diff options
author | Alexander Aring <aahringo@redhat.com> | 2023-01-12 17:18:42 -0500 |
---|---|---|
committer | David Teigland <teigland@redhat.com> | 2023-01-23 14:36:37 -0600 |
commit | 54fbe0c15d9a8072e2db7f6765f50c995834e2d7 (patch) | |
tree | c2ecf18fc518e674fa53ae6dec6fb06aa9e271c0 /fs/dlm/lowcomms.c | |
parent | 00908b3388255fc1d3782b744d07f327712f401f (diff) | |
download | linux-54fbe0c15d9a8072e2db7f6765f50c995834e2d7.tar.gz |
fs: dlm: bring back previous shutdown handling
This patch mostly reverts commit 4f567acb0b86 ("fs: dlm: remove socket
shutdown handling"). There can be situations where the dlm midcomms nodes
hash and lowcomms connection hash are not equal, but we need to guarantee
that the lowcomms are all closed on a last release of a dlm lockspace,
when a shutdown is invoked. This patch guarantees that we always close
all sockets managed by the lowcomms connection hash, and calls shutdown
for the last message sent. This ensures we don't cut the socket, which
could cause the peer to get a connection reset.
In future we should try to merge the midcomms/lowcomms hashes into one
hash and not handle both in separate hashes.
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Signed-off-by: David Teigland <teigland@redhat.com>
Diffstat (limited to 'fs/dlm/lowcomms.c')
-rw-r--r-- | fs/dlm/lowcomms.c | 77 |
1 files changed, 56 insertions, 21 deletions
diff --git a/fs/dlm/lowcomms.c b/fs/dlm/lowcomms.c index 4450721ec83c6..61cd6c2628fa5 100644 --- a/fs/dlm/lowcomms.c +++ b/fs/dlm/lowcomms.c @@ -61,6 +61,7 @@ #include "memory.h" #include "config.h" +#define DLM_SHUTDOWN_WAIT_TIMEOUT msecs_to_jiffies(5000) #define NEEDED_RMEM (4*1024*1024) struct connection { @@ -99,6 +100,7 @@ struct connection { struct connection *othercon; struct work_struct rwork; /* receive worker */ struct work_struct swork; /* send worker */ + wait_queue_head_t shutdown_wait; unsigned char rx_leftover_buf[DLM_MAX_SOCKET_BUFSIZE]; int rx_leftover; int mark; @@ -282,6 +284,7 @@ static void dlm_con_init(struct connection *con, int nodeid) INIT_WORK(&con->swork, process_send_sockets); INIT_WORK(&con->rwork, process_recv_sockets); spin_lock_init(&con->addrs_lock); + init_waitqueue_head(&con->shutdown_wait); } /* @@ -790,6 +793,43 @@ static void close_connection(struct connection *con, bool and_other) up_write(&con->sock_lock); } +static void shutdown_connection(struct connection *con, bool and_other) +{ + int ret; + + if (con->othercon && and_other) + shutdown_connection(con->othercon, false); + + flush_workqueue(io_workqueue); + down_read(&con->sock_lock); + /* nothing to shutdown */ + if (!con->sock) { + up_read(&con->sock_lock); + return; + } + + ret = kernel_sock_shutdown(con->sock, SHUT_WR); + up_read(&con->sock_lock); + if (ret) { + log_print("Connection %p failed to shutdown: %d will force close", + con, ret); + goto force_close; + } else { + ret = wait_event_timeout(con->shutdown_wait, !con->sock, + DLM_SHUTDOWN_WAIT_TIMEOUT); + if (ret == 0) { + log_print("Connection %p shutdown timed out, will force close", + con); + goto force_close; + } + } + + return; + +force_close: + close_connection(con, false); +} + static struct processqueue_entry *new_processqueue_entry(int nodeid, int buflen) { @@ -1488,6 +1528,7 @@ static void process_recv_sockets(struct work_struct *work) break; case DLM_IO_EOF: close_connection(con, false); + wake_up(&con->shutdown_wait); /* CF_RECV_PENDING cleared */ break; case DLM_IO_RESCHED: @@ -1695,6 +1736,9 @@ static int work_start(void) void dlm_lowcomms_shutdown(void) { + struct connection *con; + int i, idx; + /* stop lowcomms_listen_data_ready calls */ lock_sock(listen_con.sock->sk); listen_con.sock->sk->sk_data_ready = listen_sock.sk_data_ready; @@ -1703,29 +1747,20 @@ void dlm_lowcomms_shutdown(void) cancel_work_sync(&listen_con.rwork); dlm_close_sock(&listen_con.sock); - flush_workqueue(process_workqueue); -} - -void dlm_lowcomms_shutdown_node(int nodeid, bool force) -{ - struct connection *con; - int idx; - idx = srcu_read_lock(&connections_srcu); - con = nodeid2con(nodeid, 0); - if (WARN_ON_ONCE(!con)) { - srcu_read_unlock(&connections_srcu, idx); - return; - } + for (i = 0; i < CONN_HASH_SIZE; i++) { + hlist_for_each_entry_rcu(con, &connection_hash[i], list) { + shutdown_connection(con, true); + stop_connection_io(con); + flush_workqueue(process_workqueue); + close_connection(con, true); - flush_work(&con->swork); - stop_connection_io(con); - WARN_ON_ONCE(!force && !list_empty(&con->writequeue)); - close_connection(con, true); - clean_one_writequeue(con); - if (con->othercon) - clean_one_writequeue(con->othercon); - allow_connection_io(con); + clean_one_writequeue(con); + if (con->othercon) + clean_one_writequeue(con->othercon); + allow_connection_io(con); + } + } srcu_read_unlock(&connections_srcu, idx); } |