Adding support for MMTk (non-moving immix)

[udesou]

This PR adds the possibility of building/running Julia using MMTk, running non-moving immix.
The binding code (Rust) associated with it is in https://github.com/mmtk/mmtk-julia/tree/upstream-ready/immix. Instructions on how to build/run Julia with MMTk are described in the README file inside the binding repo.

[d-netto]

First review pass.

Will follow-up with more comments soon.

[maleadt]

In any case, I had to restart PkgEval, so re-triggering:

@nanosoldier runtests()

[nanosoldier]

The package evaluation job you requested has completed - possible new issues were detected.
The full report is available.

[NHDaly]

run a PkgEval on Julia built with MMTk.

From discussion: We should not block this PR on a PkgEval run w/ MMTk. If we are happy w/ the PkgEval results from this PR with stock GC, this should be ready to merge.

This PR is just a refactoring and introducing a new GC behind a flag. It could be that literally nothing works with the MMTk GC, and that's fine - we will iterate in future PRs. This PR is just to get the new GC option in, and unit tests passing, to keep it stable.

I'm still interested in doing that PkgEval with MMTk, but IMO that should be a follow-up step (which you could drive from a separate PR).

---

EDIT: To be clear, we should not put this in release notes or advertise it to users, nor make any stability claims, until we've run PkgEval and much more! But just merging this first step is totally fine.

[udesou]

@nanosoldier runtests(["EarlyStopping", "Malt", "NearestNeighbors", "BPGates", "IMFData", "NCEI", "OptimizationPRIMA", "JUDI", "TaylorIntegration", "KolmogorovArnold", "DynamicalSystemsBase", "DiffEqDevTools", "MicroCanonicalHMC", "CellMLToolkit", "FlightSims", "SpiDy", "QuantumAnnealing", "Modia", "MRIgeneralizedBloch", "SideKicks"])

[oscardssmith]

"Malt", and "NearestNeighbors" are clear. Gabriel and I were seeing semi-random segfaults on them on nightly.

[udesou]

"Malt", and "NearestNeighbors" are clear. Gabriel and I were seeing semi-random segfaults on them on nightly.

Good to know. I commented again because I'm not sure it actually triggered a rerun. Do I need specific privileges to run PkgEval?

[oscardssmith]

yes, and it also needs the back ticks
@nanosoldier runtests(["EarlyStopping", "BPGates", "IMFData", "NCEI", "OptimizationPRIMA", "JUDI", "TaylorIntegration", "KolmogorovArnold", "DynamicalSystemsBase", "DiffEqDevTools", "MicroCanonicalHMC", "CellMLToolkit", "FlightSims", "SpiDy", "QuantumAnnealing", "Modia", "MRIgeneralizedBloch", "SideKicks"])

[nanosoldier]

The package evaluation job you requested has completed - no new issues were detected.
The full report is available.

[udesou]

It seems like PkgEval ran successfully with the stock GC. Given that @d-netto is on holidays, I’m not sure he’ll have time to look into this. @gbaraldi if you’re happy with the makefile changes, this should be ready to merge, I believe.

[oscardssmith]

PkgEval came back fully clean.

[nanosoldier]

The package evaluation job you requested has completed - no new issues were detected.
The full report is available.