[PATCH v3 1/6] fs: Add unimplemented rename hook to FS layer

[Heinrich Schuchardt]

On 2/7/25 13:45, Daniel Venzin wrote:
> The EFI API supports file renaming, which is required by
> systemd-stub for its bootcounting implementation.
>
> Signed-off-by: Daniel Venzin <Daniel.Venzin at mt.com>
>
> ---
>
> (no changes since v2)
>
> Changes in v2:
> - Adapt variable names for better consistency
> - Add missing rename hooks
>
>   fs/fs.c      | 30 ++++++++++++++++++++++++++++++
>   include/fs.h |  9 +++++++++
>   2 files changed, 39 insertions(+)
>
> diff --git a/fs/fs.c b/fs/fs.c
> index 99ddcc5e37b..b746d05ebcd 100644
> --- a/fs/fs.c
> +++ b/fs/fs.c
> @@ -143,6 +143,12 @@ static inline int fs_mkdir_unsupported(const char *dirname)
>   	return -1;
>   }
>
> +static inline int fs_rename_unsupported(const char *old_filename,
> +					 const char *new_filename)
> +{
> +	return -1;
> +}
> +
>   struct fstype_info {
>   	int fstype;
>   	char *name;
> @@ -182,6 +188,7 @@ struct fstype_info {
>   	void (*closedir)(struct fs_dir_stream *dirs);
>   	int (*unlink)(const char *filename);
>   	int (*mkdir)(const char *dirname);

The structure and its elements should be documented.

The expected return values should be described.

> +	int (*rename)(const char *old_filename, const char *new_filename);
>   	int (*ln)(const char *filename, const char *target);
>   };
>
> @@ -201,10 +208,12 @@ static struct fstype_info fstypes[] = {
>   		.write = file_fat_write,
>   		.unlink = fat_unlink,
>   		.mkdir = fat_mkdir,
> +		.rename = fs_rename_unsupported,
>   #else
>   		.write = fs_write_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   #endif
>   		.uuid = fat_uuid,
>   		.opendir = fat_opendir,
> @@ -238,6 +247,7 @@ static struct fstype_info fstypes[] = {
>   		.closedir = ext4fs_closedir,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   	},
>   #endif
>   #if IS_ENABLED(CONFIG_SANDBOX) && !IS_ENABLED(CONFIG_XPL_BUILD)
> @@ -256,6 +266,7 @@ static struct fstype_info fstypes[] = {
>   		.opendir = fs_opendir_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   		.ln = fs_ln_unsupported,
>   	},
>   #endif
> @@ -275,6 +286,7 @@ static struct fstype_info fstypes[] = {
>   		.opendir = fs_opendir_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   		.ln = fs_ln_unsupported,
>   	},
>   #endif
> @@ -295,6 +307,7 @@ static struct fstype_info fstypes[] = {
>   		.opendir = fs_opendir_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   		.ln = fs_ln_unsupported,
>   	},
>   #endif
> @@ -316,6 +329,7 @@ static struct fstype_info fstypes[] = {
>   		.opendir = fs_opendir_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   		.ln = fs_ln_unsupported,
>   	},
>   #endif
> @@ -339,6 +353,7 @@ static struct fstype_info fstypes[] = {
>   		.ln = fs_ln_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   	},
>   #endif
>   #if IS_ENABLED(CONFIG_FS_EROFS)
> @@ -360,6 +375,7 @@ static struct fstype_info fstypes[] = {
>   		.ln = fs_ln_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   	},
>   #endif
>   	{
> @@ -377,6 +393,7 @@ static struct fstype_info fstypes[] = {
>   		.opendir = fs_opendir_unsupported,
>   		.unlink = fs_unlink_unsupported,
>   		.mkdir = fs_mkdir_unsupported,
> +		.rename = fs_rename_unsupported,
>   		.ln = fs_ln_unsupported,
>   	},
>   };
> @@ -697,6 +714,19 @@ int fs_mkdir(const char *dirname)
>   	return ret;
>   }
>
> +int fs_rename(const char *old_filename, const char *new_filename)
> +{
> +	int ret;
> +
> +	struct fstype_info *info = fs_get_info(fs_type);
> +
> +	ret = info->rename(old_filename, new_filename);
> +
> +	fs_close();
> +
> +	return ret;
> +}
> +
>   int fs_ln(const char *fname, const char *target)
>   {
>   	struct fstype_info *info = fs_get_info(fs_type);
> diff --git a/include/fs.h b/include/fs.h
> index 2474880385d..8c1aa9c1a7b 100644
> --- a/include/fs.h
> +++ b/include/fs.h
> @@ -270,6 +270,15 @@ int fs_unlink(const char *filename);
>    */
>   int fs_mkdir(const char *filename);
>
> +/*

/**

> + * fs_rename - Rename a file

fs_rename()

Please, have a look at
https://docs.kernel.org/doc-guide/kernel-doc.html#function-documentation

Please, describe that set_blk_dev() must be called before invoking
fs_rename() and that fs_close() is called implicitly.

> + *
> + * @old_filename: Name of the file to rename
> + * @new_filename: New name for the file
> + * Return: 0 on success, -1 on error conditions

Other file system functions return a negative error code like -ENOENT,
-EIO, or -ENOMEM. Why do you require -1 here?

Best regards

Heinrich

> + */
> +int fs_rename(const char *old_filename, const char *new_filename);
> +
>   /*
>    * Common implementation for various filesystem commands, optionally limited
>    * to a specific filesystem type via the fstype parameter.