feat: integrate numpy as statically-linked built-in module

.gitmodules

@@ -2,9 +2,14 @@
 	path = deps/cpython
 	url = https://github.com/nanvix/cpython.git
 	branch = nanvix/v3.12.3
-        shallow = true
+	shallow = true
 [submodule "deps/cymem"]
-        path = deps/cymem
-        url = https://github.com/nanvix/cymem.git
-        branch = nanvix/v2.0.11
-        shallow = true
+	path = deps/cymem
+	url = https://github.com/nanvix/cymem.git
+	branch = nanvix/v2.0.11
+	shallow = true
+[submodule "deps/numpy"]
+	path = deps/numpy
+	url = https://github.com/nanvix/numpy.git
+	branch = nanvix/v1.26.4
+	shallow = true

patches/nanvix_numpy_bootstrap.py

@@ -0,0 +1,89 @@
+"""
+Lazy import redirector for NumPy built-in modules on Nanvix.
+"""
+import importlib
+import importlib.abc
+import importlib.machinery
+import sys
+import types
+
+_ALIASES = {
+    "numpy.core._multiarray_umath": "_np_multiarray_umath",
+    "numpy._core._multiarray_umath": "_np_multiarray_umath",
+    "_multiarray_umath": "_np_multiarray_umath",
+}
+
+# C extension modules that are not statically linked but numpy tries
+# to import them during init.  Provide permissive stubs so init succeeds.
+_STUBS = {
+    "numpy.core._multiarray_tests",
+    "numpy.core._simd",
+    "numpy.linalg._umath_linalg",
+    "numpy.linalg.lapack_lite",
+    "numpy.fft._pocketfft_internal",
+    "numpy.random._common",
+    "numpy.random._bounded_integers",
+    "numpy.random._generator",
+    "numpy.random._mt19937",
+    "numpy.random._pcg64",
+    "numpy.random._philox",
+    "numpy.random._sfc64",
+    "numpy.random.bit_generator",
+    "numpy.random.mtrand",
+}
+
+_loading = set()
+
+
+class _StubModule(types.ModuleType):
+    __all__ = []
+    """Module stub that returns None for any missing attribute."""
+    def __getattr__(self, name):
+        return None
+
+
+class _NanvixLoader(importlib.abc.Loader):
+    def __init__(self, fullname):
+        self._fullname = fullname
+
+    def create_module(self, spec):
+        fullname = self._fullname
+
+        if fullname in _STUBS:
+            mod = _StubModule(fullname)
+            mod.__file__ = "<nanvix-stub>"
+            return mod
+
+        builtin_name = _ALIASES.get(fullname)
+        if not builtin_name:
+            return None
+
+        # Try sys.modules first (C-level pre-registration)
+        mod = sys.modules.get(builtin_name)
+        if mod is not None:
+            return mod
+
+        _loading.add(fullname)
+        try:
+            mod = importlib.import_module(builtin_name)
+        finally:
+            _loading.discard(fullname)
+        return mod
+
+    def exec_module(self, module):
+        fullname = self._fullname
+        sys.modules[fullname] = module
+        if fullname != "_multiarray_umath" and fullname in _ALIASES:
+            sys.modules["_multiarray_umath"] = module
+
+
+class _NanvixNumpyFinder(importlib.abc.MetaPathFinder):
+    def find_spec(self, fullname, path, target=None):
+        if fullname in _STUBS:
+            return importlib.machinery.ModuleSpec(fullname, _NanvixLoader(fullname))
+        if fullname in _ALIASES and fullname not in _loading:
+            return importlib.machinery.ModuleSpec(fullname, _NanvixLoader(fullname))
+        return None
+
+
+sys.meta_path.insert(0, _NanvixNumpyFinder())

patches/numpy_libm_compat.c

@@ -0,0 +1,8 @@
+#include <math.h>
+
+long double __kernel_tanl(long double x, long double y, int iy)
+{
+    (void)y;
+    (void)iy;
+    return (long double)tan((double)x);
+}

patches/numpy_makefile_nanvix.patch

@@ -0,0 +1,25 @@
+diff --git a/Makefile.nanvix b/Makefile.nanvix
+index 0f9703e9a2..66c67c2814 100644
+--- a/Makefile.nanvix
++++ b/Makefile.nanvix
+@@ -76,9 +76,9 @@ $(CROSS_FILE):
+ 	@echo '' >> $@
+ 	@echo '[built-in options]' >> $@
+ 	@echo "c_args = ['-I$(SYSROOT_PATH)/include', '-I$(SYSROOT_PATH)/include/python3.12']" >> $@
+-	@echo "c_link_args = ['-static', '-L$(SYSROOT_PATH)/lib', '-T$(SYSROOT_PATH)/lib/user.ld', '-Wl,--start-group', '$(SYSROOT_PATH)/lib/libposix.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libc.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libm.a', '-Wl,--end-group']" >> $@
++	@echo "c_link_args = ['-static', '-L$(SYSROOT_PATH)/lib', '-T$(SYSROOT_PATH)/lib/user.ld', '-Wl,--allow-multiple-definition', '-Wl,--start-group', '$(SYSROOT_PATH)/lib/libposix.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libc.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libm.a', '-Wl,--end-group']" >> $@
+ 	@echo "cpp_args = ['-I$(SYSROOT_PATH)/include', '-I$(SYSROOT_PATH)/include/python3.12']" >> $@
+-	@echo "cpp_link_args = ['-static', '-L$(SYSROOT_PATH)/lib', '-T$(SYSROOT_PATH)/lib/user.ld', '-Wl,--start-group', '$(SYSROOT_PATH)/lib/libposix.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libc.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libm.a', '-lstdc++', '-Wl,--end-group']" >> $@
++	@echo "cpp_link_args = ['-static', '-L$(SYSROOT_PATH)/lib', '-T$(SYSROOT_PATH)/lib/user.ld', '-Wl,--allow-multiple-definition', '-Wl,--start-group', '$(SYSROOT_PATH)/lib/libposix.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libc.a', '$(TOOLCHAIN_PREFIX)/i686-nanvix/lib/libm.a', '-lstdc++', '-Wl,--end-group']" >> $@
+ 	@echo '' >> $@
+ 	@echo '[host_machine]' >> $@
+ 	@echo "system = 'nanvix'" >> $@
+@@ -95,7 +95,7 @@ configure: $(CONFIGURED_MARKER)
+
+ $(CONFIGURED_MARKER): $(CROSS_FILE)
+ 	PKG_CONFIG_LIBDIR="$(SYSROOT_PATH)/lib/pkgconfig" \
+-	/usr/bin/python3 vendored-meson/meson/meson.py setup $(BUILDDIR) \
++	python3 vendored-meson/meson/meson.py setup $(BUILDDIR) \
+ 	    --cross-file $(CROSS_FILE) \
+ 	    --pkg-config-path="$(SYSROOT_PATH)/lib/pkgconfig" \
+ 	    -Dallow-noblas=true \

patches/numpy_static_builtin.md

@@ -0,0 +1,131 @@
+# NumPy Static Built-in Patch for Nanvix
+
+This patch modifies NumPy 1.26.4 so that its core C extension
+(`_multiarray_umath`) can be **statically linked into the CPython binary**
+as a built-in module on the Nanvix microkernel.
+
+Two files are touched:
+
+| File | Purpose |
+|------|---------|
+| `numpy/core/src/common/npy_cpu_features.c` | CPU feature detection |
+| `numpy/core/src/multiarray/multiarraymodule.c` | Module initialisation |
+
+---
+
+## 1. CPU feature detection (`npy_cpu_features.c`)
+
+### Problem
+
+NumPy probes the host CPU at import time with the privileged `CPUID` and
+`XGETBV` instructions (inline assembly).  On Nanvix these instructions
+are not emulated in user mode and cause the process to **hang
+indefinitely**.
+
+### Fix
+
+Guard both `npy__cpu_cpuid()` and `npy__cpu_getxcr0()` with
+`#ifdef __nanvix__` so that on Nanvix they return zeroed-out results
+instead of executing the instructions.  NumPy then treats the CPU as
+having no optional SIMD extensions, which is correct for the Nanvix
+i686 target.
+
+---
+
+## 2. Module initialisation (`multiarraymodule.c`)
+
+Three separate issues are addressed in this file.
+
+### 2a. Import-lock deadlock during `npy_cache_import`
+
+#### Problem
+
+`initialize_static_globals()` calls
+`npy_cache_import("numpy.exceptions", ...)` and
+`npy_cache_import("numpy.core._exceptions", ...)`.  These trigger
+`import numpy`, which in turn imports `numpy.core._multiarray_umath` —
+the very module whose `PyInit` is currently executing.  Because CPython
+holds a **per-module import lock** for the duration of `PyInit`, the
+re-entrant import attempt deadlocks.
+
+#### Fix
+
+Wrap both `npy_cache_import` calls in `#ifndef __nanvix__`.  The two
+global pointers (`npy_DTypePromotionError`, `npy_UFuncNoLoopError`)
+remain `NULL` after init; they are populated lazily by
+`npy_cache_import`'s own `*cache == NULL` guard on first actual use.
+
+An early-return guard (`if (npy_DTypePromotionError != NULL && …)`) is
+also added so that a re-entrant call into `initialize_static_globals()`
+is a harmless no-op instead of hitting the original `assert()` (which
+would abort the process).
+
+### 2b. Module renamed to `_np_multiarray_umath`
+
+#### Problem
+
+If the built-in is registered under the canonical name
+`_multiarray_umath`, any Python code that does
+`import _multiarray_umath` during `PyInit` will re-enter the same
+`PyInit` function (import-lock deadlock, same as 2a but from a different
+call site).
+
+#### Fix
+
+On Nanvix the module name in `PyModuleDef` is changed to
+`_np_multiarray_umath` and the init function is renamed to
+`PyInit__np_multiarray_umath`.  A meta-path finder installed in
+`sitecustomize.py` at runtime redirects all imports of the dotted name
+(`numpy.core._multiarray_umath`) to the renamed flat built-in, breaking
+the circular chain.
+
+### 2c. Pre-registration in `sys.modules`
+
+#### Problem
+
+CPython only adds a built-in module to `sys.modules` **after** `PyInit`
+returns.  Any Python-level import triggered during `PyInit` (e.g.
+`import datetime` via `PyDateTime_IMPORT`) that transitively tries to
+resolve `numpy.core._multiarray_umath` will not find the module in
+`sys.modules` yet, causing an `ImportError` or a second call to
+`PyInit`.
+
+#### Fix
+
+Immediately after `PyModule_Create`, the partially-initialised module
+object is inserted into `sys.modules` under three keys:
+
+| Key | Why |
+|-----|-----|
+| `_np_multiarray_umath` | The built-in's own name |
+| `_multiarray_umath` | The name NumPy Python code expects |
+| `numpy.core._multiarray_umath` | The fully-qualified dotted path |
+
+This lets the import machinery return the already-created (but not yet
+fully populated) module object instead of calling `PyInit` again.
+
+---
+
+## Runtime companion: `nanvix_numpy_bootstrap.py`
+
+The patch alone is not sufficient — a Python-side `sitecustomize.py`
+module (`patches/nanvix_numpy_bootstrap.py`) is installed alongside
+NumPy.  It provides:
+
+* A **`find_spec`-based meta-path finder** that intercepts
+  `import numpy.core._multiarray_umath` (and similar dotted paths) and
+  redirects them to the renamed `_np_multiarray_umath` built-in.
+* **Permissive stub modules** (`_StubModule`) for C extensions that are
+  not statically linked (e.g. `_multiarray_tests`, `_simd`, linalg,
+  fft, random sub-modules).  These stubs have `__all__ = []` and a
+  `__getattr__` that returns `None`, allowing NumPy's init to complete
+  without those extensions.
+
+---
+
+## Companion files
+
+| File | Role |
+|------|------|
+| `patches/numpy_libm_compat.c` | Provides `__kernel_tanl` stub missing from Nanvix newlib |
+| `patches/nanvix_numpy_bootstrap.py` | Runtime meta-path finder and stub modules |