drbd: allow to select specific bitmap pages for writeout

We are about to allow several changes to the active set in one activity
log transaction. We have to write out the corresponding bitmap pages as
well, if changed.

Introduce drbd_bm_mark_for_writeout(), then re-use the existing bitmap
writeout path to submit all marked pages in one go.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>

2 files changed, 52 insertions, 10 deletions

diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index 4be737055718..bc89c4a30cbc 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -188,6 +188,9 @@ void drbd_bm_unlock(struct drbd_conf *mdev)
 /* to mark for lazy writeout once syncer cleared all clearable bits,
  * we if bits have been cleared since last IO. */
 #define BM_PAGE_LAZY_WRITEOUT	28
+/* pages marked with this "HINT" will be considered for writeout
+ * on activity log transactions */
+#define BM_PAGE_HINT_WRITEOUT	27
 
 /* store_page_idx uses non-atomic assignment. It is only used directly after
  * allocating the page.  All other bm_set_page_* and bm_clear_page_* need to
@@ -237,6 +240,27 @@ static void bm_set_page_need_writeout(struct page *page)
 	set_bit(BM_PAGE_NEED_WRITEOUT, &page_private(page));
 }
 
+/**
+ * drbd_bm_mark_for_writeout() - mark a page with a "hint" to be considered for writeout
+ * @mdev:	DRBD device.
+ * @page_nr:	the bitmap page to mark with the "hint" flag
+ *
+ * From within an activity log transaction, we mark a few pages with these
+ * hints, then call drbd_bm_write_hinted(), which will only write out changed
+ * pages which are flagged with this mark.
+ */
+void drbd_bm_mark_for_writeout(struct drbd_conf *mdev, int page_nr)
+{
+	struct page *page;
+	if (page_nr >= mdev->bitmap->bm_number_of_pages) {
+		dev_warn(DEV, "BAD: page_nr: %u, number_of_pages: %u\n",
+			 page_nr, (int)mdev->bitmap->bm_number_of_pages);
+		return;
+	}
+	page = mdev->bitmap->bm_pages[page_nr];
+	set_bit(BM_PAGE_HINT_WRITEOUT, &page_private(page));
+}
+
 static int bm_test_page_unchanged(struct page *page)
 {
 	volatile const unsigned long *addr = &page_private(page);
@@ -897,6 +921,7 @@ struct bm_aio_ctx {
 	struct completion done;
 	unsigned flags;
 #define BM_AIO_COPY_PAGES	1
+#define BM_AIO_WRITE_HINTED	2
 	int error;
 };
 
@@ -1007,13 +1032,13 @@ static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must
 /*
  * bm_rw: read/write the whole bitmap from/to its on disk location.
  */
-static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_idx) __must_hold(local)
+static int bm_rw(struct drbd_conf *mdev, int rw, unsigned flags, unsigned lazy_writeout_upper_idx) __must_hold(local)
 {
 	struct bm_aio_ctx ctx = {
 		.mdev = mdev,
 		.in_flight = ATOMIC_INIT(1),
 		.done = COMPLETION_INITIALIZER_ONSTACK(ctx.done),
-		.flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0,
+		.flags = flags,
 	};
 	struct drbd_bitmap *b = mdev->bitmap;
 	int num_pages, i, count = 0;
@@ -1042,6 +1067,10 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 		if (lazy_writeout_upper_idx && i == lazy_writeout_upper_idx)
 			break;
 		if (rw & WRITE) {
+			if ((flags & BM_AIO_WRITE_HINTED) &&
+			    !test_and_clear_bit(BM_PAGE_HINT_WRITEOUT,
+				    &page_private(b->bm_pages[i])))
+				continue;
 			if (bm_test_page_unchanged(b->bm_pages[i])) {
 				dynamic_dev_dbg(DEV, "skipped bm write for idx %u\n", i);
 				continue;
@@ -1099,7 +1128,7 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
  */
 int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local)
 {
-	return bm_rw(mdev, READ, 0);
+	return bm_rw(mdev, READ, 0, 0);
 }
 
 /**
@@ -1110,7 +1139,7 @@ int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local)
  */
 int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local)
 {
-	return bm_rw(mdev, WRITE, 0);
+	return bm_rw(mdev, WRITE, 0, 0);
 }
 
 /**
@@ -1120,12 +1149,20 @@ int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local)
  */
 int drbd_bm_write_lazy(struct drbd_conf *mdev, unsigned upper_idx) __must_hold(local)
 {
-	return bm_rw(mdev, WRITE, upper_idx);
+	return bm_rw(mdev, WRITE, BM_AIO_COPY_PAGES, upper_idx);
 }
 
+/**
+ * drbd_bm_write_hinted() - Write bitmap pages with "hint" marks, if they have changed.
+ * @mdev:	DRBD device.
+ */
+int drbd_bm_write_hinted(struct drbd_conf *mdev) __must_hold(local)
+{
+	return bm_rw(mdev, WRITE, BM_AIO_WRITE_HINTED | BM_AIO_COPY_PAGES, 0);
+}
 
 /**
- * drbd_bm_write_page: Writes a PAGE_SIZE aligned piece of bitmap
+ * drbd_bm_write_page() - Writes a PAGE_SIZE aligned piece of bitmap
  * @mdev:	DRBD device.
  * @idx:	bitmap page index
  *
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 93eb3a7ac711..edfdeb62c18f 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1310,11 +1310,14 @@ struct bm_extent {
 
 #define SLEEP_TIME (HZ/10)
 
-#define BM_BLOCK_SHIFT  12			 /* 4k per bit */
+/* We do bitmap IO in units of 4k blocks.
+ * We also still have a hardcoded 4k per bit relation. */
+#define BM_BLOCK_SHIFT	12			 /* 4k per bit */
 #define BM_BLOCK_SIZE	 (1<<BM_BLOCK_SHIFT)
-/* (9+3) : 512 bytes @ 8 bits; representing 16M storage
- * per sector of on disk bitmap */
-#define BM_EXT_SHIFT	 (BM_BLOCK_SHIFT + MD_SECTOR_SHIFT + 3)  /* = 24 */
+/* mostly arbitrarily set the represented size of one bitmap extent,
+ * aka resync extent, to 16 MiB (which is also 512 Byte worth of bitmap
+ * at 4k per bit resolution) */
+#define BM_EXT_SHIFT	 24	/* 16 MiB per resync extent */
 #define BM_EXT_SIZE	 (1<<BM_EXT_SHIFT)
 
 #if (BM_EXT_SHIFT != 24) || (BM_BLOCK_SHIFT != 12)
@@ -1408,7 +1411,9 @@ extern int  drbd_bm_test_bit(struct drbd_conf *mdev, unsigned long bitnr);
 extern int  drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr);
 extern int  drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local);
 extern int  drbd_bm_read(struct drbd_conf *mdev) __must_hold(local);
+extern void drbd_bm_mark_for_writeout(struct drbd_conf *mdev, int page_nr);
 extern int  drbd_bm_write(struct drbd_conf *mdev) __must_hold(local);
+extern int  drbd_bm_write_hinted(struct drbd_conf *mdev) __must_hold(local);
 extern unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev,
 		unsigned long al_enr);
 extern size_t	     drbd_bm_words(struct drbd_conf *mdev);