Thread: BUG #16340: lo_manage() crashs DB instance

The following bug has been logged on the website:

Bug reference:      16340
Logged by:          William Crowell
Email address:      william.crowell@openlogic.com
PostgreSQL version: 11.7
Operating system:   RHEL 7.7
Description:

PostgreSQL crashes if you do the following:

$psql postgres -c 'create extension lo'
CREATE EXTENSION
$psql postgres -c 'select lo_manage()' server closed the connection
unexpectedly
       This probably means the server terminated abnormally
       before or while processing the request.
connection to server was lost

...
2020-04-02 13:42:23 CEST / postgres / xxxx / 1043 / 00000 / [local]] - LOG:
AUDIT: SESSION,1,2,FUNCTION,EXECUTE,FUNCTION,public.lo_manage,select
lo_manage(),<none>
[2020-04-02 13:42:23 CEST /  /  / 24618 / 00000 / ] - LOG:  server process
(PID 1043) was terminated by signal 11: Segmentation fault
[2020-04-02 13:42:23 CEST /  /  / 24618 / 00000 / ] - DETAIL:  Failed
process was running: select lo_manage()
[2020-04-02 13:42:23 CEST /  /  / 24618 / 00000 / ] - LOG:  terminating any
other active server processes

I do know that lo_manage requires an argument, but it should not create a
segmentation fault even though the argument is not provided.

PG Bug reporting form <noreply@postgresql.org> writes:
> PostgreSQL crashes if you do the following:
> $psql postgres -c 'create extension lo'
> CREATE EXTENSION
> $psql postgres -c 'select lo_manage()'
> server closed the connection unexpectedly

Huh, somebody was a bit too cute here:

    if (!CALLED_AS_TRIGGER(fcinfo)) /* internal error */
        elog(ERROR, "%s: not fired by trigger manager",
             trigdata->tg_trigger->tgname);

Will fix, thanks for the report!

            regards, tom lane

Tim,

No problem.  Do you happen to know what release this will be fixed in? 
Maybe 11.8? 

Regards,

Bill Crowell



--
Sent from: https://www.postgresql-archive.org/PostgreSQL-bugs-f2117394.html