Enabling the epoch serving-mode JIT globally regressed continuation-based guest
interpreters (the epoch mode is the shared command channel every loop's
conformance runner uses). Two-part fix:
1. SAFE DEFAULT GATE. register_jit_hook in the persistent server branch is now
opt-in via SX_SERVING_JIT=1 (default OFF). Default behaviour is unchanged
(no JIT in epoch serving) → zero regression for sibling loops. The
content/Smalltalk page server opts in.
2. GENERAL FIXES + per-guest interpret-only declarations:
- callable? (sx_server/run_tests/integration_tests/mcp_tree) now accepts
VmClosure. A JIT-compiled higher-order function returns its inner closure
as a VmClosure; callable? previously rejected it, so scheme-apply's
(callable? proc) guard failed with "not a procedure: <vm:anon>".
- jit-exclude! gains a trailing-"*" namespace-prefix form
(Sx_types.jit_excluded_prefixes), the robust way to mark a whole guest
interpreter interpret-only (a name-list misses functions in extra files —
it left erlang's vm/dispatcher JIT'd and 13 tests short).
- Per-guest exclusions in each guest's runtime.sx:
scheme "scheme-*" "scm-*" erlang "er-*" "erlang-*"
prolog "pl-*" common-lisp "cl-*" "clos-*"
js "js-*" haskell "hk-*"
Verified under opt-in JIT (== CEK, no hang): smalltalk 847/847, scheme/flow
166/166, erlang 530/530, prolog 590/590, apl 152/152, js 147/148. Residual
(documented, protected by the default gate): common-lisp 6 fails in advanced
suites (parser-recovery/debugger/CLOS/MOP). lua (0/16) and tcl (3/4) fail
identically on CEK — pre-existing, not JIT. run_tests --jit/no-jit unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Phase 1 of the jit-perf-regression plan reproduced and quantified the alleged
30× substrate slowdown across 5 guests (tcl, lua, erlang, prolog, haskell). On
a quiet machine all five suites pass cleanly:
tcl test.sh 57.8s wall, 16.3s user, 376/376 ✓
lua test.sh 27.3s wall, 4.2s user, 185/185 ✓
erlang conformance 3m25s wall, 36.8s user, 530/530 ✓ (needs ≥600s budget)
prolog conformance 3m54s wall, 1m08s user, 590/590 ✓
haskell conformance 6m59s wall, 2m37s user, 156/156 ✓
Per-test user-time at architecture HEAD vs pre-substrate-merge baseline
(83dbb595) is essentially flat (tcl 0.83×, lua 1.4×, prolog 0.82×). The
symptoms reported in the plan (test timeouts, OOMs, 30-min hangs) were heavy
CPU contention from concurrent loops + one undersized internal `timeout 120`
in erlang's conformance script. There is no substrate regression to bisect.
Changes:
* lib/tcl/test.sh: `timeout 2400` → `timeout 300`. The original 180s deadline
is comfortable on a quiet machine (3.1× headroom); 300s gives some safety
margin for moderate contention without masking real regressions.
* lib/erlang/conformance.sh: `timeout 120` → `timeout 600`. The 120s budget
was actually too tight for the full 9-suite chain even before this work.
* lib/erlang/scoreboard.{json,md}: 0/0 → 530/530 — populated by a successful
conformance run with the new deadline. The previous 0/0 was a stale
artefact of the run timing out before parsing any markers.
* plans/jit-perf-regression.md: full Phase 1 progress log including
per-guest perf table, quiet-machine re-measurement, and conclusion.
Phases 2–4 (bisect, diagnose, fix) skipped — there is no substrate regression
to find. Phase 6 (perf-regression alarm) still planned to catch the next
quadratic blow-up early instead of via watchdog bumps.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
