[libata] SCSI: improve FORMAT UNIT; minor code cleanups

* SAT specifies that FORMAT UNIT should be translated into a series
  of READ and WRITE commands that zero the ATA device.  That is far too
  cumbersome to bother with.

  Since we don't actually format the device, the old behavior of
  always returning success was inaccurate.  Change FORMAT UNIT from
  returning success immediately (old behavior) to always returning
  an error (new behavior).

* Add some comments around SYNCHRONIZE CACHE

* Shuffle scsi command code around a bit, so that things are close
  to alphabetic order.
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

drivers/ata/libata-scsi.c

@@ -2920,14 +2920,9 @@ void ata_scsi_simulate(struct ata_device *dev, struct scsi_cmnd *cmd,
 	args.done = done;
 
 	switch(scsicmd[0]) {
-		/* no-op's, complete with success */
-		case SYNCHRONIZE_CACHE:
-		case REZERO_UNIT:
-		case SEEK_6:
-		case SEEK_10:
-		case TEST_UNIT_READY:
-		case FORMAT_UNIT:		/* FIXME: correct? */
-			ata_scsi_rbuf_fill(&args, ata_scsiop_noop);
+		/* TODO: worth improving? */
+		case FORMAT_UNIT:
+			ata_scsi_invalid_field(cmd, done);
 			break;
 
 		case INQUIRY:
@@ -2985,6 +2980,20 @@ void ata_scsi_simulate(struct ata_device *dev, struct scsi_cmnd *cmd,
 			done(cmd);
 			break;
 
+		/* if we reach this, then writeback caching is disabled,
+		 * turning this into a no-op.
+		 */
+		case SYNCHRONIZE_CACHE:
+			/* fall through */
+
+		/* no-op's, complete with success */
+		case REZERO_UNIT:
+		case SEEK_6:
+		case SEEK_10:
+		case TEST_UNIT_READY:
+			ata_scsi_rbuf_fill(&args, ata_scsiop_noop);
+			break;
+
 		case SEND_DIAGNOSTIC:
 			tmp8 = scsicmd[1] & ~(1 << 3);
 			if ((tmp8 == 0x4) && (!scsicmd[3]) && (!scsicmd[4]))