drbd: don't count sendpage()d pages only referenced by tcp as in use

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

4 files changed, 27 insertions, 12 deletions

diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 79b877db9a3..eb1273d04ca 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1077,7 +1077,8 @@ struct drbd_conf {
 	int next_barrier_nr;
 	struct hlist_head *app_reads_hash; /* is proteced by req_lock */
 	struct list_head resync_reads;
-	atomic_t pp_in_use;
+	atomic_t pp_in_use;		/* allocated from page pool */
+	atomic_t pp_in_use_by_net;	/* sendpage()d, still referenced by tcp */
 	wait_queue_head_t ee_wait;
 	struct page *md_io_page;	/* one page buffer for md_io */
 	struct page *md_io_tmpp;	/* for logical_block_size != 512 */
@@ -1555,7 +1556,10 @@ extern struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev,
 					    sector_t sector,
 					    unsigned int data_size,
 					    gfp_t gfp_mask) __must_hold(local);
-extern void drbd_free_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e);
+extern void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e,
+		int is_net);
+#define drbd_free_ee(m,e)	drbd_free_some_ee(m, e, 0)
+#define drbd_free_net_ee(m,e)	drbd_free_some_ee(m, e, 1)
 extern void drbd_wait_ee_list_empty(struct drbd_conf *mdev,
 		struct list_head *head);
 extern void _drbd_wait_ee_list_empty(struct drbd_conf *mdev,
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 1827cf073c2..981cfd178b0 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -2753,6 +2753,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev)
 	atomic_set(&mdev->net_cnt, 0);
 	atomic_set(&mdev->packet_seq, 0);
 	atomic_set(&mdev->pp_in_use, 0);
+	atomic_set(&mdev->pp_in_use_by_net, 0);
 	atomic_set(&mdev->rs_sect_in, 0);
 	atomic_set(&mdev->rs_sect_ev, 0);
 
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index e96fbb04ea4..2c3edf0ac5c 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -241,7 +241,7 @@ static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev)
 	spin_unlock_irq(&mdev->req_lock);
 
 	list_for_each_entry_safe(e, t, &reclaimed, w.list)
-		drbd_free_ee(mdev, e);
+		drbd_free_net_ee(mdev, e);
 }
 
 /**
@@ -298,9 +298,11 @@ static struct page *drbd_pp_alloc(struct drbd_conf *mdev, unsigned number, bool
  * Is also used from inside an other spin_lock_irq(&mdev->req_lock);
  * Either links the page chain back to the global pool,
  * or returns all pages to the system. */
-static void drbd_pp_free(struct drbd_conf *mdev, struct page *page)
+static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net)
 {
+	atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use;
 	int i;
+
 	if (drbd_pp_vacant > (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE)*minor_count)
 		i = page_chain_free(page);
 	else {
@@ -311,10 +313,10 @@ static void drbd_pp_free(struct drbd_conf *mdev, struct page *page)
 		drbd_pp_vacant += i;
 		spin_unlock(&drbd_pp_lock);
 	}
-	atomic_sub(i, &mdev->pp_in_use);
-	i = atomic_read(&mdev->pp_in_use);
+	i = atomic_sub_return(i, a);
 	if (i < 0)
-		dev_warn(DEV, "ASSERTION FAILED: pp_in_use: %d < 0\n", i);
+		dev_warn(DEV, "ASSERTION FAILED: %s: %d < 0\n",
+			is_net ? "pp_in_use_by_net" : "pp_in_use", i);
 	wake_up(&drbd_pp_wait);
 }
 
@@ -374,11 +376,11 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev,
 	return NULL;
 }
 
-void drbd_free_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e)
+void drbd_free_some_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, int is_net)
 {
 	if (e->flags & EE_HAS_DIGEST)
 		kfree(e->digest);
-	drbd_pp_free(mdev, e->pages);
+	drbd_pp_free(mdev, e->pages, is_net);
 	D_ASSERT(atomic_read(&e->pending_bios) == 0);
 	D_ASSERT(hlist_unhashed(&e->colision));
 	mempool_free(e, drbd_ee_mempool);
@@ -389,13 +391,14 @@ int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list)
 	LIST_HEAD(work_list);
 	struct drbd_epoch_entry *e, *t;
 	int count = 0;
+	int is_net = list == &mdev->net_ee;
 
 	spin_lock_irq(&mdev->req_lock);
 	list_splice_init(list, &work_list);
 	spin_unlock_irq(&mdev->req_lock);
 
 	list_for_each_entry_safe(e, t, &work_list, w.list) {
-		drbd_free_ee(mdev, e);
+		drbd_free_some_ee(mdev, e, is_net);
 		count++;
 	}
 	return count;
@@ -424,7 +427,7 @@ static int drbd_process_done_ee(struct drbd_conf *mdev)
 	spin_unlock_irq(&mdev->req_lock);
 
 	list_for_each_entry_safe(e, t, &reclaimed, w.list)
-		drbd_free_ee(mdev, e);
+		drbd_free_net_ee(mdev, e);
 
 	/* possible callbacks here:
 	 * e_end_block, and e_end_resync_block, e_send_discard_ack.
@@ -1460,7 +1463,7 @@ static int drbd_drain_block(struct drbd_conf *mdev, int data_size)
 		data_size -= rr;
 	}
 	kunmap(page);
-	drbd_pp_free(mdev, page);
+	drbd_pp_free(mdev, page, 0);
 	return rv;
 }
 
@@ -3879,6 +3882,9 @@ static void drbd_disconnect(struct drbd_conf *mdev)
 	i = drbd_release_ee(mdev, &mdev->net_ee);
 	if (i)
 		dev_info(DEV, "net_ee not empty, killed %u entries\n", i);
+	i = atomic_read(&mdev->pp_in_use_by_net);
+	if (i)
+		dev_info(DEV, "pp_in_use_by_net = %d, expected 0\n", i);
 	i = atomic_read(&mdev->pp_in_use);
 	if (i)
 		dev_info(DEV, "pp_in_use = %d, expected 0\n", i);
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 0e5bf8c9829..01743193f32 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -914,9 +914,13 @@ static void move_to_net_ee_or_free(struct drbd_conf *mdev, struct drbd_epoch_ent
 {
 	if (drbd_ee_has_active_page(e)) {
 		/* This might happen if sendpage() has not finished */
+		int i = DIV_ROUND_UP(e->size, PAGE_SIZE);
+		atomic_add(i, &mdev->pp_in_use_by_net);
+		atomic_sub(i, &mdev->pp_in_use);
 		spin_lock_irq(&mdev->req_lock);
 		list_add_tail(&e->w.list, &mdev->net_ee);
 		spin_unlock_irq(&mdev->req_lock);
+		wake_up(&drbd_pp_wait);
 	} else
 		drbd_free_ee(mdev, e);
 }