netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr)
@ 2012-02-28  8:28 roy.qing.li
  2012-02-28  8:28 ` [PATCH 2/2] ipv4: fix mss comparison roy.qing.li
  2012-02-28  9:56 ` [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) Eric Dumazet
  0 siblings, 2 replies; 6+ messages in thread
From: roy.qing.li @ 2012-02-28  8:28 UTC (permalink / raw)
  To: netdev

From: RongQing.Li <roy.qing.li@gmail.com>

When mss is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr),
set mss to IPV6_MAXPLEN - sizeof(struct tcphdr), not IPV6_MAXPLEN.

Signed-off-by: RongQing.Li <roy.qing.li@gmail.com>
---
 net/ipv6/route.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index 92be12b..42cc16c 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -1054,7 +1054,7 @@ static unsigned int ip6_default_advmss(const struct dst_entry *dst)
 	 * rely only on pmtu discovery"
 	 */
 	if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
-		mtu = IPV6_MAXPLEN;
+		mtu = IPV6_MAXPLEN - sizeof(struct tcphdr);
 	return mtu;
 }
 
-- 
1.7.1

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* [PATCH 2/2] ipv4: fix mss comparison
  2012-02-28  8:28 [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) roy.qing.li
@ 2012-02-28  8:28 ` roy.qing.li
  2012-02-28  9:34   ` Francois Romieu
  2012-02-28 10:01   ` Eric Dumazet
  2012-02-28  9:56 ` [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) Eric Dumazet
  1 sibling, 2 replies; 6+ messages in thread
From: roy.qing.li @ 2012-02-28  8:28 UTC (permalink / raw)
  To: netdev

From: RongQing.Li <roy.qing.li@gmail.com>

mss should compare with 65535 - sizeof(struct tcphdr),  
the largest mss is 65535 - sizeof(struct tcphdr).  
40 is sum of sizeof(struct tcphdr) and sizeof(struct iphdr)

Signed-off-by: RongQing.Li <roy.qing.li@gmail.com>
---
 net/ipv4/route.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 0489ced..c72f765 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1914,8 +1914,8 @@ static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
 	if (advmss == 0) {
 		advmss = max_t(unsigned int, dst->dev->mtu - 40,
 			       ip_rt_min_advmss);
-		if (advmss > 65535 - 40)
-			advmss = 65535 - 40;
+		if (advmss > 65535 - sizeof(struct tcphdr))
+			advmss = 65535 - sizeof(struct tcphdr);
 	}
 	return advmss;
 }
-- 
1.7.1

^ permalink raw reply related	[flat|nested] 6+ messages in thread

* Re: [PATCH 2/2] ipv4: fix mss comparison
  2012-02-28  8:28 ` [PATCH 2/2] ipv4: fix mss comparison roy.qing.li
@ 2012-02-28  9:34   ` Francois Romieu
  2012-02-28  9:55     ` RongQing Li
  2012-02-28 10:01   ` Eric Dumazet
  1 sibling, 1 reply; 6+ messages in thread
From: Francois Romieu @ 2012-02-28  9:34 UTC (permalink / raw)
  To: roy.qing.li; +Cc: netdev

roy.qing.li@gmail.com <roy.qing.li@gmail.com> :
[...]
> diff --git a/net/ipv4/route.c b/net/ipv4/route.c
> index 0489ced..c72f765 100644
> --- a/net/ipv4/route.c
> +++ b/net/ipv4/route.c
> @@ -1914,8 +1914,8 @@ static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
>  	if (advmss == 0) {
>  		advmss = max_t(unsigned int, dst->dev->mtu - 40,
>  			       ip_rt_min_advmss);
> -		if (advmss > 65535 - 40)
> -			advmss = 65535 - 40;
> +		if (advmss > 65535 - sizeof(struct tcphdr))
> +			advmss = 65535 - sizeof(struct tcphdr);

What would prevent net/ipv4/tcp_output.c::tcp_advertise_mss returning
a value beyond 65535 - 40 ?

-- 
Ueimor

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 2/2] ipv4: fix mss comparison
  2012-02-28  9:34   ` Francois Romieu
@ 2012-02-28  9:55     ` RongQing Li
  0 siblings, 0 replies; 6+ messages in thread
From: RongQing Li @ 2012-02-28  9:55 UTC (permalink / raw)
  To: Francois Romieu; +Cc: netdev

2012/2/28 Francois Romieu <romieu@fr.zoreil.com>:
> roy.qing.li@gmail.com <roy.qing.li@gmail.com> :
> [...]
>> diff --git a/net/ipv4/route.c b/net/ipv4/route.c
>> index 0489ced..c72f765 100644
>> --- a/net/ipv4/route.c
>> +++ b/net/ipv4/route.c
>> @@ -1914,8 +1914,8 @@ static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
>>       if (advmss == 0) {
>>               advmss = max_t(unsigned int, dst->dev->mtu - 40,
>>                              ip_rt_min_advmss);
>> -             if (advmss > 65535 - 40)
>> -                     advmss = 65535 - 40;
>> +             if (advmss > 65535 - sizeof(struct tcphdr))
>> +                     advmss = 65535 - sizeof(struct tcphdr);
>
> What would prevent net/ipv4/tcp_output.c::tcp_advertise_mss returning
> a value beyond 65535 - 40 ?
>

You are right, Please ignore this patch.
The biggest MSS should be (Total Length of ipv4 header -
sizeof(iphdr)- sizeof(tcphdr))

I mix up the Total Length of ipv4 header and Payload Length of ipv6 header.

> --
> Ueimor

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr)
  2012-02-28  8:28 [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) roy.qing.li
  2012-02-28  8:28 ` [PATCH 2/2] ipv4: fix mss comparison roy.qing.li
@ 2012-02-28  9:56 ` Eric Dumazet
  1 sibling, 0 replies; 6+ messages in thread
From: Eric Dumazet @ 2012-02-28  9:56 UTC (permalink / raw)
  To: roy.qing.li; +Cc: netdev

Le mardi 28 février 2012 à 16:28 +0800, roy.qing.li@gmail.com a écrit :
> From: RongQing.Li <roy.qing.li@gmail.com>
> 
> When mss is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr),
> set mss to IPV6_MAXPLEN - sizeof(struct tcphdr), not IPV6_MAXPLEN.
> 
> Signed-off-by: RongQing.Li <roy.qing.li@gmail.com>
> ---
>  net/ipv6/route.c |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/net/ipv6/route.c b/net/ipv6/route.c
> index 92be12b..42cc16c 100644
> --- a/net/ipv6/route.c
> +++ b/net/ipv6/route.c
> @@ -1054,7 +1054,7 @@ static unsigned int ip6_default_advmss(const struct dst_entry *dst)
>  	 * rely only on pmtu discovery"
>  	 */
>  	if (mtu > IPV6_MAXPLEN - sizeof(struct tcphdr))
> -		mtu = IPV6_MAXPLEN;
> +		mtu = IPV6_MAXPLEN - sizeof(struct tcphdr);
>  	return mtu;
>  }
>  

Hmm... What about the comment above the code mentioning pmtu discovery ?

        /*
         * Maximal non-jumbo IPv6 payload is IPV6_MAXPLEN and
         * corresponding MSS is IPV6_MAXPLEN - tcp_header_size.
         * IPV6_MAXPLEN is also valid and means: "any MSS,
         * rely only on pmtu discovery"
         */


Check out code from net/ipv6/ip6_output.c, ip6_append_data() :

        if (mtu <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) {
                if (cork->length + length > sizeof(struct ipv6hdr) + IPV6_MAXPLEN - fragheaderlen) {
                        ipv6_local_error(sk, EMSGSIZE, fl6, mtu-exthdrlen);
                        return -EMSGSIZE;
                }
        }

and __ip6_local_out()

	if (len > IPV6_MAXPLEN)
		len = 0;

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: [PATCH 2/2] ipv4: fix mss comparison
  2012-02-28  8:28 ` [PATCH 2/2] ipv4: fix mss comparison roy.qing.li
  2012-02-28  9:34   ` Francois Romieu
@ 2012-02-28 10:01   ` Eric Dumazet
  1 sibling, 0 replies; 6+ messages in thread
From: Eric Dumazet @ 2012-02-28 10:01 UTC (permalink / raw)
  To: roy.qing.li; +Cc: netdev

Le mardi 28 février 2012 à 16:28 +0800, roy.qing.li@gmail.com a écrit :
> From: RongQing.Li <roy.qing.li@gmail.com>
> 
> mss should compare with 65535 - sizeof(struct tcphdr),  
> the largest mss is 65535 - sizeof(struct tcphdr).  
> 40 is sum of sizeof(struct tcphdr) and sizeof(struct iphdr)
> 
> Signed-off-by: RongQing.Li <roy.qing.li@gmail.com>
> ---
>  net/ipv4/route.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/net/ipv4/route.c b/net/ipv4/route.c
> index 0489ced..c72f765 100644
> --- a/net/ipv4/route.c
> +++ b/net/ipv4/route.c
> @@ -1914,8 +1914,8 @@ static unsigned int ipv4_default_advmss(const struct dst_entry *dst)
>  	if (advmss == 0) {
>  		advmss = max_t(unsigned int, dst->dev->mtu - 40,
>  			       ip_rt_min_advmss);
> -		if (advmss > 65535 - 40)
> -			advmss = 65535 - 40;
> +		if (advmss > 65535 - sizeof(struct tcphdr))
> +			advmss = 65535 - sizeof(struct tcphdr);
>  	}
>  	return advmss;
>  }

Hmm... I am puzzled.

1) If you touch this code, please use min_t() macro to be consistent
with prior line ? Please note I dont advocate a general use of these
min/max macros, just some consistency in the same page of code.

2) Prior line still has the magic 40 value.
    (I personally like 40 more than sizeof(struct tcphdr) +
sizeof(struct iphdr)), its not like 40 is going to change one day :)

3) What is the maximum datagram size in ipv4.
      (Say you name this IPV4_MAXPLEN)

4) What would be the meaning of having TCP MSS bigger than
    IPV4_MAXPLEN - 40 ?

^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2012-02-28 10:01 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-02-28  8:28 [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) roy.qing.li
2012-02-28  8:28 ` [PATCH 2/2] ipv4: fix mss comparison roy.qing.li
2012-02-28  9:34   ` Francois Romieu
2012-02-28  9:55     ` RongQing Li
2012-02-28 10:01   ` Eric Dumazet
2012-02-28  9:56 ` [PATCH 1/2] ipv6: fix mss when it is bigger than IPV6_MAXPLEN - sizeof(struct tcphdr) 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).