public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Andrew Burgess <aburgess@redhat.com>
To: gdb-patches@sourceware.org
Cc: Andrew Burgess <aburgess@redhat.com>
Subject: [PATCHv2 2/3] gdb/python: deallocate tui window factories at Python shut down
Date: Wed, 25 Jan 2023 13:56:10 +0000	[thread overview]
Message-ID: <80daa36b6b15e84c88df642e50e01b494cce0a04.1674654912.git.aburgess@redhat.com> (raw)
In-Reply-To: <cover.1674654912.git.aburgess@redhat.com>

The previous commit relied on spotting when a Python defined TUI
window factory was deleted.  I spotted that the window factories are
not deleted when GDB shuts down its Python environment, they are only
deleted when one window factory replaces another.  Consider this
example Python script:

  class TestWindowFactory:
      def __init__(self, msg):
          self.msg = msg
          print("Entering TestWindowFactory.__init__: %s" % self.msg)

      def __call__(self, tui_win):
          print("Entering TestWindowFactory.__call__: %s" % self.msg)
          return TestWindow(tui_win, self.msg)

      def __del__(self):
          print("Entering TestWindowFactory.__del__: %s" % self.msg)

  gdb.register_window_type("test_window", TestWindowFactory("A"))
  gdb.register_window_type("test_window", TestWindowFactory("B"))

And this GDB session:

  (gdb) source tui.py
  Entering TestWindowFactory.__init__: A
  Entering TestWindowFactory.__init__: B
  Entering TestWindowFactory.__del__: B
  (gdb) quit

Notice that when the 'B' window replaces the 'A' window we see the 'A'
object being deleted.  But, when Python is shut down (after the
'quit') the 'B' object is never deleted.

Instead, GDB retains a reference to the window factory object, which
forces the Python object to remain live even after the Python
interpreter itself has been shut down.

The references themselves are held in a dynamically allocated
std::unordered_map (in tui/tui-layout.c) which is never deallocated,
thus the underlying Python references are never decremented to zero,
and so GDB never tries to delete these Python objects.

This commit is the first half of the work to clean up this edge case.

All gdbpy_tui_window_maker objects (the objects that implement the
TUI window factory callback for Python defined TUI windows), are now
linked together into a global list using the intrusive list mechanism.

When GDB shuts down the Python interpreter we can now walk this global
list and release the reference that is held to the underlying Python
object.  By releasing this reference the Python object will now be
deleted.

I've added a new assert in gdbpy_tui_window_maker::operator(), this
will catch the case where we somehow end up in here after having
reset the reference to the underlying Python object.  I don't think
this should ever happen though as we only clear the references when
shutting down the Python interpreter, and the ::operator() function is
only called when trying to apply a new TUI layout - something that
shouldn't happen while GDB itself is shutting down.

This commit does not update the std::unordered_map in tui-layout.c,
that will be done in the next commit.
---
 gdb/python/py-tui.c                           | 52 ++++++++++++++++++-
 gdb/python/python-internal.h                  |  1 +
 gdb/python/python.c                           |  1 +
 .../gdb.python/tui-window-factory.exp         | 32 ++++++++++++
 4 files changed, 84 insertions(+), 2 deletions(-)

diff --git a/gdb/python/py-tui.c b/gdb/python/py-tui.c
index e9c91012ae9..9ce76659052 100644
--- a/gdb/python/py-tui.c
+++ b/gdb/python/py-tui.c
@@ -21,6 +21,7 @@
 #include "defs.h"
 #include "arch-utils.h"
 #include "python-internal.h"
+#include "gdbsupport/intrusive_list.h"
 
 #ifdef TUI
 
@@ -268,12 +269,14 @@ tui_py_window::output (const char *text, bool full_window)
    user-supplied window constructor.  */
 
 class gdbpy_tui_window_maker
+  : public intrusive_list_node<gdbpy_tui_window_maker>
 {
 public:
 
   explicit gdbpy_tui_window_maker (gdbpy_ref<> &&constr)
     : m_constr (std::move (constr))
   {
+    m_window_maker_list.push_back (*this);
   }
 
   ~gdbpy_tui_window_maker ();
@@ -281,12 +284,14 @@ class gdbpy_tui_window_maker
   gdbpy_tui_window_maker (gdbpy_tui_window_maker &&other) noexcept
     : m_constr (std::move (other.m_constr))
   {
+    m_window_maker_list.push_back (*this);
   }
 
   gdbpy_tui_window_maker (const gdbpy_tui_window_maker &other)
   {
     gdbpy_enter enter_py;
     m_constr = other.m_constr;
+    m_window_maker_list.push_back (*this);
   }
 
   gdbpy_tui_window_maker &operator= (gdbpy_tui_window_maker &&other)
@@ -304,16 +309,43 @@ class gdbpy_tui_window_maker
 
   tui_win_info *operator() (const char *name);
 
+  /* Reset the m_constr field of all gdbpy_tui_window_maker objects back to
+     nullptr, this will allow the Python object referenced to be
+     deallocated.  This function is intended to be called when GDB is
+     shutting down the Python interpreter to allow all Python objects to be
+     deallocated and cleaned up.  */
+  static void
+  invalidate_all ()
+  {
+    gdbpy_enter enter_py;
+    for (gdbpy_tui_window_maker &f : m_window_maker_list)
+      f.m_constr.reset (nullptr);
+  }
+
 private:
 
   /* A constructor that is called to make a TUI window.  */
   gdbpy_ref<> m_constr;
+
+  /* A global list of all gdbpy_tui_window_maker objects.  */
+  static intrusive_list<gdbpy_tui_window_maker> m_window_maker_list;
 };
 
+/* See comment in class declaration above.  */
+
+intrusive_list<gdbpy_tui_window_maker>
+  gdbpy_tui_window_maker::m_window_maker_list;
+
 gdbpy_tui_window_maker::~gdbpy_tui_window_maker ()
 {
-  gdbpy_enter enter_py;
-  m_constr.reset (nullptr);
+  /* Remove this gdbpy_tui_window_maker from the global list.  */
+  m_window_maker_list.erase (m_window_maker_list.iterator_to (*this));
+
+  if (m_constr != nullptr)
+    {
+      gdbpy_enter enter_py;
+      m_constr.reset (nullptr);
+    }
 }
 
 tui_win_info *
@@ -332,6 +364,14 @@ gdbpy_tui_window_maker::operator() (const char *win_name)
   std::unique_ptr<tui_py_window> window
     (new tui_py_window (win_name, wrapper));
 
+  /* There's only two ways that m_constr can be reset back to nullptr,
+     first when the parent gdbpy_tui_window_maker object is deleted, in
+     which case it should be impossible to call this method, or second, as
+     a result of a gdbpy_tui_window_maker::invalidate_all call, but this is
+     only called when GDB's Python interpreter is being shut down, after
+     which, this method should not be called.  */
+  gdb_assert (m_constr != nullptr);
+
   gdbpy_ref<> user_window
     (PyObject_CallFunctionObjArgs (m_constr.get (),
 				   (PyObject *) wrapper.get (),
@@ -572,3 +612,11 @@ gdbpy_initialize_tui ()
 
   return 0;
 }
+
+/* Finalize this module.  */
+
+void
+gdbpy_finalize_tui ()
+{
+  gdbpy_tui_window_maker::invalidate_all ();
+}
diff --git a/gdb/python/python-internal.h b/gdb/python/python-internal.h
index c41a43bac96..8fb09796b15 100644
--- a/gdb/python/python-internal.h
+++ b/gdb/python/python-internal.h
@@ -550,6 +550,7 @@ int gdbpy_initialize_unwind (void)
   CPYCHECKER_NEGATIVE_RESULT_SETS_EXCEPTION;
 int gdbpy_initialize_tui ()
   CPYCHECKER_NEGATIVE_RESULT_SETS_EXCEPTION;
+void gdbpy_finalize_tui ();
 int gdbpy_initialize_membuf ()
   CPYCHECKER_NEGATIVE_RESULT_SETS_EXCEPTION;
 int gdbpy_initialize_connection ()
diff --git a/gdb/python/python.c b/gdb/python/python.c
index 54623f445ed..ed466cc4511 100644
--- a/gdb/python/python.c
+++ b/gdb/python/python.c
@@ -1950,6 +1950,7 @@ finalize_python (void *ignore)
   gdbpy_enter::finalize ();
 
   gdbpy_finalize_micommands ();
+  gdbpy_finalize_tui ();
 
   Py_Finalize ();
 
diff --git a/gdb/testsuite/gdb.python/tui-window-factory.exp b/gdb/testsuite/gdb.python/tui-window-factory.exp
index 99f9fbb1bc4..3e898d01c7b 100644
--- a/gdb/testsuite/gdb.python/tui-window-factory.exp
+++ b/gdb/testsuite/gdb.python/tui-window-factory.exp
@@ -71,3 +71,35 @@ with_test_prefix "msg_3" {
     Term::check_box_contents "check test_window box" 0 0 80 15 \
 	"TestWindow \\(msg_3\\)"
 }
+
+# Restart GDB, setup a TUI window factory, and then check that the
+# Python object is deallocated when GDB exits.
+with_test_prefix "call __del__ at exit" {
+    clean_restart
+
+    gdb_test "source ${pyfile}" "Python script imported" \
+	"import python scripts"
+
+    gdb_test "python register_window_factory('msg_1')" \
+	"Entering TestWindowFactory\\.__init__: msg_1"
+
+    gdb_test "python register_window_factory('msg_2')" \
+	[multi_line \
+	     "Entering TestWindowFactory\\.__init__: msg_2" \
+	     "Entering TestWindowFactory\\.__del__: msg_1"]
+
+    set saw_window_factory_del 0
+    gdb_test_multiple "quit" "" {
+	-re "^quit\r\n" {
+	    exp_continue
+	}
+	-re "^Entering TestWindowFactory.__del__: msg_2\r\n" {
+	    incr saw_window_factory_del
+	    exp_continue
+	}
+	eof {
+	    gdb_assert { $saw_window_factory_del == 1 }
+	    pass $gdb_test_name
+	}
+    }
+}
-- 
2.25.4


  parent reply	other threads:[~2023-01-25 13:56 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-12 19:19 [PATCH 0/3] Python/TUI Window Creation / Destruction Fixes Andrew Burgess
2023-01-12 19:19 ` [PATCH 1/3] gdb/python: allow Python TUI windows to be replaced Andrew Burgess
2023-01-12 19:19 ` [PATCH 2/3] gdb/python: deallocate tui window factories at Python shut down Andrew Burgess
2023-01-12 19:19 ` [PATCH 3/3] gdb/tui: don't leak the known_window_types map Andrew Burgess
2023-01-25 13:56 ` [PATCHv2 0/3] Python/TUI Window Creation / Destruction Fixes Andrew Burgess
2023-01-25 13:56   ` [PATCHv2 1/3] gdb/python: allow Python TUI windows to be replaced Andrew Burgess
2023-01-25 13:56   ` Andrew Burgess [this message]
2023-01-25 13:56   ` [PATCHv2 3/3] gdb/tui: don't leak the known_window_types map Andrew Burgess
2023-01-27 16:34   ` [PATCHv3 0/3] Python/TUI Window Creation / Destruction Fixes Andrew Burgess
2023-01-27 16:34     ` [PATCHv3 1/3] gdb/python: allow Python TUI windows to be replaced Andrew Burgess
2023-01-27 16:34     ` [PATCHv3 2/3] gdb/python: deallocate tui window factories at Python shut down Andrew Burgess
2023-01-27 16:34     ` [PATCHv3 3/3] gdb/tui: don't leak the known_window_types map Andrew Burgess
2023-02-13 12:32     ` [PATCHv3 0/3] Python/TUI Window Creation / Destruction Fixes Andrew Burgess
2023-02-13 14:11       ` Tom Tromey
2023-02-13 14:52         ` Andrew Burgess

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=80daa36b6b15e84c88df642e50e01b494cce0a04.1674654912.git.aburgess@redhat.com \
    --to=aburgess@redhat.com \
    --cc=gdb-patches@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).