From 89d44f0a6c732db23b219be708e2fe1e03ee4842 Mon Sep 17 00:00:00 2001 From: Majd Dibbiny Date: Wed, 14 Oct 2015 17:43:46 +0300 Subject: net/mlx5_core: Add pci error handlers to mlx5_core driver This patch implement the pci_error_handlers for mlx5_core which allow the driver to recover from PCI error. Once an error is detected in the PCI, the mlx5_pci_err_detected is called and it: 1) Marks the device to be in 'Internal Error' state. 2) Dispatches an event to the mlx5_ib to flush all the outstanding cqes with error. 3) Returns all the on going commands with error. 4) Unloads the driver. Afterwards, the FW is reset and mlx5_pci_slot_reset is called and it enables the device and restore it's pci state. If the later succeeds, mlx5_pci_resume is called, and it loads the SW stack. Signed-off-by: Majd Dibbiny Signed-off-by: Eli Cohen Signed-off-by: Or Gerlitz Signed-off-by: David S. Miller --- drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 170 +++++++++++++++++++ drivers/net/ethernet/mellanox/mlx5/core/health.c | 72 ++++++++ drivers/net/ethernet/mellanox/mlx5/core/main.c | 182 ++++++++++++++++++++- .../net/ethernet/mellanox/mlx5/core/mlx5_core.h | 4 + .../net/ethernet/mellanox/mlx5/core/pagealloc.c | 13 +- 5 files changed, 429 insertions(+), 12 deletions(-) (limited to 'drivers') diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c index c3e54b7e8780..fabfc9e0a948 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c @@ -256,8 +256,154 @@ static void dump_buf(void *buf, int size, int data_only, int offset) enum { MLX5_DRIVER_STATUS_ABORTED = 0xfe, + MLX5_DRIVER_SYND = 0xbadd00de, }; +static int mlx5_internal_err_ret_value(struct mlx5_core_dev *dev, u16 op, + u32 *synd, u8 *status) +{ + *synd = 0; + *status = 0; + + switch (op) { + case MLX5_CMD_OP_TEARDOWN_HCA: + case MLX5_CMD_OP_DISABLE_HCA: + case MLX5_CMD_OP_MANAGE_PAGES: + case MLX5_CMD_OP_DESTROY_MKEY: + case MLX5_CMD_OP_DESTROY_EQ: + case MLX5_CMD_OP_DESTROY_CQ: + case MLX5_CMD_OP_DESTROY_QP: + case MLX5_CMD_OP_DESTROY_PSV: + case MLX5_CMD_OP_DESTROY_SRQ: + case MLX5_CMD_OP_DESTROY_XRC_SRQ: + case MLX5_CMD_OP_DESTROY_DCT: + case MLX5_CMD_OP_DEALLOC_Q_COUNTER: + case MLX5_CMD_OP_DEALLOC_PD: + case MLX5_CMD_OP_DEALLOC_UAR: + case MLX5_CMD_OP_DETTACH_FROM_MCG: + case MLX5_CMD_OP_DEALLOC_XRCD: + case MLX5_CMD_OP_DEALLOC_TRANSPORT_DOMAIN: + case MLX5_CMD_OP_DELETE_VXLAN_UDP_DPORT: + case MLX5_CMD_OP_DELETE_L2_TABLE_ENTRY: + case MLX5_CMD_OP_DESTROY_TIR: + case MLX5_CMD_OP_DESTROY_SQ: + case MLX5_CMD_OP_DESTROY_RQ: + case MLX5_CMD_OP_DESTROY_RMP: + case MLX5_CMD_OP_DESTROY_TIS: + case MLX5_CMD_OP_DESTROY_RQT: + case MLX5_CMD_OP_DESTROY_FLOW_TABLE: + case MLX5_CMD_OP_DESTROY_FLOW_GROUP: + case MLX5_CMD_OP_DELETE_FLOW_TABLE_ENTRY: + return MLX5_CMD_STAT_OK; + + case MLX5_CMD_OP_QUERY_HCA_CAP: + case MLX5_CMD_OP_QUERY_ADAPTER: + case MLX5_CMD_OP_INIT_HCA: + case MLX5_CMD_OP_ENABLE_HCA: + case MLX5_CMD_OP_QUERY_PAGES: + case MLX5_CMD_OP_SET_HCA_CAP: + case MLX5_CMD_OP_QUERY_ISSI: + case MLX5_CMD_OP_SET_ISSI: + case MLX5_CMD_OP_CREATE_MKEY: + case MLX5_CMD_OP_QUERY_MKEY: + case MLX5_CMD_OP_QUERY_SPECIAL_CONTEXTS: + case MLX5_CMD_OP_PAGE_FAULT_RESUME: + case MLX5_CMD_OP_CREATE_EQ: + case MLX5_CMD_OP_QUERY_EQ: + case MLX5_CMD_OP_GEN_EQE: + case MLX5_CMD_OP_CREATE_CQ: + case MLX5_CMD_OP_QUERY_CQ: + case MLX5_CMD_OP_MODIFY_CQ: + case MLX5_CMD_OP_CREATE_QP: + case MLX5_CMD_OP_RST2INIT_QP: + case MLX5_CMD_OP_INIT2RTR_QP: + case MLX5_CMD_OP_RTR2RTS_QP: + case MLX5_CMD_OP_RTS2RTS_QP: + case MLX5_CMD_OP_SQERR2RTS_QP: + case MLX5_CMD_OP_2ERR_QP: + case MLX5_CMD_OP_2RST_QP: + case MLX5_CMD_OP_QUERY_QP: + case MLX5_CMD_OP_SQD_RTS_QP: + case MLX5_CMD_OP_INIT2INIT_QP: + case MLX5_CMD_OP_CREATE_PSV: + case MLX5_CMD_OP_CREATE_SRQ: + case MLX5_CMD_OP_QUERY_SRQ: + case MLX5_CMD_OP_ARM_RQ: + case MLX5_CMD_OP_CREATE_XRC_SRQ: + case MLX5_CMD_OP_QUERY_XRC_SRQ: + case MLX5_CMD_OP_ARM_XRC_SRQ: + case MLX5_CMD_OP_CREATE_DCT: + case MLX5_CMD_OP_DRAIN_DCT: + case MLX5_CMD_OP_QUERY_DCT: + case MLX5_CMD_OP_ARM_DCT_FOR_KEY_VIOLATION: + case MLX5_CMD_OP_QUERY_VPORT_STATE: + case MLX5_CMD_OP_MODIFY_VPORT_STATE: + case MLX5_CMD_OP_QUERY_ESW_VPORT_CONTEXT: + case MLX5_CMD_OP_MODIFY_ESW_VPORT_CONTEXT: + case MLX5_CMD_OP_QUERY_NIC_VPORT_CONTEXT: + case MLX5_CMD_OP_MODIFY_NIC_VPORT_CONTEXT: + case MLX5_CMD_OP_QUERY_ROCE_ADDRESS: + case MLX5_CMD_OP_SET_ROCE_ADDRESS: + case MLX5_CMD_OP_QUERY_HCA_VPORT_CONTEXT: + case MLX5_CMD_OP_MODIFY_HCA_VPORT_CONTEXT: + case MLX5_CMD_OP_QUERY_HCA_VPORT_GID: + case MLX5_CMD_OP_QUERY_HCA_VPORT_PKEY: + case MLX5_CMD_OP_QUERY_VPORT_COUNTER: + case MLX5_CMD_OP_ALLOC_Q_COUNTER: + case MLX5_CMD_OP_QUERY_Q_COUNTER: + case MLX5_CMD_OP_ALLOC_PD: + case MLX5_CMD_OP_ALLOC_UAR: + case MLX5_CMD_OP_CONFIG_INT_MODERATION: + case MLX5_CMD_OP_ACCESS_REG: + case MLX5_CMD_OP_ATTACH_TO_MCG: + case MLX5_CMD_OP_GET_DROPPED_PACKET_LOG: + case MLX5_CMD_OP_MAD_IFC: + case MLX5_CMD_OP_QUERY_MAD_DEMUX: + case MLX5_CMD_OP_SET_MAD_DEMUX: + case MLX5_CMD_OP_NOP: + case MLX5_CMD_OP_ALLOC_XRCD: + case MLX5_CMD_OP_ALLOC_TRANSPORT_DOMAIN: + case MLX5_CMD_OP_QUERY_CONG_STATUS: + case MLX5_CMD_OP_MODIFY_CONG_STATUS: + case MLX5_CMD_OP_QUERY_CONG_PARAMS: + case MLX5_CMD_OP_MODIFY_CONG_PARAMS: + case MLX5_CMD_OP_QUERY_CONG_STATISTICS: + case MLX5_CMD_OP_ADD_VXLAN_UDP_DPORT: + case MLX5_CMD_OP_SET_L2_TABLE_ENTRY: + case MLX5_CMD_OP_QUERY_L2_TABLE_ENTRY: + case MLX5_CMD_OP_CREATE_TIR: + case MLX5_CMD_OP_MODIFY_TIR: + case MLX5_CMD_OP_QUERY_TIR: + case MLX5_CMD_OP_CREATE_SQ: + case MLX5_CMD_OP_MODIFY_SQ: + case MLX5_CMD_OP_QUERY_SQ: + case MLX5_CMD_OP_CREATE_RQ: + case MLX5_CMD_OP_MODIFY_RQ: + case MLX5_CMD_OP_QUERY_RQ: + case MLX5_CMD_OP_CREATE_RMP: + case MLX5_CMD_OP_MODIFY_RMP: + case MLX5_CMD_OP_QUERY_RMP: + case MLX5_CMD_OP_CREATE_TIS: + case MLX5_CMD_OP_MODIFY_TIS: + case MLX5_CMD_OP_QUERY_TIS: + case MLX5_CMD_OP_CREATE_RQT: + case MLX5_CMD_OP_MODIFY_RQT: + case MLX5_CMD_OP_QUERY_RQT: + case MLX5_CMD_OP_CREATE_FLOW_TABLE: + case MLX5_CMD_OP_QUERY_FLOW_TABLE: + case MLX5_CMD_OP_CREATE_FLOW_GROUP: + case MLX5_CMD_OP_QUERY_FLOW_GROUP: + case MLX5_CMD_OP_SET_FLOW_TABLE_ENTRY: + case MLX5_CMD_OP_QUERY_FLOW_TABLE_ENTRY: + *status = MLX5_DRIVER_STATUS_ABORTED; + *synd = MLX5_DRIVER_SYND; + return -EIO; + default: + mlx5_core_err(dev, "Unknown FW command (%d)\n", op); + return -EINVAL; + } +} + const char *mlx5_command_str(int command) { switch (command) { @@ -592,6 +738,16 @@ static int wait_func(struct mlx5_core_dev *dev, struct mlx5_cmd_work_ent *ent) return err; } +static __be32 *get_synd_ptr(struct mlx5_outbox_hdr *out) +{ + return &out->syndrome; +} + +static u8 *get_status_ptr(struct mlx5_outbox_hdr *out) +{ + return &out->status; +} + /* Notes: * 1. Callback functions may not sleep * 2. page queue commands do not support asynchrous completion @@ -1200,6 +1356,11 @@ static struct mlx5_cmd_msg *alloc_msg(struct mlx5_core_dev *dev, int in_size, return msg; } +static u16 opcode_from_in(struct mlx5_inbox_hdr *in) +{ + return be16_to_cpu(in->opcode); +} + static int is_manage_pages(struct mlx5_inbox_hdr *in) { return be16_to_cpu(in->opcode) == MLX5_CMD_OP_MANAGE_PAGES; @@ -1214,6 +1375,15 @@ static int cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out, gfp_t gfp; int err; u8 status = 0; + u32 drv_synd; + + if (pci_channel_offline(dev->pdev) || + dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + err = mlx5_internal_err_ret_value(dev, opcode_from_in(in), &drv_synd, &status); + *get_synd_ptr(out) = cpu_to_be32(drv_synd); + *get_status_ptr(out) = status; + return err; + } pages_queue = is_manage_pages(in); gfp = callback ? GFP_ATOMIC : GFP_KERNEL; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/health.c b/drivers/net/ethernet/mellanox/mlx5/core/health.c index f1eb686c45b1..f5deb642d0d6 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/health.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/health.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include #include "mlx5_core.h" @@ -68,6 +69,29 @@ static u8 get_nic_interface(struct mlx5_core_dev *dev) return (ioread32be(&dev->iseg->cmdq_addr_l_sz) >> 8) & 3; } +static void trigger_cmd_completions(struct mlx5_core_dev *dev) +{ + unsigned long flags; + u64 vector; + + /* wait for pending handlers to complete */ + synchronize_irq(dev->priv.msix_arr[MLX5_EQ_VEC_CMD].vector); + spin_lock_irqsave(&dev->cmd.alloc_lock, flags); + vector = ~dev->cmd.bitmask & ((1ul << (1 << dev->cmd.log_sz)) - 1); + if (!vector) + goto no_trig; + + vector |= MLX5_TRIGGERED_CMD_COMP; + spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); + + mlx5_core_dbg(dev, "vector 0x%llx\n", vector); + mlx5_cmd_comp_handler(dev, vector); + return; + +no_trig: + spin_unlock_irqrestore(&dev->cmd.alloc_lock, flags); +} + static int in_fatal(struct mlx5_core_dev *dev) { struct mlx5_core_health *health = &dev->priv.health; @@ -82,6 +106,43 @@ static int in_fatal(struct mlx5_core_dev *dev) return 0; } +void mlx5_enter_error_state(struct mlx5_core_dev *dev) +{ + if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) + return; + + mlx5_core_err(dev, "start\n"); + if (pci_channel_offline(dev->pdev) || in_fatal(dev)) + dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR; + + mlx5_core_event(dev, MLX5_DEV_EVENT_SYS_ERROR, 0); + mlx5_core_err(dev, "end\n"); +} + +static void mlx5_handle_bad_state(struct mlx5_core_dev *dev) +{ + u8 nic_interface = get_nic_interface(dev); + + switch (nic_interface) { + case MLX5_NIC_IFC_FULL: + mlx5_core_warn(dev, "Expected to see disabled NIC but it is full driver\n"); + break; + + case MLX5_NIC_IFC_DISABLED: + mlx5_core_warn(dev, "starting teardown\n"); + break; + + case MLX5_NIC_IFC_NO_DRAM_NIC: + mlx5_core_warn(dev, "Expected to see disabled NIC but it is no dram nic\n"); + break; + default: + mlx5_core_warn(dev, "Expected to see disabled NIC but it is has invalid value %d\n", + nic_interface); + } + + mlx5_disable_device(dev); +} + static void health_care(struct work_struct *work) { struct mlx5_core_health *health; @@ -92,6 +153,7 @@ static void health_care(struct work_struct *work) priv = container_of(health, struct mlx5_priv, health); dev = container_of(priv, struct mlx5_core_dev, priv); mlx5_core_warn(dev, "handling bad device here\n"); + mlx5_handle_bad_state(dev); } static const char *hsynd_str(u8 synd) @@ -147,6 +209,10 @@ static void print_health_info(struct mlx5_core_dev *dev) u32 fw; int i; + /* If the syndrom is 0, the device is OK and no need to print buffer */ + if (!ioread8(&h->synd)) + return; + for (i = 0; i < ARRAY_SIZE(h->assert_var); i++) dev_err(&dev->pdev->dev, "assert_var[%d] 0x%08x\n", i, ioread32be(h->assert_var + i)); @@ -178,6 +244,12 @@ static void poll_health(unsigned long data) struct mlx5_core_health *health = &dev->priv.health; u32 count; + if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + trigger_cmd_completions(dev); + mod_timer(&health->timer, get_next_poll_jiffies()); + return; + } + count = ioread32be(health->health_counter); if (count == health->prev) ++health->miss_counter; diff --git a/drivers/net/ethernet/mellanox/mlx5/core/main.c b/drivers/net/ethernet/mellanox/mlx5/core/main.c index b6edc58766ad..a103a54d6660 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/main.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/main.c @@ -45,6 +45,7 @@ #include #include #include +#include #include #include "mlx5_core.h" @@ -181,6 +182,34 @@ static int set_dma_caps(struct pci_dev *pdev) return err; } +static int mlx5_pci_enable_device(struct mlx5_core_dev *dev) +{ + struct pci_dev *pdev = dev->pdev; + int err = 0; + + mutex_lock(&dev->pci_status_mutex); + if (dev->pci_status == MLX5_PCI_STATUS_DISABLED) { + err = pci_enable_device(pdev); + if (!err) + dev->pci_status = MLX5_PCI_STATUS_ENABLED; + } + mutex_unlock(&dev->pci_status_mutex); + + return err; +} + +static void mlx5_pci_disable_device(struct mlx5_core_dev *dev) +{ + struct pci_dev *pdev = dev->pdev; + + mutex_lock(&dev->pci_status_mutex); + if (dev->pci_status == MLX5_PCI_STATUS_ENABLED) { + pci_disable_device(pdev); + dev->pci_status = MLX5_PCI_STATUS_DISABLED; + } + mutex_unlock(&dev->pci_status_mutex); +} + static int request_bar(struct pci_dev *pdev) { int err = 0; @@ -807,7 +836,7 @@ static int mlx5_pci_init(struct mlx5_core_dev *dev, struct mlx5_priv *priv) if (!priv->dbg_root) return -ENOMEM; - err = pci_enable_device(pdev); + err = mlx5_pci_enable_device(dev); if (err) { dev_err(&pdev->dev, "Cannot enable PCI device, aborting\n"); goto err_dbg; @@ -841,7 +870,7 @@ err_clr_master: pci_clear_master(dev->pdev); release_bar(dev->pdev); err_disable: - pci_disable_device(dev->pdev); + mlx5_pci_disable_device(dev); err_dbg: debugfs_remove(priv->dbg_root); @@ -853,7 +882,7 @@ static void mlx5_pci_close(struct mlx5_core_dev *dev, struct mlx5_priv *priv) iounmap(dev->iseg); pci_clear_master(dev->pdev); release_bar(dev->pdev); - pci_disable_device(dev->pdev); + mlx5_pci_disable_device(dev); debugfs_remove(priv->dbg_root); } @@ -863,13 +892,25 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv) struct pci_dev *pdev = dev->pdev; int err; + mutex_lock(&dev->intf_state_mutex); + if (dev->interface_state == MLX5_INTERFACE_STATE_UP) { + dev_warn(&dev->pdev->dev, "%s: interface is up, NOP\n", + __func__); + goto out; + } + dev_info(&pdev->dev, "firmware version: %d.%d.%d\n", fw_rev_maj(dev), fw_rev_min(dev), fw_rev_sub(dev)); + /* on load removing any previous indication of internal error, device is + * up + */ + dev->state = MLX5_DEVICE_STATE_UP; + err = mlx5_cmd_init(dev); if (err) { dev_err(&pdev->dev, "Failed initializing command interface, aborting\n"); - return err; + goto out_err; } mlx5_pagealloc_init(dev); @@ -994,6 +1035,10 @@ static int mlx5_load_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv) if (err) pr_info("failed request module on %s\n", MLX5_IB_MOD); + dev->interface_state = MLX5_INTERFACE_STATE_UP; +out: + mutex_unlock(&dev->intf_state_mutex); + return 0; err_reg_dev: @@ -1024,7 +1069,7 @@ err_stop_poll: mlx5_stop_health_poll(dev); if (mlx5_cmd_teardown_hca(dev)) { dev_err(&dev->pdev->dev, "tear_down_hca failed, skip cleanup\n"); - return err; + goto out_err; } err_pagealloc_stop: @@ -1040,13 +1085,23 @@ err_pagealloc_cleanup: mlx5_pagealloc_cleanup(dev); mlx5_cmd_cleanup(dev); +out_err: + dev->state = MLX5_DEVICE_STATE_INTERNAL_ERROR; + mutex_unlock(&dev->intf_state_mutex); + return err; } static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv) { - int err; + int err = 0; + mutex_lock(&dev->intf_state_mutex); + if (dev->interface_state == MLX5_INTERFACE_STATE_DOWN) { + dev_warn(&dev->pdev->dev, "%s: interface is down, NOP\n", + __func__); + goto out; + } mlx5_unregister_device(dev); mlx5_cleanup_mr_table(dev); mlx5_cleanup_srq_table(dev); @@ -1072,10 +1127,12 @@ static int mlx5_unload_one(struct mlx5_core_dev *dev, struct mlx5_priv *priv) mlx5_cmd_cleanup(dev); out: + dev->interface_state = MLX5_INTERFACE_STATE_DOWN; + mutex_unlock(&dev->intf_state_mutex); return err; } -static void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event, +void mlx5_core_event(struct mlx5_core_dev *dev, enum mlx5_dev_event event, unsigned long param) { struct mlx5_priv *priv = &dev->priv; @@ -1125,6 +1182,8 @@ static int init_one(struct pci_dev *pdev, INIT_LIST_HEAD(&priv->ctx_list); spin_lock_init(&priv->ctx_lock); + mutex_init(&dev->pci_status_mutex); + mutex_init(&dev->intf_state_mutex); err = mlx5_pci_init(dev, priv); if (err) { dev_err(&pdev->dev, "mlx5_pci_init failed with error code %d\n", err); @@ -1172,6 +1231,112 @@ static void remove_one(struct pci_dev *pdev) kfree(dev); } +static pci_ers_result_t mlx5_pci_err_detected(struct pci_dev *pdev, + pci_channel_state_t state) +{ + struct mlx5_core_dev *dev = pci_get_drvdata(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); + mlx5_pci_disable_device(dev); + return state == pci_channel_io_perm_failure ? + PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_NEED_RESET; +} + +static pci_ers_result_t mlx5_pci_slot_reset(struct pci_dev *pdev) +{ + struct mlx5_core_dev *dev = pci_get_drvdata(pdev); + int err = 0; + + dev_info(&pdev->dev, "%s was called\n", __func__); + + err = mlx5_pci_enable_device(dev); + if (err) { + dev_err(&pdev->dev, "%s: mlx5_pci_enable_device failed with error code: %d\n" + , __func__, err); + return PCI_ERS_RESULT_DISCONNECT; + } + pci_set_master(pdev); + pci_set_power_state(pdev, PCI_D0); + pci_restore_state(pdev); + + return err ? PCI_ERS_RESULT_DISCONNECT : PCI_ERS_RESULT_RECOVERED; +} + +void mlx5_disable_device(struct mlx5_core_dev *dev) +{ + mlx5_pci_err_detected(dev->pdev, 0); +} + +/* wait for the device to show vital signs. For now we check + * that we can read the device ID and that the health buffer + * shows a non zero value which is different than 0xffffffff + */ +static void wait_vital(struct pci_dev *pdev) +{ + struct mlx5_core_dev *dev = pci_get_drvdata(pdev); + struct mlx5_core_health *health = &dev->priv.health; + const int niter = 100; + u32 count; + u16 did; + int i; + + /* Wait for firmware to be ready after reset */ + msleep(1000); + for (i = 0; i < niter; i++) { + if (pci_read_config_word(pdev, 2, &did)) { + dev_warn(&pdev->dev, "failed reading config word\n"); + break; + } + if (did == pdev->device) { + dev_info(&pdev->dev, "device ID correctly read after %d iterations\n", i); + break; + } + msleep(50); + } + if (i == niter) + dev_warn(&pdev->dev, "%s-%d: could not read device ID\n", __func__, __LINE__); + + for (i = 0; i < niter; i++) { + count = ioread32be(health->health_counter); + if (count && count != 0xffffffff) { + dev_info(&pdev->dev, "Counter value 0x%x after %d iterations\n", count, i); + break; + } + msleep(50); + } + + if (i == niter) + dev_warn(&pdev->dev, "%s-%d: could not read device ID\n", __func__, __LINE__); +} + +static void mlx5_pci_resume(struct pci_dev *pdev) +{ + struct mlx5_core_dev *dev = pci_get_drvdata(pdev); + struct mlx5_priv *priv = &dev->priv; + int err; + + dev_info(&pdev->dev, "%s was called\n", __func__); + + pci_save_state(pdev); + wait_vital(pdev); + + err = mlx5_load_one(dev, priv); + if (err) + dev_err(&pdev->dev, "%s: mlx5_load_one failed with error code: %d\n" + , __func__, err); + else + dev_info(&pdev->dev, "%s: device recovered\n", __func__); +} + +static const struct pci_error_handlers mlx5_err_handler = { + .error_detected = mlx5_pci_err_detected, + .slot_reset = mlx5_pci_slot_reset, + .resume = mlx5_pci_resume +}; + static const struct pci_device_id mlx5_core_pci_table[] = { { PCI_VDEVICE(MELLANOX, 0x1011) }, /* Connect-IB */ { PCI_VDEVICE(MELLANOX, 0x1012) }, /* Connect-IB VF */ @@ -1188,7 +1353,8 @@ static struct pci_driver mlx5_core_driver = { .name = DRIVER_NAME, .id_table = mlx5_core_pci_table, .probe = init_one, - .remove = remove_one + .remove = remove_one, + .err_handler = &mlx5_err_handler }; static int __init init(void) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h index 30c0be721b08..cee5b7a839bc 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h +++ b/drivers/net/ethernet/mellanox/mlx5/core/mlx5_core.h @@ -86,6 +86,10 @@ int mlx5_query_hca_caps(struct mlx5_core_dev *dev); int mlx5_query_board_id(struct mlx5_core_dev *dev); int mlx5_cmd_init_hca(struct mlx5_core_dev *dev); int mlx5_cmd_teardown_hca(struct mlx5_core_dev *dev); +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 mlx5e_init(void); void mlx5e_cleanup(void); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c index 76432a510ac2..1cda5d268ec9 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/pagealloc.c @@ -493,15 +493,20 @@ int mlx5_reclaim_startup_pages(struct mlx5_core_dev *dev) struct fw_page *fwp; struct rb_node *p; int nclaimed = 0; - int err; + int err = 0; do { p = rb_first(&dev->priv.page_root); if (p) { fwp = rb_entry(p, struct fw_page, rb_node); - err = reclaim_pages(dev, fwp->func_id, - optimal_reclaimed_pages(), - &nclaimed); + if (dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + free_4k(dev, fwp->addr); + nclaimed = 1; + } else { + err = reclaim_pages(dev, fwp->func_id, + optimal_reclaimed_pages(), + &nclaimed); + } if (err) { mlx5_core_warn(dev, "failed reclaiming pages (%d)\n", err); -- cgit v1.2.3