We are running Ingres 9.2.0 with patch 13300. We are using a 'C'
application to execute a bunch SQL's (INS/UPD/DEL) against the
database and we ran into the following error:

XXXXXX01::[61091 , 0c7e4f00]: Mon May 4 12:02:10 2009
Segmentation Vi
olation (SIGSEGV) ade_execute_cx(0x1005ce020+0x404) @ PC 1005ce424 SP
ffffffff79
63cec1 PSR 99 G1 ffffffff7e0398c0 o0
15
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff7963dce0
: qen_execute_cx+0x28 [0x10030d0ec]
(10d10e918,10d110d88,0,fffffffffffffff9,0,10d
65e761)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff7963dd90
: qea_append+0x7c4 [0x1002a0c3c]
(10cbb0d50,10c801cd8,10d10e918,0,0,10cbb0bd8)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff7963ded0
: qeq_query+0xfb4 [0x100298b9c]
(100297a14,0,10c801cd8,100287958,0,60000)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff7963e820
: qef_call+0xd44 [0x100288f9c](100a93228,163c,
10c801cd8,650,1,60000)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff7963f1b0
: scs_sequencer+0x46f0 [0x10004f43c]
(2000,10c7e4f00,60000,2,1,5)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff79641c30
: CSMT_setup+0x574 [0x1006d3d9c]
(10002,4,ffdfbc00,100a8af00,10c7e4f00,ffdfbfff)
XXXXXX01::[61091 , 0c7e4f00]: Stack dump pid 18886:
0xffffffff79641f50
: _thr_slot_offset+0x484 [0xffffffff7e0d2268]
(0,0,0,0,0,0)
XXXXXX01 ::[61091 , 18886 ,
000000010c7e4f00]: Mon May
4 12:02:10 2009 E_SC0216_QEF_ERROR Error returned by
QEF.
XXXXXX01 ::[61091 , 18886 ,
000000010c7e4f00]: Mon May
4 12:02:10 2009 E_SC0206_CANNOT_PROCESS An internal error
prevents furth
er processing of this query.
-------------------------------------------------------------------------
Anybody else ever encounter this error? What was the cause?
Thanks.

On May 12, 2009, at 10:29 AM, jimmythegeek44720 (AT) yahoo (DOT) com wrote:



From the stack backtrace, I can tell you that it was executing an
insert statement, and was executing the CX (compiled expression)
that stages values from the select statement to line them up
as needed for the insert. Why that would segv during CX execution
is anyone's guess at this point. I don't offhand recall seeing
a bug-fix related to segv's in CX execution, but that doesn't
mean much. :-) If you have support from Ingres, I'd open an
issue with them.

Is this reproducible? If it's not easy to reproduce, I
would start getting suspicious that it's some kind of memory
corruption bug, and those are hard to track down.

Karl

On May 12, 12:02*pm, Karl & Betty Schendel <schen... (AT) kbcomputer (DOT) com>
wrote:
Thanks