Horton

Re: Fixing memory leaks in postgres_fdw

Matheus Alcantara <matheusssilv97@gmail.com>

View thread View on postgr.es
From: Matheus Alcantara <matheusssilv97@gmail.com>
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Etsuro Fujita <etsuro.fujita@gmail.com>, pgsql-hackers@lists.postgresql.org
Date: 2025-05-29T13:07:31Z
Lists: pgsql-hackers 
On 28/05/25 17:56, Tom Lane wrote:
> I wrote:
>> Having said that, the idea that this sequence is OOM-safe is pretty
>> silly anyway, considering that createNewConnection does a pstrdup,
>> and creates a new hashtable entry which might require enlarging the
>> hashtable, and for that matter might even create the hashtable.
>> So maybe rather than continuing to adhere to a half-baked coding
>> rule, we need to think of some other way to do that.
>
> Here's an attempt at fixing this properly.  I'm posting it as a
> standalone patch because I now think this part might be worth
> back-patching.  The odds of an OOM at just the wrong time aren't
> high, but losing track of an open connection seems pretty bad.
>

The v5-0001 makes sense to me. I think that now it follows a similar
approach with postgres_fdw that first create the hash entry and them set
the connection on the hash entry.

I've also reviewed the remaining patches, v4-0002, v4-0003 and v4-0004
and it all looks reasonable to me. +1 for going forward with these
patches.

The only point is that I've just tried to apply the v5-0001 on top of
the previous v4-000X patches and is raising an error:

❯❯❯ git am v5-0001-Avoid-resource-leaks-when-a-dblink-connection-fai.patch
Applying: Avoid resource leaks when a dblink connection fails.
error: patch failed: contrib/dblink/dblink.c:105
error: contrib/dblink/dblink.c: patch does not apply
Patch failed at 0001 Avoid resource leaks when a dblink connection fails.

-- 
Matheus Alcantara

Commits

  1. Silence leakage complaint about postgres_fdw's InitPgFdwOptions.

  2. Run pgindent on the changes of the previous patch.

  3. Reap the benefits of not having to avoid leaking PGresults.

  4. Create infrastructure to reliably prevent leakage of PGresults.

  5. Fix memory leakage in postgres_fdw's DirectModify code path.

  6. Avoid resource leaks when a dblink connection fails.