[PATCH] readdir: supply dir_context.count as readdir buffer size hint

[PATCH] readdir: supply dir_context.count as readdir buffer size hint

[Miklos Szeredi]

This is a preparation for large readdir buffers in fuse.

Simply setting the fuse buffer size to the userspace buffer size should
work, the record sizes are similar (fuse's is slightly larger than libc's,
so no overflow should ever happen).

Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Signed-off-by: Jaco Kroon <jaco@uls.co.za>
---
 fs/exportfs/expfs.c    |  1 +
 fs/overlayfs/readdir.c | 12 +++++++++++-
 fs/readdir.c           | 29 ++++++++++++++---------------
 include/linux/fs.h     |  7 +++++++
 4 files changed, 33 insertions(+), 16 deletions(-)

diff --git a/fs/exportfs/expfs.c b/fs/exportfs/expfs.c
index 128dd092916b..48f0c505b50f 100644
--- a/fs/exportfs/expfs.c
+++ b/fs/exportfs/expfs.c
@@ -284,6 +284,7 @@ static int get_name(const struct path *path, char *name, struct dentry *child)
 	};
 	struct getdents_callback buffer = {
 		.ctx.actor = filldir_one,
+		.ctx.count = INT_MAX,
 		.name = name,
 	};
 
diff --git a/fs/overlayfs/readdir.c b/fs/overlayfs/readdir.c
index 881ec5592da5..126c797751e9 100644
--- a/fs/overlayfs/readdir.c
+++ b/fs/overlayfs/readdir.c
@@ -351,6 +351,7 @@ static int ovl_dir_read_merged(struct dentry *dentry, struct list_head *list,
 	struct path realpath;
 	struct ovl_readdir_data rdd = {
 		.ctx.actor = ovl_fill_merge,
+		.ctx.count = INT_MAX,
 		.dentry = dentry,
 		.list = list,
 		.root = root,
@@ -571,6 +572,7 @@ static int ovl_dir_read_impure(const struct path *path,  struct list_head *list,
 	struct ovl_cache_entry *p, *n;
 	struct ovl_readdir_data rdd = {
 		.ctx.actor = ovl_fill_plain,
+		.ctx.count = INT_MAX,
 		.list = list,
 		.root = root,
 	};
@@ -672,6 +674,7 @@ static bool ovl_fill_real(struct dir_context *ctx, const char *name,
 	struct ovl_readdir_translate *rdt =
 		container_of(ctx, struct ovl_readdir_translate, ctx);
 	struct dir_context *orig_ctx = rdt->orig_ctx;
+	bool res;
 
 	if (rdt->parent_ino && strcmp(name, "..") == 0) {
 		ino = rdt->parent_ino;
@@ -686,7 +689,10 @@ static bool ovl_fill_real(struct dir_context *ctx, const char *name,
 					  name, namelen, rdt->xinowarn);
 	}
 
-	return orig_ctx->actor(orig_ctx, name, namelen, offset, ino, d_type);
+	res = orig_ctx->actor(orig_ctx, name, namelen, offset, ino, d_type);
+	ctx->count = orig_ctx->count;
+
+	return res;
 }
 
 static bool ovl_is_impure_dir(struct file *file)
@@ -713,6 +719,7 @@ static int ovl_iterate_real(struct file *file, struct dir_context *ctx)
 	const struct ovl_layer *lower_layer = ovl_layer_lower(dir);
 	struct ovl_readdir_translate rdt = {
 		.ctx.actor = ovl_fill_real,
+		.ctx.count = ctx->count,
 		.orig_ctx = ctx,
 		.xinobits = ovl_xino_bits(ofs),
 		.xinowarn = ovl_xino_warn(ofs),
@@ -1073,6 +1080,7 @@ int ovl_check_d_type_supported(const struct path *realpath)
 	int err;
 	struct ovl_readdir_data rdd = {
 		.ctx.actor = ovl_check_d_type,
+		.ctx.count = INT_MAX,
 		.d_type_supported = false,
 	};
 
@@ -1094,6 +1102,7 @@ static int ovl_workdir_cleanup_recurse(struct ovl_fs *ofs, const struct path *pa
 	struct ovl_cache_entry *p;
 	struct ovl_readdir_data rdd = {
 		.ctx.actor = ovl_fill_plain,
+		.ctx.count = INT_MAX,
 		.list = &list,
 	};
 	bool incompat = false;
@@ -1178,6 +1187,7 @@ int ovl_indexdir_cleanup(struct ovl_fs *ofs)
 	struct ovl_cache_entry *p;
 	struct ovl_readdir_data rdd = {
 		.ctx.actor = ovl_fill_plain,
+		.ctx.count = INT_MAX,
 		.list = &list,
 	};
 
diff --git a/fs/readdir.c b/fs/readdir.c
index 857d402bc531..7764b8638978 100644
--- a/fs/readdir.c
+++ b/fs/readdir.c
@@ -222,6 +222,7 @@ SYSCALL_DEFINE3(old_readdir, unsigned int, fd,
 	CLASS(fd_pos, f)(fd);
 	struct readdir_callback buf = {
 		.ctx.actor = fillonedir,
+		.ctx.count = 1, /* Hint to fs: just one entry. */
 		.dirent = dirent
 	};
 
@@ -252,7 +253,6 @@ struct getdents_callback {
 	struct dir_context ctx;
 	struct linux_dirent __user * current_dir;
 	int prev_reclen;
-	int count;
 	int error;
 };
 
@@ -275,7 +275,7 @@ static bool filldir(struct dir_context *ctx, const char *name, int namlen,
 	if (unlikely(buf->error))
 		return false;
 	buf->error = -EINVAL;	/* only used if we fail.. */
-	if (reclen > buf->count)
+	if (reclen > ctx->count)
 		return false;
 	d_ino = ino;
 	if (sizeof(d_ino) < sizeof(ino) && d_ino != ino) {
@@ -300,7 +300,7 @@ static bool filldir(struct dir_context *ctx, const char *name, int namlen,
 
 	buf->current_dir = (void __user *)dirent + reclen;
 	buf->prev_reclen = reclen;
-	buf->count -= reclen;
+	ctx->count -= reclen;
 	return true;
 efault_end:
 	user_write_access_end();
@@ -315,7 +315,7 @@ SYSCALL_DEFINE3(getdents, unsigned int, fd,
 	CLASS(fd_pos, f)(fd);
 	struct getdents_callback buf = {
 		.ctx.actor = filldir,
-		.count = count,
+		.ctx.count = count,
 		.current_dir = dirent
 	};
 	int error;
@@ -333,7 +333,7 @@ SYSCALL_DEFINE3(getdents, unsigned int, fd,
 		if (put_user(buf.ctx.pos, &lastdirent->d_off))
 			error = -EFAULT;
 		else
-			error = count - buf.count;
+			error = count - buf.ctx.count;
 	}
 	return error;
 }
@@ -342,7 +342,6 @@ struct getdents_callback64 {
 	struct dir_context ctx;
 	struct linux_dirent64 __user * current_dir;
 	int prev_reclen;
-	int count;
 	int error;
 };
 
@@ -364,7 +363,7 @@ static bool filldir64(struct dir_context *ctx, const char *name, int namlen,
 	if (unlikely(buf->error))
 		return false;
 	buf->error = -EINVAL;	/* only used if we fail.. */
-	if (reclen > buf->count)
+	if (reclen > ctx->count)
 		return false;
 	prev_reclen = buf->prev_reclen;
 	if (!(flags & FILLDIR_FLAG_NOINTR) && prev_reclen && signal_pending(current))
@@ -384,7 +383,7 @@ static bool filldir64(struct dir_context *ctx, const char *name, int namlen,
 
 	buf->prev_reclen = reclen;
 	buf->current_dir = (void __user *)dirent + reclen;
-	buf->count -= reclen;
+	ctx->count -= reclen;
 	return true;
 
 efault_end:
@@ -400,7 +399,7 @@ SYSCALL_DEFINE3(getdents64, unsigned int, fd,
 	CLASS(fd_pos, f)(fd);
 	struct getdents_callback64 buf = {
 		.ctx.actor = filldir64,
-		.count = count,
+		.ctx.count = count,
 		.current_dir = dirent
 	};
 	int error;
@@ -419,7 +418,7 @@ SYSCALL_DEFINE3(getdents64, unsigned int, fd,
 		if (put_user(d_off, &lastdirent->d_off))
 			error = -EFAULT;
 		else
-			error = count - buf.count;
+			error = count - buf.ctx.count;
 	}
 	return error;
 }
@@ -483,6 +482,7 @@ COMPAT_SYSCALL_DEFINE3(old_readdir, unsigned int, fd,
 	CLASS(fd_pos, f)(fd);
 	struct compat_readdir_callback buf = {
 		.ctx.actor = compat_fillonedir,
+		.ctx.count = 1, /* Hint to fs: just one entry. */
 		.dirent = dirent
 	};
 
@@ -507,7 +507,6 @@ struct compat_getdents_callback {
 	struct dir_context ctx;
 	struct compat_linux_dirent __user *current_dir;
 	int prev_reclen;
-	int count;
 	int error;
 };
 
@@ -530,7 +529,7 @@ static bool compat_filldir(struct dir_context *ctx, const char *name, int namlen
 	if (unlikely(buf->error))
 		return false;
 	buf->error = -EINVAL;	/* only used if we fail.. */
-	if (reclen > buf->count)
+	if (reclen > ctx->count)
 		return false;
 	d_ino = ino;
 	if (sizeof(d_ino) < sizeof(ino) && d_ino != ino) {
@@ -554,7 +553,7 @@ static bool compat_filldir(struct dir_context *ctx, const char *name, int namlen
 
 	buf->prev_reclen = reclen;
 	buf->current_dir = (void __user *)dirent + reclen;
-	buf->count -= reclen;
+	ctx->count -= reclen;
 	return true;
 efault_end:
 	user_write_access_end();
@@ -569,8 +568,8 @@ COMPAT_SYSCALL_DEFINE3(getdents, unsigned int, fd,
 	CLASS(fd_pos, f)(fd);
 	struct compat_getdents_callback buf = {
 		.ctx.actor = compat_filldir,
+		.ctx.count = count,
 		.current_dir = dirent,
-		.count = count
 	};
 	int error;
 
@@ -587,7 +586,7 @@ COMPAT_SYSCALL_DEFINE3(getdents, unsigned int, fd,
 		if (put_user(buf.ctx.pos, &lastdirent->d_off))
 			error = -EFAULT;
 		else
-			error = count - buf.count;
+			error = count - buf.ctx.count;
 	}
 	return error;
 }
diff --git a/include/linux/fs.h b/include/linux/fs.h
index 0f2a1a572e3a..dfc5a3327124 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -2071,6 +2071,13 @@ typedef bool (*filldir_t)(struct dir_context *, const char *, int, loff_t, u64,
 struct dir_context {
 	filldir_t actor;
 	loff_t pos;
+	/*
+	 * Filesystems MUST NOT MODIFY count, but may use as a hint:
+	 * 0	    unknown
+	 * > 0      space in buffer (assume at least one entry)
+	 * INT_MAX  unlimited
+	 */
+	int count;
 };
 
 /* If OR-ed with d_type, pending signals are not checked */
-- 
2.49.0

Re: [PATCH] readdir: supply dir_context.count as readdir buffer size hint

[Christian Brauner]

On Tue, 13 May 2025 17:10:08 +0200, Miklos Szeredi wrote:
> This is a preparation for large readdir buffers in fuse.
> 
> Simply setting the fuse buffer size to the userspace buffer size should
> work, the record sizes are similar (fuse's is slightly larger than libc's,
> so no overflow should ever happen).
> 
> 
> [...]

Applied to the vfs-6.16.misc branch of the vfs/vfs.git tree.
Patches in the vfs-6.16.misc branch should appear in linux-next soon.

Please report any outstanding bugs that were missed during review in a
new review to the original patch series allowing us to drop it.

It's encouraged to provide Acked-bys and Reviewed-bys even though the
patch has now been applied. If possible patch trailers will be updated.

Note that commit hashes shown below are subject to change due to rebase,
trailer updates or similar. If in doubt, please check the listed branch.

tree:   https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git
branch: vfs-6.16.misc

[1/1] readdir: supply dir_context.count as readdir buffer size hint
      https://git.kernel.org/vfs/vfs/c/e0410e956b97