Re: [PATCH 4/6] hfsplus: add iomap operations for regular file data

On Wed, Aug 19, 2026 at 04:27:50PM -0700, Viacheslav Dubeyko wrote:
> Hi Christoph,
> 
> On Fri, 2026-07-24 at 07:32 +0200, Christoph Hellwig wrote:
> 
> 
> > 
> > > +static int hfsplus_file_write_dio_end_io(struct kiocb *iocb,
> > > ssize_t size,
> > > +					 int error, unsigned int
> > > flags)
> > > +{
> > > +	struct inode *inode = file_inode(iocb->ki_filp);
> > > +
> > > +	if (error)
> > > +		return error;
> > > +
> > > +	if (size && i_size_read(inode) < iocb->ki_pos + size) {
> > > +		i_size_write(inode, iocb->ki_pos + size);
> > > +		mark_inode_dirty(inode);
> > > +	}
> > > +
> > > +	return 0;
> > > +}
> > 
> > We've have a copy of this in various places, maye add it to a header?
> > Or even into the core code based on a flag so that these file systems
> > don't need an end_io method?
> 
> I think I need more clarification here. :) By "various places" do you
> mean HFS+ code or Linux kernel as whole?
> 
> I assume that you mean this code block:
> 
> +	if (size && i_size_read(inode) < iocb->ki_pos + size) {
> +		i_size_write(inode, iocb->ki_pos + size);
> +		mark_inode_dirty(inode);
> +	}
> 
> Do you suggest to introduce this logic as a special method that can be
> used in other file systems too? Am I right?

That was the original idea.  But I think what is even better is to
do it when no iomap_dio_ops.end_io method is defined.  I think I did
an audit that there is no such current case for writes, but we should
double check it.