陈雨钒 · Chen Yufan

系统与 AI 基础设施  ·  Systems & AI Infrastructure

写编译器前端的补丁，也修别人的认证漏洞。
_{Patching compiler frontends, and fixing other people's auth bypasses.}

给自己在用的项目提交的修复。链接指向 PR —— 7 个已提交，其中 4 个已获维护者批准。 _{Fixes sent to projects I use. Seven pull requests, four approved by maintainers.}

#20245  ✅ 已批准 approved  — 修复 Relax PyTorch 前端 flatten 的参数校验缺陷。越界的负 start_dim 不在使用处报错：它被归一化成 -1，把多余的一维折进乘积，发出形状错误的 reshape，直到下游才失败。补上范围与顺序校验、0 维张量支持和回归测试。

_{Validate flatten dims in the Relax PyTorch frontend. An out-of-range negative start_dim silently computed a wrong shape that only failed further downstream in reshape. Added range and ordering checks, 0-d input support, and regression tests.}

#20254  ✅ 已批准 approved  — torch.sort / torch.argsort 返回 int64 索引，前端却发 int32。同文件的 _topk 已经显式覆盖了同一个 int32 默认值，说明这是遗漏而非约定；前端支持的其他索引类算子（argmax、argmin、max(dim)、median(dim)、bucketize）也都是 int64。一张导入图会因此对同一种值带上两种索引 dtype。

_{torch.sort / torch.argsort return int64 indices; the frontend emitted int32. _topk in the same file already overrides the identical default, and every other index-producing op comes out int64 — so one imported graph could carry two index dtypes for the same kind of value.}

#20255  ✅ 已批准 approved  — Relax 的 reshape 把目标形状里的字面 0 读作"沿用输入对应维度"（ONNX allowzero=0 语义），PyTorch 读作真实的零维。空张量上的 reshape / view / flatten / unflatten 因此报错或静默给出错误形状；flatten 在 (0,3) 上恰好正确，正是这个巧合掩盖了其余情形。经与 torch 逐例对照 2132 组静态形状 + 63 组符号形状验证。

_{Relax's reshape reads a literal 0 as "copy the input dimension" (ONNX allowzero=0) where PyTorch reads a real zero-sized one, so reshape / view / flatten / unflatten on empty tensors raised or silently produced a wrong shape. Verified against torch across 2132 static and 63 symbolic shape cases.}

#6801 — 修复 RBAC 完全绕过漏洞（公告 GHSA-h543-6vgr-fm36，issue #6785）。两个 token 解析器都从未验签的 JWT 解码中授予完全信任的内部身份，而对比值硬编码在 Feast 官方 Helm chart 里 —— 任何能访问到服务的人都能伪造它，跳过该服务上所有项目的所有权限检查。改为对内部 token 签名、密钥移入 Kubernetes Secret 且不提供默认值、未配置时 fail closed。

_{Fix an RBAC bypass. Both token parsers granted a fully trusted internal identity from an unverified JWT decode, compared against a value hardcoded in Feast's own Helm chart. Signed the internal token, moved the secret to a Kubernetes Secret with no default, and made an unset secret fail closed.}

#6803 — 修复离线服务器的项目上下文泄漏。set_current_project 设置的 ContextVar 从不重置，一个请求的项目会残留到同一 worker 上后续未指定项目的请求里。改为 try / finally 配对重置，覆盖 _call_api 与 do_get 两条路径。

_{Fix a project-context leak in the offline server. The ContextVar set per request was never reset, so one request's project persisted into later requests on the same worker. Paired it with a finally reset on both entry points.}

#4140  ✅ 已批准 approved  — 统一 resize / rescale 的零尺寸语义。它们对零尺寸输出抛裸 ZeroDivisionError，而 warp_affine、warp_perspective、center_crop 对同样的参数返回空图像。空结果还需保持在自动微分图上 —— 否则一个退化成零尺寸的 batch 元素会静默断开梯度，这是 review 中提出、复现并修掉的。经 512 组形状 / 精度 / 插值模式组合验证，对合法输入数值零影响。

_{Align zero-size semantics in resize / rescale with the warping ops, and keep the empty result attached to the autograd graph so a batch element that degenerates to a zero-sized output still contributes a zero gradient. Verified numerically inert across 512 shape / dtype / interpolation combinations.}

#4143 — kornia.core.__all__ 与 kornia.color.__all__ 列出了模块里并未绑定的名字，from kornia.core import * 直接抛 AttributeError。既有的 API 面守卫看不到这类问题 —— 它比对的就是 __all__ 本身，绑定丢失的名字仍然"在列"。恢复绑定、清理遗留条目，并加了一条覆盖全仓库每一个声明 __all__ 模块的解析守卫。

_{kornia.core.__all__ and kornia.color.__all__ listed names that were not bound, so import * raised AttributeError. The existing surface guard could not see it — it compares __all__ against itself. Restored the bindings, dropped the leftover entry, and added a resolve check across every module that declares __all__.}