> On Wed, Jul 17, 2013 at 12:11, Sergey Bronnikov wrote: > > Bug was catched by fsfuzzer. Probably that bug cannot be > > found in real life with real usecase, but anyway it is a bug. > > Sorry, but I don't think these bugs are interesting. The filesystem > code panics when it encounters a broken filesystem. That is by design. > Attempting to continue given that we know the filesystem is > inconsistent would only make things worse.
I concur. If the filesystem is that broken, I want it to panic. If you read up on the history of how filesystems work, it is as Ted says -- by design. If you don't that design, the filesystem structures and code would be substantially different from what they are now.
