[PATCH 03/12] drbd: Removing drbd_cfg_rwsem

From: Philipp Reisner
Date: Fri Oct 07 2011 - 07:29:42 EST

* Updates to all configuration items is done under genl_lock().
Including removal of mdevs or tconns.
* All read non sleeping read sides are protected by rcu
* All sleeping read sides keep reference counts to keep the
objects alive

Signed-off-by: Philipp Reisner <philipp.reisner@xxxxxxxxxx>
Signed-off-by: Lars Ellenberg <lars.ellenberg@xxxxxxxxxx>
---
drivers/block/drbd/drbd_int.h | 9 ++---
drivers/block/drbd/drbd_main.c | 20 ++++--------
drivers/block/drbd/drbd_nl.c | 43 +++++++++----------------
drivers/block/drbd/drbd_proc.c | 6 +--
drivers/block/drbd/drbd_receiver.c | 61 +++++++++++++++++++++++-------------
drivers/block/drbd/drbd_worker.c | 8 +++-
6 files changed, 72 insertions(+), 75 deletions(-)

diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 8bc604e..56b190c 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -169,11 +169,8 @@ drbd_insert_fault(struct drbd_conf *mdev, unsigned int type) {
#define DRBD_MD_MAGIC (DRBD_MAGIC+4)

extern struct ratelimit_state drbd_ratelimit_state;
-extern struct idr minors;
-extern struct list_head drbd_tconns;
-extern struct rw_semaphore drbd_cfg_rwsem;
-/* drbd_cfg_rwsem protects: drbd_tconns list, minors idr, tconn->volumes idr
- note: non sleeping iterations over the idrs are protoected by RCU */
+extern struct idr minors; /* RCU, updates: genl_lock() */
+extern struct list_head drbd_tconns; /* RCU, updates: genl_lock() */

/* on the wire */
enum drbd_packet {
@@ -1477,7 +1474,7 @@ extern struct page *drbd_alloc_pages(struct drbd_conf *, unsigned int, bool);
extern void drbd_set_recv_tcq(struct drbd_conf *mdev, int tcq_enabled);
extern void _drbd_clear_done_ee(struct drbd_conf *mdev, struct list_head *to_be_freed);
extern void conn_flush_workqueue(struct drbd_tconn *tconn);
-extern int drbd_connected(int vnr, void *p, void *data);
+extern int drbd_connected(struct drbd_conf *mdev);
static inline void drbd_flush_workqueue(struct drbd_conf *mdev)
{
conn_flush_workqueue(mdev->tconn);
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index c229448..dd5b56c 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -120,7 +120,6 @@ module_param_string(usermode_helper, usermode_helper, sizeof(usermode_helper), 0
*/
struct idr minors;
struct list_head drbd_tconns; /* list of struct drbd_tconn */
-DECLARE_RWSEM(drbd_cfg_rwsem);

struct kmem_cache *drbd_request_cache;
struct kmem_cache *drbd_ee_cache; /* peer requests */
@@ -2331,21 +2330,20 @@ static void drbd_cleanup(void)

drbd_genl_unregister();

- down_write(&drbd_cfg_rwsem);
idr_for_each_entry(&minors, mdev, i) {
idr_remove(&minors, mdev_to_minor(mdev));
idr_remove(&mdev->tconn->volumes, mdev->vnr);
del_gendisk(mdev->vdisk);
- synchronize_rcu();
+ /* synchronize_rcu(); No other threads running at this point */
kref_put(&mdev->kref, &drbd_minor_destroy);
}

+ /* not _rcu since, no other updater anymore. Genl already unregistered */
list_for_each_entry_safe(tconn, tmp, &drbd_tconns, all_tconn) {
- list_del_rcu(&tconn->all_tconn);
- synchronize_rcu();
+ list_del(&tconn->all_tconn); /* not _rcu no proc, not other threads */
+ /* synchronize_rcu(); */
kref_put(&tconn->kref, &conn_destroy);
}
- up_write(&drbd_cfg_rwsem);

drbd_destroy_mempools();
unregister_blkdev(DRBD_MAJOR, "drbd");
@@ -2408,7 +2406,7 @@ struct drbd_tconn *conn_get_by_name(const char *name)
if (!name || !name[0])
return NULL;

- down_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
list_for_each_entry_rcu(tconn, &drbd_tconns, all_tconn) {
if (!strcmp(tconn->name, name)) {
kref_get(&tconn->kref);
@@ -2417,7 +2415,7 @@ struct drbd_tconn *conn_get_by_name(const char *name)
}
tconn = NULL;
found:
- up_read(&drbd_cfg_rwsem);
+ rcu_read_unlock();
return tconn;
}

@@ -2502,10 +2500,8 @@ struct drbd_tconn *conn_create(const char *name)

drbd_set_res_opts_defaults(&tconn->res_opts);

- down_write(&drbd_cfg_rwsem);
kref_init(&tconn->kref);
list_add_tail_rcu(&tconn->all_tconn, &drbd_tconns);
- up_write(&drbd_cfg_rwsem);

return tconn;

@@ -2637,7 +2633,7 @@ enum drbd_ret_code conn_new_minor(struct drbd_tconn *tconn, unsigned int minor,
/* inherit the connection state */
mdev->state.conn = tconn->cstate;
if (mdev->state.conn == C_WF_REPORT_PARAMS)
- drbd_connected(vnr, mdev, tconn);
+ drbd_connected(mdev);

return NO_ERROR;

@@ -2913,12 +2909,10 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
}
spin_unlock_irq(&mdev->tconn->req_lock);

- mutex_lock(&mdev->tconn->conf_update);
/* This blocks wants to be get removed... */
bdev->disk_conf->al_extents = be32_to_cpu(buffer->al_nr_extents);
if (bdev->disk_conf->al_extents < DRBD_AL_EXTENTS_MIN)
bdev->disk_conf->al_extents = DRBD_AL_EXTENTS_DEF;
- mutex_unlock(&mdev->tconn->conf_update);

err:
mutex_unlock(&mdev->md_io_mutex);
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index 49f3b8d..70f7dd8 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -370,10 +370,15 @@ static void conn_md_sync(struct drbd_tconn *tconn)
struct drbd_conf *mdev;
int vnr;

- down_read(&drbd_cfg_rwsem);
- idr_for_each_entry(&tconn->volumes, mdev, vnr)
+ rcu_read_lock();
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
+ kref_get(&mdev->kref);
+ rcu_read_unlock();
drbd_md_sync(mdev);
- up_read(&drbd_cfg_rwsem);
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ rcu_read_lock();
+ }
+ rcu_read_unlock();
}

int conn_khelper(struct drbd_tconn *tconn, char *cmd)
@@ -1229,12 +1234,12 @@ int drbd_adm_disk_opts(struct sk_buff *skb, struct genl_info *info)
rcu_assign_pointer(mdev->rs_plan_s, new_plan);
}

+ mutex_unlock(&mdev->tconn->conf_update);
drbd_md_sync(mdev);

if (mdev->state.conn >= C_CONNECTED)
drbd_send_sync_param(mdev);

- mutex_unlock(&mdev->tconn->conf_update);
synchronize_rcu();
kfree(old_disk_conf);
kfree(old_plan);
@@ -2058,7 +2063,7 @@ int drbd_adm_connect(struct sk_buff *skb, struct genl_info *info)
new_my_addr = (struct sockaddr *)&new_conf->my_addr;
new_peer_addr = (struct sockaddr *)&new_conf->peer_addr;

- /* No need to take drbd_cfg_rwsem here. All reconfiguration is
+ /* No need for _rcu here. All reconfiguration is
* strictly serialized on genl_lock(). We are protected against
* concurrent reconfiguration/addition/deletion */
list_for_each_entry(oconn, &drbd_tconns, all_tconn) {
@@ -2717,7 +2722,7 @@ int get_one_status(struct sk_buff *skb, struct netlink_callback *cb)
*/

/* synchronize with conn_create()/conn_destroy() */
- down_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
/* revalidate iterator position */
list_for_each_entry_rcu(tmp, &drbd_tconns, all_tconn) {
if (pos == NULL) {
@@ -2785,7 +2790,7 @@ next_tconn:
}

out:
- up_read(&drbd_cfg_rwsem);
+ rcu_read_unlock();
/* where to start the next iteration */
cb->args[0] = (long)pos;
cb->args[1] = (pos == tconn) ? volume + 1 : 0;
@@ -3064,9 +3069,7 @@ int drbd_adm_add_minor(struct sk_buff *skb, struct genl_info *info)
goto out;
}

- down_write(&drbd_cfg_rwsem);
retcode = conn_new_minor(adm_ctx.tconn, dh->minor, adm_ctx.volume);
- up_write(&drbd_cfg_rwsem);
out:
drbd_adm_finish(info, retcode);
return 0;
@@ -3099,9 +3102,7 @@ int drbd_adm_delete_minor(struct sk_buff *skb, struct genl_info *info)
if (retcode != NO_ERROR)
goto out;

- down_write(&drbd_cfg_rwsem);
retcode = adm_delete_minor(adm_ctx.mdev);
- up_write(&drbd_cfg_rwsem);
out:
drbd_adm_finish(info, retcode);
return 0;
@@ -3124,52 +3125,43 @@ int drbd_adm_down(struct sk_buff *skb, struct genl_info *info)
goto out;
}

- down_read(&drbd_cfg_rwsem);
/* demote */
idr_for_each_entry(&adm_ctx.tconn->volumes, mdev, i) {
retcode = drbd_set_role(mdev, R_SECONDARY, 0);
if (retcode < SS_SUCCESS) {
drbd_msg_put_info("failed to demote");
- goto out_unlock;
+ goto out;
}
}
- up_read(&drbd_cfg_rwsem);

- /* disconnect; may stop the receiver;
- * must not hold the drbd_cfg_rwsem */
retcode = conn_try_disconnect(adm_ctx.tconn, 0);
if (retcode < SS_SUCCESS) {
drbd_msg_put_info("failed to disconnect");
goto out;
}

- down_read(&drbd_cfg_rwsem);
/* detach */
idr_for_each_entry(&adm_ctx.tconn->volumes, mdev, i) {
retcode = adm_detach(mdev);
if (retcode < SS_SUCCESS) {
drbd_msg_put_info("failed to detach");
- goto out_unlock;
+ goto out;
}
}
- up_read(&drbd_cfg_rwsem);

/* If we reach this, all volumes (of this tconn) are Secondary,
* Disconnected, Diskless, aka Unconfigured. Make sure all threads have
- * actually stopped, state handling only does drbd_thread_stop_nowait().
- * This needs to be done without holding drbd_cfg_rwsem. */
+ * actually stopped, state handling only does drbd_thread_stop_nowait(). */
drbd_thread_stop(&adm_ctx.tconn->worker);

/* Now, nothing can fail anymore */

/* delete volumes */
- down_write(&drbd_cfg_rwsem);
idr_for_each_entry(&adm_ctx.tconn->volumes, mdev, i) {
retcode = adm_delete_minor(mdev);
if (retcode != NO_ERROR) {
/* "can not happen" */
drbd_msg_put_info("failed to delete volume");
- up_write(&drbd_cfg_rwsem);
goto out;
}
}
@@ -3186,10 +3178,7 @@ int drbd_adm_down(struct sk_buff *skb, struct genl_info *info)
retcode = ERR_CONN_IN_USE;
drbd_msg_put_info("failed to delete connection");
}
- up_write(&drbd_cfg_rwsem);
goto out;
-out_unlock:
- up_read(&drbd_cfg_rwsem);
out:
drbd_adm_finish(info, retcode);
return 0;
@@ -3205,7 +3194,6 @@ int drbd_adm_delete_connection(struct sk_buff *skb, struct genl_info *info)
if (retcode != NO_ERROR)
goto out;

- down_write(&drbd_cfg_rwsem);
if (conn_lowest_minor(adm_ctx.tconn) < 0) {
list_del_rcu(&adm_ctx.tconn->all_tconn);
synchronize_rcu();
@@ -3215,7 +3203,6 @@ int drbd_adm_delete_connection(struct sk_buff *skb, struct genl_info *info)
} else {
retcode = ERR_CONN_IN_USE;
}
- up_write(&drbd_cfg_rwsem);

if (retcode == NO_ERROR)
drbd_thread_stop(&adm_ctx.tconn->worker);
diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c
index 792a71e..6b226cc 100644
--- a/drivers/block/drbd/drbd_proc.c
+++ b/drivers/block/drbd/drbd_proc.c
@@ -229,7 +229,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
oos .. known out-of-sync kB
*/

- down_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
idr_for_each_entry(&minors, mdev, i) {
if (prev_i != i - 1)
seq_printf(seq, "\n");
@@ -242,10 +242,8 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
mdev->state.role == R_SECONDARY) {
seq_printf(seq, "%2d: cs:Unconfigured\n", i);
} else {
- rcu_read_lock();
nc = rcu_dereference(mdev->tconn->net_conf);
wp = nc ? nc->wire_protocol - DRBD_PROT_A + 'A' : ' ';
- rcu_read_unlock();
seq_printf(seq,
"%2d: cs:%s ro:%s/%s ds:%s/%s %c %c%c%c%c%c%c\n"
" ns:%u nr:%u dw:%u dr:%u al:%u bm:%u "
@@ -299,7 +297,7 @@ static int drbd_seq_show(struct seq_file *seq, void *v)
}
}
}
- up_read(&drbd_cfg_rwsem);
+ rcu_read_unlock();

return 0;
}
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index 7156e53..aa42967 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -63,7 +63,7 @@ enum finish_epoch {

static int drbd_do_features(struct drbd_tconn *tconn);
static int drbd_do_auth(struct drbd_tconn *tconn);
-static int drbd_disconnected(int vnr, void *p, void *data);
+static int drbd_disconnected(struct drbd_conf *mdev);

static enum finish_epoch drbd_may_finish_epoch(struct drbd_conf *, struct drbd_epoch *, enum epoch_event);
static int e_end_block(struct drbd_work *, int);
@@ -811,9 +811,8 @@ static int drbd_socket_okay(struct socket **sock)
}
/* Gets called if a connection is established, or if a new minor gets created
in a connection */
-int drbd_connected(int vnr, void *p, void *data)
+int drbd_connected(struct drbd_conf *mdev)
{
- struct drbd_conf *mdev = (struct drbd_conf *)p;
int err;

atomic_set(&mdev->packet_seq, 0);
@@ -847,8 +846,9 @@ int drbd_connected(int vnr, void *p, void *data)
static int conn_connect(struct drbd_tconn *tconn)
{
struct socket *sock, *msock;
+ struct drbd_conf *mdev;
struct net_conf *nc;
- int timeout, try, h, ok;
+ int vnr, timeout, try, h, ok;

if (conn_request_state(tconn, NS(conn, C_WF_CONNECTION), CS_VERBOSE) < SS_SUCCESS)
return -2;
@@ -1001,9 +1001,16 @@ retry:
if (drbd_send_protocol(tconn) == -EOPNOTSUPP)
return -1;

- down_read(&drbd_cfg_rwsem);
- h = !idr_for_each(&tconn->volumes, drbd_connected, tconn);
- up_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
+ kref_get(&mdev->kref);
+ rcu_read_unlock();
+ drbd_connected(mdev);
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ rcu_read_lock();
+ }
+ rcu_read_unlock();
+
return h;

out_release_sockets:
@@ -4242,8 +4249,9 @@ void conn_flush_workqueue(struct drbd_tconn *tconn)

static void conn_disconnect(struct drbd_tconn *tconn)
{
+ struct drbd_conf *mdev;
enum drbd_conns oc;
- int rv = SS_UNKNOWN_ERROR;
+ int vnr, rv = SS_UNKNOWN_ERROR;

if (tconn->cstate == C_STANDALONE)
return;
@@ -4252,9 +4260,16 @@ static void conn_disconnect(struct drbd_tconn *tconn)
drbd_thread_stop(&tconn->asender);
drbd_free_sock(tconn);

- down_read(&drbd_cfg_rwsem);
- idr_for_each(&tconn->volumes, drbd_disconnected, tconn);
- up_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
+ kref_get(&mdev->kref);
+ rcu_read_unlock();
+ drbd_disconnected(mdev);
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ rcu_read_lock();
+ }
+ rcu_read_unlock();
+
conn_info(tconn, "Connection closed\n");

if (conn_highest_role(tconn) == R_PRIMARY && conn_highest_pdsk(tconn) >= D_UNKNOWN)
@@ -4271,9 +4286,8 @@ static void conn_disconnect(struct drbd_tconn *tconn)
conn_request_state(tconn, NS(conn, C_STANDALONE), CS_VERBOSE | CS_HARD);
}

-static int drbd_disconnected(int vnr, void *p, void *data)
+static int drbd_disconnected(struct drbd_conf *mdev)
{
- struct drbd_conf *mdev = (struct drbd_conf *)p;
enum drbd_fencing_p fp;
unsigned int i;

@@ -4974,30 +4988,33 @@ static int got_skip(struct drbd_tconn *tconn, struct packet_info *pi)
static int tconn_finish_peer_reqs(struct drbd_tconn *tconn)
{
struct drbd_conf *mdev;
- int i, not_empty = 0;
+ int vnr, not_empty = 0;

do {
clear_bit(SIGNAL_ASENDER, &tconn->flags);
flush_signals(current);
- down_read(&drbd_cfg_rwsem);
- idr_for_each_entry(&tconn->volumes, mdev, i) {
+
+ rcu_read_lock();
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
+ kref_get(&mdev->kref);
+ rcu_read_unlock();
if (drbd_finish_peer_reqs(mdev)) {
- up_read(&drbd_cfg_rwsem);
- return 1; /* error */
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ return 1;
}
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ rcu_read_lock();
}
- up_read(&drbd_cfg_rwsem);
set_bit(SIGNAL_ASENDER, &tconn->flags);

spin_lock_irq(&tconn->req_lock);
- rcu_read_lock();
- idr_for_each_entry(&tconn->volumes, mdev, i) {
+ idr_for_each_entry(&tconn->volumes, mdev, vnr) {
not_empty = !list_empty(&mdev->done_ee);
if (not_empty)
break;
}
- rcu_read_unlock();
spin_unlock_irq(&tconn->req_lock);
+ rcu_read_unlock();
} while (not_empty);

return 0;
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index feb12b0..46b5f7f 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -1774,12 +1774,16 @@ int drbd_worker(struct drbd_thread *thi)
*/
spin_unlock_irq(&tconn->data.work.q_lock);

- down_read(&drbd_cfg_rwsem);
+ rcu_read_lock();
idr_for_each_entry(&tconn->volumes, mdev, vnr) {
D_ASSERT(mdev->state.disk == D_DISKLESS && mdev->state.conn == C_STANDALONE);
+ kref_get(&mdev->kref);
+ rcu_read_unlock();
drbd_mdev_cleanup(mdev);
+ kref_put(&mdev->kref, &drbd_minor_destroy);
+ rcu_read_lock();
}
- up_read(&drbd_cfg_rwsem);
+ rcu_read_unlock();

return 0;
}
--
1.7.4.1

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/