Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-next

Pull sparc update from David Miller:

 1) Implement support for up to 47-bit physical addresses on sparc64.

 2) Support HAVE_CONTEXT_TRACKING on sparc64, from Kirill Tkhai.

 3) Fix Simba bridge window calculations, from Kjetil Oftedal.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-next:
  sparc64: Implement HAVE_CONTEXT_TRACKING
  sparc64: Add self-IPI support for smp_send_reschedule()
  sparc: PCI: Fix incorrect address calculation of PCI Bridge windows on Simba-bridges
  sparc64: Encode huge PMDs using PTE encoding.
  sparc64: Move to 64-bit PGDs and PMDs.
  sparc64: Move from 4MB to 8MB huge pages.
  sparc64: Make PAGE_OFFSET variable.
  sparc64: Fix inconsistent max-physical-address defines.
  sparc64: Document the shift counts used to validate linear kernel addresses.
  sparc64: Define PAGE_OFFSET in terms of physical address bits.
  sparc64: Use PAGE_OFFSET instead of a magic constant.
  sparc64: Clean up 64-bit mmap exclusion defines.

arch/sparc/Kconfig

@@ -63,6 +63,7 @@ config SPARC64
 	select HAVE_DYNAMIC_FTRACE
 	select HAVE_FTRACE_MCOUNT_RECORD
 	select HAVE_SYSCALL_TRACEPOINTS
+	select HAVE_CONTEXT_TRACKING
 	select HAVE_DEBUG_KMEMLEAK
 	select RTC_DRV_CMOS
 	select RTC_DRV_BQ4802

arch/sparc/include/asm/mmu_64.h

@@ -93,7 +93,6 @@ typedef struct {
 	spinlock_t		lock;
 	unsigned long		sparc64_ctx_val;
 	unsigned long		huge_pte_count;
-	struct page		*pgtable_page;
 	struct tsb_config	tsb_block[MM_NUM_TSBS];
 	struct hv_tsb_descr	tsb_descr[MM_NUM_TSBS];
 } mm_context_t;

arch/sparc/include/asm/page_64.h

@@ -15,7 +15,10 @@
 #define DCACHE_ALIASING_POSSIBLE
 #endif
 
-#define HPAGE_SHIFT		22
+#define HPAGE_SHIFT		23
+#define REAL_HPAGE_SHIFT	22
+
+#define REAL_HPAGE_SIZE		(_AC(1,UL) << REAL_HPAGE_SHIFT)
 
 #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
 #define HPAGE_SIZE		(_AC(1,UL) << HPAGE_SHIFT)
@@ -53,8 +56,8 @@ extern void copy_user_page(void *to, void *from, unsigned long vaddr, struct pag
 /* These are used to make use of C type-checking.. */
 typedef struct { unsigned long pte; } pte_t;
 typedef struct { unsigned long iopte; } iopte_t;
-typedef struct { unsigned int pmd; } pmd_t;
+typedef struct { unsigned long pmd; } pmd_t;
-typedef struct { unsigned int pgd; } pgd_t;
+typedef struct { unsigned long pgd; } pgd_t;
 typedef struct { unsigned long pgprot; } pgprot_t;
 
 #define pte_val(x)	((x).pte)
@@ -73,8 +76,8 @@ typedef struct { unsigned long pgprot; } pgprot_t;
 /* .. while these make it easier on the compiler */
 typedef unsigned long pte_t;
 typedef unsigned long iopte_t;
-typedef unsigned int pmd_t;
+typedef unsigned long pmd_t;
-typedef unsigned int pgd_t;
+typedef unsigned long pgd_t;
 typedef unsigned long pgprot_t;
 
 #define pte_val(x)	(x)
@@ -93,18 +96,44 @@ typedef unsigned long pgprot_t;
 
 typedef pte_t *pgtable_t;
 
+/* These two values define the virtual address space range in which we
+ * must forbid 64-bit user processes from making mappings.  It used to
+ * represent precisely the virtual address space hole present in most
+ * early sparc64 chips including UltraSPARC-I.  But now it also is
+ * further constrained by the limits of our page tables, which is
+ * 43-bits of virtual address.
+ */
+#define SPARC64_VA_HOLE_TOP	_AC(0xfffffc0000000000,UL)
+#define SPARC64_VA_HOLE_BOTTOM	_AC(0x0000040000000000,UL)
+
+/* The next two defines specify the actual exclusion region we
+ * enforce, wherein we use a 4GB red zone on each side of the VA hole.
+ */
+#define VA_EXCLUDE_START (SPARC64_VA_HOLE_BOTTOM - (1UL << 32UL))
+#define VA_EXCLUDE_END   (SPARC64_VA_HOLE_TOP + (1UL << 32UL))
+
 #define TASK_UNMAPPED_BASE	(test_thread_flag(TIF_32BIT) ? \
-				 (_AC(0x0000000070000000,UL)) : \
+				 _AC(0x0000000070000000,UL) : \
-				 (_AC(0xfffff80000000000,UL) + (1UL << 32UL)))
+				 VA_EXCLUDE_END)
 
 #include <asm-generic/memory_model.h>
 
+#define PAGE_OFFSET_BY_BITS(X)	(-(_AC(1,UL) << (X)))
+extern unsigned long PAGE_OFFSET;
+
 #endif /* !(__ASSEMBLY__) */
 
-/* We used to stick this into a hard-coded global register (%g4)
+/* The maximum number of physical memory address bits we support, this
- * but that does not make sense anymore.
+ * is used to size various tables used to manage kernel TLB misses and
+ * also the sparsemem code.
  */
-#define PAGE_OFFSET		_AC(0xFFFFF80000000000,UL)
+#define MAX_PHYS_ADDRESS_BITS	47
+
+/* These two shift counts are used when indexing sparc64_valid_addr_bitmap
+ * and kpte_linear_bitmap.
+ */
+#define ILOG2_4MB		22
+#define ILOG2_256MB		28
 
 #ifndef __ASSEMBLY__
 

arch/sparc/include/asm/pgtable_64.h

@@ -48,18 +48,18 @@
 /* PMD_SHIFT determines the size of the area a second-level page
  * table can map
  */
-#define PMD_SHIFT	(PAGE_SHIFT + (PAGE_SHIFT-4))
+#define PMD_SHIFT	(PAGE_SHIFT + (PAGE_SHIFT-3))
 #define PMD_SIZE	(_AC(1,UL) << PMD_SHIFT)
 #define PMD_MASK	(~(PMD_SIZE-1))
-#define PMD_BITS	(PAGE_SHIFT - 2)
+#define PMD_BITS	(PAGE_SHIFT - 3)
 
 /* PGDIR_SHIFT determines what a third-level page table entry can map */
-#define PGDIR_SHIFT	(PAGE_SHIFT + (PAGE_SHIFT-4) + PMD_BITS)
+#define PGDIR_SHIFT	(PAGE_SHIFT + (PAGE_SHIFT-3) + PMD_BITS)
 #define PGDIR_SIZE	(_AC(1,UL) << PGDIR_SHIFT)
 #define PGDIR_MASK	(~(PGDIR_SIZE-1))
-#define PGDIR_BITS	(PAGE_SHIFT - 2)
+#define PGDIR_BITS	(PAGE_SHIFT - 3)
 
-#if (PGDIR_SHIFT + PGDIR_BITS) != 44
+#if (PGDIR_SHIFT + PGDIR_BITS) != 43
 #error Page table parameters do not cover virtual address space properly.
 #endif
 
@@ -67,35 +67,12 @@
 #error PMD_SHIFT must equal HPAGE_SHIFT for transparent huge pages.
 #endif
 
-/* PMDs point to PTE tables which are 4K aligned.  */
-#define PMD_PADDR	_AC(0xfffffffe,UL)
-#define PMD_PADDR_SHIFT	_AC(11,UL)
-
-#define PMD_ISHUGE	_AC(0x00000001,UL)
-
-/* This is the PMD layout when PMD_ISHUGE is set.  With 4MB huge
- * pages, this frees up a bunch of bits in the layout that we can
- * use for the protection settings and software metadata.
- */
-#define PMD_HUGE_PADDR		_AC(0xfffff800,UL)
-#define PMD_HUGE_PROTBITS	_AC(0x000007ff,UL)
-#define PMD_HUGE_PRESENT	_AC(0x00000400,UL)
-#define PMD_HUGE_WRITE		_AC(0x00000200,UL)
-#define PMD_HUGE_DIRTY		_AC(0x00000100,UL)
-#define PMD_HUGE_ACCESSED	_AC(0x00000080,UL)
-#define PMD_HUGE_EXEC		_AC(0x00000040,UL)
-#define PMD_HUGE_SPLITTING	_AC(0x00000020,UL)
-
-/* PGDs point to PMD tables which are 8K aligned.  */
-#define PGD_PADDR	_AC(0xfffffffc,UL)
-#define PGD_PADDR_SHIFT	_AC(11,UL)
-
 #ifndef __ASSEMBLY__
 
 #include <linux/sched.h>
 
 /* Entries per page directory level. */
-#define PTRS_PER_PTE	(1UL << (PAGE_SHIFT-4))
+#define PTRS_PER_PTE	(1UL << (PAGE_SHIFT-3))
 #define PTRS_PER_PMD	(1UL << PMD_BITS)
 #define PTRS_PER_PGD	(1UL << PGDIR_BITS)
 
@@ -112,6 +89,7 @@
 #define _PAGE_VALID	  _AC(0x8000000000000000,UL) /* Valid TTE            */
 #define _PAGE_R	  	  _AC(0x8000000000000000,UL) /* Keep ref bit uptodate*/
 #define _PAGE_SPECIAL     _AC(0x0200000000000000,UL) /* Special page         */
+#define _PAGE_PMD_HUGE    _AC(0x0100000000000000,UL) /* Huge page            */
 
 /* Advertise support for _PAGE_SPECIAL */
 #define __HAVE_ARCH_PTE_SPECIAL
@@ -125,6 +103,7 @@
 #define _PAGE_IE_4U	  _AC(0x0800000000000000,UL) /* Invert Endianness    */
 #define _PAGE_SOFT2_4U	  _AC(0x07FC000000000000,UL) /* Software bits, set 2 */
 #define _PAGE_SPECIAL_4U  _AC(0x0200000000000000,UL) /* Special page         */
+#define _PAGE_PMD_HUGE_4U _AC(0x0100000000000000,UL) /* Huge page            */
 #define _PAGE_RES1_4U	  _AC(0x0002000000000000,UL) /* Reserved             */
 #define _PAGE_SZ32MB_4U	  _AC(0x0001000000000000,UL) /* (Panther) 32MB page  */
 #define _PAGE_SZ256MB_4U  _AC(0x2001000000000000,UL) /* (Panther) 256MB page */
@@ -155,6 +134,7 @@
 #define _PAGE_READ_4V	  _AC(0x0800000000000000,UL) /* Readable SW Bit      */
 #define _PAGE_WRITE_4V	  _AC(0x0400000000000000,UL) /* Writable SW Bit      */
 #define _PAGE_SPECIAL_4V  _AC(0x0200000000000000,UL) /* Special page         */
+#define _PAGE_PMD_HUGE_4V _AC(0x0100000000000000,UL) /* Huge page            */
 #define _PAGE_PADDR_4V	  _AC(0x00FFFFFFFFFFE000,UL) /* paddr[55:13]         */
 #define _PAGE_IE_4V	  _AC(0x0000000000001000,UL) /* Invert Endianness    */
 #define _PAGE_E_4V	  _AC(0x0000000000000800,UL) /* side-Effect          */
@@ -180,6 +160,10 @@
 #define _PAGE_SZBITS_4U	_PAGE_SZ8K_4U
 #define _PAGE_SZBITS_4V	_PAGE_SZ8K_4V
 
+#if REAL_HPAGE_SHIFT != 22
+#error REAL_HPAGE_SHIFT and _PAGE_SZHUGE_foo must match up
+#endif
+
 #define _PAGE_SZHUGE_4U	_PAGE_SZ4MB_4U
 #define _PAGE_SZHUGE_4V	_PAGE_SZ4MB_4V
 
@@ -239,16 +223,13 @@ static inline pte_t pfn_pte(unsigned long pfn, pgprot_t prot)
 #define mk_pte(page, pgprot)	pfn_pte(page_to_pfn(page), (pgprot))
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-extern pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot);
+static inline pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
-#define mk_pmd(page, pgprot)	pfn_pmd(page_to_pfn(page), (pgprot))
-
-extern pmd_t pmd_modify(pmd_t pmd, pgprot_t newprot);
-
-static inline pmd_t pmd_mkhuge(pmd_t pmd)
 {
-	/* Do nothing, mk_pmd() does this part.  */
+	pte_t pte = pfn_pte(page_nr, pgprot);
-	return pmd;
+
+	return __pmd(pte_val(pte));
 }
+#define mk_pmd(page, pgprot)	pfn_pmd(page_to_pfn(page), (pgprot))
 #endif
 
 /* This one can be done with two shifts.  */
@@ -309,14 +290,25 @@ static inline pte_t pte_modify(pte_t pte, pgprot_t prot)
 	: "=r" (mask), "=r" (tmp)
 	: "i" (_PAGE_PADDR_4U | _PAGE_MODIFIED_4U | _PAGE_ACCESSED_4U |
 	       _PAGE_CP_4U | _PAGE_CV_4U | _PAGE_E_4U | _PAGE_PRESENT_4U |
-	       _PAGE_SPECIAL),
+	       _PAGE_SPECIAL | _PAGE_PMD_HUGE | _PAGE_SZALL_4U),
 	  "i" (_PAGE_PADDR_4V | _PAGE_MODIFIED_4V | _PAGE_ACCESSED_4V |
 	       _PAGE_CP_4V | _PAGE_CV_4V | _PAGE_E_4V | _PAGE_PRESENT_4V |
-	       _PAGE_SPECIAL));
+	       _PAGE_SPECIAL | _PAGE_PMD_HUGE | _PAGE_SZALL_4V));
 
 	return __pte((pte_val(pte) & mask) | (pgprot_val(prot) & ~mask));
 }
 
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+static inline pmd_t pmd_modify(pmd_t pmd, pgprot_t newprot)
+{
+	pte_t pte = __pte(pmd_val(pmd));
+
+	pte = pte_modify(pte, newprot);
+
+	return __pmd(pte_val(pte));
+}
+#endif
+
 static inline pte_t pgoff_to_pte(unsigned long off)
 {
 	off <<= PAGE_SHIFT;
@@ -357,7 +349,7 @@ static inline pgprot_t pgprot_noncached(pgprot_t prot)
  */
 #define pgprot_noncached pgprot_noncached
 
-#ifdef CONFIG_HUGETLB_PAGE
+#if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
 static inline pte_t pte_mkhuge(pte_t pte)
 {
 	unsigned long mask;
@@ -375,6 +367,17 @@ static inline pte_t pte_mkhuge(pte_t pte)
 
 	return __pte(pte_val(pte) | mask);
 }
+#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+static inline pmd_t pmd_mkhuge(pmd_t pmd)
+{
+	pte_t pte = __pte(pmd_val(pmd));
+
+	pte = pte_mkhuge(pte);
+	pte_val(pte) |= _PAGE_PMD_HUGE;
+
+	return __pmd(pte_val(pte));
+}
+#endif
 #endif
 
 static inline pte_t pte_mkdirty(pte_t pte)
@@ -626,91 +629,130 @@ static inline unsigned long pte_special(pte_t pte)
 	return pte_val(pte) & _PAGE_SPECIAL;
 }
 
-static inline int pmd_large(pmd_t pmd)
+static inline unsigned long pmd_large(pmd_t pmd)
 {
-	return (pmd_val(pmd) & (PMD_ISHUGE | PMD_HUGE_PRESENT)) ==
+	pte_t pte = __pte(pmd_val(pmd));
-		(PMD_ISHUGE | PMD_HUGE_PRESENT);
+
+	return (pte_val(pte) & _PAGE_PMD_HUGE) && pte_present(pte);
 }
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-static inline int pmd_young(pmd_t pmd)
+static inline unsigned long pmd_young(pmd_t pmd)
 {
-	return pmd_val(pmd) & PMD_HUGE_ACCESSED;
+	pte_t pte = __pte(pmd_val(pmd));
+
+	return pte_young(pte);
 }
 
-static inline int pmd_write(pmd_t pmd)
+static inline unsigned long pmd_write(pmd_t pmd)
 {
-	return pmd_val(pmd) & PMD_HUGE_WRITE;
+	pte_t pte = __pte(pmd_val(pmd));
+
+	return pte_write(pte);
 }
 
 static inline unsigned long pmd_pfn(pmd_t pmd)
 {
-	unsigned long val = pmd_val(pmd) & PMD_HUGE_PADDR;
+	pte_t pte = __pte(pmd_val(pmd));
 
-	return val >> (PAGE_SHIFT - PMD_PADDR_SHIFT);
+	return pte_pfn(pte);
 }
 
-static inline int pmd_trans_splitting(pmd_t pmd)
+static inline unsigned long pmd_trans_huge(pmd_t pmd)
 {
-	return (pmd_val(pmd) & (PMD_ISHUGE|PMD_HUGE_SPLITTING)) ==
+	pte_t pte = __pte(pmd_val(pmd));
-		(PMD_ISHUGE|PMD_HUGE_SPLITTING);
+
+	return pte_val(pte) & _PAGE_PMD_HUGE;
 }
 
-static inline int pmd_trans_huge(pmd_t pmd)
+static inline unsigned long pmd_trans_splitting(pmd_t pmd)
 {
-	return pmd_val(pmd) & PMD_ISHUGE;
+	pte_t pte = __pte(pmd_val(pmd));
+
+	return pmd_trans_huge(pmd) && pte_special(pte);
 }
 
 #define has_transparent_hugepage() 1
 
 static inline pmd_t pmd_mkold(pmd_t pmd)
 {
-	pmd_val(pmd) &= ~PMD_HUGE_ACCESSED;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_mkold(pte);
+
+	return __pmd(pte_val(pte));
 }
 
 static inline pmd_t pmd_wrprotect(pmd_t pmd)
 {
-	pmd_val(pmd) &= ~PMD_HUGE_WRITE;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_wrprotect(pte);
+
+	return __pmd(pte_val(pte));
 }
 
 static inline pmd_t pmd_mkdirty(pmd_t pmd)
 {
-	pmd_val(pmd) |= PMD_HUGE_DIRTY;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_mkdirty(pte);
+
+	return __pmd(pte_val(pte));
 }
 
 static inline pmd_t pmd_mkyoung(pmd_t pmd)
 {
-	pmd_val(pmd) |= PMD_HUGE_ACCESSED;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_mkyoung(pte);
+
+	return __pmd(pte_val(pte));
 }
 
 static inline pmd_t pmd_mkwrite(pmd_t pmd)
 {
-	pmd_val(pmd) |= PMD_HUGE_WRITE;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_mkwrite(pte);
+
+	return __pmd(pte_val(pte));
 }
 
 static inline pmd_t pmd_mknotpresent(pmd_t pmd)
 {
-	pmd_val(pmd) &= ~PMD_HUGE_PRESENT;
+	unsigned long mask;
+
+	if (tlb_type == hypervisor)
+		mask = _PAGE_PRESENT_4V;
+	else
+		mask = _PAGE_PRESENT_4U;
+
+	pmd_val(pmd) &= ~mask;
+
 	return pmd;
 }
 
 static inline pmd_t pmd_mksplitting(pmd_t pmd)
 {
-	pmd_val(pmd) |= PMD_HUGE_SPLITTING;
+	pte_t pte = __pte(pmd_val(pmd));
-	return pmd;
+
+	pte = pte_mkspecial(pte);
+
+	return __pmd(pte_val(pte));
 }
 
-extern pgprot_t pmd_pgprot(pmd_t entry);
+static inline pgprot_t pmd_pgprot(pmd_t entry)
+{
+	unsigned long val = pmd_val(entry);
+
+	return __pgprot(val);
+}
 #endif
 
 static inline int pmd_present(pmd_t pmd)
 {
-	return pmd_val(pmd) != 0U;
+	return pmd_val(pmd) != 0UL;
 }
 
 #define pmd_none(pmd)			(!pmd_val(pmd))
@@ -728,33 +770,32 @@ static inline void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 
 static inline void pmd_set(struct mm_struct *mm, pmd_t *pmdp, pte_t *ptep)
 {
-	unsigned long val = __pa((unsigned long) (ptep)) >> PMD_PADDR_SHIFT;
+	unsigned long val = __pa((unsigned long) (ptep));
 
 	pmd_val(*pmdp) = val;
 }
 
 #define pud_set(pudp, pmdp)	\
-	(pud_val(*(pudp)) = (__pa((unsigned long) (pmdp)) >> PGD_PADDR_SHIFT))
+	(pud_val(*(pudp)) = (__pa((unsigned long) (pmdp))))
 static inline unsigned long __pmd_page(pmd_t pmd)
 {
-	unsigned long paddr = (unsigned long) pmd_val(pmd);
+	pte_t pte = __pte(pmd_val(pmd));
-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
+	unsigned long pfn;
-	if (pmd_val(pmd) & PMD_ISHUGE)
+
-		paddr &= PMD_HUGE_PADDR;
+	pfn = pte_pfn(pte);
-#endif
+
-	paddr <<= PMD_PADDR_SHIFT;
+	return ((unsigned long) __va(pfn << PAGE_SHIFT));
-	return ((unsigned long) __va(paddr));
 }
 #define pmd_page(pmd) 			virt_to_page((void *)__pmd_page(pmd))
 #define pud_page_vaddr(pud)		\
-	((unsigned long) __va((((unsigned long)pud_val(pud))<<PGD_PADDR_SHIFT)))
+	((unsigned long) __va(pud_val(pud)))
 #define pud_page(pud) 			virt_to_page((void *)pud_page_vaddr(pud))
 #define pmd_bad(pmd)			(0)
-#define pmd_clear(pmdp)			(pmd_val(*(pmdp)) = 0U)
+#define pmd_clear(pmdp)			(pmd_val(*(pmdp)) = 0UL)
 #define pud_none(pud)			(!pud_val(pud))
 #define pud_bad(pud)			(0)
 #define pud_present(pud)		(pud_val(pud) != 0U)
-#define pud_clear(pudp)			(pud_val(*(pudp)) = 0U)
+#define pud_clear(pudp)			(pud_val(*(pudp)) = 0UL)
 
 /* Same in both SUN4V and SUN4U.  */
 #define pte_none(pte) 			(!pte_val(pte))
@@ -789,7 +830,7 @@ static inline pmd_t pmdp_get_and_clear(struct mm_struct *mm,
 				       pmd_t *pmdp)
 {
 	pmd_t pmd = *pmdp;
-	set_pmd_at(mm, addr, pmdp, __pmd(0U));
+	set_pmd_at(mm, addr, pmdp, __pmd(0UL));
 	return pmd;
 }
 
@@ -837,8 +878,8 @@ static inline void __set_pte_at(struct mm_struct *mm, unsigned long addr,
 })
 #endif
 
-extern pgd_t swapper_pg_dir[2048];
+extern pgd_t swapper_pg_dir[PTRS_PER_PGD];
-extern pmd_t swapper_low_pmd_dir[2048];
+extern pmd_t swapper_low_pmd_dir[PTRS_PER_PMD];
 
 extern void paging_init(void);
 extern unsigned long find_ecache_flush_span(unsigned long size);

arch/sparc/include/asm/sparsemem.h

@@ -3,9 +3,11 @@
 
 #ifdef __KERNEL__
 
+#include <asm/page.h>
+
 #define SECTION_SIZE_BITS       30
-#define MAX_PHYSADDR_BITS       42
+#define MAX_PHYSADDR_BITS       MAX_PHYS_ADDRESS_BITS
-#define MAX_PHYSMEM_BITS        42
+#define MAX_PHYSMEM_BITS        MAX_PHYS_ADDRESS_BITS
 
 #endif /* !(__KERNEL__) */
 

arch/sparc/include/asm/thread_info_64.h

@@ -192,7 +192,7 @@ register struct thread_info *current_thread_info_reg asm("g6");
 #define TIF_UNALIGNED		5	/* allowed to do unaligned accesses */
 /* flag bit 6 is available */
 #define TIF_32BIT		7	/* 32-bit binary */
-/* flag bit 8 is available */
+#define TIF_NOHZ		8	/* in adaptive nohz mode */
 #define TIF_SECCOMP		9	/* secure computing */
 #define TIF_SYSCALL_AUDIT	10	/* syscall auditing active */
 #define TIF_SYSCALL_TRACEPOINT	11	/* syscall tracepoint instrumentation */
@@ -210,6 +210,7 @@ register struct thread_info *current_thread_info_reg asm("g6");
 #define _TIF_NEED_RESCHED	(1<<TIF_NEED_RESCHED)
 #define _TIF_UNALIGNED		(1<<TIF_UNALIGNED)
 #define _TIF_32BIT		(1<<TIF_32BIT)
+#define _TIF_NOHZ		(1<<TIF_NOHZ)
 #define _TIF_SECCOMP		(1<<TIF_SECCOMP)
 #define _TIF_SYSCALL_AUDIT	(1<<TIF_SYSCALL_AUDIT)
 #define _TIF_SYSCALL_TRACEPOINT	(1<<TIF_SYSCALL_TRACEPOINT)

arch/sparc/include/asm/tsb.h

@@ -142,98 +142,39 @@ extern struct tsb_phys_patch_entry __tsb_phys_patch, __tsb_phys_patch_end;
 	or		REG1, %lo(swapper_pg_dir), REG1; \
 	sllx		VADDR, 64 - (PGDIR_SHIFT + PGDIR_BITS), REG2; \
 	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	andn		REG2, 0x3, REG2; \
+	andn		REG2, 0x7, REG2; \
-	lduw		[REG1 + REG2], REG1; \
+	ldx		[REG1 + REG2], REG1; \
 	brz,pn		REG1, FAIL_LABEL; \
 	 sllx		VADDR, 64 - (PMD_SHIFT + PMD_BITS), REG2; \
 	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	sllx		REG1, PGD_PADDR_SHIFT, REG1; \
+	andn		REG2, 0x7, REG2; \
-	andn		REG2, 0x3, REG2; \
+	ldxa		[REG1 + REG2] ASI_PHYS_USE_EC, REG1; \
-	lduwa		[REG1 + REG2] ASI_PHYS_USE_EC, REG1; \
 	brz,pn		REG1, FAIL_LABEL; \
 	 sllx		VADDR, 64 - PMD_SHIFT, REG2; \
-	srlx		REG2, 64 - (PAGE_SHIFT - 1), REG2; \
+	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	sllx		REG1, PMD_PADDR_SHIFT, REG1; \
 	andn		REG2, 0x7, REG2; \
 	add		REG1, REG2, REG1;
 
-	/* These macros exists only to make the PMD translator below
-	 * easier to read.  It hides the ELF section switch for the
-	 * sun4v code patching.
-	 */
-#define OR_PTE_BIT_1INSN(REG, NAME)			\
-661:	or		REG, _PAGE_##NAME##_4U, REG;	\
-	.section	.sun4v_1insn_patch, "ax";	\
-	.word		661b;				\
-	or		REG, _PAGE_##NAME##_4V, REG;	\
-	.previous;
-
-#define OR_PTE_BIT_2INSN(REG, TMP, NAME)		\
-661:	sethi		%hi(_PAGE_##NAME##_4U), TMP;	\
-	or		REG, TMP, REG;			\
-	.section	.sun4v_2insn_patch, "ax";	\
-	.word		661b;				\
-	mov		-1, TMP;			\
-	or		REG, _PAGE_##NAME##_4V, REG;	\
-	.previous;
-
-	/* Load into REG the PTE value for VALID, CACHE, and SZHUGE.  */
-#define BUILD_PTE_VALID_SZHUGE_CACHE(REG)				   \
-661:	sethi		%uhi(_PAGE_VALID|_PAGE_SZHUGE_4U), REG;		   \
-	.section	.sun4v_1insn_patch, "ax";			   \
-	.word		661b;						   \
-	sethi		%uhi(_PAGE_VALID), REG;				   \
-	.previous;							   \
-	sllx		REG, 32, REG;					   \
-661:	or		REG, _PAGE_CP_4U|_PAGE_CV_4U, REG;		   \
-	.section	.sun4v_1insn_patch, "ax";			   \
-	.word		661b;						   \
-	or		REG, _PAGE_CP_4V|_PAGE_CV_4V|_PAGE_SZHUGE_4V, REG; \
-	.previous;
-
 	/* PMD has been loaded into REG1, interpret the value, seeing
 	 * if it is a HUGE PMD or a normal one.  If it is not valid
 	 * then jump to FAIL_LABEL.  If it is a HUGE PMD, and it
 	 * translates to a valid PTE, branch to PTE_LABEL.
 	 *
-	 * We translate the PMD by hand, one bit at a time,
+	 * We have to propagate the 4MB bit of the virtual address
-	 * constructing the huge PTE.
+	 * because we are fabricating 8MB pages using 4MB hw pages.
-	 *
-	 * So we construct the PTE in REG2 as follows:
-	 *
-	 * 1) Extract the PMD PFN from REG1 and place it into REG2.
-	 *
-	 * 2) Translate PMD protection bits in REG1 into REG2, one bit
-	 *    at a time using andcc tests on REG1 and OR's into REG2.
-	 *
-	 *    Only two bits to be concerned with here, EXEC and WRITE.
-	 *    Now REG1 is freed up and we can use it as a temporary.
-	 *
-	 * 3) Construct the VALID, CACHE, and page size PTE bits in
-	 *    REG1, OR with REG2 to form final PTE.
 	 */
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
 #define USER_PGTABLE_CHECK_PMD_HUGE(VADDR, REG1, REG2, FAIL_LABEL, PTE_LABEL) \
-	brz,pn		REG1, FAIL_LABEL;				      \
+	brz,pn		REG1, FAIL_LABEL;		\
-	 andcc		REG1, PMD_ISHUGE, %g0;				      \
+	 sethi		%uhi(_PAGE_PMD_HUGE), REG2;	\
-	be,pt		%xcc, 700f;					      \
+	sllx		REG2, 32, REG2;			\
-	 and		REG1, PMD_HUGE_PRESENT|PMD_HUGE_ACCESSED, REG2;	      \
+	andcc		REG1, REG2, %g0;		\
-	cmp		REG2, PMD_HUGE_PRESENT|PMD_HUGE_ACCESSED;	      \
+	be,pt		%xcc, 700f;			\
-	bne,pn		%xcc, FAIL_LABEL;				      \
+	 sethi		%hi(4 * 1024 * 1024), REG2;	\
-	 andn		REG1, PMD_HUGE_PROTBITS, REG2;			      \
+	andn		REG1, REG2, REG1;		\
-	sllx		REG2, PMD_PADDR_SHIFT, REG2;			      \
+	and		VADDR, REG2, REG2;		\
-	/* REG2 now holds PFN << PAGE_SHIFT */				      \
+	brlz,pt		REG1, PTE_LABEL;		\
-	andcc		REG1, PMD_HUGE_WRITE, %g0;			      \
+	 or		REG1, REG2, REG1;		\
-	bne,a,pt	%xcc, 1f;					      \
-	 OR_PTE_BIT_1INSN(REG2, W);					      \
-1:	andcc		REG1, PMD_HUGE_EXEC, %g0;			      \
-	be,pt		%xcc, 1f;					      \
-	 nop;								      \
-	OR_PTE_BIT_2INSN(REG2, REG1, EXEC);				      \
-	/* REG1 can now be clobbered, build final PTE */		      \
-1:	BUILD_PTE_VALID_SZHUGE_CACHE(REG1);				      \
-	ba,pt		%xcc, PTE_LABEL;				      \
-	 or		REG1, REG2, REG1;				      \
 700:
 #else
 #define USER_PGTABLE_CHECK_PMD_HUGE(VADDR, REG1, REG2, FAIL_LABEL, PTE_LABEL) \
@@ -253,18 +194,16 @@ extern struct tsb_phys_patch_entry __tsb_phys_patch, __tsb_phys_patch_end;
 #define USER_PGTABLE_WALK_TL1(VADDR, PHYS_PGD, REG1, REG2, FAIL_LABEL)	\
 	sllx		VADDR, 64 - (PGDIR_SHIFT + PGDIR_BITS), REG2; \
 	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	andn		REG2, 0x3, REG2; \
+	andn		REG2, 0x7, REG2; \
-	lduwa		[PHYS_PGD + REG2] ASI_PHYS_USE_EC, REG1; \
+	ldxa		[PHYS_PGD + REG2] ASI_PHYS_USE_EC, REG1; \
 	brz,pn		REG1, FAIL_LABEL; \
 	 sllx		VADDR, 64 - (PMD_SHIFT + PMD_BITS), REG2; \
 	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	sllx		REG1, PGD_PADDR_SHIFT, REG1; \
+	andn		REG2, 0x7, REG2; \
-	andn		REG2, 0x3, REG2; \
+	ldxa		[REG1 + REG2] ASI_PHYS_USE_EC, REG1; \
-	lduwa		[REG1 + REG2] ASI_PHYS_USE_EC, REG1; \
 	USER_PGTABLE_CHECK_PMD_HUGE(VADDR, REG1, REG2, FAIL_LABEL, 800f) \
 	sllx		VADDR, 64 - PMD_SHIFT, REG2; \
-	srlx		REG2, 64 - (PAGE_SHIFT - 1), REG2; \
+	srlx		REG2, 64 - PAGE_SHIFT, REG2; \
-	sllx		REG1, PMD_PADDR_SHIFT, REG1; \
 	andn		REG2, 0x7, REG2; \
 	add		REG1, REG2, REG1; \
 	ldxa		[REG1] ASI_PHYS_USE_EC, REG1; \

arch/sparc/kernel/entry.h

@@ -88,7 +88,6 @@ extern asmlinkage void syscall_trace_leave(struct pt_regs *regs);
 
 extern void bad_trap_tl1(struct pt_regs *regs, long lvl);
 
-extern void do_fpe_common(struct pt_regs *regs);
 extern void do_fpieee(struct pt_regs *regs);
 extern void do_fpother(struct pt_regs *regs);
 extern void do_tof(struct pt_regs *regs);

arch/sparc/kernel/kgdb_64.c

@@ -159,11 +159,12 @@ int kgdb_arch_handle_exception(int e_vector, int signo, int err_code,
 
 asmlinkage void kgdb_trap(unsigned long trap_level, struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	unsigned long flags;
 
 	if (user_mode(regs)) {
 		bad_trap(regs, trap_level);
-		return;
+		goto out;
 	}
 
 	flushw_all();
@@ -171,6 +172,8 @@ asmlinkage void kgdb_trap(unsigned long trap_level, struct pt_regs *regs)
 	local_irq_save(flags);
 	kgdb_handle_exception(0x172, SIGTRAP, 0, regs);
 	local_irq_restore(flags);
+out:
+	exception_exit(prev_state);
 }
 
 int kgdb_arch_init(void)

arch/sparc/kernel/kprobes.c

@@ -8,6 +8,7 @@
 #include <linux/module.h>
 #include <linux/kdebug.h>
 #include <linux/slab.h>
+#include <linux/context_tracking.h>
 #include <asm/signal.h>
 #include <asm/cacheflush.h>
 #include <asm/uaccess.h>
@@ -418,12 +419,14 @@ int __kprobes kprobe_exceptions_notify(struct notifier_block *self,
 asmlinkage void __kprobes kprobe_trap(unsigned long trap_level,
 				      struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
+
 	BUG_ON(trap_level != 0x170 && trap_level != 0x171);
 
 	if (user_mode(regs)) {
 		local_irq_enable();
 		bad_trap(regs, trap_level);
-		return;
+		goto out;
 	}
 
 	/* trap_level == 0x170 --> ta 0x70
@@ -433,6 +436,8 @@ asmlinkage void __kprobes kprobe_trap(unsigned long trap_level,
 		       (trap_level == 0x170) ? "debug" : "debug_2",
 		       regs, 0, trap_level, SIGTRAP) != NOTIFY_STOP)
 		bad_trap(regs, trap_level);
+out:
+	exception_exit(prev_state);
 }
 
 /* Jprobes support.  */

arch/sparc/kernel/ktlb.S

@@ -153,12 +153,19 @@ kvmap_dtlb_tsb4m_miss:
 	/* Clear the PAGE_OFFSET top virtual bits, shift
 	 * down to get PFN, and make sure PFN is in range.
 	 */
-	sllx		%g4, 21, %g5
+661:	sllx		%g4, 0, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
 
 	/* Check to see if we know about valid memory at the 4MB
 	 * chunk this physical address will reside within.
 	 */
-	srlx		%g5, 21 + 41, %g2
+661:	srlx		%g5, MAX_PHYS_ADDRESS_BITS, %g2
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	brnz,pn		%g2, kvmap_dtlb_longpath
 	 nop
 
@@ -176,7 +183,11 @@ valid_addr_bitmap_patch:
 	or		%g7, %lo(sparc64_valid_addr_bitmap), %g7
 	.previous
 
-	srlx		%g5, 21 + 22, %g2
+661:	srlx		%g5, ILOG2_4MB, %g2
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	srlx		%g2, 6, %g5
 	and		%g2, 63, %g2
 	sllx		%g5, 3, %g5
@@ -189,9 +200,18 @@ valid_addr_bitmap_patch:
 2:	 sethi		%hi(kpte_linear_bitmap), %g2
 
 	/* Get the 256MB physical address index. */
-	sllx		%g4, 21, %g5
+661:	sllx		%g4, 0, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	or		%g2, %lo(kpte_linear_bitmap), %g2
-	srlx		%g5, 21 + 28, %g5
+
+661:	srlx		%g5, ILOG2_256MB, %g5
+	.section	.page_offset_shift_patch, "ax"
+	.word		661b
+	.previous
+
 	and		%g5, (32 - 1), %g7
 
 	/* Divide by 32 to get the offset into the bitmask.  */

arch/sparc/kernel/pci.c

@@ -398,8 +398,8 @@ static void apb_fake_ranges(struct pci_dev *dev,
 	apb_calc_first_last(map, &first, &last);
 	res = bus->resource[1];
 	res->flags = IORESOURCE_MEM;
-	region.start = (first << 21);
+	region.start = (first << 29);
-	region.end = (last << 21) + ((1 << 21) - 1);
+	region.end = (last << 29) + ((1 << 29) - 1);
 	pcibios_bus_to_resource(dev, res, &region);
 }
 

arch/sparc/kernel/process_64.c

@@ -31,6 +31,7 @@
 #include <linux/elfcore.h>
 #include <linux/sysrq.h>
 #include <linux/nmi.h>
+#include <linux/context_tracking.h>
 
 #include <asm/uaccess.h>
 #include <asm/page.h>
@@ -557,6 +558,7 @@ void fault_in_user_windows(void)
 
 barf:
 	set_thread_wsaved(window + 1);
+	user_exit();
 	do_exit(SIGILL);
 }
 

arch/sparc/kernel/ptrace_64.c

@@ -27,6 +27,7 @@
 #include <trace/syscall.h>
 #include <linux/compat.h>
 #include <linux/elf.h>
+#include <linux/context_tracking.h>
 
 #include <asm/asi.h>
 #include <asm/pgtable.h>
@@ -1066,6 +1067,9 @@ asmlinkage int syscall_trace_enter(struct pt_regs *regs)
 	/* do the secure computing check first */
 	secure_computing_strict(regs->u_regs[UREG_G1]);
 
+	if (test_thread_flag(TIF_NOHZ))
+		user_exit();
+
 	if (test_thread_flag(TIF_SYSCALL_TRACE))
 		ret = tracehook_report_syscall_entry(regs);
 
@@ -1086,6 +1090,9 @@ asmlinkage int syscall_trace_enter(struct pt_regs *regs)
 
 asmlinkage void syscall_trace_leave(struct pt_regs *regs)
 {
+	if (test_thread_flag(TIF_NOHZ))
+		user_exit();
+
 	audit_syscall_exit(regs);
 
 	if (unlikely(test_thread_flag(TIF_SYSCALL_TRACEPOINT)))
@@ -1093,4 +1100,7 @@ asmlinkage void syscall_trace_leave(struct pt_regs *regs)
 
 	if (test_thread_flag(TIF_SYSCALL_TRACE))
 		tracehook_report_syscall_exit(regs, 0);
+
+	if (test_thread_flag(TIF_NOHZ))
+		user_enter();
 }

arch/sparc/kernel/rtrap_64.S

@@ -18,10 +18,16 @@
 #define		RTRAP_PSTATE_IRQOFF	(PSTATE_TSO|PSTATE_PEF|PSTATE_PRIV)
 #define		RTRAP_PSTATE_AG_IRQOFF	(PSTATE_TSO|PSTATE_PEF|PSTATE_PRIV|PSTATE_AG)
 
+#ifdef CONFIG_CONTEXT_TRACKING
+# define SCHEDULE_USER schedule_user
+#else
+# define SCHEDULE_USER schedule
+#endif
+
 		.text
 		.align			32
 __handle_preemption:
-		call			schedule
+		call			SCHEDULE_USER
 		 wrpr			%g0, RTRAP_PSTATE, %pstate
 		ba,pt			%xcc, __handle_preemption_continue
 		 wrpr			%g0, RTRAP_PSTATE_IRQOFF, %pstate

arch/sparc/kernel/signal_64.c

@@ -23,6 +23,7 @@
 #include <linux/tty.h>
 #include <linux/binfmts.h>
 #include <linux/bitops.h>
+#include <linux/context_tracking.h>
 
 #include <asm/uaccess.h>
 #include <asm/ptrace.h>
@@ -43,6 +44,7 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 {
 	struct ucontext __user *ucp = (struct ucontext __user *)
 		regs->u_regs[UREG_I0];
+	enum ctx_state prev_state = exception_enter();
 	mc_gregset_t __user *grp;
 	unsigned long pc, npc, tstate;
 	unsigned long fp, i7;
@@ -129,16 +131,19 @@ asmlinkage void sparc64_set_context(struct pt_regs *regs)
 	}
 	if (err)
 		goto do_sigsegv;
-
+out:
+	exception_exit(prev_state);
 	return;
 do_sigsegv:
 	force_sig(SIGSEGV, current);
+	goto out;
 }
 
 asmlinkage void sparc64_get_context(struct pt_regs *regs)
 {
 	struct ucontext __user *ucp = (struct ucontext __user *)
 		regs->u_regs[UREG_I0];
+	enum ctx_state prev_state = exception_enter();
 	mc_gregset_t __user *grp;
 	mcontext_t __user *mcp;
 	unsigned long fp, i7;
@@ -220,10 +225,12 @@ asmlinkage void sparc64_get_context(struct pt_regs *regs)
 	}
 	if (err)
 		goto do_sigsegv;
-
+out:
+	exception_exit(prev_state);
 	return;
 do_sigsegv:
 	force_sig(SIGSEGV, current);
+	goto out;
 }
 
 struct rt_signal_frame {
@@ -528,11 +535,13 @@ static void do_signal(struct pt_regs *regs, unsigned long orig_i0)
 
 void do_notify_resume(struct pt_regs *regs, unsigned long orig_i0, unsigned long thread_info_flags)
 {
+	user_exit();
 	if (thread_info_flags & _TIF_SIGPENDING)
 		do_signal(regs, orig_i0);
 	if (thread_info_flags & _TIF_NOTIFY_RESUME) {
 		clear_thread_flag(TIF_NOTIFY_RESUME);
 		tracehook_notify_resume(regs);
 	}
+	user_enter();
 }
 

arch/sparc/kernel/smp_64.c

@@ -1399,8 +1399,13 @@ void __init smp_cpus_done(unsigned int max_cpus)
 
 void smp_send_reschedule(int cpu)
 {
-	xcall_deliver((u64) &xcall_receive_signal, 0, 0,
+	if (cpu == smp_processor_id()) {
-		      cpumask_of(cpu));
+		WARN_ON_ONCE(preemptible());
+		set_softint(1 << PIL_SMP_RECEIVE_SIGNAL);
+	} else {
+		xcall_deliver((u64) &xcall_receive_signal,
+			      0, 0, cpumask_of(cpu));
+	}
 }
 
 void __irq_entry smp_receive_signal_client(int irq, struct pt_regs *regs)

arch/sparc/kernel/sun4v_tlb_miss.S

@@ -182,7 +182,7 @@ sun4v_tsb_miss_common:
 	cmp	%g5, -1
 	be,pt	%xcc, 80f
 	 nop
-	COMPUTE_TSB_PTR(%g5, %g4, HPAGE_SHIFT, %g2, %g7)
+	COMPUTE_TSB_PTR(%g5, %g4, REAL_HPAGE_SHIFT, %g2, %g7)
 
 	/* That clobbered %g2, reload it.  */
 	ldxa	[%g0] ASI_SCRATCHPAD, %g2

arch/sparc/kernel/sys_sparc_64.c

@@ -24,6 +24,7 @@
 #include <linux/personality.h>
 #include <linux/random.h>
 #include <linux/export.h>
+#include <linux/context_tracking.h>
 
 #include <asm/uaccess.h>
 #include <asm/utrap.h>
@@ -39,9 +40,6 @@ asmlinkage unsigned long sys_getpagesize(void)
 	return PAGE_SIZE;
 }
 
-#define VA_EXCLUDE_START (0x0000080000000000UL - (1UL << 32UL))
-#define VA_EXCLUDE_END   (0xfffff80000000000UL + (1UL << 32UL))
-
 /* Does addr --> addr+len fall within 4GB of the VA-space hole or
  * overflow past the end of the 64-bit address space?
  */
@@ -499,6 +497,7 @@ asmlinkage unsigned long c_sys_nis_syscall(struct pt_regs *regs)
 
 asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (test_thread_flag(TIF_32BIT)) {
@@ -517,6 +516,7 @@ asmlinkage void sparc_breakpoint(struct pt_regs *regs)
 #ifdef DEBUG_SPARC_BREAKPOINT
 	printk ("TRAP: Returning to space: PC=%lx nPC=%lx\n", regs->tpc, regs->tnpc);
 #endif
+	exception_exit(prev_state);
 }
 
 extern void check_pending(int signum);

arch/sparc/kernel/syscalls.S

@@ -52,7 +52,7 @@ sys32_rt_sigreturn:
 #endif
 	.align	32
 1:	ldx	[%g6 + TI_FLAGS], %l5
-	andcc	%l5, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT), %g0
+	andcc	%l5, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT|_TIF_NOHZ), %g0
 	be,pt	%icc, rtrap
 	 nop
 	call	syscall_trace_leave
@@ -184,7 +184,7 @@ linux_sparc_syscall32:
 
 	srl	%i3, 0, %o3				! IEU0
 	srl	%i2, 0, %o2				! IEU0	Group
-	andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT), %g0
+	andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT|_TIF_NOHZ), %g0
 	bne,pn	%icc, linux_syscall_trace32		! CTI
 	 mov	%i0, %l5				! IEU1
 5:	call	%l7					! CTI	Group brk forced
@@ -207,7 +207,7 @@ linux_sparc_syscall:
 
 	mov	%i3, %o3				! IEU1
 	mov	%i4, %o4				! IEU0	Group
-	andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT), %g0
+	andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT|_TIF_NOHZ), %g0
 	bne,pn	%icc, linux_syscall_trace		! CTI	Group
 	 mov	%i0, %l5				! IEU0
 2:	call	%l7					! CTI	Group brk forced
@@ -223,7 +223,7 @@ ret_sys_call:
 
 	cmp	%o0, -ERESTART_RESTARTBLOCK
 	bgeu,pn	%xcc, 1f
-	 andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT), %g0
+	 andcc	%l0, (_TIF_SYSCALL_TRACE|_TIF_SECCOMP|_TIF_SYSCALL_AUDIT|_TIF_SYSCALL_TRACEPOINT|_TIF_NOHZ), %g0
 	ldx	[%sp + PTREGS_OFF + PT_V9_TNPC], %l1 ! pc = npc
 
 2:

arch/sparc/kernel/traps_64.c

@@ -20,6 +20,7 @@
 #include <linux/ftrace.h>
 #include <linux/reboot.h>
 #include <linux/gfp.h>
+#include <linux/context_tracking.h>
 
 #include <asm/smp.h>
 #include <asm/delay.h>
@@ -186,11 +187,12 @@ EXPORT_SYMBOL_GPL(unregister_dimm_printer);
 
 void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "instruction access exception", regs,
 		       0, 0x8, SIGTRAP) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (regs->tstate & TSTATE_PRIV) {
 		printk("spitfire_insn_access_exception: SFSR[%016lx] "
@@ -207,6 +209,8 @@ void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, un
 	info.si_addr = (void __user *)regs->tpc;
 	info.si_trapno = 0;
 	force_sig_info(SIGSEGV, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 void spitfire_insn_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
@@ -260,11 +264,12 @@ void sun4v_insn_access_exception_tl1(struct pt_regs *regs, unsigned long addr, u
 
 void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "data access exception", regs,
 		       0, 0x30, SIGTRAP) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (regs->tstate & TSTATE_PRIV) {
 		/* Test if this comes from uaccess places. */
@@ -280,7 +285,7 @@ void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, un
 #endif
 			regs->tpc = entry->fixup;
 			regs->tnpc = regs->tpc + 4;
-			return;
+			goto out;
 		}
 		/* Shit... */
 		printk("spitfire_data_access_exception: SFSR[%016lx] "
@@ -294,6 +299,8 @@ void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, un
 	info.si_addr = (void __user *)sfar;
 	info.si_trapno = 0;
 	force_sig_info(SIGSEGV, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 void spitfire_data_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
@@ -1994,6 +2001,7 @@ static void sun4v_log_error(struct pt_regs *regs, struct sun4v_error_entry *ent,
  */
 void sun4v_resum_error(struct pt_regs *regs, unsigned long offset)
 {
+	enum ctx_state prev_state = exception_enter();
 	struct sun4v_error_entry *ent, local_copy;
 	struct trap_per_cpu *tb;
 	unsigned long paddr;
@@ -2022,12 +2030,14 @@ void sun4v_resum_error(struct pt_regs *regs, unsigned long offset)
 		pr_info("Shutdown request, %u seconds...\n",
 			local_copy.err_secs);
 		orderly_poweroff(true);
-		return;
+		goto out;
 	}
 
 	sun4v_log_error(regs, &local_copy, cpu,
 			KERN_ERR "RESUMABLE ERROR",
 			&sun4v_resum_oflow_cnt);
+out:
+	exception_exit(prev_state);
 }
 
 /* If we try to printk() we'll probably make matters worse, by trying
@@ -2152,7 +2162,7 @@ void hypervisor_tlbop_error_xcall(unsigned long err, unsigned long op)
 	       err, op);
 }
 
-void do_fpe_common(struct pt_regs *regs)
+static void do_fpe_common(struct pt_regs *regs)
 {
 	if (regs->tstate & TSTATE_PRIV) {
 		regs->tpc = regs->tnpc;
@@ -2188,23 +2198,28 @@ void do_fpe_common(struct pt_regs *regs)
 
 void do_fpieee(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
+
 	if (notify_die(DIE_TRAP, "fpu exception ieee", regs,
 		       0, 0x24, SIGFPE) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	do_fpe_common(regs);
+out:
+	exception_exit(prev_state);
 }
 
 extern int do_mathemu(struct pt_regs *, struct fpustate *, bool);
 
 void do_fpother(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	struct fpustate *f = FPUSTATE;
 	int ret = 0;
 
 	if (notify_die(DIE_TRAP, "fpu exception other", regs,
 		       0, 0x25, SIGFPE) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	switch ((current_thread_info()->xfsr[0] & 0x1c000)) {
 	case (2 << 14): /* unfinished_FPop */
@@ -2213,17 +2228,20 @@ void do_fpother(struct pt_regs *regs)
 		break;
 	}
 	if (ret)
-		return;
+		goto out;
 	do_fpe_common(regs);
+out:
+	exception_exit(prev_state);
 }
 
 void do_tof(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "tagged arithmetic overflow", regs,
 		       0, 0x26, SIGEMT) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (regs->tstate & TSTATE_PRIV)
 		die_if_kernel("Penguin overflow trap from kernel mode", regs);
@@ -2237,15 +2255,18 @@ void do_tof(struct pt_regs *regs)
 	info.si_addr = (void __user *)regs->tpc;
 	info.si_trapno = 0;
 	force_sig_info(SIGEMT, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 void do_div0(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "integer division by zero", regs,
 		       0, 0x28, SIGFPE) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (regs->tstate & TSTATE_PRIV)
 		die_if_kernel("TL0: Kernel divide by zero.", regs);
@@ -2259,6 +2280,8 @@ void do_div0(struct pt_regs *regs)
 	info.si_addr = (void __user *)regs->tpc;
 	info.si_trapno = 0;
 	force_sig_info(SIGFPE, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 static void instruction_dump(unsigned int *pc)
@@ -2415,6 +2438,7 @@ extern int handle_ldf_stq(u32 insn, struct pt_regs *regs);
 
 void do_illegal_instruction(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	unsigned long pc = regs->tpc;
 	unsigned long tstate = regs->tstate;
 	u32 insn;
@@ -2422,7 +2446,7 @@ void do_illegal_instruction(struct pt_regs *regs)
 
 	if (notify_die(DIE_TRAP, "illegal instruction", regs,
 		       0, 0x10, SIGILL) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (tstate & TSTATE_PRIV)
 		die_if_kernel("Kernel illegal instruction", regs);
@@ -2431,14 +2455,14 @@ void do_illegal_instruction(struct pt_regs *regs)
 	if (get_user(insn, (u32 __user *) pc) != -EFAULT) {
 		if ((insn & 0xc1ffc000) == 0x81700000) /* POPC */ {
 			if (handle_popc(insn, regs))
-				return;
+				goto out;
 		} else if ((insn & 0xc1580000) == 0xc1100000) /* LDQ/STQ */ {
 			if (handle_ldf_stq(insn, regs))
-				return;
+				goto out;
 		} else if (tlb_type == hypervisor) {
 			if ((insn & VIS_OPCODE_MASK) == VIS_OPCODE_VAL) {
 				if (!vis_emul(regs, insn))
-					return;
+					goto out;
 			} else {
 				struct fpustate *f = FPUSTATE;
 
@@ -2448,7 +2472,7 @@ void do_illegal_instruction(struct pt_regs *regs)
 				 * Trap in the %fsr to unimplemented_FPop.
 				 */
 				if (do_mathemu(regs, f, true))
-					return;
+					goto out;
 			}
 		}
 	}
@@ -2458,21 +2482,24 @@ void do_illegal_instruction(struct pt_regs *regs)
 	info.si_addr = (void __user *)pc;
 	info.si_trapno = 0;
 	force_sig_info(SIGILL, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 extern void kernel_unaligned_trap(struct pt_regs *regs, unsigned int insn);
 
 void mem_address_unaligned(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "memory address unaligned", regs,
 		       0, 0x34, SIGSEGV) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (regs->tstate & TSTATE_PRIV) {
 		kernel_unaligned_trap(regs, *((unsigned int *)regs->tpc));
-		return;
+		goto out;
 	}
 	info.si_signo = SIGBUS;
 	info.si_errno = 0;
@@ -2480,6 +2507,8 @@ void mem_address_unaligned(struct pt_regs *regs, unsigned long sfar, unsigned lo
 	info.si_addr = (void __user *)sfar;
 	info.si_trapno = 0;
 	force_sig_info(SIGBUS, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 void sun4v_do_mna(struct pt_regs *regs, unsigned long addr, unsigned long type_ctx)
@@ -2504,11 +2533,12 @@ void sun4v_do_mna(struct pt_regs *regs, unsigned long addr, unsigned long type_c
 
 void do_privop(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	siginfo_t info;
 
 	if (notify_die(DIE_TRAP, "privileged operation", regs,
 		       0, 0x11, SIGILL) == NOTIFY_STOP)
-		return;
+		goto out;
 
 	if (test_thread_flag(TIF_32BIT)) {
 		regs->tpc &= 0xffffffff;
@@ -2520,6 +2550,8 @@ void do_privop(struct pt_regs *regs)
 	info.si_addr = (void __user *)regs->tpc;
 	info.si_trapno = 0;
 	force_sig_info(SIGILL, &info, current);
+out:
+	exception_exit(prev_state);
 }
 
 void do_privact(struct pt_regs *regs)
@@ -2530,99 +2562,116 @@ void do_privact(struct pt_regs *regs)
 /* Trap level 1 stuff or other traps we should never see... */
 void do_cee(struct pt_regs *regs)
 {
+	exception_enter();
 	die_if_kernel("TL0: Cache Error Exception", regs);
 }
 
 void do_cee_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Cache Error Exception", regs);
 }
 
 void do_dae_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Data Access Exception", regs);
 }
 
 void do_iae_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Instruction Access Exception", regs);
 }
 
 void do_div0_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: DIV0 Exception", regs);
 }
 
 void do_fpdis_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: FPU Disabled", regs);
 }
 
 void do_fpieee_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: FPU IEEE Exception", regs);
 }
 
 void do_fpother_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: FPU Other Exception", regs);
 }
 
 void do_ill_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Illegal Instruction Exception", regs);
 }
 
 void do_irq_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: IRQ Exception", regs);
 }
 
 void do_lddfmna_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: LDDF Exception", regs);
 }
 
 void do_stdfmna_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: STDF Exception", regs);
 }
 
 void do_paw(struct pt_regs *regs)
 {
+	exception_enter();
 	die_if_kernel("TL0: Phys Watchpoint Exception", regs);
 }
 
 void do_paw_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Phys Watchpoint Exception", regs);
 }
 
 void do_vaw(struct pt_regs *regs)
 {
+	exception_enter();
 	die_if_kernel("TL0: Virt Watchpoint Exception", regs);
 }
 
 void do_vaw_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Virt Watchpoint Exception", regs);
 }
 
 void do_tof_tl1(struct pt_regs *regs)
 {
+	exception_enter();
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
 	die_if_kernel("TL1: Tag Overflow Exception", regs);
 }

arch/sparc/kernel/tsb.S

@@ -75,7 +75,7 @@ tsb_miss_page_table_walk:
 	mov		512, %g7
 	andn		%g5, 0x7, %g5
 	sllx		%g7, %g6, %g7
-	srlx		%g4, HPAGE_SHIFT, %g6
+	srlx		%g4, REAL_HPAGE_SHIFT, %g6
 	sub		%g7, 1, %g7
 	and		%g6, %g7, %g6
 	sllx		%g6, 4, %g6

arch/sparc/kernel/unaligned_64.c

@@ -21,9 +21,12 @@
 #include <linux/bitops.h>
 #include <linux/perf_event.h>
 #include <linux/ratelimit.h>
+#include <linux/context_tracking.h>
 #include <asm/fpumacro.h>
 #include <asm/cacheflush.h>
 
+#include "entry.h"
+
 enum direction {
 	load,    /* ld, ldd, ldh, ldsh */
 	store,   /* st, std, sth, stsh */
@@ -418,9 +421,6 @@ int handle_popc(u32 insn, struct pt_regs *regs)
 
 extern void do_fpother(struct pt_regs *regs);
 extern void do_privact(struct pt_regs *regs);
-extern void spitfire_data_access_exception(struct pt_regs *regs,
-					   unsigned long sfsr,
-					   unsigned long sfar);
 extern void sun4v_data_access_exception(struct pt_regs *regs,
 					unsigned long addr,
 					unsigned long type_ctx);
@@ -578,6 +578,7 @@ void handle_ld_nf(u32 insn, struct pt_regs *regs)
 
 void handle_lddfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr)
 {
+	enum ctx_state prev_state = exception_enter();
 	unsigned long pc = regs->tpc;
 	unsigned long tstate = regs->tstate;
 	u32 insn;
@@ -632,13 +633,16 @@ daex:
 			sun4v_data_access_exception(regs, sfar, sfsr);
 		else
 			spitfire_data_access_exception(regs, sfsr, sfar);
-		return;
+		goto out;
 	}
 	advance(regs);
+out:
+	exception_exit(prev_state);
 }
 
 void handle_stdfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr)
 {
+	enum ctx_state prev_state = exception_enter();
 	unsigned long pc = regs->tpc;
 	unsigned long tstate = regs->tstate;
 	u32 insn;
@@ -680,7 +684,9 @@ daex:
 			sun4v_data_access_exception(regs, sfar, sfsr);
 		else
 			spitfire_data_access_exception(regs, sfsr, sfar);
-		return;
+		goto out;
 	}
 	advance(regs);
+out:
+	exception_exit(prev_state);
 }

arch/sparc/kernel/vmlinux.lds.S

@@ -122,6 +122,11 @@ SECTIONS
 		*(.swapper_4m_tsb_phys_patch)
 		__swapper_4m_tsb_phys_patch_end = .;
 	}
+	.page_offset_shift_patch : {
+		__page_offset_shift_patch = .;
+		*(.page_offset_shift_patch)
+		__page_offset_shift_patch_end = .;
+	}
 	.popc_3insn_patch : {
 		__popc_3insn_patch = .;
 		*(.popc_3insn_patch)

arch/sparc/lib/clear_page.S

@@ -37,10 +37,10 @@ _clear_page:		/* %o0=dest */
 	.globl		clear_user_page
 clear_user_page:	/* %o0=dest, %o1=vaddr */
 	lduw		[%g6 + TI_PRE_COUNT], %o2
-	sethi		%uhi(PAGE_OFFSET), %g2
+	sethi		%hi(PAGE_OFFSET), %g2
 	sethi		%hi(PAGE_SIZE), %o4
 
-	sllx		%g2, 32, %g2
+	ldx		[%g2 + %lo(PAGE_OFFSET)], %g2
 	sethi		%hi(PAGE_KERNEL_LOCKED), %g3
 
 	ldx		[%g3 + %lo(PAGE_KERNEL_LOCKED)], %g3

arch/sparc/lib/copy_page.S

@@ -46,10 +46,10 @@
 	.type		copy_user_page,#function
 copy_user_page:		/* %o0=dest, %o1=src, %o2=vaddr */
 	lduw		[%g6 + TI_PRE_COUNT], %o4
-	sethi		%uhi(PAGE_OFFSET), %g2
+	sethi		%hi(PAGE_OFFSET), %g2
 	sethi		%hi(PAGE_SIZE), %o3
 
-	sllx		%g2, 32, %g2
+	ldx		[%g2 + %lo(PAGE_OFFSET)], %g2
 	sethi		%hi(PAGE_KERNEL_LOCKED), %g3
 
 	ldx		[%g3 + %lo(PAGE_KERNEL_LOCKED)], %g3

arch/sparc/mm/fault_64.c

@@ -21,6 +21,7 @@
 #include <linux/kprobes.h>
 #include <linux/kdebug.h>
 #include <linux/percpu.h>
+#include <linux/context_tracking.h>
 
 #include <asm/page.h>
 #include <asm/pgtable.h>
@@ -272,6 +273,7 @@ static void noinline __kprobes bogus_32bit_fault_address(struct pt_regs *regs,
 
 asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)
 {
+	enum ctx_state prev_state = exception_enter();
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma;
 	unsigned int insn = 0;
@@ -282,7 +284,7 @@ asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)
 	fault_code = get_thread_fault_code();
 
 	if (notify_page_fault(regs))
-		return;
+		goto exit_exception;
 
 	si_code = SEGV_MAPERR;
 	address = current_thread_info()->fault_address;
@@ -313,7 +315,7 @@ asmlinkage void __kprobes do_sparc64_fault(struct pt_regs *regs)
 			/* Valid, no problems... */
 		} else {
 			bad_kernel_pc(regs, address);
-			return;
+			goto exit_exception;
 		}
 	} else
 		flags |= FAULT_FLAG_USER;
@@ -430,7 +432,7 @@ good_area:
 	fault = handle_mm_fault(mm, vma, address, flags);
 
 	if ((fault & VM_FAULT_RETRY) && fatal_signal_pending(current))
-		return;
+		goto exit_exception;
 
 	if (unlikely(fault & VM_FAULT_ERROR)) {
 		if (fault & VM_FAULT_OOM)
@@ -482,6 +484,8 @@ good_area:
 
 	}
 #endif
+exit_exception:
+	exception_exit(prev_state);
 	return;
 
 	/*
@@ -494,7 +498,7 @@ bad_area:
 
 handle_kernel_fault:
 	do_kernel_fault(regs, si_code, fault_code, insn, address);
-	return;
+	goto exit_exception;
 
 /*
  * We ran out of memory, or some other thing happened to us that made
@@ -505,7 +509,7 @@ out_of_memory:
 	up_read(&mm->mmap_sem);
 	if (!(regs->tstate & TSTATE_PRIV)) {
 		pagefault_out_of_memory();
-		return;
+		goto exit_exception;
 	}
 	goto handle_kernel_fault;
 

arch/sparc/mm/gup.c

@@ -71,13 +71,12 @@ static int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr,
 			int *nr)
 {
 	struct page *head, *page, *tail;
-	u32 mask;
 	int refs;
 
-	mask = PMD_HUGE_PRESENT;
+	if (!pmd_large(pmd))
-	if (write)
+		return 0;
-		mask |= PMD_HUGE_WRITE;
+
-	if ((pmd_val(pmd) & mask) != mask)
+	if (write && !pmd_write(pmd))
 		return 0;
 
 	refs = 0;

arch/sparc/mm/hugetlbpage.c

@@ -21,8 +21,6 @@
 /* Slightly simplified from the non-hugepage variant because by
  * definition we don't have to worry about any page coloring stuff
  */
-#define VA_EXCLUDE_START (0x0000080000000000UL - (1UL << 32UL))
-#define VA_EXCLUDE_END   (0xfffff80000000000UL + (1UL << 32UL))
 
 static unsigned long hugetlb_get_unmapped_area_bottomup(struct file *filp,
 							unsigned long addr,

arch/sparc/mm/init_64.c

@@ -354,7 +354,7 @@ void update_mmu_cache(struct vm_area_struct *vma, unsigned long address, pte_t *
 
 #if defined(CONFIG_HUGETLB_PAGE) || defined(CONFIG_TRANSPARENT_HUGEPAGE)
 	if (mm->context.huge_pte_count && is_hugetlb_pte(pte))
-		__update_mmu_tsb_insert(mm, MM_TSB_HUGE, HPAGE_SHIFT,
+		__update_mmu_tsb_insert(mm, MM_TSB_HUGE, REAL_HPAGE_SHIFT,
 					address, pte_val(pte));
 	else
 #endif
@@ -1557,6 +1557,96 @@ unsigned long __init find_ecache_flush_span(unsigned long size)
 	return ~0UL;
 }
 
+unsigned long PAGE_OFFSET;
+EXPORT_SYMBOL(PAGE_OFFSET);
+
+static void __init page_offset_shift_patch_one(unsigned int *insn, unsigned long phys_bits)
+{
+	unsigned long final_shift;
+	unsigned int val = *insn;
+	unsigned int cnt;
+
+	/* We are patching in ilog2(max_supported_phys_address), and
+	 * we are doing so in a manner similar to a relocation addend.
+	 * That is, we are adding the shift value to whatever value
+	 * is in the shift instruction count field already.
+	 */
+	cnt = (val & 0x3f);
+	val &= ~0x3f;
+
+	/* If we are trying to shift >= 64 bits, clear the destination
+	 * register.  This can happen when phys_bits ends up being equal
+	 * to MAX_PHYS_ADDRESS_BITS.
+	 */
+	final_shift = (cnt + (64 - phys_bits));
+	if (final_shift >= 64) {
+		unsigned int rd = (val >> 25) & 0x1f;
+
+		val = 0x80100000 | (rd << 25);
+	} else {
+		val |= final_shift;
+	}
+	*insn = val;
+
+	__asm__ __volatile__("flush	%0"
+			     : /* no outputs */
+			     : "r" (insn));
+}
+
+static void __init page_offset_shift_patch(unsigned long phys_bits)
+{
+	extern unsigned int __page_offset_shift_patch;
+	extern unsigned int __page_offset_shift_patch_end;
+	unsigned int *p;
+
+	p = &__page_offset_shift_patch;
+	while (p < &__page_offset_shift_patch_end) {
+		unsigned int *insn = (unsigned int *)(unsigned long)*p;
+
+		page_offset_shift_patch_one(insn, phys_bits);
+
+		p++;
+	}
+}
+
+static void __init setup_page_offset(void)
+{
+	unsigned long max_phys_bits = 40;
+
+	if (tlb_type == cheetah || tlb_type == cheetah_plus) {
+		max_phys_bits = 42;
+	} else if (tlb_type == hypervisor) {
+		switch (sun4v_chip_type) {
+		case SUN4V_CHIP_NIAGARA1:
+		case SUN4V_CHIP_NIAGARA2:
+			max_phys_bits = 39;
+			break;
+		case SUN4V_CHIP_NIAGARA3:
+			max_phys_bits = 43;
+			break;
+		case SUN4V_CHIP_NIAGARA4:
+		case SUN4V_CHIP_NIAGARA5:
+		case SUN4V_CHIP_SPARC64X:
+		default:
+			max_phys_bits = 47;
+			break;
+		}
+	}
+
+	if (max_phys_bits > MAX_PHYS_ADDRESS_BITS) {
+		prom_printf("MAX_PHYS_ADDRESS_BITS is too small, need %lu\n",
+			    max_phys_bits);
+		prom_halt();
+	}
+
+	PAGE_OFFSET = PAGE_OFFSET_BY_BITS(max_phys_bits);
+
+	pr_info("PAGE_OFFSET is 0x%016lx (max_phys_bits == %lu)\n",
+		PAGE_OFFSET, max_phys_bits);
+
+	page_offset_shift_patch(max_phys_bits);
+}
+
 static void __init tsb_phys_patch(void)
 {
 	struct tsb_ldquad_phys_patch_entry *pquad;
@@ -1722,7 +1812,7 @@ static void __init sun4v_linear_pte_xor_finalize(void)
 #ifndef CONFIG_DEBUG_PAGEALLOC
 	if (cpu_pgsz_mask & HV_PGSZ_MASK_256MB) {
 		kern_linear_pte_xor[1] = (_PAGE_VALID | _PAGE_SZ256MB_4V) ^
-			0xfffff80000000000UL;
+			PAGE_OFFSET;
 		kern_linear_pte_xor[1] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 					   _PAGE_P_4V | _PAGE_W_4V);
 	} else {
@@ -1731,7 +1821,7 @@ static void __init sun4v_linear_pte_xor_finalize(void)
 
 	if (cpu_pgsz_mask & HV_PGSZ_MASK_2GB) {
 		kern_linear_pte_xor[2] = (_PAGE_VALID | _PAGE_SZ2GB_4V) ^
-			0xfffff80000000000UL;
+			PAGE_OFFSET;
 		kern_linear_pte_xor[2] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 					   _PAGE_P_4V | _PAGE_W_4V);
 	} else {
@@ -1740,7 +1830,7 @@ static void __init sun4v_linear_pte_xor_finalize(void)
 
 	if (cpu_pgsz_mask & HV_PGSZ_MASK_16GB) {
 		kern_linear_pte_xor[3] = (_PAGE_VALID | _PAGE_SZ16GB_4V) ^
-			0xfffff80000000000UL;
+			PAGE_OFFSET;
 		kern_linear_pte_xor[3] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 					   _PAGE_P_4V | _PAGE_W_4V);
 	} else {
@@ -1752,7 +1842,7 @@ static void __init sun4v_linear_pte_xor_finalize(void)
 /* paging_init() sets up the page tables */
 
 static unsigned long last_valid_pfn;
-pgd_t swapper_pg_dir[2048];
+pgd_t swapper_pg_dir[PTRS_PER_PGD];
 
 static void sun4u_pgprot_init(void);
 static void sun4v_pgprot_init(void);
@@ -1763,6 +1853,8 @@ void __init paging_init(void)
 	unsigned long real_end, i;
 	int node;
 
+	setup_page_offset();
+
 	/* These build time checkes make sure that the dcache_dirty_cpu()
 	 * page->flags usage will work.
 	 *
@@ -2261,10 +2353,10 @@ static void __init sun4u_pgprot_init(void)
 		     __ACCESS_BITS_4U | _PAGE_E_4U);
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
-	kern_linear_pte_xor[0] = _PAGE_VALID ^ 0xfffff80000000000UL;
+	kern_linear_pte_xor[0] = _PAGE_VALID ^ PAGE_OFFSET;
 #else
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZ4MB_4U) ^
-		0xfffff80000000000UL;
+		PAGE_OFFSET;
 #endif
 	kern_linear_pte_xor[0] |= (_PAGE_CP_4U | _PAGE_CV_4U |
 				   _PAGE_P_4U | _PAGE_W_4U);
@@ -2308,10 +2400,10 @@ static void __init sun4v_pgprot_init(void)
 	_PAGE_CACHE = _PAGE_CACHE_4V;
 
 #ifdef CONFIG_DEBUG_PAGEALLOC
-	kern_linear_pte_xor[0] = _PAGE_VALID ^ 0xfffff80000000000UL;
+	kern_linear_pte_xor[0] = _PAGE_VALID ^ PAGE_OFFSET;
 #else
 	kern_linear_pte_xor[0] = (_PAGE_VALID | _PAGE_SZ4MB_4V) ^
-		0xfffff80000000000UL;
+		PAGE_OFFSET;
 #endif
 	kern_linear_pte_xor[0] |= (_PAGE_CP_4V | _PAGE_CV_4V |
 				   _PAGE_P_4V | _PAGE_W_4V);
@@ -2455,53 +2547,13 @@ void __flush_tlb_all(void)
 			     : : "r" (pstate));
 }
 
-static pte_t *get_from_cache(struct mm_struct *mm)
-{
-	struct page *page;
-	pte_t *ret;
-
-	spin_lock(&mm->page_table_lock);
-	page = mm->context.pgtable_page;
-	ret = NULL;
-	if (page) {
-		void *p = page_address(page);
-
-		mm->context.pgtable_page = NULL;
-
-		ret = (pte_t *) (p + (PAGE_SIZE / 2));
-	}
-	spin_unlock(&mm->page_table_lock);
-
-	return ret;
-}
-
-static struct page *__alloc_for_cache(struct mm_struct *mm)
-{
-	struct page *page = alloc_page(GFP_KERNEL | __GFP_NOTRACK |
-				       __GFP_REPEAT | __GFP_ZERO);
-
-	if (page) {
-		spin_lock(&mm->page_table_lock);
-		if (!mm->context.pgtable_page) {
-			atomic_set(&page->_count, 2);
-			mm->context.pgtable_page = page;
-		}
-		spin_unlock(&mm->page_table_lock);
-	}
-	return page;
-}
-
 pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
 			    unsigned long address)
 {
-	struct page *page;
+	struct page *page = alloc_page(GFP_KERNEL | __GFP_NOTRACK |
-	pte_t *pte;
+				       __GFP_REPEAT | __GFP_ZERO);
+	pte_t *pte = NULL;
 
-	pte = get_from_cache(mm);
-	if (pte)
-		return pte;
-
-	page = __alloc_for_cache(mm);
 	if (page)
 		pte = (pte_t *) page_address(page);
 
@@ -2511,14 +2563,10 @@ pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
 pgtable_t pte_alloc_one(struct mm_struct *mm,
 			unsigned long address)
 {
-	struct page *page;
+	struct page *page = alloc_page(GFP_KERNEL | __GFP_NOTRACK |
-	pte_t *pte;
+				       __GFP_REPEAT | __GFP_ZERO);
+	pte_t *pte = NULL;
 
-	pte = get_from_cache(mm);
-	if (pte)
-		return pte;
-
-	page = __alloc_for_cache(mm);
 	if (!page)
 		return NULL;
 	if (!pgtable_page_ctor(page)) {
@@ -2530,18 +2578,15 @@ pgtable_t pte_alloc_one(struct mm_struct *mm,
 
 void pte_free_kernel(struct mm_struct *mm, pte_t *pte)
 {
-	struct page *page = virt_to_page(pte);
+	free_page((unsigned long)pte);
-	if (put_page_testzero(page))
-		free_hot_cold_page(page, 0);
 }
 
 static void __pte_free(pgtable_t pte)
 {
 	struct page *page = virt_to_page(pte);
-	if (put_page_testzero(page)) {
+
-		pgtable_page_dtor(page);
+	pgtable_page_dtor(page);
-		free_hot_cold_page(page, 0);
+	__free_page(page);
-	}
 }
 
 void pte_free(struct mm_struct *mm, pgtable_t pte)
@@ -2558,124 +2603,27 @@ void pgtable_free(void *table, bool is_page)
 }
 
 #ifdef CONFIG_TRANSPARENT_HUGEPAGE
-static pmd_t pmd_set_protbits(pmd_t pmd, pgprot_t pgprot, bool for_modify)
-{
-	if (pgprot_val(pgprot) & _PAGE_VALID)
-		pmd_val(pmd) |= PMD_HUGE_PRESENT;
-	if (tlb_type == hypervisor) {
-		if (pgprot_val(pgprot) & _PAGE_WRITE_4V)
-			pmd_val(pmd) |= PMD_HUGE_WRITE;
-		if (pgprot_val(pgprot) & _PAGE_EXEC_4V)
-			pmd_val(pmd) |= PMD_HUGE_EXEC;
-
-		if (!for_modify) {
-			if (pgprot_val(pgprot) & _PAGE_ACCESSED_4V)
-				pmd_val(pmd) |= PMD_HUGE_ACCESSED;
-			if (pgprot_val(pgprot) & _PAGE_MODIFIED_4V)
-				pmd_val(pmd) |= PMD_HUGE_DIRTY;
-		}
-	} else {
-		if (pgprot_val(pgprot) & _PAGE_WRITE_4U)
-			pmd_val(pmd) |= PMD_HUGE_WRITE;
-		if (pgprot_val(pgprot) & _PAGE_EXEC_4U)
-			pmd_val(pmd) |= PMD_HUGE_EXEC;
-
-		if (!for_modify) {
-			if (pgprot_val(pgprot) & _PAGE_ACCESSED_4U)
-				pmd_val(pmd) |= PMD_HUGE_ACCESSED;
-			if (pgprot_val(pgprot) & _PAGE_MODIFIED_4U)
-				pmd_val(pmd) |= PMD_HUGE_DIRTY;
-		}
-	}
-
-	return pmd;
-}
-
-pmd_t pfn_pmd(unsigned long page_nr, pgprot_t pgprot)
-{
-	pmd_t pmd;
-
-	pmd_val(pmd) = (page_nr << ((PAGE_SHIFT - PMD_PADDR_SHIFT)));
-	pmd_val(pmd) |= PMD_ISHUGE;
-	pmd = pmd_set_protbits(pmd, pgprot, false);
-	return pmd;
-}
-
-pmd_t pmd_modify(pmd_t pmd, pgprot_t newprot)
-{
-	pmd_val(pmd) &= ~(PMD_HUGE_PRESENT |
-			  PMD_HUGE_WRITE |
-			  PMD_HUGE_EXEC);
-	pmd = pmd_set_protbits(pmd, newprot, true);
-	return pmd;
-}
-
-pgprot_t pmd_pgprot(pmd_t entry)
-{
-	unsigned long pte = 0;
-
-	if (pmd_val(entry) & PMD_HUGE_PRESENT)
-		pte |= _PAGE_VALID;
-
-	if (tlb_type == hypervisor) {
-		if (pmd_val(entry) & PMD_HUGE_PRESENT)
-			pte |= _PAGE_PRESENT_4V;
-		if (pmd_val(entry) & PMD_HUGE_EXEC)
-			pte |= _PAGE_EXEC_4V;
-		if (pmd_val(entry) & PMD_HUGE_WRITE)
-			pte |= _PAGE_W_4V;
-		if (pmd_val(entry) & PMD_HUGE_ACCESSED)
-			pte |= _PAGE_ACCESSED_4V;
-		if (pmd_val(entry) & PMD_HUGE_DIRTY)
-			pte |= _PAGE_MODIFIED_4V;
-		pte |= _PAGE_CP_4V|_PAGE_CV_4V;
-	} else {
-		if (pmd_val(entry) & PMD_HUGE_PRESENT)
-			pte |= _PAGE_PRESENT_4U;
-		if (pmd_val(entry) & PMD_HUGE_EXEC)
-			pte |= _PAGE_EXEC_4U;
-		if (pmd_val(entry) & PMD_HUGE_WRITE)
-			pte |= _PAGE_W_4U;
-		if (pmd_val(entry) & PMD_HUGE_ACCESSED)
-			pte |= _PAGE_ACCESSED_4U;
-		if (pmd_val(entry) & PMD_HUGE_DIRTY)
-			pte |= _PAGE_MODIFIED_4U;
-		pte |= _PAGE_CP_4U|_PAGE_CV_4U;
-	}
-
-	return __pgprot(pte);
-}
-
 void update_mmu_cache_pmd(struct vm_area_struct *vma, unsigned long addr,
 			  pmd_t *pmd)
 {
 	unsigned long pte, flags;
 	struct mm_struct *mm;
 	pmd_t entry = *pmd;
-	pgprot_t prot;
 
 	if (!pmd_large(entry) || !pmd_young(entry))
 		return;
 
-	pte = (pmd_val(entry) & ~PMD_HUGE_PROTBITS);
+	pte = pmd_val(entry);
-	pte <<= PMD_PADDR_SHIFT;
-	pte |= _PAGE_VALID;
 
-	prot = pmd_pgprot(entry);
+	/* We are fabricating 8MB pages using 4MB real hw pages.  */
-
+	pte |= (addr & (1UL << REAL_HPAGE_SHIFT));
-	if (tlb_type == hypervisor)
-		pgprot_val(prot) |= _PAGE_SZHUGE_4V;
-	else
-		pgprot_val(prot) |= _PAGE_SZHUGE_4U;
-
-	pte |= pgprot_val(prot);
 
 	mm = vma->vm_mm;
 
 	spin_lock_irqsave(&mm->context.lock, flags);
 
 	if (mm->context.tsb_block[MM_TSB_HUGE].tsb != NULL)
-		__update_mmu_tsb_insert(mm, MM_TSB_HUGE, HPAGE_SHIFT,
+		__update_mmu_tsb_insert(mm, MM_TSB_HUGE, REAL_HPAGE_SHIFT,
 					addr, pte);
 
 	spin_unlock_irqrestore(&mm->context.lock, flags);

arch/sparc/mm/init_64.h

@@ -1,11 +1,13 @@
 #ifndef _SPARC64_MM_INIT_H
 #define _SPARC64_MM_INIT_H
 
+#include <asm/page.h>
+
 /* Most of the symbols in this file are defined in init.c and
  * marked non-static so that assembler code can get at them.
  */
 
-#define MAX_PHYS_ADDRESS	(1UL << 41UL)
+#define MAX_PHYS_ADDRESS	(1UL << MAX_PHYS_ADDRESS_BITS)
 #define KPTE_BITMAP_CHUNK_SZ		(256UL * 1024UL * 1024UL)
 #define KPTE_BITMAP_BYTES	\
 	((MAX_PHYS_ADDRESS / KPTE_BITMAP_CHUNK_SZ) / 4)

arch/sparc/mm/tlb.c

@@ -161,8 +161,8 @@ void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 	if (mm == &init_mm)
 		return;
 
-	if ((pmd_val(pmd) ^ pmd_val(orig)) & PMD_ISHUGE) {
+	if ((pmd_val(pmd) ^ pmd_val(orig)) & _PAGE_PMD_HUGE) {
-		if (pmd_val(pmd) & PMD_ISHUGE)
+		if (pmd_val(pmd) & _PAGE_PMD_HUGE)
 			mm->context.huge_pte_count++;
 		else
 			mm->context.huge_pte_count--;
@@ -178,13 +178,16 @@ void set_pmd_at(struct mm_struct *mm, unsigned long addr,
 	}
 
 	if (!pmd_none(orig)) {
-		bool exec = ((pmd_val(orig) & PMD_HUGE_EXEC) != 0);
+		pte_t orig_pte = __pte(pmd_val(orig));
+		bool exec = pte_exec(orig_pte);
 
 		addr &= HPAGE_MASK;
-		if (pmd_val(orig) & PMD_ISHUGE)
+		if (pmd_trans_huge(orig)) {
 			tlb_batch_add_one(mm, addr, exec);
-		else
+			tlb_batch_add_one(mm, addr + REAL_HPAGE_SIZE, exec);
+		} else {
 			tlb_batch_pmd_scan(mm, addr, orig, exec);
+		}
 	}
 }
 

arch/sparc/mm/tsb.c

@@ -87,7 +87,7 @@ void flush_tsb_user(struct tlb_batch *tb)
 		nentries = mm->context.tsb_block[MM_TSB_HUGE].tsb_nentries;
 		if (tlb_type == cheetah_plus || tlb_type == hypervisor)
 			base = __pa(base);
-		__flush_tsb_one(tb, HPAGE_SHIFT, base, nentries);
+		__flush_tsb_one(tb, REAL_HPAGE_SHIFT, base, nentries);
 	}
 #endif
 	spin_unlock_irqrestore(&mm->context.lock, flags);
@@ -111,7 +111,7 @@ void flush_tsb_user_page(struct mm_struct *mm, unsigned long vaddr)
 		nentries = mm->context.tsb_block[MM_TSB_HUGE].tsb_nentries;
 		if (tlb_type == cheetah_plus || tlb_type == hypervisor)
 			base = __pa(base);
-		__flush_tsb_one_entry(base, vaddr, HPAGE_SHIFT, nentries);
+		__flush_tsb_one_entry(base, vaddr, REAL_HPAGE_SHIFT, nentries);
 	}
 #endif
 	spin_unlock_irqrestore(&mm->context.lock, flags);
@@ -472,8 +472,6 @@ int init_new_context(struct task_struct *tsk, struct mm_struct *mm)
 	mm->context.huge_pte_count = 0;
 #endif
 
-	mm->context.pgtable_page = NULL;
-
 	/* copy_mm() copies over the parent's mm_struct before calling
 	 * us, so we need to zero out the TSB pointer or else tsb_grow()
 	 * will be confused and think there is an older TSB to free up.
@@ -512,17 +510,10 @@ static void tsb_destroy_one(struct tsb_config *tp)
 void destroy_context(struct mm_struct *mm)
 {
 	unsigned long flags, i;
-	struct page *page;
 
 	for (i = 0; i < MM_NUM_TSBS; i++)
 		tsb_destroy_one(&mm->context.tsb_block[i]);
 
-	page = mm->context.pgtable_page;
-	if (page && put_page_testzero(page)) {
-		pgtable_page_dtor(page);
-		free_hot_cold_page(page, 0);
-	}
-
 	spin_lock_irqsave(&ctx_alloc_lock, flags);
 
 	if (CTX_VALID(mm->context)) {

arch/sparc/mm/ultra.S

@@ -153,10 +153,10 @@ __spitfire_flush_tlb_mm_slow:
 	.globl		__flush_icache_page
 __flush_icache_page:	/* %o0 = phys_page */
 	srlx		%o0, PAGE_SHIFT, %o0
-	sethi		%uhi(PAGE_OFFSET), %g1
+	sethi		%hi(PAGE_OFFSET), %g1
 	sllx		%o0, PAGE_SHIFT, %o0
 	sethi		%hi(PAGE_SIZE), %g2
-	sllx		%g1, 32, %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	add		%o0, %g1, %o0
 1:	subcc		%g2, 32, %g2
 	bne,pt		%icc, 1b
@@ -178,8 +178,8 @@ __flush_icache_page:	/* %o0 = phys_page */
 	.align		64
 	.globl		__flush_dcache_page
 __flush_dcache_page:	/* %o0=kaddr, %o1=flush_icache */
-	sethi		%uhi(PAGE_OFFSET), %g1
+	sethi		%hi(PAGE_OFFSET), %g1
-	sllx		%g1, 32, %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	sub		%o0, %g1, %o0			! physical address
 	srlx		%o0, 11, %o0			! make D-cache TAG
 	sethi		%hi(1 << 14), %o2		! D-cache size
@@ -287,8 +287,8 @@ __cheetah_flush_tlb_pending:	/* 27 insns */
 
 #ifdef DCACHE_ALIASING_POSSIBLE
 __cheetah_flush_dcache_page: /* 11 insns */
-	sethi		%uhi(PAGE_OFFSET), %g1
+	sethi		%hi(PAGE_OFFSET), %g1
-	sllx		%g1, 32, %g1
+	ldx		[%g1 + %lo(PAGE_OFFSET)], %g1
 	sub		%o0, %g1, %o0
 	sethi		%hi(PAGE_SIZE), %o4
 1:	subcc		%o4, (1 << 5), %o4