Re: COPY FROM crash

On Tue, 30 Jul 2024 at 15:52, Zhang Mingli <zmlpostgres@gmail.com> wrote:
> I have a test on Postgres and it has the similar issue(different places but same function).
>
> However it’s a little hard to reproduce because it happened when inserting next tuple after a previous copy multi
insertbuffer is flushed. 
>
> To reproduce easily, change the Macros to:
>
> #define MAX_BUFFERED_TUPLES 1
> #define MAX_PARTITION_BUFFERS 0

I think you're going to need to demonstrate to us there's an actual
PostgreSQL bug here with a test case that causes a crash without
changing the above definitions.

It seems to me that it's not valid to set MAX_PARTITION_BUFFERS to
anything less than 2 due to the code inside
CopyMultiInsertInfoFlush(). If we find the CopyMultiInsertBuffer for
'curr_rri' then that code would misbehave if the list only contained a
single CopyMultiInsertBuffer due to the expectation there's another
item in the list after the list_delete_first().  If you're only able
to get it to misbehave by setting MAX_PARTITION_BUFFERS to less than
2, then my suggested fix would be to add a comment to say that values
less than to are not supported.

David