[PATCH net-next 0/7] Mellanox driver update for net-next

[PATCH net-next 0/7] Mellanox driver update for net-next

[Or Gerlitz]

Hi Dave,

Some small fixes and small enhancements from the team.

Series applies over net-next commit acb4a6b "tcp: ensure prior synack rtx behavior
with small backlogs".

Or.

Eli Cohen (4):
  net/mlx5_core: Improve mlx5 messages
  net/mlx5_core: Prepare cmd interface to system errors handling
  net/mlx5_core: Use accessor functions to read from device memory
  net/mlx5_core: Use private health thread for each device

Ido Shamay (1):
  net/mlx4_en: Add steering rules after RSS creation

Jack Morgenstein (1):
  net/mlx4_core: Fix mailbox leak in error flow when performing update qp

Saeed Mahameed (1):
  net/mlx4_core: Fix resource tracker error flow in add_res_range

 drivers/net/ethernet/mellanox/mlx4/en_netdev.c     | 144 ++++++++++++---------
 drivers/net/ethernet/mellanox/mlx4/qp.c            |   6 +-
 .../net/ethernet/mellanox/mlx4/resource_tracker.c  |   4 +-
 drivers/net/ethernet/mellanox/mlx5/core/cmd.c      |  20 ++-
 drivers/net/ethernet/mellanox/mlx5/core/health.c   |  98 +++++++-------
 drivers/net/ethernet/mellanox/mlx5/core/main.c     |  37 +++---
 .../net/ethernet/mellanox/mlx5/core/mlx5_core.h    |  22 ++--
 include/linux/mlx5/driver.h                        |  13 +-
 8 files changed, 194 insertions(+), 150 deletions(-)

-- 
2.3.7

[PATCH net-next 1/7] net/mlx5_core: Improve mlx5 messages

[Or Gerlitz]

From: Eli Cohen <eli@mellanox.com>

Improve the messages printed by the mlx5 macros to include the device
string. In addition, prefix names used by the macros with two underscores
to avoid possible name collisions.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 .../net/ethernet/mellanox/mlx5/core/mlx5_core.h    | 22 +++++++++++-----------
 1 file changed, 11 insertions(+), 11 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
index 566a704..30c0be7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
+++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h
@@ -43,25 +43,25 @@
 
 extern int mlx5_core_debug_mask;
 
-#define mlx5_core_dbg(dev, format, ...)					\
-	pr_debug("%s:%s:%d:(pid %d): " format,				\
-		 (dev)->priv.name, __func__, __LINE__, current->pid,	\
+#define mlx5_core_dbg(__dev, format, ...)				\
+	dev_dbg(&(__dev)->pdev->dev, "%s:%s:%d:(pid %d): " format,	\
+		 (__dev)->priv.name, __func__, __LINE__, current->pid,	\
 		 ##__VA_ARGS__)
 
-#define mlx5_core_dbg_mask(dev, mask, format, ...)			\
+#define mlx5_core_dbg_mask(__dev, mask, format, ...)			\
 do {									\
 	if ((mask) & mlx5_core_debug_mask)				\
-		mlx5_core_dbg(dev, format, ##__VA_ARGS__);		\
+		mlx5_core_dbg(__dev, format, ##__VA_ARGS__);		\
 } while (0)
 
-#define mlx5_core_err(dev, format, ...)					\
-	pr_err("%s:%s:%d:(pid %d): " format,				\
-	       (dev)->priv.name, __func__, __LINE__, current->pid,	\
+#define mlx5_core_err(__dev, format, ...)				\
+	dev_err(&(__dev)->pdev->dev, "%s:%s:%d:(pid %d): " format,	\
+	       (__dev)->priv.name, __func__, __LINE__, current->pid,	\
 	       ##__VA_ARGS__)
 
-#define mlx5_core_warn(dev, format, ...)				\
-	pr_warn("%s:%s:%d:(pid %d): " format,				\
-		(dev)->priv.name, __func__, __LINE__, current->pid,	\
+#define mlx5_core_warn(__dev, format, ...)				\
+	dev_warn(&(__dev)->pdev->dev, "%s:%s:%d:(pid %d): " format,	\
+		(__dev)->priv.name, __func__, __LINE__, current->pid,	\
 		##__VA_ARGS__)
 
 enum {
-- 
2.3.7

[PATCH net-next 2/7] net/mlx5_core: Prepare cmd interface to system errors handling

[Or Gerlitz]

From: Eli Cohen <eli@mellanox.com>

In preparation to handling system errors at the mlx5_core level, change the
interface of cmd_work_handler to accept a 64 bit argument for the vector.

This allows to encode a flag that signifies when the handler is called
as a result of a driver logic that wishes to terminate commands that
the hardware may not be able to terminate. Such command completions
are detected at the handler and proper return status is encoded.

To be able to terminate page handler commands, we make sure to set
the corresponding bit in the bitmask.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 20 ++++++++++++++++++--
 include/linux/mlx5/driver.h                   |  6 +++++-
 2 files changed, 23 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
index 84838c2..c3e54b7 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c
@@ -254,6 +254,10 @@ static void dump_buf(void *buf, int size, int data_only, int offset)
 		pr_debug("\n");
 }
 
+enum {
+	MLX5_DRIVER_STATUS_ABORTED = 0xfe,
+};
+
 const char *mlx5_command_str(int command)
 {
 	switch (command) {
@@ -473,6 +477,7 @@ static void cmd_work_handler(struct work_struct *work)
 	struct mlx5_core_dev *dev = container_of(cmd, struct mlx5_core_dev, cmd);
 	struct mlx5_cmd_layout *lay;
 	struct semaphore *sem;
+	unsigned long flags;
 
 	sem = ent->page_queue ? &cmd->pages_sem : &cmd->sem;
 	down(sem);
@@ -485,6 +490,9 @@ static void cmd_work_handler(struct work_struct *work)
 		}
 	} else {
 		ent->idx = cmd->max_reg_cmds;
+		spin_lock_irqsave(&cmd->alloc_lock, flags);
+		clear_bit(ent->idx, &cmd->bitmask);
+		spin_unlock_irqrestore(&cmd->alloc_lock, flags);
 	}
 
 	ent->token = alloc_token(cmd);
@@ -1081,7 +1089,7 @@ static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg)
 	}
 }
 
-void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector)
+void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec)
 {
 	struct mlx5_cmd *cmd = &dev->cmd;
 	struct mlx5_cmd_work_ent *ent;
@@ -1092,7 +1100,10 @@ void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector)
 	s64 ds;
 	struct mlx5_cmd_stats *stats;
 	unsigned long flags;
+	unsigned long vector;
 
+	/* there can be at most 32 command queues */
+	vector = vec & 0xffffffff;
 	for (i = 0; i < (1 << cmd->log_sz); i++) {
 		if (test_bit(i, &vector)) {
 			struct semaphore *sem;
@@ -1110,11 +1121,16 @@ void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector)
 					ent->ret = verify_signature(ent);
 				else
 					ent->ret = 0;
-				ent->status = ent->lay->status_own >> 1;
+				if (vec & MLX5_TRIGGERED_CMD_COMP)
+					ent->status = MLX5_DRIVER_STATUS_ABORTED;
+				else
+					ent->status = ent->lay->status_own >> 1;
+
 				mlx5_core_dbg(dev, "command completed. ret 0x%x, delivery status %s(0x%x)\n",
 					      ent->ret, deliv_status_to_str(ent->status), ent->status);
 			}
 			free_ent(cmd, ent->idx);
+
 			if (ent->callback) {
 				ds = ent->ts2 - ent->ts1;
 				if (ent->op < ARRAY_SIZE(cmd->stats)) {
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index 8b6d6f2..aa89955 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -731,7 +731,7 @@ void mlx5_eq_pagefault(struct mlx5_core_dev *dev, struct mlx5_eqe *eqe);
 #endif
 void mlx5_srq_event(struct mlx5_core_dev *dev, u32 srqn, int event_type);
 struct mlx5_core_srq *mlx5_core_get_srq(struct mlx5_core_dev *dev, u32 srqn);
-void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, unsigned long vector);
+void mlx5_cmd_comp_handler(struct mlx5_core_dev *dev, u64 vec);
 void mlx5_cq_event(struct mlx5_core_dev *dev, u32 cqn, int event_type);
 int mlx5_create_map_eq(struct mlx5_core_dev *dev, struct mlx5_eq *eq, u8 vecidx,
 		       int nent, u64 mask, const char *name, struct mlx5_uar *uar);
@@ -865,4 +865,8 @@ static inline int mlx5_get_gid_table_len(u16 param)
 	return 8 * (1 << param);
 }
 
+enum {
+	MLX5_TRIGGERED_CMD_COMP = (u64)1 << 32,
+};
+
 #endif /* MLX5_DRIVER_H */
-- 
2.3.7

[PATCH net-next 3/7] net/mlx5_core: Use accessor functions to read from device memory

[Or Gerlitz]

From: Eli Cohen <eli@mellanox.com>

Use ioread function to read health buffer data. In addition, print the
firmware version as a string for readability and also use dev_err to have
the device string to be printed.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/health.c | 35 +++++++++++++++---------
 1 file changed, 22 insertions(+), 13 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
index 1277c42..8770968 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/health.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -114,32 +114,41 @@ static const char *hsynd_str(u8 synd)
 	}
 }
 
-static u16 read_be16(__be16 __iomem *p)
+static u16 get_maj(u32 fw)
 {
-	return swab16(readl((__force u16 __iomem *) p));
+	return fw >> 28;
 }
 
-static u32 read_be32(__be32 __iomem *p)
+static u16 get_min(u32 fw)
 {
-	return swab32(readl((__force u32 __iomem *) p));
+	return fw >> 16 & 0xfff;
+}
+
+static u16 get_sub(u32 fw)
+{
+	return fw & 0xffff;
 }
 
 static void print_health_info(struct mlx5_core_dev *dev)
 {
 	struct mlx5_core_health *health = &dev->priv.health;
 	struct health_buffer __iomem *h = health->health;
+	char fw_str[18];
+	u32 fw;
 	int i;
 
 	for (i = 0; i < ARRAY_SIZE(h->assert_var); i++)
-		pr_info("assert_var[%d] 0x%08x\n", i, read_be32(h->assert_var + i));
-
-	pr_info("assert_exit_ptr 0x%08x\n", read_be32(&h->assert_exit_ptr));
-	pr_info("assert_callra 0x%08x\n", read_be32(&h->assert_callra));
-	pr_info("fw_ver 0x%08x\n", read_be32(&h->fw_ver));
-	pr_info("hw_id 0x%08x\n", read_be32(&h->hw_id));
-	pr_info("irisc_index %d\n", readb(&h->irisc_index));
-	pr_info("synd 0x%x: %s\n", readb(&h->synd), hsynd_str(readb(&h->synd)));
-	pr_info("ext_sync 0x%04x\n", read_be16(&h->ext_synd));
+		dev_err(&dev->pdev->dev, "assert_var[%d] 0x%08x\n", i, ioread32be(h->assert_var + i));
+
+	dev_err(&dev->pdev->dev, "assert_exit_ptr 0x%08x\n", ioread32be(&h->assert_exit_ptr));
+	dev_err(&dev->pdev->dev, "assert_callra 0x%08x\n", ioread32be(&h->assert_callra));
+	fw = ioread32be(&h->fw_ver);
+	sprintf(fw_str, "%d.%d.%d", get_maj(fw), get_min(fw), get_sub(fw));
+	dev_err(&dev->pdev->dev, "fw_ver %s\n", fw_str);
+	dev_err(&dev->pdev->dev, "hw_id 0x%08x\n", ioread32be(&h->hw_id));
+	dev_err(&dev->pdev->dev, "irisc_index %d\n", ioread8(&h->irisc_index));
+	dev_err(&dev->pdev->dev, "synd 0x%x: %s\n", ioread8(&h->synd), hsynd_str(ioread8(&h->synd)));
+	dev_err(&dev->pdev->dev, "ext_synd 0x%04x\n", ioread16be(&h->ext_synd));
 }
 
 static void poll_health(unsigned long data)
-- 
2.3.7

[PATCH net-next 4/7] net/mlx5_core: Use private health thread for each device

[Or Gerlitz]

From: Eli Cohen <eli@mellanox.com>

Use a single threaded work queue for each device in the system instead of
using one thread for any device. This is required so we can concurrently
process system error handling for all the devices that need that.

Signed-off-by: Eli Cohen <eli@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx5/core/health.c | 63 +++++++++++-------------
 drivers/net/ethernet/mellanox/mlx5/core/main.c   | 37 +++++++-------
 include/linux/mlx5/driver.h                      |  7 +--
 3 files changed, 52 insertions(+), 55 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c
index 8770968..9b81e1c 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/health.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c
@@ -57,31 +57,16 @@ enum {
 	MLX5_HEALTH_SYNDR_HIGH_TEMP		= 0x10
 };
 
-static DEFINE_SPINLOCK(health_lock);
-static LIST_HEAD(health_list);
-static struct work_struct health_work;
-
 static void health_care(struct work_struct *work)
 {
-	struct mlx5_core_health *health, *n;
+	struct mlx5_core_health *health;
 	struct mlx5_core_dev *dev;
 	struct mlx5_priv *priv;
-	LIST_HEAD(tlist);
-
-	spin_lock_irq(&health_lock);
-	list_splice_init(&health_list, &tlist);
-
-	spin_unlock_irq(&health_lock);
 
-	list_for_each_entry_safe(health, n, &tlist, list) {
-		priv = container_of(health, struct mlx5_priv, health);
-		dev = container_of(priv, struct mlx5_core_dev, priv);
-		mlx5_core_warn(dev, "handling bad device here\n");
-		/* nothing yet */
-		spin_lock_irq(&health_lock);
-		list_del_init(&health->list);
-		spin_unlock_irq(&health_lock);
-	}
+	health = container_of(work, struct mlx5_core_health, work);
+	priv = container_of(health, struct mlx5_priv, health);
+	dev = container_of(priv, struct mlx5_core_dev, priv);
+	mlx5_core_warn(dev, "handling bad device here\n");
 }
 
 static const char *hsynd_str(u8 synd)
@@ -168,11 +153,7 @@ static void poll_health(unsigned long data)
 	if (health->miss_counter == MAX_MISSES) {
 		mlx5_core_err(dev, "device's health compromised\n");
 		print_health_info(dev);
-		spin_lock_irq(&health_lock);
-		list_add_tail(&health->list, &health_list);
-		spin_unlock_irq(&health_lock);
-
-		queue_work(mlx5_core_wq, &health_work);
+		queue_work(health->wq, &health->work);
 	} else {
 		get_random_bytes(&next, sizeof(next));
 		next %= HZ;
@@ -185,7 +166,6 @@ void mlx5_start_health_poll(struct mlx5_core_dev *dev)
 {
 	struct mlx5_core_health *health = &dev->priv.health;
 
-	INIT_LIST_HEAD(&health->list);
 	init_timer(&health->timer);
 	health->health = &dev->iseg->health;
 	health->health_counter = &dev->iseg->health_counter;
@@ -201,18 +181,33 @@ void mlx5_stop_health_poll(struct mlx5_core_dev *dev)
 	struct mlx5_core_health *health = &dev->priv.health;
 
 	del_timer_sync(&health->timer);
-
-	spin_lock_irq(&health_lock);
-	if (!list_empty(&health->list))
-		list_del_init(&health->list);
-	spin_unlock_irq(&health_lock);
 }
 
-void mlx5_health_cleanup(void)
+void mlx5_health_cleanup(struct mlx5_core_dev *dev)
 {
+	struct mlx5_core_health *health = &dev->priv.health;
+
+	destroy_workqueue(health->wq);
 }
 
-void  __init mlx5_health_init(void)
+int mlx5_health_init(struct mlx5_core_dev *dev)
 {
-	INIT_WORK(&health_work, health_care);
+	struct mlx5_core_health *health;
+	char *name;
+
+	health = &dev->priv.health;
+	name = kmalloc(64, GFP_KERNEL);
+	if (!name)
+		return -ENOMEM;
+
+	strcpy(name, "mlx5_health");
+	strcat(name, dev_name(&dev->pdev->dev));
+	health->wq = create_singlethread_workqueue(name);
+	kfree(name);
+	if (!health->wq)
+		return -ENOMEM;
+
+	INIT_WORK(&health->work, health_care);
+
+	return 0;
 }
diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c
index 7718f6a..b6edc58 100644
--- a/drivers/net/ethernet/mellanox/mlx5/core/main.c
+++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c
@@ -62,7 +62,6 @@ static int prof_sel = MLX5_DEFAULT_PROF;
 module_param_named(prof_sel, prof_sel, int, 0444);
 MODULE_PARM_DESC(prof_sel, "profile selector. Valid range 0 - 2");
 
-struct workqueue_struct *mlx5_core_wq;
 static LIST_HEAD(intf_list);
 static LIST_HEAD(dev_list);
 static DEFINE_MUTEX(intf_mutex);
@@ -1046,6 +1045,7 @@ err_pagealloc_cleanup:
 
 static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 {
+	int err;
 
 	mlx5_unregister_device(dev);
 	mlx5_cleanup_mr_table(dev);
@@ -1060,9 +1060,10 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 	mlx5_eq_cleanup(dev);
 	mlx5_disable_msix(dev);
 	mlx5_stop_health_poll(dev);
-	if (mlx5_cmd_teardown_hca(dev)) {
+	err = mlx5_cmd_teardown_hca(dev);
+	if (err) {
 		dev_err(&dev->pdev->dev, "tear_down_hca failed, skip cleanup\n");
-		return 1;
+		goto out;
 	}
 	mlx5_pagealloc_stop(dev);
 	mlx5_reclaim_startup_pages(dev);
@@ -1070,11 +1071,12 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv)
 	mlx5_pagealloc_cleanup(dev);
 	mlx5_cmd_cleanup(dev);
 
-	return 0;
+out:
+	return err;
 }
 
 static void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event,
-			    unsigned long param)
+		     unsigned long param)
 {
 	struct mlx5_priv *priv = &dev->priv;
 	struct mlx5_device_context *dev_ctx;
@@ -1129,14 +1131,22 @@ static int init_one(struct pci_dev *pdev,
 		goto clean_dev;
 	}
 
+	err = mlx5_health_init(dev);
+	if (err) {
+		dev_err(&pdev->dev, "mlx5_health_init failed with error code %d\n", err);
+		goto close_pci;
+	}
+
 	err = mlx5_load_one(dev, priv);
 	if (err) {
 		dev_err(&pdev->dev, "mlx5_load_one failed with error code %d\n", err);
-		goto close_pci;
+		goto clean_health;
 	}
 
 	return 0;
 
+clean_health:
+	mlx5_health_cleanup(dev);
 close_pci:
 	mlx5_pci_close(dev, priv);
 clean_dev:
@@ -1153,8 +1163,10 @@ static void remove_one(struct pci_dev *pdev)
 
 	if (mlx5_unload_one(dev, priv)) {
 		dev_err(&dev->pdev->dev, "mlx5_unload_one failed\n");
+		mlx5_health_cleanup(dev);
 		return;
 	}
+	mlx5_health_cleanup(dev);
 	mlx5_pci_close(dev, priv);
 	pci_set_drvdata(pdev, NULL);
 	kfree(dev);
@@ -1184,16 +1196,10 @@ static int __init init(void)
 	int err;
 
 	mlx5_register_debugfs();
-	mlx5_core_wq = create_singlethread_workqueue("mlx5_core_wq");
-	if (!mlx5_core_wq) {
-		err = -ENOMEM;
-		goto err_debug;
-	}
-	mlx5_health_init();
 
 	err = pci_register_driver(&mlx5_core_driver);
 	if (err)
-		goto err_health;
+		goto err_debug;
 
 #ifdef CONFIG_MLX5_CORE_EN
 	mlx5e_init();
@@ -1201,9 +1207,6 @@ static int __init init(void)
 
 	return 0;
 
-err_health:
-	mlx5_health_cleanup();
-	destroy_workqueue(mlx5_core_wq);
 err_debug:
 	mlx5_unregister_debugfs();
 	return err;
@@ -1215,8 +1218,6 @@ static void __exit cleanup(void)
 	mlx5e_cleanup();
 #endif
 	pci_unregister_driver(&mlx5_core_driver);
-	mlx5_health_cleanup();
-	destroy_workqueue(mlx5_core_wq);
 	mlx5_unregister_debugfs();
 }
 
diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h
index aa89955..41a3287 100644
--- a/include/linux/mlx5/driver.h
+++ b/include/linux/mlx5/driver.h
@@ -391,9 +391,10 @@ struct mlx5_core_health {
 	struct health_buffer __iomem   *health;
 	__be32 __iomem		       *health_counter;
 	struct timer_list		timer;
-	struct list_head		list;
 	u32				prev;
 	int				miss_counter;
+	struct workqueue_struct	       *wq;
+	struct work_struct		work;
 };
 
 struct mlx5_cq_table {
@@ -676,8 +677,8 @@ int mlx5_alloc_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari);
 int mlx5_free_uuars(struct mlx5_core_dev *dev, struct mlx5_uuar_info *uuari);
 int mlx5_alloc_map_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar);
 void mlx5_unmap_free_uar(struct mlx5_core_dev *mdev, struct mlx5_uar *uar);
-void mlx5_health_cleanup(void);
-void  __init mlx5_health_init(void);
+void mlx5_health_cleanup(struct mlx5_core_dev *dev);
+int mlx5_health_init(struct mlx5_core_dev *dev);
 void mlx5_start_health_poll(struct mlx5_core_dev *dev);
 void mlx5_stop_health_poll(struct mlx5_core_dev *dev);
 int mlx5_buf_alloc_node(struct mlx5_core_dev *dev, int size,
-- 
2.3.7

[PATCH net-next 5/7] net/mlx4_en: Add steering rules after RSS creation

[Or Gerlitz]

From: Ido Shamay <idos@mellanox.com>

Changed the receive control flow in a way that steering
rules are added only when the RSS object is already in RTR/RTS mode.
Some optimization features, which are enabled by the device firmware,
require this condition in order to be effective.

Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx4/en_netdev.c | 144 ++++++++++++++-----------
 1 file changed, 80 insertions(+), 64 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
index 4726122..597d892 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_netdev.c
@@ -573,10 +573,8 @@ static int mlx4_en_get_qp(struct mlx4_en_priv *priv)
 {
 	struct mlx4_en_dev *mdev = priv->mdev;
 	struct mlx4_dev *dev = mdev->dev;
-	struct mlx4_mac_entry *entry;
 	int index = 0;
 	int err = 0;
-	u64 reg_id = 0;
 	int *qpn = &priv->base_qpn;
 	u64 mac = mlx4_mac_to_u64(priv->dev->dev_addr);
 
@@ -600,44 +598,11 @@ static int mlx4_en_get_qp(struct mlx4_en_priv *priv)
 	en_dbg(DRV, priv, "Reserved qp %d\n", *qpn);
 	if (err) {
 		en_err(priv, "Failed to reserve qp for mac registration\n");
-		goto qp_err;
-	}
-
-	err = mlx4_en_uc_steer_add(priv, priv->dev->dev_addr, qpn, &reg_id);
-	if (err)
-		goto steer_err;
-
-	err = mlx4_en_tunnel_steer_add(priv, priv->dev->dev_addr, *qpn,
-				       &priv->tunnel_reg_id);
-	if (err)
-		goto tunnel_err;
-
-	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
-	if (!entry) {
-		err = -ENOMEM;
-		goto alloc_err;
+		mlx4_unregister_mac(dev, priv->port, mac);
+		return err;
 	}
-	memcpy(entry->mac, priv->dev->dev_addr, sizeof(entry->mac));
-	memcpy(priv->current_mac, entry->mac, sizeof(priv->current_mac));
-	entry->reg_id = reg_id;
-
-	hlist_add_head_rcu(&entry->hlist,
-			   &priv->mac_hash[entry->mac[MLX4_EN_MAC_HASH_IDX]]);
 
 	return 0;
-
-alloc_err:
-	if (priv->tunnel_reg_id)
-		mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
-tunnel_err:
-	mlx4_en_uc_steer_release(priv, priv->dev->dev_addr, *qpn, reg_id);
-
-steer_err:
-	mlx4_qp_release_range(dev, *qpn, 1);
-
-qp_err:
-	mlx4_unregister_mac(dev, priv->port, mac);
-	return err;
 }
 
 static void mlx4_en_put_qp(struct mlx4_en_priv *priv)
@@ -645,39 +610,13 @@ static void mlx4_en_put_qp(struct mlx4_en_priv *priv)
 	struct mlx4_en_dev *mdev = priv->mdev;
 	struct mlx4_dev *dev = mdev->dev;
 	int qpn = priv->base_qpn;
-	u64 mac;
 
 	if (dev->caps.steering_mode == MLX4_STEERING_MODE_A0) {
-		mac = mlx4_mac_to_u64(priv->dev->dev_addr);
+		u64 mac = mlx4_mac_to_u64(priv->dev->dev_addr);
 		en_dbg(DRV, priv, "Registering MAC: %pM for deleting\n",
 		       priv->dev->dev_addr);
 		mlx4_unregister_mac(dev, priv->port, mac);
 	} else {
-		struct mlx4_mac_entry *entry;
-		struct hlist_node *tmp;
-		struct hlist_head *bucket;
-		unsigned int i;
-
-		for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) {
-			bucket = &priv->mac_hash[i];
-			hlist_for_each_entry_safe(entry, tmp, bucket, hlist) {
-				mac = mlx4_mac_to_u64(entry->mac);
-				en_dbg(DRV, priv, "Registering MAC: %pM for deleting\n",
-				       entry->mac);
-				mlx4_en_uc_steer_release(priv, entry->mac,
-							 qpn, entry->reg_id);
-
-				mlx4_unregister_mac(dev, priv->port, mac);
-				hlist_del_rcu(&entry->hlist);
-				kfree_rcu(entry, rcu);
-			}
-		}
-
-		if (priv->tunnel_reg_id) {
-			mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
-			priv->tunnel_reg_id = 0;
-		}
-
 		en_dbg(DRV, priv, "Releasing qp: port %d, qpn %d\n",
 		       priv->port, qpn);
 		mlx4_qp_release_range(dev, qpn, 1);
@@ -1283,6 +1222,75 @@ static void mlx4_en_netpoll(struct net_device *dev)
 }
 #endif
 
+static int mlx4_en_set_rss_steer_rules(struct mlx4_en_priv *priv)
+{
+	u64 reg_id;
+	int err = 0;
+	int *qpn = &priv->base_qpn;
+	struct mlx4_mac_entry *entry;
+
+	err = mlx4_en_uc_steer_add(priv, priv->dev->dev_addr, qpn, &reg_id);
+	if (err)
+		return err;
+
+	err = mlx4_en_tunnel_steer_add(priv, priv->dev->dev_addr, *qpn,
+				       &priv->tunnel_reg_id);
+	if (err)
+		goto tunnel_err;
+
+	entry = kmalloc(sizeof(*entry), GFP_KERNEL);
+	if (!entry) {
+		err = -ENOMEM;
+		goto alloc_err;
+	}
+
+	memcpy(entry->mac, priv->dev->dev_addr, sizeof(entry->mac));
+	memcpy(priv->current_mac, entry->mac, sizeof(priv->current_mac));
+	entry->reg_id = reg_id;
+	hlist_add_head_rcu(&entry->hlist,
+			   &priv->mac_hash[entry->mac[MLX4_EN_MAC_HASH_IDX]]);
+
+	return 0;
+
+alloc_err:
+	if (priv->tunnel_reg_id)
+		mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
+
+tunnel_err:
+	mlx4_en_uc_steer_release(priv, priv->dev->dev_addr, *qpn, reg_id);
+	return err;
+}
+
+static void mlx4_en_delete_rss_steer_rules(struct mlx4_en_priv *priv)
+{
+	u64 mac;
+	unsigned int i;
+	int qpn = priv->base_qpn;
+	struct hlist_head *bucket;
+	struct hlist_node *tmp;
+	struct mlx4_mac_entry *entry;
+
+	for (i = 0; i < MLX4_EN_MAC_HASH_SIZE; ++i) {
+		bucket = &priv->mac_hash[i];
+		hlist_for_each_entry_safe(entry, tmp, bucket, hlist) {
+			mac = mlx4_mac_to_u64(entry->mac);
+			en_dbg(DRV, priv, "Registering MAC:%pM for deleting\n",
+			       entry->mac);
+			mlx4_en_uc_steer_release(priv, entry->mac,
+						 qpn, entry->reg_id);
+
+			mlx4_unregister_mac(priv->mdev->dev, priv->port, mac);
+			hlist_del_rcu(&entry->hlist);
+			kfree_rcu(entry, rcu);
+		}
+	}
+
+	if (priv->tunnel_reg_id) {
+		mlx4_flow_detach(priv->mdev->dev, priv->tunnel_reg_id);
+		priv->tunnel_reg_id = 0;
+	}
+}
+
 static void mlx4_en_tx_timeout(struct net_device *dev)
 {
 	struct mlx4_en_priv *priv = netdev_priv(dev);
@@ -1684,6 +1692,11 @@ int mlx4_en_start_port(struct net_device *dev)
 		goto tx_err;
 	}
 
+	/* Set Unicast and VXLAN steering rules */
+	if (mdev->dev->caps.steering_mode != MLX4_STEERING_MODE_A0 &&
+	    mlx4_en_set_rss_steer_rules(priv))
+		mlx4_warn(mdev, "Failed setting steering rules\n");
+
 	/* Attach rx QP to bradcast address */
 	eth_broadcast_addr(&mc_list[10]);
 	mc_list[5] = priv->port; /* needed for B0 steering support */
@@ -1831,6 +1844,9 @@ void mlx4_en_stop_port(struct net_device *dev, int detach)
 	for (i = 0; i < priv->tx_ring_num; i++)
 		mlx4_en_free_tx_buf(dev, priv->tx_ring[i]);
 
+	if (mdev->dev->caps.steering_mode != MLX4_STEERING_MODE_A0)
+		mlx4_en_delete_rss_steer_rules(priv);
+
 	/* Free RSS qps */
 	mlx4_en_release_rss_steer(priv);
 
-- 
2.3.7

[PATCH net-next 6/7] net/mlx4_core: Fix mailbox leak in error flow when performing update qp

[Or Gerlitz]

From: Jack Morgenstein <jackm@dev.mellanox.co.il>

The procedure mlx4_update_qp leaks mailboxes in its error-flow, fix that.

Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx4/qp.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/mellanox/mlx4/qp.c b/drivers/net/ethernet/mellanox/mlx4/qp.c
index 2026863..3311f35 100644
--- a/drivers/net/ethernet/mellanox/mlx4/qp.c
+++ b/drivers/net/ethernet/mellanox/mlx4/qp.c
@@ -422,15 +422,15 @@ int mlx4_update_qp(struct mlx4_dev *dev, u32 qpn,
 	u64 qp_mask = 0;
 	int err = 0;
 
+	if (!attr || (attr & ~MLX4_UPDATE_QP_SUPPORTED_ATTRS))
+		return -EINVAL;
+
 	mailbox = mlx4_alloc_cmd_mailbox(dev);
 	if (IS_ERR(mailbox))
 		return PTR_ERR(mailbox);
 
 	cmd = (struct mlx4_update_qp_context *)mailbox->buf;
 
-	if (!attr || (attr & ~MLX4_UPDATE_QP_SUPPORTED_ATTRS))
-		return -EINVAL;
-
 	if (attr & MLX4_UPDATE_QP_SMAC) {
 		pri_addr_path_mask |= 1ULL << MLX4_UPD_QP_PATH_MASK_MAC_INDEX;
 		cmd->qp_context.pri_path.grh_mylmc = params->smac_index;
-- 
2.3.7

[PATCH net-next 7/7] net/mlx4_core: Fix resource tracker error flow in add_res_range

[Or Gerlitz]

From: Saeed Mahameed <saeedm@mellanox.com>

The 'for' loop when undoing rb-tree insertions and list-adds in the
error flow in add_res_range had errors, fix them.

Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
---
 drivers/net/ethernet/mellanox/mlx4/resource_tracker.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
index 731423c..ac4b99a 100644
--- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
+++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c
@@ -1238,8 +1238,10 @@ static int add_res_range(struct mlx4_dev *dev, int slave, u64 base, int count,
 	return 0;
 
 undo:
-	for (--i; i >= base; --i)
+	for (--i; i >= 0; --i) {
 		rb_erase(&res_arr[i]->node, root);
+		list_del_init(&res_arr[i]->list);
+	}
 
 	spin_unlock_irq(mlx4_tlock(dev));
 
-- 
2.3.7

Re: [PATCH net-next 0/7] Mellanox driver update for net-next

[David Miller]

From: Or Gerlitz <ogerlitz@mellanox.com>
Date: Thu,  8 Oct 2015 17:13:56 +0300

> Some small fixes and small enhancements from the team.
> 
> Series applies over net-next commit acb4a6b "tcp: ensure prior synack rtx behavior
> with small backlogs".

Series applied, thanks.