core dump in forms30

From: Kim Ng <kimmng_at_uswnvg.com>
Date: 29 Jun 94 21:43:17 GMT
Message-ID: <32512_at_uswnvg.uswnvg.com>


Hi everyone

Has anyone experience an intermittent core dump when doing "runform30" ? If yes (see details below), how do you solve it --- if solvable.

My client is using Unix on Sequent (ptx V2.0.1 Release 3.2.0), SQL*FORMS 3.0.16.12.7 (production), PL/SQL 2.0.17.1.0 (production) and ORACLE 7.0.15.4.0 (production).

The problem we had was:
(1) An existing form which has been in use for months suddenly did a

    core dump.
(2) This core dump can be duplicated on another machine (same ptx) but

    it occured at another location when a different key was pressed.
(3) Another test caused another core dump with a different key and

    location. For example: the first one occured when "KEY-NXTFLD" was     pressed while in field "a"; the second one when "KEY-UP" in field     "z" and the third when "KEY-PRVFLD" in field "x". NONE of the value     of ANY fields had been changed.
(4) This happened any time, any field (database, non-database, single-

    record block, multi-record-block).
(5) This had happened to use about 6 months ago and then it mysteriously

    went away --- until now.
(6) It seemed that once the first core dump occurred, the subsequent ones

    took less keystrokes for it to happened. For example, when we tried to     recreate the problem many, many hours later, it took a lot of keystrokes     (maybe 40 to 50 of all navigation keys I could think of) before it core     dumped on exit (I sort of gave up before it decided to give me a break).     Immediately, I reran the form and upon the 10th or 20th next field, it     core dumped.

We suspected strongly that it is a memory problem, but we do not know what actually caused it. Also, once, we got the message "Assertion failed:
(struct header *)lblk < arenaend, file malloc.c, line 445" before it went
into infinite loop. We forced it into core dump and got 180 megabytes of "core" file. When we ran "dbx" on all "core" files, "malloc" seemed to be the last statement.

Note: this core dump happened the first time when we were in the form, not

      after repeated call.

Any hint or help would be greatly appreciated. Thanks



 Kim Ng
 (Just a low life contract programmer. Thus, my clients won't adopt my views.) Received on Wed Jun 29 1994 - 23:43:17 CEST

Original text of this message