On Monday, 11/16/2009 at 08:17 EST, "J. Cassidy" <s...@jdcassidy.net> wrote: > am trying to run (believe it or not) an OS/390 2.4 guest under z/VM 5.3 > (0703) on a CP (z990). ... > As soon as I reply with >>r 00,sysp=00,clpa<< everything dies with a > HCPGIR453W CP entered; program interrupt loop message.
As Rob says, HCP453W indicates that the PROGRAM NEW PSW was corrupted or otherwise not valid when a Program Interrupt (program check) occurred. While it's fairly easy to find out what caused the fatal program interrupt, it won't help to explain why the program new PSW is bad. You can try #CP TRACE PROG RUN CMD D PSW PROG and then re-IPL. You will likely get more than one hit, but the last one will be the fatal one and you will will see from the output that the PNPSW is bogus. Wiping out of the PNPSW isn't usually a surgical strike, but is more often part of the larger mass destruction of low core (that's page 0 for you young whippersnappers out there), usually from picking up an unexpectedly zero pointer. > Anyone any ideas?? It's anyone's guess. Unfortunately you are trying to run an unsupported operating system in an unsupported configuration (OS/390 2.4 on a z990), so I can't even suggest that you open a PMR. The z990 had a minimum requirement for OS/390 2.10 (with PTFs). Even the z900 required OS/390 2.8 (with PTFs). But before you give up, make absolutely certain that there is nothing "new" in the virtual machine configuration. Where was this OS/390 running prior to your attempt to bring it up as a guest of z/VM 5.3 on the z990? Is there anything different about the virtual machine config from it's prior home? And take a look at those TRACE PROG hits. One of them might give you a hint. In case it hasn't been said often or loud enough: Make sure Management knows that z/VM does not hide the hardware from the guest. The guest sees any new chpid types, new disks, new cryptos, and new CPU features (and the loss of old ones, in rare cases). So when the h/w says that some particular level of s/w is required, there's a reason fir it. Sometimes is just the validation and other times it is a technical dependency. If you let the software get *too* old, then you create a void: The newer software won't run on the old box and the older software won't run on the new box. It's a painful, and potentially expensive, chasm to cross. Alan Altmark z/VM Development IBM Endicott