- Reinstall the "dynamic page table length" that was removed some
years ago.
- Limit the user page table submap to max 5% of available memory.
- Free the page table space when a process is swapped out.
- If the UPT submap runs out of space, throw away pmap mappings
using the same algorithm as for swapping processes.
As a result of this, 4MB machines are useable again and it's even possible
to compile a kernel for 2MB machines (but it will be slow... :-)
Still to do:
- Multiprocessor fixes.
- More profiling.
Fix botch on my part and make the IPL_* match reality on VAX.
Redefine spl macro using the symbolic IPL_ instead of being hardcoded.
Move schedsoftnet, schedsoftclock from <machine/cpu.h> to <machine/intr.h>
Add a _setsirr macro for schedsoft*.
Add softintr function and framework.
both uniprocessor and multiprocessor environments. Use the otherwise
unused internal CPU register SSP to store the cpu_info pointer.
The macros curcpu(), curproc, cpu_number() and need_resched() are now the
same in both uniprocessor and multiprocessor environments.
state into global and per-CPU scheduler state:
- Global state: sched_qs (run queues), sched_whichqs (bitmap
of non-empty run queues), sched_slpque (sleep queues).
NOTE: These may collectively move into a struct schedstate
at some point in the future.
- Per-CPU state, struct schedstate_percpu: spc_runtime
(time process on this CPU started running), spc_flags
(replaces struct proc's p_schedflags), and
spc_curpriority (usrpri of processes on this CPU).
- Every platform must now supply a struct cpu_info and
a curcpu() macro. Simplify existing cpu_info declarations
where appropriate.
- All references to per-CPU scheduler state now made through
curcpu(). NOTE: this will likely be adjusted in the future
after further changes to struct proc are made.
Tested on i386 and Alpha. Changes are mostly mechanical, but apologies
in advance if it doesn't compile on a particular platform.
This is done on a physical page size basis, instead of virtual (as the
(on vax yet non-existing) bus_* routines does). This is similar to the
way uba allocation is done.