s390 patches for the 3.6 merge window #2

index 5c63615..b749c57 100644
@@ -11,7 +11,6 @@
 #include <asm/uaccess.h>
 #include <asm/tlbflush.h>
 #include <asm/ctl_reg.h>
-#include <asm-generic/mm_hooks.h>
 
 static inline int init_new_context(struct task_struct *tsk,
 				   struct mm_struct *mm)

@@ -58,7 +57,7 @@ static inline void update_mm(struct mm_struct *mm, struct task_struct *tsk)
 	pgd_t *pgd = mm->pgd;
 
 	S390_lowcore.user_asce = mm->context.asce_bits | __pa(pgd);
-	if (user_mode != HOME_SPACE_MODE) {
+	if (addressing_mode != HOME_SPACE_MODE) {
 		/* Load primary space page table origin. */
 		asm volatile(LCTL_OPCODE" 1,1,%0\n"
 			     : : "m" (S390_lowcore.user_asce) );

@@ -91,4 +90,17 @@ static inline void activate_mm(struct mm_struct *prev,
         switch_mm(prev, next, current);
 }
 
+static inline void arch_dup_mmap(struct mm_struct *oldmm,
+				 struct mm_struct *mm)
+{
+#ifdef CONFIG_64BIT
+	if (oldmm->context.asce_limit < mm->context.asce_limit)
+		crst_table_downgrade(mm, oldmm->context.asce_limit);
+#endif
+}
+
+static inline void arch_exit_mmap(struct mm_struct *mm)
+{
+}
+
 #endif /* __S390_MMU_CONTEXT_H */

index c40fa91..11e4e32 100644
@@ -120,7 +120,9 @@ struct stack_frame {
 	regs->psw.mask	= psw_user_bits | PSW_MASK_BA;			\
 	regs->psw.addr	= new_psw | PSW_ADDR_AMODE;			\
 	regs->gprs[15]	= new_stackp;					\
+	__tlb_flush_mm(current->mm);					\
 	crst_table_downgrade(current->mm, 1UL << 31);			\
+	update_mm(current->mm, current);				\
 } while (0)
 
 /* Forward declaration, a strange C thing */

index 57e8053..e6859d1 100644
@@ -60,7 +60,7 @@ void create_mem_hole(struct mem_chunk memory_chunk[], unsigned long addr,
 #define SECONDARY_SPACE_MODE	2
 #define HOME_SPACE_MODE		3
 
-extern unsigned int user_mode;
+extern unsigned int addressing_mode;
 
 /*
  * Machine features detected in head.S

161 lines not on this page · raw message

index 1f6b428..619c5d3 100644
@@ -1531,7 +1531,7 @@ static int print_insn(char *buffer, unsigned char *code, unsigned long addr)
 
 void show_code(struct pt_regs *regs)
 {
-	char *mode = (regs->psw.mask & PSW_MASK_PSTATE) ? "User" : "Krnl";
+	char *mode = user_mode(regs) ? "User" : "Krnl";
 	unsigned char code[64];
 	char buffer[64], *ptr;
 	mm_segment_t old_fs;

@@ -1540,7 +1540,7 @@ void show_code(struct pt_regs *regs)
 
 	/* Get a snapshot of the 64 bytes surrounding the fault address. */
 	old_fs = get_fs();
-	set_fs((regs->psw.mask & PSW_MASK_PSTATE) ? USER_DS : KERNEL_DS);
+	set_fs(user_mode(regs) ? USER_DS : KERNEL_DS);
 	for (start = 32; start && regs->psw.addr >= 34 - start; start -= 2) {
 		addr = regs->psw.addr - 34 + start;
 		if (__copy_from_user(code + start - 2,

index bc95a8e..83c3271 100644
@@ -455,7 +455,6 @@ void __init startup_init(void)
 	init_kernel_storage_key();
 	lockdep_init();
 	lockdep_off();
-	sort_main_extable();
 	setup_lowcore_early();
 	setup_facility_list();
 	detect_machine_type();

index e64d141..6ffcd32 100644
@@ -1583,7 +1583,7 @@ static struct kset *vmcmd_kset;
 
 static void vmcmd_run(struct shutdown_trigger *trigger)
 {
-	char *cmd, *next_cmd;
+	char *cmd;
 
 	if (strcmp(trigger->name, ON_REIPL_STR) == 0)
 		cmd = vmcmd_on_reboot;

@@ -1600,15 +1600,7 @@ static void vmcmd_run(struct shutdown_trigger *trigger)
 
 	if (strlen(cmd) == 0)
 		return;
-	do {
-		next_cmd = strchr(cmd, '\n');
-		if (next_cmd) {
-			next_cmd[0] = 0;
-			next_cmd += 1;
-		}
-		__cpcmd(cmd, NULL, 0, NULL);
-		cmd = next_cmd;
-	} while (cmd != NULL);
+	__cpcmd(cmd, NULL, 0, NULL);
 }
 
 static int vmcmd_init(void)

index 743c0f3..f86c81e 100644
@@ -302,8 +302,8 @@ static int __init parse_vmalloc(char *arg)
 }
 early_param("vmalloc", parse_vmalloc);
 
-unsigned int user_mode = HOME_SPACE_MODE;
-EXPORT_SYMBOL_GPL(user_mode);
+unsigned int addressing_mode = HOME_SPACE_MODE;
+EXPORT_SYMBOL_GPL(addressing_mode);
 
 static int set_amode_primary(void)
 {

@@ -328,7 +328,7 @@ static int set_amode_primary(void)
  */
 static int __init early_parse_switch_amode(char *p)
 {
-	user_mode = PRIMARY_SPACE_MODE;
+	addressing_mode = PRIMARY_SPACE_MODE;
 	return 0;
 }
 early_param("switch_amode", early_parse_switch_amode);

@@ -336,9 +336,9 @@ early_param("switch_amode", early_parse_switch_amode);
 static int __init early_parse_user_mode(char *p)
 {
 	if (p && strcmp(p, "primary") == 0)
-		user_mode = PRIMARY_SPACE_MODE;
+		addressing_mode = PRIMARY_SPACE_MODE;
 	else if (!p || strcmp(p, "home") == 0)
-		user_mode = HOME_SPACE_MODE;
+		addressing_mode = HOME_SPACE_MODE;
 	else
 		return 1;
 	return 0;

@@ -347,7 +347,7 @@ early_param("user_mode", early_parse_user_mode);
 
 static void setup_addressing_mode(void)
 {
-	if (user_mode == PRIMARY_SPACE_MODE) {
+	if (addressing_mode == PRIMARY_SPACE_MODE) {
 		if (set_amode_primary())
 			pr_info("Address spaces switched, "
 				"mvcos available\n");

index af2421a..01775c0 100644
@@ -185,7 +185,7 @@ void show_registers(struct pt_regs *regs)
 {
 	char *mode;
 
-	mode = (regs->psw.mask & PSW_MASK_PSTATE) ? "User" : "Krnl";
+	mode = user_mode(regs) ? "User" : "Krnl";
 	printk("%s PSW : %p %p",
 	       mode, (void *) regs->psw.mask,
 	       (void *) regs->psw.addr);

@@ -225,7 +225,7 @@ void show_regs(struct pt_regs *regs)
 	       (void *) current->thread.ksp);
 	show_registers(regs);
 	/* Show stack backtrace if pt_regs is from kernel mode */
-	if (!(regs->psw.mask & PSW_MASK_PSTATE))
+	if (!user_mode(regs))
 		show_trace(NULL, (unsigned long *) regs->gprs[15]);
 	show_last_breaking_event(regs);
 }

@@ -300,7 +300,7 @@ static void __kprobes do_trap(struct pt_regs *regs,
 		       regs->int_code, si_signo) == NOTIFY_STOP)
 		return;
 
-        if (regs->psw.mask & PSW_MASK_PSTATE) {
+	if (user_mode(regs)) {
 		info.si_signo = si_signo;
 		info.si_errno = 0;
 		info.si_code = si_code;

@@ -341,7 +341,7 @@ void __kprobes do_per_trap(struct pt_regs *regs)
 
 static void default_trap_handler(struct pt_regs *regs)
 {
-        if (regs->psw.mask & PSW_MASK_PSTATE) {
+	if (user_mode(regs)) {
 		report_user_fault(regs, SIGSEGV);
 		do_exit(SIGSEGV);
 	} else

@@ -410,7 +410,7 @@ static void __kprobes illegal_op(struct pt_regs *regs)
 
 	location = get_psw_address(regs);
 
-	if (regs->psw.mask & PSW_MASK_PSTATE) {
+	if (user_mode(regs)) {
 		if (get_user(*((__u16 *) opcode), (__u16 __user *) location))
 			return;
 		if (*((__u16 *) opcode) == S390_BREAKPOINT_U16) {

@@ -478,7 +478,7 @@ void specification_exception(struct pt_regs *regs)
 
 	location = (__u16 __user *) get_psw_address(regs);
 
-        if (regs->psw.mask & PSW_MASK_PSTATE) {
+	if (user_mode(regs)) {
 		get_user(*((__u16 *) opcode), location);
 		switch (opcode[0]) {
 		case 0x28: /* LDR Rx,Ry   */

@@ -531,7 +531,7 @@ static void data_exception(struct pt_regs *regs)
 		asm volatile("stfpc %0" : "=m" (current->thread.fp_regs.fpc));
 
 #ifdef CONFIG_MATHEMU
-        else if (regs->psw.mask & PSW_MASK_PSTATE) {
+	else if (user_mode(regs)) {
         	__u8 opcode[6];
 		get_user(*((__u16 *) opcode), location);
 		switch (opcode[0]) {

@@ -598,7 +598,7 @@ static void data_exception(struct pt_regs *regs)
 static void space_switch_exception(struct pt_regs *regs)
 {
 	/* Set user psw back to home space mode. */
-	if (regs->psw.mask & PSW_MASK_PSTATE)
+	if (user_mode(regs))
 		regs->psw.mask |= PSW_ASC_HOME;
 	/* Send SIGILL. */
 	do_trap(regs, SIGILL, ILL_PRVOPC, "space switch event");

index ea5590f..9a19ca3 100644
@@ -84,7 +84,8 @@ struct vdso_data *vdso_data = &vdso_data_store.data;
  */
 static void vdso_init_data(struct vdso_data *vd)
 {
-	vd->ectg_available = user_mode != HOME_SPACE_MODE && test_facility(31);
+	vd->ectg_available =
+		addressing_mode != HOME_SPACE_MODE && test_facility(31);
 }
 
 #ifdef CONFIG_64BIT

@@ -101,7 +102,7 @@ int vdso_alloc_per_cpu(struct _lowcore *lowcore)
 
 	lowcore->vdso_per_cpu_data = __LC_PASTE;
 
-	if (user_mode == HOME_SPACE_MODE || !vdso_enabled)
+	if (addressing_mode == HOME_SPACE_MODE || !vdso_enabled)
 		return 0;
 
 	segment_table = __get_free_pages(GFP_KERNEL, SEGMENT_ORDER);

@@ -146,7 +147,7 @@ void vdso_free_per_cpu(struct _lowcore *lowcore)
 	unsigned long segment_table, page_table, page_frame;
 	u32 *psal, *aste;
 
-	if (user_mode == HOME_SPACE_MODE || !vdso_enabled)
+	if (addressing_mode == HOME_SPACE_MODE || !vdso_enabled)
 		return;
 
 	psal = (u32 *)(addr_t) lowcore->paste[4];

@@ -164,7 +165,7 @@ static void vdso_init_cr5(void)
 {
 	unsigned long cr5;
 
-	if (user_mode == HOME_SPACE_MODE || !vdso_enabled)
+	if (addressing_mode == HOME_SPACE_MODE || !vdso_enabled)
 		return;
 	cr5 = offsetof(struct _lowcore, paste);
 	__ctl_load(cr5, 5, 5);

index 6a12d1b..6c013f5 100644
@@ -49,6 +49,7 @@
 #define VM_FAULT_BADCONTEXT	0x010000
 #define VM_FAULT_BADMAP		0x020000
 #define VM_FAULT_BADACCESS	0x040000
+#define VM_FAULT_SIGNAL	0x080000
 
 static unsigned long store_indication;
 

@@ -110,7 +111,7 @@ static inline int user_space_fault(unsigned long trans_exc_code)
 	if (trans_exc_code == 2)
 		/* Access via secondary space, set_fs setting decides */
 		return current->thread.mm_segment.ar4;
-	if (user_mode == HOME_SPACE_MODE)
+	if (addressing_mode == HOME_SPACE_MODE)
 		/* User space if the access has been done via home space. */
 		return trans_exc_code == 3;
 	/*

@@ -219,7 +220,7 @@ static noinline void do_fault_error(struct pt_regs *regs, int fault)
 	case VM_FAULT_BADACCESS:
 	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;

@@ -229,15 +230,19 @@ static noinline void do_fault_error(struct pt_regs *regs, int fault)
 	case VM_FAULT_BADCONTEXT:
 		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) {
-			if (!(regs->psw.mask & PSW_MASK_PSTATE))
+			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))
+			if (!user_mode(regs))
 				do_no_context(regs);
 			else
 				do_sigbus(regs);

@@ -286,7 +291,7 @@ static inline int do_exception(struct pt_regs *regs, int access)
 
 	address = trans_exc_code & __FAIL_ADDR_MASK;
 	perf_sw_event(PERF_COUNT_SW_PAGE_FAULTS, 1, regs, address);
-	flags = FAULT_FLAG_ALLOW_RETRY;
+	flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE;
 	if (access == VM_WRITE || (trans_exc_code & store_indication) == 0x400)
 		flags |= FAULT_FLAG_WRITE;
 	down_read(&mm->mmap_sem);

@@ -335,6 +340,11 @@ retry:
 	 * the fault.
 	 */
 	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;
 

@@ -426,7 +436,7 @@ void __kprobes do_asce_exception(struct pt_regs *regs)
 	}
 
 	/* User mode accesses just cause a SIGSEGV */
-	if (regs->psw.mask & PSW_MASK_PSTATE) {
+	if (user_mode(regs)) {
 		do_sigsegv(regs, SEGV_MAPERR);
 		return;
 	}

@@ -441,6 +451,7 @@ int __handle_fault(unsigned long uaddr, unsigned long pgm_int_code, int write)
 	struct pt_regs regs;
 	int access, fault;
 
+	/* Emulate a uaccess fault from kernel mode. */
 	regs.psw.mask = psw_kernel_bits | PSW_MASK_DAT | PSW_MASK_MCHECK;
 	if (!irqs_disabled())
 		regs.psw.mask |= PSW_MASK_IO | PSW_MASK_EXT;

@@ -450,12 +461,12 @@ int __handle_fault(unsigned long uaddr, unsigned long pgm_int_code, int write)
 	regs.int_parm_long = (uaddr & PAGE_MASK) | 2;
 	access = write ? VM_WRITE : VM_READ;
 	fault = do_exception(&regs, access);
-	if (unlikely(fault)) {
-		if (fault & VM_FAULT_OOM)
-			return -EFAULT;
-		else if (fault & VM_FAULT_SIGBUS)
-			do_sigbus(&regs);
-	}
+	/*
+	 * Since the fault happened in kernel mode while performing a uaccess
+	 * all we need to do now is emulating a fixup in case "fault" is not
+	 * zero.
+	 * For the calling uaccess functions this results always in -EFAULT.
+	 */
 	return fault ? -EFAULT : 0;
 }
 

index 5733842..c59a5ef 100644
@@ -103,9 +103,15 @@ void arch_pick_mmap_layout(struct mm_struct *mm)
 
 int s390_mmap_check(unsigned long addr, unsigned long len)
 {
+	int rc;
+
 	if (!is_compat_task() &&
-	    len >= TASK_SIZE && TASK_SIZE < (1UL << 53))
-		return crst_table_upgrade(current->mm, 1UL << 53);
+	    len >= TASK_SIZE && TASK_SIZE < (1UL << 53)) {
+		rc = crst_table_upgrade(current->mm, 1UL << 53);
+		if (rc)
+			return rc;
+		update_mm(current->mm, current);
+	}
 	return 0;
 }
 

@@ -125,6 +131,7 @@ s390_get_unmapped_area(struct file *filp, unsigned long addr,
 		rc = crst_table_upgrade(mm, 1UL << 53);
 		if (rc)
 			return (unsigned long) rc;
+		update_mm(mm, current);
 		area = arch_get_unmapped_area(filp, addr, len, pgoff, flags);
 	}
 	return area;

@@ -147,6 +154,7 @@ s390_get_unmapped_area_topdown(struct file *filp, const unsigned long addr,
 		rc = crst_table_upgrade(mm, 1UL << 53);
 		if (rc)
 			return (unsigned long) rc;
+		update_mm(mm, current);
 		area = arch_get_unmapped_area_topdown(filp, addr, len,
 						      pgoff, flags);
 	}

index 1cab221..18df31d 100644
@@ -85,7 +85,6 @@ repeat:
 		crst_table_free(mm, table);
 	if (mm->context.asce_limit < limit)
 		goto repeat;
-	update_mm(mm, current);
 	return 0;
 }
 

@@ -93,9 +92,6 @@ void crst_table_downgrade(struct mm_struct *mm, unsigned long limit)
 {
 	pgd_t *pgd;
 
-	if (mm->context.asce_limit <= limit)
-		return;
-	__tlb_flush_mm(mm);
 	while (mm->context.asce_limit > limit) {
 		pgd = mm->pgd;
 		switch (pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) {

@@ -118,7 +114,6 @@ void crst_table_downgrade(struct mm_struct *mm, unsigned long limit)
 		mm->task_size = mm->context.asce_limit;
 		crst_table_free(mm, (unsigned long *) pgd);
 	}
-	update_mm(mm, current);
 }
 #endif
 

@@ -801,7 +796,7 @@ int s390_enable_sie(void)
 	struct mm_struct *mm, *old_mm;
 
 	/* Do we have switched amode? If no, we cannot do sie */
-	if (user_mode == HOME_SPACE_MODE)
+	if (addressing_mode == HOME_SPACE_MODE)
 		return -EINVAL;
 
 	/* Do we have pgstes? if yes, we are done */

index c82f62f..8a6811b 100644
@@ -58,7 +58,7 @@ void s390_backtrace(struct pt_regs * const regs, unsigned int depth)
 	unsigned long head;
 	struct stack_frame* head_sf;
 
-	if (user_mode (regs))
+	if (user_mode(regs))
 		return;
 
 	head = regs->gprs[15];

index 1ca9ceb..6acf834 100644
@@ -247,6 +247,7 @@ do_file(char const *const fname)
 	case EM_X86_64:
 		custom_sort = sort_x86_table;
 		break;
+	case EM_S390:
 	case EM_MIPS:
 		break;
 	}  /* end switch */