[Info-vax] Crashing a system with PCA/DTM ?

[Marc Van Dyck]

Now time for a weird question.

Long time ago, I developped some code that later on became an important
component of our production environment. This was in Alpha times. So
important that I implemented a regression testing environment with DTM
and PCA, to ensure I didn't break anything with new versions.

This code was in fact quite stable so the regression environment was
not used very often.

10 years later we migrated to Itanium and I just rebuilt, without
running the regression tests - all worked perfectly.

Now we migrated to VSI OpenVMS and my PCSI kit is broken, because it
insists on having HP OpenVMS installed. At the same time, I finally got
a request to implement a new function in my code (it's written in
Pascal).

So I dusted off the development environment of this code, made some
modifications, changed the PCF file to make a new kit for VSI, and
rebuilt. So far, so good.

Now I'd like to run my regression testing again. It's painful enough to
remember how DTM and PCA work, after having used it only once 15 years
ago. But my problem is that when I did that 15 years ago, I crashed the
development system. Twice. I'd like not to repeat that, but I do not
remember exactly how it happened.

I tend to remember it has something to do with images installed, and
possibly with privileges. My code is made of a shareable image,
installed /OPEN /HEADER /SHARED, and a main image that calls it,
installed with CMKRNL (it creates kernel mode logicals). Something
about that being not compatible with PCA or DTM (more probably PCA).
But the details escape me, it's too long ago.

Does that ring a bell to anyone ?

Thank you,

Marc.

-- 
Marc Van Dyck