[PATCH] Small segfault fix when there is no python

[PATCH] Small segfault fix when there is no python

[Daniel Gutson]

[-- Attachment #1: Type: text/plain, Size: 639 bytes --]

Hi,

   the small attached patch prevents gdb to segfault when an extension
language definition has no ops,
which e.g. occurs when HAVE_PYTHON is not defined so
extension_language_python remains with ops in NULL.
This causes the line
   if (extlang->ops->eval_from_control_command != NULL)
(in eval_ext_lang_from_control_command) to dereference a null pointer.

Please commit it for me if approved since I don't have write access.

Thanks,

   Daniel.


    2014-03-20    Daniel Gutson (daniel.gutson@tallertechnologies.com)

gdb/
    * extension.c: (eval_ext_lang_from_control_command) Added check to
prevent dereference of null pointer.

[-- Attachment #2: python.patch --]
[-- Type: text/x-patch, Size: 768 bytes --]

diff --git a/gdb/extension.c b/gdb/extension.c
index c2f502b..8357ee8 100644
--- a/gdb/extension.c
+++ b/gdb/extension.c
@@ -342,11 +342,14 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
     {
       if (extlang->cli_control_type == cmd->control_type)
 	{
-	  if (extlang->ops->eval_from_control_command != NULL)
-	    {
-	      extlang->ops->eval_from_control_command (extlang, cmd);
-	      return;
-	    }
+      if (extlang->ops != NULL)
+        {  
+	    if (extlang->ops->eval_from_control_command != NULL)
+	      {
+	        extlang->ops->eval_from_control_command (extlang, cmd);
+	        return;
+	      }
+        }
 	  /* The requested extension language is not supported in this GDB.  */
 	  throw_ext_lang_unsupported (extlang);
 	}

Re: [PATCH] Small segfault fix when there is no python

[Sergio Durigan Junior]

On Thursday, March 20 2014, Daniel Gutson wrote:

> Hi,
>
>    the small attached patch prevents gdb to segfault when an extension
> language definition has no ops,
> which e.g. occurs when HAVE_PYTHON is not defined so
> extension_language_python remains with ops in NULL.
> This causes the line
>    if (extlang->ops->eval_from_control_command != NULL)
> (in eval_ext_lang_from_control_command) to dereference a null pointer.

Hi Daniel,

Thanks for the patch.  This is a simple patch so it doesn't need a
copyright assignment from you.  However, if you intend to continue
contributing to GDB, please e-mail me offlist and I can send you the
papers to obtain the assignment.

Just a few comments about formatting issues.

>     2014-03-20    Daniel Gutson (daniel.gutson@tallertechnologies.com)
>
> gdb/
>     * extension.c: (eval_ext_lang_from_control_command) Added check to
> prevent dereference of null pointer.

The ChangeLog format is wrong.  Take a look at the ChangeLog file for
lots of examples, but basically you need to write:

2014-03-20  Your Name  <your@email>

	* file.c (function): Added check to prevent blabla...

Pay attention to the 2 spaces between the date, the name and the e-mail,
and also to the TAB character indenting the description.

> diff --git a/gdb/extension.c b/gdb/extension.c
> index c2f502b..8357ee8 100644
> --- a/gdb/extension.c
> +++ b/gdb/extension.c
> @@ -342,11 +342,14 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
>      {
>        if (extlang->cli_control_type == cmd->control_type)
>  	{
> -	  if (extlang->ops->eval_from_control_command != NULL)
> -	    {
> -	      extlang->ops->eval_from_control_command (extlang, cmd);
> -	      return;
> -	    }
> +      if (extlang->ops != NULL)
> +        {  
> +	    if (extlang->ops->eval_from_control_command != NULL)
> +	      {
> +	        extlang->ops->eval_from_control_command (extlang, cmd);
> +	        return;
> +	      }
> +        }

You could simplify this by writing:

  if (extlang->ops != NULL && extlang->ops->eval_from_control_command != NULL)

If you don't want to join the two "if"s, then you don't need to put the
braces on the outter "if", because it has one single statement.

The patch looks good to me, but I'm not a maintainer and can't approve
it.

Thanks,

-- 
Sergio

Re: [PATCH] Small segfault fix when there is no python

[Doug Evans]

On Thu, Mar 20, 2014 at 10:31 AM, Sergio Durigan Junior
<sergiodj@redhat.com> wrote:
> On Thursday, March 20 2014, Daniel Gutson wrote:
>
>> Hi,
>>
>>    the small attached patch prevents gdb to segfault when an extension
>> language definition has no ops,
>> which e.g. occurs when HAVE_PYTHON is not defined so
>> extension_language_python remains with ops in NULL.
>> This causes the line
>>    if (extlang->ops->eval_from_control_command != NULL)
>> (in eval_ext_lang_from_control_command) to dereference a null pointer.
>
> Hi Daniel,
>
> Thanks for the patch.  This is a simple patch so it doesn't need a
> copyright assignment from you.  However, if you intend to continue
> contributing to GDB, please e-mail me offlist and I can send you the
> papers to obtain the assignment.
>
> Just a few comments about formatting issues.
>
>>     2014-03-20    Daniel Gutson (daniel.gutson@tallertechnologies.com)
>>
>> gdb/
>>     * extension.c: (eval_ext_lang_from_control_command) Added check to
>> prevent dereference of null pointer.
>
> The ChangeLog format is wrong.  Take a look at the ChangeLog file for
> lots of examples, but basically you need to write:
>
> 2014-03-20  Your Name  <your@email>
>
>         * file.c (function): Added check to prevent blabla...
>
> Pay attention to the 2 spaces between the date, the name and the e-mail,
> and also to the TAB character indenting the description.
>
>> diff --git a/gdb/extension.c b/gdb/extension.c
>> index c2f502b..8357ee8 100644
>> --- a/gdb/extension.c
>> +++ b/gdb/extension.c
>> @@ -342,11 +342,14 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
>>      {
>>        if (extlang->cli_control_type == cmd->control_type)
>>       {
>> -       if (extlang->ops->eval_from_control_command != NULL)
>> -         {
>> -           extlang->ops->eval_from_control_command (extlang, cmd);
>> -           return;
>> -         }
>> +      if (extlang->ops != NULL)
>> +        {
>> +         if (extlang->ops->eval_from_control_command != NULL)
>> +           {
>> +             extlang->ops->eval_from_control_command (extlang, cmd);
>> +             return;
>> +           }
>> +        }
>
> You could simplify this by writing:
>
>   if (extlang->ops != NULL && extlang->ops->eval_from_control_command != NULL)
>
> If you don't want to join the two "if"s, then you don't need to put the
> braces on the outter "if", because it has one single statement.
>
> The patch looks good to me, but I'm not a maintainer and can't approve
> it.
>
> Thanks,
>
> --
> Sergio


I did an audit of all the uses of ->ops and think this is the only one I missed.
I will commit with the needed changes.
Thanks for the patch!

Re: [PATCH] Small segfault fix when there is no python

[Daniel Gutson]

Thanks Doug and Sergio!

   Daniel.


On Thu, Mar 20, 2014 at 2:52 PM, Doug Evans <dje@google.com> wrote:
> On Thu, Mar 20, 2014 at 10:31 AM, Sergio Durigan Junior
> <sergiodj@redhat.com> wrote:
>> On Thursday, March 20 2014, Daniel Gutson wrote:
>>
>>> Hi,
>>>
>>>    the small attached patch prevents gdb to segfault when an extension
>>> language definition has no ops,
>>> which e.g. occurs when HAVE_PYTHON is not defined so
>>> extension_language_python remains with ops in NULL.
>>> This causes the line
>>>    if (extlang->ops->eval_from_control_command != NULL)
>>> (in eval_ext_lang_from_control_command) to dereference a null pointer.
>>
>> Hi Daniel,
>>
>> Thanks for the patch.  This is a simple patch so it doesn't need a
>> copyright assignment from you.  However, if you intend to continue
>> contributing to GDB, please e-mail me offlist and I can send you the
>> papers to obtain the assignment.
>>
>> Just a few comments about formatting issues.
>>
>>>     2014-03-20    Daniel Gutson (daniel.gutson@tallertechnologies.com)
>>>
>>> gdb/
>>>     * extension.c: (eval_ext_lang_from_control_command) Added check to
>>> prevent dereference of null pointer.
>>
>> The ChangeLog format is wrong.  Take a look at the ChangeLog file for
>> lots of examples, but basically you need to write:
>>
>> 2014-03-20  Your Name  <your@email>
>>
>>         * file.c (function): Added check to prevent blabla...
>>
>> Pay attention to the 2 spaces between the date, the name and the e-mail,
>> and also to the TAB character indenting the description.
>>
>>> diff --git a/gdb/extension.c b/gdb/extension.c
>>> index c2f502b..8357ee8 100644
>>> --- a/gdb/extension.c
>>> +++ b/gdb/extension.c
>>> @@ -342,11 +342,14 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
>>>      {
>>>        if (extlang->cli_control_type == cmd->control_type)
>>>       {
>>> -       if (extlang->ops->eval_from_control_command != NULL)
>>> -         {
>>> -           extlang->ops->eval_from_control_command (extlang, cmd);
>>> -           return;
>>> -         }
>>> +      if (extlang->ops != NULL)
>>> +        {
>>> +         if (extlang->ops->eval_from_control_command != NULL)
>>> +           {
>>> +             extlang->ops->eval_from_control_command (extlang, cmd);
>>> +             return;
>>> +           }
>>> +        }
>>
>> You could simplify this by writing:
>>
>>   if (extlang->ops != NULL && extlang->ops->eval_from_control_command != NULL)
>>
>> If you don't want to join the two "if"s, then you don't need to put the
>> braces on the outter "if", because it has one single statement.
>>
>> The patch looks good to me, but I'm not a maintainer and can't approve
>> it.
>>
>> Thanks,
>>
>> --
>> Sergio
>
>
> I did an audit of all the uses of ->ops and think this is the only one I missed.
> I will commit with the needed changes.
> Thanks for the patch!



-- 

Daniel F. Gutson
Chief Engineering Officer, SPD


San Lorenzo 47, 3rd Floor, Office 5

Córdoba, Argentina


Phone: +54 351 4217888 / +54 351 4218211

Skype: dgutson

Re: [PATCH] Small segfault fix when there is no python

[Doug Evans]

Daniel Gutson <daniel.gutson@tallertechnologies.com> writes:
> Thanks Doug and Sergio!
>
>    Daniel.
>
>
> On Thu, Mar 20, 2014 at 2:52 PM, Doug Evans <dje@google.com> wrote:
>> On Thu, Mar 20, 2014 at 10:31 AM, Sergio Durigan Junior
>> <sergiodj@redhat.com> wrote:
>>> On Thursday, March 20 2014, Daniel Gutson wrote:
>>>
>>>> Hi,
>>>>
>>>>    the small attached patch prevents gdb to segfault when an extension
>>>> language definition has no ops,
>>>> which e.g. occurs when HAVE_PYTHON is not defined so
>>>> extension_language_python remains with ops in NULL.
>>>> This causes the line
>>>>    if (extlang->ops->eval_from_control_command != NULL)
>>>> (in eval_ext_lang_from_control_command) to dereference a null pointer.
>>>
>>> Hi Daniel,
>>>
>>> Thanks for the patch.  This is a simple patch so it doesn't need a
>>> copyright assignment from you.  However, if you intend to continue
>>> contributing to GDB, please e-mail me offlist and I can send you the
>>> papers to obtain the assignment.
>>>
>>> Just a few comments about formatting issues.
>>>
>>>>     2014-03-20    Daniel Gutson (daniel.gutson@tallertechnologies.com)
>>>>
>>>> gdb/
>>>>     * extension.c: (eval_ext_lang_from_control_command) Added check to
>>>> prevent dereference of null pointer.
>>>
>>> The ChangeLog format is wrong.  Take a look at the ChangeLog file for
>>> lots of examples, but basically you need to write:
>>>
>>> 2014-03-20  Your Name  <your@email>
>>>
>>>         * file.c (function): Added check to prevent blabla...
>>>
>>> Pay attention to the 2 spaces between the date, the name and the e-mail,
>>> and also to the TAB character indenting the description.
>>>
>>>> diff --git a/gdb/extension.c b/gdb/extension.c
>>>> index c2f502b..8357ee8 100644
>>>> --- a/gdb/extension.c
>>>> +++ b/gdb/extension.c
>>>> @@ -342,11 +342,14 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
>>>>      {
>>>>        if (extlang->cli_control_type == cmd->control_type)
>>>>       {
>>>> -       if (extlang->ops->eval_from_control_command != NULL)
>>>> -         {
>>>> -           extlang->ops->eval_from_control_command (extlang, cmd);
>>>> -           return;
>>>> -         }
>>>> +      if (extlang->ops != NULL)
>>>> +        {
>>>> +         if (extlang->ops->eval_from_control_command != NULL)
>>>> +           {
>>>> +             extlang->ops->eval_from_control_command (extlang, cmd);
>>>> +             return;
>>>> +           }
>>>> +        }
>>>
>>> You could simplify this by writing:
>>>
>>>   if (extlang->ops != NULL &&
>>> extlang->ops->eval_from_control_command != NULL)
>>>
>>> If you don't want to join the two "if"s, then you don't need to put the
>>> braces on the outter "if", because it has one single statement.
>>>
>>> The patch looks good to me, but I'm not a maintainer and can't approve
>>> it.
>>>
>>> Thanks,
>>>
>>> --
>>> Sergio
>>
>>
>> I did an audit of all the uses of ->ops and think this is the only
>> one I missed.
>> I will commit with the needed changes.
>> Thanks for the patch!

Here's what I committed.
I added a testcase.

2014-03-21  Daniel Gutson  <daniel.gutson@tallertechnologies.com>

	* extension.c (eval_ext_lang_from_control_command): Avoid dereferencing
	NULL pointer.

testsuite/

2014-03-22  Doug Evans  <xdje42@gmail.com>

	* gdb.python/python.exp (python not supported): Verify multi-line
	python command issues an error.
	* gdb.guile/guile.exp (guile not supported): Verify multi-line
	guile command issues an error.

diff --git a/gdb/extension.c b/gdb/extension.c
index c2f502b..1146cc7 100644
--- a/gdb/extension.c
+++ b/gdb/extension.c
@@ -342,7 +342,8 @@ eval_ext_lang_from_control_command (struct command_line *cmd)
     {
       if (extlang->cli_control_type == cmd->control_type)
 	{
-	  if (extlang->ops->eval_from_control_command != NULL)
+	  if (extlang->ops != NULL
+	      && extlang->ops->eval_from_control_command != NULL)
 	    {
 	      extlang->ops->eval_from_control_command (extlang, cmd);
 	      return;

diff --git a/gdb/testsuite/gdb.python/python.exp b/gdb/testsuite/gdb.python/python.exp
index a470427..5b8d34a 100644
--- a/gdb/testsuite/gdb.python/python.exp
+++ b/gdb/testsuite/gdb.python/python.exp
@@ -41,6 +41,13 @@ gdb_test_multiple "python print (23)" "verify python support" {
       gdb_test "source $srcdir/$subdir/source2.py" \
 	  "Error in sourced command file:.*" \
 	  "source source2.py when python disabled"
+
+      # Verify multi-line python commands cause an error.
+      gdb_py_test_multiple "multi-line python command" \
+	  "python" "" \
+	  "print (23)" "" \
+	  "end" "not supported.*"
+
       return -1
     }
     -re "$gdb_prompt $"	{}
diff --git a/gdb/testsuite/gdb.guile/guile.exp b/gdb/testsuite/gdb.guile/guile.exp
index 2a171fe..f300d72 100644
--- a/gdb/testsuite/gdb.guile/guile.exp
+++ b/gdb/testsuite/gdb.guile/guile.exp
@@ -38,6 +38,13 @@ gdb_test_multiple "guile (display 23) (newline)" "verify guile support" {
       gdb_test "source $srcdir/$subdir/source2.scm" \
 	  "Error in sourced command file:.*" \
 	  "source source2.scm when guile disabled"
+
+      # Verify multi-line guile commands cause an error.
+      gdb_test_multiline "multi-line guile command" \
+	  "guile" "" \
+	  "(print 23)" "" \
+	  "end" "not supported.*"
+
       return
     }
     -re "$gdb_prompt $"	{}