test(coverage): stat_print / DB_STAT_ALL coverage (statprint001 + wire env020 into the suite)

[gburd]

What

Raises coverage of the statistics subsystem — the biggest tractable single-process gap (~2k lines across 8 files at 0-29%). These are the verbose stat_print / DB_STAT_ALL formatters (__*_print_*) that format every field to a message file; functional tests read stat() values but almost never call stat_print(DB_STAT_ALL).

How

Two levers, both wired into the default COV_TESTS:

1. env020 (already existed, was just never in the coverage subset) exercises every Tcl *_stat_print binding — env / lock / log / mpool / mutex / txn / rep / repmgr / db / seq — with each flag (default, -clear, -all, -subsystem, -lk_*, -hash, …).

2. statprint001 (new) closes the two spots env020 misses and adds the db_stat utility entry path:

   • heap_stat.c — env020 opens no heap DB. Sweeps stat_print default / -fast / -all.
   • dbreg_stat.c __dbreg_print_all — reached only with DB_STAT_ALL | DB_STAT_SUBSYSTEM set together and databases open (so the LOG FNAME list has entries to iterate); env020 passes those flags separately. Calls env stat_print -all -subsystem with live open DBs and asserts the FNAME list is printed.
   • db_stat utility sweep — util/db_stat.c -> __*_stat_print (a separate read-only on-disk path the Tcl suite never touches). Runs db_stat with every meaningful flag: -e -E -c -C -l -L -m -M -x -X -r -R -t -Z -d -f over a populated all-subsystems env.

Registered statprint in testparams.tcl (subs + test_names), added a statprint case to the r dispatcher, and added env020 + statprint001 to run_coverage.sh COV_TESTS (README updated).

Measured lift (env020 + statprint001 vs full-run-2/cov-ranking.txt baseline)

file	before	after
env/env_stat.c	16.8%	79.9%
lock/lock_stat.c	17.4%	73.6%
rep/rep_stat.c	19.4%	72.4%
db/db_stati.c	22.0%	61.8%
log/log_stat.c	22.9%	87.1%
mutex/mut_stat.c	29.4%	86.8%
sequence/seq_stat.c	0.0%	63.0%
dbreg/dbreg_stat.c	0.0%	69.6%
heap/heap_stat.c	0.0%	80.3%
qam/qam_stat.c	57.6%	78.0%
txn/txn_stat.c	56.8%	69.0%

The two 0% files (seq_stat.c, dbreg_stat.c, heap_stat.c) are now substantially covered.

Validation

Built with CC=gcc ... CFLAGS="-O0 -g --coverage", ran env020 + statprint001 under timeout in the nix dev shell, captured with lcov from .libs. Both tests PASS. r statprint also runs the new test via the dispatcher. No coverage artifacts / TESTDIR / cores committed.

Still cold (documented, not fixed)

• seq_stat.c __seq_print_all is a no-op stub, and the Tcl seq stat_print binding exposes only -clear, not -all — reaching the DB_STAT_ALL branch would need a binding change for a 2-line stub. Left alone.
• repmgr_stat.c send-path counters need a live repmgr transport (repmgr009+, COV_REP=1).

No bugs found

All stat_print paths behaved correctly under the sweep.

[github-actions]

Coccinelle convention checks

No new violations. ✅

Resolved since baseline (2) -- update dist/cocci/baseline.txt to lock these in.

rule_mutex_unbalanced|MUTEX_UNBALANCED|src/crypto/mersenne/mt19937db.c|return (ret);
rule_mutex_unbalanced|MUTEX_UNBALANCED|src/mp/mp_register.c|return (ret);

[github-actions]

ABI diff vs v5.3.31 (libabigail — authoritative)

Removed exported symbols (nm -D, _NNNN version suffix normalized)

None.

---

Advisory: libabigail/nm is the authoritative binary-ABI check; Coccinelle is complementary source-level early warning. See dist/cocci/README.md.