Patch Detail
get:
Show a patch.
patch:
Update a patch.
put:
Update a patch.
GET /api/patches/748/?format=api
{ "id": 748, "url": "http://patchwork.ozlabs.org/api/patches/748/?format=api", "web_url": "http://patchwork.ozlabs.org/project/netdev/patch/1221947278-16715-5-git-send-email-ilpo.jarvinen@helsinki.fi/", "project": { "id": 7, "url": "http://patchwork.ozlabs.org/api/projects/7/?format=api", "name": "Linux network development", "link_name": "netdev", "list_id": "netdev.vger.kernel.org", "list_email": "netdev@vger.kernel.org", "web_url": null, "scm_url": null, "webscm_url": null, "list_archive_url": "", "list_archive_url_format": "", "commit_url_format": "" }, "msgid": "<1221947278-16715-5-git-send-email-ilpo.jarvinen@helsinki.fi>", "list_archive_url": null, "date": "2008-09-20T21:47:47", "name": "[PATCHv2,net-next,04/15] tcp: convert retransmit_cnt_hint to seqno", "commit_ref": null, "pull_url": null, "state": "accepted", "archived": true, "hash": "9561ba2876f1fb2ba00be564a3ba53d1f81962a4", "submitter": { "id": 255, "url": "http://patchwork.ozlabs.org/api/people/255/?format=api", "name": "Ilpo Järvinen", "email": "ilpo.jarvinen@helsinki.fi" }, "delegate": { "id": 34, "url": "http://patchwork.ozlabs.org/api/users/34/?format=api", "username": "davem", "first_name": "David", "last_name": "Miller", "email": "davem@davemloft.net" }, "mbox": "http://patchwork.ozlabs.org/project/netdev/patch/1221947278-16715-5-git-send-email-ilpo.jarvinen@helsinki.fi/mbox/", "series": [], "comments": "http://patchwork.ozlabs.org/api/patches/748/comments/", "check": "pending", "checks": "http://patchwork.ozlabs.org/api/patches/748/checks/", "tags": {}, "related": [], "headers": { "Return-Path": "<netdev-owner@vger.kernel.org>", "X-Original-To": "patchwork-incoming@ozlabs.org", "Delivered-To": "patchwork-incoming@ozlabs.org", "Received": [ "from vger.kernel.org (vger.kernel.org [209.132.176.167])\n\tby ozlabs.org (Postfix) with ESMTP id B9106DDE24\n\tfor <patchwork-incoming@ozlabs.org>;\n\tSun, 21 Sep 2008 07:48:46 +1000 (EST)", "(majordomo@vger.kernel.org) by vger.kernel.org via listexpand\n\tid S1752611AbYITVsN (ORCPT <rfc822;patchwork-incoming@ozlabs.org>);\n\tSat, 20 Sep 2008 17:48:13 -0400", "(majordomo@vger.kernel.org) by vger.kernel.org id S1752112AbYITVsL\n\t(ORCPT <rfc822; netdev-outgoing>); Sat, 20 Sep 2008 17:48:11 -0400", "from courier.cs.helsinki.fi ([128.214.9.1]:36807 \"EHLO\n\tmail.cs.helsinki.fi\" rhost-flags-OK-OK-OK-OK) by vger.kernel.org\n\twith ESMTP id S1751576AbYITVsC (ORCPT\n\t<rfc822;netdev@vger.kernel.org>); Sat, 20 Sep 2008 17:48:02 -0400", "from wrl-59.cs.helsinki.fi (wrl-59.cs.helsinki.fi\n\t[128.214.166.179])\n\t(AUTH: PLAIN cs-relay, TLS: TLSv1/SSLv3,256bits,AES256-SHA)\n\tby mail.cs.helsinki.fi with esmtp; Sun, 21 Sep 2008 00:47:59 +0300\n\tid 0005BEDE.48D56F8F.00004FF8", "by wrl-59.cs.helsinki.fi (Postfix, from userid 50795)\n\tid A93A2A00A5; Sun, 21 Sep 2008 00:47:58 +0300 (EEST)" ], "From": "\"=?ISO-8859-1?Q?Ilpo_J=E4rvinen?=\" <ilpo.jarvinen@helsinki.fi>", "To": "David Miller <davem@davemloft.net>", "Cc": "netdev@vger.kernel.org,\n\t\"=?utf-8?q?Ilpo=20J=E4rvinen?=\" <ilpo.jarvinen@helsinki.fi>", "Subject": "[PATCHv2 net-next 04/15] tcp: convert retransmit_cnt_hint to seqno", "Date": "Sun, 21 Sep 2008 00:47:47 +0300", "Message-Id": "<1221947278-16715-5-git-send-email-ilpo.jarvinen@helsinki.fi>", "X-Mailer": "git-send-email 1.5.4.2.156.ge3c5", "In-Reply-To": "<1221947278-16715-4-git-send-email-ilpo.jarvinen@helsinki.fi>", "References": "<1221947278-16715-1-git-send-email-ilpo.jarvinen@helsinki.fi>\n\t<1221947278-16715-2-git-send-email-ilpo.jarvinen@helsinki.fi>\n\t<1221947278-16715-3-git-send-email-ilpo.jarvinen@helsinki.fi>\n\t<1221947278-16715-4-git-send-email-ilpo.jarvinen@helsinki.fi>", "MIME-Version": "1.0", "Content-Type": "text/plain; charset=ISO-8859-1", "Content-Transfer-Encoding": "8bit", "Sender": "netdev-owner@vger.kernel.org", "Precedence": "bulk", "List-ID": "<netdev.vger.kernel.org>", "X-Mailing-List": "netdev@vger.kernel.org" }, "content": "Main benefit in this is that we can then freely point\nthe retransmit_skb_hint to anywhere we want to because\nthere's no longer need to know what would be the count\nchanges involve, and since this is really used only as a\nterminator, unnecessary work is one time walk at most,\nand if some retransmissions are necessary after that\npoint later on, the walk is not full waste of time\nanyway.\n\nSince retransmit_high must be kept valid, all lost\nmarkers must ensure that.\n\nNow I also have learned how those \"holes\" in the\nrexmittable skbs can appear, mtu probe does them. So\nI removed the misleading comment as well.\n\nSigned-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>\n---\n include/linux/tcp.h | 2 +-\n include/net/tcp.h | 2 ++\n net/ipv4/tcp_input.c | 34 ++++++++++++++++++++--------------\n net/ipv4/tcp_output.c | 25 +++++++------------------\n 4 files changed, 30 insertions(+), 33 deletions(-)", "diff": "diff --git a/include/linux/tcp.h b/include/linux/tcp.h\nindex 2e25573..d7637c4 100644\n--- a/include/linux/tcp.h\n+++ b/include/linux/tcp.h\n@@ -358,7 +358,7 @@ struct tcp_sock {\n \t\t\t\t\t */\n \n \tint lost_cnt_hint;\n-\tint retransmit_cnt_hint;\n+\tu32 retransmit_high;\t/* L-bits may be on up to this seqno */\n \n \tu32\tlost_retrans_low;\t/* Sent seq after any rxmit (lowest) */\n \ndiff --git a/include/net/tcp.h b/include/net/tcp.h\nindex b716763..d0e90c5 100644\n--- a/include/net/tcp.h\n+++ b/include/net/tcp.h\n@@ -472,6 +472,8 @@ extern void tcp_send_delayed_ack(struct sock *sk);\n \n /* tcp_input.c */\n extern void tcp_cwnd_application_limited(struct sock *sk);\n+extern void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp,\n+\t\t\t\t\t struct sk_buff *skb);\n \n /* tcp_timer.c */\n extern void tcp_init_xmit_timers(struct sock *);\ndiff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c\nindex 1251233..d271cc8 100644\n--- a/net/ipv4/tcp_input.c\n+++ b/net/ipv4/tcp_input.c\n@@ -979,17 +979,17 @@ static void tcp_update_reordering(struct sock *sk, const int metric,\n \t}\n }\n \n-/* RFC: This is from the original, I doubt that this is necessary at all:\n- * clear xmit_retrans hint if seq of this skb is beyond hint. How could we\n- * retransmitted past LOST markings in the first place? I'm not fully sure\n- * about undo and end of connection cases, which can cause R without L?\n- */\n+/* This must be called before lost_out is incremented */\n static void tcp_verify_retransmit_hint(struct tcp_sock *tp, struct sk_buff *skb)\n {\n-\tif ((tp->retransmit_skb_hint != NULL) &&\n+\tif ((tp->retransmit_skb_hint == NULL) ||\n \t before(TCP_SKB_CB(skb)->seq,\n \t\t TCP_SKB_CB(tp->retransmit_skb_hint)->seq))\n-\t\ttp->retransmit_skb_hint = NULL;\n+\t\ttp->retransmit_skb_hint = skb;\n+\n+\tif (!tp->lost_out ||\n+\t after(TCP_SKB_CB(skb)->end_seq, tp->retransmit_high))\n+\t\ttp->retransmit_high = TCP_SKB_CB(skb)->end_seq;\n }\n \n static void tcp_skb_mark_lost(struct tcp_sock *tp, struct sk_buff *skb)\n@@ -1002,6 +1002,16 @@ static void tcp_skb_mark_lost(struct tcp_sock *tp, struct sk_buff *skb)\n \t}\n }\n \n+void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb)\n+{\n+\ttcp_verify_retransmit_hint(tp, skb);\n+\n+\tif (!(TCP_SKB_CB(skb)->sacked & (TCPCB_LOST|TCPCB_SACKED_ACKED))) {\n+\t\ttp->lost_out += tcp_skb_pcount(skb);\n+\t\tTCP_SKB_CB(skb)->sacked |= TCPCB_LOST;\n+\t}\n+}\n+\n /* This procedure tags the retransmission queue when SACKs arrive.\n *\n * We have three tag bits: SACKED(S), RETRANS(R) and LOST(L).\n@@ -1178,13 +1188,7 @@ static void tcp_mark_lost_retrans(struct sock *sk)\n \t\t\tTCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;\n \t\t\ttp->retrans_out -= tcp_skb_pcount(skb);\n \n-\t\t\t/* clear lost hint */\n-\t\t\ttp->retransmit_skb_hint = NULL;\n-\n-\t\t\tif (!(TCP_SKB_CB(skb)->sacked & (TCPCB_LOST|TCPCB_SACKED_ACKED))) {\n-\t\t\t\ttp->lost_out += tcp_skb_pcount(skb);\n-\t\t\t\tTCP_SKB_CB(skb)->sacked |= TCPCB_LOST;\n-\t\t\t}\n+\t\t\ttcp_skb_mark_lost_uncond_verify(tp, skb);\n \t\t\tNET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPLOSTRETRANSMIT);\n \t\t} else {\n \t\t\tif (before(ack_seq, new_low_seq))\n@@ -1890,6 +1894,7 @@ static void tcp_enter_frto_loss(struct sock *sk, int allowed_segments, int flag)\n \t\tif (!(TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) {\n \t\t\tTCP_SKB_CB(skb)->sacked |= TCPCB_LOST;\n \t\t\ttp->lost_out += tcp_skb_pcount(skb);\n+\t\t\ttp->retransmit_high = TCP_SKB_CB(skb)->end_seq;\n \t\t}\n \t}\n \ttcp_verify_left_out(tp);\n@@ -1974,6 +1979,7 @@ void tcp_enter_loss(struct sock *sk, int how)\n \t\t\tTCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_ACKED;\n \t\t\tTCP_SKB_CB(skb)->sacked |= TCPCB_LOST;\n \t\t\ttp->lost_out += tcp_skb_pcount(skb);\n+\t\t\ttp->retransmit_high = TCP_SKB_CB(skb)->end_seq;\n \t\t}\n \t}\n \ttcp_verify_left_out(tp);\ndiff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c\nindex 2242c2c..bb30e68 100644\n--- a/net/ipv4/tcp_output.c\n+++ b/net/ipv4/tcp_output.c\n@@ -1836,7 +1836,7 @@ void tcp_simple_retransmit(struct sock *sk)\n \tstruct tcp_sock *tp = tcp_sk(sk);\n \tstruct sk_buff *skb;\n \tunsigned int mss = tcp_current_mss(sk, 0);\n-\tint lost = 0;\n+\tu32 prior_lost = tp->lost_out;\n \n \ttcp_for_write_queue(skb, sk) {\n \t\tif (skb == tcp_send_head(sk))\n@@ -1847,17 +1847,13 @@ void tcp_simple_retransmit(struct sock *sk)\n \t\t\t\tTCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;\n \t\t\t\ttp->retrans_out -= tcp_skb_pcount(skb);\n \t\t\t}\n-\t\t\tif (!(TCP_SKB_CB(skb)->sacked & TCPCB_LOST)) {\n-\t\t\t\tTCP_SKB_CB(skb)->sacked |= TCPCB_LOST;\n-\t\t\t\ttp->lost_out += tcp_skb_pcount(skb);\n-\t\t\t\tlost = 1;\n-\t\t\t}\n+\t\t\ttcp_skb_mark_lost_uncond_verify(tp, skb);\n \t\t}\n \t}\n \n \ttcp_clear_all_retrans_hints(tp);\n \n-\tif (!lost)\n+\tif (prior_lost == tp->lost_out)\n \t\treturn;\n \n \tif (tcp_is_reno(tp))\n@@ -2007,15 +2003,11 @@ void tcp_xmit_retransmit_queue(struct sock *sk)\n \tconst struct inet_connection_sock *icsk = inet_csk(sk);\n \tstruct tcp_sock *tp = tcp_sk(sk);\n \tstruct sk_buff *skb;\n-\tint packet_cnt;\n \n-\tif (tp->retransmit_skb_hint) {\n+\tif (tp->retransmit_skb_hint)\n \t\tskb = tp->retransmit_skb_hint;\n-\t\tpacket_cnt = tp->retransmit_cnt_hint;\n-\t} else {\n+\telse\n \t\tskb = tcp_write_queue_head(sk);\n-\t\tpacket_cnt = 0;\n-\t}\n \n \t/* First pass: retransmit lost packets. */\n \tif (tp->lost_out) {\n@@ -2026,7 +2018,6 @@ void tcp_xmit_retransmit_queue(struct sock *sk)\n \t\t\t\tbreak;\n \t\t\t/* we could do better than to assign each time */\n \t\t\ttp->retransmit_skb_hint = skb;\n-\t\t\ttp->retransmit_cnt_hint = packet_cnt;\n \n \t\t\t/* Assume this retransmit will generate\n \t\t\t * only one packet for congestion window\n@@ -2037,6 +2028,8 @@ void tcp_xmit_retransmit_queue(struct sock *sk)\n \t\t\t */\n \t\t\tif (tcp_packets_in_flight(tp) >= tp->snd_cwnd)\n \t\t\t\treturn;\n+\t\t\tif (!before(TCP_SKB_CB(skb)->seq, tp->retransmit_high))\n+\t\t\t\tbreak;\n \n \t\t\tif (sacked & TCPCB_LOST) {\n \t\t\t\tif (!(sacked & (TCPCB_SACKED_ACKED|TCPCB_SACKED_RETRANS))) {\n@@ -2057,10 +2050,6 @@ void tcp_xmit_retransmit_queue(struct sock *sk)\n \t\t\t\t\t\t\t\t\t inet_csk(sk)->icsk_rto,\n \t\t\t\t\t\t\t\t\t TCP_RTO_MAX);\n \t\t\t\t}\n-\n-\t\t\t\tpacket_cnt += tcp_skb_pcount(skb);\n-\t\t\t\tif (packet_cnt >= tp->lost_out)\n-\t\t\t\t\tbreak;\n \t\t\t}\n \t\t}\n \t}\n", "prefixes": [ "PATCHv2", "net-next", "04/15" ] }