Re: [PATCH 3/3] btrfs: disguise single-data-RAID56 as RAID1/RAID1C3

On Mon, May 25, 2026 at 12:17:07PM +0200, David Sterba wrote:
> > It absolutely is.  Adding fast path workarounds for this is completely
> > stupid when it can be trivially handled on the mount side.
> 
> It's a lazy hack at best and fix on absolutely wrong layer. The library
> should provide the support for the edge case. But we disagree on that.

No, the library should not provide hacks for something it never intended
to support, and which is a bad idea.  It's also something the
implementation that the library was factored out from for btrfs use
never supported.

So adding the workaround to btrfs, which caused the problem is a
requirement.  Doing it to only create mount time overhead instead of
burdening the I/O path is a very smart way to solve the problem,
while adding hacks to the I/O fastpath is everything but smart.

But in the end this is for the btrfs maintainers to decide.