Prev: MonetDB test says that PostgreSQL often has errorsor missing results
Next: [HACKERS] Serializable implementation milestone: table SIREAD locks without correct lifespan
From: Andrew Dunstan on 20 Jan 2010 17:07

It seems like Custom GUCs are still in need of some work, as shown in my
recent email. In particular, they are not transaction safe - if a
transaction attempts to do DefineCustomFooVariable() and that
transaction aborts, the placeholder setting that it used is already gone
by the time it tries to roll back GUC settings. I think this code at the
end of define_custom_variable()

/*
* Free up as much as we conveniently can of the placeholder
structure
* (this neglects any stack items...)
*/
set_string_field(pHolder, pHolder->variable, NULL);
set_string_field(pHolder, &pHolder->reset_val, NULL);

free(pHolder);


needs to be removed and instead we need to save pHolder in a list along
with the GUC level, to be processed later by AtEOXact_GUC(), which would
do the right thing according to whether or not it had a commit or an abort.

I want to get this fixed before we consider custom settings for plperl
that have possible security implications.

Thoughts?

cheers

andrew



--
Sent via pgsql-hackers mailing list (pgsql-hackers(a)postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

 | 
Pages: 1
Prev: MonetDB test says that PostgreSQL often has errorsor missing results
Next: [HACKERS] Serializable implementation milestone: table SIREAD locks without correct lifespan