It makes more sense to have a scheduler (or runqueue pair) per-cpu
rather than per-container. This provides more flexible global scheduling
policy that is also simpler due to all scheduling details being controlled
from a single point.
Status:
- Capability initialization is a bit hacky with dummy current etc.
- All container caps belong to the pager
- Tasks refer to their pager's capabilities for mutex allocation - Hacky.
- Kernel container keeps quantitative caps and memory caps in separate lists - Hacky.
These will all evolve and get fixed.
Previously virt_to_phys/phys_to_virt macros were used such that they did a
blind offset translation for creating a pager internal virtual address for
user mappings.
This is now changed such that a properly bookkeeped virtual address pool is
used which will avoid any clashes on the virtual space.
