xboard-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [XBoard-devel] Buffer overflow


From: h.g. muller
Subject: Re: [XBoard-devel] Buffer overflow
Date: Fri, 23 Dec 2011 00:38:51 +0100

At 20:22 22-12-2011 +0100, Byrial Jensen wrote:
I think that the function PGNTagsStatic() in pgntags.c is unsafe. The problem is that gameInfo->extraTags which can have any size is copied without size control to a buffer which can overflow as the result. This is rather serious as it is easy to construct a pgn file that can cause the overflow.

You cannot have something of possibly unlimited size in a static buffer, so I suggest that PGNTagsStatic is removed. PGNTags() can then allocate a big eneogh buffer and do the work itself. PrintPGNTags() doesn't need a buffer, but can print directly to the file.

I think I was responsible for this, because the problem was that the original malloc'ed code caused a devastating memory leak in WinBoard on my Win XP laptop. Although I could not find anything wrong with the code; it seemedthe free() was just not working. But it led to WB crashing when you loaded a couple of large PGN files, (~40k games a piece), because it could no longer allocate memory. So I made it static based on the philosophy that it is better to have a hypothetical problem that never ocurs in practice, than having a certain crash on commonly done tasks...


The "guilty" function is strcat() which is used in many places, so it is probably a good idea to check all uses of strcat.

However there is no reason to replace all occurences of strcat with something else like it is done with strcpy(). I find it a little silly to see calls to safeStrCpy instead of strcpy when you have just allocated a new buffer of the required size, so strcpy would be perfectly safe to use.

True. This whole safeStrCpy business is a folly IMO, and it has already caused a lot of grief. But it seems we cannot build for some distributions when we use strcpy, because it is a blacklisted function. I already proposed to subvert this ridiculous demand by simply ading a "#define strcpy myStrCpy" in common.h, and providing our own version of strcpy not detected by these compilers, but Arun wouldn't have it!






reply via email to

[Prev in Thread] Current Thread [Next in Thread]