cannon: Remove support for singlethreaded and 32-bit multithreaded versions

[ajsutton]

Description

Governance has now approved 64 bit cannon so it should be used for all deployments going forward and we can remove support for the single threaded and 32-bit multithreaded versions. This is a pretty brute force removal - all the build tags are removed (though things like the Word alias are preserved). Lots of things still have 64 and multithreaded in the name which isn't really necessary anymore. We can choose if we want to do all those renames in follow up PRs.

Builds on #15487

Tests

Removed lots.

Metadata
Part of #13447

[codecov]

Codecov Report

Attention: Patch coverage is 56.25000% with 7 lines in your changes missing coverage. Please review.

Project coverage is 45.95%. Comparing base (24080ea) to head (aa00e6a).
Report is 1 commits behind head on develop.

Files with missing lines	Patch %	Lines
cannon/mipsevm/testutil/evm.go	20.00%	3 Missing and 1 partial ⚠️
cannon/cmd/load_elf.go	0.00%	1 Missing ⚠️
cannon/mipsevm/tests/helpers.go	50.00%	1 Missing ⚠️
cannon/mipsevm/versions/state.go	66.66%	1 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop   #15443      +/-   ##
===========================================
- Coverage    46.76%   45.95%   -0.81%     
===========================================
  Files         1286     1279       -7     
  Lines       106661   105478    -1183     
===========================================
- Hits         49875    48474    -1401     
- Misses       53262    53515     +253     
+ Partials      3524     3489      -35     

Flag	Coverage Δ
cannon-go-tests-32	?
cannon-go-tests-64	64.87% <56.25%> (+7.31%)	⬆️
contracts-bedrock-tests	86.74% <ø> (-7.55%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
cannon/mipsevm/arch/arch64.go	100.00% <ø> (ø)
cannon/mipsevm/program/patch.go	93.75% <ø> (+26.58%)	⬆️
cannon/mipsevm/testutil/arch.go	65.71% <ø> (-6.38%)	⬇️
cannon/mipsevm/testutil/elf.go	100.00% <100.00%> (+19.04%)	⬆️
cannon/mipsevm/testutil/vmtests.go	100.00% <100.00%> (+9.00%)	⬆️
cannon/mipsevm/versions/version.go	80.00% <ø> (-1.49%)	⬇️
cannon/cmd/load_elf.go	0.00% <0.00%> (ø)
cannon/mipsevm/tests/helpers.go	81.66% <50.00%> (-2.65%)	⬇️
cannon/mipsevm/versions/state.go	52.63% <66.66%> (-8.37%)	⬇️
cannon/mipsevm/testutil/evm.go	69.49% <20.00%> (-1.10%)	⬇️

... and 19 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[ajsutton]

@mbaxter I think I've come unstuck on this one. It's failing in the contract tests (https://app.circleci.com/pipelines/github/ethereum-optimism/optimism/87722/workflows/9d9cfc81-5c44-4ee6-953e-14234977d6b0/jobs/3455600) because of a memory alignment problem. I think that's because it's running MIPS2.t.sol which is the 32-bit cannon implementation but the FFI calls are winding up calling into 64-bit cannon code now (whereas previously it was 32-bit). I'm not really sure why they only started failing now - maybe because I rebased...

Any chance you could give me some hints? We could extend this to remove the 32-bit solidity implementations as well but we don't have a MIPS64.t.sol so not sure if we need to migrate some more solidity tests over to 64 bit to maintain coverage.

[mbaxter]

I think that's because it's running MIPS2.t.sol which is the 32-bit cannon implementation but the FFI calls are winding up calling into 64-bit cannon code now (whereas previously it was 32-bit).

Yeah - looks like the go-ffi script depends on the cannon64 build tags. Since we've removed the build tags, the 32-bit tests are failing because they are now using the 64-bit logic. I think we didn't see these earlier because the bedrock-go-tests were blocked because of other failing tests.

Looks like we will have to go ahead and remove the 32-bit solidity MIPS tests, which probably mean we should just go ahead and cut out the 32-bit contracts here as well.

we don't have a MIPS64.t.sol so not sure if we need to migrate some more solidity tests

We moved away from solidity tests because they're hard to work with and provide less coverage (no way to do differential testing in solidity). These tests have been implemented as differential tests in cannon/mipsevm/tests. So, there's no need to migrate anything else - everything should be covered there.

[ajsutton]

Ok, got this passing everything by deleting the 32-bit solidity implementations and their tests as well and changing mipsVersion from 1 to 6 in about a bajillion places. We're going to need to follow up with some way to avoid having such an insane number of places hard code the latest mips version, but I think this PR is already way too big. The test failures show very clearly what needs to be updated when the state version updates, it's just time consuming to go around them all.

[ajsutton]

The failing tests are all ones that connect to an external node and something seems to have changed with that node causing failures. Getting similar failures on develop as well: https://app.circleci.com/pipelines/github/ethereum-optimism/optimism/87917/workflows/b9c2ee0b-e389-4d01-92ca-8513fea99541/jobs/3462290

[mbaxter]

lgtm - if it's helpful, I rebased and pushed a few more small fixes here.

[axelKingsley]

LGTM, I focused on the circle ci file, as the rest has had attention from the proofs team. Left one question but otherwise G2G.