Re: rwd0a and label sector marked read-only

[Jean-Yves Migeon <jean-yves.migeon%espci.fr@localhost>]

Izumi Tsutsui wrote:
> All partitions (except RAW_PART) are defined in the disklabel,
> so it might be problematic to define behavior when
> the disklabel is updated via a partition which is being changed. 
>
> > - why EROFS and not EPERM?
>
> Because it requires a proper procedure (DIOCWLABEL), not privilege?

Well, disklabel(8) mentions for -W option:

"Allow writes to the disk sector that contains the label."

For me "allow/disallow" is more tied to a permission than a read/write 
state, that's why I am asking. But I am no native speaker :)

Besides, returning EROFS might be misleading for the user, as it informs 
that the fs is read-only, which is partly correct here (having one 
sector marked read-only does not necessarily mean that the rest is in 
the same state).

Thanks for the answers.

--
Jean-Yves Migeon
jean-yves.migeon%espci.fr@localhost