Protocol layers and other parts refactoring (+ nits)

Nit: if I read it right, the proxyv2 commit also removes PROTOLAYER_PAYLOAD_BUFFER support from tcp unwrap. That might be worth noting in the commit message, as it's hard to see due to the code being moved.

I've passed through all of the commit diffs and noticed nothing wrong. But some kinds of errors could be missed. Really lots of lines have changed.

changed title from Nits to Protocol layers and other parts refactoring (+ nits)

added 1 commit

bda116f9 - amend! daemon/session2: support multiple short-lived buffers

Compare with previous version

added 1 commit

d7d8fead - amend! daemon/proxyv2: move PROXY protocol into its own layer

Compare with previous version

added 1 commit

b69ead10 - amend! daemon/proxyv2: move PROXY protocol into its own layer

Compare with previous version

added 2 commits

2a757895 - .clang-tidy: add quotation marks to avoid YAML parse errors
9a9ce1f1 - Some more Clang-Tidy silencing

Compare with previous version

added 1 commit

ab343f1f - Some more Clang-Tidy silencing

Compare with previous version

added 1 commit

0f8f0be6 - Some more Clang-Tidy silencing

Compare with previous version

added 1 commit

99009219 - .gitlab-ci: use Arch for linting

Compare with previous version

marked this merge request as draft

added 18 commits

99009219...6241c371 - 7 commits from branch master
080d05a6 - 1 earlier commit
52a767c6 - daemon/http: fix typo
d50ba83b - daemon/io: remove minor inefficiency
bfe6b008 - daemon/session2: documentation improvements
6483ce1a - daemon, lib: unify protolayer_grp and kr_proto enums
8fa91a73 - daemon/session2: protocol layer refactors + docs
195736cb - daemon/proxyv2: move PROXY protocol into its own layer
9c43619d - daemon/session2: optimize allocations
9465e1ef - Some more Clang-Tidy silencing
c8a51d1c - daemon/session2: fix asynchronous layer iterations
b89bb8e5 - .gitlab-ci: use Arch for linting

Compare with previous version

added 1 commit

d694af29 - .gitlab-ci: use Arch for linting

Compare with previous version

added 1 commit

5c59f588 - .gitlab-ci: use Arch for linting

Compare with previous version

added 1 commit

4bd8bc19 - .gitlab-ci: use Arch for linting

Compare with previous version

added 1 commit

9df6b23e - .gitlab-ci: use Arch for linting

Compare with previous version

mentioned in merge request knot-resolver-ci!1 (merged)

added 1 commit

94ee9bef - .gitlab-ci: use Arch for linting

Compare with previous version

added 1 commit

7e68b12f - .gitlab-ci: use Arch for linting

Compare with previous version

resolved all threads

Alright. A few more things happened here since the last comment - the more serious ones are in new commits. Some commits were fixed-up, because there were typos (namely, I for some reason managed to use mm_ctx_init() which uses malloc/free instead of the mm_ctx_mempool(), which makes an actual memory pool - that would be some ugly memory leaks).

There were a few bugs in protolayers pointed out to me by @lukas.ondracek, so I (hopefully) fixed them. I'm going to do a final rebase of his commits from rrl-wip-broken and put that into rrl-wip-unbroken, resolving all conflicts, once this is merged. If there are further bugs, I'll address them separately.

So, I believe this is just waiting for the merge of knot-resolver-ci!1 (merged), plus a tag in that repo.