Re: [PATCH 2/4] scsi: add transport host byte errors

[Mike Christie <michaelc@cs.wisc.edu>]

michaelc@cs.wisc.edu wrote:
> When using multipath and the fast_io_fail_tmo fires then the class
> can fail commands with DID_TRANSPORT_FAILFAST or drivers can use

Bah, that is going to be wrong or ugly for tape commands.

We would want to add another new value, DID_TRANSPORT_ABORTED which
would check the retries or device type. Drivers would use this when
failing IO that was possibly running on the device.

> -
> +	case DID_TRANSPORT_BLOCKED:
> +		/*
> +		 * LLD/transport was disrupted during processing the IO.
> +		 * The transport is now blocked and attempting to recover,
> +		 * and the transport will decide what to do with the IO
> +		 * based on its timers and recovery capablilities.
> +		 */
> +		return NEEDS_RETRY;
> +	case DID_TRANSPORT_FAILFAST:
> +		/*
> +		 * The transport decided to failfast the IO (most likely
> +		 * the fast io fail tmo fired), so send IO directly upwards.
> +		 */
> +		return SUCCESS;
>  	case DID_ERROR:
>  		if (msg_byte(scmd->result) == COMMAND_COMPLETE &&
>  		    status_byte(scmd->result) == RESERVATION_CONFLICT)