Message-ID: <20260126192655.GS5910@frogsfrogsfrogs>
Date: Mon, 26 Jan 2026 11:26:55 -0800
From: "Darrick J. Wong" <djwong@...nel.org>
To: Christoph Hellwig <hch@....de>
Cc: Eric Biggers <ebiggers@...nel.org>, Al Viro <viro@...iv.linux.org.uk>,
	Christian Brauner <brauner@...nel.org>, Jan Kara <jack@...e.cz>,
	David Sterba <dsterba@...e.com>, Theodore Ts'o <tytso@....edu>,
	Jaegeuk Kim <jaegeuk@...nel.org>, Chao Yu <chao@...nel.org>,
	Andrey Albershteyn <aalbersh@...hat.com>,
	Matthew Wilcox <willy@...radead.org>, linux-fsdevel@...r.kernel.org,
	linux-btrfs@...r.kernel.org, linux-ext4@...r.kernel.org,
	linux-f2fs-devel@...ts.sourceforge.net, fsverity@...ts.linux.dev
Subject: Re: [PATCH 08/16] fsverity: kick off hash readahead at data I/O
 submission time

On Mon, Jan 26, 2026 at 05:50:54AM +0100, Christoph Hellwig wrote:
> Currently all reads of the fsverity hashes is kicked off from the data
> I/O completion handler, leading to needlessly dependent I/O.  This is
> worked around a bit by performing readahead on the level 0 nodes, but
> still fairly ineffective.
> 
> Switch to a model where the ->read_folio and ->readahead methods instead
> kick off explicit readahead of the fsverity hashed so they are usually
> available at I/O completion time.
> 
> For 64k sequential reads on my test VM this improves read performance
> from 2.4GB/s - 2.6GB/s to 3.5GB/s - 3.9GB/s.  The improvements for
> random reads are likely to be even bigger.
> 
> Signed-off-by: Christoph Hellwig <hch@....de>
> Acked-by: David Sterba <dsterba@...e.com> [btrfs]
> ---
>  fs/btrfs/verity.c         |  4 +--
>  fs/ext4/readpage.c        |  7 ++++
>  fs/ext4/verity.c          | 13 +++++--
>  fs/f2fs/data.c            |  7 ++++
>  fs/f2fs/verity.c          | 13 +++++--
>  fs/verity/pagecache.c     | 39 +++++++++++++-------
>  fs/verity/read_metadata.c | 17 ++++++---
>  fs/verity/verify.c        | 76 +++++++++++++++++++++++++--------------
>  include/linux/fsverity.h  | 29 ++++++++++-----
>  9 files changed, 145 insertions(+), 60 deletions(-)
> 

<snip to the vfs stuff>

> diff --git a/fs/verity/pagecache.c b/fs/verity/pagecache.c
> index 63393f0f5834..ee99d07754ff 100644
> --- a/fs/verity/pagecache.c
> +++ b/fs/verity/pagecache.c
> @@ -10,14 +10,34 @@
>   * generic_read_merkle_tree_page - generic ->read_merkle_tree_page helper
>   * @inode:	inode containing the Merkle tree
>   * @index:	0-based index of the page in the inode
> - * @num_ra_pages: The number of Merkle tree pages that should be prefetched.
>   *
>   * The caller needs to adjust @index from the Merkle-tree relative index passed
>   * to ->read_merkle_tree_page to the actual index where the Merkle tree is
>   * stored in the page cache for @inode.
>   */
> -struct page *generic_read_merkle_tree_page(struct inode *inode, pgoff_t index,
> -		unsigned long num_ra_pages)
> +struct page *generic_read_merkle_tree_page(struct inode *inode, pgoff_t index)
> +{
> +	struct folio *folio;
> +
> +	folio = read_mapping_folio(inode->i_mapping, index, NULL);
> +	if (IS_ERR(folio))
> +		return ERR_CAST(folio);
> +	return folio_file_page(folio, index);
> +}
> +EXPORT_SYMBOL_GPL(generic_read_merkle_tree_page);
> +
> +/**
> + * generic_readahead_merkle_tree() - generic ->readahead_merkle_tree helper
> + * @inode:	inode containing the Merkle tree
> + * @index:	0-based index of the first page to read ahead in the inode
> + * @nr_pages:	number of data pages to read ahead

On second examination: *data* pages?  I thought @index/@...ex are the
range of merkle tree pages to read into the pagecache?

I'd have thought the kerneldoc would read "number of merkle tree pages
to read ahead".

> + *
> + * The caller needs to adjust @index from the Merkle-tree relative index passed
> + * to ->read_merkle_tree_page to the actual index where the Merkle tree is
> + * stored in the page cache for @inode.
> + */
> +void generic_readahead_merkle_tree(struct inode *inode, pgoff_t index,
> +		unsigned long nr_pages)
>  {
>  	struct folio *folio;
>  
> @@ -26,14 +46,9 @@ struct page *generic_read_merkle_tree_page(struct inode *inode, pgoff_t index,
>  	    !(IS_ERR(folio) && !folio_test_uptodate(folio))) {
>  		DEFINE_READAHEAD(ractl, NULL, NULL, inode->i_mapping, index);
>  
> -		if (!IS_ERR(folio))
> -			folio_put(folio);
> -		else if (num_ra_pages > 1)
> -			page_cache_ra_unbounded(&ractl, num_ra_pages, 0);
> -		folio = read_mapping_folio(inode->i_mapping, index, NULL);
> -		if (IS_ERR(folio))
> -			return ERR_CAST(folio);
> +		page_cache_ra_unbounded(&ractl, nr_pages, 0);
>  	}
> -	return folio_file_page(folio, index);
> +	if (!IS_ERR(folio))
> +		folio_put(folio);
>  }
> -EXPORT_SYMBOL_GPL(generic_read_merkle_tree_page);
> +EXPORT_SYMBOL_GPL(generic_readahead_merkle_tree);
> diff --git a/fs/verity/read_metadata.c b/fs/verity/read_metadata.c
> index cba5d6af4e04..81b82e9ddb1d 100644
> --- a/fs/verity/read_metadata.c
> +++ b/fs/verity/read_metadata.c
> @@ -28,24 +28,31 @@ static int fsverity_read_merkle_tree(struct inode *inode,
>  	if (offset >= end_offset)
>  		return 0;
>  	offs_in_page = offset_in_page(offset);
> +	index = offset >> PAGE_SHIFT;
>  	last_index = (end_offset - 1) >> PAGE_SHIFT;
>  
> +	/*
> +	 * Kick off readahead for the range we are going to read to ensure a
> +	 * single large sequential read instead of lots of small ones.
> +	 */
> +	if (inode->i_sb->s_vop->readahead_merkle_tree) {
> +		inode->i_sb->s_vop->readahead_merkle_tree(inode, index,
> +				last_index - index + 1);
> +	}
> +
>  	/*
>  	 * Iterate through each Merkle tree page in the requested range and copy
>  	 * the requested portion to userspace.  Note that the Merkle tree block
>  	 * size isn't important here, as we are returning a byte stream; i.e.,
>  	 * we can just work with pages even if the tree block size != PAGE_SIZE.
>  	 */
> -	for (index = offset >> PAGE_SHIFT; index <= last_index; index++) {
> -		unsigned long num_ra_pages =
> -			min_t(unsigned long, last_index - index + 1,
> -			      inode->i_sb->s_bdi->io_pages);
> +	for (; index <= last_index; index++) {
>  		unsigned int bytes_to_copy = min_t(u64, end_offset - offset,
>  						   PAGE_SIZE - offs_in_page);
>  		struct page *page;
>  		const void *virt;
>  
> -		page = vops->read_merkle_tree_page(inode, index, num_ra_pages);
> +		page = vops->read_merkle_tree_page(inode, index);
>  		if (IS_ERR(page)) {
>  			err = PTR_ERR(page);
>  			fsverity_err(inode,
> diff --git a/fs/verity/verify.c b/fs/verity/verify.c
> index 86067c8b40cf..32cadb71953c 100644
> --- a/fs/verity/verify.c
> +++ b/fs/verity/verify.c
> @@ -9,6 +9,7 @@
>  
>  #include <linux/bio.h>
>  #include <linux/export.h>
> +#include <linux/pagemap.h>
>  
>  #define FS_VERITY_MAX_PENDING_BLOCKS 2
>  
> @@ -21,7 +22,6 @@ struct fsverity_pending_block {
>  struct fsverity_verification_context {
>  	struct inode *inode;
>  	struct fsverity_info *vi;
> -	unsigned long max_ra_pages;
>  
>  	/*
>  	 * This is the queue of data blocks that are pending verification.  When
> @@ -37,6 +37,49 @@ struct fsverity_verification_context {
>  
>  static struct workqueue_struct *fsverity_read_workqueue;
>  
> +/**
> + * fsverity_readahead() - kick off readahead on fsverity hashes
> + * @folio:		first file data folio that is being read
> + * @nr_pages:		number of file data pages to be read
> + *
> + * Start readahead on the fsverity hashes that are needed to verity the file
> + * data in the range from folio->index to folio->index + nr_pages.
> + *
> + * To be called from the file systems' ->read_folio and ->readahead methods to
> + * ensure that the hashes are already cached on completion of the file data
> + * read if possible.
> + */
> +void fsverity_readahead(struct folio *folio, unsigned long nr_pages)
> +{
> +	struct inode *inode = folio->mapping->host;
> +	const struct fsverity_info *vi = *fsverity_info_addr(inode);
> +	const struct merkle_tree_params *params = &vi->tree_params;
> +	u64 start_hidx = (u64)folio->index << params->log_blocks_per_page;
> +	u64 end_hidx = (((u64)folio->index + nr_pages) <<
> +			params->log_blocks_per_page) - 1;
> +	int level;
> +
> +	if (!inode->i_sb->s_vop->readahead_merkle_tree)
> +		return;
> +
> +	for (level = 0; level < params->num_levels; level++) {
> +		unsigned long level_start = params->level_start[level];
> +		unsigned long next_start_hidx = start_hidx >> params->log_arity;
> +		unsigned long next_end_hidx = end_hidx >> params->log_arity;
> +		pgoff_t long start_idx = (level_start + next_start_hidx) >>
> +				params->log_blocks_per_page;
> +		pgoff_t end_idx = (level_start + next_end_hidx) >>
> +				params->log_blocks_per_page;

The pgoff_t usage here makes the unit analysis a bit easier, thanks.

--D

> +
> +		inode->i_sb->s_vop->readahead_merkle_tree(inode, start_idx,
> +				end_idx - start_idx + 1);
> +
> +		start_hidx = next_start_hidx;
> +		end_hidx = next_end_hidx;
> +	}
> +}
> +EXPORT_SYMBOL_GPL(fsverity_readahead);
> +
>  /*
>   * Returns true if the hash block with index @hblock_idx in the tree, located in
>   * @hpage, has already been verified.
> @@ -114,8 +157,7 @@ static bool is_hash_block_verified(struct fsverity_info *vi, struct page *hpage,
>   * Return: %true if the data block is valid, else %false.
>   */
>  static bool verify_data_block(struct inode *inode, struct fsverity_info *vi,
> -			      const struct fsverity_pending_block *dblock,
> -			      unsigned long max_ra_pages)
> +			      const struct fsverity_pending_block *dblock)
>  {
>  	const u64 data_pos = dblock->pos;
>  	const struct merkle_tree_params *params = &vi->tree_params;
> @@ -200,8 +242,7 @@ static bool verify_data_block(struct inode *inode, struct fsverity_info *vi,
>  			  (params->block_size - 1);
>  
>  		hpage = inode->i_sb->s_vop->read_merkle_tree_page(inode,
> -				hpage_idx, level == 0 ? min(max_ra_pages,
> -					params->tree_pages - hpage_idx) : 0);
> +				hpage_idx);
>  		if (IS_ERR(hpage)) {
>  			fsverity_err(inode,
>  				     "Error %ld reading Merkle tree page %lu",
> @@ -272,14 +313,12 @@ static bool verify_data_block(struct inode *inode, struct fsverity_info *vi,
>  
>  static void
>  fsverity_init_verification_context(struct fsverity_verification_context *ctx,
> -				   struct inode *inode,
> -				   unsigned long max_ra_pages)
> +				   struct inode *inode)
>  {
>  	struct fsverity_info *vi = *fsverity_info_addr(inode);
>  
>  	ctx->inode = inode;
>  	ctx->vi = vi;
> -	ctx->max_ra_pages = max_ra_pages;
>  	ctx->num_pending = 0;
>  	if (vi->tree_params.hash_alg->algo_id == HASH_ALGO_SHA256 &&
>  	    sha256_finup_2x_is_optimized())
> @@ -322,8 +361,7 @@ fsverity_verify_pending_blocks(struct fsverity_verification_context *ctx)
>  	}
>  
>  	for (i = 0; i < ctx->num_pending; i++) {
> -		if (!verify_data_block(ctx->inode, vi, &ctx->pending_blocks[i],
> -				       ctx->max_ra_pages))
> +		if (!verify_data_block(ctx->inode, vi, &ctx->pending_blocks[i]))
>  			return false;
>  	}
>  	fsverity_clear_pending_blocks(ctx);
> @@ -373,7 +411,7 @@ bool fsverity_verify_blocks(struct folio *folio, size_t len, size_t offset)
>  {
>  	struct fsverity_verification_context ctx;
>  
> -	fsverity_init_verification_context(&ctx, folio->mapping->host, 0);
> +	fsverity_init_verification_context(&ctx, folio->mapping->host);
>  
>  	if (fsverity_add_data_blocks(&ctx, folio, len, offset) &&
>  	    fsverity_verify_pending_blocks(&ctx))
> @@ -403,22 +441,8 @@ void fsverity_verify_bio(struct bio *bio)
>  	struct inode *inode = bio_first_folio_all(bio)->mapping->host;
>  	struct fsverity_verification_context ctx;
>  	struct folio_iter fi;
> -	unsigned long max_ra_pages = 0;
> -
> -	if (bio->bi_opf & REQ_RAHEAD) {
> -		/*
> -		 * If this bio is for data readahead, then we also do readahead
> -		 * of the first (largest) level of the Merkle tree.  Namely,
> -		 * when a Merkle tree page is read, we also try to piggy-back on
> -		 * some additional pages -- up to 1/4 the number of data pages.
> -		 *
> -		 * This improves sequential read performance, as it greatly
> -		 * reduces the number of I/O requests made to the Merkle tree.
> -		 */
> -		max_ra_pages = bio->bi_iter.bi_size >> (PAGE_SHIFT + 2);
> -	}
>  
> -	fsverity_init_verification_context(&ctx, inode, max_ra_pages);
> +	fsverity_init_verification_context(&ctx, inode);
>  
>  	bio_for_each_folio_all(fi, bio) {
>  		if (!fsverity_add_data_blocks(&ctx, fi.folio, fi.length,
> diff --git a/include/linux/fsverity.h b/include/linux/fsverity.h
> index 121703625cc8..bade511cf3aa 100644
> --- a/include/linux/fsverity.h
> +++ b/include/linux/fsverity.h
> @@ -97,10 +97,6 @@ struct fsverity_operations {
>  	 *
>  	 * @inode: the inode
>  	 * @index: 0-based index of the page within the Merkle tree
> -	 * @num_ra_pages: The number of Merkle tree pages that should be
> -	 *		  prefetched starting at @index if the page at @index
> -	 *		  isn't already cached.  Implementations may ignore this
> -	 *		  argument; it's only a performance optimization.
>  	 *
>  	 * This can be called at any time on an open verity file.  It may be
>  	 * called by multiple processes concurrently, even with the same page.
> @@ -110,8 +106,23 @@ struct fsverity_operations {
>  	 * Return: the page on success, ERR_PTR() on failure
>  	 */
>  	struct page *(*read_merkle_tree_page)(struct inode *inode,
> -					      pgoff_t index,
> -					      unsigned long num_ra_pages);
> +					      pgoff_t index);
> +
> +	/**
> +	 * Perform readahead of a Merkle tree for the given inode.
> +	 *
> +	 * @inode: the inode
> +	 * @index: 0-based index of the first page within the Merkle tree
> +	 * @nr_pages: number of pages to be read ahead.
> +	 *
> +	 * This can be called at any time on an open verity file.  It may be
> +	 * called by multiple processes concurrently, even with the same range.
> +	 *
> +	 * Optional method so that ->read_merkle_tree_page preferably finds
> +	 * cached data instead of issuing dependent I/O.
> +	 */
> +	void (*readahead_merkle_tree)(struct inode *inode, pgoff_t index,
> +			unsigned long nr_pages);
>  
>  	/**
>  	 * Write a Merkle tree block to the given inode.
> @@ -308,8 +319,10 @@ static inline int fsverity_file_open(struct inode *inode, struct file *filp)
>  }
>  
>  void fsverity_cleanup_inode(struct inode *inode);
> +void fsverity_readahead(struct folio *folio, unsigned long nr_pages);
>  
> -struct page *generic_read_merkle_tree_page(struct inode *inode, pgoff_t index,
> -		unsigned long num_ra_pages);
> +struct page *generic_read_merkle_tree_page(struct inode *inode, pgoff_t index);
> +void generic_readahead_merkle_tree(struct inode *inode, pgoff_t index,
> +		unsigned long nr_pages);
>  
>  #endif	/* _LINUX_FSVERITY_H */
> -- 
> 2.47.3
> 
> 

Powered by blists - more mailing lists