Re: [BUG] CRASH: ECPGprepared_statement() and ECPGdeallocate_all() when connection is NULL

Andrew Dunstan <andrew@dunslane.net>

From: Andrew Dunstan <andrew@dunslane.net>

To: Tom Lane <tgl@sss.pgh.pa.us>

Cc: Alexander Lakhin <exclusion@gmail.com>, Nishant Sharma <nishant.sharma@enterprisedb.com>, Shruthi Gowda <gowdashru@gmail.com>, Mahendra Singh Thalor <mahi6run@gmail.com>, Fujii Masao <masao.fujii@gmail.com>, PostgreSQL Development <pgsql-hackers@postgresql.org>

Date: 2026-05-06T15:50:04Z

Lists: pgsql-hackers

On 2026-05-06 We 10:58 AM, Tom Lane wrote:
> Andrew Dunstan <andrew@dunslane.net> writes:
>> Attached is a patch with the fix, courtesy of claude. It's a slight
>> behaviour change:
> Yeah.  So we could either do something like this, or say that the
> test case is buggy and needs to provide its own mutexes, per the
> existing comment
>
> -         * if no connection in TSD for this thread, get the global default
> -         * connection and hope the user knows what they're doing (i.e. using
> -         * their own mutex to protect that connection from concurrent accesses
>
> On the whole I think I favor the behavior change.  We might get some
> complaints, but it just seems a lot safer to redefine it like this.


I agree.


>
> Either way, it seems like some documentation adjustments are called
> for.
>
> As far as the patch itself goes, I'd be inclined to pull the
> preparatory step
>
>           ecpg_pthreads_init();    /* ensure actual_connection_key is valid */
>
> into the new ecpg_default_connection() subroutine, especially since
> its proposed comment doesn't mention that prerequisite.
>
> 			


Right.


Given the lack of field complaints, maybe we should wait until after 
next week's releases?


cheers


andrew


--
Andrew Dunstan
EDB: https://www.enterprisedb.com