[Geany-Devel] CodeAi Fixes a Null Pointer Dereference

[Matthew Brush]

On 2017-05-11 12:40 PM, Benjamin Bales wrote:
> Lex and Matthew,
>
> You seem to disagree on this issue.  Is this a valid issue?  And if so, do
> you like the fix?  I can batch a few of them (<= 10) as a single PR.  No
> problem.
>

If they're valid bugs (even minor), it certainly doesn't hurt to make a 
PR (or even an Issue just giving the info). They'll probably require 
more than an automated fix though, to account for formatting/indentation 
as well as looking at how it affects the code. Likely they'll need a 
human to evaluate what the correct fix is, for example the one given 
before is probably not very good as it just guards out code that was 
expected to run, rather than figuring out how the code was meant to work 
and make it do that.

Regards,
Matthew Brush

> -Ben
>
> On Wed, May 10, 2017 at 8:22 PM, Matthew Brush <mbrush at codebrainz.ca> wrote:
>
>> On 2017-05-10 04:09 PM, Lex Trotman wrote:
>>
>>> On 11 May 2017 at 08:10, Benjamin Bales <benjamin.bales at qbitlogic.com>
>>> wrote:
>>>
>>> CodeAi (https://github.com/C0deAi), an automated repair tool developed at
>>>> QbitLogic (www.qbitlogic.com), suggested the following fix. Could I
>>>> submit it as a patch if it looks alright?
>>>>
>>>> plugins/saveactions.c: “doc->file_type” pointer might be dereferenced
>>>> when
>>>> null on line 283.  Initialization may be provided by “doc” passed in as a
>>>> function argument, but a null check would be prudent just in case. The
>>>> fix
>>>> checks “doc->file_type” for null before allowing a dereference on the
>>>> following line.  A snapshot of the bug report generated by CodeAi is
>>>> attached.  A full report is available upon request.
>>>>
>>>>
>>> This function is called (via the signal framework) by the function that
>>> created `doc` and as such cannot be null.  The design of the application
>>> uses the signal framework to decouple caller and callee and this is likely
>>> to confuse your tool since it cannot see where functions are called.
>>> Whilst any contributions are welcome, a report with a lot of similar false
>>> positives may end up being ignored and be a bad advertisement for your
>>> tool.
>>>
>>>
>> Naw, I think it's technically a real bug, albeit very minor. It's the
>> `file_type` member of the `doc` that can be NULL. IIUC tools like this look
>> to see if you checked the NULL-ness of something and then proceed to
>> dereference it outside of that check later, which this code does (checks if
>> `ft == NULL` several lines up and then unconditionally dereferences it on
>> the line given by the OP).
>>
>> Regards,
>> Matthew Brush
>>
>>
>> _______________________________________________
>> Devel mailing list
>> Devel at lists.geany.org
>> https://lists.geany.org/cgi-bin/mailman/listinfo/devel
>>
>
>
>
>
>
> _______________________________________________
> Devel mailing list
> Devel at lists.geany.org
> https://lists.geany.org/cgi-bin/mailman/listinfo/devel
>