Horton

Re: BUG #14808: V10-beta4, backend abort

Michael Paquier <michael.paquier@gmail.com>

View thread View on postgr.es
From: Michael Paquier <michael.paquier@gmail.com>
To: phb07@apra.asso.fr
Cc: PostgreSQL mailing lists <pgsql-bugs@postgresql.org>
Date: 2017-09-09T11:27:05Z
Lists: pgsql-bugs 
On Sat, Sep 9, 2017 at 3:48 PM,  <phb07@apra.asso.fr> wrote:
> INSERT 0 1101
> UPDATE 0
> server closed the connection unexpectedly
>         This probably means the server terminated abnormally
>         before or while processing the request.
> connection to server was lost

Crash is confirmed here so I am adding an open item. I have not dug
into the issue seriously, but at short glance we have a code path
trying to access something that has already been free'd:
#0  0x0000561bfe0f0959 in tuplestore_tuple_count
(state=0x7f7f7f7f7f7f7f7f) at tuplestore.c:548
548        return state->tuples;
(gdb) bt
#0  0x0000561bfe0f0959 in tuplestore_tuple_count
(state=0x7f7f7f7f7f7f7f7f) at tuplestore.c:548
#1  0x0000561bfdd8ca22 in SPI_register_trigger_data
(tdata=0x7ffc92083860) at spi.c:2764
#2  0x00007f7075da7156 in plpgsql_exec_trigger (func=0x561bfe7bc5a8,
trigdata=0x7ffc92083860) at pl_exec.c:692
#3  0x00007f7075da08e7 in plpgsql_call_handler (fcinfo=0x7ffc920833d0)
at pl_handler.c:24
(gdb) p state
$1 = (Tuplestorestate *) 0x7f7f7f7f7f7f7f7
-- 
Michael

Commits

  1. Fix SQL-spec incompatibilities in new transition table feature.

  2. Quick-hack fix for foreign key cascade vs triggers with transition tables.

  3. Fix transition tables for ON CONFLICT.

  4. Fix transition tables for wCTEs.

  5. Repair problems occurring when multiple RI updates have to be done to the same