Simple question for comment in btrfs-progs qgroup code

Simple question for comment in btrfs-progs qgroup code

[Sidong Yang]

Hello!

I see an comment in qgroup cmds below.

diff --git a/cmds/qgroup.c b/cmds/qgroup.c
index 2da83ffd..a71089e9 100644
--- a/cmds/qgroup.c
+++ b/cmds/qgroup.c
@@ -84,6 +84,7 @@ static int _cmd_qgroup_assign(const struct cmd_struct
*cmd, int assign,
	/*
	* FIXME src should accept subvol path
	*/

This comment is little confusing for me. src was parsed with
parse_qgroupid() and it also accepts subvol path. is it outdated
comment? or src should accept only lowest level of qgroup?

Thanks,
Sidong

Re: Simple question for comment in btrfs-progs qgroup code

[Qu Wenruo]

On 2021/3/21 下午7:33, Sidong Yang wrote:
> Hello!
>
> I see an comment in qgroup cmds below.
>
> diff --git a/cmds/qgroup.c b/cmds/qgroup.c
> index 2da83ffd..a71089e9 100644
> --- a/cmds/qgroup.c
> +++ b/cmds/qgroup.c
> @@ -84,6 +84,7 @@ static int _cmd_qgroup_assign(const struct cmd_struct
> *cmd, int assign,
> 	/*
> 	* FIXME src should accept subvol path
> 	*/
>
> This comment is little confusing for me. src was parsed with
> parse_qgroupid() and it also accepts subvol path. is it outdated
> comment? or src should accept only lowest level of qgroup?

It's outdated and confusing.

For the assign ioctl, it only accepts 3 u64 parameters,
assign:bool to indicate if it's adding or deleting qgroup relationship.
src/dst to specify the qgroupid (u64).

There is no room for path at all.

The qgroupid resolve is done completely in user space.

Feel free to delete the comment.

>
> Thanks,
> Sidong
>