block: fix blk_rq_get_max_sectors for driver private requests
authorChristoph Hellwig <hch@lst.de>
Thu, 3 Mar 2016 21:43:45 +0000 (14:43 -0700)
committerJens Axboe <axboe@fb.com>
Thu, 3 Mar 2016 21:43:45 +0000 (14:43 -0700)
Driver private request types should not get the artifical cap for the
FS requests.  This is important to use the full device capabilities
for internal command or NVMe pass through commands.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reported-by: Jeff Lien <Jeff.Lien@hgst.com>
Tested-by: Jeff Lien <Jeff.Lien@hgst.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Updated by me to use an explicit check for the one command type that
does support extended checking, instead of relying on the ordering
of the enum command values - as suggested by Keith.

Signed-off-by: Jens Axboe <axboe@fb.com>
include/linux/blkdev.h

index d7f6bca707ef39101e2fad5e035c66d316eeb875..413c84fbc4edafc582c765630ccb3f3117dbc1fc 100644 (file)
@@ -895,7 +895,7 @@ static inline unsigned int blk_rq_get_max_sectors(struct request *rq)
 {
        struct request_queue *q = rq->q;
 
-       if (unlikely(rq->cmd_type == REQ_TYPE_BLOCK_PC))
+       if (unlikely(rq->cmd_type != REQ_TYPE_FS))
                return q->limits.max_hw_sectors;
 
        if (!q->limits.chunk_sectors || (rq->cmd_flags & REQ_DISCARD))