regression in CVS HEAD

Started by Nonameover 23 years ago4 messages
#1Noname
nconway@klamath.dyndns.org

Running CVS HEAD from this morning, 'make check' fails due to an
initdb failure when initializing pg_depend:

initializing pg_depend...
/home/nconway/pgsql/src/test/regress/./tmp_check/install//data/pgsql/bin/initdb:
line 718: 31128 Segmentation fault (core dumped) "$PGPATH"/postgres
$PGSQL_OPT template1 >/dev/null <<EOF

That corresponds to the following lines in initdb, which lead to the
core dump:

DELETE FROM pg_depend;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_class;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_proc;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_type;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_constraint;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_attrdef;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_language;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_operator;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_rewrite;
INSERT INTO pg_depend SELECT 0,0,0, tableoid,oid,0, 'p' FROM pg_trigger;

Here's the backtrace from the core dump:

(gdb) bt
#0 0x0812ff4d in expression_tree_walker (node=0xffffffff,
walker=0x812b0ac <fix_opids_walker>, context=0x0) at clauses.c:1814
#1 0x081300cf in expression_tree_walker (node=0x82cb308,
walker=0x812b0ac <fix_opids_walker>, context=0x0) at clauses.c:1832
#2 0x0812b0f7 in fix_opids_walker (node=0x82cb308, context=0x0) at setrefs.c:579
#3 0x081303c6 in expression_tree_walker (node=0x82cb3a0,
walker=0x812b0ac <fix_opids_walker>, context=0x0) at clauses.c:1930
#4 0x0812b0f7 in fix_opids_walker (node=0x82cb3a0, context=0x0) at setrefs.c:579
#5 0x08130397 in expression_tree_walker (node=0x82cb3e0,
walker=0x812b0ac <fix_opids_walker>, context=0x0) at clauses.c:1925
#6 0x0812b0f7 in fix_opids_walker (node=0x82cb3e0, context=0x0) at setrefs.c:579
#7 0x0812b0a7 in fix_opids (node=0x82cb3e0) at setrefs.c:569
#8 0x0812ac0e in fix_expr_references (plan=0x82cbb38, node=0x82cb3e0)
at setrefs.c:258
#9 0x0812a805 in set_plan_references (plan=0x82cbb38, rtable=0x82be5d8)
at setrefs.c:97
#10 0x08129057 in planner (parse=0x82bdf18) at planner.c:104
#11 0x08155f81 in pg_plan_query (querytree=0x82bdf18) at postgres.c:493
#12 0x081561fd in pg_exec_query_string (query_string=0x82bdc90, dest=Debug,
parse_context=0x8291d60) at postgres.c:759
#13 0x0815745a in PostgresMain (argc=9, argv=0x827b388,
username=0x827b9f0 "nconway") at postgres.c:1924
#14 0x081070cd in main (argc=9, argv=0xbffff9a4) at main.c:229

Can someone fix this please?

Cheers,

Neil

--
Neil Conway <neilconway@rogers.com>
PGP Key ID: DB3C29FC

#2Tom Lane
tgl@sss.pgh.pa.us
In reply to: Noname (#1)
Re: regression in CVS HEAD

nconway@klamath.dyndns.org (Neil Conway) writes:

Running CVS HEAD from this morning, 'make check' fails due to an
initdb failure when initializing pg_depend:

Did you do a full recompile after your last update? I notice Bruce
has arbitrarily renumbered expression nodes ...

regards, tom lane

#3Bruce Momjian
pgman@candle.pha.pa.us
In reply to: Tom Lane (#2)
Re: regression in CVS HEAD

Tom Lane wrote:

nconway@klamath.dyndns.org (Neil Conway) writes:

Running CVS HEAD from this morning, 'make check' fails due to an
initdb failure when initializing pg_depend:

Did you do a full recompile after your last update? I notice Bruce
has arbitrarily renumbered expression nodes ...

Yes, I am seeing a failure too. I didn't have time to run a full
regression after applying those patches last night. I was working on my
own patch at the same time. I will investigate.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026
#4Bruce Momjian
pgman@candle.pha.pa.us
In reply to: Tom Lane (#2)
Re: regression in CVS HEAD

Tom Lane wrote:

nconway@klamath.dyndns.org (Neil Conway) writes:

Running CVS HEAD from this morning, 'make check' fails due to an
initdb failure when initializing pg_depend:

Did you do a full recompile after your last update? I notice Bruce
has arbitrarily renumbered expression nodes ...

OK, it was the BETWEEN patch that was causing the failure. I asked Rod
to retest and resubmit.

-- 
  Bruce Momjian                        |  http://candle.pha.pa.us
  pgman@candle.pha.pa.us               |  (610) 853-3000
  +  If your life is a hard drive,     |  830 Blythe Avenue
  +  Christ can be your backup.        |  Drexel Hill, Pennsylvania 19026