Re: BUG #18486: Is there something wrong with the calculation in ReorderBufferChangeSize()?

On Thu, May 30, 2024 at 2:30 AM PG Bug reporting form
<noreply(at)postgresql(dot)org> wrote:
>
> The following bug has been logged on the website:
>
> Bug reference: 18486
> Logged by: Xingwang Xu
> Email address: xu(dot)xw2008(at)163(dot)com
> PostgreSQL version: 17beta1
> Operating system: CentOS7.9
> Description:
>
> In the code related to logical replication, there is a function
> ReorderBufferChangeSize(), which is used to calculate the size of a change
> in memory.
>
> When looking at the ReorderBufferChangeSize() function, I saw the following
> code：
>
> case REORDER_BUFFER_CHANGE_MESSAGE:
> {
> Size prefix_size = strlen(change->data.msg.prefix) + 1;
>
> sz += prefix_size + change->data.msg.message_size +
> sizeof(Size) + sizeof(Size);
>
> break;
> }
>
> When calculating the change size of the message type, there are two
> “sizeof(Size)” in the code. It is not clear why these two “sizeof(Size)” are
> added and whether these two “sizeof(Size)” are redundant.
>

These two sizeof(Size) are added as while serializing or restoring
message change, we explicitly serialize/restore the size of the prefix
and the actual message. See
ReorderBufferSerializeChange()/ReorderBufferRestoreChange().

--
With Regards,
Amit Kapila.