Re: [PATCH 02/12] fs_parse: fix fs_param_v_optional handling

[Andrew Price <anprice@redhat.com>]

On 28/11/2019 15:59, Miklos Szeredi wrote:
> String options always have parameters, hence the check for optional
> parameter will never trigger.
> 
> Check for param type being a flag first (flag is the only type that does
> not have a parameter) and report "Missing value" if the parameter is
> mandatory.
> 
> Tested with gfs2's "quota" option, which is currently the only user of
> fs_param_v_optional.

It's not clear to me what the bug is here. My tests with the quota 
option are giving expected results. Perhaps I missed a case?

Andy

> Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
> Cc: Andrew Price <anprice@redhat.com>
> Cc: David Howells <dhowells@redhat.com>
> Fixes: 31d921c7fb96 ("vfs: Add configuration parser helpers")
> Cc: <stable@vger.kernel.org> # v5.4
> ---
>   fs/fs_parser.c | 10 ++++++----
>   1 file changed, 6 insertions(+), 4 deletions(-)
> 
> diff --git a/fs/fs_parser.c b/fs/fs_parser.c
> index d1930adce68d..5d8833d71b37 100644
> --- a/fs/fs_parser.c
> +++ b/fs/fs_parser.c
> @@ -127,13 +127,15 @@ int fs_parse(struct fs_context *fc,
>   	case fs_param_is_u64:
>   	case fs_param_is_enum:
>   	case fs_param_is_string:
> -		if (param->type != fs_value_is_string)
> -			goto bad_value;
> -		if (!result->has_value) {
> +		if (param->type == fs_value_is_flag) {
>   			if (p->flags & fs_param_v_optional)
>   				goto okay;
> -			goto bad_value;
> +
> +			return invalf(fc, "%s: Missing value for '%s'",
> +				      desc->name, param->key);
>   		}
> +		if (param->type != fs_value_is_string)
> +			goto bad_value;
>   		/* Fall through */
>   	default:
>   		break;
>