Re: [PATCH 7/9] xfs: walk b_addr for buffer I/O

[Christoph Hellwig <hch@infradead.org>]

> Huh.  So ... where does b_offset come into play here?
> 
> OH.  Since we're starting with b_addr and working our way /back/ to
> folios, we don't need b_offset anymore since we can compute that from
> (b_addr - folio_address()).  So then the @offset variable in
> _xfs_buf_ioapply is really a cursor into how far into the xfs_buf we've
> ioapply'd.

Yes.

> Would you mind adding a sentence to the commit message?
> 
> "Instead of walking the folio array just walk the kernel virtual address
> in ->b_addr.  This prepares for using vmalloc for buffers and removing
> the b_folio array.  Furthermore, b_offset goes away since we can compute
> that from b_addr and the folio."

Fine with me.