Re: dnscurve updates

The standard C malloc is supposed to return a pointer to memory
suitably aligned for anything. What else is needed ?


On 1/7/10, David Nicol <[email protected]> wrote:
> after looking at
> http://www.opengroup.org/onlinepubs/000095399/functions/posix_memalign.html
>
> one wonders, esp. since its only needed for command line tools, if the
> following
> wouldn't do the trick by simply ignoring the alignment request:
>
> #ifndef HAVE_POSIX_MEMALIGN /*or equivalent*/
> /* int posix_memalign(void **memptr, size_t alignment, size_t size); */
> #define posix_memalign(mp,a,s) (((*mp = malloc(s)) == 0 ) ? errno : 0)
> #endif
>
> What alternatives (to ignoring alignment) exist? malloc() until you
> get one that is aligned right? The requirement that posix_memalign's
> memory can go to free() rules out wrappers that get too much and then
> return the first aligned address within.
>