On 10/06/2017 08:48 PM, Philippe Mathieu-Daudé wrote:
> Signed-off-by: Philippe Mathieu-Daudé <f4...@amsat.org>
> ---
> scripts/coccinelle/g_new.cocci | 46
> ++++++++++++++++++++++++++++++++++++++++++
> 1 file changed, 46 insertions(+)
>
> diff --git a/scripts/coccinelle/g_new.cocci b/scripts/coccinelle/g_new.cocci
> index 368f0c5651..9643ce3f23 100644
> --- a/scripts/coccinelle/g_new.cocci
> +++ b/scripts/coccinelle/g_new.cocci
> @@ -127,6 +127,52 @@ T **m;
> - *m = g_malloc0(sizeof(**m));
> + *m = g_new0(T *, 1);
>
> +////////////////////////////////////////
> +//
> +// no point in zeroing the allocation
> +//
> +
> +// the loop right afterwards fully initializes the memory
> +@@
> +type T;
> +identifier a, i;
> +expression c, e;
> +@@
> +(
> +-a = g_new0(T, c);
> + ... when != a, c
> ++a = g_new(T, c);
> + for (i = 0; i < c; i++) {
> + a[i] = e;
> + }
> +|
> +-a = g_try_new0(T, c);
> + ... when != a, c
> ++a = g_try_new(T, c);
> + for (i = 0; i < c; i++) {
> + a[i] = e;
> + }
> +)
> +
> +// compound literals
> +@@
> +type T;
> +identifier p;
> +expression s, e;
> +@@
> +(
> +- p = g_malloc0(s);
> ++ p = g_malloc(s);
> +|
> +- p = g_try_malloc0(s);
> ++ p = g_try_malloc(s);
> +|
> +- p = g_new0(T, 1);
> ++ p = g_new(T, 1);
> +)
> + ... when != p
> + *p = e;
It seems I misunderstood compound literals, I somehow believed this
would imply uninitialized fields would be zeroed.
So this transformation leads to incorrect code.
> +
> ////////////////////////////////////////
> //
> // last transformations: cleanups
>
