Merge branch 'master' into for-linus

62 files changed, 533 insertions, 392 deletions

diff --git a/arch/alpha/include/asm/tlb.h b/arch/alpha/include/asm/tlb.h
index c13636575fb..42866759f3f 100644
--- a/arch/alpha/include/asm/tlb.h
+++ b/arch/alpha/include/asm/tlb.h
@@ -9,7 +9,7 @@
 
 #include <asm-generic/tlb.h>
 
-#define __pte_free_tlb(tlb, pte)			pte_free((tlb)->mm, pte)
-#define __pmd_free_tlb(tlb, pmd)			pmd_free((tlb)->mm, pmd)
+#define __pte_free_tlb(tlb, pte, address)		pte_free((tlb)->mm, pte)
+#define __pmd_free_tlb(tlb, pmd, address)		pmd_free((tlb)->mm, pmd)
  
 #endif
diff --git a/arch/arm/include/asm/tlb.h b/arch/arm/include/asm/tlb.h
index 321c83e43a1..f41a6f57cd1 100644
--- a/arch/arm/include/asm/tlb.h
+++ b/arch/arm/include/asm/tlb.h
@@ -102,8 +102,8 @@ tlb_end_vma(struct mmu_gather *tlb, struct vm_area_struct *vma)
 }
 
 #define tlb_remove_page(tlb,page)	free_page_and_swap_cache(page)
-#define pte_free_tlb(tlb, ptep)		pte_free((tlb)->mm, ptep)
-#define pmd_free_tlb(tlb, pmdp)		pmd_free((tlb)->mm, pmdp)
+#define pte_free_tlb(tlb, ptep, addr)	pte_free((tlb)->mm, ptep)
+#define pmd_free_tlb(tlb, pmdp, addr)	pmd_free((tlb)->mm, pmdp)
 
 #define tlb_migrate_finish(mm)		do { } while (0)
 
diff --git a/arch/avr32/include/asm/pgalloc.h b/arch/avr32/include/asm/pgalloc.h
index 64082132394..92ecd8446ef 100644
--- a/arch/avr32/include/asm/pgalloc.h
+++ b/arch/avr32/include/asm/pgalloc.h
@@ -83,7 +83,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	quicklist_free_page(QUICK_PT, NULL, pte);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,addr)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb), pte);			\
diff --git a/arch/cris/include/asm/pgalloc.h b/arch/cris/include/asm/pgalloc.h
index a1ba761d057..6da975db112 100644
--- a/arch/cris/include/asm/pgalloc.h
+++ b/arch/cris/include/asm/pgalloc.h
@@ -47,7 +47,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	__free_page(pte);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,address)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb), pte);			\
diff --git a/arch/frv/include/asm/pgalloc.h b/arch/frv/include/asm/pgalloc.h
index 971e6addb00..416d19a632f 100644
--- a/arch/frv/include/asm/pgalloc.h
+++ b/arch/frv/include/asm/pgalloc.h
@@ -49,7 +49,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	__free_page(pte);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,address)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb),(pte));			\
@@ -62,7 +62,7 @@ do {							\
  */
 #define pmd_alloc_one(mm, addr)		({ BUG(); ((pmd_t *) 2); })
 #define pmd_free(mm, x)			do { } while (0)
-#define __pmd_free_tlb(tlb,x)		do { } while (0)
+#define __pmd_free_tlb(tlb,x,a)		do { } while (0)
 
 #endif /* CONFIG_MMU */
 
diff --git a/arch/frv/include/asm/pgtable.h b/arch/frv/include/asm/pgtable.h
index 33233011b1c..22c60692b55 100644
--- a/arch/frv/include/asm/pgtable.h
+++ b/arch/frv/include/asm/pgtable.h
@@ -225,7 +225,7 @@ static inline pud_t *pud_offset(pgd_t *pgd, unsigned long address)
  */
 #define pud_alloc_one(mm, address)		NULL
 #define pud_free(mm, x)				do { } while (0)
-#define __pud_free_tlb(tlb, x)			do { } while (0)
+#define __pud_free_tlb(tlb, x, address)		do { } while (0)
 
 /*
  * The "pud_xxx()" functions here are trivial for a folded two-level
diff --git a/arch/ia64/include/asm/pgalloc.h b/arch/ia64/include/asm/pgalloc.h
index b9ac1a6fc21..96a8d927db2 100644
--- a/arch/ia64/include/asm/pgalloc.h
+++ b/arch/ia64/include/asm/pgalloc.h
@@ -48,7 +48,7 @@ static inline void pud_free(struct mm_struct *mm, pud_t *pud)
 {
 	quicklist_free(0, NULL, pud);
 }
-#define __pud_free_tlb(tlb, pud)	pud_free((tlb)->mm, pud)
+#define __pud_free_tlb(tlb, pud, address)	pud_free((tlb)->mm, pud)
 #endif /* CONFIG_PGTABLE_4 */
 
 static inline void
@@ -67,7 +67,7 @@ static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
 	quicklist_free(0, NULL, pmd);
 }
 
-#define __pmd_free_tlb(tlb, pmd)	pmd_free((tlb)->mm, pmd)
+#define __pmd_free_tlb(tlb, pmd, address)	pmd_free((tlb)->mm, pmd)
 
 static inline void
 pmd_populate(struct mm_struct *mm, pmd_t * pmd_entry, pgtable_t pte)
@@ -117,6 +117,6 @@ static inline void check_pgt_cache(void)
 	quicklist_trim(0, NULL, 25, 16);
 }
 
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, pte)
+#define __pte_free_tlb(tlb, pte, address)	pte_free((tlb)->mm, pte)
 
 #endif				/* _ASM_IA64_PGALLOC_H */
diff --git a/arch/ia64/include/asm/tlb.h b/arch/ia64/include/asm/tlb.h
index 20d8a39680c..85d965cb19a 100644
--- a/arch/ia64/include/asm/tlb.h
+++ b/arch/ia64/include/asm/tlb.h
@@ -236,22 +236,22 @@ do {							\
 	__tlb_remove_tlb_entry(tlb, ptep, addr);	\
 } while (0)
 
-#define pte_free_tlb(tlb, ptep)				\
+#define pte_free_tlb(tlb, ptep, address)		\
 do {							\
 	tlb->need_flush = 1;				\
-	__pte_free_tlb(tlb, ptep);			\
+	__pte_free_tlb(tlb, ptep, address);		\
 } while (0)
 
-#define pmd_free_tlb(tlb, ptep)				\
+#define pmd_free_tlb(tlb, ptep, address)		\
 do {							\
 	tlb->need_flush = 1;				\
-	__pmd_free_tlb(tlb, ptep);			\
+	__pmd_free_tlb(tlb, ptep, address);		\
 } while (0)
 
-#define pud_free_tlb(tlb, pudp)				\
+#define pud_free_tlb(tlb, pudp, address)		\
 do {							\
 	tlb->need_flush = 1;				\
-	__pud_free_tlb(tlb, pudp);			\
+	__pud_free_tlb(tlb, pudp, address);		\
 } while (0)
 
 #endif /* _ASM_IA64_TLB_H */
diff --git a/arch/m32r/include/asm/pgalloc.h b/arch/m32r/include/asm/pgalloc.h
index f11a2b909cd..0fc73619897 100644
--- a/arch/m32r/include/asm/pgalloc.h
+++ b/arch/m32r/include/asm/pgalloc.h
@@ -58,7 +58,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	__free_page(pte);
 }
 
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, (pte))
+#define __pte_free_tlb(tlb, pte, addr)	pte_free((tlb)->mm, (pte))
 
 /*
  * allocating and freeing a pmd is trivial: the 1-entry pmd is
@@ -68,7 +68,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 
 #define pmd_alloc_one(mm, addr)		({ BUG(); ((pmd_t *)2); })
 #define pmd_free(mm, x)			do { } while (0)
-#define __pmd_free_tlb(tlb, x)		do { } while (0)
+#define __pmd_free_tlb(tlb, x, addr)	do { } while (0)
 #define pgd_populate(mm, pmd, pte)	BUG()
 
 #define check_pgt_cache()	do { } while (0)
diff --git a/arch/m68k/include/asm/motorola_pgalloc.h b/arch/m68k/include/asm/motorola_pgalloc.h
index d08bf6261df..15ee4c74a9f 100644
--- a/arch/m68k/include/asm/motorola_pgalloc.h
+++ b/arch/m68k/include/asm/motorola_pgalloc.h
@@ -54,7 +54,8 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t page)
 	__free_page(page);
 }
 
-static inline void __pte_free_tlb(struct mmu_gather *tlb, pgtable_t page)
+static inline void __pte_free_tlb(struct mmu_gather *tlb, pgtable_t page,
+				  unsigned long address)
 {
 	pgtable_page_dtor(page);
 	cache_page(kmap(page));
@@ -73,7 +74,8 @@ static inline int pmd_free(struct mm_struct *mm, pmd_t *pmd)
 	return free_pointer_table(pmd);
 }
 
-static inline int __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd)
+static inline int __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
+				 unsigned long address)
 {
 	return free_pointer_table(pmd);
 }
diff --git a/arch/m68k/include/asm/sun3_pgalloc.h b/arch/m68k/include/asm/sun3_pgalloc.h
index d4c83f14381..48d80d5a666 100644
--- a/arch/m68k/include/asm/sun3_pgalloc.h
+++ b/arch/m68k/include/asm/sun3_pgalloc.h
@@ -32,7 +32,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t page)
         __free_page(page);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,addr)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb), pte);			\
@@ -80,7 +80,7 @@ static inline void pmd_populate(struct mm_struct *mm, pmd_t *pmd, pgtable_t page
  * inside the pgd, so has no extra memory associated with it.
  */
 #define pmd_free(mm, x)			do { } while (0)
-#define __pmd_free_tlb(tlb, x)		do { } while (0)
+#define __pmd_free_tlb(tlb, x, addr)	do { } while (0)
 
 static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
 {
diff --git a/arch/microblaze/Makefile b/arch/microblaze/Makefile
index d0bcf80a113..8439598d465 100644
--- a/arch/microblaze/Makefile
+++ b/arch/microblaze/Makefile
@@ -6,14 +6,16 @@ endif
 
 # What CPU vesion are we building for, and crack it open
 # as major.minor.rev
-CPU_VER=$(subst ",,$(CONFIG_XILINX_MICROBLAZE0_HW_VER) )
-CPU_MAJOR=$(shell echo $(CPU_VER) | cut -d '.' -f 1)
-CPU_MINOR=$(shell echo $(CPU_VER) | cut -d '.' -f 2)
-CPU_REV=$(shell echo $(CPU_VER) | cut -d '.' -f 3)
+CPU_VER   := $(shell echo $(CONFIG_XILINX_MICROBLAZE0_HW_VER))
+CPU_MAJOR := $(shell echo $(CPU_VER) | cut -d '.' -f 1)
+CPU_MINOR := $(shell echo $(CPU_VER) | cut -d '.' -f 2)
+CPU_REV   := $(shell echo $(CPU_VER) | cut -d '.' -f 3)
 
 export CPU_VER CPU_MAJOR CPU_MINOR CPU_REV
 
 # Use cpu-related CONFIG_ vars to set compile options.
+# The various CONFIG_XILINX cpu features options are integers 0/1/2...
+# rather than bools y/n
 
 # Work out HW multipler support.  This is icky.
 # 1. Spartan2 has no HW multiplers.
@@ -34,30 +36,29 @@ CPUFLAGS-$(CONFIG_XILINX_MICROBLAZE0_USE_PCMP_INSTR) += -mxl-pattern-compare
 
 CPUFLAGS-1 += $(call cc-option,-mcpu=v$(CPU_VER))
 
-# The various CONFIG_XILINX cpu features options are integers 0/1/2...
-# rather than bools y/n
-
 # r31 holds current when in kernel mode
-CFLAGS_KERNEL += -ffixed-r31 $(CPUFLAGS-1) $(CPUFLAGS-2)
+KBUILD_KERNEL += -ffixed-r31 $(CPUFLAGS-1) $(CPUFLAGS-2)
 
 LDFLAGS		:=
 LDFLAGS_vmlinux	:=
-LDFLAGS_BLOB := --format binary --oformat elf32-microblaze
 
-LIBGCC := $(shell $(CC) $(CFLAGS_KERNEL) -print-libgcc-file-name)
+LIBGCC := $(shell $(CC) $(KBUILD_KERNEL) -print-libgcc-file-name)
 
-head-y		:= arch/microblaze/kernel/head.o
-libs-y		+= arch/microblaze/lib/ $(LIBGCC)
-core-y		+= arch/microblaze/kernel/ arch/microblaze/mm/ \
-		   arch/microblaze/platform/
+head-y := arch/microblaze/kernel/head.o
+libs-y += arch/microblaze/lib/
+libs-y += $(LIBGCC)
+core-y += arch/microblaze/kernel/
+core-y += arch/microblaze/mm/
+core-y += arch/microblaze/platform/
 
-boot := arch/$(ARCH)/boot
+boot := arch/microblaze/boot
 
 # defines filename extension depending memory management type
 ifeq ($(CONFIG_MMU),)
-MMUEXT		:= -nommu
+MMU := -nommu
 endif
-export	MMUEXT
+
+export MMU
 
 all: linux.bin
 
diff --git a/arch/microblaze/include/asm/io.h b/arch/microblaze/include/asm/io.h
index 5c173424d07..7c3ec13b44d 100644
--- a/arch/microblaze/include/asm/io.h
+++ b/arch/microblaze/include/asm/io.h
@@ -14,7 +14,6 @@
 #include <asm/byteorder.h>
 #include <asm/page.h>
 #include <linux/types.h>
-#include <asm/byteorder.h>
 #include <linux/mm.h>          /* Get struct page {...} */
 
 
diff --git a/arch/microblaze/include/asm/pgalloc.h b/arch/microblaze/include/asm/pgalloc.h
index 59a757e46ba..b0131da1387 100644
--- a/arch/microblaze/include/asm/pgalloc.h
+++ b/arch/microblaze/include/asm/pgalloc.h
@@ -180,7 +180,7 @@ extern inline void pte_free(struct mm_struct *mm, struct page *ptepage)
 	__free_page(ptepage);
 }
 
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, (pte))
+#define __pte_free_tlb(tlb, pte, addr)	pte_free((tlb)->mm, (pte))
 
 #define pmd_populate(mm, pmd, pte)	(pmd_val(*(pmd)) = page_address(pte))
 
@@ -193,7 +193,7 @@ extern inline void pte_free(struct mm_struct *mm, struct page *ptepage)
  */
 #define pmd_alloc_one(mm, address)	({ BUG(); ((pmd_t *)2); })
 /*#define pmd_free(mm, x)			do { } while (0)*/
-#define __pmd_free_tlb(tlb, x)		do { } while (0)
+#define __pmd_free_tlb(tlb, x, addr)	do { } while (0)
 #define pgd_populate(mm, pmd, pte)	BUG()
 
 extern int do_check_pgt_cache(int, int);
diff --git a/arch/microblaze/include/asm/pgtable.h b/arch/microblaze/include/asm/pgtable.h
index 4c57a586a98..cc3a4dfc3ea 100644
--- a/arch/microblaze/include/asm/pgtable.h
+++ b/arch/microblaze/include/asm/pgtable.h
@@ -185,6 +185,7 @@ static inline pte_t pte_mkspecial(pte_t pte)	{ return pte; }
 
 /* Definitions for MicroBlaze. */
 #define	_PAGE_GUARDED	0x001	/* G: page is guarded from prefetch */
+#define _PAGE_FILE	0x001	/* when !present: nonlinear file mapping */
 #define _PAGE_PRESENT	0x002	/* software: PTE contains a translation */
 #define	_PAGE_NO_CACHE	0x004	/* I: caching is inhibited */
 #define	_PAGE_WRITETHRU	0x008	/* W: caching is write-through */
@@ -320,8 +321,7 @@ static inline int pte_write(pte_t pte) { return pte_val(pte) & _PAGE_RW; }
 static inline int pte_exec(pte_t pte)  { return pte_val(pte) & _PAGE_EXEC; }
 static inline int pte_dirty(pte_t pte) { return pte_val(pte) & _PAGE_DIRTY; }
 static inline int pte_young(pte_t pte) { return pte_val(pte) & _PAGE_ACCESSED; }
-/* FIXME */
-static inline int pte_file(pte_t pte)		{ return 0; }
+static inline int pte_file(pte_t pte)  { return pte_val(pte) & _PAGE_FILE; }
 
 static inline void pte_uncache(pte_t pte) { pte_val(pte) |= _PAGE_NO_CACHE; }
 static inline void pte_cache(pte_t pte)   { pte_val(pte) &= ~_PAGE_NO_CACHE; }
@@ -488,7 +488,7 @@ static inline pmd_t *pmd_offset(pgd_t *dir, unsigned long address)
 /* Encode and decode a nonlinear file mapping entry */
 #define PTE_FILE_MAX_BITS	29
 #define pte_to_pgoff(pte)	(pte_val(pte) >> 3)
-#define pgoff_to_pte(off)	((pte_t) { ((off) << 3) })
+#define pgoff_to_pte(off)	((pte_t) { ((off) << 3) | _PAGE_FILE })
 
 extern pgd_t swapper_pg_dir[PTRS_PER_PGD];
 
diff --git a/arch/microblaze/include/asm/prom.h b/arch/microblaze/include/asm/prom.h
index 20f7b3a926e..37e6f305a68 100644
--- a/arch/microblaze/include/asm/prom.h
+++ b/arch/microblaze/include/asm/prom.h
@@ -16,6 +16,18 @@
 #define _ASM_MICROBLAZE_PROM_H
 #ifdef __KERNEL__
 
+/* Definitions used by the flattened device tree */
+#define OF_DT_HEADER		0xd00dfeed /* marker */
+#define OF_DT_BEGIN_NODE	0x1 /* Start of node, full name */
+#define OF_DT_END_NODE		0x2 /* End node */
+#define OF_DT_PROP		0x3 /* Property: name off, size, content */
+#define OF_DT_NOP		0x4 /* nop */
+#define OF_DT_END		0x9
+
+#define OF_DT_VERSION		0x10
+
+#ifndef __ASSEMBLY__
+
 #include <linux/types.h>
 #include <linux/proc_fs.h>
 #include <linux/platform_device.h>
@@ -29,16 +41,6 @@
 #define of_prop_cmp(s1, s2)		strcmp((s1), (s2))
 #define of_node_cmp(s1, s2)		strcasecmp((s1), (s2))
 
-/* Definitions used by the flattened device tree */
-#define OF_DT_HEADER		0xd00dfeed /* marker */
-#define OF_DT_BEGIN_NODE	0x1 /* Start of node, full name */
-#define OF_DT_END_NODE		0x2 /* End node */
-#define OF_DT_PROP		0x3 /* Property: name off, size, content */
-#define OF_DT_NOP		0x4 /* nop */
-#define OF_DT_END		0x9
-
-#define OF_DT_VERSION		0x10
-
 /*
  * This is what gets passed to the kernel by prom_init or kexec
  *
@@ -309,5 +311,6 @@ extern void __iomem *of_iomap(struct device_node *device, int index);
  */
 #include <linux/of.h>
 
+#endif /* __ASSEMBLY__ */
 #endif /* __KERNEL__ */
 #endif /* _ASM_MICROBLAZE_PROM_H */
diff --git a/arch/microblaze/include/asm/tlb.h b/arch/microblaze/include/asm/tlb.h
index c472d280113..e8abd4a0349 100644
--- a/arch/microblaze/include/asm/tlb.h
+++ b/arch/microblaze/include/asm/tlb.h
@@ -11,7 +11,7 @@
 #ifndef _ASM_MICROBLAZE_TLB_H
 #define _ASM_MICROBLAZE_TLB_H
 
-#define tlb_flush(tlb)	do {} while (0)
+#define tlb_flush(tlb)	flush_tlb_mm((tlb)->mm)
 
 #include <asm-generic/tlb.h>
 
diff --git a/arch/microblaze/include/asm/uaccess.h b/arch/microblaze/include/asm/uaccess.h
index 65adad61e7e..5431b4631a7 100644
--- a/arch/microblaze/include/asm/uaccess.h
+++ b/arch/microblaze/include/asm/uaccess.h
@@ -189,7 +189,7 @@ extern long strnlen_user(const char *src, long count);
 
 #define __put_user(x, ptr)						\
 ({									\
-	__typeof__(*(ptr)) __gu_val = x;				\
+	__typeof__(*(ptr)) volatile __gu_val = (x);			\
 	long __gu_err = 0;						\
 	switch (sizeof(__gu_val)) {					\
 	case 1:								\
diff --git a/arch/microblaze/kernel/Makefile b/arch/microblaze/kernel/Makefile
index f4a5e19a20e..d487729683d 100644
--- a/arch/microblaze/kernel/Makefile
+++ b/arch/microblaze/kernel/Makefile
@@ -17,4 +17,4 @@ obj-$(CONFIG_HEART_BEAT)	+= heartbeat.o
 obj-$(CONFIG_MODULES)		+= microblaze_ksyms.o module.o
 obj-$(CONFIG_MMU)		+= misc.o
 
-obj-y	+= entry$(MMUEXT).o
+obj-y	+= entry$(MMU).o
diff --git a/arch/microblaze/kernel/cpu/cpuinfo-pvr-full.c b/arch/microblaze/kernel/cpu/cpuinfo-pvr-full.c
index 153f57c57b6..c259786e7fa 100644
--- a/arch/microblaze/kernel/cpu/cpuinfo-pvr-full.c
+++ b/arch/microblaze/kernel/cpu/cpuinfo-pvr-full.c
@@ -22,7 +22,7 @@
 
 #define CI(c, p) { ci->c = PVR_##p(pvr); }
 #define err_printk(x) \
-	early_printk("ERROR: Microblaze " x " - different for PVR and DTS\n");
+	early_printk("ERROR: Microblaze " x "-different for PVR and DTS\n");
 
 void set_cpuinfo_pvr_full(struct cpuinfo *ci, struct device_node *cpu)
 {
diff --git a/arch/microblaze/kernel/cpu/cpuinfo-static.c b/arch/microblaze/kernel/cpu/cpuinfo-static.c
index 450ca6bb828..adb448f93d5 100644
--- a/arch/microblaze/kernel/cpu/cpuinfo-static.c
+++ b/arch/microblaze/kernel/cpu/cpuinfo-static.c
@@ -18,7 +18,7 @@ static const char family_string[] = CONFIG_XILINX_MICROBLAZE0_FAMILY;
 static const char cpu_ver_string[] = CONFIG_XILINX_MICROBLAZE0_HW_VER;
 
 #define err_printk(x) \
-	early_printk("ERROR: Microblaze " x "- different for kernel and DTS\n");
+	early_printk("ERROR: Microblaze " x "-different for kernel and DTS\n");
 
 void __init set_cpuinfo_static(struct cpuinfo *ci, struct device_node *cpu)
 {
diff --git a/arch/microblaze/kernel/cpu/cpuinfo.c b/arch/microblaze/kernel/cpu/cpuinfo.c
index a10bea119b9..c411c6757de 100644
--- a/arch/microblaze/kernel/cpu/cpuinfo.c
+++ b/arch/microblaze/kernel/cpu/cpuinfo.c
@@ -26,6 +26,8 @@ const struct cpu_ver_key cpu_ver_lookup[] = {
 	{"7.10.b", 0x09},
 	{"7.10.c", 0x0a},
 	{"7.10.d", 0x0b},
+	{"7.20.a", 0x0c},
+	{"7.20.b", 0x0d},
 	/* FIXME There is no keycode defined in MBV for these versions */
 	{"2.10.a", 0x10},
 	{"3.00.a", 0x20},
diff --git a/arch/microblaze/kernel/head.S b/arch/microblaze/kernel/head.S
index e568d6ec621..e41c6ce2a7b 100644
--- a/arch/microblaze/kernel/head.S
+++ b/arch/microblaze/kernel/head.S
@@ -31,6 +31,7 @@
 #include <linux/linkage.h>
 #include <asm/thread_info.h>
 #include <asm/page.h>
+#include <asm/prom.h>		/* for OF_DT_HEADER */
 
 #ifdef CONFIG_MMU
 #include <asm/setup.h> /* COMMAND_LINE_SIZE */
@@ -54,11 +55,19 @@ ENTRY(_start)
 	andi	r1, r1, ~2
 	mts	rmsr, r1
 
-/* save fdt to kernel location */
-/* r7 stores pointer to fdt blob */
-	beqi	r7, no_fdt_arg
+/* r7 may point to an FDT, or there may be one linked in.
+   if it's in r7, we've got to save it away ASAP.
+   We ensure r7 points to a valid FDT, just in case the bootloader
+   is broken or non-existent */
+	beqi	r7, no_fdt_arg			/* NULL pointer?  don't copy */
+	lw	r11, r0, r7			/* Does r7 point to a */
+	rsubi	r11, r11, OF_DT_HEADER		/* valid FDT? */
+	beqi	r11, _prepare_copy_fdt
+	or	r7, r0, r0		/* clear R7 when not valid DTB */
+	bnei	r11, no_fdt_arg			/* No - get out of here */
+_prepare_copy_fdt:
 	or	r11, r0, r0 /* incremment */
-	ori	r4, r0, TOPHYS(_fdt_start) /* save bram context */
+	ori	r4, r0, TOPHYS(_fdt_start)
 	ori	r3, r0, (0x4000 - 4)
 _copy_fdt:
 	lw	r12, r7, r11 /* r12 = r7 + r11 */
diff --git a/arch/microblaze/kernel/hw_exception_handler.S b/arch/microblaze/kernel/hw_exception_handler.S
index 9d591cd74fc..3288c973767 100644
--- a/arch/microblaze/kernel/hw_exception_handler.S
+++ b/arch/microblaze/kernel/hw_exception_handler.S
@@ -74,6 +74,7 @@
 
 #include <asm/mmu.h>
 #include <asm/pgtable.h>
+#include <asm/signal.h>
 #include <asm/asm-offsets.h>
 
 /* Helpful Macros */
@@ -428,19 +429,9 @@ handle_unaligned_ex:
 	mfs	r17, rbtr;	/* ESR[DS] set - return address in BTR */
 	nop
 _no_delayslot:
-#endif
-
-#ifdef CONFIG_MMU
-	/* Check if unaligned address is last on a 4k page */
-		andi	r5, r4, 0xffc
-		xori	r5, r5, 0xffc
-		bnei	r5, _unaligned_ex2
-	_unaligned_ex1:
-		RESTORE_STATE;
-/* Another page must be accessed or physical address not in page table */
-		bri	unaligned_data_trap
-
-	_unaligned_ex2:
+	/* jump to high level unaligned handler */
+	RESTORE_STATE;
+	bri	unaligned_data_trap
 #endif
 	andi	r6, r3, 0x3E0; /* Mask and extract the register operand */
 	srl	r6, r6; /* r6 >> 5 */
@@ -450,45 +441,6 @@ _no_delayslot:
 	srl	r6, r6;
 	/* Store the register operand in a temporary location */
 	sbi	r6, r0, TOPHYS(ex_reg_op);
-#ifdef CONFIG_MMU
-	/* Get physical address */
-	/* If we are faulting a kernel address, we have to use the
-	 * kernel page tables.
-	 */
-	ori	r5, r0, CONFIG_KERNEL_START
-	cmpu	r5, r4, r5
-	bgti	r5, _unaligned_ex3
-	ori	r5, r0, swapper_pg_dir
-	bri	_unaligned_ex4
-
-	/* Get the PGD for the current thread. */
-_unaligned_ex3: /* user thread */
-	addi	r5 ,CURRENT_TASK, TOPHYS(0); /* get current task address */
-	lwi	r5, r5, TASK_THREAD + PGDIR
-_unaligned_ex4:
-	tophys(r5,r5)
-	BSRLI(r6,r4,20)			/* Create L1 (pgdir/pmd) address */
-	andi	r6, r6, 0xffc
-/* Assume pgdir aligned on 4K boundary, no need for "andi r5,r5,0xfffff003" */
-	or	r5, r5, r6
-	lwi	r6, r5, 0		/* Get L1 entry */
-	andi	r5, r6, 0xfffff000	/* Extract L2 (pte) base address. */
-	beqi	r5, _unaligned_ex1	/* Bail if no table */
-
-	tophys(r5,r5)
-	BSRLI(r6,r4,10)			/* Compute PTE address */
-	andi	r6, r6, 0xffc
-	andi	r5, r5, 0xfffff003
-	or	r5, r5, r6
-	lwi	r5, r5, 0		/* Get Linux PTE */
-
-	andi	r6, r5, _PAGE_PRESENT
-	beqi	r6, _unaligned_ex1	/* Bail if no page */
-
-	andi	r5, r5, 0xfffff000	/* Extract RPN */
-	andi	r4, r4, 0x00000fff	/* Extract offset */
-	or	r4, r4, r5		/* Create physical address */
-#endif /* CONFIG_MMU */
 
 	andi	r6, r3, 0x400; /* Extract ESR[S] */
 	bnei	r6, ex_sw;
@@ -959,15 +911,15 @@ _unaligned_data_exception:
 	andi	r6, r3, 0x800;	/* Extract ESR[W] - delay slot */
 ex_lw_vm:
 	beqid	r6, ex_lhw_vm;
-	lbui	r5, r4, 0;	/* Exception address in r4 - delay slot */
+load1:	lbui	r5, r4, 0;	/* Exception address in r4 - delay slot */
 /* Load a word, byte-by-byte from destination address and save it in tmp space*/
 	la	r6, r0, ex_tmp_data_loc_0;
 	sbi	r5, r6, 0;
-	lbui	r5, r4, 1;
+load2:	lbui	r5, r4, 1;
 	sbi	r5, r6, 1;
-	lbui	r5, r4, 2;
+load3:	lbui	r5, r4, 2;
 	sbi	r5, r6, 2;
-	lbui	r5, r4, 3;
+load4:	lbui	r5, r4, 3;
 	sbi	r5, r6, 3;
 	brid	ex_lw_tail_vm;
 /* Get the destination register value into r3 - delay slot */
@@ -977,7 +929,7 @@ ex_lhw_vm:
 	 * save it in tmp space */
 	la	r6, r0, ex_tmp_data_loc_0;
 	sbi	r5, r6, 0;
-	lbui	r5, r4, 1;
+load5:	lbui	r5, r4, 1;
 	sbi	r5, r6, 1;
 	lhui	r3, r6, 0;	/* Get the destination register value into r3 */
 ex_lw_tail_vm:
@@ -996,22 +948,53 @@ ex_sw_tail_vm:
 	swi	r3, r5, 0;	/* Get the word - delay slot */
 	/* Store the word, byte-by-byte into destination address */
 	lbui	r3, r5, 0;
-	sbi	r3, r4, 0;
+store1:	sbi	r3, r4, 0;
 	lbui	r3, r5, 1;
-	sbi	r3, r4, 1;
+store2:	sbi	r3, r4, 1;
 	lbui	r3, r5, 2;
-	sbi	r3, r4, 2;
+store3:	sbi	r3, r4, 2;
 	lbui	r3, r5, 3;
 	brid	ret_from_exc;
-	sbi	r3, r4, 3;	/* Delay slot */
+store4:	sbi	r3, r4, 3;	/* Delay slot */
 ex_shw_vm:
 	/* Store the lower half-word, byte-by-byte into destination address */
 	lbui	r3, r5, 2;
-	sbi	r3, r4, 0;
+store5:	sbi	r3, r4, 0;
 	lbui	r3, r5, 3;
 	brid	ret_from_exc;
-	sbi	r3, r4, 1;	/* Delay slot */
+store6:	sbi	r3, r4, 1;	/* Delay slot */
 ex_sw_end_vm:			/* Exception handling of store word, ends. */
+
+/* We have to prevent cases that get/put_user macros get unaligned pointer
+ * to bad page area. We have to find out which origin instruction caused it
+ * and called fixup for that origin instruction not instruction in unaligned
+ * handler */
+ex_unaligned_fixup:
+	ori	r5, r7, 0 /* setup pointer to pt_regs */
+	lwi	r6, r7, PT_PC; /* faulting address is one instruction above */
+	addik	r6, r6, -4 /* for finding proper fixup */
+	swi	r6, r7, PT_PC; /* a save back it to PT_PC */
+	addik	r7, r0, SIGSEGV
+	/* call bad_page_fault for finding aligned fixup, fixup address is saved
+	 * in PT_PC which is used as return address from exception */
+	la	r15, r0, ret_from_exc-8 /* setup return address */
+	brid	bad_page_fault
+	nop
+
+/* We prevent all load/store because it could failed any attempt to access */
+.section __ex_table,"a";
+	.word	load1,ex_unaligned_fixup;
+	.word	load2,ex_unaligned_fixup;
+	.word	load3,ex_unaligned_fixup;
+	.word	load4,ex_unaligned_fixup;
+	.word	load5,ex_unaligned_fixup;
+	.word	store1,ex_unaligned_fixup;
+	.word	store2,ex_unaligned_fixup;
+	.word	store3,ex_unaligned_fixup;
+	.word	store4,ex_unaligned_fixup;
+	.word	store5,ex_unaligned_fixup;
+	.word	store6,ex_unaligned_fixup;
+.previous;
 .end _unaligned_data_exception
 #endif /* CONFIG_MMU */
 
diff --git a/arch/microblaze/kernel/module.c b/arch/microblaze/kernel/module.c
index 51414171326..5a45b1adfef 100644
--- a/arch/microblaze/kernel/module.c
+++ b/arch/microblaze/kernel/module.c
@@ -57,7 +57,6 @@ int apply_relocate_add(Elf32_Shdr *sechdrs, const char *strtab,
 	Elf32_Rela *rela = (void *)sechdrs[relsec].sh_addr;
 	Elf32_Sym *sym;
 	unsigned long int *location;
-	unsigned long int locoffs;
 	unsigned long int value;
 #if __GNUC__ < 4
 	unsigned long int old_value;
@@ -113,10 +112,12 @@ int apply_relocate_add(Elf32_Shdr *sechdrs, const char *strtab,
 			break;
 
 		case R_MICROBLAZE_64_PCREL:
-			locoffs = (location[0] & 0xFFFF) << 16 |
+#if __GNUC__ < 4
+			old_value = (location[0] & 0xFFFF) << 16 |
 				(location[1] & 0xFFFF);
-			value -= (unsigned long int)(location) + 4 +
-				locoffs;
+			value -= old_value;
+#endif
+			value -= (unsigned long int)(location) + 4;
 			location[0] = (location[0] & 0xFFFF0000) |
 					(value >> 16);
 			location[1] = (location[1] & 0xFFFF0000) |
@@ -125,6 +126,14 @@ int apply_relocate_add(Elf32_Shdr *sechdrs, const char *strtab,
 				value);
 			break;
 
+		case R_MICROBLAZE_32_PCREL_LO:
+			pr_debug("R_MICROBLAZE_32_PCREL_LO\n");
+			break;
+
+		case R_MICROBLAZE_64_NONE:
+			pr_debug("R_MICROBLAZE_NONE\n");
+			break;
+
 		case R_MICROBLAZE_NONE:
 			pr_debug("R_MICROBLAZE_NONE\n");
 			break;
@@ -133,7 +142,7 @@ int apply_relocate_add(Elf32_Shdr *sechdrs, const char *strtab,
 			printk(KERN_ERR "module %s: "
 				"Unknown relocation: %u\n",
 				module->name,
-				ELF32_R_TYPE(rela->r_info));
+				ELF32_R_TYPE(rela[i].r_info));
 			return -ENOEXEC;
 		}
 	}
diff --git a/arch/microblaze/kernel/setup.c b/arch/microblaze/kernel/setup.c
index 8709bea0960..2a97bf513b6 100644
--- a/arch/microblaze/kernel/setup.c
+++ b/arch/microblaze/kernel/setup.c
@@ -138,8 +138,12 @@ void __init machine_early_init(const char *cmdline, unsigned int ram,
 	setup_early_printk(NULL);
 #endif
 
-	early_printk("Ramdisk addr 0x%08x, FDT 0x%08x\n", ram, fdt);
-	printk(KERN_NOTICE "Found FDT at 0x%08x\n", fdt);
+	early_printk("Ramdisk addr 0x%08x, ", ram);
+	if (fdt)
+		early_printk("FDT at 0x%08x\n", fdt);
+	else
+		early_printk("Compiled-in FDT at 0x%08x\n",
+					(unsigned int)_fdt_start);
 
 #ifdef CONFIG_MTD_UCLINUX
 	early_printk("Found romfs @ 0x%08x (0x%08x)\n",
diff --git a/arch/microblaze/kernel/sys_microblaze.c b/arch/microblaze/kernel/sys_microblaze.c
index e000bce09b2..b96f1682bb2 100644
--- a/arch/microblaze/kernel/sys_microblaze.c
+++ b/arch/microblaze/kernel/sys_microblaze.c
@@ -33,105 +33,6 @@
 #include <linux/unistd.h>
 
 #include <asm/syscalls.h>
-/*
- * sys_ipc() is the de-multiplexer for the SysV IPC calls..
- *
- * This is really horribly ugly. This will be remove with new toolchain.
- */
-asmlinkage long
-sys_ipc(uint call, int first, int second, int third, void *ptr, long fifth)
-{
-	int version, ret;
-
-	version = call >> 16; /* hack for backward compatibility */
-	call &= 0xffff;
-
-	ret = -EINVAL;
-	switch (call) {
-	case SEMOP:
-		ret = sys_semop(first, (struct sembuf *)ptr, second);
-		break;
-	case SEMGET:
-		ret = sys_semget(first, second, third);
-		break;
-	case SEMCTL:
-	{
-		union semun fourth;
-
-		if (!ptr)
-			break;
-		ret = (access_ok(VERIFY_READ, ptr, sizeof(long)) ? 0 : -EFAULT)
-				|| (get_user(fourth.__pad, (void **)ptr)) ;
-		if (ret)
-			break;
-		ret = sys_semctl(first, second, third, fourth);
-		break;
-	}
-	case MSGSND:
-		ret = sys_msgsnd(first, (struct msgbuf *) ptr, second, third);
-		break;
-	case MSGRCV:
-		switch (version) {
-		case 0: {
-			struct ipc_kludge tmp;
-
-			if (!ptr)
-				break;
-			ret = (access_ok(VERIFY_READ, ptr, sizeof(tmp))
-				? 0 : -EFAULT) || copy_from_user(&tmp,
-				(struct ipc_kludge *) ptr, sizeof(tmp));
-			if (ret)
-				break;
-			ret = sys_msgrcv(first, tmp.msgp, second, tmp.msgtyp,
-					third);
-			break;
-			}
-		default:
-			ret = sys_msgrcv(first, (struct msgbuf *) ptr,
-					second, fifth, third);
-			break;
-		}
-		break;
-	case MSGGET:
-		ret = sys_msgget((key_t) first, second);
-		break;
-	case MSGCTL:
-		ret = sys_msgctl(first, second, (struct msqid_ds *) ptr);
-		break;
-	case SHMAT:
-		switch (version) {
-		default: {
-			ulong raddr;
-			ret = access_ok(VERIFY_WRITE, (ulong *) third,
-					sizeof(ulong)) ? 0 : -EFAULT;
-			if (ret)
-				break;
-			ret = do_shmat(first, (char *) ptr, second, &raddr);
-			if (ret)
-				break;
-			ret = put_user(raddr, (ulong *) third);
-			break;
-			}
-		case 1:	/* iBCS2 emulator entry point */
-			if (!segment_eq(get_fs(), get_ds()))
-				break;
-			ret = do_shmat(first, (char *) ptr, second,
-					(ulong *) third);
-			break;
-		}
-		break;
-	case SHMDT:
-		ret = sys_shmdt((char *)ptr);
-		break;
-	case SHMGET:
-		ret = sys_shmget(first, second, third);
-		break;
-	case SHMCTL:
-		ret = sys_shmctl(first, second, (struct shmid_ds *) ptr);
-		break;
-	}
-	return ret;
-}
 
 asmlinkage long microblaze_vfork(struct pt_regs *regs)
 {
diff --git a/arch/microblaze/kernel/syscall_table.S b/arch/microblaze/kernel/syscall_table.S
index 31b32a6c5f4..216db817beb 100644
--- a/arch/microblaze/kernel/syscall_table.S
+++ b/arch/microblaze/kernel/syscall_table.S
@@ -121,7 +121,7 @@ ENTRY(sys_call_table)
 	.long sys_wait4
 	.long sys_swapoff		/* 115 */
 	.long sys_sysinfo
-	.long sys_ipc
+	.long sys_ni_syscall		/* old sys_ipc */
 	.long sys_fsync
 	.long sys_ni_syscall		/* sys_sigreturn_wrapper */
 	.long sys_clone		/* 120 */
diff --git a/arch/microblaze/mm/fault.c b/arch/microblaze/mm/fault.c
index 956607a63f4..d9d249a66ff 100644
--- a/arch/microblaze/mm/fault.c
+++ b/arch/microblaze/mm/fault.c
@@ -69,7 +69,7 @@ static int store_updates_sp(struct pt_regs *regs)
  * It is called from do_page_fault above and from some of the procedures
  * in traps.c.
  */
-static void bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
+void bad_page_fault(struct pt_regs *regs, unsigned long address, int sig)
 {
 	const struct exception_table_entry *fixup;
 /* MS: no context */
@@ -122,15 +122,10 @@ void do_page_fault(struct pt_regs *regs, unsigned long address,
 	}
 #endif /* CONFIG_KGDB */
 
-	if (in_atomic() || mm == NULL) {
-		/* FIXME */
-		if (kernel_mode(regs)) {
-			printk(KERN_EMERG
-				"Page fault in kernel mode - Oooou!!! pid %d\n",
-				current->pid);
-			_exception(SIGSEGV, regs, code, address);
-			return;
-		}
+	if (in_atomic() || !mm) {
+		if (kernel_mode(regs))
+			goto bad_area_nosemaphore;
+
 		/* in_atomic() in user mode is really bad,
 		   as is current->mm == NULL. */
 		printk(KERN_EMERG "Page fault in user mode with "
diff --git a/arch/mips/include/asm/pgalloc.h b/arch/mips/include/asm/pgalloc.h
index 1275831dda2..3738f4b48cb 100644
--- a/arch/mips/include/asm/pgalloc.h
+++ b/arch/mips/include/asm/pgalloc.h
@@ -98,23 +98,12 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	__free_pages(pte, PTE_ORDER);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,address)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb), pte);			\
 } while (0)
 
-#ifdef CONFIG_32BIT
-
-/*
- * allocating and freeing a pmd is trivial: the 1-entry pmd is
- * inside the pgd, so has no extra memory associated with it.
- */
-#define pmd_free(mm, x)			do { } while (0)
-#define __pmd_free_tlb(tlb, x)		do { } while (0)
-
-#endif
-
 #ifdef CONFIG_64BIT
 
 static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
@@ -132,7 +121,7 @@ static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
 	free_pages((unsigned long)pmd, PMD_ORDER);
 }
 
-#define __pmd_free_tlb(tlb, x)	pmd_free((tlb)->mm, x)
+#define __pmd_free_tlb(tlb, x, addr)	pmd_free((tlb)->mm, x)
 
 #endif
 
diff --git a/arch/mn10300/include/asm/pgalloc.h b/arch/mn10300/include/asm/pgalloc.h
index ec057e1bd4c..a19f11327cd 100644
--- a/arch/mn10300/include/asm/pgalloc.h
+++ b/arch/mn10300/include/asm/pgalloc.h
@@ -51,6 +51,6 @@ static inline void pte_free(struct mm_struct *mm, struct page *pte)
 }
 
 
-#define __pte_free_tlb(tlb, pte) tlb_remove_page((tlb), (pte))
+#define __pte_free_tlb(tlb, pte, addr) tlb_remove_page((tlb), (pte))
 
 #endif /* _ASM_PGALLOC_H */
diff --git a/arch/parisc/include/asm/tlb.h b/arch/parisc/include/asm/tlb.h
index 383b1db310e..07924903989 100644
--- a/arch/parisc/include/asm/tlb.h
+++ b/arch/parisc/include/asm/tlb.h
@@ -21,7 +21,7 @@ do {	if (!(tlb)->fullmm)	\
 
 #include <asm-generic/tlb.h>
 
-#define __pmd_free_tlb(tlb, pmd)	pmd_free((tlb)->mm, pmd)
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, pte)
+#define __pmd_free_tlb(tlb, pmd, addr)	pmd_free((tlb)->mm, pmd)
+#define __pte_free_tlb(tlb, pte, addr)	pte_free((tlb)->mm, pte)
 
 #endif
diff --git a/arch/powerpc/include/asm/pgalloc-32.h b/arch/powerpc/include/asm/pgalloc-32.h
index 0815eb40aca..c9500d666a1 100644
--- a/arch/powerpc/include/asm/pgalloc-32.h
+++ b/arch/powerpc/include/asm/pgalloc-32.h
@@ -16,7 +16,7 @@ extern void pgd_free(struct mm_struct *mm, pgd_t *pgd);
  */
 /* #define pmd_alloc_one(mm,address)       ({ BUG(); ((pmd_t *)2); }) */
 #define pmd_free(mm, x) 		do { } while (0)
-#define __pmd_free_tlb(tlb,x)		do { } while (0)
+#define __pmd_free_tlb(tlb,x,a)		do { } while (0)
 /* #define pgd_populate(mm, pmd, pte)      BUG() */
 
 #ifndef CONFIG_BOOKE
diff --git a/arch/powerpc/include/asm/pgalloc-64.h b/arch/powerpc/include/asm/pgalloc-64.h
index afda2bdd860..e6f069c4f71 100644
--- a/arch/powerpc/include/asm/pgalloc-64.h
+++ b/arch/powerpc/include/asm/pgalloc-64.h
@@ -118,11 +118,11 @@ static inline void pgtable_free(pgtable_free_t pgf)
 		kmem_cache_free(pgtable_cache[cachenum], p);
 }
 
-#define __pmd_free_tlb(tlb, pmd) 	\
+#define __pmd_free_tlb(tlb, pmd,addr)		      \
 	pgtable_free_tlb(tlb, pgtable_free_cache(pmd, \
 		PMD_CACHE_NUM, PMD_TABLE_SIZE-1))
 #ifndef CONFIG_PPC_64K_PAGES
-#define __pud_free_tlb(tlb, pud)	\
+#define __pud_free_tlb(tlb, pud, addr)		      \
 	pgtable_free_tlb(tlb, pgtable_free_cache(pud, \
 		PUD_CACHE_NUM, PUD_TABLE_SIZE-1))
 #endif /* CONFIG_PPC_64K_PAGES */
diff --git a/arch/powerpc/include/asm/pgalloc.h b/arch/powerpc/include/asm/pgalloc.h
index 5d8480265a7..1730e5e298d 100644
--- a/arch/powerpc/include/asm/pgalloc.h
+++ b/arch/powerpc/include/asm/pgalloc.h
@@ -38,14 +38,14 @@ static inline pgtable_free_t pgtable_free_cache(void *p, int cachenum,
 extern void pgtable_free_tlb(struct mmu_gather *tlb, pgtable_free_t pgf);
 
 #ifdef CONFIG_SMP
-#define __pte_free_tlb(tlb,ptepage)	\
+#define __pte_free_tlb(tlb,ptepage,address)		\
 do { \
 	pgtable_page_dtor(ptepage); \
 	pgtable_free_tlb(tlb, pgtable_free_cache(page_address(ptepage), \
-		PTE_NONCACHE_NUM, PTE_TABLE_SIZE-1)); \
+					PTE_NONCACHE_NUM, PTE_TABLE_SIZE-1)); \
 } while (0)
 #else
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, (pte))
+#define __pte_free_tlb(tlb, pte, address)	pte_free((tlb)->mm, (pte))
 #endif
 
 
diff --git a/arch/powerpc/mm/hugetlbpage.c b/arch/powerpc/mm/hugetlbpage.c
index 9920d6a7cf2..c46ef2ffa3d 100644
--- a/arch/powerpc/mm/hugetlbpage.c
+++ b/arch/powerpc/mm/hugetlbpage.c
@@ -305,7 +305,7 @@ static void hugetlb_free_pmd_range(struct mmu_gather *tlb, pud_t *pud,
 
 	pmd = pmd_offset(pud, start);
 	pud_clear(pud);
-	pmd_free_tlb(tlb, pmd);
+	pmd_free_tlb(tlb, pmd, start);
 }
 
 static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd,
@@ -348,7 +348,7 @@ static void hugetlb_free_pud_range(struct mmu_gather *tlb, pgd_t *pgd,
 
 	pud = pud_offset(pgd, start);
 	pgd_clear(pgd);
-	pud_free_tlb(tlb, pud);
+	pud_free_tlb(tlb, pud, start);
 }
 
 /*
diff --git a/arch/s390/include/asm/tlb.h b/arch/s390/include/asm/tlb.h
index 3d8a96d39d9..81150b05368 100644
--- a/arch/s390/include/asm/tlb.h
+++ b/arch/s390/include/asm/tlb.h
@@ -96,7 +96,8 @@ static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
  * pte_free_tlb frees a pte table and clears the CRSTE for the
  * page table from the tlb.
  */
-static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte)
+static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte,
+				unsigned long address)
 {
 	if (!tlb->fullmm) {
 		tlb->array[tlb->nr_ptes++] = pte;
@@ -113,7 +114,8 @@ static inline void pte_free_tlb(struct mmu_gather *tlb, pgtable_t pte)
  * as the pgd. pmd_free_tlb checks the asce_limit against 2GB
  * to avoid the double free of the pmd in this case.
  */
-static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd)
+static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
+				unsigned long address)
 {
 #ifdef __s390x__
 	if (tlb->mm->context.asce_limit <= (1UL << 31))
@@ -134,7 +136,8 @@ static inline void pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd)
  * as the pgd. pud_free_tlb checks the asce_limit against 4TB
  * to avoid the double free of the pud in this case.
  */
-static inline void pud_free_tlb(struct mmu_gather *tlb, pud_t *pud)
+static inline void pud_free_tlb(struct mmu_gather *tlb, pud_t *pud,
+				unsigned long address)
 {
 #ifdef __s390x__
 	if (tlb->mm->context.asce_limit <= (1UL << 42))
diff --git a/arch/s390/kernel/early.c b/arch/s390/kernel/early.c
index f9b144049dc..8d15314381e 100644
--- a/arch/s390/kernel/early.c
+++ b/arch/s390/kernel/early.c
@@ -210,7 +210,7 @@ static noinline __init void detect_machine_type(void)
 		machine_flags |= MACHINE_FLAG_VM;
 }
 
-static void early_pgm_check_handler(void)
+static __init void early_pgm_check_handler(void)
 {
 	unsigned long addr;
 	const struct exception_table_entry *fixup;
@@ -222,7 +222,7 @@ static void early_pgm_check_handler(void)
 	S390_lowcore.program_old_psw.addr = fixup->fixup | PSW_ADDR_AMODE;
 }
 
-void setup_lowcore_early(void)
+static noinline __init void setup_lowcore_early(void)
 {
 	psw_t psw;
 
diff --git a/arch/s390/kernel/smp.c b/arch/s390/kernel/smp.c
index 2270730f535..be2cae08340 100644
--- a/arch/s390/kernel/smp.c
+++ b/arch/s390/kernel/smp.c
@@ -687,13 +687,14 @@ void __init smp_prepare_cpus(unsigned int max_cpus)
 #ifndef CONFIG_64BIT
 	if (MACHINE_HAS_IEEE)
 		lowcore->extended_save_area_addr = (u32) save_area;
-#else
-	if (vdso_alloc_per_cpu(smp_processor_id(), lowcore))
-		BUG();
 #endif
 	set_prefix((u32)(unsigned long) lowcore);
 	local_mcck_enable();
 	local_irq_enable();
+#ifdef CONFIG_64BIT
+	if (vdso_alloc_per_cpu(smp_processor_id(), &S390_lowcore))
+		BUG();
+#endif
 	for_each_possible_cpu(cpu)
 		if (cpu != smp_processor_id())
 			smp_create_idle(cpu);
diff --git a/arch/s390/kernel/vdso64/clock_gettime.S b/arch/s390/kernel/vdso64/clock_gettime.S
index 79dbfee831e..49106c6e6f8 100644
--- a/arch/s390/kernel/vdso64/clock_gettime.S
+++ b/arch/s390/kernel/vdso64/clock_gettime.S
@@ -88,10 +88,17 @@ __kernel_clock_gettime:
 	llilh	%r4,0x0100
 	sar	%a4,%r4
 	lghi	%r4,0
+	epsw	%r5,0
 	sacf	512				/* Magic ectg instruction */
 	.insn	ssf,0xc80100000000,__VDSO_ECTG_BASE(4),__VDSO_ECTG_USER(4),4
-	sacf	0
-	sar	%a4,%r2
+	tml	%r5,0x4000
+	jo	11f
+	tml	%r5,0x8000
+	jno	10f
+	sacf	256
+	j	11f
+10:	sacf	0
+11:	sar	%a4,%r2
 	algr	%r1,%r0				/* r1 = cputime as TOD value */
 	mghi	%r1,1000			/* convert to nanoseconds */
 	srlg	%r1,%r1,12			/* r1 = cputime in nanosec */
diff --git a/arch/s390/power/swsusp.c b/arch/s390/power/swsusp.c
index e6a4fe9f5f2..bd1f5c6b0b8 100644
--- a/arch/s390/power/swsusp.c
+++ b/arch/s390/power/swsusp.c
@@ -7,24 +7,36 @@
  *
  */
 
+#include <asm/system.h>
 
-/*
- * save CPU registers before creating a hibernation image and before
- * restoring the memory state from it
- */
 void save_processor_state(void)
 {
-	/* implentation contained in the
-	 * swsusp_arch_suspend function
+	/* swsusp_arch_suspend() actually saves all cpu register contents.
+	 * Machine checks must be disabled since swsusp_arch_suspend() stores
+	 * register contents to their lowcore save areas. That's the same
+	 * place where register contents on machine checks would be saved.
+	 * To avoid register corruption disable machine checks.
+	 * We must also disable machine checks in the new psw mask for
+	 * program checks, since swsusp_arch_suspend() may generate program
+	 * checks. Disabling machine checks for all other new psw masks is
+	 * just paranoia.
 	 */
+	local_mcck_disable();
+	/* Disable lowcore protection */
+	__ctl_clear_bit(0,28);
+	S390_lowcore.external_new_psw.mask &= ~PSW_MASK_MCHECK;
+	S390_lowcore.svc_new_psw.mask &= ~PSW_MASK_MCHECK;
+	S390_lowcore.io_new_psw.mask &= ~PSW_MASK_MCHECK;
+	S390_lowcore.program_new_psw.mask &= ~PSW_MASK_MCHECK;
 }
 
-/*
- * restore the contents of CPU registers
- */
 void restore_processor_state(void)
 {
-	/* implentation contained in the
-	 * swsusp_arch_resume function
-	 */
+	S390_lowcore.external_new_psw.mask |= PSW_MASK_MCHECK;
+	S390_lowcore.svc_new_psw.mask |= PSW_MASK_MCHECK;
+	S390_lowcore.io_new_psw.mask |= PSW_MASK_MCHECK;
+	S390_lowcore.program_new_psw.mask |= PSW_MASK_MCHECK;
+	/* Enable lowcore protection */
+	__ctl_set_bit(0,28);
+	local_mcck_enable();
 }
diff --git a/arch/s390/power/swsusp_asm64.S b/arch/s390/power/swsusp_asm64.S
index 76d688da32f..b26df5c5933 100644
--- a/arch/s390/power/swsusp_asm64.S
+++ b/arch/s390/power/swsusp_asm64.S
@@ -32,19 +32,14 @@ swsusp_arch_suspend:
 	/* Deactivate DAT */
 	stnsm	__SF_EMPTY(%r15),0xfb
 
-	/* Switch off lowcore protection */
-	stctg	%c0,%c0,__SF_EMPTY(%r15)
-	ni	__SF_EMPTY+4(%r15),0xef
-	lctlg	%c0,%c0,__SF_EMPTY(%r15)
-
 	/* Store prefix register on stack */
 	stpx	__SF_EMPTY(%r15)
 
-	/* Setup base register for lowcore (absolute 0) */
-	llgf	%r1,__SF_EMPTY(%r15)
+	/* Save prefix register contents for lowcore */
+	llgf	%r4,__SF_EMPTY(%r15)
 
 	/* Get pointer to save area */
-	aghi	%r1,0x1000
+	lghi	%r1,0x1000
 
 	/* Store registers */
 	mvc	0x318(4,%r1),__SF_EMPTY(%r15)	/* move prefix to lowcore */
@@ -79,17 +74,15 @@ swsusp_arch_suspend:
 	xc	__SF_EMPTY(4,%r15),__SF_EMPTY(%r15)
 	spx	__SF_EMPTY(%r15)
 
-	/* Setup lowcore */
-	brasl	%r14,setup_lowcore_early
+	lghi	%r2,0
+	lghi	%r3,2*PAGE_SIZE
+	lghi	%r5,2*PAGE_SIZE
+1:	mvcle	%r2,%r4,0
+	jo	1b
 
 	/* Save image */
 	brasl	%r14,swsusp_save
 
-	/* Switch on lowcore protection */
-	stctg	%c0,%c0,__SF_EMPTY(%r15)
-	oi	__SF_EMPTY+4(%r15),0x10
-	lctlg	%c0,%c0,__SF_EMPTY(%r15)
-
 	/* Restore prefix register and return */
 	lghi	%r1,0x1000
 	spx	0x318(%r1)
@@ -117,11 +110,6 @@ swsusp_arch_resume:
 	/* Deactivate DAT */
 	stnsm	__SF_EMPTY(%r15),0xfb
 
-	/* Switch off lowcore protection */
-	stctg	%c0,%c0,__SF_EMPTY(%r15)
-	ni	__SF_EMPTY+4(%r15),0xef
-	lctlg	%c0,%c0,__SF_EMPTY(%r15)
-
 	/* Set prefix page to zero */
 	xc	__SF_EMPTY(4,%r15),__SF_EMPTY(%r15)
 	spx	__SF_EMPTY(%r15)
@@ -175,7 +163,7 @@ swsusp_arch_resume:
 	/* Load old stack */
 	lg	%r15,0x2f8(%r13)
 
-	/* Pointer to save arae */
+	/* Pointer to save area */
 	lghi	%r13,0x1000
 
 #ifdef CONFIG_SMP
@@ -187,11 +175,6 @@ swsusp_arch_resume:
 	/* Restore prefix register */
 	spx	0x318(%r13)
 
-	/* Switch on lowcore protection */
-	stctg	%c0,%c0,__SF_EMPTY(%r15)
-	oi	__SF_EMPTY+4(%r15),0x10
-	lctlg	%c0,%c0,__SF_EMPTY(%r15)
-
 	/* Activate DAT */
 	stosm	__SF_EMPTY(%r15),0x04
 
diff --git a/arch/sh/include/asm/pgalloc.h b/arch/sh/include/asm/pgalloc.h
index 84dd2db7104..63ca37bd9a9 100644
--- a/arch/sh/include/asm/pgalloc.h
+++ b/arch/sh/include/asm/pgalloc.h
@@ -73,20 +73,12 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	quicklist_free_page(QUICK_PT, NULL, pte);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte,addr)			\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb), (pte));			\
 } while (0)
 
-/*
- * allocating and freeing a pmd is trivial: the 1-entry pmd is
- * inside the pgd, so has no extra memory associated with it.
- */
-
-#define pmd_free(mm, x)			do { } while (0)
-#define __pmd_free_tlb(tlb,x)		do { } while (0)
-
 static inline void check_pgt_cache(void)
 {
 	quicklist_trim(QUICK_PGD, NULL, 25, 16);
diff --git a/arch/sh/include/asm/tlb.h b/arch/sh/include/asm/tlb.h
index 9c16f737074..da8fe7ab872 100644
--- a/arch/sh/include/asm/tlb.h
+++ b/arch/sh/include/asm/tlb.h
@@ -91,9 +91,9 @@ tlb_end_vma(struct mmu_gather *tlb, struct vm_area_struct *vma)
 }
 
 #define tlb_remove_page(tlb,page)	free_page_and_swap_cache(page)
-#define pte_free_tlb(tlb, ptep)		pte_free((tlb)->mm, ptep)
-#define pmd_free_tlb(tlb, pmdp)		pmd_free((tlb)->mm, pmdp)
-#define pud_free_tlb(tlb, pudp)		pud_free((tlb)->mm, pudp)
+#define pte_free_tlb(tlb, ptep, addr)	pte_free((tlb)->mm, ptep)
+#define pmd_free_tlb(tlb, pmdp, addr)	pmd_free((tlb)->mm, pmdp)
+#define pud_free_tlb(tlb, pudp, addr)	pud_free((tlb)->mm, pudp)
 
 #define tlb_migrate_finish(mm)		do { } while (0)
 
diff --git a/arch/sparc/include/asm/pgalloc_32.h b/arch/sparc/include/asm/pgalloc_32.h
index 681582d2696..ca2b34456c4 100644
--- a/arch/sparc/include/asm/pgalloc_32.h
+++ b/arch/sparc/include/asm/pgalloc_32.h
@@ -44,8 +44,8 @@ BTFIXUPDEF_CALL(pmd_t *, pmd_alloc_one, struct mm_struct *, unsigned long)
 BTFIXUPDEF_CALL(void, free_pmd_fast, pmd_t *)
 #define free_pmd_fast(pmd)	BTFIXUP_CALL(free_pmd_fast)(pmd)
 
-#define pmd_free(mm, pmd)	free_pmd_fast(pmd)
-#define __pmd_free_tlb(tlb, pmd) pmd_free((tlb)->mm, pmd)
+#define pmd_free(mm, pmd)		free_pmd_fast(pmd)
+#define __pmd_free_tlb(tlb, pmd, addr)	pmd_free((tlb)->mm, pmd)
 
 BTFIXUPDEF_CALL(void, pmd_populate, pmd_t *, struct page *)
 #define pmd_populate(MM, PMD, PTE)        BTFIXUP_CALL(pmd_populate)(PMD, PTE)
@@ -62,7 +62,7 @@ BTFIXUPDEF_CALL(void, free_pte_fast, pte_t *)
 #define pte_free_kernel(mm, pte)	BTFIXUP_CALL(free_pte_fast)(pte)
 
 BTFIXUPDEF_CALL(void, pte_free, pgtable_t )
-#define pte_free(mm, pte)	BTFIXUP_CALL(pte_free)(pte)
-#define __pte_free_tlb(tlb, pte)	pte_free((tlb)->mm, pte)
+#define pte_free(mm, pte)		BTFIXUP_CALL(pte_free)(pte)
+#define __pte_free_tlb(tlb, pte, addr)	pte_free((tlb)->mm, pte)
 
 #endif /* _SPARC_PGALLOC_H */
diff --git a/arch/sparc/include/asm/tlb_64.h b/arch/sparc/include/asm/tlb_64.h
index ee38e731bfa..dca406b9b6f 100644
--- a/arch/sparc/include/asm/tlb_64.h
+++ b/arch/sparc/include/asm/tlb_64.h
@@ -100,9 +100,9 @@ static inline void tlb_remove_page(struct mmu_gather *mp, struct page *page)
 }
 
 #define tlb_remove_tlb_entry(mp,ptep,addr) do { } while (0)
-#define pte_free_tlb(mp, ptepage) pte_free((mp)->mm, ptepage)
-#define pmd_free_tlb(mp, pmdp) pmd_free((mp)->mm, pmdp)
-#define pud_free_tlb(tlb,pudp) __pud_free_tlb(tlb,pudp)
+#define pte_free_tlb(mp, ptepage, addr) pte_free((mp)->mm, ptepage)
+#define pmd_free_tlb(mp, pmdp, addr) pmd_free((mp)->mm, pmdp)
+#define pud_free_tlb(tlb,pudp, addr) __pud_free_tlb(tlb,pudp,addr)
 
 #define tlb_migrate_finish(mm)	do { } while (0)
 #define tlb_start_vma(tlb, vma) do { } while (0)
diff --git a/arch/um/include/asm/pgalloc.h b/arch/um/include/asm/pgalloc.h
index 718984359f8..32c8ce4e151 100644
--- a/arch/um/include/asm/pgalloc.h
+++ b/arch/um/include/asm/pgalloc.h
@@ -40,7 +40,7 @@ static inline void pte_free(struct mm_struct *mm, pgtable_t pte)
 	__free_page(pte);
 }
 
-#define __pte_free_tlb(tlb,pte)				\
+#define __pte_free_tlb(tlb,pte, address)		\
 do {							\
 	pgtable_page_dtor(pte);				\
 	tlb_remove_page((tlb),(pte));			\
@@ -53,7 +53,7 @@ static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
 	free_page((unsigned long)pmd);
 }
 
-#define __pmd_free_tlb(tlb,x)   tlb_remove_page((tlb),virt_to_page(x))
+#define __pmd_free_tlb(tlb,x, address)   tlb_remove_page((tlb),virt_to_page(x))
 #endif
 
 #define check_pgt_cache()	do { } while (0)
diff --git a/arch/um/include/asm/tlb.h b/arch/um/include/asm/tlb.h
index 5240fa1c5e0..660caedac9e 100644
--- a/arch/um/include/asm/tlb.h
+++ b/arch/um/include/asm/tlb.h
@@ -116,11 +116,11 @@ static inline void tlb_remove_page(struct mmu_gather *tlb, struct page *page)
 		__tlb_remove_tlb_entry(tlb, ptep, address);	\
 	} while (0)
 
-#define pte_free_tlb(tlb, ptep) __pte_free_tlb(tlb, ptep)
+#define pte_free_tlb(tlb, ptep, addr) __pte_free_tlb(tlb, ptep, addr)
 
-#define pud_free_tlb(tlb, pudp) __pud_free_tlb(tlb, pudp)
+#define pud_free_tlb(tlb, pudp, addr) __pud_free_tlb(tlb, pudp, addr)
 
-#define pmd_free_tlb(tlb, pmdp) __pmd_free_tlb(tlb, pmdp)
+#define pmd_free_tlb(tlb, pmdp, addr) __pmd_free_tlb(tlb, pmdp, addr)
 
 #define tlb_migrate_finish(mm) do {} while (0)
 
diff --git a/arch/x86/include/asm/pgalloc.h b/arch/x86/include/asm/pgalloc.h
index dd14c54ac71..0e8c2a0fd92 100644
--- a/arch/x86/include/asm/pgalloc.h
+++ b/arch/x86/include/asm/pgalloc.h
@@ -46,7 +46,13 @@ static inline void pte_free(struct mm_struct *mm, struct page *pte)
 	__free_page(pte);
 }
 
-extern void __pte_free_tlb(struct mmu_gather *tlb, struct page *pte);
+extern void ___pte_free_tlb(struct mmu_gather *tlb, struct page *pte);
+
+static inline void __pte_free_tlb(struct mmu_gather *tlb, struct page *pte,
+				  unsigned long address)
+{
+	___pte_free_tlb(tlb, pte);
+}
 
 static inline void pmd_populate_kernel(struct mm_struct *mm,
 				       pmd_t *pmd, pte_t *pte)
@@ -78,7 +84,13 @@ static inline void pmd_free(struct mm_struct *mm, pmd_t *pmd)
 	free_page((unsigned long)pmd);
 }
 
-extern void __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd);
+extern void ___pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd);
+
+static inline void __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd,
+				  unsigned long adddress)
+{
+	___pmd_free_tlb(tlb, pmd);
+}
 
 #ifdef CONFIG_X86_PAE
 extern void pud_populate(struct mm_struct *mm, pud_t *pudp, pmd_t *pmd);
@@ -108,7 +120,14 @@ static inline void pud_free(struct mm_struct *mm, pud_t *pud)
 	free_page((unsigned long)pud);
 }
 
-extern void __pud_free_tlb(struct mmu_gather *tlb, pud_t *pud);
+extern void ___pud_free_tlb(struct mmu_gather *tlb, pud_t *pud);
+
+static inline void __pud_free_tlb(struct mmu_gather *tlb, pud_t *pud,
+				  unsigned long address)
+{
+	___pud_free_tlb(tlb, pud);
+}
+
 #endif	/* PAGETABLE_LEVELS > 3 */
 #endif	/* PAGETABLE_LEVELS > 2 */
 
diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
index 20e6a795e16..d2c6c930b49 100644
--- a/arch/x86/include/asm/uaccess.h
+++ b/arch/x86/include/asm/uaccess.h
@@ -212,9 +212,9 @@ extern int __get_user_bad(void);
 		     : "A" ((typeof(*(ptr)))(x)), "c" (ptr) : "ebx")
 #else
 #define __put_user_asm_u64(x, ptr, retval, errret) \
-	__put_user_asm(x, ptr, retval, "q", "", "Zr", errret)
+	__put_user_asm(x, ptr, retval, "q", "", "er", errret)
 #define __put_user_asm_ex_u64(x, addr)	\
-	__put_user_asm_ex(x, addr, "q", "", "Zr")
+	__put_user_asm_ex(x, addr, "q", "", "er")
 #define __put_user_x8(x, ptr, __ret_pu) __put_user_x(8, x, ptr, __ret_pu)
 #endif
 
diff --git a/arch/x86/include/asm/uaccess_64.h b/arch/x86/include/asm/uaccess_64.h
index 8cc687326eb..db24b215fc5 100644
--- a/arch/x86/include/asm/uaccess_64.h
+++ b/arch/x86/include/asm/uaccess_64.h
@@ -88,11 +88,11 @@ int __copy_to_user(void __user *dst, const void *src, unsigned size)
 			      ret, "l", "k", "ir", 4);
 		return ret;
 	case 8:__put_user_asm(*(u64 *)src, (u64 __user *)dst,
-			      ret, "q", "", "ir", 8);
+			      ret, "q", "", "er", 8);
 		return ret;
 	case 10:
 		__put_user_asm(*(u64 *)src, (u64 __user *)dst,
-			       ret, "q", "", "ir", 10);
+			       ret, "q", "", "er", 10);
 		if (unlikely(ret))
 			return ret;
 		asm("":::"memory");
@@ -101,12 +101,12 @@ int __copy_to_user(void __user *dst, const void *src, unsigned size)
 		return ret;
 	case 16:
 		__put_user_asm(*(u64 *)src, (u64 __user *)dst,
-			       ret, "q", "", "ir", 16);
+			       ret, "q", "", "er", 16);
 		if (unlikely(ret))
 			return ret;
 		asm("":::"memory");
 		__put_user_asm(1[(u64 *)src], 1 + (u64 __user *)dst,
-			       ret, "q", "", "ir", 8);
+			       ret, "q", "", "er", 8);
 		return ret;
 	default:
 		return copy_user_generic((__force void *)dst, src, size);
@@ -157,7 +157,7 @@ int __copy_in_user(void __user *dst, const void __user *src, unsigned size)
 			       ret, "q", "", "=r", 8);
 		if (likely(!ret))
 			__put_user_asm(tmp, (u64 __user *)dst,
-				       ret, "q", "", "ir", 8);
+				       ret, "q", "", "er", 8);
 		return ret;
 	}
 	default:
diff --git a/arch/x86/kernel/cpu/amd.c b/arch/x86/kernel/cpu/amd.c
index 28e5f595604..e2485b03f1c 100644
--- a/arch/x86/kernel/cpu/amd.c
+++ b/arch/x86/kernel/cpu/amd.c
@@ -356,7 +356,7 @@ static void __cpuinit early_init_amd(struct cpuinfo_x86 *c)
 #endif
 #if defined(CONFIG_X86_LOCAL_APIC) && defined(CONFIG_PCI)
 	/* check CPU config space for extended APIC ID */
-	if (c->x86 >= 0xf) {
+	if (cpu_has_apic && c->x86 >= 0xf) {
 		unsigned int val;
 		val = read_pci_config(0, 24, 0, 0x68);
 		if ((val & ((1 << 17) | (1 << 18))) == ((1 << 17) | (1 << 18)))
diff --git a/arch/x86/kernel/cpu/mcheck/mce.c b/arch/x86/kernel/cpu/mcheck/mce.c
index 484c1e5f658..1cfb623ce11 100644
--- a/arch/x86/kernel/cpu/mcheck/mce.c
+++ b/arch/x86/kernel/cpu/mcheck/mce.c
@@ -1692,17 +1692,15 @@ static ssize_t set_trigger(struct sys_device *s, struct sysdev_attribute *attr,
 				const char *buf, size_t siz)
 {
 	char *p;
-	int len;
 
 	strncpy(mce_helper, buf, sizeof(mce_helper));
 	mce_helper[sizeof(mce_helper)-1] = 0;
-	len = strlen(mce_helper);
 	p = strchr(mce_helper, '\n');
 
-	if (*p)
+	if (p)
 		*p = 0;
 
-	return len;
+	return strlen(mce_helper) + !!p;
 }
 
 static ssize_t set_ignore_ce(struct sys_device *s,
diff --git a/arch/x86/kernel/cpu/perf_counter.c b/arch/x86/kernel/cpu/perf_counter.c
index 36c3dc7b899..a7aa8f90095 100644
--- a/arch/x86/kernel/cpu/perf_counter.c
+++ b/arch/x86/kernel/cpu/perf_counter.c
@@ -66,6 +66,52 @@ static DEFINE_PER_CPU(struct cpu_hw_counters, cpu_hw_counters) = {
 };
 
 /*
+ * Not sure about some of these
+ */
+static const u64 p6_perfmon_event_map[] =
+{
+  [PERF_COUNT_HW_CPU_CYCLES]		= 0x0079,
+  [PERF_COUNT_HW_INSTRUCTIONS]		= 0x00c0,
+  [PERF_COUNT_HW_CACHE_REFERENCES]	= 0x0000,
+  [PERF_COUNT_HW_CACHE_MISSES]		= 0x0000,
+  [PERF_COUNT_HW_BRANCH_INSTRUCTIONS]	= 0x00c4,
+  [PERF_COUNT_HW_BRANCH_MISSES]		= 0x00c5,
+  [PERF_COUNT_HW_BUS_CYCLES]		= 0x0062,
+};
+
+static u64 p6_pmu_event_map(int event)
+{
+	return p6_perfmon_event_map[event];
+}
+
+/*
+ * Counter setting that is specified not to count anything.
+ * We use this to effectively disable a counter.
+ *
+ * L2_RQSTS with 0 MESI unit mask.
+ */
+#define P6_NOP_COUNTER			0x0000002EULL
+
+static u64 p6_pmu_raw_event(u64 event)
+{
+#define P6_EVNTSEL_EVENT_MASK		0x000000FFULL
+#define P6_EVNTSEL_UNIT_MASK		0x0000FF00ULL
+#define P6_EVNTSEL_EDGE_MASK		0x00040000ULL
+#define P6_EVNTSEL_INV_MASK		0x00800000ULL
+#define P6_EVNTSEL_COUNTER_MASK		0xFF000000ULL
+
+#define P6_EVNTSEL_MASK			\
+	(P6_EVNTSEL_EVENT_MASK |	\
+	 P6_EVNTSEL_UNIT_MASK  |	\
+	 P6_EVNTSEL_EDGE_MASK  |	\
+	 P6_EVNTSEL_INV_MASK   |	\
+	 P6_EVNTSEL_COUNTER_MASK)
+
+	return event & P6_EVNTSEL_MASK;
+}
+
+
+/*
  * Intel PerfMon v3. Used on Core2 and later.
  */
 static const u64 intel_perfmon_event_map[] =
@@ -666,6 +712,7 @@ static int __hw_perf_counter_init(struct perf_counter *counter)
 {
 	struct perf_counter_attr *attr = &counter->attr;
 	struct hw_perf_counter *hwc = &counter->hw;
+	u64 config;
 	int err;
 
 	if (!x86_pmu_initialized())
@@ -718,14 +765,40 @@ static int __hw_perf_counter_init(struct perf_counter *counter)
 
 	if (attr->config >= x86_pmu.max_events)
 		return -EINVAL;
+
 	/*
 	 * The generic map:
 	 */
-	hwc->config |= x86_pmu.event_map(attr->config);
+	config = x86_pmu.event_map(attr->config);
+
+	if (config == 0)
+		return -ENOENT;
+
+	if (config == -1LL)
+		return -EINVAL;
+
+	hwc->config |= config;
 
 	return 0;
 }
 
+static void p6_pmu_disable_all(void)
+{
+	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
+	u64 val;
+
+	if (!cpuc->enabled)
+		return;
+
+	cpuc->enabled = 0;
+	barrier();
+
+	/* p6 only has one enable register */
+	rdmsrl(MSR_P6_EVNTSEL0, val);
+	val &= ~ARCH_PERFMON_EVENTSEL0_ENABLE;
+	wrmsrl(MSR_P6_EVNTSEL0, val);
+}
+
 static void intel_pmu_disable_all(void)
 {
 	wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, 0);
@@ -767,6 +840,23 @@ void hw_perf_disable(void)
 	return x86_pmu.disable_all();
 }
 
+static void p6_pmu_enable_all(void)
+{
+	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
+	unsigned long val;
+
+	if (cpuc->enabled)
+		return;
+
+	cpuc->enabled = 1;
+	barrier();
+
+	/* p6 only has one enable register */
+	rdmsrl(MSR_P6_EVNTSEL0, val);
+	val |= ARCH_PERFMON_EVENTSEL0_ENABLE;
+	wrmsrl(MSR_P6_EVNTSEL0, val);
+}
+
 static void intel_pmu_enable_all(void)
 {
 	wrmsrl(MSR_CORE_PERF_GLOBAL_CTRL, x86_pmu.intel_ctrl);
@@ -784,13 +874,13 @@ static void amd_pmu_enable_all(void)
 	barrier();
 
 	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
+		struct perf_counter *counter = cpuc->counters[idx];
 		u64 val;
 
 		if (!test_bit(idx, cpuc->active_mask))
 			continue;
-		rdmsrl(MSR_K7_EVNTSEL0 + idx, val);
-		if (val & ARCH_PERFMON_EVENTSEL0_ENABLE)
-			continue;
+
+		val = counter->hw.config;
 		val |= ARCH_PERFMON_EVENTSEL0_ENABLE;
 		wrmsrl(MSR_K7_EVNTSEL0 + idx, val);
 	}
@@ -819,16 +909,13 @@ static inline void intel_pmu_ack_status(u64 ack)
 
 static inline void x86_pmu_enable_counter(struct hw_perf_counter *hwc, int idx)
 {
-	int err;
-	err = checking_wrmsrl(hwc->config_base + idx,
+	(void)checking_wrmsrl(hwc->config_base + idx,
 			      hwc->config | ARCH_PERFMON_EVENTSEL0_ENABLE);
 }
 
 static inline void x86_pmu_disable_counter(struct hw_perf_counter *hwc, int idx)
 {
-	int err;
-	err = checking_wrmsrl(hwc->config_base + idx,
-			      hwc->config);
+	(void)checking_wrmsrl(hwc->config_base + idx, hwc->config);
 }
 
 static inline void
@@ -836,13 +923,24 @@ intel_pmu_disable_fixed(struct hw_perf_counter *hwc, int __idx)
 {
 	int idx = __idx - X86_PMC_IDX_FIXED;
 	u64 ctrl_val, mask;
-	int err;
 
 	mask = 0xfULL << (idx * 4);
 
 	rdmsrl(hwc->config_base, ctrl_val);
 	ctrl_val &= ~mask;
-	err = checking_wrmsrl(hwc->config_base, ctrl_val);
+	(void)checking_wrmsrl(hwc->config_base, ctrl_val);
+}
+
+static inline void
+p6_pmu_disable_counter(struct hw_perf_counter *hwc, int idx)
+{
+	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
+	u64 val = P6_NOP_COUNTER;
+
+	if (cpuc->enabled)
+		val |= ARCH_PERFMON_EVENTSEL0_ENABLE;
+
+	(void)checking_wrmsrl(hwc->config_base + idx, val);
 }
 
 static inline void
@@ -943,6 +1041,19 @@ intel_pmu_enable_fixed(struct hw_perf_counter *hwc, int __idx)
 	err = checking_wrmsrl(hwc->config_base, ctrl_val);
 }
 
+static void p6_pmu_enable_counter(struct hw_perf_counter *hwc, int idx)
+{
+	struct cpu_hw_counters *cpuc = &__get_cpu_var(cpu_hw_counters);
+	u64 val;
+
+	val = hwc->config;
+	if (cpuc->enabled)
+		val |= ARCH_PERFMON_EVENTSEL0_ENABLE;
+
+	(void)checking_wrmsrl(hwc->config_base + idx, val);
+}
+
+
 static void intel_pmu_enable_counter(struct hw_perf_counter *hwc, int idx)
 {
 	if (unlikely(hwc->config_base == MSR_ARCH_PERFMON_FIXED_CTR_CTRL)) {
@@ -959,8 +1070,6 @@ static void amd_pmu_enable_counter(struct hw_perf_counter *hwc, int idx)
 
 	if (cpuc->enabled)
 		x86_pmu_enable_counter(hwc, idx);
-	else
-		x86_pmu_disable_counter(hwc, idx);
 }
 
 static int
@@ -1176,6 +1285,49 @@ static void intel_pmu_reset(void)
 	local_irq_restore(flags);
 }
 
+static int p6_pmu_handle_irq(struct pt_regs *regs)
+{
+	struct perf_sample_data data;
+	struct cpu_hw_counters *cpuc;
+	struct perf_counter *counter;
+	struct hw_perf_counter *hwc;
+	int idx, handled = 0;
+	u64 val;
+
+	data.regs = regs;
+	data.addr = 0;
+
+	cpuc = &__get_cpu_var(cpu_hw_counters);
+
+	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
+		if (!test_bit(idx, cpuc->active_mask))
+			continue;
+
+		counter = cpuc->counters[idx];
+		hwc = &counter->hw;
+
+		val = x86_perf_counter_update(counter, hwc, idx);
+		if (val & (1ULL << (x86_pmu.counter_bits - 1)))
+			continue;
+
+		/*
+		 * counter overflow
+		 */
+		handled		= 1;
+		data.period	= counter->hw.last_period;
+
+		if (!x86_perf_counter_set_period(counter, hwc, idx))
+			continue;
+
+		if (perf_counter_overflow(counter, 1, &data))
+			p6_pmu_disable_counter(hwc, idx);
+	}
+
+	if (handled)
+		inc_irq_stat(apic_perf_irqs);
+
+	return handled;
+}
 
 /*
  * This handler is triggered by the local APIC, so the APIC IRQ handling
@@ -1185,14 +1337,13 @@ static int intel_pmu_handle_irq(struct pt_regs *regs)
 {
 	struct perf_sample_data data;
 	struct cpu_hw_counters *cpuc;
-	int bit, cpu, loops;
+	int bit, loops;
 	u64 ack, status;
 
 	data.regs = regs;
 	data.addr = 0;
 
-	cpu = smp_processor_id();
-	cpuc = &per_cpu(cpu_hw_counters, cpu);
+	cpuc = &__get_cpu_var(cpu_hw_counters);
 
 	perf_disable();
 	status = intel_pmu_get_status();
@@ -1249,14 +1400,13 @@ static int amd_pmu_handle_irq(struct pt_regs *regs)
 	struct cpu_hw_counters *cpuc;
 	struct perf_counter *counter;
 	struct hw_perf_counter *hwc;
-	int cpu, idx, handled = 0;
+	int idx, handled = 0;
 	u64 val;
 
 	data.regs = regs;
 	data.addr = 0;
 
-	cpu = smp_processor_id();
-	cpuc = &per_cpu(cpu_hw_counters, cpu);
+	cpuc = &__get_cpu_var(cpu_hw_counters);
 
 	for (idx = 0; idx < x86_pmu.num_counters; idx++) {
 		if (!test_bit(idx, cpuc->active_mask))
@@ -1353,6 +1503,32 @@ static __read_mostly struct notifier_block perf_counter_nmi_notifier = {
 	.priority		= 1
 };
 
+static struct x86_pmu p6_pmu = {
+	.name			= "p6",
+	.handle_irq		= p6_pmu_handle_irq,
+	.disable_all		= p6_pmu_disable_all,
+	.enable_all		= p6_pmu_enable_all,
+	.enable			= p6_pmu_enable_counter,
+	.disable		= p6_pmu_disable_counter,
+	.eventsel		= MSR_P6_EVNTSEL0,
+	.perfctr		= MSR_P6_PERFCTR0,
+	.event_map		= p6_pmu_event_map,
+	.raw_event		= p6_pmu_raw_event,
+	.max_events		= ARRAY_SIZE(p6_perfmon_event_map),
+	.max_period		= (1ULL << 31) - 1,
+	.version		= 0,
+	.num_counters		= 2,
+	/*
+	 * Counters have 40 bits implemented. However they are designed such
+	 * that bits [32-39] are sign extensions of bit 31. As such the
+	 * effective width of a counter for P6-like PMU is 32 bits only.
+	 *
+	 * See IA-32 Intel Architecture Software developer manual Vol 3B
+	 */
+	.counter_bits		= 32,
+	.counter_mask		= (1ULL << 32) - 1,
+};
+
 static struct x86_pmu intel_pmu = {
 	.name			= "Intel",
 	.handle_irq		= intel_pmu_handle_irq,
@@ -1392,6 +1568,37 @@ static struct x86_pmu amd_pmu = {
 	.max_period		= (1ULL << 47) - 1,
 };
 
+static int p6_pmu_init(void)
+{
+	switch (boot_cpu_data.x86_model) {
+	case 1:
+	case 3:  /* Pentium Pro */
+	case 5:
+	case 6:  /* Pentium II */
+	case 7:
+	case 8:
+	case 11: /* Pentium III */
+		break;
+	case 9:
+	case 13:
+		/* Pentium M */
+		break;
+	default:
+		pr_cont("unsupported p6 CPU model %d ",
+			boot_cpu_data.x86_model);
+		return -ENODEV;
+	}
+
+	if (!cpu_has_apic) {
+		pr_info("no Local APIC, try rebooting with lapic");
+		return -ENODEV;
+	}
+
+	x86_pmu				= p6_pmu;
+
+	return 0;
+}
+
 static int intel_pmu_init(void)
 {
 	union cpuid10_edx edx;
@@ -1400,8 +1607,14 @@ static int intel_pmu_init(void)
 	unsigned int ebx;
 	int version;
 
-	if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON))
+	if (!cpu_has(&boot_cpu_data, X86_FEATURE_ARCH_PERFMON)) {
+		/* check for P6 processor family */
+	   if (boot_cpu_data.x86 == 6) {
+		return p6_pmu_init();
+	   } else {
 		return -ENODEV;
+	   }
+	}
 
 	/*
 	 * Check whether the Architectural PerfMon supports
diff --git a/arch/x86/kernel/irqinit.c b/arch/x86/kernel/irqinit.c
index 696f0e475c2..92b7703d3d5 100644
--- a/arch/x86/kernel/irqinit.c
+++ b/arch/x86/kernel/irqinit.c
@@ -187,7 +187,7 @@ static void __init apic_intr_init(void)
 #ifdef CONFIG_X86_THERMAL_VECTOR
 	alloc_intr_gate(THERMAL_APIC_VECTOR, thermal_interrupt);
 #endif
-#ifdef CONFIG_X86_THRESHOLD
+#ifdef CONFIG_X86_MCE_THRESHOLD
 	alloc_intr_gate(THRESHOLD_APIC_VECTOR, threshold_interrupt);
 #endif
 #if defined(CONFIG_X86_NEW_MCE) && defined(CONFIG_X86_LOCAL_APIC)
diff --git a/arch/x86/kernel/mfgpt_32.c b/arch/x86/kernel/mfgpt_32.c
index 846510b78a0..2a62d843f01 100644
--- a/arch/x86/kernel/mfgpt_32.c
+++ b/arch/x86/kernel/mfgpt_32.c
@@ -347,7 +347,7 @@ static irqreturn_t mfgpt_tick(int irq, void *dev_id)
 
 static struct irqaction mfgptirq  = {
 	.handler = mfgpt_tick,
-	.flags = IRQF_DISABLED | IRQF_NOBALANCING,
+	.flags = IRQF_DISABLED | IRQF_NOBALANCING | IRQF_TIMER,
 	.name = "mfgpt-timer"
 };
 
diff --git a/arch/x86/kernel/reboot.c b/arch/x86/kernel/reboot.c
index d2d1ce8170f..508e982dd07 100644
--- a/arch/x86/kernel/reboot.c
+++ b/arch/x86/kernel/reboot.c
@@ -249,6 +249,14 @@ static struct dmi_system_id __initdata reboot_dmi_table[] = {
 			DMI_MATCH(DMI_PRODUCT_NAME, "VGN-Z540N"),
 		},
 	},
+	{	/* Handle problems with rebooting on CompuLab SBC-FITPC2 */
+		.callback = set_bios_reboot,
+		.ident = "CompuLab SBC-FITPC2",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "CompuLab"),
+			DMI_MATCH(DMI_PRODUCT_NAME, "SBC-FITPC2"),
+		},
+	},
 	{ }
 };
 
diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index de2cab13284..63f32d220ef 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -672,6 +672,19 @@ static struct dmi_system_id __initdata bad_bios_dmi_table[] = {
 			DMI_MATCH(DMI_BIOS_VENDOR, "Phoenix Technologies"),
 		},
 	},
+	{
+	/*
+	 * AMI BIOS with low memory corruption was found on Intel DG45ID board.
+	 * It hase different DMI_BIOS_VENDOR = "Intel Corp.", for now we will
+	 * match only DMI_BOARD_NAME and see if there is more bad products
+	 * with this vendor.
+	 */
+		.callback = dmi_low_memory_corruption,
+		.ident = "AMI BIOS",
+		.matches = {
+			DMI_MATCH(DMI_BOARD_NAME, "DG45ID"),
+		},
+	},
 #endif
 	{}
 };
diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
index 367e8788204..59f31d2dd43 100644
--- a/arch/x86/kernel/vmlinux.lds.S
+++ b/arch/x86/kernel/vmlinux.lds.S
@@ -112,11 +112,6 @@ SECTIONS
 		_sdata = .;
 		DATA_DATA
 		CONSTRUCTORS
-
-#ifdef CONFIG_X86_64
-		/* End of data section */
-		_edata = .;
-#endif
 	} :data
 
 #ifdef CONFIG_X86_32
@@ -156,10 +151,8 @@ SECTIONS
 	.data.read_mostly : AT(ADDR(.data.read_mostly) - LOAD_OFFSET) {
 		*(.data.read_mostly)
 
-#ifdef CONFIG_X86_32
 		/* End of data section */
 		_edata = .;
-#endif
 	}
 
 #ifdef CONFIG_X86_64
diff --git a/arch/x86/mm/pgtable.c b/arch/x86/mm/pgtable.c
index 8e43bdd4545..af8f9650058 100644
--- a/arch/x86/mm/pgtable.c
+++ b/arch/x86/mm/pgtable.c
@@ -25,7 +25,7 @@ pgtable_t pte_alloc_one(struct mm_struct *mm, unsigned long address)
 	return pte;
 }
 
-void __pte_free_tlb(struct mmu_gather *tlb, struct page *pte)
+void ___pte_free_tlb(struct mmu_gather *tlb, struct page *pte)
 {
 	pgtable_page_dtor(pte);
 	paravirt_release_pte(page_to_pfn(pte));
@@ -33,14 +33,14 @@ void __pte_free_tlb(struct mmu_gather *tlb, struct page *pte)
 }
 
 #if PAGETABLE_LEVELS > 2
-void __pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd)
+void ___pmd_free_tlb(struct mmu_gather *tlb, pmd_t *pmd)
 {
 	paravirt_release_pmd(__pa(pmd) >> PAGE_SHIFT);
 	tlb_remove_page(tlb, virt_to_page(pmd));
 }
 
 #if PAGETABLE_LEVELS > 3
-void __pud_free_tlb(struct mmu_gather *tlb, pud_t *pud)
+void ___pud_free_tlb(struct mmu_gather *tlb, pud_t *pud)
 {
 	paravirt_release_pud(__pa(pud) >> PAGE_SHIFT);
 	tlb_remove_page(tlb, virt_to_page(pud));
diff --git a/arch/x86/mm/srat_64.c b/arch/x86/mm/srat_64.c
index 2dfcbf9df2a..dbb5381f7b3 100644
--- a/arch/x86/mm/srat_64.c
+++ b/arch/x86/mm/srat_64.c
@@ -79,8 +79,10 @@ static __init void bad_srat(void)
 	acpi_numa = -1;
 	for (i = 0; i < MAX_LOCAL_APIC; i++)
 		apicid_to_node[i] = NUMA_NO_NODE;
-	for (i = 0; i < MAX_NUMNODES; i++)
-		nodes_add[i].start = nodes[i].end = 0;
+	for (i = 0; i < MAX_NUMNODES; i++) {
+		nodes[i].start = nodes[i].end = 0;
+		nodes_add[i].start = nodes_add[i].end = 0;
+	}
 	remove_all_active_ranges();
 }
 
diff --git a/arch/xtensa/include/asm/tlb.h b/arch/xtensa/include/asm/tlb.h
index 31c220faca0..0d766f9c108 100644
--- a/arch/xtensa/include/asm/tlb.h
+++ b/arch/xtensa/include/asm/tlb.h
@@ -42,6 +42,6 @@
 
 #include <asm-generic/tlb.h>
 
-#define __pte_free_tlb(tlb, pte)		pte_free((tlb)->mm, pte)
+#define __pte_free_tlb(tlb, pte, address)	pte_free((tlb)->mm, pte)
 
 #endif	/* _XTENSA_TLB_H */