Re: [PATCH v2 3/6] fs/ntfs3: Use new api for mounting

[Kari Argillander <kari.argillander@xxxxxxxxx>]

On Thu, Aug 19, 2021 at 03:26:30AM +0300, Kari Argillander wrote:
> We have now new mount api as described in Documentation/filesystems. We
> should use it as it gives us some benefits which are desribed here
> lore.kernel.org/linux-fsdevel/159646178122.1784947.11705396571718464082.stgit@xxxxxxxxxxxxxxxxxxxxxx/
> 
> Nls loading is changed a to load with string. This did make code also
> little cleaner.
> 
> Also try to use fsparam_flag_no as much as possible. This is just nice
> little touch and is not mandatory but it should not make any harm. It
> is just convenient that we can use example acl/noacl mount options.
> 
> Signed-off-by: Kari Argillander <kari.argillander@xxxxxxxxx>

I will send new patches when Komarov has take a look of these. I have
found some bugs and how to improve things. Please take a look below my
second comment.

>  fs/ntfs3/super.c   | 392 +++++++++++++++++++++++----------------------

> +static void ntfs_fs_free(struct fs_context *fc)
> +{
> +	struct ntfs_sb_info *sbi = fc->s_fs_info;
> +
> +	if (sbi)
> +		put_ntfs(sbi);
> +}
> +
> +static const struct fs_context_operations ntfs_context_ops = {
> +	.parse_param	= ntfs_fs_parse_param,
> +	.get_tree	= ntfs_fs_get_tree,
> +	.reconfigure	= ntfs_fs_reconfigure,
> +	.free		= ntfs_fs_free,
> +};
> +
> +static int ntfs_init_fs_context(struct fs_context *fc)
>  {
> -	return mount_bdev(fs_type, flags, dev_name, data, ntfs_fill_super);
> +	struct ntfs_sb_info *sbi;
> +
> +	sbi = ntfs_zalloc(sizeof(struct ntfs_sb_info));
> +	if (!sbi)
> +		return -ENOMEM;
> +
> +	/* Default options */
> +	sbi->options.fs_uid = current_uid();
> +	sbi->options.fs_gid = current_gid();
> +	sbi->options.fs_fmask_inv = ~current_umask();
> +	sbi->options.fs_dmask_inv = ~current_umask();
> +
> +	fc->s_fs_info = sbi;
> +	fc->ops = &ntfs_context_ops;
> +
> +	return 0;
>  }
 
In this code I did not like that we make whole new sbi everytime.
Especially because we do zalloc. So when we regonfigure then new sbi
is allocated just for options. I notice that example xfs does allocate
their "sbi" everytime. Then I notice that example squashfs allocate
just mount options.

I have impression that we should allocate sbi if it first time so I
did "between code". I would like to do things like they are intended
with api so can Christoph comment that is this "right" thing to do
and is there any draw backs which I should know.

static void ntfs_fs_free(struct fs_context *fc)
{
	struct ntfs_mount_options *opts = fc->fs_private;
	struct ntfs_sb_info *sbi = fc->s_fs_info;

	if (sbi)
		put_ntfs(sbi);

	if (opts)
		clear_mount_options(opts);
}

static int ntfs_init_fs_context(struct fs_context *fc)
{
	struct ntfs_mount_options *opts;
	struct ntfs_sb_info *sbi = NULL;

	fc->ops = &ntfs_context_ops;

	opts = ntfs_zalloc(sizeof(struct ntfs_mount_options));
	if (!opts)
		return -ENOMEM;

	/* Default options. */
	opts->fs_uid = current_uid();
	opts->fs_gid = current_gid();
	opts->fs_fmask_inv = ~current_umask();
	opts->fs_dmask_inv = ~current_umask();

	fc->fs_private = opts;

	/* No need to initialize sbi if we just reconf. */
	if (fc->purpose == FS_CONTEXT_FOR_RECONFIGURE)
		return 0;

	sbi = ntfs_zalloc(sizeof(struct ntfs_sb_info));
	if (!sbi) {
		ntfs_free(opts);
		return -ENOMEM;
	}

	mutex_init(&sbi->compress.mtx_lznt);
#ifdef CONFIG_NTFS3_LZX_XPRESS
	mutex_init(&sbi->compress.mtx_xpress);
	mutex_init(&sbi->compress.mtx_lzx);
#endif

	sbi->options = opts;
	fc->s_fs_info = sbi;

	return 0;
}