2008-11-19 13:26:10 +01:00
|
|
|
#
|
2005-04-21 16:53:53 +02:00
|
|
|
! This file, mpx386.s, is included by mpx.s when Minix is compiled for
|
|
|
|
|
! 32-bit Intel CPUs. The alternative mpx88.s is compiled for 16-bit CPUs.
|
|
|
|
|
|
2005-05-18 12:36:23 +02:00
|
|
|
! This file is part of the lowest layer of the MINIX kernel. (The other part
|
|
|
|
|
! is "proc.c".) The lowest layer does process switching and message handling.
|
|
|
|
|
! Furthermore it contains the assembler startup code for Minix and the 32-bit
|
|
|
|
|
! interrupt handlers. It cooperates with the code in "start.c" to set up a
|
|
|
|
|
! good environment for main().
|
2005-04-21 16:53:53 +02:00
|
|
|
|
2005-05-18 12:36:23 +02:00
|
|
|
! Every transition to the kernel goes through this file. Transitions to the
|
|
|
|
|
! kernel may be nested. The initial entry may be with a system call (i.e.,
|
|
|
|
|
! send or receive a message), an exception or a hardware interrupt; kernel
|
|
|
|
|
! reentries may only be made by hardware interrupts. The count of reentries
|
|
|
|
|
! is kept in "k_reenter". It is important for deciding whether to switch to
|
|
|
|
|
! the kernel stack and for protecting the message passing code in "proc.c".
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
! For the message passing trap, most of the machine state is saved in the
|
|
|
|
|
! proc table. (Some of the registers need not be saved.) Then the stack is
|
|
|
|
|
! switched to "k_stack", and interrupts are reenabled. Finally, the system
|
|
|
|
|
! call handler (in C) is called. When it returns, interrupts are disabled
|
|
|
|
|
! again and the code falls into the restart routine, to finish off held-up
|
|
|
|
|
! interrupts and run the process or task whose pointer is in "proc_ptr".
|
|
|
|
|
|
|
|
|
|
! Hardware interrupt handlers do the same, except (1) The entire state must
|
|
|
|
|
! be saved. (2) There are too many handlers to do this inline, so the save
|
|
|
|
|
! routine is called. A few cycles are saved by pushing the address of the
|
|
|
|
|
! appropiate restart routine for a return later. (3) A stack switch is
|
|
|
|
|
! avoided when the stack is already switched. (4) The (master) 8259 interrupt
|
|
|
|
|
! controller is reenabled centrally in save(). (5) Each interrupt handler
|
|
|
|
|
! masks its interrupt line using the 8259 before enabling (other unmasked)
|
|
|
|
|
! interrupts, and unmasks it after servicing the interrupt. This limits the
|
|
|
|
|
! nest level to the number of lines and protects the handler from itself.
|
|
|
|
|
|
|
|
|
|
! For communication with the boot monitor at startup time some constant
|
|
|
|
|
! data are compiled into the beginning of the text segment. This facilitates
|
|
|
|
|
! reading the data at the start of the boot process, since only the first
|
|
|
|
|
! sector of the file needs to be read.
|
|
|
|
|
|
|
|
|
|
! Some data storage is also allocated at the end of this file. This data
|
|
|
|
|
! will be at the start of the data segment of the kernel and will be read
|
|
|
|
|
! and modified by the boot monitor before the kernel starts.
|
|
|
|
|
|
|
|
|
|
! sections
|
|
|
|
|
|
2008-12-19 16:46:29 +01:00
|
|
|
#include <sys/vm_i386.h>
|
|
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
.sect .text
|
|
|
|
|
begtext:
|
|
|
|
|
.sect .rom
|
|
|
|
|
begrom:
|
|
|
|
|
.sect .data
|
|
|
|
|
begdata:
|
|
|
|
|
.sect .bss
|
|
|
|
|
begbss:
|
|
|
|
|
|
|
|
|
|
#include <minix/config.h>
|
|
|
|
|
#include <minix/const.h>
|
|
|
|
|
#include <minix/com.h>
|
2005-05-24 12:06:17 +02:00
|
|
|
#include <ibm/interrupt.h>
|
Split of architecture-dependent and -independent functions for i386,
mainly in the kernel and headers. This split based on work by
Ingmar Alting <iaalting@cs.vu.nl> done for his Minix PowerPC architecture
port.
. kernel does not program the interrupt controller directly, do any
other architecture-dependent operations, or contain assembly any more,
but uses architecture-dependent functions in arch/$(ARCH)/.
. architecture-dependent constants and types defined in arch/$(ARCH)/include.
. <ibm/portio.h> moved to <minix/portio.h>, as they have become, for now,
architecture-independent functions.
. int86, sdevio, readbios, and iopenable are now i386-specific kernel calls
and live in arch/i386/do_* now.
. i386 arch now supports even less 86 code; e.g. mpx86.s and klib86.s have
gone, and 'machine.protected' is gone (and always taken to be 1 in i386).
If 86 support is to return, it should be a new architecture.
. prototypes for the architecture-dependent functions defined in
kernel/arch/$(ARCH)/*.c but used in kernel/ are in kernel/proto.h
. /etc/make.conf included in makefiles and shell scripts that need to
know the building architecture; it defines ARCH=<arch>, currently only
i386.
. some basic per-architecture build support outside of the kernel (lib)
. in clock.c, only dequeue a process if it was ready
. fixes for new include files
files deleted:
. mpx/klib.s - only for choosing between mpx/klib86 and -386
. klib86.s - only for 86
i386-specific files files moved (or arch-dependent stuff moved) to arch/i386/:
. mpx386.s (entry point)
. klib386.s
. sconst.h
. exception.c
. protect.c
. protect.h
. i8269.c
2006-12-22 16:22:27 +01:00
|
|
|
#include <archconst.h>
|
|
|
|
|
#include "../../const.h"
|
2005-04-21 16:53:53 +02:00
|
|
|
#include "sconst.h"
|
|
|
|
|
|
|
|
|
|
/* Selected 386 tss offsets. */
|
|
|
|
|
#define TSS3_S_SP0 4
|
|
|
|
|
|
|
|
|
|
! Exported functions
|
|
|
|
|
! Note: in assembly language the .define statement applied to a function name
|
|
|
|
|
! is loosely equivalent to a prototype in C code -- it makes it possible to
|
|
|
|
|
! link to an entity declared in the assembly code but does not create
|
|
|
|
|
! the entity.
|
|
|
|
|
|
|
|
|
|
.define _restart
|
|
|
|
|
.define save
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
.define _reload_cr3
|
|
|
|
|
.define _write_cr3 ! write cr3
|
2008-11-19 13:26:10 +01:00
|
|
|
|
|
|
|
|
.define errexception
|
|
|
|
|
.define exception1
|
|
|
|
|
.define exception
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
.define _divide_error
|
|
|
|
|
.define _single_step_exception
|
|
|
|
|
.define _nmi
|
|
|
|
|
.define _breakpoint_exception
|
|
|
|
|
.define _overflow
|
|
|
|
|
.define _bounds_check
|
|
|
|
|
.define _inval_opcode
|
|
|
|
|
.define _copr_not_available
|
|
|
|
|
.define _double_fault
|
|
|
|
|
.define _copr_seg_overrun
|
|
|
|
|
.define _inval_tss
|
|
|
|
|
.define _segment_not_present
|
|
|
|
|
.define _stack_exception
|
|
|
|
|
.define _general_protection
|
|
|
|
|
.define _page_fault
|
|
|
|
|
.define _copr_error
|
2008-11-19 13:26:10 +01:00
|
|
|
.define _params_size
|
|
|
|
|
.define _params_offset
|
|
|
|
|
.define _mon_ds
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
.define _schedcheck
|
|
|
|
|
.define _dirtypde
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
.define _hwint00 ! handlers for hardware interrupts
|
|
|
|
|
.define _hwint01
|
|
|
|
|
.define _hwint02
|
|
|
|
|
.define _hwint03
|
|
|
|
|
.define _hwint04
|
|
|
|
|
.define _hwint05
|
|
|
|
|
.define _hwint06
|
|
|
|
|
.define _hwint07
|
|
|
|
|
.define _hwint08
|
|
|
|
|
.define _hwint09
|
|
|
|
|
.define _hwint10
|
|
|
|
|
.define _hwint11
|
|
|
|
|
.define _hwint12
|
|
|
|
|
.define _hwint13
|
|
|
|
|
.define _hwint14
|
|
|
|
|
.define _hwint15
|
|
|
|
|
|
|
|
|
|
.define _s_call
|
|
|
|
|
.define _p_s_call
|
|
|
|
|
.define _level0_call
|
|
|
|
|
|
|
|
|
|
! Exported variables.
|
|
|
|
|
.define begbss
|
|
|
|
|
.define begdata
|
|
|
|
|
|
|
|
|
|
.sect .text
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* MINIX *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
MINIX: ! this is the entry point for the MINIX kernel
|
|
|
|
|
jmp over_flags ! skip over the next few bytes
|
|
|
|
|
.data2 CLICK_SHIFT ! for the monitor: memory granularity
|
|
|
|
|
flags:
|
|
|
|
|
.data2 0x01FD ! boot monitor flags:
|
|
|
|
|
! call in 386 mode, make bss, make stack,
|
2005-09-07 20:15:21 +02:00
|
|
|
! load high, don't patch, will return,
|
2005-04-21 16:53:53 +02:00
|
|
|
! uses generic INT, memory vector,
|
|
|
|
|
! new boot code return
|
|
|
|
|
nop ! extra byte to sync up disassembler
|
|
|
|
|
over_flags:
|
|
|
|
|
|
|
|
|
|
! Set up a C stack frame on the monitor stack. (The monitor sets cs and ds
|
|
|
|
|
! right. The ss descriptor still references the monitor data segment.)
|
|
|
|
|
movzx esp, sp ! monitor stack is a 16 bit stack
|
|
|
|
|
push ebp
|
|
|
|
|
mov ebp, esp
|
|
|
|
|
push esi
|
|
|
|
|
push edi
|
|
|
|
|
cmp 4(ebp), 0 ! monitor return vector is
|
|
|
|
|
jz noret ! nonzero if return possible
|
|
|
|
|
inc (_mon_return)
|
|
|
|
|
noret: mov (_mon_sp), esp ! save stack pointer for later return
|
|
|
|
|
|
|
|
|
|
! Copy the monitor global descriptor table to the address space of kernel and
|
|
|
|
|
! switch over to it. Prot_init() can then update it with immediate effect.
|
|
|
|
|
|
|
|
|
|
sgdt (_gdt+GDT_SELECTOR) ! get the monitor gdtr
|
|
|
|
|
mov esi, (_gdt+GDT_SELECTOR+2) ! absolute address of GDT
|
|
|
|
|
mov ebx, _gdt ! address of kernel GDT
|
|
|
|
|
mov ecx, 8*8 ! copying eight descriptors
|
|
|
|
|
copygdt:
|
|
|
|
|
eseg movb al, (esi)
|
|
|
|
|
movb (ebx), al
|
|
|
|
|
inc esi
|
|
|
|
|
inc ebx
|
|
|
|
|
loop copygdt
|
|
|
|
|
mov eax, (_gdt+DS_SELECTOR+2) ! base of kernel data
|
|
|
|
|
and eax, 0x00FFFFFF ! only 24 bits
|
|
|
|
|
add eax, _gdt ! eax = vir2phys(gdt)
|
|
|
|
|
mov (_gdt+GDT_SELECTOR+2), eax ! set base of GDT
|
|
|
|
|
lgdt (_gdt+GDT_SELECTOR) ! switch over to kernel GDT
|
|
|
|
|
|
|
|
|
|
! Locate boot parameters, set up kernel segment registers and stack.
|
|
|
|
|
mov ebx, 8(ebp) ! boot parameters offset
|
|
|
|
|
mov edx, 12(ebp) ! boot parameters length
|
|
|
|
|
mov eax, 16(ebp) ! address of a.out headers
|
|
|
|
|
mov (_aout), eax
|
|
|
|
|
mov ax, ds ! kernel data
|
|
|
|
|
mov es, ax
|
|
|
|
|
mov fs, ax
|
|
|
|
|
mov gs, ax
|
|
|
|
|
mov ss, ax
|
|
|
|
|
mov esp, k_stktop ! set sp to point to the top of kernel stack
|
|
|
|
|
|
2008-11-19 13:26:10 +01:00
|
|
|
! Save boot parameters into these global variables for i386 code
|
|
|
|
|
mov (_params_size), edx
|
|
|
|
|
mov (_params_offset), ebx
|
|
|
|
|
mov (_mon_ds), SS_SELECTOR
|
|
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
! Call C startup code to set up a proper environment to run main().
|
|
|
|
|
push edx
|
|
|
|
|
push ebx
|
|
|
|
|
push SS_SELECTOR
|
|
|
|
|
push DS_SELECTOR
|
|
|
|
|
push CS_SELECTOR
|
|
|
|
|
call _cstart ! cstart(cs, ds, mds, parmoff, parmlen)
|
|
|
|
|
add esp, 5*4
|
|
|
|
|
|
|
|
|
|
! Reload gdtr, idtr and the segment registers to global descriptor table set
|
|
|
|
|
! up by prot_init().
|
|
|
|
|
|
|
|
|
|
lgdt (_gdt+GDT_SELECTOR)
|
|
|
|
|
lidt (_gdt+IDT_SELECTOR)
|
|
|
|
|
|
|
|
|
|
jmpf CS_SELECTOR:csinit
|
|
|
|
|
csinit:
|
|
|
|
|
o16 mov ax, DS_SELECTOR
|
|
|
|
|
mov ds, ax
|
|
|
|
|
mov es, ax
|
|
|
|
|
mov fs, ax
|
|
|
|
|
mov gs, ax
|
|
|
|
|
mov ss, ax
|
|
|
|
|
o16 mov ax, TSS_SELECTOR ! no other TSS is used
|
|
|
|
|
ltr ax
|
|
|
|
|
push 0 ! set flags to known good state
|
|
|
|
|
popf ! esp, clear nested task and int enable
|
|
|
|
|
jmp _main ! main()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* interrupt handlers *
|
|
|
|
|
!* interrupt handlers for 386 32-bit protected mode *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* hwint00 - 07 *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
! Note this is a macro, it just looks like a subroutine.
|
2005-05-24 12:06:17 +02:00
|
|
|
#define hwint_master(irq) \
|
|
|
|
|
call save /* save interrupted process state */;\
|
2005-05-02 16:30:04 +02:00
|
|
|
push (_irq_handlers+4*irq) /* irq_handlers[irq] */;\
|
2005-05-18 12:36:23 +02:00
|
|
|
call _intr_handle /* intr_handle(irq_handlers[irq]) */;\
|
2005-04-21 16:53:53 +02:00
|
|
|
pop ecx ;\
|
|
|
|
|
cmp (_irq_actids+4*irq), 0 /* interrupt still active? */;\
|
2005-05-24 12:06:17 +02:00
|
|
|
jz 0f ;\
|
|
|
|
|
inb INT_CTLMASK /* get current mask */ ;\
|
|
|
|
|
orb al, [1<<irq] /* mask irq */ ;\
|
|
|
|
|
outb INT_CTLMASK /* disable the irq */;\
|
|
|
|
|
0: movb al, END_OF_INT ;\
|
|
|
|
|
outb INT_CTL /* reenable master 8259 */;\
|
|
|
|
|
ret /* restart (another) process */
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
! Each of these entry points is an expansion of the hwint_master macro
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint00: ! Interrupt routine for irq 0 (the clock).
|
|
|
|
|
hwint_master(0)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint01: ! Interrupt routine for irq 1 (keyboard)
|
|
|
|
|
hwint_master(1)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint02: ! Interrupt routine for irq 2 (cascade!)
|
|
|
|
|
hwint_master(2)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint03: ! Interrupt routine for irq 3 (second serial)
|
|
|
|
|
hwint_master(3)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint04: ! Interrupt routine for irq 4 (first serial)
|
|
|
|
|
hwint_master(4)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint05: ! Interrupt routine for irq 5 (XT winchester)
|
|
|
|
|
hwint_master(5)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint06: ! Interrupt routine for irq 6 (floppy)
|
|
|
|
|
hwint_master(6)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint07: ! Interrupt routine for irq 7 (printer)
|
|
|
|
|
hwint_master(7)
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* hwint08 - 15 *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
! Note this is a macro, it just looks like a subroutine.
|
|
|
|
|
#define hwint_slave(irq) \
|
|
|
|
|
call save /* save interrupted process state */;\
|
2005-05-02 16:30:04 +02:00
|
|
|
push (_irq_handlers+4*irq) /* irq_handlers[irq] */;\
|
2005-09-11 18:44:06 +02:00
|
|
|
call _intr_handle /* intr_handle(irq_handlers[irq]) */;\
|
2005-04-21 16:53:53 +02:00
|
|
|
pop ecx ;\
|
|
|
|
|
cmp (_irq_actids+4*irq), 0 /* interrupt still active? */;\
|
2005-05-24 12:06:17 +02:00
|
|
|
jz 0f ;\
|
2005-04-21 16:53:53 +02:00
|
|
|
inb INT2_CTLMASK ;\
|
2005-05-24 12:06:17 +02:00
|
|
|
orb al, [1<<[irq-8]] ;\
|
|
|
|
|
outb INT2_CTLMASK /* disable the irq */;\
|
|
|
|
|
0: movb al, END_OF_INT ;\
|
|
|
|
|
outb INT_CTL /* reenable master 8259 */;\
|
|
|
|
|
outb INT2_CTL /* reenable slave 8259 */;\
|
|
|
|
|
ret /* restart (another) process */
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
! Each of these entry points is an expansion of the hwint_slave macro
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint08: ! Interrupt routine for irq 8 (realtime clock)
|
|
|
|
|
hwint_slave(8)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint09: ! Interrupt routine for irq 9 (irq 2 redirected)
|
|
|
|
|
hwint_slave(9)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint10: ! Interrupt routine for irq 10
|
|
|
|
|
hwint_slave(10)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint11: ! Interrupt routine for irq 11
|
|
|
|
|
hwint_slave(11)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint12: ! Interrupt routine for irq 12
|
|
|
|
|
hwint_slave(12)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint13: ! Interrupt routine for irq 13 (FPU exception)
|
|
|
|
|
hwint_slave(13)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint14: ! Interrupt routine for irq 14 (AT winchester)
|
|
|
|
|
hwint_slave(14)
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
_hwint15: ! Interrupt routine for irq 15
|
|
|
|
|
hwint_slave(15)
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* save *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
! Save for protected mode.
|
|
|
|
|
! This is much simpler than for 8086 mode, because the stack already points
|
|
|
|
|
! into the process table, or has already been switched to the kernel stack.
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
save:
|
|
|
|
|
cld ! set direction flag to a known value
|
|
|
|
|
pushad ! save "general" registers
|
|
|
|
|
o16 push ds ! save ds
|
|
|
|
|
o16 push es ! save es
|
|
|
|
|
o16 push fs ! save fs
|
|
|
|
|
o16 push gs ! save gs
|
|
|
|
|
mov dx, ss ! ss is kernel data segment
|
|
|
|
|
mov ds, dx ! load rest of kernel segments
|
|
|
|
|
mov es, dx ! kernel does not use fs, gs
|
|
|
|
|
mov eax, esp ! prepare to return
|
|
|
|
|
incb (_k_reenter) ! from -1 if not reentering
|
|
|
|
|
jnz set_restart1 ! stack is already kernel stack
|
|
|
|
|
mov esp, k_stktop
|
|
|
|
|
push _restart ! build return address for int handler
|
|
|
|
|
xor ebp, ebp ! for stacktrace
|
|
|
|
|
jmp RETADR-P_STACKBASE(eax)
|
|
|
|
|
|
|
|
|
|
.align 4
|
|
|
|
|
set_restart1:
|
|
|
|
|
push restart1
|
|
|
|
|
jmp RETADR-P_STACKBASE(eax)
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* _s_call *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
.align 16
|
|
|
|
|
_s_call:
|
|
|
|
|
_p_s_call:
|
|
|
|
|
cld ! set direction flag to a known value
|
Merge of David's ptrace branch. Summary:
o Support for ptrace T_ATTACH/T_DETACH and T_SYSCALL
o PM signal handling logic should now work properly, even with debuggers
being present
o Asynchronous PM/VFS protocol, full IPC support for senda(), and
AMF_NOREPLY senda() flag
DETAILS
Process stop and delay call handling of PM:
o Added sys_runctl() kernel call with sys_stop() and sys_resume()
aliases, for PM to stop and resume a process
o Added exception for sending/syscall-traced processes to sys_runctl(),
and matching SIGKREADY pseudo-signal to PM
o Fixed PM signal logic to deal with requests from a process after
stopping it (so-called "delay calls"), using the SIGKREADY facility
o Fixed various PM panics due to race conditions with delay calls versus
VFS calls
o Removed special PRIO_STOP priority value
o Added SYS_LOCK RTS kernel flag, to stop an individual process from
running while modifying its process structure
Signal and debugger handling in PM:
o Fixed debugger signals being dropped if a second signal arrives when
the debugger has not retrieved the first one
o Fixed debugger signals being sent to the debugger more than once
o Fixed debugger signals unpausing process in VFS; removed PM_UNPAUSE_TR
protocol message
o Detached debugger signals from general signal logic and from being
blocked on VFS calls, meaning that even VFS can now be traced
o Fixed debugger being unable to receive more than one pending signal in
one process stop
o Fixed signal delivery being delayed needlessly when multiple signals
are pending
o Fixed wait test for tracer, which was returning for children that were
not waited for
o Removed second parallel pending call from PM to VFS for any process
o Fixed process becoming runnable between exec() and debugger trap
o Added support for notifying the debugger before the parent when a
debugged child exits
o Fixed debugger death causing child to remain stopped forever
o Fixed consistently incorrect use of _NSIG
Extensions to ptrace():
o Added T_ATTACH and T_DETACH ptrace request, to attach and detach a
debugger to and from a process
o Added T_SYSCALL ptrace request, to trace system calls
o Added T_SETOPT ptrace request, to set trace options
o Added TO_TRACEFORK trace option, to attach automatically to children
of a traced process
o Added TO_ALTEXEC trace option, to send SIGSTOP instead of SIGTRAP upon
a successful exec() of the tracee
o Extended T_GETUSER ptrace support to allow retrieving a process's priv
structure
o Removed T_STOP ptrace request again, as it does not help implementing
debuggers properly
o Added MINIX3-specific ptrace test (test42)
o Added proper manual page for ptrace(2)
Asynchronous PM/VFS interface:
o Fixed asynchronous messages not being checked when receive() is called
with an endpoint other than ANY
o Added AMF_NOREPLY senda() flag, preventing such messages from
satisfying the receive part of a sendrec()
o Added asynsend3() that takes optional flags; asynsend() is now a
#define passing in 0 as third parameter
o Made PM/VFS protocol asynchronous; reintroduced tell_fs()
o Made PM_BASE request/reply number range unique
o Hacked in a horrible temporary workaround into RS to deal with newly
revealed RS-PM-VFS race condition triangle until VFS is asynchronous
System signal handling:
o Fixed shutdown logic of device drivers; removed old SIGKSTOP signal
o Removed is-superuser check from PM's do_procstat() (aka getsigset())
o Added sigset macros to allow system processes to deal with the full
signal set, rather than just the POSIX subset
Miscellaneous PM fixes:
o Split do_getset into do_get and do_set, merging common code and making
structure clearer
o Fixed setpriority() being able to put to sleep processes using an
invalid parameter, or revive zombie processes
o Made find_proc() global; removed obsolete proc_from_pid()
o Cleanup here and there
Also included:
o Fixed false-positive boot order kernel warning
o Removed last traces of old NOTIFY_FROM code
THINGS OF POSSIBLE INTEREST
o It should now be possible to run PM at any priority, even lower than
user processes
o No assumptions are made about communication speed between PM and VFS,
although communication must be FIFO
o A debugger will now receive incoming debuggee signals at kill time
only; the process may not yet be fully stopped
o A first step has been made towards making the SYSTEM task preemptible
2009-09-30 11:57:22 +02:00
|
|
|
sub esp, 4 ! skip RETADR
|
|
|
|
|
pushad ! save "general" registers
|
2005-04-21 16:53:53 +02:00
|
|
|
o16 push ds
|
|
|
|
|
o16 push es
|
|
|
|
|
o16 push fs
|
|
|
|
|
o16 push gs
|
2008-11-19 13:26:10 +01:00
|
|
|
|
2006-03-10 17:10:05 +01:00
|
|
|
mov si, ss ! ss is kernel data segment
|
|
|
|
|
mov ds, si ! load rest of kernel segments
|
|
|
|
|
mov es, si ! kernel does not use fs, gs
|
|
|
|
|
incb (_k_reenter) ! increment kernel entry count
|
2005-04-21 16:53:53 +02:00
|
|
|
mov esi, esp ! assumes P_STACKBASE == 0
|
|
|
|
|
mov esp, k_stktop
|
|
|
|
|
xor ebp, ebp ! for stacktrace
|
|
|
|
|
! end of inline save
|
|
|
|
|
! now set up parameters for sys_call()
|
2006-03-10 17:10:05 +01:00
|
|
|
push edx ! event set or flags bit map
|
2005-04-21 16:53:53 +02:00
|
|
|
push ebx ! pointer to user message
|
2006-03-10 17:10:05 +01:00
|
|
|
push eax ! source / destination
|
|
|
|
|
push ecx ! call number (ipc primitive to use)
|
2008-11-19 13:26:10 +01:00
|
|
|
|
2006-03-10 17:10:05 +01:00
|
|
|
call _sys_call ! sys_call(call_nr, src_dst, m_ptr, bit_map)
|
2005-04-21 16:53:53 +02:00
|
|
|
! caller is now explicitly in proc_ptr
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
mov AXREG(esi), eax
|
2005-04-21 16:53:53 +02:00
|
|
|
|
|
|
|
|
! Fall into code to restart proc/task running.
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* restart *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
_restart:
|
|
|
|
|
|
2005-05-24 12:06:17 +02:00
|
|
|
! Restart the current process or the next process if it is set.
|
|
|
|
|
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
cli
|
|
|
|
|
call _schedcheck ! ask C function who we're running
|
|
|
|
|
mov esp, (_proc_ptr) ! will assume P_STACKBASE == 0
|
2005-05-24 16:35:58 +02:00
|
|
|
lldt P_LDT_SEL(esp) ! enable process' segment descriptors
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
cmp P_CR3(esp), 0 ! process does not have its own PT
|
|
|
|
|
jz 0f
|
|
|
|
|
mov eax, P_CR3(esp)
|
|
|
|
|
cmp eax, (loadedcr3)
|
|
|
|
|
jz 0f
|
|
|
|
|
mov cr3, eax
|
|
|
|
|
mov (loadedcr3), eax
|
|
|
|
|
mov eax, (_proc_ptr)
|
|
|
|
|
mov (_ptproc), eax
|
|
|
|
|
mov (_dirtypde), 0
|
|
|
|
|
0:
|
2005-04-21 16:53:53 +02:00
|
|
|
lea eax, P_STACKTOP(esp) ! arrange for next interrupt
|
|
|
|
|
mov (_tss+TSS3_S_SP0), eax ! to save state in process table
|
|
|
|
|
restart1:
|
|
|
|
|
decb (_k_reenter)
|
|
|
|
|
o16 pop gs
|
|
|
|
|
o16 pop fs
|
|
|
|
|
o16 pop es
|
|
|
|
|
o16 pop ds
|
|
|
|
|
popad
|
|
|
|
|
add esp, 4 ! skip return adr
|
|
|
|
|
iretd ! continue process
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* exception handlers *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
_divide_error:
|
|
|
|
|
push DIVIDE_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_single_step_exception:
|
|
|
|
|
push DEBUG_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_nmi:
|
|
|
|
|
push NMI_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_breakpoint_exception:
|
|
|
|
|
push BREAKPOINT_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_overflow:
|
|
|
|
|
push OVERFLOW_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_bounds_check:
|
|
|
|
|
push BOUNDS_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_inval_opcode:
|
|
|
|
|
push INVAL_OP_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_copr_not_available:
|
|
|
|
|
push COPROC_NOT_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_double_fault:
|
|
|
|
|
push DOUBLE_FAULT_VECTOR
|
|
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_copr_seg_overrun:
|
|
|
|
|
push COPROC_SEG_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
_inval_tss:
|
|
|
|
|
push INVAL_TSS_VECTOR
|
|
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_segment_not_present:
|
|
|
|
|
push SEG_NOT_VECTOR
|
|
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_stack_exception:
|
|
|
|
|
push STACK_FAULT_VECTOR
|
|
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_general_protection:
|
|
|
|
|
push PROTECTION_VECTOR
|
|
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_page_fault:
|
|
|
|
|
push PAGE_FAULT_VECTOR
|
2008-11-19 13:26:10 +01:00
|
|
|
push eax
|
|
|
|
|
mov eax, cr2
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
sseg mov (pagefaultcr2), eax
|
2008-11-19 13:26:10 +01:00
|
|
|
pop eax
|
2005-04-21 16:53:53 +02:00
|
|
|
jmp errexception
|
|
|
|
|
|
|
|
|
|
_copr_error:
|
|
|
|
|
push COPROC_ERR_VECTOR
|
|
|
|
|
jmp exception
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* exception *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
! This is called for all exceptions which do not push an error code.
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
exception:
|
|
|
|
|
sseg mov (trap_errno), 0 ! clear trap_errno
|
|
|
|
|
sseg pop (ex_number)
|
|
|
|
|
jmp exception1
|
|
|
|
|
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
!* errexception *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
! This is called for all exceptions which push an error code.
|
|
|
|
|
|
|
|
|
|
.align 16
|
|
|
|
|
errexception:
|
|
|
|
|
sseg pop (ex_number)
|
|
|
|
|
sseg pop (trap_errno)
|
|
|
|
|
exception1: ! Common for all exceptions.
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
sseg mov (old_eax_ptr), esp ! where will eax be saved?
|
|
|
|
|
sseg sub (old_eax_ptr), PCREG-AXREG ! here
|
|
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
push eax ! eax is scratch register
|
2008-11-19 13:26:10 +01:00
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
mov eax, 0+4(esp) ! old eip
|
|
|
|
|
sseg mov (old_eip), eax
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
mov eax, esp
|
|
|
|
|
add eax, 4
|
|
|
|
|
sseg mov (old_eip_ptr), eax
|
2005-04-21 16:53:53 +02:00
|
|
|
movzx eax, 4+4(esp) ! old cs
|
|
|
|
|
sseg mov (old_cs), eax
|
|
|
|
|
mov eax, 8+4(esp) ! old eflags
|
|
|
|
|
sseg mov (old_eflags), eax
|
2008-11-19 13:26:10 +01:00
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
pop eax
|
|
|
|
|
call save
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
push (pagefaultcr2)
|
|
|
|
|
push (old_eax_ptr)
|
|
|
|
|
push (old_eip_ptr)
|
2005-04-21 16:53:53 +02:00
|
|
|
push (old_eflags)
|
|
|
|
|
push (old_cs)
|
|
|
|
|
push (old_eip)
|
|
|
|
|
push (trap_errno)
|
|
|
|
|
push (ex_number)
|
|
|
|
|
call _exception ! (ex_number, trap_errno, old_eip,
|
|
|
|
|
! old_cs, old_eflags)
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
add esp, 8*4
|
2005-04-21 16:53:53 +02:00
|
|
|
ret
|
|
|
|
|
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
!*===========================================================================*
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
!* write_cr3 *
|
2005-04-21 16:53:53 +02:00
|
|
|
!*===========================================================================*
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
! PUBLIC void write_cr3(unsigned long value);
|
|
|
|
|
_write_cr3:
|
|
|
|
|
push ebp
|
|
|
|
|
mov ebp, esp
|
|
|
|
|
mov eax, 8(ebp)
|
|
|
|
|
cmp eax, (loadedcr3)
|
|
|
|
|
jz 0f
|
|
|
|
|
mov cr3, eax
|
|
|
|
|
mov (loadedcr3), eax
|
|
|
|
|
mov (_dirtypde), 0
|
|
|
|
|
0:
|
|
|
|
|
pop ebp
|
|
|
|
|
ret
|
2005-04-21 16:53:53 +02:00
|
|
|
|
2008-11-19 13:26:10 +01:00
|
|
|
!*===========================================================================*
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
!* reload_cr3 *
|
2008-11-19 13:26:10 +01:00
|
|
|
!*===========================================================================*
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
! PUBLIC void reload_cr3(void);
|
|
|
|
|
_reload_cr3:
|
|
|
|
|
push ebp
|
|
|
|
|
mov ebp, esp
|
|
|
|
|
mov (_dirtypde), 0
|
|
|
|
|
mov eax, cr3
|
|
|
|
|
mov cr3, eax
|
|
|
|
|
pop ebp
|
2008-11-19 13:26:10 +01:00
|
|
|
ret
|
|
|
|
|
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
!*===========================================================================*
|
|
|
|
|
!* level0_call *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
_level0_call:
|
|
|
|
|
call save
|
|
|
|
|
jmp (_level0_func)
|
|
|
|
|
|
2005-04-21 16:53:53 +02:00
|
|
|
!*===========================================================================*
|
|
|
|
|
!* data *
|
|
|
|
|
!*===========================================================================*
|
|
|
|
|
|
|
|
|
|
.sect .rom ! Before the string table please
|
|
|
|
|
.data2 0x526F ! this must be the first data entry (magic #)
|
|
|
|
|
|
|
|
|
|
.sect .bss
|
|
|
|
|
k_stack:
|
|
|
|
|
.space K_STACK_BYTES ! kernel stack
|
|
|
|
|
k_stktop: ! top of kernel stack
|
|
|
|
|
.comm ex_number, 4
|
|
|
|
|
.comm trap_errno, 4
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
.comm old_eip_ptr, 4
|
|
|
|
|
.comm old_eax_ptr, 4
|
2005-04-21 16:53:53 +02:00
|
|
|
.comm old_eip, 4
|
|
|
|
|
.comm old_cs, 4
|
|
|
|
|
.comm old_eflags, 4
|
Primary goal for these changes is:
- no longer have kernel have its own page table that is loaded
on every kernel entry (trap, interrupt, exception). the primary
purpose is to reduce the number of required reloads.
Result:
- kernel can only access memory of process that was running when
kernel was entered
- kernel must be mapped into every process page table, so traps to
kernel keep working
Problem:
- kernel must often access memory of arbitrary processes (e.g. send
arbitrary processes messages); this can't happen directly any more;
usually because that process' page table isn't loaded at all, sometimes
because that memory isn't mapped in at all, sometimes because it isn't
mapped in read-write.
So:
- kernel must be able to map in memory of any process, in its own
address space.
Implementation:
- VM and kernel share a range of memory in which addresses of
all page tables of all processes are available. This has two purposes:
. Kernel has to know what data to copy in order to map in a range
. Kernel has to know where to write the data in order to map it in
That last point is because kernel has to write in the currently loaded
page table.
- Processes and kernel are separated through segments; kernel segments
haven't changed.
- The kernel keeps the process whose page table is currently loaded
in 'ptproc.'
- If it wants to map in a range of memory, it writes the value of the
page directory entry for that range into the page directory entry
in the currently loaded map. There is a slot reserved for such
purposes. The kernel can then access this memory directly.
- In order to do this, its segment has been increased (and the
segments of processes start where it ends).
- In the pagefault handler, detect if the kernel is doing
'trappable' memory access (i.e. a pagefault isn't a fatal
error) and if so,
- set the saved instruction pointer to phys_copy_fault,
breaking out of phys_copy
- set the saved eax register to the address of the page
fault, both for sanity checking and for checking in
which of the two ranges that phys_copy was called
with the fault occured
- Some boot-time processes do not have their own page table,
and are mapped in with the kernel, and separated with
segments. The kernel detects this using HASPT. If such a
process has to be scheduled, any page table will work and
no page table switch is done.
Major changes in kernel are
- When accessing user processes memory, kernel no longer
explicitly checks before it does so if that memory is OK.
It simply makes the mapping (if necessary), tries to do the
operation, and traps the pagefault if that memory isn't present;
if that happens, the copy function returns EFAULT.
So all of the CHECKRANGE_OR_SUSPEND macros are gone.
- Kernel no longer has to copy/read and parse page tables.
- A message copying optimisation: when messages are copied, and
the recipient isn't mapped in, they are copied into a buffer
in the kernel. This is done in QueueMess. The next time
the recipient is scheduled, this message is copied into
its memory. This happens in schedcheck().
This eliminates the mapping/copying step for messages, and makes
it easier to deliver messages. This eliminates soft_notify.
- Kernel no longer creates a page table at all, so the vm_setbuf
and pagetable writing in memory.c is gone.
Minor changes in kernel are
- ipc_stats thrown out, wasn't used
- misc flags all renamed to MF_*
- NOREC_* macros to enter and leave functions that should not
be called recursively; just sanity checks really
- code to fully decode segment selectors and descriptors
to print on exceptions
- lots of vmassert()s added, only executed if DEBUG_VMASSERT is 1
2009-09-21 16:31:52 +02:00
|
|
|
.comm pagefaultcr2, 4
|
|
|
|
|
.comm loadedcr3, 4
|
2008-11-19 13:26:10 +01:00
|
|
|
|
