SOURCES (rpm-4_4_9): rpm-lualeak.patch - more _free() compatible, ...

[Jeff Johnson]

On Mar 17, 2008, at 5:44 PM, Tomasz Wittner wrote:

>
> IMO construction like `if(foo) free(foo);' has little value (maybe  
> is useful
> for performance reasons - conditional instruction vs. calling  
> function - I
> don't know). I didn't write any piece of C code for ages, but it  
> seems to me
> that you would like achieve rather sth like:
> if(_lua_init) { free(_lua_init); _lua_init = NULL; }
>

Heh, Real Men use reference counted garbage collected slab memory  
memory with
a SAT solving loop breaker.

Meanwhile, you're looking at code out-of-context.

The snippet that is extracted came from rpm code that looks like
	ptr = _free(ptr);

Whether free(NULL) is permitted (or not) is hardly the issue. The
stoopid _free() wrapper immediately poisons (by setting to NULL)
the pointer that used to reference malloc'd memory, thereby preventing
any possibility of accidentally referencing already free'd memory.

Those bugs are _REALLY_ hard to find, segfault's from looking through
a NULL pointer are easy pie fixing in comparison.

73 de Jeff