public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
* [PATCH] Fix value chain use-after-free
@ 2023-02-11  1:08 Tom Tromey
  2023-02-14 17:05 ` Alexandra Petlanova Hajkova
  2023-02-27 22:45 ` Tom Tromey
  0 siblings, 2 replies; 4+ messages in thread
From: Tom Tromey @ 2023-02-11  1:08 UTC (permalink / raw)
  To: gdb-patches; +Cc: Tom Tromey

Hannes filed a bug showing a crash, where a pretty-printer written in
Python could cause a use-after-free.  He sent a patch, but I thought a
different approach was needed.

In a much earlier patch (see bug #12533), we changed the Python code
to release new values from the value chain when constructing a
gdb.Value.  The rationale for this is that if you write a command that
does a lot of computations in a loop, all the values will be kept live
by the value chain, resulting in gdb using a large amount of memory.

However, suppose a value is passed to Python from some code in gdb
that needs to use the value after the call into Python.  In this
scenario, value_to_value_object will still release the value -- and
because gdb code doesn't generally keep strong references to values (a
consequence of the ancient decision to use the value chain to avoid
memory management), this will result in a use-after-free.

This scenario can happen, as it turns out, when a value is passed to
Python for pretty-printing.  Now, normally this route boxes the value
via value_to_value_object_no_release, avoiding the problematic release
from the value chain.  However, if you then call Value.cast, the
underlying value API might return the same value, when is then
released from the chain.

This patch fixes the problem by changing how value boxing is done.
value_to_value_object no longer removes a value from the chain.
Instead, every spot in gdb that might construct new values uses a
scoped_value_mark to ensure that the requirements of bug #12533 are
met.  And, because incoming values aren't ever released from the chain
(the Value.cast one comes earlier on the chain than the
scoped_value_mark), the bug can no longer occur.  (Note that many
spots in the Python layer already take this approach, so not many
places needed to be touched.)

In the future I think we should replace the use of raw "value *" with
value_ref_ptr pretty much everywhere.  This will ensure lifetime
safety throughout gdb.

The test case in this patch comes from Hannes' original patch.  I only
made a trivial ("require") change to it.  However, while this fails
for him, I can't make it fail on this machine; nevertheless, he tried
my patch and reported the bug as being fixed.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=30044
---
 gdb/python/py-finishbreakpoint.c        |  4 ++
 gdb/python/py-frame.c                   | 17 +++++---
 gdb/python/py-lazy-string.c             |  9 ++++-
 gdb/python/py-prettyprint.c             |  2 +-
 gdb/python/py-symbol.c                  |  8 ++--
 gdb/python/py-type.c                    |  9 +++--
 gdb/python/py-unwind.c                  | 11 +++--
 gdb/python/py-value.c                   | 54 ++++++++++---------------
 gdb/python/py-xmethods.c                |  1 +
 gdb/python/python-internal.h            |  1 -
 gdb/python/python.c                     |  8 ++--
 gdb/testsuite/gdb.python/py-pp-cast.c   | 35 ++++++++++++++++
 gdb/testsuite/gdb.python/py-pp-cast.exp | 40 ++++++++++++++++++
 gdb/testsuite/gdb.python/py-pp-cast.py  | 28 +++++++++++++
 14 files changed, 173 insertions(+), 54 deletions(-)
 create mode 100644 gdb/testsuite/gdb.python/py-pp-cast.c
 create mode 100644 gdb/testsuite/gdb.python/py-pp-cast.exp
 create mode 100644 gdb/testsuite/gdb.python/py-pp-cast.py

diff --git a/gdb/python/py-finishbreakpoint.c b/gdb/python/py-finishbreakpoint.c
index d4d129110e9..3b682f5ad36 100644
--- a/gdb/python/py-finishbreakpoint.c
+++ b/gdb/python/py-finishbreakpoint.c
@@ -112,6 +112,8 @@ bpfinishpy_pre_stop_hook (struct gdbpy_breakpoint_object *bp_obj)
 
   try
     {
+      scoped_value_mark free_values;
+
       struct symbol *func_symbol =
 	symbol_object_to_symbol (self_finishbp->func_symbol);
       struct value *function =
@@ -258,6 +260,8 @@ bpfinishpy_init (PyObject *self, PyObject *args, PyObject *kwargs)
 	      /* Remember only non-void return types.  */
 	      if (ret_type->code () != TYPE_CODE_VOID)
 		{
+		  scoped_value_mark free_values;
+
 		  /* Ignore Python errors at this stage.  */
 		  value *func_value = read_var_value (function, NULL, frame);
 		  self_bpfinish->function_value
diff --git a/gdb/python/py-frame.c b/gdb/python/py-frame.c
index f66d22bfa3d..ecd55d2e568 100644
--- a/gdb/python/py-frame.c
+++ b/gdb/python/py-frame.c
@@ -241,12 +241,13 @@ static PyObject *
 frapy_read_register (PyObject *self, PyObject *args)
 {
   PyObject *pyo_reg_id;
-  struct value *val = NULL;
+  PyObject *result = nullptr;
 
   if (!PyArg_UnpackTuple (args, "read_register", 1, 1, &pyo_reg_id))
     return NULL;
   try
     {
+      scoped_value_mark free_values;
       frame_info_ptr frame;
       int regnum;
 
@@ -257,17 +258,19 @@ frapy_read_register (PyObject *self, PyObject *args)
 	return nullptr;
 
       gdb_assert (regnum >= 0);
-      val = value_of_register (regnum, frame);
+      struct value *val = value_of_register (regnum, frame);
 
       if (val == NULL)
 	PyErr_SetString (PyExc_ValueError, _("Can't read register."));
+      else
+	result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return val == NULL ? NULL : value_to_value_object (val);
+  return result;
 }
 
 /* Implementation of gdb.Frame.block (self) -> gdb.Block.
@@ -482,7 +485,6 @@ frapy_read_var (PyObject *self, PyObject *args)
   PyObject *sym_obj, *block_obj = NULL;
   struct symbol *var = NULL;	/* gcc-4.3.2 false warning.  */
   const struct block *block = NULL;
-  struct value *val = NULL;
 
   if (!PyArg_ParseTuple (args, "O|O", &sym_obj, &block_obj))
     return NULL;
@@ -540,18 +542,21 @@ frapy_read_var (PyObject *self, PyObject *args)
       return NULL;
     }
 
+  PyObject *result = nullptr;
   try
     {
       FRAPY_REQUIRE_VALID (self, frame);
 
-      val = read_var_value (var, block, frame);
+      scoped_value_mark free_values;
+      struct value *val = read_var_value (var, block, frame);
+      result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (val);
+  return result;
 }
 
 /* Select this frame.  */
diff --git a/gdb/python/py-lazy-string.c b/gdb/python/py-lazy-string.c
index 7177d74c32e..9b1205463d4 100644
--- a/gdb/python/py-lazy-string.c
+++ b/gdb/python/py-lazy-string.c
@@ -104,7 +104,6 @@ static PyObject *
 stpy_convert_to_value (PyObject *self, PyObject *args)
 {
   lazy_string_object *self_string = (lazy_string_object *) self;
-  struct value *val = NULL;
 
   if (self_string->address == 0)
     {
@@ -113,10 +112,14 @@ stpy_convert_to_value (PyObject *self, PyObject *args)
       return NULL;
     }
 
+  PyObject *result = nullptr;
   try
     {
+      scoped_value_mark free_values;
+
       struct type *type = type_object_to_type (self_string->type);
       struct type *realtype;
+      struct value *val;
 
       gdb_assert (type != NULL);
       realtype = check_typedef (type);
@@ -141,13 +144,15 @@ stpy_convert_to_value (PyObject *self, PyObject *args)
 	  val = value_at_lazy (type, self_string->address);
 	  break;
 	}
+
+      result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (val);
+  return result;
 }
 
 static void
diff --git a/gdb/python/py-prettyprint.c b/gdb/python/py-prettyprint.c
index 1a5ce08c477..17dd648ff9e 100644
--- a/gdb/python/py-prettyprint.c
+++ b/gdb/python/py-prettyprint.c
@@ -590,7 +590,7 @@ gdbpy_apply_val_pretty_printer (const struct extension_language_defn *extlang,
 
   gdbpy_enter enter_py (gdbarch, language);
 
-  gdbpy_ref<> val_obj (value_to_value_object_no_release (value));
+  gdbpy_ref<> val_obj (value_to_value_object (value));
   if (val_obj == NULL)
     {
       print_stack_unless_memory_error (stream);
diff --git a/gdb/python/py-symbol.c b/gdb/python/py-symbol.c
index b8777966c47..e6497ed473f 100644
--- a/gdb/python/py-symbol.c
+++ b/gdb/python/py-symbol.c
@@ -267,7 +267,6 @@ sympy_value (PyObject *self, PyObject *args)
   struct symbol *symbol = NULL;
   frame_info_ptr frame_info = NULL;
   PyObject *frame_obj = NULL;
-  struct value *value = NULL;
 
   if (!PyArg_ParseTuple (args, "|O", &frame_obj))
     return NULL;
@@ -285,6 +284,7 @@ sympy_value (PyObject *self, PyObject *args)
       return NULL;
     }
 
+  PyObject *result = nullptr;
   try
     {
       if (frame_obj != NULL)
@@ -301,14 +301,16 @@ sympy_value (PyObject *self, PyObject *args)
 	 was found, so we have no block to pass to read_var_value.  This will
 	 yield an incorrect value when symbol is not local to FRAME_INFO (this
 	 can happen with nested functions).  */
-      value = read_var_value (symbol, NULL, frame_info);
+      scoped_value_mark free_values;
+      struct value *value = read_var_value (symbol, NULL, frame_info);
+      result = value_to_value_object (value);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (value);
+  return result;
 }
 
 /* Given a symbol, and a symbol_object that has previously been
diff --git a/gdb/python/py-type.c b/gdb/python/py-type.c
index 81b595140af..051d77477a0 100644
--- a/gdb/python/py-type.c
+++ b/gdb/python/py-type.c
@@ -957,7 +957,6 @@ typy_template_argument (PyObject *self, PyObject *args)
   const struct block *block = NULL;
   PyObject *block_obj = NULL;
   struct symbol *sym;
-  struct value *val = NULL;
 
   if (! PyArg_ParseTuple (args, "i|O", &argno, &block_obj))
     return NULL;
@@ -1014,16 +1013,19 @@ typy_template_argument (PyObject *self, PyObject *args)
       return NULL;
     }
 
+  PyObject *result = nullptr;
   try
     {
-      val = value_of_variable (sym, block);
+      scoped_value_mark free_values;
+      struct value *val = value_of_variable (sym, block);
+      result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (val);
+  return result;
 }
 
 static PyObject *
@@ -1189,6 +1191,7 @@ typy_optimized_out (PyObject *self, PyObject *args)
 {
   struct type *type = ((type_object *) self)->type;
 
+  scoped_value_mark free_values;
   return value_to_value_object (allocate_optimized_out_value (type));
 }
 
diff --git a/gdb/python/py-unwind.c b/gdb/python/py-unwind.c
index e7ec660e150..bafecc23ea9 100644
--- a/gdb/python/py-unwind.c
+++ b/gdb/python/py-unwind.c
@@ -365,7 +365,6 @@ static PyObject *
 pending_framepy_read_register (PyObject *self, PyObject *args)
 {
   pending_frame_object *pending_frame = (pending_frame_object *) self;
-  struct value *val = NULL;
   int regnum;
   PyObject *pyo_reg_id;
 
@@ -380,25 +379,31 @@ pending_framepy_read_register (PyObject *self, PyObject *args)
   if (!gdbpy_parse_register_id (pending_frame->gdbarch, pyo_reg_id, &regnum))
     return nullptr;
 
+  PyObject *result = nullptr;
   try
     {
+      scoped_value_mark free_values;
+
       /* Fetch the value associated with a register, whether it's
 	 a real register or a so called "user" register, like "pc",
 	 which maps to a real register.  In the past,
 	 get_frame_register_value() was used here, which did not
 	 handle the user register case.  */
-      val = value_of_register (regnum, pending_frame->frame_info);
+      struct value *val = value_of_register (regnum,
+					     pending_frame->frame_info);
       if (val == NULL)
 	PyErr_Format (PyExc_ValueError,
 		      "Cannot read register %d from frame.",
 		      regnum);
+      else
+	result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return val == NULL ? NULL : value_to_value_object (val);
+  return result;
 }
 
 /* Implementation of
diff --git a/gdb/python/py-value.c b/gdb/python/py-value.c
index da2565081e6..14d5f1ad2d2 100644
--- a/gdb/python/py-value.c
+++ b/gdb/python/py-value.c
@@ -1559,18 +1559,20 @@ valpy_nonzero (PyObject *self)
 static PyObject *
 valpy_invert (PyObject *self)
 {
-  struct value *val = NULL;
+  PyObject *result = nullptr;
 
   try
     {
-      val = value_complement (((value_object *) self)->value);
+      scoped_value_mark free_values;
+      struct value *val = value_complement (((value_object *) self)->value);
+      result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (val);
+  return result;
 }
 
 /* Implements left shift for value objects.  */
@@ -1774,32 +1776,10 @@ valpy_float (PyObject *self)
   return PyFloat_FromDouble (d);
 }
 
-/* Returns an object for a value which is released from the all_values chain,
-   so its lifetime is not bound to the execution of a command.  */
-PyObject *
-value_to_value_object (struct value *val)
-{
-  value_object *val_obj;
-
-  val_obj = PyObject_New (value_object, &value_object_type);
-  if (val_obj != NULL)
-    {
-      val_obj->value = release_value (val).release ();
-      val_obj->next = nullptr;
-      val_obj->prev = nullptr;
-      val_obj->address = NULL;
-      val_obj->type = NULL;
-      val_obj->dynamic_type = NULL;
-      note_value (val_obj);
-    }
-
-  return (PyObject *) val_obj;
-}
-
-/* Returns an object for a value, but without releasing it from the
+/* Returns an object for a value, without releasing it from the
    all_values chain.  */
 PyObject *
-value_to_value_object_no_release (struct value *val)
+value_to_value_object (struct value *val)
 {
   value_object *val_obj;
 
@@ -1926,21 +1906,23 @@ PyObject *
 gdbpy_history (PyObject *self, PyObject *args)
 {
   int i;
-  struct value *res_val = NULL;	  /* Initialize to appease gcc warning.  */
 
   if (!PyArg_ParseTuple (args, "i", &i))
     return NULL;
 
+  PyObject *result = nullptr;
   try
     {
-      res_val = access_value_history (i);
+      scoped_value_mark free_values;
+      struct value *res_val = access_value_history (i);
+      result = value_to_value_object (res_val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (res_val);
+  return result;
 }
 
 /* Add a gdb.Value into GDB's history, and return (as an integer) the
@@ -1988,15 +1970,23 @@ gdbpy_convenience_variable (PyObject *self, PyObject *args)
   if (!PyArg_ParseTuple (args, "s", &varname))
     return NULL;
 
+  PyObject *result = nullptr;
+  bool found = false;
   try
     {
       struct internalvar *var = lookup_only_internalvar (varname);
 
       if (var != NULL)
 	{
+	  scoped_value_mark free_values;
 	  res_val = value_of_internalvar (gdbpy_enter::get_gdbarch (), var);
 	  if (value_type (res_val)->code () == TYPE_CODE_VOID)
 	    res_val = NULL;
+	  else
+	    {
+	      found = true;
+	      result = value_to_value_object (res_val);
+	    }
 	}
     }
   catch (const gdb_exception &except)
@@ -2004,10 +1994,10 @@ gdbpy_convenience_variable (PyObject *self, PyObject *args)
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  if (res_val == NULL)
+  if (result == nullptr && !found)
     Py_RETURN_NONE;
 
-  return value_to_value_object (res_val);
+  return result;
 }
 
 /* Set the value of a convenience variable.  */
diff --git a/gdb/python/py-xmethods.c b/gdb/python/py-xmethods.c
index d6d607d99d4..d03a2272868 100644
--- a/gdb/python/py-xmethods.c
+++ b/gdb/python/py-xmethods.c
@@ -423,6 +423,7 @@ python_xmethod_worker::do_get_result_type (value *obj,
       return EXT_LANG_RC_OK;
     }
 
+  scoped_value_mark free_values;
   obj_type = check_typedef (value_type (obj));
   this_type = check_typedef (type_object_to_type (m_this_type));
   if (obj_type->code () == TYPE_CODE_PTR)
diff --git a/gdb/python/python-internal.h b/gdb/python/python-internal.h
index c41a43bac96..39f47d6b1e2 100644
--- a/gdb/python/python-internal.h
+++ b/gdb/python/python-internal.h
@@ -437,7 +437,6 @@ PyObject *symbol_to_symbol_object (struct symbol *sym);
 PyObject *block_to_block_object (const struct block *block,
 				 struct objfile *objfile);
 PyObject *value_to_value_object (struct value *v);
-PyObject *value_to_value_object_no_release (struct value *v);
 PyObject *type_to_type_object (struct type *);
 PyObject *frame_info_to_frame_object (frame_info_ptr frame);
 PyObject *symtab_to_linetable_object (PyObject *symtab);
diff --git a/gdb/python/python.c b/gdb/python/python.c
index 54623f445ed..8c23e0cd05e 100644
--- a/gdb/python/python.c
+++ b/gdb/python/python.c
@@ -969,22 +969,24 @@ static PyObject *
 gdbpy_parse_and_eval (PyObject *self, PyObject *args)
 {
   const char *expr_str;
-  struct value *result = NULL;
 
   if (!PyArg_ParseTuple (args, "s", &expr_str))
     return NULL;
 
+  PyObject *result = nullptr;
   try
     {
       gdbpy_allow_threads allow_threads;
-      result = parse_and_eval (expr_str);
+      scoped_value_mark free_values;
+      struct value *val = parse_and_eval (expr_str);
+      result = value_to_value_object (val);
     }
   catch (const gdb_exception &except)
     {
       GDB_PY_HANDLE_EXCEPTION (except);
     }
 
-  return value_to_value_object (result);
+  return result;
 }
 
 /* Implementation of gdb.invalidate_cached_frames.  */
diff --git a/gdb/testsuite/gdb.python/py-pp-cast.c b/gdb/testsuite/gdb.python/py-pp-cast.c
new file mode 100644
index 00000000000..72716a79dd4
--- /dev/null
+++ b/gdb/testsuite/gdb.python/py-pp-cast.c
@@ -0,0 +1,35 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2023 Free Software Foundation, Inc.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see  <http://www.gnu.org/licenses/>.  */
+
+typedef int pp_int;
+
+int break_function()
+{
+  return 0;
+}
+
+struct container
+{
+  pp_int p_i;
+  int i;
+};
+
+int main()
+{
+  struct container c = { 10, 5 };
+  return break_function();
+}
diff --git a/gdb/testsuite/gdb.python/py-pp-cast.exp b/gdb/testsuite/gdb.python/py-pp-cast.exp
new file mode 100644
index 00000000000..8a6f2976046
--- /dev/null
+++ b/gdb/testsuite/gdb.python/py-pp-cast.exp
@@ -0,0 +1,40 @@
+# Copyright (C) 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+# Test casting of a gdb.Value inside a pretty printer.
+
+load_lib gdb-python.exp
+
+require allow_python_tests
+
+standard_testfile
+
+if { [prepare_for_testing "failed to prepare" ${testfile} ${srcfile}] } {
+    return -1
+}
+
+if ![runto break_function] {
+    return -1
+}
+
+set remote_python_file [gdb_remote_download host \
+			    ${srcdir}/${subdir}/${testfile}.py]
+
+gdb_test_no_output "source ${remote_python_file}" \
+    "source ${testfile}.py"
+
+gdb_test "up" "#1.*main.*"
+
+gdb_test "info locals" "c = {p_i = 10p, i = 5}"
diff --git a/gdb/testsuite/gdb.python/py-pp-cast.py b/gdb/testsuite/gdb.python/py-pp-cast.py
new file mode 100644
index 00000000000..b171a919c70
--- /dev/null
+++ b/gdb/testsuite/gdb.python/py-pp-cast.py
@@ -0,0 +1,28 @@
+# Copyright (C) 2023 Free Software Foundation, Inc.
+
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+
+class PpIntPrinter(object):
+    def __init__(self, val):
+        self.val = val
+
+    def to_string(self):
+        val = self.val.cast(self.val.type)
+        return "%dp" % int(val)
+
+
+pp = gdb.printing.RegexpCollectionPrettyPrinter("pp-cast")
+pp.add_printer("pp_int", "^pp_int$", PpIntPrinter)
+gdb.printing.register_pretty_printer(gdb.current_objfile(), pp)
-- 
2.39.1


^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] Fix value chain use-after-free
  2023-02-11  1:08 [PATCH] Fix value chain use-after-free Tom Tromey
@ 2023-02-14 17:05 ` Alexandra Petlanova Hajkova
  2023-02-14 22:16   ` Tom Tromey
  2023-02-27 22:45 ` Tom Tromey
  1 sibling, 1 reply; 4+ messages in thread
From: Alexandra Petlanova Hajkova @ 2023-02-14 17:05 UTC (permalink / raw)
  To: Tom Tromey; +Cc: gdb-patches

[-- Attachment #1: Type: text/plain, Size: 2575 bytes --]

On Sat, Feb 11, 2023 at 2:08 AM Tom Tromey <tom@tromey.com> wrote:

> Hannes filed a bug showing a crash, where a pretty-printer written in
> Python could cause a use-after-free.  He sent a patch, but I thought a
> different approach was needed.
>
> In a much earlier patch (see bug #12533), we changed the Python code
> to release new values from the value chain when constructing a
> gdb.Value.  The rationale for this is that if you write a command that
> does a lot of computations in a loop, all the values will be kept live
> by the value chain, resulting in gdb using a large amount of memory.
>
> However, suppose a value is passed to Python from some code in gdb
> that needs to use the value after the call into Python.  In this
> scenario, value_to_value_object will still release the value -- and
> because gdb code doesn't generally keep strong references to values (a
> consequence of the ancient decision to use the value chain to avoid
> memory management), this will result in a use-after-free.
>
> This scenario can happen, as it turns out, when a value is passed to
> Python for pretty-printing.  Now, normally this route boxes the value
> via value_to_value_object_no_release, avoiding the problematic release
> from the value chain.  However, if you then call Value.cast, the
> underlying value API might return the same value, when is then
> released from the chain.
>
> This patch fixes the problem by changing how value boxing is done.
> value_to_value_object no longer removes a value from the chain.
> Instead, every spot in gdb that might construct new values uses a
> scoped_value_mark to ensure that the requirements of bug #12533 are
> met.  And, because incoming values aren't ever released from the chain
> (the Value.cast one comes earlier on the chain than the
> scoped_value_mark), the bug can no longer occur.  (Note that many
> spots in the Python layer already take this approach, so not many
> places needed to be touched.)
>
> In the future I think we should replace the use of raw "value *" with
> value_ref_ptr pretty much everywhere.  This will ensure lifetime
> safety throughout gdb.
>
> The test case in this patch comes from Hannes' original patch.  I only
> made a trivial ("require") change to it.  However, while this fails
> for him, I can't make it fail on this machine; nevertheless, he tried
> my patch and reported the bug as being fixed.
>
> 2.39.1
>
>
I think it's a good solution and I can confirm gdb.python/py-pp-cast.exp
passes for me. But I wasn't able to apply the patch cleanly, I think it
needs to be rebased.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] Fix value chain use-after-free
  2023-02-14 17:05 ` Alexandra Petlanova Hajkova
@ 2023-02-14 22:16   ` Tom Tromey
  0 siblings, 0 replies; 4+ messages in thread
From: Tom Tromey @ 2023-02-14 22:16 UTC (permalink / raw)
  To: Alexandra Petlanova Hajkova; +Cc: Tom Tromey, gdb-patches

Alexandra> I think it's a good solution and I can confirm
Alexandra> gdb.python/py-pp-cast.exp passes for me. But I wasn't able to
Alexandra> apply the patch cleanly, I think it needs to be rebased.

Thank you.  I usually rebase if there are either changes required or
before checking in.  I also push all my patches to my github though it
can be hard sometimes to figure out exactly what branch they're on.

Tom

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] Fix value chain use-after-free
  2023-02-11  1:08 [PATCH] Fix value chain use-after-free Tom Tromey
  2023-02-14 17:05 ` Alexandra Petlanova Hajkova
@ 2023-02-27 22:45 ` Tom Tromey
  1 sibling, 0 replies; 4+ messages in thread
From: Tom Tromey @ 2023-02-27 22:45 UTC (permalink / raw)
  To: Tom Tromey; +Cc: gdb-patches

>>>>> "Tom" == Tom Tromey <tom@tromey.com> writes:

Tom> This patch fixes the problem by changing how value boxing is done.
[...]

I am checking this in now.

Tom

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2023-02-27 22:45 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-02-11  1:08 [PATCH] Fix value chain use-after-free Tom Tromey
2023-02-14 17:05 ` Alexandra Petlanova Hajkova
2023-02-14 22:16   ` Tom Tromey
2023-02-27 22:45 ` Tom Tromey

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).