[3.14] gh-144777: Fix data races in IncrementalNewlineDecoder (gh-144971) (#145143)

Author: colesbury

Lib/test/test_free_threading/test_io.py

@@ -1,10 +1,15 @@
+import codecs
+import io
 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 io import BytesIO
 from sys import getsizeof
 
+threading_helper.requires_working_threading(module=True)
+
 
 class TestBytesIO(TestCase):
     # Test pretty much everything that can break under free-threading.
@@ -107,3 +112,54 @@ def sizeof(barrier, b, *ignore):
         self.check([truncate] + [sizeof] * 10, BytesIO(b'0\n'*204800))
 
         # no tests for seek or tell because they don't break anything
+
+
+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)

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`.

Modules/_io/clinic/textio.c.h

@@ -518,6 +518,7 @@ _PyIncrementalNewlineDecoder_decode(PyObject *myself,
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.decode
     input: object
     final: bool = False
@@ -526,18 +527,19 @@ _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;
@@ -575,6 +577,7 @@ _io_IncrementalNewlineDecoder_getstate_impl(nldecoder_object *self)
 }
 
 /*[clinic input]
+@critical_section
 _io.IncrementalNewlineDecoder.setstate
     state: object
     /
@@ -583,7 +586,7 @@ _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;
@@ -613,12 +616,13 @@ _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);