From f4df233157339147539324ac3e86b5ec44513e83 Mon Sep 17 00:00:00 2001 From: Alaa Hleihel Date: Mon, 1 Jun 2020 15:40:32 -0400 Subject: [PATCH 279/312] [netdrv] net/mlx5: Fix a race when moving command interface to events mode Message-id: <20200601154102.25980-10-ahleihel@redhat.com> Patchwork-id: 315714 Patchwork-instance: patchwork O-Subject: [RHEL8.3 BZ 1842258 09/39] net/mlx5: Fix a race when moving command interface to events mode Bugzilla: 1842258 RH-Acked-by: Honggang Li RH-Acked-by: Kamal Heib RH-Acked-by: Marcelo Leitner RH-Acked-by: Jarod Wilson Bugzilla: http://bugzilla.redhat.com/1842258 Upstream: v5.7-rc7 commit d43b7007dbd1195a5b6b83213e49b1516aaf6f5e Author: Eran Ben Elisha Date: Wed Mar 18 21:44:32 2020 +0200 net/mlx5: Fix a race when moving command interface to events mode After driver creates (via FW command) an EQ for commands, the driver will be informed on new commands completion by EQE. However, due to a race in driver's internal command mode metadata update, some new commands will still be miss-handled by driver as if we are in polling mode. Such commands can get two non forced completion, leading to already freed command entry access. CREATE_EQ command, that maps EQ to the command queue must be posted to the command queue while it is empty and no other command should be posted. Add SW mechanism that once the CREATE_EQ command is about to be executed, all other commands will return error without being sent to the FW. Allow sending other commands only after successfully changing the driver's internal command mode metadata. We can safely return error to all other commands while creating the command EQ, as all other commands might be sent from the user/application during driver load. Application can rerun them later after driver's load was finished. Fixes: e126ba97dba9 ("mlx5: Add driver for Mellanox Connect-IB adapters") Signed-off-by: Eran Ben Elisha Signed-off-by: Moshe Shemesh Signed-off-by: Saeed Mahameed Signed-off-by: Alaa Hleihel Signed-off-by: Frantisek Hrbata --- drivers/net/ethernet/mellanox/mlx5/core/cmd.c | 35 ++++++++++++++++++++++++--- drivers/net/ethernet/mellanox/mlx5/core/eq.c | 3 +++ include/linux/mlx5/driver.h | 8 ++++++ 3 files changed, 42 insertions(+), 4 deletions(-) diff --git a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c index 50783828d2e8..2280bb7e748d 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/cmd.c @@ -848,6 +848,14 @@ static void free_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg); static void mlx5_free_cmd_msg(struct mlx5_core_dev *dev, struct mlx5_cmd_msg *msg); +static bool opcode_allowed(struct mlx5_cmd *cmd, u16 opcode) +{ + if (cmd->allowed_opcode == CMD_ALLOWED_OPCODE_ALL) + return true; + + return cmd->allowed_opcode == opcode; +} + static void cmd_work_handler(struct work_struct *work) { struct mlx5_cmd_work_ent *ent = container_of(work, struct mlx5_cmd_work_ent, work); @@ -914,7 +922,8 @@ static void cmd_work_handler(struct work_struct *work) /* Skip sending command to fw if internal error */ if (pci_channel_offline(dev->pdev) || - dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { + dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR || + !opcode_allowed(&dev->cmd, ent->op)) { u8 status = 0; u32 drv_synd; @@ -1405,6 +1414,22 @@ static void create_debugfs_files(struct mlx5_core_dev *dev) mlx5_cmdif_debugfs_init(dev); } +void mlx5_cmd_allowed_opcode(struct mlx5_core_dev *dev, u16 opcode) +{ + struct mlx5_cmd *cmd = &dev->cmd; + int i; + + for (i = 0; i < cmd->max_reg_cmds; i++) + down(&cmd->sem); + down(&cmd->pages_sem); + + cmd->allowed_opcode = opcode; + + up(&cmd->pages_sem); + for (i = 0; i < cmd->max_reg_cmds; i++) + up(&cmd->sem); +} + static void mlx5_cmd_change_mod(struct mlx5_core_dev *dev, int mode) { struct mlx5_cmd *cmd = &dev->cmd; @@ -1681,12 +1706,13 @@ static int cmd_exec(struct mlx5_core_dev *dev, void *in, int in_size, void *out, int err; u8 status = 0; u32 drv_synd; + u16 opcode; u8 token; + opcode = MLX5_GET(mbox_in, in, opcode); if (pci_channel_offline(dev->pdev) || - dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR) { - u16 opcode = MLX5_GET(mbox_in, in, opcode); - + dev->state == MLX5_DEVICE_STATE_INTERNAL_ERROR || + !opcode_allowed(&dev->cmd, opcode)) { err = mlx5_internal_err_ret_value(dev, opcode, &drv_synd, &status); MLX5_SET(mbox_out, out, status, status); MLX5_SET(mbox_out, out, syndrome, drv_synd); @@ -1988,6 +2014,7 @@ int mlx5_cmd_init(struct mlx5_core_dev *dev) mlx5_core_dbg(dev, "descriptor at dma 0x%llx\n", (unsigned long long)(cmd->dma)); cmd->mode = CMD_MODE_POLLING; + cmd->allowed_opcode = CMD_ALLOWED_OPCODE_ALL; create_msg_cache(dev); diff --git a/drivers/net/ethernet/mellanox/mlx5/core/eq.c b/drivers/net/ethernet/mellanox/mlx5/core/eq.c index cccea3a8eddd..ce6c621af043 100644 --- a/drivers/net/ethernet/mellanox/mlx5/core/eq.c +++ b/drivers/net/ethernet/mellanox/mlx5/core/eq.c @@ -611,11 +611,13 @@ static int create_async_eqs(struct mlx5_core_dev *dev) .nent = MLX5_NUM_CMD_EQE, .mask[0] = 1ull << MLX5_EVENT_TYPE_CMD, }; + mlx5_cmd_allowed_opcode(dev, MLX5_CMD_OP_CREATE_EQ); err = setup_async_eq(dev, &table->cmd_eq, ¶m, "cmd"); if (err) goto err1; mlx5_cmd_use_events(dev); + mlx5_cmd_allowed_opcode(dev, CMD_ALLOWED_OPCODE_ALL); param = (struct mlx5_eq_param) { .irq_index = 0, @@ -645,6 +647,7 @@ static int create_async_eqs(struct mlx5_core_dev *dev) mlx5_cmd_use_polling(dev); cleanup_async_eq(dev, &table->cmd_eq, "cmd"); err1: + mlx5_cmd_allowed_opcode(dev, CMD_ALLOWED_OPCODE_ALL); mlx5_eq_notifier_unregister(dev, &table->cq_err_nb); return err; } diff --git a/include/linux/mlx5/driver.h b/include/linux/mlx5/driver.h index df47476d6fca..c2009064805a 100644 --- a/include/linux/mlx5/driver.h +++ b/include/linux/mlx5/driver.h @@ -300,6 +300,9 @@ struct mlx5_cmd { */ spinlock_t token_lock; u8 token; +#ifndef __GENKSYMS__ + u16 allowed_opcode; +#endif unsigned long bitmask; char wq_name[MLX5_CMD_WQ_MAX_NAME]; struct workqueue_struct *wq; @@ -895,10 +898,15 @@ mlx5_frag_buf_get_idx_last_contig_stride(struct mlx5_frag_buf_ctrl *fbc, u32 ix) return min_t(u32, last_frag_stride_idx - fbc->strides_offset, fbc->sz_m1); } +enum { + CMD_ALLOWED_OPCODE_ALL, +}; + int mlx5_cmd_init(struct mlx5_core_dev *dev); void mlx5_cmd_cleanup(struct mlx5_core_dev *dev); void mlx5_cmd_use_events(struct mlx5_core_dev *dev); void mlx5_cmd_use_polling(struct mlx5_core_dev *dev); +void mlx5_cmd_allowed_opcode(struct mlx5_core_dev *dev, u16 opcode); struct mlx5_async_ctx { struct mlx5_core_dev *dev; -- 2.13.6