* [PATCH] tcp: fix retransmission in repair mode
@ 2012-11-15 14:03 Andrey Vagin
2012-11-15 14:10 ` Pavel Emelyanov
2012-11-15 15:03 ` Eric Dumazet
0 siblings, 2 replies; 4+ messages in thread
From: Andrey Vagin @ 2012-11-15 14:03 UTC (permalink / raw)
To: netdev
Cc: criu, linux-kernel, Andrew Vagin, Pavel Emelyanov,
David S. Miller, Alexey Kuznetsov, James Morris,
Hideaki YOSHIFUJI, Patrick McHardy
From: Andrew Vagin <avagin@openvz.org>
Currently if a socket was repaired with a few packet in a write queue,
a kernel bug may be triggered:
kernel BUG at net/ipv4/tcp_output.c:2330!
RIP: 0010:[<ffffffff8155784f>] tcp_retransmit_skb+0x5ff/0x610
According to the initial realization v3.4-rc2-963-gc0e88ff,
all skb-s should look like already posted. This patch fixes code
according with this sentence.
Here are three points, which were not done in the initial patch:
1. A tcp send head should not be changed
2. Initialize TSO state of a skb
3. Reset the retransmission time
This patch moves logic from tcp_sendmsg to tcp_write_xmit. A packet
passes the ussual way, but isn't sent to network. This patch solves
all described problems and handles tcp_sendpages.
Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: James Morris <jmorris@namei.org>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: Patrick McHardy <kaber@trash.net>
Signed-off-by: Andrey Vagin <avagin@openvz.org>
---
net/ipv4/tcp.c | 4 ++--
net/ipv4/tcp_output.c | 4 ++++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 197c000..083092e 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -1212,7 +1212,7 @@ new_segment:
wait_for_sndbuf:
set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
wait_for_memory:
- if (copied && likely(!tp->repair))
+ if (copied)
tcp_push(sk, flags & ~MSG_MORE, mss_now, TCP_NAGLE_PUSH);
if ((err = sk_stream_wait_memory(sk, &timeo)) != 0)
@@ -1223,7 +1223,7 @@ wait_for_memory:
}
out:
- if (copied && likely(!tp->repair))
+ if (copied)
tcp_push(sk, flags, mss_now, tp->nonagle);
release_sock(sk);
return copied + copied_syn;
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index cfe6ffe..2798706 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1986,6 +1986,9 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
tso_segs = tcp_init_tso_segs(sk, skb, mss_now);
BUG_ON(!tso_segs);
+ if (unlikely(tp->repair) && tp->repair_queue == TCP_SEND_QUEUE)
+ goto repair; /* Skip network transmission */
+
cwnd_quota = tcp_cwnd_test(tp, skb);
if (!cwnd_quota)
break;
@@ -2026,6 +2029,7 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
if (unlikely(tcp_transmit_skb(sk, skb, 1, gfp)))
break;
+repair:
/* Advance the send_head. This one is sent out.
* This call will increment packets_out.
*/
--
1.7.11.7
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH] tcp: fix retransmission in repair mode
2012-11-15 14:03 [PATCH] tcp: fix retransmission in repair mode Andrey Vagin
@ 2012-11-15 14:10 ` Pavel Emelyanov
2012-11-15 22:45 ` David Miller
2012-11-15 15:03 ` Eric Dumazet
1 sibling, 1 reply; 4+ messages in thread
From: Pavel Emelyanov @ 2012-11-15 14:10 UTC (permalink / raw)
To: Andrey Vagin, David S. Miller
Cc: netdev, criu, linux-kernel, Alexey Kuznetsov, James Morris,
Hideaki YOSHIFUJI, Patrick McHardy
On 11/15/2012 06:03 PM, Andrey Vagin wrote:
> From: Andrew Vagin <avagin@openvz.org>
>
> Currently if a socket was repaired with a few packet in a write queue,
> a kernel bug may be triggered:
>
> kernel BUG at net/ipv4/tcp_output.c:2330!
> RIP: 0010:[<ffffffff8155784f>] tcp_retransmit_skb+0x5ff/0x610
>
> According to the initial realization v3.4-rc2-963-gc0e88ff,
> all skb-s should look like already posted. This patch fixes code
> according with this sentence.
>
> Here are three points, which were not done in the initial patch:
> 1. A tcp send head should not be changed
> 2. Initialize TSO state of a skb
> 3. Reset the retransmission time
>
> This patch moves logic from tcp_sendmsg to tcp_write_xmit. A packet
> passes the ussual way, but isn't sent to network. This patch solves
> all described problems and handles tcp_sendpages.
>
> Cc: Pavel Emelyanov <xemul@parallels.com>
> Cc: "David S. Miller" <davem@davemloft.net>
> Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
> Cc: James Morris <jmorris@namei.org>
> Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
> Cc: Patrick McHardy <kaber@trash.net>
> Signed-off-by: Andrey Vagin <avagin@openvz.org>
Acked-by: Pavel Emelyanov <xemul@parallels.com>
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] tcp: fix retransmission in repair mode
2012-11-15 14:03 [PATCH] tcp: fix retransmission in repair mode Andrey Vagin
2012-11-15 14:10 ` Pavel Emelyanov
@ 2012-11-15 15:03 ` Eric Dumazet
1 sibling, 0 replies; 4+ messages in thread
From: Eric Dumazet @ 2012-11-15 15:03 UTC (permalink / raw)
To: Andrey Vagin
Cc: netdev, criu, linux-kernel, Pavel Emelyanov, David S. Miller,
Alexey Kuznetsov, James Morris, Hideaki YOSHIFUJI,
Patrick McHardy
On Thu, 2012-11-15 at 18:03 +0400, Andrey Vagin wrote:
> From: Andrew Vagin <avagin@openvz.org>
>
> Currently if a socket was repaired with a few packet in a write queue,
> a kernel bug may be triggered:
>
> kernel BUG at net/ipv4/tcp_output.c:2330!
> RIP: 0010:[<ffffffff8155784f>] tcp_retransmit_skb+0x5ff/0x610
>
> According to the initial realization v3.4-rc2-963-gc0e88ff,
> all skb-s should look like already posted. This patch fixes code
> according with this sentence.
>
> Here are three points, which were not done in the initial patch:
> 1. A tcp send head should not be changed
> 2. Initialize TSO state of a skb
> 3. Reset the retransmission time
>
> This patch moves logic from tcp_sendmsg to tcp_write_xmit. A packet
> passes the ussual way, but isn't sent to network. This patch solves
> all described problems and handles tcp_sendpages.
>
> Cc: Pavel Emelyanov <xemul@parallels.com>
> Cc: "David S. Miller" <davem@davemloft.net>
> Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
> Cc: James Morris <jmorris@namei.org>
> Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
> Cc: Patrick McHardy <kaber@trash.net>
> Signed-off-by: Andrey Vagin <avagin@openvz.org>
Any chance these tcp repair hacks could be done outside of tcp fast
path ?
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH] tcp: fix retransmission in repair mode
2012-11-15 14:10 ` Pavel Emelyanov
@ 2012-11-15 22:45 ` David Miller
0 siblings, 0 replies; 4+ messages in thread
From: David Miller @ 2012-11-15 22:45 UTC (permalink / raw)
To: xemul
Cc: avagin, netdev, criu, linux-kernel, kuznet, jmorris, yoshfuji, kaber
From: Pavel Emelyanov <xemul@parallels.com>
Date: Thu, 15 Nov 2012 18:10:59 +0400
> On 11/15/2012 06:03 PM, Andrey Vagin wrote:
>> From: Andrew Vagin <avagin@openvz.org>
>>
>> Currently if a socket was repaired with a few packet in a write queue,
>> a kernel bug may be triggered:
>>
>> kernel BUG at net/ipv4/tcp_output.c:2330!
>> RIP: 0010:[<ffffffff8155784f>] tcp_retransmit_skb+0x5ff/0x610
>>
>> According to the initial realization v3.4-rc2-963-gc0e88ff,
>> all skb-s should look like already posted. This patch fixes code
>> according with this sentence.
>>
>> Here are three points, which were not done in the initial patch:
>> 1. A tcp send head should not be changed
>> 2. Initialize TSO state of a skb
>> 3. Reset the retransmission time
>>
>> This patch moves logic from tcp_sendmsg to tcp_write_xmit. A packet
>> passes the ussual way, but isn't sent to network. This patch solves
>> all described problems and handles tcp_sendpages.
...
>> Signed-off-by: Andrey Vagin <avagin@openvz.org>
>
> Acked-by: Pavel Emelyanov <xemul@parallels.com>
Applied and queued up for -stable, thanks.
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2012-11-15 22:45 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-11-15 14:03 [PATCH] tcp: fix retransmission in repair mode Andrey Vagin
2012-11-15 14:10 ` Pavel Emelyanov
2012-11-15 22:45 ` David Miller
2012-11-15 15:03 ` Eric Dumazet
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).