[GitHub/mt8127/android_kernel_alcatel_ttab.git] / arch / x86 / mm / fault_32.c

/*
 *  Copyright (C) 1995  Linus Torvalds
 */

#include <linux/signal.h>
#include <linux/sched.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/types.h>
#include <linux/ptrace.h>
#include <linux/mman.h>
#include <linux/mm.h>
#include <linux/smp.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/tty.h>
#include <linux/vt_kern.h>		/* For unblank_screen() */
#include <linux/highmem.h>
#include <linux/bootmem.h>		/* for max_low_pfn */
#include <linux/vmalloc.h>
#include <linux/module.h>
#include <linux/kprobes.h>
#include <linux/uaccess.h>
#include <linux/kdebug.h>

#include <asm/system.h>
#include <asm/desc.h>
#include <asm/segment.h>

/*
 * Page fault error code bits
 *	bit 0 == 0 means no page found, 1 means protection fault
 *	bit 1 == 0 means read, 1 means write
 *	bit 2 == 0 means kernel, 1 means user-mode
 *	bit 3 == 1 means use of reserved bit detected
 *	bit 4 == 1 means fault was an instruction fetch
 */
#define PF_PROT	(1<<0)
#define PF_WRITE	(1<<1)
#define PF_USER	(1<<2)
#define PF_RSVD	(1<<3)
#define PF_INSTR	(1<<4)

static inline int notify_page_fault(struct pt_regs *regs)
{
#ifdef CONFIG_KPROBES
	int ret = 0;

	/* kprobe_running() needs smp_processor_id() */
	if (!user_mode_vm(regs)) {
		preempt_disable();
		if (kprobe_running() && kprobe_fault_handler(regs, 14))
			ret = 1;
		preempt_enable();
	}

	return ret;
#else
	return 0;
#endif
}

/*
 * X86_32
 * Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch.
 * Check that here and ignore it.
 *
 * X86_64
 * Sometimes the CPU reports invalid exceptions on prefetch.
 * Check that here and ignore it.
 *
 * Opcode checker based on code by Richard Brunner
 */
static int is_prefetch(struct pt_regs *regs, unsigned long addr,
		       unsigned long error_code)
{
	unsigned char *instr;
	int scan_more = 1;
	int prefetch = 0;
	unsigned char *max_instr;

#ifdef CONFIG_X86_32
	unsigned long limit;
	if (unlikely(boot_cpu_data.x86_vendor == X86_VENDOR_AMD &&
		     boot_cpu_data.x86 >= 6)) {
		/* Catch an obscure case of prefetch inside an NX page. */
		if (nx_enabled && (error_code & PF_INSTR))
			return 0;
	} else {
		return 0;
	}
	instr = (unsigned char *)get_segment_eip(regs, &limit);
#else
	/* If it was a exec fault ignore */
	if (error_code & PF_INSTR)
		return 0;
	instr = (unsigned char __user *)convert_ip_to_linear(current, regs);
#endif

	max_instr = instr + 15;

#ifdef CONFIG_X86_64
	if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE)
		return 0;
#endif

	while (scan_more && instr < max_instr) {
		unsigned char opcode;
		unsigned char instr_hi;
		unsigned char instr_lo;

#ifdef CONFIG_X86_32
		if (instr > (unsigned char *)limit)
			break;
#endif
		if (probe_kernel_address(instr, opcode))
			break;

		instr_hi = opcode & 0xf0;
		instr_lo = opcode & 0x0f;
		instr++;

		switch (instr_hi) {
		case 0x20:
		case 0x30:
			/*
			 * Values 0x26,0x2E,0x36,0x3E are valid x86 prefixes.
			 * In X86_64 long mode, the CPU will signal invalid
			 * opcode if some of these prefixes are present so
			 * X86_64 will never get here anyway
			 */
			scan_more = ((instr_lo & 7) == 0x6);
			break;
#ifdef CONFIG_X86_64
		case 0x40:
			/*
			 * In AMD64 long mode 0x40..0x4F are valid REX prefixes
			 * Need to figure out under what instruction mode the
			 * instruction was issued. Could check the LDT for lm,
			 * but for now it's good enough to assume that long
			 * mode only uses well known segments or kernel.
			 */
			scan_more = (!user_mode(regs)) || (regs->cs == __USER_CS);
			break;
#endif
		case 0x60:
			/* 0x64 thru 0x67 are valid prefixes in all modes. */
			scan_more = (instr_lo & 0xC) == 0x4;
			break;
		case 0xF0:
			/* 0xF0, 0xF2, 0xF3 are valid prefixes in all modes. */
			scan_more = !instr_lo || (instr_lo>>1) == 1;
			break;
		case 0x00:
			/* Prefetch instruction is 0x0F0D or 0x0F18 */
			scan_more = 0;
#ifdef CONFIG_X86_32
			if (instr > (unsigned char *)limit)
				break;
#endif
			if (probe_kernel_address(instr, opcode))
				break;
			prefetch = (instr_lo == 0xF) &&
				(opcode == 0x0D || opcode == 0x18);
			break;
		default:
			scan_more = 0;
			break;
		}
	}
	return prefetch;
}

static void force_sig_info_fault(int si_signo, int si_code,
	unsigned long address, struct task_struct *tsk)
{
	siginfo_t info;

	info.si_signo = si_signo;
	info.si_errno = 0;
	info.si_code = si_code;
	info.si_addr = (void __user *)address;
	force_sig_info(si_signo, &info, tsk);
}

void do_invalid_op(struct pt_regs *, unsigned long);

static inline pmd_t *vmalloc_sync_one(pgd_t *pgd, unsigned long address)
{
	unsigned index = pgd_index(address);
	pgd_t *pgd_k;
	pud_t *pud, *pud_k;
	pmd_t *pmd, *pmd_k;

	pgd += index;
	pgd_k = init_mm.pgd + index;

	if (!pgd_present(*pgd_k))
		return NULL;

	/*
	 * set_pgd(pgd, *pgd_k); here would be useless on PAE
	 * and redundant with the set_pmd() on non-PAE. As would
	 * set_pud.
	 */

	pud = pud_offset(pgd, address);
	pud_k = pud_offset(pgd_k, address);
	if (!pud_present(*pud_k))
		return NULL;

	pmd = pmd_offset(pud, address);
	pmd_k = pmd_offset(pud_k, address);
	if (!pmd_present(*pmd_k))
		return NULL;
	if (!pmd_present(*pmd)) {
		set_pmd(pmd, *pmd_k);
		arch_flush_lazy_mmu_mode();
	} else
		BUG_ON(pmd_page(*pmd) != pmd_page(*pmd_k));
	return pmd_k;
}

#ifdef CONFIG_X86_64
static const char errata93_warning[] =
KERN_ERR "******* Your BIOS seems to not contain a fix for K8 errata #93\n"
KERN_ERR "******* Working around it, but it may cause SEGVs or burn power.\n"
KERN_ERR "******* Please consider a BIOS update.\n"
KERN_ERR "******* Disabling USB legacy in the BIOS may also help.\n";

/* Workaround for K8 erratum #93 & buggy BIOS.
   BIOS SMM functions are required to use a specific workaround
   to avoid corruption of the 64bit RIP register on C stepping K8.
   A lot of BIOS that didn't get tested properly miss this.
   The OS sees this as a page fault with the upper 32bits of RIP cleared.
   Try to work around it here.
   Note we only handle faults in kernel here. */

static int is_errata93(struct pt_regs *regs, unsigned long address)
{
	static int warned;
	if (address != regs->ip)
		return 0;
	if ((address >> 32) != 0)
		return 0;
	address |= 0xffffffffUL << 32;
	if ((address >= (u64)_stext && address <= (u64)_etext) ||
	    (address >= MODULES_VADDR && address <= MODULES_END)) {
		if (!warned) {
			printk(errata93_warning);
			warned = 1;
		}
		regs->ip = address;
		return 1;
	}
	return 0;
}
#endif

/*
 * Handle a fault on the vmalloc or module mapping area
 *
 * This assumes no large pages in there.
 */
static inline int vmalloc_fault(unsigned long address)
{
	unsigned long pgd_paddr;
	pmd_t *pmd_k;
	pte_t *pte_k;
	/*
	 * Synchronize this task's top level page-table
	 * with the 'reference' page table.
	 *
	 * Do _not_ use "current" here. We might be inside
	 * an interrupt in the middle of a task switch..
	 */
	pgd_paddr = read_cr3();
	pmd_k = vmalloc_sync_one(__va(pgd_paddr), address);
	if (!pmd_k)
		return -1;
	pte_k = pte_offset_kernel(pmd_k, address);
	if (!pte_present(*pte_k))
		return -1;
	return 0;
}

int show_unhandled_signals = 1;

/*
 * This routine handles page faults.  It determines the address,
 * and the problem, and then passes it off to one of the appropriate
 * routines.
 */
void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
{
	struct task_struct *tsk;
	struct mm_struct *mm;
	struct vm_area_struct *vma;
	unsigned long address;
	int write, si_code;
	int fault;

	/*
	 * We can fault from pretty much anywhere, with unknown IRQ state.
	 */
	trace_hardirqs_fixup();

	/* get the address */
	address = read_cr2();

	tsk = current;

	si_code = SEGV_MAPERR;

	/*
	 * We fault-in kernel-space virtual memory on-demand. The
	 * 'reference' page table is init_mm.pgd.
	 *
	 * NOTE! We MUST NOT take any locks for this case. We may
	 * be in an interrupt or a critical region, and should
	 * only copy the information from the master page table,
	 * nothing more.
	 *
	 * This verifies that the fault happens in kernel space
	 * (error_code & 4) == 0, and that the fault was not a
	 * protection error (error_code & 9) == 0.
	 */
	if (unlikely(address >= TASK_SIZE)) {
		if (!(error_code & (PF_RSVD|PF_USER|PF_PROT)) &&
		    vmalloc_fault(address) >= 0)
			return;
		if (notify_page_fault(regs))
			return;
		/*
		 * Don't take the mm semaphore here. If we fixup a prefetch
		 * fault we could otherwise deadlock.
		 */
		goto bad_area_nosemaphore;
	}

	if (notify_page_fault(regs))
		return;

	/* It's safe to allow irq's after cr2 has been saved and the vmalloc
	   fault has been handled. */
	if (regs->flags & (X86_EFLAGS_IF|VM_MASK))
		local_irq_enable();

	mm = tsk->mm;

	/*
	 * If we're in an interrupt, have no user context or are running in an
	 * atomic region then we must not take the fault.
	 */
	if (in_atomic() || !mm)
		goto bad_area_nosemaphore;

	/* When running in the kernel we expect faults to occur only to
	 * addresses in user space.  All other faults represent errors in the
	 * kernel and should generate an OOPS.  Unfortunately, in the case of an
	 * erroneous fault occurring in a code path which already holds mmap_sem
	 * we will deadlock attempting to validate the fault against the
	 * address space.  Luckily the kernel only validly references user
	 * space from well defined areas of code, which are listed in the
	 * exceptions table.
	 *
	 * As the vast majority of faults will be valid we will only perform
	 * the source reference check when there is a possibility of a deadlock.
	 * Attempt to lock the address space, if we cannot we then validate the
	 * source.  If this is invalid we can skip the address space check,
	 * thus avoiding the deadlock.
	 */
	if (!down_read_trylock(&mm->mmap_sem)) {
		if ((error_code & PF_USER) == 0 &&
		    !search_exception_tables(regs->ip))
			goto bad_area_nosemaphore;
		down_read(&mm->mmap_sem);
	}

	vma = find_vma(mm, address);
	if (!vma)
		goto bad_area;
	if (vma->vm_start <= address)
		goto good_area;
	if (!(vma->vm_flags & VM_GROWSDOWN))
		goto bad_area;
	if (error_code & PF_USER) {
		/*
		 * Accessing the stack below %sp is always a bug.
		 * The large cushion allows instructions like enter
		 * and pusha to work.  ("enter $65535,$31" pushes
		 * 32 pointers and then decrements %sp by 65535.)
		 */
		if (address + 65536 + 32 * sizeof(unsigned long) < regs->sp)
			goto bad_area;
	}
	if (expand_stack(vma, address))
		goto bad_area;
/*
 * Ok, we have a good vm_area for this memory access, so
 * we can handle it..
 */
good_area:
	si_code = SEGV_ACCERR;
	write = 0;
	switch (error_code & (PF_PROT|PF_WRITE)) {
	default:	/* 3: write, present */
		/* fall through */
	case PF_WRITE:		/* write, not present */
		if (!(vma->vm_flags & VM_WRITE))
			goto bad_area;
		write++;
		break;
	case PF_PROT:		/* read, present */
		goto bad_area;
	case 0:			/* read, not present */
		if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)))
			goto bad_area;
	}

 survive:
	/*
	 * If for any reason at all we couldn't handle the fault,
	 * make sure we exit gracefully rather than endlessly redo
	 * the fault.
	 */
	fault = handle_mm_fault(mm, vma, address, write);
	if (unlikely(fault & VM_FAULT_ERROR)) {
		if (fault & VM_FAULT_OOM)
			goto out_of_memory;
		else if (fault & VM_FAULT_SIGBUS)
			goto do_sigbus;
		BUG();
	}
	if (fault & VM_FAULT_MAJOR)
		tsk->maj_flt++;
	else
		tsk->min_flt++;

	/*
	 * Did it hit the DOS screen memory VA from vm86 mode?
	 */
	if (regs->flags & VM_MASK) {
		unsigned long bit = (address - 0xA0000) >> PAGE_SHIFT;
		if (bit < 32)
			tsk->thread.screen_bitmap |= 1 << bit;
	}
	up_read(&mm->mmap_sem);
	return;

/*
 * Something tried to access memory that isn't in our memory map..
 * Fix it, but check if it's kernel or user first..
 */
bad_area:
	up_read(&mm->mmap_sem);

bad_area_nosemaphore:
	/* User mode accesses just cause a SIGSEGV */
	if (error_code & PF_USER) {
		/*
		 * It's possible to have interrupts off here.
		 */
		local_irq_enable();

		/*
		 * Valid to do another page fault here because this one came
		 * from user space.
		 */
		if (is_prefetch(regs, address, error_code))
			return;

		if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
		    printk_ratelimit()) {
			printk("%s%s[%d]: segfault at %08lx ip %08lx "
			    "sp %08lx error %lx\n",
			    task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
			    tsk->comm, task_pid_nr(tsk), address, regs->ip,
			    regs->sp, error_code);
		}
		tsk->thread.cr2 = address;
		/* Kernel addresses are always protection faults */
		tsk->thread.error_code = error_code | (address >= TASK_SIZE);
		tsk->thread.trap_no = 14;
		force_sig_info_fault(SIGSEGV, si_code, address, tsk);
		return;
	}

#ifdef CONFIG_X86_F00F_BUG
	/*
	 * Pentium F0 0F C7 C8 bug workaround.
	 */
	if (boot_cpu_data.f00f_bug) {
		unsigned long nr;

		nr = (address - idt_descr.address) >> 3;

		if (nr == 6) {
			do_invalid_op(regs, 0);
			return;
		}
	}
#endif

no_context:
	/* Are we prepared to handle this kernel fault?  */
	if (fixup_exception(regs))
		return;

	/*
	 * Valid to do another page fault here, because if this fault
	 * had been triggered by is_prefetch fixup_exception would have
	 * handled it.
	 */
	if (is_prefetch(regs, address, error_code))
		return;

/*
 * Oops. The kernel tried to access some bad page. We'll have to
 * terminate things with extreme prejudice.
 */

	bust_spinlocks(1);

	if (oops_may_print()) {
		__typeof__(pte_val(__pte(0))) page;

#ifdef CONFIG_X86_PAE
		if (error_code & PF_INSTR) {
			pte_t *pte = lookup_address(address);

			if (pte && pte_present(*pte) && !pte_exec_kernel(*pte))
				printk(KERN_CRIT "kernel tried to execute "
					"NX-protected page - exploit attempt? "
					"(uid: %d)\n", current->uid);
		}
#endif
		if (address < PAGE_SIZE)
			printk(KERN_ALERT "BUG: unable to handle kernel NULL "
					"pointer dereference");
		else
			printk(KERN_ALERT "BUG: unable to handle kernel paging"
					" request");
		printk(" at virtual address %08lx\n", address);
		printk(KERN_ALERT "printing ip: %08lx ", regs->ip);

		page = read_cr3();
		page = ((__typeof__(page) *) __va(page))[address >> PGDIR_SHIFT];
#ifdef CONFIG_X86_PAE
		printk("*pdpt = %016Lx ", page);
		if ((page >> PAGE_SHIFT) < max_low_pfn
		    && page & _PAGE_PRESENT) {
			page &= PAGE_MASK;
			page = ((__typeof__(page) *) __va(page))[(address >> PMD_SHIFT)
			                                         & (PTRS_PER_PMD - 1)];
			printk(KERN_CONT "*pde = %016Lx ", page);
			page &= ~_PAGE_NX;
		}
#else
		printk("*pde = %08lx ", page);
#endif

		/*
		 * We must not directly access the pte in the highpte
		 * case if the page table is located in highmem.
		 * And let's rather not kmap-atomic the pte, just in case
		 * it's allocated already.
		 */
		if ((page >> PAGE_SHIFT) < max_low_pfn
		    && (page & _PAGE_PRESENT)
		    && !(page & _PAGE_PSE)) {
			page &= PAGE_MASK;
			page = ((__typeof__(page) *) __va(page))[(address >> PAGE_SHIFT)
			                                         & (PTRS_PER_PTE - 1)];
			printk("*pte = %0*Lx ", sizeof(page)*2, (u64)page);
		}

		printk("\n");
	}

	tsk->thread.cr2 = address;
	tsk->thread.trap_no = 14;
	tsk->thread.error_code = error_code;
	die("Oops", regs, error_code);
	bust_spinlocks(0);
	do_exit(SIGKILL);

/*
 * We ran out of memory, or some other thing happened to us that made
 * us unable to handle the page fault gracefully.
 */
out_of_memory:
	up_read(&mm->mmap_sem);
	if (is_global_init(tsk)) {
		yield();
		down_read(&mm->mmap_sem);
		goto survive;
	}
	printk("VM: killing process %s\n", tsk->comm);
	if (error_code & PF_USER)
		do_group_exit(SIGKILL);
	goto no_context;

do_sigbus:
	up_read(&mm->mmap_sem);

	/* Kernel mode? Handle exceptions or die */
	if (!(error_code & PF_USER))
		goto no_context;

	/* User space => ok to do another page fault */
	if (is_prefetch(regs, address, error_code))
		return;

	tsk->thread.cr2 = address;
	tsk->thread.error_code = error_code;
	tsk->thread.trap_no = 14;
	force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk);
}

void vmalloc_sync_all(void)
{
	/*
	 * Note that races in the updates of insync and start aren't
	 * problematic: insync can only get set bits added, and updates to
	 * start are only improving performance (without affecting correctness
	 * if undone).
	 */
	static DECLARE_BITMAP(insync, PTRS_PER_PGD);
	static unsigned long start = TASK_SIZE;
	unsigned long address;

	if (SHARED_KERNEL_PMD)
		return;

	BUILD_BUG_ON(TASK_SIZE & ~PGDIR_MASK);
	for (address = start; address >= TASK_SIZE; address += PGDIR_SIZE) {
		if (!test_bit(pgd_index(address), insync)) {
			unsigned long flags;
			struct page *page;

			spin_lock_irqsave(&pgd_lock, flags);
			for (page = pgd_list; page; page =
					(struct page *)page->index)
				if (!vmalloc_sync_one(page_address(page),
								address)) {
					BUG_ON(page != pgd_list);
					break;
				}
			spin_unlock_irqrestore(&pgd_lock, flags);
			if (!page)
				set_bit(pgd_index(address), insync);
		}
		if (address == start && test_bit(pgd_index(address), insync))
			start = address + PGDIR_SIZE;
	}
}
Commit	Line	Data
1da177e4	1	/*
1da177e4 LT	2	* Copyright (C) 1995 Linus Torvalds
	3	*/
	4
	5	#include <linux/signal.h>
	6	#include <linux/sched.h>
	7	#include <linux/kernel.h>
	8	#include <linux/errno.h>
	9	#include <linux/string.h>
	10	#include <linux/types.h>
	11	#include <linux/ptrace.h>
	12	#include <linux/mman.h>
	13	#include <linux/mm.h>
	14	#include <linux/smp.h>
1da177e4 LT	15	#include <linux/interrupt.h>
	16	#include <linux/init.h>
	17	#include <linux/tty.h>
	18	#include <linux/vt_kern.h> /* For unblank_screen() */
	19	#include <linux/highmem.h>
28609f6e	20	#include <linux/bootmem.h> /* for max_low_pfn */
1eeb66a1	21	#include <linux/vmalloc.h>
1da177e4	22	#include <linux/module.h>
3d97ae5b	23	#include <linux/kprobes.h>
11a4180c	24	#include <linux/uaccess.h>
1eeb66a1	25	#include <linux/kdebug.h>
1da177e4 LT	26
1da177e4 LT	27	#include <asm/system.h>
1da177e4	28	#include <asm/desc.h>
78be3706	29	#include <asm/segment.h>
1da177e4	30
33cb5243 HH	31	/*
	32	* Page fault error code bits
	33	* bit 0 == 0 means no page found, 1 means protection fault
	34	* bit 1 == 0 means read, 1 means write
	35	* bit 2 == 0 means kernel, 1 means user-mode
	36	* bit 3 == 1 means use of reserved bit detected
	37	* bit 4 == 1 means fault was an instruction fetch
	38	*/
	39	#define PF_PROT (1<<0)
	40	#define PF_WRITE (1<<1)
	41	#define PF_USER (1<<2)
	42	#define PF_RSVD (1<<3)
	43	#define PF_INSTR (1<<4)
	44
74a0b576	45	static inline int notify_page_fault(struct pt_regs *regs)
b71b5b65	46	{
33cb5243	47	#ifdef CONFIG_KPROBES
74a0b576 CH	48	int ret = 0;
	49
	50	/* kprobe_running() needs smp_processor_id() */
	51	if (!user_mode_vm(regs)) {
	52	preempt_disable();
	53	if (kprobe_running() && kprobe_fault_handler(regs, 14))
	54	ret = 1;
	55	preempt_enable();
	56	}
b71b5b65	57
74a0b576	58	return ret;
74a0b576	59	#else
74a0b576	60	return 0;
74a0b576	61	#endif
33cb5243	62	}
b71b5b65	63
33cb5243	64	/*
1dc85be0	65	* X86_32
1da177e4 LT	66	* Sometimes AMD Athlon/Opteron CPUs report invalid exceptions on prefetch.
1da177e4 LT	67	* Check that here and ignore it.
1dc85be0 HH	68	*
	69	* X86_64
	70	* Sometimes the CPU reports invalid exceptions on prefetch.
	71	* Check that here and ignore it.
	72	*
	73	* Opcode checker based on code by Richard Brunner
1da177e4	74	*/
1dc85be0 HH	75	static int is_prefetch(struct pt_regs *regs, unsigned long addr,
1dc85be0 HH	76	unsigned long error_code)
33cb5243	77	{
1dc85be0	78	unsigned char *instr;
1da177e4	79	int scan_more = 1;
33cb5243	80	int prefetch = 0;
1dc85be0 HH	81	unsigned char *max_instr;
	82
	83	#ifdef CONFIG_X86_32
	84	unsigned long limit;
	85	if (unlikely(boot_cpu_data.x86_vendor == X86_VENDOR_AMD &&
	86	boot_cpu_data.x86 >= 6)) {
	87	/* Catch an obscure case of prefetch inside an NX page. */
	88	if (nx_enabled && (error_code & PF_INSTR))
	89	return 0;
	90	} else {
	91	return 0;
	92	}
	93	instr = (unsigned char *)get_segment_eip(regs, &limit);
	94	#else
	95	/* If it was a exec fault ignore */
	96	if (error_code & PF_INSTR)
	97	return 0;
37cd9cf3	98	instr = (unsigned char __user *)convert_ip_to_linear(current, regs);
1dc85be0	99	#endif
1da177e4	100
1dc85be0 HH	101	max_instr = instr + 15;
	102
	103	#ifdef CONFIG_X86_64
	104	if (user_mode(regs) && instr >= (unsigned char *)TASK_SIZE)
	105	return 0;
	106	#endif
	107
	108	while (scan_more && instr < max_instr) {
1da177e4 LT	109	unsigned char opcode;
	110	unsigned char instr_hi;
	111	unsigned char instr_lo;
	112
1dc85be0	113	#ifdef CONFIG_X86_32
11a4180c	114	if (instr > (unsigned char *)limit)
1da177e4	115	break;
1dc85be0	116	#endif
11a4180c	117	if (probe_kernel_address(instr, opcode))
33cb5243	118	break;
1da177e4	119
33cb5243 HH	120	instr_hi = opcode & 0xf0;
33cb5243 HH	121	instr_lo = opcode & 0x0f;
1da177e4 LT	122	instr++;
1da177e4 LT	123
33cb5243	124	switch (instr_hi) {
1da177e4 LT	125	case 0x20:
1da177e4 LT	126	case 0x30:
33cb5243 HH	127	/*
	128	* Values 0x26,0x2E,0x36,0x3E are valid x86 prefixes.
	129	* In X86_64 long mode, the CPU will signal invalid
	130	* opcode if some of these prefixes are present so
	131	* X86_64 will never get here anyway
	132	*/
1da177e4 LT	133	scan_more = ((instr_lo & 7) == 0x6);
1da177e4 LT	134	break;
33cb5243 HH	135	#ifdef CONFIG_X86_64
	136	case 0x40:
	137	/*
	138	* In AMD64 long mode 0x40..0x4F are valid REX prefixes
	139	* Need to figure out under what instruction mode the
	140	* instruction was issued. Could check the LDT for lm,
	141	* but for now it's good enough to assume that long
	142	* mode only uses well known segments or kernel.
	143	*/
	144	scan_more = (!user_mode(regs)) \|\| (regs->cs == __USER_CS);
	145	break;
	146	#endif
1da177e4 LT	147	case 0x60:
	148	/* 0x64 thru 0x67 are valid prefixes in all modes. */
	149	scan_more = (instr_lo & 0xC) == 0x4;
33cb5243	150	break;
1da177e4	151	case 0xF0:
33cb5243	152	/* 0xF0, 0xF2, 0xF3 are valid prefixes in all modes. */
1da177e4	153	scan_more = !instr_lo \|\| (instr_lo>>1) == 1;
33cb5243	154	break;
1da177e4 LT	155	case 0x00:
	156	/* Prefetch instruction is 0x0F0D or 0x0F18 */
	157	scan_more = 0;
1dc85be0	158	#ifdef CONFIG_X86_32
11a4180c	159	if (instr > (unsigned char *)limit)
1da177e4	160	break;
1dc85be0	161	#endif
11a4180c	162	if (probe_kernel_address(instr, opcode))
1da177e4 LT	163	break;
	164	prefetch = (instr_lo == 0xF) &&
	165	(opcode == 0x0D \|\| opcode == 0x18);
33cb5243	166	break;
1da177e4 LT	167	default:
	168	scan_more = 0;
	169	break;
33cb5243	170	}
1da177e4 LT	171	}
	172	return prefetch;
	173	}
	174
c4aba4a8	175	static void force_sig_info_fault(int si_signo, int si_code,
869f96a0 IM	176	unsigned long address, struct task_struct *tsk)
	177	{
	178	siginfo_t info;
	179
	180	info.si_signo = si_signo;
	181	info.si_errno = 0;
	182	info.si_code = si_code;
	183	info.si_addr = (void __user *)address;
	184	force_sig_info(si_signo, &info, tsk);
	185	}
	186
75604d7f	187	void do_invalid_op(struct pt_regs *, unsigned long);
1da177e4	188
101f12af JB	189	static inline pmd_t vmalloc_sync_one(pgd_t pgd, unsigned long address)
	190	{
	191	unsigned index = pgd_index(address);
	192	pgd_t *pgd_k;
	193	pud_t pud, pud_k;
	194	pmd_t pmd, pmd_k;
	195
	196	pgd += index;
	197	pgd_k = init_mm.pgd + index;
	198
	199	if (!pgd_present(*pgd_k))
	200	return NULL;
	201
	202	/*
	203	* set_pgd(pgd, *pgd_k); here would be useless on PAE
	204	* and redundant with the set_pmd() on non-PAE. As would
	205	* set_pud.
	206	*/
	207
	208	pud = pud_offset(pgd, address);
	209	pud_k = pud_offset(pgd_k, address);
	210	if (!pud_present(*pud_k))
	211	return NULL;
	212
	213	pmd = pmd_offset(pud, address);
	214	pmd_k = pmd_offset(pud_k, address);
	215	if (!pmd_present(*pmd_k))
	216	return NULL;
8b14cb99	217	if (!pmd_present(*pmd)) {
101f12af	218	set_pmd(pmd, *pmd_k);
8b14cb99 ZA	219	arch_flush_lazy_mmu_mode();
8b14cb99 ZA	220	} else
101f12af JB	221	BUG_ON(pmd_page(pmd) != pmd_page(pmd_k));
	222	return pmd_k;
	223	}
	224
1dc85be0 HH	225	#ifdef CONFIG_X86_64
	226	static const char errata93_warning[] =
	227	KERN_ERR "******* Your BIOS seems to not contain a fix for K8 errata #93\n"
	228	KERN_ERR "******* Working around it, but it may cause SEGVs or burn power.\n"
	229	KERN_ERR "******* Please consider a BIOS update.\n"
	230	KERN_ERR "******* Disabling USB legacy in the BIOS may also help.\n";
	231
	232	/* Workaround for K8 erratum #93 & buggy BIOS.
	233	BIOS SMM functions are required to use a specific workaround
	234	to avoid corruption of the 64bit RIP register on C stepping K8.
	235	A lot of BIOS that didn't get tested properly miss this.
	236	The OS sees this as a page fault with the upper 32bits of RIP cleared.
	237	Try to work around it here.
	238	Note we only handle faults in kernel here. */
	239
	240	static int is_errata93(struct pt_regs *regs, unsigned long address)
	241	{
	242	static int warned;
	243	if (address != regs->ip)
	244	return 0;
	245	if ((address >> 32) != 0)
	246	return 0;
	247	address \|= 0xffffffffUL << 32;
	248	if ((address >= (u64)_stext && address <= (u64)_etext) \|\|
	249	(address >= MODULES_VADDR && address <= MODULES_END)) {
	250	if (!warned) {
	251	printk(errata93_warning);
	252	warned = 1;
	253	}
	254	regs->ip = address;
	255	return 1;
	256	}
	257	return 0;
	258	}
	259	#endif
	260
101f12af JB	261	/*
	262	* Handle a fault on the vmalloc or module mapping area
	263	*
	264	* This assumes no large pages in there.
	265	*/
	266	static inline int vmalloc_fault(unsigned long address)
	267	{
	268	unsigned long pgd_paddr;
	269	pmd_t *pmd_k;
	270	pte_t *pte_k;
	271	/*
	272	* Synchronize this task's top level page-table
	273	* with the 'reference' page table.
	274	*
	275	* Do _not_ use "current" here. We might be inside
	276	* an interrupt in the middle of a task switch..
	277	*/
	278	pgd_paddr = read_cr3();
	279	pmd_k = vmalloc_sync_one(__va(pgd_paddr), address);
	280	if (!pmd_k)
	281	return -1;
	282	pte_k = pte_offset_kernel(pmd_k, address);
	283	if (!pte_present(*pte_k))
	284	return -1;
	285	return 0;
	286	}
	287
abd4f750 MAS	288	int show_unhandled_signals = 1;
abd4f750 MAS	289
1da177e4 LT	290	/*
	291	* This routine handles page faults. It determines the address,
	292	* and the problem, and then passes it off to one of the appropriate
	293	* routines.
1da177e4	294	*/
75604d7f	295	void __kprobes do_page_fault(struct pt_regs *regs, unsigned long error_code)
1da177e4 LT	296	{
	297	struct task_struct *tsk;
	298	struct mm_struct *mm;
33cb5243	299	struct vm_area_struct *vma;
1da177e4	300	unsigned long address;
869f96a0	301	int write, si_code;
83c54070	302	int fault;
1da177e4	303
143a5d32 PZ	304	/*
	305	* We can fault from pretty much anywhere, with unknown IRQ state.
	306	*/
	307	trace_hardirqs_fixup();
	308
1da177e4	309	/* get the address */
33cb5243	310	address = read_cr2();
1da177e4	311
1da177e4 LT	312	tsk = current;
1da177e4 LT	313
869f96a0	314	si_code = SEGV_MAPERR;
1da177e4 LT	315
	316	/*
	317	* We fault-in kernel-space virtual memory on-demand. The
	318	* 'reference' page table is init_mm.pgd.
	319	*
	320	* NOTE! We MUST NOT take any locks for this case. We may
	321	* be in an interrupt or a critical region, and should
	322	* only copy the information from the master page table,
	323	* nothing more.
	324	*
	325	* This verifies that the fault happens in kernel space
	326	* (error_code & 4) == 0, and that the fault was not a
101f12af	327	* protection error (error_code & 9) == 0.
1da177e4	328	*/
101f12af	329	if (unlikely(address >= TASK_SIZE)) {
318aa296 HH	330	if (!(error_code & (PF_RSVD\|PF_USER\|PF_PROT)) &&
318aa296 HH	331	vmalloc_fault(address) >= 0)
101f12af	332	return;
74a0b576	333	if (notify_page_fault(regs))
101f12af JB	334	return;
101f12af JB	335	/*
1da177e4 LT	336	* Don't take the mm semaphore here. If we fixup a prefetch
	337	* fault we could otherwise deadlock.
	338	*/
	339	goto bad_area_nosemaphore;
101f12af JB	340	}
101f12af JB	341
74a0b576	342	if (notify_page_fault(regs))
101f12af JB	343	return;
	344
	345	/* It's safe to allow irq's after cr2 has been saved and the vmalloc
	346	fault has been handled. */
65ea5b03	347	if (regs->flags & (X86_EFLAGS_IF\|VM_MASK))
101f12af	348	local_irq_enable();
1da177e4 LT	349
	350	mm = tsk->mm;
	351
	352	/*
	353	* If we're in an interrupt, have no user context or are running in an
33cb5243	354	* atomic region then we must not take the fault.
1da177e4 LT	355	*/
	356	if (in_atomic() \|\| !mm)
	357	goto bad_area_nosemaphore;
	358
	359	/* When running in the kernel we expect faults to occur only to
	360	* addresses in user space. All other faults represent errors in the
27b46d76	361	* kernel and should generate an OOPS. Unfortunately, in the case of an
80f7228b	362	* erroneous fault occurring in a code path which already holds mmap_sem
1da177e4 LT	363	* we will deadlock attempting to validate the fault against the
	364	* address space. Luckily the kernel only validly references user
	365	* space from well defined areas of code, which are listed in the
	366	* exceptions table.
	367	*
	368	* As the vast majority of faults will be valid we will only perform
27b46d76	369	* the source reference check when there is a possibility of a deadlock.
1da177e4 LT	370	* Attempt to lock the address space, if we cannot we then validate the
	371	* source. If this is invalid we can skip the address space check,
	372	* thus avoiding the deadlock.
	373	*/
	374	if (!down_read_trylock(&mm->mmap_sem)) {
33cb5243	375	if ((error_code & PF_USER) == 0 &&
65ea5b03	376	!search_exception_tables(regs->ip))
1da177e4 LT	377	goto bad_area_nosemaphore;
	378	down_read(&mm->mmap_sem);
	379	}
	380
	381	vma = find_vma(mm, address);
	382	if (!vma)
	383	goto bad_area;
	384	if (vma->vm_start <= address)
	385	goto good_area;
	386	if (!(vma->vm_flags & VM_GROWSDOWN))
	387	goto bad_area;
33cb5243	388	if (error_code & PF_USER) {
1da177e4	389	/*
65ea5b03	390	* Accessing the stack below %sp is always a bug.
21528454 CE	391	* The large cushion allows instructions like enter
21528454 CE	392	* and pusha to work. ("enter $65535,$31" pushes
65ea5b03	393	* 32 pointers and then decrements %sp by 65535.)
1da177e4	394	*/
65ea5b03	395	if (address + 65536 + 32 * sizeof(unsigned long) < regs->sp)
1da177e4 LT	396	goto bad_area;
	397	}
	398	if (expand_stack(vma, address))
	399	goto bad_area;
	400	/*
	401	* Ok, we have a good vm_area for this memory access, so
	402	* we can handle it..
	403	*/
	404	good_area:
869f96a0	405	si_code = SEGV_ACCERR;
1da177e4	406	write = 0;
33cb5243 HH	407	switch (error_code & (PF_PROT\|PF_WRITE)) {
	408	default: /* 3: write, present */
	409	/* fall through */
	410	case PF_WRITE: /* write, not present */
	411	if (!(vma->vm_flags & VM_WRITE))
	412	goto bad_area;
	413	write++;
	414	break;
	415	case PF_PROT: /* read, present */
	416	goto bad_area;
	417	case 0: /* read, not present */
	418	if (!(vma->vm_flags & (VM_READ \| VM_EXEC \| VM_WRITE)))
1da177e4	419	goto bad_area;
1da177e4 LT	420	}
	421
	422	survive:
	423	/*
	424	* If for any reason at all we couldn't handle the fault,
	425	* make sure we exit gracefully rather than endlessly redo
	426	* the fault.
	427	*/
83c54070 NP	428	fault = handle_mm_fault(mm, vma, address, write);
	429	if (unlikely(fault & VM_FAULT_ERROR)) {
	430	if (fault & VM_FAULT_OOM)
1da177e4	431	goto out_of_memory;
83c54070 NP	432	else if (fault & VM_FAULT_SIGBUS)
	433	goto do_sigbus;
	434	BUG();
1da177e4	435	}
83c54070 NP	436	if (fault & VM_FAULT_MAJOR)
	437	tsk->maj_flt++;
	438	else
	439	tsk->min_flt++;
1da177e4 LT	440
	441	/*
	442	* Did it hit the DOS screen memory VA from vm86 mode?
	443	*/
65ea5b03	444	if (regs->flags & VM_MASK) {
1da177e4 LT	445	unsigned long bit = (address - 0xA0000) >> PAGE_SHIFT;
	446	if (bit < 32)
	447	tsk->thread.screen_bitmap \|= 1 << bit;
	448	}
	449	up_read(&mm->mmap_sem);
	450	return;
	451
	452	/*
	453	* Something tried to access memory that isn't in our memory map..
	454	* Fix it, but check if it's kernel or user first..
	455	*/
	456	bad_area:
	457	up_read(&mm->mmap_sem);
	458
	459	bad_area_nosemaphore:
	460	/* User mode accesses just cause a SIGSEGV */
33cb5243	461	if (error_code & PF_USER) {
e5e3c84b SR	462	/*
	463	* It's possible to have interrupts off here.
	464	*/
	465	local_irq_enable();
	466
33cb5243 HH	467	/*
33cb5243 HH	468	* Valid to do another page fault here because this one came
1da177e4 LT	469	* from user space.
	470	*/
	471	if (is_prefetch(regs, address, error_code))
	472	return;
	473
abd4f750 MAS	474	if (show_unhandled_signals && unhandled_signal(tsk, SIGSEGV) &&
abd4f750 MAS	475	printk_ratelimit()) {
65ea5b03 PA	476	printk("%s%s[%d]: segfault at %08lx ip %08lx "
65ea5b03 PA	477	"sp %08lx error %lx\n",
19c5870c	478	task_pid_nr(tsk) > 1 ? KERN_INFO : KERN_EMERG,
65ea5b03 PA	479	tsk->comm, task_pid_nr(tsk), address, regs->ip,
65ea5b03 PA	480	regs->sp, error_code);
abd4f750	481	}
1da177e4 LT	482	tsk->thread.cr2 = address;
	483	/* Kernel addresses are always protection faults */
	484	tsk->thread.error_code = error_code \| (address >= TASK_SIZE);
	485	tsk->thread.trap_no = 14;
869f96a0	486	force_sig_info_fault(SIGSEGV, si_code, address, tsk);
1da177e4 LT	487	return;
	488	}
	489
	490	#ifdef CONFIG_X86_F00F_BUG
	491	/*
	492	* Pentium F0 0F C7 C8 bug workaround.
	493	*/
	494	if (boot_cpu_data.f00f_bug) {
	495	unsigned long nr;
33cb5243	496
1da177e4 LT	497	nr = (address - idt_descr.address) >> 3;
	498
	499	if (nr == 6) {
	500	do_invalid_op(regs, 0);
	501	return;
	502	}
	503	}
	504	#endif
	505
	506	no_context:
	507	/* Are we prepared to handle this kernel fault? */
	508	if (fixup_exception(regs))
	509	return;
	510
33cb5243	511	/*
1da177e4	512	* Valid to do another page fault here, because if this fault
33cb5243	513	* had been triggered by is_prefetch fixup_exception would have
1da177e4 LT	514	* handled it.
1da177e4 LT	515	*/
33cb5243 HH	516	if (is_prefetch(regs, address, error_code))
33cb5243 HH	517	return;
1da177e4 LT	518
	519	/*
	520	* Oops. The kernel tried to access some bad page. We'll have to
	521	* terminate things with extreme prejudice.
	522	*/
	523
	524	bust_spinlocks(1);
	525
dd287796	526	if (oops_may_print()) {
28609f6e JB	527	__typeof__(pte_val(__pte(0))) page;
	528
	529	#ifdef CONFIG_X86_PAE
318aa296	530	if (error_code & PF_INSTR) {
dd287796 AM	531	pte_t *pte = lookup_address(address);
	532
	533	if (pte && pte_present(pte) && !pte_exec_kernel(pte))
	534	printk(KERN_CRIT "kernel tried to execute "
	535	"NX-protected page - exploit attempt? "
	536	"(uid: %d)\n", current->uid);
	537	}
28609f6e	538	#endif
dd287796 AM	539	if (address < PAGE_SIZE)
	540	printk(KERN_ALERT "BUG: unable to handle kernel NULL "
	541	"pointer dereference");
	542	else
	543	printk(KERN_ALERT "BUG: unable to handle kernel paging"
	544	" request");
33cb5243	545	printk(" at virtual address %08lx\n", address);
65ea5b03	546	printk(KERN_ALERT "printing ip: %08lx ", regs->ip);
28609f6e JB	547
	548	page = read_cr3();
	549	page = ((__typeof__(page) *) __va(page))[address >> PGDIR_SHIFT];
	550	#ifdef CONFIG_X86_PAE
9aa8d719	551	printk("*pdpt = %016Lx ", page);
28609f6e JB	552	if ((page >> PAGE_SHIFT) < max_low_pfn
	553	&& page & _PAGE_PRESENT) {
	554	page &= PAGE_MASK;
	555	page = ((__typeof__(page) *) __va(page))[(address >> PMD_SHIFT)
	556	& (PTRS_PER_PMD - 1)];
eec407c9	557	printk(KERN_CONT "*pde = %016Lx ", page);
28609f6e JB	558	page &= ~_PAGE_NX;
	559	}
	560	#else
9aa8d719	561	printk("*pde = %08lx ", page);
1da177e4	562	#endif
28609f6e JB	563
	564	/*
	565	* We must not directly access the pte in the highpte
	566	* case if the page table is located in highmem.
	567	* And let's rather not kmap-atomic the pte, just in case
	568	* it's allocated already.
	569	*/
	570	if ((page >> PAGE_SHIFT) < max_low_pfn
b1992df3 JB	571	&& (page & _PAGE_PRESENT)
b1992df3 JB	572	&& !(page & _PAGE_PSE)) {
28609f6e JB	573	page &= PAGE_MASK;
	574	page = ((__typeof__(page) *) __va(page))[(address >> PAGE_SHIFT)
	575	& (PTRS_PER_PTE - 1)];
9aa8d719	576	printk("pte = %0Lx ", sizeof(page)*2, (u64)page);
28609f6e	577	}
9aa8d719 PE	578
9aa8d719 PE	579	printk("\n");
28609f6e JB	580	}
28609f6e JB	581
4f339ecb AN	582	tsk->thread.cr2 = address;
	583	tsk->thread.trap_no = 14;
	584	tsk->thread.error_code = error_code;
1da177e4 LT	585	die("Oops", regs, error_code);
	586	bust_spinlocks(0);
	587	do_exit(SIGKILL);
	588
	589	/*
	590	* We ran out of memory, or some other thing happened to us that made
	591	* us unable to handle the page fault gracefully.
	592	*/
	593	out_of_memory:
	594	up_read(&mm->mmap_sem);
b460cbc5	595	if (is_global_init(tsk)) {
1da177e4 LT	596	yield();
	597	down_read(&mm->mmap_sem);
	598	goto survive;
	599	}
	600	printk("VM: killing process %s\n", tsk->comm);
318aa296	601	if (error_code & PF_USER)
dcca2bde	602	do_group_exit(SIGKILL);
1da177e4 LT	603	goto no_context;
	604
	605	do_sigbus:
	606	up_read(&mm->mmap_sem);
	607
	608	/* Kernel mode? Handle exceptions or die */
33cb5243	609	if (!(error_code & PF_USER))
1da177e4 LT	610	goto no_context;
	611
	612	/* User space => ok to do another page fault */
	613	if (is_prefetch(regs, address, error_code))
	614	return;
	615
	616	tsk->thread.cr2 = address;
	617	tsk->thread.error_code = error_code;
	618	tsk->thread.trap_no = 14;
869f96a0	619	force_sig_info_fault(SIGBUS, BUS_ADRERR, address, tsk);
101f12af	620	}
1da177e4	621
101f12af JB	622	void vmalloc_sync_all(void)
	623	{
	624	/*
	625	* Note that races in the updates of insync and start aren't
	626	* problematic: insync can only get set bits added, and updates to
	627	* start are only improving performance (without affecting correctness
	628	* if undone).
	629	*/
	630	static DECLARE_BITMAP(insync, PTRS_PER_PGD);
	631	static unsigned long start = TASK_SIZE;
	632	unsigned long address;
1da177e4	633
5311ab62 JF	634	if (SHARED_KERNEL_PMD)
	635	return;
	636
101f12af JB	637	BUILD_BUG_ON(TASK_SIZE & ~PGDIR_MASK);
	638	for (address = start; address >= TASK_SIZE; address += PGDIR_SIZE) {
	639	if (!test_bit(pgd_index(address), insync)) {
	640	unsigned long flags;
	641	struct page *page;
	642
	643	spin_lock_irqsave(&pgd_lock, flags);
	644	for (page = pgd_list; page; page =
	645	(struct page *)page->index)
	646	if (!vmalloc_sync_one(page_address(page),
	647	address)) {
	648	BUG_ON(page != pgd_list);
	649	break;
	650	}
	651	spin_unlock_irqrestore(&pgd_lock, flags);
	652	if (!page)
	653	set_bit(pgd_index(address), insync);
	654	}
	655	if (address == start && test_bit(pgd_index(address), insync))
	656	start = address + PGDIR_SIZE;
1da177e4 LT	657	}
1da177e4 LT	658	}