Fixes for the WASI cross-compile build.

Author: nascheme

Tools/configure/conf_init.py

@@ -59,7 +59,11 @@ def setup_source_dirs(v):
     srcdir_abs = pyconf.abspath(srcdir_raw)
     builddir = pyconf.abspath(".")
     out_of_tree = srcdir_abs != builddir
-    v.export("srcdir", "." if not out_of_tree else srcdir_abs)
+    # Use a relative path for out-of-tree builds.  An absolute srcdir would
+    # break WASI/Emscripten where VPATH is baked into the binary but the
+    # host path doesn't exist in the guest sandbox.
+    srcdir_rel = pyconf.relpath(srcdir_abs, builddir) if out_of_tree else "."
+    v.export("srcdir", srcdir_rel)
     v.export("abs_srcdir", srcdir_abs)
     v.export("abs_builddir", builddir)
     pyconf.srcdir = srcdir_abs

Tools/configure/conf_optimization.py

@@ -186,7 +186,7 @@ def setup_pgo_flags(v):
     v.export("LLVM_PROFDATA")
 
     v.LLVM_PROFDATA = pyconf.check_prog(
-        "llvm-profdata", path=v.llvm_path, default=""
+        "llvm-profdata", path=v.llvm_path, default="''"
     )
     v.export("LLVM_PROFDATA")
     LLVM_PROF_FOUND = (

Tools/configure/conf_output.py

@@ -42,10 +42,9 @@ def generate_output(v):
                 "# Edit this file for local setup changes\n",
             )
 
-        # Use relative paths when building in-place (srcdir == builddir)
-        # so that generated files like Modules/config.c get a relative path
-        # comment, matching the autoconf configure behaviour.
-        srcdir_rel = pyconf.relpath(pyconf.srcdir) if pyconf.srcdir else "."
+        # Use the srcdir substitution value (e.g. "." for in-place,
+        # absolute path for out-of-tree) to match autoconf behaviour.
+        srcdir_rel = v.srcdir or "."
         if not pyconf.cmd(
             [
                 "/bin/sh",

Tools/configure/conf_platform.py

@@ -91,13 +91,15 @@ def setup_host_prefix(v):
     if not v.host_prefix:
         v.export(
             "host_prefix",
-            "/" if v.ac_sys_system == "Emscripten" else "${prefix}",
+            "/"
+            if v.ac_sys_system in ("Emscripten", "WASI")
+            else "${prefix}",
         )
 
     if not v.host_exec_prefix:
         v.host_exec_prefix = (
             v.host_prefix
-            if v.ac_sys_system == "Emscripten"
+            if v.ac_sys_system in ("Emscripten", "WASI")
             else "${exec_prefix}"
         )
     v.export("host_exec_prefix")
@@ -431,8 +433,10 @@ def check_declarations(v):
         extra_includes=["sys/param.h"],
     )
 
+    # AC_CHECK_DECLS always defines HAVE_DECL_<NAME> to 0 or 1.
     pyconf.check_decl(
         "UT_NAMESIZE",
+        define_name="HAVE_DECL_UT_NAMESIZE",
         on_found=_define_ut_namesize,
         extra_includes=["utmp.h"],
     )
@@ -441,6 +445,7 @@ def check_declarations(v):
     if v.ac_cv_libc != "musl":
         pyconf.check_decl(
             "PR_SET_VMA_ANON_NAME",
+            define_name="HAVE_DECL_PR_SET_VMA_ANON_NAME",
             on_found=_define_pr_set_vma_anon_name,
             extra_includes=["linux/prctl.h", "sys/prctl.h"],
         )

Tools/configure/conf_syslibs.py

@@ -50,11 +50,14 @@ def detect_dbm(v):
     v.have_gdbm_compat = False
 
     # Only search for dbm_open if ndbm.h is present (mirrors AC_CHECK_HEADERS guard)
+    # Wrap in save_env() to avoid leaking -lgdbm_compat into LIBS
+    # (matches autoconf's WITH_SAVE_ENV).
     ac_cv_search_dbm_open = False
     if pyconf.check_header("ndbm.h"):
-        ac_cv_search_dbm_open = pyconf.search_libs(
-            "dbm_open", ["ndbm", "gdbm_compat"], required=False
-        )
+        with pyconf.save_env():
+            ac_cv_search_dbm_open = pyconf.search_libs(
+                "dbm_open", ["ndbm", "gdbm_compat"], required=False
+            )
     if ac_cv_search_dbm_open and ac_cv_search_dbm_open is not False:
         if (
             "ndbm" in ac_cv_search_dbm_open
@@ -78,15 +81,10 @@ def detect_dbm(v):
         )
 
     ac_cv_header_gdbm_dash_ndbm_h = pyconf.check_header("gdbm-ndbm.h")
-    if ac_cv_header_gdbm_dash_ndbm_h:
-        pyconf.define(
-            "HAVE_GDBM_DASH_NDBM_H",
-            1,
-            "Define to 1 if you have the <gdbm-ndbm.h> header file.",
-        )
 
     if ac_cv_header_gdbm_slash_ndbm_h or ac_cv_header_gdbm_dash_ndbm_h:
-        r = pyconf.search_libs("dbm_open", ["gdbm_compat"], required=False)
+        with pyconf.save_env():
+            r = pyconf.search_libs("dbm_open", ["gdbm_compat"], required=False)
         if r and r is not False:
             v.have_gdbm_compat = True
         else:

Tools/configure/conf_wasm.py

@@ -113,10 +113,10 @@ def _setup_emscripten_flags(v):
     v.LINKFORSHARED += " -sSTACK_SIZE=5MB"
     v.LINKFORSHARED += " -sTEXTDECODER=2"
 
-    if v.enable_wasm_dynamic_linking:
+    if v.enable_wasm_dynamic_linking is True:
         v.LINKFORSHARED += " -sMAIN_MODULE"
 
-    if v.enable_wasm_pthreads:
+    if v.enable_wasm_pthreads is True:
         v.CFLAGS_NODIST += " -pthread"
         v.LDFLAGS_NODIST += " -sUSE_PTHREADS"
         v.LINKFORSHARED += " -sPROXY_TO_PTHREAD"
@@ -150,8 +150,7 @@ def _setup_wasi_flags(v):
     )
     v.LIBS += " -lwasi-emulated-signal -lwasi-emulated-getpid -lwasi-emulated-process-clocks"
 
-    if v.enable_wasm_pthreads:
-        v.CFLAGS += " -target wasm32-wasi-threads -pthread"
+    if v.enable_wasm_pthreads is True:
         v.CFLAGS_NODIST += " -target wasm32-wasi-threads -pthread"
         v.LDFLAGS_NODIST += (
             " -target wasm32-wasi-threads -pthread"

Tools/configure/pyconf.py

@@ -75,13 +75,19 @@ def export(
         """Mark a variable for Makefile substitution (AC_SUBST).
 
         If *value* is given, assigns it to this Vars instance first.
-        Otherwise looks up the value from this Vars instance.
+        Otherwise looks up the value from this Vars instance (including
+        environment variables via ``__getattr__``).
         The name is recorded in ``self._exports``.
         """
         self._exports.add(name)
         if value is not self._SENTINEL:
             setattr(self, name, value)
         val = self.__dict__.get(name)
+        if val is None:
+            # Fall back to environment variable (mirrors __getattr__),
+            # matching autoconf where AC_SUBST reads shell variables
+            # which include inherited environment.
+            val = os.environ.get(name, None)
         if val is not None:
             substs[name] = (
                 format_yn(val) if isinstance(val, bool) else str(val)
@@ -1000,12 +1006,10 @@ def _populate_module_substs() -> None:
         na = info.get("na", False)
         if na:
             state = "n/a"
-        elif not supported and enabled:
-            state = "missing"
-        elif not supported:
-            state = "n/a"
         elif not enabled:
             state = "disabled"
+        elif not supported:
+            state = "missing"
         else:
             state = "yes"
         # _TRUE="" means "build this module"; only "yes" state builds
@@ -2343,10 +2347,15 @@ def _try(boolean_expr: str) -> bool:
 
 
 def _header_name_to_define(header: str) -> str:
-    """Convert 'sys/types.h' → 'HAVE_SYS_TYPES_H'."""
-    return "HAVE_" + header.upper().replace("/", "_").replace(
-        ".", "_"
-    ).replace("-", "_")
+    """Convert 'sys/types.h' → 'HAVE_SYS_TYPES_H'.
+
+    Dashes are mapped to '_DASH_' to match autoconf's convention where
+    'gdbm-ndbm.h' produces HAVE_GDBM_DASH_NDBM_H (distinct from
+    'gdbm/ndbm.h' → HAVE_GDBM_NDBM_H).
+    """
+    return "HAVE_" + header.upper().replace("-", "_DASH_").replace(
+        "/", "_"
+    ).replace(".", "_")
 
 
 def _ac_includes_default() -> str:
@@ -2399,9 +2408,9 @@ def check_header(
             f"check_header: invalid header name {header!r} "
             f"(must match [a-z0-9][a-z0-9_./+-]*)"
         )
-    cache_key = "ac_cv_header_" + header.replace("/", "_").replace(
-        ".", "_"
-    ).replace("-", "_")
+    cache_key = "ac_cv_header_" + header.replace("-", "_dash_").replace(
+        "/", "_"
+    ).replace(".", "_")
     define_name = _header_name_to_define(header)
     # Print "checking for <header>..." unless the caller already started one.
     own_checking = not _result_pending
@@ -2606,8 +2615,10 @@ def check_sizeof(
         if output_str.strip().isdigit():
             size = int(output_str.strip())
     if size is None:
-        # Cross-compiling or run failed: compile-time binary search
-        inc = "#include <stddef.h>\n#include <stdint.h>\n" + extra_includes
+        # Cross-compiling or run failed: compile-time binary search.
+        # Use AC_INCLUDES_DEFAULT (which includes stdio.h etc.) so that
+        # types like fpos_t are visible, matching autoconf behaviour.
+        inc = _ac_includes_default() + extra_includes
         size = _compute_int(f"(long int)(sizeof({type_}))", inc)
     if size is None:
         size = default if default is not None else 0
@@ -3173,7 +3184,10 @@ def replace_funcs(funcs: list[str]) -> None:
     missing = substs.get("LIBOBJS", "").split()
     for func in funcs:
         if not check_func(func):
-            missing.append(f"{func}.o")
+            # Match autoconf's final LIBOBJS fixup: prepend ${LIBOBJDIR}
+            # and insert $U before the extension so Makefile.pre.in can
+            # locate the replacement source under Python/.
+            missing.append(f"${{LIBOBJDIR}}{func}$U.o")
     substs["LIBOBJS"] = " ".join(missing)
 
 

Tools/configure/test_pyconf.py

@@ -1485,6 +1485,31 @@ def test_loading_cache_message(self, capsys, monkeypatch, tmp_path):
         assert f"loading cache {cache_path}" in captured.out
 
 
+# ---------------------------------------------------------------------------
+# replace_funcs / LIBOBJS
+# ---------------------------------------------------------------------------
+
+
+@pytest.mark.skipif(not HAS_CC, reason="no C compiler available")
+class TestReplaceFuncs:
+    def test_missing_func_added_to_libobjs(self, with_cc):
+        pyconf.replace_funcs(["__no_such_func_xyz_123__"])
+        libobjs = pyconf.substs.get("LIBOBJS", "")
+        assert "${LIBOBJDIR}__no_such_func_xyz_123__$U.o" in libobjs
+
+    def test_existing_func_not_in_libobjs(self, with_cc):
+        pyconf.replace_funcs(["printf"])
+        assert pyconf.substs.get("LIBOBJS", "") == ""
+
+    def test_libobjs_format_matches_autoconf(self, with_cc):
+        """LIBOBJS entries must use ${LIBOBJDIR}name$U.o for Makefile.pre.in."""
+        pyconf.replace_funcs(["__no_such_a__", "__no_such_b__"])
+        entries = pyconf.substs["LIBOBJS"].split()
+        for entry in entries:
+            assert entry.startswith("${LIBOBJDIR}"), entry
+            assert "$U.o" in entry, entry
+
+
 # ---------------------------------------------------------------------------
 
 if __name__ == "__main__":

Tools/configure/transpiler/pyconf.awk

@@ -211,6 +211,10 @@ function pyconf_export(name, value) {
 }
 
 function v_export(name) {
+        # Fall back to ENVIRON if V[name] is not set, matching
+        # autoconf where AC_SUBST reads inherited environment variables.
+        if (!(name in V) && name in ENVIRON)
+                V[name] = ENVIRON[name]
         ENV[name] = V[name]
         MODIFIED_ENV[name] = 1
 }
@@ -562,7 +566,11 @@ function _pyconf_includes_from_list(hdr_list, n, arr, i, result) {
 
 function _pyconf_header_to_define(header, d) {
         d = "HAVE_" header
-        gsub(/[\/\.\+\-]/, "_", d)
+        # Map "-" to "_DASH_" before other replacements so that
+        # gdbm-ndbm.h -> HAVE_GDBM_DASH_NDBM_H (distinct from
+        # gdbm/ndbm.h -> HAVE_GDBM_NDBM_H), matching autoconf.
+        gsub(/-/, "_DASH_", d)
+        gsub(/[\/\.\+]/, "_", d)
         d = toupper(d)
         return d
 }
@@ -586,7 +594,8 @@ function _pyconf_ac_includes_default(    result) {
 function pyconf_check_header(header, prologue, default_inc, define, cache_key, source, cv, rc) {
         if (define == "") define = _pyconf_header_to_define(header)
         if (cache_key == "") cache_key = "ac_cv_header_" header
-        gsub(/[\/\.\+\-]/, "_", cache_key)
+        gsub(/-/, "_dash_", cache_key)
+        gsub(/[\/\.\+]/, "_", cache_key)
 
         if (cache_key in CACHE) {
                 rc = (CACHE[cache_key] == "yes")
@@ -688,9 +697,9 @@ function pyconf_replace_funcs(funcs, n, i) {
                 if (funcs[i] != "") {
                         if (!pyconf_check_func(funcs[i])) {
                                 if (SUBST["LIBOBJS"] != "")
-                                        SUBST["LIBOBJS"] = SUBST["LIBOBJS"] " " funcs[i] ".o"
+                                        SUBST["LIBOBJS"] = SUBST["LIBOBJS"] " ${LIBOBJDIR}" funcs[i] "$U.o"
                                 else
-                                        SUBST["LIBOBJS"] = funcs[i] ".o"
+                                        SUBST["LIBOBJS"] = "${LIBOBJDIR}" funcs[i] "$U.o"
                         }
                 }
         }
@@ -1488,16 +1497,12 @@ function pyconf_stdlib_module(name, supported, enabled, cflags, ldflags, has_cfl
         }
         if (prev_na)
                 state = "n/a"
-        else if (supported == "yes" && enabled == "yes")
-                state = "yes"
-        else if (supported != "yes" && enabled == "yes")
-                state = "missing"
-        else if (supported != "yes")
-                state = "n/a"
         else if (enabled != "yes")
                 state = "disabled"
+        else if (supported != "yes")
+                state = "missing"
         else
-                state = "n/a"
+                state = "yes"
         SUBST[key "_STATE"] = state
         SUBST[key "_TRUE"] = (state == "yes") ? "" : "#"
         if (has_cflags == "yes")