]> err.no Git - linux-2.6/commitdiff
x86: CPA, fix alias checks
authorThomas Gleixner <tglx@linutronix.de>
Mon, 18 Feb 2008 19:54:14 +0000 (20:54 +0100)
committerThomas Gleixner <tglx@linutronix.de>
Mon, 18 Feb 2008 19:54:14 +0000 (20:54 +0100)
c_p_a() did not discover all aliases correctly. (such as when called
on vmalloc()-ed areas or ioremap()-ed areas)

Push the alias checks to the lower, physical level and consistently
discover all aliases that might exist: the low direct mappings and
the high linear kernel-text mappings (on 64-bit).

Thanks to Andi Kleen for pointing out that this was buggy.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
arch/x86/mm/pageattr.c

index 4119379f80fff553de99fd30facb39485659f250..7b4a17236972bc90815204f766e9e8bfc389dc10 100644 (file)
@@ -16,6 +16,7 @@
 #include <asm/sections.h>
 #include <asm/uaccess.h>
 #include <asm/pgalloc.h>
+#include <asm/proto.h>
 
 /*
  * The current flushing context - we pass it instead of 5 arguments:
@@ -26,8 +27,23 @@ struct cpa_data {
        pgprot_t        mask_clr;
        int             numpages;
        int             flushtlb;
+       unsigned long   pfn;
 };
 
+#ifdef CONFIG_X86_64
+
+static inline unsigned long highmap_start_pfn(void)
+{
+       return __pa(_text) >> PAGE_SHIFT;
+}
+
+static inline unsigned long highmap_end_pfn(void)
+{
+       return __pa(round_up((unsigned long)_end, PMD_SIZE)) >> PAGE_SHIFT;
+}
+
+#endif
+
 static inline int
 within(unsigned long addr, unsigned long start, unsigned long end)
 {
@@ -123,29 +139,14 @@ static void cpa_flush_range(unsigned long start, int numpages, int cache)
        }
 }
 
-#define HIGH_MAP_START __START_KERNEL_map
-#define HIGH_MAP_END   (__START_KERNEL_map + KERNEL_TEXT_SIZE)
-
-
-/*
- * Converts a virtual address to a X86-64 highmap address
- */
-static unsigned long virt_to_highmap(void *address)
-{
-#ifdef CONFIG_X86_64
-       return __pa((unsigned long)address) + HIGH_MAP_START - phys_base;
-#else
-       return (unsigned long)address;
-#endif
-}
-
 /*
  * Certain areas of memory on x86 require very specific protection flags,
  * for example the BIOS area or kernel text. Callers don't always get this
  * right (again, ioremap() on BIOS memory is not uncommon) so this function
  * checks and fixes these known static required protection bits.
  */
-static inline pgprot_t static_protections(pgprot_t prot, unsigned long address)
+static inline pgprot_t static_protections(pgprot_t prot, unsigned long address,
+                                  unsigned long pfn)
 {
        pgprot_t forbidden = __pgprot(0);
 
@@ -153,30 +154,23 @@ static inline pgprot_t static_protections(pgprot_t prot, unsigned long address)
         * The BIOS area between 640k and 1Mb needs to be executable for
         * PCI BIOS based config access (CONFIG_PCI_GOBIOS) support.
         */
-       if (within(__pa(address), BIOS_BEGIN, BIOS_END))
+       if (within(pfn, BIOS_BEGIN >> PAGE_SHIFT, BIOS_END >> PAGE_SHIFT))
                pgprot_val(forbidden) |= _PAGE_NX;
 
        /*
         * The kernel text needs to be executable for obvious reasons
-        * Does not cover __inittext since that is gone later on
+        * Does not cover __inittext since that is gone later on. On
+        * 64bit we do not enforce !NX on the low mapping
         */
        if (within(address, (unsigned long)_text, (unsigned long)_etext))
                pgprot_val(forbidden) |= _PAGE_NX;
-       /*
-        * Do the same for the x86-64 high kernel mapping
-        */
-       if (within(address, virt_to_highmap(_text), virt_to_highmap(_etext)))
-               pgprot_val(forbidden) |= _PAGE_NX;
 
-       /* The .rodata section needs to be read-only */
-       if (within(address, (unsigned long)__start_rodata,
-                               (unsigned long)__end_rodata))
-               pgprot_val(forbidden) |= _PAGE_RW;
        /*
-        * Do the same for the x86-64 high kernel mapping
+        * The .rodata section needs to be read-only. Using the pfn
+        * catches all aliases.
         */
-       if (within(address, virt_to_highmap(__start_rodata),
-                               virt_to_highmap(__end_rodata)))
+       if (within(pfn, __pa((unsigned long)__start_rodata) >> PAGE_SHIFT,
+                  __pa((unsigned long)__end_rodata) >> PAGE_SHIFT))
                pgprot_val(forbidden) |= _PAGE_RW;
 
        prot = __pgprot(pgprot_val(prot) & ~pgprot_val(forbidden));
@@ -253,7 +247,7 @@ static int
 try_preserve_large_page(pte_t *kpte, unsigned long address,
                        struct cpa_data *cpa)
 {
-       unsigned long nextpage_addr, numpages, pmask, psize, flags, addr;
+       unsigned long nextpage_addr, numpages, pmask, psize, flags, addr, pfn;
        pte_t new_pte, old_pte, *tmp;
        pgprot_t old_prot, new_prot;
        int i, do_split = 1;
@@ -301,7 +295,15 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
 
        pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
        pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
-       new_prot = static_protections(new_prot, address);
+
+       /*
+        * old_pte points to the large page base address. So we need
+        * to add the offset of the virtual address:
+        */
+       pfn = pte_pfn(old_pte) + ((address & (psize - 1)) >> PAGE_SHIFT);
+       cpa->pfn = pfn;
+
+       new_prot = static_protections(new_prot, address, pfn);
 
        /*
         * We need to check the full range, whether
@@ -309,8 +311,9 @@ try_preserve_large_page(pte_t *kpte, unsigned long address,
         * the pages in the range we try to preserve:
         */
        addr = address + PAGE_SIZE;
-       for (i = 1; i < cpa->numpages; i++, addr += PAGE_SIZE) {
-               pgprot_t chk_prot = static_protections(new_prot, addr);
+       pfn++;
+       for (i = 1; i < cpa->numpages; i++, addr += PAGE_SIZE, pfn++) {
+               pgprot_t chk_prot = static_protections(new_prot, addr, pfn);
 
                if (pgprot_val(chk_prot) != pgprot_val(new_prot))
                        goto out_unlock;
@@ -505,46 +508,51 @@ out_unlock:
        return 0;
 }
 
-static int __change_page_attr(unsigned long address, struct cpa_data *cpa)
+static int __change_page_attr(struct cpa_data *cpa, int primary)
 {
+       unsigned long address = cpa->vaddr;
        int do_split, err;
        unsigned int level;
        struct page *kpte_page;
-       pte_t *kpte;
+       pte_t *kpte, old_pte;
 
 repeat:
        kpte = lookup_address(address, &level);
        if (!kpte)
+               return primary ? -EINVAL : 0;
+
+       old_pte = *kpte;
+       if (!pte_val(old_pte)) {
+               if (!primary)
+                       return 0;
+               printk(KERN_WARNING "CPA: called for zero pte. "
+                      "vaddr = %lx cpa->vaddr = %lx\n", address,
+                      cpa->vaddr);
+               WARN_ON(1);
                return -EINVAL;
+       }
 
        kpte_page = virt_to_page(kpte);
        BUG_ON(PageLRU(kpte_page));
        BUG_ON(PageCompound(kpte_page));
 
        if (level == PG_LEVEL_4K) {
-               pte_t new_pte, old_pte = *kpte;
+               pte_t new_pte;
                pgprot_t new_prot = pte_pgprot(old_pte);
-
-               if(!pte_val(old_pte)) {
-                       printk(KERN_WARNING "CPA: called for zero pte. "
-                              "vaddr = %lx cpa->vaddr = %lx\n", address,
-                               cpa->vaddr);
-                       WARN_ON(1);
-                       return -EINVAL;
-               }
+               unsigned long pfn = pte_pfn(old_pte);
 
                pgprot_val(new_prot) &= ~pgprot_val(cpa->mask_clr);
                pgprot_val(new_prot) |= pgprot_val(cpa->mask_set);
 
-               new_prot = static_protections(new_prot, address);
+               new_prot = static_protections(new_prot, address, pfn);
 
                /*
                 * We need to keep the pfn from the existing PTE,
                 * after all we're only going to change it's attributes
                 * not the memory it points to
                 */
-               new_pte = pfn_pte(pte_pfn(old_pte), canon_pgprot(new_prot));
-
+               new_pte = pfn_pte(pfn, canon_pgprot(new_prot));
+               cpa->pfn = pfn;
                /*
                 * Do we really change anything ?
                 */
@@ -581,67 +589,44 @@ repeat:
        return err;
 }
 
-/**
- * change_page_attr_addr - Change page table attributes in linear mapping
- * @address: Virtual address in linear mapping.
- * @prot:    New page table attribute (PAGE_*)
- *
- * Change page attributes of a page in the direct mapping. This is a variant
- * of change_page_attr() that also works on memory holes that do not have
- * mem_map entry (pfn_valid() is false).
- *
- * See change_page_attr() documentation for more details.
- *
- * Modules and drivers should use the set_memory_* APIs instead.
- */
-static int change_page_attr_addr(struct cpa_data *cpa)
+static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias);
+
+static int cpa_process_alias(struct cpa_data *cpa)
 {
-       int err;
-       unsigned long address = cpa->vaddr;
+       struct cpa_data alias_cpa;
+       int ret;
 
-#ifdef CONFIG_X86_64
-       unsigned long phys_addr = __pa(address);
+       if (cpa->pfn > max_pfn_mapped)
+               return 0;
 
-       /*
-        * If we are inside the high mapped kernel range, then we
-        * fixup the low mapping first. __va() returns the virtual
-        * address in the linear mapping:
-        */
-       if (within(address, HIGH_MAP_START, HIGH_MAP_END))
-               address = (unsigned long) __va(phys_addr);
-#endif
+       alias_cpa = *cpa;
+       alias_cpa.vaddr = (unsigned long) __va(cpa->pfn << PAGE_SHIFT);
 
-       err = __change_page_attr(address, cpa);
-       if (err)
-               return err;
+       ret = __change_page_attr_set_clr(&alias_cpa, 0);
 
 #ifdef CONFIG_X86_64
+       if (ret)
+               return ret;
        /*
         * If the physical address is inside the kernel map, we need
         * to touch the high mapped kernel as well:
         */
-       if (within(phys_addr, 0, KERNEL_TEXT_SIZE)) {
-               /*
-                * Calc the high mapping address. See __phys_addr()
-                * for the non obvious details.
-                *
-                * Note that NX and other required permissions are
-                * checked in static_protections().
-                */
-               address = phys_addr + HIGH_MAP_START - phys_base;
+       if (!within(cpa->pfn, highmap_start_pfn(), highmap_end_pfn()))
+               return 0;
 
-               /*
-                * Our high aliases are imprecise, because we check
-                * everything between 0 and KERNEL_TEXT_SIZE, so do
-                * not propagate lookup failures back to users:
-                */
-               __change_page_attr(address, cpa);
-       }
+       alias_cpa = *cpa;
+       alias_cpa.vaddr =
+               (cpa->pfn << PAGE_SHIFT) + __START_KERNEL_map - phys_base;
+
+       /*
+        * The high mapping range is imprecise, so ignore the return value.
+        */
+       __change_page_attr_set_clr(&alias_cpa, 0);
 #endif
-       return err;
+       return ret;
 }
 
-static int __change_page_attr_set_clr(struct cpa_data *cpa)
+static int __change_page_attr_set_clr(struct cpa_data *cpa, int checkalias)
 {
        int ret, numpages = cpa->numpages;
 
@@ -651,10 +636,17 @@ static int __change_page_attr_set_clr(struct cpa_data *cpa)
                 * preservation check.
                 */
                cpa->numpages = numpages;
-               ret = change_page_attr_addr(cpa);
+
+               ret = __change_page_attr(cpa, checkalias);
                if (ret)
                        return ret;
 
+               if (checkalias) {
+                       ret = cpa_process_alias(cpa);
+                       if (ret)
+                               return ret;
+               }
+
                /*
                 * Adjust the number of pages with the result of the
                 * CPA operation. Either a large page has been
@@ -703,7 +695,7 @@ static int change_page_attr_set_clr(unsigned long addr, int numpages,
        cpa.mask_clr = mask_clr;
        cpa.flushtlb = 0;
 
-       ret = __change_page_attr_set_clr(&cpa);
+       ret = __change_page_attr_set_clr(&cpa, 1);
 
        /*
         * Check whether we really changed something:
@@ -841,7 +833,7 @@ static int __set_pages_p(struct page *page, int numpages)
                                .mask_set = __pgprot(_PAGE_PRESENT | _PAGE_RW),
                                .mask_clr = __pgprot(0)};
 
-       return __change_page_attr_set_clr(&cpa);
+       return __change_page_attr_set_clr(&cpa, 1);
 }
 
 static int __set_pages_np(struct page *page, int numpages)
@@ -851,7 +843,7 @@ static int __set_pages_np(struct page *page, int numpages)
                                .mask_set = __pgprot(0),
                                .mask_clr = __pgprot(_PAGE_PRESENT | _PAGE_RW)};
 
-       return __change_page_attr_set_clr(&cpa);
+       return __change_page_attr_set_clr(&cpa, 1);
 }
 
 void kernel_map_pages(struct page *page, int numpages, int enable)