From 04c0c1ab38e95105d950db5b84e727637e149ce7 Mon Sep 17 00:00:00 2001 From: Mohamad Haj Yahia Date: Tue, 25 Oct 2016 18:36:34 +0300 Subject: [PATCH] net/mlx5: PCI error recovery health care simulation In case that the kernel PCI error handlers are not called, we will trigger our own recovery flow. The health work will give priority to the kernel pci error handlers to recover the PCI by waiting for a small period, if the pci error handlers are not triggered the manual recovery flow will be executed. We don't save pci state in case of manual recovery because it will ruin the pci configuration space and we will lose dma sync. Fixes: 89d44f0a6c73 ('net/mlx5_core: Add pci error handlers to mlx5_core driver') Signed-off-by: Mohamad Haj Yahia Signed-off-by: Saeed Mahameed Signed-off-by: David S. Miller --- .../net/ethernet/mellanox/mlx5/core/health.c | 45 +++++++++++++++++-- .../net/ethernet/mellanox/mlx5/core/main.c | 18 +++++--- .../ethernet/mellanox/mlx5/core/mlx5_core.h | 1 + include/linux/mlx5/driver.h | 1 + 4 files changed, 56 insertions(+), 9 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c index 2d00022c92d8..5bcf93422ee0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/health.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c @@ -61,14 +61,15 @@ enum { enum { MLX5_NIC_IFC_FULL = 0, MLX5_NIC_IFC_DISABLED = 1, - MLX5_NIC_IFC_NO_DRAM_NIC = 2 + MLX5_NIC_IFC_NO_DRAM_NIC = 2, + MLX5_NIC_IFC_INVALID = 3 }; enum { MLX5_DROP_NEW_HEALTH_WORK, }; -static u8 get_nic_interface(struct mlx5_core_dev *dev) +static u8 get_nic_state(struct mlx5_core_dev *dev) { return (ioread32be(&dev->iseg->cmdq_addr_l_sz) >> 8) & 3; } @@ -101,7 +102,7 @@ static int in_fatal(struct mlx5_core_dev *dev) struct mlx5_core_health *health = &dev->priv.health; struct health_buffer __iomem *h = health->health; - if (get_nic_interface(dev) == MLX5_NIC_IFC_DISABLED) + if (get_nic_state(dev) == MLX5_NIC_IFC_DISABLED) return 1; if (ioread32be(&h->fw_ver) == 0xffffffff) @@ -131,7 +132,7 @@ unlock: static void mlx5_handle_bad_state(struct mlx5_core_dev *dev) { - u8 nic_interface = get_nic_interface(dev); + u8 nic_interface = get_nic_state(dev); switch (nic_interface) { case MLX5_NIC_IFC_FULL: @@ -153,8 +154,34 @@ static void mlx5_handle_bad_state(struct mlx5_core_dev *dev) mlx5_disable_device(dev); } +static void health_recover(struct work_struct *work) +{ + struct mlx5_core_health *health; + struct delayed_work *dwork; + struct mlx5_core_dev *dev; + struct mlx5_priv *priv; + u8 nic_state; + + dwork = container_of(work, struct delayed_work, work); + health = container_of(dwork, struct mlx5_core_health, recover_work); + priv = container_of(health, struct mlx5_priv, health); + dev = container_of(priv, struct mlx5_core_dev, priv); + + nic_state = get_nic_state(dev); + if (nic_state == MLX5_NIC_IFC_INVALID) { + dev_err(&dev->pdev->dev, "health recovery flow aborted since the nic state is invalid\n"); + return; + } + + dev_err(&dev->pdev->dev, "starting health recovery flow\n"); + mlx5_recover_device(dev); +} + +/* How much time to wait until health resetting the driver (in msecs) */ +#define MLX5_RECOVERY_DELAY_MSECS 60000 static void health_care(struct work_struct *work) { + unsigned long recover_delay = msecs_to_jiffies(MLX5_RECOVERY_DELAY_MSECS); struct mlx5_core_health *health; struct mlx5_core_dev *dev; struct mlx5_priv *priv; @@ -164,6 +191,14 @@ static void health_care(struct work_struct *work) dev = container_of(priv, struct mlx5_core_dev, priv); mlx5_core_warn(dev, "handling bad device here\n"); mlx5_handle_bad_state(dev); + + spin_lock(&health->wq_lock); + if (!test_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags)) + schedule_delayed_work(&health->recover_work, recover_delay); + else + dev_err(&dev->pdev->dev, + "new health works are not permitted at this stage\n"); + spin_unlock(&health->wq_lock); } static const char *hsynd_str(u8 synd) @@ -316,6 +351,7 @@ void mlx5_drain_health_wq(struct mlx5_core_dev *dev) spin_lock(&health->wq_lock); set_bit(MLX5_DROP_NEW_HEALTH_WORK, &health->flags); spin_unlock(&health->wq_lock); + cancel_delayed_work_sync(&health->recover_work); cancel_work_sync(&health->work); } @@ -344,6 +380,7 @@ int mlx5_health_init(struct mlx5_core_dev *dev) return -ENOMEM; spin_lock_init(&health->wq_lock); INIT_WORK(&health->work, health_care); + INIT_DELAYED_WORK(&health->recover_work, health_recover); return 0; } diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c index 9f90226bc120..d5433c49b2b0 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c @@ -1313,6 +1313,7 @@ static pci_ers_result_t mlx5_pci_err_detected(struct pci_dev *pdev, struct mlx5_priv *priv = &dev->priv; dev_info(&pdev->dev, "%s was called\n", __func__); + mlx5_enter_error_state(dev); mlx5_unload_one(dev, priv, false); /* In case of kernel call save the pci state and drain health wq */ @@ -1378,11 +1379,6 @@ static pci_ers_result_t mlx5_pci_slot_reset(struct pci_dev *pdev) return PCI_ERS_RESULT_RECOVERED; } -void mlx5_disable_device(struct mlx5_core_dev *dev) -{ - mlx5_pci_err_detected(dev->pdev, 0); -} - static void mlx5_pci_resume(struct pci_dev *pdev) { struct mlx5_core_dev *dev = pci_get_drvdata(pdev); @@ -1432,6 +1428,18 @@ static const struct pci_device_id mlx5_core_pci_table[] = { MODULE_DEVICE_TABLE(pci, mlx5_core_pci_table); +void mlx5_disable_device(struct mlx5_core_dev *dev) +{ + mlx5_pci_err_detected(dev->pdev, 0); +} + +void mlx5_recover_device(struct mlx5_core_dev *dev) +{ + mlx5_pci_disable_device(dev); + if (mlx5_pci_slot_reset(dev->pdev) == PCI_ERS_RESULT_RECOVERED) + mlx5_pci_resume(dev->pdev); +} + static struct pci_driver mlx5_core_driver = { .name = DRIVER_NAME, .id_table = mlx5_core_pci_table, diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h index 3d0cfb9f18f9..187662c8ea96 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h @@ -83,6 +83,7 @@ void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event, unsigned long param); void mlx5_enter_error_state(struct mlx5_core_dev *dev); void mlx5_disable_device(struct mlx5_core_dev *dev); +void mlx5_recover_device(struct mlx5_core_dev *dev); int mlx5_sriov_init(struct mlx5_core_dev *dev); void mlx5_sriov_cleanup(struct mlx5_core_dev *dev); int mlx5_sriov_attach(struct mlx5_core_dev *dev); diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index 7d9a5d08eb59..ecc451d89ccd 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -423,6 +423,7 @@ struct mlx5_core_health { struct workqueue_struct *wq; unsigned long flags; struct work_struct work; + struct delayed_work recover_work; }; struct mlx5_cq_table { -- 2.20.1