Re: [PATCH 2/4] xfs: factor out xfs_da3_node_entry_remove

["Darrick J. Wong" <djwong@kernel.org>]

On Tue, Mar 10, 2026 at 07:58:23PM +0800, Long Li wrote:
> On Mon, Mar 09, 2026 at 09:42:29AM -0700, Darrick J. Wong wrote:
> > 
> > > +	__xfs_da3_node_remove(state->args->trans, state->args->dp,
> > > +			state->args->geo, drop_blk);
> > > +}
> > > +
> > > +/*
> > > + * Remove an entry from a node at the specified index, this is an exported
> > > + * wrapper for removing entries from intermediate nodes.
> > > + */
> > > +void
> > > +xfs_da3_node_entry_remove(
> > 
> > This only applies to attr(ibute) structures, as evidenced by m_attr_geo
> > below.  I think this ought to be named xfs_attr3_node_entry_remove.
> > 
> 
> Considering that xfs_da_btree.c implements functions related to xfs_da*, 
> it might not be appropriate to place xfs_attr3_node_entry_remove here either.
> I think we could add an additional `struct xfs_da_geometry *geo` parameter
> to xfs_da3_node_entry_remove() and let it be specified externally, which
> would increase the function's generality. What do you think?

Why does it need to be general?  Don't go adding code for a user
(directories) that doesn't exist.

--D

> 
> Thanks,
> Long Li
>