> On Jan 9, 2023, at 11:34 AM, Andres Freund <andres@anarazel.de> wrote:
>
> 1) Because ctx->next_xid is set after the XidFromFullTransactionId() call in
> update_cached_xid_range(), we end up using the xid 0 (or an outdated value in
> subsequent calls) to determine whether epoch needs to be reduced.
Can you say a bit more about your analysis here, preferably with pointers to the lines of code you are analyzing? Does
theproblem exist in amcheck as currently committed, or are you thinking about a problem that arises only after applying
yourpatch? I'm a bit fuzzy on where xid 0 gets used.
Thanks
—
Mark Dilger
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company