pgsql: Fix memory leak in plpgsql's CALL processing.
Fix memory leak in plpgsql's CALL processing.
When executing a CALL or DO in a non-atomic context (i.e., not inside
a function or query), plpgsql creates a new plan each time through,
as a rather hacky solution to some resource management issues. But
it failed to free this plan until exit of the current procedure or DO
block, resulting in serious memory bloat in procedures that called
other procedures many times. Fix by remembering to free the plan,
and by being more honest about restoring the previous state (otherwise,
recursive procedure calls have a problem).
There was also a smaller leak associated with recalculation of the
"target" list of output variables. Fix that by using the statement-
lifespan context to hold non-permanent values.
Back-patch to v11 where procedures were introduced.
Pavel Stehule and Tom Lane
Discussion: /messages/by-id/CAFj8pRDiiU1dqym+_P4_GuTWm76knJu7z9opWayBJTC0nQGUUA@mail.gmail.com
Branch
------
master
Details
-------
https://git.postgresql.org/pg/commitdiff/a6b1f5365d58356b5d42829e9cd89a6c725d7a0a
Modified Files
--------------
src/pl/plpgsql/src/pl_exec.c | 91 ++++++++++++++++++++++++++++++++++----------
1 file changed, 70 insertions(+), 21 deletions(-)
This change has broken the following test case (reduced from actual
production code):
CREATE OR REPLACE PROCEDURE prc_inout_test_inner(INOUT x integer)
LANGUAGE plpgsql
AS
$procedure$
BEGIN
COMMIT;
x := 2;
END;
$procedure$;
CREATE OR REPLACE PROCEDURE prc_inout_test_main()
LANGUAGE plpgsql
AS
$procedure$
DECLARE
y INTEGER := 1;
BEGIN
CALL prc_inout_test_inner(y);
END;
$procedure$;
CALL prc_inout_test_main();
With the patch, it fails like this:
ERROR: XX000: unsupported target type: 0
CONTEXT: PL/pgSQL function prc_inout_test_main() line 5 at CALL
LOCATION: exec_move_row_from_fields, pl_exec.c:7196
Show quoted text
On 2020-09-29 17:18, Tom Lane wrote:
Fix memory leak in plpgsql's CALL processing.
When executing a CALL or DO in a non-atomic context (i.e., not inside
a function or query), plpgsql creates a new plan each time through,
as a rather hacky solution to some resource management issues. But
it failed to free this plan until exit of the current procedure or DO
block, resulting in serious memory bloat in procedures that called
other procedures many times. Fix by remembering to free the plan,
and by being more honest about restoring the previous state (otherwise,
recursive procedure calls have a problem).There was also a smaller leak associated with recalculation of the
"target" list of output variables. Fix that by using the statement-
lifespan context to hold non-permanent values.Back-patch to v11 where procedures were introduced.
Pavel Stehule and Tom Lane
Discussion: /messages/by-id/CAFj8pRDiiU1dqym+_P4_GuTWm76knJu7z9opWayBJTC0nQGUUA@mail.gmail.com
Branch
------
masterDetails
-------
https://git.postgresql.org/pg/commitdiff/a6b1f5365d58356b5d42829e9cd89a6c725d7a0aModified Files
--------------
src/pl/plpgsql/src/pl_exec.c | 91 ++++++++++++++++++++++++++++++++++----------
1 file changed, 70 insertions(+), 21 deletions(-)
Peter Eisentraut <peter.eisentraut@enterprisedb.com> writes:
This change has broken the following test case (reduced from actual
production code):
Thanks for the report. What I see here is
ERROR: unsupported target type: 2139062143
so it looks like a memory clobber --- probably some data structure is
being built in the wrong memory context. Will look closer shortly.
regards, tom lane