Optimise the module init function for small code size regardless of the externally provided CFLAGS.

See #2102.

CHANGES.rst

@@ -8,6 +8,10 @@ Cython Changelog
 Features added
 --------------
 
+* When compiling with gcc, the module init function is now tuned for small
+  code size instead of whatever compile flags were provided externally.
+  (Github issue #2102)
+
 * Cdef classes can now multiply inherit from ordinary Python classes.
   (The primary base must still be a c class, possibly ``object``, and
   the other bases must *not* be cdef classes.)

Cython/Compiler/ModuleNode.py

@@ -2279,10 +2279,11 @@ class ModuleNode(Nodes.Node, Nodes.BlockNode):
         header2 = "__Pyx_PyMODINIT_FUNC init%s(void)" % env.module_name
         header3 = "__Pyx_PyMODINIT_FUNC %s(void)" % self.mod_init_func_cname('PyInit', env)
         code.putln("#if PY_MAJOR_VERSION < 3")
-        code.putln("%s; /*proto*/" % header2)
+        # Optimise for small code size as the module init function is only executed once.
+        code.putln("CYTHON_SMALL_CODE %s; /*proto*/" % header2)
         code.putln(header2)
         code.putln("#else")
-        code.putln("%s; /*proto*/" % header3)
+        code.putln("CYTHON_SMALL_CODE %s; /*proto*/" % header3)
         code.putln(header3)
 
         # CPython 3.5+ supports multi-phase module initialisation (gives access to __spec__, __file__, etc.)
@@ -2296,7 +2297,7 @@ class ModuleNode(Nodes.Node, Nodes.BlockNode):
 
         code.putln("")
         # main module init code lives in Py_mod_exec function, not in PyInit function
-        code.putln("static int %s(PyObject *%s)" % (
+        code.putln("CYTHON_SMALL_CODE static int %s(PyObject *%s)" % (
             self.mod_init_func_cname(Naming.pymodule_exec_func_cname, env),
             Naming.pymodinit_module_arg))
         code.putln("#endif")  # PEP489

Cython/Utility/ModuleSetupCode.c

@@ -665,6 +665,12 @@ static CYTHON_INLINE void * PyThread_tss_get(Py_tss_t *key) {
 
 #endif
 
+#if defined(__GNUC__)
+    #define CYTHON_SMALL_CODE __attribute__((optimize("Os")))
+#else
+    #define CYTHON_SMALL_CODE
+#endif
+
 
 /////////////// FastTypeChecks.proto ///////////////