That Tk app is cool. ++. I'm inspired to learn Tk now.
I don't want to let you get away with speaking ex cathedra, so... which experts and where? :)
Probably should have mentioned earlier that I'm familiar with Linux (this is not to say your answer wasn't helpful). As far as I know, you can do better than using losetup's encryption facilities. According to this paper (PDF), CBC mode (used by - correct me if wrong - losetup) has a few known problems. Whether these are more than academic is likely dependent on whom you're trying to hide data from. :) In any case (as I perceive it), the state of the art with Linux disk encryption right now is LUKS and dm-crypt.
Nevertheless, I think you're right... encrypted loopback is the best solution. The "far more complex" situation means that
memset(sensitive_buf, 0, sizeof sensitive_buf);... is not sufficient, because to assert it is would be assuming the code is running on some ethereal Turing machine in the sky. But it's not. You must also consider the broader environment this code runs in (namely, one which includes modern OSs with virtual memory systems), which means taking into account memory being swapped out, etc.
I think I have more of a handle on the problem now. Cheers. :)
In reply to Re^4: Is it possible to sanitize Perl memory that holds sensitive data? (crypto implications)
by missingthepoint
in thread Is it possible to sanitize Perl memory that holds sensitive data? (crypto implications)
by missingthepoint
| For: | Use: | ||
| & | & | ||
| < | < | ||
| > | > | ||
| [ | [ | ||
| ] | ] |