From c69453e294c9f16da977b68e658a8028b854c209 Mon Sep 17 00:00:00 2001 From: Yishai Hadas Date: Sun, 25 Jan 2015 16:59:40 +0200 Subject: [PATCH] net/mlx4_core: Manage interface state for Reset flow cases We need to manage interface state to sync between reset flow and some other relative cases such as remove_one. This has to be done to prevent certain races. For example in case software stack is down as a result of unload call, the remove_one should skip the unload phase. Implement the remove_one case, handling AER and other cases comes next. The interface can be up/down, upon remove_one, the state will include an extra bit indicating that the device is cleaned-up, forcing other tasks to finish before the final cleanup. Signed-off-by: Yishai Hadas Signed-off-by: Or Gerlitz Signed-off-by: David S. Miller --- drivers/net/ethernet/mellanox/mlx4/catas.c | 13 +++++++++++-- drivers/net/ethernet/mellanox/mlx4/intf.c | 2 ++ drivers/net/ethernet/mellanox/mlx4/main.c | 13 ++++++++++++- include/linux/mlx4/device.h | 7 +++++++ 4 files changed, 32 insertions(+), 3 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx4/catas.c b/drivers/net/ethernet/mellanox/mlx4/catas.c index 63f14ffcc906..3fcf3cfaedfc 100644 --- a/drivers/net/ethernet/mellanox/mlx4/catas.c +++ b/drivers/net/ethernet/mellanox/mlx4/catas.c @@ -122,8 +122,14 @@ static void mlx4_handle_error_state(struct mlx4_dev_persistent *persist) int err = 0; mlx4_enter_error_state(persist); - err = mlx4_restart_one(persist->pdev); - mlx4_info(persist->dev, "mlx4_restart_one was ended, ret=%d\n", err); + mutex_lock(&persist->interface_state_mutex); + if (persist->interface_state & MLX4_INTERFACE_STATE_UP && + !(persist->interface_state & MLX4_INTERFACE_STATE_DELETION)) { + err = mlx4_restart_one(persist->pdev); + mlx4_info(persist->dev, "mlx4_restart_one was ended, ret=%d\n", + err); + } + mutex_unlock(&persist->interface_state_mutex); } static void dump_err_buf(struct mlx4_dev *dev) @@ -211,6 +217,9 @@ void mlx4_stop_catas_poll(struct mlx4_dev *dev) iounmap(priv->catas_err.map); priv->catas_err.map = NULL; } + + if (dev->persist->interface_state & MLX4_INTERFACE_STATE_DELETION) + flush_workqueue(dev->persist->catas_wq); } int mlx4_catas_init(struct mlx4_dev *dev) diff --git a/drivers/net/ethernet/mellanox/mlx4/intf.c b/drivers/net/ethernet/mellanox/mlx4/intf.c index 116895ac8b35..fba0b96a6f28 100644 --- a/drivers/net/ethernet/mellanox/mlx4/intf.c +++ b/drivers/net/ethernet/mellanox/mlx4/intf.c @@ -138,6 +138,7 @@ int mlx4_register_device(struct mlx4_dev *dev) mutex_lock(&intf_mutex); + dev->persist->interface_state |= MLX4_INTERFACE_STATE_UP; list_add_tail(&priv->dev_list, &dev_list); list_for_each_entry(intf, &intf_list, list) mlx4_add_device(intf, priv); @@ -162,6 +163,7 @@ void mlx4_unregister_device(struct mlx4_dev *dev) mlx4_remove_device(intf, priv); list_del(&priv->dev_list); + dev->persist->interface_state &= ~MLX4_INTERFACE_STATE_UP; mutex_unlock(&intf_mutex); } diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c index dc2d910fcc88..d59cae5da3f0 100644 --- a/drivers/net/ethernet/mellanox/mlx4/main.c +++ b/drivers/net/ethernet/mellanox/mlx4/main.c @@ -3114,6 +3114,7 @@ static int mlx4_init_one(struct pci_dev *pdev, const struct pci_device_id *id) pci_set_drvdata(pdev, dev->persist); priv->pci_dev_data = id->driver_data; mutex_init(&dev->persist->device_state_mutex); + mutex_init(&dev->persist->interface_state_mutex); ret = __mlx4_init_one(pdev, id->driver_data, priv); if (ret) { @@ -3232,7 +3233,17 @@ static void mlx4_remove_one(struct pci_dev *pdev) struct mlx4_dev *dev = persist->dev; struct mlx4_priv *priv = mlx4_priv(dev); - mlx4_unload_one(pdev); + mutex_lock(&persist->interface_state_mutex); + persist->interface_state |= MLX4_INTERFACE_STATE_DELETION; + mutex_unlock(&persist->interface_state_mutex); + + /* device marked to be under deletion running now without the lock + * letting other tasks to be terminated + */ + if (persist->interface_state & MLX4_INTERFACE_STATE_UP) + mlx4_unload_one(pdev); + else + mlx4_info(dev, "%s: interface is down\n", __func__); mlx4_catas_end(dev); pci_release_regions(pdev); pci_disable_device(pdev); diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h index 7d5d317cb7a6..33f9ca71925c 100644 --- a/include/linux/mlx4/device.h +++ b/include/linux/mlx4/device.h @@ -416,6 +416,11 @@ enum { MLX4_DEVICE_STATE_INTERNAL_ERROR = 1 << 1, }; +enum { + MLX4_INTERFACE_STATE_UP = 1 << 0, + MLX4_INTERFACE_STATE_DELETION = 1 << 1, +}; + #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \ MLX4_EQ_PORT_INFO_MSTR_SM_LID_CHANGE_MASK) @@ -760,6 +765,8 @@ struct mlx4_dev_persistent { struct workqueue_struct *catas_wq; struct mutex device_state_mutex; /* protect HW state */ u8 state; + struct mutex interface_state_mutex; /* protect SW state */ + u8 interface_state; }; struct mlx4_dev { -- 2.20.1