Take Monocypher or TweetNaCL, and rip off any primitive you don't need. This may be enough. Even when you take the whole thing, Monocypher only needs 30kb of x86-64 machine code when compiled with -Os. If you only keep authenticated encryption and x25519, which are enough for many uses, I think you should be able to halve that down to 15kb or less.

If speed doesn't matter, TweetNaCl is even smaller.

If those aren't enough still, you may want to dive in, learn a ton about crypto (starting with https://www.crypto101.io/), and investigate the sponge construction, whose versatility may allow you to shrink the code even more. Perhaps. I'm in over my head at this point.

avr has aes extensions, surely arm cortex-m has similar. are those any good, are they used by the libs you mention?

Neither of those libraries uses any extensions. TweetNaCl is a famously minimal implementation of NaCl in portable C by the NaCl authors. Monocypher is a library 'loup-vaillant wrote as a sort of side project. TweetNaCl does see some use, but libsodium is the most popular NaCl implementation, and libsodium's author provides his own low-footprint misuse-resistant crypto API separately.

Neither TweetNaCl nor 'loup-vaillant's library really addresses the low-footprint concern.

If they have hardware AES support, forget everything I said. Monocypher and TweetNacl are portable C, they don't use extensions of any kind.

When you're that small it really requires more context. What are you trying to do?

I've deployed TLS in 50 KiB flash and 6 Kb RAM. But I could rely on hardware support for some public key operations and could pick a single ciphersuite, and controlled the server end as well.

The boring answer is that you'll need to identify the set of primitives you need and then find a library that matches. I can't think of any single primitive that would break that budget, so it'll depend on how many you need!

You need to provide more info.

  1. How much of your flash budget is eaten up by the application?

  2. What performance do you need for different use cases and operations?

Generally, RAM is often the least of your problems. And if you are into Cortex M0 or even M0+ territory, your MCU will quite probably sport an AES-128 core. Even if it is slow you should try and use that.

libhydrogen might be an option.