[PATCH] tcp_cubic: fix the range of delayed_ack

[PATCH] tcp_cubic: fix the range of delayed_ack

[Liu Yu]

From: Liu Yu <allanyuliu@tencent.com>

commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
divide error) try to prevent divide error, but there is still a little
chance that delayed_ack can reach zero. In case the param cnt get
negative value, then ratio+cnt would overflow and may happen to be zero.
As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.

In some old kernels, such as 2.6.32, there is a bug that would
pass negative param, which then ultimately leads to this divide error.

commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
with skb MSS) fixed the negative param issue. However, 
it's safe that we fix the range of delayed_ack as well,
to make sure we do not hit a divide by zero.

CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Liu Yu <allanyuliu@tencent.com>
---
 net/ipv4/tcp_cubic.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/net/ipv4/tcp_cubic.c b/net/ipv4/tcp_cubic.c
index 8bf2245..b4f1b29 100644
--- a/net/ipv4/tcp_cubic.c
+++ b/net/ipv4/tcp_cubic.c
@@ -409,7 +409,7 @@ static void bictcp_acked(struct sock *sk, u32 cnt, s32 rtt_us)
 		ratio -= ca->delayed_ack >> ACK_RATIO_SHIFT;
 		ratio += cnt;
 
-		ca->delayed_ack = min(ratio, ACK_RATIO_LIMIT);
+		ca->delayed_ack = clamp(ratio, 1U, ACK_RATIO_LIMIT);
 	}
 
 	/* Some calls are for duplicates without timetamps */
-- 
1.7.1

Re: [PATCH] tcp_cubic: fix the range of delayed_ack

[Eric Dumazet]

On Wed, 2014-04-30 at 17:34 +0800, Liu Yu wrote:
> From: Liu Yu <allanyuliu@tencent.com>
> 
> commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
> divide error) try to prevent divide error, but there is still a little
> chance that delayed_ack can reach zero. In case the param cnt get
> negative value, then ratio+cnt would overflow and may happen to be zero.
> As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.
> 
> In some old kernels, such as 2.6.32, there is a bug that would
> pass negative param, which then ultimately leads to this divide error.
> 
> commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
> with skb MSS) fixed the negative param issue. However, 
> it's safe that we fix the range of delayed_ack as well,
> to make sure we do not hit a divide by zero.
> 
> CC: Stephen Hemminger <shemminger@vyatta.com>
> Signed-off-by: Liu Yu <allanyuliu@tencent.com>
> ---
>  net/ipv4/tcp_cubic.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)

Signed-off-by: Eric Dumazet <edumazet@google.com>

Thanks !

Re: [PATCH] tcp_cubic: fix the range of delayed_ack

[Neal Cardwell]

On Wed, Apr 30, 2014 at 10:18 AM, Eric Dumazet <eric.dumazet@gmail.com> wrote:
> On Wed, 2014-04-30 at 17:34 +0800, Liu Yu wrote:
>> From: Liu Yu <allanyuliu@tencent.com>
>>
>> commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
>> divide error) try to prevent divide error, but there is still a little
>> chance that delayed_ack can reach zero. In case the param cnt get
>> negative value, then ratio+cnt would overflow and may happen to be zero.
>> As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.
>>
>> In some old kernels, such as 2.6.32, there is a bug that would
>> pass negative param, which then ultimately leads to this divide error.
>>
>> commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
>> with skb MSS) fixed the negative param issue. However,
>> it's safe that we fix the range of delayed_ack as well,
>> to make sure we do not hit a divide by zero.
>>
>> CC: Stephen Hemminger <shemminger@vyatta.com>
>> Signed-off-by: Liu Yu <allanyuliu@tencent.com>
>> ---
>>  net/ipv4/tcp_cubic.c |    2 +-
>>  1 files changed, 1 insertions(+), 1 deletions(-)
>
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Acked-by: Neal Cardwell <ncardwell@google.com>

Indeed, thanks!

neal

Re: [PATCH] tcp_cubic: fix the range of delayed_ack

[David Miller]

From: Liu Yu <allanliuyu@gmail.com>
Date: Wed, 30 Apr 2014 17:34:09 +0800

> 
> From: Liu Yu <allanyuliu@tencent.com>
> 
> commit b9f47a3aaeab (tcp_cubic: limit delayed_ack ratio to prevent
> divide error) try to prevent divide error, but there is still a little
> chance that delayed_ack can reach zero. In case the param cnt get
> negative value, then ratio+cnt would overflow and may happen to be zero.
> As a result, min(ratio, ACK_RATIO_LIMIT) will calculate to be zero.
> 
> In some old kernels, such as 2.6.32, there is a bug that would
> pass negative param, which then ultimately leads to this divide error.
> 
> commit 5b35e1e6e9c (tcp: fix tcp_trim_head() to adjust segment count
> with skb MSS) fixed the negative param issue. However, 
> it's safe that we fix the range of delayed_ack as well,
> to make sure we do not hit a divide by zero.
> 
> CC: Stephen Hemminger <shemminger@vyatta.com>
> Signed-off-by: Liu Yu <allanyuliu@tencent.com>

Applied and queued up for -stable, thanks.