]> git.ipfire.org Git - thirdparty/postgresql.git/commit
Use standard die() handler for SIGTERM in bgworkers
authorHeikki Linnakangas <heikki.linnakangas@iki.fi>
Wed, 18 Feb 2026 17:59:34 +0000 (19:59 +0200)
committerHeikki Linnakangas <heikki.linnakangas@iki.fi>
Wed, 18 Feb 2026 17:59:34 +0000 (19:59 +0200)
commitd62dca3b297413a11a594c9675f2fb22e931d01b
treecb54a056ddacdb1277fab068dd0f53682b1b607f
parent3894f08abe490d31518d038b0a8eaa46ce971b03
Use standard die() handler for SIGTERM in bgworkers

The previous default bgworker_die() signal would exit with elog(FATAL)
directly from the signal handler. That could cause deadlocks or
crashes if the signal handler runs while we're e.g holding a spinlock
or in the middle of a memory allocation.

All the built-in background workers overrode that to use the normal
die() handler and CHECK_FOR_INTERRUPTS(). Let's make that the default
for all background workers. Some extensions relying on the old
behavior might need to adapt, but the new default is much safer and is
the right thing to do for most background workers.

Reviewed-by: Nathan Bossart <nathandbossart@gmail.com>
Reviewed-by: Kirill Reshke <reshkekirill@gmail.com>
Discussion: https://www.postgresql.org/message-id/5238fe45-e486-4c62-a7f3-c7d8d416e812@iki.fi
doc/src/sgml/bgworker.sgml
src/backend/access/transam/parallel.c
src/backend/postmaster/bgworker.c
src/backend/replication/logical/applyparallelworker.c
src/backend/replication/logical/launcher.c
src/backend/replication/logical/worker.c
src/test/modules/test_shm_mq/worker.c