1 files changed, 407 insertions, 252 deletions

diff --git a/arch/s390/mm/fault.c b/arch/s390/mm/fault.c
index 2c57806c085..3f3b35403d0 100644
--- a/arch/s390/mm/fault.c
+++ b/arch/s390/mm/fault.c
@@ -1,8 +1,6 @@
 /*
- *  arch/s390/mm/fault.c
- *
  *  S390 version
- *    Copyright (C) 1999 IBM Deutschland Entwicklung GmbH, IBM Corporation
+ *    Copyright IBM Corp. 1999
  *    Author(s): Hartmut Penner (hp@de.ibm.com)
  *               Ulrich Weigand (uweigand@de.ibm.com)
  *
@@ -32,11 +30,10 @@
 #include <linux/uaccess.h>
 #include <linux/hugetlb.h>
 #include <asm/asm-offsets.h>
-#include <asm/system.h>
 #include <asm/pgtable.h>
-#include <asm/s390_ext.h>
+#include <asm/irq.h>
 #include <asm/mmu_context.h>
-#include <asm/compat.h>
+#include <asm/facility.h>
 #include "../kernel/entry.h"
 
 #ifndef CONFIG_64BIT
@@ -52,14 +49,20 @@
 #define VM_FAULT_BADCONTEXT	0x010000
 #define VM_FAULT_BADMAP		0x020000
 #define VM_FAULT_BADACCESS	0x040000
+#define VM_FAULT_SIGNAL		0x080000
+#define VM_FAULT_PFAULT		0x100000
 
-static unsigned long store_indication;
+static unsigned long store_indication __read_mostly;
 
-void fault_init(void)
+#ifdef CONFIG_64BIT
+static int __init fault_init(void)
 {
-	if (test_facility(2) && test_facility(75))
+	if (test_facility(75))
 		store_indication = 0xc00;
+	return 0;
 }
+early_initcall(fault_init);
+#endif
 
 static inline int notify_page_fault(struct pt_regs *regs)
 {
@@ -103,30 +106,154 @@ void bust_spinlocks(int yes)
  * Returns the address space associated with the fault.
  * Returns 0 for kernel space and 1 for user space.
  */
-static inline int user_space_fault(unsigned long trans_exc_code)
+static inline int user_space_fault(struct pt_regs *regs)
 {
+	unsigned long trans_exc_code;
+
 	/*
 	 * The lowest two bits of the translation exception
 	 * identification indicate which paging table was used.
 	 */
-	trans_exc_code &= 3;
-	if (trans_exc_code == 2)
-		/* Access via secondary space, set_fs setting decides */
+	trans_exc_code = regs->int_parm_long & 3;
+	if (trans_exc_code == 3) /* home space -> kernel */
+		return 0;
+	if (user_mode(regs))
+		return 1;
+	if (trans_exc_code == 2) /* secondary space -> set_fs */
 		return current->thread.mm_segment.ar4;
-	if (user_mode == HOME_SPACE_MODE)
-		/* User space if the access has been done via home space. */
-		return trans_exc_code == 3;
-	/*
-	 * If the user space is not the home space the kernel runs in home
-	 * space. Access via secondary space has already been covered,
-	 * access via primary space or access register is from user space
-	 * and access via home space is from the kernel.
-	 */
-	return trans_exc_code != 3;
+	if (current->flags & PF_VCPU)
+		return 1;
+	return 0;
+}
+
+static int bad_address(void *p)
+{
+	unsigned long dummy;
+
+	return probe_kernel_address((unsigned long *)p, dummy);
+}
+
+#ifdef CONFIG_64BIT
+static void dump_pagetable(unsigned long asce, unsigned long address)
+{
+	unsigned long *table = __va(asce & PAGE_MASK);
+
+	pr_alert("AS:%016lx ", asce);
+	switch (asce & _ASCE_TYPE_MASK) {
+	case _ASCE_TYPE_REGION1:
+		table = table + ((address >> 53) & 0x7ff);
+		if (bad_address(table))
+			goto bad;
+		pr_cont("R1:%016lx ", *table);
+		if (*table & _REGION_ENTRY_INVALID)
+			goto out;
+		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
+		/* fallthrough */
+	case _ASCE_TYPE_REGION2:
+		table = table + ((address >> 42) & 0x7ff);
+		if (bad_address(table))
+			goto bad;
+		pr_cont("R2:%016lx ", *table);
+		if (*table & _REGION_ENTRY_INVALID)
+			goto out;
+		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
+		/* fallthrough */
+	case _ASCE_TYPE_REGION3:
+		table = table + ((address >> 31) & 0x7ff);
+		if (bad_address(table))
+			goto bad;
+		pr_cont("R3:%016lx ", *table);
+		if (*table & (_REGION_ENTRY_INVALID | _REGION3_ENTRY_LARGE))
+			goto out;
+		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
+		/* fallthrough */
+	case _ASCE_TYPE_SEGMENT:
+		table = table + ((address >> 20) & 0x7ff);
+		if (bad_address(table))
+			goto bad;
+		pr_cont(KERN_CONT "S:%016lx ", *table);
+		if (*table & (_SEGMENT_ENTRY_INVALID | _SEGMENT_ENTRY_LARGE))
+			goto out;
+		table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
+	}
+	table = table + ((address >> 12) & 0xff);
+	if (bad_address(table))
+		goto bad;
+	pr_cont("P:%016lx ", *table);
+out:
+	pr_cont("\n");
+	return;
+bad:
+	pr_cont("BAD\n");
 }
 
-static inline void report_user_fault(struct pt_regs *regs, long int_code,
-				     int signr, unsigned long address)
+#else /* CONFIG_64BIT */
+
+static void dump_pagetable(unsigned long asce, unsigned long address)
+{
+	unsigned long *table = __va(asce & PAGE_MASK);
+
+	pr_alert("AS:%08lx ", asce);
+	table = table + ((address >> 20) & 0x7ff);
+	if (bad_address(table))
+		goto bad;
+	pr_cont("S:%08lx ", *table);
+	if (*table & _SEGMENT_ENTRY_INVALID)
+		goto out;
+	table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
+	table = table + ((address >> 12) & 0xff);
+	if (bad_address(table))
+		goto bad;
+	pr_cont("P:%08lx ", *table);
+out:
+	pr_cont("\n");
+	return;
+bad:
+	pr_cont("BAD\n");
+}
+
+#endif /* CONFIG_64BIT */
+
+static void dump_fault_info(struct pt_regs *regs)
+{
+	unsigned long asce;
+
+	pr_alert("Fault in ");
+	switch (regs->int_parm_long & 3) {
+	case 3:
+		pr_cont("home space ");
+		break;
+	case 2:
+		pr_cont("secondary space ");
+		break;
+	case 1:
+		pr_cont("access register ");
+		break;
+	case 0:
+		pr_cont("primary space ");
+		break;
+	}
+	pr_cont("mode while using ");
+	if (!user_space_fault(regs)) {
+		asce = S390_lowcore.kernel_asce;
+		pr_cont("kernel ");
+	}
+#ifdef CONFIG_PGSTE
+	else if ((current->flags & PF_VCPU) && S390_lowcore.gmap) {
+		struct gmap *gmap = (struct gmap *)S390_lowcore.gmap;
+		asce = gmap->asce;
+		pr_cont("gmap ");
+	}
+#endif
+	else {
+		asce = S390_lowcore.user_asce;
+		pr_cont("user ");
+	}
+	pr_cont("ASCE.\n");
+	dump_pagetable(asce, regs->int_parm_long & __FAIL_ADDR_MASK);
+}
+
+static inline void report_user_fault(struct pt_regs *regs, long signr)
 {
 	if ((task_pid_nr(current) > 1) && !show_unhandled_signals)
 		return;
@@ -134,10 +261,13 @@ static inline void report_user_fault(struct pt_regs *regs, long int_code,
 		return;
 	if (!printk_ratelimit())
 		return;
-	printk("User process fault: interruption code 0x%lX ", int_code);
+	printk(KERN_ALERT "User process fault: interruption code 0x%X ",
+	       regs->int_code);
 	print_vma_addr(KERN_CONT "in ", regs->psw.addr & PSW_ADDR_INSN);
-	printk("\n");
-	printk("failing address: %lX\n", address);
+	printk(KERN_CONT "\n");
+	printk(KERN_ALERT "failing address: %016lx TEID: %016lx\n",
+	       regs->int_parm_long & __FAIL_ADDR_MASK, regs->int_parm_long);
+	dump_fault_info(regs);
 	show_regs(regs);
 }
 
@@ -145,24 +275,18 @@ static inline void report_user_fault(struct pt_regs *regs, long int_code,
  * Send SIGSEGV to task.  This is an external routine
  * to keep the stack usage of do_page_fault small.
  */
-static noinline void do_sigsegv(struct pt_regs *regs, long int_code,
-				int si_code, unsigned long trans_exc_code)
+static noinline void do_sigsegv(struct pt_regs *regs, int si_code)
 {
 	struct siginfo si;
-	unsigned long address;
 
-	address = trans_exc_code & __FAIL_ADDR_MASK;
-	current->thread.prot_addr = address;
-	current->thread.trap_no = int_code;
-	report_user_fault(regs, int_code, SIGSEGV, address);
+	report_user_fault(regs, SIGSEGV);
 	si.si_signo = SIGSEGV;
 	si.si_code = si_code;
-	si.si_addr = (void __user *) address;
+	si.si_addr = (void __user *)(regs->int_parm_long & __FAIL_ADDR_MASK);
 	force_sig_info(SIGSEGV, &si, current);
 }
 
-static noinline void do_no_context(struct pt_regs *regs, long int_code,
-				   unsigned long trans_exc_code)
+static noinline void do_no_context(struct pt_regs *regs)
 {
 	const struct exception_table_entry *fixup;
 	unsigned long address;
@@ -170,7 +294,7 @@ static noinline void do_no_context(struct pt_regs *regs, long int_code,
 	/* Are we prepared to handle this kernel fault?  */
 	fixup = search_exception_tables(regs->psw.addr & PSW_ADDR_INSN);
 	if (fixup) {
-		regs->psw.addr = fixup->fixup | PSW_ADDR_AMODE;
+		regs->psw.addr = extable_fixup(fixup) | PSW_ADDR_AMODE;
 		return;
 	}
 
@@ -178,115 +302,84 @@ static noinline void do_no_context(struct pt_regs *regs, long int_code,
 	 * Oops. The kernel tried to access some bad page. We'll have to
 	 * terminate things with extreme prejudice.
 	 */
-	address = trans_exc_code & __FAIL_ADDR_MASK;
-	if (!user_space_fault(trans_exc_code))
+	address = regs->int_parm_long & __FAIL_ADDR_MASK;
+	if (!user_space_fault(regs))
 		printk(KERN_ALERT "Unable to handle kernel pointer dereference"
-		       " at virtual kernel address %p\n", (void *)address);
+		       " in virtual kernel address space\n");
 	else
 		printk(KERN_ALERT "Unable to handle kernel paging request"
-		       " at virtual user address %p\n", (void *)address);
-
-	die("Oops", regs, int_code);
+		       " in virtual user address space\n");
+	printk(KERN_ALERT "failing address: %016lx TEID: %016lx\n",
+	       regs->int_parm_long & __FAIL_ADDR_MASK, regs->int_parm_long);
+	dump_fault_info(regs);
+	die(regs, "Oops");
 	do_exit(SIGKILL);
 }
 
-static noinline void do_low_address(struct pt_regs *regs, long int_code,
-				    unsigned long trans_exc_code)
+static noinline void do_low_address(struct pt_regs *regs)
 {
 	/* Low-address protection hit in kernel mode means
 	   NULL pointer write access in kernel mode.  */
 	if (regs->psw.mask & PSW_MASK_PSTATE) {
 		/* Low-address protection hit in user mode 'cannot happen'. */
-		die ("Low-address protection", regs, int_code);
+		die (regs, "Low-address protection");
 		do_exit(SIGKILL);
 	}
 
-	do_no_context(regs, int_code, trans_exc_code);
+	do_no_context(regs);
 }
 
-static noinline void do_sigbus(struct pt_regs *regs, long int_code,
-			       unsigned long trans_exc_code)
+static noinline void do_sigbus(struct pt_regs *regs)
 {
 	struct task_struct *tsk = current;
-	unsigned long address;
 	struct siginfo si;
 
 	/*
 	 * Send a sigbus, regardless of whether we were in kernel
 	 * or user mode.
 	 */
-	address = trans_exc_code & __FAIL_ADDR_MASK;
-	tsk->thread.prot_addr = address;
-	tsk->thread.trap_no = int_code;
 	si.si_signo = SIGBUS;
 	si.si_errno = 0;
 	si.si_code = BUS_ADRERR;
-	si.si_addr = (void __user *) address;
+	si.si_addr = (void __user *)(regs->int_parm_long & __FAIL_ADDR_MASK);
 	force_sig_info(SIGBUS, &si, tsk);
 }
 
-#ifdef CONFIG_S390_EXEC_PROTECT
-static noinline int signal_return(struct pt_regs *regs, long int_code,
-				  unsigned long trans_exc_code)
-{
-	u16 instruction;
-	int rc;
-
-	rc = __get_user(instruction, (u16 __user *) regs->psw.addr);
-
-	if (!rc && instruction == 0x0a77) {
-		clear_tsk_thread_flag(current, TIF_PER_TRAP);
-		if (is_compat_task())
-			sys32_sigreturn();
-		else
-			sys_sigreturn();
-	} else if (!rc && instruction == 0x0aad) {
-		clear_tsk_thread_flag(current, TIF_PER_TRAP);
-		if (is_compat_task())
-			sys32_rt_sigreturn();
-		else
-			sys_rt_sigreturn();
-	} else
-		do_sigsegv(regs, int_code, SEGV_MAPERR, trans_exc_code);
-	return 0;
-}
-#endif /* CONFIG_S390_EXEC_PROTECT */
-
-static noinline void do_fault_error(struct pt_regs *regs, long int_code,
-				    unsigned long trans_exc_code, int fault)
+static noinline void do_fault_error(struct pt_regs *regs, int fault)
 {
 	int si_code;
 
 	switch (fault) {
 	case VM_FAULT_BADACCESS:
-#ifdef CONFIG_S390_EXEC_PROTECT
-		if ((regs->psw.mask & PSW_MASK_ASC) == PSW_ASC_SECONDARY &&
-		    (trans_exc_code & 3) == 0) {
-			signal_return(regs, int_code, trans_exc_code);
-			break;
-		}
-#endif /* CONFIG_S390_EXEC_PROTECT */
 	case VM_FAULT_BADMAP:
 		/* Bad memory access. Check if it is kernel or user space. */
-		if (regs->psw.mask & PSW_MASK_PSTATE) {
+		if (user_mode(regs)) {
 			/* User mode accesses just cause a SIGSEGV */
 			si_code = (fault == VM_FAULT_BADMAP) ?
 				SEGV_MAPERR : SEGV_ACCERR;
-			do_sigsegv(regs, int_code, si_code, trans_exc_code);
+			do_sigsegv(regs, si_code);
 			return;
 		}
 	case VM_FAULT_BADCONTEXT:
-		do_no_context(regs, int_code, trans_exc_code);
+	case VM_FAULT_PFAULT:
+		do_no_context(regs);
+		break;
+	case VM_FAULT_SIGNAL:
+		if (!user_mode(regs))
+			do_no_context(regs);
 		break;
 	default: /* fault & VM_FAULT_ERROR */
-		if (fault & VM_FAULT_OOM)
-			pagefault_out_of_memory();
-		else if (fault & VM_FAULT_SIGBUS) {
+		if (fault & VM_FAULT_OOM) {
+			if (!user_mode(regs))
+				do_no_context(regs);
+			else
+				pagefault_out_of_memory();
+		} else if (fault & VM_FAULT_SIGBUS) {
 			/* Kernel mode? Handle exceptions or die */
-			if (!(regs->psw.mask & PSW_MASK_PSTATE))
-				do_no_context(regs, int_code, trans_exc_code);
+			if (!user_mode(regs))
+				do_no_context(regs);
 			else
-				do_sigbus(regs, int_code, trans_exc_code);
+				do_sigbus(regs);
 		} else
 			BUG();
 		break;
@@ -304,20 +397,31 @@ static noinline void do_fault_error(struct pt_regs *regs, long int_code,
  *   11       Page translation     ->  Not present       (nullification)
  *   3b       Region third trans.  ->  Not present       (nullification)
  */
-static inline int do_exception(struct pt_regs *regs, int access,
-			       unsigned long trans_exc_code)
+static inline int do_exception(struct pt_regs *regs, int access)
 {
+#ifdef CONFIG_PGSTE
+	struct gmap *gmap;
+#endif
 	struct task_struct *tsk;
 	struct mm_struct *mm;
 	struct vm_area_struct *vma;
+	unsigned long trans_exc_code;
 	unsigned long address;
-	int fault, write;
+	unsigned int flags;
+	int fault;
+
+	tsk = current;
+	/*
+	 * The instruction that caused the program check has
+	 * been nullified. Don't signal single step via SIGTRAP.
+	 */
+	clear_pt_regs_flag(regs, PIF_PER_TRAP);
 
 	if (notify_page_fault(regs))
 		return 0;
 
-	tsk = current;
 	mm = tsk->mm;
+	trans_exc_code = regs->int_parm_long;
 
 	/*
 	 * Verify that the fault happened in user space, that
@@ -325,13 +429,37 @@ static inline int do_exception(struct pt_regs *regs, int access,
 	 * user context.
 	 */
 	fault = VM_FAULT_BADCONTEXT;
-	if (unlikely(!user_space_fault(trans_exc_code) || in_atomic() || !mm))
+	if (unlikely(!user_space_fault(regs) || in_atomic() || !mm))
 		goto out;
 
 	address = trans_exc_code & __FAIL_ADDR_MASK;
-	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, 0, regs, address);
+	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
+	flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
+	if (user_mode(regs))
+		flags |= FAULT_FLAG_USER;
+	if (access == VM_WRITE || (trans_exc_code & store_indication) == 0x400)
+		flags |= FAULT_FLAG_WRITE;
 	down_read(&mm->mmap_sem);
 
+#ifdef CONFIG_PGSTE
+	gmap = (struct gmap *)
+		((current->flags & PF_VCPU) ? S390_lowcore.gmap : 0);
+	if (gmap) {
+		address = __gmap_fault(address, gmap);
+		if (address == -EFAULT) {
+			fault = VM_FAULT_BADMAP;
+			goto out_up;
+		}
+		if (address == -ENOMEM) {
+			fault = VM_FAULT_OOM;
+			goto out_up;
+		}
+		if (gmap->pfault_enabled)
+			flags |= FAULT_FLAG_RETRY_NOWAIT;
+	}
+#endif
+
+retry:
 	fault = VM_FAULT_BADMAP;
 	vma = find_vma(mm, address);
 	if (!vma)
@@ -359,27 +487,49 @@ static inline int do_exception(struct pt_regs *regs, int access,
 	 * make sure we exit gracefully rather than endlessly redo
 	 * the fault.
 	 */
-	write = (access == VM_WRITE ||
-		 (trans_exc_code & store_indication) == 0x400) ?
-		FAULT_FLAG_WRITE : 0;
-	fault = handle_mm_fault(mm, vma, address, write);
+	fault = handle_mm_fault(mm, vma, address, flags);
+	/* No reason to continue if interrupted by SIGKILL. */
+	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current)) {
+		fault = VM_FAULT_SIGNAL;
+		goto out;
+	}
 	if (unlikely(fault & VM_FAULT_ERROR))
 		goto out_up;
 
-	if (fault & VM_FAULT_MAJOR) {
-		tsk->maj_flt++;
-		perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ, 1, 0,
-				     regs, address);
-	} else {
-		tsk->min_flt++;
-		perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN, 1, 0,
-				     regs, address);
-	}
 	/*
-	 * The instruction that caused the program check will
-	 * be repeated. Don't signal single step via SIGTRAP.
+	 * Major/minor page fault accounting is only done on the
+	 * initial attempt. If we go through a retry, it is extremely
+	 * likely that the page will be found in page cache at that point.
 	 */
-	clear_tsk_thread_flag(tsk, TIF_PER_TRAP);
+	if (flags & FAULT_FLAG_ALLOW_RETRY) {
+		if (fault & VM_FAULT_MAJOR) {
+			tsk->maj_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MAJ, 1,
+				      regs, address);
+		} else {
+			tsk->min_flt++;
+			perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS_MIN, 1,
+				      regs, address);
+		}
+		if (fault & VM_FAULT_RETRY) {
+#ifdef CONFIG_PGSTE
+			if (gmap && (flags & FAULT_FLAG_RETRY_NOWAIT)) {
+				/* FAULT_FLAG_RETRY_NOWAIT has been set,
+				 * mmap_sem has not been released */
+				current->thread.gmap_pfault = 1;
+				fault = VM_FAULT_PFAULT;
+				goto out_up;
+			}
+#endif
+			/* Clear FAULT_FLAG_ALLOW_RETRY to avoid any risk
+			 * of starvation. */
+			flags &= ~(FAULT_FLAG_ALLOW_RETRY |
+				   FAULT_FLAG_RETRY_NOWAIT);
+			flags |= FAULT_FLAG_TRIED;
+			down_read(&mm->mmap_sem);
+			goto retry;
+		}
+	}
 	fault = 0;
 out_up:
 	up_read(&mm->mmap_sem);
@@ -387,94 +537,41 @@ out:
 	return fault;
 }
 
-void __kprobes do_protection_exception(struct pt_regs *regs, long pgm_int_code,
-				       unsigned long trans_exc_code)
+void __kprobes do_protection_exception(struct pt_regs *regs)
 {
+	unsigned long trans_exc_code;
 	int fault;
 
-	/* Protection exception is supressing, decrement psw address. */
-	regs->psw.addr -= (pgm_int_code >> 16);
+	trans_exc_code = regs->int_parm_long;
+	/*
+	 * Protection exceptions are suppressing, decrement psw address.
+	 * The exception to this rule are aborted transactions, for these
+	 * the PSW already points to the correct location.
+	 */
+	if (!(regs->int_code & 0x200))
+		regs->psw.addr = __rewind_psw(regs->psw, regs->int_code >> 16);
 	/*
 	 * 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 (unlikely(!(trans_exc_code & 4))) {
-		do_low_address(regs, pgm_int_code, trans_exc_code);
+		do_low_address(regs);
 		return;
 	}
-	fault = do_exception(regs, VM_WRITE, trans_exc_code);
+	fault = do_exception(regs, VM_WRITE);
 	if (unlikely(fault))
-		do_fault_error(regs, 4, trans_exc_code, fault);
+		do_fault_error(regs, fault);
 }
 
-void __kprobes do_dat_exception(struct pt_regs *regs, long pgm_int_code,
-				unsigned long trans_exc_code)
+void __kprobes do_dat_exception(struct pt_regs *regs)
 {
 	int access, fault;
 
 	access = VM_READ | VM_EXEC | VM_WRITE;
-#ifdef CONFIG_S390_EXEC_PROTECT
-	if ((regs->psw.mask & PSW_MASK_ASC) == PSW_ASC_SECONDARY &&
-	    (trans_exc_code & 3) == 0)
-		access = VM_EXEC;
-#endif
-	fault = do_exception(regs, access, trans_exc_code);
+	fault = do_exception(regs, access);
 	if (unlikely(fault))
-		do_fault_error(regs, pgm_int_code & 255, trans_exc_code, fault);
-}
-
-#ifdef CONFIG_64BIT
-void __kprobes do_asce_exception(struct pt_regs *regs, long pgm_int_code,
-				 unsigned long trans_exc_code)
-{
-	struct mm_struct *mm = current->mm;
-	struct vm_area_struct *vma;
-
-	if (unlikely(!user_space_fault(trans_exc_code) || in_atomic() || !mm))
-		goto no_context;
-
-	down_read(&mm->mmap_sem);
-	vma = find_vma(mm, trans_exc_code & __FAIL_ADDR_MASK);
-	up_read(&mm->mmap_sem);
-
-	if (vma) {
-		update_mm(mm, current);
-		return;
-	}
-
-	/* User mode accesses just cause a SIGSEGV */
-	if (regs->psw.mask & PSW_MASK_PSTATE) {
-		do_sigsegv(regs, pgm_int_code, SEGV_MAPERR, trans_exc_code);
-		return;
-	}
-
-no_context:
-	do_no_context(regs, pgm_int_code, trans_exc_code);
-}
-#endif
-
-int __handle_fault(unsigned long uaddr, unsigned long pgm_int_code, int write)
-{
-	struct pt_regs regs;
-	int access, fault;
-
-	regs.psw.mask = psw_kernel_bits;
-	if (!irqs_disabled())
-		regs.psw.mask |= PSW_MASK_IO | PSW_MASK_EXT;
-	regs.psw.addr = (unsigned long) __builtin_return_address(0);
-	regs.psw.addr |= PSW_ADDR_AMODE;
-	uaddr &= PAGE_MASK;
-	access = write ? VM_WRITE : VM_READ;
-	fault = do_exception(&regs, access, uaddr | 2);
-	if (unlikely(fault)) {
-		if (fault & VM_FAULT_OOM) {
-			pagefault_out_of_memory();
-			fault = 0;
-		} else if (fault & VM_FAULT_SIGBUS)
-			do_sigbus(&regs, pgm_int_code, uaddr);
-	}
-	return fault ? -EFAULT : 0;
+		do_fault_error(regs, fault);
 }
 
 #ifdef CONFIG_PFAULT 
@@ -491,25 +588,31 @@ static int __init nopfault(char *str)
 
 __setup("nopfault", nopfault);
 
-typedef struct {
-	__u16 refdiagc;
-	__u16 reffcode;
-	__u16 refdwlen;
-	__u16 refversn;
-	__u64 refgaddr;
-	__u64 refselmk;
-	__u64 refcmpmk;
-	__u64 reserved;
-} __attribute__ ((packed, aligned(8))) pfault_refbk_t;
+struct pfault_refbk {
+	u16 refdiagc;
+	u16 reffcode;
+	u16 refdwlen;
+	u16 refversn;
+	u64 refgaddr;
+	u64 refselmk;
+	u64 refcmpmk;
+	u64 reserved;
+} __attribute__ ((packed, aligned(8)));
 
 int pfault_init(void)
 {
-	pfault_refbk_t refbk =
-		{ 0x258, 0, 5, 2, __LC_CURRENT, 1ULL << 48, 1ULL << 48,
-		  __PF_RES_FIELD };
+	struct pfault_refbk refbk = {
+		.refdiagc = 0x258,
+		.reffcode = 0,
+		.refdwlen = 5,
+		.refversn = 2,
+		.refgaddr = __LC_CURRENT_PID,
+		.refselmk = 1ULL << 48,
+		.refcmpmk = 1ULL << 48,
+		.reserved = __PF_RES_FIELD };
         int rc;
 
-	if (!MACHINE_IS_VM || pfault_disable)
+	if (pfault_disable)
 		return -1;
 	asm volatile(
 		"	diag	%1,%0,0x258\n"
@@ -518,18 +621,20 @@ int pfault_init(void)
 		"2:\n"
 		EX_TABLE(0b,1b)
 		: "=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 (!MACHINE_IS_VM || pfault_disable)
+	struct pfault_refbk refbk = {
+		.refdiagc = 0x258,
+		.reffcode = 1,
+		.refdwlen = 5,
+		.refversn = 2,
+	};
+
+	if (pfault_disable)
 		return;
-	__ctl_clear_bit(0,9);
 	asm volatile(
 		"	diag	%0,0,0x258\n"
 		"0:\n"
@@ -537,84 +642,134 @@ void pfault_fini(void)
 		: : "a" (&refbk), "m" (refbk) : "cc");
 }
 
-static void pfault_interrupt(unsigned int ext_int_code,
+static DEFINE_SPINLOCK(pfault_lock);
+static LIST_HEAD(pfault_list);
+
+static void pfault_interrupt(struct ext_code ext_code,
 			     unsigned int param32, unsigned long param64)
 {
 	struct task_struct *tsk;
 	__u16 subcode;
+	pid_t pid;
 
-	kstat_cpu(smp_processor_id()).irqs[EXTINT_PFL]++;
 	/*
 	 * Get the external interruption subcode & pfault
 	 * initial/completion signal bit. VM stores this 
 	 * in the 'cpu address' field associated with the
          * external interrupt. 
 	 */
-	subcode = ext_int_code >> 16;
+	subcode = ext_code.subcode;
 	if ((subcode & 0xff00) != __SUBCODE_MASK)
 		return;
-
-	/*
-	 * Get the token (= address of the task structure of the affected task).
-	 */
-#ifdef CONFIG_64BIT
-	tsk = *(struct task_struct **) param64;
-#else
-	tsk = *(struct task_struct **) param32;
-#endif
-
+	inc_irq_stat(IRQEXT_PFL);
+	/* Get the token (= pid of the affected task). */
+	pid = sizeof(void *) == 4 ? param32 : param64;
+	rcu_read_lock();
+	tsk = find_task_by_pid_ns(pid, &init_pid_ns);
+	if (tsk)
+		get_task_struct(tsk);
+	rcu_read_unlock();
+	if (!tsk)
+		return;
+	spin_lock(&pfault_lock);
 	if (subcode & 0x0080) {
 		/* signal bit is set -> a page has been swapped in by VM */
-		if (xchg(&tsk->thread.pfault_wait, -1) != 0) {
+		if (tsk->thread.pfault_wait == 1) {
 			/* 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;
+			list_del(&tsk->thread.list);
 			wake_up_process(tsk);
 			put_task_struct(tsk);
+		} else {
+			/* Completion interrupt was faster than initial
+			 * interrupt. Set pfault_wait to -1 so the initial
+			 * interrupt doesn't put the task to sleep.
+			 * If the task is not running, ignore the completion
+			 * interrupt since it must be a leftover of a PFAULT
+			 * CANCEL operation which didn't remove all pending
+			 * completion interrupts. */
+			if (tsk->state == TASK_RUNNING)
+				tsk->thread.pfault_wait = -1;
 		}
 	} 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) {
+		if (WARN_ON_ONCE(tsk != current))
+			goto out;
+		if (tsk->thread.pfault_wait == 1) {
+			/* Already on the list with a reference: put to sleep */
+			__set_task_state(tsk, TASK_UNINTERRUPTIBLE);
+			set_tsk_need_resched(tsk);
+		} else if (tsk->thread.pfault_wait == -1) {
 			/* 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. */
+			 * interrupt (pfault_wait == -1). Set pfault_wait
+			 * back to zero and exit. */
 			tsk->thread.pfault_wait = 0;
-			set_task_state(tsk, TASK_RUNNING);
-			put_task_struct(tsk);
-		} else
+		} else {
+			/* Initial interrupt arrived before completion
+			 * interrupt. Let the task sleep.
+			 * An extra task reference is needed since a different
+			 * cpu may set the task state to TASK_RUNNING again
+			 * before the scheduler is reached. */
+			get_task_struct(tsk);
+			tsk->thread.pfault_wait = 1;
+			list_add(&tsk->thread.list, &pfault_list);
+			__set_task_state(tsk, TASK_UNINTERRUPTIBLE);
 			set_tsk_need_resched(tsk);
+		}
 	}
+out:
+	spin_unlock(&pfault_lock);
+	put_task_struct(tsk);
+}
+
+static int pfault_cpu_notify(struct notifier_block *self, unsigned long action,
+			     void *hcpu)
+{
+	struct thread_struct *thread, *next;
+	struct task_struct *tsk;
+
+	switch (action & ~CPU_TASKS_FROZEN) {
+	case CPU_DEAD:
+		spin_lock_irq(&pfault_lock);
+		list_for_each_entry_safe(thread, next, &pfault_list, list) {
+			thread->pfault_wait = 0;
+			list_del(&thread->list);
+			tsk = container_of(thread, struct task_struct, thread);
+			wake_up_process(tsk);
+			put_task_struct(tsk);
+		}
+		spin_unlock_irq(&pfault_lock);
+		break;
+	default:
+		break;
+	}
+	return NOTIFY_OK;
 }
 
 static int __init pfault_irq_init(void)
 {
 	int rc;
 
-	if (!MACHINE_IS_VM)
-		return 0;
-	/*
-	 * Try to get pfault pseudo page faults going.
-	 */
-	rc = register_external_interrupt(0x2603, pfault_interrupt);
-	if (rc) {
-		pfault_disable = 1;
-		return rc;
-	}
-	if (pfault_init() == 0)
-		return 0;
+	rc = register_external_irq(EXT_IRQ_CP_SERVICE, pfault_interrupt);
+	if (rc)
+		goto out_extint;
+	rc = pfault_init() == 0 ? 0 : -EOPNOTSUPP;
+	if (rc)
+		goto out_pfault;
+	irq_subclass_register(IRQ_SUBCLASS_SERVICE_SIGNAL);
+	hotcpu_notifier(pfault_cpu_notify, 0);
+	return 0;
 
-	/* Tough luck, no pfault. */
+out_pfault:
+	unregister_external_irq(EXT_IRQ_CP_SERVICE, pfault_interrupt);
+out_extint:
 	pfault_disable = 1;
-	unregister_external_interrupt(0x2603, pfault_interrupt);
-	return 0;
+	return rc;
 }
 early_initcall(pfault_irq_init);
 
-#endif
+#endif /* CONFIG_PFAULT */