> -----Original Message----- > From: Eric Biggers <ebiggers@xxxxxxxxxx> > Sent: Saturday, August 10, 2019 12:05 AM > To: Pascal Van Leeuwen <pvanleeuwen@xxxxxxxxxxxxxx> > Cc: linux-crypto@xxxxxxxxxxxxxxx > Subject: Re: [RFC PATCH v2] md/dm-crypt - reuse eboiv skcipher for IV generation > > On Fri, Aug 09, 2019 at 09:33:14PM +0000, Pascal Van Leeuwen wrote: > > Real life designs require all kinds of trade-offs and compromises. > > If you want to make something twice as expensive, you'd better have a > > really solid reason for doing so. So yes, I do believe it is useful to > > be sceptical and question these things. But I always listen to good > > arguments, so just convince me I got it wrong *for my particular use > > case* (I'm not generally interested in the generic case). > > Or rather, if you want to take shortcuts and incorrectly implement a crypto > construction, you'd better have a really solid reason for doing so. > Sure. It's called cost. Or power consumption. Or performance. Or any other reason that's often considered important, possibly even (much) more important than security. Welcome to the real world. The best possible security is rarely affordable. And a weaker implementation is not necessarily incorrect by the way, as long as it still meets your security goals under a relevant and realistic attack model. As an architect, I don't set the requirements or the priorities. I just try to make the best possible compromise. And fortunately, I do have a real cryptographer running around here somewhere, looking over my shoulder to ensure I don't do anything really stupid ;-) But I make him work for his money. (So no, I did not myself create any XTS solution using a single key, in case you were wondering. I just know it has been done.) But, yes, that's why a lot of bad crypto and other security vulnerabilities exist. And, newsflash, it's not going to change any day soon. > It's on you to show that your crypto is okay, not me. > Reality check: unless you work for some government (been there, done that), usually, nobody cares (until the shit hits the fan) > - Eric Regards, Pascal van Leeuwen Silicon IP Architect, Multi-Protocol Engines @ Verimatrix www.insidesecure.com