perf(gnovm): recycle runtime blocks through a per-machine pool

[thehowl]

Note

Part 3 of 6 of the gnovm performance stack (split from #5800), to be merged in order:

1. perf(gnovm): parallelize test suites and add gno test -jobs #5811 — perf(gnovm): parallelize test suites and add gno test -jobs
2. perf(gnovm): avoid heap-boxed byte access in copy, range and index reads #5812 — perf(gnovm): avoid heap-boxed byte access in copy, range and index reads
3. perf(gnovm): recycle runtime blocks through a per-machine pool #5813 — perf(gnovm): recycle runtime blocks through a per-machine pool
4. perf(gnovm): share interface-held values when copying arrays #5814 — perf(gnovm): share interface-held values when copying arrays (gas-visible)
5. ci(gnovm): skip print-only coverage instrumentation #5815 — ci(gnovm): skip print-only coverage instrumentation
6. perf(gnovm): reduce per-call and per-op allocations #5816 — perf(gnovm): reduce per-call and per-op allocations

Each PR is based on the previous one's branch; this one diffs against part 2. Together: ci / gnovm ~14m → 6m18s; pkg/gnolang test time −64%; VM heap allocations −84% on the heaviest suite.

Summary

After the byte-access fixes, NewBlock was the dominant allocation site: 45% of remaining heap objects (135M in the bytes suite — 75M per-scope blocks from doOpExec for if/for/range/switch/block, 60M call blocks from doOpCall, each also allocating a Values slice).

The enabler is an invariant the heap-items design already established: closures capture HeapItemValues rather than blocks (doOpFuncLit sets Parent: nil and copies Captures), &local is always heap-promoted by the preprocessor, and frames/stacktraces store only indices and locations — so a runtime block provably dies when discarded from the machine's block stack. acquireBlock/releaseBlock implement a small per-machine pool on that invariant, routed through every block discard site (OpPopBlock, GotoJump, PopFrameAndReset/Return, PeekFrameAndContinueFor/Range). Released blocks are zeroed so they retain no references.

releaseBlock excludes the three block populations that also travel the stack but must never be pooled:

• node-owned static blocks (Eval/RunStatement flows push them) — identified by static-block identity (b.Source.GetStaticBlock().GetBlock() == b);
• file/package blocks (referenced by FuncValue.Parent) — identified by Source node type;
• defer-site blocks — Defer.Parent is visited by the VM's GC until the defer runs, so doOpDefer marks them via a flag stored in bodyStmt's trailing padding, keeping unsafe.Sizeof(Block{}) — and with it the _allocBlock gas constant — unchanged (the alloc-constants init assert enforces this).

Panic unwinding skips pooling entirely as cheap conservatism. Gas and VM-GC accounting are unchanged: acquireBlock charges AllocateBlock exactly like Allocator.NewBlock, and pooled blocks are unreachable from GC roots just like dead blocks today.

One negative result is documented in Machine.Release: carrying the pool through the cross-goroutine machinePool measured 10–25% slower on parallel workloads (extra live heap across GC cycles, lost cache locality) without helping machine-churn workloads, so the pool is deliberately per-machine-lifetime only.

Measurements

	before	after
heap objects (bytes suite)	300M	165M
TestStdlibs/bytes solo	105.2s	94.9s
full pkg/gnolang long mode, 16 cores	184.6s	154.0s
4-core + coverage CI simulation	509.7s	424.9s (600.0s on master)

Verification: all 2344 filetest goldens byte-identical (incl. Gas:/Realm:/Storage:/MAXALLOC tests), vm gas tests, txtar, 220 example packages, cmd/gno, and the closure/defer/recover/heap/goto escape-pattern filetests specifically.

[Gno2D2]

🛠 PR Checks Summary

All Automated Checks passed. ✅

Manual Checks (for Reviewers):

• IGNORE the bot requirements for this PR (force green CI check)

Read More

🤖 This bot helps streamline PR reviews by verifying automated checks and providing guidance for contributors and reviewers.

✅ Automated Checks (for Contributors):

No automated checks match this pull request.

☑️ Contributor Actions:

1. Fix any issues flagged by automated checks.
2. Follow the Contributor Checklist to ensure your PR is ready for review.
   • Add new tests, or document why they are unnecessary.
   • Provide clear examples/screenshots, if necessary.
   • Update documentation, if required.
   • Ensure no breaking changes, or include BREAKING CHANGE notes.
   • Link related issues/PRs, where applicable.

☑️ Reviewer Actions:

1. Complete manual checks for the PR, including the guidelines and additional checks if applicable.

📚 Resources:

• Report a bug with the bot.
• View the bot’s configuration file.

Debug

Manual Checks

**IGNORE** the bot requirements for this PR (force green CI check)

If

🟢 Condition met
└── 🟢 On every pull request

Can be checked by

• Any user with comment edit permission