[PULL 22/34] nbd/client-connection: implement connection retry

[Eric Blake <eblake@redhat.com>]

From: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

Add an option for a thread to retry connecting until it succeeds. We'll
use nbd/client-connection both for reconnect and for initial connection
in nbd_open(), so we need a possibility to use same NBDClientConnection
instance to connect once in nbd_open() and then use retry semantics for
reconnect.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20210610100802.5888-21-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
[eblake: grammar tweak]
Signed-off-by: Eric Blake <eblake@redhat.com>
---
 include/block/nbd.h     |  2 ++
 nbd/client-connection.c | 56 +++++++++++++++++++++++++++++++----------
 2 files changed, 45 insertions(+), 13 deletions(-)

diff --git a/include/block/nbd.h b/include/block/nbd.h
index 5d86e6a393ff..5bb54d831c8a 100644
--- a/include/block/nbd.h
+++ b/include/block/nbd.h
@@ -409,6 +409,8 @@ const char *nbd_err_lookup(int err);
 /* nbd/client-connection.c */
 typedef struct NBDClientConnection NBDClientConnection;

+void nbd_client_connection_enable_retry(NBDClientConnection *conn);
+
 NBDClientConnection *nbd_client_connection_new(const SocketAddress *saddr,
                                                bool do_negotiation,
                                                const char *export_name,
diff --git a/nbd/client-connection.c b/nbd/client-connection.c
index 4ed37cd73fa1..032b38ed3e9b 100644
--- a/nbd/client-connection.c
+++ b/nbd/client-connection.c
@@ -35,6 +35,7 @@ struct NBDClientConnection {
     QCryptoTLSCreds *tlscreds;
     NBDExportInfo initial_info;
     bool do_negotiation;
+    bool do_retry;

     QemuMutex mutex;

@@ -61,6 +62,15 @@ struct NBDClientConnection {
     Coroutine *wait_co;
 };

+/*
+ * The function isn't protected by any mutex, only call it when the client
+ * connection attempt has not yet started.
+ */
+void nbd_client_connection_enable_retry(NBDClientConnection *conn)
+{
+    conn->do_retry = true;
+}
+
 NBDClientConnection *nbd_client_connection_new(const SocketAddress *saddr,
                                                bool do_negotiation,
                                                const char *export_name,
@@ -155,24 +165,44 @@ static void *connect_thread_func(void *opaque)
     NBDClientConnection *conn = opaque;
     int ret;
     bool do_free;
+    uint64_t timeout = 1;
+    uint64_t max_timeout = 16;

-    conn->sioc = qio_channel_socket_new();
+    while (true) {
+        conn->sioc = qio_channel_socket_new();

-    error_free(conn->err);
-    conn->err = NULL;
-    conn->updated_info = conn->initial_info;
+        error_free(conn->err);
+        conn->err = NULL;
+        conn->updated_info = conn->initial_info;

-    ret = nbd_connect(conn->sioc, conn->saddr,
-                      conn->do_negotiation ? &conn->updated_info : NULL,
-                      conn->tlscreds, &conn->ioc, &conn->err);
-    if (ret < 0) {
-        object_unref(OBJECT(conn->sioc));
-        conn->sioc = NULL;
+        ret = nbd_connect(conn->sioc, conn->saddr,
+                          conn->do_negotiation ? &conn->updated_info : NULL,
+                          conn->tlscreds, &conn->ioc, &conn->err);
+
+        /*
+         * conn->updated_info will finally be returned to the user. Clear the
+         * pointers to our internally allocated strings, which are IN parameters
+         * of nbd_receive_negotiate() and therefore nbd_connect(). Caller
+         * shoudn't be interested in these fields.
+         */
+        conn->updated_info.x_dirty_bitmap = NULL;
+        conn->updated_info.name = NULL;
+
+        if (ret < 0) {
+            object_unref(OBJECT(conn->sioc));
+            conn->sioc = NULL;
+            if (conn->do_retry) {
+                sleep(timeout);
+                if (timeout < max_timeout) {
+                    timeout *= 2;
+                }
+                continue;
+            }
+        }
+
+        break;
     }

-    conn->updated_info.x_dirty_bitmap = NULL;
-    conn->updated_info.name = NULL;
-
     qemu_mutex_lock(&conn->mutex);

     assert(conn->running);
-- 
2.31.1