perf(mpool): #2/#7 prototype — measured that false-sharing isn't the cap

[gburd]

Prototypes and measures the first candidate fix for the read-scaling ceiling identified in docs/design/scaling-findings.md.

Hypothesis

struct __bh packs the write-hot fields (pin ref count, LRU priority — written on every __memp_fget/__memp_fput) into the same cache line as the read-mostly identity fields (pgno/mf_offset/flags/hq) that every concurrent hash-chain walk of a hot (btree root) buffer reads. So each pin would invalidate the line all readers need just to traverse/match.

Change

Isolate the write-hot fields on their own cache line, behind MPOOL_HOTFIELDS_ISOLATED (one-line A/B). Off by default.

Measured (controlled interleaved A/B, medians, 12-core)

workload	t	packed	isolated	delta
rrand	8	486,745	489,564	+0.6%
rrand	12	390,927	390,416	-0.1%
snap	8	518,422	514,260	-0.8%
snap	12	408,213	409,415	+0.3%

No effect (±0.6%). The cap is true sharing of the atomic counters (bhp->ref + the shared-latch share-counts), not false sharing — relocating the words can't help.

Why it's still useful

It rules out the cheap fix with data and refines #2/#7: the per-read shared-counter RMW must be removed (optimistic/versioned access — needs epoch reclamation BDB lacks — or a sharded pin count), not relocated. Kept guarded + off to re-A/B on the 24-core Linux box (currently unreachable) where the futex-dominated ceiling was characterized. Smoke-tested write + MVCC-freeze paths; full TCL regression required before any default-on change.

Builds clean default (off). See docs/design/scaling-findings.md → Prototype 1.

[gburd]

Closing as a superseded negative result.

This prototype tested whether cache-line isolation / false-sharing of the per-buffer pin and latch fields was the multi-core read-scaling bottleneck. The controlled A/B showed it was not: isolating those fields changed throughput by <1%. The cost is true sharing of the words themselves (every thread does atomic RMWs on the same root/internal-page cache lines on every descent), which padding cannot fix.

That finding directly motivated the chosen direction — write-free optimistic reads of the hot pages — now implemented as Stage 1b (lock-free root snapshot, #22) and the swip/optimistic-descent design (#20). The measured next bottleneck is the shared-handle cursor allocation mutex, tracked separately.

Keeping the branch for the record; no longer a merge candidate.