Fix config.status for awk version.

Rather than re-execing configure, make config.status regen the outputs.
Change the Python version to store the data inside the config.status
file, rather than using a separate .json file.

Author: nascheme

Tools/configure/pyconf.py

@@ -1209,33 +1209,17 @@ def _process_config_files() -> None:
 
 
 def _write_config_status() -> None:
-    """Write config.status script and its saved state.
+    """Write config.status with saved state embedded after ``exit 0``.
 
-    Generates:
-    - config.status.json: serialized substs and defines so config.status
-      can regenerate output files without re-running the full configure.
-    - config.status: a shell script that handles --recheck (re-runs configure)
-      and file regeneration (re-does @VAR@ substitution).
+    The shell preamble handles ``--recheck`` and delegates file
+    regeneration to ``pyconf.py --config-status``.  All SUBST/DEFINES
+    data is appended after the ``@CONFIG_STATUS_DATA@`` marker so the
+    shell never sees it but Python can read it back.
     """
-    import json as _json
-
-    # Serialize defines: convert tuples to lists for JSON
-    serialized_defines = {}
-    for name, entry in defines.items():
-        value, description, quoted = entry
-        serialized_defines[name] = [value, description, quoted]
-
-    state = {
-        "substs": dict(substs),
-        "defines": serialized_defines,
-        "srcdir": srcdir,
-    }
-    with open("config.status.json", "w") as fh:
-        _json.dump(state, fh, indent=1)
-
-    # Write config.status as a shell script
     config_args = substs.get("CONFIG_ARGS", "")
-    # The script delegates to pyconf._config_status_main() for file generation
+    sd = srcdir or "."
+    pyconf_path = os.path.join(sd, "Tools/configure/pyconf.py")
+
     script = f"""\
 #!/bin/sh
 # Generated by configure.  Do not edit.
@@ -1255,14 +1239,31 @@ def _write_config_status() -> None:
   esac
 done
 
-# Regenerate files.  If CONFIG_FILES is set in the environment, only
-# those files are processed.  If positional arguments are given, they
-# are treated as the list of files.  Otherwise all config files are
-# regenerated.
-exec python3 {os.path.join(srcdir or ".", "Tools/configure/pyconf.py")} --config-status "$@"
+# Regenerate files.  pyconf.py reads saved data from the end of $0.
+exec python3 {pyconf_path} --config-status "$0" "$@"
+
+exit 0
 """
+
     with open("config.status", "w") as fh:
         fh.write(script)
+
+        # Append data section after exit 0.
+        fh.write("@CONFIG_STATUS_DATA@\n")
+        fh.write("@SUBST@\n")
+        for key, val in substs.items():
+            lines = val.split("\n") if val else [""]
+            fh.write(f"{key}\n{len(lines)}\n")
+            for line in lines:
+                fh.write(f"{line}\n")
+        fh.write("@DEFINES@\n")
+        for name, entry in defines.items():
+            value, description, quoted = entry
+            # Use repr() so ast.literal_eval() restores the exact type
+            fh.write(f"{name}\n{repr(value)}\n{description}\n")
+            fh.write(f"{repr(quoted)}\n")
+        fh.write("@END@\n")
+
     import stat
 
     current = os.stat("config.status").st_mode
@@ -1271,38 +1272,75 @@ def _write_config_status() -> None:
     )
 
 
+def _load_config_status_data(path: str) -> None:
+    """Read saved SUBST/DEFINES data from the end of *path*.
+
+    The file is expected to contain a ``@CONFIG_STATUS_DATA@`` marker
+    followed by ``@SUBST@``, key/nlines/value blocks, ``@DEFINES@``,
+    key/value/desc/quoted blocks, and ``@END@``.
+    """
+    import ast
+
+    global substs, defines, srcdir
+
+    with open(path) as fh:
+        # Skip to the data marker
+        for line in fh:
+            if line.rstrip("\n") == "@CONFIG_STATUS_DATA@":
+                break
+        else:
+            raise RuntimeError(f"no @CONFIG_STATUS_DATA@ marker in {path}")
+
+        # Expect @SUBST@
+        tag = fh.readline().rstrip("\n")
+        assert tag == "@SUBST@", f"expected @SUBST@, got {tag!r}"
+
+        substs = {}
+        while True:
+            key = fh.readline().rstrip("\n")
+            if key == "@DEFINES@":
+                break
+            nlines = int(fh.readline().rstrip("\n"))
+            val_lines = [fh.readline().rstrip("\n") for _ in range(nlines)]
+            substs[key] = "\n".join(val_lines)
+
+        defines = {}
+        while True:
+            name = fh.readline().rstrip("\n")
+            if name == "@END@":
+                break
+            value = ast.literal_eval(fh.readline().rstrip("\n"))
+            description = fh.readline().rstrip("\n")
+            quoted = ast.literal_eval(fh.readline().rstrip("\n"))
+            defines[name] = (value, description, quoted)
+
+    srcdir = substs.get("srcdir", ".")
+
+
 def _config_status_main(argv: list[str]) -> None:
     """Entry point when config.status invokes pyconf for file regeneration.
 
-    Loads saved state from config.status.json and regenerates the requested
-    files (or all config files if none are specified).
+    Loads saved state from config.status (embedded after the
+    ``@CONFIG_STATUS_DATA@`` marker) and regenerates the requested files.
 
     Supports:
     - ``./config.status``  — regenerate all config files and pyconfig.h
     - ``./config.status Makefile.pre``  — regenerate only Makefile.pre
     - ``CONFIG_FILES=Makefile.pre CONFIG_HEADERS= ./config.status``
     """
-    import json as _json
-
     global substs, defines, srcdir
 
-    with open("config.status.json") as fh:
-        state = _json.load(fh)
+    # First positional arg is the path to config.status itself
+    cs_path = argv[0] if argv else "config.status"
+    rest = argv[1:]
 
-    substs = state["substs"]
-    # Restore defines: convert lists back to tuples
-    defines = {}
-    for name, entry in state["defines"].items():
-        defines[name] = tuple(entry)
-    srcdir = state["srcdir"]
+    _load_config_status_data(cs_path)
 
     # Determine which files to process
+    file_args = [a for a in rest if not a.startswith("-")]
     env_files = os.environ.get("CONFIG_FILES")
     env_headers = os.environ.get("CONFIG_HEADERS")
 
-    # Positional arguments (non-flag) override the file list
-    file_args = [a for a in argv if not a.startswith("-")]
-
     if file_args:
         # Regenerate only the specified files
         requested = set(file_args)

Tools/configure/transpiler/pyconf.awk

@@ -1857,36 +1857,118 @@ function _pyconf_process_config_files(    i, inf, outf, outdir, saved_abs_srcdir
         }
 }
 
-function _pyconf_write_config_status(    outf, config_args, sd) {
+function _pyconf_write_config_status(    outf, config_args, sd, k, val, i, cf, cfx, sep) {
         outf = "config.status"
         config_args = SUBST["CONFIG_ARGS"]
         if (config_args == "")
                 config_args = _pyconf_config_args
         sd = SUBST["srcdir"]
         if (sd == "")
                 sd = "."
+
+        # Save config file lists into SUBST so they get serialized.
+        cf = ""
+        sep = ""
+        for (i = 1; i <= _pyconf_config_file_count; i++) {
+                cf = cf sep CONFIG_FILES[i]
+                sep = " "
+        }
+        SUBST["_config_files"] = cf
+        cfx = ""
+        sep = ""
+        for (i = 1; i <= _pyconf_config_file_x_count; i++) {
+                cfx = cfx sep CONFIG_FILES_X[i]
+                sep = " "
+        }
+        SUBST["_config_files_x"] = cfx
+
+        # Write config.status as a thin shell wrapper.
+        # --recheck re-runs configure; everything else delegates to
+        # regen_status.awk which reads saved data from after the
+        # @CONFIG_STATUS_DATA@ marker at the end of this file.
         printf "#!/bin/sh\n" > outf
         printf "# Generated by configure.  Do not edit.\n" >> outf
         printf "#\n" >> outf
-        printf "# This script re-runs configure to regenerate output files.\n" >> outf
+        printf "# This script can regenerate output files from templates\n" >> outf
+        printf "# without re-running the full configure.\n" >> outf
         printf "\n" >> outf
-        printf "CONFIG_ARGS=\"%s\"\n", config_args >> outf
+        printf "CONFIG_ARGS='%s'\n", config_args >> outf
+        printf "SRCDIR='%s'\n", sd >> outf
         printf "\n" >> outf
         printf "for arg in \"$@\"; do\n" >> outf
         printf "  case \"$arg\" in\n" >> outf
         printf "    --recheck)\n" >> outf
         printf "      echo \"running configure with args: $CONFIG_ARGS\"\n" >> outf
-        printf "      exec %s/configure $CONFIG_ARGS\n", sd >> outf
+        printf "      exec \"$SRCDIR\"/configure $CONFIG_ARGS\n" >> outf
         printf "      ;;\n" >> outf
         printf "  esac\n" >> outf
         printf "done\n" >> outf
         printf "\n" >> outf
-        printf "# Regenerate all output files by re-running configure.\n" >> outf
-        printf "exec %s/configure $CONFIG_ARGS\n", sd >> outf
+        # Emit _find_awk: prefer mawk/nawk, avoid gawk crash bug.
+        printf "_find_awk() {\n" >> outf
+        printf "    for _c in mawk nawk awk; do\n" >> outf
+        printf "        IFS=:\n" >> outf
+        printf "        for _d in $PATH; do\n" >> outf
+        printf "            if [ -x \"${_d}/${_c}\" ]; then\n" >> outf
+        printf "                case $(\"${_d}/${_c}\" --version < /dev/null 2>&1) in\n" >> outf
+        printf "                    *\"GNU Awk\"*) ;;\n" >> outf
+        printf "                    *) echo \"${_d}/${_c}\"; return 0 ;;\n" >> outf
+        printf "                esac\n" >> outf
+        printf "            fi\n" >> outf
+        printf "        done\n" >> outf
+        printf "        unset IFS\n" >> outf
+        printf "    done\n" >> outf
+        printf "    echo awk\n" >> outf
+        printf "}\n" >> outf
+        printf "_awk=$(_find_awk)\n" >> outf
+        printf "\n" >> outf
+        printf "exec \"$_awk\" -f \"$SRCDIR\"/Tools/configure/transpiler/regen_status.awk \\\n" >> outf
+        printf "    -v cs_args=\"$*\" \\\n" >> outf
+        printf "    -v cs_config_files=\"${CONFIG_FILES-@UNSET@}\" \\\n" >> outf
+        printf "    -v cs_config_headers=\"${CONFIG_HEADERS-@UNSET@}\" \\\n" >> outf
+        printf "    -v srcdir=\"$SRCDIR\" \\\n" >> outf
+        printf "    \"$0\"\n" >> outf
+        printf "\n" >> outf
+        printf "exit 0\n" >> outf
+
+        # Data section: the shell never reaches here (exit above).
+        # regen_status.awk scans for @CONFIG_STATUS_DATA@ to find this data.
+        printf "@CONFIG_STATUS_DATA@\n" >> outf
+        printf "@SUBST@\n" >> outf
+        for (k in SUBST) {
+                val = SUBST[k]
+                _cs_write_data_entry(outf, k, val)
+        }
+        printf "@DEFINES@\n" >> outf
+        for (i = 1; i <= _pyconf_defines_order_n; i++) {
+                k = _pyconf_defines_order[i]
+                if (k in DEFINES) {
+                        printf "%s\n", k >> outf
+                        printf "%s\n", DEFINES[k] >> outf
+                        printf "%s\n", (k in DEFINE_DESC ? DEFINE_DESC[k] : "") >> outf
+                        printf "%s\n", (DEFINE_QUOTED[k] ? "1" : "0") >> outf
+                }
+        }
+        printf "@END@\n" >> outf
+
         close(outf)
         system("chmod +x " _shell_quote(outf))
 }
 
+# Write a SUBST entry: key, line_count, value lines
+function _cs_write_data_entry(outf, key, val,    n, lines, i) {
+        n = split(val, lines, "\n")
+        if (n == 0) n = 1
+        printf "%s\n%d\n", key, n >> outf
+        if (val == "") {
+                printf "\n" >> outf
+        } else {
+                for (i = 1; i <= n; i++)
+                        printf "%s\n", lines[i] >> outf
+        }
+}
+
+
 function _last_index(s, ch, i, last) {
         last = 0
         for (i = 1; i <= length(s); i++)