drbd: Delay/reject other state changes while establishing a connection

Changes to the role and disk state should be delayed or rejected
while we establish a connection.

This is necessary, since the peer will base its resync decision
on the UUIDs and the state we sent in the drbd_connect() function.

The most prominent example for this race is becoming primary after
sending state and UUIDs and before the state changes to C_WF_CONNECTION.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
This commit is contained in:
Philipp Reisner 2012-03-26 16:47:11 +02:00
parent 46385c84ac
commit 197296ffed
4 changed files with 24 additions and 2 deletions

View file

@ -850,6 +850,7 @@ enum {
NEW_CUR_UUID, /* Create new current UUID when thawing IO */
AL_SUSPENDED, /* Activity logging is currently suspended. */
AHEAD_TO_SYNC_SOURCE, /* Ahead -> SyncSource queued */
STATE_SENT, /* Do not change state/UUIDs while this is set */
};
struct drbd_bitmap; /* opaque for drbd_conf */

View file

@ -841,6 +841,13 @@ is_valid_state_transition(struct drbd_conf *mdev, union drbd_state ns,
if (ns.conn == os.conn && ns.conn == C_WF_REPORT_PARAMS)
rv = SS_IN_TRANSIENT_STATE;
/* While establishing a connection only allow cstate to change.
Delay/refuse role changes, detach attach etc... */
if (test_bit(STATE_SENT, &mdev->flags) &&
!(os.conn == C_WF_REPORT_PARAMS ||
(ns.conn == C_WF_REPORT_PARAMS && os.conn == C_WF_CONNECTION)))
rv = SS_IN_TRANSIENT_STATE;
if ((ns.conn == C_VERIFY_S || ns.conn == C_VERIFY_T) && os.conn < C_CONNECTED)
rv = SS_NEED_CONNECTION;
@ -1668,6 +1675,12 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
if (os.disk < D_UP_TO_DATE && os.conn >= C_SYNC_SOURCE && ns.conn == C_CONNECTED)
drbd_send_state(mdev, ns);
/* Wake up role changes, that were delayed because of connection establishing */
if (os.conn == C_WF_REPORT_PARAMS && ns.conn != C_WF_REPORT_PARAMS) {
clear_bit(STATE_SENT, &mdev->flags);
wake_up(&mdev->state_wait);
}
/* This triggers bitmap writeout of potentially still unwritten pages
* if the resync finished cleanly, or aborted because of peer disk
* failure, or because of connection loss.

View file

@ -289,7 +289,7 @@ static int _try_outdate_peer_async(void *data)
*/
spin_lock_irq(&mdev->req_lock);
ns = mdev->state;
if (ns.conn < C_WF_REPORT_PARAMS) {
if (ns.conn < C_WF_REPORT_PARAMS && !test_bit(STATE_SENT, &mdev->flags)) {
ns.pdsk = nps;
_drbd_set_state(mdev, ns, CS_VERBOSE, NULL);
}

View file

@ -751,6 +751,7 @@ static int drbd_connect(struct drbd_conf *mdev)
{
struct socket *s, *sock, *msock;
int try, h, ok;
enum drbd_state_rv rv;
D_ASSERT(!mdev->data.socket);
@ -897,6 +898,7 @@ static int drbd_connect(struct drbd_conf *mdev)
if (drbd_send_protocol(mdev) == -1)
return -1;
set_bit(STATE_SENT, &mdev->flags);
drbd_send_sync_param(mdev, &mdev->sync_conf);
drbd_send_sizes(mdev, 0, 0);
drbd_send_uuids(mdev);
@ -904,7 +906,13 @@ static int drbd_connect(struct drbd_conf *mdev)
clear_bit(USE_DEGR_WFC_T, &mdev->flags);
clear_bit(RESIZE_PENDING, &mdev->flags);
if (drbd_request_state(mdev, NS(conn, C_WF_REPORT_PARAMS)) < SS_SUCCESS)
spin_lock_irq(&mdev->req_lock);
rv = _drbd_set_state(_NS(mdev, conn, C_WF_REPORT_PARAMS), CS_VERBOSE, NULL);
if (mdev->state.conn != C_WF_REPORT_PARAMS)
clear_bit(STATE_SENT, &mdev->flags);
spin_unlock_irq(&mdev->req_lock);
if (rv < SS_SUCCESS)
return 0;
drbd_thread_start(&mdev->asender);