From: Arnaldo Carvalho de Melo <acme@redhat.com>
To: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org, dccp@vger.kernel.org,
Gerrit Renker <gerrit@erg.abdn.ac.uk>,
Ian McDonald <ian.mcdonald@jandi.co.nz>,
Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 4/5] [CCID3]: Nofeedback timer according to rfc3448bis
Date: Mon, 17 Dec 2007 13:00:18 -0200 [thread overview]
Message-ID: <1197903619-12301-5-git-send-email-acme@redhat.com> (raw)
In-Reply-To: <1197903619-12301-4-git-send-email-acme@redhat.com>
From: Gerrit Renker <gerrit@erg.abdn.ac.uk>
This implements the changes to the nofeedback timer handling suggested
in draft rfc3448bis00, section 4.4. In particular, these changes mean:
* better handling of the lossless case (p == 0)
* the timestamp for computing t_ld becomes obsolete
* much more recent document (RFC 3448 is almost 5 years old)
* concepts in rfc3448bis arose from a real, working implementation
(cf. sec. 12)
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
net/dccp/ccids/ccid3.c | 63 ++++++++++++++++++++++--------------------------
1 files changed, 29 insertions(+), 34 deletions(-)
diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c
index 7c8e9ad..00b5f11 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -131,12 +131,11 @@ static u32 ccid3_hc_tx_idle_rtt(struct ccid3_hc_tx_sock *hctx, ktime_t now)
*
*/
static void ccid3_hc_tx_update_x(struct sock *sk, ktime_t *stamp)
-
{
struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
__u64 min_rate = 2 * hctx->ccid3hctx_x_recv;
const __u64 old_x = hctx->ccid3hctx_x;
- ktime_t now = stamp? *stamp : ktime_get_real();
+ ktime_t now = stamp ? *stamp : ktime_get_real();
/*
* Handle IDLE periods: do not reduce below RFC3390 initial sending rate
@@ -230,27 +229,27 @@ static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
ccid3_pr_debug("%s(%p, state=%s) - entry \n", dccp_role(sk), sk,
ccid3_tx_state_name(hctx->ccid3hctx_state));
- switch (hctx->ccid3hctx_state) {
- case TFRC_SSTATE_NO_FBACK:
- /* RFC 3448, 4.4: Halve send rate directly */
+ if (hctx->ccid3hctx_state == TFRC_SSTATE_FBACK)
+ ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK);
+ else if (hctx->ccid3hctx_state != TFRC_SSTATE_NO_FBACK)
+ goto out;
+
+ /*
+ * Determine new allowed sending rate X as per draft rfc3448bis-00, 4.4
+ */
+ if (hctx->ccid3hctx_t_rto == 0 || /* no feedback received yet */
+ hctx->ccid3hctx_p == 0) {
+
+ /* halve send rate directly */
hctx->ccid3hctx_x = max(hctx->ccid3hctx_x / 2,
(((__u64)hctx->ccid3hctx_s) << 6) /
TFRC_T_MBI);
-
- ccid3_pr_debug("%s(%p, state=%s), updated tx rate to %u "
- "bytes/s\n", dccp_role(sk), sk,
- ccid3_tx_state_name(hctx->ccid3hctx_state),
- (unsigned)(hctx->ccid3hctx_x >> 6));
- /* The value of R is still undefined and so we can not recompute
- * the timout value. Keep initial value as per [RFC 4342, 5]. */
- t_nfb = TFRC_INITIAL_TIMEOUT;
ccid3_update_send_interval(hctx);
- break;
- case TFRC_SSTATE_FBACK:
+ } else {
/*
- * Modify the cached value of X_recv [RFC 3448, 4.4]
+ * Modify the cached value of X_recv
*
- * If (p == 0 || X_calc > 2 * X_recv)
+ * If (X_calc > 2 * X_recv)
* X_recv = max(X_recv / 2, s / (2 * t_mbi));
* Else
* X_recv = X_calc / 4;
@@ -259,32 +258,28 @@ static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
*/
BUG_ON(hctx->ccid3hctx_p && !hctx->ccid3hctx_x_calc);
- if (hctx->ccid3hctx_p == 0 ||
- (hctx->ccid3hctx_x_calc > (hctx->ccid3hctx_x_recv >> 5))) {
-
+ if (hctx->ccid3hctx_x_calc > (hctx->ccid3hctx_x_recv >> 5))
hctx->ccid3hctx_x_recv =
max(hctx->ccid3hctx_x_recv / 2,
(((__u64)hctx->ccid3hctx_s) << 6) /
(2 * TFRC_T_MBI));
- } else {
+ else {
hctx->ccid3hctx_x_recv = hctx->ccid3hctx_x_calc;
hctx->ccid3hctx_x_recv <<= 4;
}
- /* Now recalculate X [RFC 3448, 4.3, step (4)] */
ccid3_hc_tx_update_x(sk, NULL);
- /*
- * Schedule no feedback timer to expire in
- * max(t_RTO, 2 * s/X) = max(t_RTO, 2 * t_ipi)
- * See comments in packet_recv() regarding the value of t_RTO.
- */
- t_nfb = max(hctx->ccid3hctx_t_rto, 2 * hctx->ccid3hctx_t_ipi);
- break;
- case TFRC_SSTATE_NO_SENT:
- DCCP_BUG("%s(%p) - Illegal state NO_SENT", dccp_role(sk), sk);
- /* fall through */
- case TFRC_SSTATE_TERM:
- goto out;
}
+ ccid3_pr_debug("Reduced X to %llu/64 bytes/sec\n",
+ (unsigned long long)hctx->ccid3hctx_x);
+
+ /*
+ * Set new timeout for the nofeedback timer.
+ * See comments in packet_recv() regarding the value of t_RTO.
+ */
+ if (unlikely(hctx->ccid3hctx_t_rto == 0)) /* no feedback yet */
+ t_nfb = TFRC_INITIAL_TIMEOUT;
+ else
+ t_nfb = max(hctx->ccid3hctx_t_rto, 2 * hctx->ccid3hctx_t_ipi);
restart_timer:
sk_reset_timer(sk, &hctx->ccid3hctx_no_feedback_timer,
--
1.5.3.6
next prev parent reply other threads:[~2007-12-17 15:00 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-12-17 15:00 [PATCHES 0/5]: DCCP patches for 2.6.25 Arnaldo Carvalho de Melo
2007-12-17 15:00 ` [PATCH 1/5] [CCID3]: Use a function to update p_inv, and p is never used Arnaldo Carvalho de Melo
2007-12-17 15:00 ` [PATCH 2/5] [CCID3]: Remove two irrelevant states in TX feedback handling Arnaldo Carvalho de Melo
2007-12-17 15:00 ` [PATCH 3/5] [CCID3]: Implement rfc3448bis changes to feedback reception Arnaldo Carvalho de Melo
2007-12-17 15:00 ` Arnaldo Carvalho de Melo [this message]
2007-12-17 15:00 ` [PATCH 5/5] [DCCP]: Remove unused inline function Arnaldo Carvalho de Melo
2007-12-18 7:02 ` [PATCHES 0/5]: DCCP patches for 2.6.25 David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1197903619-12301-5-git-send-email-acme@redhat.com \
--to=acme@redhat.com \
--cc=davem@davemloft.net \
--cc=dccp@vger.kernel.org \
--cc=gerrit@erg.abdn.ac.uk \
--cc=ian.mcdonald@jandi.co.nz \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox