dm: fix REQ_OP_ZONE_REPORT bio handling
authorDamien Le Moal <damien.lemoal@wdc.com>
Mon, 8 May 2017 23:40:47 +0000 (16:40 -0700)
committerMike Snitzer <snitzer@redhat.com>
Mon, 19 Jun 2017 15:03:51 +0000 (11:03 -0400)
A REQ_OP_ZONE_REPORT bio is not a medium access command.  Its number of
sectors indicates the maximum size allowed for the report reply size and
not an amount of sectors accessed from the device.  REQ_OP_ZONE_REPORT
bios should thus not be split depending on the target device maximum I/O
length but passed as-is.  Note that it is the responsability of the
target to remap and format the report reply.

Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Reviewed-by: Hannes Reinecke <hare@suse.com>
Reviewed-by: Bart Van Assche <bart.vanassche@sandisk.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
drivers/md/dm.c

index d85adec..e38d1d7 100644 (file)
@@ -1152,7 +1152,8 @@ static int clone_bio(struct dm_target_io *tio, struct bio *bio,
                        return r;
        }
 
-       bio_advance(clone, to_bytes(sector - clone->bi_iter.bi_sector));
+       if (bio_op(bio) != REQ_OP_ZONE_REPORT)
+               bio_advance(clone, to_bytes(sector - clone->bi_iter.bi_sector));
        clone->bi_iter.bi_size = to_bytes(len);
 
        if (unlikely(bio_integrity(bio) != NULL))
@@ -1341,7 +1342,11 @@ static int __split_and_process_non_flush(struct clone_info *ci)
        if (!dm_target_is_valid(ti))
                return -EIO;
 
-       len = min_t(sector_t, max_io_len(ci->sector, ti), ci->sector_count);
+       if (bio_op(bio) == REQ_OP_ZONE_REPORT)
+               len = ci->sector_count;
+       else
+               len = min_t(sector_t, max_io_len(ci->sector, ti),
+                           ci->sector_count);
 
        r = __clone_and_map_data_bio(ci, ti, ci->sector, &len);
        if (r < 0)