drbd: remove struct drbd_tl_epoch objects (barrier works)
cherry-picked and adapted from drbd 9 devel branch
DRBD requests (struct drbd_request) are already on the per resource
transfer log list, and carry their epoch number. We do not need to
additionally link them on other ring lists in other structs.
The drbd sender thread can recognize itself when to send a P_BARRIER,
by tracking the currently processed epoch, and how many writes
have been processed for that epoch.
If the epoch of the request to be processed does not match the currently
processed epoch, any writes have been processed in it, a P_BARRIER for
this last processed epoch is send out first.
The new epoch then becomes the currently processed epoch.
To not get stuck in drbd_al_begin_io() waiting for P_BARRIER_ACK,
the sender thread also needs to handle the case when the current
epoch was closed already, but no new requests are queued yet,
and send out P_BARRIER as soon as possible.
This is done by comparing the per resource "current transfer log epoch"
(tconn->current_tle_nr) with the per connection "currently processed
epoch number" (tconn->send.current_epoch_nr), while waiting for
new requests to be processed in wait_for_work().
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index c5d4fac..bbc5c2f 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -622,6 +622,8 @@
/* Wait until nothing is on the fly :) */
wait_event(mdev->misc_wait, atomic_read(&mdev->ap_pending_cnt) == 0);
+ /* FIXME also wait for all pending P_BARRIER_ACK? */
+
if (new_role == R_SECONDARY) {
set_disk_ro(mdev->vdisk, true);
if (get_ldev(mdev)) {
@@ -1436,6 +1438,12 @@
drbd_suspend_io(mdev);
/* also wait for the last barrier ack. */
+ /* FIXME see also https://daiquiri.linbit/cgi-bin/bugzilla/show_bug.cgi?id=171
+ * We need a way to either ignore barrier acks for barriers sent before a device
+ * was attached, or a way to wait for all pending barrier acks to come in.
+ * As barriers are counted per resource,
+ * we'd need to suspend io on all devices of a resource.
+ */
wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_pending_cnt) || drbd_suspended(mdev));
/* and for any other previously queued work */
drbd_flush_workqueue(mdev);