Re: Bad nsec conversion in svc_udp_recvfrom()

[Andrew Morton <akpm@osdl.org>]

Steve Dickson <SteveD@redhat.com> wrote:
>
> In patch-2.6.14-rc5 there is the following:
> @@ -584,13 +583,16 @@ svc_udp_recvfrom(struct svc_rqst *rqstp)
>          /* possibly an icmp error */
>          dprintk("svc: recvfrom returned error %d\n", -err);
>      }
> -   if (skb->stamp.tv_sec == 0) {
> -       skb->stamp.tv_sec = xtime.tv_sec;
> -       skb->stamp.tv_usec = xtime.tv_nsec / NSEC_PER_USEC;
> +   if (skb->tstamp.off_sec == 0) {
> +       struct timeval tv;
> +
> +       tv.tv_sec = xtime.tv_sec;
> +       tv.tv_usec = xtime.tv_nsec * 1000;
> +       skb_set_timestamp(skb, &tv);
>          /* Don't enable netstamp, sunrpc doesn't
>             need that much accuracy */
>      }
> -   svsk->sk_sk->sk_stamp = skb->stamp;
> +   skb_get_timestamp(skb, &svsk->sk_sk->sk_stamp);
>      set_bit(SK_DATA, &svsk->sk_flags); /* there may be more data... */
> 
>      /*
> Shouldn't tv.tv_usec = xtime.tv_nsec * 1000
> be tv.tv_usec = xtime.tv_nsec / 1000 or possible
> tv.tv_usec = xtime.tv_nsec / NSEC_PER_USEC ?
> 
> The was fixed by a previous patch
> (see http://lkml.org/lkml/2005/8/1/251)
> but now it seems to be broken again...
> 

Yes, that's screwed up - well spotted.

Patrick, please be more careful about these things.

I'll fix it up.