Cost values are unsigned int, perhaps should be uint64_t

[magnumripper]

The ZIP format, for example, has the HMAC data size as a cost but it supports sizes beyond 32-bit, which makes the reporting very confusing.

The actual cost here is 4,294,968,320:

Loaded 1 password hash (ZIP-opencl, WinZip [PBKDF2-SHA1 OpenCL])
Cost 1 (HMAC size) is 1024 for all loaded hashes

[magnumripper]

We could of course scale it in the output, so it reports KB. Maybe that's the way for now?

[magnumripper]

We could of course scale it in the output, so it reports KB. Maybe that's the way for now?

That will however make it show as 0 for many archives, which is also confusing.

[magnumripper]

An alternative hack would be to have the ZIP cost function return UINT_MAX for such (unusal) costs. Not ideal but better than nothing.

[solardiz]

We could of course scale it in the output, so it reports KB. Maybe that's the way for now?

That will however make it show as 0 for many archives, which is also confusing.

We could round always up. So make it the number of full or partial KiB.

An alternative hack would be to have the ZIP cost function return UINT_MAX for such (unusal) costs. Not ideal but better than nothing.

Yes, and we'd need something like this for above 4 TiB even with the first hack.