[PATCH] Allow VFS readahead to fall to zero

Some workloads really, really want to have no readahead.  Databases which are
perfoming small synchronous I/Os against a file which has extremely poor
layout.  Any readahead at all is a lose here.

But the current readahead code refuses to adapt that low.

Fix it up so that we can indeed adaptively disable readahead altogether, and
do not start it again until we have seen max_readahead()'s worth of
consecutive reads.

include/linux/mm.h

@@ -558,6 +558,7 @@ int write_one_page(struct page *page, int wait);
 /* readahead.c */
 #define VM_MAX_READAHEAD	128	/* kbytes */
 #define VM_MIN_READAHEAD	16	/* kbytes (includes current page) */
+
 int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
 			unsigned long offset, unsigned long nr_to_read);
 void page_cache_readahead(struct address_space *mapping, 
@@ -569,7 +570,7 @@ void page_cache_readaround(struct address_space *mapping,
 			   struct file *filp,
 			   unsigned long offset);
 void handle_ra_miss(struct address_space *mapping, 
-		    struct file_ra_state *ra);
+		    struct file_ra_state *ra, pgoff_t offset);
 unsigned long max_sane_readahead(unsigned long nr);
 
 /* Do stack extension */

mm/filemap.c

@@ -562,7 +562,7 @@ void do_generic_mapping_read(struct address_space *mapping,
 find_page:
 		page = find_get_page(mapping, index);
 		if (unlikely(page == NULL)) {
-			handle_ra_miss(mapping, ra);
+			handle_ra_miss(mapping, ra, index);
 			goto no_cached_page;
 		}
 		if (!PageUptodate(page))
@@ -978,7 +978,7 @@ struct page * filemap_nopage(struct vm_area_struct * area, unsigned long address
 	page = find_get_page(mapping, pgoff);
 	if (!page) {
 		if (did_readahead) {
-			handle_ra_miss(mapping,ra);
+			handle_ra_miss(mapping, ra, pgoff);
 			did_readahead = 0;
 		}
 		goto no_cached_page;

mm/readahead.c

@@ -136,6 +136,12 @@ static int read_pages(struct address_space *mapping, struct file *filp,
  * ahead_size:  Together, these form the "ahead window".
  * ra_pages:	The externally controlled max readahead for this fd.
  *
+ * When readahead is in the "maximally shrunk" state (next_size == -1UL),
+ * readahead is disabled.  In this state, prev_page and size are used, inside
+ * handle_ra_miss(), to detect the resumption of sequential I/O.  Once there
+ * has been a decent run of sequential I/O (defined by get_min_readahead),
+ * readahead is reenabled.
+ *
  * The readahead code manages two windows - the "current" and the "ahead"
  * windows.  The intent is that while the application is walking the pages
  * in the current window, I/O is underway on the ahead window.  When the
@@ -168,6 +174,8 @@ static int read_pages(struct address_space *mapping, struct file *filp,
  * will continue to perform linear reads.  Either at the new file position, or
  * at the old one after another seek.
  *
+ * After enough misses, readahead is fully disabled. (next_size = -1UL).
+ *
  * There is a special-case: if the first page which the application tries to
  * read happens to be the first page of the file, it is assumed that a linear
  * read is about to happen and the window is immediately set to half of the
@@ -253,14 +261,19 @@ int do_page_cache_readahead(struct address_space *mapping, struct file *filp,
 	int ret = 0;
 
 	while (nr_to_read) {
+		int err;
+
 		unsigned long this_chunk = (2 * 1024 * 1024) / PAGE_CACHE_SIZE;
 
 		if (this_chunk > nr_to_read)
 			this_chunk = nr_to_read;
-		ret = __do_page_cache_readahead(mapping, filp,
+		err = __do_page_cache_readahead(mapping, filp,
 						offset, this_chunk);
-		if (ret < 0)
+		if (err < 0) {
+			ret = err;
 			break;
+		}
+		ret += err;
 		offset += this_chunk;
 		nr_to_read -= this_chunk;
 	}
@@ -286,6 +299,7 @@ check_ra_success(struct file_ra_state *ra, pgoff_t attempt,
 				ra->ahead_size = ra->next_size;
 		} else {
 			ra->next_size = -1UL;
+			ra->size = 0;
 		}
 	}
 }
@@ -345,16 +359,19 @@ page_cache_readahead(struct address_space *mapping, struct file_ra_state *ra,
 		ra->next_size += 2;
 	} else {
 		/*
-		 * A miss - lseek, pread, etc.  Shrink the readahead
+		 * A miss - lseek, pagefault, pread, etc.  Shrink the readahead
 		 * window by 25%.
 		 */
-		ra->next_size -= ra->next_size / 4;
+		ra->next_size -= ra->next_size / 4 + 2;
 	}
 
-	if (ra->next_size > max)
+	if ((long)ra->next_size > (long)max)
 		ra->next_size = max;
-	if (ra->next_size < min)
-		ra->next_size = min;
+	if ((long)ra->next_size <= 0L) {
+		ra->next_size = -1UL;
+		ra->size = 0;
+		goto out;		/* Readahead is off */
+	}
 
 	/*
 	 * Is this request outside the current window?
@@ -374,6 +391,7 @@ page_cache_readahead(struct address_space *mapping, struct file_ra_state *ra,
 			ra->prev_page = ra->start;
 			ra->ahead_start = 0;
 			ra->ahead_size = 0;
+
 			/*
 			 * Control now returns, probably to sleep until I/O
 			 * completes against the first ahead page.
@@ -394,7 +412,6 @@ page_cache_readahead(struct address_space *mapping, struct file_ra_state *ra,
 		ra->size = ra->next_size;
 		ra->ahead_start = 0;		/* Invalidate these */
 		ra->ahead_size = 0;
-
 		actual = do_page_cache_readahead(mapping, filp, offset,
 						 ra->size);
 		check_ra_success(ra, ra->size, actual, orig_next_size);
@@ -455,21 +472,37 @@ page_cache_readaround(struct address_space *mapping, struct file_ra_state *ra,
  * not found.  This will happen if it was evicted by the VM (readahead
  * thrashing) or if the readahead window is maximally shrunk.
  *
- * If the window has been maximally shrunk (next_size == 0) then bump it up
- * again to resume readahead.
+ * If the window has been maximally shrunk (next_size == -1UL) then look to see
+ * if we are getting misses against sequential file offsets.  If so, and this
+ * persists then resume readahead.
  *
  * Otherwise we're thrashing, so shrink the readahead window by three pages.
  * This is because it is grown by two pages on a readahead hit.  Theory being
  * that the readahead window size will stabilise around the maximum level at
  * which there is no thrashing.
  */
-void handle_ra_miss(struct address_space *mapping, struct file_ra_state *ra)
+void handle_ra_miss(struct address_space *mapping,
+		struct file_ra_state *ra, pgoff_t offset)
 {
-	const unsigned long min = get_min_readahead(ra);
-
 	if (ra->next_size == -1UL) {
-		ra->next_size = min;
+		const unsigned long max = get_max_readahead(ra);
+
+		if (offset != ra->prev_page + 1) {
+			ra->size = 0;			/* Not sequential */
+		} else {
+			ra->size++;			/* A sequential read */
+			if (ra->size >= max) {		/* Resume readahead */
+				ra->start = offset - max;
+				ra->next_size = max;
+				ra->size = max;
+				ra->ahead_start = 0;
+				ra->ahead_size = 0;
+			}
+		}
+		ra->prev_page = offset;
 	} else {
+		const unsigned long min = get_min_readahead(ra);
+
 		ra->next_size -= 3;
 		if (ra->next_size < min)
 			ra->next_size = min;