Fix: 修正单元素维 stride 的连续性判定

[nalinaly]

shape=(64, 1)、stride=(1, 64) 是合法连续列向量，但上游将
单元素维的 stride 与行主序期望值比较，导致 ChipTensor 及
HBG/TMR Tensor 误判为不连续，零拷贝 reshape 随之触发断言。

• 公共 ChipTensor 和 HBG/TMR 派生缓存统一忽略 size=1 维的 stride；其他维度继续严格检查，保持存储跨度与 ABI 不变。
• HBG 图元数据校验使用相同规则，继续拒绝伪造连续标志、 跨步存储的错误标志及错误 extent，避免合法图记录被拒绝。
• 增加 10 项 C++ 回归，覆盖外部参数转换、转置列向量、 零拷贝 reshape、真实空隙拒绝及图元数据校验；更新 ABI 文档。

上游验证：

• PyPTO 747e1e4 固定 runtime 4e4d3a4 与 simpler 主线 d79c88c 均能复现：新增 10 项用例中 6 项失败、4 项通过。
• 修复后新增 10 项全部通过；同时运行既有 buffer 和 child_memory 的 23 项用例，总计 33 项通过。
• 使用当前独立工作树重新编译 CPU C++ 测试，无需 NPU。 本分支未执行 A5 真机、Python 套件或完整系统回归。

这是 0b852b1 修复语义在上游重构后的 Tensor 架构上的适配。
分支直接基于上游主线，不包含 fork 中其他 A5 L1 功能修改。

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Team

Run ID: af5c36b6-b3ef-4c17-9f11-088513f320ea

📥 Commits

Reviewing files that changed from the base of the PR and between d79c88c and 247689f.

📒 Files selected for processing (8)

• docs/buffer-abi.md
• src/common/host_build_graph/graph_execution.h
• src/common/host_build_graph/tensor.h
• src/common/task_interface/tensor.h
• src/common/tensormap_and_ringbuffer/tensor.h
• tests/ut/cpp/CMakeLists.txt
• tests/ut/cpp/types/test_tensor_contiguity.cpp
• tests/ut/cpp/types/test_tmr_tensor_contiguity.cpp

Included review availability: Your plan provides up to 1 included review per hour; 0 remain after this review.

---

📝 Walkthrough

Walkthrough

The change aligns tensor contiguity checks across runtime types and graph validation. Singleton dimensions no longer constrain strides. New tests cover storage extent, zero-copy reshape, wire validation, and rejection of gapped storage.

Changes

Contiguity rule alignment

Layer / File(s)	Summary
Runtime contiguity semantics docs/buffer-abi.md, src/common/host_build_graph/tensor.h, src/common/task_interface/tensor.h, src/common/tensormap_and_ringbuffer/tensor.h	Contiguity checks now ignore strides for dimensions of size one. Non-singleton dimensions still require row-major strides. Documentation and cached-field comments describe the updated rule.
Graph validation and test coverage src/common/host_build_graph/graph_execution.h, tests/ut/cpp/CMakeLists.txt, tests/ut/cpp/types/test_*contiguity.cpp	Graph wire validation accepts valid singleton strides and rejects forged flags or invalid extents. Tests cover runtime metadata, zero-copy reshaping, transposed singleton dimensions, and gapped storage.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Merge Risk: ⚪ Minimal · up to 24768

Singleton-stride tensor views now use the intended contiguity semantics without changing storage layout or ABI. Covered runtime, reshape, extent, and graph-validation cases show no remaining merge-blocking risk.

Poem

A rabbit checks each stride in line
Singleton gaps are now benign
The graph accepts the matching view
Zero-copy paths stay true
Gapped storage still says “not you”

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 30.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 20 functions across 6 files. (2 skipped: …	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description check	✅ Passed	The description clearly explains the singleton-dimension stride fix, affected tensor layers, validation rules, regression tests, and verification scope. It is directly related to the changeset.
Title check	✅ Passed	The title concisely identifies the fix for contiguity判定 of singleton-dimension strides. It accurately summarizes the main change.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

Full details: Docstring Coverage

Explanation

Docstring coverage is 30.00% which is insufficient. The required threshold is 80.00%. Docstring coverage is scoped to functions touched by this diff. Analyzed 20 functions across 6 files. (2 skipped: 2 unsupported.)

• Fix all pre-merge checks with AI

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}