On Tue 06 Mar 2018 03:06:46 PM CET, Kevin Wolf wrote:
>> So although having a way to repair this kind of corruption would be
>> nice, we should still allow the user to try to open the image and
>> read data from it without requiring a potentially destructive
>> operation first.
> Generally speaking, there is also always the option to refuse opening
> a corrupted image read-write, but to allow opening it read-only. This
> would probably still require some checks on use, though (or maybe just
> hiding all snapshots).
You're right. The checks are in this series anyway.