Re: [PATCH 2/4] libm/stdlib: don't read past source in nano_realloc

[Corinna Vinschen <vinschen@redhat.com>]

On Aug 11 16:05, Keith Packard via Newlib wrote:
> Save the computed block size and use it to avoid reading past
> the end of the source block.
> 
> Signed-off-by: Keith Packard <keithp@keithp.com>
> ---
>  newlib/libc/stdlib/nano-mallocr.c | 6 +++++-
>  1 file changed, 5 insertions(+), 1 deletion(-)
> 
> diff --git a/newlib/libc/stdlib/nano-mallocr.c b/newlib/libc/stdlib/nano-mallocr.c
> index 04465eb9e..cef23977e 100644
> --- a/newlib/libc/stdlib/nano-mallocr.c
> +++ b/newlib/libc/stdlib/nano-mallocr.c
> @@ -466,6 +466,7 @@ void * nano_realloc(RARG void * ptr, malloc_size_t size)
>  {
>      void * mem;
>      chunk * p_to_realloc;
> +    malloc_size_t old_size;
>  
>      if (ptr == NULL) return nano_malloc(RCALL size);
>  
> @@ -477,12 +478,15 @@ void * nano_realloc(RARG void * ptr, malloc_size_t size)
>  
>      /* TODO: There is chance to shrink the chunk if newly requested
>       * size is much small */
> -    if (nano_malloc_usable_size(RCALL ptr) >= size)
> +    old_size = nano_malloc_usable_size(RCALL ptr);
> +    if (old_size >= size)
>        return ptr;

So, after this statement, we can be sure that size > old_size, right?

>      mem = nano_malloc(RCALL size);
>      if (mem != NULL)
>      {
> +	if (size > old_size)

...which makes this condition useless.

> +	    size = old_size;
>          memcpy(mem, ptr, size);

why not just

           memcpy(mem, ptr, old_size);

instead?

>          nano_free(RCALL ptr);
>      }
> -- 
> 2.28.0

Thanks,
Corinna


-- 
Corinna Vinschen
Cygwin Maintainer
Red Hat