[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <1315405061-1473-7-git-send-email-philipp.reisner@linbit.com>
Date: Wed, 7 Sep 2011 16:17:33 +0200
From: Philipp Reisner <philipp.reisner@...bit.com>
To: linux-kernel@...r.kernel.org, Jens Axboe <axboe@...nel.dk>
Cc: drbd-dev@...ts.linbit.com
Subject: [PATCH 06/14] drbd: Converted drbd_try_outdate_peer() from mdev to tconn
Signed-off-by: Philipp Reisner <philipp.reisner@...bit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@...bit.com>
---
drivers/block/drbd/drbd_int.h | 4 +-
drivers/block/drbd/drbd_nl.c | 139 ++++++++++++++++--------------------
drivers/block/drbd/drbd_receiver.c | 7 +-
drivers/block/drbd/drbd_state.c | 62 +++++++++++++----
drivers/block/drbd/drbd_state.h | 5 ++
include/linux/drbd.h | 3 +-
6 files changed, 124 insertions(+), 96 deletions(-)
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 04716df..47a435a 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1472,8 +1472,8 @@ extern void drbd_reconsider_max_bio_size(struct drbd_conf *mdev);
extern enum drbd_state_rv drbd_set_role(struct drbd_conf *mdev,
enum drbd_role new_role,
int force);
-extern enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev);
-extern void drbd_try_outdate_peer_async(struct drbd_conf *mdev);
+extern bool conn_try_outdate_peer(struct drbd_tconn *tconn);
+extern void conn_try_outdate_peer_async(struct drbd_tconn *tconn);
extern int drbd_khelper(struct drbd_conf *mdev, char *cmd);
/* drbd_worker.c */
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index 9a0e446..931ffc4 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -401,116 +401,113 @@ int conn_khelper(struct drbd_tconn *tconn, char *cmd)
return ret;
}
-enum drbd_disk_state drbd_try_outdate_peer(struct drbd_conf *mdev)
+static enum drbd_fencing_p highest_fencing_policy(struct drbd_tconn *tconn)
{
+ enum drbd_fencing_p fp = FP_NOT_AVAIL;
+ struct drbd_conf *mdev;
+ int vnr;
+
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
+ if (get_ldev_if_state(mdev, D_CONSISTENT)) {
+ fp = max_t(enum drbd_fencing_p, fp, mdev->ldev->dc.fencing);
+ put_ldev(mdev);
+ }
+ }
+
+ return fp;
+}
+
+bool conn_try_outdate_peer(struct drbd_tconn *tconn)
+{
+ union drbd_state mask = { { .susp_fen = 1 } };
+ union drbd_state val = { };
+ enum drbd_fencing_p fp;
char *ex_to_string;
int r;
- enum drbd_disk_state nps;
- enum drbd_fencing_p fp;
- D_ASSERT(mdev->state.pdsk == D_UNKNOWN);
+ if (tconn->cstate >= C_WF_REPORT_PARAMS) {
+ conn_err(tconn, "Expected cstate < C_WF_REPORT_PARAMS\n");
+ return false;
+ }
- if (get_ldev_if_state(mdev, D_CONSISTENT)) {
- fp = mdev->ldev->dc.fencing;
- put_ldev(mdev);
- } else {
- dev_warn(DEV, "Not fencing peer, I'm not even Consistent myself.\n");
- nps = mdev->state.pdsk;
+ fp = highest_fencing_policy(tconn);
+ switch (fp) {
+ case FP_NOT_AVAIL:
+ conn_warn(tconn, "Not fencing peer, I'm not even Consistent myself.\n");
goto out;
+ case FP_DONT_CARE:
+ return true;
+ default: ;
}
- r = drbd_khelper(mdev, "fence-peer");
+ r = conn_khelper(tconn, "fence-peer");
switch ((r>>8) & 0xff) {
case 3: /* peer is inconsistent */
ex_to_string = "peer is inconsistent or worse";
- nps = D_INCONSISTENT;
+ mask.pdsk = D_MASK;
+ val.pdsk = D_INCONSISTENT;
break;
case 4: /* peer got outdated, or was already outdated */
ex_to_string = "peer was fenced";
- nps = D_OUTDATED;
+ mask.pdsk = D_MASK;
+ val.pdsk = D_OUTDATED;
break;
case 5: /* peer was down */
- if (mdev->state.disk == D_UP_TO_DATE) {
+ if (conn_highest_disk(tconn) == D_UP_TO_DATE) {
/* we will(have) create(d) a new UUID anyways... */
ex_to_string = "peer is unreachable, assumed to be dead";
- nps = D_OUTDATED;
+ mask.pdsk = D_MASK;
+ val.pdsk = D_OUTDATED;
} else {
ex_to_string = "peer unreachable, doing nothing since disk != UpToDate";
- nps = mdev->state.pdsk;
}
break;
case 6: /* Peer is primary, voluntarily outdate myself.
* This is useful when an unconnected R_SECONDARY is asked to
* become R_PRIMARY, but finds the other peer being active. */
ex_to_string = "peer is active";
- dev_warn(DEV, "Peer is primary, outdating myself.\n");
- nps = D_UNKNOWN;
- _drbd_request_state(mdev, NS(disk, D_OUTDATED), CS_WAIT_COMPLETE);
+ conn_warn(tconn, "Peer is primary, outdating myself.\n");
+ mask.disk = D_MASK;
+ val.disk = D_OUTDATED;
break;
case 7:
if (fp != FP_STONITH)
- dev_err(DEV, "fence-peer() = 7 && fencing != Stonith !!!\n");
+ conn_err(tconn, "fence-peer() = 7 && fencing != Stonith !!!\n");
ex_to_string = "peer was stonithed";
- nps = D_OUTDATED;
+ mask.pdsk = D_MASK;
+ val.pdsk = D_OUTDATED;
break;
default:
/* The script is broken ... */
- nps = D_UNKNOWN;
- dev_err(DEV, "fence-peer helper broken, returned %d\n", (r>>8)&0xff);
- return nps;
+ conn_err(tconn, "fence-peer helper broken, returned %d\n", (r>>8)&0xff);
+ return false; /* Eventually leave IO frozen */
}
- dev_info(DEV, "fence-peer helper returned %d (%s)\n",
- (r>>8) & 0xff, ex_to_string);
-
-out:
- if (mdev->state.susp_fen && nps >= D_UNKNOWN) {
- /* The handler was not successful... unfreeze here, the
- state engine can not unfreeze... */
- _drbd_request_state(mdev, NS(susp_fen, 0), CS_VERBOSE);
- }
+ conn_info(tconn, "fence-peer helper returned %d (%s)\n",
+ (r>>8) & 0xff, ex_to_string);
- return nps;
+ out:
+ conn_request_state(tconn, mask, val, CS_VERBOSE);
+ return conn_highest_pdsk(tconn) <= D_OUTDATED;
}
static int _try_outdate_peer_async(void *data)
{
- struct drbd_conf *mdev = (struct drbd_conf *)data;
- enum drbd_disk_state nps;
- union drbd_state ns;
-
- nps = drbd_try_outdate_peer(mdev);
-
- /* Not using
- drbd_request_state(mdev, NS(pdsk, nps));
- here, because we might were able to re-establish the connection
- in the meantime. This can only partially be solved in the state's
- engine is_valid_state() and is_valid_state_transition()
- functions.
+ struct drbd_tconn *tconn = (struct drbd_tconn *)data;
- nps can be D_INCONSISTENT, D_OUTDATED or D_UNKNOWN.
- pdsk == D_INCONSISTENT while conn >= C_CONNECTED is valid,
- therefore we have to have the pre state change check here.
- */
- spin_lock_irq(&mdev->tconn->req_lock);
- ns = mdev->state;
- if (ns.conn < C_WF_REPORT_PARAMS) {
- ns.pdsk = nps;
- _drbd_set_state(mdev, ns, CS_VERBOSE, NULL);
- }
- spin_unlock_irq(&mdev->tconn->req_lock);
+ conn_try_outdate_peer(tconn);
return 0;
}
-void drbd_try_outdate_peer_async(struct drbd_conf *mdev)
+void conn_try_outdate_peer_async(struct drbd_tconn *tconn)
{
struct task_struct *opa;
- opa = kthread_run(_try_outdate_peer_async, mdev, "drbd%d_a_helper", mdev_to_minor(mdev));
+ opa = kthread_run(_try_outdate_peer_async, tconn, "drbd_async_h");
if (IS_ERR(opa))
- dev_err(DEV, "out of mem, failed to invoke fence-peer helper\n");
+ conn_err(tconn, "out of mem, failed to invoke fence-peer helper\n");
}
enum drbd_state_rv
@@ -521,7 +518,6 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force)
int try = 0;
int forced = 0;
union drbd_state mask, val;
- enum drbd_disk_state nps;
if (new_role == R_PRIMARY)
request_ping(mdev->tconn); /* Detect a dead peer ASAP */
@@ -554,32 +550,23 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force)
if (rv == SS_NO_UP_TO_DATE_DISK &&
mdev->state.disk == D_CONSISTENT && mask.pdsk == 0) {
D_ASSERT(mdev->state.pdsk == D_UNKNOWN);
- nps = drbd_try_outdate_peer(mdev);
- if (nps == D_OUTDATED || nps == D_INCONSISTENT) {
+ if (conn_try_outdate_peer(mdev->tconn)) {
val.disk = D_UP_TO_DATE;
mask.disk = D_MASK;
}
-
- val.pdsk = nps;
- mask.pdsk = D_MASK;
-
continue;
}
if (rv == SS_NOTHING_TO_DO)
goto out;
if (rv == SS_PRIMARY_NOP && mask.pdsk == 0) {
- nps = drbd_try_outdate_peer(mdev);
-
- if (force && nps > D_OUTDATED) {
+ if (!conn_try_outdate_peer(mdev->tconn) && force) {
dev_warn(DEV, "Forced into split brain situation!\n");
- nps = D_OUTDATED;
- }
-
- mask.pdsk = D_MASK;
- val.pdsk = nps;
+ mask.pdsk = D_MASK;
+ val.pdsk = D_OUTDATED;
+ }
continue;
}
if (rv == SS_TWO_PRIMARIES) {
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index 449637f..08140d9 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -4028,9 +4028,11 @@ static void drbd_disconnect(struct drbd_tconn *tconn)
drbd_free_sock(tconn);
idr_for_each(&tconn->volumes, drbd_disconnected, tconn);
-
conn_info(tconn, "Connection closed\n");
+ if (conn_highest_role(tconn) == R_PRIMARY && conn_highest_pdsk(tconn) >= D_UNKNOWN)
+ conn_try_outdate_peer_async(tconn);
+
spin_lock_irq(&tconn->req_lock);
oc = tconn->cstate;
if (oc >= C_UNCONNECTED)
@@ -4107,9 +4109,6 @@ static int drbd_disconnected(int vnr, void *p, void *data)
put_ldev(mdev);
}
- if (mdev->state.role == R_PRIMARY && fp >= FP_RESOURCE && mdev->state.pdsk >= D_UNKNOWN)
- drbd_try_outdate_peer_async(mdev);
-
/* serialize with bitmap writeout triggered by the state change,
* if any. */
wait_event(mdev->misc_wait, !test_bit(BITMAP_IO, &mdev->flags));
diff --git a/drivers/block/drbd/drbd_state.c b/drivers/block/drbd/drbd_state.c
index a14511b..280b31c 100644
--- a/drivers/block/drbd/drbd_state.c
+++ b/drivers/block/drbd/drbd_state.c
@@ -62,6 +62,54 @@ bool conn_all_vols_unconf(struct drbd_tconn *tconn)
return true;
}
+enum drbd_role conn_highest_role(struct drbd_tconn *tconn)
+{
+ enum drbd_role role = R_UNKNOWN;
+ struct drbd_conf *mdev;
+ int vnr;
+
+ idr_for_each_entry(&tconn->volumes, mdev, vnr)
+ role = max_t(enum drbd_role, role, mdev->state.role);
+
+ return role;
+}
+
+enum drbd_role conn_highest_peer(struct drbd_tconn *tconn)
+{
+ enum drbd_role peer = R_UNKNOWN;
+ struct drbd_conf *mdev;
+ int vnr;
+
+ idr_for_each_entry(&tconn->volumes, mdev, vnr)
+ peer = max_t(enum drbd_role, peer, mdev->state.role);
+
+ return peer;
+}
+
+enum drbd_disk_state conn_highest_disk(struct drbd_tconn *tconn)
+{
+ enum drbd_disk_state ds = D_DISKLESS;
+ struct drbd_conf *mdev;
+ int vnr;
+
+ idr_for_each_entry(&tconn->volumes, mdev, vnr)
+ ds = max_t(enum drbd_disk_state, ds, mdev->state.disk);
+
+ return ds;
+}
+
+enum drbd_disk_state conn_highest_pdsk(struct drbd_tconn *tconn)
+{
+ enum drbd_disk_state ds = D_DISKLESS;
+ struct drbd_conf *mdev;
+ int vnr;
+
+ idr_for_each_entry(&tconn->volumes, mdev, vnr)
+ ds = max_t(enum drbd_disk_state, ds, mdev->state.pdsk);
+
+ return ds;
+}
+
/**
* cl_wide_st_chg() - true if the state change is a cluster wide one
* @mdev: DRBD device.
@@ -330,18 +378,6 @@ static void print_state_change(struct drbd_conf *mdev, union drbd_state os, unio
dev_info(DEV, "%s\n", pb);
}
-static bool vol_has_primary_peer(struct drbd_tconn *tconn)
-{
- struct drbd_conf *mdev;
- int vnr;
-
- idr_for_each_entry(&tconn->volumes, mdev, vnr) {
- if (mdev->state.peer == R_PRIMARY)
- return true;
- }
- return false;
-}
-
/**
* is_valid_state() - Returns an SS_ error code if ns is not valid
* @mdev: DRBD device.
@@ -365,7 +401,7 @@ is_valid_state(struct drbd_conf *mdev, union drbd_state ns)
if (!mdev->tconn->net_conf->two_primaries && ns.role == R_PRIMARY) {
if (ns.peer == R_PRIMARY)
rv = SS_TWO_PRIMARIES;
- else if (vol_has_primary_peer(mdev->tconn))
+ else if (conn_highest_peer(mdev->tconn) == R_PRIMARY)
rv = SS_O_VOL_PEER_PRI;
}
put_net_conf(mdev->tconn);
diff --git a/drivers/block/drbd/drbd_state.h b/drivers/block/drbd/drbd_state.h
index 55df072..394a199 100644
--- a/drivers/block/drbd/drbd_state.h
+++ b/drivers/block/drbd/drbd_state.h
@@ -110,4 +110,9 @@ static inline int drbd_request_state(struct drbd_conf *mdev,
return _drbd_request_state(mdev, mask, val, CS_VERBOSE + CS_ORDERED);
}
+enum drbd_role conn_highest_role(struct drbd_tconn *tconn);
+enum drbd_role conn_highest_peer(struct drbd_tconn *tconn);
+enum drbd_disk_state conn_highest_disk(struct drbd_tconn *tconn);
+enum drbd_disk_state conn_highest_pdsk(struct drbd_tconn *tconn);
+
#endif
diff --git a/include/linux/drbd.h b/include/linux/drbd.h
index 9cdb888..60d3088 100644
--- a/include/linux/drbd.h
+++ b/include/linux/drbd.h
@@ -65,7 +65,8 @@ enum drbd_io_error_p {
};
enum drbd_fencing_p {
- FP_DONT_CARE,
+ FP_NOT_AVAIL = -1, /* Not a policy */
+ FP_DONT_CARE = 0,
FP_RESOURCE,
FP_STONITH
};
--
1.7.4.1
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists