UPSTREAM: mm: don't cap request size based on read-ahead setting

We ran into a funky issue, where someone doing 256K buffered reads saw
128K requests at the device level.  Turns out it is read-ahead capping
the request size, since we use 128K as the default setting.  This
doesn't make a lot of sense - if someone is issuing 256K reads, they
should see 256K reads, regardless of the read-ahead setting, if the
underlying device can support a 256K read in a single command.

This patch introduces a bdi hint, io_pages.  This is the soft max IO
size for the lower level, I've hooked it up to the bdev settings here.
Read-ahead is modified to issue the maximum of the user request size,
and the read-ahead max size, but capped to the max request size on the
device side.  The latter is done to avoid reading ahead too much, if the
application asks for a huge read.  With this patch, the kernel behaves
like the application expects.

Change-Id: Ibe52ffac7a6e1ac86ed0c6a59a0f7a32d651ee5f
Link: http://lkml.kernel.org/r/1479498073-8657-1-git-send-email-axboe@fb.com
Signed-off-by: Jens Axboe <axboe@fb.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>

block/blk-settings.c

@@ -274,6 +274,8 @@ EXPORT_SYMBOL(blk_limits_max_hw_sectors);
 void blk_queue_max_hw_sectors(struct request_queue *q, unsigned int max_hw_sectors)
 {
 	blk_limits_max_hw_sectors(&q->limits, max_hw_sectors);
+	q->backing_dev_info.io_pages =
+			q->limits.max_sectors >> (PAGE_SHIFT - 9);
 }
 EXPORT_SYMBOL(blk_queue_max_hw_sectors);
 

block/blk-sysfs.c

@@ -178,6 +178,7 @@ queue_max_sectors_store(struct request_queue *q, const char *page, size_t count)
 
 	spin_lock_irq(q->queue_lock);
 	q->limits.max_sectors = max_sectors_kb << 1;
+	q->backing_dev_info.io_pages = max_sectors_kb >> (PAGE_SHIFT - 10);
 	spin_unlock_irq(q->queue_lock);
 
 	return ret;

include/linux/backing-dev.h

@@ -62,6 +62,7 @@ struct bdi_writeback {
 struct backing_dev_info {
 	struct list_head bdi_list;
 	unsigned long ra_pages;	/* max readahead in PAGE_CACHE_SIZE units */
+	unsigned long io_pages;	/* max allowed IO size */
 	unsigned long state;	/* Always use atomic bitops on this */
 	unsigned int capabilities; /* Device capabilities */
 	congested_fn *congested_fn; /* Function pointer if device is md/dm */

mm/readahead.c

@@ -203,17 +203,38 @@ int __do_page_cache_readahead(struct address_space *mapping, struct file *filp,
 	return ret;
 }
 
+/* Copied from fs/fs-writeback.c to avoid backport conflict. */
+static inline struct backing_dev_info *inode_to_bdi(struct inode *inode)
+{
+	struct super_block *sb = inode->i_sb;
+
+	if (sb_is_blkdev_sb(sb))
+		return inode->i_mapping->backing_dev_info;
+
+	return sb->s_bdi;
+}
+
 /*
  * Chunk the readahead into 2 megabyte units, so that we don't pin too much
  * memory at once.
  */
 int force_page_cache_readahead(struct address_space *mapping, struct file *filp,
-		pgoff_t offset, unsigned long nr_to_read)
+			       pgoff_t offset, unsigned long nr_to_read)
 {
+	struct backing_dev_info *bdi = inode_to_bdi(mapping->host);
+	struct file_ra_state *ra = &filp->f_ra;
+	unsigned long max_pages;
+
 	if (unlikely(!mapping->a_ops->readpage && !mapping->a_ops->readpages))
 		return -EINVAL;
 
-	nr_to_read = max_sane_readahead(nr_to_read);
+	/*
+	 * If the request exceeds the readahead window, allow the read to
+	 * be up to the optimal hardware IO size
+	 */
+	max_pages = max_t(unsigned long, bdi->io_pages, ra->ra_pages);
+	nr_to_read = min(nr_to_read, max_pages);
+
 	while (nr_to_read) {
 		int err;
 
@@ -380,9 +401,17 @@ ondemand_readahead(struct address_space *mapping,
 		   bool hit_readahead_marker, pgoff_t offset,
 		   unsigned long req_size)
 {
-	unsigned long max = max_sane_readahead(ra->ra_pages);
+	struct backing_dev_info *bdi = inode_to_bdi(mapping->host);
+	unsigned long max_pages = max_sane_readahead(ra->ra_pages);
 	pgoff_t prev_offset;
 
+	/*
+	 * If the request exceeds the readahead window, allow the read to
+	 * be up to the optimal hardware IO size
+	 */
+	if (req_size > max_pages && bdi->io_pages > max_pages)
+		max_pages = min(req_size, bdi->io_pages);
+
 	/*
 	 * start of file
 	 */
@@ -396,7 +425,7 @@ ondemand_readahead(struct address_space *mapping,
 	if ((offset == (ra->start + ra->size - ra->async_size) ||
 	     offset == (ra->start + ra->size))) {
 		ra->start += ra->size;
-		ra->size = get_next_ra_size(ra, max);
+		ra->size = get_next_ra_size(ra, max_pages);
 		ra->async_size = ra->size;
 		goto readit;
 	}
@@ -411,24 +440,24 @@ ondemand_readahead(struct address_space *mapping,
 		pgoff_t start;
 
 		rcu_read_lock();
-		start = page_cache_next_hole(mapping, offset + 1, max);
+		start = page_cache_next_hole(mapping, offset + 1, max_pages);
 		rcu_read_unlock();
 
-		if (!start || start - offset > max)
+		if (!start || start - offset > max_pages)
 			return 0;
 
 		ra->start = start;
 		ra->size = start - offset;	/* old async_size */
 		ra->size += req_size;
-		ra->size = get_next_ra_size(ra, max);
+		ra->size = get_next_ra_size(ra, max_pages);
 		ra->async_size = ra->size;
 		goto readit;
 	}
 
 	/*
 	 * oversize read
 	 */
-	if (req_size > max)
+	if (req_size > max_pages)
 		goto initial_readahead;
 
 	/*
@@ -444,7 +473,7 @@ ondemand_readahead(struct address_space *mapping,
 	 * Query the page cache and look for the traces(cached history pages)
 	 * that a sequential stream would leave behind.
 	 */
-	if (try_context_readahead(mapping, ra, offset, req_size, max))
+	if (try_context_readahead(mapping, ra, offset, req_size, max_pages))
 		goto readit;
 
 	/*
@@ -455,7 +484,7 @@ ondemand_readahead(struct address_space *mapping,
 
 initial_readahead:
 	ra->start = offset;
-	ra->size = get_init_ra_size(req_size, max);
+	ra->size = get_init_ra_size(req_size, max_pages);
 	ra->async_size = ra->size > req_size ? ra->size - req_size : ra->size;
 
 readit:
@@ -465,7 +494,7 @@ ondemand_readahead(struct address_space *mapping,
 	 * the resulted next readahead window into the current one.
 	 */
 	if (offset == ra->start && ra->size == ra->async_size) {
-		ra->async_size = get_next_ra_size(ra, max);
+		ra->async_size = get_next_ra_size(ra, max_pages);
 		ra->size += ra->async_size;
 	}