block/nbd: ensure ->connection_thread is always valid

Simplify lifetime management of BDRVNBDState->connect_thread by
delaying the possible cleanup of it until the BDRVNBDState itself goes
away.

This also reverts
 0267101af6 "block/nbd: fix possible use after free of s->connect_thread"
as now s->connect_thread can't be cleared until the very end.

Signed-off-by: Roman Kagan <rvkagan@yandex-team.ru>
 [vsementsov: rebase, revert 0267101af6 changes]
Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
 [eblake: tweak comment]
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20210610100802.5888-8-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

1 files changed, 21 insertions, 35 deletions

diff --git a/block/nbd.c b/block/nbd.c
index 1c99654ef7..08ae47d83c 100644
--- a/block/nbd.c
+++ b/block/nbd.c
@@ -144,17 +144,31 @@ typedef struct BDRVNBDState {
     NBDConnectThread *connect_thread;
 } BDRVNBDState;
 
+static void nbd_free_connect_thread(NBDConnectThread *thr);
 static int nbd_establish_connection(BlockDriverState *bs, SocketAddress *saddr,
                                     Error **errp);
 static int nbd_co_establish_connection(BlockDriverState *bs, Error **errp);
-static void nbd_co_establish_connection_cancel(BlockDriverState *bs,
-                                               bool detach);
+static void nbd_co_establish_connection_cancel(BlockDriverState *bs);
 static int nbd_client_handshake(BlockDriverState *bs, Error **errp);
 static void nbd_yank(void *opaque);
 
 static void nbd_clear_bdrvstate(BlockDriverState *bs)
 {
     BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
+    NBDConnectThread *thr = s->connect_thread;
+    bool thr_running;
+
+    qemu_mutex_lock(&thr->mutex);
+    thr_running = thr->state == CONNECT_THREAD_RUNNING;
+    if (thr_running) {
+        thr->state = CONNECT_THREAD_RUNNING_DETACHED;
+    }
+    qemu_mutex_unlock(&thr->mutex);
+
+    /* the runaway thread will clean up itself */
+    if (!thr_running) {
+        nbd_free_connect_thread(thr);
+    }
 
     yank_unregister_instance(BLOCKDEV_YANK_INSTANCE(bs->node_name));
 
@@ -295,7 +309,7 @@ static void coroutine_fn nbd_client_co_drain_begin(BlockDriverState *bs)
     s->drained = true;
     qemu_co_sleep_wake(&s->reconnect_sleep);
 
-    nbd_co_establish_connection_cancel(bs, false);
+    nbd_co_establish_connection_cancel(bs);
 
     reconnect_delay_timer_del(s);
 
@@ -333,7 +347,7 @@ static void nbd_teardown_connection(BlockDriverState *bs)
     s->state = NBD_CLIENT_QUIT;
     if (s->connection_co) {
         qemu_co_sleep_wake(&s->reconnect_sleep);
-        nbd_co_establish_connection_cancel(bs, true);
+        nbd_co_establish_connection_cancel(bs);
     }
     if (qemu_in_coroutine()) {
         s->teardown_co = qemu_coroutine_self();
@@ -446,11 +460,6 @@ nbd_co_establish_connection(BlockDriverState *bs, Error **errp)
     BDRVNBDState *s = bs->opaque;
     NBDConnectThread *thr = s->connect_thread;
 
-    if (!thr) {
-        /* detached */
-        return -1;
-    }
-
     qemu_mutex_lock(&thr->mutex);
 
     switch (thr->state) {
@@ -494,12 +503,6 @@ nbd_co_establish_connection(BlockDriverState *bs, Error **errp)
     s->wait_connect = true;
     qemu_coroutine_yield();
 
-    if (!s->connect_thread) {
-        /* detached */
-        return -1;
-    }
-    assert(thr == s->connect_thread);
-
     qemu_mutex_lock(&thr->mutex);
 
     switch (thr->state) {
@@ -547,18 +550,12 @@ nbd_co_establish_connection(BlockDriverState *bs, Error **errp)
  * nbd_co_establish_connection_cancel
  * Cancel nbd_co_establish_connection asynchronously: it will finish soon, to
  * allow drained section to begin.
- *
- * If detach is true, also cleanup the state (or if thread is running, move it
- * to CONNECT_THREAD_RUNNING_DETACHED state). s->connect_thread becomes NULL if
- * detach is true.
  */
-static void nbd_co_establish_connection_cancel(BlockDriverState *bs,
-                                               bool detach)
+static void nbd_co_establish_connection_cancel(BlockDriverState *bs)
 {
     BDRVNBDState *s = bs->opaque;
     NBDConnectThread *thr = s->connect_thread;
     bool wake = false;
-    bool do_free = false;
 
     qemu_mutex_lock(&thr->mutex);
 
@@ -569,21 +566,10 @@ static void nbd_co_establish_connection_cancel(BlockDriverState *bs,
             s->wait_connect = false;
             wake = true;
         }
-        if (detach) {
-            thr->state = CONNECT_THREAD_RUNNING_DETACHED;
-            s->connect_thread = NULL;
-        }
-    } else if (detach) {
-        do_free = true;
     }
 
     qemu_mutex_unlock(&thr->mutex);
 
-    if (do_free) {
-        nbd_free_connect_thread(thr);
-        s->connect_thread = NULL;
-    }
-
     if (wake) {
         aio_co_wake(s->connection_co);
     }
@@ -2310,6 +2296,8 @@ static int nbd_open(BlockDriverState *bs, QDict *options, int flags,
         goto fail;
     }
 
+    nbd_init_connect_thread(s);
+
     /*
      * establish TCP connection, return error if it fails
      * TODO: Configurable retry-until-timeout behaviour.
@@ -2326,8 +2314,6 @@ static int nbd_open(BlockDriverState *bs, QDict *options, int flags,
     /* successfully connected */
     s->state = NBD_CLIENT_CONNECTED;
 
-    nbd_init_connect_thread(s);
-
     s->connection_co = qemu_coroutine_create(nbd_connection_entry, s);
     bdrv_inc_in_flight(bs);
     aio_co_schedule(bdrv_get_aio_context(bs), s->connection_co);