Horton

Re: [PATCH] Add additional extended protocol commands to psql: \parse and \bind

Anthonin Bonnefoy <anthonin.bonnefoy@datadoghq.com>

View thread View on postgr.es
From: Anthonin Bonnefoy <anthonin.bonnefoy@datadoghq.com>
To: Alexander Lakhin <exclusion@gmail.com>
Cc: Michael Paquier <michael@paquier.xyz>, Jelte Fennema-Nio <postgres@jeltef.nl>, Peter Eisentraut <peter@eisentraut.org>, Tomas Vondra <tomas.vondra@enterprisedb.com>, vignesh C <vignesh21@gmail.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>
Date: 2024-09-18T07:42:43Z
Lists: pgsql-hackers

Attachments

On Tue, Sep 17, 2024 at 5:00 PM Alexander Lakhin <exclusion@gmail.com> wrote:
>
> Please look at an assertion failure, caused by \bind_named:
> regression=# SELECT $1 \parse s
> \bind_named s
>
> regression=# \bind_named
> \bind_named: missing required argument
> regression=# 1 \g
> psql: common.c:1501: ExecQueryAndProcessResults: Assertion `pset.stmtName != ((void *)0)' failed.

Thanks for the report.

Looking at the failure, it seems like the issue was already present
with \bind, though there was no assertion failure: repeatedly calling
\bind would allocate new stmtName/bind_params and leak them at the
start of exec_command_bind.

I've joined a patch to clean the psql extended state at the start of
every extended protocol backslash command, freeing the allocated
variables and resetting the send_mode. Another possible approach would
be to return an error when there's already an existing state instead
of overwriting it.

Commits

  1. psql: Rename meta-command \close to \close_prepared

  2. psql: Add tests for repeated calls of \bind[_named]

  3. psql: Fix memory leak with repeated calls of \bind

  4. psql: Clean up more aggressively state of \bind[_named], \parse and \close

  5. psql: Add more meta-commands able to use the extended protocol

  6. psql: Add ignore_slash_options in bind's inactive branch