* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 01/16] fs: dlm: always run complete for possible waiters
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 02/16] fs: dlm: add dlm macros for ratelimit log Alexander Aring
` (14 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch changes the ping_members() result that we always run
complete() for possible waiters. We handle the -EINTR error code as
successful. This error code is returned if the recovery is stopped which
is likely that a new recovery is triggered with a new members
configuration and ping_members() runs again.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/member.c | 18 ++++++++++++------
1 file changed, 12 insertions(+), 6 deletions(-)
diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index ceef3f2074ff..48245ba3c640 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -576,12 +576,18 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
*neg_out = neg;
error = ping_members(ls);
- if (!error || error == -EPROTO) {
- /* new_lockspace() may be waiting to know if the config
- is good or bad */
- ls->ls_members_result = error;
- complete(&ls->ls_members_done);
- }
+ /* error -EINTR means that a new recovery action is triggered.
+ * We ignore this recovery action and let run the new one which might
+ * have new member configuration.
+ */
+ if (error == -EINTR)
+ error = 0;
+
+ /* new_lockspace() may be waiting to know if the config
+ * is good or bad
+ */
+ ls->ls_members_result = error;
+ complete(&ls->ls_members_done);
log_rinfo(ls, "dlm_recover_members %d nodes", ls->ls_num_nodes);
return error;
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 02/16] fs: dlm: add dlm macros for ratelimit log
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 01/16] fs: dlm: always run complete for possible waiters Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 03/16] fs: dlm: fix srcu read lock usage Alexander Aring
` (13 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch add ratelimit macro to dlm subsystem and will set the
connecting log message to ratelimit. In non blocking connecting cases it
will print out this message a lot.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/dlm_internal.h | 2 ++
fs/dlm/lowcomms.c | 4 ++--
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 04fe9f525ac7..ae3fdf6d9cda 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -60,6 +60,8 @@ struct dlm_mhandle;
#define log_print(fmt, args...) \
printk(KERN_ERR "dlm: "fmt"\n" , ##args)
+#define log_print_ratelimited(fmt, args...) \
+ printk_ratelimited(KERN_ERR "dlm: "fmt"\n", ##args)
#define log_error(ls, fmt, args...) \
printk(KERN_ERR "dlm: %s: " fmt "\n", (ls)->ls_name , ##args)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 166e36fcf3e4..15810701b13e 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -1075,7 +1075,7 @@ static void sctp_connect_to_sock(struct connection *con)
make_sockaddr(&daddr, dlm_config.ci_tcp_port, &addr_len);
- log_print("connecting to %d", con->nodeid);
+ log_print_ratelimited("connecting to %d", con->nodeid);
/* Turn off Nagle's algorithm */
sctp_sock_set_nodelay(sock->sk);
@@ -1171,7 +1171,7 @@ static void tcp_connect_to_sock(struct connection *con)
make_sockaddr(&saddr, dlm_config.ci_tcp_port, &addr_len);
- log_print("connecting to %d", con->nodeid);
+ log_print_ratelimited("connecting to %d", con->nodeid);
/* Turn off Nagle's algorithm */
tcp_sock_set_nodelay(sock->sk);
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 03/16] fs: dlm: fix srcu read lock usage
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 01/16] fs: dlm: always run complete for possible waiters Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 02/16] fs: dlm: add dlm macros for ratelimit log Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 04/16] fs: dlm: set is othercon flag Alexander Aring
` (12 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch holds the srcu connection read lock in cases where we lookup
the connections and accessing it. We don't hold the srcu lock in workers
function where the scheduled worker is part of the connection itself.
The connection should not be freed if any worker is scheduled or
pending.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 75 ++++++++++++++++++++++++++++++++---------------
1 file changed, 52 insertions(+), 23 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 15810701b13e..7c7a31377f27 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -113,6 +113,7 @@ struct writequeue_entry {
int len;
int end;
int users;
+ int idx; /* get()/commit() idx exchange */
struct connection *con;
};
@@ -163,21 +164,14 @@ static inline int nodeid_hash(int nodeid)
return nodeid & (CONN_HASH_SIZE-1);
}
-static struct connection *__find_con(int nodeid)
+static struct connection *__find_con(int nodeid, int r)
{
- int r, idx;
struct connection *con;
- r = nodeid_hash(nodeid);
-
- idx = srcu_read_lock(&connections_srcu);
hlist_for_each_entry_rcu(con, &connection_hash[r], list) {
- if (con->nodeid == nodeid) {
- srcu_read_unlock(&connections_srcu, idx);
+ if (con->nodeid == nodeid)
return con;
- }
}
- srcu_read_unlock(&connections_srcu, idx);
return NULL;
}
@@ -216,7 +210,8 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
struct connection *con, *tmp;
int r, ret;
- con = __find_con(nodeid);
+ r = nodeid_hash(nodeid);
+ con = __find_con(nodeid, r);
if (con || !alloc)
return con;
@@ -230,8 +225,6 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
return NULL;
}
- r = nodeid_hash(nodeid);
-
spin_lock(&connections_lock);
/* Because multiple workqueues/threads calls this function it can
* race on multiple cpu's. Instead of locking hot path __find_con()
@@ -239,7 +232,7 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
* under protection of connections_lock. If this is the case we
* abort our connection creation and return the existing connection.
*/
- tmp = __find_con(nodeid);
+ tmp = __find_con(nodeid, r);
if (tmp) {
spin_unlock(&connections_lock);
kfree(con->rx_buf);
@@ -256,15 +249,13 @@ static struct connection *nodeid2con(int nodeid, gfp_t alloc)
/* Loop round all connections */
static void foreach_conn(void (*conn_func)(struct connection *c))
{
- int i, idx;
+ int i;
struct connection *con;
- idx = srcu_read_lock(&connections_srcu);
for (i = 0; i < CONN_HASH_SIZE; i++) {
hlist_for_each_entry_rcu(con, &connection_hash[i], list)
conn_func(con);
}
- srcu_read_unlock(&connections_srcu, idx);
}
static struct dlm_node_addr *find_node_addr(int nodeid)
@@ -518,14 +509,21 @@ static void lowcomms_state_change(struct sock *sk)
int dlm_lowcomms_connect_node(int nodeid)
{
struct connection *con;
+ int idx;
if (nodeid == dlm_our_nodeid())
return 0;
+ idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, GFP_NOFS);
- if (!con)
+ if (!con) {
+ srcu_read_unlock(&connections_srcu, idx);
return -ENOMEM;
+ }
+
lowcomms_connect_sock(con);
+ srcu_read_unlock(&connections_srcu, idx);
+
return 0;
}
@@ -864,7 +862,7 @@ static int accept_from_sock(struct listen_connection *con)
int result;
struct sockaddr_storage peeraddr;
struct socket *newsock;
- int len;
+ int len, idx;
int nodeid;
struct connection *newcon;
struct connection *addcon;
@@ -907,8 +905,10 @@ static int accept_from_sock(struct listen_connection *con)
* the same time and the connections cross on the wire.
* In this case we store the incoming one in "othercon"
*/
+ idx = srcu_read_lock(&connections_srcu);
newcon = nodeid2con(nodeid, GFP_NOFS);
if (!newcon) {
+ srcu_read_unlock(&connections_srcu, idx);
result = -ENOMEM;
goto accept_err;
}
@@ -924,6 +924,7 @@ static int accept_from_sock(struct listen_connection *con)
if (!othercon) {
log_print("failed to allocate incoming socket");
mutex_unlock(&newcon->sock_mutex);
+ srcu_read_unlock(&connections_srcu, idx);
result = -ENOMEM;
goto accept_err;
}
@@ -932,6 +933,7 @@ static int accept_from_sock(struct listen_connection *con)
if (result < 0) {
kfree(othercon);
mutex_unlock(&newcon->sock_mutex);
+ srcu_read_unlock(&connections_srcu, idx);
goto accept_err;
}
@@ -966,6 +968,8 @@ static int accept_from_sock(struct listen_connection *con)
if (!test_and_set_bit(CF_READ_PENDING, &addcon->flags))
queue_work(recv_workqueue, &addcon->rwork);
+ srcu_read_unlock(&connections_srcu, idx);
+
return 0;
accept_err:
@@ -1403,7 +1407,9 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
{
+ struct writequeue_entry *e;
struct connection *con;
+ int idx;
if (len > DEFAULT_BUFFER_SIZE ||
len < sizeof(struct dlm_header)) {
@@ -1413,11 +1419,23 @@ void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
return NULL;
}
+ idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, allocation);
- if (!con)
+ if (!con) {
+ srcu_read_unlock(&connections_srcu, idx);
return NULL;
+ }
+
+ e = new_wq_entry(con, len, allocation, ppc);
+ if (!e) {
+ srcu_read_unlock(&connections_srcu, idx);
+ return NULL;
+ }
+
+ /* we assume if successful commit must called */
+ e->idx = idx;
- return new_wq_entry(con, len, allocation, ppc);
+ return e;
}
void dlm_lowcomms_commit_buffer(void *mh)
@@ -1435,10 +1453,12 @@ void dlm_lowcomms_commit_buffer(void *mh)
spin_unlock(&con->writequeue_lock);
queue_work(send_workqueue, &con->swork);
+ srcu_read_unlock(&connections_srcu, e->idx);
return;
out:
spin_unlock(&con->writequeue_lock);
+ srcu_read_unlock(&connections_srcu, e->idx);
return;
}
@@ -1532,8 +1552,10 @@ int dlm_lowcomms_close(int nodeid)
{
struct connection *con;
struct dlm_node_addr *na;
+ int idx;
log_print("closing connection to node %d", nodeid);
+ idx = srcu_read_lock(&connections_srcu);
con = nodeid2con(nodeid, 0);
if (con) {
set_bit(CF_CLOSE, &con->flags);
@@ -1542,6 +1564,7 @@ int dlm_lowcomms_close(int nodeid)
if (con->othercon)
clean_one_writequeue(con->othercon);
}
+ srcu_read_unlock(&connections_srcu, idx);
spin_lock(&dlm_node_addrs_spin);
na = find_node_addr(nodeid);
@@ -1621,6 +1644,8 @@ static void shutdown_conn(struct connection *con)
void dlm_lowcomms_shutdown(void)
{
+ int idx;
+
/* Set all the flags to prevent any
* socket activity.
*/
@@ -1633,7 +1658,9 @@ void dlm_lowcomms_shutdown(void)
dlm_close_sock(&listen_con.sock);
+ idx = srcu_read_lock(&connections_srcu);
foreach_conn(shutdown_conn);
+ srcu_read_unlock(&connections_srcu, idx);
}
static void _stop_conn(struct connection *con, bool and_other)
@@ -1682,7 +1709,7 @@ static void free_conn(struct connection *con)
static void work_flush(void)
{
- int ok, idx;
+ int ok;
int i;
struct connection *con;
@@ -1693,7 +1720,6 @@ static void work_flush(void)
flush_workqueue(recv_workqueue);
if (send_workqueue)
flush_workqueue(send_workqueue);
- idx = srcu_read_lock(&connections_srcu);
for (i = 0; i < CONN_HASH_SIZE && ok; i++) {
hlist_for_each_entry_rcu(con, &connection_hash[i],
list) {
@@ -1707,14 +1733,17 @@ static void work_flush(void)
}
}
}
- srcu_read_unlock(&connections_srcu, idx);
} while (!ok);
}
void dlm_lowcomms_stop(void)
{
+ int idx;
+
+ idx = srcu_read_lock(&connections_srcu);
work_flush();
foreach_conn(free_conn);
+ srcu_read_unlock(&connections_srcu, idx);
work_stop();
deinit_local();
}
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 04/16] fs: dlm: set is othercon flag
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (2 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 03/16] fs: dlm: fix srcu read lock usage Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 05/16] fs: dlm: reconnect if socket error report occurs Alexander Aring
` (11 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
There is a is othercon flag which is never used, this patch will set it
and printout a warning if the othercon ever sends a dlm message which
should never be the case.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 7c7a31377f27..4944aef24aa5 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -938,6 +938,7 @@ static int accept_from_sock(struct listen_connection *con)
}
lockdep_set_subclass(&othercon->sock_mutex, 1);
+ set_bit(CF_IS_OTHERCON, &othercon->flags);
newcon->othercon = othercon;
} else {
/* close other sock con if we have something new */
@@ -1601,6 +1602,8 @@ static void process_send_sockets(struct work_struct *work)
{
struct connection *con = container_of(work, struct connection, swork);
+ WARN_ON(test_bit(CF_IS_OTHERCON, &con->flags));
+
clear_bit(CF_WRITE_PENDING, &con->flags);
if (con->sock == NULL) /* not mutex protected so check it inside too */
con->connect_action(con);
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 05/16] fs: dlm: reconnect if socket error report occurs
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (3 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 04/16] fs: dlm: set is othercon flag Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 06/16] fs: dlm: cancel work sync othercon Alexander Aring
` (10 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch will change the reconnect handling that if an error occurs
if a socket error callback is occurred. This will also handle reconnects
in a non blocking connecting case which is currently missing. If error
ECONNREFUSED is reported we delay the reconnect by one second.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 60 ++++++++++++++++++++++++++++++-----------------
1 file changed, 39 insertions(+), 21 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 4944aef24aa5..051f22dbb83a 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -79,6 +79,8 @@ struct connection {
#define CF_CLOSING 8
#define CF_SHUTDOWN 9
#define CF_CONNECTED 10
+#define CF_RECONNECT 11
+#define CF_DELAY_CONNECT 12
struct list_head writequeue; /* List of outgoing writequeue_entries */
spinlock_t writequeue_lock;
void (*connect_action) (struct connection *); /* What to do to connect */
@@ -87,6 +89,7 @@ struct connection {
#define MAX_CONNECT_RETRIES 3
struct hlist_node list;
struct connection *othercon;
+ struct connection *sendcon;
struct work_struct rwork; /* Receive workqueue */
struct work_struct swork; /* Send workqueue */
wait_queue_head_t shutdown_wait; /* wait for graceful shutdown */
@@ -585,6 +588,22 @@ static void lowcomms_error_report(struct sock *sk)
dlm_config.ci_tcp_port, sk->sk_err,
sk->sk_err_soft);
}
+
+ /* below sendcon only handling */
+ if (test_bit(CF_IS_OTHERCON, &con->flags))
+ con = con->sendcon;
+
+ switch (sk->sk_err) {
+ case ECONNREFUSED:
+ set_bit(CF_DELAY_CONNECT, &con->flags);
+ break;
+ default:
+ break;
+ }
+
+ if (!test_and_set_bit(CF_RECONNECT, &con->flags))
+ queue_work(send_workqueue, &con->swork);
+
out:
read_unlock_bh(&sk->sk_callback_lock);
if (orig_report)
@@ -702,6 +721,8 @@ static void close_connection(struct connection *con, bool and_other,
con->rx_leftover = 0;
con->retries = 0;
clear_bit(CF_CONNECTED, &con->flags);
+ clear_bit(CF_DELAY_CONNECT, &con->flags);
+ clear_bit(CF_RECONNECT, &con->flags);
mutex_unlock(&con->sock_mutex);
clear_bit(CF_CLOSING, &con->flags);
}
@@ -840,18 +861,15 @@ static int receive_from_sock(struct connection *con)
out_close:
mutex_unlock(&con->sock_mutex);
- if (ret != -EAGAIN) {
- /* Reconnect when there is something to send */
+ if (ret == 0) {
close_connection(con, false, true, false);
- if (ret == 0) {
- log_print("connection %p got EOF from %d",
- con, con->nodeid);
- /* handling for tcp shutdown */
- clear_bit(CF_SHUTDOWN, &con->flags);
- wake_up(&con->shutdown_wait);
- /* signal to breaking receive worker */
- ret = -1;
- }
+ log_print("connection %p got EOF from %d",
+ con, con->nodeid);
+ /* handling for tcp shutdown */
+ clear_bit(CF_SHUTDOWN, &con->flags);
+ wake_up(&con->shutdown_wait);
+ /* signal to breaking receive worker */
+ ret = -1;
}
return ret;
}
@@ -940,6 +958,7 @@ static int accept_from_sock(struct listen_connection *con)
lockdep_set_subclass(&othercon->sock_mutex, 1);
set_bit(CF_IS_OTHERCON, &othercon->flags);
newcon->othercon = othercon;
+ othercon->sendcon = newcon;
} else {
/* close other sock con if we have something new */
close_connection(othercon, false, true, false);
@@ -1504,7 +1523,7 @@ static void send_to_sock(struct connection *con)
cond_resched();
goto out;
} else if (ret < 0)
- goto send_error;
+ goto out;
}
/* Don't starve people filling buffers */
@@ -1521,14 +1540,6 @@ static void send_to_sock(struct connection *con)
mutex_unlock(&con->sock_mutex);
return;
-send_error:
- mutex_unlock(&con->sock_mutex);
- close_connection(con, false, false, true);
- /* Requeue the send work. When the work daemon runs again, it will try
- a new connection, then call this function again. */
- queue_work(send_workqueue, &con->swork);
- return;
-
out_connect:
mutex_unlock(&con->sock_mutex);
queue_work(send_workqueue, &con->swork);
@@ -1605,8 +1616,15 @@ static void process_send_sockets(struct work_struct *work)
WARN_ON(test_bit(CF_IS_OTHERCON, &con->flags));
clear_bit(CF_WRITE_PENDING, &con->flags);
- if (con->sock == NULL) /* not mutex protected so check it inside too */
+
+ if (test_and_clear_bit(CF_RECONNECT, &con->flags))
+ close_connection(con, false, false, true);
+
+ if (con->sock == NULL) { /* not mutex protected so check it inside too */
+ if (test_and_clear_bit(CF_DELAY_CONNECT, &con->flags))
+ msleep(1000);
con->connect_action(con);
+ }
if (!list_empty(&con->writequeue))
send_to_sock(con);
}
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 06/16] fs: dlm: cancel work sync othercon
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (4 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 05/16] fs: dlm: reconnect if socket error report occurs Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 07/16] fs: dlm: fix connection tcp EOF handling Alexander Aring
` (9 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
These rx tx flags arguments are for signaling close_connection() from
which worker they are called. Obviously the receive worker cannot cancel
itself and vice versa for swork. For the othercon the receive worker
should only be used, however to avoid deadlocks we should pass the same
flags as the original close_connection() was called.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 051f22dbb83a..0a4851b3cd4b 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -715,7 +715,7 @@ static void close_connection(struct connection *con, bool and_other,
if (con->othercon && and_other) {
/* Will only re-enter once. */
- close_connection(con->othercon, false, true, true);
+ close_connection(con->othercon, false, tx, rx);
}
con->rx_leftover = 0;
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 07/16] fs: dlm: fix connection tcp EOF handling
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (5 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 06/16] fs: dlm: cancel work sync othercon Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 08/16] fs: dlm: public header in out utility Alexander Aring
` (8 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch fixes the EOF handling for TCP that if and EOF is received we
will close the socket next time the writequeue runs empty. This is a
half-closed socket functionality which doesn't exists in SCTP. The
midcomms layer will do a half closed socket functionality on DLM side to
solve this problem for the SCTP case. However there is still the last ack
flying around but other reset functionality will take care of it if it got
lost.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 48 ++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 43 insertions(+), 5 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 0a4851b3cd4b..14ca3eda6a83 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -81,10 +81,13 @@ struct connection {
#define CF_CONNECTED 10
#define CF_RECONNECT 11
#define CF_DELAY_CONNECT 12
+#define CF_EOF 13
struct list_head writequeue; /* List of outgoing writequeue_entries */
spinlock_t writequeue_lock;
+ atomic_t writequeue_cnt;
void (*connect_action) (struct connection *); /* What to do to connect */
void (*shutdown_action)(struct connection *con); /* What to do to shutdown */
+ bool (*eof_condition)(struct connection *con); /* What to do to eof check */
int retries;
#define MAX_CONNECT_RETRIES 3
struct hlist_node list;
@@ -179,6 +182,11 @@ static struct connection *__find_con(int nodeid, int r)
return NULL;
}
+static bool tcp_eof_condition(struct connection *con)
+{
+ return atomic_read(&con->writequeue_cnt);
+}
+
static int dlm_con_init(struct connection *con, int nodeid)
{
con->rx_buflen = dlm_config.ci_buffer_size;
@@ -190,6 +198,7 @@ static int dlm_con_init(struct connection *con, int nodeid)
mutex_init(&con->sock_mutex);
INIT_LIST_HEAD(&con->writequeue);
spin_lock_init(&con->writequeue_lock);
+ atomic_set(&con->writequeue_cnt, 0);
INIT_WORK(&con->swork, process_send_sockets);
INIT_WORK(&con->rwork, process_recv_sockets);
init_waitqueue_head(&con->shutdown_wait);
@@ -197,6 +206,7 @@ static int dlm_con_init(struct connection *con, int nodeid)
if (dlm_config.ci_protocol == 0) {
con->connect_action = tcp_connect_to_sock;
con->shutdown_action = dlm_tcp_shutdown;
+ con->eof_condition = tcp_eof_condition;
} else {
con->connect_action = sctp_connect_to_sock;
}
@@ -723,6 +733,7 @@ static void close_connection(struct connection *con, bool and_other,
clear_bit(CF_CONNECTED, &con->flags);
clear_bit(CF_DELAY_CONNECT, &con->flags);
clear_bit(CF_RECONNECT, &con->flags);
+ clear_bit(CF_EOF, &con->flags);
mutex_unlock(&con->sock_mutex);
clear_bit(CF_CLOSING, &con->flags);
}
@@ -860,16 +871,26 @@ static int receive_from_sock(struct connection *con)
return -EAGAIN;
out_close:
- mutex_unlock(&con->sock_mutex);
if (ret == 0) {
- close_connection(con, false, true, false);
log_print("connection %p got EOF from %d",
con, con->nodeid);
- /* handling for tcp shutdown */
- clear_bit(CF_SHUTDOWN, &con->flags);
- wake_up(&con->shutdown_wait);
+
+ if (con->eof_condition && con->eof_condition(con)) {
+ set_bit(CF_EOF, &con->flags);
+ mutex_unlock(&con->sock_mutex);
+ } else {
+ mutex_unlock(&con->sock_mutex);
+ close_connection(con, false, true, false);
+
+ /* handling for tcp shutdown */
+ clear_bit(CF_SHUTDOWN, &con->flags);
+ wake_up(&con->shutdown_wait);
+ }
+
/* signal to breaking receive worker */
ret = -1;
+ } else {
+ mutex_unlock(&con->sock_mutex);
}
return ret;
}
@@ -1021,6 +1042,7 @@ static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
if (e->len == 0 && e->users == 0) {
list_del(&e->list);
+ atomic_dec(&e->con->writequeue_cnt);
free_entry(e);
}
}
@@ -1417,6 +1439,7 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
*ppc = page_address(e->page);
e->end += len;
+ atomic_inc(&con->writequeue_cnt);
spin_lock(&con->writequeue_lock);
list_add_tail(&e->list, &con->writequeue);
@@ -1536,6 +1559,21 @@ static void send_to_sock(struct connection *con)
writequeue_entry_complete(e, ret);
}
spin_unlock(&con->writequeue_lock);
+
+ /* close if we got EOF */
+ if (test_and_clear_bit(CF_EOF, &con->flags)) {
+ mutex_unlock(&con->sock_mutex);
+ close_connection(con, false, false, true);
+
+ /* handling for tcp shutdown */
+ clear_bit(CF_SHUTDOWN, &con->flags);
+ wake_up(&con->shutdown_wait);
+ } else {
+ mutex_unlock(&con->sock_mutex);
+ }
+
+ return;
+
out:
mutex_unlock(&con->sock_mutex);
return;
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 08/16] fs: dlm: public header in out utility
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (6 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 07/16] fs: dlm: fix connection tcp EOF handling Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 09/16] fs: dlm: add more midcomms hooks Alexander Aring
` (7 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch allows to use header_out() and header_in() outside of dlm
util functionality.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/util.c | 4 ++--
fs/dlm/util.h | 2 ++
2 files changed, 4 insertions(+), 2 deletions(-)
diff --git a/fs/dlm/util.c b/fs/dlm/util.c
index cfd0d00b19ae..74a8c5bfe9b5 100644
--- a/fs/dlm/util.c
+++ b/fs/dlm/util.c
@@ -20,7 +20,7 @@
#define DLM_ERRNO_ETIMEDOUT 110
#define DLM_ERRNO_EINPROGRESS 115
-static void header_out(struct dlm_header *hd)
+void header_out(struct dlm_header *hd)
{
hd->h_version = cpu_to_le32(hd->h_version);
hd->h_lockspace = cpu_to_le32(hd->h_lockspace);
@@ -28,7 +28,7 @@ static void header_out(struct dlm_header *hd)
hd->h_length = cpu_to_le16(hd->h_length);
}
-static void header_in(struct dlm_header *hd)
+void header_in(struct dlm_header *hd)
{
hd->h_version = le32_to_cpu(hd->h_version);
hd->h_lockspace = le32_to_cpu(hd->h_lockspace);
diff --git a/fs/dlm/util.h b/fs/dlm/util.h
index cc719ca9397e..d46f23c7a6a0 100644
--- a/fs/dlm/util.h
+++ b/fs/dlm/util.h
@@ -15,6 +15,8 @@ void dlm_message_out(struct dlm_message *ms);
void dlm_message_in(struct dlm_message *ms);
void dlm_rcom_out(struct dlm_rcom *rc);
void dlm_rcom_in(struct dlm_rcom *rc);
+void header_out(struct dlm_header *hd);
+void header_in(struct dlm_header *hd);
#endif
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 09/16] fs: dlm: add more midcomms hooks
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (7 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 08/16] fs: dlm: public header in out utility Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 10/16] fs: dlm: make buffer handling per msg Alexander Aring
` (6 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch prepares hooks to redirect to the midcomms layer which will
be used by the midcomms re-transmit handling.
There exists the new concept of stateless buffers allocation and
commits. This can be used to bypass the midcomms re-transmit handling. It
is used by RCOM_STATUS and RCOM_NAMES messages, because they have their
own ping-like re-transmit handling. As well these two messages will be
used to determine the DLM version per node, because these two messages
are per observation the first messages which are exchanged.
Cluster manager events for node membership are added to add support for
half-closed connections in cases that the peer connection get to
an end of file but DLM still holds membership of the node. In
this time DLM can still trigger new message which we should allow. After
the cluster manager node removal event occurs it safe to close the
connection.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/config.c | 3 +-
fs/dlm/lock.c | 8 ++--
fs/dlm/lockspace.c | 7 ++--
fs/dlm/member.c | 17 ++++++--
fs/dlm/midcomms.c | 31 +++++++++++++-
fs/dlm/midcomms.h | 8 ++++
fs/dlm/rcom.c | 101 +++++++++++++++++++++++++++++++--------------
7 files changed, 133 insertions(+), 42 deletions(-)
diff --git a/fs/dlm/config.c b/fs/dlm/config.c
index 88d95d96e36c..01ae294743e9 100644
--- a/fs/dlm/config.c
+++ b/fs/dlm/config.c
@@ -20,6 +20,7 @@
#include <net/sock.h>
#include "config.h"
+#include "midcomms.h"
#include "lowcomms.h"
/*
@@ -532,7 +533,7 @@ static void drop_comm(struct config_group *g, struct config_item *i)
struct dlm_comm *cm = config_item_to_comm(i);
if (local_comm == cm)
local_comm = NULL;
- dlm_lowcomms_close(cm->nodeid);
+ dlm_midcomms_close(cm->nodeid);
while (cm->addr_count--)
kfree(cm->addr[cm->addr_count]);
config_item_put(i);
diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c
index b93df39d0915..b625ce92464a 100644
--- a/fs/dlm/lock.c
+++ b/fs/dlm/lock.c
@@ -59,7 +59,7 @@
#include "dlm_internal.h"
#include <linux/dlm_device.h>
#include "memory.h"
-#include "lowcomms.h"
+#include "midcomms.h"
#include "requestqueue.h"
#include "util.h"
#include "dir.h"
@@ -3534,10 +3534,10 @@ static int _create_message(struct dlm_ls *ls, int mb_len,
char *mb;
/* get_buffer gives us a message handle (mh) that we need to
- pass into lowcomms_commit and a message buffer (mb) that we
+ pass into midcomms_commit and a message buffer (mb) that we
write our data into */
- mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
+ mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, GFP_NOFS, &mb);
if (!mh)
return -ENOBUFS;
@@ -3589,7 +3589,7 @@ static int create_message(struct dlm_rsb *r, struct dlm_lkb *lkb,
static int send_message(struct dlm_mhandle *mh, struct dlm_message *ms)
{
dlm_message_out(ms);
- dlm_lowcomms_commit_buffer(mh);
+ dlm_midcomms_commit_mhandle(mh);
return 0;
}
diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index c14cf2b7faab..bf5c55ef9d0d 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -16,6 +16,7 @@
#include "member.h"
#include "recoverd.h"
#include "dir.h"
+#include "midcomms.h"
#include "lowcomms.h"
#include "config.h"
#include "memory.h"
@@ -390,7 +391,7 @@ static int threads_start(void)
}
/* Thread for sending/receiving messages for all lockspace's */
- error = dlm_lowcomms_start();
+ error = dlm_midcomms_start();
if (error) {
log_print("cannot start dlm lowcomms %d", error);
goto scand_fail;
@@ -698,7 +699,7 @@ int dlm_new_lockspace(const char *name, const char *cluster,
error = 0;
if (!ls_count) {
dlm_scand_stop();
- dlm_lowcomms_shutdown();
+ dlm_midcomms_shutdown();
dlm_lowcomms_stop();
}
out:
@@ -787,7 +788,7 @@ static int release_lockspace(struct dlm_ls *ls, int force)
if (ls_count == 1) {
dlm_scand_stop();
- dlm_lowcomms_shutdown();
+ dlm_midcomms_shutdown();
}
dlm_callback_stop(ls);
diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index 48245ba3c640..63971c594bdc 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -15,6 +15,7 @@
#include "recover.h"
#include "rcom.h"
#include "config.h"
+#include "midcomms.h"
#include "lowcomms.h"
int dlm_slots_version(struct dlm_header *h)
@@ -329,6 +330,7 @@ static int dlm_add_member(struct dlm_ls *ls, struct dlm_config_node *node)
memb->nodeid = node->nodeid;
memb->weight = node->weight;
memb->comm_seq = node->comm_seq;
+ dlm_midcomms_add_member(node->nodeid);
add_ordered_member(ls, memb);
ls->ls_num_nodes++;
return 0;
@@ -359,26 +361,34 @@ int dlm_is_removed(struct dlm_ls *ls, int nodeid)
return 0;
}
-static void clear_memb_list(struct list_head *head)
+static void clear_memb_list(struct list_head *head,
+ void (*after_del)(int nodeid))
{
struct dlm_member *memb;
while (!list_empty(head)) {
memb = list_entry(head->next, struct dlm_member, list);
list_del(&memb->list);
+ if (after_del)
+ after_del(memb->nodeid);
kfree(memb);
}
}
+static void clear_members_cb(int nodeid)
+{
+ dlm_midcomms_remove_member(nodeid);
+}
+
void dlm_clear_members(struct dlm_ls *ls)
{
- clear_memb_list(&ls->ls_nodes);
+ clear_memb_list(&ls->ls_nodes, clear_members_cb);
ls->ls_num_nodes = 0;
}
void dlm_clear_members_gone(struct dlm_ls *ls)
{
- clear_memb_list(&ls->ls_nodes_gone);
+ clear_memb_list(&ls->ls_nodes_gone, NULL);
}
static void make_member_array(struct dlm_ls *ls)
@@ -552,6 +562,7 @@ int dlm_recover_members(struct dlm_ls *ls, struct dlm_recover *rv, int *neg_out)
neg++;
list_move(&memb->list, &ls->ls_nodes_gone);
+ dlm_midcomms_remove_member(memb->nodeid);
ls->ls_num_nodes--;
dlm_lsop_recover_slot(ls, memb);
}
diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c
index 1c6654a21ec4..1a280dda99d3 100644
--- a/fs/dlm/midcomms.c
+++ b/fs/dlm/midcomms.c
@@ -28,6 +28,36 @@
#include "lock.h"
#include "midcomms.h"
+struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
+ gfp_t allocation, char **ppc)
+{
+ return dlm_lowcomms_get_buffer(nodeid, len, allocation, ppc);
+}
+
+void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh)
+{
+ dlm_lowcomms_commit_buffer(mh);
+}
+
+void dlm_midcomms_add_member(int nodeid) { }
+
+void dlm_midcomms_remove_member(int nodeid) { }
+
+int dlm_midcomms_start(void)
+{
+ return dlm_lowcomms_start();
+}
+
+void dlm_midcomms_shutdown(void)
+{
+ dlm_lowcomms_shutdown();
+}
+
+int dlm_midcomms_close(int nodeid)
+{
+ return dlm_lowcomms_close(nodeid);
+}
+
/*
* Called from the low-level comms layer to process a buffer of
* commands.
@@ -101,4 +131,3 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
return ret;
}
-
diff --git a/fs/dlm/midcomms.h b/fs/dlm/midcomms.h
index 61e90a921849..9ac1190ce277 100644
--- a/fs/dlm/midcomms.h
+++ b/fs/dlm/midcomms.h
@@ -13,6 +13,14 @@
#define __MIDCOMMS_DOT_H__
int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int buflen);
+struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
+ gfp_t allocation, char **ppc);
+void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh);
+int dlm_midcomms_close(int nodeid);
+int dlm_midcomms_start(void);
+void dlm_midcomms_shutdown(void);
+void dlm_midcomms_add_member(int nodeid);
+void dlm_midcomms_remove_member(int nodeid);
#endif /* __MIDCOMMS_DOT_H__ */
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index f5b1bd65728d..2661674364af 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -27,20 +27,10 @@ static int rcom_response(struct dlm_ls *ls)
return test_bit(LSFL_RCOM_READY, &ls->ls_flags);
}
-static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
- struct dlm_rcom **rc_ret, struct dlm_mhandle **mh_ret)
+static void _create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
+ struct dlm_rcom **rc_ret, char *mb, int mb_len)
{
struct dlm_rcom *rc;
- struct dlm_mhandle *mh;
- char *mb;
- int mb_len = sizeof(struct dlm_rcom) + len;
-
- mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
- if (!mh) {
- log_print("create_rcom to %d type %d len %d ENOBUFS",
- to_nodeid, type, len);
- return -ENOBUFS;
- }
rc = (struct dlm_rcom *) mb;
@@ -56,15 +46,64 @@ static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
rc->rc_seq = ls->ls_recover_seq;
spin_unlock(&ls->ls_recover_lock);
- *mh_ret = mh;
*rc_ret = rc;
+}
+
+static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
+ struct dlm_rcom **rc_ret, struct dlm_mhandle **mh_ret)
+{
+ int mb_len = sizeof(struct dlm_rcom) + len;
+ struct dlm_mhandle *mh;
+ char *mb;
+
+ mh = dlm_midcomms_get_mhandle(to_nodeid, mb_len, GFP_NOFS, &mb);
+ if (!mh) {
+ log_print("%s to %d type %d len %d ENOBUFS",
+ __func__, to_nodeid, type, len);
+ return -ENOBUFS;
+ }
+
+ _create_rcom(ls, to_nodeid, type, len, rc_ret, mb, mb_len);
+ *mh_ret = mh;
return 0;
}
+static int create_rcom_stateless(struct dlm_ls *ls, int to_nodeid, int type,
+ int len, struct dlm_rcom **rc_ret,
+ void **mh_ret)
+{
+ int mb_len = sizeof(struct dlm_rcom) + len;
+ void *mh;
+ char *mb;
+
+ mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
+ if (!mh) {
+ log_print("create_rcom to %d type %d len %d ENOBUFS",
+ to_nodeid, type, len);
+ return -ENOBUFS;
+ }
+
+ _create_rcom(ls, to_nodeid, type, len, rc_ret, mb, mb_len);
+ *mh_ret = mh;
+ return 0;
+}
+
+static void _send_rcom(struct dlm_ls *ls, struct dlm_rcom *rc)
+{
+ dlm_rcom_out(rc);
+}
+
static void send_rcom(struct dlm_ls *ls, struct dlm_mhandle *mh,
struct dlm_rcom *rc)
{
- dlm_rcom_out(rc);
+ _send_rcom(ls, rc);
+ dlm_midcomms_commit_mhandle(mh);
+}
+
+static void send_rcom_stateless(struct dlm_ls *ls, void *mh,
+ struct dlm_rcom *rc)
+{
+ _send_rcom(ls, rc);
dlm_lowcomms_commit_buffer(mh);
}
@@ -141,8 +180,8 @@ static void disallow_sync_reply(struct dlm_ls *ls)
int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
{
struct dlm_rcom *rc;
- struct dlm_mhandle *mh;
int error = 0;
+ void *mh;
ls->ls_recover_nodeid = nodeid;
@@ -153,8 +192,8 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
}
retry:
- error = create_rcom(ls, nodeid, DLM_RCOM_STATUS,
- sizeof(struct rcom_status), &rc, &mh);
+ error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS,
+ sizeof(struct rcom_status), &rc, &mh);
if (error)
goto out;
@@ -163,7 +202,7 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
allow_sync_reply(ls, &rc->rc_id);
memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
- send_rcom(ls, mh, rc);
+ send_rcom_stateless(ls, mh, rc);
error = dlm_wait_function(ls, &rcom_response);
disallow_sync_reply(ls);
@@ -191,13 +230,13 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
{
struct dlm_rcom *rc;
- struct dlm_mhandle *mh;
struct rcom_status *rs;
uint32_t status;
int nodeid = rc_in->rc_header.h_nodeid;
int len = sizeof(struct rcom_config);
int num_slots = 0;
int error;
+ void *mh;
if (!dlm_slots_version(&rc_in->rc_header)) {
status = dlm_recover_status(ls);
@@ -218,8 +257,8 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
len += num_slots * sizeof(struct rcom_slot);
do_create:
- error = create_rcom(ls, nodeid, DLM_RCOM_STATUS_REPLY,
- len, &rc, &mh);
+ error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS_REPLY,
+ len, &rc, &mh);
if (error)
return;
@@ -246,7 +285,7 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
spin_unlock(&ls->ls_recover_lock);
do_send:
- send_rcom(ls, mh, rc);
+ send_rcom_stateless(ls, mh, rc);
}
static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
@@ -271,13 +310,14 @@ static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
{
struct dlm_rcom *rc;
- struct dlm_mhandle *mh;
int error = 0;
+ void *mh;
ls->ls_recover_nodeid = nodeid;
retry:
- error = create_rcom(ls, nodeid, DLM_RCOM_NAMES, last_len, &rc, &mh);
+ error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES, last_len,
+ &rc, &mh);
if (error)
goto out;
memcpy(rc->rc_buf, last_name, last_len);
@@ -285,7 +325,7 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
allow_sync_reply(ls, &rc->rc_id);
memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
- send_rcom(ls, mh, rc);
+ send_rcom_stateless(ls, mh, rc);
error = dlm_wait_function(ls, &rcom_response);
disallow_sync_reply(ls);
@@ -298,14 +338,15 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
{
struct dlm_rcom *rc;
- struct dlm_mhandle *mh;
int error, inlen, outlen, nodeid;
+ void *mh;
nodeid = rc_in->rc_header.h_nodeid;
inlen = rc_in->rc_header.h_length - sizeof(struct dlm_rcom);
outlen = LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom);
- error = create_rcom(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen, &rc, &mh);
+ error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen,
+ &rc, &mh);
if (error)
return;
rc->rc_id = rc_in->rc_id;
@@ -313,7 +354,7 @@ static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
dlm_copy_master_names(ls, rc_in->rc_buf, inlen, rc->rc_buf, outlen,
nodeid);
- send_rcom(ls, mh, rc);
+ send_rcom_stateless(ls, mh, rc);
}
int dlm_send_rcom_lookup(struct dlm_rsb *r, int dir_nodeid)
@@ -458,7 +499,7 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
char *mb;
int mb_len = sizeof(struct dlm_rcom) + sizeof(struct rcom_config);
- mh = dlm_lowcomms_get_buffer(nodeid, mb_len, GFP_NOFS, &mb);
+ mh = dlm_midcomms_get_mhandle(nodeid, mb_len, GFP_NOFS, &mb);
if (!mh)
return -ENOBUFS;
@@ -479,7 +520,7 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
rf->rf_lvblen = cpu_to_le32(~0U);
dlm_rcom_out(rc);
- dlm_lowcomms_commit_buffer(mh);
+ dlm_midcomms_commit_mhandle(mh);
return 0;
}
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 10/16] fs: dlm: make buffer handling per msg
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (8 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 09/16] fs: dlm: add more midcomms hooks Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 11/16] fs: dlm: add functionality to re-transmit a message Alexander Aring
` (5 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch makes the void pointer handle for lowcomms functionality per
message and not per page allocation entry. A refcount handling for the
handle was added to keep the message alive until the user doesn't need
it anymore.
There exists now a per message callback which will be called when
allocating a new buffer. This callback will be guaranteed to be called
according the order of the sending buffer, which can be used that the
caller increments a sequence number for the dlm message handle.
For transition process we cast the dlm_mhandle to dlm_msg and vice versa
until the midcomms layer will implement a specific dlm_mhandle structure.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/dlm_internal.h | 1 +
fs/dlm/lowcomms.c | 101 +++++++++++++++++++++++++++++++++++-------
fs/dlm/lowcomms.h | 7 ++-
fs/dlm/midcomms.c | 7 ++-
fs/dlm/rcom.c | 40 +++++++++--------
5 files changed, 117 insertions(+), 39 deletions(-)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index ae3fdf6d9cda..e8dc5f4f1f9e 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -57,6 +57,7 @@ struct dlm_header;
struct dlm_message;
struct dlm_rcom;
struct dlm_mhandle;
+struct dlm_msg;
#define log_print(fmt, args...) \
printk(KERN_ERR "dlm: "fmt"\n" , ##args)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 14ca3eda6a83..d222e6088ab2 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -119,8 +119,19 @@ struct writequeue_entry {
int len;
int end;
int users;
- int idx; /* get()/commit() idx exchange */
struct connection *con;
+ struct list_head msgs;
+ struct kref ref;
+};
+
+struct dlm_msg {
+ struct writequeue_entry *entry;
+ void *ppc;
+ int len;
+ int idx; /* new()/commit() idx exchange */
+
+ struct list_head list;
+ struct kref ref;
};
struct dlm_node_addr {
@@ -1022,12 +1033,37 @@ static int accept_from_sock(struct listen_connection *con)
return result;
}
-static void free_entry(struct writequeue_entry *e)
+static void dlm_page_release(struct kref *kref)
{
+ struct writequeue_entry *e = container_of(kref, struct writequeue_entry,
+ ref);
+
__free_page(e->page);
kfree(e);
}
+static void dlm_msg_release(struct kref *kref)
+{
+ struct dlm_msg *msg = container_of(kref, struct dlm_msg, ref);
+
+ kref_put(&msg->entry->ref, dlm_page_release);
+ kfree(msg);
+}
+
+static void free_entry(struct writequeue_entry *e)
+{
+ struct dlm_msg *msg, *tmp;
+
+ list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
+ list_del(&msg->list);
+ kref_put(&msg->ref, dlm_msg_release);
+ }
+
+ list_del(&e->list);
+ atomic_dec(&e->con->writequeue_cnt);
+ kref_put(&e->ref, dlm_page_release);
+}
+
/*
* writequeue_entry_complete - try to delete and free write queue entry
* @e: write queue entry to try to delete
@@ -1040,11 +1076,8 @@ static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
e->offset += completed;
e->len -= completed;
- if (e->len == 0 && e->users == 0) {
- list_del(&e->list);
- atomic_dec(&e->con->writequeue_cnt);
+ if (e->len == 0 && e->users == 0)
free_entry(e);
- }
}
/*
@@ -1410,12 +1443,16 @@ static struct writequeue_entry *new_writequeue_entry(struct connection *con,
entry->con = con;
entry->users = 1;
+ kref_init(&entry->ref);
+ INIT_LIST_HEAD(&entry->msgs);
return entry;
}
static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
- gfp_t allocation, char **ppc)
+ gfp_t allocation, char **ppc,
+ void (*cb)(struct dlm_mhandle *mh),
+ struct dlm_mhandle *mh)
{
struct writequeue_entry *e;
@@ -1423,7 +1460,12 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
if (!list_empty(&con->writequeue)) {
e = list_last_entry(&con->writequeue, struct writequeue_entry, list);
if (DLM_WQ_REMAIN_BYTES(e) >= len) {
+ kref_get(&e->ref);
+
*ppc = page_address(e->page) + e->end;
+ if (cb)
+ cb(mh);
+
e->end += len;
e->users++;
spin_unlock(&con->writequeue_lock);
@@ -1437,21 +1479,28 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
if (!e)
return NULL;
+ kref_get(&e->ref);
*ppc = page_address(e->page);
e->end += len;
atomic_inc(&con->writequeue_cnt);
spin_lock(&con->writequeue_lock);
+ if (cb)
+ cb(mh);
+
list_add_tail(&e->list, &con->writequeue);
spin_unlock(&con->writequeue_lock);
return e;
};
-void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
+struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
+ char **ppc, void (*cb)(struct dlm_mhandle *mh),
+ struct dlm_mhandle *mh)
{
struct writequeue_entry *e;
struct connection *con;
+ struct dlm_msg *msg;
int idx;
if (len > DEFAULT_BUFFER_SIZE ||
@@ -1469,25 +1518,41 @@ void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc)
return NULL;
}
- e = new_wq_entry(con, len, allocation, ppc);
+ msg = kzalloc(sizeof(*msg), allocation);
+ if (!msg) {
+ srcu_read_unlock(&connections_srcu, idx);
+ return NULL;
+ }
+
+ kref_init(&msg->ref);
+
+ e = new_wq_entry(con, len, allocation, ppc, cb, mh);
if (!e) {
srcu_read_unlock(&connections_srcu, idx);
+ kfree(msg);
return NULL;
}
+ msg->ppc = *ppc;
+ msg->len = len;
+ msg->entry = e;
+
/* we assume if successful commit must called */
- e->idx = idx;
+ msg->idx = idx;
- return e;
+ return msg;
}
-void dlm_lowcomms_commit_buffer(void *mh)
+void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
{
- struct writequeue_entry *e = (struct writequeue_entry *)mh;
+ struct writequeue_entry *e = msg->entry;
struct connection *con = e->con;
int users;
spin_lock(&con->writequeue_lock);
+ kref_get(&msg->ref);
+ list_add(&msg->list, &e->msgs);
+
users = --e->users;
if (users)
goto out;
@@ -1496,15 +1561,20 @@ void dlm_lowcomms_commit_buffer(void *mh)
spin_unlock(&con->writequeue_lock);
queue_work(send_workqueue, &con->swork);
- srcu_read_unlock(&connections_srcu, e->idx);
+ srcu_read_unlock(&connections_srcu, msg->idx);
return;
out:
spin_unlock(&con->writequeue_lock);
- srcu_read_unlock(&connections_srcu, e->idx);
+ srcu_read_unlock(&connections_srcu, msg->idx);
return;
}
+void dlm_lowcomms_put_msg(struct dlm_msg *msg)
+{
+ kref_put(&msg->ref, dlm_msg_release);
+}
+
/* Send a message */
static void send_to_sock(struct connection *con)
{
@@ -1590,7 +1660,6 @@ static void clean_one_writequeue(struct connection *con)
spin_lock(&con->writequeue_lock);
list_for_each_entry_safe(e, safe, &con->writequeue, list) {
- list_del(&e->list);
free_entry(e);
}
spin_unlock(&con->writequeue_lock);
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index 48bbc4e18761..cdb8f066f0d8 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -22,8 +22,11 @@ void dlm_lowcomms_shutdown(void);
void dlm_lowcomms_stop(void);
void dlm_lowcomms_exit(void);
int dlm_lowcomms_close(int nodeid);
-void *dlm_lowcomms_get_buffer(int nodeid, int len, gfp_t allocation, char **ppc);
-void dlm_lowcomms_commit_buffer(void *mh);
+struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
+ char **ppc, void (*cb)(struct dlm_mhandle *mh),
+ struct dlm_mhandle *mh);
+void dlm_lowcomms_commit_msg(struct dlm_msg *msg);
+void dlm_lowcomms_put_msg(struct dlm_msg *msg);
int dlm_lowcomms_connect_node(int nodeid);
int dlm_lowcomms_nodes_set_mark(int nodeid, unsigned int mark);
int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len);
diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c
index 1a280dda99d3..aadb3781bebe 100644
--- a/fs/dlm/midcomms.c
+++ b/fs/dlm/midcomms.c
@@ -31,12 +31,15 @@
struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
gfp_t allocation, char **ppc)
{
- return dlm_lowcomms_get_buffer(nodeid, len, allocation, ppc);
+ return (struct dlm_mhandle *)dlm_lowcomms_new_msg(nodeid, len,
+ allocation, ppc,
+ NULL, NULL);
}
void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh)
{
- dlm_lowcomms_commit_buffer(mh);
+ dlm_lowcomms_commit_msg((struct dlm_msg *)mh);
+ dlm_lowcomms_put_msg((struct dlm_msg *)mh);
}
void dlm_midcomms_add_member(int nodeid) { }
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index 2661674364af..6f653a339bea 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -70,21 +70,22 @@ static int create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
static int create_rcom_stateless(struct dlm_ls *ls, int to_nodeid, int type,
int len, struct dlm_rcom **rc_ret,
- void **mh_ret)
+ struct dlm_msg **msg_ret)
{
int mb_len = sizeof(struct dlm_rcom) + len;
- void *mh;
+ struct dlm_msg *msg;
char *mb;
- mh = dlm_lowcomms_get_buffer(to_nodeid, mb_len, GFP_NOFS, &mb);
- if (!mh) {
+ msg = dlm_lowcomms_new_msg(to_nodeid, mb_len, GFP_NOFS, &mb,
+ NULL, NULL);
+ if (!msg) {
log_print("create_rcom to %d type %d len %d ENOBUFS",
to_nodeid, type, len);
return -ENOBUFS;
}
_create_rcom(ls, to_nodeid, type, len, rc_ret, mb, mb_len);
- *mh_ret = mh;
+ *msg_ret = msg;
return 0;
}
@@ -100,11 +101,12 @@ static void send_rcom(struct dlm_ls *ls, struct dlm_mhandle *mh,
dlm_midcomms_commit_mhandle(mh);
}
-static void send_rcom_stateless(struct dlm_ls *ls, void *mh,
+static void send_rcom_stateless(struct dlm_ls *ls, struct dlm_msg *msg,
struct dlm_rcom *rc)
{
_send_rcom(ls, rc);
- dlm_lowcomms_commit_buffer(mh);
+ dlm_lowcomms_commit_msg(msg);
+ dlm_lowcomms_put_msg(msg);
}
static void set_rcom_status(struct dlm_ls *ls, struct rcom_status *rs,
@@ -180,8 +182,8 @@ static void disallow_sync_reply(struct dlm_ls *ls)
int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
{
struct dlm_rcom *rc;
+ struct dlm_msg *msg;
int error = 0;
- void *mh;
ls->ls_recover_nodeid = nodeid;
@@ -193,7 +195,7 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
retry:
error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS,
- sizeof(struct rcom_status), &rc, &mh);
+ sizeof(struct rcom_status), &rc, &msg);
if (error)
goto out;
@@ -202,7 +204,7 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
allow_sync_reply(ls, &rc->rc_id);
memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
- send_rcom_stateless(ls, mh, rc);
+ send_rcom_stateless(ls, msg, rc);
error = dlm_wait_function(ls, &rcom_response);
disallow_sync_reply(ls);
@@ -234,9 +236,9 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
uint32_t status;
int nodeid = rc_in->rc_header.h_nodeid;
int len = sizeof(struct rcom_config);
+ struct dlm_msg *msg;
int num_slots = 0;
int error;
- void *mh;
if (!dlm_slots_version(&rc_in->rc_header)) {
status = dlm_recover_status(ls);
@@ -258,7 +260,7 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
do_create:
error = create_rcom_stateless(ls, nodeid, DLM_RCOM_STATUS_REPLY,
- len, &rc, &mh);
+ len, &rc, &msg);
if (error)
return;
@@ -285,7 +287,7 @@ static void receive_rcom_status(struct dlm_ls *ls, struct dlm_rcom *rc_in)
spin_unlock(&ls->ls_recover_lock);
do_send:
- send_rcom_stateless(ls, mh, rc);
+ send_rcom_stateless(ls, msg, rc);
}
static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
@@ -310,14 +312,14 @@ static void receive_sync_reply(struct dlm_ls *ls, struct dlm_rcom *rc_in)
int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
{
struct dlm_rcom *rc;
+ struct dlm_msg *msg;
int error = 0;
- void *mh;
ls->ls_recover_nodeid = nodeid;
retry:
error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES, last_len,
- &rc, &mh);
+ &rc, &msg);
if (error)
goto out;
memcpy(rc->rc_buf, last_name, last_len);
@@ -325,7 +327,7 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
allow_sync_reply(ls, &rc->rc_id);
memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
- send_rcom_stateless(ls, mh, rc);
+ send_rcom_stateless(ls, msg, rc);
error = dlm_wait_function(ls, &rcom_response);
disallow_sync_reply(ls);
@@ -339,14 +341,14 @@ static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
{
struct dlm_rcom *rc;
int error, inlen, outlen, nodeid;
- void *mh;
+ struct dlm_msg *msg;
nodeid = rc_in->rc_header.h_nodeid;
inlen = rc_in->rc_header.h_length - sizeof(struct dlm_rcom);
outlen = LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom);
error = create_rcom_stateless(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen,
- &rc, &mh);
+ &rc, &msg);
if (error)
return;
rc->rc_id = rc_in->rc_id;
@@ -354,7 +356,7 @@ static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
dlm_copy_master_names(ls, rc_in->rc_buf, inlen, rc->rc_buf, outlen,
nodeid);
- send_rcom_stateless(ls, mh, rc);
+ send_rcom_stateless(ls, msg, rc);
}
int dlm_send_rcom_lookup(struct dlm_rsb *r, int dir_nodeid)
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 11/16] fs: dlm: add functionality to re-transmit a message
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (9 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 10/16] fs: dlm: make buffer handling per msg Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 12/16] fs: dlm: move out some hash functionality Alexander Aring
` (4 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch introduces a retransmit functionality for a lowcomms message
handle. It's just allocates a new buffer and transmit it again, no
special handling about prioritize it because keeping bytestream in order.
To avoid another connection look some refactor was done to make a new
buffer allocation with a preexisting connection pointer.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 85 ++++++++++++++++++++++++++++++++++++-----------
fs/dlm/lowcomms.h | 1 +
2 files changed, 67 insertions(+), 19 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index d222e6088ab2..df9827ec12f3 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -126,6 +126,8 @@ struct writequeue_entry {
struct dlm_msg {
struct writequeue_entry *entry;
+ struct dlm_msg *orig_msg;
+ bool retransmit;
void *ppc;
int len;
int idx; /* new()/commit() idx exchange */
@@ -1055,6 +1057,10 @@ static void free_entry(struct writequeue_entry *e)
struct dlm_msg *msg, *tmp;
list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
+ if (msg->orig_msg) {
+ msg->orig_msg->retransmit = false;
+ kref_put(&msg->orig_msg->ref, dlm_msg_release);
+ }
list_del(&msg->list);
kref_put(&msg->ref, dlm_msg_release);
}
@@ -1494,11 +1500,37 @@ static struct writequeue_entry *new_wq_entry(struct connection *con, int len,
return e;
};
+static struct dlm_msg *dlm_lowcomms_new_msg_con(struct connection *con, int len,
+ gfp_t allocation, char **ppc,
+ void (*cb)(struct dlm_mhandle *mh),
+ struct dlm_mhandle *mh)
+{
+ struct writequeue_entry *e;
+ struct dlm_msg *msg;
+
+ msg = kzalloc(sizeof(*msg), allocation);
+ if (!msg)
+ return NULL;
+
+ kref_init(&msg->ref);
+
+ e = new_wq_entry(con, len, allocation, ppc, cb, mh);
+ if (!e) {
+ kfree(msg);
+ return NULL;
+ }
+
+ msg->ppc = *ppc;
+ msg->len = len;
+ msg->entry = e;
+
+ return msg;
+}
+
struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
char **ppc, void (*cb)(struct dlm_mhandle *mh),
struct dlm_mhandle *mh)
{
- struct writequeue_entry *e;
struct connection *con;
struct dlm_msg *msg;
int idx;
@@ -1518,32 +1550,18 @@ struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
return NULL;
}
- msg = kzalloc(sizeof(*msg), allocation);
+ msg = dlm_lowcomms_new_msg_con(con, len, allocation, ppc, cb, mh);
if (!msg) {
srcu_read_unlock(&connections_srcu, idx);
return NULL;
}
- kref_init(&msg->ref);
-
- e = new_wq_entry(con, len, allocation, ppc, cb, mh);
- if (!e) {
- srcu_read_unlock(&connections_srcu, idx);
- kfree(msg);
- return NULL;
- }
-
- msg->ppc = *ppc;
- msg->len = len;
- msg->entry = e;
-
/* we assume if successful commit must called */
msg->idx = idx;
-
return msg;
}
-void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
+static void _dlm_lowcomms_commit_msg(struct dlm_msg *msg)
{
struct writequeue_entry *e = msg->entry;
struct connection *con = e->con;
@@ -1561,20 +1579,49 @@ void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
spin_unlock(&con->writequeue_lock);
queue_work(send_workqueue, &con->swork);
- srcu_read_unlock(&connections_srcu, msg->idx);
return;
out:
spin_unlock(&con->writequeue_lock);
- srcu_read_unlock(&connections_srcu, msg->idx);
return;
}
+void dlm_lowcomms_commit_msg(struct dlm_msg *msg)
+{
+ _dlm_lowcomms_commit_msg(msg);
+ srcu_read_unlock(&connections_srcu, msg->idx);
+}
+
void dlm_lowcomms_put_msg(struct dlm_msg *msg)
{
kref_put(&msg->ref, dlm_msg_release);
}
+/* does not held connections_srcu, usage workqueue only */
+int dlm_lowcomms_resend_msg(struct dlm_msg *msg)
+{
+ struct dlm_msg *msg_resend;
+ char *ppc;
+
+ if (msg->retransmit)
+ return 1;
+
+ msg_resend = dlm_lowcomms_new_msg_con(msg->entry->con, msg->len,
+ GFP_ATOMIC, &ppc, NULL, NULL);
+ if (!msg_resend)
+ return -ENOMEM;
+
+ msg->retransmit = true;
+ kref_get(&msg->ref);
+ msg_resend->orig_msg = msg;
+
+ memcpy(ppc, msg->ppc, msg->len);
+ _dlm_lowcomms_commit_msg(msg_resend);
+ dlm_lowcomms_put_msg(msg_resend);
+
+ return 0;
+}
+
/* Send a message */
static void send_to_sock(struct connection *con)
{
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index cdb8f066f0d8..a4384826442c 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -27,6 +27,7 @@ struct dlm_msg *dlm_lowcomms_new_msg(int nodeid, int len, gfp_t allocation,
struct dlm_mhandle *mh);
void dlm_lowcomms_commit_msg(struct dlm_msg *msg);
void dlm_lowcomms_put_msg(struct dlm_msg *msg);
+int dlm_lowcomms_resend_msg(struct dlm_msg *msg);
int dlm_lowcomms_connect_node(int nodeid);
int dlm_lowcomms_nodes_set_mark(int nodeid, unsigned int mark);
int dlm_lowcomms_addr(int nodeid, struct sockaddr_storage *addr, int len);
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 12/16] fs: dlm: move out some hash functionality
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (10 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 11/16] fs: dlm: add functionality to re-transmit a message Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 13/16] fs: dlm: add union in dlm header for lockspace id Alexander Aring
` (3 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch moves out some lowcomms hash functionality into lowcomms
header to provide them to other layers like midcomms as well.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 9 ---------
fs/dlm/lowcomms.h | 10 ++++++++++
2 files changed, 10 insertions(+), 9 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index df9827ec12f3..1f2759cfda09 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -59,7 +59,6 @@
#include "config.h"
#define NEEDED_RMEM (4*1024*1024)
-#define CONN_HASH_SIZE 32
/* Number of messages to send before rescheduling */
#define MAX_SEND_MSG_COUNT 25
@@ -175,14 +174,6 @@ static void sctp_connect_to_sock(struct connection *con);
static void tcp_connect_to_sock(struct connection *con);
static void dlm_tcp_shutdown(struct connection *con);
-/* This is deliberately very simple because most clusters have simple
- sequential nodeids, so we should be able to go straight to a connection
- struct in the array */
-static inline int nodeid_hash(int nodeid)
-{
- return nodeid & (CONN_HASH_SIZE-1);
-}
-
static struct connection *__find_con(int nodeid, int r)
{
struct connection *con;
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index a4384826442c..66dc1bb3de7f 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -13,6 +13,16 @@
#define __LOWCOMMS_DOT_H__
#define LOWCOMMS_MAX_TX_BUFFER_LEN 4096
+#define CONN_HASH_SIZE 32
+
+/* This is deliberately very simple because most clusters have simple
+ * sequential nodeids, so we should be able to go straight to a connection
+ * struct in the array
+ */
+static inline int nodeid_hash(int nodeid)
+{
+ return nodeid & (CONN_HASH_SIZE-1);
+}
/* switch to check if dlm is running */
extern int dlm_allow_conn;
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 13/16] fs: dlm: add union in dlm header for lockspace id
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (11 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 12/16] fs: dlm: move out some hash functionality Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 14/16] fs: dlm: add reliable connection if reconnect Alexander Aring
` (2 subsequent siblings)
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch adds union inside the lockspace id to handle it also for
another use case for a different dlm command.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/dlm_internal.h | 5 ++++-
fs/dlm/lock.c | 8 ++++----
fs/dlm/rcom.c | 4 ++--
fs/dlm/util.c | 6 ++++--
4 files changed, 14 insertions(+), 9 deletions(-)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index e8dc5f4f1f9e..8f5980909d80 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -380,7 +380,10 @@ static inline int rsb_flag(struct dlm_rsb *r, enum rsb_flags flag)
struct dlm_header {
uint32_t h_version;
- uint32_t h_lockspace;
+ union {
+ /* for DLM_MSG and DLM_RCOM */
+ uint32_t h_lockspace;
+ } u;
uint32_t h_nodeid; /* nodeid of sender */
uint16_t h_length;
uint8_t h_cmd; /* DLM_MSG, DLM_RCOM */
diff --git a/fs/dlm/lock.c b/fs/dlm/lock.c
index b625ce92464a..c502c065d007 100644
--- a/fs/dlm/lock.c
+++ b/fs/dlm/lock.c
@@ -3544,7 +3544,7 @@ static int _create_message(struct dlm_ls *ls, int mb_len,
ms = (struct dlm_message *) mb;
ms->m_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
- ms->m_header.h_lockspace = ls->ls_global_id;
+ ms->m_header.u.h_lockspace = ls->ls_global_id;
ms->m_header.h_nodeid = dlm_our_nodeid();
ms->m_header.h_length = mb_len;
ms->m_header.h_cmd = DLM_MSG;
@@ -5038,16 +5038,16 @@ void dlm_receive_buffer(union dlm_packet *p, int nodeid)
if (hd->h_nodeid != nodeid) {
log_print("invalid h_nodeid %d from %d lockspace %x",
- hd->h_nodeid, nodeid, hd->h_lockspace);
+ hd->h_nodeid, nodeid, hd->u.h_lockspace);
return;
}
- ls = dlm_find_lockspace_global(hd->h_lockspace);
+ ls = dlm_find_lockspace_global(hd->u.h_lockspace);
if (!ls) {
if (dlm_config.ci_log_debug) {
printk_ratelimited(KERN_DEBUG "dlm: invalid lockspace "
"%u from %d cmd %d type %d\n",
- hd->h_lockspace, nodeid, hd->h_cmd, type);
+ hd->u.h_lockspace, nodeid, hd->h_cmd, type);
}
if (hd->h_cmd == DLM_RCOM && type == DLM_RCOM_STATUS)
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index 6f653a339bea..7c1a06bcd418 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -35,7 +35,7 @@ static void _create_rcom(struct dlm_ls *ls, int to_nodeid, int type, int len,
rc = (struct dlm_rcom *) mb;
rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
- rc->rc_header.h_lockspace = ls->ls_global_id;
+ rc->rc_header.u.h_lockspace = ls->ls_global_id;
rc->rc_header.h_nodeid = dlm_our_nodeid();
rc->rc_header.h_length = mb_len;
rc->rc_header.h_cmd = DLM_RCOM;
@@ -508,7 +508,7 @@ int dlm_send_ls_not_ready(int nodeid, struct dlm_rcom *rc_in)
rc = (struct dlm_rcom *) mb;
rc->rc_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
- rc->rc_header.h_lockspace = rc_in->rc_header.h_lockspace;
+ rc->rc_header.u.h_lockspace = rc_in->rc_header.u.h_lockspace;
rc->rc_header.h_nodeid = dlm_our_nodeid();
rc->rc_header.h_length = mb_len;
rc->rc_header.h_cmd = DLM_RCOM;
diff --git a/fs/dlm/util.c b/fs/dlm/util.c
index 74a8c5bfe9b5..58acbcc2081a 100644
--- a/fs/dlm/util.c
+++ b/fs/dlm/util.c
@@ -23,7 +23,8 @@
void header_out(struct dlm_header *hd)
{
hd->h_version = cpu_to_le32(hd->h_version);
- hd->h_lockspace = cpu_to_le32(hd->h_lockspace);
+ /* does it for others u32 in union as well */
+ hd->u.h_lockspace = cpu_to_le32(hd->u.h_lockspace);
hd->h_nodeid = cpu_to_le32(hd->h_nodeid);
hd->h_length = cpu_to_le16(hd->h_length);
}
@@ -31,7 +32,8 @@ void header_out(struct dlm_header *hd)
void header_in(struct dlm_header *hd)
{
hd->h_version = le32_to_cpu(hd->h_version);
- hd->h_lockspace = le32_to_cpu(hd->h_lockspace);
+ /* does it for others u32 in union as well */
+ hd->u.h_lockspace = le32_to_cpu(hd->u.h_lockspace);
hd->h_nodeid = le32_to_cpu(hd->h_nodeid);
hd->h_length = le16_to_cpu(hd->h_length);
}
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 14/16] fs: dlm: add reliable connection if reconnect
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (12 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 13/16] fs: dlm: add union in dlm header for lockspace id Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 15/16] fs: dlm: add midcomms debugfs functionality Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 16/16] fs: dlm: don't allow half transmitted messages Alexander Aring
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch introduce to make a tcp lowcomms connection reliable even if
reconnects occurs. This is done by an application layer re-transmission
handling and sequence numbers in dlm protocols. There are three new dlm
commands:
DLM_OPTS:
This will encapsulate an existing dlm message (and rcom message if they
don't have an own application side re-transmission handling). As optional
handling additional tlv's (type length fields) can be appended. This can
be for example a sequence number field. However because in DLM_OPTS the
lockspace field is unused and a sequence number is a mandatory field it
isn't made as a tlv and we put the sequence number inside the lockspace
id. The possibility to add optional options are still there for future
purposes.
DLM_ACK:
Just a dlm header to acknowledge the receive of a DLM_OPTS message to
it's sender.
DLM_FIN:
This provides a 4 way handshake for connection termination inclusive
support for half-closed connections. It's provided on application layer
because SCTP doesn't support half-closed sockets, the shutdown() call
can interrupted by e.g. TCP resets itself and a hard logic to implement
it because the othercon paradigm in lowcomms. The 4-way termination
handshake also solve problems to synchronize peer EOF arrival and that
the cluster manager removes the peer in the node membership handling of
DLM. In some cases messages can be still transmitted in this time and we
need to wait for the node membership event.
To provide a reliable connection the node will retransmit all
unacknowledges message to it's peer on reconnect. The receiver will then
filtering out the next received message and drop all messages which are
duplicates.
As RCOM_STATUS and RCOM_NAMES messages are the first messages which are
exchanged and they have they own re-transmission handling, there exists
logic that these messages must be first. If these messages arrives we
store the dlm version field. This handling is on DLM 3.1 and after this
patch 3.2 the same. A backwards compatibility handling has been added
which seems to work on tests without tcpkill, however it's not recommended
to use DLM 3.1 and 3.2 at the same time, because DLM 3.2 tries to fix long
term bugs in the DLM protocol.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/dlm_internal.h | 30 +-
fs/dlm/lockspace.c | 7 +-
fs/dlm/lowcomms.c | 4 +-
fs/dlm/lowcomms.h | 7 +-
fs/dlm/midcomms.c | 1290 +++++++++++++++++++++++++++++++++++++++--
fs/dlm/midcomms.h | 1 +
fs/dlm/rcom.c | 4 +-
7 files changed, 1292 insertions(+), 51 deletions(-)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 8f5980909d80..40917c878370 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -371,18 +371,26 @@ static inline int rsb_flag(struct dlm_rsb *r, enum rsb_flags flag)
/* dlm_header is first element of all structs sent between nodes */
#define DLM_HEADER_MAJOR 0x00030000
-#define DLM_HEADER_MINOR 0x00000001
+#define DLM_HEADER_MINOR 0x00000002
+
+#define DLM_VERSION_3_1 0x00030001
+#define DLM_VERSION_3_2 0x00030002
#define DLM_HEADER_SLOTS 0x00000001
#define DLM_MSG 1
#define DLM_RCOM 2
+#define DLM_OPTS 3
+#define DLM_ACK 4
+#define DLM_FIN 5
struct dlm_header {
uint32_t h_version;
union {
/* for DLM_MSG and DLM_RCOM */
uint32_t h_lockspace;
+ /* for DLM_ACK and DLM_OPTS */
+ uint32_t h_seq;
} u;
uint32_t h_nodeid; /* nodeid of sender */
uint16_t h_length;
@@ -390,7 +398,6 @@ struct dlm_header {
uint8_t h_pad;
};
-
#define DLM_MSG_REQUEST 1
#define DLM_MSG_CONVERT 2
#define DLM_MSG_UNLOCK 3
@@ -458,10 +465,29 @@ struct dlm_rcom {
char rc_buf[];
};
+struct dlm_opt_header {
+ uint16_t t_type;
+ uint16_t t_length;
+ uint32_t o_pad;
+ /* need to be 8 byte aligned */
+ char t_value[];
+};
+
+/* encapsulation header */
+struct dlm_opts {
+ struct dlm_header o_header;
+ uint8_t o_nextcmd;
+ uint8_t o_pad;
+ uint16_t o_optlen;
+ uint32_t o_pad2;
+ char o_opts[];
+};
+
union dlm_packet {
struct dlm_header header; /* common to other two */
struct dlm_message message;
struct dlm_rcom rcom;
+ struct dlm_opts opts;
};
#define DLM_RSF_NEED_SLOTS 0x00000001
diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index bf5c55ef9d0d..2b738be8d7e4 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -567,7 +567,12 @@ static int new_lockspace(const char *name, const char *cluster,
mutex_init(&ls->ls_requestqueue_mutex);
mutex_init(&ls->ls_clear_proc_locks);
- ls->ls_recover_buf = kmalloc(LOWCOMMS_MAX_TX_BUFFER_LEN, GFP_NOFS);
+ /* Due backwards compatibility with 3.1 we need to use maximum
+ * possible dlm message size to be sure the message will fit and
+ * not having out of bounds issues. However on sending side 3.2
+ * might send less.
+ */
+ ls->ls_recover_buf = kmalloc(DEFAULT_BUFFER_SIZE, GFP_NOFS);
if (!ls->ls_recover_buf)
goto out_lkbidr;
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index 1f2759cfda09..fe9113bd5ba0 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -1762,8 +1762,10 @@ static void process_send_sockets(struct work_struct *work)
clear_bit(CF_WRITE_PENDING, &con->flags);
- if (test_and_clear_bit(CF_RECONNECT, &con->flags))
+ if (test_and_clear_bit(CF_RECONNECT, &con->flags)) {
close_connection(con, false, false, true);
+ dlm_midcomms_unack_msg_resend(con->nodeid);
+ }
if (con->sock == NULL) { /* not mutex protected so check it inside too */
if (test_and_clear_bit(CF_DELAY_CONNECT, &con->flags))
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index 66dc1bb3de7f..730c34317183 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -12,7 +12,12 @@
#ifndef __LOWCOMMS_DOT_H__
#define __LOWCOMMS_DOT_H__
-#define LOWCOMMS_MAX_TX_BUFFER_LEN 4096
+#include "dlm_internal.h"
+
+#define DLM_MIDCOMMS_OPT_LEN sizeof(struct dlm_opts)
+#define LOWCOMMS_MAX_TX_BUFFER_LEN (DEFAULT_BUFFER_SIZE - \
+ DLM_MIDCOMMS_OPT_LEN)
+
#define CONN_HASH_SIZE 32
/* This is deliberately very simple because most clusters have simple
diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c
index aadb3781bebe..eef3938a363e 100644
--- a/fs/dlm/midcomms.c
+++ b/fs/dlm/midcomms.c
@@ -3,7 +3,7 @@
*******************************************************************************
**
** Copyright (C) Sistina Software, Inc. 1997-2003 All rights reserved.
-** Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
+** Copyright (C) 2004-2021 Red Hat, Inc. All rights reserved.
**
**
*******************************************************************************
@@ -12,53 +12,817 @@
/*
* midcomms.c
*
- * This is the appallingly named "mid-level" comms layer.
+ * This is the appallingly named "mid-level" comms layer. It takes care about
+ * deliver an on application layer "reliable" communication above the used
+ * lowcomms transport layer.
*
- * Its purpose is to take packets from the "real" comms layer,
- * split them up into packets and pass them to the interested
- * part of the locking mechanism.
+ * How it works:
*
- * It also takes messages from the locking layer, formats them
- * into packets and sends them to the comms layer.
+ * Each nodes keeps track of all send DLM messages in send_queue with a sequence
+ * number. The receive will send an DLM_ACK message back for every DLM message
+ * received at the other side. If a reconnect happens in lowcomms we will send
+ * all unacknowledged dlm messages again. The receiving side might drop any already
+ * received message by comparing sequence numbers.
+ *
+ * How version detection works:
+ *
+ * Due the fact that dlm has pre-configured node addresses on every side
+ * it is in it's nature that every side connects at starts to transmit
+ * dlm messages which ends in a race. However DLM_RCOM_NAMES, DLM_RCOM_STATUS
+ * and their replies are the first messages which are exchanges. Due backwards
+ * compatibility these messages are not covered by the midcomms re-transmission
+ * layer. These messages have their own re-transmission handling in the dlm
+ * application layer. The version field of every node will be set on these RCOM
+ * messages as soon as they arrived and the node isn't yet part of the nodes
+ * hash. There exists also logic to detect version mismatched if something weird
+ * going on or the first messages isn't an expected one.
+ *
+ * Termination:
+ *
+ * The midcomms layer does a 4 way handshake for termination on DLM protocol
+ * like TCP supports it with half-closed socket support. SCTP doesn't support
+ * half-closed socket, so we do it on DLM layer. Also socket shutdown() can be
+ * interrupted by .e.g. tcp reset itself. Additional there exists the othercon
+ * paradigm in lowcomms which cannot be easily without breaking backwards
+ * compatibility. A node cannot send anything to another node when a DLM_FIN
+ * message was send. There exists additional logic to print a warning if
+ * DLM wants to do it. There exists a state handling like RFC 793 but reduced
+ * to termination only. The event "member removal event" describes the cluster
+ * manager removed the node from internal lists, at this point DLM does not
+ * send any message to the other node. There exists two cases:
+ *
+ * 1. The cluster member was removed and we received a FIN
+ * OR
+ * 2. We received a FIN but the member was not removed yet
+ *
+ * One of these cases will do the CLOSE_WAIT to LAST_ACK change.
+ *
+ *
+ * +---------+
+ * | CLOSED |
+ * +---------+
+ * | add member/receive RCOM version
+ * | detection msg
+ * V
+ * +---------+
+ * | ESTAB |
+ * +---------+
+ * CLOSE | | rcv FIN
+ * ------- | | -------
+ * +---------+ snd FIN / \ snd ACK +---------+
+ * | FIN |<----------------- ------------------>| CLOSE |
+ * | WAIT-1 |------------------ | WAIT |
+ * +---------+ rcv FIN \ +---------+
+ * | rcv ACK of FIN ------- | CLOSE | member
+ * | -------------- snd ACK | ------- | removal
+ * V x V snd FIN V event
+ * +---------+ +---------+ +---------+
+ * |FINWAIT-2| | CLOSING | | LAST-ACK|
+ * +---------+ +---------+ +---------+
+ * | rcv ACK of FIN | rcv ACK of FIN |
+ * | rcv FIN -------------- | -------------- |
+ * | ------- x V x V
+ * \ snd ACK +---------+ +---------+
+ * ------------------------>| CLOSED | | CLOSED |
+ * +---------+ +---------+
+ *
+ * NOTE: any state can interrupted by midcomms_close() and state will be
+ * switched to CLOSED in case of fencing. There exists also some timeout
+ * handling when we receive the version detection RCOM messages which is
+ * made by observation.
+ *
+ * Future improvements:
+ *
+ * There exists some known issues/improvements of the dlm handling. Some
+ * of them should be done in a next major dlm version bump which makes
+ * it incompatible with previous versions.
+ *
+ * Unaligned memory access:
+ *
+ * There exists cases when the dlm message buffer length is not aligned
+ * to 8 byte. However seems nobody detected any problem with it. This
+ * can be fixed in the next major version bump of dlm.
+ *
+ * Version detection:
+ *
+ * The version detection and how it's done is related to backwards
+ * compatibility. There exists better ways to make a better handling.
+ * However this should be changed in the next major version bump of dlm.
+ *
+ * Ack handling:
+ *
+ * Currently we send an ack message for every dlm message. However we
+ * can ack multiple dlm messages with one ack by just delaying the ack
+ * message. Will reduce some traffic but makes the drop detection slower.
+ *
+ * Tail Size checking:
+ *
+ * There exists a message tail payload in e.g. DLM_MSG however we don't
+ * check it against the message length yet regarding to the receive buffer
+ * length. That need to be validated.
+ *
+ * Fencing bad nodes:
+ *
+ * At timeout places or weird sequence number behaviours we should send
+ * a fencing request to the cluster manager.
*/
+/* Debug switch to enable a 5 seconds sleep waiting of a termination.
+ * This can be useful to test fencing while termination is running.
+ * This requires a setup with only gfs2 as dlm user, so that the
+ * last umount will terminate the connection.
+ *
+ * However it became useful to test, while the 5 seconds block in umount
+ * just press the reset button. In a lot of dropping the termination
+ * process can could take several seconds.
+ */
+#define DLM_DEBUG_FENCE_TERMINATION 0
+
+#include <net/tcp.h>
+
#include "dlm_internal.h"
#include "lowcomms.h"
#include "config.h"
#include "lock.h"
+#include "util.h"
#include "midcomms.h"
-struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
- gfp_t allocation, char **ppc)
+/* init value for sequence numbers for testing purpose only e.g. overflows */
+#define DLM_SEQ_INIT 0
+/* 3 minutes wait to sync ending of dlm */
+#define DLM_SHUTDOWN_TIMEOUT msecs_to_jiffies(3 * 60 * 1000)
+#define DLM_VERSION_NOT_SET 0
+
+struct midcomms_node {
+ int nodeid;
+ uint32_t version;
+ uint32_t seq_send;
+ uint32_t seq_next;
+ /* These queues are unbound because we cannot drop any message in dlm.
+ * We could send a fence signal for a specific node to the cluster
+ * manager if queues hits some maximum value, however this handling
+ * not supported yet.
+ */
+ struct list_head send_queue;
+ spinlock_t send_queue_lock;
+ atomic_t send_queue_cnt;
+#define DLM_NODE_FLAG_CLOSE 1
+#define DLM_NODE_FLAG_STOP_TX 2
+#define DLM_NODE_FLAG_STOP_RX 3
+ unsigned long flags;
+ wait_queue_head_t shutdown_wait;
+
+ /* dlm tcp termination state */
+#define DLM_CLOSED 1
+#define DLM_ESTABLISHED 2
+#define DLM_FIN_WAIT1 3
+#define DLM_FIN_WAIT2 4
+#define DLM_CLOSE_WAIT 5
+#define DLM_LAST_ACK 6
+#define DLM_CLOSING 7
+ int state;
+ spinlock_t state_lock;
+
+ /* counts how many lockspaces are using this node
+ * this refcount is necessary to determine if the
+ * node wants to disconnect.
+ */
+ int users;
+
+ struct hlist_node hlist;
+ struct rcu_head rcu;
+};
+
+struct dlm_mhandle {
+ const struct dlm_header *inner_hd;
+ struct midcomms_node *node;
+ struct dlm_opts *opts;
+ struct dlm_msg *msg;
+ bool committed;
+ uint32_t seq;
+
+ void (*ack_rcv)(struct midcomms_node *node);
+
+ /* get_mhandle/commit srcu idx exchange */
+ int idx;
+
+ struct list_head list;
+ struct rcu_head rcu;
+};
+
+static struct hlist_head node_hash[CONN_HASH_SIZE];
+static DEFINE_SPINLOCK(nodes_lock);
+DEFINE_STATIC_SRCU(nodes_srcu);
+
+/* This mutex prevents that midcomms_close() is running while
+ * stop() or remove(). As I experienced invalid memory access
+ * behaviours when DLM_DEBUG_FENCE_TERMINATION is enabled and
+ * resetting machines. I will end in some double deletion in nodes
+ * datastructure.
+ */
+static DEFINE_MUTEX(close_lock);
+
+static inline const char *dlm_state_str(int state)
{
- return (struct dlm_mhandle *)dlm_lowcomms_new_msg(nodeid, len,
- allocation, ppc,
- NULL, NULL);
+ switch (state) {
+ case DLM_CLOSED:
+ return "CLOSED";
+ case DLM_ESTABLISHED:
+ return "ESTABLISHED";
+ case DLM_FIN_WAIT1:
+ return "FIN_WAIT1";
+ case DLM_FIN_WAIT2:
+ return "FIN_WAIT2";
+ case DLM_CLOSE_WAIT:
+ return "CLOSE_WAIT";
+ case DLM_LAST_ACK:
+ return "LAST_ACK";
+ case DLM_CLOSING:
+ return "CLOSING";
+ default:
+ return "UNKNOWN";
+ }
}
-void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh)
+static struct midcomms_node *__find_node(int nodeid, int r)
{
- dlm_lowcomms_commit_msg((struct dlm_msg *)mh);
- dlm_lowcomms_put_msg((struct dlm_msg *)mh);
+ struct midcomms_node *node;
+
+ hlist_for_each_entry_rcu(node, &node_hash[r], hlist) {
+ if (node->nodeid == nodeid)
+ return node;
+ }
+
+ return NULL;
}
-void dlm_midcomms_add_member(int nodeid) { }
+static void dlm_mhandle_release(struct rcu_head *rcu)
+{
+ struct dlm_mhandle *mh = container_of(rcu, struct dlm_mhandle, rcu);
-void dlm_midcomms_remove_member(int nodeid) { }
+ dlm_lowcomms_put_msg(mh->msg);
+ kfree(mh);
+}
-int dlm_midcomms_start(void)
+static void dlm_send_queue_flush(struct midcomms_node *node)
{
- return dlm_lowcomms_start();
+ struct dlm_mhandle *mh;
+
+ pr_debug("flush midcomms send queue of node %d\n", node->nodeid);
+
+ rcu_read_lock();
+ list_for_each_entry_rcu(mh, &node->send_queue, list) {
+ spin_lock(&node->send_queue_lock);
+ list_del_rcu(&mh->list);
+ spin_unlock(&node->send_queue_lock);
+
+ atomic_dec(&node->send_queue_cnt);
+
+ call_rcu(&mh->rcu, dlm_mhandle_release);
+ }
+ rcu_read_unlock();
}
-void dlm_midcomms_shutdown(void)
+static void midcomms_node_reset(struct midcomms_node *node)
{
- dlm_lowcomms_shutdown();
+ pr_debug("reset node %d\n", node->nodeid);
+
+ node->seq_next = DLM_SEQ_INIT;
+ node->seq_send = DLM_SEQ_INIT;
+ node->version = DLM_VERSION_NOT_SET;
+ node->flags = 0;
+
+ dlm_send_queue_flush(node);
+ node->state = DLM_CLOSED;
+ wake_up(&node->shutdown_wait);
}
-int dlm_midcomms_close(int nodeid)
+static struct midcomms_node *nodeid2node(int nodeid, gfp_t alloc)
+{
+ struct midcomms_node *node, *tmp;
+ int r = nodeid_hash(nodeid);
+
+ node = __find_node(nodeid, r);
+ if (node || !alloc)
+ return node;
+
+ node = kmalloc(sizeof(*node), alloc);
+ if (!node)
+ return NULL;
+
+ node->nodeid = nodeid;
+ spin_lock_init(&node->state_lock);
+ spin_lock_init(&node->send_queue_lock);
+ atomic_set(&node->send_queue_cnt, 0);
+ INIT_LIST_HEAD(&node->send_queue);
+ init_waitqueue_head(&node->shutdown_wait);
+ node->users = 0;
+ midcomms_node_reset(node);
+
+ spin_lock(&nodes_lock);
+ /* check again if there was somebody else
+ * earlier here to add the node
+ */
+ tmp = __find_node(nodeid, r);
+ if (tmp) {
+ spin_unlock(&nodes_lock);
+ kfree(node);
+ return tmp;
+ }
+
+ hlist_add_head_rcu(&node->hlist, &node_hash[r]);
+ spin_unlock(&nodes_lock);
+ return node;
+}
+
+static int dlm_send_ack(int nodeid, uint32_t seq)
+{
+ int mb_len = sizeof(struct dlm_header);
+ struct dlm_header *m_header;
+ struct dlm_msg *msg;
+ char *ppc;
+
+ msg = dlm_lowcomms_new_msg(nodeid, mb_len, GFP_NOFS, &ppc,
+ NULL, NULL);
+ if (!msg)
+ return -ENOMEM;
+
+ m_header = (struct dlm_header *)ppc;
+
+ m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+ m_header->h_nodeid = dlm_our_nodeid();
+ m_header->h_length = mb_len;
+ m_header->h_cmd = DLM_ACK;
+ m_header->u.h_seq = seq;
+
+ header_out(m_header);
+ dlm_lowcomms_commit_msg(msg);
+ dlm_lowcomms_put_msg(msg);
+
+ return 0;
+}
+
+static int dlm_send_fin(struct midcomms_node *node,
+ void (*ack_rcv)(struct midcomms_node *node))
+{
+ int mb_len = sizeof(struct dlm_header);
+ struct dlm_header *m_header;
+ struct dlm_mhandle *mh;
+ char *ppc;
+
+ mh = dlm_midcomms_get_mhandle(node->nodeid, mb_len, GFP_NOFS, &ppc);
+ if (!mh)
+ return -ENOMEM;
+
+ mh->ack_rcv = ack_rcv;
+
+ m_header = (struct dlm_header *)ppc;
+
+ m_header->h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+ m_header->h_nodeid = dlm_our_nodeid();
+ m_header->h_length = mb_len;
+ m_header->h_cmd = DLM_FIN;
+
+ header_out(m_header);
+
+ pr_debug("sending fin msg to node %d\n", node->nodeid);
+ dlm_midcomms_commit_mhandle(mh);
+ set_bit(DLM_NODE_FLAG_STOP_TX, &node->flags);
+
+ return 0;
+}
+
+static void dlm_receive_ack(struct midcomms_node *node, uint32_t seq)
+{
+ struct dlm_mhandle *mh;
+
+ rcu_read_lock();
+ list_for_each_entry_rcu(mh, &node->send_queue, list) {
+ if (before(mh->seq, seq)) {
+ spin_lock(&node->send_queue_lock);
+ list_del_rcu(&mh->list);
+ spin_unlock(&node->send_queue_lock);
+
+ atomic_dec(&node->send_queue_cnt);
+
+ if (mh->ack_rcv)
+ mh->ack_rcv(node);
+
+ call_rcu(&mh->rcu, dlm_mhandle_release);
+ } else {
+ /* send queue should be ordered */
+ break;
+ }
+ }
+ rcu_read_unlock();
+}
+
+static void dlm_pas_fin_ack_rcv(struct midcomms_node *node)
+{
+ spin_lock(&node->state_lock);
+ pr_debug("receive passive fin ack from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+
+ switch (node->state) {
+ case DLM_LAST_ACK:
+ /* DLM_CLOSED */
+ midcomms_node_reset(node);
+ break;
+ case DLM_CLOSED:
+ /* not valid but somehow we got what we want */
+ wake_up(&node->shutdown_wait);
+ break;
+ default:
+ spin_unlock(&node->state_lock);
+ log_print("%s: unexpected state: %d\n",
+ __func__, node->state);
+ WARN_ON(1);
+ return;
+ }
+ spin_unlock(&node->state_lock);
+}
+
+static void dlm_midcomms_receive_buffer(union dlm_packet *p,
+ struct midcomms_node *node,
+ uint32_t seq)
+{
+ if (seq == node->seq_next) {
+ node->seq_next++;
+ /* send ack before fin */
+ dlm_send_ack(node->nodeid, node->seq_next);
+
+ switch (p->header.h_cmd) {
+ case DLM_FIN:
+ spin_lock(&node->state_lock);
+ pr_debug("receive fin msg from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+
+ switch (node->state) {
+ case DLM_ESTABLISHED:
+ node->state = DLM_CLOSE_WAIT;
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ /* passive shutdown DLM_LAST_ACK case 1
+ * additional we check if the node is used by
+ * cluster manager events at all.
+ */
+ if (node->users == 0) {
+ node->state = DLM_LAST_ACK;
+ pr_debug("switch node %d to state %s case 1\n",
+ node->nodeid, dlm_state_str(node->state));
+ spin_unlock(&node->state_lock);
+ goto send_fin;
+ }
+ break;
+ case DLM_FIN_WAIT1:
+ node->state = DLM_CLOSING;
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ break;
+ case DLM_FIN_WAIT2:
+ midcomms_node_reset(node);
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ wake_up(&node->shutdown_wait);
+ break;
+ case DLM_LAST_ACK:
+ /* probably remove_member caught it, do nothing */
+ break;
+ default:
+ spin_unlock(&node->state_lock);
+ log_print("%s: unexpected state: %d\n",
+ __func__, node->state);
+ WARN_ON(1);
+ return;
+ }
+ spin_unlock(&node->state_lock);
+
+ set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+ break;
+ default:
+ WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags));
+ dlm_receive_buffer(p, node->nodeid);
+ break;
+ }
+ } else {
+ /* retry to ack message which we already have by sending back
+ * current node->seq_next number as ack.
+ */
+ if (seq < node->seq_next)
+ dlm_send_ack(node->nodeid, node->seq_next);
+
+ log_print_ratelimited("ignore dlm msg because seq mismatch, seq: %u, expected: %u, nodeid: %d",
+ seq, node->seq_next, node->nodeid);
+ }
+
+ return;
+
+send_fin:
+ set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+ dlm_send_fin(node, dlm_pas_fin_ack_rcv);
+}
+
+static struct midcomms_node *
+dlm_midcomms_recv_node_lookup(int nodeid, const union dlm_packet *p,
+ uint16_t msglen, int (*cb)(struct midcomms_node *node))
+{
+ struct midcomms_node *node = NULL;
+ gfp_t allocation = 0;
+ int ret;
+
+ switch (p->header.h_cmd) {
+ case DLM_RCOM:
+ if (msglen < sizeof(struct dlm_rcom)) {
+ log_print("rcom msg too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ return NULL;
+ }
+
+ switch (le32_to_cpu(p->rcom.rc_type)) {
+ case DLM_RCOM_NAMES:
+ fallthrough;
+ case DLM_RCOM_NAMES_REPLY:
+ fallthrough;
+ case DLM_RCOM_STATUS:
+ fallthrough;
+ case DLM_RCOM_STATUS_REPLY:
+ node = nodeid2node(nodeid, 0);
+ if (node) {
+ spin_lock(&node->state_lock);
+ if (node->state != DLM_ESTABLISHED)
+ pr_debug("receive begin RCOM msg from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+
+ switch (node->state) {
+ case DLM_CLOSED:
+ node->state = DLM_ESTABLISHED;
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ break;
+ case DLM_ESTABLISHED:
+ break;
+ default:
+ /* some invalid state passive shutdown
+ * was failed, we try to reset and
+ * hope it will go on.
+ */
+ log_print("reset node %d because shutdown stucked",
+ node->nodeid);
+
+ midcomms_node_reset(node);
+ node->state = DLM_ESTABLISHED;
+ break;
+ }
+ spin_unlock(&node->state_lock);
+ }
+
+ allocation = GFP_NOFS;
+ break;
+ default:
+ break;
+ }
+
+ break;
+ default:
+ break;
+ }
+
+ node = nodeid2node(nodeid, allocation);
+ if (!node) {
+ log_print_ratelimited("received dlm message cmd %d nextcmd %d from node %d in an invalid sequence",
+ p->header.h_cmd, p->opts.o_nextcmd, nodeid);
+ return NULL;
+ }
+
+ ret = cb(node);
+ if (ret < 0)
+ return NULL;
+
+ return node;
+}
+
+static int dlm_midcomms_version_check_3_2(struct midcomms_node *node)
+{
+ switch (node->version) {
+ case DLM_VERSION_NOT_SET:
+ node->version = DLM_VERSION_3_2;
+ log_print("version 0x%08x for node %d detected", DLM_VERSION_3_2,
+ node->nodeid);
+ break;
+ case DLM_VERSION_3_2:
+ break;
+ default:
+ log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x",
+ DLM_VERSION_3_2, node->nodeid, node->version);
+ return -1;
+ }
+
+ return 0;
+}
+
+static int dlm_opts_check_msglen(union dlm_packet *p, uint16_t msglen, int nodeid)
+{
+ int len = msglen;
+
+ /* we only trust outer header msglen because
+ * it's checked against receive buffer length.
+ */
+ if (len < sizeof(struct dlm_opts))
+ return -1;
+ len -= sizeof(struct dlm_opts);
+
+ if (len < le16_to_cpu(p->opts.o_optlen))
+ return -1;
+ len -= le16_to_cpu(p->opts.o_optlen);
+
+ switch (p->opts.o_nextcmd) {
+ case DLM_FIN:
+ if (len < sizeof(struct dlm_header)) {
+ log_print("fin too small: %d, will skip this message from node %d",
+ len, nodeid);
+ return -1;
+ }
+
+ break;
+ case DLM_MSG:
+ if (len < sizeof(struct dlm_message)) {
+ log_print("msg too small: %d, will skip this message from node %d",
+ msglen, nodeid);
+ return -1;
+ }
+
+ break;
+ case DLM_RCOM:
+ if (len < sizeof(struct dlm_rcom)) {
+ log_print("rcom msg too small: %d, will skip this message from node %d",
+ len, nodeid);
+ return -1;
+ }
+
+ break;
+ default:
+ log_print("unsupported o_nextcmd received: %u, will skip this message from node %d",
+ p->opts.o_nextcmd, nodeid);
+ return -1;
+ }
+
+ return 0;
+}
+
+static void dlm_midcomms_receive_buffer_3_2(union dlm_packet *p, int nodeid)
+{
+ uint16_t msglen = le16_to_cpu(p->header.h_length);
+ struct midcomms_node *node;
+ uint32_t seq;
+ int ret, idx;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen,
+ dlm_midcomms_version_check_3_2);
+ if (!node)
+ goto out;
+
+ switch (p->header.h_cmd) {
+ case DLM_RCOM:
+ /* these rcom message we use to determine version.
+ * they have their own retransmission handling and
+ * are the first messages of dlm.
+ *
+ * length already checked.
+ */
+ switch (le32_to_cpu(p->rcom.rc_type)) {
+ case DLM_RCOM_NAMES:
+ fallthrough;
+ case DLM_RCOM_NAMES_REPLY:
+ fallthrough;
+ case DLM_RCOM_STATUS:
+ fallthrough;
+ case DLM_RCOM_STATUS_REPLY:
+ break;
+ default:
+ log_print("unsupported rcom type received: %u, will skip this message from node %d",
+ le32_to_cpu(p->rcom.rc_type), nodeid);
+ goto out;
+ }
+
+ WARN_ON(test_bit(DLM_NODE_FLAG_STOP_RX, &node->flags));
+ dlm_receive_buffer(p, nodeid);
+ break;
+ case DLM_OPTS:
+ seq = le32_to_cpu(p->header.u.h_seq);
+
+ ret = dlm_opts_check_msglen(p, msglen, nodeid);
+ if (ret < 0) {
+ log_print("opts msg too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ goto out;
+ }
+
+ p = (union dlm_packet *)((unsigned char *)p->opts.o_opts +
+ le16_to_cpu(p->opts.o_optlen));
+
+ /* recheck inner msglen just if it's not garbage */
+ msglen = le16_to_cpu(p->header.h_length);
+ switch (p->header.h_cmd) {
+ case DLM_RCOM:
+ if (msglen < sizeof(struct dlm_rcom)) {
+ log_print("inner rcom msg too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ goto out;
+ }
+
+ break;
+ case DLM_MSG:
+ if (msglen < sizeof(struct dlm_message)) {
+ log_print("inner msg too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ goto out;
+ }
+
+ break;
+ case DLM_FIN:
+ if (msglen < sizeof(struct dlm_header)) {
+ log_print("inner fin too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ goto out;
+ }
+
+ break;
+ default:
+ log_print("unsupported inner h_cmd received: %u, will skip this message from node %d",
+ msglen, nodeid);
+ goto out;
+ }
+
+ dlm_midcomms_receive_buffer(p, node, seq);
+ break;
+ case DLM_ACK:
+ seq = le32_to_cpu(p->header.u.h_seq);
+ dlm_receive_ack(node, seq);
+ break;
+ default:
+ log_print("unsupported h_cmd received: %u, will skip this message from node %d",
+ p->header.h_cmd, nodeid);
+ break;
+ }
+
+out:
+ srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static int dlm_midcomms_version_check_3_1(struct midcomms_node *node)
{
- return dlm_lowcomms_close(nodeid);
+ switch (node->version) {
+ case DLM_VERSION_NOT_SET:
+ node->version = DLM_VERSION_3_1;
+ log_print("version 0x%08x for node %d detected", DLM_VERSION_3_1,
+ node->nodeid);
+ break;
+ case DLM_VERSION_3_1:
+ break;
+ default:
+ log_print_ratelimited("version mismatch detected, assumed 0x%08x but node %d has 0x%08x",
+ DLM_VERSION_3_1, node->nodeid, node->version);
+ return -1;
+ }
+
+ return 0;
+}
+
+static void dlm_midcomms_receive_buffer_3_1(union dlm_packet *p, int nodeid)
+{
+ uint16_t msglen = le16_to_cpu(p->header.h_length);
+ struct midcomms_node *node;
+ int idx;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = dlm_midcomms_recv_node_lookup(nodeid, p, msglen,
+ dlm_midcomms_version_check_3_1);
+ if (!node) {
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+ }
+ srcu_read_unlock(&nodes_srcu, idx);
+
+ switch (p->header.h_cmd) {
+ case DLM_RCOM:
+ /* length already checked */
+ break;
+ case DLM_MSG:
+ if (msglen < sizeof(struct dlm_message)) {
+ log_print("msg too small: %u, will skip this message from node %d",
+ msglen, nodeid);
+ return;
+ }
+
+ break;
+ default:
+ log_print("unsupported h_cmd received: %u, will skip this message from node %d",
+ p->header.h_cmd, nodeid);
+ return;
+ }
+
+ dlm_receive_buffer(p, nodeid);
}
/*
@@ -101,32 +865,19 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
if (msglen > len)
break;
- switch (hd->h_cmd) {
- case DLM_MSG:
- if (msglen < sizeof(struct dlm_message)) {
- log_print("dlm msg too small: %u, will skip this message",
- msglen);
- goto skip;
- }
-
+ switch (le32_to_cpu(hd->h_version)) {
+ case DLM_VERSION_3_1:
+ dlm_midcomms_receive_buffer_3_1((union dlm_packet *)ptr, nodeid);
break;
- case DLM_RCOM:
- if (msglen < sizeof(struct dlm_rcom)) {
- log_print("dlm rcom msg too small: %u, will skip this message",
- msglen);
- goto skip;
- }
-
+ case DLM_VERSION_3_2:
+ dlm_midcomms_receive_buffer_3_2((union dlm_packet *)ptr, nodeid);
break;
default:
- log_print("unsupported h_cmd received: %u, will skip this message",
- hd->h_cmd);
- goto skip;
+ log_print("received invalid version header: %u from node %d, will skip this message",
+ le32_to_cpu(hd->h_version), nodeid);
+ break;
}
- dlm_receive_buffer((union dlm_packet *)ptr, nodeid);
-
-skip:
ret += msglen;
len -= msglen;
ptr += msglen;
@@ -134,3 +885,454 @@ int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int len)
return ret;
}
+
+void dlm_midcomms_unack_msg_resend(int nodeid)
+{
+ struct midcomms_node *node;
+ struct dlm_mhandle *mh;
+ int idx, ret;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = nodeid2node(nodeid, 0);
+ if (!node) {
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+ }
+
+ /* old protocol, we don't support to retransmit on failure */
+ switch (node->version) {
+ case DLM_VERSION_3_2:
+ break;
+ default:
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+ }
+
+ rcu_read_lock();
+ list_for_each_entry_rcu(mh, &node->send_queue, list) {
+ if (!mh->committed)
+ continue;
+
+ ret = dlm_lowcomms_resend_msg(mh->msg);
+ if (!ret)
+ log_print_ratelimited("retransmit dlm msg, seq %u, nodeid %d",
+ mh->seq, node->nodeid);
+ }
+ rcu_read_unlock();
+ srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static void dlm_fill_opts_header(struct dlm_opts *opts, uint16_t inner_len,
+ uint32_t seq)
+{
+ opts->o_header.h_cmd = DLM_OPTS;
+ opts->o_header.h_version = (DLM_HEADER_MAJOR | DLM_HEADER_MINOR);
+ opts->o_header.h_nodeid = dlm_our_nodeid();
+ opts->o_header.h_length = DLM_MIDCOMMS_OPT_LEN + inner_len;
+ opts->o_header.u.h_seq = seq;
+ header_out(&opts->o_header);
+}
+
+static void midcomms_new_msg_cb(struct dlm_mhandle *mh)
+{
+ atomic_inc(&mh->node->send_queue_cnt);
+
+ spin_lock(&mh->node->send_queue_lock);
+ list_add_tail_rcu(&mh->list, &mh->node->send_queue);
+ spin_unlock(&mh->node->send_queue_lock);
+
+ mh->seq = mh->node->seq_send++;
+}
+
+static struct dlm_msg *dlm_midcomms_get_msg_3_2(struct dlm_mhandle *mh, int nodeid,
+ int len, gfp_t allocation, char **ppc)
+{
+ struct dlm_opts *opts;
+ struct dlm_msg *msg;
+
+ msg = dlm_lowcomms_new_msg(nodeid, len + DLM_MIDCOMMS_OPT_LEN,
+ allocation, ppc, midcomms_new_msg_cb, mh);
+ if (!msg)
+ return NULL;
+
+ opts = (struct dlm_opts *)*ppc;
+ mh->opts = opts;
+
+ /* add possible options here */
+ dlm_fill_opts_header(opts, len, mh->seq);
+
+ *ppc += sizeof(*opts);
+ mh->inner_hd = (const struct dlm_header *)*ppc;
+ return msg;
+}
+
+struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
+ gfp_t allocation, char **ppc)
+{
+ struct midcomms_node *node;
+ struct dlm_mhandle *mh;
+ struct dlm_msg *msg;
+ int idx;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = nodeid2node(nodeid, 0);
+ if (!node) {
+ WARN_ON_ONCE(1);
+ goto err;
+ }
+
+ /* this is a bug, however we going on and hope it will be resolved */
+ WARN_ON(test_bit(DLM_NODE_FLAG_STOP_TX, &node->flags));
+
+ mh = kzalloc(sizeof(*mh), GFP_NOFS);
+ if (!mh)
+ goto err;
+
+ mh->idx = idx;
+ mh->node = node;
+
+ switch (node->version) {
+ case DLM_VERSION_3_1:
+ msg = dlm_lowcomms_new_msg(nodeid, len, allocation, ppc,
+ NULL, NULL);
+ if (!msg) {
+ kfree(mh);
+ goto err;
+ }
+
+ break;
+ case DLM_VERSION_3_2:
+ msg = dlm_midcomms_get_msg_3_2(mh, nodeid, len, allocation,
+ ppc);
+ if (!msg) {
+ kfree(mh);
+ goto err;
+ }
+
+ break;
+ default:
+ kfree(mh);
+ WARN_ON(1);
+ goto err;
+ }
+
+ mh->msg = msg;
+
+ /* keep in mind that is a must to call
+ * dlm_midcomms_commit_msg() which releases
+ * nodes_srcu using mh->idx which is assumed
+ * here that the application will call it.
+ */
+ return mh;
+
+err:
+ srcu_read_unlock(&nodes_srcu, idx);
+ return NULL;
+}
+
+static void dlm_midcomms_commit_msg_3_2(struct dlm_mhandle *mh)
+{
+ /* nexthdr chain for fast lookup */
+ mh->opts->o_nextcmd = mh->inner_hd->h_cmd;
+ mh->committed = true;
+ dlm_lowcomms_commit_msg(mh->msg);
+}
+
+void dlm_midcomms_commit_mhandle(struct dlm_mhandle *mh)
+{
+ switch (mh->node->version) {
+ case DLM_VERSION_3_1:
+ srcu_read_unlock(&nodes_srcu, mh->idx);
+
+ dlm_lowcomms_commit_msg(mh->msg);
+ dlm_lowcomms_put_msg(mh->msg);
+ /* mh is not part of rcu list in this case */
+ kfree(mh);
+ break;
+ case DLM_VERSION_3_2:
+ dlm_midcomms_commit_msg_3_2(mh);
+ srcu_read_unlock(&nodes_srcu, mh->idx);
+ break;
+ default:
+ srcu_read_unlock(&nodes_srcu, mh->idx);
+ WARN_ON(1);
+ break;
+ }
+}
+
+int dlm_midcomms_start(void)
+{
+ int i;
+
+ for (i = 0; i < CONN_HASH_SIZE; i++)
+ INIT_HLIST_HEAD(&node_hash[i]);
+
+ return dlm_lowcomms_start();
+}
+
+static void dlm_act_fin_ack_rcv(struct midcomms_node *node)
+{
+ spin_lock(&node->state_lock);
+ pr_debug("receive active fin ack from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+
+ switch (node->state) {
+ case DLM_FIN_WAIT1:
+ node->state = DLM_FIN_WAIT2;
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ break;
+ case DLM_CLOSING:
+ midcomms_node_reset(node);
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ wake_up(&node->shutdown_wait);
+ break;
+ case DLM_CLOSED:
+ /* not valid but somehow we got what we want */
+ wake_up(&node->shutdown_wait);
+ break;
+ default:
+ spin_unlock(&node->state_lock);
+ log_print("%s: unexpected state: %d\n",
+ __func__, node->state);
+ WARN_ON(1);
+ return;
+ }
+ spin_unlock(&node->state_lock);
+}
+
+void dlm_midcomms_add_member(int nodeid)
+{
+ struct midcomms_node *node;
+ int idx;
+
+ if (nodeid == dlm_our_nodeid())
+ return;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = nodeid2node(nodeid, GFP_NOFS);
+ if (!node) {
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+ }
+
+ spin_lock(&node->state_lock);
+ if (!node->users) {
+ pr_debug("receive add member from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ switch (node->state) {
+ case DLM_ESTABLISHED:
+ break;
+ case DLM_CLOSED:
+ node->state = DLM_ESTABLISHED;
+ pr_debug("switch node %d to state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ break;
+ default:
+ /* some invalid state passive shutdown
+ * was failed, we try to reset and
+ * hope it will go on.
+ */
+ log_print("reset node %d because shutdown stucked",
+ node->nodeid);
+
+ midcomms_node_reset(node);
+ node->state = DLM_ESTABLISHED;
+ break;
+ }
+ }
+
+ node->users++;
+ pr_debug("users inc count %d\n", node->users);
+ spin_unlock(&node->state_lock);
+
+ srcu_read_unlock(&nodes_srcu, idx);
+}
+
+void dlm_midcomms_remove_member(int nodeid)
+{
+ struct midcomms_node *node;
+ int idx;
+
+ if (nodeid == dlm_our_nodeid())
+ return;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ node = nodeid2node(nodeid, 0);
+ if (!node) {
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+ }
+
+ spin_lock(&node->state_lock);
+ node->users--;
+ pr_debug("users dec count %d\n", node->users);
+
+ /* hitting users count to zero means the
+ * other side is running dlm_midcomms_stop()
+ * we meet us to have a clean disconnect.
+ */
+ if (node->users == 0) {
+ pr_debug("receive remove member from node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ switch (node->state) {
+ case DLM_ESTABLISHED:
+ break;
+ case DLM_CLOSE_WAIT:
+ /* passive shutdown DLM_LAST_ACK case 2 */
+ node->state = DLM_LAST_ACK;
+ spin_unlock(&node->state_lock);
+
+ pr_debug("switch node %d to state %s case 2\n",
+ node->nodeid, dlm_state_str(node->state));
+ goto send_fin;
+ case DLM_LAST_ACK:
+ /* probably receive fin caught it, do nothing */
+ break;
+ case DLM_CLOSED:
+ /* already gone, do nothing */
+ break;
+ default:
+ log_print("%s: unexpected state: %d\n",
+ __func__, node->state);
+ break;
+ }
+ }
+ spin_unlock(&node->state_lock);
+
+ srcu_read_unlock(&nodes_srcu, idx);
+ return;
+
+send_fin:
+ set_bit(DLM_NODE_FLAG_STOP_RX, &node->flags);
+ dlm_send_fin(node, dlm_pas_fin_ack_rcv);
+ srcu_read_unlock(&nodes_srcu, idx);
+}
+
+static void midcomms_node_release(struct rcu_head *rcu)
+{
+ struct midcomms_node *node = container_of(rcu, struct midcomms_node, rcu);
+
+ WARN_ON(atomic_read(&node->send_queue_cnt));
+ kfree(node);
+}
+
+static void midcomms_shutdown(struct midcomms_node *node)
+{
+ int ret;
+
+ /* old protocol, we don't wait for pending operations */
+ switch (node->version) {
+ case DLM_VERSION_3_2:
+ break;
+ default:
+ return;
+ }
+
+ spin_lock(&node->state_lock);
+ pr_debug("receive active shutdown for node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ switch (node->state) {
+ case DLM_ESTABLISHED:
+ node->state = DLM_FIN_WAIT1;
+ pr_debug("switch node %d to state %s case 2\n",
+ node->nodeid, dlm_state_str(node->state));
+ break;
+ case DLM_CLOSED:
+ /* we have what we want */
+ spin_unlock(&node->state_lock);
+ return;
+ default:
+ /* busy to enter DLM_FIN_WAIT1, wait until passive
+ * done in shutdown_wait to enter DLM_CLOSED.
+ */
+ break;
+ }
+ spin_unlock(&node->state_lock);
+
+ if (node->state == DLM_FIN_WAIT1) {
+ dlm_send_fin(node, dlm_act_fin_ack_rcv);
+
+ if (DLM_DEBUG_FENCE_TERMINATION)
+ msleep(5000);
+ }
+
+ /* wait for other side dlm + fin */
+ ret = wait_event_timeout(node->shutdown_wait,
+ node->state == DLM_CLOSED ||
+ test_bit(DLM_NODE_FLAG_CLOSE, &node->flags),
+ DLM_SHUTDOWN_TIMEOUT);
+ if (!ret || test_bit(DLM_NODE_FLAG_CLOSE, &node->flags)) {
+ pr_debug("active shutdown timed out for node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+ midcomms_node_reset(node);
+ return;
+ }
+
+ pr_debug("active shutdown done for node %d with state %s\n",
+ node->nodeid, dlm_state_str(node->state));
+}
+
+void dlm_midcomms_shutdown(void)
+{
+ struct midcomms_node *node;
+ int i, idx;
+
+ mutex_lock(&close_lock);
+ idx = srcu_read_lock(&nodes_srcu);
+ for (i = 0; i < CONN_HASH_SIZE; i++) {
+ hlist_for_each_entry_rcu(node, &node_hash[i], hlist) {
+ midcomms_shutdown(node);
+
+ spin_lock(&nodes_lock);
+ hlist_del_rcu(&node->hlist);
+ spin_unlock(&nodes_lock);
+
+ call_srcu(&nodes_srcu, &node->rcu, midcomms_node_release);
+ }
+ }
+ srcu_read_unlock(&nodes_srcu, idx);
+ mutex_unlock(&close_lock);
+
+ dlm_lowcomms_shutdown();
+}
+
+int dlm_midcomms_close(int nodeid)
+{
+ struct midcomms_node *node;
+ int idx, ret;
+
+ if (nodeid == dlm_our_nodeid())
+ return 0;
+
+ idx = srcu_read_lock(&nodes_srcu);
+ /* Abort pending close/remove operation */
+ node = nodeid2node(nodeid, 0);
+ if (node) {
+ /* let shutdown waiters leave */
+ set_bit(DLM_NODE_FLAG_CLOSE, &node->flags);
+ wake_up(&node->shutdown_wait);
+ }
+ srcu_read_unlock(&nodes_srcu, idx);
+
+ synchronize_srcu(&nodes_srcu);
+
+ idx = srcu_read_lock(&nodes_srcu);
+ mutex_lock(&close_lock);
+ node = nodeid2node(nodeid, 0);
+ if (!node) {
+ mutex_unlock(&close_lock);
+ srcu_read_unlock(&nodes_srcu, idx);
+ return dlm_lowcomms_close(nodeid);
+ }
+
+ ret = dlm_lowcomms_close(nodeid);
+ spin_lock(&node->state_lock);
+ midcomms_node_reset(node);
+ spin_unlock(&node->state_lock);
+ srcu_read_unlock(&nodes_srcu, idx);
+ mutex_unlock(&close_lock);
+
+ return ret;
+}
diff --git a/fs/dlm/midcomms.h b/fs/dlm/midcomms.h
index 9ac1190ce277..1178b836315b 100644
--- a/fs/dlm/midcomms.h
+++ b/fs/dlm/midcomms.h
@@ -21,6 +21,7 @@ int dlm_midcomms_start(void);
void dlm_midcomms_shutdown(void);
void dlm_midcomms_add_member(int nodeid);
void dlm_midcomms_remove_member(int nodeid);
+void dlm_midcomms_unack_msg_resend(int nodeid);
#endif /* __MIDCOMMS_DOT_H__ */
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index 7c1a06bcd418..085f21966c72 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -202,7 +202,7 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
set_rcom_status(ls, (struct rcom_status *)rc->rc_buf, status_flags);
allow_sync_reply(ls, &rc->rc_id);
- memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
+ memset(ls->ls_recover_buf, 0, DEFAULT_BUFFER_SIZE);
send_rcom_stateless(ls, msg, rc);
@@ -325,7 +325,7 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
memcpy(rc->rc_buf, last_name, last_len);
allow_sync_reply(ls, &rc->rc_id);
- memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
+ memset(ls->ls_recover_buf, 0, DEFAULT_BUFFER_SIZE);
send_rcom_stateless(ls, msg, rc);
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 15/16] fs: dlm: add midcomms debugfs functionality
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (13 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 14/16] fs: dlm: add reliable connection if reconnect Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 16/16] fs: dlm: don't allow half transmitted messages Alexander Aring
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch adds functionality to debug midcomms per connection state
inside a comms directory which is similar like dlm configfs. Currently
there exists the possibility to read out two attributes which is the
send queue counter and the version of each midcomms node state.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/debug_fs.c | 54 +++++++++++++++++++++++++++++++++++++++++++
fs/dlm/dlm_internal.h | 4 ++++
fs/dlm/midcomms.c | 27 ++++++++++++++++++++++
fs/dlm/midcomms.h | 6 +++++
4 files changed, 91 insertions(+)
diff --git a/fs/dlm/debug_fs.c b/fs/dlm/debug_fs.c
index d5bd990bcab8..47e9d57e4cae 100644
--- a/fs/dlm/debug_fs.c
+++ b/fs/dlm/debug_fs.c
@@ -16,6 +16,7 @@
#include <linux/slab.h>
#include "dlm_internal.h"
+#include "midcomms.h"
#include "lock.h"
#define DLM_DEBUG_BUF_LEN 4096
@@ -23,6 +24,7 @@ static char debug_buf[DLM_DEBUG_BUF_LEN];
static struct mutex debug_buf_lock;
static struct dentry *dlm_root;
+static struct dentry *dlm_comms;
static char *print_lockmode(int mode)
{
@@ -738,6 +740,57 @@ void dlm_delete_debug_file(struct dlm_ls *ls)
debugfs_remove(ls->ls_debug_toss_dentry);
}
+static int dlm_state_show(struct seq_file *file, void *offset)
+{
+ seq_printf(file, "%s\n", dlm_midcomms_state(file->private));
+ return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_state);
+
+static int dlm_flags_show(struct seq_file *file, void *offset)
+{
+ seq_printf(file, "%lu\n", dlm_midcomms_flags(file->private));
+ return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_flags);
+
+static int dlm_send_queue_cnt_show(struct seq_file *file, void *offset)
+{
+ seq_printf(file, "%d\n", dlm_midcomms_send_queue_cnt(file->private));
+ return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_send_queue_cnt);
+
+static int dlm_version_show(struct seq_file *file, void *offset)
+{
+ seq_printf(file, "0x%08x\n", dlm_midcomms_version(file->private));
+ return 0;
+}
+DEFINE_SHOW_ATTRIBUTE(dlm_version);
+
+void *dlm_create_debug_comms_file(int nodeid, void *data)
+{
+ struct dentry *d_node;
+ char name[256];
+
+ memset(name, 0, sizeof(name));
+ snprintf(name, 256, "%d", nodeid);
+
+ d_node = debugfs_create_dir(name, dlm_comms);
+ debugfs_create_file("state", 0444, d_node, data, &dlm_state_fops);
+ debugfs_create_file("flags", 0444, d_node, data, &dlm_flags_fops);
+ debugfs_create_file("send_queue_count", 0444, d_node, data,
+ &dlm_send_queue_cnt_fops);
+ debugfs_create_file("version", 0444, d_node, data, &dlm_version_fops);
+
+ return d_node;
+}
+
+void dlm_delete_debug_comms_file(void *ctx)
+{
+ debugfs_remove(ctx);
+}
+
void dlm_create_debug_file(struct dlm_ls *ls)
{
char name[DLM_LOCKSPACE_LEN + 8];
@@ -797,6 +850,7 @@ void __init dlm_register_debugfs(void)
{
mutex_init(&debug_buf_lock);
dlm_root = debugfs_create_dir("dlm", NULL);
+ dlm_comms = debugfs_create_dir("comms", dlm_root);
}
void dlm_unregister_debugfs(void)
diff --git a/fs/dlm/dlm_internal.h b/fs/dlm/dlm_internal.h
index 40917c878370..91d1ca3a121a 100644
--- a/fs/dlm/dlm_internal.h
+++ b/fs/dlm/dlm_internal.h
@@ -754,11 +754,15 @@ void dlm_register_debugfs(void);
void dlm_unregister_debugfs(void);
void dlm_create_debug_file(struct dlm_ls *ls);
void dlm_delete_debug_file(struct dlm_ls *ls);
+void *dlm_create_debug_comms_file(int nodeid, void *data);
+void dlm_delete_debug_comms_file(void *ctx);
#else
static inline void dlm_register_debugfs(void) { }
static inline void dlm_unregister_debugfs(void) { }
static inline void dlm_create_debug_file(struct dlm_ls *ls) { }
static inline void dlm_delete_debug_file(struct dlm_ls *ls) { }
+static inline void *dlm_create_debug_comms_file(int nodeid, void *data) { return NULL; }
+static inline void dlm_delete_debug_comms_file(void *ctx) { }
#endif
#endif /* __DLM_INTERNAL_DOT_H__ */
diff --git a/fs/dlm/midcomms.c b/fs/dlm/midcomms.c
index eef3938a363e..35664950f6b7 100644
--- a/fs/dlm/midcomms.c
+++ b/fs/dlm/midcomms.c
@@ -189,6 +189,9 @@ struct midcomms_node {
*/
int users;
+ /* not protected by srcu, node_hash lifetime */
+ void *debugfs;
+
struct hlist_node hlist;
struct rcu_head rcu;
};
@@ -244,6 +247,26 @@ static inline const char *dlm_state_str(int state)
}
}
+const char *dlm_midcomms_state(struct midcomms_node *node)
+{
+ return dlm_state_str(node->state);
+}
+
+unsigned long dlm_midcomms_flags(struct midcomms_node *node)
+{
+ return node->flags;
+}
+
+int dlm_midcomms_send_queue_cnt(struct midcomms_node *node)
+{
+ return atomic_read(&node->send_queue_cnt);
+}
+
+uint32_t dlm_midcomms_version(struct midcomms_node *node)
+{
+ return node->version;
+}
+
static struct midcomms_node *__find_node(int nodeid, int r)
{
struct midcomms_node *node;
@@ -332,6 +355,8 @@ static struct midcomms_node *nodeid2node(int nodeid, gfp_t alloc)
hlist_add_head_rcu(&node->hlist, &node_hash[r]);
spin_unlock(&nodes_lock);
+
+ node->debugfs = dlm_create_debug_comms_file(nodeid, node);
return node;
}
@@ -1285,6 +1310,8 @@ void dlm_midcomms_shutdown(void)
hlist_for_each_entry_rcu(node, &node_hash[i], hlist) {
midcomms_shutdown(node);
+ dlm_delete_debug_comms_file(node->debugfs);
+
spin_lock(&nodes_lock);
hlist_del_rcu(&node->hlist);
spin_unlock(&nodes_lock);
diff --git a/fs/dlm/midcomms.h b/fs/dlm/midcomms.h
index 1178b836315b..579abc6929be 100644
--- a/fs/dlm/midcomms.h
+++ b/fs/dlm/midcomms.h
@@ -12,6 +12,8 @@
#ifndef __MIDCOMMS_DOT_H__
#define __MIDCOMMS_DOT_H__
+struct midcomms_node;
+
int dlm_process_incoming_buffer(int nodeid, unsigned char *buf, int buflen);
struct dlm_mhandle *dlm_midcomms_get_mhandle(int nodeid, int len,
gfp_t allocation, char **ppc);
@@ -22,6 +24,10 @@ void dlm_midcomms_shutdown(void);
void dlm_midcomms_add_member(int nodeid);
void dlm_midcomms_remove_member(int nodeid);
void dlm_midcomms_unack_msg_resend(int nodeid);
+const char *dlm_midcomms_state(struct midcomms_node *node);
+unsigned long dlm_midcomms_flags(struct midcomms_node *node);
+int dlm_midcomms_send_queue_cnt(struct midcomms_node *node);
+uint32_t dlm_midcomms_version(struct midcomms_node *node);
#endif /* __MIDCOMMS_DOT_H__ */
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread* [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 16/16] fs: dlm: don't allow half transmitted messages
2021-05-21 19:08 [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 00/16] fs: dlm: introduce dlm re-transmission layer Alexander Aring
` (14 preceding siblings ...)
2021-05-21 19:08 ` [Cluster-devel] [PATCHv6 v5.13-rc1 dlm/next 15/16] fs: dlm: add midcomms debugfs functionality Alexander Aring
@ 2021-05-21 19:08 ` Alexander Aring
15 siblings, 0 replies; 17+ messages in thread
From: Alexander Aring @ 2021-05-21 19:08 UTC (permalink / raw)
To: cluster-devel.redhat.com
This patch will clean a dirty page buffer if a reconnect occurs. If a page
buffer was half transmitted we cannot start inside the middle of a dlm
message if a node connects again. I observed invalid length receptions
errors and was guessing that this behaviour occurs, after this patch I
never saw an invalid message length again. This patch might drops more
messages for dlm version 3.1 but 3.1 can't deal with half messages as
well, for 3.2 it might trigger more re-transmissions but will not leave dlm
in a broken state.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
fs/dlm/lowcomms.c | 95 ++++++++++++++++++++++++++++++-----------------
1 file changed, 60 insertions(+), 35 deletions(-)
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c
index fe9113bd5ba0..36adccc4f849 100644
--- a/fs/dlm/lowcomms.c
+++ b/fs/dlm/lowcomms.c
@@ -118,6 +118,7 @@ struct writequeue_entry {
int len;
int end;
int users;
+ bool dirty;
struct connection *con;
struct list_head msgs;
struct kref ref;
@@ -700,6 +701,42 @@ static void make_sockaddr(struct sockaddr_storage *saddr, uint16_t port,
memset((char *)saddr + *addr_len, 0, sizeof(struct sockaddr_storage) - *addr_len);
}
+static void dlm_page_release(struct kref *kref)
+{
+ struct writequeue_entry *e = container_of(kref, struct writequeue_entry,
+ ref);
+
+ __free_page(e->page);
+ kfree(e);
+}
+
+static void dlm_msg_release(struct kref *kref)
+{
+ struct dlm_msg *msg = container_of(kref, struct dlm_msg, ref);
+
+ kref_put(&msg->entry->ref, dlm_page_release);
+ kfree(msg);
+}
+
+static void free_entry(struct writequeue_entry *e)
+{
+ struct dlm_msg *msg, *tmp;
+
+ list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
+ if (msg->orig_msg) {
+ msg->orig_msg->retransmit = false;
+ kref_put(&msg->orig_msg->ref, dlm_msg_release);
+ }
+
+ list_del(&msg->list);
+ kref_put(&msg->ref, dlm_msg_release);
+ }
+
+ list_del(&e->list);
+ atomic_dec(&e->con->writequeue_cnt);
+ kref_put(&e->ref, dlm_page_release);
+}
+
static void dlm_close_sock(struct socket **sock)
{
if (*sock) {
@@ -714,6 +751,7 @@ static void close_connection(struct connection *con, bool and_other,
bool tx, bool rx)
{
bool closing = test_and_set_bit(CF_CLOSING, &con->flags);
+ struct writequeue_entry *e;
if (tx && !closing && cancel_work_sync(&con->swork)) {
log_print("canceled swork for node %d", con->nodeid);
@@ -732,6 +770,26 @@ static void close_connection(struct connection *con, bool and_other,
close_connection(con->othercon, false, tx, rx);
}
+ /* if we send a writequeue entry only a half way, we drop the
+ * whole entry because reconnection and that we not start of the
+ * middle of a msg which will confuse the other end.
+ *
+ * we can always drop messages because retransmits, but what we
+ * cannot allow is to transmit half messages which may be processed
+ * at the other side.
+ *
+ * our policy is to start on a clean state when disconnects, we don't
+ * know what's send/received on transport layer in this case.
+ */
+ spin_lock(&con->writequeue_lock);
+ if (!list_empty(&con->writequeue)) {
+ e = list_first_entry(&con->writequeue, struct writequeue_entry,
+ list);
+ if (e->dirty)
+ free_entry(e);
+ }
+ spin_unlock(&con->writequeue_lock);
+
con->rx_leftover = 0;
con->retries = 0;
clear_bit(CF_CONNECTED, &con->flags);
@@ -1026,41 +1084,6 @@ static int accept_from_sock(struct listen_connection *con)
return result;
}
-static void dlm_page_release(struct kref *kref)
-{
- struct writequeue_entry *e = container_of(kref, struct writequeue_entry,
- ref);
-
- __free_page(e->page);
- kfree(e);
-}
-
-static void dlm_msg_release(struct kref *kref)
-{
- struct dlm_msg *msg = container_of(kref, struct dlm_msg, ref);
-
- kref_put(&msg->entry->ref, dlm_page_release);
- kfree(msg);
-}
-
-static void free_entry(struct writequeue_entry *e)
-{
- struct dlm_msg *msg, *tmp;
-
- list_for_each_entry_safe(msg, tmp, &e->msgs, list) {
- if (msg->orig_msg) {
- msg->orig_msg->retransmit = false;
- kref_put(&msg->orig_msg->ref, dlm_msg_release);
- }
- list_del(&msg->list);
- kref_put(&msg->ref, dlm_msg_release);
- }
-
- list_del(&e->list);
- atomic_dec(&e->con->writequeue_cnt);
- kref_put(&e->ref, dlm_page_release);
-}
-
/*
* writequeue_entry_complete - try to delete and free write queue entry
* @e: write queue entry to try to delete
@@ -1072,6 +1095,8 @@ static void writequeue_entry_complete(struct writequeue_entry *e, int completed)
{
e->offset += completed;
e->len -= completed;
+ /* signal that page was half way transmitted */
+ e->dirty = true;
if (e->len == 0 && e->users == 0)
free_entry(e);
--
2.26.3
^ permalink raw reply related [flat|nested] 17+ messages in thread