kernel.git/drivers/block/drbd/drbd_state.c, branch linux-4.5.y

drbd: make suspend_io() / resume_io() must be thread and recursion safe

2015-11-25T16:22:03Z

Avoid to prematurely resume application IO: don't set/clear a single bit, but inc/dec an atomic counter. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: fix NULL deref in remember_new_state

2015-11-25T16:22:01Z

The recent (not yet released) backport of the extended state broadcasts to support the "events2" subcommand of drbdsetup had some glitches. remember_old_state() would first count all connections with a net_conf != NULL, then allocate a suitable array, then populate that array with all connections found to have net_conf != NULL. This races with the state change to C_STANDALONE, and the NULL assignment there. remember_new_state() then iterates over said connection array, assuming that it would be fully populated. But rcu_lock() just makes sure the thing some pointer points to, if any, won't go away. It does not make the pointer itself immutable. In fact there is no need to "filter" connections based on whether or not they have a currently valid configuration. Just record them always, if they don't have a config, that's fine, there will be no change then. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: add comment why we want to first call local-io-error, then send state

2015-11-25T16:22:01Z

Even though we really want to get the state information about our bad disk to the peer as soon as possible, it is useful to first call the local-io-error handler. People may chose to hard-reset the box from there. If that looks and behaves exactly like a "regular node crash", without bumping the data generation UUIDs on the peer in between, it makes it easier to deal with. If you intend to return from the local-io-error handler, then better return as quickly as possible to avoid triggering other timeouts. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: also bump UUIDs if a diskless primary connects

2015-11-25T16:22:01Z

If for some reason the primary lost its disk *and* the replication link before it is able to communicate the disk loss, probably blocked IO, then later is able to re-establish the connection, the peer needs to bump its UUIDs just like it does when peer only loses the disk and is able to communicate this in time. Otherwise, a later re-attach of the disk on the primary may start a resync in the "wrong" direction. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: Backport the "events2" command

2015-11-25T16:22:00Z

The events2 command originates from drbd-9 development. It features more information but requires a incompatible change in output format. Therefore the previous events command continues to exist, the new improved events2 command becomes available now. This prepares the user-base for a later switch to the complete drbd9 code base. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: Fix locking across all resources

2015-11-25T16:22:00Z

Instead of using a rwlock for synchronizing state changes across resources, take the request locks of all resources for global state changes. Use resources_mutex to serialize global state changes. This means that taking the request lock of a resource is now enough to prevent changes of that resource. (Previously, a read lock on the global state lock was needed as well.) Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: fix resync throttling initialization

2014-11-10T16:27:37Z

If for some reason DRBD resync was the only activity on a backend device, drbd_rs_c_min_rate_throttle() would mistakenly decide that it is still initialization time, and keep throttling the resync. This patch explicitly initializes ->rs_last_events to the current backend event counters, and drops the rs_last_events == 0 from the throttle condition. Reported-by: Mikhail Sugakov Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: fix race between role change and handshake

2014-11-10T16:27:35Z

Symptoms: If DRBD was "cleanly shut down" (all in sync, both Secondary before disconnect, identical data generation uuids), and then one side was promoted *during* the next connection handshake, the role change could confuse the handshake. The Primary would get stuck in WFBitmapS, the Secondary would log unexpected cstate (Connected) in receive_bitmap and get stuck in WFBitmapT. Fix: The test in is_valid_soft_transition wrong. It works because the not allowed actions (promote/attach) do not touch the cstate. The previous condition failed to demand a cstate change in one clause. In order to avoid deadlocks give up the state_mutex while waiting for the transient state to go away. Conflicts: drbd/drbd_state.c drbd/drbd_state.h drbd/drbd_wrappers.h Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: Use better variable names

2014-09-11T14:41:29Z

Rename local variable 'ds' to 'disk_state' or 'data_size'. 'dgs' to 'digest_size' Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg Signed-off-by: Jens Axboe

drbd: clear CRASHED_PRIMARY only after successful resync

2014-07-10T16:35:05Z

If we lost a disk during the first resync after primary crash, we could have prematurely cleared the CRASHED_PRIMARY flag. Testing on C_CONNECTED is not what we meant there, but testing for both peers to become D_UP_TO_DATE. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg