scsi_error.c

来自「Linux Kernel 2.6.9 for OMAP1710」· C语言 代码 · 共 1,846 行 · 第 1/4 页

	struct scsi_device *sdev;

	shost_for_each_device(sdev, shost) {
		stu_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry)
			if (scmd->device == sdev && SCSI_SENSE_VALID(scmd) &&
			    scsi_check_sense(scmd) == FAILED ) {
				stu_scmd = scmd;
				break;
			}

		if (!stu_scmd)
			continue;

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending START_UNIT to sdev:"
						  " 0x%p\n", current->comm, sdev));

		if (!scsi_eh_try_stu(stu_scmd)) {
			if (!scsi_device_online(sdev) ||
			    !scsi_eh_tur(stu_scmd)) {
				list_for_each_safe(lh, lh_sf, work_q) {
					scmd = list_entry(lh, struct scsi_cmnd, eh_entry);
					if (scmd->device == sdev)
						scsi_eh_finish_cmd(scmd, done_q);
				}
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3,
						printk("%s: START_UNIT failed to sdev:"
						       " 0x%p\n", current->comm, sdev));
		}
	}

	return list_empty(work_q);
}


/**
 * scsi_eh_bus_device_reset - send bdr if needed
 * @shost:	scsi host being recovered.
 * @eh_done_q:	list_head for processed commands.
 *
 * Notes:
 *    Try a bus device reset.  still, look to see whether we have multiple
 *    devices that are jammed or not - if we have multiple devices, it
 *    makes no sense to try bus_device_reset - we really would need to try
 *    a bus_reset instead. 
 **/
static int scsi_eh_bus_device_reset(struct Scsi_Host *shost,
				    struct list_head *work_q,
				    struct list_head *done_q)
{
	struct list_head *lh, *lh_sf;
	struct scsi_cmnd *scmd, *bdr_scmd;
	struct scsi_device *sdev;
	int rtn;

	shost_for_each_device(sdev, shost) {
		bdr_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry)
			if (scmd->device == sdev) {
				bdr_scmd = scmd;
				break;
			}

		if (!bdr_scmd)
			continue;

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending BDR sdev:"
						  " 0x%p\n", current->comm,
						  sdev));
		rtn = scsi_try_bus_device_reset(bdr_scmd);
		if (rtn == SUCCESS) {
			if (!scsi_device_online(sdev) ||
			    !scsi_eh_tur(bdr_scmd)) {
				list_for_each_safe(lh, lh_sf,
						   work_q) {
					scmd = list_entry(lh, struct
							  scsi_cmnd,
							  eh_entry);
					if (scmd->device == sdev)
						scsi_eh_finish_cmd(scmd,
								   done_q);
				}
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: BDR"
							  " failed sdev:"
							  "0x%p\n",
							  current->comm,
							   sdev));
		}
	}

	return list_empty(work_q);
}

/**
 * scsi_try_bus_reset - ask host to perform a bus reset
 * @scmd:	SCSI cmd to send bus reset.
 **/
static int scsi_try_bus_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;

	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Snd Bus RST\n",
					  __FUNCTION__));
	scmd->owner = SCSI_OWNER_LOWLEVEL;
	scmd->serial_number_at_timeout = scmd->serial_number;

	if (!scmd->device->host->hostt->eh_bus_reset_handler)
		return FAILED;

	spin_lock_irqsave(scmd->device->host->host_lock, flags);
	rtn = scmd->device->host->hostt->eh_bus_reset_handler(scmd);
	spin_unlock_irqrestore(scmd->device->host->host_lock, flags);

	if (rtn == SUCCESS) {
		if (!scmd->device->host->hostt->skip_settle_delay)
			ssleep(BUS_RESET_SETTLE_TIME);
		spin_lock_irqsave(scmd->device->host->host_lock, flags);
		scsi_report_bus_reset(scmd->device->host, scmd->device->channel);
		spin_unlock_irqrestore(scmd->device->host->host_lock, flags);
	}

	return rtn;
}

/**
 * scsi_try_host_reset - ask host adapter to reset itself
 * @scmd:	SCSI cmd to send hsot reset.
 **/
static int scsi_try_host_reset(struct scsi_cmnd *scmd)
{
	unsigned long flags;
	int rtn;

	SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Snd Host RST\n",
					  __FUNCTION__));
	scmd->owner = SCSI_OWNER_LOWLEVEL;
	scmd->serial_number_at_timeout = scmd->serial_number;

	if (!scmd->device->host->hostt->eh_host_reset_handler)
		return FAILED;

	spin_lock_irqsave(scmd->device->host->host_lock, flags);
	rtn = scmd->device->host->hostt->eh_host_reset_handler(scmd);
	spin_unlock_irqrestore(scmd->device->host->host_lock, flags);

	if (rtn == SUCCESS) {
		if (!scmd->device->host->hostt->skip_settle_delay)
			ssleep(HOST_RESET_SETTLE_TIME);
		spin_lock_irqsave(scmd->device->host->host_lock, flags);
		scsi_report_bus_reset(scmd->device->host, scmd->device->channel);
		spin_unlock_irqrestore(scmd->device->host->host_lock, flags);
	}

	return rtn;
}

/**
 * scsi_eh_bus_reset - send a bus reset 
 * @shost:	scsi host being recovered.
 * @eh_done_q:	list_head for processed commands.
 **/
static int scsi_eh_bus_reset(struct Scsi_Host *shost,
			     struct list_head *work_q,
			     struct list_head *done_q)
{
	struct list_head *lh, *lh_sf;
	struct scsi_cmnd *scmd;
	struct scsi_cmnd *chan_scmd;
	unsigned int channel;
	int rtn;

	/*
	 * we really want to loop over the various channels, and do this on
	 * a channel by channel basis.  we should also check to see if any
	 * of the failed commands are on soft_reset devices, and if so, skip
	 * the reset.  
	 */

	for (channel = 0; channel <= shost->max_channel; channel++) {
		chan_scmd = NULL;
		list_for_each_entry(scmd, work_q, eh_entry) {
			if (channel == scmd->device->channel) {
				chan_scmd = scmd;
				break;
				/*
				 * FIXME add back in some support for
				 * soft_reset devices.
				 */
			}
		}

		if (!chan_scmd)
			continue;
		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending BRST chan:"
						  " %d\n", current->comm,
						  channel));
		rtn = scsi_try_bus_reset(chan_scmd);
		if (rtn == SUCCESS) {
			list_for_each_safe(lh, lh_sf, work_q) {
				scmd = list_entry(lh, struct scsi_cmnd,
						  eh_entry);
				if (channel == scmd->device->channel)
					if (!scsi_device_online(scmd->device) ||
					    !scsi_eh_tur(scmd))
						scsi_eh_finish_cmd(scmd,
								   done_q);
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: BRST"
							  " failed chan: %d\n",
							  current->comm,
							  channel));
		}
	}
	return list_empty(work_q);
}

/**
 * scsi_eh_host_reset - send a host reset 
 * @work_q:	list_head for processed commands.
 * @done_q:	list_head for processed commands.
 **/
static int scsi_eh_host_reset(struct list_head *work_q,
			      struct list_head *done_q)
{
	int rtn;
	struct list_head *lh, *lh_sf;
	struct scsi_cmnd *scmd;

	if (!list_empty(work_q)) {
		scmd = list_entry(work_q->next,
				  struct scsi_cmnd, eh_entry);

		SCSI_LOG_ERROR_RECOVERY(3, printk("%s: Sending HRST\n"
						  , current->comm));

		rtn = scsi_try_host_reset(scmd);
		if (rtn == SUCCESS) {
			list_for_each_safe(lh, lh_sf, work_q) {
				scmd = list_entry(lh, struct scsi_cmnd, eh_entry);
				if (!scsi_device_online(scmd->device) ||
				    (!scsi_eh_try_stu(scmd) && !scsi_eh_tur(scmd)) ||
				    !scsi_eh_tur(scmd))
					scsi_eh_finish_cmd(scmd, done_q);
			}
		} else {
			SCSI_LOG_ERROR_RECOVERY(3, printk("%s: HRST"
							  " failed\n",
							  current->comm));
		}
	}
	return list_empty(work_q);
}

/**
 * scsi_eh_offline_sdevs - offline scsi devices that fail to recover
 * @work_q:	list_head for processed commands.
 * @done_q:	list_head for processed commands.
 *
 **/
static void scsi_eh_offline_sdevs(struct list_head *work_q,
				  struct list_head *done_q)
{
	struct list_head *lh, *lh_sf;
	struct scsi_cmnd *scmd;

	list_for_each_safe(lh, lh_sf, work_q) {
		scmd = list_entry(lh, struct scsi_cmnd, eh_entry);
		printk(KERN_INFO "scsi: Device offlined - not"
		       		" ready after error recovery: host"
				" %d channel %d id %d lun %d\n",
				scmd->device->host->host_no,
				scmd->device->channel,
				scmd->device->id,
				scmd->device->lun);
		scsi_device_set_state(scmd->device, SDEV_OFFLINE);
		if (scsi_eh_eflags_chk(scmd, SCSI_EH_CANCEL_CMD)) {
			/*
			 * FIXME: Handle lost cmds.
			 */
		}
		scsi_eh_finish_cmd(scmd, done_q);
	}
	return;
}

/**
 * scsi_decide_disposition - Disposition a cmd on return from LLD.
 * @scmd:	SCSI cmd to examine.
 *
 * Notes:
 *    This is *only* called when we are examining the status after sending
 *    out the actual data command.  any commands that are queued for error
 *    recovery (e.g. test_unit_ready) do *not* come through here.
 *
 *    When this routine returns failed, it means the error handler thread
 *    is woken.  In cases where the error code indicates an error that
 *    doesn't require the error handler read (i.e. we don't need to
 *    abort/reset), this function should return SUCCESS.
 **/
int scsi_decide_disposition(struct scsi_cmnd *scmd)
{
	int rtn;

	/*
	 * if the device is offline, then we clearly just pass the result back
	 * up to the top level.
	 */
	if (!scsi_device_online(scmd->device)) {
		SCSI_LOG_ERROR_RECOVERY(5, printk("%s: device offline - report"
						  " as SUCCESS\n",
						  __FUNCTION__));
		return SUCCESS;
	}

	/*
	 * first check the host byte, to see if there is anything in there
	 * that would indicate what we need to do.
	 */
	switch (host_byte(scmd->result)) {
	case DID_PASSTHROUGH:
		/*
		 * no matter what, pass this through to the upper layer.
		 * nuke this special code so that it looks like we are saying
		 * did_ok.
		 */
		scmd->result &= 0xff00ffff;
		return SUCCESS;
	case DID_OK:
		/*
		 * looks good.  drop through, and check the next byte.
		 */
		break;
	case DID_NO_CONNECT:
	case DID_BAD_TARGET:
	case DID_ABORT:
		/*
		 * note - this means that we just report the status back
		 * to the top level driver, not that we actually think
		 * that it indicates SUCCESS.
		 */
		return SUCCESS;
		/*
		 * when the low level driver returns did_soft_error,
		 * it is responsible for keeping an internal retry counter 
		 * in order to avoid endless loops (db)
		 *
		 * actually this is a bug in this function here.  we should
		 * be mindful of the maximum number of retries specified
		 * and not get stuck in a loop.
		 */
	case DID_SOFT_ERROR:
		goto maybe_retry;
	case DID_IMM_RETRY:
		return NEEDS_RETRY;

	case DID_ERROR:
		if (msg_byte(scmd->result) == COMMAND_COMPLETE &&
		    status_byte(scmd->result) == RESERVATION_CONFLICT)
			/*
			 * execute reservation conflict processing code
			 * lower down
			 */
			break;
		/* fallthrough */

	case DID_BUS_BUSY:
	case DID_PARITY:
		goto maybe_retry;
	case DID_TIME_OUT:
		/*
		 * when we scan the bus, we get timeout messages for
		 * these commands if there is no device available.
		 * other hosts report did_no_connect for the same thing.
		 */
		if ((scmd->cmnd[0] == TEST_UNIT_READY ||
		     scmd->cmnd[0] == INQUIRY)) {
			return SUCCESS;
		} else {
			return FAILED;
		}
	case DID_RESET:
		return SUCCESS;
	default:
		return FAILED;
	}

	/*
	 * next, check the message byte.
	 */
	if (msg_byte(scmd->result) != COMMAND_COMPLETE)
		return FAILED;

	/*
	 * check the status byte to see if this indicates anything special.
	 */
	switch (status_byte(scmd->result)) {
	case QUEUE_FULL:
		/*
		 * the case of trying to send too many commands to a
		 * tagged queueing device.
		 */
	case BUSY:
		/*
		 * device can't talk to us at the moment.  Should only
		 * occur (SAM-3) when the task queue is empty, so will cause
		 * the empty queue handling to trigger a stall in the
		 * device.
		 */
		return ADD_TO_MLQUEUE;
	case GOOD:
	case COMMAND_TERMINATED:
		return SUCCESS;
	case CHECK_CONDITION:
		rtn = scsi_check_sense(scmd);
		if (rtn == NEEDS_RETRY)
			goto maybe_retry;
		/* if rtn == FAILED, we have no sense information;
		 * returning FAILED will wake the error handler thread
		 * to collect the sense and redo the decide
		 * disposition */
		return rtn;
	case CONDITION_GOOD:
	case INTERMEDIATE_GOOD:
	case INTERMEDIATE_C_GOOD:
		/*
		 * who knows?  FIXME(eric)
		 */
		return SUCCESS;

	case RESERVATION_CONFLICT:
		printk("scsi%d (%d,%d,%d) : reservation conflict\n",
		       scmd->device->host->host_no, scmd->device->channel,
		       scmd->device->id, scmd->device->lun);
		return SUCCESS; /* causes immediate i/o error */
	default:
		return FAILED;
	}
	return FAILED;

      maybe_retry:

	/* we requeue for retry because the error was retryable, and
	 * the request was not marked fast fail.  Note that above,
	 * even if the request is marked fast fail, we still requeue
	 * for queue congestion conditions (QUEUE_FULL or BUSY) */
	if ((++scmd->retries) < scmd->allowed 
	    && !blk_noretry_request(scmd->request)) {
		return NEEDS_RETRY;
	} else {
		/*
		 * no more retries - report this one back to upper level.
		 */
		return SUCCESS;
	}
}