[3.13] gh-122581: Avoid data races when collecting parser statistics (GH-122694) (#122733)

miss-islington · lysnikolaou · web-flow · commit 9fb6bfbef027 · 2024-08-07T18:29:03.000+05:30
gh-122581: Avoid data races when collecting parser statistics (GH-122694) (cherry picked from commit ce0d66c) Co-authored-by: Lysandros Nikolaou <lisandrosnik@gmail.com>
diff --git a/Include/internal/pycore_parser.h b/Include/internal/pycore_parser.h
@@ -21,15 +21,20 @@ extern "C" {
 struct _parser_runtime_state {
 #ifdef Py_DEBUG
     long memo_statistics[_PYPEGEN_NSTATISTICS];
+#ifdef Py_GIL_DISABLED
+    PyMutex mutex;
+#endif
 #else
     int _not_used;
 #endif
     struct _expr dummy_name;
 };
 
 _Py_DECLARE_STR(empty, "")
+#if defined(Py_DEBUG) && defined(Py_GIL_DISABLED)
 #define _parser_runtime_state_INIT \
     { \
+        .mutex = {0}, \
         .dummy_name = { \
             .kind = Name_kind, \
             .v.Name.id = &_Py_STR(empty), \
@@ -40,6 +45,20 @@ _Py_DECLARE_STR(empty, "")
             .end_col_offset = 0, \
         }, \
     }
+#else
+#define _parser_runtime_state_INIT \
+    { \
+        .dummy_name = { \
+            .kind = Name_kind, \
+            .v.Name.id = &_Py_STR(empty), \
+            .v.Name.ctx = Load, \
+            .lineno = 1, \
+            .col_offset = 0, \
+            .end_lineno = 1, \
+            .end_col_offset = 0, \
+        }, \
+    }
+#endif
 
 extern struct _mod* _PyParser_ASTFromString(
     const char *str,
diff --git a/Parser/pegen.c b/Parser/pegen.c
@@ -296,12 +296,22 @@ _PyPegen_fill_token(Parser *p)
 #define NSTATISTICS _PYPEGEN_NSTATISTICS
 #define memo_statistics _PyRuntime.parser.memo_statistics
 
+#ifdef Py_GIL_DISABLED
+#define MUTEX_LOCK() PyMutex_Lock(&_PyRuntime.parser.mutex)
+#define MUTEX_UNLOCK() PyMutex_Unlock(&_PyRuntime.parser.mutex)
+#else
+#define MUTEX_LOCK()
+#define MUTEX_UNLOCK()
+#endif
+
 void
 _PyPegen_clear_memo_statistics(void)
 {
+    MUTEX_LOCK();
     for (int i = 0; i < NSTATISTICS; i++) {
         memo_statistics[i] = 0;
     }
+    MUTEX_UNLOCK();
 }
 
 PyObject *
@@ -311,18 +321,23 @@ _PyPegen_get_memo_statistics(void)
     if (ret == NULL) {
         return NULL;
     }
+
+    MUTEX_LOCK();
     for (int i = 0; i < NSTATISTICS; i++) {
         PyObject *value = PyLong_FromLong(memo_statistics[i]);
         if (value == NULL) {
+            MUTEX_UNLOCK();
             Py_DECREF(ret);
             return NULL;
         }
         // PyList_SetItem borrows a reference to value.
         if (PyList_SetItem(ret, i, value) < 0) {
+            MUTEX_UNLOCK();
             Py_DECREF(ret);
             return NULL;
         }
     }
+    MUTEX_UNLOCK();
     return ret;
 }
 #endif
@@ -348,7 +363,9 @@ _PyPegen_is_memoized(Parser *p, int type, void *pres)
                 if (count <= 0) {
                     count = 1;
                 }
+                MUTEX_LOCK();
                 memo_statistics[type] += count;
+                MUTEX_UNLOCK();
             }
 #endif
             p->mark = m->mark;

Original file line number	Diff line number	Diff line change
`@@ -296,12 +296,22 @@ _PyPegen_fill_token(Parser *p)`
`296`	`296`	`#define NSTATISTICS _PYPEGEN_NSTATISTICS`
`297`	`297`	`#define memo_statistics _PyRuntime.parser.memo_statistics`
`298`	`298`
	`299`	`+#ifdef Py_GIL_DISABLED`
	`300`	`+#define MUTEX_LOCK() PyMutex_Lock(&_PyRuntime.parser.mutex)`
	`301`	`+#define MUTEX_UNLOCK() PyMutex_Unlock(&_PyRuntime.parser.mutex)`
	`302`	`+#else`
	`303`	`+#define MUTEX_LOCK()`
	`304`	`+#define MUTEX_UNLOCK()`
	`305`	`+#endif`
	`306`	`+`
`299`	`307`	`void`
`300`	`308`	`_PyPegen_clear_memo_statistics(void)`
`301`	`309`	`{`
	`310`	`+ MUTEX_LOCK();`
`302`	`311`	`for (int i = 0; i < NSTATISTICS; i++) {`
`303`	`312`	`memo_statistics[i] = 0;`
`304`	`313`	`}`
	`314`	`+ MUTEX_UNLOCK();`
`305`	`315`	`}`
`306`	`316`
`307`	`317`	`PyObject *`
`@@ -311,18 +321,23 @@ _PyPegen_get_memo_statistics(void)`
`311`	`321`	`if (ret == NULL) {`
`312`	`322`	`return NULL;`
`313`	`323`	`}`
	`324`	`+`
	`325`	`+ MUTEX_LOCK();`
`314`	`326`	`for (int i = 0; i < NSTATISTICS; i++) {`
`315`	`327`	`PyObject *value = PyLong_FromLong(memo_statistics[i]);`
`316`	`328`	`if (value == NULL) {`
	`329`	`+ MUTEX_UNLOCK();`
`317`	`330`	`Py_DECREF(ret);`
`318`	`331`	`return NULL;`
`319`	`332`	`}`
`320`	`333`	`// PyList_SetItem borrows a reference to value.`
`321`	`334`	`if (PyList_SetItem(ret, i, value) < 0) {`
	`335`	`+ MUTEX_UNLOCK();`
`322`	`336`	`Py_DECREF(ret);`
`323`	`337`	`return NULL;`
`324`	`338`	`}`
`325`	`339`	`}`
	`340`	`+ MUTEX_UNLOCK();`
`326`	`341`	`return ret;`
`327`	`342`	`}`
`328`	`343`	`#endif`
`@@ -348,7 +363,9 @@ _PyPegen_is_memoized(Parser p, int type, void pres)`
`348`	`363`	`if (count <= 0) {`
`349`	`364`	`count = 1;`
`350`	`365`	`}`
	`366`	`+ MUTEX_LOCK();`
`351`	`367`	`memo_statistics[type] += count;`
	`368`	`+ MUTEX_UNLOCK();`
`352`	`369`	`}`
`353`	`370`	`#endif`
`354`	`371`	`p->mark = m->mark;`