Re: [PATCH v2] Catch exception in value_rtti_indirect_type

[Simon Marchi <simon dot marchi at ericsson dot com>]

On 2014-12-04 12:47 PM, Pedro Alves wrote:
> On 09/20/2014 02:02 AM, Simon Marchi wrote:
>> In the situation described in bug 17416 [1], 
> 
> Could you paste that in the commit log please?

Done.

>> an exception thrown in
>> value_ind can propagate too far and leave an half-built variable object,
>> leading to a wrong state. This patch adds a TRY_CATCH to catch it and
>> makes value_rtti_indirect_type return NULL in that case, meaning
>> that the type of the pointed object could not be found.
>>
>> If you want, I can also integrate the test case provided in the bug
> 
> Yes please.

Done.

>> description. I just don't know how to name it without giving it a
>> ridiculously long name such as
>> mi-var-list-children-with-print-object-on-and-a-null-pointer-to-a-structure-that-contains-a-pointer-to-a-structure.exp.
> 
> Perhaps mi-var-list-children-invalid-grandchild.exp.

Perfect, thanks.

>>
>> I tested the change on my machine, Ubuntu 14.10 x86-64.
>>
>> gdb/Changelog:
>>
>> 	* valops.c (value_rtti_indirect_type): Catch exception thrown by
>> 	value_ind.
>>
>> [1] https://sourceware.org/bugzilla/show_bug.cgi?id=17416
>> ---
>>  gdb/valops.c | 13 ++++++++++++-
>>  1 file changed, 12 insertions(+), 1 deletion(-)
>>
>> diff --git a/gdb/valops.c b/gdb/valops.c
>> index e1decf0..c1a0c86 100644
>> --- a/gdb/valops.c
>> +++ b/gdb/valops.c
>> @@ -3609,7 +3609,18 @@ value_rtti_indirect_type (struct value *v, int *full,
>>    if (TYPE_CODE (type) == TYPE_CODE_REF)
>>      target = coerce_ref (v);
>>    else if (TYPE_CODE (type) == TYPE_CODE_PTR)
>> -    target = value_ind (v);
>> +    {
>> +      volatile struct gdb_exception except;
>> +
>> +      TRY_CATCH (except, RETURN_MASK_ERROR)
>> +        {
>> +	  target = value_ind (v);
>> +        }
>> +      if (except.error == MEMORY_ERROR)
>> +	  return NULL;
>> +      else if (except.error != GDB_NO_ERROR)
>> +	throw_exception (except);
> 
> Indentation looks odd.  But, could you write it like this instead ?
> 
>       volatile struct gdb_exception except;
> 
>       TRY_CATCH (except, RETURN_MASK_ERROR)
>         {
> 	  target = value_ind (v);
>         }
>       if (except.reason < 0)
>         {
>           if (except.error == MEMORY_ERROR)
>             {
>               /* Add comment here.  */
> 	      return NULL;
>             }
>   	  throw_exception (except);
>         }

Indeed it was odd. I took your suggestion. Here is the updated v2:

>From 4f3924e505d6343c430292e2261d146e66d8799a Mon Sep 17 00:00:00 2001
From: Simon Marchi <simon.marchi@ericsson.com>
Date: Fri, 19 Sep 2014 21:02:33 -0400
Subject: [PATCH v2] Catch exception in value_rtti_indirect_type

In the situation described in bug 17416 [1]:

  * "set print object" is on;
  * The variable object is a pointer to a struct, and it contains an
    invalid value (e.g. NULL, or random uninitialized value);
  * The variable object (struct) has a child which is also a pointer to a
    struct;
  * We try to use "-var-list-children".

... an exception thrown in value_ind can propagate too far and leave an
half-built variable object, leading to a wrong state. This patch adds a
TRY_CATCH to catch it and makes value_rtti_indirect_type return NULL in
that case, meaning that the type of the pointed object could not be
found.

A test for the fix is also added.

New in v2:

  * Added test.
  * Restructured "catch" code.
  * Added details about the bug in commit log.

gdb/Changelog:

	* valops.c (value_rtti_indirect_type): Catch exception thrown by
	value_ind.

gdb/testsuite/ChangeLog

	* gdb.mi/mi-var-list-children-invalid-grandchild.c: New file.
	* gdb.mi/mi-var-list-children-invalid-grandchild.exp: New file.

[1] https://sourceware.org/bugzilla/show_bug.cgi?id=17416
---
 .../mi-var-list-children-invalid-grandchild.c      | 46 ++++++++++++++++
 .../mi-var-list-children-invalid-grandchild.exp    | 63 ++++++++++++++++++++++
 gdb/valops.c                                       | 20 ++++++-
 3 files changed, 128 insertions(+), 1 deletion(-)
 create mode 100644 gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
 create mode 100644 gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp

diff --git a/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
new file mode 100644
index 0000000..b67d875
--- /dev/null
+++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.c
@@ -0,0 +1,46 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+   Copyright 2014 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/>.  */
+
+struct inner
+{
+  int a;
+};
+
+struct outer
+{
+  struct inner *inner;
+};
+
+int main (void)
+{
+  struct inner inner;
+  struct outer outer;
+  struct outer *p_outer;
+
+  inner.a = 42;
+  outer.inner = &inner;
+
+  /* We force p_outer to an invalid value, but this also happens naturally
+   * when a variable has not been initialized. */
+
+  p_outer = 0;
+  /* p_outer set to invalid value */
+  p_outer = &outer;
+  /* p_outer set to valid value */
+
+  return 0;
+}
diff --git a/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
new file mode 100644
index 0000000..ed8460d
--- /dev/null
+++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
@@ -0,0 +1,63 @@
+# Copyright 2014 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/>.
+
+# This is a regression test for bug 17416
+# https://sourceware.org/bugzilla/show_bug.cgi?id=17416
+
+load_lib mi-support.exp
+set MIFLAGS "-i=mi"
+
+gdb_exit
+if [mi_gdb_start] {
+    continue
+}
+
+standard_testfile
+
+if  { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } {
+    untested mi-var-list-children-invalid-grandchild.exp
+    return -1
+}
+
+mi_delete_breakpoints
+mi_gdb_reinitialize_dir $srcdir/$subdir
+mi_gdb_load ${binfile}
+
+set line_invalid_pointer_value [gdb_get_line_number "p_outer set to invalid value"]
+set line_valid_pointer_value [gdb_get_line_number "p_outer set to valid value"]
+
+# set print object on
+mi_gdb_test "-interpreter-exec console \"set print object on\"" \
+    "(.*=cmd-param-changed,param=\"print object\",value=\"on\".*|)\\^done" \
+    "set print object on"
+
+mi_runto main
+
+mi_continue_to_line $line_invalid_pointer_value "continue to invalid pointer value"
+
+# Create variable object
+mi_gdb_test "-var-create var1 * p_outer" "\\^done,name=\"var1\",numchild=\"1\",value=\"0x0\",type=\"struct outer \\*\",thread-id=\"1\",has_more=\"0\"" "create variable object"
+
+# List children while the pointer is uninitialized.
+# Note: if you remove this -var-list-children, the second one will work.
+mi_gdb_test "-var-list-children var1" "\\^done,numchild=\"1\",children=\\\[child={name=\"var1.inner\",exp=\"inner\",numchild=\"1\",type=\"struct inner \\*\",thread-id=\"1\"}\\\],has_more=\"0\"" "list children #1"
+
+mi_continue_to_line $line_valid_pointer_value "continue to valid pointer value"
+
+# Update and list children again
+mi_gdb_test "-var-update var1" "\\^done,changelist=\\\[{name=\"var1\",in_scope=\"true\",type_changed=\"false\",has_more=\"0\"},{name=\"var1.inner\",in_scope=\"true\",type_changed=\"false\",has_more=\"0\"}\\\]" "update variable object"
+
+mi_gdb_test "-var-list-children var1" "\\^done,numchild=\"1\",children=\\\[child={name=\"var1.inner\",exp=\"inner\",numchild=\"1\",type=\"struct inner \\*\",thread-id=\"1\"}\\\],has_more=\"0\"" "list children #2"
+
diff --git a/gdb/valops.c b/gdb/valops.c
index 4125fc0..d806f27 100644
--- a/gdb/valops.c
+++ b/gdb/valops.c
@@ -3609,7 +3609,25 @@ value_rtti_indirect_type (struct value *v, int *full,
   if (TYPE_CODE (type) == TYPE_CODE_REF)
     target = coerce_ref (v);
   else if (TYPE_CODE (type) == TYPE_CODE_PTR)
-    target = value_ind (v);
+    {
+      volatile struct gdb_exception except;
+
+      TRY_CATCH (except, RETURN_MASK_ERROR)
+        {
+	  target = value_ind (v);
+        }
+      if (except.reason < 0)
+	{
+	  if (except.error == MEMORY_ERROR)
+	    {
+	      /* value_ind threw a memory error. The pointer is NULL or
+	         contains an uninitialized value: we can't determine any
+	         type.  */
+	      return NULL;
+	    }
+	  throw_exception (except);
+	}
+    }
   else
     return NULL;

-- 
2.1.3