Message ID | ZOxnTFhchkTvKpZV@gondor.apana.org.au |
---|---|
State | New |
Headers | show |
Series | [GIT,PULL] Crypto Update for 6.6 | expand |
The pull request you sent on Mon, 28 Aug 2023 17:22:20 +0800:
> git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6.git v6.6-p1
has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/68cf01760bc0891074e813b9bb06d2696cac1c01
Thank you!
On Wed, 1 Nov 2023 at 20:56, Herbert Xu <herbert@gondor.apana.org.au> wrote: > > Stephan Müller (5): > crypto: jitter - add RCT/APT support for different OSRs > crypto: jitter - Allow configuration of memory size > crypto: jitter - Allow configuration of oversampling rate > crypto: jitter - reuse allocated entropy collector > crypto: jitter - use permanent health test storage This is beyond annoying. These are adding Kconfig questions that don't make sense. The whole jitter thing is debatably useful in the first place, and now you just annoy users with random questions. And I mean truly random - the whole jitter entropy is voodoo programming to begin with, and having some crazy 8MB buffer for it is just ridiculous. Honestly, this all smells like somebody's PhD thesis, not a real life thing. And no, we don't make our Kconfig questions more annoying for some PhD thesis. We also don't ask people questions that don't have valid answers. Just because the whole "what is entropy in the first place" isn't clear-cut, we don't then punt some tweaking question to the user. We have a very simple and stupid jitter entropy thing AT BOOT TIME just to try to generate some amount of entropy to make boots non-repeatable (see "try_to_generate_entropy()" in drivers/char/random.c). Honestly, the whole crypto layer one is ridiculous overkill in the first place, but the annoying new questions have now literally made me consider just removing it entirely. Because no, IT IS NOT OK TO ASK CRAZY QUESTIONS. If some developer cannot come up with a reasonable answer, a random user sure has hell cannot. And no, any question that says "do you want to use 8MB of memory for jitter entropy" is just batsh*t crazy. This kind of crap needs to stop. If somebody wants to do this kind of thing, just do it in user space. It's ridiculously pointless in the kernel. Convince me I'm wrong. But there is no way in *hell* you will convince me that we should ask users about some jitter memory sizing. Allocating memory for timing analysis is silly to begin with, since any kernel thing could just use the physical memory mapping we already have in the kernel. I suspect strongly that all this code has been influenced by code running in user space, where it belongs, and where you do need to allocate memory to have it available. Please just make this noise go away. Linus
On Thu, 2 Nov 2023 at 19:52, Herbert Xu <herbert@gondor.apana.org.au> wrote: > > Fair enough. How about adding an EXPERT dependency on this? I think that would help the situation, but I assume the sizing for the jitter buffer is at least partly due to trying to account for cache sizing or similar issues? Which really means that I assume any static compile-time answer to that question is always wrong - whether you are an expert or not. Unless you are just building the thing for one particular machine. So I do think the problem is deeper than "this is a question only for experts". I definitely don't think you should ask a regular user (or even a distro kernel package manager). I suspect it's likely that the question is just wrong in general - because any particular one buffer size for any number of machines simply cannot be the right answer. I realize that the commit says "*allow* for configuration of memory size", but I really question the whole approach. But yes - hiding these questions from any reasonable normal user is at least a good first step. Linus