Merge branch 'master' into skylion007/classmethod

.github/ISSUE_TEMPLATE/bug-report.yml

@@ -6,7 +6,8 @@ body:
   - type: markdown
     attributes:
       value: |
-        Maintainers will only make a best effort to triage PRs. Please do your best to make the issue as easy to act on as possible, and only open if clearly a problem with pybind11 (ask first if unsure).
+        Please do your best to make the issue as easy to act on as possible, and only submit here if there is clearly a problem with pybind11 (ask first if unsure). **Note that a reproducer in a PR is much more likely to get immediate attention.**
+
   - type: checkboxes
     id: steps
     attributes:
@@ -20,6 +21,12 @@ body:
         - label: Consider asking first in the [Gitter chat room](https://gitter.im/pybind/Lobby) or in a [Discussion](https:/pybind/pybind11/discussions/new).
           required: false
 
+  - type: Input
+    id: version
+    attributes:
+      label: What version (or hash if on master) of pybind11 are you using?
+      required: true
+
   - type: textarea
     id: description
     attributes:
@@ -40,6 +47,14 @@ body:
         The code should be minimal, have no external dependencies, isolate the
         function(s) that cause breakage. Submit matched and complete C++ and
         Python snippets that can be easily compiled and run to diagnose the
-        issue. If possible, make a PR with a new, failing test to give us a
+        issue. — Note that a reproducer in a PR is much more likely to get
-        starting point to work on!
+        immediate attention: failing tests in the pybind11 CI are the best
+        starting point for working out fixes.
       render: text
+
+  - type: Input
+    id: regression
+    attributes:
+      label: Is this a regression? Put the last known working version here if it is.
+      description: Put the last known working version here if this is a regression.
+      value: Not a regression

.github/workflows/ci.yml

@@ -30,7 +30,7 @@ jobs:
         - '3.6'
         - '3.9'
         - '3.10'
-        - '3.11-dev'
+        - '3.11'
         - 'pypy-3.7'
         - 'pypy-3.8'
         - 'pypy-3.9'
@@ -102,10 +102,12 @@ jobs:
       run: python -m pip install pytest-github-actions-annotate-failures
 
     # First build - C++11 mode and inplace
+    # More-or-less randomly adding -DPYBIND11_SIMPLE_GIL_MANAGEMENT=ON here.
     - name: Configure C++11 ${{ matrix.args }}
       run: >
         cmake -S . -B .
         -DPYBIND11_WERROR=ON
+        -DPYBIND11_SIMPLE_GIL_MANAGEMENT=ON
         -DDOWNLOAD_CATCH=ON
         -DDOWNLOAD_EIGEN=ON
         -DCMAKE_CXX_STANDARD=11
@@ -119,7 +121,7 @@ jobs:
 
     - name: C++11 tests
       # TODO: Figure out how to load the DLL on Python 3.8+
-      if: "!(runner.os == 'Windows' && (matrix.python == 3.8 || matrix.python == 3.9 || matrix.python == '3.10' || matrix.python == '3.11-dev' || matrix.python == 'pypy-3.8'))"
+      if: "!(runner.os == 'Windows' && (matrix.python == 3.8 || matrix.python == 3.9 || matrix.python == '3.10' || matrix.python == '3.11' || matrix.python == 'pypy-3.8'))"
       run: cmake --build .  --target cpptest -j 2
 
     - name: Interface test C++11
@@ -129,10 +131,12 @@ jobs:
       run: git clean -fdx
 
     # Second build - C++17 mode and in a build directory
+    # More-or-less randomly adding -DPYBIND11_SIMPLE_GIL_MANAGEMENT=OFF here.
     - name: Configure C++17
       run: >
         cmake -S . -B build2
         -DPYBIND11_WERROR=ON
+        -DPYBIND11_SIMPLE_GIL_MANAGEMENT=OFF
         -DDOWNLOAD_CATCH=ON
         -DDOWNLOAD_EIGEN=ON
         -DCMAKE_CXX_STANDARD=17
@@ -146,7 +150,7 @@ jobs:
 
     - name: C++ tests
       # TODO: Figure out how to load the DLL on Python 3.8+
-      if: "!(runner.os == 'Windows' && (matrix.python == 3.8 || matrix.python == 3.9 || matrix.python == '3.10' || matrix.python == '3.11-dev' || matrix.python == 'pypy-3.8'))"
+      if: "!(runner.os == 'Windows' && (matrix.python == 3.8 || matrix.python == 3.9 || matrix.python == '3.10' || matrix.python == '3.11' || matrix.python == 'pypy-3.8'))"
       run: cmake --build build2 --target cpptest
 
     # Third build - C++17 mode with unstable ABI
@@ -186,7 +190,7 @@ jobs:
         - python-version: "3.9"
           python-debug: true
           valgrind: true
-        - python-version: "3.11-dev"
+        - python-version: "3.11"
           python-debug: false
 
     name: "🐍 ${{ matrix.python-version }}${{ matrix.python-debug && '-dbg' || '' }} (deadsnakes)${{ matrix.valgrind && ' • Valgrind' || '' }} • x64"
@@ -391,7 +395,7 @@ jobs:
   # Testing on CentOS 7 + PGI compilers, which seems to require more workarounds
   centos-nvhpc7:
     runs-on: ubuntu-latest
-    name: "🐍 3 • CentOS7 / PGI 22.3 • x64"
+    name: "🐍 3 • CentOS7 / PGI 22.9 • x64"
     container: centos:7
 
     steps:
@@ -401,7 +405,7 @@ jobs:
       run: yum update -y && yum install -y epel-release && yum install -y git python3-devel make environment-modules cmake3 yum-utils
 
     - name: Install NVidia HPC SDK
-      run: yum-config-manager --add-repo https://developer.download.nvidia.com/hpc-sdk/rhel/nvhpc.repo && yum -y install nvhpc-22.3
+      run: yum-config-manager --add-repo https://developer.download.nvidia.com/hpc-sdk/rhel/nvhpc.repo && yum -y install nvhpc-22.9
 
     # On CentOS 7, we have to filter a few tests (compiler internal error)
     # and allow deeper template recursion (not needed on CentOS 8 with a newer
@@ -411,12 +415,12 @@ jobs:
       shell: bash
       run: |
         source /etc/profile.d/modules.sh
-        module load /opt/nvidia/hpc_sdk/modulefiles/nvhpc/22.3
+        module load /opt/nvidia/hpc_sdk/modulefiles/nvhpc/22.9
         cmake3 -S . -B build -DDOWNLOAD_CATCH=ON \
                             -DCMAKE_CXX_STANDARD=11 \
                             -DPYTHON_EXECUTABLE=$(python3 -c "import sys; print(sys.executable)") \
                             -DCMAKE_CXX_FLAGS="-Wc,--pending_instantiations=0" \
-                            -DPYBIND11_TEST_FILTER="test_smart_ptr.cpp;test_virtual_functions.cpp"
+                            -DPYBIND11_TEST_FILTER="test_smart_ptr.cpp"
 
     # Building before installing Pip should produce a warning but not an error
     - name: Build
@@ -757,7 +761,7 @@ jobs:
       uses: jwlawson/actions-setup-cmake@v1.13
 
     - name: Prepare MSVC
-      uses: ilammy/msvc-dev-cmd@v1.11.0
+      uses: ilammy/msvc-dev-cmd@v1.12.0
       with:
         arch: x86
 
@@ -810,7 +814,7 @@ jobs:
       uses: jwlawson/actions-setup-cmake@v1.13
 
     - name: Prepare MSVC
-      uses: ilammy/msvc-dev-cmd@v1.11.0
+      uses: ilammy/msvc-dev-cmd@v1.12.0
       with:
         arch: x86
 

.github/workflows/labeler.yml

@@ -10,7 +10,11 @@ jobs:
     steps:
 
     - uses: actions/labeler@main
-      if: github.event.pull_request.merged == true
+      if: >
+        github.event.pull_request.merged == true &&
+        !startsWith(github.event.pull_request.title, 'chore(deps):') &&
+        !startsWith(github.event.pull_request.title, 'ci(fix):') &&
+        !startsWith(github.event.pull_request.title, 'docs(changelog):')
       with:
         repo-token: ${{ secrets.GITHUB_TOKEN }}
         configuration-path: .github/labeler_merged.yml

.gitignore

@@ -43,3 +43,4 @@ pybind11Targets.cmake
 /pybind11/share/*
 /docs/_build/*
 .ipynb_checkpoints/
+tests/main.cpp

.pre-commit-config.yaml

@@ -41,7 +41,7 @@ repos:
 
 # Upgrade old Python syntax
 - repo: https://github.com/asottile/pyupgrade
-  rev: "v2.38.2"
+  rev: "v3.2.0"
   hooks:
   - id: pyupgrade
     args: [--py36-plus]
@@ -54,7 +54,7 @@ repos:
 
 # Black, the code formatter, natively supports pre-commit
 - repo: https://github.com/psf/black
-  rev: "22.8.0" # Keep in sync with blacken-docs
+  rev: "22.10.0" # Keep in sync with blacken-docs
   hooks:
   - id: black
 
@@ -64,7 +64,7 @@ repos:
   hooks:
   - id: blacken-docs
     additional_dependencies:
-    - black==22.8.0 # keep in sync with black hook
+    - black==22.10.0 # keep in sync with black hook
 
 # Changes tabs to spaces
 - repo: https://github.com/Lucas-C/pre-commit-hooks
@@ -116,7 +116,7 @@ repos:
 
 # PyLint has native support - not always usable, but works for us
 - repo: https://github.com/PyCQA/pylint
-  rev: "v2.15.3"
+  rev: "v2.15.5"
   hooks:
   - id: pylint
     files: ^pybind11
@@ -132,7 +132,7 @@ repos:
 
 # Check static types with mypy
 - repo: https://github.com/pre-commit/mirrors-mypy
-  rev: "v0.981"
+  rev: "v0.982"
   hooks:
   - id: mypy
     args: []
@@ -152,7 +152,7 @@ repos:
 # Use tools/codespell_ignore_lines_from_errors.py
 # to rebuild .codespell-ignore-lines
 - repo: https://github.com/codespell-project/codespell
-  rev: "v2.2.1"
+  rev: "v2.2.2"
   hooks:
   - id: codespell
     exclude: ".supp$"

CMakeLists.txt

@@ -91,10 +91,16 @@ endif()
 option(PYBIND11_INSTALL "Install pybind11 header files?" ${PYBIND11_MASTER_PROJECT})
 option(PYBIND11_TEST "Build pybind11 test suite?" ${PYBIND11_MASTER_PROJECT})
 option(PYBIND11_NOPYTHON "Disable search for Python" OFF)
+option(PYBIND11_SIMPLE_GIL_MANAGEMENT
+       "Use simpler GIL management logic that does not support disassociation" OFF)
 set(PYBIND11_INTERNALS_VERSION
     ""
     CACHE STRING "Override the ABI version, may be used to enable the unstable ABI.")
 
+if(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+  add_compile_definitions(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+endif()
+
 cmake_dependent_option(
   USE_PYTHON_INCLUDE_DIR
   "Install pybind11 headers in Python include directory instead of default installation prefix"
@@ -120,6 +126,8 @@ set(PYBIND11_HEADERS
     include/pybind11/complex.h
     include/pybind11/options.h
     include/pybind11/eigen.h
+    include/pybind11/eigen/matrix.h
+    include/pybind11/eigen/tensor.h
     include/pybind11/embed.h
     include/pybind11/eval.h
     include/pybind11/gil.h

docs/advanced/exceptions.rst

@@ -177,9 +177,12 @@ section.
     may be explicitly (re-)thrown to delegate it to the other,
     previously-declared existing exception translators.
 
-    Note that ``libc++`` and ``libstdc++`` `behave differently <https://stackoverflow.com/questions/19496643/using-clang-fvisibility-hidden-and-typeinfo-and-type-erasure/28827430>`_
+    Note that ``libc++`` and ``libstdc++`` `behave differently under macOS
-    with ``-fvisibility=hidden``. Therefore exceptions that are used across ABI boundaries need to be explicitly exported, as exercised in ``tests/test_exceptions.h``.
+    <https://stackoverflow.com/questions/19496643/using-clang-fvisibility-hidden-and-typeinfo-and-type-erasure/28827430>`_
-    See also: "Problems with C++ exceptions" under `GCC Wiki <https://gcc.gnu.org/wiki/Visibility>`_.
+    with ``-fvisibility=hidden``. Therefore exceptions that are used across ABI
+    boundaries need to be explicitly exported, as exercised in
+    ``tests/test_exceptions.h``. See also:
+    "Problems with C++ exceptions" under `GCC Wiki <https://gcc.gnu.org/wiki/Visibility>`_.
 
 
 Local vs Global Exception Translators

docs/changelog.rst

@@ -10,12 +10,147 @@ Changes will be added here periodically from the "Suggested changelog entry"
 block in pull request descriptions.
 
 
-
 IN DEVELOPMENT
 --------------
 
 Changes will be summarized here periodically.
 
+Version 2.10.1 (Oct 31, 2022)
+-----------------------------
+
+This is the first version to fully support embedding the newly released Python 3.11.
+
+Changes:
+
+* Allow ``pybind11::capsule`` constructor to take null destructor pointers.
+  `#4221 <https://github.com/pybind/pybind11/pull/4221>`_
+
+* ``embed.h`` was changed so that ``PYTHONPATH`` is used also with Python 3.11
+  (established behavior).
+  `#4119 <https://github.com/pybind/pybind11/pull/4119>`_
+
+* A ``PYBIND11_SIMPLE_GIL_MANAGEMENT`` option was added (cmake, C++ define),
+  along with many additional tests in ``test_gil_scoped.py``. The option may be
+  useful to try when debugging GIL-related issues, to determine if the more
+  complex default implementation is or is not to blame. See #4216 for
+  background. WARNING: Please be careful to not create ODR violations when
+  using the option: everything that is linked together with mutual symbol
+  visibility needs to be rebuilt.
+  `#4216 <https://github.com/pybind/pybind11/pull/4216>`_
+
+* ``PYBIND11_EXPORT_EXCEPTION`` was made non-empty only under macOS. This makes
+  Linux builds safer, and enables the removal of warning suppression pragmas for
+  Windows.
+  `#4298 <https://github.com/pybind/pybind11/pull/4298>`_
+
+Bug fixes:
+
+* Fixed a bug where ``UnicodeDecodeError`` was not propagated from various
+  ``py::str`` ctors when decoding surrogate utf characters.
+  `#4294 <https://github.com/pybind/pybind11/pull/4294>`_
+
+* Revert perfect forwarding for ``make_iterator``. This broke at least one
+  valid use case. May revisit later.
+  `#4234 <https://github.com/pybind/pybind11/pull/4234>`_
+
+* Fix support for safe casts to ``void*`` (regression in 2.10.0).
+  `#4275 <https://github.com/pybind/pybind11/pull/4275>`_
+
+* Fix ``char8_t`` support (regression in 2.9).
+  `#4278 <https://github.com/pybind/pybind11/pull/4278>`_
+
+* Unicode surrogate character in Python exception message leads to process
+  termination in ``error_already_set::what()``.
+  `#4297 <https://github.com/pybind/pybind11/pull/4297>`_
+
+* Fix MSVC 2019 v.1924 & C++14 mode error for ``overload_cast``.
+  `#4188 <https://github.com/pybind/pybind11/pull/4188>`_
+
+* Make augmented assignment operators non-const for the object-api. Behavior
+  was previously broken for augmented assignment operators.
+  `#4065 <https://github.com/pybind/pybind11/pull/4065>`_
+
+* Add proper error checking to C++ bindings for Python list append and insert.
+  `#4208 <https://github.com/pybind/pybind11/pull/4208>`_
+
+* Work-around for Nvidia's CUDA nvcc compiler in versions 11.4.0 - 11.8.0.
+  `#4220 <https://github.com/pybind/pybind11/pull/4220>`_
+
+* A workaround for PyPy was added in the ``py::error_already_set``
+  implementation, related to PR `#1895 <https://github.com/pybind/pybind11/pull/1895>`_
+  released with v2.10.0.
+  `#4079 <https://github.com/pybind/pybind11/pull/4079>`_
+
+* Fixed compiler errors when C++23 ``std::forward_like`` is available.
+  `#4136 <https://github.com/pybind/pybind11/pull/4136>`_
+
+* Properly raise exceptions in contains methods (like when an object in unhashable).
+  `#4209 <https://github.com/pybind/pybind11/pull/4209>`_
+
+* Further improve another error in exception handling.
+  `#4232 <https://github.com/pybind/pybind11/pull/4232>`_
+
+* ``get_local_internals()`` was made compatible with
+  ``finalize_interpreter()``, fixing potential freezes during interpreter
+  finalization.
+  `#4192 <https://github.com/pybind/pybind11/pull/4192>`_
+
+
+Performance and style:
+
+* Reserve space in set and STL map casters if possible. This will prevent
+  unnecessary rehashing / resizing by knowing the number of keys ahead of time
+  for Python to C++ casting. This improvement will greatly speed up the casting
+  of large unordered maps and sets.
+  `#4194 <https://github.com/pybind/pybind11/pull/4194>`_
+
+* GIL RAII scopes are non-copyable to avoid potential bugs.
+  `#4183 <https://github.com/pybind/pybind11/pull/4183>`_
+
+* Explicitly default all relevant ctors for pytypes in the ``PYBIND11_OBJECT``
+  macros and enforce the clang-tidy checks ``modernize-use-equals-default`` in
+  macros as well.
+  `#4017 <https://github.com/pybind/pybind11/pull/4017>`_
+
+* Optimize iterator advancement in C++ bindings.
+  `#4237 <https://github.com/pybind/pybind11/pull/4237>`_
+
+* Use the modern ``PyObject_GenericGetDict`` and ``PyObject_GenericSetDict``
+  for handling dynamic attribute dictionaries.
+  `#4106 <https://github.com/pybind/pybind11/pull/4106>`_
+
+* Document that users should use ``PYBIND11_NAMESPACE`` instead of using ``pybind11`` when
+  opening namespaces. Using namespace declarations and namespace qualification
+  remain the same as ``pybind11``. This is done to ensure consistent symbol
+  visibility.
+  `#4098 <https://github.com/pybind/pybind11/pull/4098>`_
+
+* Mark ``detail::forward_like`` as constexpr.
+  `#4147 <https://github.com/pybind/pybind11/pull/4147>`_
+
+* Optimize unpacking_collector when processing ``arg_v`` arguments.
+  `#4219 <https://github.com/pybind/pybind11/pull/4219>`_
+
+* Optimize casting C++ object to ``None``.
+  `#4269 <https://github.com/pybind/pybind11/pull/4269>`_
+
+
+Build system improvements:
+
+* CMake: revert overwrite behavior, now opt-in with ``PYBIND11_PYTHONLIBS_OVERRWRITE OFF``.
+  `#4195 <https://github.com/pybind/pybind11/pull/4195>`_
+
+* Include a pkg-config file when installing pybind11, such as in the Python
+  package.
+  `#4077 <https://github.com/pybind/pybind11/pull/4077>`_
+
+* Avoid stripping debug symbols when ``CMAKE_BUILD_TYPE`` is set to ``DEBUG``
+  instead of ``Debug``.
+  `#4078 <https://github.com/pybind/pybind11/pull/4078>`_
+
+* Followup to `#3948 <https://github.com/pybind/pybind11/pull/3948>`_, fixing vcpkg again.
+  `#4123 <https://github.com/pybind/pybind11/pull/4123>`_
+
 Version 2.10.0 (Jul 15, 2022)
 -----------------------------
 

include/pybind11/cast.h

@@ -248,7 +248,7 @@ public:
         return false;
     }
     static handle cast(T, return_value_policy /* policy */, handle /* parent */) {
-        return none().inc_ref();
+        return none().release();
     }
     PYBIND11_TYPE_CASTER(T, const_name("None"));
 };
@@ -291,7 +291,7 @@ public:
         if (ptr) {
             return capsule(ptr).release();
         }
-        return none().inc_ref();
+        return none().release();
     }
 
     template <typename T>
@@ -537,7 +537,7 @@ public:
 
     static handle cast(const CharT *src, return_value_policy policy, handle parent) {
         if (src == nullptr) {
-            return pybind11::none().inc_ref();
+            return pybind11::none().release();
         }
         return StringCaster::cast(StringType(src), policy, parent);
     }
@@ -1179,11 +1179,9 @@ enable_if_t<cast_is_temporary_value_reference<T>::value, T> cast_safe(object &&)
     pybind11_fail("Internal error: cast_safe fallback invoked");
 }
 template <typename T>
-enable_if_t<std::is_same<void, intrinsic_t<T>>::value, void> cast_safe(object &&) {}
+enable_if_t<std::is_void<T>::value, void> cast_safe(object &&) {}
 template <typename T>
-enable_if_t<detail::none_of<cast_is_temporary_value_reference<T>,
+enable_if_t<detail::none_of<cast_is_temporary_value_reference<T>, std::is_void<T>>::value, T>
-                            std::is_same<void, intrinsic_t<T>>>::value,
-            T>
 cast_safe(object &&o) {
     return pybind11::cast<T>(std::move(o));
 }

include/pybind11/detail/common.h

@@ -96,13 +96,10 @@
 #endif
 
 #if !defined(PYBIND11_EXPORT_EXCEPTION)
-#    ifdef __MINGW32__
+#    if defined(__apple_build_version__)
-// workaround for:
-// error: 'dllexport' implies default visibility, but xxx has already been declared with a
-// different visibility
-#        define PYBIND11_EXPORT_EXCEPTION
-#    else
 #        define PYBIND11_EXPORT_EXCEPTION PYBIND11_EXPORT
+#    else
+#        define PYBIND11_EXPORT_EXCEPTION
 #    endif
 #endif
 
@@ -205,11 +202,8 @@
 #    endif
 #endif
 
-#if defined(__cpp_lib_char8_t) && __cpp_lib_char8_t >= 201811L
-#    define PYBIND11_HAS_U8STRING
-#endif
-
 #include <Python.h>
+// Reminder: WITH_THREAD is always defined if PY_VERSION_HEX >= 0x03070000
 #if PY_VERSION_HEX < 0x03060000
 #    error "PYTHON < 3.6 IS UNSUPPORTED. pybind11 v2.9 was the last to support Python 2 and 3.5."
 #endif
@@ -233,6 +227,10 @@
 #    undef copysign
 #endif
 
+#if defined(PYPY_VERSION) && !defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+#    define PYBIND11_SIMPLE_GIL_MANAGEMENT
+#endif
+
 #if defined(_MSC_VER)
 #    if defined(PYBIND11_DEBUG_MARKER)
 #        define _DEBUG
@@ -259,6 +257,11 @@
 #    endif
 #endif
 
+// Must be after including <version> or one of the other headers specified by the standard
+#if defined(__cpp_lib_char8_t) && __cpp_lib_char8_t >= 201811L
+#    define PYBIND11_HAS_U8STRING
+#endif
+
 // #define PYBIND11_STR_LEGACY_PERMISSIVE
 // If DEFINED, pybind11::str can hold PyUnicodeObject or PyBytesObject
 //             (probably surprising and never documented, but this was the
@@ -898,12 +901,6 @@ using expand_side_effects = bool[];
 
 PYBIND11_NAMESPACE_END(detail)
 
-#if defined(_MSC_VER)
-#    pragma warning(push)
-#    pragma warning(disable : 4275)
-//     warning C4275: An exported class was derived from a class that wasn't exported.
-//     Can be ignored when derived from a STL class.
-#endif
 /// C++ bindings of builtin Python exceptions
 class PYBIND11_EXPORT_EXCEPTION builtin_exception : public std::runtime_error {
 public:
@@ -911,9 +908,6 @@ public:
     /// Set the error using the Python C API
     virtual void set_error() const = 0;
 };
-#if defined(_MSC_VER)
-#    pragma warning(pop)
-#endif
 
 #define PYBIND11_RUNTIME_EXCEPTION(name, type)                                                    \
     class PYBIND11_EXPORT_EXCEPTION name : public builtin_exception {                             \

include/pybind11/detail/internals.h

@@ -9,6 +9,12 @@
 
 #pragma once
 
+#include "common.h"
+
+#if defined(WITH_THREAD) && defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+#    include "../gil.h"
+#endif
+
 #include "../pytypes.h"
 
 #include <exception>
@@ -37,6 +43,8 @@ using ExceptionTranslator = void (*)(std::exception_ptr);
 
 PYBIND11_NAMESPACE_BEGIN(detail)
 
+constexpr const char *internals_function_record_capsule_name = "pybind11_function_record_capsule";
+
 // Forward declarations
 inline PyTypeObject *make_static_property_type();
 inline PyTypeObject *make_default_metaclass();
@@ -49,7 +57,7 @@ inline PyObject *make_object_base_type(PyTypeObject *metaclass);
 // `Py_LIMITED_API` anyway.
 #    if PYBIND11_INTERNALS_VERSION > 4
 #        define PYBIND11_TLS_KEY_REF Py_tss_t &
-#        ifdef __GNUC__
+#        if defined(__GNUC__) && !defined(__INTEL_COMPILER)
 // Clang on macOS warns due to `Py_tss_NEEDS_INIT` not specifying an initializer
 // for every field.
 #            define PYBIND11_TLS_KEY_INIT(var)                                                    \
@@ -169,11 +177,23 @@ struct internals {
     PyTypeObject *default_metaclass;
     PyObject *instance_base;
 #if defined(WITH_THREAD)
+    // Unused if PYBIND11_SIMPLE_GIL_MANAGEMENT is defined:
     PYBIND11_TLS_KEY_INIT(tstate)
 #    if PYBIND11_INTERNALS_VERSION > 4
     PYBIND11_TLS_KEY_INIT(loader_life_support_tls_key)
 #    endif // PYBIND11_INTERNALS_VERSION > 4
+    // Unused if PYBIND11_SIMPLE_GIL_MANAGEMENT is defined:
     PyInterpreterState *istate = nullptr;
+
+#    if PYBIND11_INTERNALS_VERSION > 4
+    // Note that we have to use a std::string to allocate memory to ensure a unique address
+    // We want unique addresses since we use pointer equality to compare function records
+    std::string function_record_capsule_name = internals_function_record_capsule_name;
+#    endif
+
+    internals() = default;
+    internals(const internals &other) = delete;
+    internals &operator=(const internals &other) = delete;
     ~internals() {
 #    if PYBIND11_INTERNALS_VERSION > 4
         PYBIND11_TLS_FREE(loader_life_support_tls_key);
@@ -408,6 +428,10 @@ PYBIND11_NOINLINE internals &get_internals() {
         return **internals_pp;
     }
 
+#if defined(WITH_THREAD)
+#    if defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+    gil_scoped_acquire gil;
+#    else
     // Ensure that the GIL is held since we will need to make Python calls.
     // Cannot use py::gil_scoped_acquire here since that constructor calls get_internals.
     struct gil_scoped_acquire_local {
@@ -417,6 +441,8 @@ PYBIND11_NOINLINE internals &get_internals() {
         ~gil_scoped_acquire_local() { PyGILState_Release(state); }
         const PyGILState_STATE state;
     } gil;
+#    endif
+#endif
     error_scope err_scope;
 
     PYBIND11_STR_TYPE id(PYBIND11_INTERNALS_ID);
@@ -534,6 +560,25 @@ const char *c_str(Args &&...args) {
     return strings.front().c_str();
 }
 
+inline const char *get_function_record_capsule_name() {
+#if PYBIND11_INTERNALS_VERSION > 4
+    return get_internals().function_record_capsule_name.c_str();
+#else
+    return nullptr;
+#endif
+}
+
+// Determine whether or not the following capsule contains a pybind11 function record.
+// Note that we use `internals` to make sure that only ABI compatible records are touched.
+//
+// This check is currently used in two places:
+// - An important optimization in functional.h to avoid overhead in C++ -> Python -> C++
+// - The sibling feature of cpp_function to allow overloads
+inline bool is_function_record_capsule(const capsule &cap) {
+    // Pointer equality as we rely on internals() to ensure unique pointers
+    return cap.name() == get_function_record_capsule_name();
+}
+
 PYBIND11_NAMESPACE_END(detail)
 
 /// Returns a named pointer that is shared among all extension modules (using the same

include/pybind11/eigen.h

@@ -9,705 +9,4 @@
 
 #pragma once
 
-/* HINT: To suppress warnings originating from the Eigen headers, use -isystem.
+#include "eigen/matrix.h"
-   See also:
-       https://stackoverflow.com/questions/2579576/i-dir-vs-isystem-dir
-       https://stackoverflow.com/questions/1741816/isystem-for-ms-visual-studio-c-compiler
-*/
-
-#include "numpy.h"
-
-// The C4127 suppression was introduced for Eigen 3.4.0. In theory we could
-// make it version specific, or even remove it later, but considering that
-// 1. C4127 is generally far more distracting than useful for modern template code, and
-// 2. we definitely want to ignore any MSVC warnings originating from Eigen code,
-//    it is probably best to keep this around indefinitely.
-#if defined(_MSC_VER)
-#    pragma warning(push)
-#    pragma warning(disable : 4127) // C4127: conditional expression is constant
-#    pragma warning(disable : 5054) // https://github.com/pybind/pybind11/pull/3741
-//       C5054: operator '&': deprecated between enumerations of different types
-#elif defined(__MINGW32__)
-#    pragma GCC diagnostic push
-#    pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
-#endif
-
-#include <Eigen/Core>
-#include <Eigen/SparseCore>
-
-#if defined(_MSC_VER)
-#    pragma warning(pop)
-#elif defined(__MINGW32__)
-#    pragma GCC diagnostic pop
-#endif
-
-// Eigen prior to 3.2.7 doesn't have proper move constructors--but worse, some classes get implicit
-// move constructors that break things.  We could detect this an explicitly copy, but an extra copy
-// of matrices seems highly undesirable.
-static_assert(EIGEN_VERSION_AT_LEAST(3, 2, 7),
-              "Eigen support in pybind11 requires Eigen >= 3.2.7");
-
-PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
-
-// Provide a convenience alias for easier pass-by-ref usage with fully dynamic strides:
-using EigenDStride = Eigen::Stride<Eigen::Dynamic, Eigen::Dynamic>;
-template <typename MatrixType>
-using EigenDRef = Eigen::Ref<MatrixType, 0, EigenDStride>;
-template <typename MatrixType>
-using EigenDMap = Eigen::Map<MatrixType, 0, EigenDStride>;
-
-PYBIND11_NAMESPACE_BEGIN(detail)
-
-#if EIGEN_VERSION_AT_LEAST(3, 3, 0)
-using EigenIndex = Eigen::Index;
-template <typename Scalar, int Flags, typename StorageIndex>
-using EigenMapSparseMatrix = Eigen::Map<Eigen::SparseMatrix<Scalar, Flags, StorageIndex>>;
-#else
-using EigenIndex = EIGEN_DEFAULT_DENSE_INDEX_TYPE;
-template <typename Scalar, int Flags, typename StorageIndex>
-using EigenMapSparseMatrix = Eigen::MappedSparseMatrix<Scalar, Flags, StorageIndex>;
-#endif
-
-// Matches Eigen::Map, Eigen::Ref, blocks, etc:
-template <typename T>
-using is_eigen_dense_map = all_of<is_template_base_of<Eigen::DenseBase, T>,
-                                  std::is_base_of<Eigen::MapBase<T, Eigen::ReadOnlyAccessors>, T>>;
-template <typename T>
-using is_eigen_mutable_map = std::is_base_of<Eigen::MapBase<T, Eigen::WriteAccessors>, T>;
-template <typename T>
-using is_eigen_dense_plain
-    = all_of<negation<is_eigen_dense_map<T>>, is_template_base_of<Eigen::PlainObjectBase, T>>;
-template <typename T>
-using is_eigen_sparse = is_template_base_of<Eigen::SparseMatrixBase, T>;
-// Test for objects inheriting from EigenBase<Derived> that aren't captured by the above.  This
-// basically covers anything that can be assigned to a dense matrix but that don't have a typical
-// matrix data layout that can be copied from their .data().  For example, DiagonalMatrix and
-// SelfAdjointView fall into this category.
-template <typename T>
-using is_eigen_other
-    = all_of<is_template_base_of<Eigen::EigenBase, T>,
-             negation<any_of<is_eigen_dense_map<T>, is_eigen_dense_plain<T>, is_eigen_sparse<T>>>>;
-
-// Captures numpy/eigen conformability status (returned by EigenProps::conformable()):
-template <bool EigenRowMajor>
-struct EigenConformable {
-    bool conformable = false;
-    EigenIndex rows = 0, cols = 0;
-    EigenDStride stride{0, 0};    // Only valid if negativestrides is false!
-    bool negativestrides = false; // If true, do not use stride!
-
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    EigenConformable(bool fits = false) : conformable{fits} {}
-    // Matrix type:
-    EigenConformable(EigenIndex r, EigenIndex c, EigenIndex rstride, EigenIndex cstride)
-        : conformable{true}, rows{r}, cols{c},
-          // TODO: when Eigen bug #747 is fixed, remove the tests for non-negativity.
-          // http://eigen.tuxfamily.org/bz/show_bug.cgi?id=747
-          stride{EigenRowMajor ? (rstride > 0 ? rstride : 0)
-                               : (cstride > 0 ? cstride : 0) /* outer stride */,
-                 EigenRowMajor ? (cstride > 0 ? cstride : 0)
-                               : (rstride > 0 ? rstride : 0) /* inner stride */},
-          negativestrides{rstride < 0 || cstride < 0} {}
-    // Vector type:
-    EigenConformable(EigenIndex r, EigenIndex c, EigenIndex stride)
-        : EigenConformable(r, c, r == 1 ? c * stride : stride, c == 1 ? r : r * stride) {}
-
-    template <typename props>
-    bool stride_compatible() const {
-        // To have compatible strides, we need (on both dimensions) one of fully dynamic strides,
-        // matching strides, or a dimension size of 1 (in which case the stride value is
-        // irrelevant). Alternatively, if any dimension size is 0, the strides are not relevant
-        // (and numpy ≥ 1.23 sets the strides to 0 in that case, so we need to check explicitly).
-        if (negativestrides) {
-            return false;
-        }
-        if (rows == 0 || cols == 0) {
-            return true;
-        }
-        return (props::inner_stride == Eigen::Dynamic || props::inner_stride == stride.inner()
-                || (EigenRowMajor ? cols : rows) == 1)
-               && (props::outer_stride == Eigen::Dynamic || props::outer_stride == stride.outer()
-                   || (EigenRowMajor ? rows : cols) == 1);
-    }
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator bool() const { return conformable; }
-};
-
-template <typename Type>
-struct eigen_extract_stride {
-    using type = Type;
-};
-template <typename PlainObjectType, int MapOptions, typename StrideType>
-struct eigen_extract_stride<Eigen::Map<PlainObjectType, MapOptions, StrideType>> {
-    using type = StrideType;
-};
-template <typename PlainObjectType, int Options, typename StrideType>
-struct eigen_extract_stride<Eigen::Ref<PlainObjectType, Options, StrideType>> {
-    using type = StrideType;
-};
-
-// Helper struct for extracting information from an Eigen type
-template <typename Type_>
-struct EigenProps {
-    using Type = Type_;
-    using Scalar = typename Type::Scalar;
-    using StrideType = typename eigen_extract_stride<Type>::type;
-    static constexpr EigenIndex rows = Type::RowsAtCompileTime, cols = Type::ColsAtCompileTime,
-                                size = Type::SizeAtCompileTime;
-    static constexpr bool row_major = Type::IsRowMajor,
-                          vector
-                          = Type::IsVectorAtCompileTime, // At least one dimension has fixed size 1
-        fixed_rows = rows != Eigen::Dynamic, fixed_cols = cols != Eigen::Dynamic,
-                          fixed = size != Eigen::Dynamic, // Fully-fixed size
-        dynamic = !fixed_rows && !fixed_cols;             // Fully-dynamic size
-
-    template <EigenIndex i, EigenIndex ifzero>
-    using if_zero = std::integral_constant<EigenIndex, i == 0 ? ifzero : i>;
-    static constexpr EigenIndex inner_stride
-        = if_zero<StrideType::InnerStrideAtCompileTime, 1>::value,
-        outer_stride = if_zero < StrideType::OuterStrideAtCompileTime,
-        vector      ? size
-        : row_major ? cols
-                    : rows > ::value;
-    static constexpr bool dynamic_stride
-        = inner_stride == Eigen::Dynamic && outer_stride == Eigen::Dynamic;
-    static constexpr bool requires_row_major
-        = !dynamic_stride && !vector && (row_major ? inner_stride : outer_stride) == 1;
-    static constexpr bool requires_col_major
-        = !dynamic_stride && !vector && (row_major ? outer_stride : inner_stride) == 1;
-
-    // Takes an input array and determines whether we can make it fit into the Eigen type.  If
-    // the array is a vector, we attempt to fit it into either an Eigen 1xN or Nx1 vector
-    // (preferring the latter if it will fit in either, i.e. for a fully dynamic matrix type).
-    static EigenConformable<row_major> conformable(const array &a) {
-        const auto dims = a.ndim();
-        if (dims < 1 || dims > 2) {
-            return false;
-        }
-
-        if (dims == 2) { // Matrix type: require exact match (or dynamic)
-
-            EigenIndex np_rows = a.shape(0), np_cols = a.shape(1),
-                       np_rstride = a.strides(0) / static_cast<ssize_t>(sizeof(Scalar)),
-                       np_cstride = a.strides(1) / static_cast<ssize_t>(sizeof(Scalar));
-            if ((PYBIND11_SILENCE_MSVC_C4127(fixed_rows) && np_rows != rows)
-                || (PYBIND11_SILENCE_MSVC_C4127(fixed_cols) && np_cols != cols)) {
-                return false;
-            }
-
-            return {np_rows, np_cols, np_rstride, np_cstride};
-        }
-
-        // Otherwise we're storing an n-vector.  Only one of the strides will be used, but
-        // whichever is used, we want the (single) numpy stride value.
-        const EigenIndex n = a.shape(0),
-                         stride = a.strides(0) / static_cast<ssize_t>(sizeof(Scalar));
-
-        if (vector) { // Eigen type is a compile-time vector
-            if (PYBIND11_SILENCE_MSVC_C4127(fixed) && size != n) {
-                return false; // Vector size mismatch
-            }
-            return {rows == 1 ? 1 : n, cols == 1 ? 1 : n, stride};
-        }
-        if (fixed) {
-            // The type has a fixed size, but is not a vector: abort
-            return false;
-        }
-        if (fixed_cols) {
-            // Since this isn't a vector, cols must be != 1.  We allow this only if it exactly
-            // equals the number of elements (rows is Dynamic, and so 1 row is allowed).
-            if (cols != n) {
-                return false;
-            }
-            return {1, n, stride};
-        } // Otherwise it's either fully dynamic, or column dynamic; both become a column vector
-        if (PYBIND11_SILENCE_MSVC_C4127(fixed_rows) && rows != n) {
-            return false;
-        }
-        return {n, 1, stride};
-    }
-
-    static constexpr bool show_writeable
-        = is_eigen_dense_map<Type>::value && is_eigen_mutable_map<Type>::value;
-    static constexpr bool show_order = is_eigen_dense_map<Type>::value;
-    static constexpr bool show_c_contiguous = show_order && requires_row_major;
-    static constexpr bool show_f_contiguous
-        = !show_c_contiguous && show_order && requires_col_major;
-
-    static constexpr auto descriptor
-        = const_name("numpy.ndarray[") + npy_format_descriptor<Scalar>::name + const_name("[")
-          + const_name<fixed_rows>(const_name<(size_t) rows>(), const_name("m")) + const_name(", ")
-          + const_name<fixed_cols>(const_name<(size_t) cols>(), const_name("n")) + const_name("]")
-          +
-          // For a reference type (e.g. Ref<MatrixXd>) we have other constraints that might need to
-          // be satisfied: writeable=True (for a mutable reference), and, depending on the map's
-          // stride options, possibly f_contiguous or c_contiguous.  We include them in the
-          // descriptor output to provide some hint as to why a TypeError is occurring (otherwise
-          // it can be confusing to see that a function accepts a 'numpy.ndarray[float64[3,2]]' and
-          // an error message that you *gave* a numpy.ndarray of the right type and dimensions.
-          const_name<show_writeable>(", flags.writeable", "")
-          + const_name<show_c_contiguous>(", flags.c_contiguous", "")
-          + const_name<show_f_contiguous>(", flags.f_contiguous", "") + const_name("]");
-};
-
-// Casts an Eigen type to numpy array.  If given a base, the numpy array references the src data,
-// otherwise it'll make a copy.  writeable lets you turn off the writeable flag for the array.
-template <typename props>
-handle
-eigen_array_cast(typename props::Type const &src, handle base = handle(), bool writeable = true) {
-    constexpr ssize_t elem_size = sizeof(typename props::Scalar);
-    array a;
-    if (props::vector) {
-        a = array({src.size()}, {elem_size * src.innerStride()}, src.data(), base);
-    } else {
-        a = array({src.rows(), src.cols()},
-                  {elem_size * src.rowStride(), elem_size * src.colStride()},
-                  src.data(),
-                  base);
-    }
-
-    if (!writeable) {
-        array_proxy(a.ptr())->flags &= ~detail::npy_api::NPY_ARRAY_WRITEABLE_;
-    }
-
-    return a.release();
-}
-
-// Takes an lvalue ref to some Eigen type and a (python) base object, creating a numpy array that
-// reference the Eigen object's data with `base` as the python-registered base class (if omitted,
-// the base will be set to None, and lifetime management is up to the caller).  The numpy array is
-// non-writeable if the given type is const.
-template <typename props, typename Type>
-handle eigen_ref_array(Type &src, handle parent = none()) {
-    // none here is to get past array's should-we-copy detection, which currently always
-    // copies when there is no base.  Setting the base to None should be harmless.
-    return eigen_array_cast<props>(src, parent, !std::is_const<Type>::value);
-}
-
-// Takes a pointer to some dense, plain Eigen type, builds a capsule around it, then returns a
-// numpy array that references the encapsulated data with a python-side reference to the capsule to
-// tie its destruction to that of any dependent python objects.  Const-ness is determined by
-// whether or not the Type of the pointer given is const.
-template <typename props, typename Type, typename = enable_if_t<is_eigen_dense_plain<Type>::value>>
-handle eigen_encapsulate(Type *src) {
-    capsule base(src, [](void *o) { delete static_cast<Type *>(o); });
-    return eigen_ref_array<props>(*src, base);
-}
-
-// Type caster for regular, dense matrix types (e.g. MatrixXd), but not maps/refs/etc. of dense
-// types.
-template <typename Type>
-struct type_caster<Type, enable_if_t<is_eigen_dense_plain<Type>::value>> {
-    using Scalar = typename Type::Scalar;
-    using props = EigenProps<Type>;
-
-    bool load(handle src, bool convert) {
-        // If we're in no-convert mode, only load if given an array of the correct type
-        if (!convert && !isinstance<array_t<Scalar>>(src)) {
-            return false;
-        }
-
-        // Coerce into an array, but don't do type conversion yet; the copy below handles it.
-        auto buf = array::ensure(src);
-
-        if (!buf) {
-            return false;
-        }
-
-        auto dims = buf.ndim();
-        if (dims < 1 || dims > 2) {
-            return false;
-        }
-
-        auto fits = props::conformable(buf);
-        if (!fits) {
-            return false;
-        }
-
-        // Allocate the new type, then build a numpy reference into it
-        value = Type(fits.rows, fits.cols);
-        auto ref = reinterpret_steal<array>(eigen_ref_array<props>(value));
-        if (dims == 1) {
-            ref = ref.squeeze();
-        } else if (ref.ndim() == 1) {
-            buf = buf.squeeze();
-        }
-
-        int result = detail::npy_api::get().PyArray_CopyInto_(ref.ptr(), buf.ptr());
-
-        if (result < 0) { // Copy failed!
-            PyErr_Clear();
-            return false;
-        }
-
-        return true;
-    }
-
-private:
-    // Cast implementation
-    template <typename CType>
-    static handle cast_impl(CType *src, return_value_policy policy, handle parent) {
-        switch (policy) {
-            case return_value_policy::take_ownership:
-            case return_value_policy::automatic:
-                return eigen_encapsulate<props>(src);
-            case return_value_policy::move:
-                return eigen_encapsulate<props>(new CType(std::move(*src)));
-            case return_value_policy::copy:
-                return eigen_array_cast<props>(*src);
-            case return_value_policy::reference:
-            case return_value_policy::automatic_reference:
-                return eigen_ref_array<props>(*src);
-            case return_value_policy::reference_internal:
-                return eigen_ref_array<props>(*src, parent);
-            default:
-                throw cast_error("unhandled return_value_policy: should not happen!");
-        };
-    }
-
-public:
-    // Normal returned non-reference, non-const value:
-    static handle cast(Type &&src, return_value_policy /* policy */, handle parent) {
-        return cast_impl(&src, return_value_policy::move, parent);
-    }
-    // If you return a non-reference const, we mark the numpy array readonly:
-    static handle cast(const Type &&src, return_value_policy /* policy */, handle parent) {
-        return cast_impl(&src, return_value_policy::move, parent);
-    }
-    // lvalue reference return; default (automatic) becomes copy
-    static handle cast(Type &src, return_value_policy policy, handle parent) {
-        if (policy == return_value_policy::automatic
-            || policy == return_value_policy::automatic_reference) {
-            policy = return_value_policy::copy;
-        }
-        return cast_impl(&src, policy, parent);
-    }
-    // const lvalue reference return; default (automatic) becomes copy
-    static handle cast(const Type &src, return_value_policy policy, handle parent) {
-        if (policy == return_value_policy::automatic
-            || policy == return_value_policy::automatic_reference) {
-            policy = return_value_policy::copy;
-        }
-        return cast(&src, policy, parent);
-    }
-    // non-const pointer return
-    static handle cast(Type *src, return_value_policy policy, handle parent) {
-        return cast_impl(src, policy, parent);
-    }
-    // const pointer return
-    static handle cast(const Type *src, return_value_policy policy, handle parent) {
-        return cast_impl(src, policy, parent);
-    }
-
-    static constexpr auto name = props::descriptor;
-
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator Type *() { return &value; }
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator Type &() { return value; }
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator Type &&() && { return std::move(value); }
-    template <typename T>
-    using cast_op_type = movable_cast_op_type<T>;
-
-private:
-    Type value;
-};
-
-// Base class for casting reference/map/block/etc. objects back to python.
-template <typename MapType>
-struct eigen_map_caster {
-private:
-    using props = EigenProps<MapType>;
-
-public:
-    // Directly referencing a ref/map's data is a bit dangerous (whatever the map/ref points to has
-    // to stay around), but we'll allow it under the assumption that you know what you're doing
-    // (and have an appropriate keep_alive in place).  We return a numpy array pointing directly at
-    // the ref's data (The numpy array ends up read-only if the ref was to a const matrix type.)
-    // Note that this means you need to ensure you don't destroy the object in some other way (e.g.
-    // with an appropriate keep_alive, or with a reference to a statically allocated matrix).
-    static handle cast(const MapType &src, return_value_policy policy, handle parent) {
-        switch (policy) {
-            case return_value_policy::copy:
-                return eigen_array_cast<props>(src);
-            case return_value_policy::reference_internal:
-                return eigen_array_cast<props>(src, parent, is_eigen_mutable_map<MapType>::value);
-            case return_value_policy::reference:
-            case return_value_policy::automatic:
-            case return_value_policy::automatic_reference:
-                return eigen_array_cast<props>(src, none(), is_eigen_mutable_map<MapType>::value);
-            default:
-                // move, take_ownership don't make any sense for a ref/map:
-                pybind11_fail("Invalid return_value_policy for Eigen Map/Ref/Block type");
-        }
-    }
-
-    static constexpr auto name = props::descriptor;
-
-    // Explicitly delete these: support python -> C++ conversion on these (i.e. these can be return
-    // types but not bound arguments).  We still provide them (with an explicitly delete) so that
-    // you end up here if you try anyway.
-    bool load(handle, bool) = delete;
-    operator MapType() = delete;
-    template <typename>
-    using cast_op_type = MapType;
-};
-
-// We can return any map-like object (but can only load Refs, specialized next):
-template <typename Type>
-struct type_caster<Type, enable_if_t<is_eigen_dense_map<Type>::value>> : eigen_map_caster<Type> {};
-
-// Loader for Ref<...> arguments.  See the documentation for info on how to make this work without
-// copying (it requires some extra effort in many cases).
-template <typename PlainObjectType, typename StrideType>
-struct type_caster<
-    Eigen::Ref<PlainObjectType, 0, StrideType>,
-    enable_if_t<is_eigen_dense_map<Eigen::Ref<PlainObjectType, 0, StrideType>>::value>>
-    : public eigen_map_caster<Eigen::Ref<PlainObjectType, 0, StrideType>> {
-private:
-    using Type = Eigen::Ref<PlainObjectType, 0, StrideType>;
-    using props = EigenProps<Type>;
-    using Scalar = typename props::Scalar;
-    using MapType = Eigen::Map<PlainObjectType, 0, StrideType>;
-    using Array
-        = array_t<Scalar,
-                  array::forcecast
-                      | ((props::row_major ? props::inner_stride : props::outer_stride) == 1
-                             ? array::c_style
-                         : (props::row_major ? props::outer_stride : props::inner_stride) == 1
-                             ? array::f_style
-                             : 0)>;
-    static constexpr bool need_writeable = is_eigen_mutable_map<Type>::value;
-    // Delay construction (these have no default constructor)
-    std::unique_ptr<MapType> map;
-    std::unique_ptr<Type> ref;
-    // Our array.  When possible, this is just a numpy array pointing to the source data, but
-    // sometimes we can't avoid copying (e.g. input is not a numpy array at all, has an
-    // incompatible layout, or is an array of a type that needs to be converted).  Using a numpy
-    // temporary (rather than an Eigen temporary) saves an extra copy when we need both type
-    // conversion and storage order conversion.  (Note that we refuse to use this temporary copy
-    // when loading an argument for a Ref<M> with M non-const, i.e. a read-write reference).
-    Array copy_or_ref;
-
-public:
-    bool load(handle src, bool convert) {
-        // First check whether what we have is already an array of the right type.  If not, we
-        // can't avoid a copy (because the copy is also going to do type conversion).
-        bool need_copy = !isinstance<Array>(src);
-
-        EigenConformable<props::row_major> fits;
-        if (!need_copy) {
-            // We don't need a converting copy, but we also need to check whether the strides are
-            // compatible with the Ref's stride requirements
-            auto aref = reinterpret_borrow<Array>(src);
-
-            if (aref && (!need_writeable || aref.writeable())) {
-                fits = props::conformable(aref);
-                if (!fits) {
-                    return false; // Incompatible dimensions
-                }
-                if (!fits.template stride_compatible<props>()) {
-                    need_copy = true;
-                } else {
-                    copy_or_ref = std::move(aref);
-                }
-            } else {
-                need_copy = true;
-            }
-        }
-
-        if (need_copy) {
-            // We need to copy: If we need a mutable reference, or we're not supposed to convert
-            // (either because we're in the no-convert overload pass, or because we're explicitly
-            // instructed not to copy (via `py::arg().noconvert()`) we have to fail loading.
-            if (!convert || need_writeable) {
-                return false;
-            }
-
-            Array copy = Array::ensure(src);
-            if (!copy) {
-                return false;
-            }
-            fits = props::conformable(copy);
-            if (!fits || !fits.template stride_compatible<props>()) {
-                return false;
-            }
-            copy_or_ref = std::move(copy);
-            loader_life_support::add_patient(copy_or_ref);
-        }
-
-        ref.reset();
-        map.reset(new MapType(data(copy_or_ref),
-                              fits.rows,
-                              fits.cols,
-                              make_stride(fits.stride.outer(), fits.stride.inner())));
-        ref.reset(new Type(*map));
-
-        return true;
-    }
-
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator Type *() { return ref.get(); }
-    // NOLINTNEXTLINE(google-explicit-constructor)
-    operator Type &() { return *ref; }
-    template <typename _T>
-    using cast_op_type = pybind11::detail::cast_op_type<_T>;
-
-private:
-    template <typename T = Type, enable_if_t<is_eigen_mutable_map<T>::value, int> = 0>
-    Scalar *data(Array &a) {
-        return a.mutable_data();
-    }
-
-    template <typename T = Type, enable_if_t<!is_eigen_mutable_map<T>::value, int> = 0>
-    const Scalar *data(Array &a) {
-        return a.data();
-    }
-
-    // Attempt to figure out a constructor of `Stride` that will work.
-    // If both strides are fixed, use a default constructor:
-    template <typename S>
-    using stride_ctor_default = bool_constant<S::InnerStrideAtCompileTime != Eigen::Dynamic
-                                              && S::OuterStrideAtCompileTime != Eigen::Dynamic
-                                              && std::is_default_constructible<S>::value>;
-    // Otherwise, if there is a two-index constructor, assume it is (outer,inner) like
-    // Eigen::Stride, and use it:
-    template <typename S>
-    using stride_ctor_dual
-        = bool_constant<!stride_ctor_default<S>::value
-                        && std::is_constructible<S, EigenIndex, EigenIndex>::value>;
-    // Otherwise, if there is a one-index constructor, and just one of the strides is dynamic, use
-    // it (passing whichever stride is dynamic).
-    template <typename S>
-    using stride_ctor_outer
-        = bool_constant<!any_of<stride_ctor_default<S>, stride_ctor_dual<S>>::value
-                        && S::OuterStrideAtCompileTime == Eigen::Dynamic
-                        && S::InnerStrideAtCompileTime != Eigen::Dynamic
-                        && std::is_constructible<S, EigenIndex>::value>;
-    template <typename S>
-    using stride_ctor_inner
-        = bool_constant<!any_of<stride_ctor_default<S>, stride_ctor_dual<S>>::value
-                        && S::InnerStrideAtCompileTime == Eigen::Dynamic
-                        && S::OuterStrideAtCompileTime != Eigen::Dynamic
-                        && std::is_constructible<S, EigenIndex>::value>;
-
-    template <typename S = StrideType, enable_if_t<stride_ctor_default<S>::value, int> = 0>
-    static S make_stride(EigenIndex, EigenIndex) {
-        return S();
-    }
-    template <typename S = StrideType, enable_if_t<stride_ctor_dual<S>::value, int> = 0>
-    static S make_stride(EigenIndex outer, EigenIndex inner) {
-        return S(outer, inner);
-    }
-    template <typename S = StrideType, enable_if_t<stride_ctor_outer<S>::value, int> = 0>
-    static S make_stride(EigenIndex outer, EigenIndex) {
-        return S(outer);
-    }
-    template <typename S = StrideType, enable_if_t<stride_ctor_inner<S>::value, int> = 0>
-    static S make_stride(EigenIndex, EigenIndex inner) {
-        return S(inner);
-    }
-};
-
-// type_caster for special matrix types (e.g. DiagonalMatrix), which are EigenBase, but not
-// EigenDense (i.e. they don't have a data(), at least not with the usual matrix layout).
-// load() is not supported, but we can cast them into the python domain by first copying to a
-// regular Eigen::Matrix, then casting that.
-template <typename Type>
-struct type_caster<Type, enable_if_t<is_eigen_other<Type>::value>> {
-protected:
-    using Matrix
-        = Eigen::Matrix<typename Type::Scalar, Type::RowsAtCompileTime, Type::ColsAtCompileTime>;
-    using props = EigenProps<Matrix>;
-
-public:
-    static handle cast(const Type &src, return_value_policy /* policy */, handle /* parent */) {
-        handle h = eigen_encapsulate<props>(new Matrix(src));
-        return h;
-    }
-    static handle cast(const Type *src, return_value_policy policy, handle parent) {
-        return cast(*src, policy, parent);
-    }
-
-    static constexpr auto name = props::descriptor;
-
-    // Explicitly delete these: support python -> C++ conversion on these (i.e. these can be return
-    // types but not bound arguments).  We still provide them (with an explicitly delete) so that
-    // you end up here if you try anyway.
-    bool load(handle, bool) = delete;
-    operator Type() = delete;
-    template <typename>
-    using cast_op_type = Type;
-};
-
-template <typename Type>
-struct type_caster<Type, enable_if_t<is_eigen_sparse<Type>::value>> {
-    using Scalar = typename Type::Scalar;
-    using StorageIndex = remove_reference_t<decltype(*std::declval<Type>().outerIndexPtr())>;
-    using Index = typename Type::Index;
-    static constexpr bool rowMajor = Type::IsRowMajor;
-
-    bool load(handle src, bool) {
-        if (!src) {
-            return false;
-        }
-
-        auto obj = reinterpret_borrow<object>(src);
-        object sparse_module = module_::import("scipy.sparse");
-        object matrix_type = sparse_module.attr(rowMajor ? "csr_matrix" : "csc_matrix");
-
-        if (!type::handle_of(obj).is(matrix_type)) {
-            try {
-                obj = matrix_type(obj);
-            } catch (const error_already_set &) {
-                return false;
-            }
-        }
-
-        auto values = array_t<Scalar>((object) obj.attr("data"));
-        auto innerIndices = array_t<StorageIndex>((object) obj.attr("indices"));
-        auto outerIndices = array_t<StorageIndex>((object) obj.attr("indptr"));
-        auto shape = pybind11::tuple((pybind11::object) obj.attr("shape"));
-        auto nnz = obj.attr("nnz").cast<Index>();
-
-        if (!values || !innerIndices || !outerIndices) {
-            return false;
-        }
-
-        value = EigenMapSparseMatrix<Scalar,
-                                     Type::Flags &(Eigen::RowMajor | Eigen::ColMajor),
-                                     StorageIndex>(shape[0].cast<Index>(),
-                                                   shape[1].cast<Index>(),
-                                                   std::move(nnz),
-                                                   outerIndices.mutable_data(),
-                                                   innerIndices.mutable_data(),
-                                                   values.mutable_data());
-
-        return true;
-    }
-
-    static handle cast(const Type &src, return_value_policy /* policy */, handle /* parent */) {
-        const_cast<Type &>(src).makeCompressed();
-
-        object matrix_type
-            = module_::import("scipy.sparse").attr(rowMajor ? "csr_matrix" : "csc_matrix");
-
-        array data(src.nonZeros(), src.valuePtr());
-        array outerIndices((rowMajor ? src.rows() : src.cols()) + 1, src.outerIndexPtr());
-        array innerIndices(src.nonZeros(), src.innerIndexPtr());
-
-        return matrix_type(pybind11::make_tuple(
-                               std::move(data), std::move(innerIndices), std::move(outerIndices)),
-                           pybind11::make_tuple(src.rows(), src.cols()))
-            .release();
-    }
-
-    PYBIND11_TYPE_CASTER(Type,
-                         const_name<(Type::IsRowMajor) != 0>("scipy.sparse.csr_matrix[",
-                                                             "scipy.sparse.csc_matrix[")
-                             + npy_format_descriptor<Scalar>::name + const_name("]"));
-};
-
-PYBIND11_NAMESPACE_END(detail)
-PYBIND11_NAMESPACE_END(PYBIND11_NAMESPACE)

include/pybind11/eigen/matrix.h

@@ -0,0 +1,712 @@
+/*
+    pybind11/eigen/matrix.h: Transparent conversion for dense and sparse Eigen matrices
+
+    Copyright (c) 2016 Wenzel Jakob <wenzel.jakob@epfl.ch>
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+#pragma once
+
+#include "../numpy.h"
+
+/* HINT: To suppress warnings originating from the Eigen headers, use -isystem.
+   See also:
+       https://stackoverflow.com/questions/2579576/i-dir-vs-isystem-dir
+       https://stackoverflow.com/questions/1741816/isystem-for-ms-visual-studio-c-compiler
+*/
+// The C4127 suppression was introduced for Eigen 3.4.0. In theory we could
+// make it version specific, or even remove it later, but considering that
+// 1. C4127 is generally far more distracting than useful for modern template code, and
+// 2. we definitely want to ignore any MSVC warnings originating from Eigen code,
+//    it is probably best to keep this around indefinitely.
+#if defined(_MSC_VER)
+#    pragma warning(push)
+#    pragma warning(disable : 4127) // C4127: conditional expression is constant
+#    pragma warning(disable : 5054) // https://github.com/pybind/pybind11/pull/3741
+//       C5054: operator '&': deprecated between enumerations of different types
+#elif defined(__MINGW32__)
+#    pragma GCC diagnostic push
+#    pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
+#endif
+
+#include <Eigen/Core>
+#include <Eigen/SparseCore>
+
+#if defined(_MSC_VER)
+#    pragma warning(pop)
+#elif defined(__MINGW32__)
+#    pragma GCC diagnostic pop
+#endif
+
+// Eigen prior to 3.2.7 doesn't have proper move constructors--but worse, some classes get implicit
+// move constructors that break things.  We could detect this an explicitly copy, but an extra copy
+// of matrices seems highly undesirable.
+static_assert(EIGEN_VERSION_AT_LEAST(3, 2, 7),
+              "Eigen matrix support in pybind11 requires Eigen >= 3.2.7");
+
+PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
+
+// Provide a convenience alias for easier pass-by-ref usage with fully dynamic strides:
+using EigenDStride = Eigen::Stride<Eigen::Dynamic, Eigen::Dynamic>;
+template <typename MatrixType>
+using EigenDRef = Eigen::Ref<MatrixType, 0, EigenDStride>;
+template <typename MatrixType>
+using EigenDMap = Eigen::Map<MatrixType, 0, EigenDStride>;
+
+PYBIND11_NAMESPACE_BEGIN(detail)
+
+#if EIGEN_VERSION_AT_LEAST(3, 3, 0)
+using EigenIndex = Eigen::Index;
+template <typename Scalar, int Flags, typename StorageIndex>
+using EigenMapSparseMatrix = Eigen::Map<Eigen::SparseMatrix<Scalar, Flags, StorageIndex>>;
+#else
+using EigenIndex = EIGEN_DEFAULT_DENSE_INDEX_TYPE;
+template <typename Scalar, int Flags, typename StorageIndex>
+using EigenMapSparseMatrix = Eigen::MappedSparseMatrix<Scalar, Flags, StorageIndex>;
+#endif
+
+// Matches Eigen::Map, Eigen::Ref, blocks, etc:
+template <typename T>
+using is_eigen_dense_map = all_of<is_template_base_of<Eigen::DenseBase, T>,
+                                  std::is_base_of<Eigen::MapBase<T, Eigen::ReadOnlyAccessors>, T>>;
+template <typename T>
+using is_eigen_mutable_map = std::is_base_of<Eigen::MapBase<T, Eigen::WriteAccessors>, T>;
+template <typename T>
+using is_eigen_dense_plain
+    = all_of<negation<is_eigen_dense_map<T>>, is_template_base_of<Eigen::PlainObjectBase, T>>;
+template <typename T>
+using is_eigen_sparse = is_template_base_of<Eigen::SparseMatrixBase, T>;
+// Test for objects inheriting from EigenBase<Derived> that aren't captured by the above.  This
+// basically covers anything that can be assigned to a dense matrix but that don't have a typical
+// matrix data layout that can be copied from their .data().  For example, DiagonalMatrix and
+// SelfAdjointView fall into this category.
+template <typename T>
+using is_eigen_other
+    = all_of<is_template_base_of<Eigen::EigenBase, T>,
+             negation<any_of<is_eigen_dense_map<T>, is_eigen_dense_plain<T>, is_eigen_sparse<T>>>>;
+
+// Captures numpy/eigen conformability status (returned by EigenProps::conformable()):
+template <bool EigenRowMajor>
+struct EigenConformable {
+    bool conformable = false;
+    EigenIndex rows = 0, cols = 0;
+    EigenDStride stride{0, 0};    // Only valid if negativestrides is false!
+    bool negativestrides = false; // If true, do not use stride!
+
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    EigenConformable(bool fits = false) : conformable{fits} {}
+    // Matrix type:
+    EigenConformable(EigenIndex r, EigenIndex c, EigenIndex rstride, EigenIndex cstride)
+        : conformable{true}, rows{r}, cols{c},
+          // TODO: when Eigen bug #747 is fixed, remove the tests for non-negativity.
+          // http://eigen.tuxfamily.org/bz/show_bug.cgi?id=747
+          stride{EigenRowMajor ? (rstride > 0 ? rstride : 0)
+                               : (cstride > 0 ? cstride : 0) /* outer stride */,
+                 EigenRowMajor ? (cstride > 0 ? cstride : 0)
+                               : (rstride > 0 ? rstride : 0) /* inner stride */},
+          negativestrides{rstride < 0 || cstride < 0} {}
+    // Vector type:
+    EigenConformable(EigenIndex r, EigenIndex c, EigenIndex stride)
+        : EigenConformable(r, c, r == 1 ? c * stride : stride, c == 1 ? r : r * stride) {}
+
+    template <typename props>
+    bool stride_compatible() const {
+        // To have compatible strides, we need (on both dimensions) one of fully dynamic strides,
+        // matching strides, or a dimension size of 1 (in which case the stride value is
+        // irrelevant). Alternatively, if any dimension size is 0, the strides are not relevant
+        // (and numpy ≥ 1.23 sets the strides to 0 in that case, so we need to check explicitly).
+        if (negativestrides) {
+            return false;
+        }
+        if (rows == 0 || cols == 0) {
+            return true;
+        }
+        return (props::inner_stride == Eigen::Dynamic || props::inner_stride == stride.inner()
+                || (EigenRowMajor ? cols : rows) == 1)
+               && (props::outer_stride == Eigen::Dynamic || props::outer_stride == stride.outer()
+                   || (EigenRowMajor ? rows : cols) == 1);
+    }
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator bool() const { return conformable; }
+};
+
+template <typename Type>
+struct eigen_extract_stride {
+    using type = Type;
+};
+template <typename PlainObjectType, int MapOptions, typename StrideType>
+struct eigen_extract_stride<Eigen::Map<PlainObjectType, MapOptions, StrideType>> {
+    using type = StrideType;
+};
+template <typename PlainObjectType, int Options, typename StrideType>
+struct eigen_extract_stride<Eigen::Ref<PlainObjectType, Options, StrideType>> {
+    using type = StrideType;
+};
+
+// Helper struct for extracting information from an Eigen type
+template <typename Type_>
+struct EigenProps {
+    using Type = Type_;
+    using Scalar = typename Type::Scalar;
+    using StrideType = typename eigen_extract_stride<Type>::type;
+    static constexpr EigenIndex rows = Type::RowsAtCompileTime, cols = Type::ColsAtCompileTime,
+                                size = Type::SizeAtCompileTime;
+    static constexpr bool row_major = Type::IsRowMajor,
+                          vector
+                          = Type::IsVectorAtCompileTime, // At least one dimension has fixed size 1
+        fixed_rows = rows != Eigen::Dynamic, fixed_cols = cols != Eigen::Dynamic,
+                          fixed = size != Eigen::Dynamic, // Fully-fixed size
+        dynamic = !fixed_rows && !fixed_cols;             // Fully-dynamic size
+
+    template <EigenIndex i, EigenIndex ifzero>
+    using if_zero = std::integral_constant<EigenIndex, i == 0 ? ifzero : i>;
+    static constexpr EigenIndex inner_stride
+        = if_zero<StrideType::InnerStrideAtCompileTime, 1>::value,
+        outer_stride = if_zero < StrideType::OuterStrideAtCompileTime,
+        vector      ? size
+        : row_major ? cols
+                    : rows > ::value;
+    static constexpr bool dynamic_stride
+        = inner_stride == Eigen::Dynamic && outer_stride == Eigen::Dynamic;
+    static constexpr bool requires_row_major
+        = !dynamic_stride && !vector && (row_major ? inner_stride : outer_stride) == 1;
+    static constexpr bool requires_col_major
+        = !dynamic_stride && !vector && (row_major ? outer_stride : inner_stride) == 1;
+
+    // Takes an input array and determines whether we can make it fit into the Eigen type.  If
+    // the array is a vector, we attempt to fit it into either an Eigen 1xN or Nx1 vector
+    // (preferring the latter if it will fit in either, i.e. for a fully dynamic matrix type).
+    static EigenConformable<row_major> conformable(const array &a) {
+        const auto dims = a.ndim();
+        if (dims < 1 || dims > 2) {
+            return false;
+        }
+
+        if (dims == 2) { // Matrix type: require exact match (or dynamic)
+
+            EigenIndex np_rows = a.shape(0), np_cols = a.shape(1),
+                       np_rstride = a.strides(0) / static_cast<ssize_t>(sizeof(Scalar)),
+                       np_cstride = a.strides(1) / static_cast<ssize_t>(sizeof(Scalar));
+            if ((PYBIND11_SILENCE_MSVC_C4127(fixed_rows) && np_rows != rows)
+                || (PYBIND11_SILENCE_MSVC_C4127(fixed_cols) && np_cols != cols)) {
+                return false;
+            }
+
+            return {np_rows, np_cols, np_rstride, np_cstride};
+        }
+
+        // Otherwise we're storing an n-vector.  Only one of the strides will be used, but
+        // whichever is used, we want the (single) numpy stride value.
+        const EigenIndex n = a.shape(0),
+                         stride = a.strides(0) / static_cast<ssize_t>(sizeof(Scalar));
+
+        if (vector) { // Eigen type is a compile-time vector
+            if (PYBIND11_SILENCE_MSVC_C4127(fixed) && size != n) {
+                return false; // Vector size mismatch
+            }
+            return {rows == 1 ? 1 : n, cols == 1 ? 1 : n, stride};
+        }
+        if (fixed) {
+            // The type has a fixed size, but is not a vector: abort
+            return false;
+        }
+        if (fixed_cols) {
+            // Since this isn't a vector, cols must be != 1.  We allow this only if it exactly
+            // equals the number of elements (rows is Dynamic, and so 1 row is allowed).
+            if (cols != n) {
+                return false;
+            }
+            return {1, n, stride};
+        } // Otherwise it's either fully dynamic, or column dynamic; both become a column vector
+        if (PYBIND11_SILENCE_MSVC_C4127(fixed_rows) && rows != n) {
+            return false;
+        }
+        return {n, 1, stride};
+    }
+
+    static constexpr bool show_writeable
+        = is_eigen_dense_map<Type>::value && is_eigen_mutable_map<Type>::value;
+    static constexpr bool show_order = is_eigen_dense_map<Type>::value;
+    static constexpr bool show_c_contiguous = show_order && requires_row_major;
+    static constexpr bool show_f_contiguous
+        = !show_c_contiguous && show_order && requires_col_major;
+
+    static constexpr auto descriptor
+        = const_name("numpy.ndarray[") + npy_format_descriptor<Scalar>::name + const_name("[")
+          + const_name<fixed_rows>(const_name<(size_t) rows>(), const_name("m")) + const_name(", ")
+          + const_name<fixed_cols>(const_name<(size_t) cols>(), const_name("n")) + const_name("]")
+          +
+          // For a reference type (e.g. Ref<MatrixXd>) we have other constraints that might need to
+          // be satisfied: writeable=True (for a mutable reference), and, depending on the map's
+          // stride options, possibly f_contiguous or c_contiguous.  We include them in the
+          // descriptor output to provide some hint as to why a TypeError is occurring (otherwise
+          // it can be confusing to see that a function accepts a 'numpy.ndarray[float64[3,2]]' and
+          // an error message that you *gave* a numpy.ndarray of the right type and dimensions.
+          const_name<show_writeable>(", flags.writeable", "")
+          + const_name<show_c_contiguous>(", flags.c_contiguous", "")
+          + const_name<show_f_contiguous>(", flags.f_contiguous", "") + const_name("]");
+};
+
+// Casts an Eigen type to numpy array.  If given a base, the numpy array references the src data,
+// otherwise it'll make a copy.  writeable lets you turn off the writeable flag for the array.
+template <typename props>
+handle
+eigen_array_cast(typename props::Type const &src, handle base = handle(), bool writeable = true) {
+    constexpr ssize_t elem_size = sizeof(typename props::Scalar);
+    array a;
+    if (props::vector) {
+        a = array({src.size()}, {elem_size * src.innerStride()}, src.data(), base);
+    } else {
+        a = array({src.rows(), src.cols()},
+                  {elem_size * src.rowStride(), elem_size * src.colStride()},
+                  src.data(),
+                  base);
+    }
+
+    if (!writeable) {
+        array_proxy(a.ptr())->flags &= ~detail::npy_api::NPY_ARRAY_WRITEABLE_;
+    }
+
+    return a.release();
+}
+
+// Takes an lvalue ref to some Eigen type and a (python) base object, creating a numpy array that
+// reference the Eigen object's data with `base` as the python-registered base class (if omitted,
+// the base will be set to None, and lifetime management is up to the caller).  The numpy array is
+// non-writeable if the given type is const.
+template <typename props, typename Type>
+handle eigen_ref_array(Type &src, handle parent = none()) {
+    // none here is to get past array's should-we-copy detection, which currently always
+    // copies when there is no base.  Setting the base to None should be harmless.
+    return eigen_array_cast<props>(src, parent, !std::is_const<Type>::value);
+}
+
+// Takes a pointer to some dense, plain Eigen type, builds a capsule around it, then returns a
+// numpy array that references the encapsulated data with a python-side reference to the capsule to
+// tie its destruction to that of any dependent python objects.  Const-ness is determined by
+// whether or not the Type of the pointer given is const.
+template <typename props, typename Type, typename = enable_if_t<is_eigen_dense_plain<Type>::value>>
+handle eigen_encapsulate(Type *src) {
+    capsule base(src, [](void *o) { delete static_cast<Type *>(o); });
+    return eigen_ref_array<props>(*src, base);
+}
+
+// Type caster for regular, dense matrix types (e.g. MatrixXd), but not maps/refs/etc. of dense
+// types.
+template <typename Type>
+struct type_caster<Type, enable_if_t<is_eigen_dense_plain<Type>::value>> {
+    using Scalar = typename Type::Scalar;
+    using props = EigenProps<Type>;
+
+    bool load(handle src, bool convert) {
+        // If we're in no-convert mode, only load if given an array of the correct type
+        if (!convert && !isinstance<array_t<Scalar>>(src)) {
+            return false;
+        }
+
+        // Coerce into an array, but don't do type conversion yet; the copy below handles it.
+        auto buf = array::ensure(src);
+
+        if (!buf) {
+            return false;
+        }
+
+        auto dims = buf.ndim();
+        if (dims < 1 || dims > 2) {
+            return false;
+        }
+
+        auto fits = props::conformable(buf);
+        if (!fits) {
+            return false;
+        }
+
+        // Allocate the new type, then build a numpy reference into it
+        value = Type(fits.rows, fits.cols);
+        auto ref = reinterpret_steal<array>(eigen_ref_array<props>(value));
+        if (dims == 1) {
+            ref = ref.squeeze();
+        } else if (ref.ndim() == 1) {
+            buf = buf.squeeze();
+        }
+
+        int result = detail::npy_api::get().PyArray_CopyInto_(ref.ptr(), buf.ptr());
+
+        if (result < 0) { // Copy failed!
+            PyErr_Clear();
+            return false;
+        }
+
+        return true;
+    }
+
+private:
+    // Cast implementation
+    template <typename CType>
+    static handle cast_impl(CType *src, return_value_policy policy, handle parent) {
+        switch (policy) {
+            case return_value_policy::take_ownership:
+            case return_value_policy::automatic:
+                return eigen_encapsulate<props>(src);
+            case return_value_policy::move:
+                return eigen_encapsulate<props>(new CType(std::move(*src)));
+            case return_value_policy::copy:
+                return eigen_array_cast<props>(*src);
+            case return_value_policy::reference:
+            case return_value_policy::automatic_reference:
+                return eigen_ref_array<props>(*src);
+            case return_value_policy::reference_internal:
+                return eigen_ref_array<props>(*src, parent);
+            default:
+                throw cast_error("unhandled return_value_policy: should not happen!");
+        };
+    }
+
+public:
+    // Normal returned non-reference, non-const value:
+    static handle cast(Type &&src, return_value_policy /* policy */, handle parent) {
+        return cast_impl(&src, return_value_policy::move, parent);
+    }
+    // If you return a non-reference const, we mark the numpy array readonly:
+    static handle cast(const Type &&src, return_value_policy /* policy */, handle parent) {
+        return cast_impl(&src, return_value_policy::move, parent);
+    }
+    // lvalue reference return; default (automatic) becomes copy
+    static handle cast(Type &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast_impl(&src, policy, parent);
+    }
+    // const lvalue reference return; default (automatic) becomes copy
+    static handle cast(const Type &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast(&src, policy, parent);
+    }
+    // non-const pointer return
+    static handle cast(Type *src, return_value_policy policy, handle parent) {
+        return cast_impl(src, policy, parent);
+    }
+    // const pointer return
+    static handle cast(const Type *src, return_value_policy policy, handle parent) {
+        return cast_impl(src, policy, parent);
+    }
+
+    static constexpr auto name = props::descriptor;
+
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator Type *() { return &value; }
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator Type &() { return value; }
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator Type &&() && { return std::move(value); }
+    template <typename T>
+    using cast_op_type = movable_cast_op_type<T>;
+
+private:
+    Type value;
+};
+
+// Base class for casting reference/map/block/etc. objects back to python.
+template <typename MapType>
+struct eigen_map_caster {
+private:
+    using props = EigenProps<MapType>;
+
+public:
+    // Directly referencing a ref/map's data is a bit dangerous (whatever the map/ref points to has
+    // to stay around), but we'll allow it under the assumption that you know what you're doing
+    // (and have an appropriate keep_alive in place).  We return a numpy array pointing directly at
+    // the ref's data (The numpy array ends up read-only if the ref was to a const matrix type.)
+    // Note that this means you need to ensure you don't destroy the object in some other way (e.g.
+    // with an appropriate keep_alive, or with a reference to a statically allocated matrix).
+    static handle cast(const MapType &src, return_value_policy policy, handle parent) {
+        switch (policy) {
+            case return_value_policy::copy:
+                return eigen_array_cast<props>(src);
+            case return_value_policy::reference_internal:
+                return eigen_array_cast<props>(src, parent, is_eigen_mutable_map<MapType>::value);
+            case return_value_policy::reference:
+            case return_value_policy::automatic:
+            case return_value_policy::automatic_reference:
+                return eigen_array_cast<props>(src, none(), is_eigen_mutable_map<MapType>::value);
+            default:
+                // move, take_ownership don't make any sense for a ref/map:
+                pybind11_fail("Invalid return_value_policy for Eigen Map/Ref/Block type");
+        }
+    }
+
+    static constexpr auto name = props::descriptor;
+
+    // Explicitly delete these: support python -> C++ conversion on these (i.e. these can be return
+    // types but not bound arguments).  We still provide them (with an explicitly delete) so that
+    // you end up here if you try anyway.
+    bool load(handle, bool) = delete;
+    operator MapType() = delete;
+    template <typename>
+    using cast_op_type = MapType;
+};
+
+// We can return any map-like object (but can only load Refs, specialized next):
+template <typename Type>
+struct type_caster<Type, enable_if_t<is_eigen_dense_map<Type>::value>> : eigen_map_caster<Type> {};
+
+// Loader for Ref<...> arguments.  See the documentation for info on how to make this work without
+// copying (it requires some extra effort in many cases).
+template <typename PlainObjectType, typename StrideType>
+struct type_caster<
+    Eigen::Ref<PlainObjectType, 0, StrideType>,
+    enable_if_t<is_eigen_dense_map<Eigen::Ref<PlainObjectType, 0, StrideType>>::value>>
+    : public eigen_map_caster<Eigen::Ref<PlainObjectType, 0, StrideType>> {
+private:
+    using Type = Eigen::Ref<PlainObjectType, 0, StrideType>;
+    using props = EigenProps<Type>;
+    using Scalar = typename props::Scalar;
+    using MapType = Eigen::Map<PlainObjectType, 0, StrideType>;
+    using Array
+        = array_t<Scalar,
+                  array::forcecast
+                      | ((props::row_major ? props::inner_stride : props::outer_stride) == 1
+                             ? array::c_style
+                         : (props::row_major ? props::outer_stride : props::inner_stride) == 1
+                             ? array::f_style
+                             : 0)>;
+    static constexpr bool need_writeable = is_eigen_mutable_map<Type>::value;
+    // Delay construction (these have no default constructor)
+    std::unique_ptr<MapType> map;
+    std::unique_ptr<Type> ref;
+    // Our array.  When possible, this is just a numpy array pointing to the source data, but
+    // sometimes we can't avoid copying (e.g. input is not a numpy array at all, has an
+    // incompatible layout, or is an array of a type that needs to be converted).  Using a numpy
+    // temporary (rather than an Eigen temporary) saves an extra copy when we need both type
+    // conversion and storage order conversion.  (Note that we refuse to use this temporary copy
+    // when loading an argument for a Ref<M> with M non-const, i.e. a read-write reference).
+    Array copy_or_ref;
+
+public:
+    bool load(handle src, bool convert) {
+        // First check whether what we have is already an array of the right type.  If not, we
+        // can't avoid a copy (because the copy is also going to do type conversion).
+        bool need_copy = !isinstance<Array>(src);
+
+        EigenConformable<props::row_major> fits;
+        if (!need_copy) {
+            // We don't need a converting copy, but we also need to check whether the strides are
+            // compatible with the Ref's stride requirements
+            auto aref = reinterpret_borrow<Array>(src);
+
+            if (aref && (!need_writeable || aref.writeable())) {
+                fits = props::conformable(aref);
+                if (!fits) {
+                    return false; // Incompatible dimensions
+                }
+                if (!fits.template stride_compatible<props>()) {
+                    need_copy = true;
+                } else {
+                    copy_or_ref = std::move(aref);
+                }
+            } else {
+                need_copy = true;
+            }
+        }
+
+        if (need_copy) {
+            // We need to copy: If we need a mutable reference, or we're not supposed to convert
+            // (either because we're in the no-convert overload pass, or because we're explicitly
+            // instructed not to copy (via `py::arg().noconvert()`) we have to fail loading.
+            if (!convert || need_writeable) {
+                return false;
+            }
+
+            Array copy = Array::ensure(src);
+            if (!copy) {
+                return false;
+            }
+            fits = props::conformable(copy);
+            if (!fits || !fits.template stride_compatible<props>()) {
+                return false;
+            }
+            copy_or_ref = std::move(copy);
+            loader_life_support::add_patient(copy_or_ref);
+        }
+
+        ref.reset();
+        map.reset(new MapType(data(copy_or_ref),
+                              fits.rows,
+                              fits.cols,
+                              make_stride(fits.stride.outer(), fits.stride.inner())));
+        ref.reset(new Type(*map));
+
+        return true;
+    }
+
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator Type *() { return ref.get(); }
+    // NOLINTNEXTLINE(google-explicit-constructor)
+    operator Type &() { return *ref; }
+    template <typename _T>
+    using cast_op_type = pybind11::detail::cast_op_type<_T>;
+
+private:
+    template <typename T = Type, enable_if_t<is_eigen_mutable_map<T>::value, int> = 0>
+    Scalar *data(Array &a) {
+        return a.mutable_data();
+    }
+
+    template <typename T = Type, enable_if_t<!is_eigen_mutable_map<T>::value, int> = 0>
+    const Scalar *data(Array &a) {
+        return a.data();
+    }
+
+    // Attempt to figure out a constructor of `Stride` that will work.
+    // If both strides are fixed, use a default constructor:
+    template <typename S>
+    using stride_ctor_default = bool_constant<S::InnerStrideAtCompileTime != Eigen::Dynamic
+                                              && S::OuterStrideAtCompileTime != Eigen::Dynamic
+                                              && std::is_default_constructible<S>::value>;
+    // Otherwise, if there is a two-index constructor, assume it is (outer,inner) like
+    // Eigen::Stride, and use it:
+    template <typename S>
+    using stride_ctor_dual
+        = bool_constant<!stride_ctor_default<S>::value
+                        && std::is_constructible<S, EigenIndex, EigenIndex>::value>;
+    // Otherwise, if there is a one-index constructor, and just one of the strides is dynamic, use
+    // it (passing whichever stride is dynamic).
+    template <typename S>
+    using stride_ctor_outer
+        = bool_constant<!any_of<stride_ctor_default<S>, stride_ctor_dual<S>>::value
+                        && S::OuterStrideAtCompileTime == Eigen::Dynamic
+                        && S::InnerStrideAtCompileTime != Eigen::Dynamic
+                        && std::is_constructible<S, EigenIndex>::value>;
+    template <typename S>
+    using stride_ctor_inner
+        = bool_constant<!any_of<stride_ctor_default<S>, stride_ctor_dual<S>>::value
+                        && S::InnerStrideAtCompileTime == Eigen::Dynamic
+                        && S::OuterStrideAtCompileTime != Eigen::Dynamic
+                        && std::is_constructible<S, EigenIndex>::value>;
+
+    template <typename S = StrideType, enable_if_t<stride_ctor_default<S>::value, int> = 0>
+    static S make_stride(EigenIndex, EigenIndex) {
+        return S();
+    }
+    template <typename S = StrideType, enable_if_t<stride_ctor_dual<S>::value, int> = 0>
+    static S make_stride(EigenIndex outer, EigenIndex inner) {
+        return S(outer, inner);
+    }
+    template <typename S = StrideType, enable_if_t<stride_ctor_outer<S>::value, int> = 0>
+    static S make_stride(EigenIndex outer, EigenIndex) {
+        return S(outer);
+    }
+    template <typename S = StrideType, enable_if_t<stride_ctor_inner<S>::value, int> = 0>
+    static S make_stride(EigenIndex, EigenIndex inner) {
+        return S(inner);
+    }
+};
+
+// type_caster for special matrix types (e.g. DiagonalMatrix), which are EigenBase, but not
+// EigenDense (i.e. they don't have a data(), at least not with the usual matrix layout).
+// load() is not supported, but we can cast them into the python domain by first copying to a
+// regular Eigen::Matrix, then casting that.
+template <typename Type>
+struct type_caster<Type, enable_if_t<is_eigen_other<Type>::value>> {
+protected:
+    using Matrix
+        = Eigen::Matrix<typename Type::Scalar, Type::RowsAtCompileTime, Type::ColsAtCompileTime>;
+    using props = EigenProps<Matrix>;
+
+public:
+    static handle cast(const Type &src, return_value_policy /* policy */, handle /* parent */) {
+        handle h = eigen_encapsulate<props>(new Matrix(src));
+        return h;
+    }
+    static handle cast(const Type *src, return_value_policy policy, handle parent) {
+        return cast(*src, policy, parent);
+    }
+
+    static constexpr auto name = props::descriptor;
+
+    // Explicitly delete these: support python -> C++ conversion on these (i.e. these can be return
+    // types but not bound arguments).  We still provide them (with an explicitly delete) so that
+    // you end up here if you try anyway.
+    bool load(handle, bool) = delete;
+    operator Type() = delete;
+    template <typename>
+    using cast_op_type = Type;
+};
+
+template <typename Type>
+struct type_caster<Type, enable_if_t<is_eigen_sparse<Type>::value>> {
+    using Scalar = typename Type::Scalar;
+    using StorageIndex = remove_reference_t<decltype(*std::declval<Type>().outerIndexPtr())>;
+    using Index = typename Type::Index;
+    static constexpr bool rowMajor = Type::IsRowMajor;
+
+    bool load(handle src, bool) {
+        if (!src) {
+            return false;
+        }
+
+        auto obj = reinterpret_borrow<object>(src);
+        object sparse_module = module_::import("scipy.sparse");
+        object matrix_type = sparse_module.attr(rowMajor ? "csr_matrix" : "csc_matrix");
+
+        if (!type::handle_of(obj).is(matrix_type)) {
+            try {
+                obj = matrix_type(obj);
+            } catch (const error_already_set &) {
+                return false;
+            }
+        }
+
+        auto values = array_t<Scalar>((object) obj.attr("data"));
+        auto innerIndices = array_t<StorageIndex>((object) obj.attr("indices"));
+        auto outerIndices = array_t<StorageIndex>((object) obj.attr("indptr"));
+        auto shape = pybind11::tuple((pybind11::object) obj.attr("shape"));
+        auto nnz = obj.attr("nnz").cast<Index>();
+
+        if (!values || !innerIndices || !outerIndices) {
+            return false;
+        }
+
+        value = EigenMapSparseMatrix<Scalar,
+                                     Type::Flags &(Eigen::RowMajor | Eigen::ColMajor),
+                                     StorageIndex>(shape[0].cast<Index>(),
+                                                   shape[1].cast<Index>(),
+                                                   std::move(nnz),
+                                                   outerIndices.mutable_data(),
+                                                   innerIndices.mutable_data(),
+                                                   values.mutable_data());
+
+        return true;
+    }
+
+    static handle cast(const Type &src, return_value_policy /* policy */, handle /* parent */) {
+        const_cast<Type &>(src).makeCompressed();
+
+        object matrix_type
+            = module_::import("scipy.sparse").attr(rowMajor ? "csr_matrix" : "csc_matrix");
+
+        array data(src.nonZeros(), src.valuePtr());
+        array outerIndices((rowMajor ? src.rows() : src.cols()) + 1, src.outerIndexPtr());
+        array innerIndices(src.nonZeros(), src.innerIndexPtr());
+
+        return matrix_type(pybind11::make_tuple(
+                               std::move(data), std::move(innerIndices), std::move(outerIndices)),
+                           pybind11::make_tuple(src.rows(), src.cols()))
+            .release();
+    }
+
+    PYBIND11_TYPE_CASTER(Type,
+                         const_name<(Type::IsRowMajor) != 0>("scipy.sparse.csr_matrix[",
+                                                             "scipy.sparse.csc_matrix[")
+                             + npy_format_descriptor<Scalar>::name + const_name("]"));
+};
+
+PYBIND11_NAMESPACE_END(detail)
+PYBIND11_NAMESPACE_END(PYBIND11_NAMESPACE)

include/pybind11/eigen/tensor.h

@@ -0,0 +1,518 @@
+/*
+    pybind11/eigen/tensor.h: Transparent conversion for Eigen tensors
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+#pragma once
+
+#include "../numpy.h"
+
+#if defined(__GNUC__) && !defined(__clang__) && !defined(__INTEL_COMPILER)
+static_assert(__GNUC__ > 5, "Eigen Tensor support in pybind11 requires GCC > 5.0");
+#endif
+
+#if defined(_MSC_VER)
+#    pragma warning(push)
+#    pragma warning(disable : 4554) // Tensor.h warning
+#    pragma warning(disable : 4127) // Tensor.h warning
+#elif defined(__MINGW32__)
+#    pragma GCC diagnostic push
+#    pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
+#endif
+
+#include <unsupported/Eigen/CXX11/Tensor>
+
+#if defined(_MSC_VER)
+#    pragma warning(pop)
+#elif defined(__MINGW32__)
+#    pragma GCC diagnostic pop
+#endif
+
+static_assert(EIGEN_VERSION_AT_LEAST(3, 3, 0),
+              "Eigen Tensor support in pybind11 requires Eigen >= 3.3.0");
+
+PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
+
+PYBIND11_NAMESPACE_BEGIN(detail)
+
+inline bool is_tensor_aligned(const void *data) {
+    return (reinterpret_cast<std::size_t>(data) % EIGEN_DEFAULT_ALIGN_BYTES) == 0;
+}
+
+template <typename T>
+constexpr int compute_array_flag_from_tensor() {
+    static_assert((static_cast<int>(T::Layout) == static_cast<int>(Eigen::RowMajor))
+                      || (static_cast<int>(T::Layout) == static_cast<int>(Eigen::ColMajor)),
+                  "Layout must be row or column major");
+    return (static_cast<int>(T::Layout) == static_cast<int>(Eigen::RowMajor)) ? array::c_style
+                                                                              : array::f_style;
+}
+
+template <typename T>
+struct eigen_tensor_helper {};
+
+template <typename Scalar_, int NumIndices_, int Options_, typename IndexType>
+struct eigen_tensor_helper<Eigen::Tensor<Scalar_, NumIndices_, Options_, IndexType>> {
+    using Type = Eigen::Tensor<Scalar_, NumIndices_, Options_, IndexType>;
+    using ValidType = void;
+
+    static Eigen::DSizes<typename Type::Index, Type::NumIndices> get_shape(const Type &f) {
+        return f.dimensions();
+    }
+
+    static constexpr bool
+    is_correct_shape(const Eigen::DSizes<typename Type::Index, Type::NumIndices> & /*shape*/) {
+        return true;
+    }
+
+    template <typename T>
+    struct helper {};
+
+    template <size_t... Is>
+    struct helper<index_sequence<Is...>> {
+        static constexpr auto value = concat(const_name(((void) Is, "?"))...);
+    };
+
+    static constexpr auto dimensions_descriptor
+        = helper<decltype(make_index_sequence<Type::NumIndices>())>::value;
+
+    template <typename... Args>
+    static Type *alloc(Args &&...args) {
+        return new Type(std::forward<Args>(args)...);
+    }
+
+    static void free(Type *tensor) { delete tensor; }
+};
+
+template <typename Scalar_, typename std::ptrdiff_t... Indices, int Options_, typename IndexType>
+struct eigen_tensor_helper<
+    Eigen::TensorFixedSize<Scalar_, Eigen::Sizes<Indices...>, Options_, IndexType>> {
+    using Type = Eigen::TensorFixedSize<Scalar_, Eigen::Sizes<Indices...>, Options_, IndexType>;
+    using ValidType = void;
+
+    static constexpr Eigen::DSizes<typename Type::Index, Type::NumIndices>
+    get_shape(const Type & /*f*/) {
+        return get_shape();
+    }
+
+    static constexpr Eigen::DSizes<typename Type::Index, Type::NumIndices> get_shape() {
+        return Eigen::DSizes<typename Type::Index, Type::NumIndices>(Indices...);
+    }
+
+    static bool
+    is_correct_shape(const Eigen::DSizes<typename Type::Index, Type::NumIndices> &shape) {
+        return get_shape() == shape;
+    }
+
+    static constexpr auto dimensions_descriptor = concat(const_name<Indices>()...);
+
+    template <typename... Args>
+    static Type *alloc(Args &&...args) {
+        Eigen::aligned_allocator<Type> allocator;
+        return ::new (allocator.allocate(1)) Type(std::forward<Args>(args)...);
+    }
+
+    static void free(Type *tensor) {
+        Eigen::aligned_allocator<Type> allocator;
+        tensor->~Type();
+        allocator.deallocate(tensor, 1);
+    }
+};
+
+template <typename Type, bool ShowDetails, bool NeedsWriteable = false>
+struct get_tensor_descriptor {
+    static constexpr auto details
+        = const_name<NeedsWriteable>(", flags.writeable", "")
+          + const_name<static_cast<int>(Type::Layout) == static_cast<int>(Eigen::RowMajor)>(
+              ", flags.c_contiguous", ", flags.f_contiguous");
+    static constexpr auto value
+        = const_name("numpy.ndarray[") + npy_format_descriptor<typename Type::Scalar>::name
+          + const_name("[") + eigen_tensor_helper<remove_cv_t<Type>>::dimensions_descriptor
+          + const_name("]") + const_name<ShowDetails>(details, const_name("")) + const_name("]");
+};
+
+// When EIGEN_AVOID_STL_ARRAY is defined, Eigen::DSizes<T, 0> does not have the begin() member
+// function. Falling back to a simple loop works around this issue.
+//
+// We need to disable the type-limits warning for the inner loop when size = 0.
+
+#if defined(__GNUC__)
+#    pragma GCC diagnostic push
+#    pragma GCC diagnostic ignored "-Wtype-limits"
+#endif
+
+template <typename T, int size>
+std::vector<T> convert_dsizes_to_vector(const Eigen::DSizes<T, size> &arr) {
+    std::vector<T> result(size);
+
+    for (size_t i = 0; i < size; i++) {
+        result[i] = arr[i];
+    }
+
+    return result;
+}
+
+template <typename T, int size>
+Eigen::DSizes<T, size> get_shape_for_array(const array &arr) {
+    Eigen::DSizes<T, size> result;
+    const T *shape = arr.shape();
+    for (size_t i = 0; i < size; i++) {
+        result[i] = shape[i];
+    }
+
+    return result;
+}
+
+#if defined(__GNUC__)
+#    pragma GCC diagnostic pop
+#endif
+
+template <typename Type>
+struct type_caster<Type, typename eigen_tensor_helper<Type>::ValidType> {
+    using Helper = eigen_tensor_helper<Type>;
+    static constexpr auto temp_name = get_tensor_descriptor<Type, false>::value;
+    PYBIND11_TYPE_CASTER(Type, temp_name);
+
+    bool load(handle src, bool convert) {
+        if (!convert) {
+            if (!isinstance<array>(src)) {
+                return false;
+            }
+            array temp = array::ensure(src);
+            if (!temp) {
+                return false;
+            }
+
+            if (!convert && !temp.dtype().is(dtype::of<typename Type::Scalar>())) {
+                return false;
+            }
+        }
+
+        array_t<typename Type::Scalar, compute_array_flag_from_tensor<Type>()> arr(
+            reinterpret_borrow<object>(src));
+
+        if (arr.ndim() != Type::NumIndices) {
+            return false;
+        }
+        auto shape = get_shape_for_array<typename Type::Index, Type::NumIndices>(arr);
+
+        if (!Helper::is_correct_shape(shape)) {
+            return false;
+        }
+
+#if EIGEN_VERSION_AT_LEAST(3, 4, 0)
+        auto data_pointer = arr.data();
+#else
+        // Handle Eigen bug
+        auto data_pointer = const_cast<typename Type::Scalar *>(arr.data());
+#endif
+
+        if (is_tensor_aligned(arr.data())) {
+            value = Eigen::TensorMap<const Type, Eigen::Aligned>(data_pointer, shape);
+        } else {
+            value = Eigen::TensorMap<const Type>(data_pointer, shape);
+        }
+
+        return true;
+    }
+
+    static handle cast(Type &&src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::reference
+            || policy == return_value_policy::reference_internal) {
+            pybind11_fail("Cannot use a reference return value policy for an rvalue");
+        }
+        return cast_impl(&src, return_value_policy::move, parent);
+    }
+
+    static handle cast(const Type &&src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::reference
+            || policy == return_value_policy::reference_internal) {
+            pybind11_fail("Cannot use a reference return value policy for an rvalue");
+        }
+        return cast_impl(&src, return_value_policy::move, parent);
+    }
+
+    static handle cast(Type &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast_impl(&src, policy, parent);
+    }
+
+    static handle cast(const Type &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast(&src, policy, parent);
+    }
+
+    static handle cast(Type *src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic) {
+            policy = return_value_policy::take_ownership;
+        } else if (policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::reference;
+        }
+        return cast_impl(src, policy, parent);
+    }
+
+    static handle cast(const Type *src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic) {
+            policy = return_value_policy::take_ownership;
+        } else if (policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::reference;
+        }
+        return cast_impl(src, policy, parent);
+    }
+
+    template <typename C>
+    static handle cast_impl(C *src, return_value_policy policy, handle parent) {
+        object parent_object;
+        bool writeable = false;
+        switch (policy) {
+            case return_value_policy::move:
+                if (std::is_const<C>::value) {
+                    pybind11_fail("Cannot move from a constant reference");
+                }
+
+                src = Helper::alloc(std::move(*src));
+
+                parent_object
+                    = capsule(src, [](void *ptr) { Helper::free(reinterpret_cast<Type *>(ptr)); });
+                writeable = true;
+                break;
+
+            case return_value_policy::take_ownership:
+                if (std::is_const<C>::value) {
+                    // This cast is ugly, and might be UB in some cases, but we don't have an
+                    // alterantive here as we must free that memory
+                    Helper::free(const_cast<Type *>(src));
+                    pybind11_fail("Cannot take ownership of a const reference");
+                }
+
+                parent_object
+                    = capsule(src, [](void *ptr) { Helper::free(reinterpret_cast<Type *>(ptr)); });
+                writeable = true;
+                break;
+
+            case return_value_policy::copy:
+                writeable = true;
+                break;
+
+            case return_value_policy::reference:
+                parent_object = none();
+                writeable = !std::is_const<C>::value;
+                break;
+
+            case return_value_policy::reference_internal:
+                // Default should do the right thing
+                if (!parent) {
+                    pybind11_fail("Cannot use reference internal when there is no parent");
+                }
+                parent_object = reinterpret_borrow<object>(parent);
+                writeable = !std::is_const<C>::value;
+                break;
+
+            default:
+                pybind11_fail("pybind11 bug in eigen.h, please file a bug report");
+        }
+
+        auto result = array_t<typename Type::Scalar, compute_array_flag_from_tensor<Type>()>(
+            convert_dsizes_to_vector(Helper::get_shape(*src)), src->data(), parent_object);
+
+        if (!writeable) {
+            array_proxy(result.ptr())->flags &= ~detail::npy_api::NPY_ARRAY_WRITEABLE_;
+        }
+
+        return result.release();
+    }
+};
+
+template <typename StoragePointerType,
+          bool needs_writeable,
+          enable_if_t<!needs_writeable, bool> = true>
+StoragePointerType get_array_data_for_type(array &arr) {
+#if EIGEN_VERSION_AT_LEAST(3, 4, 0)
+    return reinterpret_cast<StoragePointerType>(arr.data());
+#else
+    // Handle Eigen bug
+    return reinterpret_cast<StoragePointerType>(const_cast<void *>(arr.data()));
+#endif
+}
+
+template <typename StoragePointerType,
+          bool needs_writeable,
+          enable_if_t<needs_writeable, bool> = true>
+StoragePointerType get_array_data_for_type(array &arr) {
+    return reinterpret_cast<StoragePointerType>(arr.mutable_data());
+}
+
+template <typename T, typename = void>
+struct get_storage_pointer_type;
+
+template <typename MapType>
+struct get_storage_pointer_type<MapType, void_t<typename MapType::StoragePointerType>> {
+    using SPT = typename MapType::StoragePointerType;
+};
+
+template <typename MapType>
+struct get_storage_pointer_type<MapType, void_t<typename MapType::PointerArgType>> {
+    using SPT = typename MapType::PointerArgType;
+};
+
+template <typename Type, int Options>
+struct type_caster<Eigen::TensorMap<Type, Options>,
+                   typename eigen_tensor_helper<remove_cv_t<Type>>::ValidType> {
+    using MapType = Eigen::TensorMap<Type, Options>;
+    using Helper = eigen_tensor_helper<remove_cv_t<Type>>;
+
+    bool load(handle src, bool /*convert*/) {
+        // Note that we have a lot more checks here as we want to make sure to avoid copies
+        if (!isinstance<array>(src)) {
+            return false;
+        }
+        auto arr = reinterpret_borrow<array>(src);
+        if ((arr.flags() & compute_array_flag_from_tensor<Type>()) == 0) {
+            return false;
+        }
+
+        if (!arr.dtype().is(dtype::of<typename Type::Scalar>())) {
+            return false;
+        }
+
+        if (arr.ndim() != Type::NumIndices) {
+            return false;
+        }
+
+        constexpr bool is_aligned = (Options & Eigen::Aligned) != 0;
+
+        if (PYBIND11_SILENCE_MSVC_C4127(is_aligned) && !is_tensor_aligned(arr.data())) {
+            return false;
+        }
+
+        auto shape = get_shape_for_array<typename Type::Index, Type::NumIndices>(arr);
+
+        if (!Helper::is_correct_shape(shape)) {
+            return false;
+        }
+
+        if (PYBIND11_SILENCE_MSVC_C4127(needs_writeable) && !arr.writeable()) {
+            return false;
+        }
+
+        auto result = get_array_data_for_type<typename get_storage_pointer_type<MapType>::SPT,
+                                              needs_writeable>(arr);
+
+        value.reset(new MapType(std::move(result), std::move(shape)));
+
+        return true;
+    }
+
+    static handle cast(MapType &&src, return_value_policy policy, handle parent) {
+        return cast_impl(&src, policy, parent);
+    }
+
+    static handle cast(const MapType &&src, return_value_policy policy, handle parent) {
+        return cast_impl(&src, policy, parent);
+    }
+
+    static handle cast(MapType &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast_impl(&src, policy, parent);
+    }
+
+    static handle cast(const MapType &src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic
+            || policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::copy;
+        }
+        return cast(&src, policy, parent);
+    }
+
+    static handle cast(MapType *src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic) {
+            policy = return_value_policy::take_ownership;
+        } else if (policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::reference;
+        }
+        return cast_impl(src, policy, parent);
+    }
+
+    static handle cast(const MapType *src, return_value_policy policy, handle parent) {
+        if (policy == return_value_policy::automatic) {
+            policy = return_value_policy::take_ownership;
+        } else if (policy == return_value_policy::automatic_reference) {
+            policy = return_value_policy::reference;
+        }
+        return cast_impl(src, policy, parent);
+    }
+
+    template <typename C>
+    static handle cast_impl(C *src, return_value_policy policy, handle parent) {
+        object parent_object;
+        constexpr bool writeable = !std::is_const<C>::value;
+        switch (policy) {
+            case return_value_policy::reference:
+                parent_object = none();
+                break;
+
+            case return_value_policy::reference_internal:
+                // Default should do the right thing
+                if (!parent) {
+                    pybind11_fail("Cannot use reference internal when there is no parent");
+                }
+                parent_object = reinterpret_borrow<object>(parent);
+                break;
+
+            case return_value_policy::take_ownership:
+                delete src;
+                // fallthrough
+            default:
+                // move, take_ownership don't make any sense for a ref/map:
+                pybind11_fail("Invalid return_value_policy for Eigen Map type, must be either "
+                              "reference or reference_internal");
+        }
+
+        auto result = array_t<typename Type::Scalar, compute_array_flag_from_tensor<Type>()>(
+            convert_dsizes_to_vector(Helper::get_shape(*src)),
+            src->data(),
+            std::move(parent_object));
+
+        if (!writeable) {
+            array_proxy(result.ptr())->flags &= ~detail::npy_api::NPY_ARRAY_WRITEABLE_;
+        }
+
+        return result.release();
+    }
+
+#if EIGEN_VERSION_AT_LEAST(3, 4, 0)
+
+    static constexpr bool needs_writeable = !std::is_const<typename std::remove_pointer<
+        typename get_storage_pointer_type<MapType>::SPT>::type>::value;
+#else
+    // Handle Eigen bug
+    static constexpr bool needs_writeable = !std::is_const<Type>::value;
+#endif
+
+protected:
+    // TODO: Move to std::optional once std::optional has more support
+    std::unique_ptr<MapType> value;
+
+public:
+    static constexpr auto name = get_tensor_descriptor<Type, true, needs_writeable>::value;
+    explicit operator MapType *() { return value.get(); }
+    explicit operator MapType &() { return *value; }
+    explicit operator MapType &&() && { return std::move(*value); }
+
+    template <typename T_>
+    using cast_op_type = ::pybind11::detail::movable_cast_op_type<T_>;
+};
+
+PYBIND11_NAMESPACE_END(detail)
+PYBIND11_NAMESPACE_END(PYBIND11_NAMESPACE)

include/pybind11/functional.h

@@ -48,9 +48,16 @@ public:
          */
         if (auto cfunc = func.cpp_function()) {
             auto *cfunc_self = PyCFunction_GET_SELF(cfunc.ptr());
-            if (isinstance<capsule>(cfunc_self)) {
+            if (cfunc_self == nullptr) {
+                PyErr_Clear();
+            } else if (isinstance<capsule>(cfunc_self)) {
                 auto c = reinterpret_borrow<capsule>(cfunc_self);
-                auto *rec = (function_record *) c;
+
+                function_record *rec = nullptr;
+                // Check that we can safely reinterpret the capsule into a function_record
+                if (detail::is_function_record_capsule(c)) {
+                    rec = c.get_pointer<function_record>();
+                }
 
                 while (rec != nullptr) {
                     if (rec->is_stateless
@@ -110,7 +117,7 @@ public:
     template <typename Func>
     static handle cast(Func &&f_, return_value_policy policy, handle /* parent */) {
         if (!f_) {
-            return none().inc_ref();
+            return none().release();
         }
 
         auto result = f_.template target<function_type>();

include/pybind11/gil.h

@@ -10,7 +10,10 @@
 #pragma once
 
 #include "detail/common.h"
+
+#if defined(WITH_THREAD) && !defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
 #    include "detail/internals.h"
+#endif
 
 PYBIND11_NAMESPACE_BEGIN(PYBIND11_NAMESPACE)
 
@@ -21,7 +24,9 @@ PyThreadState *get_thread_state_unchecked();
 
 PYBIND11_NAMESPACE_END(detail)
 
-#if defined(WITH_THREAD) && !defined(PYPY_VERSION)
+#if defined(WITH_THREAD)
+
+#    if !defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
 
 /* The functions below essentially reproduce the PyGILState_* API using a RAII
  * pattern, but there are a few important differences:
@@ -178,12 +183,14 @@ private:
     bool disassoc;
     bool active = true;
 };
-#elif defined(PYPY_VERSION)
+
+#    else // PYBIND11_SIMPLE_GIL_MANAGEMENT
+
 class gil_scoped_acquire {
     PyGILState_STATE state;
 
 public:
-    gil_scoped_acquire() { state = PyGILState_Ensure(); }
+    gil_scoped_acquire() : state{PyGILState_Ensure()} {}
     gil_scoped_acquire(const gil_scoped_acquire &) = delete;
     gil_scoped_acquire &operator=(const gil_scoped_acquire &) = delete;
     ~gil_scoped_acquire() { PyGILState_Release(state); }
@@ -194,19 +201,39 @@ class gil_scoped_release {
     PyThreadState *state;
 
 public:
-    gil_scoped_release() { state = PyEval_SaveThread(); }
+    gil_scoped_release() : state{PyEval_SaveThread()} {}
     gil_scoped_release(const gil_scoped_release &) = delete;
     gil_scoped_release &operator=(const gil_scoped_acquire &) = delete;
     ~gil_scoped_release() { PyEval_RestoreThread(state); }
     void disarm() {}
 };
-#else
+
+#    endif // PYBIND11_SIMPLE_GIL_MANAGEMENT
+
+#else // WITH_THREAD
+
 class gil_scoped_acquire {
+public:
+    gil_scoped_acquire() {
+        // Trick to suppress `unused variable` error messages (at call sites).
+        (void) (this != (this + 1));
+    }
+    gil_scoped_acquire(const gil_scoped_acquire &) = delete;
+    gil_scoped_acquire &operator=(const gil_scoped_acquire &) = delete;
     void disarm() {}
 };
+
 class gil_scoped_release {
+public:
+    gil_scoped_release() {
+        // Trick to suppress `unused variable` error messages (at call sites).
+        (void) (this != (this + 1));
+    }
+    gil_scoped_release(const gil_scoped_release &) = delete;
+    gil_scoped_release &operator=(const gil_scoped_acquire &) = delete;
     void disarm() {}
 };
-#endif
+
+#endif // WITH_THREAD
 
 PYBIND11_NAMESPACE_END(PYBIND11_NAMESPACE)

include/pybind11/pybind11.h

@@ -468,14 +468,21 @@ protected:
         if (rec->sibling) {
             if (PyCFunction_Check(rec->sibling.ptr())) {
                 auto *self = PyCFunction_GET_SELF(rec->sibling.ptr());
-                capsule rec_capsule = isinstance<capsule>(self) ? reinterpret_borrow<capsule>(self)
+                if (!isinstance<capsule>(self)) {
-                                                                : capsule(self);
+                    chain = nullptr;
-                chain = (detail::function_record *) rec_capsule;
+                } else {
+                    auto rec_capsule = reinterpret_borrow<capsule>(self);
+                    if (detail::is_function_record_capsule(rec_capsule)) {
+                        chain = rec_capsule.get_pointer<detail::function_record>();
                         /* Never append a method to an overload chain of a parent class;
                            instead, hide the parent's overloads in this case */
                         if (!chain->scope.is(rec->scope)) {
                             chain = nullptr;
                         }
+                    } else {
+                        chain = nullptr;
+                    }
+                }
             }
             // Don't trigger for things like the default __init__, which are wrapper_descriptors
             // that we are intentionally replacing
@@ -496,6 +503,7 @@ protected:
 
             capsule rec_capsule(unique_rec.release(),
                                 [](void *ptr) { destruct((detail::function_record *) ptr); });
+            rec_capsule.set_name(detail::get_function_record_capsule_name());
             guarded_strdup.release();
 
             object scope_module;
@@ -661,10 +669,13 @@ protected:
     /// Main dispatch logic for calls to functions bound using pybind11
     static PyObject *dispatcher(PyObject *self, PyObject *args_in, PyObject *kwargs_in) {
         using namespace detail;
+        assert(isinstance<capsule>(self));
 
         /* Iterator over the list of potentially admissible overloads */
-        const function_record *overloads = (function_record *) PyCapsule_GetPointer(self, nullptr),
+        const function_record *overloads = reinterpret_cast<function_record *>(
+                                  PyCapsule_GetPointer(self, get_function_record_capsule_name())),
                               *it = overloads;
+        assert(overloads != nullptr);
 
         /* Need to know how many arguments + keyword arguments there are to pick the right
            overload */
@@ -1883,9 +1894,22 @@ private:
 
     static detail::function_record *get_function_record(handle h) {
         h = detail::get_function(h);
-        return h ? (detail::function_record *) reinterpret_borrow<capsule>(
+        if (!h) {
-                   PyCFunction_GET_SELF(h.ptr()))
+            return nullptr;
-                 : nullptr;
+        }
+
+        handle func_self = PyCFunction_GET_SELF(h.ptr());
+        if (!func_self) {
+            throw error_already_set();
+        }
+        if (!isinstance<capsule>(func_self)) {
+            return nullptr;
+        }
+        auto cap = reinterpret_borrow<capsule>(func_self);
+        if (!detail::is_function_record_capsule(cap)) {
+            return nullptr;
+        }
+        return cap.get_pointer<detail::function_record>();
     }
 };
 
@@ -2345,7 +2369,7 @@ template <typename Access,
           typename Sentinel,
           typename ValueType,
           typename... Extra>
-iterator make_iterator_impl(Iterator &&first, Sentinel &&last, Extra &&...extra) {
+iterator make_iterator_impl(Iterator first, Sentinel last, Extra &&...extra) {
     using state = detail::iterator_state<Access, Policy, Iterator, Sentinel, ValueType, Extra...>;
     // TODO: state captures only the types of Extra, not the values
 
@@ -2371,7 +2395,7 @@ iterator make_iterator_impl(Iterator &&first, Sentinel &&last, Extra &&...extra)
                 Policy);
     }
 
-    return cast(state{std::forward<Iterator>(first), std::forward<Sentinel>(last), true});
+    return cast(state{first, last, true});
 }
 
 PYBIND11_NAMESPACE_END(detail)
@@ -2382,15 +2406,13 @@ template <return_value_policy Policy = return_value_policy::reference_internal,
           typename Sentinel,
           typename ValueType = typename detail::iterator_access<Iterator>::result_type,
           typename... Extra>
-iterator make_iterator(Iterator &&first, Sentinel &&last, Extra &&...extra) {
+iterator make_iterator(Iterator first, Sentinel last, Extra &&...extra) {
     return detail::make_iterator_impl<detail::iterator_access<Iterator>,
                                       Policy,
                                       Iterator,
                                       Sentinel,
                                       ValueType,
-                                      Extra...>(std::forward<Iterator>(first),
+                                      Extra...>(first, last, std::forward<Extra>(extra)...);
-                                                std::forward<Sentinel>(last),
-                                                std::forward<Extra>(extra)...);
 }
 
 /// Makes a python iterator over the keys (`.first`) of a iterator over pairs from a
@@ -2400,15 +2422,13 @@ template <return_value_policy Policy = return_value_policy::reference_internal,
           typename Sentinel,
           typename KeyType = typename detail::iterator_key_access<Iterator>::result_type,
           typename... Extra>
-iterator make_key_iterator(Iterator &&first, Sentinel &&last, Extra &&...extra) {
+iterator make_key_iterator(Iterator first, Sentinel last, Extra &&...extra) {
     return detail::make_iterator_impl<detail::iterator_key_access<Iterator>,
                                       Policy,
                                       Iterator,
                                       Sentinel,
                                       KeyType,
-                                      Extra...>(std::forward<Iterator>(first),
+                                      Extra...>(first, last, std::forward<Extra>(extra)...);
-                                                std::forward<Sentinel>(last),
-                                                std::forward<Extra>(extra)...);
 }
 
 /// Makes a python iterator over the values (`.second`) of a iterator over pairs from a
@@ -2418,15 +2438,13 @@ template <return_value_policy Policy = return_value_policy::reference_internal,
           typename Sentinel,
           typename ValueType = typename detail::iterator_value_access<Iterator>::result_type,
           typename... Extra>
-iterator make_value_iterator(Iterator &&first, Sentinel &&last, Extra &&...extra) {
+iterator make_value_iterator(Iterator first, Sentinel last, Extra &&...extra) {
     return detail::make_iterator_impl<detail::iterator_value_access<Iterator>,
                                       Policy,
                                       Iterator,
                                       Sentinel,
                                       ValueType,
-                                      Extra...>(std::forward<Iterator>(first),
+                                      Extra...>(first, last, std::forward<Extra>(extra)...);
-                                                std::forward<Sentinel>(last),
-                                                std::forward<Extra>(extra)...);
 }
 
 /// Makes an iterator over values of an stl container or other container supporting

include/pybind11/pytypes.h

@@ -501,11 +501,29 @@ struct error_fetch_and_normalize {
         std::string message_error_string;
         if (m_value) {
             auto value_str = reinterpret_steal<object>(PyObject_Str(m_value.ptr()));
+            constexpr const char *message_unavailable_exc
+                = "<MESSAGE UNAVAILABLE DUE TO ANOTHER EXCEPTION>";
             if (!value_str) {
                 message_error_string = detail::error_string();
-                result = "<MESSAGE UNAVAILABLE DUE TO ANOTHER EXCEPTION>";
+                result = message_unavailable_exc;
             } else {
-                result = value_str.cast<std::string>();
+                // Not using `value_str.cast<std::string>()`, to not potentially throw a secondary
+                // error_already_set that will then result in process termination (#4288).
+                auto value_bytes = reinterpret_steal<object>(
+                    PyUnicode_AsEncodedString(value_str.ptr(), "utf-8", "backslashreplace"));
+                if (!value_bytes) {
+                    message_error_string = detail::error_string();
+                    result = message_unavailable_exc;
+                } else {
+                    char *buffer = nullptr;
+                    Py_ssize_t length = 0;
+                    if (PyBytes_AsStringAndSize(value_bytes.ptr(), &buffer, &length) == -1) {
+                        message_error_string = detail::error_string();
+                        result = message_unavailable_exc;
+                    } else {
+                        result = std::string(buffer, static_cast<std::size_t>(length));
+                    }
+                }
             }
         } else {
             result = "<MESSAGE UNAVAILABLE>";
@@ -605,12 +623,6 @@ inline std::string error_string() {
 
 PYBIND11_NAMESPACE_END(detail)
 
-#if defined(_MSC_VER)
-#    pragma warning(push)
-#    pragma warning(disable : 4275 4251)
-//     warning C4275: An exported class was derived from a class that wasn't exported.
-//     Can be ignored when derived from a STL class.
-#endif
 /// Fetch and hold an error which was already set in Python.  An instance of this is typically
 /// thrown to propagate python-side errors back through C++ which can either be caught manually or
 /// else falls back to the function dispatcher (which then raises the captured error back to
@@ -670,9 +682,6 @@ private:
     ///          crashes (undefined behavior) if the Python interpreter is finalizing.
     static void m_fetched_error_deleter(detail::error_fetch_and_normalize *raw_ptr);
 };
-#if defined(_MSC_VER)
-#    pragma warning(pop)
-#endif
 
 /// Replaces the current Python error indicator with the chosen error, performing a
 /// 'raise from' to indicate that the chosen error was caused by the original error.
@@ -1383,7 +1392,7 @@ public:
 private:
     void advance() {
         value = reinterpret_steal<object>(PyIter_Next(m_ptr));
-        if (PyErr_Occurred()) {
+        if (value.ptr() == nullptr && PyErr_Occurred()) {
             throw error_already_set();
         }
     }
@@ -1433,6 +1442,9 @@ public:
     str(const char *c, const SzType &n)
         : object(PyUnicode_FromStringAndSize(c, ssize_t_cast(n)), stolen_t{}) {
         if (!m_ptr) {
+            if (PyErr_Occurred()) {
+                throw error_already_set();
+            }
             pybind11_fail("Could not allocate string object!");
         }
     }
@@ -1442,6 +1454,9 @@ public:
     // NOLINTNEXTLINE(google-explicit-constructor)
     str(const char *c = "") : object(PyUnicode_FromString(c), stolen_t{}) {
         if (!m_ptr) {
+            if (PyErr_Occurred()) {
+                throw error_already_set();
+            }
             pybind11_fail("Could not allocate string object!");
         }
     }
@@ -1599,6 +1614,9 @@ inline str::str(const bytes &b) {
     }
     auto obj = reinterpret_steal<object>(PyUnicode_FromStringAndSize(buffer, length));
     if (!obj) {
+        if (PyErr_Occurred()) {
+            throw error_already_set();
+        }
         pybind11_fail("Could not allocate string object!");
     }
     m_ptr = obj.release().ptr();
@@ -1810,16 +1828,16 @@ public:
 
     explicit capsule(const void *value,
                      const char *name = nullptr,
-                     void (*destructor)(PyObject *) = nullptr)
+                     PyCapsule_Destructor destructor = nullptr)
         : object(PyCapsule_New(const_cast<void *>(value), name, destructor), stolen_t{}) {
         if (!m_ptr) {
             throw error_already_set();
         }
     }
 
-    PYBIND11_DEPRECATED("Please pass a destructor that takes a void pointer as input")
+    PYBIND11_DEPRECATED("Please use the ctor with value, name, destructor args")
-    capsule(const void *value, void (*destruct)(PyObject *))
+    capsule(const void *value, PyCapsule_Destructor destructor)
-        : object(PyCapsule_New(const_cast<void *>(value), nullptr, destruct), stolen_t{}) {
+        : object(PyCapsule_New(const_cast<void *>(value), nullptr, destructor), stolen_t{}) {
         if (!m_ptr) {
             throw error_already_set();
         }
@@ -1830,7 +1848,7 @@ public:
             // guard if destructor called while err indicator is set
             error_scope error_guard;
             auto destructor = reinterpret_cast<void (*)(void *)>(PyCapsule_GetContext(o));
-            if (PyErr_Occurred()) {
+            if (destructor == nullptr && PyErr_Occurred()) {
                 throw error_already_set();
             }
             const char *name = get_name_in_error_scope(o);
@@ -1844,7 +1862,7 @@ public:
             }
         });
 
-        if (!m_ptr || PyCapsule_SetContext(m_ptr, (void *) destructor) != 0) {
+        if (!m_ptr || PyCapsule_SetContext(m_ptr, reinterpret_cast<void *>(destructor)) != 0) {
             throw error_already_set();
         }
     }
@@ -1968,7 +1986,11 @@ public:
     void clear() /* py-non-const */ { PyDict_Clear(ptr()); }
     template <typename T>
     bool contains(T &&key) const {
-        return PyDict_Contains(m_ptr, detail::object_or_cast(std::forward<T>(key)).ptr()) == 1;
+        auto result = PyDict_Contains(m_ptr, detail::object_or_cast(std::forward<T>(key)).ptr());
+        if (result == -1) {
+            throw error_already_set();
+        }
+        return result == 1;
     }
 
 private:
@@ -2054,7 +2076,11 @@ public:
     bool empty() const { return size() == 0; }
     template <typename T>
     bool contains(T &&val) const {
-        return PySet_Contains(m_ptr, detail::object_or_cast(std::forward<T>(val)).ptr()) == 1;
+        auto result = PySet_Contains(m_ptr, detail::object_or_cast(std::forward<T>(val)).ptr());
+        if (result == -1) {
+            throw error_already_set();
+        }
+        return result == 1;
     }
 };
 

include/pybind11/stl.h

@@ -311,7 +311,7 @@ struct optional_caster {
     template <typename T>
     static handle cast(T &&src, return_value_policy policy, handle parent) {
         if (!src) {
-            return none().inc_ref();
+            return none().release();
         }
         if (!std::is_lvalue_reference<T>::value) {
             policy = return_value_policy_override<Value>::policy(policy);

tests/CMakeLists.txt

@@ -128,7 +128,9 @@ set(PYBIND11_TEST_FILES
     test_custom_type_casters
     test_custom_type_setup
     test_docstring_options
-    test_eigen
+    test_eigen_matrix
+    test_eigen_tensor
+    test_eigen_tensor_avoid_stl_array.cpp
     test_enum
     test_eval
     test_exceptions
@@ -233,7 +235,10 @@ list(GET PYBIND11_EIGEN_VERSION_AND_HASH 1 PYBIND11_EIGEN_VERSION_HASH)
 # Check if Eigen is available; if not, remove from PYBIND11_TEST_FILES (but
 # keep it in PYBIND11_PYTEST_FILES, so that we get the "eigen is not installed"
 # skip message).
-list(FIND PYBIND11_TEST_FILES test_eigen.cpp PYBIND11_TEST_FILES_EIGEN_I)
+list(FIND PYBIND11_TEST_FILES test_eigen_matrix.cpp PYBIND11_TEST_FILES_EIGEN_I)
+if(PYBIND11_TEST_FILES_EIGEN_I EQUAL -1)
+  list(FIND PYBIND11_TEST_FILES test_eigen_tensor.cpp PYBIND11_TEST_FILES_EIGEN_I)
+endif()
 if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
   # Try loading via newer Eigen's Eigen3Config first (bypassing tools/FindEigen3.cmake).
   # Eigen 3.3.1+ exports a cmake 3.0+ target for handling dependency requirements, but also
@@ -289,12 +294,37 @@ if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
     endif()
     message(STATUS "Building tests with Eigen v${EIGEN3_VERSION}")
   else()
+    list(FIND PYBIND11_TEST_FILES test_eigen_matrix.cpp PYBIND11_TEST_FILES_EIGEN_I)
+    if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
       list(REMOVE_AT PYBIND11_TEST_FILES ${PYBIND11_TEST_FILES_EIGEN_I})
+    endif()
+
+    list(FIND PYBIND11_TEST_FILES test_eigen_tensor.cpp PYBIND11_TEST_FILES_EIGEN_I)
+    if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
+      list(REMOVE_AT PYBIND11_TEST_FILES ${PYBIND11_TEST_FILES_EIGEN_I})
+    endif()
+    list(FIND PYBIND11_TEST_FILES test_eigen_tensor_avoid_stl_array.cpp
+         PYBIND11_TEST_FILES_EIGEN_I)
+    if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
+      list(REMOVE_AT PYBIND11_TEST_FILES ${PYBIND11_TEST_FILES_EIGEN_I})
+    endif()
     message(
       STATUS "Building tests WITHOUT Eigen, use -DDOWNLOAD_EIGEN=ON on CMake 3.11+ to download")
   endif()
 endif()
 
+# Some code doesn't support gcc 4
+if(CMAKE_CXX_COMPILER_ID STREQUAL "GNU" AND CMAKE_CXX_COMPILER_VERSION VERSION_LESS 5.0)
+  list(FIND PYBIND11_TEST_FILES test_eigen_tensor.cpp PYBIND11_TEST_FILES_EIGEN_I)
+  if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
+    list(REMOVE_AT PYBIND11_TEST_FILES ${PYBIND11_TEST_FILES_EIGEN_I})
+  endif()
+  list(FIND PYBIND11_TEST_FILES test_eigen_tensor_avoid_stl_array.cpp PYBIND11_TEST_FILES_EIGEN_I)
+  if(PYBIND11_TEST_FILES_EIGEN_I GREATER -1)
+    list(REMOVE_AT PYBIND11_TEST_FILES ${PYBIND11_TEST_FILES_EIGEN_I})
+  endif()
+endif()
+
 # Optional dependency for some tests (boost::variant is only supported with version >= 1.56)
 find_package(Boost 1.56)
 

tests/conftest.py

@@ -210,4 +210,5 @@ def pytest_report_header(config):
         f" {pybind11_tests.compiler_info}"
         f" {pybind11_tests.cpp_std}"
         f" {pybind11_tests.PYBIND11_INTERNALS_ID}"
+        f" PYBIND11_SIMPLE_GIL_MANAGEMENT={pybind11_tests.PYBIND11_SIMPLE_GIL_MANAGEMENT}"
     )

tests/cross_module_gil_utils.cpp

@@ -6,9 +6,15 @@
     All rights reserved. Use of this source code is governed by a
     BSD-style license that can be found in the LICENSE file.
 */
+#if defined(PYBIND11_INTERNALS_VERSION)
+#    undef PYBIND11_INTERNALS_VERSION
+#endif
+#define PYBIND11_INTERNALS_VERSION 21814642 // Ensure this module has its own `internals` instance.
 #include <pybind11/pybind11.h>
 
 #include <cstdint>
+#include <string>
+#include <thread>
 
 // This file mimics a DSO that makes pybind11 calls but does not define a
 // PYBIND11_MODULE. The purpose is to test that such a DSO can create a
@@ -21,8 +27,54 @@
 namespace {
 
 namespace py = pybind11;
+
 void gil_acquire() { py::gil_scoped_acquire gil; }
 
+std::string gil_multi_acquire_release(unsigned bits) {
+    if ((bits & 0x1u) != 0u) {
+        py::gil_scoped_acquire gil;
+    }
+    if ((bits & 0x2u) != 0u) {
+        py::gil_scoped_release gil;
+    }
+    if ((bits & 0x4u) != 0u) {
+        py::gil_scoped_acquire gil;
+    }
+    if ((bits & 0x8u) != 0u) {
+        py::gil_scoped_release gil;
+    }
+    return PYBIND11_INTERNALS_ID;
+}
+
+struct CustomAutoGIL {
+    CustomAutoGIL() : gstate(PyGILState_Ensure()) {}
+    ~CustomAutoGIL() { PyGILState_Release(gstate); }
+
+    PyGILState_STATE gstate;
+};
+struct CustomAutoNoGIL {
+    CustomAutoNoGIL() : save(PyEval_SaveThread()) {}
+    ~CustomAutoNoGIL() { PyEval_RestoreThread(save); }
+
+    PyThreadState *save;
+};
+
+template <typename Acquire, typename Release>
+void gil_acquire_inner() {
+    Acquire acquire_outer;
+    Acquire acquire_inner;
+    Release release;
+}
+
+template <typename Acquire, typename Release>
+void gil_acquire_nested() {
+    Acquire acquire_outer;
+    Acquire acquire_inner;
+    Release release;
+    auto thread = std::thread(&gil_acquire_inner<Acquire, Release>);
+    thread.join();
+}
+
 constexpr char kModuleName[] = "cross_module_gil_utils";
 
 struct PyModuleDef moduledef = {
@@ -30,6 +82,9 @@ struct PyModuleDef moduledef = {
 
 } // namespace
 
+#define ADD_FUNCTION(Name, ...)                                                                   \
+    PyModule_AddObject(m, Name, PyLong_FromVoidPtr(reinterpret_cast<void *>(&__VA_ARGS__)));
+
 extern "C" PYBIND11_EXPORT PyObject *PyInit_cross_module_gil_utils() {
 
     PyObject *m = PyModule_Create(&moduledef);
@@ -37,8 +92,16 @@ extern "C" PYBIND11_EXPORT PyObject *PyInit_cross_module_gil_utils() {
     if (m != nullptr) {
         static_assert(sizeof(&gil_acquire) == sizeof(void *),
                       "Function pointer must have the same size as void*");
-        PyModule_AddObject(
+        ADD_FUNCTION("gil_acquire_funcaddr", gil_acquire)
-            m, "gil_acquire_funcaddr", PyLong_FromVoidPtr(reinterpret_cast<void *>(&gil_acquire)));
+        ADD_FUNCTION("gil_multi_acquire_release_funcaddr", gil_multi_acquire_release)
+        ADD_FUNCTION("gil_acquire_inner_custom_funcaddr",
+                     gil_acquire_inner<CustomAutoGIL, CustomAutoNoGIL>)
+        ADD_FUNCTION("gil_acquire_nested_custom_funcaddr",
+                     gil_acquire_nested<CustomAutoGIL, CustomAutoNoGIL>)
+        ADD_FUNCTION("gil_acquire_inner_pybind11_funcaddr",
+                     gil_acquire_inner<py::gil_scoped_acquire, py::gil_scoped_release>)
+        ADD_FUNCTION("gil_acquire_nested_pybind11_funcaddr",
+                     gil_acquire_nested<py::gil_scoped_acquire, py::gil_scoped_release>)
     }
 
     return m;

tests/extra_python_package/test_files.py

@@ -55,6 +55,11 @@ detail_headers = {
     "include/pybind11/detail/typeid.h",
 }
 
+eigen_headers = {
+    "include/pybind11/eigen/matrix.h",
+    "include/pybind11/eigen/tensor.h",
+}
+
 stl_headers = {
     "include/pybind11/stl/filesystem.h",
 }
@@ -82,7 +87,7 @@ py_files = {
     "setup_helpers.py",
 }
 
-headers = main_headers | detail_headers | stl_headers
+headers = main_headers | detail_headers | eigen_headers | stl_headers
 src_files = headers | cmake_files | pkgconfig_files
 all_files = src_files | py_files
 
@@ -92,6 +97,7 @@ sdist_files = {
     "pybind11/include",
     "pybind11/include/pybind11",
     "pybind11/include/pybind11/detail",
+    "pybind11/include/pybind11/eigen",
     "pybind11/include/pybind11/stl",
     "pybind11/share",
     "pybind11/share/cmake",

tests/pybind11_tests.cpp

@@ -89,6 +89,12 @@ PYBIND11_MODULE(pybind11_tests, m) {
 #endif
     m.attr("cpp_std") = cpp_std();
     m.attr("PYBIND11_INTERNALS_ID") = PYBIND11_INTERNALS_ID;
+    m.attr("PYBIND11_SIMPLE_GIL_MANAGEMENT") =
+#if defined(PYBIND11_SIMPLE_GIL_MANAGEMENT)
+        true;
+#else
+        false;
+#endif
 
     bind_ConstructorStats(m);
 

tests/test_callbacks.cpp

@@ -240,4 +240,41 @@ TEST_SUBMODULE(callbacks, m) {
             f();
         }
     });
+
+    auto *custom_def = []() {
+        static PyMethodDef def;
+        def.ml_name = "example_name";
+        def.ml_doc = "Example doc";
+        def.ml_meth = [](PyObject *, PyObject *args) -> PyObject * {
+            if (PyTuple_Size(args) != 1) {
+                throw std::runtime_error("Invalid number of arguments for example_name");
+            }
+            PyObject *first = PyTuple_GetItem(args, 0);
+            if (!PyLong_Check(first)) {
+                throw std::runtime_error("Invalid argument to example_name");
+            }
+            auto result = py::cast(PyLong_AsLong(first) * 9);
+            return result.release().ptr();
+        };
+        def.ml_flags = METH_VARARGS;
+        return &def;
+    }();
+
+    // rec_capsule with name that has the same value (but not pointer) as our internal one
+    // This capsule should be detected by our code as foreign and not inspected as the pointers
+    // shouldn't match
+    constexpr const char *rec_capsule_name
+        = pybind11::detail::internals_function_record_capsule_name;
+    py::capsule rec_capsule(std::malloc(1), [](void *data) { std::free(data); });
+    rec_capsule.set_name(rec_capsule_name);
+    m.add_object("custom_function", PyCFunction_New(custom_def, rec_capsule.ptr()));
+
+    // This test requires a new ABI version to pass
+#if PYBIND11_INTERNALS_VERSION > 4
+    // rec_capsule with nullptr name
+    py::capsule rec_capsule2(std::malloc(1), [](void *data) { std::free(data); });
+    m.add_object("custom_function2", PyCFunction_New(custom_def, rec_capsule2.ptr()));
+#else
+    m.add_object("custom_function2", py::none());
+#endif
 }

tests/test_callbacks.py

@@ -193,3 +193,16 @@ def test_callback_num_times():
     if len(rates) > 1:
         print("Min    Mean   Max")
         print(f"{min(rates):6.3f} {sum(rates) / len(rates):6.3f} {max(rates):6.3f}")
+
+
+def test_custom_func():
+    assert m.custom_function(4) == 36
+    assert m.roundtrip(m.custom_function)(4) == 36
+
+
+@pytest.mark.skipif(
+    m.custom_function2 is None, reason="Current PYBIND11_INTERNALS_VERSION too low"
+)
+def test_custom_func2():
+    assert m.custom_function2(3) == 27
+    assert m.roundtrip(m.custom_function2)(3) == 27

tests/test_class.cpp

@@ -392,6 +392,8 @@ TEST_SUBMODULE(class_, m) {
 
     protected:
         virtual int foo() const { return value; }
+        virtual void *void_foo() { return static_cast<void *>(&value); }
+        virtual void *get_self() { return static_cast<void *>(this); }
 
     private:
         int value = 42;
@@ -400,6 +402,8 @@ TEST_SUBMODULE(class_, m) {
     class TrampolineB : public ProtectedB {
     public:
         int foo() const override { PYBIND11_OVERRIDE(int, ProtectedB, foo, ); }
+        void *void_foo() override { PYBIND11_OVERRIDE(void *, ProtectedB, void_foo, ); }
+        void *get_self() override { PYBIND11_OVERRIDE(void *, ProtectedB, get_self, ); }
     };
 
     class PublicistB : public ProtectedB {
@@ -409,11 +413,23 @@ TEST_SUBMODULE(class_, m) {
         // (in Debug builds only, tested with icpc (ICC) 2021.1 Beta 20200827)
         ~PublicistB() override{}; // NOLINT(modernize-use-equals-default)
         using ProtectedB::foo;
+        using ProtectedB::get_self;
+        using ProtectedB::void_foo;
     };
 
+    m.def("read_foo", [](const void *original) {
+        const int *ptr = reinterpret_cast<const int *>(original);
+        return *ptr;
+    });
+
+    m.def("pointers_equal",
+          [](const void *original, const void *comparison) { return original == comparison; });
+
     py::class_<ProtectedB, TrampolineB>(m, "ProtectedB")
         .def(py::init<>())
-        .def("foo", &PublicistB::foo);
+        .def("foo", &PublicistB::foo)
+        .def("void_foo", &PublicistB::void_foo)
+        .def("get_self", &PublicistB::get_self);
 
     // test_brace_initialization
     struct BraceInitialization {

tests/test_class.py

@@ -320,6 +320,8 @@ def test_bind_protected_functions():
 
     b = m.ProtectedB()
     assert b.foo() == 42
+    assert m.read_foo(b.void_foo()) == 42
+    assert m.pointers_equal(b.get_self(), b)
 
     class C(m.ProtectedB):
         def __init__(self):

tests/test_eigen_matrix.cpp

@@ -7,7 +7,7 @@
     BSD-style license that can be found in the LICENSE file.
 */
 
-#include <pybind11/eigen.h>
+#include <pybind11/eigen/matrix.h>
 #include <pybind11/stl.h>
 
 #include "constructor_stats.h"
@@ -81,7 +81,7 @@ struct CustomOperatorNew {
     EIGEN_MAKE_ALIGNED_OPERATOR_NEW;
 };
 
-TEST_SUBMODULE(eigen, m) {
+TEST_SUBMODULE(eigen_matrix, m) {
     using FixedMatrixR = Eigen::Matrix<float, 5, 6, Eigen::RowMajor>;
     using FixedMatrixC = Eigen::Matrix<float, 5, 6>;
     using DenseMatrixR = Eigen::Matrix<float, Eigen::Dynamic, Eigen::Dynamic, Eigen::RowMajor>;

tests/test_eigen_matrix.py

@@ -3,7 +3,7 @@ import pytest
 from pybind11_tests import ConstructorStats
 
 np = pytest.importorskip("numpy")
-m = pytest.importorskip("pybind11_tests.eigen")
+m = pytest.importorskip("pybind11_tests.eigen_matrix")
 
 
 ref = np.array(

tests/test_eigen_tensor.cpp

@@ -0,0 +1,16 @@
+/*
+    tests/eigen_tensor.cpp -- automatic conversion of Eigen Tensor
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+constexpr const char *test_eigen_tensor_module_name = "eigen_tensor";
+
+#define PYBIND11_TEST_EIGEN_TENSOR_NAMESPACE eigen_tensor
+
+#ifdef EIGEN_AVOID_STL_ARRAY
+#    undef EIGEN_AVOID_STL_ARRAY
+#endif
+
+#include "test_eigen_tensor.inl"

tests/test_eigen_tensor.inl

@@ -0,0 +1,333 @@
+/*
+    tests/eigen_tensor.cpp -- automatic conversion of Eigen Tensor
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+#include <pybind11/eigen/tensor.h>
+
+#include "pybind11_tests.h"
+
+namespace PYBIND11_TEST_EIGEN_TENSOR_NAMESPACE {
+
+template <typename M>
+void reset_tensor(M &x) {
+    for (int i = 0; i < x.dimension(0); i++) {
+        for (int j = 0; j < x.dimension(1); j++) {
+            for (int k = 0; k < x.dimension(2); k++) {
+                x(i, j, k) = i * (5 * 2) + j * 2 + k;
+            }
+        }
+    }
+}
+
+template <typename M>
+bool check_tensor(M &x) {
+    for (int i = 0; i < x.dimension(0); i++) {
+        for (int j = 0; j < x.dimension(1); j++) {
+            for (int k = 0; k < x.dimension(2); k++) {
+                if (x(i, j, k) != (i * (5 * 2) + j * 2 + k)) {
+                    return false;
+                }
+            }
+        }
+    }
+    return true;
+}
+
+template <int Options>
+Eigen::Tensor<double, 3, Options> &get_tensor() {
+    static Eigen::Tensor<double, 3, Options> *x;
+
+    if (!x) {
+        x = new Eigen::Tensor<double, 3, Options>(3, 5, 2);
+        reset_tensor(*x);
+    }
+
+    return *x;
+}
+
+template <int Options>
+Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> &get_tensor_map() {
+    static Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> *x;
+
+    if (!x) {
+        x = new Eigen::TensorMap<Eigen::Tensor<double, 3, Options>>(get_tensor<Options>());
+    }
+
+    return *x;
+}
+
+template <int Options>
+Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options> &get_fixed_tensor() {
+    static Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options> *x;
+
+    if (!x) {
+        Eigen::aligned_allocator<Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options>>
+            allocator;
+        x = new (allocator.allocate(1))
+            Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options>();
+        reset_tensor(*x);
+    }
+
+    return *x;
+}
+
+template <int Options>
+const Eigen::Tensor<double, 3, Options> &get_const_tensor() {
+    return get_tensor<Options>();
+}
+
+template <int Options>
+struct CustomExample {
+    CustomExample() : member(get_tensor<Options>()), view_member(member) {}
+
+    Eigen::Tensor<double, 3, Options> member;
+    Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> view_member;
+};
+
+template <int Options>
+void init_tensor_module(pybind11::module &m) {
+    const char *needed_options = "";
+    if (PYBIND11_SILENCE_MSVC_C4127(Options == Eigen::ColMajor)) {
+        needed_options = "F";
+    } else {
+        needed_options = "C";
+    }
+    m.attr("needed_options") = needed_options;
+
+    m.def("setup", []() {
+        reset_tensor(get_tensor<Options>());
+        reset_tensor(get_fixed_tensor<Options>());
+    });
+
+    m.def("is_ok", []() {
+        return check_tensor(get_tensor<Options>()) && check_tensor(get_fixed_tensor<Options>());
+    });
+
+    py::class_<CustomExample<Options>>(m, "CustomExample")
+        .def(py::init<>())
+        .def_readonly(
+            "member", &CustomExample<Options>::member, py::return_value_policy::reference_internal)
+        .def_readonly("member_view",
+                      &CustomExample<Options>::view_member,
+                      py::return_value_policy::reference_internal);
+
+    m.def(
+        "copy_fixed_tensor",
+        []() { return &get_fixed_tensor<Options>(); },
+        py::return_value_policy::copy);
+
+    m.def(
+        "copy_tensor", []() { return &get_tensor<Options>(); }, py::return_value_policy::copy);
+
+    m.def(
+        "copy_const_tensor",
+        []() { return &get_const_tensor<Options>(); },
+        py::return_value_policy::copy);
+
+    m.def(
+        "move_fixed_tensor_copy",
+        []() -> Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options> {
+            return get_fixed_tensor<Options>();
+        },
+        py::return_value_policy::move);
+
+    m.def(
+        "move_tensor_copy",
+        []() -> Eigen::Tensor<double, 3, Options> { return get_tensor<Options>(); },
+        py::return_value_policy::move);
+
+    m.def(
+        "move_const_tensor",
+        []() -> const Eigen::Tensor<double, 3, Options> & { return get_const_tensor<Options>(); },
+        py::return_value_policy::move);
+
+    m.def(
+        "take_fixed_tensor",
+
+        []() {
+            Eigen::aligned_allocator<
+                Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options>>
+                allocator;
+            return new (allocator.allocate(1))
+                Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options>(
+                    get_fixed_tensor<Options>());
+        },
+        py::return_value_policy::take_ownership);
+
+    m.def(
+        "take_tensor",
+        []() { return new Eigen::Tensor<double, 3, Options>(get_tensor<Options>()); },
+        py::return_value_policy::take_ownership);
+
+    m.def(
+        "take_const_tensor",
+        []() -> const Eigen::Tensor<double, 3, Options> * {
+            return new Eigen::Tensor<double, 3, Options>(get_tensor<Options>());
+        },
+        py::return_value_policy::take_ownership);
+
+    m.def(
+        "take_view_tensor",
+        []() -> const Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> * {
+            return new Eigen::TensorMap<Eigen::Tensor<double, 3, Options>>(get_tensor<Options>());
+        },
+        py::return_value_policy::take_ownership);
+
+    m.def(
+        "reference_tensor",
+        []() { return &get_tensor<Options>(); },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_tensor_v2",
+        []() -> Eigen::Tensor<double, 3, Options> & { return get_tensor<Options>(); },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_tensor_internal",
+        []() { return &get_tensor<Options>(); },
+        py::return_value_policy::reference_internal);
+
+    m.def(
+        "reference_fixed_tensor",
+        []() { return &get_tensor<Options>(); },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_const_tensor",
+        []() { return &get_const_tensor<Options>(); },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_const_tensor_v2",
+        []() -> const Eigen::Tensor<double, 3, Options> & { return get_const_tensor<Options>(); },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor",
+        []() -> Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> {
+            return get_tensor_map<Options>();
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor_v2",
+        // NOLINTNEXTLINE(readability-const-return-type)
+        []() -> const Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> {
+            return get_tensor_map<Options>(); // NOLINT(readability-const-return-type)
+        },                                    // NOLINT(readability-const-return-type)
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor_v3",
+        []() -> Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> * {
+            return &get_tensor_map<Options>();
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor_v4",
+        []() -> const Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> * {
+            return &get_tensor_map<Options>();
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor_v5",
+        []() -> Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> & {
+            return get_tensor_map<Options>();
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_tensor_v6",
+        []() -> const Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> & {
+            return get_tensor_map<Options>();
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "reference_view_of_fixed_tensor",
+        []() {
+            return Eigen::TensorMap<
+                Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options>>(
+                get_fixed_tensor<Options>());
+        },
+        py::return_value_policy::reference);
+
+    m.def("round_trip_tensor",
+          [](const Eigen::Tensor<double, 3, Options> &tensor) { return tensor; });
+
+    m.def(
+        "round_trip_tensor_noconvert",
+        [](const Eigen::Tensor<double, 3, Options> &tensor) { return tensor; },
+        py::arg("tensor").noconvert());
+
+    m.def("round_trip_tensor2",
+          [](const Eigen::Tensor<int32_t, 3, Options> &tensor) { return tensor; });
+
+    m.def("round_trip_fixed_tensor",
+          [](const Eigen::TensorFixedSize<double, Eigen::Sizes<3, 5, 2>, Options> &tensor) {
+              return tensor;
+          });
+
+    m.def(
+        "round_trip_view_tensor",
+        [](Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> view) { return view; },
+        py::return_value_policy::reference);
+
+    m.def(
+        "round_trip_view_tensor_ref",
+        [](Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> &view) { return view; },
+        py::return_value_policy::reference);
+
+    m.def(
+        "round_trip_view_tensor_ptr",
+        [](Eigen::TensorMap<Eigen::Tensor<double, 3, Options>> *view) { return view; },
+        py::return_value_policy::reference);
+
+    m.def(
+        "round_trip_aligned_view_tensor",
+        [](Eigen::TensorMap<Eigen::Tensor<double, 3, Options>, Eigen::Aligned> view) {
+            return view;
+        },
+        py::return_value_policy::reference);
+
+    m.def(
+        "round_trip_const_view_tensor",
+        [](Eigen::TensorMap<const Eigen::Tensor<double, 3, Options>> view) {
+            return Eigen::Tensor<double, 3, Options>(view);
+        },
+        py::return_value_policy::move);
+
+    m.def(
+        "round_trip_rank_0",
+        [](const Eigen::Tensor<double, 0, Options> &tensor) { return tensor; },
+        py::return_value_policy::move);
+
+    m.def(
+        "round_trip_rank_0_noconvert",
+        [](const Eigen::Tensor<double, 0, Options> &tensor) { return tensor; },
+        py::arg("tensor").noconvert(),
+        py::return_value_policy::move);
+
+    m.def(
+        "round_trip_rank_0_view",
+        [](Eigen::TensorMap<Eigen::Tensor<double, 0, Options>> &tensor) { return tensor; },
+        py::return_value_policy::reference);
+}
+
+void test_module(py::module_ &);
+test_initializer name(test_eigen_tensor_module_name, test_module);
+void test_module(py::module_ &m) {
+    auto f_style = m.def_submodule("f_style");
+    auto c_style = m.def_submodule("c_style");
+
+    init_tensor_module<Eigen::ColMajor>(f_style);
+    init_tensor_module<Eigen::RowMajor>(c_style);
+}
+
+} // namespace PYBIND11_TEST_EIGEN_TENSOR_NAMESPACE

tests/test_eigen_tensor.py

@@ -0,0 +1,296 @@
+import sys
+
+import pytest
+
+np = pytest.importorskip("numpy")
+eigen_tensor = pytest.importorskip("pybind11_tests.eigen_tensor")
+submodules = [eigen_tensor.c_style, eigen_tensor.f_style]
+try:
+    from pybind11_tests import eigen_tensor_avoid_stl_array as avoid
+
+    submodules += [avoid.c_style, avoid.f_style]
+except ImportError as e:
+    # Ensure config, build, toolchain, etc. issues are not masked here:
+    raise RuntimeError(
+        "import pybind11_tests.eigen_tensor_avoid_stl_array FAILED, while "
+        "import pybind11_tests.eigen_tensor succeeded. "
+        "Please ensure that "
+        "test_eigen_tensor.cpp & "
+        "test_eigen_tensor_avoid_stl_array.cpp "
+        "are built together (or both are not built if Eigen is not available)."
+    ) from e
+
+tensor_ref = np.empty((3, 5, 2), dtype=np.int64)
+
+for i in range(tensor_ref.shape[0]):
+    for j in range(tensor_ref.shape[1]):
+        for k in range(tensor_ref.shape[2]):
+            tensor_ref[i, j, k] = i * (5 * 2) + j * 2 + k
+
+indices = (2, 3, 1)
+
+
+@pytest.fixture(autouse=True)
+def cleanup():
+    for module in submodules:
+        module.setup()
+
+    yield
+
+    for module in submodules:
+        assert module.is_ok()
+
+
+def test_import_avoid_stl_array():
+    pytest.importorskip("pybind11_tests.eigen_tensor_avoid_stl_array")
+    assert len(submodules) == 4
+
+
+def assert_equal_tensor_ref(mat, writeable=True, modified=None):
+    assert mat.flags.writeable == writeable
+
+    copy = np.array(tensor_ref)
+    if modified is not None:
+        copy[indices] = modified
+
+    np.testing.assert_array_equal(mat, copy)
+
+
+@pytest.mark.parametrize("m", submodules)
+@pytest.mark.parametrize("member_name", ["member", "member_view"])
+def test_reference_internal(m, member_name):
+
+    if not hasattr(sys, "getrefcount"):
+        pytest.skip("No reference counting")
+    foo = m.CustomExample()
+    counts = sys.getrefcount(foo)
+    mem = getattr(foo, member_name)
+    assert_equal_tensor_ref(mem, writeable=False)
+    new_counts = sys.getrefcount(foo)
+    assert new_counts == counts + 1
+    assert_equal_tensor_ref(mem, writeable=False)
+    del mem
+    assert sys.getrefcount(foo) == counts
+
+
+assert_equal_funcs = [
+    "copy_tensor",
+    "copy_fixed_tensor",
+    "copy_const_tensor",
+    "move_tensor_copy",
+    "move_fixed_tensor_copy",
+    "take_tensor",
+    "take_fixed_tensor",
+    "reference_tensor",
+    "reference_tensor_v2",
+    "reference_fixed_tensor",
+    "reference_view_of_tensor",
+    "reference_view_of_tensor_v3",
+    "reference_view_of_tensor_v5",
+    "reference_view_of_fixed_tensor",
+]
+
+assert_equal_const_funcs = [
+    "reference_view_of_tensor_v2",
+    "reference_view_of_tensor_v4",
+    "reference_view_of_tensor_v6",
+    "reference_const_tensor",
+    "reference_const_tensor_v2",
+]
+
+
+@pytest.mark.parametrize("m", submodules)
+@pytest.mark.parametrize("func_name", assert_equal_funcs + assert_equal_const_funcs)
+def test_convert_tensor_to_py(m, func_name):
+    writeable = func_name in assert_equal_funcs
+    assert_equal_tensor_ref(getattr(m, func_name)(), writeable=writeable)
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_bad_cpp_to_python_casts(m):
+
+    with pytest.raises(
+        RuntimeError, match="Cannot use reference internal when there is no parent"
+    ):
+        m.reference_tensor_internal()
+
+    with pytest.raises(RuntimeError, match="Cannot move from a constant reference"):
+        m.move_const_tensor()
+
+    with pytest.raises(
+        RuntimeError, match="Cannot take ownership of a const reference"
+    ):
+        m.take_const_tensor()
+
+    with pytest.raises(
+        RuntimeError,
+        match="Invalid return_value_policy for Eigen Map type, must be either reference or reference_internal",
+    ):
+        m.take_view_tensor()
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_bad_python_to_cpp_casts(m):
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_tensor\(\): incompatible function arguments"
+    ):
+        m.round_trip_tensor(np.zeros((2, 3)))
+
+    with pytest.raises(TypeError, match=r"^Cannot cast array data from dtype"):
+        m.round_trip_tensor(np.zeros(dtype=np.str_, shape=(2, 3, 1)))
+
+    with pytest.raises(
+        TypeError,
+        match=r"^round_trip_tensor_noconvert\(\): incompatible function arguments",
+    ):
+        m.round_trip_tensor_noconvert(tensor_ref)
+
+    assert_equal_tensor_ref(
+        m.round_trip_tensor_noconvert(tensor_ref.astype(np.float64))
+    )
+
+    if m.needed_options == "F":
+        bad_options = "C"
+    else:
+        bad_options = "F"
+    # Shape, dtype and the order need to be correct for a TensorMap cast
+    with pytest.raises(
+        TypeError, match=r"^round_trip_view_tensor\(\): incompatible function arguments"
+    ):
+        m.round_trip_view_tensor(
+            np.zeros((3, 5, 2), dtype=np.float64, order=bad_options)
+        )
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_view_tensor\(\): incompatible function arguments"
+    ):
+        m.round_trip_view_tensor(
+            np.zeros((3, 5, 2), dtype=np.float32, order=m.needed_options)
+        )
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_view_tensor\(\): incompatible function arguments"
+    ):
+        m.round_trip_view_tensor(
+            np.zeros((3, 5), dtype=np.float64, order=m.needed_options)
+        )
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_view_tensor\(\): incompatible function arguments"
+    ):
+        temp = np.zeros((3, 5, 2), dtype=np.float64, order=m.needed_options)
+        m.round_trip_view_tensor(
+            temp[:, ::-1, :],
+        )
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_view_tensor\(\): incompatible function arguments"
+    ):
+        temp = np.zeros((3, 5, 2), dtype=np.float64, order=m.needed_options)
+        temp.setflags(write=False)
+        m.round_trip_view_tensor(temp)
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_references_actually_refer(m):
+
+    a = m.reference_tensor()
+    temp = a[indices]
+    a[indices] = 100
+    assert_equal_tensor_ref(m.copy_const_tensor(), modified=100)
+    a[indices] = temp
+    assert_equal_tensor_ref(m.copy_const_tensor())
+
+    a = m.reference_view_of_tensor()
+    a[indices] = 100
+    assert_equal_tensor_ref(m.copy_const_tensor(), modified=100)
+    a[indices] = temp
+    assert_equal_tensor_ref(m.copy_const_tensor())
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_round_trip(m):
+
+    assert_equal_tensor_ref(m.round_trip_tensor(tensor_ref))
+
+    with pytest.raises(TypeError, match="^Cannot cast array data from"):
+        assert_equal_tensor_ref(m.round_trip_tensor2(tensor_ref))
+
+    assert_equal_tensor_ref(m.round_trip_tensor2(np.array(tensor_ref, dtype=np.int32)))
+    assert_equal_tensor_ref(m.round_trip_fixed_tensor(tensor_ref))
+    assert_equal_tensor_ref(m.round_trip_aligned_view_tensor(m.reference_tensor()))
+
+    copy = np.array(tensor_ref, dtype=np.float64, order=m.needed_options)
+    assert_equal_tensor_ref(m.round_trip_view_tensor(copy))
+    assert_equal_tensor_ref(m.round_trip_view_tensor_ref(copy))
+    assert_equal_tensor_ref(m.round_trip_view_tensor_ptr(copy))
+    copy.setflags(write=False)
+    assert_equal_tensor_ref(m.round_trip_const_view_tensor(copy))
+
+    np.testing.assert_array_equal(
+        tensor_ref[:, ::-1, :], m.round_trip_tensor(tensor_ref[:, ::-1, :])
+    )
+
+    assert m.round_trip_rank_0(np.float64(3.5)) == 3.5
+    assert m.round_trip_rank_0(3.5) == 3.5
+
+    with pytest.raises(
+        TypeError,
+        match=r"^round_trip_rank_0_noconvert\(\): incompatible function arguments",
+    ):
+        m.round_trip_rank_0_noconvert(np.float64(3.5))
+
+    with pytest.raises(
+        TypeError,
+        match=r"^round_trip_rank_0_noconvert\(\): incompatible function arguments",
+    ):
+        m.round_trip_rank_0_noconvert(3.5)
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_rank_0_view\(\): incompatible function arguments"
+    ):
+        m.round_trip_rank_0_view(np.float64(3.5))
+
+    with pytest.raises(
+        TypeError, match=r"^round_trip_rank_0_view\(\): incompatible function arguments"
+    ):
+        m.round_trip_rank_0_view(3.5)
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_round_trip_references_actually_refer(m):
+
+    # Need to create a copy that matches the type on the C side
+    copy = np.array(tensor_ref, dtype=np.float64, order=m.needed_options)
+    a = m.round_trip_view_tensor(copy)
+    temp = a[indices]
+    a[indices] = 100
+    assert_equal_tensor_ref(copy, modified=100)
+    a[indices] = temp
+    assert_equal_tensor_ref(copy)
+
+
+@pytest.mark.parametrize("m", submodules)
+def test_doc_string(m, doc):
+    assert (
+        doc(m.copy_tensor) == "copy_tensor() -> numpy.ndarray[numpy.float64[?, ?, ?]]"
+    )
+    assert (
+        doc(m.copy_fixed_tensor)
+        == "copy_fixed_tensor() -> numpy.ndarray[numpy.float64[3, 5, 2]]"
+    )
+    assert (
+        doc(m.reference_const_tensor)
+        == "reference_const_tensor() -> numpy.ndarray[numpy.float64[?, ?, ?]]"
+    )
+
+    order_flag = f"flags.{m.needed_options.lower()}_contiguous"
+    assert doc(m.round_trip_view_tensor) == (
+        f"round_trip_view_tensor(arg0: numpy.ndarray[numpy.float64[?, ?, ?], flags.writeable, {order_flag}])"
+        + f" -> numpy.ndarray[numpy.float64[?, ?, ?], flags.writeable, {order_flag}]"
+    )
+    assert doc(m.round_trip_const_view_tensor) == (
+        f"round_trip_const_view_tensor(arg0: numpy.ndarray[numpy.float64[?, ?, ?], {order_flag}])"
+        + " -> numpy.ndarray[numpy.float64[?, ?, ?]]"
+    )

tests/test_eigen_tensor_avoid_stl_array.cpp

@@ -0,0 +1,16 @@
+/*
+    tests/eigen_tensor.cpp -- automatic conversion of Eigen Tensor
+
+    All rights reserved. Use of this source code is governed by a
+    BSD-style license that can be found in the LICENSE file.
+*/
+
+constexpr const char *test_eigen_tensor_module_name = "eigen_tensor_avoid_stl_array";
+
+#ifndef EIGEN_AVOID_STL_ARRAY
+#    define EIGEN_AVOID_STL_ARRAY
+#endif
+
+#define PYBIND11_TEST_EIGEN_TENSOR_NAMESPACE eigen_tensor_avoid_stl_array
+
+#include "test_eigen_tensor.inl"

tests/test_embed/test_interpreter.cpp

@@ -293,7 +293,6 @@ TEST_CASE("Threads") {
 
     {
         py::gil_scoped_release gil_release{};
-        REQUIRE(has_pybind11_internals_static());
 
         auto threads = std::vector<std::thread>();
         for (auto i = 0; i < num_threads; ++i) {

tests/test_exceptions.cpp

@@ -105,11 +105,6 @@ struct PythonAlreadySetInDestructor {
     py::str s;
 };
 
-std::string error_already_set_what(const py::object &exc_type, const py::object &exc_value) {
-    PyErr_SetObject(exc_type.ptr(), exc_value.ptr());
-    return py::error_already_set().what();
-}
-
 TEST_SUBMODULE(exceptions, m) {
     m.def("throw_std_exception",
           []() { throw std::runtime_error("This exception was intentionally thrown."); });

tests/test_exceptions.py

@@ -275,6 +275,20 @@ def test_local_translator(msg):
     assert msg(excinfo.value) == "this mod"
 
 
+def test_error_already_set_message_with_unicode_surrogate():  # Issue #4288
+    assert m.error_already_set_what(RuntimeError, "\ud927") == (
+        "RuntimeError: \\ud927",
+        False,
+    )
+
+
+def test_error_already_set_message_with_malformed_utf8():
+    assert m.error_already_set_what(RuntimeError, b"\x80") == (
+        "RuntimeError: b'\\x80'",
+        False,
+    )
+
+
 class FlakyException(Exception):
     def __init__(self, failure_point):
         if failure_point == "failure_point_init":

tests/test_gil_scoped.cpp

@@ -11,6 +11,13 @@
 
 #include "pybind11_tests.h"
 
+#include <string>
+#include <thread>
+
+#define CROSS_MODULE(Function)                                                                    \
+    auto cm = py::module_::import("cross_module_gil_utils");                                      \
+    auto target = reinterpret_cast<void (*)()>(PyLong_AsVoidPtr(cm.attr(Function).ptr()));
+
 class VirtClass {
 public:
     virtual ~VirtClass() = default;
@@ -28,6 +35,16 @@ class PyVirtClass : public VirtClass {
 };
 
 TEST_SUBMODULE(gil_scoped, m) {
+    m.attr("defined_THREAD_SANITIZER") =
+#if defined(THREAD_SANITIZER)
+        true;
+#else
+        false;
+#endif
+
+    m.def("intentional_deadlock",
+          []() { std::thread([]() { py::gil_scoped_acquire gil_acquired; }).join(); });
+
     py::class_<VirtClass, PyVirtClass>(m, "VirtClass")
         .def(py::init<>())
         .def("virtual_func", &VirtClass::virtual_func)
@@ -37,11 +54,91 @@ TEST_SUBMODULE(gil_scoped, m) {
     m.def("test_callback_std_func", [](const std::function<void()> &func) { func(); });
     m.def("test_callback_virtual_func", [](VirtClass &virt) { virt.virtual_func(); });
     m.def("test_callback_pure_virtual_func", [](VirtClass &virt) { virt.pure_virtual_func(); });
-    m.def("test_cross_module_gil", []() {
+    m.def("test_cross_module_gil_released", []() {
-        auto cm = py::module_::import("cross_module_gil_utils");
+        CROSS_MODULE("gil_acquire_funcaddr")
-        auto gil_acquire = reinterpret_cast<void (*)()>(
-            PyLong_AsVoidPtr(cm.attr("gil_acquire_funcaddr").ptr()));
         py::gil_scoped_release gil_release;
-        gil_acquire();
+        target();
+    });
+    m.def("test_cross_module_gil_acquired", []() {
+        CROSS_MODULE("gil_acquire_funcaddr")
+        py::gil_scoped_acquire gil_acquire;
+        target();
+    });
+    m.def("test_cross_module_gil_inner_custom_released", []() {
+        CROSS_MODULE("gil_acquire_inner_custom_funcaddr")
+        py::gil_scoped_release gil_release;
+        target();
+    });
+    m.def("test_cross_module_gil_inner_custom_acquired", []() {
+        CROSS_MODULE("gil_acquire_inner_custom_funcaddr")
+        py::gil_scoped_acquire gil_acquire;
+        target();
+    });
+    m.def("test_cross_module_gil_inner_pybind11_released", []() {
+        CROSS_MODULE("gil_acquire_inner_pybind11_funcaddr")
+        py::gil_scoped_release gil_release;
+        target();
+    });
+    m.def("test_cross_module_gil_inner_pybind11_acquired", []() {
+        CROSS_MODULE("gil_acquire_inner_pybind11_funcaddr")
+        py::gil_scoped_acquire gil_acquire;
+        target();
+    });
+    m.def("test_cross_module_gil_nested_custom_released", []() {
+        CROSS_MODULE("gil_acquire_nested_custom_funcaddr")
+        py::gil_scoped_release gil_release;
+        target();
+    });
+    m.def("test_cross_module_gil_nested_custom_acquired", []() {
+        CROSS_MODULE("gil_acquire_nested_custom_funcaddr")
+        py::gil_scoped_acquire gil_acquire;
+        target();
+    });
+    m.def("test_cross_module_gil_nested_pybind11_released", []() {
+        CROSS_MODULE("gil_acquire_nested_pybind11_funcaddr")
+        py::gil_scoped_release gil_release;
+        target();
+    });
+    m.def("test_cross_module_gil_nested_pybind11_acquired", []() {
+        CROSS_MODULE("gil_acquire_nested_pybind11_funcaddr")
+        py::gil_scoped_acquire gil_acquire;
+        target();
+    });
+    m.def("test_release_acquire", [](const py::object &obj) {
+        py::gil_scoped_release gil_released;
+        py::gil_scoped_acquire gil_acquired;
+        return py::str(obj);
+    });
+    m.def("test_nested_acquire", [](const py::object &obj) {
+        py::gil_scoped_release gil_released;
+        py::gil_scoped_acquire gil_acquired_outer;
+        py::gil_scoped_acquire gil_acquired_inner;
+        return py::str(obj);
+    });
+    m.def("test_multi_acquire_release_cross_module", [](unsigned bits) {
+        py::set internals_ids;
+        internals_ids.add(PYBIND11_INTERNALS_ID);
+        {
+            py::gil_scoped_release gil_released;
+            auto thread_f = [bits, &internals_ids]() {
+                py::gil_scoped_acquire gil_acquired;
+                auto cm = py::module_::import("cross_module_gil_utils");
+                auto target = reinterpret_cast<std::string (*)(unsigned)>(
+                    PyLong_AsVoidPtr(cm.attr("gil_multi_acquire_release_funcaddr").ptr()));
+                std::string cm_internals_id = target(bits >> 3);
+                internals_ids.add(cm_internals_id);
+            };
+            if ((bits & 0x1u) != 0u) {
+                thread_f();
+            }
+            if ((bits & 0x2u) != 0u) {
+                std::thread non_python_thread(thread_f);
+                non_python_thread.join();
+            }
+            if ((bits & 0x4u) != 0u) {
+                thread_f();
+            }
+        }
+        return internals_ids;
     });
 }

tests/test_gil_scoped.py

@@ -1,26 +1,13 @@
 import multiprocessing
+import sys
 import threading
+import time
+
+import pytest
 
 from pybind11_tests import gil_scoped as m
 
 
-def _run_in_process(target, *args, **kwargs):
-    """Runs target in process and returns its exitcode after 10s (None if still alive)."""
-    process = multiprocessing.Process(target=target, args=args, kwargs=kwargs)
-    process.daemon = True
-    try:
-        process.start()
-        # Do not need to wait much, 10s should be more than enough.
-        process.join(timeout=10)
-        return process.exitcode
-    finally:
-        if process.is_alive():
-            process.terminate()
-
-
-def _python_to_cpp_to_python():
-    """Calls different C++ functions that come back to Python."""
-
 class ExtendedVirtClass(m.VirtClass):
     def virtual_func(self):
         pass
@@ -28,18 +15,185 @@ def _python_to_cpp_to_python():
     def pure_virtual_func(self):
         pass
 
-    extended = ExtendedVirtClass()
+
+def test_callback_py_obj():
     m.test_callback_py_obj(lambda: None)
+
+
+def test_callback_std_func():
     m.test_callback_std_func(lambda: None)
+
+
+def test_callback_virtual_func():
+    extended = ExtendedVirtClass()
     m.test_callback_virtual_func(extended)
+
+
+def test_callback_pure_virtual_func():
+    extended = ExtendedVirtClass()
     m.test_callback_pure_virtual_func(extended)
 
 
-def _python_to_cpp_to_python_from_threads(num_threads, parallel=False):
+def test_cross_module_gil_released():
-    """Calls different C++ functions that come back to Python, from Python threads."""
+    """Makes sure that the GIL can be acquired by another module from a GIL-released state."""
+    m.test_cross_module_gil_released()  # Should not raise a SIGSEGV
+
+
+def test_cross_module_gil_acquired():
+    """Makes sure that the GIL can be acquired by another module from a GIL-acquired state."""
+    m.test_cross_module_gil_acquired()  # Should not raise a SIGSEGV
+
+
+def test_cross_module_gil_inner_custom_released():
+    """Makes sure that the GIL can be acquired/released by another module
+    from a GIL-released state using custom locking logic."""
+    m.test_cross_module_gil_inner_custom_released()
+
+
+def test_cross_module_gil_inner_custom_acquired():
+    """Makes sure that the GIL can be acquired/acquired by another module
+    from a GIL-acquired state using custom locking logic."""
+    m.test_cross_module_gil_inner_custom_acquired()
+
+
+def test_cross_module_gil_inner_pybind11_released():
+    """Makes sure that the GIL can be acquired/released by another module
+    from a GIL-released state using pybind11 locking logic."""
+    m.test_cross_module_gil_inner_pybind11_released()
+
+
+def test_cross_module_gil_inner_pybind11_acquired():
+    """Makes sure that the GIL can be acquired/acquired by another module
+    from a GIL-acquired state using pybind11 locking logic."""
+    m.test_cross_module_gil_inner_pybind11_acquired()
+
+
+def test_cross_module_gil_nested_custom_released():
+    """Makes sure that the GIL can be nested acquired/released by another module
+    from a GIL-released state using custom locking logic."""
+    m.test_cross_module_gil_nested_custom_released()
+
+
+def test_cross_module_gil_nested_custom_acquired():
+    """Makes sure that the GIL can be nested acquired/acquired by another module
+    from a GIL-acquired state using custom locking logic."""
+    m.test_cross_module_gil_nested_custom_acquired()
+
+
+def test_cross_module_gil_nested_pybind11_released():
+    """Makes sure that the GIL can be nested acquired/released by another module
+    from a GIL-released state using pybind11 locking logic."""
+    m.test_cross_module_gil_nested_pybind11_released()
+
+
+def test_cross_module_gil_nested_pybind11_acquired():
+    """Makes sure that the GIL can be nested acquired/acquired by another module
+    from a GIL-acquired state using pybind11 locking logic."""
+    m.test_cross_module_gil_nested_pybind11_acquired()
+
+
+def test_release_acquire():
+    assert m.test_release_acquire(0xAB) == "171"
+
+
+def test_nested_acquire():
+    assert m.test_nested_acquire(0xAB) == "171"
+
+
+def test_multi_acquire_release_cross_module():
+    for bits in range(16 * 8):
+        internals_ids = m.test_multi_acquire_release_cross_module(bits)
+        assert len(internals_ids) == 2 if bits % 8 else 1
+
+
+# Intentionally putting human review in the loop here, to guard against accidents.
+VARS_BEFORE_ALL_BASIC_TESTS = dict(vars())  # Make a copy of the dict (critical).
+ALL_BASIC_TESTS = (
+    test_callback_py_obj,
+    test_callback_std_func,
+    test_callback_virtual_func,
+    test_callback_pure_virtual_func,
+    test_cross_module_gil_released,
+    test_cross_module_gil_acquired,
+    test_cross_module_gil_inner_custom_released,
+    test_cross_module_gil_inner_custom_acquired,
+    test_cross_module_gil_inner_pybind11_released,
+    test_cross_module_gil_inner_pybind11_acquired,
+    test_cross_module_gil_nested_custom_released,
+    test_cross_module_gil_nested_custom_acquired,
+    test_cross_module_gil_nested_pybind11_released,
+    test_cross_module_gil_nested_pybind11_acquired,
+    test_release_acquire,
+    test_nested_acquire,
+    test_multi_acquire_release_cross_module,
+)
+
+
+def test_all_basic_tests_completeness():
+    num_found = 0
+    for key, value in VARS_BEFORE_ALL_BASIC_TESTS.items():
+        if not key.startswith("test_"):
+            continue
+        assert value in ALL_BASIC_TESTS
+        num_found += 1
+    assert len(ALL_BASIC_TESTS) == num_found
+
+
+def _intentional_deadlock():
+    m.intentional_deadlock()
+
+
+ALL_BASIC_TESTS_PLUS_INTENTIONAL_DEADLOCK = ALL_BASIC_TESTS + (_intentional_deadlock,)
+SKIP_IF_DEADLOCK = True  # See PR #4216
+
+
+def _run_in_process(target, *args, **kwargs):
+    if len(args) == 0:
+        test_fn = target
+    else:
+        test_fn = args[0]
+    # Do not need to wait much, 10s should be more than enough.
+    timeout = 0.1 if test_fn is _intentional_deadlock else 10
+    process = multiprocessing.Process(target=target, args=args, kwargs=kwargs)
+    process.daemon = True
+    try:
+        t_start = time.time()
+        process.start()
+        if timeout >= 100:  # For debugging.
+            print(
+                "\nprocess.pid STARTED", process.pid, (sys.argv, target, args, kwargs)
+            )
+            print(f"COPY-PASTE-THIS: gdb {sys.argv[0]} -p {process.pid}", flush=True)
+        process.join(timeout=timeout)
+        if timeout >= 100:
+            print("\nprocess.pid JOINED", process.pid, flush=True)
+        t_delta = time.time() - t_start
+        if process.exitcode == 66 and m.defined_THREAD_SANITIZER:  # Issue #2754
+            # WOULD-BE-NICE-TO-HAVE: Check that the message below is actually in the output.
+            # Maybe this could work:
+            # https://gist.github.com/alexeygrigorev/01ce847f2e721b513b42ea4a6c96905e
+            pytest.skip(
+                "ThreadSanitizer: starting new threads after multi-threaded fork is not supported."
+            )
+        elif test_fn is _intentional_deadlock:
+            assert process.exitcode is None
+            return 0
+        elif process.exitcode is None:
+            assert t_delta > 0.9 * timeout
+            msg = "DEADLOCK, most likely, exactly what this test is meant to detect."
+            if SKIP_IF_DEADLOCK:
+                pytest.skip(msg)
+            raise RuntimeError(msg)
+        return process.exitcode
+    finally:
+        if process.is_alive():
+            process.terminate()
+
+
+def _run_in_threads(test_fn, num_threads, parallel):
     threads = []
     for _ in range(num_threads):
-        thread = threading.Thread(target=_python_to_cpp_to_python)
+        thread = threading.Thread(target=test_fn)
         thread.daemon = True
         thread.start()
         if parallel:
@@ -51,43 +205,40 @@ def _python_to_cpp_to_python_from_threads(num_threads, parallel=False):
 
 
 # TODO: FIXME, sometimes returns -11 (segfault) instead of 0 on macOS Python 3.9
-def test_python_to_cpp_to_python_from_thread():
+@pytest.mark.parametrize("test_fn", ALL_BASIC_TESTS_PLUS_INTENTIONAL_DEADLOCK)
+def test_run_in_process_one_thread(test_fn):
     """Makes sure there is no GIL deadlock when running in a thread.
 
     It runs in a separate process to be able to stop and assert if it deadlocks.
     """
-    assert _run_in_process(_python_to_cpp_to_python_from_threads, 1) == 0
+    assert _run_in_process(_run_in_threads, test_fn, num_threads=1, parallel=False) == 0
 
 
 # TODO: FIXME on macOS Python 3.9
-def test_python_to_cpp_to_python_from_thread_multiple_parallel():
+@pytest.mark.parametrize("test_fn", ALL_BASIC_TESTS_PLUS_INTENTIONAL_DEADLOCK)
+def test_run_in_process_multiple_threads_parallel(test_fn):
     """Makes sure there is no GIL deadlock when running in a thread multiple times in parallel.
 
     It runs in a separate process to be able to stop and assert if it deadlocks.
     """
-    assert _run_in_process(_python_to_cpp_to_python_from_threads, 8, parallel=True) == 0
+    assert _run_in_process(_run_in_threads, test_fn, num_threads=8, parallel=True) == 0
 
 
 # TODO: FIXME on macOS Python 3.9
-def test_python_to_cpp_to_python_from_thread_multiple_sequential():
+@pytest.mark.parametrize("test_fn", ALL_BASIC_TESTS_PLUS_INTENTIONAL_DEADLOCK)
+def test_run_in_process_multiple_threads_sequential(test_fn):
     """Makes sure there is no GIL deadlock when running in a thread multiple times sequentially.
 
     It runs in a separate process to be able to stop and assert if it deadlocks.
     """
-    assert (
+    assert _run_in_process(_run_in_threads, test_fn, num_threads=8, parallel=False) == 0
-        _run_in_process(_python_to_cpp_to_python_from_threads, 8, parallel=False) == 0
-    )
 
 
 # TODO: FIXME on macOS Python 3.9
-def test_python_to_cpp_to_python_from_process():
+@pytest.mark.parametrize("test_fn", ALL_BASIC_TESTS_PLUS_INTENTIONAL_DEADLOCK)
+def test_run_in_process_direct(test_fn):
     """Makes sure there is no GIL deadlock when using processes.
 
     This test is for completion, but it was never an issue.
     """
-    assert _run_in_process(_python_to_cpp_to_python) == 0
+    assert _run_in_process(test_fn) == 0
-
-
-def test_cross_module_gil():
-    """Makes sure that the GIL can be acquired by another module from a GIL-released state."""
-    m.test_cross_module_gil()  # Should not raise a SIGSEGV

tests/test_numpy_array.cpp

@@ -521,4 +521,6 @@ TEST_SUBMODULE(numpy_array, sm) {
     sm.def("test_fmt_desc_double", [](const py::array_t<double> &) {});
     sm.def("test_fmt_desc_const_float", [](const py::array_t<const float> &) {});
     sm.def("test_fmt_desc_const_double", [](const py::array_t<const double> &) {});
+
+    sm.def("round_trip_float", [](double d) { return d; });
 }

tests/test_numpy_array.py

@@ -585,3 +585,9 @@ def test_dtype_refcount_leak():
     m.ndim(a)
     after = getrefcount(dtype)
     assert after == before
+
+
+def test_round_trip_float():
+    arr = np.zeros((), np.float64)
+    arr[()] = 37.2
+    assert m.round_trip_float(arr) == 37.2

tests/test_pytypes.cpp

@@ -183,7 +183,7 @@ TEST_SUBMODULE(pytypes, m) {
         return d2;
     });
     m.def("dict_contains",
-          [](const py::dict &dict, py::object val) { return dict.contains(val); });
+          [](const py::dict &dict, const py::object &val) { return dict.contains(val); });
     m.def("dict_contains",
           [](const py::dict &dict, const char *val) { return dict.contains(val); });
 
@@ -206,7 +206,12 @@ TEST_SUBMODULE(pytypes, m) {
     m.def("str_from_char_ssize_t", []() { return py::str{"red", (py::ssize_t) 3}; });
     m.def("str_from_char_size_t", []() { return py::str{"blue", (py::size_t) 4}; });
     m.def("str_from_string", []() { return py::str(std::string("baz")); });
+    m.def("str_from_std_string_input", [](const std::string &stri) { return py::str(stri); });
+    m.def("str_from_cstr_input", [](const char *c_str) { return py::str(c_str); });
     m.def("str_from_bytes", []() { return py::str(py::bytes("boo", 3)); });
+    m.def("str_from_bytes_input",
+          [](const py::bytes &encoded_str) { return py::str(encoded_str); });
+
     m.def("str_from_object", [](const py::object &obj) { return py::str(obj); });
     m.def("repr_from_object", [](const py::object &obj) { return py::repr(obj); });
     m.def("str_from_handle", [](py::handle h) { return py::str(h); });
@@ -538,6 +543,9 @@ TEST_SUBMODULE(pytypes, m) {
 
     m.def("hash_function", [](py::object obj) { return py::hash(std::move(obj)); });
 
+    m.def("obj_contains",
+          [](py::object &obj, const py::object &key) { return obj.contains(key); });
+
     m.def("test_number_protocol", [](const py::object &a, const py::object &b) {
         py::list l;
         l.append(a.equal(b));

tests/test_pytypes.py

@@ -168,6 +168,31 @@ def test_dict(capture, doc):
     assert m.dict_keyword_constructor() == {"x": 1, "y": 2, "z": 3}
 
 
+class CustomContains:
+    d = {"key": None}
+
+    def __contains__(self, m):
+        return m in self.d
+
+
+@pytest.mark.parametrize(
+    "arg,func",
+    [
+        (set(), m.anyset_contains),
+        (dict(), m.dict_contains),
+        (CustomContains(), m.obj_contains),
+    ],
+)
+@pytest.mark.xfail("env.PYPY and sys.pypy_version_info < (7, 3, 10)", strict=False)
+def test_unhashable_exceptions(arg, func):
+    class Unhashable:
+        __hash__ = None
+
+    with pytest.raises(TypeError) as exc_info:
+        func(arg, Unhashable())
+    assert "unhashable type:" in str(exc_info.value)
+
+
 def test_tuple():
     assert m.tuple_no_args() == ()
     assert m.tuple_ssize_t() == ()
@@ -219,6 +244,20 @@ def test_str(doc):
         m.str_from_string_from_str(ucs_surrogates_str)
 
 
+@pytest.mark.parametrize(
+    "func",
+    [
+        m.str_from_bytes_input,
+        m.str_from_cstr_input,
+        m.str_from_std_string_input,
+    ],
+)
+def test_surrogate_pairs_unicode_error(func):
+    input_str = "\ud83d\ude4f".encode("utf-8", "surrogatepass")
+    with pytest.raises(UnicodeDecodeError):
+        func(input_str)
+
+
 def test_bytes(doc):
     assert m.bytes_from_char_ssize_t().decode() == "green"
     assert m.bytes_from_char_size_t().decode() == "purple"

tests/test_sequences_and_iterators.cpp

@@ -559,4 +559,23 @@ TEST_SUBMODULE(sequences_and_iterators, m) {
           []() { return py::make_iterator<py::return_value_policy::copy>(list); });
     m.def("make_iterator_2",
           []() { return py::make_iterator<py::return_value_policy::automatic>(list); });
+
+    // test_iterator on c arrays
+    // #4100: ensure lvalue required as increment operand
+    class CArrayHolder {
+    public:
+        CArrayHolder(double x, double y, double z) {
+            values[0] = x;
+            values[1] = y;
+            values[2] = z;
+        };
+        double values[3];
+    };
+
+    py::class_<CArrayHolder>(m, "CArrayHolder")
+        .def(py::init<double, double, double>())
+        .def(
+            "__iter__",
+            [](const CArrayHolder &v) { return py::make_iterator(v.values, v.values + 3); },
+            py::keep_alive<0, 1>());
 }

tests/test_sequences_and_iterators.py

@@ -241,3 +241,11 @@ def test_iterator_rvp():
     assert list(m.make_iterator_1()) == [1, 2, 3]
     assert list(m.make_iterator_2()) == [1, 2, 3]
     assert not isinstance(m.make_iterator_1(), type(m.make_iterator_2()))
+
+
+def test_carray_iterator():
+    """#4100: Check for proper iterator overload with C-Arrays"""
+    args_gt = list(float(i) for i in range(3))
+    arr_h = m.CArrayHolder(*args_gt)
+    args = list(arr_h)
+    assert args_gt == args

tools/FindPythonLibsNew.cmake

@@ -151,9 +151,13 @@ if(NOT _PYTHON_SUCCESS MATCHES 0)
   return()
 endif()
 
+option(
+  PYBIND11_PYTHONLIBS_OVERWRITE
+  "Overwrite cached values read from Python library (classic search). Turn off if cross-compiling and manually setting these values."
+  ON)
 # Can manually set values when cross-compiling
 macro(_PYBIND11_GET_IF_UNDEF lst index name)
-  if(NOT DEFINED "${name}")
+  if(PYBIND11_PYTHONLIBS_OVERWRITE OR NOT DEFINED "${name}")
     list(GET "${lst}" "${index}" "${name}")
   endif()
 endmacro()

tools/setup_global.py.in

@@ -27,10 +27,11 @@ class InstallHeadersNested(install_headers):
 
 main_headers = glob.glob("pybind11/include/pybind11/*.h")
 detail_headers = glob.glob("pybind11/include/pybind11/detail/*.h")
+eigen_headers = glob.glob("pybind11/include/pybind11/eigen/*.h")
 stl_headers = glob.glob("pybind11/include/pybind11/stl/*.h")
 cmake_files = glob.glob("pybind11/share/cmake/pybind11/*.cmake")
 pkgconfig_files = glob.glob("pybind11/share/pkgconfig/*.pc")
-headers = main_headers + detail_headers + stl_headers
+headers = main_headers + detail_headers + stl_headers + eigen_headers
 
 cmdclass = {"install_headers": InstallHeadersNested}
 $extra_cmd
@@ -55,6 +56,7 @@ setup(
         (base + "share/pkgconfig", pkgconfig_files),
         (base + "include/pybind11", main_headers),
         (base + "include/pybind11/detail", detail_headers),
+        (base + "include/pybind11/eigen", eigen_headers),
         (base + "include/pybind11/stl", stl_headers),
     ],
     cmdclass=cmdclass,

tools/setup_main.py.in

@@ -15,6 +15,7 @@ setup(
         "pybind11",
         "pybind11.include.pybind11",
         "pybind11.include.pybind11.detail",
+        "pybind11.include.pybind11.eigen",
         "pybind11.include.pybind11.stl",
         "pybind11.share.cmake.pybind11",
         "pybind11.share.pkgconfig",
@@ -23,6 +24,7 @@ setup(
         "pybind11": ["py.typed"],
         "pybind11.include.pybind11": ["*.h"],
         "pybind11.include.pybind11.detail": ["*.h"],
+        "pybind11.include.pybind11.eigen": ["*.h"],
         "pybind11.include.pybind11.stl": ["*.h"],
         "pybind11.share.cmake.pybind11": ["*.cmake"],
         "pybind11.share.pkgconfig": ["*.pc"],