[PATCH] TCP_MIB_OUTSEGS increased including retransmitted octets
Linux kernel increased the TCP_MIB_OUTSEGS even if only retransmitted
octets is send out. RFC2012 saied that it excluding those containing
only retransmitted octets.
Refer to RFC2012, tcpOutSegs is defined as following:
tcpOutSegs OBJECT-TYPE
SYNTAX Counter32
MAX-ACCESS read-only
STATUS current
DESCRIPTION
The total number of segments sent, including those on
current connections but excluding those containing only
retransmitted octets.
::= { tcp 11 }
--- a/net/ipv4/tcp_output.c 2006-02-23 09:20:24.660261904 +0900
+++ b/net/ipv4/tcp_output.c 2006-02-23 09:41:16.393969456 +0900
@@ -379,7 +379,8 @@
if (skb-len != tcp_header_size)
tcp_event_data_sent(tp, skb, sk);
- TCP_INC_STATS(TCP_MIB_OUTSEGS);
+ if (!(tcb-sacked TCPCB_LOST))
+ TCP_INC_STATS(TCP_MIB_OUTSEGS);
err = tp-af_specific-queue_xmit(skb, 0);
if (unlikely(err = 0))
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Uninline kfree_skb and allow NULL argument
On Thu, 23 February 2006 12:48:01 -0800, David S. Miller wrote:
If you wish to contribute to a software project, you should adhere to
the coding style and conventions of that project when submitting
changes. It doesn't matter what the reasons are for those
conventions, you should follow them until the projects decides to
change them.
Agreed.
If you wish to discuss the merits of putting extern there or not in
function declarations, you can start a thread about that and make
proposals on linux-kernel.
Not really. According to my limited understanding of C, the presence
or absence of extern shouldn't make a difference either way. If this
understanding if correct and people simply prefer to have the extern,
that's perfectly fine with me. But if it does indeed make a
difference, I'd like to learn something new.
So place add extern here, thanks a lot.
Done. Sorry about that.
Jörn
--
The cost of changing business rules is much more expensive for software
than for a secretaty.
-- unknown
o Uninline kfree_skb, which saves some 15k of object code on my notebook.
o Allow kfree_skb to be called with a NULL argument.
Subsequent patches can remove conditional from drivers and further
reduce source and object size.
Signed-off-by: Jörn Engel [EMAIL PROTECTED]
---
include/linux/skbuff.h | 17 +
net/core/skbuff.c | 18 ++
2 files changed, 19 insertions(+), 16 deletions(-)
--- kfree_skb/include/linux/skbuff.h~kfree_skb_uninline_null2006-02-23
13:35:05.0 +0100
+++ kfree_skb/include/linux/skbuff.h2006-02-23 13:36:23.0 +0100
@@ -306,6 +306,7 @@ struct sk_buff {
#include asm/system.h
+extern void kfree_skb(struct sk_buff *skb);
extern void __kfree_skb(struct sk_buff *skb);
extern struct sk_buff *__alloc_skb(unsigned int size,
gfp_t priority, int fclone);
@@ -406,22 +407,6 @@ static inline struct sk_buff *skb_get(st
*/
/**
- * kfree_skb - free an sk_buff
- * @skb: buffer to free
- *
- * Drop a reference to the buffer and free it if the usage count has
- * hit zero.
- */
-static inline void kfree_skb(struct sk_buff *skb)
-{
- if (likely(atomic_read(skb-users) == 1))
- smp_rmb();
- else if (likely(!atomic_dec_and_test(skb-users)))
- return;
- __kfree_skb(skb);
-}
-
-/**
* skb_cloned - is the buffer a clone
* @skb: buffer to check
*
--- kfree_skb/net/core/skbuff.c~kfree_skb_uninline_null 2006-02-23
13:35:05.0 +0100
+++ kfree_skb/net/core/skbuff.c 2006-02-23 13:37:01.0 +0100
@@ -355,6 +355,24 @@ void __kfree_skb(struct sk_buff *skb)
}
/**
+ * kfree_skb - free an sk_buff
+ * @skb: buffer to free
+ *
+ * Drop a reference to the buffer and free it if the usage count has
+ * hit zero.
+ */
+void kfree_skb(struct sk_buff *skb)
+{
+ if (unlikely(!skb))
+ return;
+ if (likely(atomic_read(skb-users) == 1))
+ smp_rmb();
+ else if (likely(!atomic_dec_and_test(skb-users)))
+ return;
+ __kfree_skb(skb);
+}
+
+/**
* skb_clone - duplicate an sk_buff
* @skb: buffer to clone
* @gfp_mask: allocation priority
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Uninline kfree_skb and allow NULL argument
From: Jörn Engel [EMAIL PROTECTED] Date: Fri, 24 Feb 2006 09:18:38 +0100 On Thu, 23 February 2006 12:48:01 -0800, David S. Miller wrote: So place add extern here, thanks a lot. Done. Sorry about that. Thanks. One more problem :-) Please export the kfree_skb() symbol so networking modules don't break. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Uninline kfree_skb and allow NULL argument
On Fri, 24 February 2006 00:32:04 -0800, David S. Miller wrote:
One more problem :-) Please export the kfree_skb() symbol so
networking modules don't break.
Doh! Done.
Any more reasons to wear a brown paperbag?
Jörn
--
When in doubt, punt. When somebody actually complains, go back and fix it...
The 90% solution is a good thing.
-- Rob Landley
o Uninline kfree_skb, which saves some 15k of object code on my notebook.
o Allow kfree_skb to be called with a NULL argument.
Subsequent patches can remove conditional from drivers and further
reduce source and object size.
Signed-off-by: Jörn Engel [EMAIL PROTECTED]
---
include/linux/skbuff.h | 17 +
net/core/skbuff.c | 19 +++
2 files changed, 20 insertions(+), 16 deletions(-)
--- kfree_skb/include/linux/skbuff.h~kfree_skb_uninline_null2006-02-23
13:35:05.0 +0100
+++ kfree_skb/include/linux/skbuff.h2006-02-24 09:35:47.0 +0100
@@ -306,6 +306,7 @@ struct sk_buff {
#include asm/system.h
+extern void kfree_skb(struct sk_buff *skb);
extern void __kfree_skb(struct sk_buff *skb);
extern struct sk_buff *__alloc_skb(unsigned int size,
gfp_t priority, int fclone);
@@ -406,22 +407,6 @@ static inline struct sk_buff *skb_get(st
*/
/**
- * kfree_skb - free an sk_buff
- * @skb: buffer to free
- *
- * Drop a reference to the buffer and free it if the usage count has
- * hit zero.
- */
-static inline void kfree_skb(struct sk_buff *skb)
-{
- if (likely(atomic_read(skb-users) == 1))
- smp_rmb();
- else if (likely(!atomic_dec_and_test(skb-users)))
- return;
- __kfree_skb(skb);
-}
-
-/**
* skb_cloned - is the buffer a clone
* @skb: buffer to check
*
--- kfree_skb/net/core/skbuff.c~kfree_skb_uninline_null 2006-02-23
13:35:05.0 +0100
+++ kfree_skb/net/core/skbuff.c 2006-02-24 09:36:15.0 +0100
@@ -355,6 +355,24 @@ void __kfree_skb(struct sk_buff *skb)
}
/**
+ * kfree_skb - free an sk_buff
+ * @skb: buffer to free
+ *
+ * Drop a reference to the buffer and free it if the usage count has
+ * hit zero.
+ */
+void kfree_skb(struct sk_buff *skb)
+{
+ if (unlikely(!skb))
+ return;
+ if (likely(atomic_read(skb-users) == 1))
+ smp_rmb();
+ else if (likely(!atomic_dec_and_test(skb-users)))
+ return;
+ __kfree_skb(skb);
+}
+
+/**
* skb_clone - duplicate an sk_buff
* @skb: buffer to clone
* @gfp_mask: allocation priority
@@ -1807,6 +1825,7 @@ void __init skb_init(void)
EXPORT_SYMBOL(___pskb_trim);
EXPORT_SYMBOL(__kfree_skb);
+EXPORT_SYMBOL(kfree_skb);
EXPORT_SYMBOL(__pskb_pull_tail);
EXPORT_SYMBOL(__alloc_skb);
EXPORT_SYMBOL(pskb_copy);
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: IPv6 setsockopt software MTU patch
Hello.
In article [EMAIL PROTECTED] (at Thu, 23 Feb 2006 17:10:18 -0800 (PST)),
David S. Miller [EMAIL PROTECTED] says:
From: YOSHIFUJI Hideaki [EMAIL PROTECTED]
Date: Fri, 24 Feb 2006 00:23:51 +0900 (JST)
David, please apply. Thank you.
Can you please resend the patch with a full changelog
entry and Signed-off-by lines for me? Thank you.
This is for net-2.6 right? Or net-2.6.17?
Here it is. I think net-2.6 is fine. Thanks.
---
2.6.x have ignored IPV6_MTU socket option, which control per-socket
Path MTU. This patch, for net-2.6, implements that.
Based on patch by Hoerdt Mickael [EMAIL PROTECTED].
Signed-off-by: YOSHIFUJI Hideaki [EMAIL PROTECTED]
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index efa3e72..f999edd 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -494,6 +494,7 @@ static int ip6_fragment(struct sk_buff *
struct net_device *dev;
struct sk_buff *frag;
struct rt6_info *rt = (struct rt6_info*)skb-dst;
+ struct ipv6_pinfo *np = skb-sk ? inet6_sk(skb-sk) : NULL;
struct ipv6hdr *tmp_hdr;
struct frag_hdr *fh;
unsigned int mtu, hlen, left, len;
@@ -505,7 +506,12 @@ static int ip6_fragment(struct sk_buff *
hlen = ip6_find_1stfragopt(skb, prevhdr);
nexthdr = *prevhdr;
- mtu = dst_mtu(rt-u.dst) - hlen - sizeof(struct frag_hdr);
+ mtu = dst_mtu(rt-u.dst);
+ if (np np-frag_size mtu) {
+ if (np-frag_size)
+ mtu = np-frag_size;
+ }
+ mtu -= hlen + sizeof(struct frag_hdr);
if (skb_shinfo(skb)-frag_list) {
int first_len = skb_pagelen(skb);
@@ -882,7 +888,12 @@ int ip6_append_data(struct sock *sk, int
inet-cork.fl = *fl;
np-cork.hop_limit = hlimit;
np-cork.tclass = tclass;
- inet-cork.fragsize = mtu = dst_mtu(rt-u.dst.path);
+ mtu = dst_mtu(rt-u.dst.path);
+ if (np np-frag_size mtu) {
+ if (np-frag_size)
+ mtu = np-frag_size;
+ }
+ inet-cork.fragsize = mtu;
if (dst_allfrag(rt-u.dst.path))
inet-cork.flags |= IPCORK_ALLFRAG;
inet-cork.length = 0;
--
YOSHIFUJI Hideaki @ USAGI Project [EMAIL PROTECTED]
GPG-FP : 9022 65EB 1ECF 3AD1 0BDF 80D8 4807 F894 E062 0EEA
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Add Wake on LAN support to sis900 (2)
Attached you find the patch that fixes two bugs in the WoL implementation of sis900. The first causes hangs on some system on driver load, the second causes troubles when disabling WoL support. Both fixes are one liner and really simple. Patch is against latest netdev-2.6 tree. Signed-off-by: Lennert Buytenhek [EMAIL PROTECTED] Signed-off-by: Daniele Venzano [EMAIL PROTECTED] sis900_wol_fix.diff Description: Binary data
Re: [PATCH] Add Wake on LAN support to sis900 (2)
Here's the full patch that adds WoL support to vanilla kernel version 2.6.15.4 with the latest fixes. For reference, comments and general review. I hope my mailer doesn't feel too smart today. Thanks. Signed-off-by: Daniele Venzano [EMAIL PROTECTED] sis900_wol.diff Description: Binary data -- Daniele Venzano http://www.brownhat.org
Re: ip6_tunnel keeping dst_cache after change of params
Hugo Santos wrote: Hi, Hi Hugo! ip6_tunnel keeps a cached dst (dst_cache in ip6_tnl) per tunnel instance. This cached dst is re-used while it's not marked obsolete. A change of the tunnel's parameters (via SIOCCHGTUNNEL) does not invalidate the dst_cache directly, which results on it being used by ip6ip6_tnl_xmit after the tunnel is configured with new parameters. Shouldn't ip6ip6_tnl_change dst_release() the cached dst and leave ip6ip6_tnl_xmit to pick a new one based on the new local/remote addresses etc? I can provide a patch to fix this, meanwhile just wanted to confirm the expected behaviour. Yes, you are right. Regards, Ville - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] ip6_tunnel: release cached dst on change of tunnel params
Hugo Santos wrote: Hi, The included patch fixes ip6_tunnel to release the cached dst entry when the tunnel parameters (such as tunnel endpoints) are changed so they are used immediatly for the next encapsulated packets. Signed-off-by: Hugo Santos [EMAIL PROTECTED] --- linux-2.6.16-rc4/net/ipv6/ip6_tunnel.c 2006-02-17 22:23:45.0 + +++ linux-2.6.16-rc4-new/net/ipv6/ip6_tunnel.c 2006-02-24 01:40:17.0 + @@ -884,6 +884,7 @@ ip6ip6_tnl_change(struct ip6_tnl *t, str t-parms.encap_limit = p-encap_limit; t-parms.flowinfo = p-flowinfo; t-parms.link = p-link; + ip6_tnl_dst_reset(t); ip6ip6_tnl_link_config(t); return 0; } Acked-by: Ville Nuorvala [EMAIL PROTECTED] - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Problem with Ipsec transport mode over NAT
Patrick McHardy wrote: I don't know what correct fix is. Adding an extra call to xfrm4_policy_check in tcp_v4_rcv before the checksum check fixes this problem and doesn't seem to break anything else. On the other hand, moving some of the code in esp_post_input into esp_input (especially line 298) will work, too. So we could move checksum validation behind xfrm4_policy_check or already set ip_summed to CHECKSUM_UNNECESSARY in esp_input. Already setting ip_summed in esp4_input looks easier. But this still leaves one problem. With netfilter and local NAT, a decapsulated transport mode packet might be forwarded to another host. In that case the checksum contained in the packet is invalid. Any ideas how to fix this anyone? I don't know what the functional separation or difference between a packet input function and a packet post input function is but the entire code in esp_post_input doesn't seem like it would cause any problem just by placing it at end of esp_input instead of current location. A forwarded decapsulated packet would have the destination IP changed from server S to another IP. Shouldn't that cause the stack to automatically recalculate the checksum? Anyway, enough speculation. I will leave the solution to those who know the linux kernel networking code. Thanks. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [git patches] net driver fixes
Wolfgang Hoffmann wrote: On Friday 24 February 2006 06:22, Jeff Garzik wrote: Please pull from 'upstream-fixes' branch of master.kernel.org:/pub/scm/linux/kernel/git/jgarzik/netdev-2.6.git [...] Stephen Hemminger: sky2: yukon-ec-u chipset initialization sky2: limit coalescing values to ring size sky2: poke coalescing timer to fix hang sky2: force early transmit status sky2: use device iomem to access PCI config sky2: close race on IRQ mask update. [...] Thanks for the update. Still I'm seeing reproducable hangs with this version of sky2 (as reported in bugzilla 6084 and discussed on netdev). Stephen, if there is anything I can do to narrow down my hangs a bit more systematically, please let me know, I'd be happy to help. Wolfgang There is an outstanding bug where the sky2 will hang if it receives a packet larger than the MTU. At this point, there isn't enough information on chip behavior to fix. You could try using a larger mut or patching the driver so that the rx_buffer_size is always big (like 4k). - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [git patches] net driver fixes
On Friday 24 February 2006 17:14, Stephen Hemminger wrote: There is an outstanding bug where the sky2 will hang if it receives a packet larger than the MTU. At this point, there isn't enough information on chip behavior to fix. You could try using a larger mut or patching the driver so that the rx_buffer_size is always big (like 4k). I've raised the MTU from 1500 to 3000 and still reproduced the hang. Would you mind sending me a patch for forcing rx_buffer_size to 4k, so I can try that, or is no sense in that, given that raising the MTU didn't help? Concerning information on chip behavior, are you missing vendor specs, or could I be helpful by reproducing the hang with an instrumented driver that gives more information about the chip status at hang time? Another thing that may be worth to find out is why 0.13 with Carl-Daniel Hailfingers fix works. I didn't see a single hang with that version. I'm currently resorting to that version, but well ... I'd really like to help get this driver working robustly. I'm not enough into networking to help by coding, but yes I'll give feedback on any driver version you send me. I'd just like to provide you more helpful data than repeating new version still hangs ;-) Wolfgang - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] ip6_tunnel: release cached dst on change of tunnel params
From: Ville Nuorvala [EMAIL PROTECTED] Date: Fri, 24 Feb 2006 15:55:23 +0200 Hugo Santos wrote: Hi, The included patch fixes ip6_tunnel to release the cached dst entry when the tunnel parameters (such as tunnel endpoints) are changed so they are used immediatly for the next encapsulated packets. Signed-off-by: Hugo Santos [EMAIL PROTECTED] --- linux-2.6.16-rc4/net/ipv6/ip6_tunnel.c 2006-02-17 22:23:45.0 + +++ linux-2.6.16-rc4-new/net/ipv6/ip6_tunnel.c 2006-02-24 01:40:17.0 + @@ -884,6 +884,7 @@ ip6ip6_tnl_change(struct ip6_tnl *t, str t-parms.encap_limit = p-encap_limit; t-parms.flowinfo = p-flowinfo; t-parms.link = p-link; + ip6_tnl_dst_reset(t); ip6ip6_tnl_link_config(t); return 0; } Acked-by: Ville Nuorvala [EMAIL PROTECTED] Applied, although the patch was tab/whitespace damaged by Hugo's email client so I had to apply it by hand. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: IPv6 setsockopt software MTU patch
From: YOSHIFUJI Hideaki [EMAIL PROTECTED] Date: Fri, 24 Feb 2006 18:22:25 +0900 (JST) Here it is. I think net-2.6 is fine. Thanks. Applied, thanks a lot. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
[PATCH 3/3] [LLC]: Fix sap refcounting
Thanks to Leslie Harlley Watter [EMAIL PROTECTED] for reporting
the problem an testing this patch.
Signed-off-by: Arnaldo Carvalho de Melo [EMAIL PROTECTED]
---
net/llc/af_llc.c |4 +++-
net/llc/llc_core.c |1 -
2 files changed, 3 insertions(+), 2 deletions(-)
42b30504aadc0e1e45eb915573e1d2c07d133b00
diff --git a/net/llc/af_llc.c b/net/llc/af_llc.c
index 2337944..5a04db7 100644
--- a/net/llc/af_llc.c
+++ b/net/llc/af_llc.c
@@ -187,8 +187,10 @@ static int llc_ui_release(struct socket
llc-laddr.lsap, llc-daddr.lsap);
if (!llc_send_disc(sk))
llc_ui_wait_for_disc(sk, sk-sk_rcvtimeo);
- if (!sock_flag(sk, SOCK_ZAPPED))
+ if (!sock_flag(sk, SOCK_ZAPPED)) {
+ llc_sap_put(llc-sap);
llc_sap_remove_socket(llc-sap, sk);
+ }
release_sock(sk);
if (llc-dev)
dev_put(llc-dev);
diff --git a/net/llc/llc_core.c b/net/llc/llc_core.c
index ab0fcd3..bd242a4 100644
--- a/net/llc/llc_core.c
+++ b/net/llc/llc_core.c
@@ -127,7 +127,6 @@ struct llc_sap *llc_sap_open(unsigned ch
goto out;
sap-laddr.lsap = lsap;
sap-rcv_func = func;
- llc_sap_hold(sap);
llc_add_sap(sap);
out:
write_unlock_bh(llc_sap_list_lock);
--
1.2.2.gd27d
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
[PATCH 2/3] [LLC]: Replace __inline__ with inline
Signed-off-by: Arnaldo Carvalho de Melo [EMAIL PROTECTED]
---
net/llc/af_llc.c |9 -
1 files changed, 4 insertions(+), 5 deletions(-)
bac4eba541726216405d373edae1b285de230e0d
diff --git a/net/llc/af_llc.c b/net/llc/af_llc.c
index 490b3b4..2337944 100644
--- a/net/llc/af_llc.c
+++ b/net/llc/af_llc.c
@@ -54,7 +54,7 @@ static int llc_ui_wait_for_busy_core(str
*
* Return the next unused link number for a given sap.
*/
-static __inline__ u16 llc_ui_next_link_no(int sap)
+static inline u16 llc_ui_next_link_no(int sap)
{
return llc_ui_sap_link_no_max[sap]++;
}
@@ -65,7 +65,7 @@ static __inline__ u16 llc_ui_next_link_n
*
* Given an ARP header type return the corresponding ethernet protocol.
*/
-static __inline__ u16 llc_proto_type(u16 arphrd)
+static inline u16 llc_proto_type(u16 arphrd)
{
return arphrd == ARPHRD_IEEE802_TR ?
htons(ETH_P_TR_802_2) : htons(ETH_P_802_2);
@@ -75,7 +75,7 @@ static __inline__ u16 llc_proto_type(u16
* llc_ui_addr_null - determines if a address structure is null
* @addr: Address to test if null.
*/
-static __inline__ u8 llc_ui_addr_null(struct sockaddr_llc *addr)
+static inline u8 llc_ui_addr_null(struct sockaddr_llc *addr)
{
return !memcmp(addr, llc_ui_addrnull, sizeof(*addr));
}
@@ -89,8 +89,7 @@ static __inline__ u8 llc_ui_addr_null(st
* operation the user would like to perform and the type of socket.
* Returns the correct llc header length.
*/
-static __inline__ u8 llc_ui_header_len(struct sock *sk,
- struct sockaddr_llc *addr)
+static inline u8 llc_ui_header_len(struct sock *sk, struct sockaddr_llc *addr)
{
u8 rc = LLC_PDU_LEN_U;
--
1.2.2.gd27d
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Uninline kfree_skb and allow NULL argument
From: Jörn Engel [EMAIL PROTECTED] Date: Fri, 24 Feb 2006 09:37:59 +0100 On Fri, 24 February 2006 00:32:04 -0800, David S. Miller wrote: One more problem :-) Please export the kfree_skb() symbol so networking modules don't break. Doh! Done. Any more reasons to wear a brown paperbag? Not that I can see :-) Patch applied, thanks a lot. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] af_unix: use shift instead of integer division
From: Benjamin LaHaise [EMAIL PROTECTED] Date: Tue, 7 Feb 2006 06:54:44 -0800 The patch below replaces a divide by 2 with a shift -- sk_sndbuf is an integer, so gcc emits an idiv, which takes 10x longer than a shift by 1. This improves af_unix bandwidth by ~6-10K/s. Also, tidy up the comment to fit in 80 columns while we're at it. Signed-off-by: Benjamin LaHaise [EMAIL PROTECTED] Applied, thanks. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 2.6.17 2/2] tg3: Fix tg3_get_ringparam()
Both patches applied to net-2.6.17, thanks a lot Michael. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: fib_trie initialzation fix [PATCH]
From: Robert Olsson [EMAIL PROTECTED] Date: Wed, 22 Feb 2006 14:23:35 +0100 In some kernel configs /proc functions seems to be accessed before the trie is initialized. The patch below checks for this. Signed-off-by: Robert Olsson [EMAIL PROTECTED] Applied, thanks Robert. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] [IPV6]: Ensure to join all-routers multicast address.
From: YOSHIFUJI Hideaki [EMAIL PROTECTED] Date: Thu, 23 Feb 2006 03:18:26 +0900 (JST) If net.ipv6.conf.default.forwarding is !0 when bringing up the interface, we failed to join all routers multicast address, while we join/leave if we enable/disable net.ipv6.conf.ethX.forwarding later. This is inconsistent because we assume that we join all routers multicast if and only if net.ipv6.conf.ethX.forwarding is !0. So, make sure to join all routers multicast address if forwarding is enabled on the interface. Signed-off-by: YOSHIFUJI Hideaki [EMAIL PROTECTED] Applied, thanks a lot. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: fib_trie stats fix [PATCH]
From: Robert Olsson [EMAIL PROTECTED] Date: Wed, 22 Feb 2006 14:32:43 +0100 fib_triestats has been buggy and caused oopses some platforms as openwrt. The patch below should cure those problems. Signed-off-by: Robert Olsson [EMAIL PROTECTED] Also applied, thanks Robert. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/3][LLC]: Cleanups Fix
From: Arnaldo Carvalho de Melo [EMAIL PROTECTED] Date: Fri, 24 Feb 2006 18:31:35 -0300 Please consider pulling from: master.kernel.org:/pub/scm/linux/kernel/git/acme/net-2.6.17.git Looks good, pulled, thanks Arnaldo. - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
[Announce] Intel PRO/Wireless 3945ABG Network Connection
Intel is pleased to announce the launch of an open source project to support the Intel PRO/Wireless 3945ABG Network Connection mini-PCI express adapter (IPW3945). The project is hosted at http://ipw3945.sourceforge.net. A development mailing list is available (linked from the top of the IPW3945 project page.) You can find the current development release for the adapter by following the links on the project home page. A stable [end user targetted] version is not yet available. Those interested in using the development version should review the notice linked to from the project page. A stable version should be available in the next few weeks. Aside from a form factor change (our prior wireless cards were mini PCI while this one is mini PCI express), this project has also changed the division of work between what occurs on the adapter and what the host is responsible for performing. The microcode and hardware provide lower level MAC services (timings, backoffs, transmit queue management, etc.) The host is responsible for middle and upper layer MAC services. As a result of this change, some of the capabilities currently required to be provided on the host include enforcement of regulatory limits for the radio transmitter (radio calibration, transmit power, valid channels, 802.11h, etc.) In order to meet the requirements of all geographies into which our adapters ship (over 100 countries) we have placed the regulatory enforcement logic into a user space daemon that we provide as a binary under the same license agreement as the microcode. We provide that binary pre-compiled as both a 32-bit and 64-bit application. The daemon utilizes a sysfs interface exposed by the driver in order to communicate with the hardware and configure the required regulatory parameters. Those familiar with our prior projects may be pleased with the changes we have made with the license agreement for binary portions of this new project. We were able to provide a more easily understood agreement for the binary components required for the adapter to function. While this new license still restricts against reverse engineering and modification, it has been changed to allow easier redistribution. You can find the terms of the agreement accessible from the microcode and daemon download page linked to from the project site. The current development snapshot contains backward compatibility code to allow the driver to work in older kernels. We will be removing that code prior to submitting the driver for inclusion in the kernel. Thanks, James - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [Announce] Intel PRO/Wireless 3945ABG Network Connection
Awesome. Now all we need is someone to write the bcm series for wireless and ndiswrapper can go away. Jeff James Ketrenos wrote: Intel is pleased to announce the launch of an open source project to support the Intel PRO/Wireless 3945ABG Network Connection mini-PCI express adapter (IPW3945). The project is hosted at http://ipw3945.sourceforge.net. A development mailing list is available (linked from the top of the IPW3945 project page.) You can find the current development release for the adapter by following the links on the project home page. A stable [end user targetted] version is not yet available. Those interested in using the development version should review the notice linked to from the project page. A stable version should be available in the next few weeks. Aside from a form factor change (our prior wireless cards were mini PCI while this one is mini PCI express), this project has also changed the division of work between what occurs on the adapter and what the host is responsible for performing. The microcode and hardware provide lower level MAC services (timings, backoffs, transmit queue management, etc.) The host is responsible for middle and upper layer MAC services. As a result of this change, some of the capabilities currently required to be provided on the host include enforcement of regulatory limits for the radio transmitter (radio calibration, transmit power, valid channels, 802.11h, etc.) In order to meet the requirements of all geographies into which our adapters ship (over 100 countries) we have placed the regulatory enforcement logic into a user space daemon that we provide as a binary under the same license agreement as the microcode. We provide that binary pre-compiled as both a 32-bit and 64-bit application. The daemon utilizes a sysfs interface exposed by the driver in order to communicate with the hardware and configure the required regulatory parameters. Those familiar with our prior projects may be pleased with the changes we have made with the license agreement for binary portions of this new project. We were able to provide a more easily understood agreement for the binary components required for the adapter to function. While this new license still restricts against reverse engineering and modification, it has been changed to allow easier redistribution. You can find the terms of the agreement accessible from the microcode and daemon download page linked to from the project site. The current development snapshot contains backward compatibility code to allow the driver to work in older kernels. We will be removing that code prior to submitting the driver for inclusion in the kernel. Thanks, James - To unsubscribe from this list: send the line unsubscribe linux-kernel in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/ - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
question about struct ipv6hdr
Hello everybody,
As a comment says in include/linux/ipv6.h, the ipv6hdr structure is not
correct. The variables priority and flow_lbl do not match the IPv6
header fields as described in the RFC.
/*
* IPv6 fixed header
*
* BEWARE, it is incorrect. The first 4 bits of flow_lbl
* are glued to priority now, forming class.
*/
struct ipv6hdr {
#if defined(__LITTLE_ENDIAN_BITFIELD)
__u8priority:4,
version:4;
#elif defined(__BIG_ENDIAN_BITFIELD)
__u8version:4,
priority:4;
#else
#error Please fix asm/byteorder.h
#endif
__u8flow_lbl[3];
[snip]
Is there some specific reason (other than historical one) ?
Didier Barvaux
signature.asc
Description: PGP signature
Re: [Announce] Intel PRO/Wireless 3945ABG Network Connection
On Fri, 2006-02-24 at 16:29 -0600, James Ketrenos wrote: Intel is pleased to announce the launch of an open source project to support the Intel PRO/Wireless 3945ABG Network Connection mini-PCI express adapter (IPW3945). Cool! In order to meet the requirements of all geographies into which our adapters ship (over 100 countries) we have placed the regulatory enforcement logic into a user space daemon that we provide as a binary under the same license agreement as the microcode. We provide that binary pre-compiled as both a 32-bit and 64-bit application. The daemon utilizes a sysfs interface exposed by the driver in order to communicate with the hardware and configure the required regulatory parameters. It was exciting to watch the centrino wireless cards go from the least supported cards in the Linux to the near the best G and A cards from a feature and licensing stand point (modulo the firmware restart issues). I have a ipw2200 and have recommended it and now the ipw2915 to anyone who has asked (myself and ipw2xxx using co-workers have taught thousands of students and decision makers in Linux classes worldwide). It is very disappointing to see this binary user space daemon (that must run as root, presumably to write into /sys/) requirement. I recognize that it is a better poison than a binary kernel module. At the point when I'm in the market for a mini-PCI express wireless adapter I hope there are other cards available that don't require any kernel or userland binary pieces. I'll vote with my wallet so to speak. Dax Kelson Guru Labs - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH] Add Wake on LAN support to sis900 (2)
Daniele Venzano wrote: Attached you find the patch that fixes two bugs in the WoL implementation of sis900. The first causes hangs on some system on driver load, the second causes troubles when disabling WoL support. Both fixes are one liner and really simple. Patch is against latest netdev-2.6 tree. Thank you for your prompt attention. The patch works for me (my SiS 730 board now boots again) when applied to Fedora Core kernel-2.6.15-1.1977_FC5 which claims to be 2.6.16rc4-git6 of 2006-02-23. -- - To unsubscribe from this list: send the line unsubscribe netdev in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html
SOLUÇÕES GRÁFICAS
SOLUÇÕES GRÁFICAS
Re: Problem with Ipsec transport mode over NAT
On Fri, Feb 24, 2006 at 04:57:33AM +, Patrick McHardy wrote:
So we could move checksum validation behind xfrm4_policy_check or
already set ip_summed to CHECKSUM_UNNECESSARY in esp_input. Already
setting ip_summed in esp4_input looks easier. But this still leaves
Absolutely.
The only reason post_input exists at all is that it gives us the
potential to adjust the checksums incrementally in future which
we ought to do.
However, after thinking about it for a bit we can adjust the
checksums without using this post_input stuff at all. The crucial
point is that only the inner-most NAT-T SA needs to be considered
when adjusting checksums. What's more, the checksum adjustment
comes down to a single u32 due to the linearity of IP checksums.
We just happen to have a spare u32 lying around in our skb structure :)
When ip_summed is set to CHECKSUM_NONE on input, the value of skb-csum
is currently unused. All we have to do is to make that the checksum
adjustment and voila, there goes all the post_input and decap structures!
I'll send patches to get rid of post_input now.
one problem. With netfilter and local NAT, a decapsulated transport
mode packet might be forwarded to another host. In that case the
checksum contained in the packet is invalid. Any ideas how to fix
this anyone?
I suppose you should treat CHECKSUM_UNNECESSARY as an indication that
you need to recompute the checksum from scratch instead of adjusting
it. So start by getting skb_checksum_help to only zap CHECKSUM_HW,
and then test on this in the *_manip_pkt functions.
BTW, the original address (nat_oa) structure is wrong. We need the
original src as well as the original dst addresses to incrementally
adjust the checksum. I wonder why everyone keeps getting this wrong.
Fortunately it shouldn't be too hard to fix up, for netlink at least.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
[1/2] [IPSEC] esp: Kill unnecessary block and indentation
Hi Dave:
This is the first of two patches which should fix the recent report about
transport mode NAT-T and TCP checksums.
[IPSEC] esp: Kill unnecessary block and indentation
We used to keep sg on the stack which is why the extra block was useful.
We've long since stopped doing that so let's kill the block and save
some indentation.
Signed-off-by: Herbert Xu [EMAIL PROTECTED]
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c
--- a/net/ipv4/esp4.c
+++ b/net/ipv4/esp4.c
@@ -150,6 +150,10 @@ static int esp_input(struct xfrm_state *
int elen = skb-len - sizeof(struct ip_esp_hdr) - esp-conf.ivlen -
alen;
int nfrags;
int encap_len = 0;
+ u8 nexthdr[2];
+ struct scatterlist *sg;
+ u8 workbuf[60];
+ int padlen;
if (!pskb_may_pull(skb, sizeof(struct ip_esp_hdr)))
goto out;
@@ -185,61 +189,56 @@ static int esp_input(struct xfrm_state *
if (esp-conf.ivlen)
crypto_cipher_set_iv(esp-conf.tfm, esph-enc_data,
crypto_tfm_alg_ivsize(esp-conf.tfm));
-{
- u8 nexthdr[2];
- struct scatterlist *sg = esp-sgbuf[0];
- u8 workbuf[60];
- int padlen;
-
- if (unlikely(nfrags ESP_NUM_FAST_SG)) {
- sg = kmalloc(sizeof(struct scatterlist)*nfrags,
GFP_ATOMIC);
- if (!sg)
- goto out;
- }
- skb_to_sgvec(skb, sg, sizeof(struct ip_esp_hdr) +
esp-conf.ivlen, elen);
- crypto_cipher_decrypt(esp-conf.tfm, sg, sg, elen);
- if (unlikely(sg != esp-sgbuf[0]))
- kfree(sg);
-
- if (skb_copy_bits(skb, skb-len-alen-2, nexthdr, 2))
- BUG();
+ sg = esp-sgbuf[0];
- padlen = nexthdr[0];
- if (padlen+2 = elen)
+ if (unlikely(nfrags ESP_NUM_FAST_SG)) {
+ sg = kmalloc(sizeof(struct scatterlist)*nfrags, GFP_ATOMIC);
+ if (!sg)
goto out;
+ }
+ skb_to_sgvec(skb, sg, sizeof(struct ip_esp_hdr) + esp-conf.ivlen,
elen);
+ crypto_cipher_decrypt(esp-conf.tfm, sg, sg, elen);
+ if (unlikely(sg != esp-sgbuf[0]))
+ kfree(sg);
- /* ... check padding bits here. Silly. :-) */
+ if (skb_copy_bits(skb, skb-len-alen-2, nexthdr, 2))
+ BUG();
- if (x-encap decap decap-decap_type) {
- struct esp_decap_data *encap_data;
- struct udphdr *uh = (struct udphdr *) (iph+1);
-
- encap_data = (struct esp_decap_data *)
(decap-decap_data);
- encap_data-proto = 0;
-
- switch (decap-decap_type) {
- case UDP_ENCAP_ESPINUDP:
- case UDP_ENCAP_ESPINUDP_NON_IKE:
- encap_data-proto = AF_INET;
- encap_data-saddr.a4 = iph-saddr;
- encap_data-sport = uh-source;
- encap_len = (void*)esph - (void*)uh;
- break;
+ padlen = nexthdr[0];
+ if (padlen+2 = elen)
+ goto out;
- default:
- goto out;
- }
- }
+ /* ... check padding bits here. Silly. :-) */
+
+ if (x-encap decap decap-decap_type) {
+ struct esp_decap_data *encap_data;
+ struct udphdr *uh = (struct udphdr *) (iph+1);
+
+ encap_data = (struct esp_decap_data *) (decap-decap_data);
+ encap_data-proto = 0;
- iph-protocol = nexthdr[1];
- pskb_trim(skb, skb-len - alen - padlen - 2);
- memcpy(workbuf, skb-nh.raw, iph-ihl*4);
- skb-h.raw = skb_pull(skb, sizeof(struct ip_esp_hdr) +
esp-conf.ivlen);
- skb-nh.raw += encap_len + sizeof(struct ip_esp_hdr) +
esp-conf.ivlen;
- memcpy(skb-nh.raw, workbuf, iph-ihl*4);
- skb-nh.iph-tot_len = htons(skb-len);
+ switch (decap-decap_type) {
+ case UDP_ENCAP_ESPINUDP:
+ case UDP_ENCAP_ESPINUDP_NON_IKE:
+ encap_data-proto = AF_INET;
+ encap_data-saddr.a4 = iph-saddr;
+ encap_data-sport = uh-source;
+ encap_len = (void*)esph - (void*)uh;
+ break;
+
+ default:
+ goto out;
+ }
}
+ iph-protocol = nexthdr[1];
+ pskb_trim(skb, skb-len - alen - padlen - 2);
+
[2/2] [IPSEC]: Kill post_input hook and do NAT-T in esp_input directly
Hi Dave:
Here is the patch that kills the post_input stuff.
[IPSEC]: Kill post_input hook and do NAT-T in esp_input directly
The only reason post_input exists at all is that it gives us the
potential to adjust the checksums incrementally in future which
we ought to do.
However, after thinking about it for a bit we can adjust the
checksums without using this post_input stuff at all. The crucial
point is that only the inner-most NAT-T SA needs to be considered
when adjusting checksums. What's more, the checksum adjustment
comes down to a single u32 due to the linearity of IP checksums.
We just happen to have a spare u32 lying around in our skb structure :)
When ip_summed is set to CHECKSUM_NONE on input, the value of skb-csum
is currently unused. All we have to do is to make that the checksum
adjustment and voila, there goes all the post_input and decap structures!
I've left in the decap data structures for now since it's intricately
woven into the sec_path stuff. We can kill them later too.
Signed-off-by: Herbert Xu [EMAIL PROTECTED]
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmVHI~} [EMAIL PROTECTED]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
diff --git a/include/net/xfrm.h b/include/net/xfrm.h
--- a/include/net/xfrm.h
+++ b/include/net/xfrm.h
@@ -233,7 +233,6 @@ struct xfrm_type
int (*init_state)(struct xfrm_state *x);
void(*destructor)(struct xfrm_state *);
int (*input)(struct xfrm_state *, struct
xfrm_decap_state *, struct sk_buff *skb);
- int (*post_input)(struct xfrm_state *, struct
xfrm_decap_state *, struct sk_buff *skb);
int (*output)(struct xfrm_state *, struct sk_buff
*pskb);
/* Estimate maximal size of result of transformation of a dgram */
u32 (*get_max_size)(struct xfrm_state *, int size);
diff --git a/net/ipv4/esp4.c b/net/ipv4/esp4.c
--- a/net/ipv4/esp4.c
+++ b/net/ipv4/esp4.c
@@ -12,13 +12,6 @@
#include net/protocol.h
#include net/udp.h
-/* decapsulation data for use when post-processing */
-struct esp_decap_data {
- xfrm_address_t saddr;
- __u16 sport;
- __u8proto;
-};
-
static int esp_output(struct xfrm_state *x, struct sk_buff *skb)
{
int err;
@@ -210,25 +203,47 @@ static int esp_input(struct xfrm_state *
/* ... check padding bits here. Silly. :-) */
- if (x-encap decap decap-decap_type) {
- struct esp_decap_data *encap_data;
- struct udphdr *uh = (struct udphdr *) (iph+1);
+ if (x-encap) {
+ struct xfrm_encap_tmpl *encap = x-encap;
+ struct udphdr *uh;
- encap_data = (struct esp_decap_data *) (decap-decap_data);
- encap_data-proto = 0;
+ if (encap-encap_type != decap-decap_type)
+ goto out;
- switch (decap-decap_type) {
- case UDP_ENCAP_ESPINUDP:
- case UDP_ENCAP_ESPINUDP_NON_IKE:
- encap_data-proto = AF_INET;
- encap_data-saddr.a4 = iph-saddr;
- encap_data-sport = uh-source;
- encap_len = (void*)esph - (void*)uh;
- break;
+ uh = (struct udphdr *)(iph + 1);
+ encap_len = (void*)esph - (void*)uh;
- default:
- goto out;
+ /*
+* 1) if the NAT-T peer's IP or port changed then
+*advertize the change to the keying daemon.
+*This is an inbound SA, so just compare
+*SRC ports.
+*/
+ if (iph-saddr != x-props.saddr.a4 ||
+ uh-source != encap-encap_sport) {
+ xfrm_address_t ipaddr;
+
+ ipaddr.a4 = iph-saddr;
+ km_new_mapping(x, ipaddr, uh-source);
+
+ /* XXX: perhaps add an extra
+* policy check here, to see
+* if we should allow or
+* reject a packet from a
+* different source
+* address/port.
+*/
}
+
+ /*
+* 2) ignore UDP/TCP checksums in case
+*of NAT-T in Transport Mode, or
+*perform other post-processing fixes
+*as per draft-ietf-ipsec-udp-encaps-06,
+*section 3.1.2
+*/
+ if (!x-props.mode)
+ skb-ip_summed = CHECKSUM_UNNECESSARY;
}
iph-protocol = nexthdr[1];
@@ -245,63 +260,6 @@ out:
return -EINVAL;
