[email protected] (Paul Gilmartin) writes: > If each ADCON is made relative to a location in the module (paired > relocation), the entire module is location-independent and can be > mapped to any location in various address spaces. I know of no > z/OS facility to exploit such a design for general programs. > > I understand the convention has been used for transient areas.
tss/360 (& tss/370) had a different convention for handling address constants that allowed for same module to appear concurrently at different addresses in different virtual address spaces. ... in the early 70s when I first did paged mapped filesystem for cp67/cms (and then ported to vm370/cms) ... I had enormous problems with the OS/360 adcon convention ... because CMS made heavy use of os/360 compilers and software conventions. periodic past tirades on the subject http://www.garlic.com/~lynn/submain.html#adcon one of the people responsible for HASP ... then did a page-mapped filesystem for MFT-II (that he called RASP). Old reference to when it was decided to add virtual memory to MVT ... both the cp67/cms and the MFT-II/RASP work was ignored ... past post on the subject (I was asked to track down somebody that was involved in the decision ... some of his comments): http://www.garlic.com/~lynn/2011d.html#73 Multiple Virtual Memory note the basic original decision was that MVT memory management was so inefficient that (on the avg) only 25% of allocated storage was typically used. adding virtual memory to MVT would allow running 16 initiators on a 1mbyte physical machine (that would ordinarily require 4mbytes, just another on my long list of if you can't figure out how to fix the software, throw hardware at it). note reference also has comments about OS/VS2 release 1 (SVS) and OS/VS2 release 2 (MVS) being on the glide-path to OS/VS2 release 3 ... the operating system for "Future System" that would completely replace 370 ... some past FS posts http://www.garlic.com/~lynn/submain.html#futuresys -- virtualization experience starting Jan1968, online at home since Mar1970 ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
