Dennis, I'm still here, John is working with us directly. Guy has done qite a bit of work on the VM and he is debugging it now on the target. I have asked him to write up an overview and status and we'll post it to the task map.
Tom Dennis Clarke wrote: > Cyril, I have spent some time over the past few days reading up and thinking > about page table schemes. Keep in mind that I am working in the dark with > no source to refer to other than the AMD Opteron NUMA-smart sources. > > As a reference we have this from Guy Shaw : > > http://polaris.blastwave.org/wiki/PolarisVirtualmemdetails > > http://polaris.blastwave.org/wiki/PolarisVirtualmem > > and a textbook picture : > > http://polaris.blastwave.org/attachment/wiki/PolarisVirtualmem/sol_vm.JPG > >>From a long long time ago I have this : > > http://www.blastwave.org/articles/BLS-0057/index.html > I resureccted it and cleaned it up. > > I don't know what the current VA space architecture is and I can only assume > we have a multi-level page table scheme. My hope is that we can architect a > page table scheme with levels of indirection that grant a full address space > to processes. A possible downside being fragmentation in PTE groups. These > are axioms however. > > I have spent some time over the past months sketching thoughts and reading > up things from the "OpenSolaris Community: Performance" pages. > > Hopefully we will be able to look at the memory management architecture > implemented by Tom and Guy Shaw and then go forwards into either 32-bit or > 64-bit address spaces. > > Dennis > > _______________________________________________ > powerpc-discuss mailing list > powerpc-discuss at opensolaris.org -- Thomas Riddle Sun Microsystems Laboratories 16 Network Circle Menlo Park, CA 94025 Mail-Stop MPK16-162 Email: tom.riddle at sun.com Phone: (650) 786-2433 Fax: (650) 786-6020
