[TFRC]: Let calling module compute First Loss Interval; convert to ktime_t
TFRC [RFC 3448] requires a synthetic value of the first loss interval length.
Since computing this value requires access to the internals of the calling
module (parameters s, RTT, and X_recv), it is better to let the calling module
compute the First Loss Interval length.
Therefore, to highlight this, the function has been renamed into
`ccid3_first_li' (due to a max line length of 80 characters); the code has
been arranged and comments tidied up, but no functional change results.
Secondly, the patch implements the remaining changes to support the newer
timeofday/ktime_t interface (reduces to a one-liner).
Signed-off-by: Gerrit Renker <gerrit@xxxxxxxxxxxxxx>
---
net/dccp/ccids/ccid3.c | 34 +++++++++++-----------------------
1 file changed, 11 insertions(+), 23 deletions(-)
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -730,39 +730,27 @@ static int ccid3_hc_rx_insert_options(st
return 0;
}
-/* calculate first loss interval
+/** ccid3_first_li - Implements [RFC 3448, 6.3.1]
*
- * returns estimated loss interval in usecs */
-
-static u32 ccid3_hc_rx_calc_first_li(struct sock *sk)
+ * Determine the length of the first loss interval via inverse lookup.
+ * Assume that X_recv can be computed by the throughput equation
+ * s
+ * X_recv = --------
+ * R * fval
+ * Find some p such that f(p) = fval; return 1/p (scaled).
+ */
+static u32 ccid3_first_li(struct sock *sk)
{
struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
- u32 x_recv, p;
- suseconds_t delta;
- struct timeval tstamp;
+ u32 x_recv, p, delta;
u64 fval;
- /*
- * Determine the length of the first loss interval via inverse lookup.
- * Assume that X_recv can be computed by the throughput equation
- * s
- * X_recv = --------
- * R * fval
- * Find some p such that f(p) = fval; return 1/p [RFC 3448, 6.3.1].
- */
if (hcrx->ccid3hcrx_rtt == 0) {
DCCP_WARN("No RTT estimate available, using fallback RTT\n");
hcrx->ccid3hcrx_rtt = DCCP_FALLBACK_RTT;
}
-/* XXX this also has a forward dependency, which is resolved in the patch
- * that does receiver RTT sampling. The problem with the code below is
- * that it still used the old, struct timeval based, interface, and that
- * the name of the variable `_last_feedback' is not yet adjusted.
- dccp_timestamp(sk, &tstamp);
- delta = timeval_delta(&tstamp, &hcrx->ccid3hcrx_tstamp_last_feedback);
- DCCP_BUG_ON(delta <= 0);
-*/
+ delta = ktime_to_us(net_timedelta(hcrx->ccid3hcrx_last_feedback));
x_recv = scaled_div32(hcrx->ccid3hcrx_bytes_recv, delta);
if (x_recv == 0) { /* would also trigger divide-by-zero */
DCCP_WARN("X_recv==0\n");
-
To unsubscribe from this list: send the line "unsubscribe dccp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
