hellyberry

+1

Such a subtree-detection would be a tremendeous time-saving and also a security-feature for efficiently handling backups of backups(-of-b....), a frequent use case.

And the good thing, there's ready-to-use parallelized code for SMP systems (on github), so running 4/8 threads is possible. This should speed up quite a bit. > QUOTE: BLAKE2 is...

I see. So fdupes and jdupes work same in this regard, but are different in the hashing-algorithm used. What do you mean by exclusionary comparison? Is first the file-size checked,...

You cannot safely detect modification by looking at the modification time. This is easily reset/modified ("touched") in filesystem.

So such a "feature" is contrary to fdupes aim - detecting securely and fast any real duplicates.

Maybe the BLAKE2-algorithm solves both needs in software. It is 256 bit in size, cryptographically secure, so low chance of 'false' collisions and relatively high speed. Maybe for examining MD5...

Most former OpenWrt developers switched to LEDE by now.

Is there any fuzzer for ARM?

What would be basically needed to be changed to support ARM ( e.g. Raspberry Pi)? different instruction sets/prefix-lengths?