INFORMIX SERVER / 5725A3900 / C10 - IDS 12.10

Following the submission of a request to read ahead index or
data pages the requesting thread may issue an abort request for
that request if the pages are no longer required. The request is
processed by one of the readahead threads but this may not
notice the abort request either during or prior to commencement
of the request processing. When this happens there is the
potential for the unintentional modification of memory in the
thread associated with the original request.

Depending on the activities of the originating thread such
modification may go unnoticed or may lead to an Assert Failure.
There could be many different stacks seen for the AF but two
scenarios have been seen during the exit processing of the
thread. One is an Assert Failure for a bad block header and the
other is a crash with a stack such as:

(oninit) afstack
(oninit) mt_ex_throw_sig
(oninit) afsig_handler
(Linux) 
(oninit) opfree
(oninit) rscleanup
(oninit) sqscb_cleanup
(oninit) destroy_session
(oninit) sq_exit
(oninit) sqmain
(oninit) spawn_thread

****************************************************************
* USERS AFFECTED:                                              *
* When readahead is active                                     *
****************************************************************
* PROBLEM DESCRIPTION:                                         *
* See Error Description                                        *
****************************************************************
* RECOMMENDATION:                                              *
* Upgrade to 12.10.xC6                                         *
****************************************************************

not known / see Local fix

Date  - problem reported    :
Date  - problem closed      :
Date  - last modified       :

07.10.2015
19.01.2016
19.01.2016

12.10.xC6