drbd: don't BUG_ON, if bio_add_page of a single page to an empty bio fails

Just deal with it more gracefully, if we fail to add even a single page
to an empty bio. We used to BUG_ON() there, but it has been observed in
some Xen deployment, so we need to handle that case more robustly now.

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

2 files changed, 34 insertions, 18 deletions

diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index a56b107e01eb..9e9fc3413137 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -1073,6 +1073,16 @@ void drbd_bump_write_ordering(struct drbd_conf *mdev, enum write_ordering_e wo)
  * @mdev:       DRBD device.
  * @e:          epoch entry
  * @rw:         flag field, see bio->bi_rw
+ *
+ * May spread the pages to multiple bios,
+ * depending on bio_add_page restrictions.
+ *
+ * Returns 0 if all bios have been submitted,
+ * -ENOMEM if we could not allocate enough bios,
+ * -ENOSPC (any better suggestion?) if we have not been able to bio_add_page a
+ *  single page to an empty bio (which should never happen and likely indicates
+ *  that the lower level IO stack is in some way broken). This has been observed
+ *  on certain Xen deployments.
  */
 /* TODO allocate from our own bio_set. */
 int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e,
@@ -1085,6 +1095,7 @@ int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e,
         unsigned ds = e->size;
         unsigned n_bios = 0;
         unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT;
+        int err = -ENOMEM;
 
         /* In most cases, we will only need one bio.  But in case the lower
          * level restrictions happen to be different at this offset on this
@@ -1110,8 +1121,17 @@ next_bio:
         page_chain_for_each(page) {
                 unsigned len = min_t(unsigned, ds, PAGE_SIZE);
                 if (!bio_add_page(bio, page, len, 0)) {
-                        /* a single page must always be possible! */
-                        BUG_ON(bio->bi_vcnt == 0);
+                        /* A single page must always be possible!
+                         * But in case it fails anyways,
+                         * we deal with it, and complain (below). */
+                        if (bio->bi_vcnt == 0) {
+                                dev_err(DEV,
+                                        "bio_add_page failed for len=%u, "
+                                        "bi_vcnt=0 (bi_sector=%llu)\n",
+                                        len, (unsigned long long)bio->bi_sector);
+                                err = -ENOSPC;
+                                goto fail;
+                        }
                         goto next_bio;
                 }
                 ds -= len;
@@ -1137,7 +1157,7 @@ fail:
                 bios = bios->bi_next;
                 bio_put(bio);
         }
-        return -ENOMEM;
+        return err;
 }
 
 static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size)
@@ -1436,9 +1456,8 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si
         if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_RS_WR) == 0)
                 return true;
 
-        /* drbd_submit_ee currently fails for one reason only:
-         * not being able to allocate enough bios.
-         * Is dropping the connection going to help? */
+        /* don't care for the reason here */
+        dev_err(DEV, "submit failed, triggering re-connect\n");
         spin_lock_irq(&mdev->req_lock);
         list_del(&e->w.list);
         spin_unlock_irq(&mdev->req_lock);
@@ -1837,9 +1856,8 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned
         if (drbd_submit_ee(mdev, e, rw, DRBD_FAULT_DT_WR) == 0)
                 return true;
 
-        /* drbd_submit_ee currently fails for one reason only:
-         * not being able to allocate enough bios.
-         * Is dropping the connection going to help? */
+        /* don't care for the reason here */
+        dev_err(DEV, "submit failed, triggering re-connect\n");
         spin_lock_irq(&mdev->req_lock);
         list_del(&e->w.list);
         hlist_del_init(&e->colision);
@@ -1848,9 +1866,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned
                 drbd_al_complete_io(mdev, e->sector);
 
 out_interrupted:
-        /* yes, the epoch_size now is imbalanced.
-         * but we drop the connection anyways, so we don't have a chance to
-         * receive a barrier... atomic_inc(&mdev->epoch_size); */
+        drbd_may_finish_epoch(mdev, e->epoch, EV_PUT + EV_CLEANUP);
         put_ldev(mdev);
         drbd_free_ee(mdev, e);
         return false;
@@ -2096,9 +2112,8 @@ submit:
         if (drbd_submit_ee(mdev, e, READ, fault_type) == 0)
                 return true;
 
-        /* drbd_submit_ee currently fails for one reason only:
-         * not being able to allocate enough bios.
-         * Is dropping the connection going to help? */
+        /* don't care for the reason here */
+        dev_err(DEV, "submit failed, triggering re-connect\n");
         spin_lock_irq(&mdev->req_lock);
         list_del(&e->w.list);
         spin_unlock_irq(&mdev->req_lock);
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index ff0eb308ee4a..cfd324b9f95b 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -369,9 +369,10 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size)
         if (drbd_submit_ee(mdev, e, READ, DRBD_FAULT_RS_RD) == 0)
                 return 0;
 
-        /* drbd_submit_ee currently fails for one reason only:
-         * not being able to allocate enough bios.
-         * Is dropping the connection going to help? */
+        /* If it failed because of ENOMEM, retry should help.  If it failed
+         * because bio_add_page failed (probably broken lower level driver),
+         * retry may or may not help.
+         * If it does not, you may need to force disconnect. */
         spin_lock_irq(&mdev->req_lock);
         list_del(&e->w.list);
         spin_unlock_irq(&mdev->req_lock);