gh-144777: Fix data races in IncrementalNewlineDecoder (gh-144971)

diff --git a/Lib/test/test_free_threading/test_io.py b/Lib/test/test_free_threading/test_io.py
index c67aaff31b3f5b627c8bf32cbed5fcb0798e2414..8a0ad30c4bc770b790edebfb9b01b76d4597e15e 100644 (file)
--- a/Lib/test/test_free_threading/test_io.py
+++ b/Lib/test/test_free_threading/test_io.py
@@ -1,11 +1,15 @@
+import codecs
 import io
 import _pyio as pyio
 import threading
 from unittest import TestCase
 from test.support import threading_helper
+from test.support.threading_helper import run_concurrently
 from random import randint
 from sys import getsizeof
 
+threading_helper.requires_working_threading(module=True)
+
 
 class ThreadSafetyMixin:
     # Test pretty much everything that can break under free-threading.
@@ -115,3 +119,54 @@ class CBytesIOTest(ThreadSafetyMixin, TestCase):
 
 class PyBytesIOTest(ThreadSafetyMixin, TestCase):
      ioclass = pyio.BytesIO
+
+
+class IncrementalNewlineDecoderTest(TestCase):
+    def make_decoder(self):
+        utf8_decoder = codecs.getincrementaldecoder('utf-8')()
+        return io.IncrementalNewlineDecoder(utf8_decoder, translate=True)
+
+    def test_concurrent_reset(self):
+        decoder = self.make_decoder()
+
+        def worker():
+            for _ in range(100):
+                decoder.reset()
+
+        run_concurrently(worker_func=worker, nthreads=2)
+
+    def test_concurrent_decode(self):
+        decoder = self.make_decoder()
+
+        def worker():
+            for _ in range(100):
+                decoder.decode(b"line\r\n", final=False)
+
+        run_concurrently(worker_func=worker, nthreads=2)
+
+    def test_concurrent_getstate_setstate(self):
+        decoder = self.make_decoder()
+        state = decoder.getstate()
+
+        def getstate_worker():
+            for _ in range(100):
+                decoder.getstate()
+
+        def setstate_worker():
+            for _ in range(100):
+                decoder.setstate(state)
+
+        run_concurrently([getstate_worker] * 2 + [setstate_worker] * 2)
+
+    def test_concurrent_decode_and_reset(self):
+        decoder = self.make_decoder()
+
+        def decode_worker():
+            for _ in range(100):
+                decoder.decode(b"line\r\n", final=False)
+
+        def reset_worker():
+            for _ in range(100):
+                decoder.reset()
+
+        run_concurrently([decode_worker] * 2 + [reset_worker] * 2)

diff --git a/Misc/NEWS.d/next/Library/2026-02-18-13-45-00.gh-issue-144777.R97q0a.rst b/Misc/NEWS.d/next/Library/2026-02-18-13-45-00.gh-issue-144777.R97q0a.rst
new file mode 100644 (file)
index 0000000..fd720bf
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-02-18-13-45-00.gh-issue-144777.R97q0a.rst
@@ -0,0 +1 @@
+Fix data races in :class:`io.IncrementalNewlineDecoder` in the :term:`free-threaded build`.

diff --git a/Modules/_io/clinic/textio.c.h b/Modules/_io/clinic/textio.c.h
index 128a5ad1678f26a3256c23eaef8c2cf9c3a3f3c9..3898a9c29824364223bf6daf15fa50abf66ec493 100644 (file)
--- a/Modules/_io/clinic/textio.c.h
+++ b/Modules/_io/clinic/textio.c.h
@@ -430,7 +430,9 @@ _io_IncrementalNewlineDecoder_decode(PyObject *self, PyObject *const *args, Py_s
         goto exit;
     }
 skip_optional_pos:
+    Py_BEGIN_CRITICAL_SECTION(self);
     return_value = _io_IncrementalNewlineDecoder_decode_impl((nldecoder_object *)self, input, final);
+    Py_END_CRITICAL_SECTION();
 
 exit:
     return return_value;
@@ -450,7 +452,13 @@ _io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self);
 static PyObject *
 _io_IncrementalNewlineDecoder_getstate(PyObject *self, PyObject *Py_UNUSED(ignored))
 {
-    return _io_IncrementalNewlineDecoder_getstate_impl((nldecoder_object *)self);
+    PyObject *return_value = NULL;
+
+    Py_BEGIN_CRITICAL_SECTION(self);
+    return_value = _io_IncrementalNewlineDecoder_getstate_impl((nldecoder_object *)self);
+    Py_END_CRITICAL_SECTION();
+
+    return return_value;
 }
 
 PyDoc_STRVAR(_io_IncrementalNewlineDecoder_setstate__doc__,
@@ -470,7 +478,9 @@ _io_IncrementalNewlineDecoder_setstate(PyObject *self, PyObject *state)
 {
     PyObject *return_value = NULL;
 
+    Py_BEGIN_CRITICAL_SECTION(self);
     return_value = _io_IncrementalNewlineDecoder_setstate_impl((nldecoder_object *)self, state);
+    Py_END_CRITICAL_SECTION();
 
     return return_value;
 }
@@ -489,7 +499,13 @@ _io_IncrementalNewlineDecoder_reset_impl(nldecoder_object *self);
 static PyObject *
 _io_IncrementalNewlineDecoder_reset(PyObject *self, PyObject *Py_UNUSED(ignored))
 {
-    return _io_IncrementalNewlineDecoder_reset_impl((nldecoder_object *)self);
+    PyObject *return_value = NULL;
+
+    Py_BEGIN_CRITICAL_SECTION(self);
+    return_value = _io_IncrementalNewlineDecoder_reset_impl((nldecoder_object *)self);
+    Py_END_CRITICAL_SECTION();
+
+    return return_value;
 }
 
 PyDoc_STRVAR(_io_TextIOWrapper___init____doc__,
@@ -1312,4 +1328,4 @@ _io_TextIOWrapper__CHUNK_SIZE_set(PyObject *self, PyObject *value, void *Py_UNUS
 
     return return_value;
 }
-/*[clinic end generated code: output=30404271a1151056 input=a9049054013a1b77]*/
+/*[clinic end generated code: output=c38e6cd5ff4b7eea input=a9049054013a1b77]*/

diff --git a/Modules/_io/textio.c b/Modules/_io/textio.c
index f9881952561292e2d829a3752485c8703c19d69d..347bfe976619e89aef4a40a6cf13dbe1ba22fa74 100644 (file)
--- a/Modules/_io/textio.c
+++ b/Modules/_io/textio.c
@@ -519,6 +519,7 @@ _PyIncrementalNewlineDecoder_decode(PyObject *myself,
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.decode
     input: object
     final: bool = False
@@ -527,18 +528,19 @@ _io.IncrementalNewlineDecoder.decode
 static PyObject *
 _io_IncrementalNewlineDecoder_decode_impl(nldecoder_object *self,
                                           PyObject *input, int final)
-/*[clinic end generated code: output=0d486755bb37a66e input=90e223c70322c5cd]*/
+/*[clinic end generated code: output=0d486755bb37a66e input=9475d16a73168504]*/
 {
     return _PyIncrementalNewlineDecoder_decode((PyObject *) self, input, final);
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.getstate
 [clinic start generated code]*/
 
 static PyObject *
 _io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self)
-/*[clinic end generated code: output=f0d2c9c136f4e0d0 input=f8ff101825e32e7f]*/
+/*[clinic end generated code: output=f0d2c9c136f4e0d0 input=dc3e1f27aa850f12]*/
 {
     PyObject *buffer;
     unsigned long long flag;
@@ -576,6 +578,7 @@ _io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self)
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.setstate
     state: object
     /
@@ -584,7 +587,7 @@ _io.IncrementalNewlineDecoder.setstate
 static PyObject *
 _io_IncrementalNewlineDecoder_setstate_impl(nldecoder_object *self,
                                             PyObject *state)
-/*[clinic end generated code: output=09135cb6e78a1dc8 input=c53fb505a76dbbe2]*/
+/*[clinic end generated code: output=09135cb6e78a1dc8 input=275fd3982d2b08cb]*/
 {
     PyObject *buffer;
     unsigned long long flag;
@@ -614,12 +617,13 @@ _io_IncrementalNewlineDecoder_setstate_impl(nldecoder_object *self,
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.reset
 [clinic start generated code]*/
 
 static PyObject *
 _io_IncrementalNewlineDecoder_reset_impl(nldecoder_object *self)
-/*[clinic end generated code: output=32fa40c7462aa8ff input=728678ddaea776df]*/
+/*[clinic end generated code: output=32fa40c7462aa8ff input=31bd8ae4e36cec83]*/
 {
     CHECK_INITIALIZED_DECODER(self);