> It's been several years since I had a hand in that code (as the team > leader of the VSM development team for the first release of MVS back in > 1972-1974), but someone mentioned the existence of the FQE (Free Queue > Element). The FQE described the number of bytes that were free within > the block of storage obtained. If the storage size requested was less > than a page or some multiple of pages, there would be some free space, > which would always be in the first page of the obtained area and an FQE > would be put at the start of that first page to identify the amount of > free space. > > Storage requests were always oriented to the high end of of the last > page of the requested area, so that an attempt to access storage beyond > the end of the area would result in a page fault. The idea being that > the more common storage addressing error resulted from having a > legitimate storage address, but an offset that exceeded the size of the > actual obtained area. > > Mike Myers > Mentor Services Corporation
VSM was completely rewritten in PL/S for the first release of MVS/XA. As of that rewrite, all VSM control blocks are in cell pools in ELSQA, so none are imbedded in the storage they describe. Taking a look at the ancient MVS/370 IEAVGM00 assembler module, it appears that FQEs for LSQA and SQA were imbedded in the free space that they described. AQEs (representing TCB ownership for LSQA subpools 253 and 254) were also imbedded in the storage that they described. Jim Mulder z/OS System Test IBM Corp. Poughkeepsie, NY ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
