diff options
author | Gerrit Renker <gerrit@erg.abdn.ac.uk> | 2009-02-27 17:38:28 -0500 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2009-03-02 06:07:23 -0500 |
commit | 361a5c1dd0bd7bb2b90e7fe9127b366d3566522e (patch) | |
tree | 440150dfa03f976d4ef6d3062744f2185373820d /net/dccp/output.c | |
parent | 9ce01461028d595a6f1cd724fbd7a0dd70464fe4 (diff) |
dccp: Minimise header option overhead in setting the MPS
This patch resolves a long-standing FIXME to dynamically update the Maximum
Packet Size depending on actual options usage.
It uses the flags set by the feature-negotiation infrastructure to compute
the required header option size.
Most options are fixed-size, a notable exception are Ack Vectors (required
currently only by CCID-2). These can have any length between 3 and 1020
bytes. As a result of testing, 16 bytes (2 bytes for type/length plus 14 Ack
Vector cells) have been found to be sufficient for loss-free situations.
There are currently no CCID-specific header options which may appear on data
packets, thus it is not necessary to define a corresponding CCID field as
suggested in the old comment.
Further changes:
----------------
Adjusted the type of 'cur_mps' to match the unsigned return type of the
function.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/dccp/output.c')
-rw-r--r-- | net/dccp/output.c | 22 |
1 files changed, 14 insertions, 8 deletions
diff --git a/net/dccp/output.c b/net/dccp/output.c index 22a618af4893..27c79bcc6a1e 100644 --- a/net/dccp/output.c +++ b/net/dccp/output.c | |||
@@ -161,21 +161,27 @@ unsigned int dccp_sync_mss(struct sock *sk, u32 pmtu) | |||
161 | struct inet_connection_sock *icsk = inet_csk(sk); | 161 | struct inet_connection_sock *icsk = inet_csk(sk); |
162 | struct dccp_sock *dp = dccp_sk(sk); | 162 | struct dccp_sock *dp = dccp_sk(sk); |
163 | u32 ccmps = dccp_determine_ccmps(dp); | 163 | u32 ccmps = dccp_determine_ccmps(dp); |
164 | int cur_mps = ccmps ? min(pmtu, ccmps) : pmtu; | 164 | u32 cur_mps = ccmps ? min(pmtu, ccmps) : pmtu; |
165 | 165 | ||
166 | /* Account for header lengths and IPv4/v6 option overhead */ | 166 | /* Account for header lengths and IPv4/v6 option overhead */ |
167 | cur_mps -= (icsk->icsk_af_ops->net_header_len + icsk->icsk_ext_hdr_len + | 167 | cur_mps -= (icsk->icsk_af_ops->net_header_len + icsk->icsk_ext_hdr_len + |
168 | sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext)); | 168 | sizeof(struct dccp_hdr) + sizeof(struct dccp_hdr_ext)); |
169 | 169 | ||
170 | /* | 170 | /* |
171 | * FIXME: this should come from the CCID infrastructure, where, say, | 171 | * Leave enough headroom for common DCCP header options. |
172 | * TFRC will say it wants TIMESTAMPS, ELAPSED time, etc, for now lets | 172 | * This only considers options which may appear on DCCP-Data packets, as |
173 | * put a rough estimate for NDP + TIMESTAMP + TIMESTAMP_ECHO + ELAPSED | 173 | * per table 3 in RFC 4340, 5.8. When running out of space for other |
174 | * TIME + TFRC_OPT_LOSS_EVENT_RATE + TFRC_OPT_RECEIVE_RATE + padding to | 174 | * options (eg. Ack Vector which can take up to 255 bytes), it is better |
175 | * make it a multiple of 4 | 175 | * to schedule a separate Ack. Thus we leave headroom for the following: |
176 | * - 1 byte for Slow Receiver (11.6) | ||
177 | * - 6 bytes for Timestamp (13.1) | ||
178 | * - 10 bytes for Timestamp Echo (13.3) | ||
179 | * - 8 bytes for NDP count (7.7, when activated) | ||
180 | * - 6 bytes for Data Checksum (9.3) | ||
181 | * - %DCCPAV_MIN_OPTLEN bytes for Ack Vector size (11.4, when enabled) | ||
176 | */ | 182 | */ |
177 | 183 | cur_mps -= roundup(1 + 6 + 10 + dp->dccps_send_ndp_count * 8 + 6 + | |
178 | cur_mps -= roundup(5 + 6 + 10 + 6 + 6 + 6, 4); | 184 | (dp->dccps_hc_rx_ackvec ? DCCPAV_MIN_OPTLEN : 0), 4); |
179 | 185 | ||
180 | /* And store cached results */ | 186 | /* And store cached results */ |
181 | icsk->icsk_pmtu_cookie = pmtu; | 187 | icsk->icsk_pmtu_cookie = pmtu; |