Re: [PATCH v7 04/13] btrfs: add support for inserting raid stripe extents

["hch@infradead.org" <hch@infradead.org>]

On Fri, Mar 03, 2023 at 11:15:00AM +0000, Johannes Thumshirn wrote:
> There's two possibilities how to handle it:
> 1) Have a common workfn that handles all the calls in the correct order
> 2) Do the RST update in btrfs_finish_ordered_io()
> 
> To me both are valuable options, so I don't care. Both need a bit of
> preparation work before, but that's the nature of the beast.
> 
> For 2) we need a pointer to the bioc in ordered_extent, so we need to
> make sure the lifetimes are in sync. Or the other way around, have
> ordered_stripe hold enough information for the RST updates and the
> end_io handler insert it in the ordered_stripe (that needs to be
> passed into the bioc or bbio).
> 
> *Iff* I interpret Christoph's proposal in [1] correctly, options 1) is
> easier to implement.

1 is probably easier, and should be done for other reasons.  But 2
really feels like the right thing to do in addition to 1.