fix: page in .py sources during warmup for snapshot restore

Signed-off-by: danbugs <danilochiarlone@gmail.com>

Author: danbugs

examples/python-agent-driver/hl_pydriver.c

@@ -259,6 +259,34 @@ static void py_initialize_once(void)
 		"  except Exception as _e:\n"
 		"    sys.stderr.write(f'warn: preload {_mod} failed: {_e}\\n')\n");
 
+	/* Exercise pyarrow's pandas shim so it imports pandas_compat,
+	 * vendored.version, etc. during warmup. Without this, the shim
+	 * lazily imports these on first use, but after snapshot restore
+	 * (without initrd re-mapping) the .py files are zeroed pages. */
+	PyRun_SimpleString(
+		"try:\n"
+		"    import pandas as _pd\n"
+		"    _pd.DataFrame({'_': ['x']})\n"
+		"    del _pd\n"
+		"except Exception:\n"
+		"    pass\n");
+
+	/* Page in all .py sources so they survive snapshot restore.
+	 * The snapshot captures only pages faulted during warmup; any
+	 * .py file not yet read would appear as null bytes after restore
+	 * (since we skip initrd re-mapping to avoid overlap). */
+	PyRun_SimpleString(
+		"import os as _os\n"
+		"for _r, _d, _ff in _os.walk('/usr/local/lib/python3.12'):\n"
+		"    for _fn in _ff:\n"
+		"        if _fn.endswith('.py'):\n"
+		"            try:\n"
+		"                with open(_os.path.join(_r, _fn), 'rb') as _fh:\n"
+		"                    _fh.read()\n"
+		"            except Exception:\n"
+		"                pass\n"
+		"del _os\n");
+
 	/* Monkey-patch zipfile.ZipInfo.__init__ to clamp pre-1980 timestamps.
 	 * Unikraft's ramfs reports epoch-0 (1970) for file mtimes; Python's
 	 * zipfile rejects timestamps before 1980, breaking openpyxl's XLSX