[SCSI] sg: fix a bug in st_map_user_pages failure path

sg's st_map_user_pages is modelled on an earlier version of st's
sgl_map_user_pages, and has the same bug: if get_user_pages got some but
not all of the pages, then those got were released, but the positive res
code returned implied that they were still to be freed.

Signed-off-by: Hugh Dickins <[e-mail]>
Signed-off-by: Douglas Gilbert <[e-mail]>
Signed-off-by: James Bottomley <[e-mail]>

drivers/scsi/sg.c

···
 	   unlock_page(pages[j]); */
 	res = 0;
  out_unmap:
-	if (res > 0)
+	if (res > 0) {
 		for (j=0; j < res; j++)
 			page_cache_release(pages[j]);
+		res = 0;
+	}
 	kfree(pages);
 	return res;
 }