X-Git-Url: http://git.cascardo.eti.br/?a=blobdiff_plain;f=drivers%2Fnet%2Fethernet%2Fmellanox%2Fmlx4%2Fcmd.c;h=7652eed4bbc823806b3e87f25dbab32e4709b152;hb=55ad359225b2;hp=5c93d1451c449e328fe05505bc848a146af3df9e;hpb=64ec45bff6b3dade2643ed4c0f688a15ecf46ea2;p=cascardo%2Flinux.git

diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index 5c93d1451c44..7652eed4bbc8 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -42,6 +42,7 @@
 #include <linux/mlx4/device.h>
 #include <linux/semaphore.h>
 #include <rdma/ib_smi.h>
+#include <linux/delay.h>
 
 #include <asm/io.h>
 
@@ -182,6 +183,72 @@ static u8 mlx4_errno_to_status(int errno)
 	}
 }
 
+static int mlx4_internal_err_ret_value(struct mlx4_dev *dev, u16 op,
+				       u8 op_modifier)
+{
+	switch (op) {
+	case MLX4_CMD_UNMAP_ICM:
+	case MLX4_CMD_UNMAP_ICM_AUX:
+	case MLX4_CMD_UNMAP_FA:
+	case MLX4_CMD_2RST_QP:
+	case MLX4_CMD_HW2SW_EQ:
+	case MLX4_CMD_HW2SW_CQ:
+	case MLX4_CMD_HW2SW_SRQ:
+	case MLX4_CMD_HW2SW_MPT:
+	case MLX4_CMD_CLOSE_HCA:
+	case MLX4_QP_FLOW_STEERING_DETACH:
+	case MLX4_CMD_FREE_RES:
+	case MLX4_CMD_CLOSE_PORT:
+		return CMD_STAT_OK;
+
+	case MLX4_CMD_QP_ATTACH:
+		/* On Detach case return success */
+		if (op_modifier == 0)
+			return CMD_STAT_OK;
+		return mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+
+	default:
+		return mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+	}
+}
+
+static int mlx4_closing_cmd_fatal_error(u16 op, u8 fw_status)
+{
+	/* Any error during the closing commands below is considered fatal */
+	if (op == MLX4_CMD_CLOSE_HCA ||
+	    op == MLX4_CMD_HW2SW_EQ ||
+	    op == MLX4_CMD_HW2SW_CQ ||
+	    op == MLX4_CMD_2RST_QP ||
+	    op == MLX4_CMD_HW2SW_SRQ ||
+	    op == MLX4_CMD_SYNC_TPT ||
+	    op == MLX4_CMD_UNMAP_ICM ||
+	    op == MLX4_CMD_UNMAP_ICM_AUX ||
+	    op == MLX4_CMD_UNMAP_FA)
+		return 1;
+	/* Error on MLX4_CMD_HW2SW_MPT is fatal except when fw status equals
+	  * CMD_STAT_REG_BOUND.
+	  * This status indicates that memory region has memory windows bound to it
+	  * which may result from invalid user space usage and is not fatal.
+	  */
+	if (op == MLX4_CMD_HW2SW_MPT && fw_status != CMD_STAT_REG_BOUND)
+		return 1;
+	return 0;
+}
+
+static int mlx4_cmd_reset_flow(struct mlx4_dev *dev, u16 op, u8 op_modifier,
+			       int err)
+{
+	/* Only if reset flow is really active return code is based on
+	  * command, otherwise current error code is returned.
+	  */
+	if (mlx4_internal_err_reset) {
+		mlx4_enter_error_state(dev->persist);
+		err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+	}
+
+	return err;
+}
+
 static int comm_pending(struct mlx4_dev *dev)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
@@ -258,7 +325,7 @@ static int mlx4_comm_cmd_wait(struct mlx4_dev *dev, u8 op,
 	cmd->free_head = context->next;
 	spin_unlock(&cmd->context_lock);
 
-	init_completion(&context->done);
+	reinit_completion(&context->done);
 
 	mlx4_comm_cmd_post(dev, op, param);
 
@@ -307,7 +374,7 @@ static int cmd_pending(struct mlx4_dev *dev)
 {
 	u32 status;
 
-	if (pci_channel_offline(dev->pdev))
+	if (pci_channel_offline(dev->persist->pdev))
 		return -EIO;
 
 	status = readl(mlx4_priv(dev)->cmd.hcr + HCR_STATUS_OFFSET);
@@ -323,17 +390,21 @@ static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
 {
 	struct mlx4_cmd *cmd = &mlx4_priv(dev)->cmd;
 	u32 __iomem *hcr = cmd->hcr;
-	int ret = -EAGAIN;
+	int ret = -EIO;
 	unsigned long end;
 
-	mutex_lock(&cmd->hcr_mutex);
-
-	if (pci_channel_offline(dev->pdev)) {
+	mutex_lock(&dev->persist->device_state_mutex);
+	/* To avoid writing to unknown addresses after the device state was
+	  * changed to internal error and the chip was reset,
+	  * check the INTERNAL_ERROR flag which is updated under
+	  * device_state_mutex lock.
+	  */
+	if (pci_channel_offline(dev->persist->pdev) ||
+	    (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)) {
 		/*
 		 * Device is going through error recovery
 		 * and cannot accept commands.
 		 */
-		ret = -EIO;
 		goto out;
 	}
 
@@ -342,12 +413,11 @@ static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
 		end += msecs_to_jiffies(GO_BIT_TIMEOUT_MSECS);
 
 	while (cmd_pending(dev)) {
-		if (pci_channel_offline(dev->pdev)) {
+		if (pci_channel_offline(dev->persist->pdev)) {
 			/*
 			 * Device is going through error recovery
 			 * and cannot accept commands.
 			 */
-			ret = -EIO;
 			goto out;
 		}
 
@@ -391,7 +461,11 @@ static int mlx4_cmd_post(struct mlx4_dev *dev, u64 in_param, u64 out_param,
 	ret = 0;
 
 out:
-	mutex_unlock(&cmd->hcr_mutex);
+	if (ret)
+		mlx4_warn(dev, "Could not post command 0x%x: ret=%d, in_param=0x%llx, in_mod=0x%x, op_mod=0x%x\n",
+			  op, ret, in_param, in_modifier, op_modifier);
+	mutex_unlock(&dev->persist->device_state_mutex);
+
 	return ret;
 }
 
@@ -464,12 +538,12 @@ static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 
 	down(&priv->cmd.poll_sem);
 
-	if (pci_channel_offline(dev->pdev)) {
+	if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
 		/*
 		 * Device is going through error recovery
 		 * and cannot accept commands.
 		 */
-		err = -EIO;
+		err = mlx4_internal_err_ret_value(dev, op, op_modifier);
 		goto out;
 	}
 
@@ -483,16 +557,21 @@ static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
 			    in_modifier, op_modifier, op, CMD_POLL_TOKEN, 0);
 	if (err)
-		goto out;
+		goto out_reset;
 
 	end = msecs_to_jiffies(timeout) + jiffies;
 	while (cmd_pending(dev) && time_before(jiffies, end)) {
-		if (pci_channel_offline(dev->pdev)) {
+		if (pci_channel_offline(dev->persist->pdev)) {
 			/*
 			 * Device is going through error recovery
 			 * and cannot accept commands.
 			 */
 			err = -EIO;
+			goto out_reset;
+		}
+
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR) {
+			err = mlx4_internal_err_ret_value(dev, op, op_modifier);
 			goto out;
 		}
 
@@ -502,8 +581,8 @@ static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 	if (cmd_pending(dev)) {
 		mlx4_warn(dev, "command 0x%x timed out (go bit not cleared)\n",
 			  op);
-		err = -ETIMEDOUT;
-		goto out;
+		err = -EIO;
+		goto out_reset;
 	}
 
 	if (out_is_imm)
@@ -515,10 +594,17 @@ static int mlx4_cmd_poll(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 	stat = be32_to_cpu((__force __be32)
 			   __raw_readl(hcr + HCR_STATUS_OFFSET)) >> 24;
 	err = mlx4_status_to_errno(stat);
-	if (err)
+	if (err) {
 		mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n",
 			 op, stat);
+		if (mlx4_closing_cmd_fatal_error(op, stat))
+			goto out_reset;
+		goto out;
+	}
 
+out_reset:
+	if (err)
+		err = mlx4_cmd_reset_flow(dev, op, op_modifier, err);
 out:
 	up(&priv->cmd.poll_sem);
 	return err;
@@ -565,17 +651,19 @@ static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 		goto out;
 	}
 
-	init_completion(&context->done);
+	reinit_completion(&context->done);
 
-	mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
-		      in_modifier, op_modifier, op, context->token, 1);
+	err = mlx4_cmd_post(dev, in_param, out_param ? *out_param : 0,
+			    in_modifier, op_modifier, op, context->token, 1);
+	if (err)
+		goto out_reset;
 
 	if (!wait_for_completion_timeout(&context->done,
 					 msecs_to_jiffies(timeout))) {
 		mlx4_warn(dev, "command 0x%x timed out (go bit not cleared)\n",
 			  op);
-		err = -EBUSY;
-		goto out;
+		err = -EIO;
+		goto out_reset;
 	}
 
 	err = context->result;
@@ -592,12 +680,20 @@ static int mlx4_cmd_wait(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 		else
 			mlx4_err(dev, "command 0x%x failed: fw status = 0x%x\n",
 				 op, context->fw_status);
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+			err = mlx4_internal_err_ret_value(dev, op, op_modifier);
+		else if (mlx4_closing_cmd_fatal_error(op, context->fw_status))
+			goto out_reset;
+
 		goto out;
 	}
 
 	if (out_is_imm)
 		*out_param = context->out_param;
 
+out_reset:
+	if (err)
+		err = mlx4_cmd_reset_flow(dev, op, op_modifier, err);
 out:
 	spin_lock(&cmd->context_lock);
 	context->next = cmd->free_head;
@@ -612,10 +708,13 @@ int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 	       int out_is_imm, u32 in_modifier, u8 op_modifier,
 	       u16 op, unsigned long timeout, int native)
 {
-	if (pci_channel_offline(dev->pdev))
-		return -EIO;
+	if (pci_channel_offline(dev->persist->pdev))
+		return mlx4_cmd_reset_flow(dev, op, op_modifier, -EIO);
 
 	if (!mlx4_is_mfunc(dev) || (native && mlx4_is_master(dev))) {
+		if (dev->persist->state & MLX4_DEVICE_STATE_INTERNAL_ERROR)
+			return mlx4_internal_err_ret_value(dev, op,
+							  op_modifier);
 		if (mlx4_priv(dev)->cmd.use_events)
 			return mlx4_cmd_wait(dev, in_param, out_param,
 					     out_is_imm, in_modifier,
@@ -631,7 +730,7 @@ int __mlx4_cmd(struct mlx4_dev *dev, u64 in_param, u64 *out_param,
 EXPORT_SYMBOL_GPL(__mlx4_cmd);
 
 
-static int mlx4_ARM_COMM_CHANNEL(struct mlx4_dev *dev)
+int mlx4_ARM_COMM_CHANNEL(struct mlx4_dev *dev)
 {
 	return mlx4_cmd(dev, 0, 0, 0, MLX4_CMD_ARM_COMM_CHANNEL,
 			MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE);
@@ -1847,8 +1946,11 @@ static void mlx4_master_do_cmd(struct mlx4_dev *dev, int slave, u8 cmd,
 		break;
 	case MLX4_COMM_CMD_VHCR_POST:
 		if ((slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_EN) &&
-		    (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_POST))
+		    (slave_state[slave].last_cmd != MLX4_COMM_CMD_VHCR_POST)) {
+			mlx4_warn(dev, "slave:%d is out of sync, cmd=0x%x, last command=0x%x, reset is needed\n",
+				  slave, cmd, slave_state[slave].last_cmd);
 			goto reset_slave;
+		}
 
 		mutex_lock(&priv->cmd.slave_cmd_mutex);
 		if (mlx4_master_process_vhcr(dev, slave, NULL)) {
@@ -1882,7 +1984,18 @@ static void mlx4_master_do_cmd(struct mlx4_dev *dev, int slave, u8 cmd,
 
 reset_slave:
 	/* cleanup any slave resources */
-	mlx4_delete_all_resources_for_slave(dev, slave);
+	if (dev->persist->interface_state & MLX4_INTERFACE_STATE_UP)
+		mlx4_delete_all_resources_for_slave(dev, slave);
+
+	if (cmd != MLX4_COMM_CMD_RESET) {
+		mlx4_warn(dev, "Turn on internal error to force reset, slave=%d, cmd=0x%x\n",
+			  slave, cmd);
+		/* Turn on internal error letting slave reset itself immeditaly,
+		 * otherwise it might take till timeout on command is passed
+		 */
+		reply |= ((u32)COMM_CHAN_EVENT_INTERNAL_ERR);
+	}
+
 	spin_lock_irqsave(&priv->mfunc.master.slave_state_lock, flags);
 	if (!slave_state[slave].is_slave_going_down)
 		slave_state[slave].last_cmd = MLX4_COMM_CMD_RESET;
@@ -1958,17 +2071,28 @@ void mlx4_master_comm_channel(struct work_struct *work)
 static int sync_toggles(struct mlx4_dev *dev)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
-	int wr_toggle;
-	int rd_toggle;
+	u32 wr_toggle;
+	u32 rd_toggle;
 	unsigned long end;
 
-	wr_toggle = swab32(readl(&priv->mfunc.comm->slave_write)) >> 31;
-	end = jiffies + msecs_to_jiffies(5000);
+	wr_toggle = swab32(readl(&priv->mfunc.comm->slave_write));
+	if (wr_toggle == 0xffffffff)
+		end = jiffies + msecs_to_jiffies(30000);
+	else
+		end = jiffies + msecs_to_jiffies(5000);
 
 	while (time_before(jiffies, end)) {
-		rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read)) >> 31;
-		if (rd_toggle == wr_toggle) {
-			priv->cmd.comm_toggle = rd_toggle;
+		rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
+		if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
+			/* PCI might be offline */
+			msleep(100);
+			wr_toggle = swab32(readl(&priv->mfunc.comm->
+					   slave_write));
+			continue;
+		}
+
+		if (rd_toggle >> 31 == wr_toggle >> 31) {
+			priv->cmd.comm_toggle = rd_toggle >> 31;
 			return 0;
 		}
 
@@ -1997,11 +2121,12 @@ int mlx4_multi_func_init(struct mlx4_dev *dev)
 
 	if (mlx4_is_master(dev))
 		priv->mfunc.comm =
-		ioremap(pci_resource_start(dev->pdev, priv->fw.comm_bar) +
+		ioremap(pci_resource_start(dev->persist->pdev,
+					   priv->fw.comm_bar) +
 			priv->fw.comm_base, MLX4_COMM_PAGESIZE);
 	else
 		priv->mfunc.comm =
-		ioremap(pci_resource_start(dev->pdev, 2) +
+		ioremap(pci_resource_start(dev->persist->pdev, 2) +
 			MLX4_SLAVE_COMM_BASE, MLX4_COMM_PAGESIZE);
 	if (!priv->mfunc.comm) {
 		mlx4_err(dev, "Couldn't map communication vector\n");
@@ -2073,13 +2198,6 @@ int mlx4_multi_func_init(struct mlx4_dev *dev)
 		if (mlx4_init_resource_tracker(dev))
 			goto err_thread;
 
-		err = mlx4_ARM_COMM_CHANNEL(dev);
-		if (err) {
-			mlx4_err(dev, " Failed to arm comm channel eq: %x\n",
-				 err);
-			goto err_resource;
-		}
-
 	} else {
 		err = sync_toggles(dev);
 		if (err) {
@@ -2089,8 +2207,6 @@ int mlx4_multi_func_init(struct mlx4_dev *dev)
 	}
 	return 0;
 
-err_resource:
-	mlx4_free_resource_tracker(dev, RES_TR_FREE_ALL);
 err_thread:
 	flush_workqueue(priv->mfunc.master.comm_wq);
 	destroy_workqueue(priv->mfunc.master.comm_wq);
@@ -2107,9 +2223,9 @@ err_comm_admin:
 err_comm:
 	iounmap(priv->mfunc.comm);
 err_vhcr:
-	dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE,
-					     priv->mfunc.vhcr,
-					     priv->mfunc.vhcr_dma);
+	dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
+			  priv->mfunc.vhcr,
+			  priv->mfunc.vhcr_dma);
 	priv->mfunc.vhcr = NULL;
 	return -ENOMEM;
 }
@@ -2120,7 +2236,6 @@ int mlx4_cmd_init(struct mlx4_dev *dev)
 	int flags = 0;
 
 	if (!priv->cmd.initialized) {
-		mutex_init(&priv->cmd.hcr_mutex);
 		mutex_init(&priv->cmd.slave_cmd_mutex);
 		sema_init(&priv->cmd.poll_sem, 1);
 		priv->cmd.use_events = 0;
@@ -2130,8 +2245,8 @@ int mlx4_cmd_init(struct mlx4_dev *dev)
 	}
 
 	if (!mlx4_is_slave(dev) && !priv->cmd.hcr) {
-		priv->cmd.hcr = ioremap(pci_resource_start(dev->pdev, 0) +
-					MLX4_HCR_BASE, MLX4_HCR_SIZE);
+		priv->cmd.hcr = ioremap(pci_resource_start(dev->persist->pdev,
+					0) + MLX4_HCR_BASE, MLX4_HCR_SIZE);
 		if (!priv->cmd.hcr) {
 			mlx4_err(dev, "Couldn't map command register\n");
 			goto err;
@@ -2140,7 +2255,8 @@ int mlx4_cmd_init(struct mlx4_dev *dev)
 	}
 
 	if (mlx4_is_mfunc(dev) && !priv->mfunc.vhcr) {
-		priv->mfunc.vhcr = dma_alloc_coherent(&(dev->pdev->dev), PAGE_SIZE,
+		priv->mfunc.vhcr = dma_alloc_coherent(&dev->persist->pdev->dev,
+						      PAGE_SIZE,
 						      &priv->mfunc.vhcr_dma,
 						      GFP_KERNEL);
 		if (!priv->mfunc.vhcr)
@@ -2150,7 +2266,8 @@ int mlx4_cmd_init(struct mlx4_dev *dev)
 	}
 
 	if (!priv->cmd.pool) {
-		priv->cmd.pool = pci_pool_create("mlx4_cmd", dev->pdev,
+		priv->cmd.pool = pci_pool_create("mlx4_cmd",
+						 dev->persist->pdev,
 						 MLX4_MAILBOX_SIZE,
 						 MLX4_MAILBOX_SIZE, 0);
 		if (!priv->cmd.pool)
@@ -2166,6 +2283,27 @@ err:
 	return -ENOMEM;
 }
 
+void mlx4_report_internal_err_comm_event(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	int slave;
+	u32 slave_read;
+
+	/* Report an internal error event to all
+	 * communication channels.
+	 */
+	for (slave = 0; slave < dev->num_slaves; slave++) {
+		slave_read = swab32(readl(&priv->mfunc.comm[slave].slave_read));
+		slave_read |= (u32)COMM_CHAN_EVENT_INTERNAL_ERR;
+		__raw_writel((__force u32)cpu_to_be32(slave_read),
+			     &priv->mfunc.comm[slave].slave_read);
+		/* Make sure that our comm channel write doesn't
+		 * get mixed in with writes from another CPU.
+		 */
+		mmiowb();
+	}
+}
+
 void mlx4_multi_func_cleanup(struct mlx4_dev *dev)
 {
 	struct mlx4_priv *priv = mlx4_priv(dev);
@@ -2181,6 +2319,7 @@ void mlx4_multi_func_cleanup(struct mlx4_dev *dev)
 		kfree(priv->mfunc.master.slave_state);
 		kfree(priv->mfunc.master.vf_admin);
 		kfree(priv->mfunc.master.vf_oper);
+		dev->num_slaves = 0;
 	}
 
 	iounmap(priv->mfunc.comm);
@@ -2202,7 +2341,7 @@ void mlx4_cmd_cleanup(struct mlx4_dev *dev, int cleanup_mask)
 	}
 	if (mlx4_is_mfunc(dev) && priv->mfunc.vhcr &&
 	    (cleanup_mask & MLX4_CMD_CLEANUP_VHCR)) {
-		dma_free_coherent(&(dev->pdev->dev), PAGE_SIZE,
+		dma_free_coherent(&dev->persist->pdev->dev, PAGE_SIZE,
 				  priv->mfunc.vhcr, priv->mfunc.vhcr_dma);
 		priv->mfunc.vhcr = NULL;
 	}
@@ -2229,6 +2368,11 @@ int mlx4_cmd_use_events(struct mlx4_dev *dev)
 	for (i = 0; i < priv->cmd.max_cmds; ++i) {
 		priv->cmd.context[i].token = i;
 		priv->cmd.context[i].next  = i + 1;
+		/* To support fatal error flow, initialize all
+		 * cmd contexts to allow simulating completions
+		 * with complete() at any time.
+		 */
+		init_completion(&priv->cmd.context[i].done);
 	}
 
 	priv->cmd.context[priv->cmd.max_cmds - 1].next = -1;
@@ -2306,8 +2450,9 @@ u32 mlx4_comm_get_version(void)
 
 static int mlx4_get_slave_indx(struct mlx4_dev *dev, int vf)
 {
-	if ((vf < 0) || (vf >= dev->num_vfs)) {
-		mlx4_err(dev, "Bad vf number:%d (number of activated vf: %d)\n", vf, dev->num_vfs);
+	if ((vf < 0) || (vf >= dev->persist->num_vfs)) {
+		mlx4_err(dev, "Bad vf number:%d (number of activated vf: %d)\n",
+			 vf, dev->persist->num_vfs);
 		return -EINVAL;
 	}
 
@@ -2316,7 +2461,7 @@ static int mlx4_get_slave_indx(struct mlx4_dev *dev, int vf)
 
 int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave)
 {
-	if (slave < 1 || slave > dev->num_vfs) {
+	if (slave < 1 || slave > dev->persist->num_vfs) {
 		mlx4_err(dev,
 			 "Bad slave number:%d (number of activated slaves: %lu)\n",
 			 slave, dev->num_slaves);
@@ -2325,6 +2470,25 @@ int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave)
 	return slave - 1;
 }
 
+void mlx4_cmd_wake_completions(struct mlx4_dev *dev)
+{
+	struct mlx4_priv *priv = mlx4_priv(dev);
+	struct mlx4_cmd_context *context;
+	int i;
+
+	spin_lock(&priv->cmd.context_lock);
+	if (priv->cmd.context) {
+		for (i = 0; i < priv->cmd.max_cmds; ++i) {
+			context = &priv->cmd.context[i];
+			context->fw_status = CMD_STAT_INTERNAL_ERR;
+			context->result    =
+				mlx4_status_to_errno(CMD_STAT_INTERNAL_ERR);
+			complete(&context->done);
+		}
+	}
+	spin_unlock(&priv->cmd.context_lock);
+}
+
 struct mlx4_active_ports mlx4_get_active_ports(struct mlx4_dev *dev, int slave)
 {
 	struct mlx4_active_ports actv_ports;
@@ -2388,7 +2552,7 @@ struct mlx4_slaves_pport mlx4_phys_to_slaves_pport(struct mlx4_dev *dev,
 	if (port <= 0 || port > dev->caps.num_ports)
 		return slaves_pport;
 
-	for (i = 0; i < dev->num_vfs + 1; i++) {
+	for (i = 0; i < dev->persist->num_vfs + 1; i++) {
 		struct mlx4_active_ports actv_ports =
 			mlx4_get_active_ports(dev, i);
 		if (test_bit(port - 1, actv_ports.ports))
@@ -2408,7 +2572,7 @@ struct mlx4_slaves_pport mlx4_phys_to_slaves_pport_actv(
 
 	bitmap_zero(slaves_pport.slaves, MLX4_MFUNC_MAX);
 
-	for (i = 0; i < dev->num_vfs + 1; i++) {
+	for (i = 0; i < dev->persist->num_vfs + 1; i++) {
 		struct mlx4_active_ports actv_ports =
 			mlx4_get_active_ports(dev, i);
 		if (bitmap_equal(crit_ports->ports, actv_ports.ports,