Re: [Sqlrelay-discussion] memory corruption, apache segfault

[Firstworks/4access <dav...@fi...>]

Try http://www.firstworks.com/sqlrelay-0.37pre2.tar.gz and see what you
get.  I didn't specifically fix the problem, but some of the changes I
made for handling non-ascii data may have fixed it.

Dave
dav...@fi...

On Wed, 2005-09-14 at 16:09 +0300, Ionut Ciocirlan wrote:
> Yes, still happening.
>=20
> Should i try CVS?
>=20
> Ionut
>=20
> ----- Original Message -----=20
> From: "David Muse" <dav...@fi...>
> To: <sql...@li...>
> Sent: Wednesday, September 14, 2005 4:27 AM
> Subject: Re: [Sqlrelay-discussion] memory corruption, apache segfault
>=20
>=20
> | Well, I said I'd look into this problem, then I misplaced the email
> | until today :/
> |
> | Are you still having this problem?
> |
> | David Muse
> | dav...@fi...
> |
> |
> | On Fri, 2005-08-19 at 07:48 +0300, Ionut Ciocirlan wrote:
> | > Hi,
> | >
> | > Nasty problem I ran into:
> | >
> | > I'm running sqlrelay-0.36.4 compiled against oracle instant=20
> client-10.1.0.4,
> | > on fedora core 3, glibc-2.3.5; apache-2.0.53 / php-4.3.11.
> | >
> | > Trying to fetch text containing several utf-8 characters (namely=20
> diacritics)
> | > from clobs sometimes crashes apache, with glibc memory errors.
> | >
> | > NLS_LANG is .AL32UTF8 for both Oracle and SQL Relay; data posted in=
to=20
> Apache
> | > is in UTF-8.
> | >
> | > Apache crashes with various messages:
> | >
> | > *** glibc detected *** malloc(): memory corruption: 0x094d6f20 ***
> | > [notice] child pid 4005 exit signal Aborted (6)
> | >
> | > *** glibc detected *** free(): invalid next size (fast): 0x0a174a80=
 ***
> | > [notice] child pid 15277 exit signal Aborted (6)
> | >
> | > *** glibc detected *** double free or corruption (!prev): 0x094d0ee=
0 ***
> | > [notice] child pid 4004 exit signal Aborted (6)
> | >
> | > *** glibc detected *** double free or corruption (out): 0x0948c598 =
***
> | > [notice] child pid 4062 exit signal Aborted (6)
> | > [notice] child pid 4009 exit signal Segmentation fault (11)
> | >
> | > The most common error is the first (memory corruption). On almost a=
ll
> | > occasions, the client browser gets a zero-sized reply. I'd say it's=
=20
> random,
> | > but on fast reruns it happens ~ 5 out of 6 times. Sometimes (rarely=
) i=20
> get
> | > no crash, but apache hangs without replying, and the browser hangs=20
> waiting
> | > for an answer.
> | >
> | > Reproduce this by inserting characters 'z =C4=83=C3=AE=C5=9F=C5=A3=C3=
=A2 y =C4=82=C3=8E=C5=9E=C5=A2=C3=82 x' in a clob=20
> column.
> | > If it doesn't reproduce add this strign several more times and mix =
it=20
> with
> | > various other text.
> | >
> | > So, this happens only on clob columns (never on char/varchar):
> | >
> | > Fetching (illegally?) in non-lob mode crashes at sqlrcur_sendQuery:
> | >
> | > $query =3D 'select column from table where id =3D xx';
> | > sqlrcur_sendQuery($cur,$query);
> | > ---- crash
> | >
> | > Clob fetching crashes at sqlrcur_executeQuery:
> | >
> | > $query =3D 'begin
> | > select column into :outvar from table where where id =3D xx;
> | > end;'
> | > sqlrcur_prepareQuery($cur, $query);
> | > sqlrcur_defineOutputBindClob($cur, 'outvar');
> | > sqlrcur_executeQuery($cur);
> | > ---- crash
> | >
> | >
> | > The memory corruption issues seem to be pretty real: it reflects ma=
inly=20
> in a
> | > flash communication server running on the same machine, that after=20
> several
> | > such events starts behaving erratically. Sshd and samba get erratic=
 too
> | > (connections lost etc.). Also, after several hours of testing/debug=
ging,=20
> my
> | > kernel Oopsed. And just now network connectivity went into oblivion=
. The
> | > only solution to restoring normal operation seems to be rebooting.
> | >
> | > Conclusions so far:
> | > - not a machine problem; memtested the hell out of the ram;
> | > - not an oracle/oci/glibc problem; i recompiled php's oci8 extensio=
n=20
> against
> | > the same oracle instant client libs, and i get the expected output,=
 plus=20
> no
> | > crashes;
> | > - i also recompiled sqlrelay against the 9.1 oracle installation; s=
ame
> | > results;
> | > - might be a problem that appeared since 0.34 (we did some extensiv=
e=20
> testing
> | > back then, and it all worked out fine); i wanted to test now on 0.3=
4=20
> too,
> | > but for some reason it now fails to compile right from the start
> | >
> | > I hope this will be easy to reproduce / repair.....
> | >
> | > Regards,
> | > Ionut
> | >
> | >
> | >
> | > -------------------------------------------------------
> | > SF.Net email is Sponsored by the Better Software Conference & EXPO
> | > September 19-22, 2005 * San Francisco, CA * Development Lifecycle=20
> Practices
> | > Agile & Plan-Driven Development * Managing Projects & Teams * Testi=
ng &=20
> QA
> | > Security * Process Improvement & Measurement *=20
> http://www.sqe.com/bsce5sf
> | > _______________________________________________
> | > Sqlrelay-discussion mailing list
> | > Sql...@li...
> | > https://lists.sourceforge.net/lists/listinfo/sqlrelay-discussion
> | >
> |
> |
> |
> | -------------------------------------------------------
> | SF.Net email is sponsored by:
> | Tame your development challenges with Apache's Geronimo App Server.=20
> Download
> | it for free - -and be entered to win a 42" plasma tv or your very own
> | Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php
> | _______________________________________________
> | Sqlrelay-discussion mailing list
> | Sql...@li...
> | https://lists.sourceforge.net/lists/listinfo/sqlrelay-discussion=20
>=20
>=20
>=20
> -------------------------------------------------------
> SF.Net email is sponsored by:
> Tame your development challenges with Apache's Geronimo App Server. Dow=
nload
> it for free - -and be entered to win a 42" plasma tv or your very own
> Sony(tm)PSP.  Click here to play: http://sourceforge.net/geronimo.php
> _______________________________________________
> Sqlrelay-discussion mailing list
> Sql...@li...
> https://lists.sourceforge.net/lists/listinfo/sqlrelay-discussion
>=20