[Spice-devel] [spice-server 4/4] worker: Simplify flush_commands()

[Frediano Ziglio]

> 
> red_disconnect_display() is duplicating what red_channel_disconnect()
> already does, so red_disconnect_display() and red_disconnect_cursor()
> are actually identical code-wise. We can directly call
> red_channel_disconnect() from flush_commands() rather than passing a
> 'red_disconnect_t disconnect' argument to that function.
> 
> Signed-off-by: Christophe Fergeau <cfergeau at redhat.com>
> ---
>  server/red-worker.c | 28 +++++-----------------------
>  1 file changed, 5 insertions(+), 23 deletions(-)
> 
> diff --git a/server/red-worker.c b/server/red-worker.c
> index 109db63e0..fc009454c 100644
> --- a/server/red-worker.c
> +++ b/server/red-worker.c
> @@ -284,17 +284,6 @@ static bool red_process_is_blocked(RedWorker *worker)
>             red_channel_max_pipe_size(RED_CHANNEL(worker->display_channel)) >
>             MAX_PIPE_SIZE;
>  }
>  
> -static void red_disconnect_display(RedWorker *worker)
> -{
> -    spice_warning("update timeout");
> -
> -    // TODO: we need to record the client that actually causes the timeout.
> So
> -    // we need to check the locations of the various pipe heads when
> counting,
> -    // and disconnect only those/that.

I would personally keep this comment and put before the spice_warning below

> -    red_channel_apply_clients(RED_CHANNEL(worker->display_channel),
> -                              red_channel_client_disconnect);
> -}
> -
>  static void red_migrate_display(DisplayChannel *display, RedChannelClient
>  *rcc)
>  {
>      /* We need to stop the streams, and to send upgrade_items to the client.
> @@ -314,10 +303,8 @@ static void red_migrate_display(DisplayChannel *display,
> RedChannelClient *rcc)
>  }
>  
>  typedef int (*red_process_t)(RedWorker *worker, int *ring_is_empty);
> -typedef void (*red_disconnect_t)(RedWorker *worker);
> -
>  static void flush_commands(RedWorker *worker, RedChannel *red_channel,
> -                           red_process_t process, red_disconnect_t
> disconnect)
> +                           red_process_t process)
>  {
>      for (;;) {
>          uint64_t end_time;
> @@ -346,7 +333,8 @@ static void flush_commands(RedWorker *worker, RedChannel
> *red_channel,
>              // TODO: MC: the whole timeout will break since it takes lowest
>              timeout, should
>              // do it client by client.
>              if (spice_get_monotonic_time_ns() >= end_time) {
> -                disconnect(worker);
> +		spice_warning("flush timeout");

this is indented with tab. I would put the old comment before this warning.

> +                red_channel_disconnect(red_channel);
>              } else {
>                  usleep(DISPLAY_CLIENT_RETRY_INTERVAL);
>              }
> @@ -357,19 +345,13 @@ static void flush_commands(RedWorker *worker,
> RedChannel *red_channel,
>  static void flush_display_commands(RedWorker *worker)
>  {
>      flush_commands(worker, RED_CHANNEL(worker->display_channel),
> -                   red_process_display, red_disconnect_display);
> -}
> -
> -static void red_disconnect_cursor(RedWorker *worker)
> -{
> -    spice_warning("flush cursor timeout");
> -    red_channel_disconnect(RED_CHANNEL(worker->cursor_channel));
> +                   red_process_display);
>  }
>  
>  static void flush_cursor_commands(RedWorker *worker)
>  {
>      flush_commands(worker, RED_CHANNEL(worker->cursor_channel),
> -                   red_process_cursor, red_disconnect_cursor);
> +                   red_process_cursor);
>  }
>  
>  // TODO: on timeout, don't disconnect all channels immediatly - try to

OT: typo in comment

>  disconnect the slowest ones

Beside these small comments I would ack the entire series.

Frediano