sparc64: implement get_user_pages_fast()
authorDavid S. Miller <davem@davemloft.net>
Tue, 26 Jul 2011 00:12:22 +0000 (17:12 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 26 Jul 2011 03:57:10 +0000 (20:57 -0700)
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
arch/sparc/mm/Makefile
arch/sparc/mm/gup.c [new file with mode: 0644]

index 79836a7dd00c3ba78c446fd43a73edc9d9bd60ed..e3cda21b5ee994200b7d09e7e14c33704363e67c 100644 (file)
@@ -4,7 +4,7 @@
 asflags-y := -ansi
 ccflags-y := -Werror
 
-obj-$(CONFIG_SPARC64)   += ultra.o tlb.o tsb.o
+obj-$(CONFIG_SPARC64)   += ultra.o tlb.o tsb.o gup.o
 obj-y                   += fault_$(BITS).o
 obj-y                   += init_$(BITS).o
 obj-$(CONFIG_SPARC32)   += loadmmu.o
diff --git a/arch/sparc/mm/gup.c b/arch/sparc/mm/gup.c
new file mode 100644 (file)
index 0000000..a986b5d
--- /dev/null
@@ -0,0 +1,181 @@
+/*
+ * Lockless get_user_pages_fast for sparc, cribbed from powerpc
+ *
+ * Copyright (C) 2008 Nick Piggin
+ * Copyright (C) 2008 Novell Inc.
+ */
+
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/vmstat.h>
+#include <linux/pagemap.h>
+#include <linux/rwsem.h>
+#include <asm/pgtable.h>
+
+/*
+ * The performance critical leaf functions are made noinline otherwise gcc
+ * inlines everything into a single function which results in too much
+ * register pressure.
+ */
+static noinline int gup_pte_range(pmd_t pmd, unsigned long addr,
+               unsigned long end, int write, struct page **pages, int *nr)
+{
+       unsigned long mask, result;
+       pte_t *ptep;
+
+       if (tlb_type == hypervisor) {
+               result = _PAGE_PRESENT_4V|_PAGE_P_4V;
+               if (write)
+                       result |= _PAGE_WRITE_4V;
+       } else {
+               result = _PAGE_PRESENT_4U|_PAGE_P_4U;
+               if (write)
+                       result |= _PAGE_WRITE_4U;
+       }
+       mask = result | _PAGE_SPECIAL;
+
+       ptep = pte_offset_kernel(&pmd, addr);
+       do {
+               struct page *page, *head;
+               pte_t pte = *ptep;
+
+               if ((pte_val(pte) & mask) != result)
+                       return 0;
+               VM_BUG_ON(!pfn_valid(pte_pfn(pte)));
+
+               /* The hugepage case is simplified on sparc64 because
+                * we encode the sub-page pfn offsets into the
+                * hugepage PTEs.  We could optimize this in the future
+                * use page_cache_add_speculative() for the hugepage case.
+                */
+               page = pte_page(pte);
+               head = compound_head(page);
+               if (!page_cache_get_speculative(head))
+                       return 0;
+               if (unlikely(pte_val(pte) != pte_val(*ptep))) {
+                       put_page(head);
+                       return 0;
+               }
+
+               pages[*nr] = page;
+               (*nr)++;
+       } while (ptep++, addr += PAGE_SIZE, addr != end);
+
+       return 1;
+}
+
+static int gup_pmd_range(pud_t pud, unsigned long addr, unsigned long end,
+               int write, struct page **pages, int *nr)
+{
+       unsigned long next;
+       pmd_t *pmdp;
+
+       pmdp = pmd_offset(&pud, addr);
+       do {
+               pmd_t pmd = *pmdp;
+
+               next = pmd_addr_end(addr, end);
+               if (pmd_none(pmd))
+                       return 0;
+               if (!gup_pte_range(pmd, addr, next, write, pages, nr))
+                       return 0;
+       } while (pmdp++, addr = next, addr != end);
+
+       return 1;
+}
+
+static int gup_pud_range(pgd_t pgd, unsigned long addr, unsigned long end,
+               int write, struct page **pages, int *nr)
+{
+       unsigned long next;
+       pud_t *pudp;
+
+       pudp = pud_offset(&pgd, addr);
+       do {
+               pud_t pud = *pudp;
+
+               next = pud_addr_end(addr, end);
+               if (pud_none(pud))
+                       return 0;
+               if (!gup_pmd_range(pud, addr, next, write, pages, nr))
+                       return 0;
+       } while (pudp++, addr = next, addr != end);
+
+       return 1;
+}
+
+int get_user_pages_fast(unsigned long start, int nr_pages, int write,
+                       struct page **pages)
+{
+       struct mm_struct *mm = current->mm;
+       unsigned long addr, len, end;
+       unsigned long next;
+       pgd_t *pgdp;
+       int nr = 0;
+
+       start &= PAGE_MASK;
+       addr = start;
+       len = (unsigned long) nr_pages << PAGE_SHIFT;
+       end = start + len;
+
+       /*
+        * XXX: batch / limit 'nr', to avoid large irq off latency
+        * needs some instrumenting to determine the common sizes used by
+        * important workloads (eg. DB2), and whether limiting the batch size
+        * will decrease performance.
+        *
+        * It seems like we're in the clear for the moment. Direct-IO is
+        * the main guy that batches up lots of get_user_pages, and even
+        * they are limited to 64-at-a-time which is not so many.
+        */
+       /*
+        * This doesn't prevent pagetable teardown, but does prevent
+        * the pagetables from being freed on sparc.
+        *
+        * So long as we atomically load page table pointers versus teardown,
+        * we can follow the address down to the the page and take a ref on it.
+        */
+       local_irq_disable();
+
+       pgdp = pgd_offset(mm, addr);
+       do {
+               pgd_t pgd = *pgdp;
+
+               next = pgd_addr_end(addr, end);
+               if (pgd_none(pgd))
+                       goto slow;
+               if (!gup_pud_range(pgd, addr, next, write, pages, &nr))
+                       goto slow;
+       } while (pgdp++, addr = next, addr != end);
+
+       local_irq_enable();
+
+       VM_BUG_ON(nr != (end - start) >> PAGE_SHIFT);
+       return nr;
+
+       {
+               int ret;
+
+slow:
+               local_irq_enable();
+
+               /* Try to get the remaining pages with get_user_pages */
+               start += nr << PAGE_SHIFT;
+               pages += nr;
+
+               down_read(&mm->mmap_sem);
+               ret = get_user_pages(current, mm, start,
+                       (end - start) >> PAGE_SHIFT, write, 0, pages, NULL);
+               up_read(&mm->mmap_sem);
+
+               /* Have to be a bit careful with return values */
+               if (nr > 0) {
+                       if (ret < 0)
+                               ret = nr;
+                       else
+                               ret += nr;
+               }
+
+               return ret;
+       }
+}