[v2] Catch exception in value_rtti_indirect_type

  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
  

On 14-12-05 03:32 PM, Simon Marchi wrote:
> 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;

Ping.

(I realized I had forgotten about this patch while reviewing my opened bugs)
  

On 12/05/2014 08:32 PM, Simon Marchi wrote:

> 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.

This is OK, thanks.

> +/* This testcase is part of GDB, the GNU debugger.
> +
> +   Copyright 2014 Free Software Foundation, Inc.

Recall to update this to "2014-2015" before pushing.

> +++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
> @@ -0,0 +1,63 @@
> +# Copyright 2014 Free Software Foundation, Inc.
> +

Please update these to "2014-2015" before pushing.

Thanks,
Pedro Alves
  

On 15-01-23 07:33 AM, Pedro Alves wrote:
> On 12/05/2014 08:32 PM, Simon Marchi wrote:
> 
>> 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.
> 
> This is OK, thanks.
> 
>> +/* This testcase is part of GDB, the GNU debugger.
>> +
>> +   Copyright 2014 Free Software Foundation, Inc.
> 
> Recall to update this to "2014-2015" before pushing.
> 
>> +++ b/gdb/testsuite/gdb.mi/mi-var-list-children-invalid-grandchild.exp
>> @@ -0,0 +1,63 @@
>> +# Copyright 2014 Free Software Foundation, Inc.
>> +
> 
> Please update these to "2014-2015" before pushing.
> 
> Thanks,
> Pedro Alves

Hmm I am now getting a

/home/emaisin/src/binutils-gdb/gdb/valops.c:3625:13: error: ‘target’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
   real_type = value_rtti_type (target, full, top, using_enc);

Logically, target can't be uninitialized there because of the exception throw. I would expect gcc to be
smarter about it, since throw_error is marked as noreturn. I can obviously fix it by assigning it to NULL
at declaration time, but is there a better way of getting rid of the error in these cases?
  

On 01/23/2015 04:00 PM, Simon Marchi wrote:

> Hmm I am now getting a
> 
> /home/emaisin/src/binutils-gdb/gdb/valops.c:3625:13: error: ‘target’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
>    real_type = value_rtti_type (target, full, top, using_enc);
> 
> Logically, target can't be uninitialized there because of the exception throw. I would expect gcc to be
> smarter about it, since throw_error is marked as noreturn. I can obviously fix it by assigning it to NULL
> at declaration time, but is there a better way of getting rid of the error in these cases?

Bah.  Yeah, initializing as NULL is fine.

Thanks,
Pedro Alves
  

On 15-01-23 11:48 AM, Pedro Alves wrote:
> On 01/23/2015 04:00 PM, Simon Marchi wrote:
> 
>> Hmm I am now getting a
>>
>> /home/emaisin/src/binutils-gdb/gdb/valops.c:3625:13: error: ‘target’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
>>    real_type = value_rtti_type (target, full, top, using_enc);
>>
>> Logically, target can't be uninitialized there because of the exception throw. I would expect gcc to be
>> smarter about it, since throw_error is marked as noreturn. I can obviously fix it by assigning it to NULL
>> at declaration time, but is there a better way of getting rid of the error in these cases?
> 
> Bah.  Yeah, initializing as NULL is fine.
> 
> Thanks,
> Pedro Alves

Thanks, pushed.