s2n VS mbedTLS

It seems to support multiple options but requires you pick at least one of them. https://github.com/aws/s2n-tls/blob/main/docs/BUILD.md#build...

I think GnuTLS is probably the second most popular TLS library, after openssl.

I'll also mentions s2n and rustls-ffi for completeness as C libraries, though the former isn't widely used, and the latter is very experimental still. https://github.com/aws/s2n-tls and https://github.com/rustls/rustls-ffi respectively.

I've seen operating on unauthenticated plaintext enough times to list it as my own pet peeve with AES-GCM. But it's a problem for chunked messages too. A few years ago we released a SCRAM mode that makes very minimal changes to AES-GCM so that it mathematically can't operate on unauthenticated plaintext. https://github.com/aws/s2n-tls/tree/main/scram

> The documentation is kind of vague, but apparently you have to re-enable it regularly.[3]

This is correct. And in the end it means more or less that setting the socket option is more of a way of sending an explicit ACK from userspace than a real setting.

It's not great for common use-cases, because making userspace care about ACKs will obviously degrade efficiency (more syscalls).

However it can make sense for some use-cases. E.g. I saw the s2n TLS library using QUICKACK to avoid the TLS handshake being stuck [1]. Maybe also worthwhile to be set in some specific RPC scenarios where the server might not immediately send a response on receiving the request, and where the client could send additional frames (e.g. gRPC client side streaming, or in pipelined HTTP requests if the server would really process those in parallel and not just let them sit in socket buffers).

[1] https://github.com/aws/s2n-tls/blob/46c47a71e637cabc312ce843...

It looks like by default s2n-quic uses this TLS implementation, which is not based on the ring crate (though it is written in C)

https://github.com/aws/s2n-tls

I would be interested in the other SSL implementations:

- https://github.com/awslabs/s2n

- https://boringssl.googlesource.com/boringssl

- https://bearssl.org/

Are these subpar implementations or there are other reasons not to use these?

It's work in progress: https://github.com/Mbed-TLS/mbedtls/blob/development/docs/ar...

Newer version have okay-ish support, I'd guess the next OpenWRT release will have it again.

It's related to this Mbed-TLS issue

i've used the entropy generator from MbedTLS which uses multiple sources of randomness, and seeding it additionally with the chip's mac address and the free heap size at time of creation. it would take a lot to recreate all that "noise" consistently, but i'm open to suggestions on how to make it more secure

However on the github page and assuming I found the right source - and this is a HUGE ASSUMPTION - there is a section on porting the library to other platforms.

Rebean is a single executable written in ANSI C that embeds Lua 5.4, MbedTLS and SQLite. The crazy thing about this is that it is a C binary that runs on six operating systems without any cross compiling thanks to leveraging Justine Tunney's cosmopolitan.

it looks like the commands are all built under https://github.com/Mbed-TLS/mbedtls/tree/development/programs and the arch mbedtls package does some extra work to rename them and copy them into /usr/local/bin: https://github.com/archlinux/svntogit-community/blob/packages/mbedtls/trunk/PKGBUILD#L55

If you want to see a real implemention of arbitrary sized integer math, mbedTLS is a great example:

https://github.com/ARMmbed/mbedtls/blob/development/library/...

All of the ECC code in that library relies on this code, which can be accelerated by dedicated hardware.

mbed TLS - Another crypto implementation. Apache-2.0

Definitely, FreeRTOS + LwIP + Mbed TLS is a fairly common combo these days, and Mbed TLS has hardware acceleration where possible.