[CRIU] [PATCH 2/2] rst: only restore dgram socket queue once

[Pavel Emelyanov]

On 07/23/2015 02:51 AM, Tycho Andersen wrote:
> In the case where there were multiple clients for a dgram socket, we were
> restoring the queue for each client. Instead, we should pick one client and
> she should restore the queue while the rest skip it.
> 
> Signed-off-by: Tycho Andersen <tycho.andersen at canonical.com>
> ---
>  sk-unix.c | 12 +++++++++++-
>  1 file changed, 11 insertions(+), 1 deletion(-)
> 
> diff --git a/sk-unix.c b/sk-unix.c
> index 6c9ec25..e72fcbb 100644
> --- a/sk-unix.c
> +++ b/sk-unix.c
> @@ -638,6 +638,13 @@ struct unix_sk_info {
>  	 * after listen() if the socket is in TCP_LISTEN.
>  	 */
>  	futex_t prepared;
> +
> +	/*
> +	 * For DGRAM sockets with queues, we should only restore the queue
> +	 * once although it may be open by more than one tid. This is the peer
> +	 * that should do the queueing.
> +	 */
> +	u32 queuer;

How about just putting a SK_QUEUE_RESTORED flag on the peer's flag?

>  };
>  
>  #define USK_PAIR_MASTER		0x1
> @@ -708,7 +715,7 @@ static int post_open_unix_sk(struct file_desc *d, int fd)
>  		return -1;
>  	}
>  
> -	if (restore_sk_queue(fd, peer->ue->id))
> +	if (peer->queuer == ui->ue->ino && restore_sk_queue(fd, peer->ue->id))

There are more places that call restore_sk_queue(). The ones in open_unixsk_pair_master
are OK, they are restoring queues for each other, but how about the call in
open_unixsk_standalone()? Should we patch it as well?

>  		return -1;
>  
>  	if (rst_file_params(fd, ui->ue->fown, ui->ue->flags))
> @@ -1021,6 +1028,7 @@ static int collect_one_unixsk(void *o, ProtobufCMessage *base)
>  		ui->name = NULL;
>  
>  	futex_init(&ui->prepared);
> +	ui->queuer = 0;
>  	ui->peer = NULL;
>  	ui->flags = 0;
>  	pr_info(" `- Got %#x peer %#x (name %s)\n",
> @@ -1063,6 +1071,8 @@ int resolve_unix_peers(void)
>  		}
>  
>  		ui->peer = peer;
> +		if (!ui->queuer)
> +			ui->queuer = ui->ue->ino;
>  		if (ui == peer)
>  			/* socket connected to self %) */
>  			continue;
>