public inbox for [email protected]
help / color / mirror / Atom feedFrom: Michael Paquier <[email protected]>
To: Postgres hackers <[email protected]>
Subject: NULL pointer dereference in syslogger with load_libraries() and -DEXEC_BACKEND at startup
Date: Mon, 25 May 2026 16:45:41 +0900
Message-ID: <[email protected]> (raw)
Hi all,
While doing some tests today for a different patch, I had the surprise
to see the following failure on HEAD (not on REL_18_STABLE or older)
while loading one of the test modules (I was playing with test_slru,
one of the custom pgstats module fails equally). Reproducing this
problem requires the following setup:
- Module loaded by shared_preload_libraries
- logging_collector active, which we don't do in most of the tests run
by the buildfarm.
- -DEXEC_BACKEND.
- Use at least log_min_messages = debug1 to make load_libraries
generate at least one message
Then one is greeted by:
==465718==Using libbacktrace symbolizer. syslogger.c:1118:7: runtime
error: null pointer passed as argument 4, which is declared to never
be null
#0 0x0000012b9331 in write_syslogger_file /home/ioltas/git/postgres/src/backend/postmaster/syslogger.c:1118
#1 0x000001aa353a in send_message_to_server_log /home/ioltas/git/postgres/src/backend/utils/error/elog.c:3889
#2 0x000001a9a3ce in EmitErrorReport /home/ioltas/git/postgres/src/backend/utils/error/elog.c:1924
#3 0x000001a93b8b in errfinish /home/ioltas/git/postgres/src/backend/utils/error/elog.c:555
#4 0x0000015ed3d7 in pgstat_register_kind /home/ioltas/git/postgres/src/backend/utils/activity/pgstat.c:1574
#5 0x7ffff7fb3bcf in _PG_init /home/ioltas/git/postgres/src/test/modules/test_custom_stats/test_custom_fixed_stats.c:80
#6 0x000001aa6fd5 in internal_load_library /home/ioltas/git/postgres/src/backend/utils/fmgr/dfmgr.c:299
#7 0x000001aa653e in load_file /home/ioltas/git/postgres/src/backend/utils/fmgr/dfmgr.c:161
#8 0x000001acbb15 in load_libraries /home/ioltas/git/postgres/src/backend/utils/init/miscinit.c:1838
#9 0x000001acbcb8 in process_shared_preload_libraries /home/ioltas/git/postgres/src/backend/utils/init/miscinit.c:1856
#10 0x0000012a907b in SubPostmasterMain /home/ioltas/git/postgres/src/backend/postmaster/launch_backend.c:678
Looking at a backtrace:
#1 0x00000000012b934d in write_syslogger_file ( buffer=0x2fba5d0
"2026-05-25 16:21:58.516 JST syslogger[465718]: [1-1]
db=,user=,app=,client= LOG: registered custom cumulative statistics
\"test_custom_fixed_stats\" with ID 26\n", count=159, destination=1)
at syslogger.c:1118 1118 rc = fwrite(buffer, 1, count, logfile);
(gdb) p logfile
$1 = (FILE *) 0x0
It is pretty clear that we have the idea to log an entry through the
syslogger but it is too early to do so and its file is not opened, so
my bet is that something in the startup logic has changed.
I did not take the cycles necessary for a bisect, but it looks like
this has been around for a few months at least. I have pinged
f3c9e341cdf1 as a safe startup point for now, so that's a 2026 issue.
Does this ring a bell to somebody?
--
Michael
Attachments:
[application/pgp-signature] signature.asc (833B, 2-signature.asc)
download
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected]
Subject: Re: NULL pointer dereference in syslogger with load_libraries() and -DEXEC_BACKEND at startup
In-Reply-To: <[email protected]>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox