This improves the slow-start phase by using the MSS
(as suggested in RFC 4342, sec. 5) instead of the packet size s.
Also figured out that __u32 is ample resource enough.
After applying, I got the following in the logs:
ccid3_hc_tx_packet_recv: client(f7421700), s=6, MSS=1424, w_init=4380,
R_sample=176us, X=24886363
Had the previous variant been used, w_init would have been as low as 24.
Committer note: removed unneeded cast to unsigned long long that was
causing a compiler warning on 64bit architectures.
Signed-off-by: Gerrit Renker <[EMAIL PROTECTED]>
Acked-by: Ian McDonald <[EMAIL PROTECTED]>
Signed-off-by: Arnaldo Carvalho de Melo <[EMAIL PROTECTED]>
---
net/dccp/ccids/ccid3.c | 10 +++++-----
1 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c
index 0009b39..52dae40 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -466,20 +466,20 @@ static void ccid3_hc_tx_packet_recv(struct sock *sk,
struct sk_buff *skb)
if (hctx->ccid3hctx_state == TFRC_SSTATE_NO_FBACK) {
/*
* Larger Initial Windows [RFC 4342, sec. 5]
- * We deviate in that we use `s' instead of `MSS'.
*/
- __u64 w_init = min(4 * hctx->ccid3hctx_s,
- max(2 * hctx->ccid3hctx_s, 4380));
+ __u32 w_init = min(4 * dp->dccps_mss_cache,
+ max(2 * dp->dccps_mss_cache, 4380U));
hctx->ccid3hctx_rtt = r_sample;
hctx->ccid3hctx_x = scaled_div(w_init << 6,
r_sample);
hctx->ccid3hctx_t_ld = now;
ccid3_update_send_time(hctx);
- ccid3_pr_debug("%s(%p), s=%u, w_init=%llu, "
+ ccid3_pr_debug("%s(%p), s=%u, MSS=%u, w_init=%u, "
"R_sample=%dus, X=%u\n", dccp_role(sk),
sk, hctx->ccid3hctx_s,
- (unsigned long long)w_init,
+ dp->dccps_mss_cache,
+ w_init,
(int)r_sample,
(unsigned)(hctx->ccid3hctx_x >> 6));
--
1.5.0.3
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
