[Gajesh2007]

Swift admission code needs one coherent allocator snapshot and checked backing-size information, but the C boundary currently offers only separate counters and no detached footprint-policy value.

This change bridges the native snapshot and immutable sizing policy, adds checked bound helpers, and exposes metadata for an already evaluated array buffer. Unavailable backing stays explicitly unavailable; inspection does not evaluate an array or synchronize a stream. The generator override preserves the API when bindings are regenerated.

Before

flowchart LR
  Swift[Swift memory caller] --> Separate[Independent C counter calls]
  Array[Lazy MLX array] --> Missing[No checked backing metadata bridge]
  Separate --> Estimate[Caller combines observations]
  Missing --> Estimate

Loading

After

flowchart LR
  Swift[Swift memory caller] --> Snapshot[mlx_get_memory_snapshot]
  Swift --> Policy[mlx_get_allocation_footprint_policy]
  Policy --> Bound[Checked C bound / maximum-extra helpers]
  Array[Already evaluated array] --> Info[mlx_array_get_buffer_info]
  Info --> Availability[Explicit availability and backing metadata]
  Snapshot --> Result[Coherent native observation]

Loading

Reviewed source and signed publication

• Reviewed source: d4328f2d8d54d711d5419e07ab9fa2f07b512a48.
• Signed publication head: 8bdeb0f3c5793c54139e86c5437b09abc9ab9b42 on release/0.9.0-validation.
• This existing signed head is retained because its source tree is already current; no empty commit was added.
• Normalizing only the signed dependency pins recovers the exact reviewed source tree. Both parent Vercel suppressions remain branch-specific.

Validation and its scope

• The C implementation is unchanged from the previous signed publication. The reviewed core companion now includes the quantized-bias correction; Swift pins both signed dependencies.
• Earlier source-matched Apple M3 Max evidence compiled this C source with core 560f57cc and Swift 4b006d8, exercised the C snapshot/footprint/buffer bridge, and passed 83 unique functions / 100 expanded cases in the combined lane. Those results retain the old lower-tuple identity.
• The corrected runtime artifact's source/resource/build-output review is separate from pending full-model gates. No new C ABI claim, full upstream-suite result or CUDA execution is inferred from that review.

Frozen Gemma controls and their limits

• On the reviewed 53f3c3d0 / native de30ef98 / Swift c06149b4 / core fab0f39f runtime, cache-off B1 Gemma QAT4 is now exact across contiguous and paged backends in ordinary, automatic and serial-target modes. Serial-target also matches ordinary.
• Automatic still differs from ordinary on both backends at zero-based index 30. The whole-verifier comparison therefore remains a failure/open gate. These seven complete trajectories reuse one fixture prompt; they are not seven independent quality prompts.
• The later recurrent-teacher runtime is outside those frozen model controls. No new throughput, B2/B4, SSD-hit, pressure, fleet-model or 0.9.0 default-posture result is inferred from them.

Remaining gates — keep draft

• Full-model, whole-verifier and 0.9.0 default/release gates remain open. In particular, Gemma automatic-versus-ordinary index 30 and the outstanding Qwen trajectory/state investigations are not waived by backend equality or operator proximity.
• Complete the exact fleet-model/backend/cache/MTP matrix, capacity/co-residency and pressure coverage, uncontended B1/B2/B4 comparisons, connected/two-host recovery and persistent-key restart evidence on the final bound artifacts.
• Complete full provider CI compilation and GPU regression for the public-API test fix, applicable platform review, and immutable signed-artifact checks. CUDA execution remains outside the macOS evidence. Historical failed checks are preserved; no new full-suite pass is claimed here.
• Explicit release, deployment, signing-workflow activation/dispatch and benchmark approval remain separate. Both Vercel configs suppress only release/0.9.0-validation. No production budget, equality criterion or tolerance is relaxed.

Companion review stack

Draft PR	Signed head	Reviewed source
mlx #15	f8eec2476cdb5263fb6e2a19dd71f2c8b7c0a7d3	fab0f39f69140393b454c32d6f4bf7a9b32f9dcc
mlx-c #8	8bdeb0f3c5793c54139e86c5437b09abc9ab9b42	d4328f2d8d54d711d5419e07ab9fa2f07b512a48
mlx-swift #21	c3ac44766c3e0be2066bbd87a3e26aaa6f410ab0	9561227d55a07db29f70a78aadc5d6b5aaeb10bf
mlx-swift-lm #141	67ac8607b5846e8acfbbdd8f36c0d252d69d0d01	ba5c03ea0ef495aca02c99c7a2dd2b7a32bc7668
d-inference #851	e1d68db84cb1a1e4a6dd10306383bd64bc5d6062	d01b82078e157ae1e66511d2d8db1f2ecf95e857

Evidence navigation

Reports retain their original source, artifact and result identities. Current links are pinned to this signed parent publication.

• Strict validation procedures.
• Recurrent teacher state and peak-capacity evidence.
• Corrected Gemma backend controls and remaining verifier failure.
• Allocator footprint and backing ownership.
• Qwen common-input operator evidence.
• Two-host fixture evidence.