25.07.2024 20:07, Alvaro Herrera wrote:
> Maybe for sanity (and perhaps for Svace compliance) we could do it the
> other way around, i.e. by testing events->tail for nullness instead of
> events->head, then add the assertion:
>
> if (events->tail == NULL)
> {
> Assert(events->head == NULL);
> events->head = chunk;
> }
> else
> events->tail->next = chunk;
>
> This way, it's not wholly redundant.
Thanks for your response!
I agree with the proposed changes and have updated the patch accordingly. Version 2 is attached.
> That said, I'm not sure we actually *need* to change this.
I understand and partly agree. But it appears that with these changes, the dereference of a null pointer is impossible even in builds where assertions are disabled. Previously, this issue could theoretically occur. Consequently, these changes slightly enhance overall security.
--
Best regards,
Alexander Kuznetsov