On 9/14/18 8:46 AM, Christoph Hellwig wrote: > On Wed, Sep 12, 2018 at 08:55:35PM -0500, Eric Sandeen wrote: >> The commit: >> >> db1327b1 xfs: report shared extent mappings to userspace correctly >> >> disabled FIBMAP on reflinked files, with this very clear rationale >> regarding swapfiles: >> >> "Have xfs_vm_bmap return zero for reflinked files because the >> bmap-based swap code requires static block mappings, which is >> incompatible with copy on write." >> >> However, this broke the FIBMAP interface for existing userspace >> apps, including bootloaders. > > It didn't break any interface because the reflink flag has been > experimental and a mkfs-time opt in. > > So NAK for this patch. That's a really weak argument, but I guess it shouldn't be surprising because this whole episode has been a string of weak arguments.
