On 18/12/2013 08:54, Ben Pfaff wrote:
On Mon, Dec 09, 2013 at 06:41:13PM +0800, Alexander Wu wrote:
v4:
No change.
v3:
No change.
v2:
No change
v1:
Now the cli we implement is very crude. Maybe it could
display better.
Signed-off-by: Alexander Wu
Reviewed-by: Simon Horman
On 18/12/2013 08:51, Ben Pfaff wrote:
On Mon, Dec 09, 2013 at 06:41:11PM +0800, Alexander Wu wrote:
v4.3:
Rollback table id 255 to 253 according to OpenFlow spec and current implement
Update AT.
v4.2:
Fix last table features id to 255.
Add more comments.
Fix wrong print of last t
On 18/12/2013 08:53, Ben Pfaff wrote:
On Mon, Dec 09, 2013 at 06:41:12PM +0800, Alexander Wu wrote:
v4.3:
Fix last table features id to 253 according to current
imlement and OpenFlow spec.
v4.2:
Fix last table features id to 255.
v4:
1. update abstract table-features to bitmap
2
On 18/12/2013 08:50, Ben Pfaff wrote:
On Mon, Dec 09, 2013 at 06:41:09PM +0800, Alexander Wu wrote:
v4:
Add enums of ofp13_action_type and make them work.
v3:
Add func to get instruction/action name by OpenFlow instruction type
directly.
Can I have a Signed-off-by?
Sure!
GCC repor
On Wed, Dec 18, 2013 at 07:39:00AM +0200, Aleksi Suhonen wrote:
> >OVS 1.9 doesn't support Linux 3.11 as it predates the release of that
> >kernel. However, in many cases you can use the in-tree version of the
> >kernel module directly (namely, if you aren't using tunneling).
> >Otherwise, the fix
Hi,
OVS 1.9 doesn't support Linux 3.11 as it predates the release of that
kernel. However, in many cases you can use the in-tree version of the
kernel module directly (namely, if you aren't using tunneling).
Otherwise, the fix you suggested is the one used in later versions of
OVS.
Oh wait, OV
On Tue, Dec 17, 2013 at 05:29:44PM -0800, Jarno Rajahalme wrote:
>
> On Dec 17, 2013, at 5:01 PM, Ben Pfaff wrote:
>
> > I think that some developers haven't noticed this, but it's useful, so
> > mention it.
> >
> > Signed-off-by: Ben Pfaff
> > ---
> > INSTALL |5 +
> > 1 file changed,
On Tue, Dec 17, 2013 at 05:13:38PM -0800, Ben Pfaff wrote:
> On Mon, Dec 16, 2013 at 05:53:22PM +0900, Simon Horman wrote:
> > Add table config to to struct ofproto and set it
> > when a table mod message is received.
> >
> > This is in preparation for changing the behaviour of the switch
> > base
On Tue, Dec 17, 2013 at 05:05:40PM -0800, Ben Pfaff wrote:
> On Mon, Dec 16, 2013 at 05:57:51PM +0900, Simon Horman wrote:
> > Update the netdev_class so that struct ofpbuf * is passed to rx_recv()
> > to provide both the data and size of the data to read a packet into.
> >
> > On success, update
On Tue, Dec 17, 2013 at 05:08:35PM -0800, Ben Pfaff wrote:
> On Mon, Dec 16, 2013 at 05:57:52PM +0900, Simon Horman wrote:
> > If VLAN acceleration is used when the kernel receives a packet
> > then the outer-most VLAN tag will not be present in the packet
> > when it is received by netdev-linux. R
On Dec 17, 2013, at 5:01 PM, Ben Pfaff wrote:
> I think that some developers haven't noticed this, but it's useful, so
> mention it.
>
> Signed-off-by: Ben Pfaff
> ---
> INSTALL |5 +
> 1 file changed, 5 insertions(+)
>
> diff --git a/INSTALL b/INSTALL
> index c8b1aa7..27c0e5f 100644
>
On Mon, Dec 16, 2013 at 05:53:22PM +0900, Simon Horman wrote:
> Add table config to to struct ofproto and set it
> when a table mod message is received.
>
> This is in preparation for changing the behaviour of the switch
> based on table config.
>
> Cc: Andy Zhou
> Signed-off-by: Simon Horman
Thanks for the review. Replies under .
Linda
- Original Message -
From: "Ben Pfaff"
To: "Linda Sun"
Cc: dev@openvswitch.org
Sent: Tuesday, December 17, 2013 4:46:44 PM
Subject: Re: [ovs-dev] [PATCH] pool-loop: windows poll_block implementation
On Thu, Dec 12, 2013 at 04:28:12PM -0800
On Mon, Dec 16, 2013 at 05:57:52PM +0900, Simon Horman wrote:
> If VLAN acceleration is used when the kernel receives a packet
> then the outer-most VLAN tag will not be present in the packet
> when it is received by netdev-linux. Rather, it will be present
> in auxdata.
>
> This patch uses recvms
On Mon, Dec 16, 2013 at 05:57:51PM +0900, Simon Horman wrote:
> Update the netdev_class so that struct ofpbuf * is passed to rx_recv()
> to provide both the data and size of the data to read a packet into.
>
> On success, update struct ofpbuf size inside netdev_class rx_recv
> implementation and r
I think that some developers haven't noticed this, but it's useful, so
mention it.
Signed-off-by: Ben Pfaff
---
INSTALL |5 +
1 file changed, 5 insertions(+)
diff --git a/INSTALL b/INSTALL
index c8b1aa7..27c0e5f 100644
--- a/INSTALL
+++ b/INSTALL
@@ -198,6 +198,11 @@ Prerequisites secti
On Mon, Dec 09, 2013 at 06:41:13PM +0800, Alexander Wu wrote:
> v4:
> No change.
>
> v3:
> No change.
>
> v2:
> No change
>
> v1:
> Now the cli we implement is very crude. Maybe it could
> display better.
>
> Signed-off-by: Alexander Wu
> Reviewed-by: Simon Horman
Will you add docu
On Mon, Dec 09, 2013 at 06:41:12PM +0800, Alexander Wu wrote:
> v4.3:
> Fix last table features id to 253 according to current
> imlement and OpenFlow spec.
>
> v4.2:
> Fix last table features id to 255.
>
> v4:
> 1. update abstract table-features to bitmap
> 2. update ofproto, reduce c
On Mon, Dec 09, 2013 at 06:41:11PM +0800, Alexander Wu wrote:
> v4.3:
> Rollback table id 255 to 253 according to OpenFlow spec and current
> implement
> Update AT.
>
> v4.2:
> Fix last table features id to 255.
> Add more comments.
> Fix wrong print of last table feature.
>
> v4:
>
On Mon, Dec 09, 2013 at 06:41:09PM +0800, Alexander Wu wrote:
> v4:
> Add enums of ofp13_action_type and make them work.
>
> v3:
> Add func to get instruction/action name by OpenFlow instruction type
> directly.
Can I have a Signed-off-by?
GCC reports:
../lib/ofp-parse.c: In function 'pars
On Mon, Dec 09, 2013 at 06:41:10PM +0800, Alexander Wu wrote:
> v4.1:
> Update bitmap_count1 function: call count_1bits to improve
> performance.
>
> v4:
> Add bitmap_count1 function to count all 1 bits.
Acked-by: Ben Pfaff
Can I have a Signed-off-by, please?
_
On Thu, Dec 12, 2013 at 04:28:12PM -0800, Linda Sun wrote:
> Use WaitForMultipleObjects for polling on windows. This works on all kinds
> of objects, e.g. sockets, files, especially ioctl calls to the kernel.
> One additional paramater is passed down in poll_fd_wait() to help
> WaitForMultipleO
On Wed, Nov 27, 2013 at 02:08:41PM +0900, Simon Horman wrote:
> The skb_priority, pkt_mark and tunl parameters of dp_netdev_port_input()
> are always passed as 0, 0 and NULL respectively. So rather than
> passing these values to dp_netdev_port_input() just use them directly.
>
> Signed-off-by: Sim
On Fri, Dec 13, 2013 at 06:54:28PM +0100, Helmut Schaa wrote:
> Currently openvswitch builds all libraries static only. However,
> libopenvswitch is linked into nearly all openvswitch executables
> making it hardly possible to run openvswitch on embedded devices
> (for example running OpenWrt).
>
Pushed,
Jarno
On Dec 17, 2013, at 1:56 PM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 10:43:11AM -0800, Jarno Rajahalme wrote:
>> We already set all the fields of the upcall, so memsetting right before
>> is unnecessary.
>>
>> Signed-off-by: Jarno Rajahalme
>
> Acked-by: Ben Pfaff
__
On Tue, Dec 17, 2013 at 3:43 PM, Pravin B Shelar wrote:
> Stats-read needs to lock stats but same lock is taken in stats
> update in irq context. Therefore it needs to disable irq to
> avoid following deadlock :-
>
> BUG: soft lockup - CPU#1 stuck for 23s! [ovs-vswitchd:1425]
> CPU 1
> Pid: 1425,
On Tue, Dec 17, 2013 at 03:59:52PM -0800, Jarno Rajahalme wrote:
> Pushed, thanks for reviews!
>
> Some comments below,
>
> Jarno
>
> On Dec 17, 2013, at 1:03 PM, Ben Pfaff wrote:
>
> > On Tue, Dec 17, 2013 at 10:22:19AM -0800, Jarno Rajahalme wrote:
> >> Commit da546e0 (dpif: Allow execute
Pushed, thanks for reviews!
Some comments below,
Jarno
On Dec 17, 2013, at 1:03 PM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 10:22:19AM -0800, Jarno Rajahalme wrote:
>> Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
>> the "dpif_upcall.packet" of "struct upcall" w
On Mon, Dec 16, 2013 at 09:30:11AM -0800, Alex Wang wrote:
> Commit 307464a11 (ofproto-dpif-monitor: Use heap to order the mport
> wakeup time.) makes bfd only send packet at specified periodic instant.
> This fails to meet the RFC5880 requirement, which requires bfd send
> FINAL immediately after
On Tue, Dec 17, 2013 at 1:59 PM, Jesse Gross wrote:
> On Tue, Dec 17, 2013 at 1:48 PM, Pravin B Shelar wrote:
>> diff --git a/datapath/flow.c b/datapath/flow.c
>> index 9b3d3a7..3a50589 100644
>> --- a/datapath/flow.c
>> +++ b/datapath/flow.c
>> @@ -110,7 +110,9 @@ void ovs_flow_stats_get(struct
On Fri, Dec 13, 2013 at 11:29:09AM -0800, Alex Wang wrote:
> Commit 307464a1 (ofproto-dpif-monitor: Use heap to order the mport
> wakeup time.) re-heapifies the heap in monitor_run(). So the
> monitor_run() should be protected by the write lock, rather than
> the read lock.
>
> This commit fixes
Stats-read needs to lock stats but same lock is taken in stats
update in irq context. Therefore it needs to disable irq to
avoid following deadlock :-
BUG: soft lockup - CPU#1 stuck for 23s! [ovs-vswitchd:1425]
CPU 1
Pid: 1425, comm: ovs-vswitchd Tainted: G O 3.2.39-server-nn23 #1
VMwar
On Mon, Dec 09, 2013 at 04:21:23AM -0800, Arun Sharma wrote:
> Signed-off-by: Nandan Nivgune
> Signed-off-by: Abhijit Bhopatkar
> Signed-off-by: Arun Sharma
Thanks!
I think that we should update vswitch.xml, too.
We should also add an item to NEWS.
I would fold this patch in with patch 1.
T
On Mon, Dec 09, 2013 at 04:21:22AM -0800, Arun Sharma wrote:
> Also added tests for NetFlow and sFlow using IPv6 sockets.
>
> Signed-off-by: Nandan Nivgune
> Signed-off-by: Abhijit Bhopatkar
> Signed-off-by: Arun Sharma
Thanks!
I think we should fold these into patch 1.
__
On Mon, Dec 09, 2013 at 04:21:21AM -0800, Arun Sharma wrote:
> Allows all socket operations to use ipv6 network addresses,
> except in-band control which is not supported for ipv6
>
> Signed-off-by: Nandan Nivgune
> Signed-off-by: Abhijit Bhopatkar
> Signed-off-by: Arun Sharma
Thank you for t
sure, thanks,
On Tue, Dec 17, 2013 at 3:12 PM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 02:37:10PM -0800, Alex Wang wrote:
> > This commit adds check of sock pointer in dpif_linux_port_get_pid().
> > If the pointer is NULL, do not call nl_sock_pid().
> >
> > Signed-off-by: Alex Wang
> > ---
On Tue, Dec 17, 2013 at 02:37:10PM -0800, Alex Wang wrote:
> This commit adds check of sock pointer in dpif_linux_port_get_pid().
> If the pointer is NULL, do not call nl_sock_pid().
>
> Signed-off-by: Alex Wang
> ---
> lib/dpif-linux.c |3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-
On Tue, Dec 17, 2013 at 11:22 AM, Stephen Hemminger
wrote:
> Several functions and datastructures could be local
> Found with 'make namespacecheck'
>
> Signed-off-by: Stephen Hemminger
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
http://op
Acked-by: Ethan Jackson
On Tue, Dec 17, 2013 at 4:53 PM, Ben Pfaff wrote:
> These functions don't have any ultimate users. The in-band control code
> used to use them, but not anymore, so we might as well delete them all.
>
> Signed-off-by: Ben Pfaff
> ---
> lib/rconn.c | 57
> -
These functions don't have any ultimate users. The in-band control code
used to use them, but not anymore, so we might as well delete them all.
Signed-off-by: Ben Pfaff
---
lib/rconn.c | 57 -
lib/rconn.h |4
lib/str
Looks good to me, & tested on my vm,
Thanks,
Alex Wang,
On Tue, Dec 17, 2013 at 1:03 PM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 10:22:19AM -0800, Jarno Rajahalme wrote:
> > Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
> > the "dpif_upcall.packet" of "struct upca
Commit 222837 (dpif-linux: Factor out port dumping helper functions.)
introduced a bug by making dpif_linux_port_dump_next__() return 'bool'
instead of 'int' as defined in dpif-provider.h. This bug causes ovs-
vswitchd failure with SEGFAULT when processing slow-path packet.
This commit fixes the
This commit adds check of sock pointer in dpif_linux_port_get_pid().
If the pointer is NULL, do not call nl_sock_pid().
Signed-off-by: Alex Wang
---
lib/dpif-linux.c |3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/lib/dpif-linux.c b/lib/dpif-linux.c
index 52c3f65..a88f8a
On Fri, Dec 06, 2013 at 12:22:07PM -0800, Joe Stringer wrote:
> In the case where we have not yet sent a control packet for a bfd
> connection, and we receive a control packet from the remote host,
> bfd->next_tx is updated to an unusual value. This causes the logging to
> incorrectly report that t
Thanks for the reviews! I applied these to master.
On Tue, Dec 17, 2013 at 11:26:34AM -0800, Andy Zhou wrote:
> Looks good.
>
> Acked-by: Andy Zhou
>
>
> On Fri, Dec 13, 2013 at 12:47 PM, Ben Pfaff wrote:
>
> > These tests configured LACP and then immediately dumped out its state.
> > Most
On Tue, Dec 17, 2013 at 1:48 PM, Pravin B Shelar wrote:
> diff --git a/datapath/flow.c b/datapath/flow.c
> index 9b3d3a7..3a50589 100644
> --- a/datapath/flow.c
> +++ b/datapath/flow.c
> @@ -110,7 +110,9 @@ void ovs_flow_stats_get(struct sw_flow *flow, struct
> ovs_flow_stats *ovs_stats,
>
The input variable of ovs_scan is changed from 'template' to
'format'. template is a keyword in C++.
Signed-off-by: Harold Lim
---
lib/util.c |8
lib/util.h |2 +-
2 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/lib/util.c b/lib/util.c
index 0227822..9f37350 10064
On Tue, Dec 17, 2013 at 10:43:11AM -0800, Jarno Rajahalme wrote:
> We already set all the fields of the upcall, so memsetting right before
> is unnecessary.
>
> Signed-off-by: Jarno Rajahalme
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.or
OK.
On Tue, Dec 17, 2013 at 01:26:43PM -0800, Jesse Gross wrote:
> At this point things are being added too frequently for it to be of
> much use (for example, currently the most recent kernel that adds new
> OVS features is the unreleased 3.13 RC). To me, it seems better to
> include it in the do
Thanks for the review, I applied this to master.
On Tue, Dec 17, 2013 at 11:21 AM, Pritesh Kothari (pritkoth)
wrote:
> looks good to me.
>
> Acked-by: pritesh
>
> On Dec 17, 2013, at 10:51 AM, Jesse Gross wrote:
>
>> Reported-by: Ben Pfaff
>> Signed-off-by: Jesse Gross
>> ---
>> FAQ | 23
On Tue, Dec 17, 2013 at 10:08 AM, Thomas Graf wrote:
> On 12/17/2013 06:49 PM, Jesse Gross wrote:
>>
>> I think there's also a potential downgrade issue if we add a new
>> feature to the list of capabilities - it won't automatically reset
>> since userspace is now using v2 of the netlink protocol.
Stats-read needs to lock stats but same lock is taken in stats
update in irq context. Therefore it needs to disable irq to
avoid following deadlock :-
BUG: soft lockup - CPU#1 stuck for 23s! [ovs-vswitchd:1425]
CPU 1
Pid: 1425, comm: ovs-vswitchd Tainted: G O 3.2.39-server-nn23 #1
VMwar
On Mon, Dec 16, 2013 at 10:17:24PM -0800, Ben Pfaff wrote:
> On Mon, Nov 25, 2013 at 08:19:50AM -0800, Bruce Davie wrote:
> > Add two new tables to the VTEP schema in support of distributed L3.
> > Each table contains MAC addresses to be used by VTEPs (both hardware
> > and software) when issuing A
On Tue, Dec 17, 2013 at 08:18:18PM +0100, Francesco Fusco wrote:
> The command ovs-dpctl can wrongly output the masks even if the
> datapath does not implement mega flows. In this case the output
> will be similar to the following:
>
> system@ovs-system:
> lookups: hit:14 missed:41 lost:0
>
At this point things are being added too frequently for it to be of
much use (for example, currently the most recent kernel that adds new
OVS features is the unreleased 3.13 RC). To me, it seems better to
include it in the documentation/log messages for individual features.
On Tue, Dec 17, 2013 at
Looks good.
Acked-by: Andy Zhou
On Fri, Dec 13, 2013 at 12:47 PM, Ben Pfaff wrote:
> These tests configured LACP and then immediately dumped out its state.
> Most of the time, this worked, but there was a brief race window in which
> the "negotiated" flag could be missing because this took on
On Tue, Dec 17, 2013 at 10:32:11AM -0800, Harold Lim wrote:
> The input variable of ovs_scan is changed from 'template' to
> 'format'. template is a keyword in C++.
>
> Signed-off-by: Harold Lim
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
On Tue, Dec 17, 2013 at 10:32:12AM -0800, Harold Lim wrote:
> This allows other libraries to use util.h that has already
> defined NOT_REACHED.
>
> Signed-off-by: Harold Lim
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.o
On Dec 17, 2013, at 1:00 PM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 12:53:36PM -0800, Justin Pettit wrote:
>> Signed-off-by: Justin Pettit
>
> Acked-by: Ben Pfaff
Thanks. I pushed it.
--Justin
___
dev mailing list
dev@openvswitch.org
http:/
This is apparently used by CentOS 6.5.
Reported-by: Phil Daws
Reported-by: Madko
Signed-off-by: Jesse Gross
---
v2: Check for netdev_features_t in netdevice.h instead as this works
on all kernel versions.
---
acinclude.m4| 1 +
datapath/linux/compat/inc
From: Francesco Fusco
Date: Thu, 12 Dec 2013 16:09:04 +0100
> From: Daniel Borkmann
>
> We are introducing a fast hash function (see patch1) that can be
> used in the context of OpenVSwitch to reduce the hashing footprint
> (patch2). For details, please see individual patches!
>
> Thanks !
>
On Tue, Dec 17, 2013 at 10:22:19AM -0800, Jarno Rajahalme wrote:
> Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
> the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
> The packet ofpbuf is likely uninitialized in this case, hence calling
> ofpbuf_
On Tue, Dec 17, 2013 at 12:53:36PM -0800, Justin Pettit wrote:
> Signed-off-by: Justin Pettit
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
I see your point, but I'm still sure users will want to know. Do we
add new features in every kernel version? That is, can we say
something like "3.x is the latest kernel version that adds new OVS
features" or does that not make any sense?
On Tue, Dec 17, 2013 at 12:53:25PM -0800, Jesse Gross wr
We already set all the fields of the upcall, so memsetting right before
is unnecessary.
Signed-off-by: Jarno Rajahalme
---
lib/dpif-linux.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/dpif-linux.c b/lib/dpif-linux.c
index 8fa42a9..9e901be 100644
--- a/lib/dpif-linu
I think we would need to have a pretty large matrix of features and
kernel versions as this is no longer just about closing a gap between
upstream and out-of-tree. For example, IPFIX support went into both
places at the same time but not all upstream modules support it. Same
thing for TCP flags and
Signed-off-by: Justin Pettit
---
FAQ |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/FAQ b/FAQ
index 3df1c54..1edcd94 100644
--- a/FAQ
+++ b/FAQ
@@ -1082,7 +1082,7 @@ A: Open vSwitch 1.9 and earlier support only OpenFlow 1.0
(plus
ovs-vsctl set bridge br0 prot
I see, thanks for the explanation.
On 17 December 2013 10:02, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 08:48:15AM -0800, Joe Stringer wrote:
>> On 15 December 2013 17:57, Ethan Jackson wrote:
>> > +if (!revalidator->n_udumps) {
>> > +if (revalidator->dump_seq != seq_read(ud
Reported-by: Ben Pfaff
Signed-off-by: Jesse Gross
---
FAQ | 23 +--
1 file changed, 9 insertions(+), 14 deletions(-)
diff --git a/FAQ b/FAQ
index 2508c8c..3df1c54 100644
--- a/FAQ
+++ b/FAQ
@@ -173,22 +173,17 @@ Q: Should userspace or kernel be upgraded first to
minimize do
The command ovs-dpctl can wrongly output the masks even if the
datapath does not implement mega flows. In this case the output
will be similar to the following:
system@ovs-system:
lookups: hit:14 missed:41 lost:0
flows: 0
masks: hit:18446744073709551615 total:4294967295
looks good to me.
Acked-by: pritesh
On Dec 17, 2013, at 10:51 AM, Jesse Gross wrote:
> Reported-by: Ben Pfaff
> Signed-off-by: Jesse Gross
> ---
> FAQ | 23 +--
> 1 file changed, 9 insertions(+), 14 deletions(-)
>
> diff --git a/FAQ b/FAQ
> index 2508c8c..3df1c54 100644
>
Could we mention the kernel version that added support? It's going to
be the next question people ask.
On Tue, Dec 17, 2013 at 11:27:35AM -0800, Jesse Gross wrote:
> Thanks for the review, I applied this to master.
>
> On Tue, Dec 17, 2013 at 11:21 AM, Pritesh Kothari (pritkoth)
> wrote:
> > lo
The change looks fine.
On Fri, Dec 13, 2013 at 12:47 PM, Ben Pfaff wrote:
> Occasionally in the unit tests the following race can happen:
>
>1. ovs-vsctl updates database
>2. ovs-vswitchd reconfigures, notifies ovs-vsctl that it is complete
>3. ovs-appctl ofproto/trace fails to see
Several functions and datastructures could be local
Found with 'make namespacecheck'
Signed-off-by: Stephen Hemminger
--- a/net/openvswitch/datapath.c2013-12-05 14:47:26.092495160 -0800
+++ b/net/openvswitch/datapath.c2013-12-17 09:11:42.744892746 -0800
@@ -133,7 +133,7 @@ stati
On Dec 17, 2013, at 10:10 AM, Ben Pfaff wrote:
> On Tue, Dec 17, 2013 at 12:16:23AM -0800, Alex Wang wrote:
>> Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
>> the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
>> Since the "struct upcall" is a
Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
The packet ofpbuf is likely uninitialized in this case, hence calling
ofpbuf_uninit() on it will likely cause a SEGFAULT.
This commit fixes this bug
On Tue, Dec 17, 2013 at 12:16:23AM -0800, Alex Wang wrote:
> Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
> the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
> Since the "struct upcall" is allocated via xmalloc, this will cause SEGFAULT
> if dpi
On 12/17/2013 06:49 PM, Jesse Gross wrote:
I think there's also a potential downgrade issue if we add a new
feature to the list of capabilities - it won't automatically reset
since userspace is now using v2 of the netlink protocol. Obviously,
this isn't an issue yet but it we should make sure tha
On Tue, Dec 17, 2013 at 9:58 AM, Ben Pfaff wrote:
> The FAQ says:
>
> Q: What features are not available in the Open vSwitch kernel datapath
>that ships as part of the upstream Linux kernel?
>
> A: The kernel module in upstream Linux 3.3 and later does not include
>tunnel v
On Tue, Dec 17, 2013 at 08:48:15AM -0800, Joe Stringer wrote:
> On 15 December 2013 17:57, Ethan Jackson wrote:
> > +if (!revalidator->n_udumps) {
> > +if (revalidator->dump_seq != seq_read(udpif->dump_seq)) {
> > +revalidator->dump_seq = seq_read(udpif->dump_se
The FAQ says:
Q: What features are not available in the Open vSwitch kernel datapath
that ships as part of the upstream Linux kernel?
A: The kernel module in upstream Linux 3.3 and later does not include
tunnel virtual ports, that is, interfaces with type "gre",
"ipse
On Tue, Dec 17, 2013 at 1:52 AM, Thomas Graf wrote:
> On 12/17/2013 02:23 AM, Jesse Gross wrote:
>>
>> On Sat, Nov 30, 2013 at 4:25 AM, Thomas Graf wrote:
>>>
>>> diff --git a/lib/dpif-linux.c b/lib/dpif-linux.c
>>> index 6c482d0..2d8a1aa 100644
>>> --- a/lib/dpif-linux.c
>>> +++ b/lib/dpif-linux
On Tue, Dec 17, 2013 at 12:58 AM, Phil Daws wrote:
> Jesse,
>
> One of the hunks failed:
>
> [compuser@kvm02 openvswitch-2.0.0]$ patch -p1 < p3.patch
> patching file datapath/linux/compat/netdevice.c
> Hunk #1 succeeded at 7 with fuzz 2 (offset 4 lines).
> Hunk #2 succeeded at 35 with fuzz 2 (offs
Alex,
Thanks for spotting this. I had momentarily missed the fact that
ofpbuf_use_stub() sets the size to zero.
Pushed to master,
Jarno
On Dec 17, 2013, at 12:16 AM, Alex Wang wrote:
> Commit da546e0 (dpif: Allow execute to modify the packet.) introduced
> a bug by subtracting the zero-val
sure,
> So the problem is that in error cases the ofpbuf packet member of upcall
> has been uninitialized. The fix you propose works because all-zero ofpbuf
> can be safely uninitialized.
Yes, my fix is more like a hack that relies on this.
> We have not documented this behavior anywhere, so
On Dec 17, 2013, at 12:16 AM, Alex Wang wrote:
> Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
> the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
> Since the "struct upcall" is allocated via xmalloc, this will cause SEGFAULT
> if dpif_recv()
On 15 December 2013 17:57, Ethan Jackson wrote:
> +if (!revalidator->n_udumps) {
> +if (revalidator->dump_seq != seq_read(udpif->dump_seq)) {
> +revalidator->dump_seq = seq_read(udpif->dump_seq);
> +revalidator_sweep(revalidator);
> +
On Mon, Dec 16, 2013 at 6:02 PM, Alin Serdean
wrote:
> I added also syslog.h as you suggested(this will help us in the future
> patches).
>
> Regarding the warnings I will send a detailed list iwth their corespondence
> in another mail and we can discuss further if we want to add them or not.
>
On 12/17/2013 02:23 AM, Jesse Gross wrote:
On Sat, Nov 30, 2013 at 4:25 AM, Thomas Graf wrote:
diff --git a/lib/dpif-linux.c b/lib/dpif-linux.c
index 6c482d0..2d8a1aa 100644
--- a/lib/dpif-linux.c
+++ b/lib/dpif-linux.c
@@ -73,6 +73,7 @@ struct dpif_linux_dp {
/* Attributes. */
cons
Jesse,
One of the hunks failed:
[compuser@kvm02 openvswitch-2.0.0]$ patch -p1 < p3.patch
patching file datapath/linux/compat/netdevice.c
Hunk #1 succeeded at 7 with fuzz 2 (offset 4 lines).
Hunk #2 succeeded at 35 with fuzz 2 (offset 4 lines).
Hunk #3 FAILED at 45.
Hunk #4 succeeded at 74 with fu
Commit da546e0 (dpif: Allow execute to modify the packet.) uninitializes
the "dpif_upcall.packet" of "struct upcall" when dpif_recv() returns error.
Since the "struct upcall" is allocated via xmalloc, this will cause SEGFAULT
if dpif_recv() returns error before memset the memory to all zero.
This
Commit da546e0 (dpif: Allow execute to modify the packet.) introduced
a bug by subtracting the zero-value ofpbuf size by "sizeof(struct
nlattr)" and assigning the result back to the ofpbuf size. This bug
causes the ovs-assert failure in facet_push_stats().
This commit fixes the bug by assigning t
92 matches
Mail list logo
