Merge branch 'master' into sh_merge_master

.pre-commit-config.yaml

@@ -25,14 +25,14 @@ repos:
 
 # Clang format the codebase automatically
 - repo: https://github.com/pre-commit/mirrors-clang-format
-  rev: "v18.1.5"
+  rev: "v18.1.8"
   hooks:
   - id: clang-format
     types_or: [c++, c, cuda]
 
 # Ruff, the Python auto-correcting linter/formatter written in Rust
 - repo: https://github.com/astral-sh/ruff-pre-commit
-  rev: v0.4.7
+  rev: v0.5.0
   hooks:
   - id: ruff
     args: ["--fix", "--show-fixes"]
@@ -40,7 +40,7 @@ repos:
 
 # Check static types with mypy
 - repo: https://github.com/pre-commit/mirrors-mypy
-  rev: "v1.10.0"
+  rev: "v1.10.1"
   hooks:
   - id: mypy
     args: []
@@ -80,7 +80,7 @@ repos:
 
 # Also code format the docs
 - repo: https://github.com/adamchainz/blacken-docs
-  rev: "1.16.0"
+  rev: "1.18.0"
   hooks:
   - id: blacken-docs
     additional_dependencies:
@@ -144,14 +144,14 @@ repos:
 
 # PyLint has native support - not always usable, but works for us
 - repo: https://github.com/PyCQA/pylint
-  rev: "v3.2.2"
+  rev: "v3.2.4"
   hooks:
   - id: pylint
     files: ^pybind11
 
 # Check schemas on some of our YAML files
 - repo: https://github.com/python-jsonschema/check-jsonschema
-  rev: 0.28.4
+  rev: 0.28.6
   hooks:
   - id: check-readthedocs
   - id: check-github-workflows

docs/requirements.txt

@@ -16,9 +16,9 @@ breathe==4.35.0 \
     --hash=sha256:5165541c3c67b6c7adde8b3ecfe895c6f7844783c4076b6d8d287e4f33d62386 \
     --hash=sha256:52c581f42ca4310737f9e435e3851c3d1f15446205a85fbc272f1f97ed74f5be
     # via -r requirements.in
-certifi==2024.2.2 \
+certifi==2024.7.4 \
-    --hash=sha256:0569859f95fc761b18b45ef421b1290a0f65f147e92a1e5eb3e635f9a5e4e66f \
+    --hash=sha256:5a1e7645bc0ec61a09e26c36f6106dd4cf40c6db3a1fb6352b0244e7fb057c7b \
-    --hash=sha256:dc383c07b76109f368f6106eee2b593b04a011ea4d55f652c6ca24a754d1cdd1
+    --hash=sha256:c198e21b1289c2ab85ee4e67bb4b4ef3ead0892059901a8d5b622f24a1101e90
     # via requests
 charset-normalizer==3.3.2 \
     --hash=sha256:06435b539f889b1f6f4ac1758871aae42dc3a8c0e24ac9e60c2384973ad73027 \

include/pybind11/detail/internals.h

@@ -149,20 +149,35 @@ struct override_hash {
 
 using instance_map = std::unordered_multimap<const void *, instance *>;
 
+#ifdef Py_GIL_DISABLED
+// Wrapper around PyMutex to provide BasicLockable semantics
+class pymutex {
+    PyMutex mutex;
+
+public:
+    pymutex() : mutex({}) {}
+    void lock() { PyMutex_Lock(&mutex); }
+    void unlock() { PyMutex_Unlock(&mutex); }
+};
+
 // Instance map shards are used to reduce mutex contention in free-threaded Python.
 struct instance_map_shard {
-    std::mutex mutex;
     instance_map registered_instances;
+    pymutex mutex;
     // alignas(64) would be better, but causes compile errors in macOS before 10.14 (see #5200)
-    char padding[64 - (sizeof(std::mutex) + sizeof(instance_map)) % 64];
+    char padding[64 - (sizeof(instance_map) + sizeof(pymutex)) % 64];
 };
 
+static_assert(sizeof(instance_map_shard) % 64 == 0,
+              "instance_map_shard size is not a multiple of 64 bytes");
+#endif
+
 /// Internal data structure used to track registered instances and types.
 /// Whenever binary incompatible changes are made to this structure,
 /// `PYBIND11_INTERNALS_VERSION` must be incremented.
 struct internals {
 #ifdef Py_GIL_DISABLED
-    std::mutex mutex;
+    pymutex mutex;
 #endif
     // std::type_index -> pybind11's type information
     type_map<type_info *> registered_types_cpp;
@@ -625,7 +640,7 @@ inline local_internals &get_local_internals() {
 }
 
 #ifdef Py_GIL_DISABLED
-#    define PYBIND11_LOCK_INTERNALS(internals) std::unique_lock<std::mutex> lock((internals).mutex)
+#    define PYBIND11_LOCK_INTERNALS(internals) std::unique_lock<pymutex> lock((internals).mutex)
 #else
 #    define PYBIND11_LOCK_INTERNALS(internals)
 #endif
@@ -662,7 +677,7 @@ inline auto with_instance_map(const void *ptr,
     auto idx = static_cast<size_t>(hash & internals.instance_shards_mask);
 
     auto &shard = internals.instance_shards[idx];
-    std::unique_lock<std::mutex> lock(shard.mutex);
+    std::unique_lock<pymutex> lock(shard.mutex);
     return cb(shard.registered_instances);
 #else
     (void) ptr;
@@ -678,7 +693,7 @@ inline size_t num_registered_instances() {
     size_t count = 0;
     for (size_t i = 0; i <= internals.instance_shards_mask; ++i) {
         auto &shard = internals.instance_shards[i];
-        std::unique_lock<std::mutex> lock(shard.mutex);
+        std::unique_lock<pymutex> lock(shard.mutex);
         count += shard.registered_instances.size();
     }
     return count;