On Tue, Feb 25, 2014 at 1:22 AM, Eric Dumazet [off-list ref] wrote:

From: Eric Dumazet <edumazet@google.com>

Upcoming congestion controls for TCP require usec resolution for RTT
estimations. Millisecond resolution is simply not enough these days.

...

---
v6: Do not use ktime_get() but skb_mstamp, as Stephen Hemminger pointed
    out some platforms have slow ktime_get()

...

-                       ca_ops->pkts_acked(sk, pkts_acked, rtt_us);
-               }
-       } else if (skb && rtt_update && sack_rtt >= 0 &&
-                  sack_rtt > (s32)(now - TCP_SKB_CB(skb)->when)) {
+       } else if (skb && rtt_update && sack_rtt_us >= 0 &&
+                  sack_rtt_us > skb_mstamp_us_delta(&now, &skb->skb_mstamp)) {

This gets rid of the (s32) cast for the result of the time
subtraction. Is that safe?

The rest looks great to me!

neal