On Mon, Nov 26, 2018 at 10:17:03AM +0800, Ming Lei wrote:
> From: Christoph Hellwig <hch@xxxxxx>
>
> The cluster flag implements some very old SCSI behavior. As far as I
> can tell the original intent was to enable or disable any kind of
> segment merging. But the actually visible effect to the LLDD is that
> it limits each segments to be inside a single page, which we can
> also affect by setting the maximum segment size and the segment
> boundary.
Reviewed-by: Omar Sandoval <osandov@xxxxxx>
One comment typo below.
> Signed-off-by: Christoph Hellwig <hch@xxxxxx>
>
> Replace virt boundary with segment boundary limit.
>
> Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxx>
> ---
> block/blk-merge.c | 20 ++++++++------------
> block/blk-settings.c | 3 ---
> block/blk-sysfs.c | 5 +----
> drivers/scsi/scsi_lib.c | 20 ++++++++++++++++----
> include/linux/blkdev.h | 6 ------
> 5 files changed, 25 insertions(+), 29 deletions(-)
>
[snip]
> diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> index 0df15cb738d2..78d6d05992b0 100644
> --- a/drivers/scsi/scsi_lib.c
> +++ b/drivers/scsi/scsi_lib.c
> @@ -1810,6 +1810,8 @@ static int scsi_map_queues(struct blk_mq_tag_set *set)
> void __scsi_init_queue(struct Scsi_Host *shost, struct request_queue *q)
> {
> struct device *dev = shost->dma_dev;
> + unsigned max_segment_size = dma_get_max_seg_size(dev);
> + unsigned long segment_boundary = shost->dma_boundary;
>
> /*
> * this limit is imposed by hardware restrictions
> @@ -1828,13 +1830,23 @@ void __scsi_init_queue(struct Scsi_Host *shost, struct request_queue *q)
> blk_queue_max_hw_sectors(q, shost->max_sectors);
> if (shost->unchecked_isa_dma)
> blk_queue_bounce_limit(q, BLK_BOUNCE_ISA);
> - blk_queue_segment_boundary(q, shost->dma_boundary);
> dma_set_seg_boundary(dev, shost->dma_boundary);
>
> - blk_queue_max_segment_size(q, dma_get_max_seg_size(dev));
> + /*
> + * Clustering is a really old concept from the stone age of Linux
> + * SCSI support. But the basic idea is that we never give the
> + * driver a segment that spans multiple pages. For that we need
> + * to limit the segment size, and set the segment boundary so that
> + * we never merge a second segment which is no page aligned.
Typo, "which is not page aligned".
