Re: Snapshot warning - Mailing list pgsql-hackers

From Alvaro Herrera
Subject Re: Snapshot warning
Date
Msg-id 20081124135213.GD3861@alvh.no-ip.org
Whole thread Raw
In response to Snapshot warning  ("Pavan Deolasee" <pavan.deolasee@gmail.com>)
Responses Re: Snapshot warning  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
Pavan Deolasee escribió:
> Following test case gives a warning of snapshot not destroyed at commit
> time.
> 
> CREATE TABLE test (a int);
> INSERT INTO test VALUES (1);
> BEGIN;
> DECLARE c CURSOR FOR SELECT * FROM test FOR update;
> SAVEPOINT A;
> FETCH -2 FROM c;
> ROLLBACK TO SAVEPOINT A;
> COMMIT;
> 
> Should we call FreeQueryDesc() even for failed portals in PortalCleanup() ?
> Or PortalDrop() is a better(right) place to do that ?

That doesn't work; doing it causes a crash:

TRAP: FailedAssertion(«!(qdesc->estate == ((void *)0))», Archivo: «/pgsql/source/00head/src/backend/tcop/pquery.c»,
Línea:126)
 

which is here:

void
FreeQueryDesc(QueryDesc *qdesc)
{   /* Can't be a live query */   Assert(qdesc->estate == NULL);

BTW I noticed that AtSubAbort_Portals() is neglecting to set
portal->cleanup to NULL after calling it.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support


pgsql-hackers by date:

Previous
From: "Dave Page"
Date:
Subject: Re: blatantly a bug in the documentation
Next
From: Andrew Dunstan
Date:
Subject: Re: blatantly a bug in the documentation