diff --git a/fs/splice.c b/fs/splice.c
index 447ebc0..a46ddd2 100644
--- a/fs/splice.c
+++ b/fs/splice.c
@@ -279,7 +279,7 @@
 	pgoff_t index, end_index;
 	loff_t isize;
 	size_t total_len;
-	int error;
+	int error, page_nr;
 	struct splice_pipe_desc spd = {
 		.pages = pages,
 		.partial = partial,
@@ -299,15 +299,64 @@
 	 * read-ahead if this is a non-zero offset (we are likely doing small
 	 * chunk splice and the page is already there) for a single page.
 	 */
-	if (!loff || spd.nr_pages > 1)
-		do_page_cache_readahead(mapping, in, index, spd.nr_pages);
+	if (!loff || nr_pages > 1)
+		page_cache_readahead(mapping, &in->f_ra, in, index, nr_pages);
 
 	/*
 	 * Now fill in the holes:
 	 */
 	error = 0;
 	total_len = 0;
-	for (spd.nr_pages = 0; spd.nr_pages < nr_pages; spd.nr_pages++, index++) {
+
+	/*
+	 * Lookup the (hopefully) full range of pages we need.
+	 */
+	spd.nr_pages = find_get_pages_contig(mapping, index, nr_pages, pages);
+
+	/*
+	 * If find_get_pages_contig() returned fewer pages than we needed,
+	 * allocate the rest.
+	 */
+	index += spd.nr_pages;
+	while (spd.nr_pages < nr_pages) {
+		/*
+		 * Page could be there, find_get_pages_contig() breaks on
+		 * the first hole.
+		 */
+		page = find_get_page(mapping, index);
+		if (!page) {
+			/*
+			 * page didn't exist, allocate one.
+			 */
+			page = page_cache_alloc_cold(mapping);
+			if (!page)
+				break;
+
+			error = add_to_page_cache_lru(page, mapping, index,
+					      mapping_gfp_mask(mapping));
+			if (unlikely(error)) {
+				page_cache_release(page);
+				break;
+			}
+			/*
+			 * add_to_page_cache() locks the page, unlock it
+			 * to avoid convoluting the logic below even more.
+			 */
+			unlock_page(page);
+		}
+
+		pages[spd.nr_pages++] = page;
+		index++;
+	}
+
+	/*
+	 * Now loop over the map and see if we need to start IO on any
+	 * pages, fill in the partial map, etc.
+	 */
+	index = *ppos >> PAGE_CACHE_SHIFT;
+	nr_pages = spd.nr_pages;
+	spd.nr_pages = 0;
+	for (page_nr = 0; page_nr < nr_pages; page_nr++) {
 		unsigned int this_len;
 
 		if (!len)
@@ -317,28 +366,7 @@
 		 * this_len is the max we'll use from this page
 		 */
 		this_len = min_t(unsigned long, len, PAGE_CACHE_SIZE - loff);
-find_page:
-		/*
-		 * lookup the page for this index
-		 */
-		page = find_get_page(mapping, index);
-		if (!page) {
-			/*
-			 * page didn't exist, allocate one
-			 */
-			page = page_cache_alloc_cold(mapping);
-			if (!page)
-				break;
-
-			error = add_to_page_cache_lru(page, mapping, index,
-						mapping_gfp_mask(mapping));
-			if (unlikely(error)) {
-				page_cache_release(page);
-				break;
-			}
-
-			goto readpage;
-		}
+		page = pages[page_nr];
 
 		/*
 		 * If the page isn't uptodate, we may need to start io on it
@@ -360,7 +388,6 @@
 			 */
 			if (!page->mapping) {
 				unlock_page(page);
-				page_cache_release(page);
 				break;
 			}
 			/*
@@ -371,16 +398,20 @@
 				goto fill_it;
 			}
 
-readpage:
 			/*
 			 * need to read in the page
 			 */
 			error = mapping->a_ops->readpage(in, page);
-
 			if (unlikely(error)) {
-				page_cache_release(page);
+				/*
+				 * We really should re-lookup the page here,
+				 * but it complicates things a lot. Instead
+				 * lets just do what we already stored, and
+				 * we'll get it the next time we are called.
+				 */
 				if (error == AOP_TRUNCATED_PAGE)
-					goto find_page;
+					error = 0;
+
 				break;
 			}
 
@@ -389,10 +420,8 @@
 			 */
 			isize = i_size_read(mapping->host);
 			end_index = (isize - 1) >> PAGE_CACHE_SHIFT;
-			if (unlikely(!isize || index > end_index)) {
-				page_cache_release(page);
+			if (unlikely(!isize || index > end_index))
 				break;
-			}
 
 			/*
 			 * if this is the last page, see if we need to shrink
@@ -400,27 +429,33 @@
 			 */
 			if (end_index == index) {
 				loff = PAGE_CACHE_SIZE - (isize & ~PAGE_CACHE_MASK);
-				if (total_len + loff > isize) {
-					page_cache_release(page);
+				if (total_len + loff > isize)
 					break;
-				}
 				/*
 				 * force quit after adding this page
 				 */
-				nr_pages = spd.nr_pages;
+				len = this_len;
 				this_len = min(this_len, loff);
 				loff = 0;
 			}
 		}
 fill_it:
-		pages[spd.nr_pages] = page;
-		partial[spd.nr_pages].offset = loff;
-		partial[spd.nr_pages].len = this_len;
+		partial[page_nr].offset = loff;
+		partial[page_nr].len = this_len;
 		len -= this_len;
 		total_len += this_len;
 		loff = 0;
+		spd.nr_pages++;
+		index++;
 	}
 
+	/*
+	 * Release any pages at the end, if we quit early. 'i' is how far
+	 * we got, 'nr_pages' is how many pages are in the map.
+	 */
+	while (page_nr < nr_pages)
+		page_cache_release(pages[page_nr++]);
+
 	if (spd.nr_pages)
 		return splice_to_pipe(pipe, &spd);
 
diff --git a/include/asm-i386/pgtable-2level.h b/include/asm-i386/pgtable-2level.h
index 27bde97..2756d4b 100644
--- a/include/asm-i386/pgtable-2level.h
+++ b/include/asm-i386/pgtable-2level.h
@@ -18,6 +18,9 @@
 #define set_pte_atomic(pteptr, pteval) set_pte(pteptr,pteval)
 #define set_pmd(pmdptr, pmdval) (*(pmdptr) = (pmdval))
 
+#define pte_clear(mm,addr,xp)	do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
+#define pmd_clear(xp)	do { set_pmd(xp, __pmd(0)); } while (0)
+
 #define ptep_get_and_clear(mm,addr,xp)	__pte(xchg(&(xp)->pte_low, 0))
 #define pte_same(a, b)		((a).pte_low == (b).pte_low)
 #define pte_page(x)		pfn_to_page(pte_pfn(x))
diff --git a/include/asm-i386/pgtable-3level.h b/include/asm-i386/pgtable-3level.h
index 36a5aa6..dccb1b3 100644
--- a/include/asm-i386/pgtable-3level.h
+++ b/include/asm-i386/pgtable-3level.h
@@ -85,6 +85,26 @@
 #define pmd_offset(pud, address) ((pmd_t *) pud_page(*(pud)) + \
 			pmd_index(address))
 
+/*
+ * For PTEs and PDEs, we must clear the P-bit first when clearing a page table
+ * entry, so clear the bottom half first and enforce ordering with a compiler
+ * barrier.
+ */
+static inline void pte_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
+{
+	ptep->pte_low = 0;
+	smp_wmb();
+	ptep->pte_high = 0;
+}
+
+static inline void pmd_clear(pmd_t *pmd)
+{
+	u32 *tmp = (u32 *)pmd;
+	*tmp = 0;
+	smp_wmb();
+	*(tmp + 1) = 0;
+}
+
 static inline pte_t ptep_get_and_clear(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 {
 	pte_t res;
diff --git a/include/asm-i386/pgtable.h b/include/asm-i386/pgtable.h
index ee056c4..672c3f7 100644
--- a/include/asm-i386/pgtable.h
+++ b/include/asm-i386/pgtable.h
@@ -204,12 +204,10 @@
 extern unsigned long pg0[];
 
 #define pte_present(x)	((x).pte_low & (_PAGE_PRESENT | _PAGE_PROTNONE))
-#define pte_clear(mm,addr,xp)	do { set_pte_at(mm, addr, xp, __pte(0)); } while (0)
 
 /* To avoid harmful races, pmd_none(x) should check only the lower when PAE */
 #define pmd_none(x)	(!(unsigned long)pmd_val(x))
 #define pmd_present(x)	(pmd_val(x) & _PAGE_PRESENT)
-#define pmd_clear(xp)	do { set_pmd(xp, __pmd(0)); } while (0)
 #define	pmd_bad(x)	((pmd_val(x) & (~PAGE_MASK & ~_PAGE_USER)) != _KERNPG_TABLE)
 
 
@@ -268,7 +266,7 @@
 	pte_t pte;
 	if (full) {
 		pte = *ptep;
-		*ptep = __pte(0);
+		pte_clear(mm, addr, ptep);
 	} else {
 		pte = ptep_get_and_clear(mm, addr, ptep);
 	}
diff --git a/include/linux/pagemap.h b/include/linux/pagemap.h
index 9539efd..7a1af57 100644
--- a/include/linux/pagemap.h
+++ b/include/linux/pagemap.h
@@ -78,6 +78,8 @@
 				unsigned long index, gfp_t gfp_mask);
 unsigned find_get_pages(struct address_space *mapping, pgoff_t start,
 			unsigned int nr_pages, struct page **pages);
+unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t start,
+			       unsigned int nr_pages, struct page **pages);
 unsigned find_get_pages_tag(struct address_space *mapping, pgoff_t *index,
 			int tag, unsigned int nr_pages, struct page **pages);
 
diff --git a/mm/filemap.c b/mm/filemap.c
index 3ef2073..fd57442 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -697,6 +697,38 @@
 	return ret;
 }
 
+/**
+ * find_get_pages_contig - gang contiguous pagecache lookup
+ * @mapping:	The address_space to search
+ * @index:	The starting page index
+ * @nr_pages:	The maximum number of pages
+ * @pages:	Where the resulting pages are placed
+ *
+ * find_get_pages_contig() works exactly like find_get_pages(), except
+ * that the returned number of pages are guaranteed to be contiguous.
+ *
+ * find_get_pages_contig() returns the number of pages which were found.
+ */
+unsigned find_get_pages_contig(struct address_space *mapping, pgoff_t index,
+			       unsigned int nr_pages, struct page **pages)
+{
+	unsigned int i;
+	unsigned int ret;
+
+	read_lock_irq(&mapping->tree_lock);
+	ret = radix_tree_gang_lookup(&mapping->page_tree,
+				(void **)pages, index, nr_pages);
+	for (i = 0; i < ret; i++) {
+		if (pages[i]->mapping == NULL || pages[i]->index != index)
+			break;
+
+		page_cache_get(pages[i]);
+		index++;
+	}
+	read_unlock_irq(&mapping->tree_lock);
+	return i;
+}
+
 /*
  * Like find_get_pages, except we only return pages which are tagged with
  * `tag'.   We update *index to index the next page for the traversal.