The idea that you should avoid grsecurity because Linus Torvalds, the "BDFL", says so is so messed up I actually don't know how to rebut it.

Either way: I'm not saying Retguard is based on RAP --- my question was, "what's the relationship between the techniques". But if there is a relationship, OpenBSD should be explicit about it, so that we can keep track of the evolution of memory corruption countermeasures.

I do not care whether you think people should or shouldn't run grsecurity.

I don't avoid grsec because Linus says so. I avoid it because he says they break things when they don't need to. Security is a trade-off and I trust Linus in judging that and in ensuring the operating system should never break user space, unless it absolutely has to. If it does, I sure don't want to be stuck going to a group that goes against the spirit of the GPL (by punishing users who redistribute by cutting them off from future access) and which sues prominent members of the community essentially for pointing this out; which pretty much ensures I will be stuck going to them for support and not redistributing their work as I am permitted to under the GPL. No thanks to that kind of coercion!

I too am curious what the technical basis is for the patch but your assertion that drama is irrelevant is dangerous when applied to community projects that exist because of the goodwill of their members. grsec should be called out, when mentioned, because of their demonstrated ability to make the code they do produce less than useful because of the encumbrance it carries due to its origin.

None of this has anything to do with my question and I'll ask that you not use my comments as a coat rack to hang your unrelated concerns about grsecurity off of.

I'm asking a research question, not a user question.

Thanks.

I was responding to your comment that the concerns raised by GP were drama and irrelevant. They are relevant to this subthread. I've said what I wish to say on the matter as well.

Oh, please. If you want to throw stones about behavior in the Linux kernel you're going to need to throw them at a hell of a lot more people than Grsec - there's decades of shit piled up.

Linus has always been a total blowhard when it comes to... everything, but in particular when it comes to security. I wouldn't take his opinions too seriously on the matter.

The fact is that grsec still maintains the state of the art for memory safety mitigations.

grsec may maintain state of the art, but it doesn't matter if one can't use their code because of the potential for breakage and toxic licensing conditions under which one would have to use it should one choose to do so. I also would not want to support a group which,according to the OSI, violates the spirit of the licence upon which their state of the art work is based - and which certainly violates generally accepted community norms.