On 3/23/25 10:51 AM, Eric Sandeen wrote: ... > But I wonder - why not call longform_dir2_entry_check_data() before we check > the header? That way it /will/ populate hashtab with any found entries in the > block, and when the header is found to be corrupt, it will rebuild it with all > entries intact, and leave nothing in lost+found. Or alternatively, since checking the header first makes intuitive sense, perhaps change the logic so that we still check the header first, set fixit++, and if we are in XFS_DIR2_FMT_BLOCK, allow longform_dir2_entry_check_data() to be called, and then if errors were found in the header, goto out_fix; at that point, with a populated hashtab. -Eric > > Thoughts? > -Eric > > > >
