Re: Bug in return_append_str

[Antonio Ceballos Roa]

Mark,

Thanks. It seems quite evident. I will try to set up a scenario to catch it with valgrind though. Do you think it would be easy? I believe you deem it unnecessary, right?

Thanks,
Antonio

El 3 jul 2017, a las 9:09, Mark Hermeling <address@hidden> escribió:

Antonio,

That is certainly true once the modification that I proposed is applied. You need to allocate the space of the string, plus one character for the EOL. However, with the current code you allocate enough memory for the string and then move the pointer one character to the right. It is a simple typo, the brackets are placed incorrectly. 

Regards,

Mark

--

Sent from my phone, please excuse any typos. 

On Jul 3, 2017, at 00:14, Antonio Ceballos <address@hidden> wrote:

Hi Mark,

Thanks for finding this potential bug and for providing a fix. However, How would you reply to the comment that is written right before the line you mentioned (I am not the author):

    /* This doesn't have buffer overflow vulnerabilities, because

       we always allocate for enough space before appending. */

Thanks,

Antonio Ceballos

On Fri, Jun 30, 2017 at 2:20 PM, Mark Hermeling <address@hidden> wrote:

Hello,

There is a buffer overrun in return_append_str in src/frontend/lexpgn.cc at line

2224	newloc = (char *) malloc(strlen(s))+1;

The line should read:

newloc = (char *) malloc(strlen(s)+1);

We found this using static analysis using CodeSonar. I don’t have an actual path that will demonstrate this bug.

Regards,

Mark

—

Mark Hermeling | GrammaTech | Senior Director Product Marketing
mobile +1 (607) 351-5719 | www.grammatech.com

_______________________________________________
Bug-gnu-chess mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/bug-gnu-chess