Re: Fix some resources leaks (src/bin/pg_basebackup/pg_createsubscriber.c)

Euler Taveira <euler@eulerto.com>

From: "Euler Taveira" <euler@eulerto.com>
To: "ranier.vf@gmail.com" <ranier.vf@gmail.com>, pgsql-hackers <pgsql-hackers@postgresql.org>
Date: 2024-03-28T02:07:44Z
Lists: pgsql-hackers
On Wed, Mar 27, 2024, at 8:50 PM, Ranier Vilela wrote:
> Coverity has some reports in the new code
> pg_createsubscriber.c
> I think that Coverity is right.

It depends on your "right" definition. If your program execution is ephemeral
and the leak is just before exiting, do you think it's worth it?

> 1.
> CID 1542682: (#1 of 1): Resource leak (RESOURCE_LEAK)
> leaked_storage: Variable buf going out of scope leaks the storage it points to.

It will exit in the next instruction.

> 2.
> CID 1542704: (#1 of 1): Resource leak (RESOURCE_LEAK)
> leaked_storage: Variable conn going out of scope leaks the storage it points to.

The connect_database function whose exit_on_error is false is used in 2 routines:

* cleanup_objects_atexit: that's about to exit;
* drop_primary_replication_slot: that will execute a few routines before exiting.

> 3.
> CID 1542691: (#1 of 1): Resource leak (RESOURCE_LEAK)
> leaked_storage: Variable str going out of scope leaks the storage it points to.

It will exit in the next instruction.

Having said that, applying this patch is just a matter of style.


--
Euler Taveira
EDB   https://www.enterprisedb.com/

Commits

  1. Fix assorted resource leaks in new pg_createsubscriber code.