Merge remote-tracking branch 'origin/main' into explicit-graph-construction

Made-with: Cursor

# Conflicts:
#	cuda_pathfinder/cuda/pathfinder/_dynamic_libs/descriptor_catalog.py

Author: Andy-Jost

.pre-commit-config.yaml

@@ -30,7 +30,7 @@ repos:
         language: python
         additional_dependencies:
           - https://files.pythonhosted.org/packages/cc/20/ff623b09d963f88bfde16306a54e12ee5ea43e9b597108672ff3a408aad6/pathspec-0.12.1-py3-none-any.whl
-        exclude: '(.*pixi\.lock)|(\.git_archival\.txt)'
+        exclude: '(.*pixi\.lock)|(\.git_archival\.txt)|(.*\.patch$)'
         args: ["--fix"]
 
       - id: no-markdown-in-docs-source
@@ -59,7 +59,13 @@ repos:
       exclude: &gen_exclude '^(?:cuda_python/README\.md|cuda_bindings/cuda/bindings/.*\.in?|cuda_bindings/docs/source/module/.*\.rst?)$'
     - id: mixed-line-ending
     - id: trailing-whitespace
-      exclude: *gen_exclude
+      exclude: |
+        (?x)^(?:
+          cuda_python/README\.md|
+          cuda_bindings/cuda/bindings/.*\.in?|
+          cuda_bindings/docs/source/module/.*\.rst?|
+          .*\.patch$
+        )$
 
   # Checking for common mistakes
   - repo: https://github.com/pre-commit/pygrep-hooks

AGENTS.md

@@ -0,0 +1 @@
+AGENTS.md

CLAUDE.md

@@ -5,6 +5,6 @@ backport_branch: "12.9.x"  # keep in sync with target-branch in .github/dependab
 
 cuda:
   build:
-    version: "13.1.1"
+    version: "13.2.0"
   prev_build:
     version: "12.9.1"

ci/versions.yml

@@ -0,0 +1,67 @@
+This file describes `cuda_bindings`, the low-level CUDA host API bindings
+subpackage in the `cuda-python` monorepo.
+
+## Scope and principles
+
+- **Role**: provide low-level, close-to-CUDA interfaces under
+  `cuda.bindings.*` with broad API coverage.
+- **Style**: prioritize correctness and API compatibility over convenience
+  wrappers. High-level ergonomics belong in `cuda_core`, not here.
+- **Cross-platform**: preserve Linux and Windows behavior unless a change is
+  intentionally platform-specific.
+
+## Package architecture
+
+- **Public module layer**: Cython modules under `cuda/bindings/` expose user
+  APIs (`driver`, `runtime`, `nvrtc`, `nvjitlink`, `nvvm`, `cufile`, etc.).
+- **Internal binding layer**: `cuda/bindings/_bindings/` provides lower-level
+  glue and loader helpers used by public modules.
+- **Platform internals**: `cuda/bindings/_internal/` contains
+  platform-specific implementation files and support code.
+- **Build/codegen backend**: `build_hooks.py` drives header parsing, template
+  expansion, extension configuration, and Cythonization.
+
+## Generated-source workflow
+
+- **Do not hand-edit generated binding files**: many files under
+  `cuda/bindings/` (including `*.pyx`, `*.pxd`, `*.pyx.in`, and `*.pxd.in`)
+  are generated artifacts.
+- **Generated files are synchronized from another repository**: changes to these
+  files in this repo are expected to be overwritten by the next sync.
+- **If generated output must change**: make the change at the generation source
+  and sync the updated artifacts back here, rather than patching generated files
+  directly in this repo.
+- **Header-driven generation**: parser behavior and required CUDA headers are
+  defined in `build_hooks.py`; update those rules when introducing new symbols.
+- **Platform split files**: keep `_linux.pyx` and `_windows.pyx` variants
+  aligned when behavior should be equivalent.
+
+## Testing expectations
+
+- **Primary tests**: `pytest tests/`
+- **Cython tests**:
+  - build: `tests/cython/build_tests.sh` (or platform equivalent)
+  - run: `pytest tests/cython/`
+- **Examples**: example coverage is pytest-based under `examples/`.
+- **Benchmarks**: run with `pytest --benchmark-only benchmarks/` when needed.
+- **Orchestrated run**: from repo root, `scripts/run_tests.sh bindings`.
+
+## Build and environment notes
+
+- `CUDA_HOME` or `CUDA_PATH` must point to a valid CUDA Toolkit for source
+  builds that parse headers.
+- `CUDA_PYTHON_PARALLEL_LEVEL` controls build parallelism.
+- `CUDA_PYTHON_PARSER_CACHING` controls parser-cache behavior during generation.
+- Runtime behavior is affected by
+  `CUDA_PYTHON_CUDA_PER_THREAD_DEFAULT_STREAM` and
+  `CUDA_PYTHON_DISABLE_MAJOR_VERSION_WARNING`.
+
+## Editing guidance
+
+- Keep CUDA return/error semantics explicit and avoid broad fallback behavior.
+- Reuse existing helper layers (`_bindings`, `_internal`, `_lib`) before adding
+  new one-off utilities.
+- If you add or change exported APIs, update relevant docs under
+  `docs/source/module/` and tests in `tests/`.
+- Prefer changes that are easy to regenerate/rebuild rather than patching
+  generated output directly.

cuda_bindings/AGENTS.md

@@ -0,0 +1 @@
+AGENTS.md

cuda_bindings/CLAUDE.md

@@ -1,7 +1,7 @@
 # SPDX-FileCopyrightText: Copyright (c) 2021-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
 # SPDX-License-Identifier: LicenseRef-NVIDIA-SOFTWARE-LICENSE
 
-# This code was automatically generated with version 13.1.0, generator version 49a8141. Do not modify it directly.
+# This code was automatically generated with version 13.2.0, generator version fa58871. Do not modify it directly.
 from cuda.bindings.cydriver cimport *
 
 {{if 'cuGetErrorString' in found_functions}}
@@ -439,6 +439,11 @@ cdef CUresult _cuKernelGetName(const char** name, CUkernel hfunc) except ?CUDA_E
 cdef CUresult _cuKernelGetParamInfo(CUkernel kernel, size_t paramIndex, size_t* paramOffset, size_t* paramSize) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuKernelGetParamCount' in found_functions}}
+
+cdef CUresult _cuKernelGetParamCount(CUkernel kernel, size_t* paramCount) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuMemGetInfo_v2' in found_functions}}
 
 cdef CUresult _cuMemGetInfo_v2(size_t* free, size_t* total) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -679,6 +684,16 @@ cdef CUresult _cuMemcpyBatchAsync_v2(CUdeviceptr* dsts, CUdeviceptr* srcs, size_
 cdef CUresult _cuMemcpy3DBatchAsync_v2(size_t numOps, CUDA_MEMCPY3D_BATCH_OP* opList, unsigned long long flags, CUstream hStream) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuMemcpyWithAttributesAsync' in found_functions}}
+
+cdef CUresult _cuMemcpyWithAttributesAsync(CUdeviceptr dst, CUdeviceptr src, size_t size, CUmemcpyAttributes* attr, CUstream hStream) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
+{{if 'cuMemcpy3DWithAttributesAsync' in found_functions}}
+
+cdef CUresult _cuMemcpy3DWithAttributesAsync(CUDA_MEMCPY3D_BATCH_OP* op, unsigned long long flags, CUstream hStream) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuMemsetD8_v2' in found_functions}}
 
 cdef CUresult _cuMemsetD8_v2(CUdeviceptr dstDevice, unsigned char uc, size_t N) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -1069,6 +1084,16 @@ cdef CUresult _cuStreamCreate(CUstream* phStream, unsigned int Flags) except ?CU
 cdef CUresult _cuStreamCreateWithPriority(CUstream* phStream, unsigned int flags, int priority) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuStreamBeginCaptureToCig' in found_functions}}
+
+cdef CUresult _cuStreamBeginCaptureToCig(CUstream hStream, CUstreamCigCaptureParams* streamCigCaptureParams) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
+{{if 'cuStreamEndCaptureToCig' in found_functions}}
+
+cdef CUresult _cuStreamEndCaptureToCig(CUstream hStream) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuStreamGetPriority' in found_functions}}
 
 cdef CUresult _cuStreamGetPriority(CUstream hStream, int* priority) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -1309,6 +1334,11 @@ cdef CUresult _cuFuncGetName(const char** name, CUfunction hfunc) except ?CUDA_E
 cdef CUresult _cuFuncGetParamInfo(CUfunction func, size_t paramIndex, size_t* paramOffset, size_t* paramSize) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuFuncGetParamCount' in found_functions}}
+
+cdef CUresult _cuFuncGetParamCount(CUfunction func, size_t* paramCount) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuFuncIsLoaded' in found_functions}}
 
 cdef CUresult _cuFuncIsLoaded(CUfunctionLoadingState* state, CUfunction function) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -1344,6 +1374,11 @@ cdef CUresult _cuLaunchCooperativeKernelMultiDevice(CUDA_LAUNCH_PARAMS* launchPa
 cdef CUresult _cuLaunchHostFunc(CUstream hStream, CUhostFn fn, void* userData) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuLaunchHostFunc_v2' in found_functions}}
+
+cdef CUresult _cuLaunchHostFunc_v2(CUstream hStream, CUhostFn fn, void* userData, unsigned int syncMode) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuFuncSetBlockShape' in found_functions}}
 
 cdef CUresult _cuFuncSetBlockShape(CUfunction hfunc, int x, int y, int z) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -1824,6 +1859,11 @@ cdef CUresult _cuGraphAddNode_v2(CUgraphNode* phGraphNode, CUgraph hGraph, const
 cdef CUresult _cuGraphNodeSetParams(CUgraphNode hNode, CUgraphNodeParams* nodeParams) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuGraphNodeGetParams' in found_functions}}
+
+cdef CUresult _cuGraphNodeGetParams(CUgraphNode hNode, CUgraphNodeParams* nodeParams) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuGraphExecNodeSetParams' in found_functions}}
 
 cdef CUresult _cuGraphExecNodeSetParams(CUgraphExec hGraphExec, CUgraphNode hNode, CUgraphNodeParams* nodeParams) except ?CUDA_ERROR_NOT_FOUND nogil
@@ -2159,6 +2199,26 @@ cdef CUresult _cuCoredumpSetAttribute(CUcoredumpSettings attrib, void* value, si
 cdef CUresult _cuCoredumpSetAttributeGlobal(CUcoredumpSettings attrib, void* value, size_t* size) except ?CUDA_ERROR_NOT_FOUND nogil
 {{endif}}
 
+{{if 'cuCoredumpRegisterStartCallback' in found_functions}}
+
+cdef CUresult _cuCoredumpRegisterStartCallback(CUcoredumpStatusCallback callback, void* userData, CUcoredumpCallbackHandle* callbackOut) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
+{{if 'cuCoredumpRegisterCompleteCallback' in found_functions}}
+
+cdef CUresult _cuCoredumpRegisterCompleteCallback(CUcoredumpStatusCallback callback, void* userData, CUcoredumpCallbackHandle* callbackOut) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
+{{if 'cuCoredumpDeregisterStartCallback' in found_functions}}
+
+cdef CUresult _cuCoredumpDeregisterStartCallback(CUcoredumpCallbackHandle callback) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
+{{if 'cuCoredumpDeregisterCompleteCallback' in found_functions}}
+
+cdef CUresult _cuCoredumpDeregisterCompleteCallback(CUcoredumpCallbackHandle callback) except ?CUDA_ERROR_NOT_FOUND nogil
+{{endif}}
+
 {{if 'cuGetExportTable' in found_functions}}
 
 cdef CUresult _cuGetExportTable(const void** ppExportTable, const CUuuid* pExportTableId) except ?CUDA_ERROR_NOT_FOUND nogil