Signed-off-by: Ed Maste
---
lib/command-line.c | 9 ++---
lib/daemon.c | 5 ++---
lib/worker.c | 3 +--
3 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/lib/command-line.c b/lib/command-line.c
index 76a4e74..eb9f6b1 100644
--- a/lib/command-line.c
+++ b/lib/command
FreeBSD libc's setproctitle provides the same interface as proctitle_set so
use it directly.
Signed-off-by: Ed Maste
---
lib/command-line.c | 3 +++
lib/command-line.h | 4
2 files changed, 7 insertions(+)
diff --git a/lib/command-line.c b/lib/command-line.c
index eb9f6b1..a018424 1
On FreeBSD I see ECONNRESET being returned from check_connection_completion
in ssl_connect. This happens before entering the SSL states, which return
EPROTO on failure.
Signed-off-by: Ed Maste
---
tests/test-vconn.c | 9 -
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a
On 22 October 2012 14:23, Ben Pfaff wrote:
> I know that I suggested uuidgen in conversation, but now I'm thinking
> that it isn't the best choice because it's not entirely portable
> (e.g. I have no idea whether FreeBSD generally has uuidgen).
It should be available on any supported FreeBSD rele
On 6 December 2012 12:44, Justin Pettit wrote:
> On Dec 6, 2012, at 8:36 AM, Ben Pfaff wrote:
>
>> Simon, does this solve your problem?
>
> I'll hold off on pushing this until we get confirmation from Simon, since you
> and I weren't completely satisfied with the solution.
Also I see that netde
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
index f8b1188..256955d 100644
--- a/lib/netdev-bsd.c
+++ b/lib/netdev-bsd.c
@@ -17,7 +17,6 @@
#include
#include
-#include
#include
#include
#include
ee above for list of violations of the rule that"; \
> + echo "every C source file must #include ."; \
> + exit 1; \
> + fi
> +.PHONY: config-h-check
> +
> # Check that "struct vlog_ratelimit" is always declared "static&
On 6 December 2012 12:52, Justin Pettit wrote:
> Would you mind testing the FreeBSD port from master? My single datapath
> changes were quite invasive, and FreeBSD is probably the biggest user of the
> userspace datapath. It would be great to know if any other problems were
> introduced.
Wil
The config.h check added in commit 9e4ba00 may bring up the pager during
build (depending on git config).
Signed-off-by: Ed Maste
---
Makefile.am | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Makefile.am b/Makefile.am
index b00b82c..fe6aa45 100644
--- a/Makefile.am
+++ b
This is a trivial port to netdev-bsd of Justin Pettit's netdev-linux fix.
Signed-off-by: Ed Maste
---
This fixes the userspace datapath on FreeBSD.
Justin's change:
Commit acf608 (ofproto-dpif: Use a single underlying datapath across
multiple bridges.) broke connectivity to
When built with Clang, vswitchd segfaulted in ovsrec_open_vswitch_init,
from calling memset() on a const struct.
Signed-off-by: Ed Maste
---
This could be addressed instead by just dropping the const in the
null_cfg definition, if preferred.
vswitchd/bridge.c | 9 +
1 file changed, 5
When built with Clang, vswitchd segfaulted in ovsrec_open_vswitch_init,
from calling memset() on a const struct.
Signed-off-by: Ed Maste
---
vswitchd/bridge.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/vswitchd/bridge.c b/vswitchd/bridge.c
index b1d2feb..cb94fa0
On 14 December 2012 10:47, Ben Pfaff wrote:
> EXIT is more readable, but I think it was invented by POSIX and is
> therefore less portable than 0.
As a point of reference FreeBSD's /bin/sh is fairly basic, and the man
page explicitly refers to EXIT for this:
| In addition, the pseudo-signal EXIT
On 28 June 2012 23:15, Ben Pfaff wrote:
> The route-table implementation is definitely not portable, but I don't
> think there's anything in the interface that can't be implemented in
> FreeBSD. One proper solution, then, would be to add an implementation
> that works in FreeBSD. I'd certainly t
This is a trivial implementation of the route-table functionality for
FreeBSD, as needed by ofproto/ofproto-dpif-sflow.c. It has not yet
been extensively tested.
Signed-off-by: Ed Maste
---
Note that my primary goal here was to get a basic implementation to allow
it to build, so that I can move
Signed-off-by: Ed Maste
---
lib/vlandev.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/vlandev.c b/lib/vlandev.c
index 77e8e93..ffb8e73 100644
--- a/lib/vlandev.c
+++ b/lib/vlandev.c
@@ -23,13 +23,13 @@
#include
#include "hash.h"
-#include "r
On 6 July 2012 08:59, Giuseppe Lettieri wrote:
> Hi,
>
> the attached patch implements new netdev classes for "system" and "tap"
> devices in FreeBSD, by using the libpcap library. With this patch (and
> some firewall rules) the user-level datapath of Open vSwitch can be made
> to work on FreeBSD.
>> It isn't clear to me though why the Linux userspace mode does not behave
>> the same way though. I had a (very) brief look at the way PF_PACKET is
>> handled in Linux and it seemed like the packet would still be passed up
>> the stack -- there must be something else that I'm missing.
>
> Linux
Signed-off-by: Ed Maste
---
lib/route-table-bsd.c |2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/route-table-bsd.c b/lib/route-table-bsd.c
index c145091..1c29071 100644
--- a/lib/route-table-bsd.c
+++ b/lib/route-table-bsd.c
@@ -29,6 +29,8 @@
#include
#include
+#include
> In FreeBSD, do you intend to add the equivalent filter rules
> automatically somewhere? Or to document that it is necessary and that
> the administrator should do so by hand? If the former, then let's
> figure out a good mechanism to hook in. Perhaps we should add a pair
> of "netdev" function
> The PCAP_SNAPLEN of 1024 surprises me. Wouldn't you want at least
> 1500, so that you can capture full Ethernet frames? (Maybe 1514 or
> 1518? I don't know whether the snaplen includes the Ethernet and VLAN
> headers.)
Yes, this appears to be an oversight. 1518 is probably reasonable.
> In
> Do you want to add an INSTALL.FreeBSD or similar file at top level?
> Feel free to copy any applicable text from INSTALL.Linux. Or, if you
> think it's better, please feel free to edit INSTALL.Linux to include
> FreeBSD instructions also, and then we can rename it to just INSTALL.
The combinati
> This looks good to me, thanks. I'll assume that you'll fold this into
> your larger patch when it's ready.
Yes, I just wanted to offer this part up for comment first (and also
because git-format-patch shows it as a delete and add so the changes
themselves are not as obvious).
> BSD make is sup
On 25 July 2012 13:31, Ben Pfaff wrote:
> Ed Maste mentioned that BSD make didn't
> work with OVS. That was unintentional; I just hadn't tested it
> with BSD make in a while.
Ok, I've applied these changes and can build with NetBSD's make (this
should be the same
invalid errno. With the addition of NLS strerror was changed
to do so for all calls.
Prior to this change I had confusing results from the test suite like
"... is 22 (Invalid argument) but should be 0 (Invalid argument)".
Signed-off-by: Ed Maste
---
tests/test-vconn.c |3 ++-
1 file
I'm working through a handful of unit test failures in the FreeBSD port
and the first issue I've come across is set_dscp failing - as it turns
out FreeBSD requires an int value for the setsockopt() call, not a char.
Looking at the Linux ip(7) man page I see the statement "TOS is a byte,"
but all o
FreeBSD requires that setsockopt(..., IP_TOS, ...) be passed an int
value. Linux accepts either int or char types (and has since at least
kernel 2.6.12) so just use int type unconditionally.
Signed-off-by: Ed Maste
---
lib/socket-util.c |6 --
1 file changed, 4 insertions(+), 2
On FreeBSD I'm seeing an intermittent failure in a number of the
vconn unit tests - e.g. 186 tcp vconn - refuse connection. The test
code:
139 static void
140 test_refuse_connection(int argc OVS_UNUSED, char *argv[])
141 {
142 const char *type = argv[1];
143 int expected_error;
144
FreeBSD's wc outputs leading whitespace before the count, so use test(1)
for numeric equality instead of AC_CHECK string match.
Signed-off-by: Ed Maste
---
tests/ovsdb-server.at |8
tests/ovsdb-tool.at |9 +++--
2 files changed, 7 insertions(+), 10 deletions(-)
I currently have 8 failing unit tests on FreeBSD:
testsuite: 186 188 474 1045 1054 1055 1056 1057 failed
Here are some notes on each of them. I have fixes / workarounds for all
but the first two and will send patches shortly.
186, 188
This is the issue in my recent post 'FreeBSD vconn refuse-co
FreeBSD sed doesn't support \b, at least in normal mode. Since the
string we're matching comes at the end of the line anyway just use $
instead.
Signed-off-by: Ed Maste
---
tests/ofproto-dpif.at |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tests/ofproto
VSCTL_CHECK_FIND_EMPTY to explicitly match an empty bridge list.
Signed-off-by: Ed Maste
---
tests/ovs-vsctl.at | 21 -
1 file changed, 12 insertions(+), 9 deletions(-)
diff --git a/tests/ovs-vsctl.at b/tests/ovs-vsctl.at
index 71c46cd..13d5329 100644
--- a/tests/ovs-vsctl.at
+++ b
Signed-off-by: Ed Maste
---
xenserver/opt_xensource_libexec_interface-reconfigure |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/xenserver/opt_xensource_libexec_interface-reconfigure
b/xenserver/opt_xensource_libexec_interface-reconfigure
index 6c6de9f..72bd5e4 100755
On 30 July 2012 20:16, Ben Pfaff wrote:
> On Mon, Jul 30, 2012 at 07:23:48PM -0400, Ed Maste wrote:
>> The FreeBSD version of xargs does not run the utility argument on empty
>> input, while GNU xargs runs it at least once, even with empty input. As
>> a result on Fre
The FreeBSD version of xargs does not run the utility argument on empty
input, while GNU xargs runs it at least once, even with empty input. As
a result on FreeBSD VSCTL_CHECK_FIND returned no output for an empty
bridge list while on Linux it returned a single blank line.
Signed-off-by: Ed Maste
on FreeBSD and
ECONNRESET on Linux.
Signed-off-by: Ed Maste
---
tests/test-vconn.c | 65 ++--
1 file changed, 58 insertions(+), 7 deletions(-)
diff --git a/tests/test-vconn.c b/tests/test-vconn.c
index f37bc38..fa244ef 100644
--- a/tests/test
On 31 July 2012 09:50, Ed Maste wrote:
> On FreeBSD some of the test-vconn.c tests failed intermittently due
> to timing variations in a TCP localhost connection.
>
> - If we receive EAGAIN for certain functions, retry.
>
> - Allow EPIPE in addition to ECONNRESET in some case
FreeBSD doesn't have "date -r" and calls its sha1 hash tool tool "sha1."
Signed-off-by: Ed Maste
---
Defining shell functions within a case statement is a little odd but I'd
rather not have several functions each with the same case statement in
them. I could
On 31 July 2012 12:51, Ed Maste wrote:
> With this change (and after enabling SSL in configure) all but 3 skipped
> tests in "make check" pass on FreeBSD.
As an aside the three skipped tests are:
library.at unix domain socket long pathname - I don't think I can do
an
> I sent out a 2-patch series that works this way:
> http://openvswitch.org/pipermail/dev/2012-July/019825.html
> http://openvswitch.org/pipermail/dev/2012-July/019824.html
>
> Please review it.
I have reviewed and tested those changes. Thanks, that is much better.
-Ed
__
FreeBSD doesn't have "date -r" and calls its sha1 hash tool "sha1".
Signed-off-by: Ed Maste
---
With indentation.
utilities/ovs-pki.in | 36 ++--
1 file changed, 34 insertions(+), 2 deletions(-)
diff --git a/utilities/ovs-pki.in b/u
Also remove another sed \b that FreeBSD sed doesn't handle.
Signed-off-by: Ed Maste
---
tests/ofproto.at | 19 ---
1 file changed, 12 insertions(+), 7 deletions(-)
diff --git a/tests/ofproto.at b/tests/ofproto.at
index 2d28131..3823b70 100644
--- a/tests/ofproto.at
101 - 142 of 142 matches
Mail list logo
