From: Jeff Garzik [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 02:42:19 -0500
Seems OK to me (though not pulled, since you didn't request such; also,
I dunno if DaveM has started being bnx2 merge master again)
I can take care of these bnx2 changes tomorrow.
-
To unsubscribe from this list: send
Hi All,
We are working on kernel version 2.4.20 and are trying to migrate to
version 2.4.31. We are doing some performance related tests to see the
impact of migration. The routing throughput seems to degrade from
2.4.20 to 2.4.31.
We have a Malta board with mips 4kc processor and are using iperf
[patch 4/7] s390: some more qeth fixes
From: Frank Pavlic [EMAIL PROTECTED]
From: Peter Tiedemann [EMAIL PROTECTED]
- possible race on list fixed by reset
list processing after every operation
- traffic hang fixed
Signed-off-by: Frank Pavlic [EMAIL PROTECTED]
[patch 6/7] s390: introduce guestLan sniffer support in qeth
From: Peter Tiedemann [EMAIL PROTECTED]
- introduce guestLan sniffer support in qeth
feature allows a linux in a virtual machine
guest to become a network LAN sniffer,
monitoring and
[patch 7/7] s390: mail address changed
From: Frank Pavlic [EMAIL PROTECTED]
- mail address changed to [EMAIL PROTECTED]
Signed-off-by: Frank Pavlic [EMAIL PROTECTED]
diffstat:
lcs.c |4 ++--
qeth_main.c |4 ++--
qeth_mpc.c |2 +-
qeth_mpc.h |2 +-
[patch 1/7] s390: synthax checking for VIPA addresses fixed
From: Peter Tiedemann [EMAIL PROTECTED]
- synthax checking for VIPA addresses fixed
Signed-off-by: Frank Pavlic [EMAIL PROTECTED]
diffstat:
qeth.h | 65 -
[patch 5/7] s390: fix recovery failure of non-guestLAN devices
From: Frank Pavlic [EMAIL PROTECTED]
- Recovery of non-guestLAN Layer 2 device failed due to
trying to register the real MAC address we got from
the READ_MAC adapter parameters command.
We have
I really believe this is a BUG, tell me if I'm wrong please.
If I add a routing info in adsl table that forces *all* packets for test_ip
to go throught isdn router I do see icmp reply.
If I don't add 'ip route add host IP ...' the packets go the same way
(becouse of 'ip rule fwmark staff' + ip
Stefan Rompf [EMAIL PROTECTED] writes:
ACK. We really should not be forced to modify all ethernet drivers in
order to
allow userspace changing dormant for 802.1X and friends.
ACK too. In fact we are at all unable to modify them all without
causing chaos.
--
Krzysztof Halasa
-
To unsubscribe
On Thu, 10 Nov 2005, James Morris wrote:
For SELinux, we'll need to track genl ID assignment and deletion, so we
can determine what the Netlink family number means when we see a Netlink
message. We'll have to assume that the sysadmin has not changed the
module name.
Forgot to add that I
On Thu, 2005-10-11 at 01:10 +0100, Thomas Graf wrote:
* Jamal Hadi Salim [EMAIL PROTECTED] 2005-11-09 18:49
Historically we point such routes to the dummy device. Remember diald
and good ole SLIP dial on demand? I would suspect it should still work
the same way.
Actually, probably use a
On Thu, 2005-10-11 at 14:17 +0100, Krzysztof Halasa wrote:
Stefan Rompf [EMAIL PROTECTED] writes:
ACK. We really should not be forced to modify all ethernet drivers in
order to
allow userspace changing dormant for 802.1X and friends.
ACK too. In fact we are at all unable to modify
On Thu, 2005-10-11 at 02:25 +0100, Thomas Graf wrote:
plain text document attachment (nl_generic)
The generic netlink family builds on top of netlink and provides
simplifies access for the less demanding netlink users. It solves
the problem of protocol numbers running out by introducing a so
* Jamal Hadi Salim [EMAIL PROTECTED] 2005-11-10 11:00
Dave - ACK on all 6. Both Thomas and I have tested these and the TIPC
folks have been playing with them for sometime now; if we get them in
for 2.6.15 it would help to encourage them to change their API on their
stable release to use this
Jamal Hadi Salim [EMAIL PROTECTED] writes:
Folks: Is this a first time where we have a violent agreement?
Shall we say we are making progress?
Not sure, looks like we've agreement WRT more obvious things and we haven't
WRT to kernel-internal netdev state representation :-(
But I think we're
Jamal Hadi Salim [EMAIL PROTECTED] writes:
Well, the arguement goes both ways actually ;-
By adding those routes the kernel _is_ making policy decisions.
In the case of standard dial on demand (any that i have seen on linux at
least), they typically have something along the lines of scripts
Fs_enet was compile-broken due to recent platform code update (a number
of platform stuff has been moved to the separate header). This fixes the
compile issue.
Signed-off-by: Vitaly Bordug [EMAIL PROTECTED]
---
diff --git a/drivers/net/fs_enet/fs_enet-main.c
* James Morris [EMAIL PROTECTED] 2005-11-10 10:15
For SELinux, we'll need to track genl ID assignment and deletion, so we
can determine what the Netlink family number means when we see a Netlink
message. We'll have to assume that the sysadmin has not changed the
module name.
Forgot to
Ben, as for your test I think something is messed up in your patches,
as no one else seems to be reporting your freezes.
I will run some tests on standard kernels if I can still reproduce on
2.6.14. If it is bugs in my patches, it's subtle..since turning off
TSO fixes it.
IIRC you said you
Rick Jones wrote:
Ben, as for your test I think something is messed up in your patches,
as no one else seems to be reporting your freezes.
I will run some tests on standard kernels if I can still reproduce on
2.6.14. If it is bugs in my patches, it's subtle..since turning off
TSO fixes it.
On Thu, 10 Nov 2005 12:35:16 -0500
Armando L. Caro, Jr. [EMAIL PROTECTED] wrote:
I have been reading through the TCP code for kernel versions 2.4 and
2.6, and have noticed that both versions not only cache cwnd state, but
also use the cached state to clamp the cwnd of subsquent TCP connections
Am Mittwoch 09 November 2005 22:12 schrieb Thomas Graf:
Something like this should do the job, although it doesn't take care
of taking things up again for now. Now all supporters of this should
tell me how to implement any case of on demand interface after taking
the routes down.
[...]
On
It is not necessarily the most efficient way to generate bulk traffic,
but I think it is a valid test.
I use non-blocking IO and poll(). It is true that both sides may have
full TX and/or RX buffers, but the code still works fine. I can adjust
my tool to request any speed any either
I can say that stock netperf has no _single-connection_ bidirectional
tests,
As I think more about it, that statement of mine is slightly incorrect. If one
configures netperf2 with --enable-burst, and are careful about the product of
the burst size and request/response sizes wrt the size of
Stephen Hemminger wrote:
On
It is not necessarily the most efficient way to generate bulk traffic,
but I think it is a valid test.
I use non-blocking IO and poll(). It is true that both sides may have
full TX and/or RX buffers, but the code still works fine. I can adjust
my tool to request
Stefan Rompf [EMAIL PROTECTED] writes:
fib_disable_ip() evicts all routes pointing to that interface, including
userspace generated ones, doesn't it? If so, we don't get away that easy.
Right. I'm no longer sure we should remove the direct route. It would be
crazy if we remove direct route
Stephen Hemminger wrote:
Network discussions like this are on netdev@vger.kernel.org not LKML
Oops. Sorry about that. I tried finding a list focusing on Linux net
development, but somehow missed this one.
You are probably talking about this code in tcp_init_metrics
if
On Thu, 10 Nov 2005, Thomas Graf wrote:
void security_netlink_genl_event(int event, unsigned int id, const char
*name);
Couldn't we have SELinux ask the controller to resolve the ID to
the name? The assignments are already stored in family_ht[], we'd
just have to export
From: Armando L. Caro, Jr. [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 14:17:14 -0500
Stephen Hemminger wrote:
It sets cwnd_clamp for the new connection only if the congestion window
has been set as part of a route entry. A normal connection will store
the last cwnd in the route metric, but
From: Herbert Xu [EMAIL PROTECTED]
Date: Sat, 5 Nov 2005 15:47:21 +1100
[NET] Detect hardware rx checksum faults correctly
Applied, thanks for following up on this Herbert.
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL PROTECTED]
More
From: John W. Linville [EMAIL PROTECTED]
Date: Wed, 09 Nov 2005 22:10:00 -0500
Still just a trio of minor fixes for bnx2. The third patch in this
series is based on a suggested fix from Michael Chan. Individual
patches to follow.
All 3 patches applied, thanks John.
-
To unsubscribe from
From: Pavel Roskin [EMAIL PROTECTED]
Date: Tue, 08 Nov 2005 14:10:36 -0500
The protocol field in ethernet headers is big-endian and should be
annotated as such. This patch allows detection of missing ntohs() calls
on the ethernet protocol field when sparse is run with __CHECK_ENDIAN__
From: Ian McDonald [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:39:41 +1300
This patch is a first go at some documentation. Please advise if gmail
has mangled patch and I will revert to an attachment:
Signed-off-by: Ian McDonald [EMAIL PROTECTED]
Patch applied, thanks Ian.
-
To unsubscribe
From: Ian McDonald [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:41:39 +1300
Website for DCCP is now hosted at OSDL
Signed-off-by: Ian McDonald [EMAIL PROTECTED]
Also applied, thanks Ian.
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to [EMAIL
Am Donnerstag 10 November 2005 01:02 schrieb Thomas Graf:
Assuming we use the overwrite schema you propose, right. The actual
representation would no longer be atomic though, e.g. when the bonding
master access the oper state from its slave you will suffer the exactly
same issues as with the
On Thu, 10 Nov 2005, Thomas Graf wrote:
This is not only about dial on demand but about on demand
interfaces. We need to have a route to the device so the
driver/manager/whatever can see the demand. If we disable routes on
!IFF_RUNNING we don't have this route and we never see the demand.
On Thu, 10 Nov 2005, Krzysztof Halasa wrote:
Seems inactive routes are really the way to go. But it should
behave as active in every aspect except being ignored for actual
routing (i.e., indistinguishable from userspace, normally added and
removed).
Yep: Userspace can clean up its own
This is an updated version of the RFC3465 ABC patch originally
for Linux 2.6.11-rc4 by Yee-Ting Li. ABC is a way of counting
bytes ack'd rather than packets when updating congestion control.
The orignal ABC described in the RFC applied to a Reno style
algorithm. For advanced congestion control
Minor spelling fixes for TCP code.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
--- net-2.6.orig/include/net/tcp.h
+++ net-2.6/include/net/tcp.h
@@ -88,10 +88,10 @@ extern void tcp_time_wait(struct sock *s
*/
#define TCP_SYN_RETRIES 5 /*
Use hits to speed up the SACK processing. Various forms
of this have been used by TCP developers (Web100, STCP, BIC)
to avoid the 2x linear search of outstanding segments.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
--- net-2.6.orig/include/linux/tcp.h
+++ net-2.6/include/linux/tcp.h
@@
This is a patch for discussion addressing some receive buffer growing issues.
This is partially related to the thread Possible BUG in IPv4 TCP window
handling... last week.
Specifically it addresses the problem of an interaction between rcvbuf
moderation (receiver autotuning) and rcv_ssthresh.
TCP congestion control update including:
TSO fix
SAC optimization
receive autotuning threshold
--
Stephen Hemminger [EMAIL PROTECTED]
OSDL http://developer.osdl.org/~shemminger
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to
TCP peformance with TSO over networks with delay is awful.
On a 100Mbit link with 150ms delay, we get 4Mbits/sec with TSO and
50Mbits/sec without TSO.
The problem is with TSO, we intentionally do not keep the maximum
number of packets in flight to fill the window, we hold out to until
we can
Simplify the code that comuputes microsecond rtt estimate used
by TCP Vegas. Move the callback out of the RTT sampler and into
the end of the ack cleanup.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
--- net-2.6.orig/net/ipv4/tcp_input.c
+++ net-2.6/net/ipv4/tcp_input.c
@@ -548,10 +548,9
Move all the code that does linear TCP slowstart to one
inline function to ease later patch to add ABC support.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
--- net-2.6.orig/include/net/tcp.h
+++ net-2.6/include/net/tcp.h
@@ -762,6 +762,16 @@ static inline __u32 tcp_current_ssthresh
* Paul Jakma [EMAIL PROTECTED] 2005-11-10 23:10
Why would an on-demand dialup device go ~IFF_RUNNING? The whole point
of such a device is to /hide/ whether the physical layer is up or
not.
The state of an on-demand interface must be visible, typically an
on-demand interface is in dormant
Hi,
you may check if /proc/sys/net/ipv4/conf/eth3/rp_filter is 0.
If it is 1 the kernel does a route lookup for an outgoing pseudo packet for
every packet arriving on eth3. This pseudo packet is the incoming packet but
with src and dst address exchanged. Only if this route goes via the same
Greg KH [EMAIL PROTECTED] writes:
Is this upstream? I need a Signed-off-by: line.
Ahh, right. Have to learn it somehow.
Care to resend it to
the stable@ address when it is accepted?
Not sure what the address is, resending. I have no doubt Jeff will apply
this as this is all one can
On Fri, 11 Nov 2005, Thomas Graf wrote:
That is a very important point, actually it's the whole truth. If
we extend it a little and say, let's give userspace control of all
directly connected routes so it can delete/replace/modify/whatever
if needed.
It may or not be better. However, it
On Fri, Nov 11, 2005 at 01:10:30AM +0100, Krzysztof Halasa wrote:
Greg KH [EMAIL PROTECTED] writes:
Is this upstream? I need a Signed-off-by: line.
Ahh, right. Have to learn it somehow.
Care to resend it to
the stable@ address when it is accepted?
Not sure what the address is,
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:08 -0800
TCP peformance with TSO over networks with delay is awful.
On a 100Mbit link with 150ms delay, we get 4Mbits/sec with TSO and
50Mbits/sec without TSO.
The problem is with TSO, we intentionally do not keep the maximum
number of
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:09 -0800
Simplify the code that comuputes microsecond rtt estimate used
by TCP Vegas. Move the callback out of the RTT sampler and into
the end of the ack cleanup.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
Nice cleanup, applied.
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:10 -0800
Simplify the TCP congestion infrastructure. Can fold the
packets acked into the cong_avoid hook.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
This patch would break at least BIC.
Before this change, the good_ack
My interface is a virtual interface which represent a radio connected to
the host using ethernet NIC. I designed my own L2 protocol on top of
802.3, which must be used, since the radio and the host are connected by
ethernet.
Now, my radio_hard_header will only add my L2 header, and my
you may check if /proc/sys/net/ipv4/conf/eth3/rp_filter is 0.
thanks a lot it works!
I never paid attention to this option and debian sets it by default to 1.
sandro
*:-)
--
Sandro Dentella *:-)
e-mail: [EMAIL PROTECTED]
http://www.tksql.orgTkSQL Home page - My GPL work
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:11 -0800
Move all the code that does linear TCP slowstart to one
inline function to ease later patch to add ABC support.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
Applied, but one question about this hunk.
---
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:12 -0800
This is an updated version of the RFC3465 ABC patch originally
for Linux 2.6.11-rc4 by Yee-Ting Li. ABC is a way of counting
bytes ack'd rather than packets when updating congestion control.
The orignal ABC described in the RFC
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:13 -0800
Specifically it addresses the problem of an interaction between rcvbuf
moderation (receiver autotuning) and rcv_ssthresh. The problem occurs when
sending small packets to a receiver with a larger MTU. (A very common case I
have
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:14 -0800
Minor spelling fixes for TCP code.
Signed-off-by: Stephen Hemminger [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
From: [EMAIL PROTECTED]
Date: Thu, 10 Nov 2005 15:15:15 -0800
Use hits to speed up the SACK processing. Various forms
of this have been used by TCP developers (Web100, STCP, BIC)
to avoid the 2x linear search of outstanding segments.
Signed-off-by: Stephen Hemminger [EMAIL PROTECTED]
Any chance we can get this patch in?
Anton
At the moment ibmveth has DEBUG enabled which is rather verbose. Disable
it.
Signed-off-by: Anton Blanchard [EMAIL PROTECTED]
---
Index: foobar2/drivers/net/ibmveth.c
===
---
[NETFILTER]: Remove okfn usage in ip_vs_core.c
okfn should only be used from different contexts to avoid deep
call stacks, i.e. by nf_queue.
Acked-by: Julian Anastasov [EMAIL PROTECTED]
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit e718fcad3d24ada7b56fdca8dd234c9ba7459219
tree
[IPV4]: Replace dst_output by ip_dst_output
Preparation for netfilter IPsec support.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit 9271210c937a55a2e5dcba05df078e1c43b81d6c
tree 7f510d89a5245630e33e429c9d690de0aa2b05c9
parent dc924f62fd0611bb349e398c80ce3ffa5c3a9025
author Patrick
[IPV6]: Replace dst_output by ip6_dst_output
Preparation for netfilter IPsec support.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit acfa963b047cbda6a8350f122da90f1e84bf4938
tree b545b44a3074377283a2b491a5f7ae60b19ed914
parent 9271210c937a55a2e5dcba05df078e1c43b81d6c
author
[NETFILTER]: Defer fragmentation in ip_output when connection tracking is used
This allows to get rid of the okfn use in ip_refrag and save the useless
fragmentation/defragmentation step when NAT is used.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit
[NETFILTER]: Use conntrack information to determine if packet was NATed
Preparation for full IPsec support for NAT:
Use conntrack information instead of saving the saving and comparing
the addresses to determine if a packet was NATed and needs to be
rerouted to make it easier to extend the key.
[NETFILTER]: Handle NAT in IPsec policy checks
Handle NAT by reconstructing the struct flowi for the original packet
from the conntrack information.
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit 536b302ffc8a31170833ca33636e9e66ff9a
tree
From: Patrick McHardy [EMAIL PROTECTED]
Date: Fri, 11 Nov 2005 04:18:52 +0100
This is the latest set patches for netfilter IPsec support.
The use of netif_rx for the innermost SA if it used transport
mode has been replaced by explicit NF_HOOK calls in
xfrm{4,6}_input.c.
Note that I consider
[NETFILTER]: Add policy match
Signed-off-by: Patrick McHardy [EMAIL PROTECTED]
---
commit 84053f60747fa31fc46ccba57df29b639e509130
tree c9f2133a0b5b6df2fce4d15349d7e6ce433fc425
parent f12c4452cdb1b7d0d33a00120145e301507458e4
author Patrick McHardy [EMAIL PROTECTED] Fri, 11 Nov 2005 06:16:45
69 matches
Mail list logo