On Thu, Jan 30, 2020 at 10:22:30AM -0800, Christoph Hellwig wrote: > While this looks functionally correct, I think the structure in > here is weird. In libxfs we usually check that magic number first > and then branch out into helper that deal with the leaf vs node > format. That is we don't do the xfs_attr3_leaf_hdr_from_disk call > for node format attrs, and also check the forward/backward pointers > based on the actual ichdr. Maybe this code should follow that > structure? Yeah, I'll refactor the case blocks into two functions. --D
