This revision was automatically updated to reflect the committed changes.
Closed by commit rL355388: Fix embedded Python initialization according to
changes in version 3.7 (authored by tkrasnukha, committed by ).
Herald added a project: LLVM.
Herald added a subscriber: llvm-commits.
Changed prior to commit:
https://reviews.llvm.org/D58833?vs=188933&id=189290#toc
Repository:
rL LLVM
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D58833/new/
https://reviews.llvm.org/D58833
Files:
lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
Index:
lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
===================================================================
---
lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
+++
lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
@@ -156,7 +156,7 @@
if (m_was_already_initialized) {
Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_SCRIPT));
LLDB_LOGV(log, "Releasing PyGILState. Returning to state = {0}locked",
- m_was_already_initialized == PyGILState_UNLOCKED ? "un" : "");
+ m_gil_state == PyGILState_UNLOCKED ? "un" : "");
PyGILState_Release(m_gil_state);
} else {
// We initialized the threads in this function, just unlock the GIL.
@@ -180,6 +180,18 @@
}
void InitializeThreadsPrivate() {
+// Since Python 3.7 `Py_Initialize` calls `PyEval_InitThreads` inside itself,
+// so there is no way to determine whether the embedded interpreter
+// was already initialized by some external code. `PyEval_ThreadsInitialized`
+// would always return `true` and `PyGILState_Ensure/Release` flow would be
+// executed instead of unlocking GIL with `PyEval_SaveThread`. When
+// an another thread calls `PyGILState_Ensure` it would get stuck in deadlock.
+#if (PY_MAJOR_VERSION == 3 && PY_MINOR_VERSION >= 7) || (PY_MAJOR_VERSION > 3)
+// The only case we should go further and acquire the GIL: it is unlocked.
+ if (PyGILState_Check())
+ return;
+#endif
+
if (PyEval_ThreadsInitialized()) {
Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_SCRIPT));
Index: lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
===================================================================
--- lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
+++ lldb/trunk/source/Plugins/ScriptInterpreter/Python/ScriptInterpreterPython.cpp
@@ -156,7 +156,7 @@
if (m_was_already_initialized) {
Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_SCRIPT));
LLDB_LOGV(log, "Releasing PyGILState. Returning to state = {0}locked",
- m_was_already_initialized == PyGILState_UNLOCKED ? "un" : "");
+ m_gil_state == PyGILState_UNLOCKED ? "un" : "");
PyGILState_Release(m_gil_state);
} else {
// We initialized the threads in this function, just unlock the GIL.
@@ -180,6 +180,18 @@
}
void InitializeThreadsPrivate() {
+// Since Python 3.7 `Py_Initialize` calls `PyEval_InitThreads` inside itself,
+// so there is no way to determine whether the embedded interpreter
+// was already initialized by some external code. `PyEval_ThreadsInitialized`
+// would always return `true` and `PyGILState_Ensure/Release` flow would be
+// executed instead of unlocking GIL with `PyEval_SaveThread`. When
+// an another thread calls `PyGILState_Ensure` it would get stuck in deadlock.
+#if (PY_MAJOR_VERSION == 3 && PY_MINOR_VERSION >= 7) || (PY_MAJOR_VERSION > 3)
+// The only case we should go further and acquire the GIL: it is unlocked.
+ if (PyGILState_Check())
+ return;
+#endif
+
if (PyEval_ThreadsInitialized()) {
Log *log(lldb_private::GetLogIfAllCategoriesSet(LIBLLDB_LOG_SCRIPT));
_______________________________________________
lldb-commits mailing list
lldb-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
