Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux

Pull s390 updates from Martin Schwidefsky:
 "The first part of the s390 updates for 4.14:

   - Add machine type 0x3906 for IBM z14

   - Add IBM z14 TLB flushing improvements for KVM guests

   - Exploit the TOD clock epoch extension to provide a continuous TOD
     clock afer 2042/09/17

   - Add NIAI spinlock hints for IBM z14

   - Rework the vmcp driver and use CMA for the respone buffer of z/VM
     CP commands

   - Drop some s390 specific asm headers and use the generic version

   - Add block discard for DASD-FBA devices under z/VM

   - Add average request times to DASD statistics

   - A few of those constify patches which seem to be in vogue right now

   - Cleanup and bug fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: (50 commits)
  s390/mm: avoid empty zero pages for KVM guests to avoid postcopy hangs
  s390/dasd: Add discard support for FBA devices
  s390/zcrypt: make CPRBX const
  s390/uaccess: avoid mvcos jump label
  s390/mm: use generic mm_hooks
  s390/facilities: fix typo
  s390/vmcp: simplify vmcp_response_free()
  s390/topology: Remove the unused parent_node() macro
  s390/dasd: Change unsigned long long to unsigned long
  s390/smp: convert cpuhp_setup_state() return code to zero on success
  s390: fix 'novx' early parameter handling
  s390/dasd: add average request times to dasd statistics
  s390/scm: use common completion path
  s390/pci: log changes to uid checking
  s390/vmcp: simplify vmcp_ioctl()
  s390/vmcp: return -ENOTTY for unknown ioctl commands
  s390/vmcp: split vmcp header file and move to uapi
  s390/vmcp: make use of contiguous memory allocator
  s390/cpcmd,vmcp: avoid GFP_DMA allocations
  s390/vmcp: fix uaccess check and avoid undefined behavior
  ...

Documentation/admin-guide/kernel-parameters.txt

@@ -4395,6 +4395,10 @@
 			decrease the size and leave more room for directly
 			mapped kernel RAM.
 
+	vmcp_cma=nn[MG]	[KNL,S390]
+			Sets the memory size reserved for contiguous memory
+			allocations for the vmcp device driver.
+
 	vmhalt=		[KNL,S390] Perform z/VM CP command after system halt.
 			Format: <command>
 

arch/s390/Kconfig

@@ -222,6 +222,10 @@ config HAVE_MARCH_Z13_FEATURES
 	def_bool n
 	select HAVE_MARCH_ZEC12_FEATURES
 
+config HAVE_MARCH_Z14_FEATURES
+	def_bool n
+	select HAVE_MARCH_Z13_FEATURES
+
 choice
 	prompt "Processor type"
 	default MARCH_Z196
@@ -282,6 +286,14 @@ config MARCH_Z13
 	  2964 series). The kernel will be slightly faster but will not work on
 	  older machines.
 
+config MARCH_Z14
+	bool "IBM z14"
+	select HAVE_MARCH_Z14_FEATURES
+	help
+	  Select this to enable optimizations for IBM z14 (3906 series).
+	  The kernel will be slightly faster but will not work on older
+	  machines.
+
 endchoice
 
 config MARCH_Z900_TUNE
@@ -305,6 +317,9 @@ config MARCH_ZEC12_TUNE
 config MARCH_Z13_TUNE
 	def_bool TUNE_Z13 || MARCH_Z13 && TUNE_DEFAULT
 
+config MARCH_Z14_TUNE
+	def_bool TUNE_Z14 || MARCH_Z14 && TUNE_DEFAULT
+
 choice
 	prompt "Tune code generation"
 	default TUNE_DEFAULT
@@ -343,6 +358,9 @@ config TUNE_ZEC12
 config TUNE_Z13
 	bool "IBM z13"
 
+config TUNE_Z14
+	bool "IBM z14"
+
 endchoice
 
 config 64BIT

arch/s390/Makefile

@@ -31,7 +31,8 @@ mflags-$(CONFIG_MARCH_Z9_109) := -march=z9-109
 mflags-$(CONFIG_MARCH_Z10)    := -march=z10
 mflags-$(CONFIG_MARCH_Z196)   := -march=z196
 mflags-$(CONFIG_MARCH_ZEC12)  := -march=zEC12
-mflags-$(CONFIG_MARCH_Z13)   := -march=z13
+mflags-$(CONFIG_MARCH_Z13)    := -march=z13
+mflags-$(CONFIG_MARCH_Z14)    := -march=z14
 
 export CC_FLAGS_MARCH := $(mflags-y)
 
@@ -44,7 +45,8 @@ cflags-$(CONFIG_MARCH_Z9_109_TUNE)	+= -mtune=z9-109
 cflags-$(CONFIG_MARCH_Z10_TUNE)		+= -mtune=z10
 cflags-$(CONFIG_MARCH_Z196_TUNE)	+= -mtune=z196
 cflags-$(CONFIG_MARCH_ZEC12_TUNE)	+= -mtune=zEC12
-cflags-$(CONFIG_MARCH_Z13_TUNE)	+= -mtune=z13
+cflags-$(CONFIG_MARCH_Z13_TUNE)		+= -mtune=z13
+cflags-$(CONFIG_MARCH_Z14_TUNE)		+= -mtune=z14
 
 cflags-y += -Wa,-I$(srctree)/arch/$(ARCH)/include
 

arch/s390/include/asm/Kbuild

@@ -16,4 +16,5 @@ generic-y += mcs_spinlock.h
 generic-y += mm-arch-hooks.h
 generic-y += preempt.h
 generic-y += trace_clock.h
+generic-y += unaligned.h
 generic-y += word-at-a-time.h

arch/s390/include/asm/cpcmd.h

@@ -10,9 +10,8 @@
 
 /*
  * the lowlevel function for cpcmd
- * the caller of __cpcmd has to ensure that the response buffer is below 2 GB
  */
-extern int __cpcmd(const char *cmd, char *response, int rlen, int *response_code);
+int __cpcmd(const char *cmd, char *response, int rlen, int *response_code);
 
 /*
  * cpcmd is the in-kernel interface for issuing CP commands
@@ -25,8 +24,8 @@ extern int __cpcmd(const char *cmd, char *response, int rlen, int *response_code
  * response_code: return pointer for VM's error code
  * return value: the size of the response. The caller can check if the buffer
  *		was large enough by comparing the return value and rlen
- * NOTE: If the response buffer is not below 2 GB, cpcmd can sleep
+ * NOTE: If the response buffer is not in real storage, cpcmd can sleep
  */
-extern int cpcmd(const char *cmd, char *response, int rlen, int *response_code);
+int cpcmd(const char *cmd, char *response, int rlen, int *response_code);
 
 #endif /* _ASM_S390_CPCMD_H */

arch/s390/include/asm/ebcdic.h

@@ -9,9 +9,7 @@
 #ifndef _EBCDIC_H
 #define _EBCDIC_H
 
-#ifndef _S390_TYPES_H
-#include <types.h>
-#endif
+#include <linux/types.h>
 
 extern __u8 _ascebc_500[256];   /* ASCII -> EBCDIC 500 conversion table */
 extern __u8 _ebcasc_500[256];   /* EBCDIC 500 -> ASCII conversion table */

arch/s390/include/asm/elf.h

@@ -191,7 +191,7 @@ struct arch_elf_state {
 	} while (0)
 
 #define CORE_DUMP_USE_REGSET
-#define ELF_EXEC_PAGESIZE	4096
+#define ELF_EXEC_PAGESIZE	PAGE_SIZE
 
 /*
  * This is the base location for PIE (ET_DYN with INTERP) loads. On

arch/s390/include/asm/ipl.h

@@ -81,7 +81,7 @@ struct ipl_parameter_block {
 		struct ipl_block_fcp fcp;
 		struct ipl_block_ccw ccw;
 	} ipl_info;
-} __attribute__((packed,aligned(4096)));
+} __packed __aligned(PAGE_SIZE);
 
 /*
  * IPL validity flags

arch/s390/include/asm/lowcore.h

@@ -95,46 +95,46 @@ struct lowcore {
 	__u64	int_clock;			/* 0x0310 */
 	__u64	mcck_clock;			/* 0x0318 */
 	__u64	clock_comparator;		/* 0x0320 */
+	__u64	boot_clock[2];			/* 0x0328 */
 
 	/* Current process. */
-	__u64	current_task;			/* 0x0328 */
-	__u8	pad_0x318[0x320-0x318];		/* 0x0330 */
-	__u64	kernel_stack;			/* 0x0338 */
+	__u64	current_task;			/* 0x0338 */
+	__u64	kernel_stack;			/* 0x0340 */
 
 	/* Interrupt, panic and restart stack. */
-	__u64	async_stack;			/* 0x0340 */
-	__u64	panic_stack;			/* 0x0348 */
-	__u64	restart_stack;			/* 0x0350 */
+	__u64	async_stack;			/* 0x0348 */
+	__u64	panic_stack;			/* 0x0350 */
+	__u64	restart_stack;			/* 0x0358 */
 
 	/* Restart function and parameter. */
-	__u64	restart_fn;			/* 0x0358 */
-	__u64	restart_data;			/* 0x0360 */
-	__u64	restart_source;			/* 0x0368 */
+	__u64	restart_fn;			/* 0x0360 */
+	__u64	restart_data;			/* 0x0368 */
+	__u64	restart_source;			/* 0x0370 */
 
 	/* Address space pointer. */
-	__u64	kernel_asce;			/* 0x0370 */
-	__u64	user_asce;			/* 0x0378 */
+	__u64	kernel_asce;			/* 0x0378 */
+	__u64	user_asce;			/* 0x0380 */
 
 	/*
 	 * The lpp and current_pid fields form a
 	 * 64-bit value that is set as program
 	 * parameter with the LPP instruction.
 	 */
-	__u32	lpp;				/* 0x0380 */
-	__u32	current_pid;			/* 0x0384 */
+	__u32	lpp;				/* 0x0388 */
+	__u32	current_pid;			/* 0x038c */
 
 	/* SMP info area */
-	__u32	cpu_nr;				/* 0x0388 */
-	__u32	softirq_pending;		/* 0x038c */
-	__u64	percpu_offset;			/* 0x0390 */
-	__u64	vdso_per_cpu_data;		/* 0x0398 */
-	__u64	machine_flags;			/* 0x03a0 */
-	__u32	preempt_count;			/* 0x03a8 */
-	__u8	pad_0x03ac[0x03b0-0x03ac];	/* 0x03ac */
-	__u64	gmap;				/* 0x03b0 */
-	__u32	spinlock_lockval;		/* 0x03b8 */
-	__u32	fpu_flags;			/* 0x03bc */
-	__u8	pad_0x03c0[0x0400-0x03c0];	/* 0x03c0 */
+	__u32	cpu_nr;				/* 0x0390 */
+	__u32	softirq_pending;		/* 0x0394 */
+	__u64	percpu_offset;			/* 0x0398 */
+	__u64	vdso_per_cpu_data;		/* 0x03a0 */
+	__u64	machine_flags;			/* 0x03a8 */
+	__u32	preempt_count;			/* 0x03b0 */
+	__u8	pad_0x03b4[0x03b8-0x03b4];	/* 0x03b4 */
+	__u64	gmap;				/* 0x03b8 */
+	__u32	spinlock_lockval;		/* 0x03c0 */
+	__u32	fpu_flags;			/* 0x03c4 */
+	__u8	pad_0x03c8[0x0400-0x03c8];	/* 0x03c8 */
 
 	/* Per cpu primary space access list */
 	__u32	paste[16];			/* 0x0400 */

arch/s390/include/asm/mman.h

@@ -1,11 +0,0 @@
-/*
- *  S390 version
- *
- *  Derived from "include/asm-i386/mman.h"
- */
-#ifndef __S390_MMAN_H__
-#define __S390_MMAN_H__
-
-#include <uapi/asm/mman.h>
-
-#endif /* __S390_MMAN_H__ */

arch/s390/include/asm/mmu_context.h

@@ -12,6 +12,7 @@
 #include <linux/mm_types.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)
@@ -33,7 +34,7 @@ static inline int init_new_context(struct task_struct *tsk,
 	mm->context.use_cmma = 0;
 #endif
 	switch (mm->context.asce_limit) {
-	case 1UL << 42:
+	case _REGION2_SIZE:
 		/*
 		 * forked 3-level task, fall through to set new asce with new
 		 * mm->pgd
@@ -49,12 +50,12 @@ static inline int init_new_context(struct task_struct *tsk,
 		mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 			_ASCE_USER_BITS | _ASCE_TYPE_REGION1;
 		break;
-	case 1UL << 53:
+	case _REGION1_SIZE:
 		/* forked 4-level task, set new asce with new mm->pgd */
 		mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 				   _ASCE_USER_BITS | _ASCE_TYPE_REGION2;
 		break;
-	case 1UL << 31:
+	case _REGION3_SIZE:
 		/* forked 2-level compat task, set new asce with new mm->pgd */
 		mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 				   _ASCE_USER_BITS | _ASCE_TYPE_SEGMENT;
@@ -138,30 +139,4 @@ static inline void activate_mm(struct mm_struct *prev,
 	set_user_asce(next);
 }
 
-static inline void arch_dup_mmap(struct mm_struct *oldmm,
-				 struct mm_struct *mm)
-{
-}
-
-static inline void arch_exit_mmap(struct mm_struct *mm)
-{
-}
-
-static inline void arch_unmap(struct mm_struct *mm,
-			struct vm_area_struct *vma,
-			unsigned long start, unsigned long end)
-{
-}
-
-static inline void arch_bprm_mm_init(struct mm_struct *mm,
-				     struct vm_area_struct *vma)
-{
-}
-
-static inline bool arch_vma_access_permitted(struct vm_area_struct *vma,
-		bool write, bool execute, bool foreign)
-{
-	/* by default, allow everything */
-	return true;
-}
 #endif /* __S390_MMU_CONTEXT_H */

arch/s390/include/asm/nmi.h

@@ -72,7 +72,7 @@ union mci {
 		u64 ar :  1; /* 33 access register validity */
 		u64 da :  1; /* 34 delayed access exception */
 		u64    :  1; /* 35 */
-		u64 gs :  1; /* 36 guarded storage registers */
+		u64 gs :  1; /* 36 guarded storage registers validity */
 		u64    :  5; /* 37-41 */
 		u64 pr :  1; /* 42 tod programmable register validity */
 		u64 fc :  1; /* 43 fp control register validity */

arch/s390/include/asm/page-states.h

@@ -13,6 +13,7 @@
 #define ESSA_SET_POT_VOLATILE		4
 #define ESSA_SET_STABLE_RESIDENT	5
 #define ESSA_SET_STABLE_IF_RESIDENT	6
+#define ESSA_SET_STABLE_NODAT		7
 
 #define ESSA_MAX	ESSA_SET_STABLE_IF_RESIDENT
 

arch/s390/include/asm/page.h

@@ -10,10 +10,14 @@
 #include <linux/const.h>
 #include <asm/types.h>
 
+#define _PAGE_SHIFT	12
+#define _PAGE_SIZE	(_AC(1, UL) << _PAGE_SHIFT)
+#define _PAGE_MASK	(~(_PAGE_SIZE - 1))
+
 /* PAGE_SHIFT determines the page size */
-#define PAGE_SHIFT      12
-#define PAGE_SIZE	(_AC(1,UL) << PAGE_SHIFT)
-#define PAGE_MASK       (~(PAGE_SIZE-1))
+#define PAGE_SHIFT	_PAGE_SHIFT
+#define PAGE_SIZE	_PAGE_SIZE
+#define PAGE_MASK	_PAGE_MASK
 #define PAGE_DEFAULT_ACC	0
 #define PAGE_DEFAULT_KEY	(PAGE_DEFAULT_ACC << 4)
 
@@ -133,6 +137,9 @@ static inline int page_reset_referenced(unsigned long addr)
 struct page;
 void arch_free_page(struct page *page, int order);
 void arch_alloc_page(struct page *page, int order);
+void arch_set_page_dat(struct page *page, int order);
+void arch_set_page_nodat(struct page *page, int order);
+int arch_test_page_nodat(struct page *page);
 void arch_set_page_states(int make_stable);
 
 static inline int devmem_is_allowed(unsigned long pfn)
@@ -145,16 +152,26 @@ static inline int devmem_is_allowed(unsigned long pfn)
 
 #endif /* !__ASSEMBLY__ */
 
-#define __PAGE_OFFSET           0x0UL
-#define PAGE_OFFSET             0x0UL
-#define __pa(x)                 (unsigned long)(x)
-#define __va(x)                 (void *)(unsigned long)(x)
-#define virt_to_page(kaddr)	pfn_to_page(__pa(kaddr) >> PAGE_SHIFT)
-#define page_to_phys(page)	(page_to_pfn(page) << PAGE_SHIFT)
-#define virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+#define __PAGE_OFFSET		0x0UL
+#define PAGE_OFFSET		0x0UL
+
+#define __pa(x)			((unsigned long)(x))
+#define __va(x)			((void *)(unsigned long)(x))
+
+#define virt_to_pfn(kaddr)	(__pa(kaddr) >> PAGE_SHIFT)
 #define pfn_to_virt(pfn)	__va((pfn) << PAGE_SHIFT)
+
+#define virt_to_page(kaddr)	pfn_to_page(virt_to_pfn(kaddr))
 #define page_to_virt(page)	pfn_to_virt(page_to_pfn(page))
 
+#define phys_to_pfn(kaddr)	((kaddr) >> PAGE_SHIFT)
+#define pfn_to_phys(pfn)	((pfn) << PAGE_SHIFT)
+
+#define phys_to_page(kaddr)	pfn_to_page(phys_to_pfn(kaddr))
+#define page_to_phys(page)	(page_to_pfn(page) << PAGE_SHIFT)
+
+#define virt_addr_valid(kaddr)	pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+
 #define VM_DATA_DEFAULT_FLAGS	(VM_READ | VM_WRITE | \
 				 VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
 

arch/s390/include/asm/pgalloc.h

@@ -15,6 +15,8 @@
 #include <linux/gfp.h>
 #include <linux/mm.h>
 
+#define CRST_ALLOC_ORDER 2
+
 unsigned long *crst_table_alloc(struct mm_struct *);
 void crst_table_free(struct mm_struct *, unsigned long *);
 
@@ -42,16 +44,16 @@ static inline void clear_table(unsigned long *s, unsigned long val, size_t n)
 
 static inline void crst_table_init(unsigned long *crst, unsigned long entry)
 {
-	clear_table(crst, entry, sizeof(unsigned long)*2048);
+	clear_table(crst, entry, _CRST_TABLE_SIZE);
 }
 
 static inline unsigned long pgd_entry_type(struct mm_struct *mm)
 {
-	if (mm->context.asce_limit <= (1UL << 31))
+	if (mm->context.asce_limit <= _REGION3_SIZE)
 		return _SEGMENT_ENTRY_EMPTY;
-	if (mm->context.asce_limit <= (1UL << 42))
+	if (mm->context.asce_limit <= _REGION2_SIZE)
 		return _REGION3_ENTRY_EMPTY;
-	if (mm->context.asce_limit <= (1UL << 53))
+	if (mm->context.asce_limit <= _REGION1_SIZE)
 		return _REGION2_ENTRY_EMPTY;
 	return _REGION1_ENTRY_EMPTY;
 }
@@ -119,7 +121,7 @@ static inline pgd_t *pgd_alloc(struct mm_struct *mm)
 
 	if (!table)
 		return NULL;
-	if (mm->context.asce_limit == (1UL << 31)) {
+	if (mm->context.asce_limit == _REGION3_SIZE) {
 		/* Forking a compat process with 2 page table levels */
 		if (!pgtable_pmd_page_ctor(virt_to_page(table))) {
 			crst_table_free(mm, table);
@@ -131,7 +133,7 @@ static inline pgd_t *pgd_alloc(struct mm_struct *mm)
 
 static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
 {
-	if (mm->context.asce_limit == (1UL << 31))
+	if (mm->context.asce_limit == _REGION3_SIZE)
 		pgtable_pmd_page_dtor(virt_to_page(pgd));
 	crst_table_free(mm, (unsigned long *) pgd);
 }
@@ -158,4 +160,8 @@ static inline void pmd_populate(struct mm_struct *mm,
 
 extern void rcu_table_freelist_finish(void);
 
+void vmem_map_init(void);
+void *vmem_crst_alloc(unsigned long val);
+pte_t *vmem_pte_alloc(void);
+
 #endif /* _S390_PGALLOC_H */

arch/s390/include/asm/pgtable.h

@@ -11,19 +11,6 @@
 #ifndef _ASM_S390_PGTABLE_H
 #define _ASM_S390_PGTABLE_H
 
-/*
- * The Linux memory management assumes a three-level page table setup.
- * For s390 64 bit we use up to four of the five levels the hardware
- * provides (region first tables are not used).
- *
- * The "pgd_xxx()" functions are trivial for a folded two-level
- * setup: the pgd is never bad, and a pmd always exists (as it's folded
- * into the pgd entry)
- *
- * This file contains the functions and defines necessary to modify and use
- * the S390 page table tree.
- */
-#ifndef __ASSEMBLY__
 #include <linux/sched.h>
 #include <linux/mm_types.h>
 #include <linux/page-flags.h>
@@ -34,9 +21,6 @@
 
 extern pgd_t swapper_pg_dir[];
 extern void paging_init(void);
-extern void vmem_map_init(void);
-pmd_t *vmem_pmd_alloc(void);
-pte_t *vmem_pte_alloc(void);
 
 enum {
 	PG_DIRECT_MAP_4K = 0,
@@ -77,38 +61,6 @@ extern unsigned long zero_page_mask;
 #define __HAVE_COLOR_ZERO_PAGE
 
 /* TODO: s390 cannot support io_remap_pfn_range... */
-#endif /* !__ASSEMBLY__ */
-
-/*
- * PMD_SHIFT determines the size of the area a second-level page
- * table can map
- * PGDIR_SHIFT determines what a third-level page table entry can map
- */
-#define PMD_SHIFT	20
-#define PUD_SHIFT	31
-#define P4D_SHIFT	42
-#define PGDIR_SHIFT	53
-
-#define PMD_SIZE        (1UL << PMD_SHIFT)
-#define PMD_MASK        (~(PMD_SIZE-1))
-#define PUD_SIZE	(1UL << PUD_SHIFT)
-#define PUD_MASK	(~(PUD_SIZE-1))
-#define P4D_SIZE	(1UL << P4D_SHIFT)
-#define P4D_MASK	(~(P4D_SIZE-1))
-#define PGDIR_SIZE	(1UL << PGDIR_SHIFT)
-#define PGDIR_MASK	(~(PGDIR_SIZE-1))
-
-/*
- * entries per page directory level: the S390 is two-level, so
- * we don't really have any PMD directory physically.
- * for S390 segment-table entries are combined to one PGD
- * that leads to 1024 pte per pgd
- */
-#define PTRS_PER_PTE	256
-#define PTRS_PER_PMD	2048
-#define PTRS_PER_PUD	2048
-#define PTRS_PER_P4D	2048
-#define PTRS_PER_PGD	2048
 
 #define FIRST_USER_ADDRESS  0UL
 
@@ -123,7 +75,6 @@ extern unsigned long zero_page_mask;
 #define pgd_ERROR(e) \
 	printk("%s:%d: bad pgd %p.\n", __FILE__, __LINE__, (void *) pgd_val(e))
 
-#ifndef __ASSEMBLY__
 /*
  * The vmalloc and module area will always be on the topmost area of the
  * kernel mapping. We reserve 128GB (64bit) for vmalloc and modules.
@@ -269,7 +220,7 @@ static inline int is_module_addr(void *addr)
  */
 
 /* Bits in the segment/region table address-space-control-element */
-#define _ASCE_ORIGIN		~0xfffUL/* segment table origin		    */
+#define _ASCE_ORIGIN		~0xfffUL/* region/segment table origin	    */
 #define _ASCE_PRIVATE_SPACE	0x100	/* private space control	    */
 #define _ASCE_ALT_EVENT		0x80	/* storage alteration event control */
 #define _ASCE_SPACE_SWITCH	0x40	/* space switch event		    */
@@ -320,9 +271,9 @@ static inline int is_module_addr(void *addr)
 #define _SEGMENT_ENTRY_BITS	0xfffffffffffffe33UL
 #define _SEGMENT_ENTRY_BITS_LARGE 0xfffffffffff0ff33UL
 #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_ORIGIN	~0x7ffUL/* page table origin		    */
+#define _SEGMENT_ENTRY_PROTECT	0x200	/* segment protection bit	    */
+#define _SEGMENT_ENTRY_NOEXEC	0x100	/* segment no-execute bit	    */
 #define _SEGMENT_ENTRY_INVALID	0x20	/* invalid segment table entry	    */
 
 #define _SEGMENT_ENTRY		(0)
@@ -340,6 +291,54 @@ static inline int is_module_addr(void *addr)
 #define _SEGMENT_ENTRY_SOFT_DIRTY 0x0000 /* SW segment soft dirty bit */
 #endif
 
+#define _CRST_ENTRIES	2048	/* number of region/segment table entries */
+#define _PAGE_ENTRIES	256	/* number of page table entries	*/
+
+#define _CRST_TABLE_SIZE (_CRST_ENTRIES * 8)
+#define _PAGE_TABLE_SIZE (_PAGE_ENTRIES * 8)
+
+#define _REGION1_SHIFT	53
+#define _REGION2_SHIFT	42
+#define _REGION3_SHIFT	31
+#define _SEGMENT_SHIFT	20
+
+#define _REGION1_INDEX	(0x7ffUL << _REGION1_SHIFT)
+#define _REGION2_INDEX	(0x7ffUL << _REGION2_SHIFT)
+#define _REGION3_INDEX	(0x7ffUL << _REGION3_SHIFT)
+#define _SEGMENT_INDEX	(0x7ffUL << _SEGMENT_SHIFT)
+#define _PAGE_INDEX	(0xffUL  << _PAGE_SHIFT)
+
+#define _REGION1_SIZE	(1UL << _REGION1_SHIFT)
+#define _REGION2_SIZE	(1UL << _REGION2_SHIFT)
+#define _REGION3_SIZE	(1UL << _REGION3_SHIFT)
+#define _SEGMENT_SIZE	(1UL << _SEGMENT_SHIFT)
+
+#define _REGION1_MASK	(~(_REGION1_SIZE - 1))
+#define _REGION2_MASK	(~(_REGION2_SIZE - 1))
+#define _REGION3_MASK	(~(_REGION3_SIZE - 1))
+#define _SEGMENT_MASK	(~(_SEGMENT_SIZE - 1))
+
+#define PMD_SHIFT	_SEGMENT_SHIFT
+#define PUD_SHIFT	_REGION3_SHIFT
+#define P4D_SHIFT	_REGION2_SHIFT
+#define PGDIR_SHIFT	_REGION1_SHIFT
+
+#define PMD_SIZE	_SEGMENT_SIZE
+#define PUD_SIZE	_REGION3_SIZE
+#define P4D_SIZE	_REGION2_SIZE
+#define PGDIR_SIZE	_REGION1_SIZE
+
+#define PMD_MASK	_SEGMENT_MASK
+#define PUD_MASK	_REGION3_MASK
+#define P4D_MASK	_REGION2_MASK
+#define PGDIR_MASK	_REGION1_MASK
+
+#define PTRS_PER_PTE	_PAGE_ENTRIES
+#define PTRS_PER_PMD	_CRST_ENTRIES
+#define PTRS_PER_PUD	_CRST_ENTRIES
+#define PTRS_PER_P4D	_CRST_ENTRIES
+#define PTRS_PER_PGD	_CRST_ENTRIES
+
 /*
  * Segment table and region3 table entry encoding
  * (R = read-only, I = invalid, y = young bit):
@@ -376,6 +375,7 @@ static inline int is_module_addr(void *addr)
 
 /* Guest Page State used for virtualization */
 #define _PGSTE_GPS_ZERO			0x0000000080000000UL
+#define _PGSTE_GPS_NODAT		0x0000000040000000UL
 #define _PGSTE_GPS_USAGE_MASK		0x0000000003000000UL
 #define _PGSTE_GPS_USAGE_STABLE		0x0000000000000000UL
 #define _PGSTE_GPS_USAGE_UNUSED		0x0000000001000000UL
@@ -505,7 +505,7 @@ static inline int mm_alloc_pgste(struct mm_struct *mm)
  * In the case that a guest uses storage keys
  * faults should no longer be backed by zero pages
  */
-#define mm_forbids_zeropage mm_use_skey
+#define mm_forbids_zeropage mm_has_pgste
 static inline int mm_use_skey(struct mm_struct *mm)
 {
 #ifdef CONFIG_PGSTE
@@ -952,15 +952,30 @@ static inline pte_t pte_mkhuge(pte_t pte)
 #define IPTE_GLOBAL	0
 #define	IPTE_LOCAL	1
 
-static inline void __ptep_ipte(unsigned long address, pte_t *ptep, int local)
+#define IPTE_NODAT	0x400
+#define IPTE_GUEST_ASCE	0x800
+
+static inline void __ptep_ipte(unsigned long address, pte_t *ptep,
+			       unsigned long opt, unsigned long asce,
+			       int local)
 {
 	unsigned long pto = (unsigned long) ptep;
 
-	/* Invalidation + TLB flush for the pte */
+	if (__builtin_constant_p(opt) && opt == 0) {
+		/* Invalidation + TLB flush for the pte */
+		asm volatile(
+			"	.insn	rrf,0xb2210000,%[r1],%[r2],0,%[m4]"
+			: "+m" (*ptep) : [r1] "a" (pto), [r2] "a" (address),
+			  [m4] "i" (local));
+		return;
+	}
+
+	/* Invalidate ptes with options + TLB flush of the ptes */
+	opt = opt | (asce & _ASCE_ORIGIN);
 	asm volatile(
-		"       .insn rrf,0xb2210000,%[r1],%[r2],0,%[m4]"
-		: "+m" (*ptep) : [r1] "a" (pto), [r2] "a" (address),
-		  [m4] "i" (local));
+		"	.insn	rrf,0xb2210000,%[r1],%[r2],%[r3],%[m4]"
+		: [r2] "+a" (address), [r3] "+a" (opt)
+		: [r1] "a" (pto), [m4] "i" (local) : "memory");
 }
 
 static inline void __ptep_ipte_range(unsigned long address, int nr,
@@ -1341,31 +1356,61 @@ static inline void __pmdp_csp(pmd_t *pmdp)
 #define IDTE_GLOBAL	0
 #define IDTE_LOCAL	1
 
-static inline void __pmdp_idte(unsigned long address, pmd_t *pmdp, int local)
+#define IDTE_PTOA	0x0800
+#define IDTE_NODAT	0x1000
+#define IDTE_GUEST_ASCE	0x2000
+
+static inline void __pmdp_idte(unsigned long addr, pmd_t *pmdp,
+			       unsigned long opt, unsigned long asce,
+			       int local)
 {
 	unsigned long sto;
 
-	sto = (unsigned long) pmdp - pmd_index(address) * sizeof(pmd_t);
-	asm volatile(
-		"	.insn	rrf,0xb98e0000,%[r1],%[r2],0,%[m4]"
-		: "+m" (*pmdp)
-		: [r1] "a" (sto), [r2] "a" ((address & HPAGE_MASK)),
-		  [m4] "i" (local)
-		: "cc" );
+	sto = (unsigned long) pmdp - pmd_index(addr) * sizeof(pmd_t);
+	if (__builtin_constant_p(opt) && opt == 0) {
+		/* flush without guest asce */
+		asm volatile(
+			"	.insn	rrf,0xb98e0000,%[r1],%[r2],0,%[m4]"
+			: "+m" (*pmdp)
+			: [r1] "a" (sto), [r2] "a" ((addr & HPAGE_MASK)),
+			  [m4] "i" (local)
+			: "cc" );
+	} else {
+		/* flush with guest asce */
+		asm volatile(
+			"	.insn	rrf,0xb98e0000,%[r1],%[r2],%[r3],%[m4]"
+			: "+m" (*pmdp)
+			: [r1] "a" (sto), [r2] "a" ((addr & HPAGE_MASK) | opt),
+			  [r3] "a" (asce), [m4] "i" (local)
+			: "cc" );
+	}
 }
 
-static inline void __pudp_idte(unsigned long address, pud_t *pudp, int local)
+static inline void __pudp_idte(unsigned long addr, pud_t *pudp,
+			       unsigned long opt, unsigned long asce,
+			       int local)
 {
 	unsigned long r3o;
 
-	r3o = (unsigned long) pudp - pud_index(address) * sizeof(pud_t);
+	r3o = (unsigned long) pudp - pud_index(addr) * sizeof(pud_t);
 	r3o |= _ASCE_TYPE_REGION3;
-	asm volatile(
-		"	.insn	rrf,0xb98e0000,%[r1],%[r2],0,%[m4]"
-		: "+m" (*pudp)
-		: [r1] "a" (r3o), [r2] "a" ((address & PUD_MASK)),
-		  [m4] "i" (local)
-		: "cc");
+	if (__builtin_constant_p(opt) && opt == 0) {
+		/* flush without guest asce */
+		asm volatile(
+			"	.insn	rrf,0xb98e0000,%[r1],%[r2],0,%[m4]"
+			: "+m" (*pudp)
+			: [r1] "a" (r3o), [r2] "a" ((addr & PUD_MASK)),
+			  [m4] "i" (local)
+			: "cc");
+	} else {
+		/* flush with guest asce */
+		asm volatile(
+			"	.insn	rrf,0xb98e0000,%[r1],%[r2],%[r3],%[m4]"
+			: "+m" (*pudp)
+			: [r1] "a" (r3o), [r2] "a" ((addr & PUD_MASK) | opt),
+			  [r3] "a" (asce), [m4] "i" (local)
+			: "cc" );
+	}
 }
 
 pmd_t pmdp_xchg_direct(struct mm_struct *, unsigned long, pmd_t *, pmd_t);
@@ -1548,8 +1593,6 @@ static inline swp_entry_t __swp_entry(unsigned long type, unsigned long offset)
 #define __pte_to_swp_entry(pte)	((swp_entry_t) { pte_val(pte) })
 #define __swp_entry_to_pte(x)	((pte_t) { (x).val })
 
-#endif /* !__ASSEMBLY__ */
-
 #define kern_addr_valid(addr)   (1)
 
 extern int vmem_add_mapping(unsigned long start, unsigned long size);

arch/s390/include/asm/qdio.h

@@ -80,7 +80,7 @@ struct qdr {
 	u32 qkey   : 4;
 	u32	   : 28;
 	struct qdesfmt0 qdf0[126];
-} __attribute__ ((packed, aligned(4096)));
+} __packed __aligned(PAGE_SIZE);
 
 #define QIB_AC_OUTBOUND_PCI_SUPPORTED	0x40
 #define QIB_RFLAGS_ENABLE_QEBSM		0x80

arch/s390/include/asm/setup.h

@@ -29,8 +29,10 @@
 #define MACHINE_FLAG_TE		_BITUL(11)
 #define MACHINE_FLAG_TLB_LC	_BITUL(12)
 #define MACHINE_FLAG_VX		_BITUL(13)
-#define MACHINE_FLAG_NX		_BITUL(14)
-#define MACHINE_FLAG_GS		_BITUL(15)
+#define MACHINE_FLAG_TLB_GUEST	_BITUL(14)
+#define MACHINE_FLAG_NX		_BITUL(15)
+#define MACHINE_FLAG_GS		_BITUL(16)
+#define MACHINE_FLAG_SCC	_BITUL(17)
 
 #define LPP_MAGIC		_BITUL(31)
 #define LPP_PFAULT_PID_MASK	_AC(0xffffffff, UL)
@@ -68,8 +70,10 @@ extern void detect_memory_memblock(void);
 #define MACHINE_HAS_TE		(S390_lowcore.machine_flags & MACHINE_FLAG_TE)
 #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_TLB_GUEST	(S390_lowcore.machine_flags & MACHINE_FLAG_TLB_GUEST)
 #define MACHINE_HAS_NX		(S390_lowcore.machine_flags & MACHINE_FLAG_NX)
 #define MACHINE_HAS_GS		(S390_lowcore.machine_flags & MACHINE_FLAG_GS)
+#define MACHINE_HAS_SCC		(S390_lowcore.machine_flags & MACHINE_FLAG_SCC)
 
 /*
  * Console mode. Override with conmode=
@@ -104,9 +108,16 @@ extern void pfault_fini(void);
 #define pfault_fini()		do { } while (0)
 #endif /* CONFIG_PFAULT */
 
+#ifdef CONFIG_VMCP
+void vmcp_cma_reserve(void);
+#else
+static inline void vmcp_cma_reserve(void) { }
+#endif
+
 void report_user_fault(struct pt_regs *regs, long signr, int is_mm_fault);
 
-extern void cmma_init(void);
+void cmma_init(void);
+void cmma_init_nodat(void);
 
 extern void (*_machine_restart)(char *command);
 extern void (*_machine_halt)(void);

arch/s390/include/asm/spinlock.h

@@ -92,10 +92,11 @@ static inline void arch_spin_unlock(arch_spinlock_t *lp)
 {
 	typecheck(int, lp->lock);
 	asm volatile(
-		"st	%1,%0\n"
-		: "+Q" (lp->lock)
-		: "d" (0)
-		: "cc", "memory");
+#ifdef CONFIG_HAVE_MARCH_ZEC12_FEATURES
+		"	.long	0xb2fa0070\n"	/* NIAI 7 */
+#endif
+		"	st	%1,%0\n"
+		: "=Q" (lp->lock) : "d" (0) : "cc", "memory");
 }
 
 /*

arch/s390/include/asm/timex.h

@@ -15,6 +15,8 @@
 /* The value of the TOD clock for 1.1.1970. */
 #define TOD_UNIX_EPOCH 0x7d91048bca000000ULL
 
+extern u64 clock_comparator_max;
+
 /* Inline functions for clock register access. */
 static inline int set_tod_clock(__u64 time)
 {
@@ -126,7 +128,7 @@ static inline unsigned long long local_tick_disable(void)
 	unsigned long long old;
 
 	old = S390_lowcore.clock_comparator;
-	S390_lowcore.clock_comparator = -1ULL;
+	S390_lowcore.clock_comparator = clock_comparator_max;
 	set_clock_comparator(S390_lowcore.clock_comparator);
 	return old;
 }
@@ -174,24 +176,24 @@ static inline cycles_t get_cycles(void)
 	return (cycles_t) get_tod_clock() >> 2;
 }
 
-int get_phys_clock(unsigned long long *clock);
+int get_phys_clock(unsigned long *clock);
 void init_cpu_timer(void);
 unsigned long long monotonic_clock(void);
 
-extern u64 sched_clock_base_cc;
+extern unsigned char tod_clock_base[16] __aligned(8);
 
 /**
  * get_clock_monotonic - returns current time in clock rate units
  *
  * The caller must ensure that preemption is disabled.
- * The clock and sched_clock_base get changed via stop_machine.
+ * The clock and tod_clock_base get changed via stop_machine.
  * Therefore preemption must be disabled when calling this
  * function, otherwise the returned value is not guaranteed to
  * be monotonic.
  */
 static inline unsigned long long get_tod_clock_monotonic(void)
 {
-	return get_tod_clock() - sched_clock_base_cc;
+	return get_tod_clock() - *(unsigned long long *) &tod_clock_base[1];
 }
 
 /**
@@ -218,4 +220,32 @@ static inline unsigned long long tod_to_ns(unsigned long long todval)
 	return ((todval >> 9) * 125) + (((todval & 0x1ff) * 125) >> 9);
 }
 
+/**
+ * tod_after - compare two 64 bit TOD values
+ * @a: first 64 bit TOD timestamp
+ * @b: second 64 bit TOD timestamp
+ *
+ * Returns: true if a is later than b
+ */
+static inline int tod_after(unsigned long long a, unsigned long long b)
+{
+	if (MACHINE_HAS_SCC)
+		return (long long) a > (long long) b;
+	return a > b;
+}
+
+/**
+ * tod_after_eq - compare two 64 bit TOD values
+ * @a: first 64 bit TOD timestamp
+ * @b: second 64 bit TOD timestamp
+ *
+ * Returns: true if a is later than b
+ */
+static inline int tod_after_eq(unsigned long long a, unsigned long long b)
+{
+	if (MACHINE_HAS_SCC)
+		return (long long) a >= (long long) b;
+	return a >= b;
+}
+
 #endif

arch/s390/include/asm/tlb.h

@@ -135,7 +135,7 @@ static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte,
 static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
 				unsigned long address)
 {
-	if (tlb->mm->context.asce_limit <= (1UL << 31))
+	if (tlb->mm->context.asce_limit <= _REGION3_SIZE)
 		return;
 	pgtable_pmd_page_dtor(virt_to_page(pmd));
 	tlb_remove_table(tlb, pmd);
@@ -151,7 +151,7 @@ static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
 static inline void p4d_free_tlb(struct mmu_gather *tlb, p4d_t *p4d,
 				unsigned long address)
 {
-	if (tlb->mm->context.asce_limit <= (1UL << 53))
+	if (tlb->mm->context.asce_limit <= _REGION1_SIZE)
 		return;
 	tlb_remove_table(tlb, p4d);
 }
@@ -166,7 +166,7 @@ static inline void p4d_free_tlb(struct mmu_gather *tlb, p4d_t *p4d,
 static inline void pud_free_tlb(struct mmu_gather *tlb, pud_t *pud,
 				unsigned long address)
 {
-	if (tlb->mm->context.asce_limit <= (1UL << 42))
+	if (tlb->mm->context.asce_limit <= _REGION2_SIZE)
 		return;
 	tlb_remove_table(tlb, pud);
 }

arch/s390/include/asm/tlbflush.h

@@ -20,10 +20,15 @@ static inline void __tlb_flush_local(void)
  */
 static inline void __tlb_flush_idte(unsigned long asce)
 {
+	unsigned long opt;
+
+	opt = IDTE_PTOA;
+	if (MACHINE_HAS_TLB_GUEST)
+		opt |= IDTE_GUEST_ASCE;
 	/* Global TLB flush for the mm */
 	asm volatile(
 		"	.insn	rrf,0xb98e0000,0,%0,%1,0"
-		: : "a" (2048), "a" (asce) : "cc");
+		: : "a" (opt), "a" (asce) : "cc");
 }
 
 #ifdef CONFIG_SMP

arch/s390/include/asm/topology.h

@@ -77,12 +77,6 @@ static inline const struct cpumask *cpumask_of_node(int node)
 	return &node_to_cpumask_map[node];
 }
 
-/*
- * Returns the number of the node containing node 'node'. This
- * architecture is flat, so it is a pretty simple function!
- */
-#define parent_node(node) (node)
-
 #define pcibus_to_node(bus) __pcibus_to_node(bus)
 
 #define node_distance(a, b) __node_distance(a, b)

arch/s390/include/asm/types.h

@@ -1,11 +0,0 @@
-/*
- *  S390 version
- *
- *  Derived from "include/asm-i386/types.h"
- */
-#ifndef _S390_TYPES_H
-#define _S390_TYPES_H
-
-#include <uapi/asm/types.h>
-
-#endif /* _S390_TYPES_H */

arch/s390/include/asm/unaligned.h

@@ -1,13 +0,0 @@
-#ifndef _ASM_S390_UNALIGNED_H
-#define _ASM_S390_UNALIGNED_H
-
-/*
- * The S390 can do unaligned accesses itself. 
- */
-#include <linux/unaligned/access_ok.h>
-#include <linux/unaligned/generic.h>
-
-#define get_unaligned	__get_unaligned_be
-#define put_unaligned	__put_unaligned_be
-
-#endif /* _ASM_S390_UNALIGNED_H */

arch/s390/include/uapi/asm/Kbuild

@@ -9,4 +9,5 @@ generic-y += param.h
 generic-y += poll.h
 generic-y += resource.h
 generic-y += sockios.h
+generic-y += swab.h
 generic-y += termbits.h

arch/s390/include/uapi/asm/dasd.h

@@ -72,7 +72,10 @@ typedef struct dasd_information2_t {
  * 0x02: use diag discipline (diag)
  * 0x04: set the device initially online (internal use only)
  * 0x08: enable ERP related logging
- * 0x20: give access to raw eckd data
+ * 0x10: allow I/O to fail on lost paths
+ * 0x20: allow I/O to fail when a lock was stolen
+ * 0x40: give access to raw eckd data
+ * 0x80: enable discard support
  */
 #define DASD_FEATURE_DEFAULT	     0x00
 #define DASD_FEATURE_READONLY	     0x01
@@ -82,6 +85,7 @@ typedef struct dasd_information2_t {
 #define DASD_FEATURE_FAILFAST	     0x10
 #define DASD_FEATURE_FAILONSLCK      0x20
 #define DASD_FEATURE_USERAW	     0x40
+#define DASD_FEATURE_DISCARD	     0x80
 
 #define DASD_PARTN_BITS 2
 

arch/s390/include/uapi/asm/swab.h

@@ -1,89 +0,0 @@
-#ifndef _S390_SWAB_H
-#define _S390_SWAB_H
-
-/*
- *  S390 version
- *    Copyright IBM Corp. 1999
- *    Author(s): Martin Schwidefsky (schwidefsky@de.ibm.com)
- */
-
-#include <linux/types.h>
-
-#ifndef __s390x__
-# define __SWAB_64_THRU_32__
-#endif
-
-#ifdef __s390x__
-static inline __u64 __arch_swab64p(const __u64 *x)
-{
-	__u64 result;
-
-	asm volatile("lrvg %0,%1" : "=d" (result) : "m" (*x));
-	return result;
-}
-#define __arch_swab64p __arch_swab64p
-
-static inline __u64 __arch_swab64(__u64 x)
-{
-	__u64 result;
-
-	asm volatile("lrvgr %0,%1" : "=d" (result) : "d" (x));
-	return result;
-}
-#define __arch_swab64 __arch_swab64
-
-static inline void __arch_swab64s(__u64 *x)
-{
-	*x = __arch_swab64p(x);
-}
-#define __arch_swab64s __arch_swab64s
-#endif /* __s390x__ */
-
-static inline __u32 __arch_swab32p(const __u32 *x)
-{
-	__u32 result;
-	
-	asm volatile(
-#ifndef __s390x__
-		"	icm	%0,8,%O1+3(%R1)\n"
-		"	icm	%0,4,%O1+2(%R1)\n"
-		"	icm	%0,2,%O1+1(%R1)\n"
-		"	ic	%0,%1"
-		: "=&d" (result) : "Q" (*x) : "cc");
-#else /* __s390x__ */
-		"	lrv	%0,%1"
-		: "=d" (result) : "m" (*x));
-#endif /* __s390x__ */
-	return result;
-}
-#define __arch_swab32p __arch_swab32p
-
-#ifdef __s390x__
-static inline __u32 __arch_swab32(__u32 x)
-{
-	__u32 result;
-	
-	asm volatile("lrvr  %0,%1" : "=d" (result) : "d" (x));
-	return result;
-}
-#define __arch_swab32 __arch_swab32
-#endif /* __s390x__ */
-
-static inline __u16 __arch_swab16p(const __u16 *x)
-{
-	__u16 result;
-	
-	asm volatile(
-#ifndef __s390x__
-		"	icm	%0,2,%O1+1(%R1)\n"
-		"	ic	%0,%1\n"
-		: "=&d" (result) : "Q" (*x) : "cc");
-#else /* __s390x__ */
-		"	lrvh	%0,%1"
-		: "=d" (result) : "m" (*x));
-#endif /* __s390x__ */
-	return result;
-}
-#define __arch_swab16p __arch_swab16p
-
-#endif /* _S390_SWAB_H */

drivers/s390/char/vmcp.h → arch/s390/include/uapi/asm/vmcp.h

@@ -12,19 +12,13 @@
  * The idea of this driver is based on cpint from Neale Ferguson
  */
 
+#ifndef _UAPI_ASM_VMCP_H
+#define _UAPI_ASM_VMCP_H
+
 #include <linux/ioctl.h>
-#include <linux/mutex.h>
 
-#define VMCP_GETCODE _IOR(0x10, 1, int)
-#define VMCP_SETBUF _IOW(0x10, 2, int)
-#define VMCP_GETSIZE _IOR(0x10, 3, int)
+#define VMCP_GETCODE	_IOR(0x10, 1, int)
+#define VMCP_SETBUF	_IOW(0x10, 2, int)
+#define VMCP_GETSIZE	_IOR(0x10, 3, int)
 
-struct vmcp_session {
-	unsigned int bufsize;
-	char *response;
-	int resp_size;
-	int resp_code;
-	/* As we use copy_from/to_user, which might     *
-	 * sleep and cannot use a spinlock              */
-	struct mutex mutex;
-};
+#endif /* _UAPI_ASM_VMCP_H */

arch/s390/kernel/asm-offsets.c

@@ -158,6 +158,7 @@ int main(void)
 	OFFSET(__LC_LAST_UPDATE_CLOCK, lowcore, last_update_clock);
 	OFFSET(__LC_INT_CLOCK, lowcore, int_clock);
 	OFFSET(__LC_MCCK_CLOCK, lowcore, mcck_clock);
+	OFFSET(__LC_BOOT_CLOCK, lowcore, boot_clock);
 	OFFSET(__LC_CURRENT, lowcore, current_task);
 	OFFSET(__LC_KERNEL_STACK, lowcore, kernel_stack);
 	OFFSET(__LC_ASYNC_STACK, lowcore, async_stack);

arch/s390/kernel/cpcmd.c

@@ -14,6 +14,7 @@
 #include <linux/spinlock.h>
 #include <linux/stddef.h>
 #include <linux/string.h>
+#include <linux/mm.h>
 #include <asm/diag.h>
 #include <asm/ebcdic.h>
 #include <asm/cpcmd.h>
@@ -28,9 +29,7 @@ static int diag8_noresponse(int cmdlen)
 	register unsigned long reg3 asm ("3") = cmdlen;
 
 	asm volatile(
-		"	sam31\n"
 		"	diag	%1,%0,0x8\n"
-		"	sam64\n"
 		: "+d" (reg3) : "d" (reg2) : "cc");
 	return reg3;
 }
@@ -43,9 +42,7 @@ static int diag8_response(int cmdlen, char *response, int *rlen)
 	register unsigned long reg5 asm ("5") = *rlen;
 
 	asm volatile(
-		"	sam31\n"
 		"	diag	%2,%0,0x8\n"
-		"	sam64\n"
 		"	brc	8,1f\n"
 		"	agr	%1,%4\n"
 		"1:\n"
@@ -57,7 +54,6 @@ static int diag8_response(int cmdlen, char *response, int *rlen)
 
 /*
  * __cpcmd has some restrictions over cpcmd
- *  - the response buffer must reside below 2GB (if any)
  *  - __cpcmd is unlocked and therefore not SMP-safe
  */
 int  __cpcmd(const char *cmd, char *response, int rlen, int *response_code)
@@ -88,13 +84,12 @@ EXPORT_SYMBOL(__cpcmd);
 
 int cpcmd(const char *cmd, char *response, int rlen, int *response_code)
 {
+	unsigned long flags;
 	char *lowbuf;
 	int len;
-	unsigned long flags;
 
-	if ((virt_to_phys(response) != (unsigned long) response) ||
-			(((unsigned long)response + rlen) >> 31)) {
-		lowbuf = kmalloc(rlen, GFP_KERNEL | GFP_DMA);
+	if (is_vmalloc_or_module_addr(response)) {
+		lowbuf = kmalloc(rlen, GFP_KERNEL);
 		if (!lowbuf) {
 			pr_warn("The cpcmd kernel function failed to allocate a response buffer\n");
 			return -ENOMEM;

arch/s390/kernel/debug.c

@@ -866,7 +866,8 @@ static inline void
 debug_finish_entry(debug_info_t * id, debug_entry_t* active, int level,
 			int exception)
 {
-	active->id.stck = get_tod_clock_fast() - sched_clock_base_cc;
+	active->id.stck = get_tod_clock_fast() -
+		*(unsigned long long *) &tod_clock_base[1];
 	active->id.fields.cpuid = smp_processor_id();
 	active->caller = __builtin_return_address(0);
 	active->id.fields.exception = exception;
@@ -1455,15 +1456,15 @@ int
 debug_dflt_header_fn(debug_info_t * id, struct debug_view *view,
 			 int area, debug_entry_t * entry, char *out_buf)
 {
-	unsigned long sec, usec;
+	unsigned long base, sec, usec;
 	char *except_str;
 	unsigned long caller;
 	int rc = 0;
 	unsigned int level;
 
 	level = entry->id.fields.level;
-	sec = (entry->id.stck >> 12) + (sched_clock_base_cc >> 12);
-	sec = sec - (TOD_UNIX_EPOCH >> 12);
+	base = (*(unsigned long *) &tod_clock_base[0]) >> 4;
+	sec = (entry->id.stck >> 12) + base - (TOD_UNIX_EPOCH >> 12);
 	usec = do_div(sec, USEC_PER_SEC);
 
 	if (entry->id.fields.exception)

arch/s390/kernel/dumpstack.c

@@ -76,7 +76,7 @@ void dump_trace(dump_trace_func_t func, void *data, struct task_struct *task,
 	frame_size = STACK_FRAME_OVERHEAD + sizeof(struct pt_regs);
 #ifdef CONFIG_CHECK_STACK
 	sp = __dump_trace(func, data, sp,
-			  S390_lowcore.panic_stack + frame_size - 4096,
+			  S390_lowcore.panic_stack + frame_size - PAGE_SIZE,
 			  S390_lowcore.panic_stack + frame_size);
 #endif
 	sp = __dump_trace(func, data, sp,

arch/s390/kernel/early.c

@@ -53,8 +53,9 @@ static void __init reset_tod_clock(void)
 	if (set_tod_clock(TOD_UNIX_EPOCH) != 0 || store_tod_clock(&time) != 0)
 		disabled_wait(0);
 
-	sched_clock_base_cc = TOD_UNIX_EPOCH;
-	S390_lowcore.last_update_clock = sched_clock_base_cc;
+	memset(tod_clock_base, 0, 16);
+	*(__u64 *) &tod_clock_base[1] = TOD_UNIX_EPOCH;
+	S390_lowcore.last_update_clock = TOD_UNIX_EPOCH;
 }
 
 #ifdef CONFIG_SHARED_KERNEL
@@ -165,8 +166,8 @@ static noinline __init void create_kernel_nss(void)
 	}
 
 	/* re-initialize cputime accounting. */
-	sched_clock_base_cc = get_tod_clock();
-	S390_lowcore.last_update_clock = sched_clock_base_cc;
+	get_tod_clock_ext(tod_clock_base);
+	S390_lowcore.last_update_clock = *(__u64 *) &tod_clock_base[1];
 	S390_lowcore.last_update_timer = 0x7fffffffffffffffULL;
 	S390_lowcore.user_timer = 0;
 	S390_lowcore.system_timer = 0;
@@ -387,6 +388,12 @@ static __init void detect_machine_facilities(void)
 	}
 	if (test_facility(133))
 		S390_lowcore.machine_flags |= MACHINE_FLAG_GS;
+	if (test_facility(139) && (tod_clock_base[1] & 0x80)) {
+		/* Enabled signed clock comparator comparisons */
+		S390_lowcore.machine_flags |= MACHINE_FLAG_SCC;
+		clock_comparator_max = -1ULL >> 1;
+		__ctl_set_bit(0, 53);
+	}
 }
 
 static inline void save_vector_registers(void)
@@ -413,7 +420,7 @@ static int __init disable_vector_extension(char *str)
 {
 	S390_lowcore.machine_flags &= ~MACHINE_FLAG_VX;
 	__ctl_clear_bit(0, 17);
-	return 1;
+	return 0;
 }
 early_param("novx", disable_vector_extension);
 

arch/s390/kernel/head.S

@@ -302,7 +302,8 @@ ENTRY(startup_kdump)
 	xc	0xe00(256),0xe00
 	xc	0xf00(256),0xf00
 	lctlg	%c0,%c15,0x200(%r0)	# initialize control registers
-	stck	__LC_LAST_UPDATE_CLOCK
+	stcke	__LC_BOOT_CLOCK
+	mvc	__LC_LAST_UPDATE_CLOCK(8),__LC_BOOT_CLOCK+1
 	spt	6f-.LPG0(%r13)
 	mvc	__LC_LAST_UPDATE_TIMER(8),6f-.LPG0(%r13)
 	l	%r15,.Lstack-.LPG0(%r13)

arch/s390/kernel/head64.S

@@ -21,8 +21,8 @@ ENTRY(startup_continue)
 	xc	__LC_LPP+1(7,0),__LC_LPP+1	# clear lpp and current_pid
 	mvi	__LC_LPP,0x80			#   and set LPP_MAGIC
 	.insn	s,0xb2800000,__LC_LPP		# load program parameter
-0:	larl	%r1,sched_clock_base_cc
-	mvc	0(8,%r1),__LC_LAST_UPDATE_CLOCK
+0:	larl	%r1,tod_clock_base
+	mvc	0(16,%r1),__LC_BOOT_CLOCK
 	larl	%r13,.LPG1		# get base
 	lctlg	%c0,%c15,.Lctl-.LPG1(%r13)	# load control registers
 	lg	%r12,.Lparmaddr-.LPG1(%r13)	# pointer to parameter area

arch/s390/kernel/irq.c

@@ -105,7 +105,8 @@ void do_IRQ(struct pt_regs *regs, int irq)
 
 	old_regs = set_irq_regs(regs);
 	irq_enter();
-	if (S390_lowcore.int_clock >= S390_lowcore.clock_comparator)
+	if (tod_after_eq(S390_lowcore.int_clock,
+			 S390_lowcore.clock_comparator))
 		/* Serve timer interrupts first. */
 		clock_comparator_work();
 	generic_handle_irq(irq);

arch/s390/kernel/relocate_kernel.S

@@ -7,6 +7,7 @@
  */
 
 #include <linux/linkage.h>
+#include <asm/page.h>
 #include <asm/sigp.h>
 
 /*
@@ -55,8 +56,8 @@ ENTRY(relocate_kernel)
 	.back_pgm:
 		lmg	%r0,%r15,gprregs-.base(%r13)
 	.top:
-		lghi	%r7,4096	# load PAGE_SIZE in r7
-		lghi	%r9,4096	# load PAGE_SIZE in r9
+		lghi	%r7,PAGE_SIZE	# load PAGE_SIZE in r7
+		lghi	%r9,PAGE_SIZE	# load PAGE_SIZE in r9
 		lg	%r5,0(%r2)	# read another word for indirection page
 		aghi	%r2,8		# increment pointer
 		tml	%r5,0x1		# is it a destination page?

arch/s390/kernel/setup.c

@@ -305,7 +305,7 @@ static void __init setup_lowcore(void)
 	/*
 	 * Setup lowcore for boot cpu
 	 */
-	BUILD_BUG_ON(sizeof(struct lowcore) != LC_PAGES * 4096);
+	BUILD_BUG_ON(sizeof(struct lowcore) != LC_PAGES * PAGE_SIZE);
 	lc = memblock_virt_alloc_low(sizeof(*lc), sizeof(*lc));
 	lc->restart_psw.mask = PSW_KERNEL_BITS;
 	lc->restart_psw.addr = (unsigned long) restart_int_handler;
@@ -323,7 +323,7 @@ static void __init setup_lowcore(void)
 	lc->io_new_psw.mask = PSW_KERNEL_BITS |
 		PSW_MASK_DAT | PSW_MASK_MCHECK;
 	lc->io_new_psw.addr = (unsigned long) io_int_handler;
-	lc->clock_comparator = -1ULL;
+	lc->clock_comparator = clock_comparator_max;
 	lc->kernel_stack = ((unsigned long) &init_thread_union)
 		+ THREAD_SIZE - STACK_FRAME_OVERHEAD - sizeof(struct pt_regs);
 	lc->async_stack = (unsigned long)
@@ -469,10 +469,10 @@ static void __init setup_memory_end(void)
 	vmalloc_size = VMALLOC_END ?: (128UL << 30) - MODULES_LEN;
 	tmp = (memory_end ?: max_physmem_end) / PAGE_SIZE;
 	tmp = tmp * (sizeof(struct page) + PAGE_SIZE);
-	if (tmp + vmalloc_size + MODULES_LEN <= (1UL << 42))
-		vmax = 1UL << 42;	/* 3-level kernel page table */
+	if (tmp + vmalloc_size + MODULES_LEN <= _REGION2_SIZE)
+		vmax = _REGION2_SIZE; /* 3-level kernel page table */
 	else
-		vmax = 1UL << 53;	/* 4-level kernel page table */
+		vmax = _REGION1_SIZE; /* 4-level kernel page table */
 	/* module area is at the end of the kernel address space. */
 	MODULES_END = vmax;
 	MODULES_VADDR = MODULES_END - MODULES_LEN;
@@ -818,6 +818,9 @@ static int __init setup_hwcaps(void)
 	case 0x2965:
 		strcpy(elf_platform, "z13");
 		break;
+	case 0x3906:
+		strcpy(elf_platform, "z14");
+		break;
 	}
 
 	/*
@@ -922,6 +925,7 @@ void __init setup_arch(char **cmdline_p)
 	setup_memory_end();
 	setup_memory();
 	dma_contiguous_reserve(memory_end);
+	vmcp_cma_reserve();
 
 	check_initrd();
 	reserve_crashkernel();

arch/s390/kernel/smp.c

@@ -1181,6 +1181,7 @@ static int __init s390_smp_init(void)
 
 	rc = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "s390/smp:online",
 			       smp_cpu_online, smp_cpu_pre_down);
+	rc = rc <= 0 ? rc : 0;
 out:
 	return rc;
 }

arch/s390/kernel/suspend.c

@@ -98,10 +98,16 @@ int page_key_alloc(unsigned long pages)
  */
 void page_key_read(unsigned long *pfn)
 {
+	struct page *page;
 	unsigned long addr;
-
-	addr = (unsigned long) page_address(pfn_to_page(*pfn));
-	*(unsigned char *) pfn = (unsigned char) page_get_storage_key(addr);
+	unsigned char key;
+
+	page = pfn_to_page(*pfn);
+	addr = (unsigned long) page_address(page);
+	key = (unsigned char) page_get_storage_key(addr) & 0x7f;
+	if (arch_test_page_nodat(page))
+		key |= 0x80;
+	*(unsigned char *) pfn = key;
 }
 
 /*
@@ -126,8 +132,16 @@ void page_key_memorize(unsigned long *pfn)
  */
 void page_key_write(void *address)
 {
-	page_set_storage_key((unsigned long) address,
-			     page_key_rp->data[page_key_rx], 0);
+	struct page *page;
+	unsigned char key;
+
+	key = page_key_rp->data[page_key_rx];
+	page_set_storage_key((unsigned long) address, key & 0x7f, 0);
+	page = virt_to_page(address);
+	if (key & 0x80)
+		arch_set_page_nodat(page, 0);
+	else
+		arch_set_page_dat(page, 0);
 	if (++page_key_rx >= PAGE_KEY_DATA_SIZE)
 		return;
 	page_key_rp = page_key_rp->next;

arch/s390/kernel/time.c

@@ -51,8 +51,15 @@
 #include <asm/cio.h>
 #include "entry.h"
 
-u64 sched_clock_base_cc = -1;	/* Force to data section. */
-EXPORT_SYMBOL_GPL(sched_clock_base_cc);
+unsigned char tod_clock_base[16] __aligned(8) = {
+	/* Force to data section. */
+	0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff,
+	0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff, 0xff
+};
+EXPORT_SYMBOL_GPL(tod_clock_base);
+
+u64 clock_comparator_max = -1ULL;
+EXPORT_SYMBOL_GPL(clock_comparator_max);
 
 static DEFINE_PER_CPU(struct clock_event_device, comparators);
 
@@ -75,7 +82,7 @@ void __init time_early_init(void)
 	struct ptff_qui qui;
 
 	/* Initialize TOD steering parameters */
-	tod_steering_end = sched_clock_base_cc;
+	tod_steering_end = *(unsigned long long *) &tod_clock_base[1];
 	vdso_data->ts_end = tod_steering_end;
 
 	if (!test_facility(28))
@@ -111,22 +118,27 @@ unsigned long long monotonic_clock(void)
 }
 EXPORT_SYMBOL(monotonic_clock);
 
-static void tod_to_timeval(__u64 todval, struct timespec64 *xt)
+static void ext_to_timespec64(unsigned char *clk, struct timespec64 *xt)
 {
-	unsigned long long sec;
+	unsigned long long high, low, rem, sec, nsec;
+
+	/* Split extendnd TOD clock to micro-seconds and sub-micro-seconds */
+	high = (*(unsigned long long *) clk) >> 4;
+	low = (*(unsigned long long *)&clk[7]) << 4;
+	/* Calculate seconds and nano-seconds */
+	sec = high;
+	rem = do_div(sec, 1000000);
+	nsec = (((low >> 32) + (rem << 32)) * 1000) >> 32;
 
-	sec = todval >> 12;
-	do_div(sec, 1000000);
 	xt->tv_sec = sec;
-	todval -= (sec * 1000000) << 12;
-	xt->tv_nsec = ((todval * 1000) >> 12);
+	xt->tv_nsec = nsec;
 }
 
 void clock_comparator_work(void)
 {
 	struct clock_event_device *cd;
 
-	S390_lowcore.clock_comparator = -1ULL;
+	S390_lowcore.clock_comparator = clock_comparator_max;
 	cd = this_cpu_ptr(&comparators);
 	cd->event_handler(cd);
 }
@@ -148,7 +160,7 @@ void init_cpu_timer(void)
 	struct clock_event_device *cd;
 	int cpu;
 
-	S390_lowcore.clock_comparator = -1ULL;
+	S390_lowcore.clock_comparator = clock_comparator_max;
 	set_clock_comparator(S390_lowcore.clock_comparator);
 
 	cpu = smp_processor_id();
@@ -179,7 +191,7 @@ static void clock_comparator_interrupt(struct ext_code ext_code,
 				       unsigned long param64)
 {
 	inc_irq_stat(IRQEXT_CLK);
-	if (S390_lowcore.clock_comparator == -1ULL)
+	if (S390_lowcore.clock_comparator == clock_comparator_max)
 		set_clock_comparator(S390_lowcore.clock_comparator);
 }
 
@@ -197,18 +209,28 @@ static void stp_reset(void);
 
 void read_persistent_clock64(struct timespec64 *ts)
 {
-	__u64 clock;
+	unsigned char clk[STORE_CLOCK_EXT_SIZE];
+	__u64 delta;
 
-	clock = get_tod_clock() - initial_leap_seconds;
-	tod_to_timeval(clock - TOD_UNIX_EPOCH, ts);
+	delta = initial_leap_seconds + TOD_UNIX_EPOCH;
+	get_tod_clock_ext(clk);
+	*(__u64 *) &clk[1] -= delta;
+	if (*(__u64 *) &clk[1] > delta)
+		clk[0]--;
+	ext_to_timespec64(clk, ts);
 }
 
 void read_boot_clock64(struct timespec64 *ts)
 {
-	__u64 clock;
+	unsigned char clk[STORE_CLOCK_EXT_SIZE];
+	__u64 delta;
 
-	clock = sched_clock_base_cc - initial_leap_seconds;
-	tod_to_timeval(clock - TOD_UNIX_EPOCH, ts);
+	delta = initial_leap_seconds + TOD_UNIX_EPOCH;
+	memcpy(clk, tod_clock_base, 16);
+	*(__u64 *) &clk[1] -= delta;
+	if (*(__u64 *) &clk[1] > delta)
+		clk[0]--;
+	ext_to_timespec64(clk, ts);
 }
 
 static u64 read_tod_clock(struct clocksource *cs)
@@ -335,7 +357,7 @@ static unsigned long clock_sync_flags;
  * source. If the clock mode is local it will return -EOPNOTSUPP and
  * -EAGAIN if the clock is not in sync with the external reference.
  */
-int get_phys_clock(unsigned long long *clock)
+int get_phys_clock(unsigned long *clock)
 {
 	atomic_t *sw_ptr;
 	unsigned int sw0, sw1;
@@ -406,7 +428,10 @@ static void clock_sync_global(unsigned long long delta)
 	struct ptff_qto qto;
 
 	/* Fixup the monotonic sched clock. */
-	sched_clock_base_cc += delta;
+	*(unsigned long long *) &tod_clock_base[1] += delta;
+	if (*(unsigned long long *) &tod_clock_base[1] < delta)
+		/* Epoch overflow */
+		tod_clock_base[0]++;
 	/* Adjust TOD steering parameters. */
 	vdso_data->tb_update_count++;
 	now = get_tod_clock();
@@ -437,7 +462,7 @@ static void clock_sync_global(unsigned long long delta)
 static void clock_sync_local(unsigned long long delta)
 {
 	/* Add the delta to the clock comparator. */
-	if (S390_lowcore.clock_comparator != -1ULL) {
+	if (S390_lowcore.clock_comparator != clock_comparator_max) {
 		S390_lowcore.clock_comparator += delta;
 		set_clock_comparator(S390_lowcore.clock_comparator);
 	}

arch/s390/kernel/vdso.c

@@ -157,6 +157,8 @@ int vdso_alloc_per_cpu(struct lowcore *lowcore)
 	page_frame = get_zeroed_page(GFP_KERNEL);
 	if (!segment_table || !page_table || !page_frame)
 		goto out;
+	arch_set_page_dat(virt_to_page(segment_table), SEGMENT_ORDER);
+	arch_set_page_dat(virt_to_page(page_table), 0);
 
 	/* Initialize per-cpu vdso data page */
 	vd = (struct vdso_per_cpu_data *) page_frame;

arch/s390/kernel/vdso32/vdso32.lds.S

@@ -2,6 +2,8 @@
  * This is the infamous ld script for the 32 bits vdso
  * library
  */
+
+#include <asm/page.h>
 #include <asm/vdso.h>
 
 OUTPUT_FORMAT("elf32-s390", "elf32-s390", "elf32-s390")
@@ -91,7 +93,7 @@ SECTIONS
 	.debug_ranges	0 : { *(.debug_ranges) }
 	.gnu.attributes 0 : { KEEP (*(.gnu.attributes)) }
 
-	. = ALIGN(4096);
+	. = ALIGN(PAGE_SIZE);
 	PROVIDE(_vdso_data = .);
 
 	/DISCARD/	: {

arch/s390/kernel/vdso64/vdso64.lds.S

@@ -2,6 +2,8 @@
  * This is the infamous ld script for the 64 bits vdso
  * library
  */
+
+#include <asm/page.h>
 #include <asm/vdso.h>
 
 OUTPUT_FORMAT("elf64-s390", "elf64-s390", "elf64-s390")
@@ -91,7 +93,7 @@ SECTIONS
 	.debug_ranges	0 : { *(.debug_ranges) }
 	.gnu.attributes 0 : { KEEP (*(.gnu.attributes)) }
 
-	. = ALIGN(4096);
+	. = ALIGN(PAGE_SIZE);
 	PROVIDE(_vdso_data = .);
 
 	/DISCARD/	: {

arch/s390/kvm/diag.c

@@ -27,7 +27,7 @@ static int diag_release_pages(struct kvm_vcpu *vcpu)
 	unsigned long prefix  = kvm_s390_get_prefix(vcpu);
 
 	start = vcpu->run->s.regs.gprs[(vcpu->arch.sie_block->ipa & 0xf0) >> 4];
-	end = vcpu->run->s.regs.gprs[vcpu->arch.sie_block->ipa & 0xf] + 4096;
+	end = vcpu->run->s.regs.gprs[vcpu->arch.sie_block->ipa & 0xf] + PAGE_SIZE;
 	vcpu->stat.diagnose_10++;
 
 	if (start & ~PAGE_MASK || end & ~PAGE_MASK || start >= end
@@ -51,9 +51,9 @@ static int diag_release_pages(struct kvm_vcpu *vcpu)
 		 */
 		gmap_discard(vcpu->arch.gmap, start, prefix);
 		if (start <= prefix)
-			gmap_discard(vcpu->arch.gmap, 0, 4096);
-		if (end > prefix + 4096)
-			gmap_discard(vcpu->arch.gmap, 4096, 8192);
+			gmap_discard(vcpu->arch.gmap, 0, PAGE_SIZE);
+		if (end > prefix + PAGE_SIZE)
+			gmap_discard(vcpu->arch.gmap, PAGE_SIZE, 2 * PAGE_SIZE);
 		gmap_discard(vcpu->arch.gmap, prefix + 2 * PAGE_SIZE, end);
 	}
 	return 0;

arch/s390/kvm/gaccess.c

@@ -629,7 +629,7 @@ static unsigned long guest_translate(struct kvm_vcpu *vcpu, unsigned long gva,
 	iep = ctlreg0.iep && test_kvm_facility(vcpu->kvm, 130);
 	if (asce.r)
 		goto real_address;
-	ptr = asce.origin * 4096;
+	ptr = asce.origin * PAGE_SIZE;
 	switch (asce.dt) {
 	case ASCE_TYPE_REGION1:
 		if (vaddr.rfx01 > asce.tl)
@@ -674,7 +674,7 @@ static unsigned long guest_translate(struct kvm_vcpu *vcpu, unsigned long gva,
 			return PGM_REGION_SECOND_TRANS;
 		if (edat1)
 			dat_protection |= rfte.p;
-		ptr = rfte.rto * 4096 + vaddr.rsx * 8;
+		ptr = rfte.rto * PAGE_SIZE + vaddr.rsx * 8;
 	}
 		/* fallthrough */
 	case ASCE_TYPE_REGION2: {
@@ -692,7 +692,7 @@ static unsigned long guest_translate(struct kvm_vcpu *vcpu, unsigned long gva,
 			return PGM_REGION_THIRD_TRANS;
 		if (edat1)
 			dat_protection |= rste.p;
-		ptr = rste.rto * 4096 + vaddr.rtx * 8;
+		ptr = rste.rto * PAGE_SIZE + vaddr.rtx * 8;
 	}
 		/* fallthrough */
 	case ASCE_TYPE_REGION3: {
@@ -720,7 +720,7 @@ static unsigned long guest_translate(struct kvm_vcpu *vcpu, unsigned long gva,
 			return PGM_SEGMENT_TRANSLATION;
 		if (edat1)
 			dat_protection |= rtte.fc0.p;
-		ptr = rtte.fc0.sto * 4096 + vaddr.sx * 8;
+		ptr = rtte.fc0.sto * PAGE_SIZE + vaddr.sx * 8;
 	}
 		/* fallthrough */
 	case ASCE_TYPE_SEGMENT: {
@@ -743,7 +743,7 @@ static unsigned long guest_translate(struct kvm_vcpu *vcpu, unsigned long gva,
 			goto absolute_address;
 		}
 		dat_protection |= ste.fc0.p;
-		ptr = ste.fc0.pto * 2048 + vaddr.px * 8;
+		ptr = ste.fc0.pto * (PAGE_SIZE / 2) + vaddr.px * 8;
 	}
 	}
 	if (kvm_is_error_gpa(vcpu->kvm, ptr))
@@ -993,7 +993,7 @@ static int kvm_s390_shadow_tables(struct gmap *sg, unsigned long saddr,
 	parent = sg->parent;
 	vaddr.addr = saddr;
 	asce.val = sg->orig_asce;
-	ptr = asce.origin * 4096;
+	ptr = asce.origin * PAGE_SIZE;
 	if (asce.r) {
 		*fake = 1;
 		ptr = 0;
@@ -1029,7 +1029,7 @@ static int kvm_s390_shadow_tables(struct gmap *sg, unsigned long saddr,
 		union region1_table_entry rfte;
 
 		if (*fake) {
-			ptr += (unsigned long) vaddr.rfx << 53;
+			ptr += vaddr.rfx * _REGION1_SIZE;
 			rfte.val = ptr;
 			goto shadow_r2t;
 		}
@@ -1044,7 +1044,7 @@ static int kvm_s390_shadow_tables(struct gmap *sg, unsigned long saddr,
 			return PGM_REGION_SECOND_TRANS;
 		if (sg->edat_level >= 1)
 			*dat_protection |= rfte.p;
-		ptr = rfte.rto << 12UL;
+		ptr = rfte.rto * PAGE_SIZE;
 shadow_r2t:
 		rc = gmap_shadow_r2t(sg, saddr, rfte.val, *fake);
 		if (rc)
@@ -1055,7 +1055,7 @@ shadow_r2t:
 		union region2_table_entry rste;
 
 		if (*fake) {
-			ptr += (unsigned long) vaddr.rsx << 42;
+			ptr += vaddr.rsx * _REGION2_SIZE;
 			rste.val = ptr;
 			goto shadow_r3t;
 		}
@@ -1070,7 +1070,7 @@ shadow_r2t:
 			return PGM_REGION_THIRD_TRANS;
 		if (sg->edat_level >= 1)
 			*dat_protection |= rste.p;
-		ptr = rste.rto << 12UL;
+		ptr = rste.rto * PAGE_SIZE;
 shadow_r3t:
 		rste.p |= *dat_protection;
 		rc = gmap_shadow_r3t(sg, saddr, rste.val, *fake);
@@ -1082,7 +1082,7 @@ shadow_r3t:
 		union region3_table_entry rtte;
 
 		if (*fake) {
-			ptr += (unsigned long) vaddr.rtx << 31;
+			ptr += vaddr.rtx * _REGION3_SIZE;
 			rtte.val = ptr;
 			goto shadow_sgt;
 		}
@@ -1098,7 +1098,7 @@ shadow_r3t:
 		if (rtte.fc && sg->edat_level >= 2) {
 			*dat_protection |= rtte.fc0.p;
 			*fake = 1;
-			ptr = rtte.fc1.rfaa << 31UL;
+			ptr = rtte.fc1.rfaa * _REGION3_SIZE;
 			rtte.val = ptr;
 			goto shadow_sgt;
 		}
@@ -1106,7 +1106,7 @@ shadow_r3t:
 			return PGM_SEGMENT_TRANSLATION;
 		if (sg->edat_level >= 1)
 			*dat_protection |= rtte.fc0.p;
-		ptr = rtte.fc0.sto << 12UL;
+		ptr = rtte.fc0.sto * PAGE_SIZE;
 shadow_sgt:
 		rtte.fc0.p |= *dat_protection;
 		rc = gmap_shadow_sgt(sg, saddr, rtte.val, *fake);
@@ -1118,7 +1118,7 @@ shadow_sgt:
 		union segment_table_entry ste;
 
 		if (*fake) {
-			ptr += (unsigned long) vaddr.sx << 20;
+			ptr += vaddr.sx * _SEGMENT_SIZE;
 			ste.val = ptr;
 			goto shadow_pgt;
 		}
@@ -1134,11 +1134,11 @@ shadow_sgt:
 		*dat_protection |= ste.fc0.p;
 		if (ste.fc && sg->edat_level >= 1) {
 			*fake = 1;
-			ptr = ste.fc1.sfaa << 20UL;
+			ptr = ste.fc1.sfaa * _SEGMENT_SIZE;
 			ste.val = ptr;
 			goto shadow_pgt;
 		}
-		ptr = ste.fc0.pto << 11UL;
+		ptr = ste.fc0.pto * (PAGE_SIZE / 2);
 shadow_pgt:
 		ste.fc0.p |= *dat_protection;
 		rc = gmap_shadow_pgt(sg, saddr, ste.val, *fake);
@@ -1187,8 +1187,7 @@ int kvm_s390_shadow_fault(struct kvm_vcpu *vcpu, struct gmap *sg,
 
 	vaddr.addr = saddr;
 	if (fake) {
-		/* offset in 1MB guest memory block */
-		pte.val = pgt + ((unsigned long) vaddr.px << 12UL);
+		pte.val = pgt + vaddr.px * PAGE_SIZE;
 		goto shadow_page;
 	}
 	if (!rc)

arch/s390/kvm/priv.c

@@ -329,7 +329,7 @@ static int handle_sske(struct kvm_vcpu *vcpu)
 	start = kvm_s390_logical_to_effective(vcpu, start);
 	if (m3 & SSKE_MB) {
 		/* start already designates an absolute address */
-		end = (start + (1UL << 20)) & ~((1UL << 20) - 1);
+		end = (start + _SEGMENT_SIZE) & ~(_SEGMENT_SIZE - 1);
 	} else {
 		start = kvm_s390_real_to_abs(vcpu, start);
 		end = start + PAGE_SIZE;
@@ -893,10 +893,10 @@ static int handle_pfmf(struct kvm_vcpu *vcpu)
 	case 0x00000000:
 		/* only 4k frames specify a real address */
 		start = kvm_s390_real_to_abs(vcpu, start);
-		end = (start + (1UL << 12)) & ~((1UL << 12) - 1);
+		end = (start + PAGE_SIZE) & ~(PAGE_SIZE - 1);
 		break;
 	case 0x00001000:
-		end = (start + (1UL << 20)) & ~((1UL << 20) - 1);
+		end = (start + _SEGMENT_SIZE) & ~(_SEGMENT_SIZE - 1);
 		break;
 	case 0x00002000:
 		/* only support 2G frame size if EDAT2 is available and we are
@@ -904,7 +904,7 @@ static int handle_pfmf(struct kvm_vcpu *vcpu)
 		if (!test_kvm_facility(vcpu->kvm, 78) ||
 		    psw_bits(vcpu->arch.sie_block->gpsw).eaba == PSW_BITS_AMODE_24BIT)
 			return kvm_s390_inject_program_int(vcpu, PGM_SPECIFICATION);
-		end = (start + (1UL << 31)) & ~((1UL << 31) - 1);
+		end = (start + _REGION3_SIZE) & ~(_REGION3_SIZE - 1);
 		break;
 	default:
 		return kvm_s390_inject_program_int(vcpu, PGM_SPECIFICATION);

arch/s390/kvm/vsie.c

@@ -1069,7 +1069,7 @@ int kvm_s390_handle_vsie(struct kvm_vcpu *vcpu)
 	if (vcpu->arch.sie_block->gpsw.mask & PSW_MASK_PSTATE)
 		return kvm_s390_inject_program_int(vcpu, PGM_PRIVILEGED_OP);
 
-	BUILD_BUG_ON(sizeof(struct vsie_page) != 4096);
+	BUILD_BUG_ON(sizeof(struct vsie_page) != PAGE_SIZE);
 	scb_addr = kvm_s390_get_base_disp_s(vcpu, NULL);
 
 	/* 512 byte alignment */

arch/s390/lib/delay.c

@@ -57,7 +57,7 @@ static void __udelay_enabled(unsigned long long usecs)
 	end = get_tod_clock_fast() + (usecs << 12);
 	do {
 		clock_saved = 0;
-		if (end < S390_lowcore.clock_comparator) {
+		if (tod_after(S390_lowcore.clock_comparator, end)) {
 			clock_saved = local_tick_disable();
 			set_clock_comparator(end);
 		}

arch/s390/lib/spinlock.c

@@ -32,42 +32,63 @@ static int __init spin_retry_setup(char *str)
 }
 __setup("spin_retry=", spin_retry_setup);
 
+static inline int arch_load_niai4(int *lock)
+{
+	int owner;
+
+	asm volatile(
+#ifdef CONFIG_HAVE_MARCH_ZEC12_FEATURES
+		"	.long	0xb2fa0040\n"	/* NIAI 4 */
+#endif
+		"	l	%0,%1\n"
+		: "=d" (owner) : "Q" (*lock) : "memory");
+       return owner;
+}
+
+static inline int arch_cmpxchg_niai8(int *lock, int old, int new)
+{
+	int expected = old;
+
+	asm volatile(
+#ifdef CONFIG_HAVE_MARCH_ZEC12_FEATURES
+		"	.long	0xb2fa0080\n"	/* NIAI 8 */
+#endif
+		"	cs	%0,%3,%1\n"
+		: "=d" (old), "=Q" (*lock)
+		: "0" (old), "d" (new), "Q" (*lock)
+		: "cc", "memory");
+	return expected == old;
+}
+
 void arch_spin_lock_wait(arch_spinlock_t *lp)
 {
 	int cpu = SPINLOCK_LOCKVAL;
-	int owner, count, first_diag;
+	int owner, count;
+
+	/* Pass the virtual CPU to the lock holder if it is not running */
+	owner = arch_load_niai4(&lp->lock);
+	if (owner && arch_vcpu_is_preempted(~owner))
+		smp_yield_cpu(~owner);
 
-	first_diag = 1;
+	count = spin_retry;
 	while (1) {
-		owner = ACCESS_ONCE(lp->lock);
+		owner = arch_load_niai4(&lp->lock);
 		/* Try to get the lock if it is free. */
 		if (!owner) {
-			if (__atomic_cmpxchg_bool(&lp->lock, 0, cpu))
+			if (arch_cmpxchg_niai8(&lp->lock, 0, cpu))
 				return;
 			continue;
 		}
-		/* First iteration: check if the lock owner is running. */
-		if (first_diag && arch_vcpu_is_preempted(~owner)) {
-			smp_yield_cpu(~owner);
-			first_diag = 0;
+		if (count-- >= 0)
 			continue;
-		}
-		/* Loop for a while on the lock value. */
 		count = spin_retry;
-		do {
-			owner = ACCESS_ONCE(lp->lock);
-		} while (owner && count-- > 0);
-		if (!owner)
-			continue;
 		/*
 		 * For multiple layers of hypervisors, e.g. z/VM + LPAR
 		 * yield the CPU unconditionally. For LPAR rely on the
 		 * sense running status.
 		 */
-		if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(~owner)) {
+		if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(~owner))
 			smp_yield_cpu(~owner);
-			first_diag = 0;
-		}
 	}
 }
 EXPORT_SYMBOL(arch_spin_lock_wait);
@@ -75,42 +96,36 @@ EXPORT_SYMBOL(arch_spin_lock_wait);
 void arch_spin_lock_wait_flags(arch_spinlock_t *lp, unsigned long flags)
 {
 	int cpu = SPINLOCK_LOCKVAL;
-	int owner, count, first_diag;
+	int owner, count;
 
 	local_irq_restore(flags);
-	first_diag = 1;
+
+	/* Pass the virtual CPU to the lock holder if it is not running */
+	owner = arch_load_niai4(&lp->lock);
+	if (owner && arch_vcpu_is_preempted(~owner))
+		smp_yield_cpu(~owner);
+
+	count = spin_retry;
 	while (1) {
-		owner = ACCESS_ONCE(lp->lock);
+		owner = arch_load_niai4(&lp->lock);
 		/* Try to get the lock if it is free. */
 		if (!owner) {
 			local_irq_disable();
-			if (__atomic_cmpxchg_bool(&lp->lock, 0, cpu))
+			if (arch_cmpxchg_niai8(&lp->lock, 0, cpu))
 				return;
 			local_irq_restore(flags);
 			continue;
 		}
-		/* Check if the lock owner is running. */
-		if (first_diag && arch_vcpu_is_preempted(~owner)) {
-			smp_yield_cpu(~owner);
-			first_diag = 0;
+		if (count-- >= 0)
 			continue;
-		}
-		/* Loop for a while on the lock value. */
 		count = spin_retry;
-		do {
-			owner = ACCESS_ONCE(lp->lock);
-		} while (owner && count-- > 0);
-		if (!owner)
-			continue;
 		/*
 		 * For multiple layers of hypervisors, e.g. z/VM + LPAR
 		 * yield the CPU unconditionally. For LPAR rely on the
 		 * sense running status.
 		 */
-		if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(~owner)) {
+		if (!MACHINE_IS_LPAR || arch_vcpu_is_preempted(~owner))
 			smp_yield_cpu(~owner);
-			first_diag = 0;
-		}
 	}
 }
 EXPORT_SYMBOL(arch_spin_lock_wait_flags);

arch/s390/lib/uaccess.c

@@ -15,8 +15,30 @@
 #include <asm/mmu_context.h>
 #include <asm/facility.h>
 
+#ifndef CONFIG_HAVE_MARCH_Z10_FEATURES
 static DEFINE_STATIC_KEY_FALSE(have_mvcos);
 
+static int __init uaccess_init(void)
+{
+	if (test_facility(27))
+		static_branch_enable(&have_mvcos);
+	return 0;
+}
+early_initcall(uaccess_init);
+
+static inline int copy_with_mvcos(void)
+{
+	if (static_branch_likely(&have_mvcos))
+		return 1;
+	return 0;
+}
+#else
+static inline int copy_with_mvcos(void)
+{
+	return 1;
+}
+#endif
+
 static inline unsigned long copy_from_user_mvcos(void *x, const void __user *ptr,
 						 unsigned long size)
 {
@@ -84,7 +106,7 @@ static inline unsigned long copy_from_user_mvcp(void *x, const void __user *ptr,
 
 unsigned long raw_copy_from_user(void *to, const void __user *from, unsigned long n)
 {
-	if (static_branch_likely(&have_mvcos))
+	if (copy_with_mvcos())
 		return copy_from_user_mvcos(to, from, n);
 	return copy_from_user_mvcp(to, from, n);
 }
@@ -157,7 +179,7 @@ static inline unsigned long copy_to_user_mvcs(void __user *ptr, const void *x,
 
 unsigned long raw_copy_to_user(void __user *to, const void *from, unsigned long n)
 {
-	if (static_branch_likely(&have_mvcos))
+	if (copy_with_mvcos())
 		return copy_to_user_mvcos(to, from, n);
 	return copy_to_user_mvcs(to, from, n);
 }
@@ -220,7 +242,7 @@ static inline unsigned long copy_in_user_mvc(void __user *to, const void __user
 
 unsigned long raw_copy_in_user(void __user *to, const void __user *from, unsigned long n)
 {
-	if (static_branch_likely(&have_mvcos))
+	if (copy_with_mvcos())
 		return copy_in_user_mvcos(to, from, n);
 	return copy_in_user_mvc(to, from, n);
 }
@@ -292,7 +314,7 @@ static inline unsigned long clear_user_xc(void __user *to, unsigned long size)
 
 unsigned long __clear_user(void __user *to, unsigned long size)
 {
-	if (static_branch_likely(&have_mvcos))
+	if (copy_with_mvcos())
 			return clear_user_mvcos(to, size);
 	return clear_user_xc(to, size);
 }
@@ -349,11 +371,3 @@ long __strncpy_from_user(char *dst, const char __user *src, long size)
 	return done;
 }
 EXPORT_SYMBOL(__strncpy_from_user);
-
-static int __init uaccess_init(void)
-{
-	if (test_facility(27))
-		static_branch_enable(&have_mvcos);
-	return 0;
-}
-early_initcall(uaccess_init);

arch/s390/mm/fault.c

@@ -135,7 +135,7 @@ static void dump_pagetable(unsigned long asce, unsigned long address)
 	pr_alert("AS:%016lx ", asce);
 	switch (asce & _ASCE_TYPE_MASK) {
 	case _ASCE_TYPE_REGION1:
-		table = table + ((address >> 53) & 0x7ff);
+		table += (address & _REGION1_INDEX) >> _REGION1_SHIFT;
 		if (bad_address(table))
 			goto bad;
 		pr_cont("R1:%016lx ", *table);
@@ -144,7 +144,7 @@ static void dump_pagetable(unsigned long asce, unsigned long address)
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* fallthrough */
 	case _ASCE_TYPE_REGION2:
-		table = table + ((address >> 42) & 0x7ff);
+		table += (address & _REGION2_INDEX) >> _REGION2_SHIFT;
 		if (bad_address(table))
 			goto bad;
 		pr_cont("R2:%016lx ", *table);
@@ -153,7 +153,7 @@ static void dump_pagetable(unsigned long asce, unsigned long address)
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* fallthrough */
 	case _ASCE_TYPE_REGION3:
-		table = table + ((address >> 31) & 0x7ff);
+		table += (address & _REGION3_INDEX) >> _REGION3_SHIFT;
 		if (bad_address(table))
 			goto bad;
 		pr_cont("R3:%016lx ", *table);
@@ -162,7 +162,7 @@ static void dump_pagetable(unsigned long asce, unsigned long address)
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* fallthrough */
 	case _ASCE_TYPE_SEGMENT:
-		table = table + ((address >> 20) & 0x7ff);
+		table += (address & _SEGMENT_INDEX) >> _SEGMENT_SHIFT;
 		if (bad_address(table))
 			goto bad;
 		pr_cont("S:%016lx ", *table);
@@ -170,7 +170,7 @@ static void dump_pagetable(unsigned long asce, unsigned long address)
 			goto out;
 		table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
 	}
-	table = table + ((address >> 12) & 0xff);
+	table += (address & _PAGE_INDEX) >> _PAGE_SHIFT;
 	if (bad_address(table))
 		goto bad;
 	pr_cont("P:%016lx ", *table);

arch/s390/mm/gmap.c

@@ -36,16 +36,16 @@ static struct gmap *gmap_alloc(unsigned long limit)
 	unsigned long *table;
 	unsigned long etype, atype;
 
-	if (limit < (1UL << 31)) {
-		limit = (1UL << 31) - 1;
+	if (limit < _REGION3_SIZE) {
+		limit = _REGION3_SIZE - 1;
 		atype = _ASCE_TYPE_SEGMENT;
 		etype = _SEGMENT_ENTRY_EMPTY;
-	} else if (limit < (1UL << 42)) {
-		limit = (1UL << 42) - 1;
+	} else if (limit < _REGION2_SIZE) {
+		limit = _REGION2_SIZE - 1;
 		atype = _ASCE_TYPE_REGION3;
 		etype = _REGION3_ENTRY_EMPTY;
-	} else if (limit < (1UL << 53)) {
-		limit = (1UL << 53) - 1;
+	} else if (limit < _REGION1_SIZE) {
+		limit = _REGION1_SIZE - 1;
 		atype = _ASCE_TYPE_REGION2;
 		etype = _REGION2_ENTRY_EMPTY;
 	} else {
@@ -65,7 +65,7 @@ static struct gmap *gmap_alloc(unsigned long limit)
 	spin_lock_init(&gmap->guest_table_lock);
 	spin_lock_init(&gmap->shadow_lock);
 	atomic_set(&gmap->ref_count, 1);
-	page = alloc_pages(GFP_KERNEL, 2);
+	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
 	if (!page)
 		goto out_free;
 	page->index = 0;
@@ -186,7 +186,7 @@ static void gmap_free(struct gmap *gmap)
 		gmap_flush_tlb(gmap);
 	/* Free all segment & region tables. */
 	list_for_each_entry_safe(page, next, &gmap->crst_list, lru)
-		__free_pages(page, 2);
+		__free_pages(page, CRST_ALLOC_ORDER);
 	gmap_radix_tree_free(&gmap->guest_to_host);
 	gmap_radix_tree_free(&gmap->host_to_guest);
 
@@ -306,7 +306,7 @@ static int gmap_alloc_table(struct gmap *gmap, unsigned long *table,
 	unsigned long *new;
 
 	/* since we dont free the gmap table until gmap_free we can unlock */
-	page = alloc_pages(GFP_KERNEL, 2);
+	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
 	if (!page)
 		return -ENOMEM;
 	new = (unsigned long *) page_to_phys(page);
@@ -321,7 +321,7 @@ static int gmap_alloc_table(struct gmap *gmap, unsigned long *table,
 	}
 	spin_unlock(&gmap->guest_table_lock);
 	if (page)
-		__free_pages(page, 2);
+		__free_pages(page, CRST_ALLOC_ORDER);
 	return 0;
 }
 
@@ -546,30 +546,30 @@ int __gmap_link(struct gmap *gmap, unsigned long gaddr, unsigned long vmaddr)
 	/* Create higher level tables in the gmap page table */
 	table = gmap->table;
 	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION1) {
-		table += (gaddr >> 53) & 0x7ff;
+		table += (gaddr & _REGION1_INDEX) >> _REGION1_SHIFT;
 		if ((*table & _REGION_ENTRY_INVALID) &&
 		    gmap_alloc_table(gmap, table, _REGION2_ENTRY_EMPTY,
-				     gaddr & 0xffe0000000000000UL))
+				     gaddr & _REGION1_MASK))
 			return -ENOMEM;
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 	}
 	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION2) {
-		table += (gaddr >> 42) & 0x7ff;
+		table += (gaddr & _REGION2_INDEX) >> _REGION2_SHIFT;
 		if ((*table & _REGION_ENTRY_INVALID) &&
 		    gmap_alloc_table(gmap, table, _REGION3_ENTRY_EMPTY,
-				     gaddr & 0xfffffc0000000000UL))
+				     gaddr & _REGION2_MASK))
 			return -ENOMEM;
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 	}
 	if ((gmap->asce & _ASCE_TYPE_MASK) >= _ASCE_TYPE_REGION3) {
-		table += (gaddr >> 31) & 0x7ff;
+		table += (gaddr & _REGION3_INDEX) >> _REGION3_SHIFT;
 		if ((*table & _REGION_ENTRY_INVALID) &&
 		    gmap_alloc_table(gmap, table, _SEGMENT_ENTRY_EMPTY,
-				     gaddr & 0xffffffff80000000UL))
+				     gaddr & _REGION3_MASK))
 			return -ENOMEM;
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 	}
-	table += (gaddr >> 20) & 0x7ff;
+	table += (gaddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT;
 	/* Walk the parent mm page table */
 	mm = gmap->mm;
 	pgd = pgd_offset(mm, vmaddr);
@@ -771,7 +771,7 @@ static inline unsigned long *gmap_table_walk(struct gmap *gmap,
 	table = gmap->table;
 	switch (gmap->asce & _ASCE_TYPE_MASK) {
 	case _ASCE_TYPE_REGION1:
-		table += (gaddr >> 53) & 0x7ff;
+		table += (gaddr & _REGION1_INDEX) >> _REGION1_SHIFT;
 		if (level == 4)
 			break;
 		if (*table & _REGION_ENTRY_INVALID)
@@ -779,7 +779,7 @@ static inline unsigned long *gmap_table_walk(struct gmap *gmap,
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* Fallthrough */
 	case _ASCE_TYPE_REGION2:
-		table += (gaddr >> 42) & 0x7ff;
+		table += (gaddr & _REGION2_INDEX) >> _REGION2_SHIFT;
 		if (level == 3)
 			break;
 		if (*table & _REGION_ENTRY_INVALID)
@@ -787,7 +787,7 @@ static inline unsigned long *gmap_table_walk(struct gmap *gmap,
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* Fallthrough */
 	case _ASCE_TYPE_REGION3:
-		table += (gaddr >> 31) & 0x7ff;
+		table += (gaddr & _REGION3_INDEX) >> _REGION3_SHIFT;
 		if (level == 2)
 			break;
 		if (*table & _REGION_ENTRY_INVALID)
@@ -795,13 +795,13 @@ static inline unsigned long *gmap_table_walk(struct gmap *gmap,
 		table = (unsigned long *)(*table & _REGION_ENTRY_ORIGIN);
 		/* Fallthrough */
 	case _ASCE_TYPE_SEGMENT:
-		table += (gaddr >> 20) & 0x7ff;
+		table += (gaddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT;
 		if (level == 1)
 			break;
 		if (*table & _REGION_ENTRY_INVALID)
 			return NULL;
 		table = (unsigned long *)(*table & _SEGMENT_ENTRY_ORIGIN);
-		table += (gaddr >> 12) & 0xff;
+		table += (gaddr & _PAGE_INDEX) >> _PAGE_SHIFT;
 	}
 	return table;
 }
@@ -1126,7 +1126,7 @@ static void gmap_unshadow_page(struct gmap *sg, unsigned long raddr)
 	table = gmap_table_walk(sg, raddr, 0); /* get page table pointer */
 	if (!table || *table & _PAGE_INVALID)
 		return;
-	gmap_call_notifier(sg, raddr, raddr + (1UL << 12) - 1);
+	gmap_call_notifier(sg, raddr, raddr + _PAGE_SIZE - 1);
 	ptep_unshadow_pte(sg->mm, raddr, (pte_t *) table);
 }
 
@@ -1144,7 +1144,7 @@ static void __gmap_unshadow_pgt(struct gmap *sg, unsigned long raddr,
 	int i;
 
 	BUG_ON(!gmap_is_shadow(sg));
-	for (i = 0; i < 256; i++, raddr += 1UL << 12)
+	for (i = 0; i < _PAGE_ENTRIES; i++, raddr += _PAGE_SIZE)
 		pgt[i] = _PAGE_INVALID;
 }
 
@@ -1164,8 +1164,8 @@ static void gmap_unshadow_pgt(struct gmap *sg, unsigned long raddr)
 	ste = gmap_table_walk(sg, raddr, 1); /* get segment pointer */
 	if (!ste || !(*ste & _SEGMENT_ENTRY_ORIGIN))
 		return;
-	gmap_call_notifier(sg, raddr, raddr + (1UL << 20) - 1);
-	sto = (unsigned long) (ste - ((raddr >> 20) & 0x7ff));
+	gmap_call_notifier(sg, raddr, raddr + _SEGMENT_SIZE - 1);
+	sto = (unsigned long) (ste - ((raddr & _SEGMENT_INDEX) >> _SEGMENT_SHIFT));
 	gmap_idte_one(sto | _ASCE_TYPE_SEGMENT, raddr);
 	pgt = (unsigned long *)(*ste & _SEGMENT_ENTRY_ORIGIN);
 	*ste = _SEGMENT_ENTRY_EMPTY;
@@ -1193,7 +1193,7 @@ static void __gmap_unshadow_sgt(struct gmap *sg, unsigned long raddr,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	asce = (unsigned long) sgt | _ASCE_TYPE_SEGMENT;
-	for (i = 0; i < 2048; i++, raddr += 1UL << 20) {
+	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _SEGMENT_SIZE) {
 		if (!(sgt[i] & _SEGMENT_ENTRY_ORIGIN))
 			continue;
 		pgt = (unsigned long *)(sgt[i] & _REGION_ENTRY_ORIGIN);
@@ -1222,8 +1222,8 @@ static void gmap_unshadow_sgt(struct gmap *sg, unsigned long raddr)
 	r3e = gmap_table_walk(sg, raddr, 2); /* get region-3 pointer */
 	if (!r3e || !(*r3e & _REGION_ENTRY_ORIGIN))
 		return;
-	gmap_call_notifier(sg, raddr, raddr + (1UL << 31) - 1);
-	r3o = (unsigned long) (r3e - ((raddr >> 31) & 0x7ff));
+	gmap_call_notifier(sg, raddr, raddr + _REGION3_SIZE - 1);
+	r3o = (unsigned long) (r3e - ((raddr & _REGION3_INDEX) >> _REGION3_SHIFT));
 	gmap_idte_one(r3o | _ASCE_TYPE_REGION3, raddr);
 	sgt = (unsigned long *)(*r3e & _REGION_ENTRY_ORIGIN);
 	*r3e = _REGION3_ENTRY_EMPTY;
@@ -1231,7 +1231,7 @@ static void gmap_unshadow_sgt(struct gmap *sg, unsigned long raddr)
 	/* Free segment table */
 	page = pfn_to_page(__pa(sgt) >> PAGE_SHIFT);
 	list_del(&page->lru);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 }
 
 /**
@@ -1251,7 +1251,7 @@ static void __gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	asce = (unsigned long) r3t | _ASCE_TYPE_REGION3;
-	for (i = 0; i < 2048; i++, raddr += 1UL << 31) {
+	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION3_SIZE) {
 		if (!(r3t[i] & _REGION_ENTRY_ORIGIN))
 			continue;
 		sgt = (unsigned long *)(r3t[i] & _REGION_ENTRY_ORIGIN);
@@ -1260,7 +1260,7 @@ static void __gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr,
 		/* Free segment table */
 		page = pfn_to_page(__pa(sgt) >> PAGE_SHIFT);
 		list_del(&page->lru);
-		__free_pages(page, 2);
+		__free_pages(page, CRST_ALLOC_ORDER);
 	}
 }
 
@@ -1280,8 +1280,8 @@ static void gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr)
 	r2e = gmap_table_walk(sg, raddr, 3); /* get region-2 pointer */
 	if (!r2e || !(*r2e & _REGION_ENTRY_ORIGIN))
 		return;
-	gmap_call_notifier(sg, raddr, raddr + (1UL << 42) - 1);
-	r2o = (unsigned long) (r2e - ((raddr >> 42) & 0x7ff));
+	gmap_call_notifier(sg, raddr, raddr + _REGION2_SIZE - 1);
+	r2o = (unsigned long) (r2e - ((raddr & _REGION2_INDEX) >> _REGION2_SHIFT));
 	gmap_idte_one(r2o | _ASCE_TYPE_REGION2, raddr);
 	r3t = (unsigned long *)(*r2e & _REGION_ENTRY_ORIGIN);
 	*r2e = _REGION2_ENTRY_EMPTY;
@@ -1289,7 +1289,7 @@ static void gmap_unshadow_r3t(struct gmap *sg, unsigned long raddr)
 	/* Free region 3 table */
 	page = pfn_to_page(__pa(r3t) >> PAGE_SHIFT);
 	list_del(&page->lru);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 }
 
 /**
@@ -1309,7 +1309,7 @@ static void __gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	asce = (unsigned long) r2t | _ASCE_TYPE_REGION2;
-	for (i = 0; i < 2048; i++, raddr += 1UL << 42) {
+	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION2_SIZE) {
 		if (!(r2t[i] & _REGION_ENTRY_ORIGIN))
 			continue;
 		r3t = (unsigned long *)(r2t[i] & _REGION_ENTRY_ORIGIN);
@@ -1318,7 +1318,7 @@ static void __gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr,
 		/* Free region 3 table */
 		page = pfn_to_page(__pa(r3t) >> PAGE_SHIFT);
 		list_del(&page->lru);
-		__free_pages(page, 2);
+		__free_pages(page, CRST_ALLOC_ORDER);
 	}
 }
 
@@ -1338,8 +1338,8 @@ static void gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr)
 	r1e = gmap_table_walk(sg, raddr, 4); /* get region-1 pointer */
 	if (!r1e || !(*r1e & _REGION_ENTRY_ORIGIN))
 		return;
-	gmap_call_notifier(sg, raddr, raddr + (1UL << 53) - 1);
-	r1o = (unsigned long) (r1e - ((raddr >> 53) & 0x7ff));
+	gmap_call_notifier(sg, raddr, raddr + _REGION1_SIZE - 1);
+	r1o = (unsigned long) (r1e - ((raddr & _REGION1_INDEX) >> _REGION1_SHIFT));
 	gmap_idte_one(r1o | _ASCE_TYPE_REGION1, raddr);
 	r2t = (unsigned long *)(*r1e & _REGION_ENTRY_ORIGIN);
 	*r1e = _REGION1_ENTRY_EMPTY;
@@ -1347,7 +1347,7 @@ static void gmap_unshadow_r2t(struct gmap *sg, unsigned long raddr)
 	/* Free region 2 table */
 	page = pfn_to_page(__pa(r2t) >> PAGE_SHIFT);
 	list_del(&page->lru);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 }
 
 /**
@@ -1367,7 +1367,7 @@ static void __gmap_unshadow_r1t(struct gmap *sg, unsigned long raddr,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	asce = (unsigned long) r1t | _ASCE_TYPE_REGION1;
-	for (i = 0; i < 2048; i++, raddr += 1UL << 53) {
+	for (i = 0; i < _CRST_ENTRIES; i++, raddr += _REGION1_SIZE) {
 		if (!(r1t[i] & _REGION_ENTRY_ORIGIN))
 			continue;
 		r2t = (unsigned long *)(r1t[i] & _REGION_ENTRY_ORIGIN);
@@ -1378,7 +1378,7 @@ static void __gmap_unshadow_r1t(struct gmap *sg, unsigned long raddr,
 		/* Free region 2 table */
 		page = pfn_to_page(__pa(r2t) >> PAGE_SHIFT);
 		list_del(&page->lru);
-		__free_pages(page, 2);
+		__free_pages(page, CRST_ALLOC_ORDER);
 	}
 }
 
@@ -1535,7 +1535,7 @@ struct gmap *gmap_shadow(struct gmap *parent, unsigned long asce,
 	/* protect after insertion, so it will get properly invalidated */
 	down_read(&parent->mm->mmap_sem);
 	rc = gmap_protect_range(parent, asce & _ASCE_ORIGIN,
-				((asce & _ASCE_TABLE_LENGTH) + 1) * 4096,
+				((asce & _ASCE_TABLE_LENGTH) + 1) * PAGE_SIZE,
 				PROT_READ, PGSTE_VSIE_BIT);
 	up_read(&parent->mm->mmap_sem);
 	spin_lock(&parent->shadow_lock);
@@ -1578,7 +1578,7 @@ int gmap_shadow_r2t(struct gmap *sg, unsigned long saddr, unsigned long r2t,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	/* Allocate a shadow region second table */
-	page = alloc_pages(GFP_KERNEL, 2);
+	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
 	if (!page)
 		return -ENOMEM;
 	page->index = r2t & _REGION_ENTRY_ORIGIN;
@@ -1614,10 +1614,10 @@ int gmap_shadow_r2t(struct gmap *sg, unsigned long saddr, unsigned long r2t,
 	}
 	spin_unlock(&sg->guest_table_lock);
 	/* Make r2t read-only in parent gmap page table */
-	raddr = (saddr & 0xffe0000000000000UL) | _SHADOW_RMAP_REGION1;
+	raddr = (saddr & _REGION1_MASK) | _SHADOW_RMAP_REGION1;
 	origin = r2t & _REGION_ENTRY_ORIGIN;
-	offset = ((r2t & _REGION_ENTRY_OFFSET) >> 6) * 4096;
-	len = ((r2t & _REGION_ENTRY_LENGTH) + 1) * 4096 - offset;
+	offset = ((r2t & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
+	len = ((r2t & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
 	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
 	spin_lock(&sg->guest_table_lock);
 	if (!rc) {
@@ -1634,7 +1634,7 @@ int gmap_shadow_r2t(struct gmap *sg, unsigned long saddr, unsigned long r2t,
 	return rc;
 out_free:
 	spin_unlock(&sg->guest_table_lock);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 	return rc;
 }
 EXPORT_SYMBOL_GPL(gmap_shadow_r2t);
@@ -1662,7 +1662,7 @@ int gmap_shadow_r3t(struct gmap *sg, unsigned long saddr, unsigned long r3t,
 
 	BUG_ON(!gmap_is_shadow(sg));
 	/* Allocate a shadow region second table */
-	page = alloc_pages(GFP_KERNEL, 2);
+	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
 	if (!page)
 		return -ENOMEM;
 	page->index = r3t & _REGION_ENTRY_ORIGIN;
@@ -1697,10 +1697,10 @@ int gmap_shadow_r3t(struct gmap *sg, unsigned long saddr, unsigned long r3t,
 	}
 	spin_unlock(&sg->guest_table_lock);
 	/* Make r3t read-only in parent gmap page table */
-	raddr = (saddr & 0xfffffc0000000000UL) | _SHADOW_RMAP_REGION2;
+	raddr = (saddr & _REGION2_MASK) | _SHADOW_RMAP_REGION2;
 	origin = r3t & _REGION_ENTRY_ORIGIN;
-	offset = ((r3t & _REGION_ENTRY_OFFSET) >> 6) * 4096;
-	len = ((r3t & _REGION_ENTRY_LENGTH) + 1) * 4096 - offset;
+	offset = ((r3t & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
+	len = ((r3t & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
 	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
 	spin_lock(&sg->guest_table_lock);
 	if (!rc) {
@@ -1717,7 +1717,7 @@ int gmap_shadow_r3t(struct gmap *sg, unsigned long saddr, unsigned long r3t,
 	return rc;
 out_free:
 	spin_unlock(&sg->guest_table_lock);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 	return rc;
 }
 EXPORT_SYMBOL_GPL(gmap_shadow_r3t);
@@ -1745,7 +1745,7 @@ int gmap_shadow_sgt(struct gmap *sg, unsigned long saddr, unsigned long sgt,
 
 	BUG_ON(!gmap_is_shadow(sg) || (sgt & _REGION3_ENTRY_LARGE));
 	/* Allocate a shadow segment table */
-	page = alloc_pages(GFP_KERNEL, 2);
+	page = alloc_pages(GFP_KERNEL, CRST_ALLOC_ORDER);
 	if (!page)
 		return -ENOMEM;
 	page->index = sgt & _REGION_ENTRY_ORIGIN;
@@ -1781,10 +1781,10 @@ int gmap_shadow_sgt(struct gmap *sg, unsigned long saddr, unsigned long sgt,
 	}
 	spin_unlock(&sg->guest_table_lock);
 	/* Make sgt read-only in parent gmap page table */
-	raddr = (saddr & 0xffffffff80000000UL) | _SHADOW_RMAP_REGION3;
+	raddr = (saddr & _REGION3_MASK) | _SHADOW_RMAP_REGION3;
 	origin = sgt & _REGION_ENTRY_ORIGIN;
-	offset = ((sgt & _REGION_ENTRY_OFFSET) >> 6) * 4096;
-	len = ((sgt & _REGION_ENTRY_LENGTH) + 1) * 4096 - offset;
+	offset = ((sgt & _REGION_ENTRY_OFFSET) >> 6) * PAGE_SIZE;
+	len = ((sgt & _REGION_ENTRY_LENGTH) + 1) * PAGE_SIZE - offset;
 	rc = gmap_protect_rmap(sg, raddr, origin + offset, len, PROT_READ);
 	spin_lock(&sg->guest_table_lock);
 	if (!rc) {
@@ -1801,7 +1801,7 @@ int gmap_shadow_sgt(struct gmap *sg, unsigned long saddr, unsigned long sgt,
 	return rc;
 out_free:
 	spin_unlock(&sg->guest_table_lock);
-	__free_pages(page, 2);
+	__free_pages(page, CRST_ALLOC_ORDER);
 	return rc;
 }
 EXPORT_SYMBOL_GPL(gmap_shadow_sgt);
@@ -1902,7 +1902,7 @@ int gmap_shadow_pgt(struct gmap *sg, unsigned long saddr, unsigned long pgt,
 	}
 	spin_unlock(&sg->guest_table_lock);
 	/* Make pgt read-only in parent gmap page table (not the pgste) */
-	raddr = (saddr & 0xfffffffffff00000UL) | _SHADOW_RMAP_SEGMENT;
+	raddr = (saddr & _SEGMENT_MASK) | _SHADOW_RMAP_SEGMENT;
 	origin = pgt & _SEGMENT_ENTRY_ORIGIN & PAGE_MASK;
 	rc = gmap_protect_rmap(sg, raddr, origin, PAGE_SIZE, PROT_READ);
 	spin_lock(&sg->guest_table_lock);
@@ -2021,7 +2021,7 @@ static void gmap_shadow_notify(struct gmap *sg, unsigned long vmaddr,
 	}
 	/* Check for top level table */
 	start = sg->orig_asce & _ASCE_ORIGIN;
-	end = start + ((sg->orig_asce & _ASCE_TABLE_LENGTH) + 1) * 4096;
+	end = start + ((sg->orig_asce & _ASCE_TABLE_LENGTH) + 1) * PAGE_SIZE;
 	if (!(sg->orig_asce & _ASCE_REAL_SPACE) && gaddr >= start &&
 	    gaddr < end) {
 		/* The complete shadow table has to go */
@@ -2032,7 +2032,7 @@ static void gmap_shadow_notify(struct gmap *sg, unsigned long vmaddr,
 		return;
 	}
 	/* Remove the page table tree from on specific entry */
-	head = radix_tree_delete(&sg->host_to_rmap, vmaddr >> 12);
+	head = radix_tree_delete(&sg->host_to_rmap, vmaddr >> PAGE_SHIFT);
 	gmap_for_each_rmap_safe(rmap, rnext, head) {
 		bits = rmap->raddr & _SHADOW_RMAP_MASK;
 		raddr = rmap->raddr ^ bits;
@@ -2076,7 +2076,7 @@ void ptep_notify(struct mm_struct *mm, unsigned long vmaddr,
 	struct gmap *gmap, *sg, *next;
 
 	offset = ((unsigned long) pte) & (255 * sizeof(pte_t));
-	offset = offset * (4096 / sizeof(pte_t));
+	offset = offset * (PAGE_SIZE / sizeof(pte_t));
 	rcu_read_lock();
 	list_for_each_entry_rcu(gmap, &mm->context.gmap_list, list) {
 		spin_lock(&gmap->guest_table_lock);
@@ -2120,6 +2120,37 @@ static inline void thp_split_mm(struct mm_struct *mm)
 #endif
 }
 
+/*
+ * Remove all empty zero pages from the mapping for lazy refaulting
+ * - This must be called after mm->context.has_pgste is set, to avoid
+ *   future creation of zero pages
+ * - This must be called after THP was enabled
+ */
+static int __zap_zero_pages(pmd_t *pmd, unsigned long start,
+			   unsigned long end, struct mm_walk *walk)
+{
+	unsigned long addr;
+
+	for (addr = start; addr != end; addr += PAGE_SIZE) {
+		pte_t *ptep;
+		spinlock_t *ptl;
+
+		ptep = pte_offset_map_lock(walk->mm, pmd, addr, &ptl);
+		if (is_zero_pfn(pte_pfn(*ptep)))
+			ptep_xchg_direct(walk->mm, addr, ptep, __pte(_PAGE_INVALID));
+		pte_unmap_unlock(ptep, ptl);
+	}
+	return 0;
+}
+
+static inline void zap_zero_pages(struct mm_struct *mm)
+{
+	struct mm_walk walk = { .pmd_entry = __zap_zero_pages };
+
+	walk.mm = mm;
+	walk_page_range(0, TASK_SIZE, &walk);
+}
+
 /*
  * switch on pgstes for its userspace process (for kvm)
  */
@@ -2137,6 +2168,7 @@ int s390_enable_sie(void)
 	mm->context.has_pgste = 1;
 	/* split thp mappings and disable thp for future mappings */
 	thp_split_mm(mm);
+	zap_zero_pages(mm);
 	up_write(&mm->mmap_sem);
 	return 0;
 }
@@ -2149,13 +2181,6 @@ EXPORT_SYMBOL_GPL(s390_enable_sie);
 static int __s390_enable_skey(pte_t *pte, unsigned long addr,
 			      unsigned long next, struct mm_walk *walk)
 {
-	/*
-	 * Remove all zero page mappings,
-	 * after establishing a policy to forbid zero page mappings
-	 * following faults for that page will get fresh anonymous pages
-	 */
-	if (is_zero_pfn(pte_pfn(*pte)))
-		ptep_xchg_direct(walk->mm, addr, pte, __pte(_PAGE_INVALID));
 	/* Clear storage key */
 	ptep_zap_key(walk->mm, addr, pte);
 	return 0;

arch/s390/mm/init.c

@@ -26,6 +26,7 @@
 #include <linux/poison.h>
 #include <linux/initrd.h>
 #include <linux/export.h>
+#include <linux/cma.h>
 #include <linux/gfp.h>
 #include <linux/memblock.h>
 #include <asm/processor.h>
@@ -84,7 +85,7 @@ void __init paging_init(void)
 	psw_t psw;
 
 	init_mm.pgd = swapper_pg_dir;
-	if (VMALLOC_END > (1UL << 42)) {
+	if (VMALLOC_END > _REGION2_SIZE) {
 		asce_bits = _ASCE_TYPE_REGION2 | _ASCE_TABLE_LENGTH;
 		pgd_type = _REGION2_ENTRY_EMPTY;
 	} else {
@@ -93,8 +94,7 @@ void __init paging_init(void)
 	}
 	init_mm.context.asce = (__pa(init_mm.pgd) & PAGE_MASK) | asce_bits;
 	S390_lowcore.kernel_asce = init_mm.context.asce;
-	clear_table((unsigned long *) init_mm.pgd, pgd_type,
-		    sizeof(unsigned long)*2048);
+	crst_table_init((unsigned long *) init_mm.pgd, pgd_type);
 	vmem_map_init();
 
         /* enable virtual mapping in kernel mode */
@@ -137,6 +137,8 @@ void __init mem_init(void)
 	free_all_bootmem();
 	setup_zero_pages();	/* Setup zeroed pages. */
 
+	cmma_init_nodat();
+
 	mem_init_print_info(NULL);
 }
 
@@ -166,6 +168,58 @@ unsigned long memory_block_size_bytes(void)
 }
 
 #ifdef CONFIG_MEMORY_HOTPLUG
+
+#ifdef CONFIG_CMA
+
+/* Prevent memory blocks which contain cma regions from going offline */
+
+struct s390_cma_mem_data {
+	unsigned long start;
+	unsigned long end;
+};
+
+static int s390_cma_check_range(struct cma *cma, void *data)
+{
+	struct s390_cma_mem_data *mem_data;
+	unsigned long start, end;
+
+	mem_data = data;
+	start = cma_get_base(cma);
+	end = start + cma_get_size(cma);
+	if (end < mem_data->start)
+		return 0;
+	if (start >= mem_data->end)
+		return 0;
+	return -EBUSY;
+}
+
+static int s390_cma_mem_notifier(struct notifier_block *nb,
+				 unsigned long action, void *data)
+{
+	struct s390_cma_mem_data mem_data;
+	struct memory_notify *arg;
+	int rc = 0;
+
+	arg = data;
+	mem_data.start = arg->start_pfn << PAGE_SHIFT;
+	mem_data.end = mem_data.start + (arg->nr_pages << PAGE_SHIFT);
+	if (action == MEM_GOING_OFFLINE)
+		rc = cma_for_each_area(s390_cma_check_range, &mem_data);
+	return notifier_from_errno(rc);
+}
+
+static struct notifier_block s390_cma_mem_nb = {
+	.notifier_call = s390_cma_mem_notifier,
+};
+
+static int __init s390_cma_mem_init(void)
+{
+	return register_memory_notifier(&s390_cma_mem_nb);
+}
+device_initcall(s390_cma_mem_init);
+
+#endif /* CONFIG_CMA */
+
 int arch_add_memory(int nid, u64 start, u64 size, bool want_memblock)
 {
 	unsigned long start_pfn = PFN_DOWN(start);

arch/s390/mm/page-states.c

@@ -10,9 +10,10 @@
 #include <linux/errno.h>
 #include <linux/types.h>
 #include <linux/mm.h>
+#include <linux/memblock.h>
 #include <linux/gfp.h>
 #include <linux/init.h>
-
+#include <asm/facility.h>
 #include <asm/page-states.h>
 
 static int cmma_flag = 1;
@@ -36,14 +37,16 @@ __setup("cmma=", cmma);
 static inline int cmma_test_essa(void)
 {
 	register unsigned long tmp asm("0") = 0;
-	register int rc asm("1") = -EOPNOTSUPP;
+	register int rc asm("1");
 
+	/* test ESSA_GET_STATE */
 	asm volatile(
-		"       .insn rrf,0xb9ab0000,%1,%1,0,0\n"
+		"	.insn	rrf,0xb9ab0000,%1,%1,%2,0\n"
 		"0:     la      %0,0\n"
 		"1:\n"
 		EX_TABLE(0b,1b)
-		: "+&d" (rc), "+&d" (tmp));
+		: "=&d" (rc), "+&d" (tmp)
+		: "i" (ESSA_GET_STATE), "0" (-EOPNOTSUPP));
 	return rc;
 }
 
@@ -51,11 +54,26 @@ void __init cmma_init(void)
 {
 	if (!cmma_flag)
 		return;
-	if (cmma_test_essa())
+	if (cmma_test_essa()) {
 		cmma_flag = 0;
+		return;
+	}
+	if (test_facility(147))
+		cmma_flag = 2;
 }
 
-static inline void set_page_unstable(struct page *page, int order)
+static inline unsigned char get_page_state(struct page *page)
+{
+	unsigned char state;
+
+	asm volatile("	.insn	rrf,0xb9ab0000,%0,%1,%2,0"
+		     : "=&d" (state)
+		     : "a" (page_to_phys(page)),
+		       "i" (ESSA_GET_STATE));
+	return state & 0x3f;
+}
+
+static inline void set_page_unused(struct page *page, int order)
 {
 	int i, rc;
 
@@ -66,14 +84,18 @@ static inline void set_page_unstable(struct page *page, int order)
 			       "i" (ESSA_SET_UNUSED));
 }
 
-void arch_free_page(struct page *page, int order)
+static inline void set_page_stable_dat(struct page *page, int order)
 {
-	if (!cmma_flag)
-		return;
-	set_page_unstable(page, order);
+	int i, rc;
+
+	for (i = 0; i < (1 << order); i++)
+		asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0"
+			     : "=&d" (rc)
+			     : "a" (page_to_phys(page + i)),
+			       "i" (ESSA_SET_STABLE));
 }
 
-static inline void set_page_stable(struct page *page, int order)
+static inline void set_page_stable_nodat(struct page *page, int order)
 {
 	int i, rc;
 
@@ -81,14 +103,154 @@ static inline void set_page_stable(struct page *page, int order)
 		asm volatile(".insn rrf,0xb9ab0000,%0,%1,%2,0"
 			     : "=&d" (rc)
 			     : "a" (page_to_phys(page + i)),
-			       "i" (ESSA_SET_STABLE));
+			       "i" (ESSA_SET_STABLE_NODAT));
+}
+
+static void mark_kernel_pmd(pud_t *pud, unsigned long addr, unsigned long end)
+{
+	unsigned long next;
+	struct page *page;
+	pmd_t *pmd;
+
+	pmd = pmd_offset(pud, addr);
+	do {
+		next = pmd_addr_end(addr, end);
+		if (pmd_none(*pmd) || pmd_large(*pmd))
+			continue;
+		page = virt_to_page(pmd_val(*pmd));
+		set_bit(PG_arch_1, &page->flags);
+	} while (pmd++, addr = next, addr != end);
+}
+
+static void mark_kernel_pud(p4d_t *p4d, unsigned long addr, unsigned long end)
+{
+	unsigned long next;
+	struct page *page;
+	pud_t *pud;
+	int i;
+
+	pud = pud_offset(p4d, addr);
+	do {
+		next = pud_addr_end(addr, end);
+		if (pud_none(*pud) || pud_large(*pud))
+			continue;
+		if (!pud_folded(*pud)) {
+			page = virt_to_page(pud_val(*pud));
+			for (i = 0; i < 3; i++)
+				set_bit(PG_arch_1, &page[i].flags);
+		}
+		mark_kernel_pmd(pud, addr, next);
+	} while (pud++, addr = next, addr != end);
+}
+
+static void mark_kernel_p4d(pgd_t *pgd, unsigned long addr, unsigned long end)
+{
+	unsigned long next;
+	struct page *page;
+	p4d_t *p4d;
+	int i;
+
+	p4d = p4d_offset(pgd, addr);
+	do {
+		next = p4d_addr_end(addr, end);
+		if (p4d_none(*p4d))
+			continue;
+		if (!p4d_folded(*p4d)) {
+			page = virt_to_page(p4d_val(*p4d));
+			for (i = 0; i < 3; i++)
+				set_bit(PG_arch_1, &page[i].flags);
+		}
+		mark_kernel_pud(p4d, addr, next);
+	} while (p4d++, addr = next, addr != end);
+}
+
+static void mark_kernel_pgd(void)
+{
+	unsigned long addr, next;
+	struct page *page;
+	pgd_t *pgd;
+	int i;
+
+	addr = 0;
+	pgd = pgd_offset_k(addr);
+	do {
+		next = pgd_addr_end(addr, MODULES_END);
+		if (pgd_none(*pgd))
+			continue;
+		if (!pgd_folded(*pgd)) {
+			page = virt_to_page(pgd_val(*pgd));
+			for (i = 0; i < 3; i++)
+				set_bit(PG_arch_1, &page[i].flags);
+		}
+		mark_kernel_p4d(pgd, addr, next);
+	} while (pgd++, addr = next, addr != MODULES_END);
+}
+
+void __init cmma_init_nodat(void)
+{
+	struct memblock_region *reg;
+	struct page *page;
+	unsigned long start, end, ix;
+
+	if (cmma_flag < 2)
+		return;
+	/* Mark pages used in kernel page tables */
+	mark_kernel_pgd();
+
+	/* Set all kernel pages not used for page tables to stable/no-dat */
+	for_each_memblock(memory, reg) {
+		start = memblock_region_memory_base_pfn(reg);
+		end = memblock_region_memory_end_pfn(reg);
+		page = pfn_to_page(start);
+		for (ix = start; ix < end; ix++, page++) {
+			if (__test_and_clear_bit(PG_arch_1, &page->flags))
+				continue;	/* skip page table pages */
+			if (!list_empty(&page->lru))
+				continue;	/* skip free pages */
+			set_page_stable_nodat(page, 0);
+		}
+	}
+}
+
+void arch_free_page(struct page *page, int order)
+{
+	if (!cmma_flag)
+		return;
+	set_page_unused(page, order);
 }
 
 void arch_alloc_page(struct page *page, int order)
 {
 	if (!cmma_flag)
 		return;
-	set_page_stable(page, order);
+	if (cmma_flag < 2)
+		set_page_stable_dat(page, order);
+	else
+		set_page_stable_nodat(page, order);
+}
+
+void arch_set_page_dat(struct page *page, int order)
+{
+	if (!cmma_flag)
+		return;
+	set_page_stable_dat(page, order);
+}
+
+void arch_set_page_nodat(struct page *page, int order)
+{
+	if (cmma_flag < 2)
+		return;
+	set_page_stable_nodat(page, order);
+}
+
+int arch_test_page_nodat(struct page *page)
+{
+	unsigned char state;
+
+	if (cmma_flag < 2)
+		return 0;
+	state = get_page_state(page);
+	return !!(state & 0x20);
 }
 
 void arch_set_page_states(int make_stable)
@@ -108,9 +270,9 @@ void arch_set_page_states(int make_stable)
 			list_for_each(l, &zone->free_area[order].free_list[t]) {
 				page = list_entry(l, struct page, lru);
 				if (make_stable)
-					set_page_stable(page, order);
+					set_page_stable_dat(page, 0);
 				else
-					set_page_unstable(page, order);
+					set_page_unused(page, order);
 			}
 		}
 		spin_unlock_irqrestore(&zone->lock, flags);

arch/s390/mm/pageattr.c

@@ -7,6 +7,7 @@
 #include <asm/cacheflush.h>
 #include <asm/facility.h>
 #include <asm/pgtable.h>
+#include <asm/pgalloc.h>
 #include <asm/page.h>
 #include <asm/set_memory.h>
 
@@ -191,7 +192,7 @@ static int split_pud_page(pud_t *pudp, unsigned long addr)
 	pud_t new;
 	int i, ro, nx;
 
-	pm_dir = vmem_pmd_alloc();
+	pm_dir = vmem_crst_alloc(_SEGMENT_ENTRY_EMPTY);
 	if (!pm_dir)
 		return -ENOMEM;
 	pmd_addr = pud_pfn(*pudp) << PAGE_SHIFT;
@@ -328,7 +329,7 @@ static void ipte_range(pte_t *pte, unsigned long address, int nr)
 		return;
 	}
 	for (i = 0; i < nr; i++) {
-		__ptep_ipte(address, pte, IPTE_GLOBAL);
+		__ptep_ipte(address, pte, 0, 0, IPTE_GLOBAL);
 		address += PAGE_SIZE;
 		pte++;
 	}

arch/s390/mm/pgalloc.c

@@ -57,6 +57,7 @@ unsigned long *crst_table_alloc(struct mm_struct *mm)
 
 	if (!page)
 		return NULL;
+	arch_set_page_dat(page, 2);
 	return (unsigned long *) page_to_phys(page);
 }
 
@@ -82,7 +83,7 @@ int crst_table_upgrade(struct mm_struct *mm, unsigned long end)
 	int rc, notify;
 
 	/* upgrade should only happen from 3 to 4, 3 to 5, or 4 to 5 levels */
-	BUG_ON(mm->context.asce_limit < (1UL << 42));
+	BUG_ON(mm->context.asce_limit < _REGION2_SIZE);
 	if (end >= TASK_SIZE_MAX)
 		return -ENOMEM;
 	rc = 0;
@@ -95,11 +96,11 @@ int crst_table_upgrade(struct mm_struct *mm, unsigned long end)
 		}
 		spin_lock_bh(&mm->page_table_lock);
 		pgd = (unsigned long *) mm->pgd;
-		if (mm->context.asce_limit == (1UL << 42)) {
+		if (mm->context.asce_limit == _REGION2_SIZE) {
 			crst_table_init(table, _REGION2_ENTRY_EMPTY);
 			p4d_populate(mm, (p4d_t *) table, (pud_t *) pgd);
 			mm->pgd = (pgd_t *) table;
-			mm->context.asce_limit = 1UL << 53;
+			mm->context.asce_limit = _REGION1_SIZE;
 			mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 				_ASCE_USER_BITS | _ASCE_TYPE_REGION2;
 		} else {
@@ -123,7 +124,7 @@ void crst_table_downgrade(struct mm_struct *mm)
 	pgd_t *pgd;
 
 	/* downgrade should only happen from 3 to 2 levels (compat only) */
-	BUG_ON(mm->context.asce_limit != (1UL << 42));
+	BUG_ON(mm->context.asce_limit != _REGION2_SIZE);
 
 	if (current->active_mm == mm) {
 		clear_user_asce();
@@ -132,7 +133,7 @@ void crst_table_downgrade(struct mm_struct *mm)
 
 	pgd = mm->pgd;
 	mm->pgd = (pgd_t *) (pgd_val(*pgd) & _REGION_ENTRY_ORIGIN);
-	mm->context.asce_limit = 1UL << 31;
+	mm->context.asce_limit = _REGION3_SIZE;
 	mm->context.asce = __pa(mm->pgd) | _ASCE_TABLE_LENGTH |
 			   _ASCE_USER_BITS | _ASCE_TYPE_SEGMENT;
 	crst_table_free(mm, (unsigned long *) pgd);
@@ -214,6 +215,7 @@ unsigned long *page_table_alloc(struct mm_struct *mm)
 		__free_page(page);
 		return NULL;
 	}
+	arch_set_page_dat(page, 0);
 	/* Initialize page table */
 	table = (unsigned long *) page_to_phys(page);
 	if (mm_alloc_pgste(mm)) {

arch/s390/mm/pgtable.c

@@ -25,8 +25,49 @@
 #include <asm/mmu_context.h>
 #include <asm/page-states.h>
 
+static inline void ptep_ipte_local(struct mm_struct *mm, unsigned long addr,
+				   pte_t *ptep, int nodat)
+{
+	unsigned long opt, asce;
+
+	if (MACHINE_HAS_TLB_GUEST) {
+		opt = 0;
+		asce = READ_ONCE(mm->context.gmap_asce);
+		if (asce == 0UL || nodat)
+			opt |= IPTE_NODAT;
+		if (asce != -1UL) {
+			asce = asce ? : mm->context.asce;
+			opt |= IPTE_GUEST_ASCE;
+		}
+		__ptep_ipte(addr, ptep, opt, asce, IPTE_LOCAL);
+	} else {
+		__ptep_ipte(addr, ptep, 0, 0, IPTE_LOCAL);
+	}
+}
+
+static inline void ptep_ipte_global(struct mm_struct *mm, unsigned long addr,
+				    pte_t *ptep, int nodat)
+{
+	unsigned long opt, asce;
+
+	if (MACHINE_HAS_TLB_GUEST) {
+		opt = 0;
+		asce = READ_ONCE(mm->context.gmap_asce);
+		if (asce == 0UL || nodat)
+			opt |= IPTE_NODAT;
+		if (asce != -1UL) {
+			asce = asce ? : mm->context.asce;
+			opt |= IPTE_GUEST_ASCE;
+		}
+		__ptep_ipte(addr, ptep, opt, asce, IPTE_GLOBAL);
+	} else {
+		__ptep_ipte(addr, ptep, 0, 0, IPTE_GLOBAL);
+	}
+}
+
 static inline pte_t ptep_flush_direct(struct mm_struct *mm,
-				      unsigned long addr, pte_t *ptep)
+				      unsigned long addr, pte_t *ptep,
+				      int nodat)
 {
 	pte_t old;
 
@@ -36,15 +77,16 @@ static inline pte_t ptep_flush_direct(struct mm_struct *mm,
 	atomic_inc(&mm->context.flush_count);
 	if (MACHINE_HAS_TLB_LC &&
 	    cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
-		__ptep_ipte(addr, ptep, IPTE_LOCAL);
+		ptep_ipte_local(mm, addr, ptep, nodat);
 	else
-		__ptep_ipte(addr, ptep, IPTE_GLOBAL);
+		ptep_ipte_global(mm, addr, ptep, nodat);
 	atomic_dec(&mm->context.flush_count);
 	return old;
 }
 
 static inline pte_t ptep_flush_lazy(struct mm_struct *mm,
-				    unsigned long addr, pte_t *ptep)
+				    unsigned long addr, pte_t *ptep,
+				    int nodat)
 {
 	pte_t old;
 
@@ -57,7 +99,7 @@ static inline pte_t ptep_flush_lazy(struct mm_struct *mm,
 		pte_val(*ptep) |= _PAGE_INVALID;
 		mm->context.flush_mm = 1;
 	} else
-		__ptep_ipte(addr, ptep, IPTE_GLOBAL);
+		ptep_ipte_global(mm, addr, ptep, nodat);
 	atomic_dec(&mm->context.flush_count);
 	return old;
 }
@@ -229,10 +271,12 @@ pte_t ptep_xchg_direct(struct mm_struct *mm, unsigned long addr,
 {
 	pgste_t pgste;
 	pte_t old;
+	int nodat;
 
 	preempt_disable();
 	pgste = ptep_xchg_start(mm, addr, ptep);
-	old = ptep_flush_direct(mm, addr, ptep);
+	nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
+	old = ptep_flush_direct(mm, addr, ptep, nodat);
 	old = ptep_xchg_commit(mm, addr, ptep, pgste, old, new);
 	preempt_enable();
 	return old;
@@ -244,10 +288,12 @@ pte_t ptep_xchg_lazy(struct mm_struct *mm, unsigned long addr,
 {
 	pgste_t pgste;
 	pte_t old;
+	int nodat;
 
 	preempt_disable();
 	pgste = ptep_xchg_start(mm, addr, ptep);
-	old = ptep_flush_lazy(mm, addr, ptep);
+	nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
+	old = ptep_flush_lazy(mm, addr, ptep, nodat);
 	old = ptep_xchg_commit(mm, addr, ptep, pgste, old, new);
 	preempt_enable();
 	return old;
@@ -259,10 +305,12 @@ pte_t ptep_modify_prot_start(struct mm_struct *mm, unsigned long addr,
 {
 	pgste_t pgste;
 	pte_t old;
+	int nodat;
 
 	preempt_disable();
 	pgste = ptep_xchg_start(mm, addr, ptep);
-	old = ptep_flush_lazy(mm, addr, ptep);
+	nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
+	old = ptep_flush_lazy(mm, addr, ptep, nodat);
 	if (mm_has_pgste(mm)) {
 		pgste = pgste_update_all(old, pgste, mm);
 		pgste_set(ptep, pgste);
@@ -290,6 +338,28 @@ void ptep_modify_prot_commit(struct mm_struct *mm, unsigned long addr,
 }
 EXPORT_SYMBOL(ptep_modify_prot_commit);
 
+static inline void pmdp_idte_local(struct mm_struct *mm,
+				   unsigned long addr, pmd_t *pmdp)
+{
+	if (MACHINE_HAS_TLB_GUEST)
+		__pmdp_idte(addr, pmdp, IDTE_NODAT | IDTE_GUEST_ASCE,
+			    mm->context.asce, IDTE_LOCAL);
+	else
+		__pmdp_idte(addr, pmdp, 0, 0, IDTE_LOCAL);
+}
+
+static inline void pmdp_idte_global(struct mm_struct *mm,
+				    unsigned long addr, pmd_t *pmdp)
+{
+	if (MACHINE_HAS_TLB_GUEST)
+		__pmdp_idte(addr, pmdp, IDTE_NODAT | IDTE_GUEST_ASCE,
+			    mm->context.asce, IDTE_GLOBAL);
+	else if (MACHINE_HAS_IDTE)
+		__pmdp_idte(addr, pmdp, 0, 0, IDTE_GLOBAL);
+	else
+		__pmdp_csp(pmdp);
+}
+
 static inline pmd_t pmdp_flush_direct(struct mm_struct *mm,
 				      unsigned long addr, pmd_t *pmdp)
 {
@@ -298,16 +368,12 @@ static inline pmd_t pmdp_flush_direct(struct mm_struct *mm,
 	old = *pmdp;
 	if (pmd_val(old) & _SEGMENT_ENTRY_INVALID)
 		return old;
-	if (!MACHINE_HAS_IDTE) {
-		__pmdp_csp(pmdp);
-		return old;
-	}
 	atomic_inc(&mm->context.flush_count);
 	if (MACHINE_HAS_TLB_LC &&
 	    cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
-		__pmdp_idte(addr, pmdp, IDTE_LOCAL);
+		pmdp_idte_local(mm, addr, pmdp);
 	else
-		__pmdp_idte(addr, pmdp, IDTE_GLOBAL);
+		pmdp_idte_global(mm, addr, pmdp);
 	atomic_dec(&mm->context.flush_count);
 	return old;
 }
@@ -325,10 +391,9 @@ static inline pmd_t pmdp_flush_lazy(struct mm_struct *mm,
 			  cpumask_of(smp_processor_id()))) {
 		pmd_val(*pmdp) |= _SEGMENT_ENTRY_INVALID;
 		mm->context.flush_mm = 1;
-	} else if (MACHINE_HAS_IDTE)
-		__pmdp_idte(addr, pmdp, IDTE_GLOBAL);
-	else
-		__pmdp_csp(pmdp);
+	} else {
+		pmdp_idte_global(mm, addr, pmdp);
+	}
 	atomic_dec(&mm->context.flush_count);
 	return old;
 }
@@ -359,28 +424,46 @@ pmd_t pmdp_xchg_lazy(struct mm_struct *mm, unsigned long addr,
 }
 EXPORT_SYMBOL(pmdp_xchg_lazy);
 
-static inline pud_t pudp_flush_direct(struct mm_struct *mm,
-				      unsigned long addr, pud_t *pudp)
+static inline void pudp_idte_local(struct mm_struct *mm,
+				   unsigned long addr, pud_t *pudp)
 {
-	pud_t old;
+	if (MACHINE_HAS_TLB_GUEST)
+		__pudp_idte(addr, pudp, IDTE_NODAT | IDTE_GUEST_ASCE,
+			    mm->context.asce, IDTE_LOCAL);
+	else
+		__pudp_idte(addr, pudp, 0, 0, IDTE_LOCAL);
+}
 
-	old = *pudp;
-	if (pud_val(old) & _REGION_ENTRY_INVALID)
-		return old;
-	if (!MACHINE_HAS_IDTE) {
+static inline void pudp_idte_global(struct mm_struct *mm,
+				    unsigned long addr, pud_t *pudp)
+{
+	if (MACHINE_HAS_TLB_GUEST)
+		__pudp_idte(addr, pudp, IDTE_NODAT | IDTE_GUEST_ASCE,
+			    mm->context.asce, IDTE_GLOBAL);
+	else if (MACHINE_HAS_IDTE)
+		__pudp_idte(addr, pudp, 0, 0, IDTE_GLOBAL);
+	else
 		/*
 		 * Invalid bit position is the same for pmd and pud, so we can
 		 * re-use _pmd_csp() here
 		 */
 		__pmdp_csp((pmd_t *) pudp);
+}
+
+static inline pud_t pudp_flush_direct(struct mm_struct *mm,
+				      unsigned long addr, pud_t *pudp)
+{
+	pud_t old;
+
+	old = *pudp;
+	if (pud_val(old) & _REGION_ENTRY_INVALID)
 		return old;
-	}
 	atomic_inc(&mm->context.flush_count);
 	if (MACHINE_HAS_TLB_LC &&
 	    cpumask_equal(mm_cpumask(mm), cpumask_of(smp_processor_id())))
-		__pudp_idte(addr, pudp, IDTE_LOCAL);
+		pudp_idte_local(mm, addr, pudp);
 	else
-		__pudp_idte(addr, pudp, IDTE_GLOBAL);
+		pudp_idte_global(mm, addr, pudp);
 	atomic_dec(&mm->context.flush_count);
 	return old;
 }
@@ -482,7 +565,7 @@ int ptep_force_prot(struct mm_struct *mm, unsigned long addr,
 {
 	pte_t entry;
 	pgste_t pgste;
-	int pte_i, pte_p;
+	int pte_i, pte_p, nodat;
 
 	pgste = pgste_get_lock(ptep);
 	entry = *ptep;
@@ -495,13 +578,14 @@ int ptep_force_prot(struct mm_struct *mm, unsigned long addr,
 		return -EAGAIN;
 	}
 	/* Change access rights and set pgste bit */
+	nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
 	if (prot == PROT_NONE && !pte_i) {
-		ptep_flush_direct(mm, addr, ptep);
+		ptep_flush_direct(mm, addr, ptep, nodat);
 		pgste = pgste_update_all(entry, pgste, mm);
 		pte_val(entry) |= _PAGE_INVALID;
 	}
 	if (prot == PROT_READ && !pte_p) {
-		ptep_flush_direct(mm, addr, ptep);
+		ptep_flush_direct(mm, addr, ptep, nodat);
 		pte_val(entry) &= ~_PAGE_INVALID;
 		pte_val(entry) |= _PAGE_PROTECT;
 	}
@@ -541,10 +625,12 @@ int ptep_shadow_pte(struct mm_struct *mm, unsigned long saddr,
 void ptep_unshadow_pte(struct mm_struct *mm, unsigned long saddr, pte_t *ptep)
 {
 	pgste_t pgste;
+	int nodat;
 
 	pgste = pgste_get_lock(ptep);
 	/* notifier is called by the caller */
-	ptep_flush_direct(mm, saddr, ptep);
+	nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
+	ptep_flush_direct(mm, saddr, ptep, nodat);
 	/* don't touch the storage key - it belongs to parent pgste */
 	pgste = pgste_set_pte(ptep, pgste, __pte(_PAGE_INVALID));
 	pgste_set_unlock(ptep, pgste);
@@ -617,6 +703,7 @@ bool test_and_clear_guest_dirty(struct mm_struct *mm, unsigned long addr)
 	pte_t *ptep;
 	pte_t pte;
 	bool dirty;
+	int nodat;
 
 	pgd = pgd_offset(mm, addr);
 	p4d = p4d_alloc(mm, pgd, addr);
@@ -645,7 +732,8 @@ bool test_and_clear_guest_dirty(struct mm_struct *mm, unsigned long addr)
 	pte = *ptep;
 	if (dirty && (pte_val(pte) & _PAGE_PRESENT)) {
 		pgste = pgste_pte_notify(mm, addr, ptep, pgste);
-		__ptep_ipte(addr, ptep, IPTE_GLOBAL);
+		nodat = !!(pgste_val(pgste) & _PGSTE_GPS_NODAT);
+		ptep_ipte_global(mm, addr, ptep, nodat);
 		if (MACHINE_HAS_ESOP || !(pte_val(pte) & _PAGE_WRITE))
 			pte_val(pte) |= _PAGE_PROTECT;
 		else

arch/s390/mm/vmem.c

@@ -38,37 +38,14 @@ static void __ref *vmem_alloc_pages(unsigned int order)
 	return (void *) memblock_alloc(size, size);
 }
 
-static inline p4d_t *vmem_p4d_alloc(void)
+void *vmem_crst_alloc(unsigned long val)
 {
-	p4d_t *p4d = NULL;
+	unsigned long *table;
 
-	p4d = vmem_alloc_pages(2);
-	if (!p4d)
-		return NULL;
-	clear_table((unsigned long *) p4d, _REGION2_ENTRY_EMPTY, PAGE_SIZE * 4);
-	return p4d;
-}
-
-static inline pud_t *vmem_pud_alloc(void)
-{
-	pud_t *pud = NULL;
-
-	pud = vmem_alloc_pages(2);
-	if (!pud)
-		return NULL;
-	clear_table((unsigned long *) pud, _REGION3_ENTRY_EMPTY, PAGE_SIZE * 4);
-	return pud;
-}
-
-pmd_t *vmem_pmd_alloc(void)
-{
-	pmd_t *pmd = NULL;
-
-	pmd = vmem_alloc_pages(2);
-	if (!pmd)
-		return NULL;
-	clear_table((unsigned long *) pmd, _SEGMENT_ENTRY_EMPTY, PAGE_SIZE * 4);
-	return pmd;
+	table = vmem_alloc_pages(CRST_ALLOC_ORDER);
+	if (table)
+		crst_table_init(table, val);
+	return table;
 }
 
 pte_t __ref *vmem_pte_alloc(void)
@@ -114,14 +91,14 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 	while (address < end) {
 		pg_dir = pgd_offset_k(address);
 		if (pgd_none(*pg_dir)) {
-			p4_dir = vmem_p4d_alloc();
+			p4_dir = vmem_crst_alloc(_REGION2_ENTRY_EMPTY);
 			if (!p4_dir)
 				goto out;
 			pgd_populate(&init_mm, pg_dir, p4_dir);
 		}
 		p4_dir = p4d_offset(pg_dir, address);
 		if (p4d_none(*p4_dir)) {
-			pu_dir = vmem_pud_alloc();
+			pu_dir = vmem_crst_alloc(_REGION3_ENTRY_EMPTY);
 			if (!pu_dir)
 				goto out;
 			p4d_populate(&init_mm, p4_dir, pu_dir);
@@ -136,7 +113,7 @@ static int vmem_add_mem(unsigned long start, unsigned long size)
 			continue;
 		}
 		if (pud_none(*pu_dir)) {
-			pm_dir = vmem_pmd_alloc();
+			pm_dir = vmem_crst_alloc(_SEGMENT_ENTRY_EMPTY);
 			if (!pm_dir)
 				goto out;
 			pud_populate(&init_mm, pu_dir, pm_dir);
@@ -253,7 +230,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 	for (address = start; address < end;) {
 		pg_dir = pgd_offset_k(address);
 		if (pgd_none(*pg_dir)) {
-			p4_dir = vmem_p4d_alloc();
+			p4_dir = vmem_crst_alloc(_REGION2_ENTRY_EMPTY);
 			if (!p4_dir)
 				goto out;
 			pgd_populate(&init_mm, pg_dir, p4_dir);
@@ -261,7 +238,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 
 		p4_dir = p4d_offset(pg_dir, address);
 		if (p4d_none(*p4_dir)) {
-			pu_dir = vmem_pud_alloc();
+			pu_dir = vmem_crst_alloc(_REGION3_ENTRY_EMPTY);
 			if (!pu_dir)
 				goto out;
 			p4d_populate(&init_mm, p4_dir, pu_dir);
@@ -269,7 +246,7 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node)
 
 		pu_dir = pud_offset(p4_dir, address);
 		if (pud_none(*pu_dir)) {
-			pm_dir = vmem_pmd_alloc();
+			pm_dir = vmem_crst_alloc(_SEGMENT_ENTRY_EMPTY);
 			if (!pm_dir)
 				goto out;
 			pud_populate(&init_mm, pu_dir, pm_dir);

arch/s390/pci/pci_clp.c

@@ -24,6 +24,14 @@
 
 bool zpci_unique_uid;
 
+static void update_uid_checking(bool new)
+{
+	if (zpci_unique_uid != new)
+		zpci_dbg(1, "uid checking:%d\n", new);
+
+	zpci_unique_uid = new;
+}
+
 static inline void zpci_err_clp(unsigned int rsp, int rc)
 {
 	struct {
@@ -319,7 +327,7 @@ static int clp_list_pci(struct clp_req_rsp_list_pci *rrb, void *data,
 			goto out;
 		}
 
-		zpci_unique_uid = rrb->response.uid_checking;
+		update_uid_checking(rrb->response.uid_checking);
 		WARN_ON_ONCE(rrb->response.entry_size !=
 			sizeof(struct clp_fh_list_entry));
 

arch/s390/tools/gen_facilities.c

@@ -41,7 +41,7 @@ static struct facility_def facility_defs[] = {
 			27, /* mvcos */
 			32, /* compare and swap and store */
 			33, /* compare and swap and store 2 */
-			34, /* general extension facility */
+			34, /* general instructions extension */
 			35, /* execute extensions */
 #endif
 #ifdef CONFIG_HAVE_MARCH_Z196_FEATURES
@@ -53,6 +53,9 @@ static struct facility_def facility_defs[] = {
 #endif
 #ifdef CONFIG_HAVE_MARCH_Z13_FEATURES
 			53, /* load-and-zero-rightmost-byte, etc. */
+#endif
+#ifdef CONFIG_HAVE_MARCH_Z14_FEATURES
+			58, /* miscellaneous-instruction-extension 2 */
 #endif
 			-1 /* END */
 		}

drivers/s390/block/dasd.c

@@ -801,11 +801,12 @@ static void dasd_profile_end(struct dasd_block *block,
 			     struct dasd_ccw_req *cqr,
 			     struct request *req)
 {
-	long strtime, irqtime, endtime, tottime;	/* in microseconds */
-	long tottimeps, sectors;
+	unsigned long strtime, irqtime, endtime, tottime;
+	unsigned long tottimeps, sectors;
 	struct dasd_device *device;
 	int sectors_ind, tottime_ind, tottimeps_ind, strtime_ind;
 	int irqtime_ind, irqtimeps_ind, endtime_ind;
+	struct dasd_profile_info *data;
 
 	device = cqr->startdev;
 	if (!(dasd_global_profile_level ||
@@ -835,6 +836,11 @@ static void dasd_profile_end(struct dasd_block *block,
 
 	spin_lock(&dasd_global_profile.lock);
 	if (dasd_global_profile.data) {
+		data = dasd_global_profile.data;
+		data->dasd_sum_times += tottime;
+		data->dasd_sum_time_str += strtime;
+		data->dasd_sum_time_irq += irqtime;
+		data->dasd_sum_time_end += endtime;
 		dasd_profile_end_add_data(dasd_global_profile.data,
 					  cqr->startdev != block->base,
 					  cqr->cpmode == 1,
@@ -847,7 +853,12 @@ static void dasd_profile_end(struct dasd_block *block,
 	spin_unlock(&dasd_global_profile.lock);
 
 	spin_lock(&block->profile.lock);
-	if (block->profile.data)
+	if (block->profile.data) {
+		data = block->profile.data;
+		data->dasd_sum_times += tottime;
+		data->dasd_sum_time_str += strtime;
+		data->dasd_sum_time_irq += irqtime;
+		data->dasd_sum_time_end += endtime;
 		dasd_profile_end_add_data(block->profile.data,
 					  cqr->startdev != block->base,
 					  cqr->cpmode == 1,
@@ -856,10 +867,16 @@ static void dasd_profile_end(struct dasd_block *block,
 					  tottimeps_ind, strtime_ind,
 					  irqtime_ind, irqtimeps_ind,
 					  endtime_ind);
+	}
 	spin_unlock(&block->profile.lock);
 
 	spin_lock(&device->profile.lock);
-	if (device->profile.data)
+	if (device->profile.data) {
+		data = device->profile.data;
+		data->dasd_sum_times += tottime;
+		data->dasd_sum_time_str += strtime;
+		data->dasd_sum_time_irq += irqtime;
+		data->dasd_sum_time_end += endtime;
 		dasd_profile_end_add_data(device->profile.data,
 					  cqr->startdev != block->base,
 					  cqr->cpmode == 1,
@@ -868,6 +885,7 @@ static void dasd_profile_end(struct dasd_block *block,
 					  tottimeps_ind, strtime_ind,
 					  irqtime_ind, irqtimeps_ind,
 					  endtime_ind);
+	}
 	spin_unlock(&device->profile.lock);
 }
 
@@ -989,6 +1007,14 @@ static void dasd_stats_seq_print(struct seq_file *m,
 	seq_printf(m, "total_sectors %u\n", data->dasd_io_sects);
 	seq_printf(m, "total_pav %u\n", data->dasd_io_alias);
 	seq_printf(m, "total_hpf %u\n", data->dasd_io_tpm);
+	seq_printf(m, "avg_total %lu\n", data->dasd_io_reqs ?
+		   data->dasd_sum_times / data->dasd_io_reqs : 0UL);
+	seq_printf(m, "avg_build_to_ssch %lu\n", data->dasd_io_reqs ?
+		   data->dasd_sum_time_str / data->dasd_io_reqs : 0UL);
+	seq_printf(m, "avg_ssch_to_irq %lu\n", data->dasd_io_reqs ?
+		   data->dasd_sum_time_irq / data->dasd_io_reqs : 0UL);
+	seq_printf(m, "avg_irq_to_end %lu\n", data->dasd_io_reqs ?
+		   data->dasd_sum_time_end / data->dasd_io_reqs : 0UL);
 	seq_puts(m, "histogram_sectors ");
 	dasd_stats_array(m, data->dasd_io_secs);
 	seq_puts(m, "histogram_io_times ");
@@ -1639,7 +1665,7 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
 {
 	struct dasd_ccw_req *cqr, *next;
 	struct dasd_device *device;
-	unsigned long long now;
+	unsigned long now;
 	int nrf_suppressed = 0;
 	int fp_suppressed = 0;
 	u8 *sense = NULL;
@@ -3152,7 +3178,9 @@ static int dasd_alloc_queue(struct dasd_block *block)
  */
 static void dasd_setup_queue(struct dasd_block *block)
 {
+	unsigned int logical_block_size = block->bp_block;
 	struct request_queue *q = block->request_queue;
+	unsigned int max_bytes, max_discard_sectors;
 	int max;
 
 	if (block->base->features & DASD_FEATURE_USERAW) {
@@ -3169,7 +3197,7 @@ static void dasd_setup_queue(struct dasd_block *block)
 	}
 	queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
 	q->limits.max_dev_sectors = max;
-	blk_queue_logical_block_size(q, block->bp_block);
+	blk_queue_logical_block_size(q, logical_block_size);
 	blk_queue_max_hw_sectors(q, max);
 	blk_queue_max_segments(q, USHRT_MAX);
 	/* with page sized segments we can translate each segement into
@@ -3177,6 +3205,21 @@ static void dasd_setup_queue(struct dasd_block *block)
 	 */
 	blk_queue_max_segment_size(q, PAGE_SIZE);
 	blk_queue_segment_boundary(q, PAGE_SIZE - 1);
+
+	/* Only activate blocklayer discard support for devices that support it */
+	if (block->base->features & DASD_FEATURE_DISCARD) {
+		q->limits.discard_granularity = logical_block_size;
+		q->limits.discard_alignment = PAGE_SIZE;
+
+		/* Calculate max_discard_sectors and make it PAGE aligned */
+		max_bytes = USHRT_MAX * logical_block_size;
+		max_bytes = ALIGN(max_bytes, PAGE_SIZE) - PAGE_SIZE;
+		max_discard_sectors = max_bytes / logical_block_size;
+
+		blk_queue_max_discard_sectors(q, max_discard_sectors);
+		blk_queue_max_write_zeroes_sectors(q, max_discard_sectors);
+		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);
+	}
 }
 
 /*

drivers/s390/block/dasd_3990_erp.c

@@ -2231,7 +2231,7 @@ static void dasd_3990_erp_account_error(struct dasd_ccw_req *erp)
 	struct dasd_device *device = erp->startdev;
 	__u8 lpum = erp->refers->irb.esw.esw1.lpum;
 	int pos = pathmask_to_pos(lpum);
-	unsigned long long clk;
+	unsigned long clk;
 
 	if (!device->path_thrhld)
 		return;

drivers/s390/block/dasd_devmap.c

@@ -1634,7 +1634,7 @@ static struct attribute * dasd_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group dasd_attr_group = {
+static const struct attribute_group dasd_attr_group = {
 	.attrs = dasd_attrs,
 };
 
@@ -1676,6 +1676,7 @@ dasd_set_feature(struct ccw_device *cdev, int feature, int flag)
 	spin_unlock(&dasd_devmap_lock);
 	return 0;
 }
+EXPORT_SYMBOL(dasd_set_feature);
 
 
 int

drivers/s390/block/dasd_diag.c

@@ -235,7 +235,7 @@ static void dasd_ext_handler(struct ext_code ext_code,
 {
 	struct dasd_ccw_req *cqr, *next;
 	struct dasd_device *device;
-	unsigned long long expires;
+	unsigned long expires;
 	unsigned long flags;
 	addr_t ip;
 	int rc;

drivers/s390/block/dasd_eckd.c

@@ -3254,11 +3254,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_track(
 	/* 1x prefix + one read/write ccw per track */
 	cplength = 1 + trkcount;
 
-	/* on 31-bit we need space for two 32 bit addresses per page
-	 * on 64-bit one 64 bit address
-	 */
-	datasize = sizeof(struct PFX_eckd_data) +
-		cidaw * sizeof(unsigned long long);
+	datasize = sizeof(struct PFX_eckd_data) + cidaw * sizeof(unsigned long);
 
 	/* Allocate the ccw request. */
 	cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength, datasize,
@@ -3856,7 +3852,7 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_raw(struct dasd_device *startdev,
 	}
 	size = ALIGN(size, 8);
 
-	datasize = size + cidaw * sizeof(unsigned long long);
+	datasize = size + cidaw * sizeof(unsigned long);
 
 	/* Allocate the ccw request. */
 	cqr = dasd_smalloc_request(DASD_ECKD_MAGIC, cplength,

drivers/s390/block/dasd_eckd.h

@@ -165,7 +165,7 @@ struct DE_eckd_data {
 	__u8 ga_extended;	/* Global Attributes Extended	*/
 	struct ch_t beg_ext;
 	struct ch_t end_ext;
-	unsigned long long ep_sys_time; /* Ext Parameter - System Time Stamp */
+	unsigned long ep_sys_time; /* Ext Parameter - System Time Stamp */
 	__u8 ep_format;        /* Extended Parameter format byte       */
 	__u8 ep_prio;          /* Extended Parameter priority I/O byte */
 	__u8 ep_reserved1;     /* Extended Parameter Reserved	       */

drivers/s390/block/dasd_erp.c

@@ -124,7 +124,7 @@ dasd_default_erp_action(struct dasd_ccw_req *cqr)
 struct dasd_ccw_req *dasd_default_erp_postaction(struct dasd_ccw_req *cqr)
 {
 	int success;
-	unsigned long long startclk, stopclk;
+	unsigned long startclk, stopclk;
 	struct dasd_device *startdev;
 
 	BUG_ON(cqr->refers == NULL || cqr->function == NULL);

drivers/s390/block/dasd_fba.c

@@ -174,6 +174,9 @@ dasd_fba_check_characteristics(struct dasd_device *device)
 	if (readonly)
 		set_bit(DASD_FLAG_DEVICE_RO, &device->flags);
 
+	/* FBA supports discard, set the according feature bit */
+	dasd_set_feature(cdev, DASD_FEATURE_DISCARD, 1);
+
 	dev_info(&device->cdev->dev,
 		 "New FBA DASD %04X/%02X (CU %04X/%02X) with %d MB "
 		 "and %d B/blk%s\n",
@@ -247,9 +250,192 @@ static void dasd_fba_check_for_device_change(struct dasd_device *device,
 		dasd_generic_handle_state_change(device);
 };
 
-static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev,
-					      struct dasd_block *block,
-					      struct request *req)
+
+/*
+ * Builds a CCW with no data payload
+ */
+static void ccw_write_no_data(struct ccw1 *ccw)
+{
+	ccw->cmd_code = DASD_FBA_CCW_WRITE;
+	ccw->flags |= CCW_FLAG_SLI;
+	ccw->count = 0;
+}
+
+/*
+ * Builds a CCW that writes only zeroes.
+ */
+static void ccw_write_zero(struct ccw1 *ccw, int count)
+{
+	ccw->cmd_code = DASD_FBA_CCW_WRITE;
+	ccw->flags |= CCW_FLAG_SLI;
+	ccw->count = count;
+	ccw->cda = (__u32) (addr_t) page_to_phys(ZERO_PAGE(0));
+}
+
+/*
+ * Helper function to count the amount of necessary CCWs within a given range
+ * with 4k alignment and command chaining in mind.
+ */
+static int count_ccws(sector_t first_rec, sector_t last_rec,
+		      unsigned int blocks_per_page)
+{
+	sector_t wz_stop = 0, d_stop = 0;
+	int cur_pos = 0;
+	int count = 0;
+
+	if (first_rec % blocks_per_page != 0) {
+		wz_stop = first_rec + blocks_per_page -
+			(first_rec % blocks_per_page) - 1;
+		if (wz_stop > last_rec)
+			wz_stop = last_rec;
+		cur_pos = wz_stop - first_rec + 1;
+		count++;
+	}
+
+	if (last_rec - (first_rec + cur_pos) + 1 >= blocks_per_page) {
+		if ((last_rec - blocks_per_page + 1) % blocks_per_page != 0)
+			d_stop = last_rec - ((last_rec - blocks_per_page + 1) %
+					     blocks_per_page);
+		else
+			d_stop = last_rec;
+
+		cur_pos += d_stop - (first_rec + cur_pos) + 1;
+		count++;
+	}
+
+	if (cur_pos == 0 || first_rec + cur_pos - 1 < last_rec)
+		count++;
+
+	return count;
+}
+
+/*
+ * This function builds a CCW request for block layer discard requests.
+ * Each page in the z/VM hypervisor that represents certain records of an FBA
+ * device will be padded with zeros. This is a special behaviour of the WRITE
+ * command which is triggered when no data payload is added to the CCW.
+ *
+ * Note: Due to issues in some z/VM versions, we can't fully utilise this
+ * special behaviour. We have to keep a 4k (or 8 block) alignment in mind to
+ * work around those issues and write actual zeroes to the unaligned parts in
+ * the request. This workaround might be removed in the future.
+ */
+static struct dasd_ccw_req *dasd_fba_build_cp_discard(
+						struct dasd_device *memdev,
+						struct dasd_block *block,
+						struct request *req)
+{
+	struct LO_fba_data *LO_data;
+	struct dasd_ccw_req *cqr;
+	struct ccw1 *ccw;
+
+	sector_t wz_stop = 0, d_stop = 0;
+	sector_t first_rec, last_rec;
+
+	unsigned int blksize = block->bp_block;
+	unsigned int blocks_per_page;
+	int wz_count = 0;
+	int d_count = 0;
+	int cur_pos = 0; /* Current position within the extent */
+	int count = 0;
+	int cplength;
+	int datasize;
+	int nr_ccws;
+
+	first_rec = blk_rq_pos(req) >> block->s2b_shift;
+	last_rec =
+		(blk_rq_pos(req) + blk_rq_sectors(req) - 1) >> block->s2b_shift;
+	count = last_rec - first_rec + 1;
+
+	blocks_per_page = BLOCKS_PER_PAGE(blksize);
+	nr_ccws = count_ccws(first_rec, last_rec, blocks_per_page);
+
+	/* define extent + nr_ccws * locate record + nr_ccws * single CCW */
+	cplength = 1 + 2 * nr_ccws;
+	datasize = sizeof(struct DE_fba_data) +
+		nr_ccws * (sizeof(struct LO_fba_data) + sizeof(struct ccw1));
+
+	cqr = dasd_smalloc_request(DASD_FBA_MAGIC, cplength, datasize, memdev);
+	if (IS_ERR(cqr))
+		return cqr;
+
+	ccw = cqr->cpaddr;
+
+	define_extent(ccw++, cqr->data, WRITE, blksize, first_rec, count);
+	LO_data = cqr->data + sizeof(struct DE_fba_data);
+
+	/* First part is not aligned. Calculate range to write zeroes. */
+	if (first_rec % blocks_per_page != 0) {
+		wz_stop = first_rec + blocks_per_page -
+			(first_rec % blocks_per_page) - 1;
+		if (wz_stop > last_rec)
+			wz_stop = last_rec;
+		wz_count = wz_stop - first_rec + 1;
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		locate_record(ccw++, LO_data++, WRITE, cur_pos, wz_count);
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		ccw_write_zero(ccw++, wz_count * blksize);
+
+		cur_pos = wz_count;
+	}
+
+	/* We can do proper discard when we've got at least blocks_per_page blocks. */
+	if (last_rec - (first_rec + cur_pos) + 1 >= blocks_per_page) {
+		/* is last record at page boundary? */
+		if ((last_rec - blocks_per_page + 1) % blocks_per_page != 0)
+			d_stop = last_rec - ((last_rec - blocks_per_page + 1) %
+					     blocks_per_page);
+		else
+			d_stop = last_rec;
+
+		d_count = d_stop - (first_rec + cur_pos) + 1;
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		locate_record(ccw++, LO_data++, WRITE, cur_pos, d_count);
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		ccw_write_no_data(ccw++);
+
+		cur_pos += d_count;
+	}
+
+	/* We might still have some bits left which need to be zeroed. */
+	if (cur_pos == 0 || first_rec + cur_pos - 1 < last_rec) {
+		if (d_stop != 0)
+			wz_count = last_rec - d_stop;
+		else if (wz_stop != 0)
+			wz_count = last_rec - wz_stop;
+		else
+			wz_count = count;
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		locate_record(ccw++, LO_data++, WRITE, cur_pos, wz_count);
+
+		ccw[-1].flags |= CCW_FLAG_CC;
+		ccw_write_zero(ccw++, wz_count * blksize);
+	}
+
+	if (blk_noretry_request(req) ||
+	    block->base->features & DASD_FEATURE_FAILFAST)
+		set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
+
+	cqr->startdev = memdev;
+	cqr->memdev = memdev;
+	cqr->block = block;
+	cqr->expires = memdev->default_expires * HZ;	/* default 5 minutes */
+	cqr->retries = memdev->default_retries;
+	cqr->buildclk = get_tod_clock();
+	cqr->status = DASD_CQR_FILLED;
+
+	return cqr;
+}
+
+static struct dasd_ccw_req *dasd_fba_build_cp_regular(
+						struct dasd_device *memdev,
+						struct dasd_block *block,
+						struct request *req)
 {
 	struct dasd_fba_private *private = block->base->private;
 	unsigned long *idaws;
@@ -372,6 +558,16 @@ static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev,
 	return cqr;
 }
 
+static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device *memdev,
+					      struct dasd_block *block,
+					      struct request *req)
+{
+	if (req_op(req) == REQ_OP_DISCARD || req_op(req) == REQ_OP_WRITE_ZEROES)
+		return dasd_fba_build_cp_discard(memdev, block, req);
+	else
+		return dasd_fba_build_cp_regular(memdev, block, req);
+}
+
 static int
 dasd_fba_free_cp(struct dasd_ccw_req *cqr, struct request *req)
 {

drivers/s390/block/dasd_int.h

@@ -167,6 +167,9 @@ do { \
 	printk(d_loglevel PRINTK_HEADER " " d_string "\n", d_args); \
 } while(0)
 
+/* Macro to calculate number of blocks per page */
+#define BLOCKS_PER_PAGE(blksize) (PAGE_SIZE / blksize)
+
 struct dasd_ccw_req {
 	unsigned int magic;		/* Eye catcher */
 	struct list_head devlist;	/* for dasd_device request queue */
@@ -196,10 +199,10 @@ struct dasd_ccw_req {
 	void *function; 		/* originating ERP action */
 
 	/* these are for statistics only */
-	unsigned long long buildclk;	/* TOD-clock of request generation */
-	unsigned long long startclk;	/* TOD-clock of request start */
-	unsigned long long stopclk;	/* TOD-clock of request interrupt */
-	unsigned long long endclk;	/* TOD-clock of request termination */
+	unsigned long buildclk;		/* TOD-clock of request generation */
+	unsigned long startclk;		/* TOD-clock of request start */
+	unsigned long stopclk;		/* TOD-clock of request interrupt */
+	unsigned long endclk;		/* TOD-clock of request termination */
 
         /* Callback that is called after reaching final status. */
 	void (*callback)(struct dasd_ccw_req *, void *data);
@@ -423,7 +426,7 @@ struct dasd_path {
 	u8 chpid;
 	struct dasd_conf_data *conf_data;
 	atomic_t error_count;
-	unsigned long long errorclk;
+	unsigned long errorclk;
 };
 
 
@@ -454,6 +457,10 @@ struct dasd_profile_info {
 	unsigned int dasd_read_time2[32];  /* hist. of time from start to irq */
 	unsigned int dasd_read_time3[32];  /* hist. of time from irq to end */
 	unsigned int dasd_read_nr_req[32]; /* hist. of # of requests in chanq */
+	unsigned long dasd_sum_times;	   /* sum of request times */
+	unsigned long dasd_sum_time_str;   /* sum of time from build to start */
+	unsigned long dasd_sum_time_irq;   /* sum of time from start to irq */
+	unsigned long dasd_sum_time_end;   /* sum of time from irq to end */
 };
 
 struct dasd_profile {
@@ -535,7 +542,7 @@ struct dasd_block {
 	struct block_device *bdev;
 	atomic_t open_count;
 
-	unsigned long long blocks; /* size of volume in blocks */
+	unsigned long blocks;	   /* size of volume in blocks */
 	unsigned int bp_block;	   /* bytes per block */
 	unsigned int s2b_shift;	   /* log2 (bp_block/512) */
 

drivers/s390/block/dasd_proc.c

@@ -90,7 +90,7 @@ dasd_devices_show(struct seq_file *m, void *v)
 			seq_printf(m, "n/f	 ");
 		else
 			seq_printf(m,
-				   "at blocksize: %d, %lld blocks, %lld MB",
+				   "at blocksize: %u, %lu blocks, %lu MB",
 				   block->bp_block, block->blocks,
 				   ((block->bp_block >> 9) *
 				    block->blocks) >> 11);

drivers/s390/block/scm_blk.c

@@ -249,13 +249,13 @@ static void scm_request_requeue(struct scm_request *scmrq)
 static void scm_request_finish(struct scm_request *scmrq)
 {
 	struct scm_blk_dev *bdev = scmrq->bdev;
+	int *error;
 	int i;
 
 	for (i = 0; i < nr_requests_per_io && scmrq->request[i]; i++) {
-		if (scmrq->error)
-			blk_mq_end_request(scmrq->request[i], scmrq->error);
-		else
-			blk_mq_complete_request(scmrq->request[i]);
+		error = blk_mq_rq_to_pdu(scmrq->request[i]);
+		*error = scmrq->error;
+		blk_mq_complete_request(scmrq->request[i]);
 	}
 
 	atomic_dec(&bdev->queued_reqs);
@@ -415,7 +415,9 @@ void scm_blk_irq(struct scm_device *scmdev, void *data, blk_status_t error)
 
 static void scm_blk_request_done(struct request *req)
 {
-	blk_mq_end_request(req, 0);
+	int *error = blk_mq_rq_to_pdu(req);
+
+	blk_mq_end_request(req, *error);
 }
 
 static const struct block_device_operations scm_blk_devops = {
@@ -448,6 +450,7 @@ int scm_blk_dev_setup(struct scm_blk_dev *bdev, struct scm_device *scmdev)
 	atomic_set(&bdev->queued_reqs, 0);
 
 	bdev->tag_set.ops = &scm_mq_ops;
+	bdev->tag_set.cmd_size = sizeof(int);
 	bdev->tag_set.nr_hw_queues = nr_requests;
 	bdev->tag_set.queue_depth = nr_requests_per_io * nr_requests;
 	bdev->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;

drivers/s390/char/Kconfig

@@ -169,10 +169,21 @@ config VMCP
 	def_bool y
 	prompt "Support for the z/VM CP interface"
 	depends on S390
+	select CMA
 	help
 	  Select this option if you want to be able to interact with the control
 	  program on z/VM
 
+config VMCP_CMA_SIZE
+	int "Memory in MiB reserved for z/VM CP interface"
+	default "4"
+	depends on VMCP
+	help
+	  Specify the default amount of memory in MiB reserved for the z/VM CP
+	  interface. If needed this memory is used for large contiguous memory
+	  allocations. The default can be changed with the kernel command line
+	  parameter "vmcp_cma".
+
 config MONREADER
 	def_tristate m
 	prompt "API for reading z/VM monitor service records"

drivers/s390/char/raw3270.c

@@ -1082,7 +1082,7 @@ static struct attribute * raw3270_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group raw3270_attr_group = {
+static const struct attribute_group raw3270_attr_group = {
 	.attrs = raw3270_attrs,
 };
 

drivers/s390/char/sclp_cmd.c

@@ -252,6 +252,7 @@ static int sclp_attach_storage(u8 id)
 	if (!sccb)
 		return -ENOMEM;
 	sccb->header.length = PAGE_SIZE;
+	sccb->header.function_code = 0x40;
 	rc = sclp_sync_request_timeout(0x00080001 | id << 8, sccb,
 				       SCLP_QUEUE_INTERVAL);
 	if (rc)

drivers/s390/char/sclp_config.c

@@ -135,7 +135,7 @@ static ssize_t sysfs_ofb_data_write(struct file *filp, struct kobject *kobj,
 	return rc ?: count;
 }
 
-static struct bin_attribute ofb_bin_attr = {
+static const struct bin_attribute ofb_bin_attr = {
 	.attr = {
 		.name = "event_data",
 		.mode = S_IWUSR,

drivers/s390/char/sclp_early.c

@@ -39,7 +39,7 @@ struct read_info_sccb {
 	u8	fac84;			/* 84 */
 	u8	fac85;			/* 85 */
 	u8	_pad_86[91 - 86];	/* 86-90 */
-	u8	flags;			/* 91 */
+	u8	fac91;			/* 91 */
 	u8	_pad_92[98 - 92];	/* 92-97 */
 	u8	fac98;			/* 98 */
 	u8	hamaxpow;		/* 99 */
@@ -103,6 +103,8 @@ static void __init sclp_early_facilities_detect(struct read_info_sccb *sccb)
 	sclp.has_kss = !!(sccb->fac98 & 0x01);
 	if (sccb->fac85 & 0x02)
 		S390_lowcore.machine_flags |= MACHINE_FLAG_ESOP;
+	if (sccb->fac91 & 0x40)
+		S390_lowcore.machine_flags |= MACHINE_FLAG_TLB_GUEST;
 	sclp.rnmax = sccb->rnmax ? sccb->rnmax : sccb->rnmax2;
 	sclp.rzm = sccb->rnsize ? sccb->rnsize : sccb->rnsize2;
 	sclp.rzm <<= 20;
@@ -139,7 +141,7 @@ static void __init sclp_early_facilities_detect(struct read_info_sccb *sccb)
 
 	/* Save IPL information */
 	sclp_ipl_info.is_valid = 1;
-	if (sccb->flags & 0x2)
+	if (sccb->fac91 & 0x2)
 		sclp_ipl_info.has_dump = 1;
 	memcpy(&sclp_ipl_info.loadparm, &sccb->loadparm, LOADPARM_LEN);
 

drivers/s390/char/sclp_ocf.c

@@ -126,7 +126,7 @@ static struct attribute *ocf_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group ocf_attr_group = {
+static const struct attribute_group ocf_attr_group = {
 	.attrs = ocf_attrs,
 };
 

drivers/s390/char/tape_core.c

@@ -175,7 +175,7 @@ static struct attribute *tape_attrs[] = {
 	NULL
 };
 
-static struct attribute_group tape_attr_group = {
+static const struct attribute_group tape_attr_group = {
 	.attrs = tape_attrs,
 };
 

drivers/s390/char/vmcp.c

@@ -17,15 +17,85 @@
 #include <linux/kernel.h>
 #include <linux/miscdevice.h>
 #include <linux/slab.h>
+#include <linux/uaccess.h>
 #include <linux/export.h>
+#include <linux/mutex.h>
+#include <linux/cma.h>
+#include <linux/mm.h>
 #include <asm/compat.h>
 #include <asm/cpcmd.h>
 #include <asm/debug.h>
-#include <linux/uaccess.h>
-#include "vmcp.h"
+#include <asm/vmcp.h>
+
+struct vmcp_session {
+	char *response;
+	unsigned int bufsize;
+	unsigned int cma_alloc : 1;
+	int resp_size;
+	int resp_code;
+	struct mutex mutex;
+};
 
 static debug_info_t *vmcp_debug;
 
+static unsigned long vmcp_cma_size __initdata = CONFIG_VMCP_CMA_SIZE * 1024 * 1024;
+static struct cma *vmcp_cma;
+
+static int __init early_parse_vmcp_cma(char *p)
+{
+	vmcp_cma_size = ALIGN(memparse(p, NULL), PAGE_SIZE);
+	return 0;
+}
+early_param("vmcp_cma", early_parse_vmcp_cma);
+
+void __init vmcp_cma_reserve(void)
+{
+	if (!MACHINE_IS_VM)
+		return;
+	cma_declare_contiguous(0, vmcp_cma_size, 0, 0, 0, false, "vmcp", &vmcp_cma);
+}
+
+static void vmcp_response_alloc(struct vmcp_session *session)
+{
+	struct page *page = NULL;
+	int nr_pages, order;
+
+	order = get_order(session->bufsize);
+	nr_pages = ALIGN(session->bufsize, PAGE_SIZE) >> PAGE_SHIFT;
+	/*
+	 * For anything below order 3 allocations rely on the buddy
+	 * allocator. If such low-order allocations can't be handled
+	 * anymore the system won't work anyway.
+	 */
+	if (order > 2)
+		page = cma_alloc(vmcp_cma, nr_pages, 0, GFP_KERNEL);
+	if (page) {
+		session->response = (char *)page_to_phys(page);
+		session->cma_alloc = 1;
+		return;
+	}
+	session->response = (char *)__get_free_pages(GFP_KERNEL | __GFP_RETRY_MAYFAIL, order);
+}
+
+static void vmcp_response_free(struct vmcp_session *session)
+{
+	int nr_pages, order;
+	struct page *page;
+
+	if (!session->response)
+		return;
+	order = get_order(session->bufsize);
+	nr_pages = ALIGN(session->bufsize, PAGE_SIZE) >> PAGE_SHIFT;
+	if (session->cma_alloc) {
+		page = phys_to_page((unsigned long)session->response);
+		cma_release(vmcp_cma, page, nr_pages);
+		session->cma_alloc = 0;
+	} else {
+		free_pages((unsigned long)session->response, order);
+	}
+	session->response = NULL;
+}
+
 static int vmcp_open(struct inode *inode, struct file *file)
 {
 	struct vmcp_session *session;
@@ -51,7 +121,7 @@ static int vmcp_release(struct inode *inode, struct file *file)
 
 	session = file->private_data;
 	file->private_data = NULL;
-	free_pages((unsigned long)session->response, get_order(session->bufsize));
+	vmcp_response_free(session);
 	kfree(session);
 	return 0;
 }
@@ -97,9 +167,7 @@ vmcp_write(struct file *file, const char __user *buff, size_t count,
 		return -ERESTARTSYS;
 	}
 	if (!session->response)
-		session->response = (char *)__get_free_pages(GFP_KERNEL
-						| __GFP_RETRY_MAYFAIL | GFP_DMA,
-						get_order(session->bufsize));
+		vmcp_response_alloc(session);
 	if (!session->response) {
 		mutex_unlock(&session->mutex);
 		kfree(cmd);
@@ -130,8 +198,8 @@ vmcp_write(struct file *file, const char __user *buff, size_t count,
 static long vmcp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 {
 	struct vmcp_session *session;
+	int ret = -ENOTTY;
 	int __user *argp;
-	int temp;
 
 	session = file->private_data;
 	if (is_compat_task())
@@ -142,28 +210,26 @@ static long vmcp_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
 		return -ERESTARTSYS;
 	switch (cmd) {
 	case VMCP_GETCODE:
-		temp = session->resp_code;
-		mutex_unlock(&session->mutex);
-		return put_user(temp, argp);
+		ret = put_user(session->resp_code, argp);
+		break;
 	case VMCP_SETBUF:
-		free_pages((unsigned long)session->response,
-				get_order(session->bufsize));
-		session->response=NULL;
-		temp = get_user(session->bufsize, argp);
-		if (get_order(session->bufsize) > 8) {
+		vmcp_response_free(session);
+		ret = get_user(session->bufsize, argp);
+		if (ret)
 			session->bufsize = PAGE_SIZE;
-			temp = -EINVAL;
+		if (!session->bufsize || get_order(session->bufsize) > 8) {
+			session->bufsize = PAGE_SIZE;
+			ret = -EINVAL;
 		}
-		mutex_unlock(&session->mutex);
-		return temp;
+		break;
 	case VMCP_GETSIZE:
-		temp = session->resp_size;
-		mutex_unlock(&session->mutex);
-		return put_user(temp, argp);
+		ret = put_user(session->resp_size, argp);
+		break;
 	default:
-		mutex_unlock(&session->mutex);
-		return -ENOIOCTLCMD;
+		break;
 	}
+	mutex_unlock(&session->mutex);
+	return ret;
 }
 
 static const struct file_operations vmcp_fops = {

drivers/s390/cio/chp.c

@@ -143,7 +143,7 @@ static ssize_t chp_measurement_chars_read(struct file *filp,
 				       sizeof(chp->cmg_chars));
 }
 
-static struct bin_attribute chp_measurement_chars_attr = {
+static const struct bin_attribute chp_measurement_chars_attr = {
 	.attr = {
 		.name = "measurement_chars",
 		.mode = S_IRUSR,
@@ -197,7 +197,7 @@ static ssize_t chp_measurement_read(struct file *filp, struct kobject *kobj,
 	return count;
 }
 
-static struct bin_attribute chp_measurement_attr = {
+static const struct bin_attribute chp_measurement_attr = {
 	.attr = {
 		.name = "measurement",
 		.mode = S_IRUSR,

drivers/s390/cio/device.c

@@ -612,7 +612,7 @@ static struct attribute *io_subchannel_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group io_subchannel_attr_group = {
+static const struct attribute_group io_subchannel_attr_group = {
 	.attrs = io_subchannel_attrs,
 };
 
@@ -626,7 +626,7 @@ static struct attribute * ccwdev_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group ccwdev_attr_group = {
+static const struct attribute_group ccwdev_attr_group = {
 	.attrs = ccwdev_attrs,
 };
 

drivers/s390/crypto/zcrypt_card.c

@@ -98,7 +98,7 @@ static struct attribute *zcrypt_card_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group zcrypt_card_attr_group = {
+static const struct attribute_group zcrypt_card_attr_group = {
 	.attrs = zcrypt_card_attrs,
 };
 

drivers/s390/crypto/zcrypt_msgtype6.c

@@ -140,7 +140,7 @@ struct function_and_rules_block {
  *   + 0x000A 'MRP     ' (MCL3 'PK' or CEX2C 'PK')
  * - VUD block
  */
-static struct CPRBX static_cprbx = {
+static const struct CPRBX static_cprbx = {
 	.cprb_len	=  0x00DC,
 	.cprb_ver_id	=  0x02,
 	.func_id	= {0x54, 0x32},

drivers/s390/crypto/zcrypt_queue.c

@@ -89,7 +89,7 @@ static struct attribute *zcrypt_queue_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group zcrypt_queue_attr_group = {
+static const struct attribute_group zcrypt_queue_attr_group = {
 	.attrs = zcrypt_queue_attrs,
 };
 

drivers/s390/net/qeth_l3_sys.c

@@ -350,7 +350,7 @@ static struct attribute *qeth_l3_device_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group qeth_l3_device_attr_group = {
+static const struct attribute_group qeth_l3_device_attr_group = {
 	.attrs = qeth_l3_device_attrs,
 };
 
@@ -680,7 +680,7 @@ static struct attribute *qeth_ipato_device_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group qeth_device_ipato_group = {
+static const struct attribute_group qeth_device_ipato_group = {
 	.name = "ipa_takeover",
 	.attrs = qeth_ipato_device_attrs,
 };
@@ -843,7 +843,7 @@ static struct attribute *qeth_vipa_device_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group qeth_device_vipa_group = {
+static const struct attribute_group qeth_device_vipa_group = {
 	.name = "vipa",
 	.attrs = qeth_vipa_device_attrs,
 };
@@ -1006,7 +1006,7 @@ static struct attribute *qeth_rxip_device_attrs[] = {
 	NULL,
 };
 
-static struct attribute_group qeth_device_rxip_group = {
+static const struct attribute_group qeth_device_rxip_group = {
 	.name = "rxip",
 	.attrs = qeth_rxip_device_attrs,
 };