From: Jesse Gross
Date: Wed, 7 Mar 2012 18:56:06 -0800
> A few last minute patches for net/3.3 if there is time.
>
> The following changes since commit dcd6c92267155e70a94b3927bce681ce74b80d1f:
>
> Linux 3.3-rc1 (2012-01-19 15:04:48 -0800)
>
> are available in the git repository at:
> git
On Wed, Mar 7, 2012 at 4:13 PM, Christian Benvenuti (benve)
wrote:
> Hi,
> since both the kernel GRE module and the OVS datapath vport-gre port
> type register with the kernel a handler for the GRE protocol
> (inet_add_protocol), does it mean that those two GRE handlers
> are mutually exclusive?
When modifying IP addresses or ports on a UDP packet we don't
correctly follow the rules for unchecksummed packets. This meant
that packets without a checksum can be given a incorrect new checksum
and packets with a checksum can become marked as being unchecksummed.
This fixes it to handle those r
From: Ben Pfaff
When OVS_VPORT_ATTR_NAME is specified and dp_ifindex is nonzero, the
logical behavior would be for the vport name lookup scope to be limited
to the specified datapath, but in fact the dp_ifindex value was ignored.
This commit causes the search scope to be honored.
Signed-off-by:
A few last minute patches for net/3.3 if there is time.
The following changes since commit dcd6c92267155e70a94b3927bce681ce74b80d1f:
Linux 3.3-rc1 (2012-01-19 15:04:48 -0800)
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jesse/openvswitch.git fixes
Be
On Wed, Mar 07, 2012 at 12:55:03PM -0800, Ben Pfaff wrote:
> On Wed, Mar 07, 2012 at 11:26:39PM +0800, Thomas Goirand wrote:
> > If you agree with my patch, I can do an NMU. Is that ok?
>
> Speaking as co-maintainer, yes, your patch looks correct, please NNU.
Likewise, please NMU.
_
Signed-off-by: Ethan Jackson
---
include/openflow/.gitignore |3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
diff --git a/include/openflow/.gitignore b/include/openflow/.gitignore
index 1a11f42..a06b563 100644
--- a/include/openflow/.gitignore
+++ b/include/openflow/.gitignore
@@ -1,2
> (It was global because I copied its usage from interface-reconfigure,
> where it is used much more extensively.)
This code has changed quite a bit too. I had assumed that a global
was once necessary and no longer is.
Ethan
___
dev mailing list
dev@op
On Wed, Mar 07, 2012 at 05:09:22PM -0800, Ethan Jackson wrote:
> There's no reason for this variable to be global, or to exist at
> all for that matter.
>
> Signed-off-by: Ethan Jackson
Looks good, thanks.
(It was global because I copied its usage from interface-reconfigure,
where it is used mu
Thanks.
I apologize that review is taking so long. I did not expect to take
this long.
On Wed, Feb 29, 2012 at 08:49:42PM +0100, ravi.ke...@telekom.com wrote:
> Hi Ben,
>
> It's based on
>
> commit dd1ba5b3f4425c8eba008d1a93b044da63466812
> Author: Justin Pettit
> Date: Fri Feb 24 10:32:36
The recently added SchemaHelper class significantly simplifies IDL
instantiation in Python. This commit converts all users of the old
method to the new method, and removes support for the old method.
Signed-off-by: Ethan Jackson
---
debian/ovs-monitor-ipsec | 52 ++--
With this patch, users can query a running ovs-monitor-ipsec
daemon's version, or ask it to exit using ovs-appctl.
Signed-off-by: Ethan Jackson
---
debian/ovs-monitor-ipsec | 25 -
1 file changed, 24 insertions(+), 1 deletion(-)
diff --git a/debian/ovs-monitor-ipsec b/
Typically Open vSwitch communicates with running processes using
unixctl. This patch converts ovs-xapi-sync to the strategy for
consistency.
Signed-off-by: Ethan Jackson
---
xenserver/etc_init.d_openvswitch |4 ++--
.../usr_share_openvswitch_scripts_ovs-xapi-sync| 19
With this patch, users can query a running ovs-xapi-sync's version
or ask it to exit using ovs-appctl.
Signed-off-by: Ethan Jackson
---
.../usr_share_openvswitch_scripts_ovs-xapi-sync| 25 +++-
1 file changed, 24 insertions(+), 1 deletion(-)
diff --git a/xenserver/usr_shar
Communicating with xapi from Python can be quite expensive so it
makes sense to cache this data if convenient.
Signed-off-by: Ethan Jackson
---
.../usr_share_openvswitch_scripts_ovs-xapi-sync| 27 ++--
1 file changed, 13 insertions(+), 14 deletions(-)
diff --git a/xenserve
There's no reason for this variable to be global, or to exist at
all for that matter.
Signed-off-by: Ethan Jackson
---
.../usr_share_openvswitch_scripts_ovs-xapi-sync|9 ++---
1 file changed, 2 insertions(+), 7 deletions(-)
diff --git a/xenserver/usr_share_openvswitch_scripts_ovs-xa
Thanks, I pushed this to master.
On Wed, Mar 07, 2012 at 04:56:22PM -0800, Ethan Jackson wrote:
> Thanks. It's good to get all of these shaken out
>
> Looks good,
> Ethan
>
> On Wed, Mar 7, 2012 at 16:38, Ben Pfaff wrote:
> > Another race condition in this test, *sigh*.
> >
> > Signed-off-by:
Thanks. It's good to get all of these shaken out
Looks good,
Ethan
On Wed, Mar 7, 2012 at 16:38, Ben Pfaff wrote:
> Another race condition in this test, *sigh*.
>
> Signed-off-by: Ben Pfaff
> ---
> tests/ofproto.at | 1 +
> 1 files changed, 1 insertions(+), 0 deletions(-)
>
> diff --git a/
Thanks, I've merged this.
On Wed, Mar 7, 2012 at 16:37, Ben Pfaff wrote:
> On Wed, Mar 07, 2012 at 04:16:21PM -0800, Ethan Jackson wrote:
>> The C code displays the build number as the empty string when 0,
>> and as +build otherwise. This commit updates version.py to be
>> consistent and tests t
Another race condition in this test, *sigh*.
Signed-off-by: Ben Pfaff
---
tests/ofproto.at |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/tests/ofproto.at b/tests/ofproto.at
index f912f31..fdc7215 100644
--- a/tests/ofproto.at
+++ b/tests/ofproto.at
@@ -434,6 +434,7 @@
On Wed, Mar 07, 2012 at 04:16:21PM -0800, Ethan Jackson wrote:
> The C code displays the build number as the empty string when 0,
> and as +build otherwise. This commit updates version.py to be
> consistent and tests that it is in the unit tests.
>
> Signed-off-by: Ethan Jackson
Looks good to m
Thanks, I pushed the revised series.
On Wed, Mar 07, 2012 at 04:15:18PM -0800, Ethan Jackson wrote:
> Looks good, thanks.
>
> Ethan
>
> On Wed, Mar 7, 2012 at 15:46, Ben Pfaff wrote:
> > The correct function to call here is "remove_file_to_unlink". That is,
> > since the file has already been
The C code displays the build number as the empty string when 0,
and as +build otherwise. This commit updates version.py to be
consistent and tests that it is in the unit tests.
Signed-off-by: Ethan Jackson
---
python/automake.mk|5 -
python/ovs/unixctl.py |6 +++---
tests/atloc
Looks good, thanks.
Ethan
On Wed, Mar 7, 2012 at 15:46, Ben Pfaff wrote:
> The correct function to call here is "remove_file_to_unlink". That is,
> since the file has already been unlinked there is no need to keep it on
> the list of files to unlink.
>
> However, "unlink_file_now" simplifies th
Looks good, thanks.
Ethan
On Wed, Mar 7, 2012 at 15:46, Ben Pfaff wrote:
> This bug is not exposed in the current tree, because no existing caller
> passes None as bind_path.
>
> Signed-off-by: Ben Pfaff
> ---
> python/ovs/socket_util.py | 10 +-
> 1 files changed, 5 insertions(+), 5
Hi,
since both the kernel GRE module and the OVS datapath vport-gre port
type register with the kernel a handler for the GRE protocol
(inet_add_protocol), does it mean that those two GRE handlers
are mutually exclusive?
/Chris
___
dev mailing list
dev
Our Ref: RTB /SNT/STB
To. Beneficiary.
Sir,
I am Special Agent David Jackson and am here in Nigeria as an FBI delegate that
has been delegated to investigate this fraudsters who are in the business of
swindling Foreigners that came for transaction in Nigeria . Please be informed
that during m
Thank you, I pushed this to master.
On Wed, Mar 07, 2012 at 03:33:32PM -0800, Ethan Jackson wrote:
> Looks good to me.
>
> Ethan
>
> On Wed, Mar 7, 2012 at 15:27, Ben Pfaff wrote:
> > The "speed_hi" member was only introduced in 2.6.27, so builds against
> > older kernel headers failed.
> >
> >
There's no reason for a Unix domain client socket to bind a name. I don't
know why we've always done that. Stevens's "Unix Network Programming"
Unix domain socket client example doesn't do a bind.
Removes the 'unlink_path' parameter from new_fd_stream() since it is now
always passed as NULL.
Si
The error handling path here failed to clean up bound sockets, by removing
them. This fixes the problem.
It was easy to observe this bug by running "ovs-vsctl" without
"ovsdb-server" running.
Bug #9811.
Bug #9769.
Reported-by: Michael
Signed-off-by: Ben Pfaff
---
lib/socket-util.c |4 ++--
The correct function to call here is "remove_file_to_unlink". That is,
since the file has already been unlinked there is no need to keep it on
the list of files to unlink.
However, "unlink_file_now" simplifies the code, so we might as well use
that.
Signed-off-by: Ben Pfaff
---
python/ovs/sock
This bug is not exposed in the current tree, because no existing caller
passes None as bind_path.
Signed-off-by: Ben Pfaff
---
python/ovs/socket_util.py | 10 +-
1 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/python/ovs/socket_util.py b/python/ovs/socket_util.py
index 8
This was previously posted as a 2-patch series here:
http://openvswitch.org/pipermail/dev/2012-February/015199.html
http://openvswitch.org/pipermail/dev/2012-February/015200.html
In the meantime, some new unit tests have uncovered bugs in the
Python implementation that must be fixed before these p
Thanks for looking at these patches.
When I rebased these onto current master, I found that there were new
test failures due to Python tests added on master in the meantime.
Avoiding these (which are genuine bugs) requires a couple of
preparatory commits. I'll re-send this as a 4-patch series soo
Looks good to me.
Ethan
On Wed, Mar 7, 2012 at 15:27, Ben Pfaff wrote:
> The "speed_hi" member was only introduced in 2.6.27, so builds against
> older kernel headers failed.
>
> speed_hi is fully backward compatible with older kernels, because older
> kernels always set it to 0, so we could eas
The "speed_hi" member was only introduced in 2.6.27, so builds against
older kernel headers failed.
speed_hi is fully backward compatible with older kernels, because older
kernels always set it to 0, so we could easily introduce a compatibility
layer here, but in fact I don't know of any OVS users
On Wed, Mar 07, 2012 at 02:51:12PM -0800, Ethan Jackson wrote:
> Without this patch, when a slave's carrier goes down, the LACP
> module (as evidenced by ovs-appctl lacp/show) would consider the
> slave current until it hadn't received LACP PDUs for the requisite
> amount of time. It should instea
Looks good, thanks for looking into this Ben.
Ethan
On Wed, Mar 7, 2012 at 14:49, Ben Pfaff wrote:
> The "ofproto - asynchronous message control" test had a race in which
> the "send: OFPT_BARRIER_REQUEST" message could get printed in different
> places because there was nothing to ensure that m
Without this patch, when a slave's carrier goes down, the LACP
module (as evidenced by ovs-appctl lacp/show) would consider the
slave current until it hadn't received LACP PDUs for the requisite
amount of time. It should instead, immediately mark the slave
expired. This shouldn't actually affect
The "ofproto - asynchronous message control" test had a race in which
the "send: OFPT_BARRIER_REQUEST" message could get printed in different
places because there was nothing to ensure that messages from the switch
were printed before messages sent to the switch, even though the actual
ordering of
On Wed, Mar 7, 2012 at 2:35 PM, Ben Pfaff wrote:
> On Wed, Mar 07, 2012 at 02:26:46PM -0800, Jesse Gross wrote:
>> On Wed, Mar 7, 2012 at 10:46 AM, Ben Pfaff wrote:
>> > On Tue, Mar 06, 2012 at 03:40:51PM -0800, Jesse Gross wrote:
>> >> When modifying IP addresses or ports on a UDP packet we don'
> (By the way, you can actually give the struct and the function the
> same name.)
Hmm, my compiler was complaining about it for some reason. It must
have been something else. Anyways, I think it's confusing to give
them the same name at the very least.
Thanks for the review, I'll merge this s
On Wed, Mar 07, 2012 at 02:29:41PM -0800, Ethan Jackson wrote:
> > Why did you change the name of "struct ofproto_trace"? I thought it
> > was an OK name.
>
> I thought it was a better name for the function than the struct. Also
> we have a precedent for using ctx in the names of these sorts of
On Wed, Mar 07, 2012 at 02:26:46PM -0800, Jesse Gross wrote:
> On Wed, Mar 7, 2012 at 10:46 AM, Ben Pfaff wrote:
> > On Tue, Mar 06, 2012 at 03:40:51PM -0800, Jesse Gross wrote:
> >> When modifying IP addresses or ports on a UDP packet we don't
> >> correctly follow the rules for unchecksummed pac
> Why did you change the name of "struct ofproto_trace"? I thought it
> was an OK name.
I thought it was a better name for the function than the struct. Also
we have a precedent for using ctx in the names of these sorts of
things. That was my reasoning at least, I really don't care what it's
na
On Wed, Mar 7, 2012 at 10:46 AM, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 03:40:51PM -0800, Jesse Gross wrote:
>> When modifying IP addresses or ports on a UDP packet we don't
>> correctly follow the rules for unchecksummed packets. This meant
>> that packets without a checksum can be given a i
On Tue, Mar 06, 2012 at 05:33:47PM -0800, Ethan Jackson wrote:
> It can be very difficult to debug xlate_actions() failures due to
> excessive resubmit recursion. In an attempt to make it easier,
> this patch adds a (severely rate-limited) full ofproto/trace to the
> logs.
>
> Suggested-by: Alan
Thanks, I removed the trailing whitespace and pushed the series.
On Wed, Mar 07, 2012 at 01:30:12PM -0800, Ethan Jackson wrote:
> There's some trailing whitespace in ovsdb-execution.at
>
> Otherwise looks good, thank you.
>
> Ethan
>
> On Mon, Feb 20, 2012 at 15:34, Ben Pfaff wrote:
> > This r
On Tue, Mar 06, 2012 at 11:31:32AM -0800, Jesse Gross wrote:
> On Mon, Mar 5, 2012 at 2:03 PM, Ben Pfaff wrote:
> > When OVS_VPORT_ATTR_NAME is specified and dp_ifindex is nonzero, the
> > logical behavior would be for the vport name lookup scope to be limited
> > to the specified datapath, but in
Thank you, I pushed this.
On Wed, Mar 07, 2012 at 01:58:09PM -0800, Ethan Jackson wrote:
> Looks good, thanks.
>
> Ethan
>
> On Wed, Mar 7, 2012 at 13:53, Ben Pfaff wrote:
> > Bug #10062.
> > Reported-by: James Schmidt
> > Signed-off-by: Ben Pfaff
> > ---
> > utilities/ovs-ofctl.c | 5 +++
On Sat, Feb 25, 2012 at 5:34 PM, Simon Horman wrote:
> On Sat, Feb 25, 2012 at 12:11:20AM -0800, Ben Pfaff wrote:
>> On Tue, Feb 21, 2012 at 03:53:48PM +0900, Simon Horman wrote:
>> > On Wed, Feb 15, 2012 at 04:37:43PM -0800, Ben Pfaff wrote:
>> > > This comprises most of the previously posted "of
On Fri, Mar 02, 2012 at 02:23:56PM -0800, Ethan Jackson wrote:
> This patch shouldn't actually affect the behavior of LACP bonds
> because the bond module won't choose to send traffic out a slave
> whose carrier is down.
>
> Signed-off-by: Ethan Jackson
I see where port->carrier_seq gets initial
Looks good, thanks.
Ethan
On Mon, Feb 27, 2012 at 11:17, Ben Pfaff wrote:
> There's no reason for a Unix domain client socket to bind a name. I don't
> know why we've always done that. Stevens's "Unix Network Programming"
> Unix domain socket client example doesn't do a bind.
>
> Removes the '
Looks good, thanks.
Ethan
On Wed, Mar 7, 2012 at 13:53, Ben Pfaff wrote:
> Bug #10062.
> Reported-by: James Schmidt
> Signed-off-by: Ben Pfaff
> ---
> utilities/ovs-ofctl.c | 5 +++--
> 1 files changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/utilities/ovs-ofctl.c b/utilities/ovs-
Bug #10062.
Reported-by: James Schmidt
Signed-off-by: Ben Pfaff
---
utilities/ovs-ofctl.c |5 +++--
1 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/utilities/ovs-ofctl.c b/utilities/ovs-ofctl.c
index fd0829e..607ec25 100644
--- a/utilities/ovs-ofctl.c
+++ b/utilities/ovs-ofctl
Looks good, thanks.
Ethan
On Mon, Feb 27, 2012 at 11:17, Ben Pfaff wrote:
> The error handling path here failed to clean up bound sockets, by removing
> them. This fixes the problem.
>
> It was easy to observe this bug by running "ovs-vsctl" without
> "ovsdb-server" running.
>
> Bug #9769.
> Re
There's some trailing whitespace in ovsdb-execution.at
Otherwise looks good, thank you.
Ethan
On Mon, Feb 20, 2012 at 15:34, Ben Pfaff wrote:
> This reduces tests/testsuite by about 70 kB, by collapsing a number of
> macro expansions into just one copy in a shell function.
>
> Signed-off-by: Be
Looks good, thanks.
Ethan
On Mon, Feb 20, 2012 at 15:33, Ben Pfaff wrote:
> This reduces tests/testsuite by about 35 kB, by collapsing a number of
> macro expansions into just one copy in a shell function.
>
> Signed-off-by: Ben Pfaff
> ---
> tests/testsuite.at | 38 -
Looks good.
Ethan
On Mon, Feb 20, 2012 at 15:33, Ben Pfaff wrote:
> This reduces tests/testsuite by about 50 kB, by collapsing a number of
> macro expansions into just one copy in a shell function.
>
> Signed-off-by: Ben Pfaff
> ---
> tests/interface-reconfigure.at | 186
> ++
On Wed, Mar 07, 2012 at 11:26:39PM +0800, Thomas Goirand wrote:
> If you agree with my patch, I can do an NMU. Is that ok?
Speaking as co-maintainer, yes, your patch looks correct, please NNU.
___
dev mailing list
dev@openvswitch.org
http://openvswitch
Thanks, I've removed the redundant double [] and merged this to master.
Ethan
On Wed, Mar 7, 2012 at 12:45, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 05:53:08PM -0800, Ethan Jackson wrote:
>> We've seen some unixctl tests hang indefinitely which makes them
>> difficult to debug. ovs-appctl and
On Tue, Mar 06, 2012 at 05:53:08PM -0800, Ethan Jackson wrote:
> We've seen some unixctl tests hang indefinitely which makes them
> difficult to debug. ovs-appctl and appctl.py calls to timeout
> instead.
>
> Signed-off-by: Ethan Jackson
I don't see any reason to double the [] in the definition
On Tue, Mar 06, 2012 at 05:53:07PM -0800, Ethan Jackson wrote:
> The transient failure in the unixctl unit tests likely still
> exists, but we've added additional instrumentation to our build
> tools to help us debug it.
>
> Signed-off-by: Ethan Jackson
Thanks, looks good.
__
On Tue, Mar 06, 2012 at 02:37:29PM -0800, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
It looks good to me, but as in a previous patch i would change:
> +if (netdev_dev->cache_valid & VALID_FEATURES) {
> +goto out;
> +}
> +
> +netdev_linux_read_features(netdev_dev);
On Wed, Mar 07, 2012 at 10:58:34AM -0800, Pravin Shelar wrote:
> On Wed, Mar 7, 2012 at 10:39 AM, Ben Pfaff wrote:
> > On Tue, Mar 06, 2012 at 02:37:29PM -0800, Pravin B Shelar wrote:
> >> Signed-off-by: Pravin B Shelar
> >
> > Did you check whether we can cache this stuff? That is, if any of th
Remove a couple stale entries:
brc_sysfs_* was renamed to dp_sysfs_* a few years ago
Add some new entries to ignore:
# Untracked file:
# (use "git add ..." to include in what will be committed)
# datapath/linux/exthdrs_core.c
# datapath/linux/genl_exec.c
# datapath/linux/ne
On Wed, Mar 7, 2012 at 10:39 AM, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 02:37:29PM -0800, Pravin B Shelar wrote:
>> Signed-off-by: Pravin B Shelar
>
> Did you check whether we can cache this stuff? That is, if any of the
> relevant data changes, do we get a Netlink notification for the
> dev
On Wed, Mar 07, 2012 at 10:42:47AM -0800, Pravin Shelar wrote:
> On Wed, Mar 7, 2012 at 9:57 AM, Ben Pfaff wrote:
> > On Tue, Mar 06, 2012 at 02:37:02PM -0800, Pravin B Shelar wrote:
> >> Rather than using ETHTOOL_GDRVINFO to check netdev-status, use
> >> ovs-get-stats since it is more consistent.
On Tue, Mar 06, 2012 at 03:40:51PM -0800, Jesse Gross wrote:
> When modifying IP addresses or ports on a UDP packet we don't
> correctly follow the rules for unchecksummed packets. This meant
> that packets without a checksum can be given a incorrect new checksum
> and packets with a checksum can
On Wed, Mar 7, 2012 at 10:17 AM, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 02:37:08PM -0800, Pravin B Shelar wrote:
>> netdev linux devices uses mtu ioctl to get and set MTU for a device.
>> By caching error code from ioctl we can reduce number of ioctl calls
>> for device which is unregistered f
On Wed, Mar 7, 2012 at 10:32 AM, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 02:37:16PM -0800, Pravin B Shelar wrote:
>> Signed-off-by: Pravin B Shelar
>
> This code in netdev_linux_get_etheraddr() looks awkward to me:
> if (netdev_dev->cache_valid & VALID_ETHERADDR) {
> goto out;
> }
On Wed, Mar 7, 2012 at 9:57 AM, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 02:37:02PM -0800, Pravin B Shelar wrote:
>> Rather than using ETHTOOL_GDRVINFO to check netdev-status, use
>> ovs-get-stats since it is more consistent. Covergage counter is
>> added for status check.
>>
>> Signed-off-by: P
On Tue, Mar 06, 2012 at 02:37:29PM -0800, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
Did you check whether we can cache this stuff? That is, if any of the
relevant data changes, do we get a Netlink notification for the
device? I have my doubts about that, because none of this data
On Tue, Mar 06, 2012 at 02:37:22PM -0800, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
All of the references to "policy" in the change log and code should
refer to "policing".
Otherwise, this looks good. Thank you.
___
dev mailing list
dev@
On Tue, Mar 06, 2012 at 02:37:16PM -0800, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
This code in netdev_linux_get_etheraddr() looks awkward to me:
if (netdev_dev->cache_valid & VALID_ETHERADDR) {
goto out;
}
get_etheraddr(netdev_dev);
out:
How about, instead:
This renames the datapath modules:
openvswitch_mod -> openvswitch
brcompat_mod -> brcompat
The first makes the module name consistent with upstream, and the latter
is just for internal consistency. This makes tools, and documentation
refer to a common module name regardless if it's coming fr
This just makes it easier to subsequently rename the brcompat_mod module
to brcompat without introducing circular dependencies.
Signed-off-by: Chris Wright
---
datapath/brcompat.c | 578 -
datapath/brcompat_main.c | 578 +++
This simply renames the datapath modules, omitting the trailing _mod.
Patch 2/2 has details on why.
Chris Wright (2):
datapath: rename brcompat.c to brcompat_main.c
datapath: omit _mod from module names
INSTALL.Linux | 14 +-
INSTALL.XenServer
On Tue, Mar 06, 2012 at 02:37:08PM -0800, Pravin B Shelar wrote:
> netdev linux devices uses mtu ioctl to get and set MTU for a device.
> By caching error code from ioctl we can reduce number of ioctl calls
> for device which is unregistered from system.
> netdev notification is used to update mtu
On Tue, Mar 06, 2012 at 02:37:02PM -0800, Pravin B Shelar wrote:
> Rather than using ETHTOOL_GDRVINFO to check netdev-status, use
> ovs-get-stats since it is more consistent. Covergage counter is
> added for status check.
>
> Signed-off-by: Pravin B Shelar
In netdev_linux_get_status(), I don't u
On Wed, Mar 07, 2012 at 09:56:13AM -0800, Ben Pfaff wrote:
> On Tue, Mar 06, 2012 at 02:37:08PM -0800, Pravin B Shelar wrote:
> > netdev linux devices uses mtu ioctl to get and set MTU for a device.
> > By caching error code from ioctl we can reduce number of ioctl calls
> > for device which is unr
On Tue, Mar 06, 2012 at 02:37:08PM -0800, Pravin B Shelar wrote:
> netdev linux devices uses mtu ioctl to get and set MTU for a device.
> By caching error code from ioctl we can reduce number of ioctl calls
> for device which is unregistered from system.
> netdev notification is used to update mtu
If you agree with my patch, I can do an NMU. Is that ok?
Thomas
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Wed, Mar 07, 2012 at 10:49:33PM +0800, Thomas Goirand wrote:
> Hi,
>
> I'm sorry, I think I didn't express myself correctly. Please read this:
> http://lintian.debian.org/tags/patch-modifying-debian-files.html
>
> As you can see, lintian did catch issues in your openvswitch package! :)
>
> So
Hi,
I'm sorry, I think I didn't express myself correctly. Please read this:
http://lintian.debian.org/tags/patch-modifying-debian-files.html
As you can see, lintian did catch issues in your openvswitch package! :)
So, if you have changes to make in let's say debian/control, then edit
the file, d
On Wed, Mar 07, 2012 at 04:48:55PM +0800, Thomas Goirand wrote:
> Hi,
>
> I had a look to the current packaging of openvswitch, in order to fix
> this bug (eg: #659685) With all due respect... it's a mess.
>
> The only reason why you are using Quilt is to patch files that are in
> your openvswitc
Hi,
I had a look to the current packaging of openvswitch, in order to fix
this bug (eg: #659685) With all due respect... it's a mess.
The only reason why you are using Quilt is to patch files that are in
your openvswitch_.debian.tar.gz. Please don't abuse quilt like
this! There's no reason to pat
--
Good day,My name is Mrs Joan Morris I have been arranging loans for many
years.3% is my interest rate and lower cost than any one else in town.Contact
me today E-mails: joanloaninvestme...@yahoo.com Loan Application
Form*Full Name*Gender...
88 matches
Mail list logo
