Allegro game programming library

On 02 Oct 2001, Angelo Mottola <a.mottola@...> wrote:
> >Btw, in both modesel.c and bgfxapi.cpp, `realloc' is being used incorrectly.
> 
> What's wrong in bgfxapi.cpp? The realloc man page says:
> 
> void *realloc(void *ptr, size_t size)

>From bgfxapi.cpp:

mode_list->mode = (GFX_MODE *)realloc(mode_list->mode,
				sizeof(GFX_MODE) * (num_modes + 1));


>From modesel.c:

temp_mode_list = realloc(temp_mode_list, sizeof(MODE_LIST) *
			(driver_list_entry->mode_count));

If realloc fails (it returns NULL), you'll have a memory leak.
Sorry about being cryptic yesterday.

I'm sorry for the non-text messages, they leave as text, but the company
uses MS Exchange, and the sysop seems to be unable to have it understand
not to convert encoding (which oddly it seems to not always perform).

--
Vincent Penquerc'h

> >realloc has some cool features. for example if ptr is NULL it works 
> >equivalent to malloc and if size is 0 it works equivalent to free.
> >
> AFAIK this is not true for every compiler out there.
> 
> it is for gcc, and I think it is for msvc, but I'm not sure if it's
> in the ansi c specs.

This behavior is ANSI standard, but some libc implementations that
claim to be ANSI compliant fail to implement this correctly. IIRC,
some versions of SunOS are faulty for this. The biggest problem with
this IMHO is that relatively few people know how it works, and are
therefore likely to treat this as a bug.
Lua, for instance, does all its memory management through realloc,
but has a special version of it for non ANSI compliant implementations.

-- 
Vincent Penquerc'h

Peter Wang <tjaden@...> wrote:

> If realloc fails (it returns NULL), you'll have a memory leak.
> Sorry about being cryptic yesterday.

I read further through the docs, the GNU libc docs are a bit cryptic, but my
old "using turbo c" book says realloc frees memory and returns NULL if
re-allocation fails. The question is if we can rely on this.

BTW! if anyone have a copy of the ANSI/ISO C specs i would be very grateful
to recieve a copy.

Sincerely Henrik Stokseth.
-----------------------------------------------------------------------
E-mail: hstokset@...  Homepage: http://hstokset.n3.net

>> If realloc fails (it returns NULL), you'll have a memory leak.
>> Sorry about being cryptic yesterday.
>
>I read further through the docs, the GNU libc docs are a bit cryptic, but my
>old "using turbo c" book says realloc frees memory and returns NULL if
>re-allocation fails. The question is if we can rely on this.
>


FYI from the gcc info file:

 - Function: void * realloc (void *PTR, size_t NEWSIZE)
     The `realloc' function changes the size of the block whose address
     is PTR to be NEWSIZE.

     <snip>

     If you pass a null pointer for PTR, `realloc' behaves just like
     `malloc (NEWSIZE)'.  This can be convenient, but beware that older
                                                      ^^^^^^^^^^^^^^^^^
     implementations (before ISO C) may not support this behavior, and
     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
     will probably crash when `realloc' is passed a null pointer.

   Like `malloc', `realloc' may return a null pointer if no memory
space is available to make the block bigger.  When this happens, the
                                                                 ^^^
original block is untouched; it has not been modified or relocated.
^^^^^^^^^^^^^^^^^^^^^^^^^^^^

   <snip>




--
Martijn Versteegh

Martijn Versteegh <m.versteegh@...> wrote:

>    Like `malloc', `realloc' may return a null pointer if no memory
> space is available to make the block bigger.  When this happens, the
>                                                                  ^^^
> original block is untouched; it has not been modified or relocated.
> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

ok, i'm convinced. i'll fix up this and another memory leak i found.

Sincerely Henrik Stokseth.
-----------------------------------------------------------------------
E-mail: hstokset@...  Homepage: http://hstokset.n3.net