block/nbd: on shutdown terminate connection attempt

On shutdown nbd driver may be in a connecting state. We should shutdown
it as well, otherwise we may hang in
nbd_teardown_connection, waiting for conneciton_co to finish in
BDRV_POLL_WHILE(bs, s->connection_co) loop if remote server is down.

How to reproduce the dead lock:

1. Create nbd-fault-injector.conf with the following contents:

[inject-error "mega1"]
event=data
io=readwrite
when=before

2. In one terminal run nbd-fault-injector in a loop, like this:

n=1; while true; do
    echo $n; ((n++));
    ./nbd-fault-injector.py 127.0.0.1:10000 nbd-fault-injector.conf;
done

3. In another terminal run qemu-io in a loop, like this:

n=1; while true; do
    echo $n; ((n++));
    ./qemu-io -c 'read 0 512' nbd://127.0.0.1:10000;
done

After some time, qemu-io will hang. Note, that this hang may be
triggered by another bug, so the whole case is fixed only together with
commit "block/nbd: allow drain during reconnect attempt".

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20200727184751.15704-4-vsementsov@virtuozzo.com>
Signed-off-by: Eric Blake <eblake@redhat.com>
This commit is contained in:
Vladimir Sementsov-Ogievskiy 2020-07-27 21:47:49 +03:00 committed by Eric Blake
parent dd1ec1a4af
commit fbeb3e63b3

View File

@ -209,11 +209,15 @@ static void nbd_teardown_connection(BlockDriverState *bs)
{
BDRVNBDState *s = (BDRVNBDState *)bs->opaque;
if (s->state == NBD_CLIENT_CONNECTED) {
if (s->ioc) {
/* finish any pending coroutines */
assert(s->ioc);
qio_channel_shutdown(s->ioc, QIO_CHANNEL_SHUTDOWN_BOTH, NULL);
} else if (s->sioc) {
/* abort negotiation */
qio_channel_shutdown(QIO_CHANNEL(s->sioc), QIO_CHANNEL_SHUTDOWN_BOTH,
NULL);
}
s->state = NBD_CLIENT_QUIT;
if (s->connection_co) {
if (s->connection_co_sleep_ns_state) {
@ -1459,6 +1463,9 @@ static int nbd_client_handshake(BlockDriverState *bs, QIOChannelSocket *sioc,
int ret;
trace_nbd_client_handshake(s->export);
s->sioc = sioc;
qio_channel_set_blocking(QIO_CHANNEL(sioc), false, NULL);
qio_channel_attach_aio_context(QIO_CHANNEL(sioc), aio_context);
@ -1473,6 +1480,7 @@ static int nbd_client_handshake(BlockDriverState *bs, QIOChannelSocket *sioc,
g_free(s->info.name);
if (ret < 0) {
object_unref(OBJECT(sioc));
s->sioc = NULL;
return ret;
}
if (s->x_dirty_bitmap && !s->info.base_allocation) {
@ -1498,8 +1506,6 @@ static int nbd_client_handshake(BlockDriverState *bs, QIOChannelSocket *sioc,
}
}
s->sioc = sioc;
if (!s->ioc) {
s->ioc = QIO_CHANNEL(sioc);
object_ref(OBJECT(s->ioc));
@ -1520,6 +1526,7 @@ static int nbd_client_handshake(BlockDriverState *bs, QIOChannelSocket *sioc,
nbd_send_request(s->ioc ?: QIO_CHANNEL(sioc), &request);
object_unref(OBJECT(sioc));
s->sioc = NULL;
return ret;
}