Thread

Re: pg_stash_advice: dump file with overlong stash name crashes worker in a restart loop

Robert Haas <robertmhaas@gmail.com> — 2026-05-29T17:25:45Z

On Sun, May 17, 2026 at 2:25 AM Ayush Tiwari
<ayushtiwari.slg01@gmail.com> wrote:
> While testing pg_stash_advice, I came across a
> crash that's reachable through the persistence dump file:

Hi,

Thanks for the report and patch. I found it to be more elaborate than
what I think we need, so I have committed a simplified version. While
it's good to fix stuff like this, the chances of anyone hitting in
practice are very low, since it requires the dump file to be
corrupted. Adding a test that could fail with a different value of
NAMEDATALEN seems likely to hurt more than it helps, and making the
test more elaborate to avoid that peril does not seem like the best
use of time and energy. I don't see this as something that needs to be
tested on every regression test run from now until forever.

-- 
Robert Haas
EDB: http://www.enterprisedb.com