[PATCH 05/10] nvme/nvme-fabrics: introduce nvmf_error_recovery_work API

[James Smart]

On 10/20/2021 3:38 AM, Max Gurtovoy wrote:
> Error recovery work is duplicated in RDMA and TCP transports. Move this
> logic to common code. For that, introduce 2 new ctrl ops to teardown IO
> and admin queue.
> 
> Also update the RDMA/TCP transport drivers to use this API and remove
> the duplicated code.
> 
> Reviewed-by: Israel Rukshin <israelr at nvidia.com>
> Reviewed-by: Chaitanya Kulkarni <kch at nvidia.com>
> Signed-off-by: Max Gurtovoy <mgurtovoy at nvidia.com>
> ---
>   drivers/nvme/host/fabrics.c | 23 +++++++++++
>   drivers/nvme/host/fabrics.h |  1 +
>   drivers/nvme/host/nvme.h    |  4 ++
>   drivers/nvme/host/rdma.c    | 78 +++++++++++++++----------------------
>   drivers/nvme/host/tcp.c     | 46 +++++++---------------
>   5 files changed, 73 insertions(+), 79 deletions(-)
> 
> diff --git a/drivers/nvme/host/fabrics.c b/drivers/nvme/host/fabrics.c
> index 2edd086fa922..5a770196eb60 100644
> --- a/drivers/nvme/host/fabrics.c
> +++ b/drivers/nvme/host/fabrics.c
> @@ -493,6 +493,29 @@ void nvmf_reconnect_or_remove(struct nvme_ctrl *ctrl)
>   }
>   EXPORT_SYMBOL_GPL(nvmf_reconnect_or_remove);
>   
> +void nvmf_error_recovery_work(struct work_struct *work)
> +{
> +	struct nvme_ctrl *ctrl = container_of(work,
> +				struct nvme_ctrl, err_work);
> +
> +	nvme_stop_keep_alive(ctrl);
> +	ctrl->ops->teardown_ctrl_io_queues(ctrl, false);
> +	/* unquiesce to fail fast pending requests */
> +	nvme_start_queues(ctrl);
> +	ctrl->ops->teardown_ctrl_admin_queue(ctrl, false);
> +	blk_mq_unquiesce_queue(ctrl->admin_q);
> +
> +	if (!nvme_change_ctrl_state(ctrl, NVME_CTRL_CONNECTING)) {
> +		/* state change failure is ok if we started ctrl delete */
> +		WARN_ON_ONCE(ctrl->state != NVME_CTRL_DELETING &&
> +			     ctrl->state != NVME_CTRL_DELETING_NOIO);
> +		return;
> +	}
> +
> +	nvmf_reconnect_or_remove(ctrl);
> +}
> +EXPORT_SYMBOL_GPL(nvmf_error_recovery_work);
> +

Same comments on just use nvme_reset_ctrl().

-- james