[SCSI] ide-scsi fails to call idescsi_check_condition for things like "Medium not present"

This patch started life as a response to fedora specific ide subsystem changes
that made error handling of my ATAPI tape drive fail; the specifics are in

https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=160868

The insertion of the statement rq->errors = err; near the end of
ide_end_drive_cmd() in drivers/ide/ide-io.c means that rq->errors does not
contain what it needs to in idescsi_end_request() in drivers/scsi/ide-scsi.c
anymore. Recent mainline kernels now also have this change.
Signed-off-by: Willem Riede <wrlk@riede.org>
Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>

drivers/scsi/ide-scsi.c

@@ -395,6 +395,7 @@ static int idescsi_end_request (ide_drive_t *drive, int uptodate, int nrsecs)
 	int log = test_bit(IDESCSI_LOG_CMD, &scsi->log);
 	struct Scsi_Host *host;
 	u8 *scsi_buf;
+	int errors = rq->errors;
 	unsigned long flags;
 
 	if (!(rq->flags & (REQ_SPECIAL|REQ_SENSE))) {
@@ -421,11 +422,11 @@ static int idescsi_end_request (ide_drive_t *drive, int uptodate, int nrsecs)
 			printk (KERN_WARNING "ide-scsi: %s: timed out for %lu\n",
 					drive->name, pc->scsi_cmd->serial_number);
 		pc->scsi_cmd->result = DID_TIME_OUT << 16;
-	} else if (rq->errors >= ERROR_MAX) {
+	} else if (errors >= ERROR_MAX) {
 		pc->scsi_cmd->result = DID_ERROR << 16;
 		if (log)
 			printk ("ide-scsi: %s: I/O error for %lu\n", drive->name, pc->scsi_cmd->serial_number);
-	} else if (rq->errors) {
+	} else if (errors) {
 		if (log)
 			printk ("ide-scsi: %s: check condition for %lu\n", drive->name, pc->scsi_cmd->serial_number);
 		if (!idescsi_check_condition(drive, rq))