Re: [E-devel] wrong usage of realloc

[Mike B. <mi...@ze...>]

On Mon, 25 Oct 2010 10:13:26 -0200
Lucas De Marchi <luc...@pr...> wrote:

> Hi all,
> 
> Find attached a file with a report of wrong usage of realloc.
> Basically, if you do "a = realloc(a, ...)" and realloc fails, you lost
> the previous pointer and then you cannot free that var anymore. There
> can be some false positives, in case you previously assigned the
> pointer to a tmp var instead of using the tmp var in the call to
> realloc.
> 
> 
> Note: this was obtained by realloc-spot-lost-pointer.cocci script in
> svn. It uses the syntax of a patch, but the "-" in the beginning of a
> line is there only to spot the callers... it's not a "-" as usually
> intended in patches.
> 
> 
> regards,
> 
> Lucas De Marchi
I feel like we went through something similar previously...

-- 
Mike Blumenkrantz
Zentific: Our boolean values are huge.