Currently, the hibmcge netdev acquires the rtnl_lock in pci_error_handlers.reset_prepare() and releases it in pci_error_handlers.reset_done(). However, in the PCI framework: pci_reset_bus - __pci_reset_slot - pci_slot_save_and_disable_locked - pci_dev_save_and_disable - err_handler->reset_prepare(dev); In pci_slot_save_and_disable_locked(): list_for_each_entry(dev, &slot->bus->devices, bus_list) { if (!dev->slot || dev->slot!= slot) continue; pci_dev_save_and_disable(dev); if (dev->subordinate) pci_bus_save_and_disable_locked(dev->subordinate); } This will iterate through all devices under the current bus and execute err_handler->reset_prepare(), causing two devices of the hibmcge driver to sequentially request the rtnl_lock, leading to a deadlock. Since the driver now executes netif_device_detach() before the reset process, it will not concurrently with other netdev APIs, so there is no need to hold the rtnl_lock now. Therefore, this patch removes the rtnl_lock during the reset process and adjusts the position of HBG_NIC_STATE_RESETTING to ensure that multiple resets are not executed concurrently. Fixes: 3f5a61f6d504f ("net: hibmcge: Add reset supported in this module") Signed-off-by: Jijie Shao --- ChangeLog: v1 -> v2: - Fix a concurrency issue, suggested by Simon Horman v1: https://lore.kernel.org/all/20250731134749.4090041-1-shaojijie@huawei.com/ --- drivers/net/ethernet/hisilicon/hibmcge/hbg_err.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-) diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_err.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_err.c index 503cfbfb4a8a..83cf75bf7a17 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_err.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_err.c @@ -53,9 +53,11 @@ static int hbg_reset_prepare(struct hbg_priv *priv, enum hbg_reset_type type) { int ret; - ASSERT_RTNL(); + if (test_and_set_bit(HBG_NIC_STATE_RESETTING, &priv->state)) + return -EBUSY; if (netif_running(priv->netdev)) { + clear_bit(HBG_NIC_STATE_RESETTING, &priv->state); dev_warn(&priv->pdev->dev, "failed to reset because port is up\n"); return -EBUSY; @@ -64,7 +66,6 @@ static int hbg_reset_prepare(struct hbg_priv *priv, enum hbg_reset_type type) netif_device_detach(priv->netdev); priv->reset_type = type; - set_bit(HBG_NIC_STATE_RESETTING, &priv->state); clear_bit(HBG_NIC_STATE_RESET_FAIL, &priv->state); ret = hbg_hw_event_notify(priv, HBG_HW_EVENT_RESET); if (ret) { @@ -84,29 +85,26 @@ static int hbg_reset_done(struct hbg_priv *priv, enum hbg_reset_type type) type != priv->reset_type) return 0; - ASSERT_RTNL(); - - clear_bit(HBG_NIC_STATE_RESETTING, &priv->state); ret = hbg_rebuild(priv); if (ret) { priv->stats.reset_fail_cnt++; set_bit(HBG_NIC_STATE_RESET_FAIL, &priv->state); + clear_bit(HBG_NIC_STATE_RESETTING, &priv->state); dev_err(&priv->pdev->dev, "failed to rebuild after reset\n"); return ret; } netif_device_attach(priv->netdev); + clear_bit(HBG_NIC_STATE_RESETTING, &priv->state); dev_info(&priv->pdev->dev, "reset done\n"); return ret; } -/* must be protected by rtnl lock */ int hbg_reset(struct hbg_priv *priv) { int ret; - ASSERT_RTNL(); ret = hbg_reset_prepare(priv, HBG_RESET_TYPE_FUNCTION); if (ret) return ret; @@ -171,7 +169,6 @@ static void hbg_pci_err_reset_prepare(struct pci_dev *pdev) struct net_device *netdev = pci_get_drvdata(pdev); struct hbg_priv *priv = netdev_priv(netdev); - rtnl_lock(); hbg_reset_prepare(priv, HBG_RESET_TYPE_FLR); } @@ -181,7 +178,6 @@ static void hbg_pci_err_reset_done(struct pci_dev *pdev) struct hbg_priv *priv = netdev_priv(netdev); hbg_reset_done(priv, HBG_RESET_TYPE_FLR); - rtnl_unlock(); } static const struct pci_error_handlers hbg_pci_err_handler = { -- 2.33.0 When the network port is down, the queue is released, and ring->len is 0. In debugfs, hbg_get_queue_used_num() will be called, which may lead to a division by zero issue. This patch adds a check, if ring->len is 0, hbg_get_queue_used_num() directly returns 0. Fixes: 40735e7543f9 ("net: hibmcge: Implement .ndo_start_xmit function") Signed-off-by: Jijie Shao Reviewed-by: Simon Horman --- drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h | 3 +++ 1 file changed, 3 insertions(+) diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h index 2883a5899ae2..2aecc73f3d49 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_txrx.h @@ -29,6 +29,9 @@ static inline bool hbg_fifo_is_full(struct hbg_priv *priv, enum hbg_dir dir) static inline u32 hbg_get_queue_used_num(struct hbg_ring *ring) { + if (!ring->len) + return 0; + return (ring->ntu + ring->len - ring->ntc) % ring->len; } -- 2.33.0 Currently, after modifying device port mode, the np_link_ok state is immediately checked. At this point, the device may not yet ready, leading to the querying of an intermediate state. This patch will poll to check if np_link is ok after modifying device port mode, and only report np_link_fail upon timeout. Fixes: e0306637e85d ("net: hibmcge: Add support for mac link exception handling feature") Signed-off-by: Jijie Shao Reviewed-by: Simon Horman --- drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c index 8cca8316ba40..d0aa0661ecd4 100644 --- a/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c +++ b/drivers/net/ethernet/hisilicon/hibmcge/hbg_hw.c @@ -12,6 +12,8 @@ #define HBG_HW_EVENT_WAIT_TIMEOUT_US (2 * 1000 * 1000) #define HBG_HW_EVENT_WAIT_INTERVAL_US (10 * 1000) +#define HBG_MAC_LINK_WAIT_TIMEOUT_US (500 * 1000) +#define HBG_MAC_LINK_WAIT_INTERVAL_US (5 * 1000) /* little endian or big endian. * ctrl means packet description, data means skb packet data */ @@ -228,6 +230,9 @@ void hbg_hw_fill_buffer(struct hbg_priv *priv, u32 buffer_dma_addr) void hbg_hw_adjust_link(struct hbg_priv *priv, u32 speed, u32 duplex) { + u32 link_status; + int ret; + hbg_hw_mac_enable(priv, HBG_STATUS_DISABLE); hbg_reg_write_field(priv, HBG_REG_PORT_MODE_ADDR, @@ -239,8 +244,14 @@ void hbg_hw_adjust_link(struct hbg_priv *priv, u32 speed, u32 duplex) hbg_hw_mac_enable(priv, HBG_STATUS_ENABLE); - if (!hbg_reg_read_field(priv, HBG_REG_AN_NEG_STATE_ADDR, - HBG_REG_AN_NEG_STATE_NP_LINK_OK_B)) + /* wait MAC link up */ + ret = readl_poll_timeout(priv->io_base + HBG_REG_AN_NEG_STATE_ADDR, + link_status, + FIELD_GET(HBG_REG_AN_NEG_STATE_NP_LINK_OK_B, + link_status), + HBG_MAC_LINK_WAIT_INTERVAL_US, + HBG_MAC_LINK_WAIT_TIMEOUT_US); + if (ret) hbg_np_link_fail_task_schedule(priv); } -- 2.33.0