minix3/kernel/arch/i386/mpx.S

/* 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(). 
 *
 * Kernel is entered either because of kernel-calls, ipc-calls, interrupts or
 * exceptions. TSS is set so that the kernel stack is loaded. The user context is
 * saved to the proc table and the handler of the event is called. Once the
 * handler is done, switch_to_user() function is called to pick a new process,
 * finish what needs to be done for the next process to run, sets its context
 * and switch to userspace.
 *
 * 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.
 */

#include "kernel/kernel.h" /* configures the kernel */

/* sections */

#include <machine/vm.h>
#include "kernel/kernel.h"
#include <minix/config.h>
#include <minix/const.h>
#include <minix/ipcconst.h>
#include <minix/com.h>
#include <machine/asm.h>
#include <machine/interrupt.h>
#include "archconst.h"
#include "kernel/const.h"
#include "kernel/proc.h"
#include "sconst.h"
#include <machine/multiboot.h>

#include "arch_proto.h" /* K_STACK_SIZE */

#ifdef CONFIG_SMP
#include "kernel/smp.h"
#endif

/* Selected 386 tss offsets. */
#define TSS3_S_SP0	4

IMPORT(usermapped_offset)
IMPORT(copr_not_available_handler)
IMPORT(params_size)
IMPORT(params_offset)
IMPORT(switch_to_user)
IMPORT(multiboot_init)

.text
/*===========================================================================*/
/*				interrupt handlers			     */
/*		interrupt handlers for 386 32-bit protected mode	     */
/*===========================================================================*/

#define PIC_IRQ_HANDLER(irq)	\
	push	$irq							    	;\
	call	_C_LABEL(irq_handle)	/* intr_handle(irq_handlers[irq]) */	;\
	add	$4, %esp						    	;

/*===========================================================================*/
/*				hwint00 - 07				     */
/*===========================================================================*/
/* Note this is a macro, it just looks like a subroutine. */

#define hwint_master(irq) \
	TEST_INT_IN_KERNEL(4, 0f)					;\
									\
	SAVE_PROCESS_CTX(0, KTS_INT_HARD)				;\
	push	%ebp							;\
	movl	$0, %ebp	/* for stack trace */			;\
	call	_C_LABEL(context_stop)					;\
	add	$4, %esp						;\
	PIC_IRQ_HANDLER(irq)						;\
	movb	$END_OF_INT, %al					;\
	outb	$INT_CTL	/* reenable interrupts in master pic */	;\
	jmp	_C_LABEL(switch_to_user)				;\
									\
0:									\
	pusha								;\
	call	_C_LABEL(context_stop_idle)				;\
	PIC_IRQ_HANDLER(irq)						;\
	movb	$END_OF_INT, %al					;\
	outb	$INT_CTL	/* reenable interrupts in master pic */	;\
	CLEAR_IF(10*4(%esp))						;\
	popa								;\
	iret								;

/* Each of these entry points is an expansion of the hwint_master macro */
ENTRY(hwint00)
/* Interrupt routine for irq 0 (the clock). */
	hwint_master(0)

ENTRY(hwint01)
/* Interrupt routine for irq 1 (keyboard) */
	hwint_master(1)

ENTRY(hwint02)
/* Interrupt routine for irq 2 (cascade!) */
	hwint_master(2)

ENTRY(hwint03)
/* Interrupt routine for irq 3 (second serial) */
	hwint_master(3)

ENTRY(hwint04)
/* Interrupt routine for irq 4 (first serial) */
	hwint_master(4)

ENTRY(hwint05)
/* Interrupt routine for irq 5 (XT winchester) */
	hwint_master(5)

ENTRY(hwint06)
/* Interrupt routine for irq 6 (floppy) */
	hwint_master(6)

ENTRY(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)	\
	TEST_INT_IN_KERNEL(4, 0f)					;\
									\
	SAVE_PROCESS_CTX(0, KTS_INT_HARD)				;\
	push	%ebp							;\
	movl	$0, %ebp	/* for stack trace */			;\
	call	_C_LABEL(context_stop)					;\
	add	$4, %esp						;\
	PIC_IRQ_HANDLER(irq)						;\
	movb	$END_OF_INT, %al					;\
	outb	$INT_CTL	/* reenable interrupts in master pic */	;\
	outb	$INT2_CTL	/* reenable slave 8259		  */	;\
	jmp	_C_LABEL(switch_to_user)				;\
									\
0:									\
	pusha								;\
	call	_C_LABEL(context_stop_idle)				;\
	PIC_IRQ_HANDLER(irq)						;\
	movb	$END_OF_INT, %al					;\
	outb	$INT_CTL	/* reenable interrupts in master pic */	;\
	outb	$INT2_CTL	/* reenable slave 8259		  */	;\
	CLEAR_IF(10*4(%esp))						;\
	popa								;\
	iret								;

/* Each of these entry points is an expansion of the hwint_slave macro */
ENTRY(hwint08)
/* Interrupt routine for irq 8 (realtime clock) */
	hwint_slave(8)

ENTRY(hwint09)
/* Interrupt routine for irq 9 (irq 2 redirected) */
	hwint_slave(9)

ENTRY(hwint10)
/* Interrupt routine for irq 10 */
	hwint_slave(10)

ENTRY(hwint11)
/* Interrupt routine for irq 11 */
	hwint_slave(11)

ENTRY(hwint12)
/* Interrupt routine for irq 12 */
	hwint_slave(12)

ENTRY(hwint13)
/* Interrupt routine for irq 13 (FPU exception) */
	hwint_slave(13)

ENTRY(hwint14)
/* Interrupt routine for irq 14 (AT winchester) */
	hwint_slave(14)

ENTRY(hwint15)
/* Interrupt routine for irq 15 */
	hwint_slave(15)

/* differences with sysenter:
 *   - we have to find our own per-cpu stack (i.e. post-SYSCALL
 *     %esp is not configured)
 *   - we have to save the post-SYSRET %eip, provided by the cpu
 *     in %ecx
 *   - the system call parameters are passed in %ecx, so we userland
 *     code that executes SYSCALL copies %ecx to %edx. So the roles
 *     of %ecx and %edx are reversed
 *   - we can use %esi as a scratch register
 */
#define ipc_entry_syscall_percpu(cpu)			;\
ENTRY(ipc_entry_syscall_cpu ## cpu)			;\
	xchg	%ecx, %edx				;\
	mov	k_percpu_stacks+4*cpu, %esi		;\
	mov	(%esi), %ebp 				;\
	movl	$KTS_SYSCALL, P_KERN_TRAP_STYLE(%ebp)	;\
	xchg	%esp, %esi				;\
	jmp	syscall_sysenter_common

ipc_entry_syscall_percpu(0)
ipc_entry_syscall_percpu(1)
ipc_entry_syscall_percpu(2)
ipc_entry_syscall_percpu(3)
ipc_entry_syscall_percpu(4)
ipc_entry_syscall_percpu(5)
ipc_entry_syscall_percpu(6)
ipc_entry_syscall_percpu(7)

ENTRY(ipc_entry_sysenter)
	/* SYSENTER simply sets kernel segments, EIP to here, and ESP
	 * to tss->sp0 (through MSR). so no automatic context saving is done.
	 * interrupts are disabled.
	 *
	 * register usage:
	 * edi: call type (IPCVEC, KERVEC)
	 * ebx, eax, ecx: syscall params, set by userland
	 * esi, edx: esp, eip to restore, set by userland
	 *
	 * no state is automatically saved; userland does all of that.
	 */
	mov	(%esp), %ebp /* get proc saved by arch_finish_switch_to_user */

	/* inform kernel we entered by sysenter and should
	 * therefore exit through restore_user_context_sysenter
	 */
	movl	$KTS_SYSENTER, P_KERN_TRAP_STYLE(%ebp)
	add	usermapped_offset, %edx	/* compensate for mapping difference */

syscall_sysenter_common:
	mov	%esi, SPREG(%ebp)	/* esi is return esp */
	mov	%edx, PCREG(%ebp)	/* edx is return eip */

	/* save PSW */
	pushf
	pop	%edx
	mov	%edx, PSWREG(%ebp)

	/* check for call type; do_ipc? */
	cmp	$IPCVEC_UM, %edi
	jz	ipc_entry_common

	/* check for kernel trap */
	cmp	$KERVEC_UM, %edi
	jz	kernel_call_entry_common

	/* unrecognized call number; restore user with error */
	movl	$-1, AXREG(%ebp)
	push	%ebp	
	call	restore_user_context	/* restore_user_context(%ebp); */

/*
 * IPC is only from a process to kernel
 */
ENTRY(ipc_entry_softint_orig)
	SAVE_PROCESS_CTX(0, KTS_INT_ORIG)
	jmp ipc_entry_common

ENTRY(ipc_entry_softint_um)
	SAVE_PROCESS_CTX(0, KTS_INT_UM)
	jmp ipc_entry_common

ENTRY(ipc_entry_common)
	/* save the pointer to the current process */
	push	%ebp

	/*
	 * pass the syscall arguments from userspace to the handler.
	 * SAVE_PROCESS_CTX() does not clobber these registers, they are still
	 * set as the userspace have set them
	 */
	push	%ebx
	push	%eax
	push	%ecx

	/* stop user process cycles */
	push	%ebp
	/* for stack trace */
	movl	$0, %ebp
	call	_C_LABEL(context_stop)
	add	$4, %esp

	call	_C_LABEL(do_ipc)

	/* restore the current process pointer and save the return value */
	add	$3 * 4, %esp
	pop	%esi
	mov     %eax, AXREG(%esi)

	jmp	_C_LABEL(switch_to_user)


/*
 * kernel call is only from a process to kernel
 */
ENTRY(kernel_call_entry_orig)
	SAVE_PROCESS_CTX(0, KTS_INT_ORIG)
	jmp	kernel_call_entry_common

ENTRY(kernel_call_entry_um)
	SAVE_PROCESS_CTX(0, KTS_INT_UM)
	jmp	kernel_call_entry_common

ENTRY(kernel_call_entry_common)
	/* save the pointer to the current process */
	push	%ebp

	/*
	 * pass the syscall arguments from userspace to the handler.
	 * SAVE_PROCESS_CTX() does not clobber these registers, they are still
	 * set as the userspace have set them
	 */
	push	%eax

	/* stop user process cycles */
	push	%ebp
	/* for stack trace */
	movl	$0, %ebp
	call	_C_LABEL(context_stop)
	add	$4, %esp

	call	_C_LABEL(kernel_call)

	/* restore the current process pointer and save the return value */
	add	$8, %esp

	jmp	_C_LABEL(switch_to_user)


.balign 16
/*
 * called by the exception interrupt vectors. If the exception does not push
 * errorcode, we assume that the vector handler pushed 0 instead. Next pushed
 * thing is the vector number. From this point on we can continue as if every
 * exception pushes an error code
 */
exception_entry:
	/*
	 * check if it is a nested trap by comparing the saved code segment
	 * descriptor with the kernel CS first
	 */
	TEST_INT_IN_KERNEL(12, exception_entry_nested)

exception_entry_from_user:
	SAVE_PROCESS_CTX(8, KTS_INT_HARD)

	/* stop user process cycles */
	push	%ebp
	/* for stack trace clear %ebp */
	movl	$0, %ebp
	call	_C_LABEL(context_stop)
	add	$4, %esp

	/*
	 * push a pointer to the interrupt state pushed by the cpu and the
	 * vector number pushed by the vector handler just before calling
	 * exception_entry and call the exception handler.
	 */
	push	%esp
	push	$0	/* it's not a nested exception */
	call 	_C_LABEL(exception_handler)

	jmp	_C_LABEL(switch_to_user)

exception_entry_nested:

	pusha
	mov	%esp, %eax
	add	$(8 * 4), %eax
	push	%eax
	pushl	$1			/* it's a nested exception */
	call	_C_LABEL(exception_handler)
	add	$8, %esp
	popa

	/* clear the error code and the exception number */
	add	$8, %esp
	/* resume execution at the point of exception */
	iret

ENTRY(restore_user_context_sysenter)
	/* return to userspace using sysexit.
	 * most of the context saving the userspace process is
	 * responsible for, we just have to take care of the right EIP
	 * and ESP restoring here to resume execution, and set EAX and
	 * EBX to the saved status values.
	 */
	mov	4(%esp), %ebp		/* retrieve proc ptr arg */
	movw	$USER_DS_SELECTOR, %ax
	movw	%ax, %ds
	mov	PCREG(%ebp), %edx	/* sysexit restores EIP using EDX */
	mov	SPREG(%ebp), %ecx	/* sysexit restores ESP using ECX */
	mov	AXREG(%ebp), %eax	/* trap return value */
	mov	BXREG(%ebp), %ebx	/* secondary return value */
	movl	PSWREG(%ebp), %edi	/* load desired PSW to EDI */
	sti				/* enable interrupts */
	sysexit				/* jump to EIP in user */

ENTRY(restore_user_context_syscall)
	/* return to userspace using sysret.
	 * the procedure is very similar to sysexit; it requires
	 * manual %esp restoring, new EIP in ECX, does not require
	 * enabling interrupts, and of course sysret instead of sysexit.
	 */
	mov	4(%esp), %ebp		/* retrieve proc ptr arg */
	mov	PCREG(%ebp), %ecx	/* sysret restores EIP using ECX */
	mov	SPREG(%ebp), %esp	/* restore ESP directly */
	mov	AXREG(%ebp), %eax	/* trap return value */
	mov	BXREG(%ebp), %ebx	/* secondary return value */
	movl	PSWREG(%ebp), %edi	/* load desired PSW to EDI */
	sysret				/* jump to EIP in user */

ENTRY(restore_user_context_int)
	mov	4(%esp), %ebp	/* will assume P_STACKBASE == 0 */

	/* reconstruct the stack for iret */
	push	$USER_DS_SELECTOR	/* ss */
	movl	SPREG(%ebp), %eax
	push	%eax
	movl	PSWREG(%ebp), %eax
	push	%eax
	push	$USER_CS_SELECTOR	/* cs */
	movl	PCREG(%ebp), %eax
	push	%eax

	/* Restore segments as the user should see them. */
	movw	$USER_DS_SELECTOR, %si
        movw    %si, %ds
        movw    %si, %es
        movw    %si, %fs
        movw    %si, %gs

	/* Same for general-purpose registers. */
	RESTORE_GP_REGS(%ebp)

	movl	BPREG(%ebp), %ebp

	iret	/* continue process */

/*===========================================================================*/
/*				exception handlers			     */
/*===========================================================================*/

#define EXCEPTION_ERR_CODE(vector)	\
	push	$vector			;\
	jmp	exception_entry

#define EXCEPTION_NO_ERR_CODE(vector)	\
	pushl	$0		;\
	EXCEPTION_ERR_CODE(vector)

LABEL(divide_error)
	EXCEPTION_NO_ERR_CODE(DIVIDE_VECTOR)

LABEL(single_step_exception)
	EXCEPTION_NO_ERR_CODE(DEBUG_VECTOR)

LABEL(nmi)
#ifndef USE_WATCHDOG
	EXCEPTION_NO_ERR_CODE(NMI_VECTOR)
#else
	/*
	 * We have to be very careful as this interrupt can occur anytime. On
	 * the other hand, if it interrupts a user process, we will resume the
	 * same process which makes things a little simpler. We know that we are
	 * already on kernel stack whenever it happened and we can be
	 * conservative and save everything as we don't need to be extremely
	 * efficient as the interrupt is infrequent and some overhead is already
	 * expected.
	 */

	/*
	 * save the important registers. We don't save %cs and %ss and they are
	 * saved and restored by CPU
	 */
	pushw	%ds
	pushw	%es
	pushw	%fs
	pushw	%gs
	pusha

	/*
	 * We cannot be sure about the state of the kernel segment register,
	 * however, we always set %ds and %es to the same as %ss
	 */
	mov	%ss, %si
	mov	%si, %ds
	mov	%si, %es

	push	%esp
	call	_C_LABEL(nmi_watchdog_handler)
	add	$4, %esp

	/* restore all the important registers as they were before the trap */
	popa
	popw	%gs
	popw	%fs
	popw	%es
	popw	%ds

	iret
#endif

LABEL(breakpoint_exception)
	EXCEPTION_NO_ERR_CODE(BREAKPOINT_VECTOR)

LABEL(overflow)
	EXCEPTION_NO_ERR_CODE(OVERFLOW_VECTOR)

LABEL(bounds_check)
	EXCEPTION_NO_ERR_CODE(BOUNDS_VECTOR)

LABEL(inval_opcode)
	EXCEPTION_NO_ERR_CODE(INVAL_OP_VECTOR)

LABEL(copr_not_available)
	TEST_INT_IN_KERNEL(4, copr_not_available_in_kernel)
	cld			/* set direction flag to a known value */
	SAVE_PROCESS_CTX(0, KTS_INT_HARD)
	/* stop user process cycles */
	push	%ebp
	mov	$0, %ebp
	call	_C_LABEL(context_stop)
	call	_C_LABEL(copr_not_available_handler)
	/* reached upon failure only */
	jmp	_C_LABEL(switch_to_user)

copr_not_available_in_kernel:
	pushl	$0
	pushl	$COPROC_NOT_VECTOR
	jmp	exception_entry_nested

LABEL(double_fault)
	EXCEPTION_ERR_CODE(DOUBLE_FAULT_VECTOR)

LABEL(copr_seg_overrun)
	EXCEPTION_NO_ERR_CODE(COPROC_SEG_VECTOR)

LABEL(inval_tss)
	EXCEPTION_ERR_CODE(INVAL_TSS_VECTOR)

LABEL(segment_not_present)
	EXCEPTION_ERR_CODE(SEG_NOT_VECTOR)

LABEL(stack_exception)
	EXCEPTION_ERR_CODE(STACK_FAULT_VECTOR)

LABEL(general_protection)
	EXCEPTION_ERR_CODE(PROTECTION_VECTOR)

LABEL(page_fault)
	EXCEPTION_ERR_CODE(PAGE_FAULT_VECTOR)

LABEL(copr_error)
	EXCEPTION_NO_ERR_CODE(COPROC_ERR_VECTOR)

LABEL(alignment_check)
	EXCEPTION_NO_ERR_CODE(ALIGNMENT_CHECK_VECTOR)

LABEL(machine_check)
	EXCEPTION_NO_ERR_CODE(MACHINE_CHECK_VECTOR)

LABEL(simd_exception)
	EXCEPTION_NO_ERR_CODE(SIMD_EXCEPTION_VECTOR)

/*===========================================================================*/
/*				reload_cr3				     */
/*===========================================================================*/
/* PUBLIC void reload_cr3(void); */
ENTRY(reload_cr3)
	push    %ebp
	mov     %esp, %ebp
	mov	%cr3, %eax
	mov	%eax, %cr3
	pop     %ebp
	ret

#ifdef CONFIG_SMP
ENTRY(startup_ap_32)
	/*
	 * we are in protected mode now, %cs is correct and we need to set the
	 * data descriptors before we can touch anything
	 *
	 * first load the regular, highly mapped idt, gdt
	 */

	/*
	 * use the boot stack for now. The running CPUs are already using their
	 * own stack, the rest is still waiting to be booted
	 */
	movw	$KERN_DS_SELECTOR, %ax
	mov	%ax, %ds
	mov	%ax, %ss
	mov	$_C_LABEL(k_boot_stktop) - 4, %esp

	/* load the highly mapped idt, gdt, per-cpu tss */
	call	_C_LABEL(prot_load_selectors)

	jmp	_C_LABEL(smp_ap_boot)
	hlt
#endif

/*===========================================================================*/
/*				data					     */
/*===========================================================================*/

.data
.short	0x526F	/* this must be the first data entry (magic #) */

.bss
k_initial_stack:
.space	K_STACK_SIZE
LABEL(__k_unpaged_k_initial_stktop)

/*
 * the kernel stack
 */
k_boot_stack:
.space	K_STACK_SIZE	/* kernel stack */ /* FIXME use macro here */
LABEL(k_boot_stktop)	/* top of kernel stack */

.balign K_STACK_SIZE
LABEL(k_stacks_start)

/* two pages for each stack, one for data, other as a sandbox */
.space	2 * (K_STACK_SIZE * CONFIG_MAX_CPUS)

LABEL(k_stacks_end)

/* top of kernel stack */