if_wg: Prefix crypto_init() & crypto_deinit() with wg_ for a cleaner namespace.

[bms]

Summary

Both these functions have non-static linkage for good reasons, however, their
naming may confuse folk when working with crypto(9) code at global scope.

[pouria]

I'm not sure whether they can be static or not, but the naming is definitely not right.

[bms]

In D58588#1344538, @pouria wrote:

I'm not sure whether they can be static or not, but the naming is definitely not right.

Thanks. I'd like to know what @jhb thinks as he did most of the heavy lifting towards the end of wg(4) landing, and so that there is no unneeded divergence from Jason Donenfeld's upstream.

[glebius]

Makes sense to wait for jhb@ for a few days.

[bms]

In D58588#1344732, @glebius wrote:

Makes sense to wait for jhb@ for a few days.

Technically sys/dev/wg is not declared in MAINTAINERS, nor is sys/crypto, but I'd be leveraging @jhb 's wg(4) work as an example for driving OCF in software mode. Sam Leffler firmed up the case for doing it in 2003, and this applies even now, despite kTLS being dependent upon kernel-space OpenSSL components; testing with the appropriate ConnectX models is out of my dev budget right now, for example.

There was expressed interest in TCP-AO by @jmg on behalf of his employer some months back. It appears they have TCP-AO in their EVO trains by way of being based on Linux, but not for the primary FreeBSD-based trains which ~30% of core Internet relies on.

I'd aim to land some infrastructure bits first, my continuing to pick up on the abandoned TCP-AO work from the early 2010s is a possibility, but I would need financial support for it to be viable for me to do this.

There's also the small matter of people happily writing blog posts about TCP-AO, yet not pointing to any production deployments, even at IXPs, and even right now.