From: | Jeevan Ladhe <jeevan(dot)ladhe(at)enterprisedb(dot)com> |
---|---|
To: | PostgreSQL Developers <pgsql-hackers(at)postgresql(dot)org> |
Subject: | Fix error in ECPG while connection handling |
Date: | 2018-03-12 08:07:21 |
Message-ID: | CAOgcT0Pnh8UUGYke16At+Fy77nq2Wb6aZFzhJ8nzONKLuO=EYw@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi,
I came across following error while working on ecpg client program.
$ install/bin/ecpg ecpg_connection_ptr.pgc
ecpg_connection_ptr.pgc:26: ERROR: AT option not allowed in WHENEVER
statement
I have attached simple ecpg program 'ecpg_connection_ptr_issue.pgc' that
reproduces the above issue.
After doing some investigation I could see that, in preproc.y, in rule of
'ClosePortalStmt', function output_statement() is called, which in turn
frees
the connection pointer. Here is the relevant trailing snippet from the
output_statement() function.
whenever_action(whenever_mode | 2);
free(stmt);
if (connection != NULL)
free(connection);
}
Now, when the ecpg parses following statement using rule 'ECPGWhenever':
EXEC SQL WHENEVER SQLERROR CONTINUE;
it checks if the connection is set, if it is then throws an error as below:
if (connection)
mmerror(PARSE_ERROR, ET_ERROR, "AT option not allowed in WHENEVER
statement");
Now, ideally the connection would have been null here, but, as the
'ClosePortalStmt'
rule freed the connection but did not set it to NULL, it still sees that
there
is a connection(which is actually having garbage in it) and throws an error.
I see similarly there are other places, which are freeing this global
connection
but not setting to NULL, and all those should be fixed. I have attached a
patch
ecpg_connection_ptr_issue_fix.patch to fix these places.
Regards,
Jeevan Ladhe
Attachment | Content-Type | Size |
---|---|---|
ecpg_connection_ptr_issue_fix.patch | application/octet-stream | 725 bytes |
ecpg_connection_ptr_issue.pgc | application/octet-stream | 619 bytes |
From | Date | Subject | |
---|---|---|---|
Next Message | Anthony Bykov | 2018-03-12 08:08:21 | Re: Transform for pl/perl |
Previous Message | Etsuro Fujita | 2018-03-12 07:55:03 | Re: [HACKERS] Another oddity in handling of WCO constraints in postgres_fdw |