Re: [PATCH v4] erofs: support to readahead dirent blocks in erofs_readdir()

[Gao Xiang <hsiangkao@linux.alibaba.com>]

On 2025/7/21 10:13, Chao Yu wrote:
> This patch supports to readahead more blocks in erofs_readdir(), it can
> enhance readdir performance in large direcotry.
> 
> readdir test in a large directory which contains 12000 sub-files.
> 
> 		files_per_second
> Before:		926385.54
> After:		2380435.562
> 
> Meanwhile, let's introduces a new sysfs entry to control readahead
> bytes to provide more flexible policy for readahead of readdir().
> - location: /sys/fs/erofs/<disk>/dir_ra_bytes
> - default value: 16384
> - disable readahead: set the value to 0
> 
> Signed-off-by: Chao Yu <chao@kernel.org>
> ---
> v4:
> - clean up codes and comments
>   Documentation/ABI/testing/sysfs-fs-erofs |  8 ++++++++
>   fs/erofs/dir.c                           | 14 ++++++++++++++
>   fs/erofs/internal.h                      |  4 ++++
>   fs/erofs/super.c                         |  2 ++
>   fs/erofs/sysfs.c                         |  2 ++
>   5 files changed, 30 insertions(+)
> 
> diff --git a/Documentation/ABI/testing/sysfs-fs-erofs b/Documentation/ABI/testing/sysfs-fs-erofs
> index bf3b6299c15e..85fa56ca092c 100644
> --- a/Documentation/ABI/testing/sysfs-fs-erofs
> +++ b/Documentation/ABI/testing/sysfs-fs-erofs
> @@ -35,3 +35,11 @@ Description:	Used to set or show hardware accelerators in effect
>   		and multiple accelerators are separated by '\n'.
>   		Supported accelerator(s): qat_deflate.
>   		Disable all accelerators with an empty string (echo > accel).
> +
> +What:		/sys/fs/erofs/<disk>/dir_ra_bytes
> +Date:		July 2025
> +Contact:	"Chao Yu" <chao@kernel.org>
> +Description:	Used to set or show readahead bytes during readdir(), by
> +		default the value is 16384.
> +
> +		- 0: disable readahead.
> diff --git a/fs/erofs/dir.c b/fs/erofs/dir.c
> index 3e4b38bec0aa..99745c272b60 100644
> --- a/fs/erofs/dir.c
> +++ b/fs/erofs/dir.c
> @@ -47,8 +47,12 @@ static int erofs_readdir(struct file *f, struct dir_context *ctx)
>   	struct inode *dir = file_inode(f);
>   	struct erofs_buf buf = __EROFS_BUF_INITIALIZER;
>   	struct super_block *sb = dir->i_sb;
> +	struct file_ra_state *ra = &f->f_ra;
>   	unsigned long bsz = sb->s_blocksize;
>   	unsigned int ofs = erofs_blkoff(sb, ctx->pos);
> +	pgoff_t ra_pages = DIV_ROUND_UP_POW2(
> +			EROFS_I_SB(dir)->dir_ra_bytes, PAGE_SIZE);
> +	pgoff_t nr_pages = DIV_ROUND_UP_POW2(dir->i_size, PAGE_SIZE);
>   	int err = 0;
>   	bool initial = true;
>   
> @@ -63,6 +67,16 @@ static int erofs_readdir(struct file *f, struct dir_context *ctx)
>   			break;
>   		}
>   
> +		/* readahead blocks to enhance performance in large directory */
> +		if (ra_pages) {
> +			pgoff_t idx = DIV_ROUND_UP(ctx->pos, PAGE_SIZE);

Can we use DIV_ROUND_UP_POW2 here too? If it's okay,
I will update the patch manually when applied.

Otherwise it looks good to me,
Reviewed-by: Gao Xiang <hsiangkao@linux.alibaba.com>

Thanks,
Gao Xiang