While hunting down oversized inlines
I stumbled on tcf_generic_walker().
It is defined in two separate files:
once as an inline in include/net/pkt_act.h
(really big one, ~750 bytes of code)
and once as a static function in net/sched/act_police.c
These two instances are not identical.
Second one
On Wed, Apr 05, 2006 at 10:25:43PM -0700, David S. Miller wrote:
From: Adrian Bunk [EMAIL PROTECTED]
Date: Tue, 4 Apr 2006 18:37:01 +0200
This patch contains the following possible cleanups:
- make the following needlessly global function static:
- arp.c: arp_rcv()
This is ok.
-
Denis Vlasenko wrote:
While hunting down oversized inlines
I stumbled on tcf_generic_walker().
It is defined in two separate files:
once as an inline in include/net/pkt_act.h
(really big one, ~750 bytes of code)
and once as a static function in net/sched/act_police.c
These two instances
On 2006-04-05 at 14:22:08, David Daney wrote:
The changes in this version are that it tests the source IP address
instead of the destination. The test now matches the test described
in the RFC. Also a small cleanup as suggested by Herbert Xu.
Some comments on the first version of the patch
Hi,
I got this oops while ejecting a Lucent Orinico Silver card:
[EMAIL PROTECTED]:~ # cardctl ident 0
product info: Lucent Technologies, WaveLAN/IEEE, Version 01.01,
manfid: 0x0156, 0x0002
function: 6 (network)
[EMAIL PROTECTED]:~ # ifdown eth2
[EMAIL PROTECTED]:~ # cardctl eject 0
Hello, Jesse Brandeburg.
On 06.04.2006 04:42 you said the following:
I built and tested the driver with patches on 2.6.16, with pci-x adapters.
I removed some workarounds for PCIe adapters, but I dont think anyone
having this problem has a PCIe adapter anyway. I saw no TX hangs and ran
Removed lkml from the CC.
On Thu, 2006-06-04 at 10:50 +0200, Patrick McHardy wrote:
Denis Vlasenko wrote:
While hunting down oversized inlines
I stumbled on tcf_generic_walker().
It is defined in two separate files:
once as an inline in include/net/pkt_act.h
(really big one, ~750
Pointed out by Denis Vlasenko. Compiles with old and new schemes.
cheers,
jamal
Rename policer specific _generic_ methods to be specific to
_act_police_
Signed-off-by: Jamal Hadi Salim [EMAIL PROTECTED]
---
diff --git a/net/sched/act_police.c b/net/sched/act_police.c
index fa877f8..24c348f
Herbert,
Both Krisztian and myself independently bumped into this problem.
Summary is: it is problematic to do reasonably timed failovers without
that timer that was initially in the original patch.
So i am attaching a patch against Linus' tree.
Dave, If Herbert acks, please apply this to Linus
spidernet uses request_firmware so it needs to select FW_LOADER.
Index: linux-2.6/drivers/net/Kconfig
===
--- linux-2.6.orig/drivers/net/Kconfig 2006-03-15 11:23:46.0 +0100
+++ linux-2.6/drivers/net/Kconfig 2006-04-04
People,
On PARISC, kernel screams if a driver uses ioremap on an io-port
region. It now strongly insists on using ioremap_nocache, which sounds
sensible enough since, in this case, cassini wants to map the card
registers.
The included one-line patch fixes it.
Please consider applying.
jamal wrote:
What needs to be done is to put a pointer to the hash, its size
and its lock in struct tc_action_ops and move the generic
functions to a seperate .c file and make them work on a struct
tcf_act_common.
This is much better than your old proposal Patrick and i have no
problem with
Hi Jamal:
On Thu, Apr 06, 2006 at 09:03:07AM -0400, jamal wrote:
Both Krisztian and myself independently bumped into this problem.
Summary is: it is problematic to do reasonably timed failovers without
that timer that was initially in the original patch.
Do you mean a case where you have
On 4/5/06, Ben Greear [EMAIL PROTECTED] wrote:
Andrew Gallatin wrote:
I'm working on a driver for a 10GbE nic. I've just gotten to the point
where I
am verifying that 802.1q vlans work without hardware vlan offload. It
seems like
the netdev features flags
On 4/5/06, David S. Miller [EMAIL PROTECTED] wrote:
From: Roland Dreier [EMAIL PROTECTED]
Date: Wed, 05 Apr 2006 14:52:24 -0700
+ case 0x1030 ... 0x1034:
Do we use the gcc case range extension in the kernel? (This is an
honest question -- I don't think I've seen it used anywhere,
On Thu, 2006-06-04 at 16:19 +0200, Patrick McHardy wrote:
jamal wrote:
This is much better than your old proposal Patrick and i have no
problem with it. Why you would need tcf_act_common
if you are going to have those elements in tc_action_ops?
So go ahead and submit the patches or you
On Fri, 2006-07-04 at 00:30 +1000, Herbert Xu wrote:
If so I see what you mean but I think a better solution is to just
set a flag when the XFRM_REPLAY_TIMEOUT fires and nothing has changed.
Then when you get XFRM_REPLAY_UPDATE you can notify unconditionally if
this flag is set.
The flag
jamal wrote:
It is actually exactly what I've always proposed. tcf_act_common
is the single action itself, tc_action_ops only includes pointers
to the hash table and the private lock.
I may have misunderstood you then or misunderstand you now. Let me be
explicit:
I like augmentation (which
On Thu, 2006-06-04 at 11:18 -0400, jamal wrote:
After a little testing (i missed the scenario where a user subscribes
and then leaves and then resubscribes; in such a case, the timer never
gets restarted) i have updated the patch to do it the brute force way.
What you propose is more optimal,
On Thu, 2006-06-04 at 17:20 +0200, Patrick McHardy wrote:
You must have misunderstood my the first time, the first variant
is what I've always proposed. tc_act_common is an abstraction for
the _members_ of the hash, the actions.
But you are still confusing me Patrick, otherwise i would agree
jamal wrote:
On Thu, 2006-06-04 at 17:20 +0200, Patrick McHardy wrote:
You must have misunderstood my the first time, the first variant
is what I've always proposed. tc_act_common is an abstraction for
the _members_ of the hash, the actions.
But you are still confusing me Patrick,
Patrick McHardy wrote:
jamal wrote:
On Thu, 2006-06-04 at 17:20 +0200, Patrick McHardy wrote:
You must have misunderstood my the first time, the first variant
is what I've always proposed. tc_act_common is an abstraction for
the _members_ of the hash, the actions.
But you are still
On Thu, 2006-06-04 at 17:26 +0200, Patrick McHardy wrote:
Patrick McHardy wrote:
jamal wrote:
On Thu, 2006-06-04 at 17:20 +0200, Patrick McHardy wrote:
You must have misunderstood my the first time, the first variant
is what I've always proposed. tc_act_common is an abstraction for
jamal wrote:
On Thu, 2006-06-04 at 17:26 +0200, Patrick McHardy wrote:
To be explicit: the hash pointer, the
size and the lock will be contained in tc_act_ops. The actions itself
(tc_act_common) of course not, that wouldn't make any sense.
Ok, go nuts then; i will volunteer to test if you
On Thu, 2006-06-04 at 17:35 +0200, Patrick McHardy wrote:
jamal wrote:
Well, I was hopeing someone else would do it, I already have lots of
TODOs on my list. One low priority item is this however, so if
noone else cares, I might do it anyway.
I could do it - it wont be in the next few days;
jamal wrote:
On Thu, 2006-06-04 at 17:35 +0200, Patrick McHardy wrote:
jamal wrote:
Well, I was hopeing someone else would do it, I already have lots of
TODOs on my list. One low priority item is this however, so if
noone else cares, I might do it anyway.
I could do it - it wont be in the
Janos Farkas wrote:
On 2006-04-05 at 14:22:08, David Daney wrote:
The changes in this version are that it tests the source IP address
instead of the destination. The test now matches the test described
in the RFC. Also a small cleanup as suggested by Herbert Xu.
Some comments on the first
On Wed, Apr 05, 2006 at 05:47:15PM -0700, Stephen Hemminger wrote:
Sky2 driver will oops referencing bad memory if used on
a dual port card. The problem is accessing past end of
MIB counter space.
Applies for both 2.6.17 and 2.6.16 (with fuzz)
Signed-off-by: Stephen Hemminger [EMAIL
The discussion about tc action reminded my of something else I wanted
to take care of in the next time. Some of the non-work-conserving
qdiscs (HFSC, TBF, netem) need to peek at the next packet when
throttling to calculate the timeout when to wake up. This is currently
done be dequeueing a packet,
Is there currently a way to be notified of an arbitrary interface
having it's netdev feature flags modified?
If not, perhaps we could add one?
I think this would be useful for allowing 802.1Q VLANs, bridges, and other
virtual
devices to support features such as TSO.
Thanks,
Ben
--
Ben Greear
On 4/6/06, David Daney [EMAIL PROTECTED] wrote:
Janos Farkas wrote:
Shouldn't it
be more correct to not depend on the ip address of the used network,
but to use the scope parameter of the given address?
RFC 3927 specifies the Ethernet arp broadcast behavior for only
169.254.0.0/16.
Jeff == Jeff Garzik [EMAIL PROTECTED] writes:
Jeff BTW, pci_iomap() will automatically select the right thing...
Fair enough. Here's an updated patch.
M.
diff --git a/drivers/net/cassini.c b/drivers/net/cassini.c
index ac48f75..2f5ea75 100644
--- a/drivers/net/cassini.c
+++
Hi,
I've been regularly losing my IRC connections going over my PPP/PPPoE/ADSL
IPv6 connection with static addresses for quite some time now. I haven't
found any obvious correlation with PPP reconnects or anything else one
would expect to cause this - now, I've had a look at it using tcpdump
and
On Thu, 06 Apr 2006 09:57:20 -0700
Ben Greear [EMAIL PROTECTED] wrote:
Is there currently a way to be notified of an arbitrary interface
having it's netdev feature flags modified?
If not, perhaps we could add one?
I think this would be useful for allowing 802.1Q VLANs, bridges, and other
Evgeniy Polyakov [EMAIL PROTECTED] wrote:
This change removed stackable output processing, which was a good idea by
design (and there are users outside the kernel, like asynchronous crypto
processing for IPsec).
Is there any reason why you can't do async crypto with the existing
framework?
Hi, Stephen and James,
Looks like the selinux_sk_ctxid() call implemented in James' patch also
requires the sk_callback_lock (see below). I am planning to introduce a
new exported fucntion selinux_sock_ctxid() which does not require any
locking. Comments?
thanks,
Catherine
Stephen Smalley
On Fri, Apr 07, 2006 at 03:50:54AM +1000, Herbert Xu ([EMAIL PROTECTED]) wrote:
Evgeniy Polyakov [EMAIL PROTECTED] wrote:
This change removed stackable output processing, which was a good idea by
design (and there are users outside the kernel, like asynchronous crypto
processing for
Below patch was developed after discussion with Daniel Drake who
mentioned to me that wireless tools expect an EAGAIN return from getscan
so that they can wait for the scan to finish before printing out the
results.
Signed-off-by: Johannes Berg [EMAIL PROTECTED]
diff --git
Evgeniy Polyakov [EMAIL PROTECTED] wrote:
I just wonder why stackable dst_output() was removed?
It's part of the netfilter/xfrm change where this change (not really
a removal though) makes things much easier.
Cheers,
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmVHI~}
On Fri, Apr 07, 2006 at 05:34:27AM +1000, Herbert Xu ([EMAIL PROTECTED]) wrote:
Evgeniy Polyakov [EMAIL PROTECTED] wrote:
I just wonder why stackable dst_output() was removed?
It's part of the netfilter/xfrm change where this change (not really
a removal though) makes things much easier.
Hi,
On Thursday 06 April 2006 17:18, jamal wrote:
On Fri, 2006-07-04 at 00:30 +1000, Herbert Xu wrote:
If so I see what you mean but I think a better solution is to just
set a flag when the XFRM_REPLAY_TIMEOUT fires and nothing has
changed. Then when you get XFRM_REPLAY_UPDATE you can
From: Adrian Bunk [EMAIL PROTECTED]
Date: Thu, 6 Apr 2006 10:49:09 +0200
On Wed, Apr 05, 2006 at 10:25:43PM -0700, David S. Miller wrote:
- remove the following unused EXPORT_SYMBO's:
- devinet.c: devinet_ioctl
Used by wan drivers, can't remove.
Used only by drivers that:
- are
On Thu, Mar 23, 2006 at 01:06:02PM +1100, Peter Chubb wrote:
I see lots of
kernel unaligned access to 0xa001009dbb6f, ip=0xa00100811591
kernel unaligned access to 0xa001009dbb6b, ip=0xa001008115c1
kernel unaligned access to 0xa001009dbb6d,
Please apply and forward upstream.
--linas
[PATCH] PCI Error Recovery: e100 network device driver
Various PCI bus errors can be signaled by newer PCI controllers. This
patch adds the PCI error recovery callbacks to the intel ethernet e100
device driver. The patch has been tested, and appears
On Thu, Apr 06, 2006 at 05:24:00PM -0500, Linas Vepstas wrote:
+ if(pci_enable_device(pdev)) {
Add a space after if and before ( please.
You do this in a few different places.
thanks,
greg k-h
-
To unsubscribe from this list: send the line unsubscribe netdev in
the body of a message to
This patch fixes several bugs in the gianfar driver, including a major one
where spinlocks were horribly broken:
* Split gianfar locks into two types: TX and RX
* Made it so gfar_start() now clears RHALT
* Fixed a bug where calling gfar_start_xmit() with interrupts off would
corrupt the
From: jamal [EMAIL PROTECTED]
Date: Thu, 06 Apr 2006 08:56:46 -0400
Pointed out by Denis Vlasenko. Compiles with old and new schemes.
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
47 matches
Mail list logo
