bmap-tools

> But I recall reading elsewhere a discussion about some userspace program which did depend on holes being present in the filesystem as actual holes (visible to SEEK_HOLE and so on) and not as runs of zeros. "treatment of on-disk segments as "what was written by programs" can cause areas of 0 to not be written by bmaptool copy": https://github.com/intel/bmap-tools/issues/75 IMO, the issue here isn't filesystem or zfs behavior, it's that bmap-tool wants an extra "don't care bit" per block, which filesystems (traditionally) don't track, and programs interacting with filesystem don't expect to exist. Some of the comments I've made in this issue describe options to make things better. (FWIW: the original hn link discusses a different issue around seek hole/data, and the bmap-tool issue is backwards from the issue the parent posits: bmap-tool

(>_<) Oh man, I knew about [0] when I posted (which is why I said it just reduces the chance of hitting the bug (by a lot)). But after spending all Saturday JST on it, I went to bed before [1] was posted. Skimming through #6958 though, it seems like it's the lesser of evils, compared to #15526... I think? It's less obvious (to me) what the impact of #6958 is. Is it silent undetectable corruption of your precious data potentially over years, or more likely to cause a crash or runtime error? Reports like https://github.com/intel/bmap-tools/issues/65 make it seem more like the latter. But I have to read more. But since the zfs_dmu_offset_next_sync setting was disabled by default until recently, I still suspect (but yeah, don't know for sure) that disabling is the safest thing we can currently do on unmodified ZFS systems.