Thanks for taking care of this, Ben. Do you not think the other two patches
from the "uswitch" series are important enough to be included in 1.1.1?
f23347e: netdev-linux: Fix netdev_send() to tap device.
b98162: netdev-linux: Fix blocking while sending packets.
Otherwise, the se
On Apr 11, 2011, at 6:33 PM, Ethan Jackson wrote:
> ---
> xenserver/openvswitch-xen.spec |8
> 1 files changed, 8 insertions(+), 0 deletions(-)
>
> diff --git a/xenserver/openvswitch-xen.spec b/xenserver/openvswitch-xen.spec
> index 7a2e176..c445f3d 100644
> --- a/xenserver/openvswitc
The patch looks good to me. I think a better title for the commit would
be "Only update vswitch config from XAPI if vswitch is enabled."
On 4/11/11 6:33 PM, Ethan Jackson wrote:
> ---
> xenserver/etc_init.d_openvswitch-xapi-update | 17 +
> 1 files changed, 17 insertions(+), 0 d
The patch looks fine to me. I'd prefer "upgrading" to "installing" in
the commit title, since this only happens on upgrade.
On 4/11/11 6:33 PM, Ethan Jackson wrote:
> ---
> xenserver/openvswitch-xen.spec |8
> 1 files changed, 8 insertions(+), 0 deletions(-)
>
> diff --git a/xenserv
This seems to be required by glibc 2.11.2.
---
tests/automake.mk |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/tests/automake.mk b/tests/automake.mk
index 7872d21..e02a951 100644
--- a/tests/automake.mk
+++ b/tests/automake.mk
@@ -226,7 +226,7 @@ tests_test_lockfile_LDA
---
lib/pcap.c |6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/pcap.c b/lib/pcap.c
index afd41fa..e860bc1 100644
--- a/lib/pcap.c
+++ b/lib/pcap.c
@@ -100,7 +100,7 @@ pcap_write_header(FILE *file)
ph.sigfigs = 0;
ph.snaplen = 1518;
ph.network = 1;
These patches are required to make Open vSwitch compile cleanly on Gentoo
systems.
Andrew Evans (2):
pcap: Silence warnings about fwrite(3) return value being ignored.
tests: Link test-multipath with libm.
lib/pcap.c|6 +++---
tests/automake.mk |2 +-
2 files changed, 4 inser
I haven't tested this series, but I will be sure to before merging.
Ethan
On Mon, Apr 11, 2011 at 6:33 PM, Ethan Jackson wrote:
> ---
> xenserver/etc_init.d_openvswitch-xapi-update | 17 +
> 1 files changed, 17 insertions(+), 0 deletions(-)
>
> diff --git a/xenserver/etc_init.
---
xenserver/openvswitch-xen.spec |8
1 files changed, 8 insertions(+), 0 deletions(-)
diff --git a/xenserver/openvswitch-xen.spec b/xenserver/openvswitch-xen.spec
index 7a2e176..c445f3d 100644
--- a/xenserver/openvswitch-xen.spec
+++ b/xenserver/openvswitch-xen.spec
@@ -235,6 +235,
---
xenserver/etc_init.d_openvswitch-xapi-update | 17 +
1 files changed, 17 insertions(+), 0 deletions(-)
diff --git a/xenserver/etc_init.d_openvswitch-xapi-update
b/xenserver/etc_init.d_openvswitch-xapi-update
index 7b2efd1..14d2cc9 100755
--- a/xenserver/etc_init.d_openvswit
This patch adds test designed to verify the correctness of the
parsing function introduced with the autopath action.
---
tests/automake.mk |1 +
tests/autopath.at | 35 +++
tests/testsuite.at |1 +
3 files changed, 37 insertions(+), 0 deletions(-)
creat
These fix a number of small, miscellaneous bugs that I noticed while
working on other issues.
Ben Pfaff (6):
ofproto: Avoid memory leak in classifier on destruction.
bridge: Initialize mirrors' uuid member.
daemon: Reduce log level of "pid file is stale" message.
bridge: Reintroduce log me
ofproto_flush_flows() flushes the flow table but then it reintroduces flows
required by fail-open or in-band. These are then leaked when the
classifier is destroyed a little later.
This fixes the problem by not reintroducing these flows when ofproto is
being destroyed.
---
ofproto/ofproto.c |
These macros expanded the LEVEL argument without protecting it with
parentheses, which meant that an argument like 'cond ? VLL_DBG : VLL_WARN'
did not have the desired effect (and caused a GCC warning).
This commit fixes the problem and avoids expanding LEVEL more than once,
too.
---
lib/vlog.h |
Otherwise mirrors get destroyed and re-created on every reconfiguration.
---
vswitchd/bridge.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/vswitchd/bridge.c b/vswitchd/bridge.c
index 59100cc..1e5caf9 100644
--- a/vswitchd/bridge.c
+++ b/vswitchd/bridge.c
@@ -3504,6 +35
This message will appear repeatedly when ovs-vswitchd is running, if there
is any stale pidfile in /var/run/openvswitch, because ovs-vswitchd reads
all of the pidfiles in that directory periodically to update statistics.
---
lib/daemon.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
Setting the 'mac' in the Interface record for a bridge's local port has
always been ineffective, but the log message was suppressed because of a
check at too high of a level. This commit fixes the problem. It also
fixes the wording of the log message, which has been obsolete since the
introductio
The logic here did not make sense. A packet arriving on a port is mirrored
if the port is a mirroring source port AND (not OR) the packet is in one of
the VLANs that is mirrored.
This test has been here since the mirroring code was introduced. It seems
to me that it was never correct.
---
vswit
ENOENT can be returned if the kernel module isn't loaded. If that's the
case then we've already logged that and there's no point in logging it
again.
---
lib/dpif-linux.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/lib/dpif-linux.c b/lib/dpif-linux.c
index fd08e64..f
---
vswitchd/vswitch.xml | 17 +
1 files changed, 13 insertions(+), 4 deletions(-)
diff --git a/vswitchd/vswitch.xml b/vswitchd/vswitch.xml
index dc6c197..a16ddfb 100644
--- a/vswitchd/vswitch.xml
+++ b/vswitchd/vswitch.xml
@@ -1522,6 +1522,12 @@
+
+T
These are cherry-picked commits from "master" that also apply
to branch-1.1, presumably for inclusion in OVS 1.1.1. All of
these have been reviewed already for master, so the goal for
this review cycle is to determine whether they are suitable
for 1.1.
Ben Pfaff (3):
vswitch: Improve schema doc
netdev_linux_get_stats() calls into netdev_vport_get_stats(), which in
turn attempts a transaction on genl_sock. If the kernel module isn't
loaded, then genl_sock won't be there, and in any case there's nothing that
guarantees that it's been initialized yet.
This fixes the problem by ensuring tha
> I guess only values in the range 1...1023 and OFPP_LOCAL are currently
> useful? I guess I don't care then. The current code is fine.
Anything else they type in is going to result in OFPP_NONE going into
the register. We do document in the ofctl man page that openflow port
numbers should be u
On Mon, Apr 11, 2011 at 10:21:27AM -0700, Andrew Evans wrote:
> On 4/8/11 4:41 PM, Ben Pfaff wrote:
> > ENOENT can be returned if the kernel module isn't loaded. If that's the
> > case then we've already logged that and there's no point in logging it
> > again.
>
> Looks good to me.
Thanks.
I f
On Mon, Apr 11, 2011 at 10:48:41AM -0700, Ethan Jackson wrote:
> > autopath_parse() rejects nonpositive 'id' values but the
> > NXAST_AUTOPATH action description doesn't say that only positive
> > values are valid.
>
> I'm not sure here whether the correct thing to do is document that
> positive I
> autopath_parse() rejects nonpositive 'id' values but the
> NXAST_AUTOPATH action description doesn't say that only positive
> values are valid.
I'm not sure here whether the correct thing to do is document that
positive ID values are required, or loosen the requirement. From the
action's persp
On Fri, Apr 08, 2011 at 06:35:22PM -0700, Ethan Jackson wrote:
> The newly created autopath action will be the way OpenFlow
> interacts with the existing bonding infrastructure.
Looks good.
In autopath_parse(), you can omit the extra " " in the first line of:
> +ovs_fatal(0, "%s: autopath
On 4/8/11 4:41 PM, Ben Pfaff wrote:
> ENOENT can be returned if the kernel module isn't loaded. If that's the
> case then we've already logged that and there's no point in logging it
> again.
Looks good to me.
___
dev mailing list
dev@openvswitch.org
ht
On 4/8/11 4:41 PM, Ben Pfaff wrote:
> netdev_linux_get_stats() calls into netdev_vport_get_stats(), which in
> turn attempts a transactino on genl_sock. If the kernel module isn't
^^^
What's a transactino? Is that a miniature Italian transaction?
Looks good to me otherw
On 4/8/11 4:41 PM, Ben Pfaff wrote:
> Commit 76c308b50d3 "netdev-linux: Support 'send' for netdevs opened with
> NETDEV_ETH_TYPE_NONE" broke sending packets to tap devices. Sending a
> packet to a tap device with an AF_PACKET socket causes that packet to be
> looped back to be received on the tap
On 4/8/11 4:41 PM, Ben Pfaff wrote:
> The AF_PACKET socket needs to be in nonblocking mode or trying to send
> a packet can take a long time.
Looks good to me.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Thanks, I pushed this one.
On Sun, Apr 10, 2011 at 11:17:39PM -0700, Justin Pettit wrote:
> Looks good to me.
>
> --Justin
>
>
> On Apr 8, 2011, at 4:45 PM, Ben Pfaff wrote:
>
> > Obviously correct code is easier on everyone. As the C FAQ says:
___
32 matches
Mail list logo
