From 3c1274bd6441600ef87b682124b4024692bf8cb5 Mon Sep 17 00:00:00 2001 From: Hans Petter Selasky Date: Fri, 30 Mar 2018 19:43:15 +0000 Subject: [PATCH] Prepare for FW dump in error state in mlx5core. - Move firmware dump prep and cleanup to init_one() and remove_one() so that the init and cleanup will happen only upon driver reload. - Add some prints to indicate firmware dump. MFC after: 3 days Submitted by: slavash@ Sponsored by: Mellanox Technologies --- sys/dev/mlx5/mlx5_core/mlx5_fwdump.c | 6 +++++- sys/dev/mlx5/mlx5_core/mlx5_main.c | 8 ++++---- 2 files changed, 9 insertions(+), 5 deletions(-) diff --git a/sys/dev/mlx5/mlx5_core/mlx5_fwdump.c b/sys/dev/mlx5/mlx5_core/mlx5_fwdump.c index 7839f4fa1a4c..4c160f6eafbe 100644 --- a/sys/dev/mlx5/mlx5_core/mlx5_fwdump.c +++ b/sys/dev/mlx5/mlx5_core/mlx5_fwdump.c @@ -116,14 +116,18 @@ mlx5_fwdump(struct mlx5_core_dev *mdev) uint32_t i, ri; int error; + dev_info(&mdev->pdev->dev, "Issuing FW dump\n"); dd = (struct mlx5_dump_data *)atomic_load_acq_ptr((uintptr_t *) &mdev->dump_data); if (dd == NULL) return; mtx_lock(&dd->dump_lock); - if (dd->dump_valid) + if (dd->dump_valid) { /* only one dump */ + dev_warn(&mdev->pdev->dev, + "Only one FW dump can be captured aborting FW dump\n"); goto failed; + } /* mlx5_vsc already warns, be silent. */ error = mlx5_vsc_lock(mdev); diff --git a/sys/dev/mlx5/mlx5_core/mlx5_main.c b/sys/dev/mlx5/mlx5_core/mlx5_main.c index a60e9c438b03..c7406d1413a2 100644 --- a/sys/dev/mlx5/mlx5_core/mlx5_main.c +++ b/sys/dev/mlx5/mlx5_core/mlx5_main.c @@ -1056,8 +1056,6 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, goto err_fs; } - mlx5_fwdump_prep(dev); - clear_bit(MLX5_INTERFACE_STATE_DOWN, &dev->intf_state); set_bit(MLX5_INTERFACE_STATE_UP, &dev->intf_state); @@ -1127,7 +1125,6 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv, goto out; } - mlx5_fwdump_clean(dev); mlx5_unregister_device(dev); mlx5_cleanup_fs(dev); @@ -1179,7 +1176,6 @@ struct mlx5_core_event_handler { void *data); }; - static int init_one(struct pci_dev *pdev, const struct pci_device_id *id) { @@ -1224,6 +1220,8 @@ static int init_one(struct pci_dev *pdev, goto clean_health; } + mlx5_fwdump_prep(dev); + pci_save_state(pdev->dev.bsddev); return 0; @@ -1248,6 +1246,7 @@ static void remove_one(struct pci_dev *pdev) return; } + mlx5_fwdump_clean(dev); mlx5_pagealloc_cleanup(dev); mlx5_health_cleanup(dev); mlx5_pci_close(dev, priv); @@ -1264,6 +1263,7 @@ static pci_ers_result_t mlx5_pci_err_detected(struct pci_dev *pdev, dev_info(&pdev->dev, "%s was called\n", __func__); mlx5_enter_error_state(dev, false); mlx5_unload_one(dev, priv, false); + if (state) { mlx5_drain_health_wq(dev); mlx5_pci_disable_device(dev);