vm86_32.c

linux 内核源代码

	return flags | (VEFLAGS & current->thread.v86mask);
}

static inline int is_revectored(int nr, struct revectored_struct * bitmap)
{
	__asm__ __volatile__("btl %2,%1\n\tsbbl %0,%0"
		:"=r" (nr)
		:"m" (*bitmap),"r" (nr));
	return nr;
}

#define val_byte(val, n) (((__u8 *)&val)[n])

#define pushb(base, ptr, val, err_label) \
	do { \
		__u8 __val = val; \
		ptr--; \
		if (put_user(__val, base + ptr) < 0) \
			goto err_label; \
	} while(0)

#define pushw(base, ptr, val, err_label) \
	do { \
		__u16 __val = val; \
		ptr--; \
		if (put_user(val_byte(__val, 1), base + ptr) < 0) \
			goto err_label; \
		ptr--; \
		if (put_user(val_byte(__val, 0), base + ptr) < 0) \
			goto err_label; \
	} while(0)

#define pushl(base, ptr, val, err_label) \
	do { \
		__u32 __val = val; \
		ptr--; \
		if (put_user(val_byte(__val, 3), base + ptr) < 0) \
			goto err_label; \
		ptr--; \
		if (put_user(val_byte(__val, 2), base + ptr) < 0) \
			goto err_label; \
		ptr--; \
		if (put_user(val_byte(__val, 1), base + ptr) < 0) \
			goto err_label; \
		ptr--; \
		if (put_user(val_byte(__val, 0), base + ptr) < 0) \
			goto err_label; \
	} while(0)

#define popb(base, ptr, err_label) \
	({ \
		__u8 __res; \
		if (get_user(__res, base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		__res; \
	})

#define popw(base, ptr, err_label) \
	({ \
		__u16 __res; \
		if (get_user(val_byte(__res, 0), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		if (get_user(val_byte(__res, 1), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		__res; \
	})

#define popl(base, ptr, err_label) \
	({ \
		__u32 __res; \
		if (get_user(val_byte(__res, 0), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		if (get_user(val_byte(__res, 1), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		if (get_user(val_byte(__res, 2), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		if (get_user(val_byte(__res, 3), base + ptr) < 0) \
			goto err_label; \
		ptr++; \
		__res; \
	})

/* There are so many possible reasons for this function to return
 * VM86_INTx, so adding another doesn't bother me. We can expect
 * userspace programs to be able to handle it. (Getting a problem
 * in userspace is always better than an Oops anyway.) [KD]
 */
static void do_int(struct kernel_vm86_regs *regs, int i,
    unsigned char __user * ssp, unsigned short sp)
{
	unsigned long __user *intr_ptr;
	unsigned long segoffs;

	if (regs->pt.xcs == BIOSSEG)
		goto cannot_handle;
	if (is_revectored(i, &KVM86->int_revectored))
		goto cannot_handle;
	if (i==0x21 && is_revectored(AH(regs),&KVM86->int21_revectored))
		goto cannot_handle;
	intr_ptr = (unsigned long __user *) (i << 2);
	if (get_user(segoffs, intr_ptr))
		goto cannot_handle;
	if ((segoffs >> 16) == BIOSSEG)
		goto cannot_handle;
	pushw(ssp, sp, get_vflags(regs), cannot_handle);
	pushw(ssp, sp, regs->pt.xcs, cannot_handle);
	pushw(ssp, sp, IP(regs), cannot_handle);
	regs->pt.xcs = segoffs >> 16;
	SP(regs) -= 6;
	IP(regs) = segoffs & 0xffff;
	clear_TF(regs);
	clear_IF(regs);
	clear_AC(regs);
	return;

cannot_handle:
	return_to_32bit(regs, VM86_INTx + (i << 8));
}

int handle_vm86_trap(struct kernel_vm86_regs * regs, long error_code, int trapno)
{
	if (VMPI.is_vm86pus) {
		if ( (trapno==3) || (trapno==1) )
			return_to_32bit(regs, VM86_TRAP + (trapno << 8));
		do_int(regs, trapno, (unsigned char __user *) (regs->pt.xss << 4), SP(regs));
		return 0;
	}
	if (trapno !=1)
		return 1; /* we let this handle by the calling routine */
	if (current->ptrace & PT_PTRACED) {
		unsigned long flags;
		spin_lock_irqsave(&current->sighand->siglock, flags);
		sigdelset(&current->blocked, SIGTRAP);
		recalc_sigpending();
		spin_unlock_irqrestore(&current->sighand->siglock, flags);
	}
	send_sig(SIGTRAP, current, 1);
	current->thread.trap_no = trapno;
	current->thread.error_code = error_code;
	return 0;
}

void handle_vm86_fault(struct kernel_vm86_regs * regs, long error_code)
{
	unsigned char opcode;
	unsigned char __user *csp;
	unsigned char __user *ssp;
	unsigned short ip, sp, orig_flags;
	int data32, pref_done;

#define CHECK_IF_IN_TRAP \
	if (VMPI.vm86dbg_active && VMPI.vm86dbg_TFpendig) \
		newflags |= TF_MASK
#define VM86_FAULT_RETURN do { \
	if (VMPI.force_return_for_pic  && (VEFLAGS & (IF_MASK | VIF_MASK))) \
		return_to_32bit(regs, VM86_PICRETURN); \
	if (orig_flags & TF_MASK) \
		handle_vm86_trap(regs, 0, 1); \
	return; } while (0)

	orig_flags = *(unsigned short *)&regs->pt.eflags;

	csp = (unsigned char __user *) (regs->pt.xcs << 4);
	ssp = (unsigned char __user *) (regs->pt.xss << 4);
	sp = SP(regs);
	ip = IP(regs);

	data32 = 0;
	pref_done = 0;
	do {
		switch (opcode = popb(csp, ip, simulate_sigsegv)) {
			case 0x66:      /* 32-bit data */     data32=1; break;
			case 0x67:      /* 32-bit address */  break;
			case 0x2e:      /* CS */              break;
			case 0x3e:      /* DS */              break;
			case 0x26:      /* ES */              break;
			case 0x36:      /* SS */              break;
			case 0x65:      /* GS */              break;
			case 0x64:      /* FS */              break;
			case 0xf2:      /* repnz */       break;
			case 0xf3:      /* rep */             break;
			default: pref_done = 1;
		}
	} while (!pref_done);

	switch (opcode) {

	/* pushf */
	case 0x9c:
		if (data32) {
			pushl(ssp, sp, get_vflags(regs), simulate_sigsegv);
			SP(regs) -= 4;
		} else {
			pushw(ssp, sp, get_vflags(regs), simulate_sigsegv);
			SP(regs) -= 2;
		}
		IP(regs) = ip;
		VM86_FAULT_RETURN;

	/* popf */
	case 0x9d:
		{
		unsigned long newflags;
		if (data32) {
			newflags=popl(ssp, sp, simulate_sigsegv);
			SP(regs) += 4;
		} else {
			newflags = popw(ssp, sp, simulate_sigsegv);
			SP(regs) += 2;
		}
		IP(regs) = ip;
		CHECK_IF_IN_TRAP;
		if (data32) {
			set_vflags_long(newflags, regs);
		} else {
			set_vflags_short(newflags, regs);
		}
		VM86_FAULT_RETURN;
		}

	/* int xx */
	case 0xcd: {
		int intno=popb(csp, ip, simulate_sigsegv);
		IP(regs) = ip;
		if (VMPI.vm86dbg_active) {
			if ( (1 << (intno &7)) & VMPI.vm86dbg_intxxtab[intno >> 3] )
				return_to_32bit(regs, VM86_INTx + (intno << 8));
		}
		do_int(regs, intno, ssp, sp);
		return;
	}

	/* iret */
	case 0xcf:
		{
		unsigned long newip;
		unsigned long newcs;
		unsigned long newflags;
		if (data32) {
			newip=popl(ssp, sp, simulate_sigsegv);
			newcs=popl(ssp, sp, simulate_sigsegv);
			newflags=popl(ssp, sp, simulate_sigsegv);
			SP(regs) += 12;
		} else {
			newip = popw(ssp, sp, simulate_sigsegv);
			newcs = popw(ssp, sp, simulate_sigsegv);
			newflags = popw(ssp, sp, simulate_sigsegv);
			SP(regs) += 6;
		}
		IP(regs) = newip;
		regs->pt.xcs = newcs;
		CHECK_IF_IN_TRAP;
		if (data32) {
			set_vflags_long(newflags, regs);
		} else {
			set_vflags_short(newflags, regs);
		}
		VM86_FAULT_RETURN;
		}

	/* cli */
	case 0xfa:
		IP(regs) = ip;
		clear_IF(regs);
		VM86_FAULT_RETURN;

	/* sti */
	/*
	 * Damn. This is incorrect: the 'sti' instruction should actually
	 * enable interrupts after the /next/ instruction. Not good.
	 *
	 * Probably needs some horsing around with the TF flag. Aiee..
	 */
	case 0xfb:
		IP(regs) = ip;
		set_IF(regs);
		VM86_FAULT_RETURN;

	default:
		return_to_32bit(regs, VM86_UNKNOWN);
	}

	return;

simulate_sigsegv:
	/* FIXME: After a long discussion with Stas we finally
	 *        agreed, that this is wrong. Here we should
	 *        really send a SIGSEGV to the user program.
	 *        But how do we create the correct context? We
	 *        are inside a general protection fault handler
	 *        and has just returned from a page fault handler.
	 *        The correct context for the signal handler
	 *        should be a mixture of the two, but how do we
	 *        get the information? [KD]
	 */
	return_to_32bit(regs, VM86_UNKNOWN);
}

/* ---------------- vm86 special IRQ passing stuff ----------------- */

#define VM86_IRQNAME		"vm86irq"

static struct vm86_irqs {
	struct task_struct *tsk;
	int sig;
} vm86_irqs[16];

static DEFINE_SPINLOCK(irqbits_lock);
static int irqbits;

#define ALLOWED_SIGS ( 1 /* 0 = don't send a signal */ \
	| (1 << SIGUSR1) | (1 << SIGUSR2) | (1 << SIGIO)  | (1 << SIGURG) \
	| (1 << SIGUNUSED) )
	
static irqreturn_t irq_handler(int intno, void *dev_id)
{
	int irq_bit;
	unsigned long flags;

	spin_lock_irqsave(&irqbits_lock, flags);	
	irq_bit = 1 << intno;
	if ((irqbits & irq_bit) || ! vm86_irqs[intno].tsk)
		goto out;
	irqbits |= irq_bit;
	if (vm86_irqs[intno].sig)
		send_sig(vm86_irqs[intno].sig, vm86_irqs[intno].tsk, 1);
	/*
	 * IRQ will be re-enabled when user asks for the irq (whether
	 * polling or as a result of the signal)
	 */
	disable_irq_nosync(intno);
	spin_unlock_irqrestore(&irqbits_lock, flags);
	return IRQ_HANDLED;

out:
	spin_unlock_irqrestore(&irqbits_lock, flags);	
	return IRQ_NONE;
}

static inline void free_vm86_irq(int irqnumber)
{
	unsigned long flags;

	free_irq(irqnumber, NULL);
	vm86_irqs[irqnumber].tsk = NULL;

	spin_lock_irqsave(&irqbits_lock, flags);	
	irqbits &= ~(1 << irqnumber);
	spin_unlock_irqrestore(&irqbits_lock, flags);	
}

void release_vm86_irqs(struct task_struct *task)
{
	int i;
	for (i = FIRST_VM86_IRQ ; i <= LAST_VM86_IRQ; i++)
	    if (vm86_irqs[i].tsk == task)
		free_vm86_irq(i);
}

static inline int get_and_reset_irq(int irqnumber)
{
	int bit;
	unsigned long flags;
	int ret = 0;
	
	if (invalid_vm86_irq(irqnumber)) return 0;
	if (vm86_irqs[irqnumber].tsk != current) return 0;
	spin_lock_irqsave(&irqbits_lock, flags);	
	bit = irqbits & (1 << irqnumber);
	irqbits &= ~bit;
	if (bit) {
		enable_irq(irqnumber);
		ret = 1;
	}

	spin_unlock_irqrestore(&irqbits_lock, flags);	
	return ret;
}


static int do_vm86_irq_handling(int subfunction, int irqnumber)
{
	int ret;
	switch (subfunction) {
		case VM86_GET_AND_RESET_IRQ: {
			return get_and_reset_irq(irqnumber);
		}
		case VM86_GET_IRQ_BITS: {
			return irqbits;
		}
		case VM86_REQUEST_IRQ: {
			int sig = irqnumber >> 8;
			int irq = irqnumber & 255;
			if (!capable(CAP_SYS_ADMIN)) return -EPERM;
			if (!((1 << sig) & ALLOWED_SIGS)) return -EPERM;
			if (invalid_vm86_irq(irq)) return -EPERM;
			if (vm86_irqs[irq].tsk) return -EPERM;
			ret = request_irq(irq, &irq_handler, 0, VM86_IRQNAME, NULL);
			if (ret) return ret;
			vm86_irqs[irq].sig = sig;
			vm86_irqs[irq].tsk = current;
			return irq;
		}
		case  VM86_FREE_IRQ: {
			if (invalid_vm86_irq(irqnumber)) return -EPERM;
			if (!vm86_irqs[irqnumber].tsk) return 0;
			if (vm86_irqs[irqnumber].tsk != current) return -EPERM;
			free_vm86_irq(irqnumber);
			return 0;
		}
	}
	return -EINVAL;
}