We didn't do anything on this so far, because the fix is easy to do (test for X'80000000') and this is at one central place in our dialog system; all module calls are processed by this control module which contains the CDE processing and the LOAD, if necessary (we call it a "interface module").
For NULL tests of addresses, we normally do something like LA R1,0(R1) cleans bit 0 in 31 bit mode and 0 to 7 in 24 bit mode IF (R1,Z) generates LTR and Branch but in this case we didn't do it, first because we assumed an address from the CDE to be valid and second because we need the first bit to know the AMODE (or for anything else, I'm not directly involved in the maintenance of these "interface modules" - I'm a freelance consultant in this area, but for some 10+ years now). Kind regards Bernd Am Donnerstag, 26. Juni 2008 19:55 schrieben Sie: > John, that may work very well, but my preference is that LOAD processing > delete the faulty CDE if the LOAD fails. Has anyone opened an incident > with IBM on this issue? It might very well be a "bug" in the code. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
