> > >Of course the nucleus is designed not to page-fault while > > disabled. One way > > >that is guaranteed is by being sure that every byte ever > > touched by disabled > > >code is fixed before doing the disabling operation. > > > > If I'm not mistaken, the nucleus is fixed and non-swappable. > > Some of it is > > even designed to run with DAT off. > > > > -- > > Tom Marchant > > I wonder why? Is there something in fixed memory which would otherwise > be unavailable because the frames are not in any page mapping? If not, > they why not just have such frames be globally mapped V=R in all address > spaces? Most curious.
The portions of the OS that don't contain implicit assumptions of disabled access are typically going to be found in the LPA if it has to be addressable everywhere, or in the linklist, or in the plethora of service address spaces. The nucleus is intended to contain only those portions of code that have the most demanding requirements. Being page-fixed and straddling the 16MB (virtual) line addresses most of that although there is still a chunk of code (a.k.a. "the DAT-off nucleus") that arrives early and runs V=R during NIP and is responsible for building the control structures that enable the rest of the system to run V=V. In reality it is likely that less and less of the OS has such demanding requirements, but there is still a lot of old code around. When you mix new code in with the implicit assumptions hidden inside the 40 year-old code, it is no-doubt safer to err on the side of caution. CC ---------------------------------------------------------------------- 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
