[deliverable/linux.git] / arch / s390 / mm / fault.c

/*
 *  arch/s390/mm/fault.c
 *
 *  S390 version
 *    Copyright (C) 1999 IBM Deutschland Entwicklung GmbH, IBM Corporation
 *    Author(s): Hartmut Penner (hp@de.ibm.com)
 *               Ulrich Weigand (uweigand@de.ibm.com)
 *
 *  Derived from "arch/i386/mm/fault.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/smp_lock.h>
#include <linux/init.h>
#include <linux/console.h>
#include <linux/module.h>
#include <linux/hardirq.h>
#include <linux/kprobes.h>

#include <asm/system.h>
#include <asm/uaccess.h>
#include <asm/pgtable.h>
#include <asm/kdebug.h>

#ifndef CONFIG_64BIT
#define __FAIL_ADDR_MASK 0x7ffff000
#define __FIXUP_MASK 0x7fffffff
#define __SUBCODE_MASK 0x0200
#define __PF_RES_FIELD 0ULL
#else /* CONFIG_64BIT */
#define __FAIL_ADDR_MASK -4096L
#define __FIXUP_MASK ~0L
#define __SUBCODE_MASK 0x0600
#define __PF_RES_FIELD 0x8000000000000000ULL
#endif /* CONFIG_64BIT */

#ifdef CONFIG_SYSCTL
extern int sysctl_userprocess_debug;
#endif

extern void die(const char *,struct pt_regs *,long);

#ifdef CONFIG_KPROBES
ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
int register_page_fault_notifier(struct notifier_block *nb)
{
	return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
}

int unregister_page_fault_notifier(struct notifier_block *nb)
{
	return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
}

static inline int notify_page_fault(enum die_val val, const char *str,
			struct pt_regs *regs, long err, int trap, int sig)
{
	struct die_args args = {
		.regs = regs,
		.str = str,
		.err = err,
		.trapnr = trap,
		.signr = sig
	};
	return atomic_notifier_call_chain(&notify_page_fault_chain, val, &args);
}
#else
static inline int notify_page_fault(enum die_val val, const char *str,
			struct pt_regs *regs, long err, int trap, int sig)
{
	return NOTIFY_DONE;
}
#endif

extern spinlock_t timerlist_lock;

/*
 * Unlock any spinlocks which will prevent us from getting the
 * message out (timerlist_lock is acquired through the
 * console unblank code)
 */
void bust_spinlocks(int yes)
{
	if (yes) {
		oops_in_progress = 1;
	} else {
		int loglevel_save = console_loglevel;
		console_unblank();
		oops_in_progress = 0;
		/*
		 * OK, the message is on the console.  Now we call printk()
		 * without oops_in_progress set so that printk will give klogd
		 * a poke.  Hold onto your hats...
		 */
		console_loglevel = 15;
		printk(" ");
		console_loglevel = loglevel_save;
	}
}

/*
 * Check which address space is addressed by the access
 * register in S390_lowcore.exc_access_id.
 * Returns 1 for user space and 0 for kernel space.
 */
static int __check_access_register(struct pt_regs *regs, int error_code)
{
	int areg = S390_lowcore.exc_access_id;

	if (areg == 0)
		/* Access via access register 0 -> kernel address */
		return 0;
	save_access_regs(current->thread.acrs);
	if (regs && areg < NUM_ACRS && current->thread.acrs[areg] <= 1)
		/*
		 * access register contains 0 -> kernel address,
		 * access register contains 1 -> user space address
		 */
		return current->thread.acrs[areg];

	/* Something unhealthy was done with the access registers... */
	die("page fault via unknown access register", regs, error_code);
	do_exit(SIGKILL);
	return 0;
}

/*
 * Check which address space the address belongs to.
 * Returns 1 for user space and 0 for kernel space.
 */
static inline int check_user_space(struct pt_regs *regs, int error_code)
{
	/*
	 * The lowest two bits of S390_lowcore.trans_exc_code indicate
	 * which paging table was used:
	 *   0: Primary Segment Table Descriptor
	 *   1: STD determined via access register
	 *   2: Secondary Segment Table Descriptor
	 *   3: Home Segment Table Descriptor
	 */
	int descriptor = S390_lowcore.trans_exc_code & 3;
	if (unlikely(descriptor == 1))
		return __check_access_register(regs, error_code);
	if (descriptor == 2)
		return current->thread.mm_segment.ar4;
	return descriptor != 0;
}

/*
 * Send SIGSEGV to task.  This is an external routine
 * to keep the stack usage of do_page_fault small.
 */
static void do_sigsegv(struct pt_regs *regs, unsigned long error_code,
		       int si_code, unsigned long address)
{
	struct siginfo si;

#if defined(CONFIG_SYSCTL) || defined(CONFIG_PROCESS_DEBUG)
#if defined(CONFIG_SYSCTL)
	if (sysctl_userprocess_debug)
#endif
	{
		printk("User process fault: interruption code 0x%lX\n",
		       error_code);
		printk("failing address: %lX\n", address);
		show_regs(regs);
	}
#endif
	si.si_signo = SIGSEGV;
	si.si_code = si_code;
	si.si_addr = (void __user *) address;
	force_sig_info(SIGSEGV, &si, current);
}

/*
 * This routine handles page faults.  It determines the address,
 * and the problem, and then passes it off to one of the appropriate
 * routines.
 *
 * error_code:
 *   04       Protection           ->  Write-Protection  (suprression)
 *   10       Segment translation  ->  Not present       (nullification)
 *   11       Page translation     ->  Not present       (nullification)
 *   3b       Region third trans.  ->  Not present       (nullification)
 */
static inline void __kprobes
do_exception(struct pt_regs *regs, unsigned long error_code, int is_protection)
{
        struct task_struct *tsk;
        struct mm_struct *mm;
        struct vm_area_struct * vma;
        unsigned long address;
	int user_address;
	const struct exception_table_entry *fixup;
	int si_code = SEGV_MAPERR;

        tsk = current;
        mm = tsk->mm;
	
	if (notify_page_fault(DIE_PAGE_FAULT, "page fault", regs, error_code, 14,
					SIGSEGV) == NOTIFY_STOP)
		return;

	/* 
         * Check for low-address protection.  This needs to be treated
	 * as a special case because the translation exception code 
	 * field is not guaranteed to contain valid data in this case.
	 */
	if (is_protection && !(S390_lowcore.trans_exc_code & 4)) {

		/* Low-address protection hit in kernel mode means 
		   NULL pointer write access in kernel mode.  */
 		if (!(regs->psw.mask & PSW_MASK_PSTATE)) {
			address = 0;
			user_address = 0;
			goto no_context;
		}

		/* Low-address protection hit in user mode 'cannot happen'.  */
		die ("Low-address protection", regs, error_code);
        	do_exit(SIGKILL);
	}

        /* 
         * get the failing address 
         * more specific the segment and page table portion of 
         * the address 
         */
        address = S390_lowcore.trans_exc_code & __FAIL_ADDR_MASK;
	user_address = check_user_space(regs, error_code);

	/*
	 * Verify that the fault happened in user space, that
	 * we are not in an interrupt and that there is a 
	 * user context.
	 */
        if (user_address == 0 || in_atomic() || !mm)
                goto no_context;

	/*
	 * When we get here, the fault happened in the current
	 * task's user address space, so we can switch on the
	 * interrupts again and then search the VMAs
	 */
	local_irq_enable();

        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 (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;
	if (!is_protection) {
		/* page not present, check vm flags */
		if (!(vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE)))
			goto bad_area;
	} else {
		if (!(vma->vm_flags & 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.
	 */
	switch (handle_mm_fault(mm, vma, address, is_protection)) {
	case VM_FAULT_MINOR:
		tsk->min_flt++;
		break;
	case VM_FAULT_MAJOR:
		tsk->maj_flt++;
		break;
	case VM_FAULT_SIGBUS:
		goto do_sigbus;
	case VM_FAULT_OOM:
		goto out_of_memory;
	default:
		BUG();
	}

        up_read(&mm->mmap_sem);
	/*
	 * The instruction that caused the program check will
	 * be repeated. Don't signal single step via SIGTRAP.
	 */
	clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
        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);

        /* User mode accesses just cause a SIGSEGV */
        if (regs->psw.mask & PSW_MASK_PSTATE) {
                tsk->thread.prot_addr = address;
                tsk->thread.trap_no = error_code;
		do_sigsegv(regs, error_code, si_code, address);
                return;
	}

no_context:
        /* Are we prepared to handle this kernel fault?  */
	fixup = search_exception_tables(regs->psw.addr & __FIXUP_MASK);
	if (fixup) {
		regs->psw.addr = fixup->fixup | PSW_ADDR_AMODE;
                return;
        }

/*
 * Oops. The kernel tried to access some bad page. We'll have to
 * terminate things with extreme prejudice.
 */
        if (user_address == 0)
                printk(KERN_ALERT "Unable to handle kernel pointer dereference"
        	       " at virtual kernel address %p\n", (void *)address);
        else
                printk(KERN_ALERT "Unable to handle kernel paging request"
		       " at virtual user address %p\n", (void *)address);

        die("Oops", regs, error_code);
        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 (tsk->pid == 1) {
		yield();
		down_read(&mm->mmap_sem);
		goto survive;
	}
	printk("VM: killing process %s\n", tsk->comm);
	if (regs->psw.mask & PSW_MASK_PSTATE)
		do_exit(SIGKILL);
	goto no_context;

do_sigbus:
	up_read(&mm->mmap_sem);

	/*
	 * Send a sigbus, regardless of whether we were in kernel
	 * or user mode.
	 */
        tsk->thread.prot_addr = address;
        tsk->thread.trap_no = error_code;
	force_sig(SIGBUS, tsk);

	/* Kernel mode? Handle exceptions or die */
	if (!(regs->psw.mask & PSW_MASK_PSTATE))
		goto no_context;
}

void do_protection_exception(struct pt_regs *regs, unsigned long error_code)
{
	regs->psw.addr -= (error_code >> 16);
	do_exception(regs, 4, 1);
}

void do_dat_exception(struct pt_regs *regs, unsigned long error_code)
{
	do_exception(regs, error_code & 0xff, 0);
}

#ifdef CONFIG_PFAULT 
/*
 * 'pfault' pseudo page faults routines.
 */
static int pfault_disable = 0;

static int __init nopfault(char *str)
{
	pfault_disable = 1;
	return 1;
}

__setup("nopfault", nopfault);

typedef struct {
	__u16 refdiagc;
	__u16 reffcode;
	__u16 refdwlen;
	__u16 refversn;
	__u64 refgaddr;
	__u64 refselmk;
	__u64 refcmpmk;
	__u64 reserved;
} __attribute__ ((packed)) pfault_refbk_t;

int pfault_init(void)
{
	pfault_refbk_t refbk =
		{ 0x258, 0, 5, 2, __LC_CURRENT, 1ULL << 48, 1ULL << 48,
		  __PF_RES_FIELD };
        int rc;

	if (pfault_disable)
		return -1;
        __asm__ __volatile__(
                "    diag  %1,%0,0x258\n"
		"0:  j     2f\n"
		"1:  la    %0,8\n"
		"2:\n"
		".section __ex_table,\"a\"\n"
		"   .align 4\n"
#ifndef CONFIG_64BIT
		"   .long  0b,1b\n"
#else /* CONFIG_64BIT */
		"   .quad  0b,1b\n"
#endif /* CONFIG_64BIT */
		".previous"
                : "=d" (rc) : "a" (&refbk), "m" (refbk) : "cc" );
        __ctl_set_bit(0, 9);
        return rc;
}

void pfault_fini(void)
{
	pfault_refbk_t refbk =
	{ 0x258, 1, 5, 2, 0ULL, 0ULL, 0ULL, 0ULL };

	if (pfault_disable)
		return;
	__ctl_clear_bit(0,9);
        __asm__ __volatile__(
                "    diag  %0,0,0x258\n"
		"0:\n"
		".section __ex_table,\"a\"\n"
		"   .align 4\n"
#ifndef CONFIG_64BIT
		"   .long  0b,0b\n"
#else /* CONFIG_64BIT */
		"   .quad  0b,0b\n"
#endif /* CONFIG_64BIT */
		".previous"
		: : "a" (&refbk), "m" (refbk) : "cc" );
}

asmlinkage void
pfault_interrupt(struct pt_regs *regs, __u16 error_code)
{
	struct task_struct *tsk;
	__u16 subcode;

	/*
	 * Get the external interruption subcode & pfault
	 * initial/completion signal bit. VM stores this 
	 * in the 'cpu address' field associated with the
         * external interrupt. 
	 */
	subcode = S390_lowcore.cpu_addr;
	if ((subcode & 0xff00) != __SUBCODE_MASK)
		return;

	/*
	 * Get the token (= address of the task structure of the affected task).
	 */
	tsk = *(struct task_struct **) __LC_PFAULT_INTPARM;

	if (subcode & 0x0080) {
		/* signal bit is set -> a page has been swapped in by VM */
		if (xchg(&tsk->thread.pfault_wait, -1) != 0) {
			/* Initial interrupt was faster than the completion
			 * interrupt. pfault_wait is valid. Set pfault_wait
			 * back to zero and wake up the process. This can
			 * safely be done because the task is still sleeping
			 * and can't produce new pfaults. */
			tsk->thread.pfault_wait = 0;
			wake_up_process(tsk);
			put_task_struct(tsk);
		}
	} else {
		/* signal bit not set -> a real page is missing. */
		get_task_struct(tsk);
		set_task_state(tsk, TASK_UNINTERRUPTIBLE);
		if (xchg(&tsk->thread.pfault_wait, 1) != 0) {
			/* Completion interrupt was faster than the initial
			 * interrupt (swapped in a -1 for pfault_wait). Set
			 * pfault_wait back to zero and exit. This can be
			 * done safely because tsk is running in kernel 
			 * mode and can't produce new pfaults. */
			tsk->thread.pfault_wait = 0;
			set_task_state(tsk, TASK_RUNNING);
			put_task_struct(tsk);
		} else
			set_tsk_need_resched(tsk);
	}
}
#endif
Commit	Line	Data
1da177e4 LT	1	/*
	2	* arch/s390/mm/fault.c
	3	*
	4	* S390 version
	5	* Copyright (C) 1999 IBM Deutschland Entwicklung GmbH, IBM Corporation
	6	* Author(s): Hartmut Penner (hp@de.ibm.com)
	7	* Ulrich Weigand (uweigand@de.ibm.com)
	8	*
	9	* Derived from "arch/i386/mm/fault.c"
	10	* Copyright (C) 1995 Linus Torvalds
	11	*/
	12
1da177e4 LT	13	#include <linux/signal.h>
	14	#include <linux/sched.h>
	15	#include <linux/kernel.h>
	16	#include <linux/errno.h>
	17	#include <linux/string.h>
	18	#include <linux/types.h>
	19	#include <linux/ptrace.h>
	20	#include <linux/mman.h>
	21	#include <linux/mm.h>
	22	#include <linux/smp.h>
	23	#include <linux/smp_lock.h>
	24	#include <linux/init.h>
	25	#include <linux/console.h>
	26	#include <linux/module.h>
	27	#include <linux/hardirq.h>
4ba069b8	28	#include <linux/kprobes.h>
1da177e4 LT	29
	30	#include <asm/system.h>
	31	#include <asm/uaccess.h>
	32	#include <asm/pgtable.h>
4ba069b8	33	#include <asm/kdebug.h>
1da177e4	34
347a8dc3	35	#ifndef CONFIG_64BIT
1da177e4 LT	36	#define __FAIL_ADDR_MASK 0x7ffff000
	37	#define __FIXUP_MASK 0x7fffffff
	38	#define __SUBCODE_MASK 0x0200
	39	#define __PF_RES_FIELD 0ULL
347a8dc3	40	#else /* CONFIG_64BIT */
1da177e4 LT	41	#define __FAIL_ADDR_MASK -4096L
	42	#define __FIXUP_MASK ~0L
	43	#define __SUBCODE_MASK 0x0600
	44	#define __PF_RES_FIELD 0x8000000000000000ULL
347a8dc3	45	#endif /* CONFIG_64BIT */
1da177e4 LT	46
	47	#ifdef CONFIG_SYSCTL
	48	extern int sysctl_userprocess_debug;
	49	#endif
	50
	51	extern void die(const char ,struct pt_regs ,long);
	52
4ba069b8 MG	53	#ifdef CONFIG_KPROBES
	54	ATOMIC_NOTIFIER_HEAD(notify_page_fault_chain);
	55	int register_page_fault_notifier(struct notifier_block *nb)
	56	{
	57	return atomic_notifier_chain_register(&notify_page_fault_chain, nb);
	58	}
	59
	60	int unregister_page_fault_notifier(struct notifier_block *nb)
	61	{
	62	return atomic_notifier_chain_unregister(&notify_page_fault_chain, nb);
	63	}
	64
	65	static inline int notify_page_fault(enum die_val val, const char *str,
	66	struct pt_regs *regs, long err, int trap, int sig)
	67	{
	68	struct die_args args = {
	69	.regs = regs,
	70	.str = str,
	71	.err = err,
	72	.trapnr = trap,
	73	.signr = sig
	74	};
	75	return atomic_notifier_call_chain(&notify_page_fault_chain, val, &args);
	76	}
	77	#else
	78	static inline int notify_page_fault(enum die_val val, const char *str,
	79	struct pt_regs *regs, long err, int trap, int sig)
	80	{
	81	return NOTIFY_DONE;
	82	}
	83	#endif
	84
1da177e4 LT	85	extern spinlock_t timerlist_lock;
	86
	87	/*
	88	* Unlock any spinlocks which will prevent us from getting the
	89	* message out (timerlist_lock is acquired through the
	90	* console unblank code)
	91	*/
	92	void bust_spinlocks(int yes)
	93	{
	94	if (yes) {
	95	oops_in_progress = 1;
	96	} else {
	97	int loglevel_save = console_loglevel;
	98	console_unblank();
	99	oops_in_progress = 0;
	100	/*
	101	* OK, the message is on the console. Now we call printk()
	102	* without oops_in_progress set so that printk will give klogd
	103	* a poke. Hold onto your hats...
	104	*/
	105	console_loglevel = 15;
	106	printk(" ");
	107	console_loglevel = loglevel_save;
	108	}
	109	}
	110
	111	/*
	112	* Check which address space is addressed by the access
	113	* register in S390_lowcore.exc_access_id.
	114	* Returns 1 for user space and 0 for kernel space.
	115	*/
	116	static int __check_access_register(struct pt_regs *regs, int error_code)
	117	{
	118	int areg = S390_lowcore.exc_access_id;
	119
	120	if (areg == 0)
	121	/* Access via access register 0 -> kernel address */
	122	return 0;
	123	save_access_regs(current->thread.acrs);
	124	if (regs && areg < NUM_ACRS && current->thread.acrs[areg] <= 1)
	125	/*
	126	* access register contains 0 -> kernel address,
	127	* access register contains 1 -> user space address
	128	*/
	129	return current->thread.acrs[areg];
	130
	131	/* Something unhealthy was done with the access registers... */
	132	die("page fault via unknown access register", regs, error_code);
	133	do_exit(SIGKILL);
	134	return 0;
	135	}
	136
	137	/*
	138	* Check which address space the address belongs to.
	139	* Returns 1 for user space and 0 for kernel space.
	140	*/
	141	static inline int check_user_space(struct pt_regs *regs, int error_code)
	142	{
	143	/*
	144	* The lowest two bits of S390_lowcore.trans_exc_code indicate
	145	* which paging table was used:
	146	* 0: Primary Segment Table Descriptor
	147	* 1: STD determined via access register
	148	* 2: Secondary Segment Table Descriptor
149	* 3: Home Segment Table Descriptor
150	*/
151	int descriptor = S390_lowcore.trans_exc_code & 3;
152	if (unlikely(descriptor == 1))
153	return __check_access_register(regs, error_code);
154	if (descriptor == 2)
155	return current->thread.mm_segment.ar4;
156	return descriptor != 0;
157	}
158
159	/*
160	* Send SIGSEGV to task. This is an external routine
161	* to keep the stack usage of do_page_fault small.
162	*/
163	static void do_sigsegv(struct pt_regs *regs, unsigned long error_code,
164	int si_code, unsigned long address)
165	{
166	struct siginfo si;
167
168	#if defined(CONFIG_SYSCTL) \|\| defined(CONFIG_PROCESS_DEBUG)
169	#if defined(CONFIG_SYSCTL)
170	if (sysctl_userprocess_debug)
171	#endif
172	{
173	printk("User process fault: interruption code 0x%lX\n",
174	error_code);
175	printk("failing address: %lX\n", address);
176	show_regs(regs);
177	}
178	#endif
179	si.si_signo = SIGSEGV;
180	si.si_code = si_code;
d2c993d8	181	si.si_addr = (void __user *) address;
1da177e4 LT	182	force_sig_info(SIGSEGV, &si, current);
	183	}
	184
	185	/*
	186	* This routine handles page faults. It determines the address,
	187	* and the problem, and then passes it off to one of the appropriate
	188	* routines.
	189	*
	190	* error_code:
	191	* 04 Protection -> Write-Protection (suprression)
	192	* 10 Segment translation -> Not present (nullification)
	193	* 11 Page translation -> Not present (nullification)
	194	* 3b Region third trans. -> Not present (nullification)
	195	*/
4ba069b8	196	static inline void __kprobes
1da177e4 LT	197	do_exception(struct pt_regs *regs, unsigned long error_code, int is_protection)
	198	{
	199	struct task_struct *tsk;
	200	struct mm_struct *mm;
	201	struct vm_area_struct * vma;
	202	unsigned long address;
	203	int user_address;
	204	const struct exception_table_entry *fixup;
	205	int si_code = SEGV_MAPERR;
	206
	207	tsk = current;
	208	mm = tsk->mm;
	209
4ba069b8 MG	210	if (notify_page_fault(DIE_PAGE_FAULT, "page fault", regs, error_code, 14,
	211	SIGSEGV) == NOTIFY_STOP)
	212	return;
	213
1da177e4 LT	214	/*
	215	* Check for low-address protection. This needs to be treated
	216	* as a special case because the translation exception code
	217	* field is not guaranteed to contain valid data in this case.
	218	*/
	219	if (is_protection && !(S390_lowcore.trans_exc_code & 4)) {
	220
	221	/* Low-address protection hit in kernel mode means
	222	NULL pointer write access in kernel mode. */
	223	if (!(regs->psw.mask & PSW_MASK_PSTATE)) {
	224	address = 0;
	225	user_address = 0;
	226	goto no_context;
	227	}
	228
	229	/* Low-address protection hit in user mode 'cannot happen'. */
	230	die ("Low-address protection", regs, error_code);
	231	do_exit(SIGKILL);
	232	}
	233
	234	/*
	235	* get the failing address
	236	* more specific the segment and page table portion of
	237	* the address
	238	*/
	239	address = S390_lowcore.trans_exc_code & __FAIL_ADDR_MASK;
	240	user_address = check_user_space(regs, error_code);
	241
	242	/*
	243	* Verify that the fault happened in user space, that
	244	* we are not in an interrupt and that there is a
	245	* user context.
	246	*/
595bf2aa	247	if (user_address == 0 \|\| in_atomic() \|\| !mm)
1da177e4 LT	248	goto no_context;
	249
	250	/*
	251	* When we get here, the fault happened in the current
	252	* task's user address space, so we can switch on the
	253	* interrupts again and then search the VMAs
	254	*/
	255	local_irq_enable();
	256
	257	down_read(&mm->mmap_sem);
	258
	259	vma = find_vma(mm, address);
	260	if (!vma)
	261	goto bad_area;
	262	if (vma->vm_start <= address)
	263	goto good_area;
	264	if (!(vma->vm_flags & VM_GROWSDOWN))
	265	goto bad_area;
	266	if (expand_stack(vma, address))
	267	goto bad_area;
	268	/*
	269	* Ok, we have a good vm_area for this memory access, so
	270	* we can handle it..
	271	*/
	272	good_area:
	273	si_code = SEGV_ACCERR;
	274	if (!is_protection) {
	275	/* page not present, check vm flags */
	276	if (!(vma->vm_flags & (VM_READ \| VM_EXEC \| VM_WRITE)))
	277	goto bad_area;
	278	} else {
	279	if (!(vma->vm_flags & VM_WRITE))
	280	goto bad_area;
	281	}
	282
	283	survive:
	284	/*
	285	* If for any reason at all we couldn't handle the fault,
	286	* make sure we exit gracefully rather than endlessly redo
	287	* the fault.
	288	*/
	289	switch (handle_mm_fault(mm, vma, address, is_protection)) {
	290	case VM_FAULT_MINOR:
	291	tsk->min_flt++;
	292	break;
	293	case VM_FAULT_MAJOR:
	294	tsk->maj_flt++;
	295	break;
	296	case VM_FAULT_SIGBUS:
	297	goto do_sigbus;
	298	case VM_FAULT_OOM:
	299	goto out_of_memory;
	300	default:
	301	BUG();
	302	}
	303
	304	up_read(&mm->mmap_sem);
	305	/*
	306	* The instruction that caused the program check will
	307	* be repeated. Don't signal single step via SIGTRAP.
	308	*/
	309	clear_tsk_thread_flag(current, TIF_SINGLE_STEP);
	310	return;
	311
312	/*
313	* Something tried to access memory that isn't in our memory map..
314	* Fix it, but check if it's kernel or user first..
315	*/
316	bad_area:
317	up_read(&mm->mmap_sem);
318
319	/* User mode accesses just cause a SIGSEGV */
320	if (regs->psw.mask & PSW_MASK_PSTATE) {
321	tsk->thread.prot_addr = address;
322	tsk->thread.trap_no = error_code;
323	do_sigsegv(regs, error_code, si_code, address);
324	return;
325	}
326
327	no_context:
328	/* Are we prepared to handle this kernel fault? */
329	fixup = search_exception_tables(regs->psw.addr & __FIXUP_MASK);
330	if (fixup) {
331	regs->psw.addr = fixup->fixup \| PSW_ADDR_AMODE;
332	return;
333	}
334
335	/*
336	* Oops. The kernel tried to access some bad page. We'll have to
337	* terminate things with extreme prejudice.
338	*/
339	if (user_address == 0)
340	printk(KERN_ALERT "Unable to handle kernel pointer dereference"
341	" at virtual kernel address %p\n", (void *)address);
342	else
343	printk(KERN_ALERT "Unable to handle kernel paging request"
344	" at virtual user address %p\n", (void *)address);
345
346	die("Oops", regs, error_code);
347	do_exit(SIGKILL);
348
349
350	/*
351	* We ran out of memory, or some other thing happened to us that made
352	* us unable to handle the page fault gracefully.
353	*/
354	out_of_memory:
355	up_read(&mm->mmap_sem);
356	if (tsk->pid == 1) {
357	yield();
bac9c66c	358	down_read(&mm->mmap_sem);
1da177e4 LT	359	goto survive;
	360	}
	361	printk("VM: killing process %s\n", tsk->comm);
	362	if (regs->psw.mask & PSW_MASK_PSTATE)
	363	do_exit(SIGKILL);
	364	goto no_context;
	365
	366	do_sigbus:
	367	up_read(&mm->mmap_sem);
	368
	369	/*
	370	* Send a sigbus, regardless of whether we were in kernel
	371	* or user mode.
	372	*/
	373	tsk->thread.prot_addr = address;
	374	tsk->thread.trap_no = error_code;
	375	force_sig(SIGBUS, tsk);
	376
	377	/* Kernel mode? Handle exceptions or die */
	378	if (!(regs->psw.mask & PSW_MASK_PSTATE))
	379	goto no_context;
	380	}
	381
	382	void do_protection_exception(struct pt_regs *regs, unsigned long error_code)
	383	{
	384	regs->psw.addr -= (error_code >> 16);
	385	do_exception(regs, 4, 1);
	386	}
	387
	388	void do_dat_exception(struct pt_regs *regs, unsigned long error_code)
	389	{
	390	do_exception(regs, error_code & 0xff, 0);
	391	}
	392
1da177e4 LT	393	#ifdef CONFIG_PFAULT
	394	/*
	395	* 'pfault' pseudo page faults routines.
	396	*/
	397	static int pfault_disable = 0;
	398
	399	static int __init nopfault(char *str)
	400	{
	401	pfault_disable = 1;
	402	return 1;
	403	}
	404
	405	__setup("nopfault", nopfault);
	406
	407	typedef struct {
	408	__u16 refdiagc;
	409	__u16 reffcode;
	410	__u16 refdwlen;
	411	__u16 refversn;
	412	__u64 refgaddr;
	413	__u64 refselmk;
	414	__u64 refcmpmk;
	415	__u64 reserved;
	416	} __attribute__ ((packed)) pfault_refbk_t;
	417
	418	int pfault_init(void)
	419	{
	420	pfault_refbk_t refbk =
	421	{ 0x258, 0, 5, 2, __LC_CURRENT, 1ULL << 48, 1ULL << 48,
	422	__PF_RES_FIELD };
	423	int rc;
	424
	425	if (pfault_disable)
	426	return -1;
	427	__asm__ __volatile__(
	428	" diag %1,%0,0x258\n"
	429	"0: j 2f\n"
	430	"1: la %0,8\n"
	431	"2:\n"
	432	".section __ex_table,\"a\"\n"
	433	" .align 4\n"
347a8dc3	434	#ifndef CONFIG_64BIT
1da177e4	435	" .long 0b,1b\n"
347a8dc3	436	#else /* CONFIG_64BIT */
1da177e4	437	" .quad 0b,1b\n"
347a8dc3	438	#endif /* CONFIG_64BIT */
1da177e4	439	".previous"
d4b68996	440	: "=d" (rc) : "a" (&refbk), "m" (refbk) : "cc" );
1da177e4 LT	441	__ctl_set_bit(0, 9);
	442	return rc;
	443	}
	444
	445	void pfault_fini(void)
	446	{
	447	pfault_refbk_t refbk =
	448	{ 0x258, 1, 5, 2, 0ULL, 0ULL, 0ULL, 0ULL };
	449
	450	if (pfault_disable)
	451	return;
	452	__ctl_clear_bit(0,9);
	453	__asm__ __volatile__(
	454	" diag %0,0,0x258\n"
	455	"0:\n"
	456	".section __ex_table,\"a\"\n"
	457	" .align 4\n"
347a8dc3	458	#ifndef CONFIG_64BIT
1da177e4	459	" .long 0b,0b\n"
347a8dc3	460	#else /* CONFIG_64BIT */
1da177e4	461	" .quad 0b,0b\n"
347a8dc3	462	#endif /* CONFIG_64BIT */
1da177e4	463	".previous"
d4b68996	464	: : "a" (&refbk), "m" (refbk) : "cc" );
1da177e4 LT	465	}
	466
	467	asmlinkage void
	468	pfault_interrupt(struct pt_regs *regs, __u16 error_code)
	469	{
	470	struct task_struct *tsk;
	471	__u16 subcode;
	472
	473	/*
	474	* Get the external interruption subcode & pfault
	475	* initial/completion signal bit. VM stores this
	476	* in the 'cpu address' field associated with the
	477	* external interrupt.
	478	*/
	479	subcode = S390_lowcore.cpu_addr;
	480	if ((subcode & 0xff00) != __SUBCODE_MASK)
	481	return;
	482
	483	/*
	484	* Get the token (= address of the task structure of the affected task).
	485	*/
	486	tsk = (struct task_struct *) __LC_PFAULT_INTPARM;
	487
	488	if (subcode & 0x0080) {
	489	/* signal bit is set -> a page has been swapped in by VM */
	490	if (xchg(&tsk->thread.pfault_wait, -1) != 0) {
	491	/* Initial interrupt was faster than the completion
	492	* interrupt. pfault_wait is valid. Set pfault_wait
	493	* back to zero and wake up the process. This can
	494	* safely be done because the task is still sleeping
b6d09449	495	* and can't produce new pfaults. */
1da177e4 LT	496	tsk->thread.pfault_wait = 0;
1da177e4 LT	497	wake_up_process(tsk);
b6d09449	498	put_task_struct(tsk);
1da177e4 LT	499	}
	500	} else {
	501	/* signal bit not set -> a real page is missing. */
b6d09449	502	get_task_struct(tsk);
1da177e4 LT	503	set_task_state(tsk, TASK_UNINTERRUPTIBLE);
	504	if (xchg(&tsk->thread.pfault_wait, 1) != 0) {
	505	/* Completion interrupt was faster than the initial
	506	* interrupt (swapped in a -1 for pfault_wait). Set
	507	* pfault_wait back to zero and exit. This can be
	508	* done safely because tsk is running in kernel
	509	* mode and can't produce new pfaults. */
	510	tsk->thread.pfault_wait = 0;
	511	set_task_state(tsk, TASK_RUNNING);
b6d09449	512	put_task_struct(tsk);
1da177e4 LT	513	} else
	514	set_tsk_need_resched(tsk);
	515	}
	516	}
	517	#endif
	518