 ++++++++++++++++++++++++++++++++-
drivers/infiniband/hw/mlx4/main.c | 17 ++++++-
drivers/infiniband/hw/mlx4/mlx4_ib.h | 1 +
drivers/infiniband/hw/mlx4/qp.c | 2 +
drivers/net/mlx4/fw.h | 1 -
drivers/net/mlx4/main.c | 22 +++++++--
include/linux/mlx4/cmd.h | 4 ++
include/linux/mlx4/device.h | 36 ++++++++++++++
8 files changed, 159 insertions(+), 10 deletions(-)
Index: ofed_kernel-fixes/drivers/infiniband/hw/mlx4/mad.c
===================================================================
--- ofed_kernel-fixes.orig/drivers/infiniband/hw/mlx4/mad.c 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/infiniband/hw/mlx4/mad.c 2010-09-01
15:33:48.571462204 +0300
@@ -229,9 +229,9 @@ static void forward_trap(struct mlx4_ib_
}
}
-int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8
port_num,
- struct ib_wc *in_wc, struct ib_grh *in_grh,
- struct ib_mad *in_mad, struct ib_mad *out_mad)
+static int ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num,
+ struct ib_wc *in_wc, struct ib_grh *in_grh,
+ struct ib_mad *in_mad, struct ib_mad *out_mad)
{
u16 slid, prev_lid = 0;
int err;
@@ -299,6 +299,87 @@ int mlx4_ib_process_mad(struct ib_device
return IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY;
}
+static __be32 be64_to_be32(__be64 b64)
+{
+ return cpu_to_be32(be64_to_cpu(b64) & 0xffffffff);
+}
+
+static void edit_counters(struct mlx4_counters *cnt, void *data)
+{
+ *(__be32 *)(data + 40 + 24) = be64_to_be32(cnt->tx_bytes);
+ *(__be32 *)(data + 40 + 28) = be64_to_be32(cnt->rx_bytes);
+ *(__be32 *)(data + 40 + 32) = be64_to_be32(cnt->tx_frames);
+ *(__be32 *)(data + 40 + 36) = be64_to_be32(cnt->rx_frames);
+}
+
+static void edit_ext_counters(struct mlx4_counters_ext *cnt, void *data)
+{
+ *(__be32 *)(data + 40 + 24) = be64_to_be32(cnt->tx_uni_bytes);
+ *(__be32 *)(data + 40 + 28) = be64_to_be32(cnt->rx_uni_bytes);
+ *(__be32 *)(data + 40 + 32) = be64_to_be32(cnt->tx_uni_frames);
+ *(__be32 *)(data + 40 + 36) = be64_to_be32(cnt->rx_uni_frames);
+ *(__be32 *)(data + 40 + 8) = be64_to_be32(cnt->rx_err_frames);
+}
+
+static int rdmaoe_process_mad(struct ib_device *ibdev, int mad_flags, u8
port_num,
+ struct ib_wc *in_wc, struct ib_grh *in_grh,
+ struct ib_mad *in_mad, struct ib_mad *out_mad)
+{
+ struct mlx4_cmd_mailbox *mailbox;
+ struct mlx4_ib_dev *dev = to_mdev(ibdev);
+ int err;
+ u32 inmod = dev->counters[port_num - 1] & 0xffff;
+ int mode;
+
+ if (in_mad->mad_hdr.mgmt_class != IB_MGMT_CLASS_PERF_MGMT)
+ return -EINVAL;
+
+ mailbox = mlx4_alloc_cmd_mailbox(dev->dev);
+ if (IS_ERR(mailbox))
+ return IB_MAD_RESULT_FAILURE;
+
+ err = mlx4_cmd_box(dev->dev, 0, mailbox->dma, inmod, 0,
+ MLX4_CMD_QUERY_IF_STAT, MLX4_CMD_TIME_CLASS_C);
+ if (err)
+ err = IB_MAD_RESULT_FAILURE;
+ else {
+ memset(out_mad->data, 0, sizeof out_mad->data);
+ mode = be32_to_cpu(((struct mlx4_counters
*)mailbox->buf)->counter_mode) & 0xf;
+ switch (mode) {
+ case 0:
+ edit_counters(mailbox->buf, out_mad->data);
+ err = IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY;
+ break;
+ case 1:
+ edit_ext_counters(mailbox->buf, out_mad->data);
+ err = IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY;
+ break;
+ default:
+ err = IB_MAD_RESULT_FAILURE;
+ }
+ }
+
+ mlx4_free_cmd_mailbox(dev->dev, mailbox);
+
+ return err;
+}
+
+int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags, u8
port_num,
+ struct ib_wc *in_wc, struct ib_grh *in_grh,
+ struct ib_mad *in_mad, struct ib_mad *out_mad)
+{
+ switch (rdma_port_link_layer(ibdev, port_num)) {
+ case IB_LINK_LAYER_INFINIBAND:
+ return ib_process_mad(ibdev, mad_flags, port_num, in_wc,
+ in_grh, in_mad, out_mad);
+ case IB_LINK_LAYER_ETHERNET:
+ return rdmaoe_process_mad(ibdev, mad_flags, port_num, in_wc,
+ in_grh, in_mad, out_mad);
+ default:
+ return -EINVAL;
+ }
+}
+
static void send_handler(struct ib_mad_agent *agent,
struct ib_mad_send_wc *mad_send_wc)
{
Index: ofed_kernel-fixes/drivers/infiniband/hw/mlx4/main.c
===================================================================
--- ofed_kernel-fixes.orig/drivers/infiniband/hw/mlx4/main.c 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/infiniband/hw/mlx4/main.c 2010-09-01
15:33:05.427464524 +0300
@@ -1169,6 +1169,7 @@ static void *mlx4_ib_add(struct mlx4_dev
int i;
int err;
struct mlx4_ib_rocee *rocee;
+ int k;
if (!mlx4_ib_version_printed) {
printk(KERN_INFO "%s", mlx4_ib_version);
@@ -1309,12 +1310,18 @@ static void *mlx4_ib_add(struct mlx4_dev
if (init_node_data(ibdev))
goto err_map;
+ for (k = 0; k < ibdev->num_ports; ++k) {
+ err = mlx4_counter_alloc(ibdev->dev, &ibdev->counters[k]);
+ if (err)
+ ibdev->counters[k] = -1;
+ }
+
spin_lock_init(&ibdev->sm_lock);
mutex_init(&ibdev->cap_mask_mutex);
mutex_init(&ibdev->xrc_reg_mutex);
if (ib_register_device(&ibdev->ib_dev))
- goto err_map;
+ goto err_counter;
if (mlx4_ib_mad_init(ibdev))
goto err_reg;
@@ -1332,7 +1339,7 @@ static void *mlx4_ib_add(struct mlx4_dev
}
if(sysfs_create_group(&ibdev->ib_dev.dev.kobj, &diag_counters_group))
- goto err_reg;
+ goto err_notif;
ibdev->ib_active = 1;
@@ -1346,6 +1353,10 @@ err_notif:
err_reg:
ib_unregister_device(&ibdev->ib_dev);
+err_counter:
+ for (; k; --k)
+ mlx4_counter_free(ibdev->dev, ibdev->counters[k - 1]);
+
err_map:
iounmap(ibdev->uar_map);
@@ -1365,11 +1376,15 @@ static void mlx4_ib_remove(struct mlx4_d
{
struct mlx4_ib_dev *ibdev = ibdev_ptr;
int p;
+ int k;
sysfs_remove_group(&ibdev->ib_dev.dev.kobj, &diag_counters_group);
mlx4_ib_mad_cleanup(ibdev);
ib_unregister_device(&ibdev->ib_dev);
+ for (k = 0; k < ibdev->num_ports; ++k)
+ mlx4_counter_free(ibdev->dev, ibdev->counters[k]);
+
if (ibdev->rocee.nb.notifier_call) {
unregister_netdevice_notifier(&ibdev->rocee.nb);
flush_workqueue(wq);
Index: ofed_kernel-fixes/drivers/infiniband/hw/mlx4/mlx4_ib.h
===================================================================
--- ofed_kernel-fixes.orig/drivers/infiniband/hw/mlx4/mlx4_ib.h 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/infiniband/hw/mlx4/mlx4_ib.h 2010-09-01
15:33:05.043465489 +0300
@@ -230,6 +230,7 @@ struct mlx4_ib_dev {
struct mutex xrc_reg_mutex;
int ib_active;
struct mlx4_ib_rocee rocee;
+ int counters[MLX4_MAX_PORTS];
};
static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev)
Index: ofed_kernel-fixes/drivers/infiniband/hw/mlx4/qp.c
===================================================================
--- ofed_kernel-fixes.orig/drivers/infiniband/hw/mlx4/qp.c 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/infiniband/hw/mlx4/qp.c 2010-09-01
15:33:05.391464429 +0300
@@ -967,7 +967,6 @@ static int mlx4_set_path(struct mlx4_ib_
--path->static_rate;
} else
path->static_rate = 0;
- path->counter_index = 0xff;
if (ah->ah_flags & IB_AH_GRH) {
if (ah->grh.sgid_index >= dev->dev->caps.gid_table_len[port]) {
@@ -1101,6 +1100,12 @@ static int __mlx4_ib_modify_qp(struct ib
}
}
+ if (cur_state == IB_QPS_INIT && new_state == IB_QPS_RTR &&
+ dev->counters[qp->port - 1] != -1) {
+ context->pri_path.counter_index = dev->counters[qp->port - 1];
+ optpar |= MLX4_QP_OPTPAR_COUNTER_INDEX;
+ }
+
if (attr_mask & IB_QP_PKEY_INDEX) {
context->pri_path.pkey_index = attr->pkey_index;
optpar |= MLX4_QP_OPTPAR_PKEY_INDEX;
Index: ofed_kernel-fixes/drivers/net/mlx4/fw.h
===================================================================
--- ofed_kernel-fixes.orig/drivers/net/mlx4/fw.h 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/net/mlx4/fw.h 2010-09-01 15:33:05.199464059
+0300
@@ -146,7 +146,6 @@ struct mlx4_init_hca_param {
u8 log_mc_table_sz;
u8 log_mpt_sz;
u8 log_uar_sz;
- u8 counter_mode;
};
struct mlx4_init_ib_param {
Index: ofed_kernel-fixes/drivers/net/mlx4/main.c
===================================================================
--- ofed_kernel-fixes.orig/drivers/net/mlx4/main.c 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/drivers/net/mlx4/main.c 2010-09-01 15:33:05.219464453
+0300
@@ -1009,17 +1009,29 @@ int mlx4_counter_alloc(struct mlx4_dev *
{
struct mlx4_priv *priv = mlx4_priv(dev);
- *idx = mlx4_bitmap_alloc(&priv->counters_bitmap);
- if (*idx == -1)
+ switch (dev->caps.counters_mode) {
+ case MLX4_CUNTERS_BASIC:
+ case MLX4_CUNTERS_EXT:
+ *idx = mlx4_bitmap_alloc(&priv->counters_bitmap);
+ if (*idx == -1)
+ return -ENOMEM;
+ return 0;
+ default:
return -ENOMEM;
-
- return 0;
+ }
}
EXPORT_SYMBOL_GPL(mlx4_counter_alloc);
void mlx4_counter_free(struct mlx4_dev *dev, u32 idx)
{
- mlx4_bitmap_free(&mlx4_priv(dev)->counters_bitmap, idx);
+ switch (dev->caps.counters_mode) {
+ case MLX4_CUNTERS_BASIC:
+ case MLX4_CUNTERS_EXT:
+ mlx4_bitmap_free(&mlx4_priv(dev)->counters_bitmap, idx);
+ return;
+ default:
+ return;
+ }
}
EXPORT_SYMBOL_GPL(mlx4_counter_free);
Index: ofed_kernel-fixes/include/linux/mlx4/cmd.h
===================================================================
--- ofed_kernel-fixes.orig/include/linux/mlx4/cmd.h 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/include/linux/mlx4/cmd.h 2010-09-01 15:30:01.000000000
+0300
@@ -122,6 +122,10 @@ enum {
/* debug commands */
MLX4_CMD_QUERY_DEBUG_MSG = 0x2a,
MLX4_CMD_SET_DEBUG_MSG = 0x2b,
+
+ /* statistics commands */
+ MLX4_CMD_QUERY_IF_STAT = 0X54,
+ MLX4_CMD_SET_IF_STAT = 0X55,
};
enum {
Index: ofed_kernel-fixes/include/linux/mlx4/device.h
===================================================================
--- ofed_kernel-fixes.orig/include/linux/mlx4/device.h 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/include/linux/mlx4/device.h 2010-09-01
15:33:05.375463893 +0300
@@ -411,6 +411,42 @@ union mlx4_ext_av {
struct mlx4_eth_av eth;
};
+struct mlx4_counters {
+ __be32 counter_mode;
+ __be32 num_ifc;
+ u32 reserved[2];
+ __be64 rx_frames;
+ __be64 rx_bytes;
+ __be64 tx_frames;
+ __be64 tx_bytes;
+};
+
+struct mlx4_counters_ext {
+ __be32 counter_mode;
+ __be32 num_ifc;
+ u32 reserved[2];
+ __be64 rx_uni_frames;
+ __be64 rx_uni_bytes;
+ __be64 rx_mcast_frames;
+ __be64 rx_mcast_bytes;
+ __be64 rx_bcast_frames;
+ __be64 rx_bcast_bytes;
+ __be64 rx_nobuf_frames;
+ __be64 rx_nobuf_bytes;
+ __be64 rx_err_frames;
+ __be64 rx_err_bytes;
+ __be64 tx_uni_frames;
+ __be64 tx_uni_bytes;
+ __be64 tx_mcast_frames;
+ __be64 tx_mcast_bytes;
+ __be64 tx_bcast_frames;
+ __be64 tx_bcast_bytes;
+ __be64 tx_nobuf_frames;
+ __be64 tx_nobuf_bytes;
+ __be64 tx_err_frames;
+ __be64 tx_err_bytes;
+};
+
struct mlx4_dev {
struct pci_dev *pdev;
unsigned long flags;
Index: ofed_kernel-fixes/include/linux/mlx4/qp.h
===================================================================
--- ofed_kernel-fixes.orig/include/linux/mlx4/qp.h 2010-09-01
15:30:01.000000000 +0300
+++ ofed_kernel-fixes/include/linux/mlx4/qp.h 2010-09-01 15:33:05.359464345
+0300
@@ -54,7 +54,8 @@ enum mlx4_qp_optpar {
MLX4_QP_OPTPAR_RETRY_COUNT = 1 << 12,
MLX4_QP_OPTPAR_RNR_RETRY = 1 << 13,
MLX4_QP_OPTPAR_ACK_TIMEOUT = 1 << 14,
- MLX4_QP_OPTPAR_SCHED_QUEUE = 1 << 16
+ MLX4_QP_OPTPAR_SCHED_QUEUE = 1 << 16,
+ MLX4_QP_OPTPAR_COUNTER_INDEX = 1 << 20
};
enum mlx4_qp_state {
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
