#include <asm/uaccess.h>
#include <asm/pgalloc.h>
+/*
+ * The current flushing context - we pass it instead of 5 arguments:
+ */
struct cpa_data {
unsigned long vaddr;
- int numpages;
pgprot_t mask_set;
pgprot_t mask_clr;
+ int numpages;
int flushtlb;
};
/*
* Only flush present addresses:
*/
- if (pte && pte_present(*pte))
+ if (pte && (pte_val(*pte) & _PAGE_PRESENT))
clflush_cache_range((void *) addr, PAGE_SIZE);
}
}
return prot;
}
+/*
+ * Lookup the page table entry for a virtual address. Return a pointer
+ * to the entry and the level of the mapping.
+ *
+ * Note: We return pud and pmd either when the entry is marked large
+ * or when the present bit is not set. Otherwise we would return a
+ * pointer to a nonexisting mapping.
+ */
pte_t *lookup_address(unsigned long address, int *level)
{
pgd_t *pgd = pgd_offset_k(address);
if (pgd_none(*pgd))
return NULL;
+
pud = pud_offset(pgd, address);
if (pud_none(*pud))
return NULL;
+
+ *level = PG_LEVEL_1G;
+ if (pud_large(*pud) || !pud_present(*pud))
+ return (pte_t *)pud;
+
pmd = pmd_offset(pud, address);
if (pmd_none(*pmd))
return NULL;
*level = PG_LEVEL_2M;
- if (pmd_large(*pmd))
+ if (pmd_large(*pmd) || !pmd_present(*pmd))
return (pte_t *)pmd;
*level = PG_LEVEL_4K;
+
return pte_offset_kernel(pmd, address);
}
+/*
+ * Set the new pmd in all the pgds we know about:
+ */
static void __set_pmd_pte(pte_t *kpte, unsigned long address, pte_t pte)
{
/* change init_mm */
if (!SHARED_KERNEL_PMD) {
struct page *page;
+ address = __pa(address);
list_for_each_entry(page, &pgd_list, lru) {
pgd_t *pgd;
pud_t *pud;
#endif
}
+static int
+try_preserve_large_page(pte_t *kpte, unsigned long address,
+ struct cpa_data *cpa)
+{
+ unsigned long nextpage_addr, numpages, pmask, psize, flags;
+ pte_t new_pte, old_pte, *tmp;
+ pgprot_t old_prot, new_prot;
+ int level, do_split = 1;
+
+ /*
+ * An Athlon 64 X2 showed hard hangs if we tried to preserve
+ * largepages and changed the PSE entry from RW to RO.
+ *
+ * As AMD CPUs have a long series of erratas in this area,
+ * (and none of the known ones seem to explain this hang),
+ * disable this code until the hang can be debugged:
+ */
+ if (boot_cpu_data.x86_vendor == X86_VENDOR_AMD)
+ return 1;
+
+ spin_lock_irqsave(&pgd_lock, flags);
+ /*
+ * Check for races, another CPU might have split this page
+ * up already:
+ */
+ tmp = lookup_address(address, &level);
+ if (tmp != kpte)
+ goto out_unlock;
+
+ switch (level) {
+ case PG_LEVEL_2M:
+ psize = PMD_PAGE_SIZE;
+ pmask = PMD_PAGE_MASK;
+ break;
+#ifdef CONFIG_X86_64
+ case PG_LEVEL_1G:
+ psize = PMD_PAGE_SIZE;
+ pmask = PMD_PAGE_MASK;
+ break;
+#endif
+ default:
+ do_split = -EINVAL;
+ goto out_unlock;
+ }
+
+ /*
+ * Calculate the number of pages, which fit into this large
+ * page starting at address:
+ */
+ nextpage_addr = (address + psize) & pmask;
+ numpages = (nextpage_addr - address) >> PAGE_SHIFT;
+ if (numpages < cpa->numpages)
+ cpa->numpages = numpages;
+
+ /*
+ * We are safe now. Check whether the new pgprot is the same:
+ */
+ old_pte = *kpte;
+ old_prot = new_prot = pte_pgprot(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);
+
+ /*
+ * If there are no changes, return. maxpages has been updated
+ * above:
+ */
+ if (pgprot_val(new_prot) == pgprot_val(old_prot)) {
+ do_split = 0;
+ goto out_unlock;
+ }
+
+ /*
+ * We need to change the attributes. Check, whether we can
+ * change the large page in one go. We request a split, when
+ * the address is not aligned and the number of pages is
+ * smaller than the number of pages in the large page. Note
+ * that we limited the number of possible pages already to
+ * the number of pages in the large page.
+ */
+ if (address == (nextpage_addr - psize) && cpa->numpages == numpages) {
+ /*
+ * The address is aligned and the number of pages
+ * covers the full page.
+ */
+ new_pte = pfn_pte(pte_pfn(old_pte), canon_pgprot(new_prot));
+ __set_pmd_pte(kpte, address, new_pte);
+ cpa->flushtlb = 1;
+ do_split = 0;
+ }
+
+out_unlock:
+ spin_unlock_irqrestore(&pgd_lock, flags);
+
+ return do_split;
+}
+
static int split_large_page(pte_t *kpte, unsigned long address)
{
- pgprot_t ref_prot = pte_pgprot(pte_clrhuge(*kpte));
+ unsigned long flags, pfn, pfninc = 1;
gfp_t gfp_flags = GFP_KERNEL;
- unsigned long flags, addr, pfn;
+ unsigned int i, level;
pte_t *pbase, *tmp;
+ pgprot_t ref_prot;
struct page *base;
- unsigned int i, level;
#ifdef CONFIG_DEBUG_PAGEALLOC
- gfp_flags = __GFP_HIGH | __GFP_NOFAIL | __GFP_NOWARN;
gfp_flags = GFP_ATOMIC | __GFP_NOWARN;
#endif
base = alloc_pages(gfp_flags, 0);
* up for us already:
*/
tmp = lookup_address(address, &level);
- if (tmp != kpte) {
- WARN_ON_ONCE(1);
+ if (tmp != kpte)
goto out_unlock;
- }
- address = __pa(address);
- addr = address & LARGE_PAGE_MASK;
pbase = (pte_t *)page_address(base);
#ifdef CONFIG_X86_32
paravirt_alloc_pt(&init_mm, page_to_pfn(base));
#endif
+ ref_prot = pte_pgprot(pte_clrhuge(*kpte));
+
+#ifdef CONFIG_X86_64
+ if (level == PG_LEVEL_1G) {
+ pfninc = PMD_PAGE_SIZE >> PAGE_SHIFT;
+ pgprot_val(ref_prot) |= _PAGE_PSE;
+ }
+#endif
/*
* Get the target pfn from the original entry:
*/
pfn = pte_pfn(*kpte);
- for (i = 0; i < PTRS_PER_PTE; i++, pfn++)
+ for (i = 0; i < PTRS_PER_PTE; i++, pfn += pfninc)
set_pte(&pbase[i], pfn_pte(pfn, ref_prot));
/*
- * Install the new, split up pagetable. Important detail here:
+ * Install the new, split up pagetable. Important details here:
*
* On Intel the NX bit of all levels must be cleared to make a
* page executable. See section 4.13.2 of Intel 64 and IA-32
* Architectures Software Developer's Manual).
+ *
+ * Mark the entry present. The current mapping might be
+ * set to not present, which we preserved above.
*/
ref_prot = pte_pgprot(pte_mkexec(pte_clrhuge(*kpte)));
+ pgprot_val(ref_prot) |= _PAGE_PRESENT;
__set_pmd_pte(kpte, address, mk_pte(base, ref_prot));
base = NULL;
static int __change_page_attr(unsigned long address, struct cpa_data *cpa)
{
+ int level, do_split, err;
struct page *kpte_page;
- int level, err = 0;
pte_t *kpte;
repeat:
set_pte_atomic(kpte, new_pte);
cpa->flushtlb = 1;
}
- } else {
- err = split_large_page(kpte, address);
- if (!err)
- goto repeat;
+ cpa->numpages = 1;
+ return 0;
+ }
+
+ /*
+ * Check, whether we can keep the large page intact
+ * and just change the pte:
+ */
+ do_split = try_preserve_large_page(kpte, address, cpa);
+ /*
+ * When the range fits into the existing large page,
+ * return. cp->numpages and cpa->tlbflush have been updated in
+ * try_large_page:
+ */
+ if (do_split <= 0)
+ return do_split;
+
+ /*
+ * We have to split the large page:
+ */
+ err = split_large_page(kpte, address);
+ if (!err) {
cpa->flushtlb = 1;
+ goto repeat;
}
+
return err;
}
*
* Modules and drivers should use the set_memory_* APIs instead.
*/
-
static int change_page_attr_addr(struct cpa_data *cpa)
{
int err;
static int __change_page_attr_set_clr(struct cpa_data *cpa)
{
- unsigned int i;
- int ret;
+ int ret, numpages = cpa->numpages;
- for (i = 0; i < cpa->numpages ; i++, cpa->vaddr += PAGE_SIZE) {
+ while (numpages) {
+ /*
+ * Store the remaining nr of pages for the large page
+ * preservation check.
+ */
+ cpa->numpages = numpages;
ret = change_page_attr_addr(cpa);
if (ret)
return ret;
- }
+ /*
+ * Adjust the number of pages with the result of the
+ * CPA operation. Either a large page has been
+ * preserved or a single page update happened.
+ */
+ BUG_ON(cpa->numpages > numpages);
+ numpages -= cpa->numpages;
+ cpa->vaddr += cpa->numpages * PAGE_SIZE;
+ }
return 0;
}