[PATCH 3.14 089/100] sparc64: Switch to 4-level page tables.

From: Greg Kroah-Hartman
Date: Mon Oct 27 2014 - 23:50:54 EST

3.14-stable review patch. If anyone has any objections, please let me know.

------------------

From: "David S. Miller" <davem@xxxxxxxxxxxxx>

[ Upstream commit ac55c768143aa34cc3789c4820cbb0809a76fd9c ]

This has become necessary with chips that support more than 43-bits
of physical addressing.

Based almost entirely upon a patch by Bob Picco.

Signed-off-by: David S. Miller <davem@xxxxxxxxxxxxx>
Acked-by: Bob Picco <bob.picco@xxxxxxxxxx>
Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
---
arch/sparc/include/asm/page_64.h | 6 +++++
arch/sparc/include/asm/pgalloc_64.h | 28 ++++++++++++++++++++++++++-
arch/sparc/include/asm/pgtable_64.h | 37 +++++++++++++++++++++++++++++++-----
arch/sparc/include/asm/tsb.h | 10 +++++++++
arch/sparc/kernel/smp_64.c | 7 ++++++
arch/sparc/mm/init_64.c | 31 ++++++++++++++++++++++++++----
6 files changed, 109 insertions(+), 10 deletions(-)

--- a/arch/sparc/include/asm/page_64.h
+++ b/arch/sparc/include/asm/page_64.h
@@ -57,18 +57,21 @@ extern void copy_user_page(void *to, voi
typedef struct { unsigned long pte; } pte_t;
typedef struct { unsigned long iopte; } iopte_t;
typedef struct { unsigned long pmd; } pmd_t;
+typedef struct { unsigned long pud; } pud_t;
typedef struct { unsigned long pgd; } pgd_t;
typedef struct { unsigned long pgprot; } pgprot_t;

#define pte_val(x) ((x).pte)
#define iopte_val(x) ((x).iopte)
#define pmd_val(x) ((x).pmd)
+#define pud_val(x) ((x).pud)
#define pgd_val(x) ((x).pgd)
#define pgprot_val(x) ((x).pgprot)

#define __pte(x) ((pte_t) { (x) } )
#define __iopte(x) ((iopte_t) { (x) } )
#define __pmd(x) ((pmd_t) { (x) } )
+#define __pud(x) ((pud_t) { (x) } )
#define __pgd(x) ((pgd_t) { (x) } )
#define __pgprot(x) ((pgprot_t) { (x) } )

@@ -77,18 +80,21 @@ typedef struct { unsigned long pgprot; }
typedef unsigned long pte_t;
typedef unsigned long iopte_t;
typedef unsigned long pmd_t;
+typedef unsigned long pud_t;
typedef unsigned long pgd_t;
typedef unsigned long pgprot_t;

#define pte_val(x) (x)
#define iopte_val(x) (x)
#define pmd_val(x) (x)
+#define pud_val(x) (x)
#define pgd_val(x) (x)
#define pgprot_val(x) (x)

#define __pte(x) (x)
#define __iopte(x) (x)
#define __pmd(x) (x)
+#define __pud(x) (x)
#define __pgd(x) (x)
#define __pgprot(x) (x)

--- a/arch/sparc/include/asm/pgalloc_64.h
+++ b/arch/sparc/include/asm/pgalloc_64.h
@@ -15,6 +15,13 @@

extern struct kmem_cache *pgtable_cache;

+static inline void __pgd_populate(pgd_t *pgd, pud_t *pud)
+{
+ pgd_set(pgd, pud);
+}
+
+#define pgd_populate(MM, PGD, PUD) __pgd_populate(PGD, PUD)
+
static inline pgd_t *pgd_alloc(struct mm_struct *mm)
{
return kmem_cache_alloc(pgtable_cache, GFP_KERNEL);
@@ -25,7 +32,23 @@ static inline void pgd_free(struct mm_st
kmem_cache_free(pgtable_cache, pgd);
}

-#define pud_populate(MM, PUD, PMD) pud_set(PUD, PMD)
+static inline void __pud_populate(pud_t *pud, pmd_t *pmd)
+{
+ pud_set(pud, pmd);
+}
+
+#define pud_populate(MM, PUD, PMD) __pud_populate(PUD, PMD)
+
+static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long addr)
+{
+ return kmem_cache_alloc(pgtable_cache,
+ GFP_KERNEL|__GFP_REPEAT);
+}
+
+static inline void pud_free(struct mm_struct *mm, pud_t *pud)
+{
+ kmem_cache_free(pgtable_cache, pud);
+}

static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long addr)
{
@@ -91,4 +114,7 @@ static inline void __pte_free_tlb(struct
#define __pmd_free_tlb(tlb, pmd, addr) \
pgtable_free_tlb(tlb, pmd, false)

+#define __pud_free_tlb(tlb, pud, addr) \
+ pgtable_free_tlb(tlb, pud, false)
+
#endif /* _SPARC64_PGALLOC_H */
--- a/arch/sparc/include/asm/pgtable_64.h
+++ b/arch/sparc/include/asm/pgtable_64.h
@@ -20,8 +20,6 @@
#include <asm/page.h>
#include <asm/processor.h>

-#include <asm-generic/pgtable-nopud.h>
-
/* The kernel image occupies 0x4000000 to 0x6000000 (4MB --> 96MB).
* The page copy blockops can use 0x6000000 to 0x8000000.
* The 8K TSB is mapped in the 0x8000000 to 0x8400000 range.
@@ -55,13 +53,21 @@
#define PMD_MASK (~(PMD_SIZE-1))
#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-3) + PMD_BITS)
+/* PUD_SHIFT determines the size of the area a third-level page
+ * table can map
+ */
+#define PUD_SHIFT (PMD_SHIFT + PMD_BITS)
+#define PUD_SIZE (_AC(1,UL) << PUD_SHIFT)
+#define PUD_MASK (~(PUD_SIZE-1))
+#define PUD_BITS (PAGE_SHIFT - 3)
+
+/* PGDIR_SHIFT determines what a fourth-level page table entry can map */
+#define PGDIR_SHIFT (PUD_SHIFT + PUD_BITS)
#define PGDIR_SIZE (_AC(1,UL) << PGDIR_SHIFT)
#define PGDIR_MASK (~(PGDIR_SIZE-1))
#define PGDIR_BITS (PAGE_SHIFT - 3)

-#if (PGDIR_SHIFT + PGDIR_BITS) != 43
+#if (PGDIR_SHIFT + PGDIR_BITS) != 53
#error Page table parameters do not cover virtual address space properly.
#endif

@@ -93,6 +99,7 @@ static inline bool kern_addr_valid(unsig
/* Entries per page directory level. */
#define PTRS_PER_PTE (1UL << (PAGE_SHIFT-3))
#define PTRS_PER_PMD (1UL << PMD_BITS)
+#define PTRS_PER_PUD (1UL << PUD_BITS)
#define PTRS_PER_PGD (1UL << PGDIR_BITS)

/* Kernel has a separate 44bit address space. */
@@ -101,6 +108,9 @@ static inline bool kern_addr_valid(unsig
#define pmd_ERROR(e) \
pr_err("%s:%d: bad pmd %p(%016lx) seen at (%pS)\n", \
__FILE__, __LINE__, &(e), pmd_val(e), __builtin_return_address(0))
+#define pud_ERROR(e) \
+ pr_err("%s:%d: bad pud %p(%016lx) seen at (%pS)\n", \
+ __FILE__, __LINE__, &(e), pud_val(e), __builtin_return_address(0))
#define pgd_ERROR(e) \
pr_err("%s:%d: bad pgd %p(%016lx) seen at (%pS)\n", \
__FILE__, __LINE__, &(e), pgd_val(e), __builtin_return_address(0))
@@ -779,6 +789,11 @@ static inline int pmd_present(pmd_t pmd)
#define pud_bad(pud) ((pud_val(pud) & ~PAGE_MASK) || \
!__kern_addr_valid(pud_val(pud)))

+#define pgd_none(pgd) (!pgd_val(pgd))
+
+#define pgd_bad(pgd) ((pgd_val(pgd) & ~PAGE_MASK) || \
+ !__kern_addr_valid(pgd_val(pgd)))
+
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
extern void set_pmd_at(struct mm_struct *mm, unsigned long addr,
pmd_t *pmdp, pmd_t pmd);
@@ -815,10 +830,17 @@ static inline unsigned long __pmd_page(p
#define pmd_clear(pmdp) (pmd_val(*(pmdp)) = 0UL)
#define pud_present(pud) (pud_val(pud) != 0U)
#define pud_clear(pudp) (pud_val(*(pudp)) = 0UL)
+#define pgd_page_vaddr(pgd) \
+ ((unsigned long) __va(pgd_val(pgd)))
+#define pgd_present(pgd) (pgd_val(pgd) != 0U)
+#define pgd_clear(pgdp) (pgd_val(*(pgd)) = 0UL)

/* Same in both SUN4V and SUN4U. */
#define pte_none(pte) (!pte_val(pte))

+#define pgd_set(pgdp, pudp) \
+ (pgd_val(*(pgdp)) = (__pa((unsigned long) (pudp))))
+
/* to find an entry in a page-table-directory. */
#define pgd_index(address) (((address) >> PGDIR_SHIFT) & (PTRS_PER_PGD - 1))
#define pgd_offset(mm, address) ((mm)->pgd + pgd_index(address))
@@ -826,6 +848,11 @@ static inline unsigned long __pmd_page(p
/* to find an entry in a kernel page-table-directory */
#define pgd_offset_k(address) pgd_offset(&init_mm, address)

+/* Find an entry in the third-level page table.. */
+#define pud_index(address) (((address) >> PUD_SHIFT) & (PTRS_PER_PUD - 1))
+#define pud_offset(pgdp, address) \
+ ((pud_t *) pgd_page_vaddr(*(pgdp)) + pud_index(address))
+
/* Find an entry in the second-level page table.. */
#define pmd_offset(pudp, address) \
((pmd_t *) pud_page_vaddr(*(pudp)) + \
--- a/arch/sparc/include/asm/tsb.h
+++ b/arch/sparc/include/asm/tsb.h
@@ -145,6 +145,11 @@ extern struct tsb_phys_patch_entry __tsb
andn REG2, 0x7, REG2; \
ldx [REG1 + REG2], REG1; \
brz,pn REG1, FAIL_LABEL; \
+ sllx VADDR, 64 - (PUD_SHIFT + PUD_BITS), REG2; \
+ srlx REG2, 64 - PAGE_SHIFT, REG2; \
+ andn REG2, 0x7, REG2; \
+ ldxa [REG1 + 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; \
andn REG2, 0x7, REG2; \
@@ -198,6 +203,11 @@ extern struct tsb_phys_patch_entry __tsb
andn REG2, 0x7, REG2; \
ldxa [PHYS_PGD + REG2] ASI_PHYS_USE_EC, REG1; \
brz,pn REG1, FAIL_LABEL; \
+ sllx VADDR, 64 - (PUD_SHIFT + PUD_BITS), REG2; \
+ srlx REG2, 64 - PAGE_SHIFT, REG2; \
+ andn REG2, 0x7, REG2; \
+ ldxa [REG1 + 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; \
andn REG2, 0x7, REG2; \
--- a/arch/sparc/kernel/smp_64.c
+++ b/arch/sparc/kernel/smp_64.c
@@ -1479,6 +1479,13 @@ static void __init pcpu_populate_pte(uns
pud_t *pud;
pmd_t *pmd;

+ if (pgd_none(*pgd)) {
+ pud_t *new;
+
+ new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE);
+ pgd_populate(&init_mm, pgd, new);
+ }
+
pud = pud_offset(pgd, addr);
if (pud_none(*pud)) {
pmd_t *new;
--- a/arch/sparc/mm/init_64.c
+++ b/arch/sparc/mm/init_64.c
@@ -1384,6 +1384,13 @@ static unsigned long __ref kernel_map_ra
pmd_t *pmd;
pte_t *pte;

+ if (pgd_none(*pgd)) {
+ pud_t *new;
+
+ new = __alloc_bootmem(PAGE_SIZE, PAGE_SIZE, PAGE_SIZE);
+ alloc_bytes += PAGE_SIZE;
+ pgd_populate(&init_mm, pgd, new);
+ }
pud = pud_offset(pgd, vstart);
if (pud_none(*pud)) {
pmd_t *new;
@@ -1850,7 +1857,12 @@ static void __init sun4v_linear_pte_xor_
/* paging_init() sets up the page tables */

static unsigned long last_valid_pfn;
-pgd_t swapper_pg_dir[PTRS_PER_PGD];
+
+/* These must be page aligned in order to not trigger the
+ * alignment tests of pgd_bad() and pud_bad().
+ */
+pgd_t swapper_pg_dir[PTRS_PER_PGD] __attribute__ ((aligned (PAGE_SIZE)));
+static pud_t swapper_pud_dir[PTRS_PER_PUD] __attribute__ ((aligned (PAGE_SIZE)));

static void sun4u_pgprot_init(void);
static void sun4v_pgprot_init(void);
@@ -1859,6 +1871,8 @@ void __init paging_init(void)
{
unsigned long end_pfn, shift, phys_base;
unsigned long real_end, i;
+ pud_t *pud;
+ pmd_t *pmd;
int node;

setup_page_offset();
@@ -1955,9 +1969,18 @@ void __init paging_init(void)

memset(swapper_low_pmd_dir, 0, sizeof(swapper_low_pmd_dir));

- /* Now can init the kernel/bad page tables. */
- pud_set(pud_offset(&swapper_pg_dir[0], 0),
- swapper_low_pmd_dir + (shift / sizeof(pgd_t)));
+ /* The kernel page tables we publish into what the rest of the
+ * world sees must be adjusted so that they see the PAGE_OFFSET
+ * address of these in-kerenel data structures. However right
+ * here we must access them from the kernel image side, because
+ * the trap tables haven't been taken over and therefore we cannot
+ * take TLB misses in the PAGE_OFFSET linear mappings yet.
+ */
+ pud = swapper_pud_dir + (shift / sizeof(pud_t));
+ pgd_set(&swapper_pg_dir[0], pud);
+
+ pmd = swapper_low_pmd_dir + (shift / sizeof(pmd_t));
+ pud_set(&swapper_pud_dir[0], pmd);

inherit_prom_mappings();



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/