s390/mm: make the pxd_offset functions more robust

Change the way how pgd_offset, p4d_offset, pud_offset and pmd_offset
walk the page tables. pgd_offset now always calculates the index for
the top-level page table and adds it to the pgd, this is either a
segment table offset for a 2-level setup, a region-3 offset for 3-levels,
region-2 offset for 4-levels, or a region-1 offset for a 5-level setup.
The other three functions p4d_offset, pud_offset and pmd_offset will
only add the respective offset if they dereference the passed pointer.

With the new way of walking the page tables a sequence like this from
mm/gup.c now works:

     pgdp = pgd_offset(current->mm, addr);
     pgd = READ_ONCE(*pgdp);
     p4dp = p4d_offset(&pgd, addr);
     p4d = READ_ONCE(*p4dp);
     pudp = pud_offset(&p4d, addr);
     pud = READ_ONCE(*pudp);
     pmdp = pmd_offset(&pud, addr);
     pmd = READ_ONCE(*pmdp);

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

1 files changed, 43 insertions, 24 deletions

diff --git a/arch/s390/include/asm/pgtable.h b/arch/s390/include/asm/pgtable.h
index 76dc344edb8c..d48be9c27df9 100644
--- a/arch/s390/include/asm/pgtable.h
+++ b/arch/s390/include/asm/pgtable.h
@@ -1204,42 +1204,67 @@ static inline pte_t mk_pte(struct page *page, pgprot_t pgprot)
 #define pmd_index(address) (((address) >> PMD_SHIFT) & (PTRS_PER_PMD-1))
 #define pte_index(address) (((address) >> PAGE_SHIFT) & (PTRS_PER_PTE-1))
 
-#define pgd_offset(mm, address) ((mm)->pgd + pgd_index(address))
-#define pgd_offset_k(address) pgd_offset(&init_mm, address)
-#define pgd_offset_raw(pgd, addr) ((pgd) + pgd_index(addr))
-
 #define pmd_deref(pmd) (pmd_val(pmd) & _SEGMENT_ENTRY_ORIGIN)
 #define pud_deref(pud) (pud_val(pud) & _REGION_ENTRY_ORIGIN)
 #define p4d_deref(pud) (p4d_val(pud) & _REGION_ENTRY_ORIGIN)
 #define pgd_deref(pgd) (pgd_val(pgd) & _REGION_ENTRY_ORIGIN)
 
-static inline p4d_t *p4d_offset(pgd_t *pgd, unsigned long address)
+/*
+ * The pgd_offset function *always* adds the index for the top-level
+ * region/segment table. This is done to get a sequence like the
+ * following to work:
+ *	pgdp = pgd_offset(current->mm, addr);
+ *	pgd = READ_ONCE(*pgdp);
+ *	p4dp = p4d_offset(&pgd, addr);
+ *	...
+ * The subsequent p4d_offset, pud_offset and pmd_offset functions
+ * only add an index if they dereferenced the pointer.
+ */
+static inline pgd_t *pgd_offset_raw(pgd_t *pgd, unsigned long address)
 {
-	p4d_t *p4d = (p4d_t *) pgd;
+	unsigned long rste;
+	unsigned int shift;
 
-	if ((pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R1)
-		p4d = (p4d_t *) pgd_deref(*pgd);
-	return p4d + p4d_index(address);
+	/* Get the first entry of the top level table */
+	rste = pgd_val(*pgd);
+	/* Pick up the shift from the table type of the first entry */
+	shift = ((rste & _REGION_ENTRY_TYPE_MASK) >> 2) * 11 + 20;
+	return pgd + ((address >> shift) & (PTRS_PER_PGD - 1));
 }
 
-static inline pud_t *pud_offset(p4d_t *p4d, unsigned long address)
+#define pgd_offset(mm, address) pgd_offset_raw(READ_ONCE((mm)->pgd), address)
+#define pgd_offset_k(address) pgd_offset(&init_mm, address)
+
+static inline p4d_t *p4d_offset(pgd_t *pgd, unsigned long address)
 {
-	pud_t *pud = (pud_t *) p4d;
+	if ((pgd_val(*pgd) & _REGION_ENTRY_TYPE_MASK) >= _REGION_ENTRY_TYPE_R1)
+		return (p4d_t *) pgd_deref(*pgd) + p4d_index(address);
+	return (p4d_t *) pgd;
+}
 
-	if ((p4d_val(*p4d) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R2)
-		pud = (pud_t *) p4d_deref(*p4d);
-	return pud + pud_index(address);
+static inline pud_t *pud_offset(p4d_t *p4d, unsigned long address)
+{
+	if ((p4d_val(*p4d) & _REGION_ENTRY_TYPE_MASK) >= _REGION_ENTRY_TYPE_R2)
+		return (pud_t *) p4d_deref(*p4d) + pud_index(address);
+	return (pud_t *) p4d;
 }
 
 static inline pmd_t *pmd_offset(pud_t *pud, unsigned long address)
 {
-	pmd_t *pmd = (pmd_t *) pud;
+	if ((pud_val(*pud) & _REGION_ENTRY_TYPE_MASK) >= _REGION_ENTRY_TYPE_R3)
+		return (pmd_t *) pud_deref(*pud) + pmd_index(address);
+	return (pmd_t *) pud;
+}
 
-	if ((pud_val(*pud) & _REGION_ENTRY_TYPE_MASK) == _REGION_ENTRY_TYPE_R3)
-		pmd = (pmd_t *) pud_deref(*pud);
-	return pmd + pmd_index(address);
+static inline pte_t *pte_offset(pmd_t *pmd, unsigned long address)
+{
+	return (pte_t *) pmd_deref(*pmd) + pte_index(address);
 }
 
+#define pte_offset_kernel(pmd, address) pte_offset(pmd, address)
+#define pte_offset_map(pmd, address) pte_offset_kernel(pmd, address)
+#define pte_unmap(pte) do { } while (0)
+
 #define pfn_pte(pfn,pgprot) mk_pte_phys(__pa((pfn) << PAGE_SHIFT),(pgprot))
 #define pte_pfn(x) (pte_val(x) >> PAGE_SHIFT)
 #define pte_page(x) pfn_to_page(pte_pfn(x))
@@ -1249,12 +1274,6 @@ static inline pmd_t *pmd_offset(pud_t *pud, unsigned long address)
 #define p4d_page(p4d) pfn_to_page(p4d_pfn(p4d))
 #define pgd_page(pgd) pfn_to_page(pgd_pfn(pgd))
 
-/* Find an entry in the lowest level page table.. */
-#define pte_offset(pmd, addr) ((pte_t *) pmd_deref(*(pmd)) + pte_index(addr))
-#define pte_offset_kernel(pmd, address) pte_offset(pmd,address)
-#define pte_offset_map(pmd, address) pte_offset_kernel(pmd, address)
-#define pte_unmap(pte) do { } while (0)
-
 static inline pmd_t pmd_wrprotect(pmd_t pmd)
 {
 	pmd_val(pmd) &= ~_SEGMENT_ENTRY_WRITE;