Re: [Qemu-devel] [RFC PATCH v4 04/15] net: resolve race of tap backend and its peer

[Stefan Hajnoczi]

On Wed, Apr 17, 2013 at 04:39:13PM +0800, Liu Ping Fan wrote:
> From: Liu Ping Fan <address@hidden>
> 
> When vhost net enabled, we should be sure that the user space
> fd handler is not in flight
> 
> Signed-off-by: Liu Ping Fan <address@hidden>
> ---
>  net/tap.c |    5 +++++
>  1 files changed, 5 insertions(+), 0 deletions(-)
> 
> diff --git a/net/tap.c b/net/tap.c
> index 35cbb6e..b5629e3 100644
> --- a/net/tap.c
> +++ b/net/tap.c
> @@ -41,6 +41,7 @@
>  #include "qemu/error-report.h"
>  
>  #include "net/tap.h"
> +#include "util/event_gsource.h"
>  
>  #include "hw/vhost_net.h"
>  
> @@ -327,6 +328,10 @@ static void tap_poll(NetClientState *nc, bool enable)
>      tap_read_poll(s, enable);
>      tap_write_poll(s, enable);
>  
> +    if (!enable) {
> +        /* need sync so vhost can take over polling */
> +        g_source_remove_poll(&s->nsrc->source, &s->nsrc->gfd);
> +    }

We must also re-enable it when .poll(nc, true) is called.

Please drop the comments the previous patch added.  In fact, this patch
can be squashed.

I think there was another vhost sync comment which you haven't addressed
here.  See the previous patch.

Stefan