[S390] merge page_test_dirty and page_clear_dirty

The page_clear_dirty primitive always sets the default storage key
which resets the access control bits and the fetch protection bit.
That will surprise a KVM guest that sets non-zero access control
bits or the fetch protection bit. Merge page_test_dirty and
page_clear_dirty back to a single function and only clear the
dirty bit from the storage key.

In addition move the function page_test_and_clear_dirty and
page_test_and_clear_young to page.h where they belong. This
requires to change the parameter from a struct page * to a page
frame number.

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

diff --git a/arch/s390/include/asm/page.h b/arch/s390/include/asm/page.h
index 3c987e9..81ee277 100644 (file)
--- a/arch/s390/include/asm/page.h
+++ b/arch/s390/include/asm/page.h
@@ -107,8 +107,8 @@ typedef pte_t *pgtable_t;
 #define __pgd(x)        ((pgd_t) { (x) } )
 #define __pgprot(x)     ((pgprot_t) { (x) } )
 
 #define __pgd(x)        ((pgd_t) { (x) } )
 #define __pgprot(x)     ((pgprot_t) { (x) } )
 

-static inline void
-page_set_storage_key(unsigned long addr, unsigned int skey, int mapped)
+static inline void page_set_storage_key(unsigned long addr,
+                                       unsigned char skey, int mapped)

 {
        if (!mapped)
                asm volatile(".insn rrf,0xb22b0000,%0,%1,8,0"
 {
        if (!mapped)
                asm volatile(".insn rrf,0xb22b0000,%0,%1,8,0"


@@ -117,15 +117,59 @@ page_set_storage_key(unsigned long addr, unsigned int skey, int mapped)
                asm volatile("sske %0,%1" : : "d" (skey), "a" (addr));
 }
 
                asm volatile("sske %0,%1" : : "d" (skey), "a" (addr));
 }
 

-static inline unsigned int
-page_get_storage_key(unsigned long addr)
+static inline unsigned char page_get_storage_key(unsigned long addr)

 {
 {

-       unsigned int skey;
+       unsigned char skey;

 
 

-       asm volatile("iske %0,%1" : "=d" (skey) : "a" (addr), "0" (0));
+       asm volatile("iske %0,%1" : "=d" (skey) : "a" (addr));

        return skey;
 }
 
        return skey;
 }
 

+static inline int page_reset_referenced(unsigned long addr)
+{
+       unsigned int ipm;
+
+       asm volatile(
+               "       rrbe    0,%1\n"
+               "       ipm     %0\n"
+               : "=d" (ipm) : "a" (addr) : "cc");
+       return !!(ipm & 0x20000000);
+}
+
+/* Bits int the storage key */
+#define _PAGE_CHANGED          0x02    /* HW changed bit               */
+#define _PAGE_REFERENCED       0x04    /* HW referenced bit            */
+#define _PAGE_FP_BIT           0x08    /* HW fetch protection bit      */
+#define _PAGE_ACC_BITS         0xf0    /* HW access control bits       */
+
+/*
+ * Test and clear dirty bit in storage key.
+ * We can't clear the changed bit atomically. This is a potential
+ * race against modification of the referenced bit. This function
+ * should therefore only be called if it is not mapped in any
+ * address space.
+ */
+#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_DIRTY
+static inline int page_test_and_clear_dirty(unsigned long pfn, int mapped)
+{
+       unsigned char skey;
+
+       skey = page_get_storage_key(pfn << PAGE_SHIFT);
+       if (!(skey & _PAGE_CHANGED))
+               return 0;
+       page_set_storage_key(pfn << PAGE_SHIFT, skey & ~_PAGE_CHANGED, mapped);
+       return 1;
+}
+
+/*
+ * Test and clear referenced bit in storage key.
+ */
+#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG
+static inline int page_test_and_clear_young(unsigned long pfn)
+{
+       return page_reset_referenced(pfn << PAGE_SHIFT);
+}
+

 struct page;
 void arch_free_page(struct page *page, int order);
 void arch_alloc_page(struct page *page, int order);
 struct page;
 void arch_free_page(struct page *page, int order);
 void arch_alloc_page(struct page *page, int order);

diff --git a/arch/s390/include/asm/pgtable.h b/arch/s390/include/asm/pgtable.h
index 763620e..4ca4dd2 100644 (file)
--- a/arch/s390/include/asm/pgtable.h
+++ b/arch/s390/include/asm/pgtable.h
@@ -373,10 +373,6 @@ extern unsigned long VMALLOC_START;
 #define _ASCE_USER_BITS                (_ASCE_SPACE_SWITCH | _ASCE_PRIVATE_SPACE | \
                                 _ASCE_ALT_EVENT)
 
 #define _ASCE_USER_BITS                (_ASCE_SPACE_SWITCH | _ASCE_PRIVATE_SPACE | \
                                 _ASCE_ALT_EVENT)
 

-/* Bits int the storage key */
-#define _PAGE_CHANGED    0x02          /* HW changed bit                   */
-#define _PAGE_REFERENCED 0x04          /* HW referenced bit                */
-

 /*
  * Page protection definitions.
  */
 /*
  * Page protection definitions.
  */


@@ -555,8 +551,6 @@ static inline void rcp_unlock(pte_t *ptep)
 #endif
 }
 
 #endif
 }
 

-/* forward declaration for SetPageUptodate in page-flags.h*/
-static inline void page_clear_dirty(struct page *page, int mapped);

 #include <linux/page-flags.h>
 
 static inline void ptep_rcp_copy(pte_t *ptep)
 #include <linux/page-flags.h>
 
 static inline void ptep_rcp_copy(pte_t *ptep)


@@ -566,7 +560,7 @@ static inline void ptep_rcp_copy(pte_t *ptep)
        unsigned int skey;
        unsigned long *pgste = (unsigned long *) (ptep + PTRS_PER_PTE);
 
        unsigned int skey;
        unsigned long *pgste = (unsigned long *) (ptep + PTRS_PER_PTE);
 

-       skey = page_get_storage_key(page_to_phys(page));
+       skey = page_get_storage_key(pte_val(*ptep) >> PAGE_SHIFT);

        if (skey & _PAGE_CHANGED) {
                set_bit_simple(RCP_GC_BIT, pgste);
                set_bit_simple(KVM_UD_BIT, pgste);
        if (skey & _PAGE_CHANGED) {
                set_bit_simple(RCP_GC_BIT, pgste);
                set_bit_simple(KVM_UD_BIT, pgste);


@@ -760,6 +754,7 @@ static inline int kvm_s390_test_and_clear_page_dirty(struct mm_struct *mm,
 {
        int dirty;
        unsigned long *pgste;
 {
        int dirty;
        unsigned long *pgste;

+       unsigned long pfn;

        struct page *page;
        unsigned int skey;
 
        struct page *page;
        unsigned int skey;
 


@@ -767,8 +762,9 @@ static inline int kvm_s390_test_and_clear_page_dirty(struct mm_struct *mm,
                return -EINVAL;
        rcp_lock(ptep);
        pgste = (unsigned long *) (ptep + PTRS_PER_PTE);
                return -EINVAL;
        rcp_lock(ptep);
        pgste = (unsigned long *) (ptep + PTRS_PER_PTE);

-       page = virt_to_page(pte_val(*ptep));
-       skey = page_get_storage_key(page_to_phys(page));
+       pfn = pte_val(*ptep) >> PAGE_SHIFT;
+       page = pfn_to_page(pfn);
+       skey = page_get_storage_key(pfn);

        if (skey & _PAGE_CHANGED) {
                set_bit_simple(RCP_GC_BIT, pgste);
                set_bit_simple(KVM_UD_BIT, pgste);
        if (skey & _PAGE_CHANGED) {
                set_bit_simple(RCP_GC_BIT, pgste);
                set_bit_simple(KVM_UD_BIT, pgste);


@@ -779,7 +775,7 @@ static inline int kvm_s390_test_and_clear_page_dirty(struct mm_struct *mm,
        }
        dirty = test_and_clear_bit_simple(KVM_UD_BIT, pgste);
        if (skey & _PAGE_CHANGED)
        }
        dirty = test_and_clear_bit_simple(KVM_UD_BIT, pgste);
        if (skey & _PAGE_CHANGED)

-               page_clear_dirty(page, 1);
+               page_set_storage_key(pfn, skey & ~_PAGE_CHANGED, 1);

        rcp_unlock(ptep);
        return dirty;
 }
        rcp_unlock(ptep);
        return dirty;
 }


@@ -790,16 +786,16 @@ static inline int ptep_test_and_clear_young(struct vm_area_struct *vma,
                                            unsigned long addr, pte_t *ptep)
 {
 #ifdef CONFIG_PGSTE
                                            unsigned long addr, pte_t *ptep)
 {
 #ifdef CONFIG_PGSTE

-       unsigned long physpage;
+       unsigned long pfn;

        int young;
        unsigned long *pgste;
 
        if (!vma->vm_mm->context.has_pgste)
                return 0;
        int young;
        unsigned long *pgste;
 
        if (!vma->vm_mm->context.has_pgste)
                return 0;

-       physpage = pte_val(*ptep) & PAGE_MASK;
+       pfn = pte_val(*ptep) >> PAGE_SHIFT;

        pgste = (unsigned long *) (ptep + PTRS_PER_PTE);
 
        pgste = (unsigned long *) (ptep + PTRS_PER_PTE);
 

-       young = ((page_get_storage_key(physpage) & _PAGE_REFERENCED) != 0);
+       young = ((page_get_storage_key(pfn) & _PAGE_REFERENCED) != 0);

        rcp_lock(ptep);
        if (young)
                set_bit_simple(RCP_GR_BIT, pgste);
        rcp_lock(ptep);
        if (young)
                set_bit_simple(RCP_GR_BIT, pgste);


@@ -936,42 +932,6 @@ static inline pte_t ptep_get_and_clear_full(struct mm_struct *mm,
        __changed;                                                      \
 })
 
        __changed;                                                      \
 })
 

-/*
- * Test and clear dirty bit in storage key.
- * We can't clear the changed bit atomically. This is a potential
- * race against modification of the referenced bit. This function
- * should therefore only be called if it is not mapped in any
- * address space.
- */
-#define __HAVE_ARCH_PAGE_TEST_DIRTY
-static inline int page_test_dirty(struct page *page)
-{
-       return (page_get_storage_key(page_to_phys(page)) & _PAGE_CHANGED) != 0;
-}
-
-#define __HAVE_ARCH_PAGE_CLEAR_DIRTY
-static inline void page_clear_dirty(struct page *page, int mapped)
-{
-       page_set_storage_key(page_to_phys(page), PAGE_DEFAULT_KEY, mapped);
-}
-
-/*
- * Test and clear referenced bit in storage key.
- */
-#define __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG
-static inline int page_test_and_clear_young(struct page *page)
-{
-       unsigned long physpage = page_to_phys(page);
-       int ccode;
-
-       asm volatile(
-               "       rrbe    0,%1\n"
-               "       ipm     %0\n"
-               "       srl     %0,28\n"
-               : "=d" (ccode) : "a" (physpage) : "cc" );
-       return ccode & 2;
-}
-

 /*
  * Conversion functions: convert a page and protection to a page entry,
  * and a page entry and page directory to the page they refer to.
 /*
  * Conversion functions: convert a page and protection to a page entry,
  * and a page entry and page directory to the page they refer to.

diff --git a/include/asm-generic/pgtable.h b/include/asm-generic/pgtable.h
index b4bfe33..e9b8e59 100644 (file)
--- a/include/asm-generic/pgtable.h
+++ b/include/asm-generic/pgtable.h
@@ -184,22 +184,18 @@ static inline int pmd_same(pmd_t pmd_a, pmd_t pmd_b)
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 
 #endif /* CONFIG_TRANSPARENT_HUGEPAGE */
 #endif
 

-#ifndef __HAVE_ARCH_PAGE_TEST_DIRTY
-#define page_test_dirty(page)          (0)
+#ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_DIRTY
+#define page_test_and_clear_dirty(pfn, mapped) (0)

 #endif
 
 #endif
 

-#ifndef __HAVE_ARCH_PAGE_CLEAR_DIRTY
-#define page_clear_dirty(page, mapped) do { } while (0)
-#endif
-
-#ifndef __HAVE_ARCH_PAGE_TEST_DIRTY
+#ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_DIRTY

 #define pte_maybe_dirty(pte)           pte_dirty(pte)
 #else
 #define pte_maybe_dirty(pte)           (1)
 #endif
 
 #ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG
 #define pte_maybe_dirty(pte)           pte_dirty(pte)
 #else
 #define pte_maybe_dirty(pte)           (1)
 #endif
 
 #ifndef __HAVE_ARCH_PAGE_TEST_AND_CLEAR_YOUNG

-#define page_test_and_clear_young(page) (0)
+#define page_test_and_clear_young(pfn) (0)

 #endif
 
 #ifndef __HAVE_ARCH_PGD_OFFSET_GATE
 #endif
 
 #ifndef __HAVE_ARCH_PGD_OFFSET_GATE

diff --git a/include/linux/page-flags.h b/include/linux/page-flags.h
index 811183d..79a6700 100644 (file)
--- a/include/linux/page-flags.h
+++ b/include/linux/page-flags.h
@@ -308,7 +308,7 @@ static inline void SetPageUptodate(struct page *page)
 {
 #ifdef CONFIG_S390
        if (!test_and_set_bit(PG_uptodate, &page->flags))
 {
 #ifdef CONFIG_S390
        if (!test_and_set_bit(PG_uptodate, &page->flags))

-               page_clear_dirty(page, 0);
+               page_set_storage_key(page_to_pfn(page), PAGE_DEFAULT_KEY, 0);

 #else
        /*
         * Memory barrier must be issued before setting the PG_uptodate bit,
 #else
        /*
         * Memory barrier must be issued before setting the PG_uptodate bit,

diff --git a/mm/rmap.c b/mm/rmap.c
index 8da044a..522e4a9 100644 (file)
--- a/mm/rmap.c
+++ b/mm/rmap.c
@@ -719,7 +719,7 @@ int page_referenced(struct page *page,
                        unlock_page(page);
        }
 out:
                        unlock_page(page);
        }
 out:

-       if (page_test_and_clear_young(page))
+       if (page_test_and_clear_young(page_to_pfn(page)))

                referenced++;
 
        return referenced;
                referenced++;
 
        return referenced;


@@ -785,10 +785,8 @@ int page_mkclean(struct page *page)
                struct address_space *mapping = page_mapping(page);
                if (mapping) {
                        ret = page_mkclean_file(mapping, page);
                struct address_space *mapping = page_mapping(page);
                if (mapping) {
                        ret = page_mkclean_file(mapping, page);

-                       if (page_test_dirty(page)) {
-                               page_clear_dirty(page, 1);
+                       if (page_test_and_clear_dirty(page_to_pfn(page), 1))

                                ret = 1;
                                ret = 1;

-                       }

                }
        }
 
                }
        }
 


@@ -981,10 +979,9 @@ void page_remove_rmap(struct page *page)
         * not if it's in swapcache - there might be another pte slot
         * containing the swap entry, but page not yet written to swap.
         */
         * not if it's in swapcache - there might be another pte slot
         * containing the swap entry, but page not yet written to swap.
         */

-       if ((!PageAnon(page) || PageSwapCache(page)) && page_test_dirty(page)) {
-               page_clear_dirty(page, 1);
+       if ((!PageAnon(page) || PageSwapCache(page)) &&
+           page_test_and_clear_dirty(page_to_pfn(page), 1))

                set_page_dirty(page);
                set_page_dirty(page);

-       }

        /*
         * Hugepages are not counted in NR_ANON_PAGES nor NR_FILE_MAPPED
         * and not charged by memcg for now.
        /*
         * Hugepages are not counted in NR_ANON_PAGES nor NR_FILE_MAPPED
         * and not charged by memcg for now.