[PATCH net-next 0/9] bnx2x: patch series

[PATCH net-next 0/9] bnx2x: patch series

[Yuval Mintz]

Hi Dave,

This patch series contains several SR-IOV related patches -
it allows setting and querying the number of vfs via sysfs,
configuring and querying VFs via iproute2 callbacks and properly
implements for VFs only the ethtool ops which they can safely perform.

In addition, it also contains several small fixes of rare corner-cases.

Please consider applying this patch-series to 'net-next'.

Thanks,
Yuval Mintz

[PATCH net-next 1/9] bnx2x: Prevent NULL pointer dereference in vlan-mac

[Yuval Mintz]

When setting the vlan mac configuration in the bnx2x driver, there is a
registry list (shadow list) which was accessed without a lock.

Under specific stress-testing tests, it was possible to have a NULL pointer
dereference as one context would have emptied the list while the other
accessed its elements.

This adds a variation on a classical single writer, multiple reader lock
in which writers don't wait for the list to be accessible but rather pend
their request to be performed once the list is available. This prevents said
issue.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c    |  433 ++++++++++++++++++++-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h    |   18 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c |    9 +-
 3 files changed, 448 insertions(+), 12 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
index 7306416..96dcc92 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
@@ -184,6 +184,7 @@ static inline void bnx2x_exe_queue_reset_pending(struct bnx2x *bp,
  * (Atomicy is ensured using the exe_queue->lock).
  */
 static inline int bnx2x_exe_queue_step(struct bnx2x *bp,
+				       struct bnx2x_vlan_mac_obj *vobj,
 				       struct bnx2x_exe_queue_obj *o,
 				       unsigned long *ramrod_flags)
 {
@@ -194,6 +195,13 @@ static inline int bnx2x_exe_queue_step(struct bnx2x *bp,
 
 	spin_lock_bh(&o->lock);
 
+	/* From here on we're under the context of the execution queue lock.
+	 * Whether or not an execution will be made, we can clear the vlan_mac
+	 * head pending execution reference, since accessing it must also be
+	 * done under this lock (there's no chance for race)
+	 */
+	vobj->head_exe = false;
+
 	/*
 	 * Next step should not be performed until the current is finished,
 	 * unless a DRV_CLEAR_ONLY bit is set. In this case we just want to
@@ -443,12 +451,333 @@ static bool bnx2x_put_credit_vlan_mac(struct bnx2x_vlan_mac_obj *o)
 	return true;
 }
 
+#define BNX2X_VLAN_MAC_H_LOCK_TIMEOUT	(5000)
+/**
+ * __bnx2x_vlan_mac_h_write_trylock - try getting the writer lock on vlan mac
+ * head list.
+ *
+ * @bp:		device handle
+ * @o:		vlan_mac object
+ *
+ * @details: Non-blocking implementation; should be called under execution
+ *           queue lock.
+ */
+static int __bnx2x_vlan_mac_h_write_trylock(struct bnx2x *bp,
+					    struct bnx2x_vlan_mac_obj *o)
+{
+	if (o->head_writer || o->head_reader) {
+		DP(BNX2X_MSG_SP, "vlan_mac_lock writer - Can't take; Busy\n");
+		return -EBUSY;
+	}
+
+	o->head_writer = true;
+	o->head_exe = true;
+	DP(BNX2X_MSG_SP, "vlan_mac_lock writer - Taken\n");
+	return 0;
+}
+
+/**
+ * __bnx2x_vlan_mac_h_write_lock - takes writer lock on vlan mac head list
+ *
+ * @bp:		device handle
+ * @o:		vlan_mac object
+ *
+ * @details May sleep. Should be called under execution queue lock; notice it
+ *          might release and reclaim it during its run.
+ */
+static int __bnx2x_vlan_mac_h_write_lock(struct bnx2x *bp,
+					   struct bnx2x_vlan_mac_obj *o)
+{
+	spinlock_t *lock = &o->exe_queue.lock;
+	int rc = 0, time_left = BNX2X_VLAN_MAC_H_LOCK_TIMEOUT;
+
+	DP(BNX2X_MSG_SP, "vlan_mac_lock - Trying to lock writer...\n");
+	while (time_left &&
+	       (__bnx2x_vlan_mac_h_write_trylock(bp, o) != 0)) {
+		spin_unlock_bh(lock);
+		usleep_range(1000, 2000);
+		time_left--;
+		spin_lock_bh(lock);
+	}
+
+	if (!time_left) {
+		BNX2X_ERR("Failed to get vlan_mac_lock (writer)\n");
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+		rc = -EBUSY;
+	} else {
+		DP(BNX2X_MSG_SP, "Got vlan_mac_lock (writer) after %d iterations\n",
+		   BNX2X_VLAN_MAC_H_LOCK_TIMEOUT - time_left);
+	}
+
+	return rc;
+}
+
+/**
+ * bnx2x_vlan_mac_h_write_lock - takes writer lock on vlan mac head list
+ *
+ * @bp:		device handle
+ * @o:		vlan_mac object
+ *
+ * @details May sleep. Notice it claims & releases the execution queue's lock.
+ */
+int bnx2x_vlan_mac_h_write_lock(struct bnx2x *bp,
+				struct bnx2x_vlan_mac_obj *o)
+{
+	int rc;
+
+	spin_lock_bh(&o->exe_queue.lock);
+	rc = __bnx2x_vlan_mac_h_write_lock(bp, o);
+	spin_unlock_bh(&o->exe_queue.lock);
+
+	return rc;
+}
+
+/**
+ * __bnx2x_vlan_mac_h_exec_pending - execute step instead of a previous step
+ * which wasn't able to run due to a taken lock on vlan mac head list.
+ *
+ * @bp:		device handle
+ * @o:		vlan_mac object
+ *
+ * @details Should be called under execution queue lock; notice it might release
+ *          and reclaim it during its run.
+ */
+static void __bnx2x_vlan_mac_h_exec_pending(struct bnx2x *bp,
+					    struct bnx2x_vlan_mac_obj *o)
+{
+	int rc;
+	unsigned long ramrod_flags = o->saved_ramrod_flags;
+
+	if (!o->head_writer) {
+		BNX2X_ERR("vlan_mac_lock - Need to execute pending command, but writer isn't marked\n");
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+	}
+
+	DP(BNX2X_MSG_SP, "vlan_mac_lock execute pending command with ramrod flags %lu\n",
+	   ramrod_flags);
+	o->head_exe = true;
+	o->head_exe_request = false;
+	o->saved_ramrod_flags = 0;
+	spin_unlock_bh(&o->exe_queue.lock);
+	rc = bnx2x_exe_queue_step(bp, o, &o->exe_queue, &ramrod_flags);
+	if (rc != 0) {
+		BNX2X_ERR("execution of pending commands failed with rc %d\n",
+			  rc);
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+	}
+	spin_lock_bh(&o->exe_queue.lock);
+}
+
+/**
+ * __bnx2x_vlan_mac_h_pend - Pend an execution step which couldn't have been
+ * called due to vlan mac head list lock being taken.
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ * @ramrod_flags:	ramrod flags of missed execution
+ *
+ * @details Should be called under execution queue lock.
+ */
+static void __bnx2x_vlan_mac_h_pend(struct bnx2x *bp,
+				    struct bnx2x_vlan_mac_obj *o,
+				    unsigned long ramrod_flags)
+{
+	if (!o->head_exe) {
+		o->head_exe_request = true;
+		o->saved_ramrod_flags = ramrod_flags;
+		DP(BNX2X_MSG_SP, "Placing pending execution with ramrod flags %lu\n",
+		   ramrod_flags);
+	} else {
+		/* Since an executor is on the field, he will step whatever
+		 * we do. We can simply cease - no need to pend a command.
+		 */
+		DP(BNX2X_MSG_SP, "Skipping pending execution with ramrod flags %lu\n",
+		   ramrod_flags);
+	}
+}
+
+/**
+ * __bnx2x_vlan_mac_h_write_unlock - unlock the vlan mac head list writer lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details Should be called under execution queue lock. Notice if a pending
+ *          execution exists, it would perform it - possibly releasing and
+ *          reclaiming the execution queue lock.
+ */
+static void __bnx2x_vlan_mac_h_write_unlock(struct bnx2x *bp,
+					    struct bnx2x_vlan_mac_obj *o)
+{
+	/* When releasing a writer lock, exec. was supposed to prev. happen */
+	if (!o->head_writer) {
+		BNX2X_ERR("Need to release vlan mac writer lock, but lock isn't taken properly (%d)\n",
+			  o->head_writer);
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+	}
+	if (o->head_exe)
+		BNX2X_ERR("Need to release vlan mac writer lock, but execution wasn't performed (%d)\n",
+			  o->head_exe);
+
+	/* It's possible a new pending execution was added since this writer
+	 * executed. If so, execute again. [Ad infinitum]
+	 */
+	while (o->head_exe_request) {
+		DP(BNX2X_MSG_SP, "vlan_mac_lock - writer release encountered a pending request\n");
+		__bnx2x_vlan_mac_h_exec_pending(bp, o);
+	}
+
+	o->head_writer = false;
+}
+
+/**
+ * bnx2x_vlan_mac_h_write_unlock - unlock the vlan mac head list writer lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details Notice if a pending execution exists, it would perform it -
+ *          possibly releasing and reclaiming the execution queue lock.
+ */
+void bnx2x_vlan_mac_h_write_unlock(struct bnx2x *bp,
+				   struct bnx2x_vlan_mac_obj *o)
+{
+	spin_lock_bh(&o->exe_queue.lock);
+	__bnx2x_vlan_mac_h_write_unlock(bp, o);
+	spin_unlock_bh(&o->exe_queue.lock);
+}
+
+/**
+ * __bnx2x_vlan_mac_h_read_lock - lock the vlan mac head list reader lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details Should be called under the execution queue lock. May sleep. May
+ *          release and reclaim execution queue lock during its run.
+ */
+static int __bnx2x_vlan_mac_h_read_lock(struct bnx2x *bp,
+					struct bnx2x_vlan_mac_obj *o)
+{
+	spinlock_t *lock = &o->exe_queue.lock;
+	int rc = 0, time_left = BNX2X_VLAN_MAC_H_LOCK_TIMEOUT;
+
+	DP(BNX2X_MSG_SP, "vlan_mac_lock - Trying to lock reader...\n");
+	while (time_left && o->head_writer) {
+		spin_unlock_bh(lock);
+		usleep_range(1000, 2000);
+		time_left--;
+		spin_lock_bh(lock);
+	}
+
+	if (!time_left) {
+		BNX2X_ERR("vlan_mac_lock - Failed to get reader lock\n");
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+		rc = -EBUSY;
+	} else {
+		o->head_reader++;
+		DP(BNX2X_MSG_SP, "vlan_mac_lock - locked reader - number %d\n",
+		   o->head_reader);
+	}
+
+	return rc;
+}
+
+/**
+ * bnx2x_vlan_mac_h_read_lock - lock the vlan mac head list reader lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details May sleep. Claims and releases execution queue lock during its run.
+ */
+int bnx2x_vlan_mac_h_read_lock(struct bnx2x *bp,
+			       struct bnx2x_vlan_mac_obj *o)
+{
+	int rc;
+
+	spin_lock_bh(&o->exe_queue.lock);
+	rc = __bnx2x_vlan_mac_h_read_lock(bp, o);
+	spin_unlock_bh(&o->exe_queue.lock);
+
+	return rc;
+}
+
+/**
+ * __bnx2x_vlan_mac_h_read_unlock - unlock the vlan mac head list reader lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details Should be called under execution queue lock. Notice if a pending
+ *          execution exists, it would be performed if this was the last
+ *          reader. possibly releasing and reclaiming the execution queue lock.
+ */
+static void __bnx2x_vlan_mac_h_read_unlock(struct bnx2x *bp,
+					  struct bnx2x_vlan_mac_obj *o)
+{
+	if (!o->head_reader) {
+		BNX2X_ERR("Need to release vlan mac reader lock, but lock isn't taken\n");
+#ifdef BNX2X_STOP_ON_ERROR
+		bnx2x_panic();
+#endif
+	} else {
+		o->head_reader--;
+		DP(BNX2X_MSG_SP, "vlan_mac_lock - decreased readers to %d\n",
+		   o->head_reader);
+	}
+
+	/* It's possible a new pending execution was added, and that this reader
+	 * was last - if so we need to execute the command.
+	 */
+	if (!o->head_reader && o->head_exe_request) {
+		DP(BNX2X_MSG_SP, "vlan_mac_lock - reader release encountered a pending request\n");
+
+		/* Writer release will do the trick */
+		o->head_writer = true;
+		__bnx2x_vlan_mac_h_write_unlock(bp, o);
+	}
+}
+
+/**
+ * bnx2x_vlan_mac_h_read_unlock - unlock the vlan mac head list reader lock
+ *
+ * @bp:			device handle
+ * @o:			vlan_mac object
+ *
+ * @details Notice if a pending execution exists, it would be performed if this
+ *          was the last reader. Claims and releases the execution queue lock
+ *          during its run.
+ */
+void bnx2x_vlan_mac_h_read_unlock(struct bnx2x *bp,
+				  struct bnx2x_vlan_mac_obj *o)
+{
+	spin_lock_bh(&o->exe_queue.lock);
+	__bnx2x_vlan_mac_h_read_unlock(bp, o);
+	spin_unlock_bh(&o->exe_queue.lock);
+}
+
 static int bnx2x_get_n_elements(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
 				int n, u8 *buf)
 {
 	struct bnx2x_vlan_mac_registry_elem *pos;
 	u8 *next = buf;
-	int counter = 0;
+	int counter = 0, read_lock;
+
+	DP(BNX2X_MSG_SP, "get_n_elements - taking vlan_mac_lock (reader)\n");
+	read_lock = bnx2x_vlan_mac_h_read_lock(bp, o);
+	if (read_lock != 0)
+		BNX2X_ERR("get_n_elements failed to get vlan mac reader lock; Access without lock\n");
 
 	/* traverse list */
 	list_for_each_entry(pos, &o->head, link) {
@@ -470,6 +799,12 @@ static int bnx2x_get_n_elements(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
 			   counter, next, pos->u.mac.mac);
 		}
 	}
+
+	if (read_lock == 0) {
+		DP(BNX2X_MSG_SP, "get_n_elements - releasing vlan_mac_lock (reader)\n");
+		bnx2x_vlan_mac_h_read_unlock(bp, o);
+	}
+
 	return counter * ETH_ALEN;
 }
 
@@ -1095,7 +1430,7 @@ static int bnx2x_vlan_mac_restore(struct bnx2x *bp,
 	/* Set a restore bit */
 	__set_bit(RAMROD_RESTORE, &p->ramrod_flags);
 
-	return bnx2x_config_vlan_mac(bp, p);
+	return bnx2x_config_vlan_mac_locked(bp, p);
 }
 
 /*
@@ -1417,6 +1752,52 @@ static int bnx2x_wait_vlan_mac(struct bnx2x *bp,
 	return -EBUSY;
 }
 
+static int __bnx2x_vlan_mac_execute_step(struct bnx2x *bp,
+					 struct bnx2x_vlan_mac_obj *o,
+					 unsigned long *ramrod_flags,
+					 bool pre_locked)
+{
+	bool current_lock;
+	int rc;
+
+	spin_lock_bh(&o->exe_queue.lock);
+
+	if (!pre_locked) {
+		DP(BNX2X_MSG_SP, "vlan_mac_execute_step - trying to take writer lock\n");
+		rc = __bnx2x_vlan_mac_h_write_trylock(bp, o);
+		current_lock = (rc == 0) ? true : false;
+	} else {
+		/* Since there's an outside lock, it's possible there've
+		 * been repeated calls to this functions. As every call to
+		 * exe_queue_step would clear the execution index, we need
+		 * to clean any pending request to execute and re-set the
+		 * execution status.
+		 */
+		o->head_exe_request = false;
+		o->saved_ramrod_flags = 0;
+		o->head_exe = true;
+
+		current_lock = true;
+	}
+
+	if (!current_lock) {
+		__bnx2x_vlan_mac_h_pend(bp, o, *ramrod_flags);
+		spin_unlock_bh(&o->exe_queue.lock);
+	} else {
+		spin_unlock_bh(&o->exe_queue.lock);
+
+		rc = bnx2x_exe_queue_step(bp, o, &o->exe_queue, ramrod_flags);
+		if (!pre_locked) {
+			DP(BNX2X_MSG_SP, "vlan_mac_execute_step - trying to release writer lock\n");
+			bnx2x_vlan_mac_h_write_unlock(bp, o);
+		}
+		if (rc < 0)
+			return rc;
+	}
+
+	return 0;
+}
+
 /**
  * bnx2x_complete_vlan_mac - complete one VLAN-MAC ramrod
  *
@@ -1446,7 +1827,9 @@ static int bnx2x_complete_vlan_mac(struct bnx2x *bp,
 
 	/* Run the next bulk of pending commands if requested */
 	if (test_bit(RAMROD_CONT, ramrod_flags)) {
-		rc = bnx2x_exe_queue_step(bp, &o->exe_queue, ramrod_flags);
+		rc = __bnx2x_vlan_mac_execute_step(bp, o,
+						   ramrod_flags, false);
+
 		if (rc < 0)
 			return rc;
 	}
@@ -1737,15 +2120,15 @@ static inline int bnx2x_vlan_mac_push_new_cmd(
 }
 
 /**
- * bnx2x_config_vlan_mac - configure VLAN/MAC/VLAN_MAC filtering rules.
+ * __bnx2x_config_vlan_mac - configure VLAN/MAC/VLAN_MAC filtering rules.
  *
  * @bp:	  device handle
  * @p:
  *
  */
-int bnx2x_config_vlan_mac(
-	struct bnx2x *bp,
-	struct bnx2x_vlan_mac_ramrod_params *p)
+static int __bnx2x_config_vlan_mac(struct bnx2x *bp,
+				   struct bnx2x_vlan_mac_ramrod_params *p,
+				   bool pre_locked)
 {
 	int rc = 0;
 	struct bnx2x_vlan_mac_obj *o = p->vlan_mac_obj;
@@ -1777,7 +2160,9 @@ int bnx2x_config_vlan_mac(
 	/* Execute commands if required */
 	if (cont || test_bit(RAMROD_EXEC, ramrod_flags) ||
 	    test_bit(RAMROD_COMP_WAIT, ramrod_flags)) {
-		rc = bnx2x_exe_queue_step(bp, &o->exe_queue, ramrod_flags);
+		rc = __bnx2x_vlan_mac_execute_step(bp, p->vlan_mac_obj,
+						   &p->ramrod_flags,
+						   pre_locked);
 		if (rc < 0)
 			return rc;
 	}
@@ -1802,8 +2187,10 @@ int bnx2x_config_vlan_mac(
 				return rc;
 
 			/* Make a next step */
-			rc = bnx2x_exe_queue_step(bp, &o->exe_queue,
-						  ramrod_flags);
+			rc = __bnx2x_vlan_mac_execute_step(bp,
+							   p->vlan_mac_obj,
+							   &p->ramrod_flags,
+							   pre_locked);
 			if (rc < 0)
 				return rc;
 		}
@@ -1814,7 +2201,17 @@ int bnx2x_config_vlan_mac(
 	return rc;
 }
 
+int bnx2x_config_vlan_mac(struct bnx2x *bp,
+			  struct bnx2x_vlan_mac_ramrod_params *p)
+{
+	return __bnx2x_config_vlan_mac(bp, p, false);
+}
 
+int bnx2x_config_vlan_mac_locked(struct bnx2x *bp,
+					 struct bnx2x_vlan_mac_ramrod_params *p)
+{
+	return __bnx2x_config_vlan_mac(bp, p, true);
+}
 
 /**
  * bnx2x_vlan_mac_del_all - delete elements with given vlan_mac_flags spec
@@ -1835,7 +2232,7 @@ static int bnx2x_vlan_mac_del_all(struct bnx2x *bp,
 				  unsigned long *ramrod_flags)
 {
 	struct bnx2x_vlan_mac_registry_elem *pos = NULL;
-	int rc = 0;
+	int rc = 0, read_lock;
 	struct bnx2x_vlan_mac_ramrod_params p;
 	struct bnx2x_exe_queue_obj *exeq = &o->exe_queue;
 	struct bnx2x_exeq_elem *exeq_pos, *exeq_pos_n;
@@ -1873,6 +2270,11 @@ static int bnx2x_vlan_mac_del_all(struct bnx2x *bp,
 	__clear_bit(RAMROD_EXEC, &p.ramrod_flags);
 	__clear_bit(RAMROD_CONT, &p.ramrod_flags);
 
+	DP(BNX2X_MSG_SP, "vlan_mac_del_all -- taking vlan_mac_lock (reader)\n");
+	read_lock = bnx2x_vlan_mac_h_read_lock(bp, o);
+	if (read_lock != 0)
+		return read_lock;
+
 	list_for_each_entry(pos, &o->head, link) {
 		if (pos->vlan_mac_flags == *vlan_mac_flags) {
 			p.user_req.vlan_mac_flags = pos->vlan_mac_flags;
@@ -1880,11 +2282,15 @@ static int bnx2x_vlan_mac_del_all(struct bnx2x *bp,
 			rc = bnx2x_config_vlan_mac(bp, &p);
 			if (rc < 0) {
 				BNX2X_ERR("Failed to add a new DEL command\n");
+				bnx2x_vlan_mac_h_read_unlock(bp, o);
 				return rc;
 			}
 		}
 	}
 
+	DP(BNX2X_MSG_SP, "vlan_mac_del_all -- releasing vlan_mac_lock (reader)\n");
+	bnx2x_vlan_mac_h_read_unlock(bp, o);
+
 	p.ramrod_flags = *ramrod_flags;
 	__set_bit(RAMROD_CONT, &p.ramrod_flags);
 
@@ -1916,6 +2322,11 @@ static inline void bnx2x_init_vlan_mac_common(struct bnx2x_vlan_mac_obj *o,
 	struct bnx2x_credit_pool_obj *vlans_pool)
 {
 	INIT_LIST_HEAD(&o->head);
+	o->head_writer = false;
+	o->head_reader = 0;
+	o->head_exe = false;
+	o->head_exe_request = false;
+	o->saved_ramrod_flags = 0;
 
 	o->macs_pool = macs_pool;
 	o->vlans_pool = vlans_pool;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
index ff90760..3ec795b 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
@@ -290,6 +290,14 @@ struct bnx2x_vlan_mac_obj {
 	 * entries.
 	 */
 	struct list_head		head;
+	/* Implement a simple reader/writer lock on the head list.
+	 * all these fields should only be accessed under the exe_queue lock
+	 */
+	bool		head_writer; /* Writer accessing head list. */
+	bool		head_exe; /* Writer will exec a step */
+	u8		head_reader; /* Num. of readers accessing head list */
+	bool		head_exe_request; /* Pending execution request. */
+	unsigned long	saved_ramrod_flags; /* Ramrods of pending execution */
 
 	/* TODO: Add it's initialization in the init functions */
 	struct bnx2x_exe_queue_obj	exe_queue;
@@ -1297,8 +1305,18 @@ void bnx2x_init_vlan_mac_obj(struct bnx2x *bp,
 			     struct bnx2x_credit_pool_obj *macs_pool,
 			     struct bnx2x_credit_pool_obj *vlans_pool);
 
+int bnx2x_vlan_mac_h_read_lock(struct bnx2x *bp,
+					struct bnx2x_vlan_mac_obj *o);
+void bnx2x_vlan_mac_h_read_unlock(struct bnx2x *bp,
+				  struct bnx2x_vlan_mac_obj *o);
+int bnx2x_vlan_mac_h_write_lock(struct bnx2x *bp,
+				struct bnx2x_vlan_mac_obj *o);
+void bnx2x_vlan_mac_h_write_unlock(struct bnx2x *bp,
+					  struct bnx2x_vlan_mac_obj *o);
 int bnx2x_config_vlan_mac(struct bnx2x *bp,
 			  struct bnx2x_vlan_mac_ramrod_params *p);
+int bnx2x_config_vlan_mac_locked(struct bnx2x *bp,
+				 struct bnx2x_vlan_mac_ramrod_params *p);
 
 int bnx2x_vlan_mac_move(struct bnx2x *bp,
 			struct bnx2x_vlan_mac_ramrod_params *p,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
index 6adfa20..89c6fa1 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
@@ -489,12 +489,19 @@ static inline void bnx2x_vfop_credit(struct bnx2x *bp,
 	 * and a valid credit counter
 	 */
 	if (!vfop->rc && args->credit) {
-		int cnt = 0;
+		int cnt = 0, read_lock;
 		struct list_head *pos;
 
+		read_lock = bnx2x_vlan_mac_h_read_lock(bp, obj);
+		if (read_lock)
+			DP(BNX2X_MSG_SP, "Failed to take vlan mac read head; continuing anyway\n");
+
 		list_for_each(pos, &obj->head)
 			cnt++;
 
+		if (!read_lock)
+			bnx2x_vlan_mac_h_read_unlock(bp, obj);
+
 		atomic_set(args->credit, cnt);
 	}
 }
-- 
1.7.9.GIT

[PATCH net-next 2/9] bnx2x: fix vlan-mac memory leak

[Yuval Mintz]

This patch releases memory previously simply omitted when elements
are removed from pending execution lists in the bnx2x driver.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
index 96dcc92..f701b72 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
@@ -2251,6 +2251,7 @@ static int bnx2x_vlan_mac_del_all(struct bnx2x *bp,
 				return rc;
 			}
 			list_del(&exeq_pos->link);
+			bnx2x_exe_queue_free_elem(bp, exeq_pos);
 		}
 	}
 
-- 
1.7.9.GIT

[PATCH net-next 3/9] bnx2x: handle spurious interrupts

[Yuval Mintz]

In scenarios in which a previous driver was removed without proper cleanup
(e.g., kdump), it is possible for the chip to generate an interrupt without
any apparent reason.
This would cause improper behaviour at best, possibly resulting in NULL
pointer dereferences as interrupts might be sent to an ISR accessing
uninitialized structures and memories.

This causes the driver to ignore any interrupt which arrives in an
inappropriate time, i.e., when the driver has not yet properly configured
its interrupt routines.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x.h      |    1 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c  |   11 +++++++++++
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c |    3 +++
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.c |    3 ++-
 4 files changed, 17 insertions(+), 1 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index e4605a9..3e26eb5 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1379,6 +1379,7 @@ struct bnx2x {
 #define USING_SINGLE_MSIX_FLAG		(1 << 20)
 #define BC_SUPPORTS_DCBX_MSG_NON_PMF	(1 << 21)
 #define IS_VF_FLAG			(1 << 22)
+#define INTERRUPTS_ENABLED_FLAG	(1 << 23)
 
 #define BP_NOMCP(bp)			((bp)->flags & NO_MCP_FLAG)
 
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index ecac04a3..525c2b5 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -1037,6 +1037,17 @@ static irqreturn_t bnx2x_msix_fp_int(int irq, void *fp_cookie)
 	DP(NETIF_MSG_INTR,
 	   "got an MSI-X interrupt on IDX:SB [fp %d fw_sd %d igusb %d]\n",
 	   fp->index, fp->fw_sb_id, fp->igu_sb_id);
+
+	/* It's possible for a spurious interrupt to be received, if the
+	 * driver was loaded above a previous configured function (e.g., kdump).
+	 * We simply ignore such interrupts.
+	 */
+	if (unlikely(!(bp->flags & INTERRUPTS_ENABLED_FLAG))) {
+		WARN_ONCE(1, "%s: Got Spurious interrupts",
+			  bp->dev ? bp->dev->name : "?");
+		return IRQ_HANDLED;
+	}
+
 	bnx2x_ack_sb(bp, fp->igu_sb_id, USTORM_ID, 0, IGU_INT_DISABLE, 0);
 
 #ifdef BNX2X_STOP_ON_ERROR
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index c4daee1..1776b8b 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -868,6 +868,7 @@ static void bnx2x_int_disable(struct bnx2x *bp)
 		bnx2x_hc_int_disable(bp);
 	else
 		bnx2x_igu_int_disable(bp);
+	bp->flags &= ~INTERRUPTS_ENABLED_FLAG;
 }
 
 void bnx2x_panic_dump(struct bnx2x *bp, bool disable_int)
@@ -1607,6 +1608,8 @@ static void bnx2x_igu_int_enable(struct bnx2x *bp)
 
 void bnx2x_int_enable(struct bnx2x *bp)
 {
+	bp->flags |= INTERRUPTS_ENABLED_FLAG;
+
 	if (bp->common.int_block == INT_BLOCK_HC)
 		bnx2x_hc_int_enable(bp);
 	else
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.c
index 3624612..df9209e 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.c
@@ -268,7 +268,8 @@ int bnx2x_vfpf_acquire(struct bnx2x *bp, u8 tx_count, u8 rx_count)
 	bp->mf_mode = 0;
 	bp->common.flash_size = 0;
 	bp->flags |=
-		NO_WOL_FLAG | NO_ISCSI_OOO_FLAG | NO_ISCSI_FLAG | NO_FCOE_FLAG;
+		NO_WOL_FLAG | NO_ISCSI_OOO_FLAG | NO_ISCSI_FLAG | NO_FCOE_FLAG |
+		INTERRUPTS_ENABLED_FLAG;
 	bp->igu_sb_cnt = 1;
 	bp->igu_base_sb = bp->acquire_resp.resc.hw_sbs[0].hw_sb_id;
 	strlcpy(bp->fw_ver, bp->acquire_resp.pfdev_info.fw_ver,
-- 
1.7.9.GIT

Re: [PATCH net-next 3/9] bnx2x: handle spurious interrupts

[David Miller]

From: "Yuval Mintz" <yuvalmin@broadcom.com>
Date: Mon, 18 Feb 2013 10:16:32 +0200

> In scenarios in which a previous driver was removed without proper cleanup
> (e.g., kdump), it is possible for the chip to generate an interrupt without
> any apparent reason.
> This would cause improper behaviour at best, possibly resulting in NULL
> pointer dereferences as interrupts might be sent to an ISR accessing
> uninitialized structures and memories.
> 
> This causes the driver to ignore any interrupt which arrives in an
> inappropriate time, i.e., when the driver has not yet properly configured
> its interrupt routines.
> 
> Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
> Signed-off-by: Ariel Elior <ariele@broadcom.com>
> Signed-off-by: Eilon Greenstein <eilong@broadcom.com>

I cannot accept this patch.

You must not request_irq() until your software state is completely
setup, there are no exceptions to this rule.

[PATCH net-next 4/9] bnx2x: Set ethtool ops for vfs

[Yuval Mintz]

From: Ariel Elior <ariele@broadcom.com>

Virtual functions don't have access to HW registers, therefore most ethtool ops
are forbidden to them. Instead of checking in each op whether the device being
driven is a virtual function or a physical function, this patch creates a
separate ethtool ops struct for virtual functions and uses it to initialize
the ethtool ops of the driver in case it is driving a virtual function device.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x.h        |    2 +-
 .../net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c    |   29 ++++++++++++++++++-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c   |    2 +-
 3 files changed, 29 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 3e26eb5..0528f3b 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -2287,7 +2287,7 @@ static const u32 dmae_reg_go_c[] = {
 	DMAE_REG_GO_C12, DMAE_REG_GO_C13, DMAE_REG_GO_C14, DMAE_REG_GO_C15
 };
 
-void bnx2x_set_ethtool_ops(struct net_device *netdev);
+void bnx2x_set_ethtool_ops(struct bnx2x *bp, struct net_device *netdev);
 void bnx2x_notify_link_changed(struct bnx2x *bp);
 
 #define BNX2X_MF_SD_PROTOCOL(bp) \
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
index 9a674b1..77173c9 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_ethtool.c
@@ -3226,7 +3226,32 @@ static const struct ethtool_ops bnx2x_ethtool_ops = {
 	.get_ts_info		= ethtool_op_get_ts_info,
 };
 
-void bnx2x_set_ethtool_ops(struct net_device *netdev)
+static const struct ethtool_ops bnx2x_vf_ethtool_ops = {
+	.get_settings		= bnx2x_get_settings,
+	.set_settings		= bnx2x_set_settings,
+	.get_drvinfo		= bnx2x_get_drvinfo,
+	.get_msglevel		= bnx2x_get_msglevel,
+	.set_msglevel		= bnx2x_set_msglevel,
+	.get_link		= bnx2x_get_link,
+	.get_coalesce		= bnx2x_get_coalesce,
+	.get_ringparam		= bnx2x_get_ringparam,
+	.set_ringparam		= bnx2x_set_ringparam,
+	.get_sset_count		= bnx2x_get_sset_count,
+	.get_strings		= bnx2x_get_strings,
+	.get_ethtool_stats	= bnx2x_get_ethtool_stats,
+	.get_rxnfc		= bnx2x_get_rxnfc,
+	.set_rxnfc		= bnx2x_set_rxnfc,
+	.get_rxfh_indir_size	= bnx2x_get_rxfh_indir_size,
+	.get_rxfh_indir		= bnx2x_get_rxfh_indir,
+	.set_rxfh_indir		= bnx2x_set_rxfh_indir,
+	.get_channels		= bnx2x_get_channels,
+	.set_channels		= bnx2x_set_channels,
+};
+
+void bnx2x_set_ethtool_ops(struct bnx2x *bp, struct net_device *netdev)
 {
-	SET_ETHTOOL_OPS(netdev, &bnx2x_ethtool_ops);
+	if (IS_PF(bp))
+		SET_ETHTOOL_OPS(netdev, &bnx2x_ethtool_ops);
+	else /* vf */
+		SET_ETHTOOL_OPS(netdev, &bnx2x_vf_ethtool_ops);
 }
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 1776b8b..d3a33ef 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -11956,7 +11956,7 @@ static int bnx2x_init_dev(struct bnx2x *bp, struct pci_dev *pdev,
 	dev->watchdog_timeo = TX_TIMEOUT;
 
 	dev->netdev_ops = &bnx2x_netdev_ops;
-	bnx2x_set_ethtool_ops(dev);
+	bnx2x_set_ethtool_ops(bp, dev);
 
 	dev->priv_flags |= IFF_UNICAST_FLT;
 
-- 
1.7.9.GIT

[PATCH net-next 5/9] bnx2x: Take chip version from MFW

[Yuval Mintz]

In latest boards, the CHIP_METAL register contains an incorrect
revision value, so the correct one needs to be obtained in a
different manner.

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c |    8 ++++++--
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h  |    6 ++----
 2 files changed, 8 insertions(+), 6 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index d3a33ef..f563fdc 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -10037,8 +10037,12 @@ static void bnx2x_get_common_hwinfo(struct bnx2x *bp)
 	id = ((val & 0xffff) << 16);
 	val = REG_RD(bp, MISC_REG_CHIP_REV);
 	id |= ((val & 0xf) << 12);
-	val = REG_RD(bp, MISC_REG_CHIP_METAL);
-	id |= ((val & 0xff) << 4);
+
+	/* Metal is read from PCI regs, but we can't access >=0x400 from
+	 * the configuration space (so we need to reg_rd)
+	 */
+	val = REG_RD(bp, PCICFG_OFFSET + PCI_ID_VAL3);
+	id |= (((val >> 24) & 0xf) << 4);
 	val = REG_RD(bp, MISC_REG_BOND_ID);
 	id |= (val & 0xf);
 	bp->common.chip_id = id;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
index 791eb2d..d22bc40 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_reg.h
@@ -1491,10 +1491,6 @@
 /* [R 4] This field indicates the type of the device. '0' - 2 Ports; '1' - 1
    Port. */
 #define MISC_REG_BOND_ID					 0xa400
-/* [R 8] These bits indicate the metal revision of the chip. This value
-   starts at 0x00 for each all-layer tape-out and increments by one for each
-   tape-out. */
-#define MISC_REG_CHIP_METAL					 0xa404
 /* [R 16] These bits indicate the part number for the chip. */
 #define MISC_REG_CHIP_NUM					 0xa408
 /* [R 4] These bits indicate the base revision of the chip. This value
@@ -6331,6 +6327,8 @@
 #define PCI_PM_DATA_B					0x414
 #define PCI_ID_VAL1					0x434
 #define PCI_ID_VAL2					0x438
+#define PCI_ID_VAL3					0x43c
+
 #define GRC_CONFIG_REG_PF_INIT_VF		0x624
 #define GRC_CR_PF_INIT_VF_PF_FIRST_VF_NUM_MASK	0xf
 /* First VF_NUM for PF is encoded in this register.
-- 
1.7.9.GIT

[PATCH net-next 6/9] bnx2x: Prevent "Unknown MF" print in SF mode

[Yuval Mintz]

From: Ariel Elior <ariele@broadcom.com>

When using a chip operating in Single Function mode, when the chip is probed
the bnx2x would print a message warning of an unknown Multi Function mode.

This patch prevents said message.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index f563fdc..2b7930c 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -11063,6 +11063,9 @@ static int bnx2x_get_hwinfo(struct bnx2x *bp)
 				} else
 					BNX2X_DEV_INFO("illegal OV for SD\n");
 				break;
+			case SHARED_FEAT_CFG_FORCE_SF_MODE_FORCED_SF:
+				bp->mf_config[vn] = 0;
+				break;
 			default:
 				/* Unknown configuration: reset mf_config */
 				bp->mf_config[vn] = 0;
-- 
1.7.9.GIT

[PATCH net-next 7/9] bnx2x: Add iproute2 support for vfs

[Yuval Mintz]

From: Ariel Elior <ariele@broadcom.com>

This patch adds support for iproute2 callbacks allowing querying a physical
function as to its child virtual functions, and setting the macs and vlans
of said virtual functions.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x.h       |    1 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h   |    3 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c  |   39 +++-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c    |   24 +--
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h    |    6 +-
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c |  231 +++++++++++++++++++--
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h |    3 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.h  |    9 +-
 8 files changed, 270 insertions(+), 46 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index 0528f3b..cbef837 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1215,6 +1215,7 @@ enum {
 	BNX2X_SP_RTNL_ENABLE_SRIOV,
 	BNX2X_SP_RTNL_VFPF_MCAST,
 	BNX2X_SP_RTNL_VFPF_STORM_RX_MODE,
+	BNX2X_SP_RTNL_HYPERVISOR_VLAN,
 };
 
 
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
index aee7671..c64b3e3 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.h
@@ -496,7 +496,10 @@ netdev_tx_t bnx2x_start_xmit(struct sk_buff *skb, struct net_device *dev);
 /* setup_tc callback */
 int bnx2x_setup_tc(struct net_device *dev, u8 num_tc);
 
+int bnx2x_get_vf_config(struct net_device *dev, int vf,
+			struct ifla_vf_info *ivi);
 int bnx2x_set_vf_mac(struct net_device *dev, int queue, u8 *mac);
+int bnx2x_set_vf_vlan(struct net_device *netdev, int vf, u16 vlan, u8 qos);
 
 /* select_queue callback */
 u16 bnx2x_select_queue(struct net_device *dev, struct sk_buff *skb);
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index 2b7930c..b643912 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -75,8 +75,6 @@
 #define FW_FILE_NAME_E1H	"bnx2x/bnx2x-e1h-" FW_FILE_VERSION ".fw"
 #define FW_FILE_NAME_E2		"bnx2x/bnx2x-e2-" FW_FILE_VERSION ".fw"
 
-#define MAC_LEADING_ZERO_CNT (ALIGN(ETH_ALEN, sizeof(u32)) - ETH_ALEN)
-
 /* Time in jiffies before concluding the transmitter is hung */
 #define TX_TIMEOUT		(5*HZ)
 
@@ -3230,16 +3228,29 @@ static void bnx2x_drv_info_ether_stat(struct bnx2x *bp)
 {
 	struct eth_stats_info *ether_stat =
 		&bp->slowpath->drv_info_to_mcp.ether_stat;
+	struct bnx2x_vlan_mac_obj *mac_obj =
+		&bp->sp_objs->mac_obj;
+	int i;
 
 	strlcpy(ether_stat->version, DRV_MODULE_VERSION,
 		ETH_STAT_INFO_VERSION_LEN);
 
-	bp->sp_objs[0].mac_obj.get_n_elements(bp, &bp->sp_objs[0].mac_obj,
-					DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED,
-					ether_stat->mac_local);
-
+	/* get DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED macs, placing them in the
+	 * mac_local field in ether_stat struct. The base address is offset by 2
+	 * bytes to account for the field being 8 bytes but a mac address is
+	 * only 6 bytes. Likewise, the stride for the get_n_elements function is
+	 * 2 bytes to compensate from the 6 bytes of a mac to the 8 bytes
+	 * allocated by the ether_stat struct, so the macs will land in their
+	 * proper positions.
+	 */
+	for (i = 0; i < DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED; i++)
+		memset(ether_stat->mac_local + i, 0,
+		       sizeof(ether_stat->mac_local[0]));
+	mac_obj->get_n_elements(bp, &bp->sp_objs[0].mac_obj,
+				DRV_INFO_ETH_STAT_NUM_MACS_REQUIRED,
+				ether_stat->mac_local + MAC_PAD, MAC_PAD,
+				ETH_ALEN);
 	ether_stat->mtu_size = bp->dev->mtu;
-
 	if (bp->dev->features & NETIF_F_RXCSUM)
 		ether_stat->feature_flags |= FEATURE_ETH_CHKSUM_OFFLOAD_MASK;
 	if (bp->dev->features & NETIF_F_TSO)
@@ -3261,8 +3272,7 @@ static void bnx2x_drv_info_fcoe_stat(struct bnx2x *bp)
 	if (!CNIC_LOADED(bp))
 		return;
 
-	memcpy(fcoe_stat->mac_local + MAC_LEADING_ZERO_CNT,
-	       bp->fip_mac, ETH_ALEN);
+	memcpy(fcoe_stat->mac_local + MAC_PAD, bp->fip_mac, ETH_ALEN);
 
 	fcoe_stat->qos_priority =
 		app->traffic_type_priority[LLFC_TRAFFIC_TYPE_FCOE];
@@ -3364,8 +3374,8 @@ static void bnx2x_drv_info_iscsi_stat(struct bnx2x *bp)
 	if (!CNIC_LOADED(bp))
 		return;
 
-	memcpy(iscsi_stat->mac_local + MAC_LEADING_ZERO_CNT,
-	       bp->cnic_eth_dev.iscsi_mac, ETH_ALEN);
+	memcpy(iscsi_stat->mac_local + MAC_PAD, bp->cnic_eth_dev.iscsi_mac,
+	       ETH_ALEN);
 
 	iscsi_stat->qos_priority =
 		app->traffic_type_priority[LLFC_TRAFFIC_TYPE_ISCSI];
@@ -9528,6 +9538,11 @@ sp_rtnl_not_reset:
 		bnx2x_vfpf_storm_rx_mode(bp);
 	}
 
+	if (test_and_clear_bit(BNX2X_SP_RTNL_HYPERVISOR_VLAN,
+			       &bp->sp_rtnl_state)) {
+		bnx2x_pf_set_vfs_vlan(bp);
+	}
+
 	/* work which needs rtnl lock not-taken (as it takes the lock itself and
 	 * can be called from other contexts as well)
 	 */
@@ -11801,6 +11816,8 @@ static const struct net_device_ops bnx2x_netdev_ops = {
 	.ndo_setup_tc		= bnx2x_setup_tc,
 #ifdef CONFIG_BNX2X_SRIOV
 	.ndo_set_vf_mac		= bnx2x_set_vf_mac,
+	.ndo_set_vf_vlan        = bnx2x_set_vf_vlan,
+	.ndo_get_vf_config	= bnx2x_get_vf_config,
 #endif
 #ifdef NETDEV_FCOE_WWNN
 	.ndo_fcoe_get_wwn	= bnx2x_fcoe_get_wwn,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
index f701b72..7011997 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.c
@@ -30,8 +30,6 @@
 
 #define BNX2X_MAX_EMUL_MULTI		16
 
-#define MAC_LEADING_ZERO_CNT (ALIGN(ETH_ALEN, sizeof(u32)) - ETH_ALEN)
-
 /**** Exe Queue interfaces ****/
 
 /**
@@ -768,10 +766,10 @@ void bnx2x_vlan_mac_h_read_unlock(struct bnx2x *bp,
 }
 
 static int bnx2x_get_n_elements(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
-				int n, u8 *buf)
+				int n, u8 *base, u8 stride, u8 size)
 {
 	struct bnx2x_vlan_mac_registry_elem *pos;
-	u8 *next = buf;
+	u8 *next = base;
 	int counter = 0, read_lock;
 
 	DP(BNX2X_MSG_SP, "get_n_elements - taking vlan_mac_lock (reader)\n");
@@ -782,21 +780,12 @@ static int bnx2x_get_n_elements(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
 	/* traverse list */
 	list_for_each_entry(pos, &o->head, link) {
 		if (counter < n) {
-			/* place leading zeroes in buffer */
-			memset(next, 0, MAC_LEADING_ZERO_CNT);
-
-			/* place mac after leading zeroes*/
-			memcpy(next + MAC_LEADING_ZERO_CNT, pos->u.mac.mac,
-			       ETH_ALEN);
-
-			/* calculate address of next element and
-			 * advance counter
-			 */
+			memcpy(next, &pos->u, size);
 			counter++;
-			next = buf + counter * ALIGN(ETH_ALEN, sizeof(u32));
+			DP(BNX2X_MSG_SP, "copied element number %d to address %p element was:\n",
+			   counter, next);
+			next += stride + size;
 
-			DP(BNX2X_MSG_SP, "copied element number %d to address %p element was %pM\n",
-			   counter, next, pos->u.mac.mac);
 		}
 	}
 
@@ -2424,6 +2413,7 @@ void bnx2x_init_vlan_obj(struct bnx2x *bp,
 		vlan_obj->check_move        = bnx2x_check_move;
 		vlan_obj->ramrod_cmd        =
 			RAMROD_CMD_ID_ETH_CLASSIFICATION_RULES;
+		vlan_obj->get_n_elements    = bnx2x_get_n_elements;
 
 		/* Exe Queue */
 		bnx2x_exe_queue_init(bp,
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
index 3ec795b..426b15e 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sp.h
@@ -321,8 +321,9 @@ struct bnx2x_vlan_mac_obj {
 	 *
 	 * @return number of copied bytes
 	 */
-	int (*get_n_elements)(struct bnx2x *bp, struct bnx2x_vlan_mac_obj *o,
-			      int n, u8 *buf);
+	int (*get_n_elements)(struct bnx2x *bp,
+			      struct bnx2x_vlan_mac_obj *o, int n, u8 *base,
+			      u8 stride, u8 size);
 
 	/**
 	 * Checks if ADD-ramrod with the given params may be performed.
@@ -850,6 +851,7 @@ enum bnx2x_q_type {
 #define BNX2X_MULTI_TX_COS_E3B0			3
 #define BNX2X_MULTI_TX_COS			3 /* Maximum possible */
 
+#define MAC_PAD (ALIGN(ETH_ALEN, sizeof(u32)) - ETH_ALEN)
 
 struct bnx2x_queue_init_params {
 	struct {
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
index 89c6fa1..80fb4dd 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
@@ -20,7 +20,9 @@
 #include "bnx2x.h"
 #include "bnx2x_init.h"
 #include "bnx2x_cmn.h"
+#include "bnx2x_sp.h"
 #include <linux/crc32.h>
+#include <linux/if_vlan.h>
 
 /* General service functions */
 static void storm_memset_vf_to_pf(struct bnx2x *bp, u16 abs_fid,
@@ -965,6 +967,12 @@ op_err:
 	BNX2X_ERR("QSETUP[%d:%d] error: rc %d\n", vf->abs_vfid, qid, vfop->rc);
 op_done:
 	case BNX2X_VFOP_QSETUP_DONE:
+		vf->cfg_flags |= VF_CFG_VLAN;
+		smp_mb__before_clear_bit();
+		set_bit(BNX2X_SP_RTNL_HYPERVISOR_VLAN,
+			&bp->sp_rtnl_state);
+		smp_mb__after_clear_bit();
+		schedule_delayed_work(&bp->sp_rtnl_task, 0);
 		bnx2x_vfop_end(bp, vf, vfop);
 		return;
 	default:
@@ -3036,6 +3044,80 @@ void bnx2x_enable_sriov(struct bnx2x *bp)
 		DP(BNX2X_MSG_IOV, "sriov enabled\n");
 }
 
+void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp)
+{
+	int vfidx;
+	struct pf_vf_bulletin_content *bulletin;
+
+	DP(BNX2X_MSG_IOV, "configuring vlan for VFs from sp-task\n");
+	for_each_vf(bp, vfidx) {
+	bulletin = BP_VF_BULLETIN(bp, vfidx);
+		if (BP_VF(bp, vfidx)->cfg_flags & VF_CFG_VLAN)
+			bnx2x_set_vf_vlan(bp->dev, vfidx, bulletin->vlan, 0);
+	}
+}
+
+int bnx2x_vf_ndo_sanity(struct bnx2x *bp, int vfidx)
+{
+	if (!IS_SRIOV(bp)) {
+		BNX2X_ERR("bnx2x_set_vf_mac called though sriov is disabled\n");
+		return -EINVAL;
+	}
+
+	if (vfidx >= BNX2X_NR_VIRTFN(bp)) {
+		BNX2X_ERR("vf ndo called for uninitialized VF. vfidx was %d BNX2X_NR_VIRTFN was %d\n",
+			  vfidx, BNX2X_NR_VIRTFN(bp));
+		return -EINVAL;
+	}
+	return 0;
+}
+
+int bnx2x_get_vf_config(struct net_device *dev, int vfidx,
+			struct ifla_vf_info *ivi)
+{
+	struct bnx2x *bp = netdev_priv(dev);
+	struct bnx2x_virtf *vf = BP_VF(bp, vfidx);
+	struct bnx2x_vlan_mac_obj *mac_obj = &bnx2x_vfq(vf, 0, mac_obj);
+	struct bnx2x_vlan_mac_obj *vlan_obj = &bnx2x_vfq(vf, 0, vlan_obj);
+	struct pf_vf_bulletin_content *bulletin = BP_VF_BULLETIN(bp, vfidx);
+	int rc;
+
+	/* sanity */
+	rc = bnx2x_vf_ndo_sanity(bp, vfidx);
+	if (rc)
+		return rc;
+
+	ivi->vf = vfidx;
+	ivi->qos = 0;
+	ivi->tx_rate = 10000; /* always 10G. TBA take from link struct */
+	ivi->spoofchk = 1; /*always enabled */
+	if (vf->state == VF_ENABLED) {
+		/* mac and vlan are in vlan_mac objects */
+		mac_obj->get_n_elements(bp, mac_obj, 1, (u8 *)&ivi->mac,
+					0, ETH_ALEN);
+		vlan_obj->get_n_elements(bp, vlan_obj, 1, (u8 *)&ivi->vlan,
+					 0, VLAN_HLEN);
+	} else {
+		/* mac */
+		if (bulletin->valid_bitmap & (1 << MAC_ADDR_VALID))
+			/* mac configured by ndo so its in bulletin board */
+			memcpy(&ivi->mac, bulletin->mac, ETH_ALEN);
+		else
+			/* funtion has not been loaded yet. Show mac as 0s */
+			memset(&ivi->mac, 0, ETH_ALEN);
+
+		/* vlan */
+		if (bulletin->valid_bitmap & (1 << VLAN_VALID))
+			/* vlan configured by ndo so its in bulletin board */
+			memcpy(&ivi->vlan, &bulletin->vlan, VLAN_HLEN);
+		else
+			/* funtion has not been loaded yet. Show vlans as 0s */
+			memset(&ivi->vlan, 0, VLAN_HLEN);
+	}
+
+	return 0;
+}
+
 /* New mac for VF. Consider these cases:
  * 1. VF hasn't been acquired yet - save the mac in local bulletin board and
  *    supply at acquire.
@@ -3051,23 +3133,19 @@ void bnx2x_enable_sriov(struct bnx2x *bp)
  * VF to configure any mac for itself except for this mac. In case of a race
  * where the VF fails to see the new post on its bulletin board before sending a
  * mac configuration request, the PF will simply fail the request and VF can try
- * again after consulting its bulletin board
+ * again after consulting its bulletin board.
  */
-int bnx2x_set_vf_mac(struct net_device *dev, int queue, u8 *mac)
+int bnx2x_set_vf_mac(struct net_device *dev, int vfidx, u8 *mac)
 {
 	struct bnx2x *bp = netdev_priv(dev);
-	int rc, q_logical_state, vfidx = queue;
+	int rc, q_logical_state;
 	struct bnx2x_virtf *vf = BP_VF(bp, vfidx);
 	struct pf_vf_bulletin_content *bulletin = BP_VF_BULLETIN(bp, vfidx);
 
-	/* if SRIOV is disabled there is nothing to do (and somewhere, someone
-	 * has erred).
-	 */
-	if (!IS_SRIOV(bp)) {
-		BNX2X_ERR("bnx2x_set_vf_mac called though sriov is disabled\n");
-		return -EINVAL;
-	}
-
+	/* sanity */
+	rc = bnx2x_vf_ndo_sanity(bp, vfidx);
+	if (rc)
+		return rc;
 	if (!is_valid_ether_addr(mac)) {
 		BNX2X_ERR("mac address invalid\n");
 		return -EINVAL;
@@ -3092,7 +3170,7 @@ int bnx2x_set_vf_mac(struct net_device *dev, int queue, u8 *mac)
 	if (vf->state == VF_ENABLED &&
 	    q_logical_state == BNX2X_Q_LOGICAL_STATE_ACTIVE) {
 		/* configure the mac in device on this vf's queue */
-		unsigned long flags = 0;
+		unsigned long ramrod_flags = 0;
 		struct bnx2x_vlan_mac_obj *mac_obj = &bnx2x_vfq(vf, 0, mac_obj);
 
 		/* must lock vfpf channel to protect against vf flows */
@@ -3113,14 +3191,133 @@ int bnx2x_set_vf_mac(struct net_device *dev, int queue, u8 *mac)
 		}
 
 		/* configure the new mac to device */
-		__set_bit(RAMROD_COMP_WAIT, &flags);
+		__set_bit(RAMROD_COMP_WAIT, &ramrod_flags);
 		bnx2x_set_mac_one(bp, (u8 *)&bulletin->mac, mac_obj, true,
-				  BNX2X_ETH_MAC, &flags);
+				  BNX2X_ETH_MAC, &ramrod_flags);
 
 		bnx2x_unlock_vf_pf_channel(bp, vf, CHANNEL_TLV_PF_SET_MAC);
 	}
 
-	return rc;
+	return 0;
+}
+
+int bnx2x_set_vf_vlan(struct net_device *dev, int vfidx, u16 vlan, u8 qos)
+{
+	struct bnx2x *bp = netdev_priv(dev);
+	int rc, q_logical_state;
+	struct bnx2x_virtf *vf = BP_VF(bp, vfidx);
+	struct pf_vf_bulletin_content *bulletin = BP_VF_BULLETIN(bp, vfidx);
+
+	/* sanity */
+	rc = bnx2x_vf_ndo_sanity(bp, vfidx);
+	if (rc)
+		return rc;
+
+	if (vlan > 4095) {
+		BNX2X_ERR("illegal vlan value %d\n", vlan);
+		return -EINVAL;
+	}
+
+	DP(BNX2X_MSG_IOV, "configuring VF %d with VLAN %d qos %d\n",
+	   vfidx, vlan, 0);
+
+	/* update PF's copy of the VF's bulletin. No point in posting the vlan
+	 * to the VF since it doesn't have anything to do with it. But it useful
+	 * to store it here in case the VF is not up yet and we can only
+	 * configure the vlan later when it does.
+	 */
+	bulletin->valid_bitmap |= 1 << VLAN_VALID;
+	bulletin->vlan = vlan;
+
+	/* is vf initialized and queue set up? */
+	q_logical_state =
+		bnx2x_get_q_logical_state(bp, &bnx2x_vfq(vf, 0, sp_obj));
+	if (vf->state == VF_ENABLED &&
+	    q_logical_state == BNX2X_Q_LOGICAL_STATE_ACTIVE) {
+		/* configure the vlan in device on this vf's queue */
+		unsigned long ramrod_flags = 0;
+		unsigned long vlan_mac_flags = 0;
+		struct bnx2x_vlan_mac_obj *vlan_obj =
+			&bnx2x_vfq(vf, 0, vlan_obj);
+		struct bnx2x_vlan_mac_ramrod_params ramrod_param;
+		struct bnx2x_queue_state_params q_params = {NULL};
+		struct bnx2x_queue_update_params *update_params;
+
+		memset(&ramrod_param, 0, sizeof(ramrod_param));
+
+		/* must lock vfpf channel to protect against vf flows */
+		bnx2x_lock_vf_pf_channel(bp, vf, CHANNEL_TLV_PF_SET_VLAN);
+
+		/* remove existing vlans */
+		__set_bit(RAMROD_COMP_WAIT, &ramrod_flags);
+		rc = vlan_obj->delete_all(bp, vlan_obj, &vlan_mac_flags,
+					  &ramrod_flags);
+		if (rc) {
+			BNX2X_ERR("failed to delete vlans\n");
+			return -EINVAL;
+		}
+
+		/* send queue update ramrod to configure default vlan and silent
+		 * vlan removal
+		 */
+		__set_bit(RAMROD_COMP_WAIT, &q_params.ramrod_flags);
+		q_params.cmd = BNX2X_Q_CMD_UPDATE;
+		q_params.q_obj = &bnx2x_vfq(vf, 0, sp_obj);
+		update_params = &q_params.params.update;
+		__set_bit(BNX2X_Q_UPDATE_DEF_VLAN_EN_CHNG,
+			  &update_params->update_flags);
+		__set_bit(BNX2X_Q_UPDATE_SILENT_VLAN_REM_CHNG,
+			  &update_params->update_flags);
+
+		if (vlan == 0) {
+			/* if vlan is 0 then we want to leave the VF traffic
+			 * untagged, and leave the incoming traffic untouched
+			 * (i.e. do not remove any vlan tags).
+			 */
+			__clear_bit(BNX2X_Q_UPDATE_DEF_VLAN_EN,
+				    &update_params->update_flags);
+			__clear_bit(BNX2X_Q_UPDATE_SILENT_VLAN_REM,
+				    &update_params->update_flags);
+		} else {
+			/* configure the new vlan to device */
+			__set_bit(RAMROD_COMP_WAIT, &ramrod_flags);
+			ramrod_param.vlan_mac_obj = vlan_obj;
+			ramrod_param.ramrod_flags = ramrod_flags;
+			ramrod_param.user_req.u.vlan.vlan = vlan;
+			ramrod_param.user_req.cmd = BNX2X_VLAN_MAC_ADD;
+			rc = bnx2x_config_vlan_mac(bp, &ramrod_param);
+			if (rc) {
+				BNX2X_ERR("failed to configure vlan\n");
+				return -EINVAL;
+			}
+
+			/* configure default vlan to vf queue and set silent
+			 * vlan removal (the vf remains unaware of this vlan).
+			 */
+			update_params = &q_params.params.update;
+			__set_bit(BNX2X_Q_UPDATE_DEF_VLAN_EN,
+				  &update_params->update_flags);
+			__set_bit(BNX2X_Q_UPDATE_SILENT_VLAN_REM,
+				  &update_params->update_flags);
+			update_params->def_vlan = vlan;
+		}
+
+		/* Update the Queue state */
+		rc = bnx2x_queue_state_change(bp, &q_params);
+		if (rc) {
+			BNX2X_ERR("Failed to configure default VLAN\n");
+			return rc;
+		}
+
+		/* clear the flag indicating that this VF needs its vlan
+		 * (will only be set if the HV configured th Vlan before vf was
+		 * and we were called because the VF came up later
+		 */
+		vf->cfg_flags &= ~VF_CFG_VLAN;
+
+		bnx2x_unlock_vf_pf_channel(bp, vf, CHANNEL_TLV_PF_SET_VLAN);
+	}
+	return 0;
 }
 
 /* crc is the first field in the bulletin board. compute the crc over the
@@ -3172,6 +3369,10 @@ enum sample_bulletin_result bnx2x_sample_bulletin(struct bnx2x *bp)
 		memcpy(bp->dev->dev_addr, bulletin.mac, ETH_ALEN);
 	}
 
+	/* the vlan in bulletin board is valid and is new */
+	if (bulletin.valid_bitmap & 1 << VLAN_VALID)
+		memcpy(&bulletin.vlan, &bp->old_bulletin.vlan, VLAN_HLEN);
+
 	/* copy new bulletin board to bp */
 	bp->old_bulletin = bulletin;
 
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
index b405017..33d4951 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
@@ -193,6 +193,7 @@ struct bnx2x_virtf {
 #define VF_CFG_TPA		0x0004
 #define VF_CFG_INT_SIMD		0x0008
 #define VF_CACHE_LINE		0x0010
+#define VF_CFG_VLAN		0x0020
 
 	u8 state;
 #define VF_FREE		0	/* VF ready to be acquired holds no resc */
@@ -757,6 +758,7 @@ static inline int bnx2x_vf_headroom(struct bnx2x *bp)
 {
 	return bp->vfdb->sriov.nr_virtfn * BNX2X_CLIENTS_PER_VF;
 }
+void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp);
 
 #else /* CONFIG_BNX2X_SRIOV */
 
@@ -804,6 +806,7 @@ static inline enum sample_bulletin_result bnx2x_sample_bulletin(struct bnx2x *bp
 
 static inline int bnx2x_vf_map_doorbells(struct bnx2x *bp) {return 0; }
 static inline int bnx2x_vf_pci_alloc(struct bnx2x *bp) {return 0; }
+static inline void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp) {}
 
 #endif /* CONFIG_BNX2X_SRIOV */
 #endif /* bnx2x_sriov.h */
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.h
index bfc80ba..41708fa 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_vfpf.h
@@ -328,9 +328,15 @@ struct pf_vf_bulletin_content {
 #define MAC_ADDR_VALID		0	/* alert the vf that a new mac address
 					 * is available for it
 					 */
+#define VLAN_VALID		1	/* when set, the vf should not access
+					 * the vfpf channel
+					 */
 
 	u8 mac[ETH_ALEN];
-	u8 padding[2];
+	u8 mac_padding[2];
+
+	u16 vlan;
+	u8 vlan_padding[6];
 };
 
 union pf_vf_bulletin {
@@ -353,6 +359,7 @@ enum channel_tlvs {
 	CHANNEL_TLV_LIST_END,
 	CHANNEL_TLV_FLR,
 	CHANNEL_TLV_PF_SET_MAC,
+	CHANNEL_TLV_PF_SET_VLAN,
 	CHANNEL_TLV_MAX
 };
 
-- 
1.7.9.GIT

[PATCH net-next 8/9] bnx2x: Control number of vfs dynamically

[Yuval Mintz]

From: Ariel Elior <ariele@broadcom.com>

1. Support sysfs interface for getting the maximal number of virtual functions
   of a given physical function.
2. Support sysfs interface for getting and setting the current number of
   virtual functions.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x.h       |    2 +
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c  |   32 ++-------
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c |   81 +++++++++++++++++----
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h |   10 ++-
 4 files changed, 83 insertions(+), 42 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
index cbef837..63e4185 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x.h
@@ -1282,6 +1282,8 @@ struct bnx2x {
 	dma_addr_t		pf2vf_bulletin_mapping;
 
 	struct pf_vf_bulletin_content	old_bulletin;
+
+	u16 requested_nr_virtfn;
 #endif /* CONFIG_BNX2X_SRIOV */
 
 	struct net_device	*dev;
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
index b643912..adc7ea4 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c
@@ -11427,26 +11427,6 @@ static int bnx2x_init_bp(struct bnx2x *bp)
  * net_device service functions
  */
 
-static int bnx2x_open_epilog(struct bnx2x *bp)
-{
-	/* Enable sriov via delayed work. This must be done via delayed work
-	 * because it causes the probe of the vf devices to be run, which invoke
-	 * register_netdevice which must have rtnl lock taken. As we are holding
-	 * the lock right now, that could only work if the probe would not take
-	 * the lock. However, as the probe of the vf may be called from other
-	 * contexts as well (such as passthrough to vm failes) it can't assume
-	 * the lock is being held for it. Using delayed work here allows the
-	 * probe code to simply take the lock (i.e. wait for it to be released
-	 * if it is being held).
-	 */
-	smp_mb__before_clear_bit();
-	set_bit(BNX2X_SP_RTNL_ENABLE_SRIOV, &bp->sp_rtnl_state);
-	smp_mb__after_clear_bit();
-	schedule_delayed_work(&bp->sp_rtnl_task, 0);
-
-	return 0;
-}
-
 /* called with rtnl_lock */
 static int bnx2x_open(struct net_device *dev)
 {
@@ -12502,13 +12482,8 @@ static int bnx2x_init_one(struct pci_dev *pdev,
 			goto init_one_exit;
 	}
 
-	/* Enable SRIOV if capability found in configuration space.
-	 * Once the generic SR-IOV framework makes it in from the
-	 * pci tree this will be revised, to allow dynamic control
-	 * over the number of VFs. Right now, change the num of vfs
-	 * param below to enable SR-IOV.
-	 */
-	rc = bnx2x_iov_init_one(bp, int_mode, 0/*num vfs*/);
+	/* Enable SRIOV if capability found in configuration space */
+	rc = bnx2x_iov_init_one(bp, int_mode, BNX2X_MAX_NUM_OF_VFS);
 	if (rc)
 		goto init_one_exit;
 
@@ -12824,6 +12799,9 @@ static struct pci_driver bnx2x_pci_driver = {
 	.suspend     = bnx2x_suspend,
 	.resume      = bnx2x_resume,
 	.err_handler = &bnx2x_err_handler,
+#ifdef CONFIG_BNX2X_SRIOV
+	.sriov_configure = bnx2x_sriov_configure,
+#endif
 };
 
 static int __init bnx2x_init(void)
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
index 80fb4dd..c05a37a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.c
@@ -1979,8 +1979,11 @@ int bnx2x_iov_init_one(struct bnx2x *bp, int int_mode_param,
 	if (iov->total == 0)
 		goto failed;
 
-	/* calculate the actual number of VFs */
-	iov->nr_virtfn = min_t(u16, iov->total, (u16)num_vfs_param);
+	/* module param can limit the resources allocated to VFs*/
+	iov->nr_virtfn = min_t(u16, iov->total, num_vfs_param);
+
+	DP(BNX2X_MSG_IOV, "num_vfs_param was %d, nr_virtfn was %d\n",
+	   num_vfs_param, iov->nr_virtfn);
 
 	/* allocate the vf array */
 	bp->vfdb->vfs = kzalloc(sizeof(struct bnx2x_virtf) *
@@ -3027,21 +3030,50 @@ void bnx2x_unlock_vf_pf_channel(struct bnx2x *bp, struct bnx2x_virtf *vf,
 	vf->op_current = CHANNEL_TLV_NONE;
 }
 
-void bnx2x_enable_sriov(struct bnx2x *bp)
+int bnx2x_sriov_configure(struct pci_dev *dev, int num_vfs_param)
 {
-	int rc = 0;
 
-	/* disbale sriov in case it is still enabled */
-	pci_disable_sriov(bp->pdev);
-	DP(BNX2X_MSG_IOV, "sriov disabled\n");
+	struct bnx2x *bp = netdev_priv(pci_get_drvdata(dev));
 
-	/* enable sriov */
-	DP(BNX2X_MSG_IOV, "vf num (%d)\n", (bp->vfdb->sriov.nr_virtfn));
-	rc = pci_enable_sriov(bp->pdev, (bp->vfdb->sriov.nr_virtfn));
-	if (rc)
+	DP(BNX2X_MSG_IOV, "bnx2x_sriov_configure called with %d, BNX2X_NR_VIRTFN(bp) was %d\n",
+	   num_vfs_param, BNX2X_NR_VIRTFN(bp));
+
+	/* pci_disable_sriov removes non passed-through vfs, releasing them */
+	if (num_vfs_param == 0) {
+		pci_disable_sriov(dev);
+		return 0;
+	}
+
+	/* we are always bound by the total_vfs in the configuration space */
+	if (num_vfs_param > BNX2X_NR_VIRTFN(bp)) {
+		BNX2X_ERR("truncating requested number of VFs (%d) down to maximum allowed (%d)\n",
+			  num_vfs_param, BNX2X_NR_VIRTFN(bp));
+		num_vfs_param = BNX2X_NR_VIRTFN(bp);
+	}
+
+	/* we want to register the requested amount whether or not we can enable
+	 * SR-IOV right now.
+	 */
+	bp->requested_nr_virtfn = num_vfs_param;
+
+	/* HW channel is only operational when PF is up */
+	if (bp->state != BNX2X_STATE_OPEN)
+		return num_vfs_param;
+
+	return bnx2x_enable_sriov(bp);
+}
+
+int bnx2x_enable_sriov(struct bnx2x *bp)
+{
+	int rc = 0, req_vfs = bp->requested_nr_virtfn;
+
+	rc = pci_enable_sriov(bp->pdev, req_vfs);
+	if (rc) {
 		BNX2X_ERR("pci_enable_sriov failed with %d\n", rc);
-	else
-		DP(BNX2X_MSG_IOV, "sriov enabled\n");
+		return rc;
+	}
+	DP(BNX2X_MSG_IOV, "sriov enabled (%d vfs)\n", req_vfs);
+	return req_vfs;
 }
 
 void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp)
@@ -3404,3 +3436,26 @@ alloc_mem_err:
 		       sizeof(union pf_vf_bulletin));
 	return -ENOMEM;
 }
+
+int bnx2x_open_epilog(struct bnx2x *bp)
+{
+	/* Enable sriov via delayed work. This must be done via delayed work
+	 * because it causes the probe of the vf devices to be run, which invoke
+	 * register_netdevice which must have rtnl lock taken. As we are holding
+	 * the lock right now, that could only work if the probe would not take
+	 * the lock. However, as the probe of the vf may be called from other
+	 * contexts as well (such as passthrough to vm failes) it can't assume
+	 * the lock is being held for it. Using delayed work here allows the
+	 * probe code to simply take the lock (i.e. wait for it to be released
+	 * if it is being held). We only want to do this if the number of VFs
+	 * was set before PF driver was loaded.
+	 */
+	if (IS_SRIOV(bp) && BNX2X_NR_VIRTFN(bp)) {
+		smp_mb__before_clear_bit();
+		set_bit(BNX2X_SP_RTNL_ENABLE_SRIOV, &bp->sp_rtnl_state);
+		smp_mb__after_clear_bit();
+		schedule_delayed_work(&bp->sp_rtnl_task, 0);
+	}
+
+	return 0;
+}
diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
index 33d4951..2d26c6a 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_sriov.h
@@ -51,6 +51,8 @@ struct bnx2x_sriov {
 	u16 stride;		/* following VF stride */
 	u32 pgsz;		/* page size for BAR alignment */
 	u8 link;		/* Function Dependency Link */
+
+	u16 requested_nr_virtfn;
 };
 
 /* bars */
@@ -753,12 +755,14 @@ static inline int bnx2x_vf_ustorm_prods_offset(struct bnx2x *bp,
 enum sample_bulletin_result bnx2x_sample_bulletin(struct bnx2x *bp);
 void bnx2x_vf_map_doorbells(struct bnx2x *bp);
 int bnx2x_vf_pci_alloc(struct bnx2x *bp);
-void bnx2x_enable_sriov(struct bnx2x *bp);
+int bnx2x_enable_sriov(struct bnx2x *bp);
 static inline int bnx2x_vf_headroom(struct bnx2x *bp)
 {
 	return bp->vfdb->sriov.nr_virtfn * BNX2X_CLIENTS_PER_VF;
 }
 void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp);
+int bnx2x_sriov_configure(struct pci_dev *dev, int num_vfs);
+int bnx2x_open_epilog(struct bnx2x *bp);
 
 #else /* CONFIG_BNX2X_SRIOV */
 
@@ -781,7 +785,7 @@ static inline void bnx2x_iov_init_dmae(struct bnx2x *bp) {}
 static inline int bnx2x_iov_init_one(struct bnx2x *bp, int int_mode_param,
 				     int num_vfs_param) {return 0; }
 static inline void bnx2x_iov_remove_one(struct bnx2x *bp) {}
-static inline void bnx2x_enable_sriov(struct bnx2x *bp) {}
+static inline int bnx2x_enable_sriov(struct bnx2x *bp) {return 0; }
 static inline int bnx2x_vfpf_acquire(struct bnx2x *bp,
 				     u8 tx_count, u8 rx_count) {return 0; }
 static inline int bnx2x_vfpf_release(struct bnx2x *bp) {return 0; }
@@ -807,6 +811,8 @@ static inline enum sample_bulletin_result bnx2x_sample_bulletin(struct bnx2x *bp
 static inline int bnx2x_vf_map_doorbells(struct bnx2x *bp) {return 0; }
 static inline int bnx2x_vf_pci_alloc(struct bnx2x *bp) {return 0; }
 static inline void bnx2x_pf_set_vfs_vlan(struct bnx2x *bp) {}
+static inline int bnx2x_sriov_configure(struct pci_dev *dev, int num_vfs) {return 0; }
+static inline int bnx2x_open_epilog(struct bnx2x *bp) {return 0; }
 
 #endif /* CONFIG_BNX2X_SRIOV */
 #endif /* bnx2x_sriov.h */
-- 
1.7.9.GIT

[PATCH net-next 9/9] bnx2x: Get gso_segs from FW

[Yuval Mintz]

Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c |    9 +++++----
 1 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index 525c2b5..dc8d135 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -451,7 +451,8 @@ static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue,
  * Compute number of aggregated segments, and gso_type.
  */
 static void bnx2x_set_gro_params(struct sk_buff *skb, u16 parsing_flags,
-				 u16 len_on_bd, unsigned int pkt_len)
+				 u16 len_on_bd, unsigned int pkt_len,
+				 u16 num_of_coalesced_segs)
 {
 	/* TPA aggregation won't have either IP options or TCP options
 	 * other than timestamp or IPv6 extension headers.
@@ -480,8 +481,7 @@ static void bnx2x_set_gro_params(struct sk_buff *skb, u16 parsing_flags,
 	/* tcp_gro_complete() will copy NAPI_GRO_CB(skb)->count
 	 * to skb_shinfo(skb)->gso_segs
 	 */
-	NAPI_GRO_CB(skb)->count = DIV_ROUND_UP(pkt_len - hdrs_len,
-					       skb_shinfo(skb)->gso_size);
+	NAPI_GRO_CB(skb)->count = num_of_coalesced_segs;
 }
 
 static int bnx2x_alloc_rx_sge(struct bnx2x *bp,
@@ -537,7 +537,8 @@ static int bnx2x_fill_frag_skb(struct bnx2x *bp, struct bnx2x_fastpath *fp,
 	/* This is needed in order to enable forwarding support */
 	if (frag_size)
 		bnx2x_set_gro_params(skb, tpa_info->parsing_flags, len_on_bd,
-				     le16_to_cpu(cqe->pkt_len));
+				     le16_to_cpu(cqe->pkt_len),
+				     le16_to_cpu(cqe->num_of_coalesced_segs));
 
 #ifdef BNX2X_STOP_ON_ERROR
 	if (pages > min_t(u32, 8, MAX_SKB_FRAGS) * SGE_PAGES) {
-- 
1.7.9.GIT