[JOS-Kernel] [kernel bytecode cache] Definition

[Gilbert Carl Herschberger II]

A kernel bytecode cache can be combined with bytecode as a resource to put
boot classes for virtual machines inside a kernel. This saves both space
and time.

It saves space because boot classes are the most often duplicated classes
in an off-the-shelf virtual machine. Boot classes should be statically
linked to a kernel and shared by multiple virtual machines and/or multiple
bytecode processes.

Also, a kernel bytecode cache is tied to the kernel's virtual memory memory
manager so that only the bytecode that's currently being used is found in
real memory. The bytecode in the kernel bytecode cache is marked
read-only/executable for optimal results.

It saves time because boot classes are loaded as part of the kernel.
Instead of "downloading" classes from a file subsystem, a virtual machine
dynamically links to its boot classes. Dynamic linking is repeated for each
bytecode process in a MPCL-compatible machine.

When using the kernel of any foreign operating system, a virtual machine
can link to the "kernel bytecode cache" implemented as a shared library.
This simulator may or may not be tied directly to the virtual memory manager.


_______________________________________________
Kernel maillist  -  [EMAIL PROTECTED]
http://jos.org/mailman/listinfo/kernel