s390: add no-execute support

Bit 0x100 of a page table, segment table of region table entry
can be used to disallow code execution for the virtual addresses
associated with the entry.

There is one tricky bit, the system call to return from a signal
is part of the signal frame written to the user stack. With a
non-executable stack this would stop working. To avoid breaking
things the protection fault handler checks the opcode that caused
the fault for 0x0a77 (sys_sigreturn) and 0x0aad (sys_rt_sigreturn)
and injects a system call. This is preferable to the alternative
solution with a stub function in the vdso because it works for
vdso=off and statically linked binaries as well.

Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/include/asm/cacheflush.h

@@ -4,9 +4,31 @@
 /* Caches aren't brain-dead on the s390. */
 #include <asm-generic/cacheflush.h>
 
-int set_memory_ro(unsigned long addr, int numpages);
-int set_memory_rw(unsigned long addr, int numpages);
-int set_memory_nx(unsigned long addr, int numpages);
-int set_memory_x(unsigned long addr, int numpages);
+#define SET_MEMORY_RO	1UL
+#define SET_MEMORY_RW	2UL
+#define SET_MEMORY_NX	4UL
+#define SET_MEMORY_X	8UL
+
+int __set_memory(unsigned long addr, int numpages, unsigned long flags);
+
+static inline int set_memory_ro(unsigned long addr, int numpages)
+{
+	return __set_memory(addr, numpages, SET_MEMORY_RO);
+}
+
+static inline int set_memory_rw(unsigned long addr, int numpages)
+{
+	return __set_memory(addr, numpages, SET_MEMORY_RW);
+}
+
+static inline int set_memory_nx(unsigned long addr, int numpages)
+{
+	return __set_memory(addr, numpages, SET_MEMORY_NX);
+}
+
+static inline int set_memory_x(unsigned long addr, int numpages)
+{
+	return __set_memory(addr, numpages, SET_MEMORY_X);
+}
 
 #endif /* _S390_CACHEFLUSH_H */

arch/s390/include/asm/pgtable.h

@@ -200,6 +200,7 @@ static inline int is_module_addr(void *addr)
  */
 
 /* Hardware bits in the page table entry */
+#define _PAGE_NOEXEC	0x100		/* HW no-execute bit  */
 #define _PAGE_PROTECT	0x200		/* HW read-only bit  */
 #define _PAGE_INVALID	0x400		/* HW invalid bit    */
 #define _PAGE_LARGE	0x800		/* Bit to mark a large pte */
@@ -277,6 +278,7 @@ static inline int is_module_addr(void *addr)
 /* Bits in the region table entry */
 #define _REGION_ENTRY_ORIGIN	~0xfffUL/* region/segment table origin	    */
 #define _REGION_ENTRY_PROTECT	0x200	/* region protection bit	    */
+#define _REGION_ENTRY_NOEXEC	0x100	/* region no-execute bit	    */
 #define _REGION_ENTRY_OFFSET	0xc0	/* region table offset		    */
 #define _REGION_ENTRY_INVALID	0x20	/* invalid region table entry	    */
 #define _REGION_ENTRY_TYPE_MASK	0x0c	/* region/segment table type mask   */
@@ -316,6 +318,7 @@ static inline int is_module_addr(void *addr)
 #define _SEGMENT_ENTRY_ORIGIN_LARGE ~0xfffffUL /* large page address	    */
 #define _SEGMENT_ENTRY_ORIGIN	~0x7ffUL/* segment table origin		    */
 #define _SEGMENT_ENTRY_PROTECT	0x200	/* page protection bit		    */
+#define _SEGMENT_ENTRY_NOEXEC	0x100	/* region no-execute bit	    */
 #define _SEGMENT_ENTRY_INVALID	0x20	/* invalid segment table entry	    */
 
 #define _SEGMENT_ENTRY		(0)
@@ -385,17 +388,23 @@ static inline int is_module_addr(void *addr)
  * Page protection definitions.
  */
 #define PAGE_NONE	__pgprot(_PAGE_PRESENT | _PAGE_INVALID | _PAGE_PROTECT)
-#define PAGE_READ	__pgprot(_PAGE_PRESENT | _PAGE_READ | \
+#define PAGE_RO		__pgprot(_PAGE_PRESENT | _PAGE_READ | \
+				 _PAGE_NOEXEC  | _PAGE_INVALID | _PAGE_PROTECT)
+#define PAGE_RX		__pgprot(_PAGE_PRESENT | _PAGE_READ | \
 				 _PAGE_INVALID | _PAGE_PROTECT)
-#define PAGE_WRITE	__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
+#define PAGE_RW		__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
+				 _PAGE_NOEXEC  | _PAGE_INVALID | _PAGE_PROTECT)
+#define PAGE_RWX	__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
 				 _PAGE_INVALID | _PAGE_PROTECT)
 
 #define PAGE_SHARED	__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
-				 _PAGE_YOUNG | _PAGE_DIRTY)
+				 _PAGE_YOUNG | _PAGE_DIRTY | _PAGE_NOEXEC)
 #define PAGE_KERNEL	__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
-				 _PAGE_YOUNG | _PAGE_DIRTY)
+				 _PAGE_YOUNG | _PAGE_DIRTY | _PAGE_NOEXEC)
 #define PAGE_KERNEL_RO	__pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_YOUNG | \
-				 _PAGE_PROTECT)
+				 _PAGE_PROTECT | _PAGE_NOEXEC)
+#define PAGE_KERNEL_EXEC __pgprot(_PAGE_PRESENT | _PAGE_READ | _PAGE_WRITE | \
+				  _PAGE_YOUNG |	_PAGE_DIRTY)
 
 /*
  * On s390 the page table entry has an invalid bit and a read-only bit.
@@ -404,43 +413,51 @@ static inline int is_module_addr(void *addr)
  */
          /*xwr*/
 #define __P000	PAGE_NONE
-#define __P001	PAGE_READ
-#define __P010	PAGE_READ
-#define __P011	PAGE_READ
-#define __P100	PAGE_READ
-#define __P101	PAGE_READ
-#define __P110	PAGE_READ
-#define __P111	PAGE_READ
+#define __P001	PAGE_RO
+#define __P010	PAGE_RO
+#define __P011	PAGE_RO
+#define __P100	PAGE_RX
+#define __P101	PAGE_RX
+#define __P110	PAGE_RX
+#define __P111	PAGE_RX
 
 #define __S000	PAGE_NONE
-#define __S001	PAGE_READ
-#define __S010	PAGE_WRITE
-#define __S011	PAGE_WRITE
-#define __S100	PAGE_READ
-#define __S101	PAGE_READ
-#define __S110	PAGE_WRITE
-#define __S111	PAGE_WRITE
+#define __S001	PAGE_RO
+#define __S010	PAGE_RW
+#define __S011	PAGE_RW
+#define __S100	PAGE_RX
+#define __S101	PAGE_RX
+#define __S110	PAGE_RWX
+#define __S111	PAGE_RWX
 
 /*
  * Segment entry (large page) protection definitions.
  */
 #define SEGMENT_NONE	__pgprot(_SEGMENT_ENTRY_INVALID | \
 				 _SEGMENT_ENTRY_PROTECT)
-#define SEGMENT_READ	__pgprot(_SEGMENT_ENTRY_PROTECT | \
+#define SEGMENT_RO	__pgprot(_SEGMENT_ENTRY_PROTECT | \
+				 _SEGMENT_ENTRY_READ | \
+				 _SEGMENT_ENTRY_NOEXEC)
+#define SEGMENT_RX	__pgprot(_SEGMENT_ENTRY_PROTECT | \
 				 _SEGMENT_ENTRY_READ)
-#define SEGMENT_WRITE	__pgprot(_SEGMENT_ENTRY_READ | \
+#define SEGMENT_RW	__pgprot(_SEGMENT_ENTRY_READ | \
+				 _SEGMENT_ENTRY_WRITE | \
+				 _SEGMENT_ENTRY_NOEXEC)
+#define SEGMENT_RWX	__pgprot(_SEGMENT_ENTRY_READ | \
 				 _SEGMENT_ENTRY_WRITE)
 #define SEGMENT_KERNEL	__pgprot(_SEGMENT_ENTRY |	\
 				 _SEGMENT_ENTRY_LARGE |	\
 				 _SEGMENT_ENTRY_READ |	\
 				 _SEGMENT_ENTRY_WRITE | \
 				 _SEGMENT_ENTRY_YOUNG | \
-				 _SEGMENT_ENTRY_DIRTY)
+				 _SEGMENT_ENTRY_DIRTY | \
+				 _SEGMENT_ENTRY_NOEXEC)
 #define SEGMENT_KERNEL_RO __pgprot(_SEGMENT_ENTRY |	\
 				 _SEGMENT_ENTRY_LARGE |	\
 				 _SEGMENT_ENTRY_READ |	\
 				 _SEGMENT_ENTRY_YOUNG |	\
-				 _SEGMENT_ENTRY_PROTECT)
+				 _SEGMENT_ENTRY_PROTECT | \
+				 _SEGMENT_ENTRY_NOEXEC)
 
 /*
  * Region3 entry (large page) protection definitions.
@@ -451,12 +468,14 @@ static inline int is_module_addr(void *addr)
 				 _REGION3_ENTRY_READ |	 \
 				 _REGION3_ENTRY_WRITE |	 \
 				 _REGION3_ENTRY_YOUNG |	 \
-				 _REGION3_ENTRY_DIRTY)
+				 _REGION3_ENTRY_DIRTY | \
+				 _REGION_ENTRY_NOEXEC)
 #define REGION3_KERNEL_RO __pgprot(_REGION_ENTRY_TYPE_R3 | \
 				   _REGION3_ENTRY_LARGE |  \
 				   _REGION3_ENTRY_READ |   \
 				   _REGION3_ENTRY_YOUNG |  \
-				   _REGION_ENTRY_PROTECT)
+				   _REGION_ENTRY_PROTECT | \
+				   _REGION_ENTRY_NOEXEC)
 
 static inline int mm_has_pgste(struct mm_struct *mm)
 {
@@ -801,14 +820,14 @@ static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
 	pte_val(pte) &= _PAGE_CHG_MASK;
 	pte_val(pte) |= pgprot_val(newprot);
 	/*
-	 * newprot for PAGE_NONE, PAGE_READ and PAGE_WRITE has the
-	 * invalid bit set, clear it again for readable, young pages
+	 * newprot for PAGE_NONE, PAGE_RO, PAGE_RX, PAGE_RW and PAGE_RWX
+	 * has the invalid bit set, clear it again for readable, young pages
 	 */
 	if ((pte_val(pte) & _PAGE_YOUNG) && (pte_val(pte) & _PAGE_READ))
 		pte_val(pte) &= ~_PAGE_INVALID;
 	/*
-	 * newprot for PAGE_READ and PAGE_WRITE has the page protection
-	 * bit set, clear it again for writable, dirty pages
+	 * newprot for PAGE_RO, PAGE_RX, PAGE_RW and PAGE_RWX has the page
+	 * protection bit set, clear it again for writable, dirty pages
 	 */
 	if ((pte_val(pte) & _PAGE_DIRTY) && (pte_val(pte) & _PAGE_WRITE))
 		pte_val(pte) &= ~_PAGE_PROTECT;
@@ -1029,6 +1048,8 @@ int get_guest_storage_key(struct mm_struct *mm, unsigned long addr,
 static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
 			      pte_t *ptep, pte_t entry)
 {
+	if (!MACHINE_HAS_NX)
+		pte_val(entry) &= ~_PAGE_NOEXEC;
 	if (mm_has_pgste(mm))
 		ptep_set_pte_at(mm, addr, ptep, entry);
 	else
@@ -1173,14 +1194,18 @@ static inline pud_t pud_mkdirty(pud_t pud)
 static inline unsigned long massage_pgprot_pmd(pgprot_t pgprot)
 {
 	/*
-	 * pgprot is PAGE_NONE, PAGE_READ, or PAGE_WRITE (see __Pxxx / __Sxxx)
-	 * Convert to segment table entry format.
+	 * pgprot is PAGE_NONE, PAGE_RO, PAGE_RX, PAGE_RW or PAGE_RWX
+	 * (see __Pxxx / __Sxxx). Convert to segment table entry format.
 	 */
 	if (pgprot_val(pgprot) == pgprot_val(PAGE_NONE))
 		return pgprot_val(SEGMENT_NONE);
-	if (pgprot_val(pgprot) == pgprot_val(PAGE_READ))
-		return pgprot_val(SEGMENT_READ);
-	return pgprot_val(SEGMENT_WRITE);
+	if (pgprot_val(pgprot) == pgprot_val(PAGE_RO))
+		return pgprot_val(SEGMENT_RO);
+	if (pgprot_val(pgprot) == pgprot_val(PAGE_RX))
+		return pgprot_val(SEGMENT_RX);
+	if (pgprot_val(pgprot) == pgprot_val(PAGE_RW))
+		return pgprot_val(SEGMENT_RW);
+	return pgprot_val(SEGMENT_RWX);
 }
 
 static inline pmd_t pmd_mkyoung(pmd_t pmd)
@@ -1315,6 +1340,8 @@ static inline int pmdp_clear_flush_young(struct vm_area_struct *vma,
 static inline void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 			      pmd_t *pmdp, pmd_t entry)
 {
+	if (!MACHINE_HAS_NX)
+		pmd_val(entry) &= ~_SEGMENT_ENTRY_NOEXEC;
 	*pmdp = entry;
 }
 

arch/s390/include/asm/setup.h

@@ -30,6 +30,7 @@
 #define MACHINE_FLAG_TLB_LC	_BITUL(12)
 #define MACHINE_FLAG_VX		_BITUL(13)
 #define MACHINE_FLAG_CAD	_BITUL(14)
+#define MACHINE_FLAG_NX		_BITUL(15)
 
 #define LPP_MAGIC		_BITUL(31)
 #define LPP_PFAULT_PID_MASK	_AC(0xffffffff, UL)
@@ -71,6 +72,7 @@ extern void detect_memory_memblock(void);
 #define MACHINE_HAS_TLB_LC	(S390_lowcore.machine_flags & MACHINE_FLAG_TLB_LC)
 #define MACHINE_HAS_VX		(S390_lowcore.machine_flags & MACHINE_FLAG_VX)
 #define MACHINE_HAS_CAD		(S390_lowcore.machine_flags & MACHINE_FLAG_CAD)
+#define MACHINE_HAS_NX		(S390_lowcore.machine_flags & MACHINE_FLAG_NX)
 
 /*
  * Console mode. Override with conmode=

arch/s390/kernel/early.c

@@ -354,6 +354,10 @@ static __init void detect_machine_facilities(void)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_VX;
 		__ctl_set_bit(0, 17);
 	}
+	if (test_facility(130)) {
+		S390_lowcore.machine_flags |= MACHINE_FLAG_NX;
+		__ctl_set_bit(0, 20);
+	}
 }
 
 static inline void save_vector_registers(void)
@@ -384,6 +388,21 @@ static int __init disable_vector_extension(char *str)
 }
 early_param("novx", disable_vector_extension);
 
+static int __init noexec_setup(char *str)
+{
+	bool enabled;
+	int rc;
+
+	rc = kstrtobool(str, &enabled);
+	if (!rc && !enabled) {
+		/* Disable no-execute support */
+		S390_lowcore.machine_flags &= ~MACHINE_FLAG_NX;
+		__ctl_clear_bit(0, 20);
+	}
+	return rc;
+}
+early_param("noexec", noexec_setup);
+
 static int __init cad_setup(char *str)
 {
 	int val;

arch/s390/kernel/entry.S

@@ -380,13 +380,11 @@ ENTRY(system_call)
 	brasl	%r14,do_signal
 	TSTMSK	__PT_FLAGS(%r11),_PIF_SYSCALL
 	jno	.Lsysc_return
+.Lsysc_do_syscall:
+	lghi	%r13,__TASK_thread
 	lmg	%r2,%r7,__PT_R2(%r11)	# load svc arguments
-	lghi	%r8,0			# svc 0 returns -ENOSYS
-	llgh	%r1,__PT_INT_CODE+2(%r11)	# load new svc number
-	cghi	%r1,NR_syscalls
-	jnl	.Lsysc_nr_ok		# invalid svc number -> do svc 0
-	slag	%r8,%r1,2
-	j	.Lsysc_nr_ok		# restart svc
+	lghi	%r1,0			# svc 0 returns -ENOSYS
+	j	.Lsysc_do_svc
 
 #
 # _TIF_NOTIFY_RESUME is set, call do_notify_resume
@@ -528,6 +526,8 @@ ENTRY(pgm_check_handler)
 	LOCKDEP_SYS_EXIT
 	tm	__PT_PSW+1(%r11),0x01	# returning to user ?
 	jno	.Lsysc_restore
+	TSTMSK	__PT_FLAGS(%r11),_PIF_SYSCALL
+	jo	.Lsysc_do_syscall
 	j	.Lsysc_tif
 
 #

arch/s390/kernel/kprobes.c

@@ -45,11 +45,17 @@ DEFINE_INSN_CACHE_OPS(dmainsn);
 
 static void *alloc_dmainsn_page(void)
 {
-	return (void *)__get_free_page(GFP_KERNEL | GFP_DMA);
+	void *page;
+
+	page = (void *) __get_free_page(GFP_KERNEL | GFP_DMA);
+	if (page)
+		set_memory_x((unsigned long) page, 1);
+	return page;
 }
 
 static void free_dmainsn_page(void *page)
 {
+	set_memory_nx((unsigned long) page, 1);
 	free_page((unsigned long)page);
 }
 

arch/s390/kernel/module.c

@@ -45,7 +45,8 @@ void *module_alloc(unsigned long size)
 	if (PAGE_ALIGN(size) > MODULES_LEN)
 		return NULL;
 	return __vmalloc_node_range(size, 1, MODULES_VADDR, MODULES_END,
-				    GFP_KERNEL, PAGE_KERNEL, 0, NUMA_NO_NODE,
+				    GFP_KERNEL, PAGE_KERNEL_EXEC,
+				    0, NUMA_NO_NODE,
 				    __builtin_return_address(0));
 }
 

arch/s390/kernel/vmlinux.lds.S

@@ -44,6 +44,7 @@ SECTIONS
 		*(.gnu.warning)
 	} :text = 0x0700
 
+	. = ALIGN(PAGE_SIZE);
 	_etext = .;		/* End of text section */
 
 	NOTES :text :note
@@ -79,7 +80,13 @@ SECTIONS
 	. = ALIGN(PAGE_SIZE);	/* Init code and data */
 	__init_begin = .;
 
-	INIT_TEXT_SECTION(PAGE_SIZE)
+	. = ALIGN(PAGE_SIZE);
+	.init.text : AT(ADDR(.init.text) - LOAD_OFFSET) {
+		VMLINUX_SYMBOL(_sinittext) = . ;
+		INIT_TEXT
+		. = ALIGN(PAGE_SIZE);
+		VMLINUX_SYMBOL(_einittext) = . ;
+	}
 
 	/*
 	 * .exit.text is discarded at runtime, not link time,

arch/s390/mm/dump_pagetables.c

@@ -49,8 +49,8 @@ static void print_prot(struct seq_file *m, unsigned int pr, int level)
 		seq_printf(m, "I\n");
 		return;
 	}
-	seq_printf(m, "%s", pr & _PAGE_PROTECT ? "RO " : "RW ");
-	seq_putc(m, '\n');
+	seq_puts(m, (pr & _PAGE_PROTECT) ? "RO " : "RW ");
+	seq_puts(m, (pr & _PAGE_NOEXEC) ? "NX\n" : "X\n");
 }
 
 static void note_page(struct seq_file *m, struct pg_state *st,
@@ -117,7 +117,8 @@ static void walk_pte_level(struct seq_file *m, struct pg_state *st,
 	for (i = 0; i < PTRS_PER_PTE && addr < max_addr; i++) {
 		st->current_address = addr;
 		pte = pte_offset_kernel(pmd, addr);
-		prot = pte_val(*pte) & (_PAGE_PROTECT | _PAGE_INVALID);
+		prot = pte_val(*pte) &
+			(_PAGE_PROTECT | _PAGE_INVALID | _PAGE_NOEXEC);
 		note_page(m, st, prot, 4);
 		addr += PAGE_SIZE;
 	}
@@ -135,7 +136,9 @@ static void walk_pmd_level(struct seq_file *m, struct pg_state *st,
 		pmd = pmd_offset(pud, addr);
 		if (!pmd_none(*pmd)) {
 			if (pmd_large(*pmd)) {
-				prot = pmd_val(*pmd) & _SEGMENT_ENTRY_PROTECT;
+				prot = pmd_val(*pmd) &
+					(_SEGMENT_ENTRY_PROTECT |
+					 _SEGMENT_ENTRY_NOEXEC);
 				note_page(m, st, prot, 3);
 			} else
 				walk_pte_level(m, st, pmd, addr);
@@ -157,7 +160,9 @@ static void walk_pud_level(struct seq_file *m, struct pg_state *st,
 		pud = pud_offset(pgd, addr);
 		if (!pud_none(*pud))
 			if (pud_large(*pud)) {
-				prot = pud_val(*pud) & _REGION_ENTRY_PROTECT;
+				prot = pud_val(*pud) &
+					(_REGION_ENTRY_PROTECT |
+					 _REGION_ENTRY_NOEXEC);
 				note_page(m, st, prot, 2);
 			} else
 				walk_pmd_level(m, st, pud, addr);

arch/s390/mm/fault.c

@@ -311,12 +311,34 @@ static noinline void do_sigbus(struct pt_regs *regs)
 	force_sig_info(SIGBUS, &si, tsk);
 }
 
-static noinline void do_fault_error(struct pt_regs *regs, int fault)
+static noinline int signal_return(struct pt_regs *regs)
+{
+	u16 instruction;
+	int rc;
+
+	rc = __get_user(instruction, (u16 __user *) regs->psw.addr);
+	if (rc)
+		return rc;
+	if (instruction == 0x0a77) {
+		set_pt_regs_flag(regs, PIF_SYSCALL);
+		regs->int_code = 0x00040077;
+		return 0;
+	} else if (instruction == 0x0aad) {
+		set_pt_regs_flag(regs, PIF_SYSCALL);
+		regs->int_code = 0x000400ad;
+		return 0;
+	}
+	return -EACCES;
+}
+
+static noinline void do_fault_error(struct pt_regs *regs, int access, int fault)
 {
 	int si_code;
 
 	switch (fault) {
 	case VM_FAULT_BADACCESS:
+		if (access == VM_EXEC && signal_return(regs) == 0)
+			break;
 	case VM_FAULT_BADMAP:
 		/* Bad memory access. Check if it is kernel or user space. */
 		if (user_mode(regs)) {
@@ -324,7 +346,7 @@ static noinline void do_fault_error(struct pt_regs *regs, int fault)
 			si_code = (fault == VM_FAULT_BADMAP) ?
 				SEGV_MAPERR : SEGV_ACCERR;
 			do_sigsegv(regs, si_code);
-			return;
+			break;
 		}
 	case VM_FAULT_BADCONTEXT:
 	case VM_FAULT_PFAULT:
@@ -525,7 +547,7 @@ out:
 void do_protection_exception(struct pt_regs *regs)
 {
 	unsigned long trans_exc_code;
-	int fault;
+	int access, fault;
 
 	trans_exc_code = regs->int_parm_long;
 	/*
@@ -544,9 +566,17 @@ void do_protection_exception(struct pt_regs *regs)
 		do_low_address(regs);
 		return;
 	}
-	fault = do_exception(regs, VM_WRITE);
+	if (unlikely(MACHINE_HAS_NX && (trans_exc_code & 0x80))) {
+		regs->int_parm_long = (trans_exc_code & ~PAGE_MASK) |
+					(regs->psw.addr & PAGE_MASK);
+		access = VM_EXEC;
+		fault = VM_FAULT_BADACCESS;
+	} else {
+		access = VM_WRITE;
+		fault = do_exception(regs, access);
+	}
 	if (unlikely(fault))
-		do_fault_error(regs, fault);
+		do_fault_error(regs, access, fault);
 }
 NOKPROBE_SYMBOL(do_protection_exception);
 
@@ -557,7 +587,7 @@ void do_dat_exception(struct pt_regs *regs)
 	access = VM_READ | VM_EXEC | VM_WRITE;
 	fault = do_exception(regs, access);
 	if (unlikely(fault))
-		do_fault_error(regs, fault);
+		do_fault_error(regs, access, fault);
 }
 NOKPROBE_SYMBOL(do_dat_exception);
 

arch/s390/mm/hugetlbpage.c

@@ -59,6 +59,8 @@ static inline unsigned long __pte_to_rste(pte_t pte)
 		rste |= move_set_bit(pte_val(pte), _PAGE_SOFT_DIRTY,
 				     _SEGMENT_ENTRY_SOFT_DIRTY);
 #endif
+		rste |= move_set_bit(pte_val(pte), _PAGE_NOEXEC,
+				     _SEGMENT_ENTRY_NOEXEC);
 	} else
 		rste = _SEGMENT_ENTRY_INVALID;
 	return rste;
@@ -113,6 +115,8 @@ static inline pte_t __rste_to_pte(unsigned long rste)
 		pte_val(pte) |= move_set_bit(rste, _SEGMENT_ENTRY_SOFT_DIRTY,
 					     _PAGE_DIRTY);
 #endif
+		pte_val(pte) |= move_set_bit(rste, _SEGMENT_ENTRY_NOEXEC,
+					     _PAGE_NOEXEC);
 	} else
 		pte_val(pte) = _PAGE_INVALID;
 	return pte;
@@ -121,7 +125,11 @@ static inline pte_t __rste_to_pte(unsigned long rste)
 void set_huge_pte_at(struct mm_struct *mm, unsigned long addr,
 		     pte_t *ptep, pte_t pte)
 {
-	unsigned long rste = __pte_to_rste(pte);
+	unsigned long rste;
+
+	rste = __pte_to_rste(pte);
+	if (!MACHINE_HAS_NX)
+		rste &= ~_SEGMENT_ENTRY_NOEXEC;
 
 	/* Set correct table type for 2G hugepages */
 	if ((pte_val(*ptep) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3)

arch/s390/mm/init.c

@@ -137,6 +137,9 @@ void __init mem_init(void)
 
 void free_initmem(void)
 {
+	__set_memory((unsigned long) _sinittext,
+		     (_einittext - _sinittext) >> PAGE_SHIFT,
+		     SET_MEMORY_RW | SET_MEMORY_NX);
 	free_initmem_default(POISON_FREE_INITMEM);
 }
 

arch/s390/mm/pageattr.c

@@ -81,24 +81,24 @@ static void pgt_set(unsigned long *old, unsigned long new, unsigned long addr,
 	}
 }
 
-struct cpa {
-	unsigned int set_ro	: 1;
-	unsigned int clear_ro	: 1;
-};
-
 static int walk_pte_level(pmd_t *pmdp, unsigned long addr, unsigned long end,
-			  struct cpa cpa)
+			  unsigned long flags)
 {
 	pte_t *ptep, new;
 
 	ptep = pte_offset(pmdp, addr);
 	do {
-		if (pte_none(*ptep))
+		new = *ptep;
+		if (pte_none(new))
 			return -EINVAL;
-		if (cpa.set_ro)
-			new = pte_wrprotect(*ptep);
-		else if (cpa.clear_ro)
-			new = pte_mkwrite(pte_mkdirty(*ptep));
+		if (flags & SET_MEMORY_RO)
+			new = pte_wrprotect(new);
+		else if (flags & SET_MEMORY_RW)
+			new = pte_mkwrite(pte_mkdirty(new));
+		if ((flags & SET_MEMORY_NX) && MACHINE_HAS_NX)
+			pte_val(new) |= _PAGE_NOEXEC;
+		else if (flags & SET_MEMORY_X)
+			pte_val(new) &= ~_PAGE_NOEXEC;
 		pgt_set((unsigned long *)ptep, pte_val(new), addr, CRDTE_DTT_PAGE);
 		ptep++;
 		addr += PAGE_SIZE;
@@ -112,14 +112,17 @@ static int split_pmd_page(pmd_t *pmdp, unsigned long addr)
 	unsigned long pte_addr, prot;
 	pte_t *pt_dir, *ptep;
 	pmd_t new;
-	int i, ro;
+	int i, ro, nx;
 
 	pt_dir = vmem_pte_alloc();
 	if (!pt_dir)
 		return -ENOMEM;
 	pte_addr = pmd_pfn(*pmdp) << PAGE_SHIFT;
 	ro = !!(pmd_val(*pmdp) & _SEGMENT_ENTRY_PROTECT);
+	nx = !!(pmd_val(*pmdp) & _SEGMENT_ENTRY_NOEXEC);
 	prot = pgprot_val(ro ? PAGE_KERNEL_RO : PAGE_KERNEL);
+	if (!nx)
+		prot &= ~_PAGE_NOEXEC;
 	ptep = pt_dir;
 	for (i = 0; i < PTRS_PER_PTE; i++) {
 		pte_val(*ptep) = pte_addr | prot;
@@ -133,19 +136,24 @@ static int split_pmd_page(pmd_t *pmdp, unsigned long addr)
 	return 0;
 }
 
-static void modify_pmd_page(pmd_t *pmdp, unsigned long addr, struct cpa cpa)
+static void modify_pmd_page(pmd_t *pmdp, unsigned long addr,
+			    unsigned long flags)
 {
-	pmd_t new;
-
-	if (cpa.set_ro)
-		new = pmd_wrprotect(*pmdp);
-	else if (cpa.clear_ro)
-		new = pmd_mkwrite(pmd_mkdirty(*pmdp));
+	pmd_t new = *pmdp;
+
+	if (flags & SET_MEMORY_RO)
+		new = pmd_wrprotect(new);
+	else if (flags & SET_MEMORY_RW)
+		new = pmd_mkwrite(pmd_mkdirty(new));
+	if ((flags & SET_MEMORY_NX) && MACHINE_HAS_NX)
+		pmd_val(new) |= _SEGMENT_ENTRY_NOEXEC;
+	else if (flags & SET_MEMORY_X)
+		pmd_val(new) &= ~_SEGMENT_ENTRY_NOEXEC;
 	pgt_set((unsigned long *)pmdp, pmd_val(new), addr, CRDTE_DTT_SEGMENT);
 }
 
 static int walk_pmd_level(pud_t *pudp, unsigned long addr, unsigned long end,
-			  struct cpa cpa)
+			  unsigned long flags)
 {
 	unsigned long next;
 	pmd_t *pmdp;
@@ -163,9 +171,9 @@ static int walk_pmd_level(pud_t *pudp, unsigned long addr, unsigned long end,
 					return rc;
 				continue;
 			}
-			modify_pmd_page(pmdp, addr, cpa);
+			modify_pmd_page(pmdp, addr, flags);
 		} else {
-			rc = walk_pte_level(pmdp, addr, next, cpa);
+			rc = walk_pte_level(pmdp, addr, next, flags);
 			if (rc)
 				return rc;
 		}
@@ -181,14 +189,17 @@ static int split_pud_page(pud_t *pudp, unsigned long addr)
 	unsigned long pmd_addr, prot;
 	pmd_t *pm_dir, *pmdp;
 	pud_t new;
-	int i, ro;
+	int i, ro, nx;
 
 	pm_dir = vmem_pmd_alloc();
 	if (!pm_dir)
 		return -ENOMEM;
 	pmd_addr = pud_pfn(*pudp) << PAGE_SHIFT;
 	ro = !!(pud_val(*pudp) & _REGION_ENTRY_PROTECT);
+	nx = !!(pud_val(*pudp) & _REGION_ENTRY_NOEXEC);
 	prot = pgprot_val(ro ? SEGMENT_KERNEL_RO : SEGMENT_KERNEL);
+	if (!nx)
+		prot &= ~_SEGMENT_ENTRY_NOEXEC;
 	pmdp = pm_dir;
 	for (i = 0; i < PTRS_PER_PMD; i++) {
 		pmd_val(*pmdp) = pmd_addr | prot;
@@ -202,19 +213,24 @@ static int split_pud_page(pud_t *pudp, unsigned long addr)
 	return 0;
 }
 
-static void modify_pud_page(pud_t *pudp, unsigned long addr, struct cpa cpa)
+static void modify_pud_page(pud_t *pudp, unsigned long addr,
+			    unsigned long flags)
 {
-	pud_t new;
-
-	if (cpa.set_ro)
-		new = pud_wrprotect(*pudp);
-	else if (cpa.clear_ro)
-		new = pud_mkwrite(pud_mkdirty(*pudp));
+	pud_t new = *pudp;
+
+	if (flags & SET_MEMORY_RO)
+		new = pud_wrprotect(new);
+	else if (flags & SET_MEMORY_RW)
+		new = pud_mkwrite(pud_mkdirty(new));
+	if ((flags & SET_MEMORY_NX) && MACHINE_HAS_NX)
+		pud_val(new) |= _REGION_ENTRY_NOEXEC;
+	else if (flags & SET_MEMORY_X)
+		pud_val(new) &= ~_REGION_ENTRY_NOEXEC;
 	pgt_set((unsigned long *)pudp, pud_val(new), addr, CRDTE_DTT_REGION3);
 }
 
 static int walk_pud_level(pgd_t *pgd, unsigned long addr, unsigned long end,
-			  struct cpa cpa)
+			  unsigned long flags)
 {
 	unsigned long next;
 	pud_t *pudp;
@@ -232,9 +248,9 @@ static int walk_pud_level(pgd_t *pgd, unsigned long addr, unsigned long end,
 					break;
 				continue;
 			}
-			modify_pud_page(pudp, addr, cpa);
+			modify_pud_page(pudp, addr, flags);
 		} else {
-			rc = walk_pmd_level(pudp, addr, next, cpa);
+			rc = walk_pmd_level(pudp, addr, next, flags);
 		}
 		pudp++;
 		addr = next;
@@ -246,7 +262,7 @@ static int walk_pud_level(pgd_t *pgd, unsigned long addr, unsigned long end,
 static DEFINE_MUTEX(cpa_mutex);
 
 static int change_page_attr(unsigned long addr, unsigned long end,
-			    struct cpa cpa)
+			    unsigned long flags)
 {
 	unsigned long next;
 	int rc = -EINVAL;
@@ -262,7 +278,7 @@ static int change_page_attr(unsigned long addr, unsigned long end,
 		if (pgd_none(*pgdp))
 			break;
 		next = pgd_addr_end(addr, end);
-		rc = walk_pud_level(pgdp, addr, next, cpa);
+		rc = walk_pud_level(pgdp, addr, next, flags);
 		if (rc)
 			break;
 		cond_resched();
@@ -271,35 +287,10 @@ static int change_page_attr(unsigned long addr, unsigned long end,
 	return rc;
 }
 
-int set_memory_ro(unsigned long addr, int numpages)
+int __set_memory(unsigned long addr, int numpages, unsigned long flags)
 {
-	struct cpa cpa = {
-		.set_ro = 1,
-	};
-
 	addr &= PAGE_MASK;
-	return change_page_attr(addr, addr + numpages * PAGE_SIZE, cpa);
-}
-
-int set_memory_rw(unsigned long addr, int numpages)
-{
-	struct cpa cpa = {
-		.clear_ro = 1,
-	};
-
-	addr &= PAGE_MASK;
-	return change_page_attr(addr, addr + numpages * PAGE_SIZE, cpa);
-}
-
-/* not possible */
-int set_memory_nx(unsigned long addr, int numpages)
-{
-	return 0;
-}
-
-int set_memory_x(unsigned long addr, int numpages)
-{
-	return 0;
+	return change_page_attr(addr, addr + numpages * PAGE_SIZE, flags);
 }
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
@@ -339,7 +330,7 @@ void __kernel_map_pages(struct page *page, int numpages, int enable)
 		nr = min(numpages - i, nr);
 		if (enable) {
 			for (j = 0; j < nr; j++) {
-				pte_val(*pte) = address | pgprot_val(PAGE_KERNEL);
+				pte_val(*pte) &= ~_PAGE_INVALID;
 				address += PAGE_SIZE;
 				pte++;
 			}

arch/s390/mm/pgtable.c

@@ -274,6 +274,8 @@ void ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
 {
 	pgste_t pgste;
 
+	if (!MACHINE_HAS_NX)
+		pte_val(pte) &= ~_PAGE_NOEXEC;
 	if (mm_has_pgste(mm)) {
 		pgste = pgste_get(ptep);
 		pgste_set_key(ptep, pgste, pte, mm);

arch/s390/mm/vmem.c

@@ -79,6 +79,7 @@ pte_t __ref *vmem_pte_alloc(void)
  */
 static int vmem_add_mem(unsigned long start, unsigned long size)
 {
+	unsigned long pgt_prot, sgt_prot, r3_prot;
 	unsigned long pages4k, pages1m, pages2g;
 	unsigned long end = start + size;
 	unsigned long address = start;
@@ -88,6 +89,14 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 	pte_t *pt_dir;
 	int ret = -ENOMEM;
 
+	pgt_prot = pgprot_val(PAGE_KERNEL);
+	sgt_prot = pgprot_val(SEGMENT_KERNEL);
+	r3_prot = pgprot_val(REGION3_KERNEL);
+	if (!MACHINE_HAS_NX) {
+		pgt_prot &= ~_PAGE_NOEXEC;
+		sgt_prot &= ~_SEGMENT_ENTRY_NOEXEC;
+		r3_prot &= ~_REGION_ENTRY_NOEXEC;
+	}
 	pages4k = pages1m = pages2g = 0;
 	while (address < end) {
 		pg_dir = pgd_offset_k(address);
@@ -101,7 +110,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 		if (MACHINE_HAS_EDAT2 && pud_none(*pu_dir) && address &&
 		    !(address & ~PUD_MASK) && (address + PUD_SIZE <= end) &&
 		     !debug_pagealloc_enabled()) {
-			pud_val(*pu_dir) = address | pgprot_val(REGION3_KERNEL);
+			pud_val(*pu_dir) = address | r3_prot;
 			address += PUD_SIZE;
 			pages2g++;
 			continue;
@@ -116,7 +125,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 		if (MACHINE_HAS_EDAT1 && pmd_none(*pm_dir) && address &&
 		    !(address & ~PMD_MASK) && (address + PMD_SIZE <= end) &&
 		    !debug_pagealloc_enabled()) {
-			pmd_val(*pm_dir) = address | pgprot_val(SEGMENT_KERNEL);
+			pmd_val(*pm_dir) = address | sgt_prot;
 			address += PMD_SIZE;
 			pages1m++;
 			continue;
@@ -129,7 +138,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 		}
 
 		pt_dir = pte_offset_kernel(pm_dir, address);
-		pte_val(*pt_dir) = address |  pgprot_val(PAGE_KERNEL);
+		pte_val(*pt_dir) = address | pgt_prot;
 		address += PAGE_SIZE;
 		pages4k++;
 	}
@@ -200,6 +209,7 @@ static void vmem_remove_range(unsigned long start, unsigned long size)
  */
 int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 {
+	unsigned long pgt_prot, sgt_prot;
 	unsigned long address = start;
 	pgd_t *pg_dir;
 	pud_t *pu_dir;
@@ -207,6 +217,12 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 	pte_t *pt_dir;
 	int ret = -ENOMEM;
 
+	pgt_prot = pgprot_val(PAGE_KERNEL);
+	sgt_prot = pgprot_val(SEGMENT_KERNEL);
+	if (!MACHINE_HAS_NX) {
+		pgt_prot &= ~_PAGE_NOEXEC;
+		sgt_prot &= ~_SEGMENT_ENTRY_NOEXEC;
+	}
 	for (address = start; address < end;) {
 		pg_dir = pgd_offset_k(address);
 		if (pgd_none(*pg_dir)) {
@@ -238,8 +254,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 				new_page = vmemmap_alloc_block(PMD_SIZE, node);
 				if (!new_page)
 					goto out;
-				pmd_val(*pm_dir) = __pa(new_page) |
-					_SEGMENT_ENTRY | _SEGMENT_ENTRY_LARGE;
+				pmd_val(*pm_dir) = __pa(new_page) | sgt_prot;
 				address = (address + PMD_SIZE) & PMD_MASK;
 				continue;
 			}
@@ -259,8 +274,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 			new_page = vmemmap_alloc_block(PAGE_SIZE, node);
 			if (!new_page)
 				goto out;
-			pte_val(*pt_dir) =
-				__pa(new_page) | pgprot_val(PAGE_KERNEL);
+			pte_val(*pt_dir) = __pa(new_page) | pgt_prot;
 		}
 		address += PAGE_SIZE;
 	}
@@ -372,13 +386,21 @@ out:
  */
 void __init vmem_map_init(void)
 {
-	unsigned long size = _eshared - _stext;
 	struct memblock_region *reg;
 
 	for_each_memblock(memory, reg)
 		vmem_add_mem(reg->base, reg->size);
-	set_memory_ro((unsigned long)_stext, size >> PAGE_SHIFT);
-	pr_info("Write protected kernel read-only data: %luk\n", size >> 10);
+	__set_memory((unsigned long) _stext,
+		     (_etext - _stext) >> PAGE_SHIFT,
+		     SET_MEMORY_RO | SET_MEMORY_X);
+	__set_memory((unsigned long) _etext,
+		     (_eshared - _etext) >> PAGE_SHIFT,
+		     SET_MEMORY_RO);
+	__set_memory((unsigned long) _sinittext,
+		     (_einittext - _sinittext) >> PAGE_SHIFT,
+		     SET_MEMORY_RO | SET_MEMORY_X);
+	pr_info("Write protected kernel read-only data: %luk\n",
+		(_eshared - _stext) >> 10);
 }
 
 /*