On Wed, Dec 09, 2020 at 05:45:27PM -0800, harshad shirwadkar wrote: > I see that makes sense. In that case, I'll rename the function to > errcode_t ext2fs_decode_extent(struct ext2fs_extent *dst, void *src). > I wonder if it's okay if we make this function return an error in case > the on-disk format is not sane. If we do that, we can add > ext2fs_validate_extent() later. Does that make sense? Sure, that works for me. Something that you should think about at some point is how much impact would be supporting an alternate on-disk extent node structure (for the leaf and/or intermediate nodes) have on Fast Commit? Obviously doing this would a new an INCOMPAT feature at the file system level, so we probably won't need any additional version negotiation in the fast commit journal header itself, but how many tags would need to be changed if we were to extend the extent tree structure sometime in the future? Cheers, - Ted
