block: use normal I/O path for discard requests

prepare_discard_fn() was being called in a place where memory allocation
was effectively impossible.  This makes it inappropriate for all but
the most trivial translations of Linux's DISCARD operation to the block
command set.  Additionally adding a payload there makes the ownership
of the bio backing unclear as it's now allocated by the device driver
and not the submitter as usual.

It is replaced with QUEUE_FLAG_DISCARD which is used to indicate whether
the queue supports discard operations or not.  blkdev_issue_discard now
allocates a one-page, sector-length payload which is the right thing
for the common ATA and SCSI implementations.

The mtd implementation of prepare_discard_fn() is replaced with simply
checking for the request being a discard.

Largely based on a previous patch from Matthew Wilcox <matthew@wil.cx>
which did the prepare_discard_fn but not the different payload allocation
yet.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>

1 files changed, 30 insertions, 5 deletions

diff --git a/block/blk-barrier.c b/block/blk-barrier.c
index 6593ab39cfe9..21f5025c3945 100644
--- a/block/blk-barrier.c
+++ b/block/blk-barrier.c
@@ -350,6 +350,7 @@ static void blkdev_discard_end_io(struct bio *bio, int err)
 
         if (bio->bi_private)
                 complete(bio->bi_private);
+        __free_page(bio_page(bio));
 
         bio_put(bio);
 }
@@ -372,26 +373,44 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
         struct request_queue *q = bdev_get_queue(bdev);
         int type = flags & DISCARD_FL_BARRIER ?
                 DISCARD_BARRIER : DISCARD_NOBARRIER;
+        struct bio *bio;
+        struct page *page;
         int ret = 0;
 
         if (!q)
                 return -ENXIO;
 
-        if (!q->prepare_discard_fn)
+        if (!blk_queue_discard(q))
                 return -EOPNOTSUPP;
 
         while (nr_sects && !ret) {
-                struct bio *bio = bio_alloc(gfp_mask, 0);
-                if (!bio)
-                        return -ENOMEM;
+                unsigned int sector_size = q->limits.logical_block_size;
 
+                bio = bio_alloc(gfp_mask, 1);
+                if (!bio)
+                        goto out;
+                bio->bi_sector = sector;
                 bio->bi_end_io = blkdev_discard_end_io;
                 bio->bi_bdev = bdev;
                 if (flags & DISCARD_FL_WAIT)
                         bio->bi_private = &wait;
 
-                bio->bi_sector = sector;
+                /*
+                 * Add a zeroed one-sector payload as that's what
+                 * our current implementations need.  If we'll ever need
+                 * more the interface will need revisiting.
+                 */
+                page = alloc_page(GFP_KERNEL | __GFP_ZERO);
+                if (!page)
+                        goto out_free_bio;
+                if (bio_add_pc_page(q, bio, page, sector_size, 0) < sector_size)
+                        goto out_free_page;
 
+                /*
+                 * And override the bio size - the way discard works we
+                 * touch many more blocks on disk than the actual payload
+                 * length.
+                 */
                 if (nr_sects > queue_max_hw_sectors(q)) {
                         bio->bi_size = queue_max_hw_sectors(q) << 9;
                         nr_sects -= queue_max_hw_sectors(q);
@@ -414,5 +433,11 @@ int blkdev_issue_discard(struct block_device *bdev, sector_t sector,
                 bio_put(bio);
         }
         return ret;
+out_free_page:
+        __free_page(page);
+out_free_bio:
+        bio_put(bio);
+out:
+        return -ENOMEM;
 }
 EXPORT_SYMBOL(blkdev_issue_discard);