On 3 April 2015 at 13:46, Ben Pfaff wrote:
>
>> > Also, the commit you name was in 2013, and we've had other
>> > contributions from FreeBSD contributors since then
>> > (namely Ed Maste ) and it seems like he
>> > would have noticed if it were totally
On 4 April 2015 at 12:56, Kevin Lo wrote:
>> Kevin, what failure mode did you observe?
>
> On FreeBSD, multicast is enabled on the interface by default, the value of
> ifr_flags could be negative if multicast is enabled.
Right, but what actually happens in this case? The sign extension bug
means
On 21 March 2014 14:03, Pravin wrote:
> Following patch adds DPDK netdev-class to userspace datapath. Now
> OVS can use DPDK port for IO by just configuring DPDK port and then
> adding dpdk type port to userspace datapath.
Intel recently released DPDK also for FreeBSD, so it may be worth
referrin
Signed-off-by: Ed Maste
---
I don't recall how I ended up with this naive anti-spam obfuscation in
AUTHORS. In any case there's no reason to have one inconsistent entry
here.
AUTHORS | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/AUTHORS b/AUTHORS
index f4e1ca
Commit f431bf7d78f3212d32bb3d122f783c5c796a1576 added new method
get_tunnel_config to netdev_class but did not update netdev-bsd.
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
index 9094d04..538f91a 100644
On 23 January 2013 15:51, Ethan Jackson wrote:
> diff --git a/FAQ b/FAQ
> index ab1c1cc..a466ca4 100644
> --- a/FAQ
> +++ b/FAQ
> @@ -172,17 +172,24 @@ A: The kernel module in upstream Linux 3.3 and later
> does not include
> vSwitch distribution instead of the upstream Linux kernel
>
On 23 January 2013 18:40, Ben Pfaff wrote:
> I think something got mixed up. Here's the change that I expected to
> see here:
>
> --
> diff --git a/FAQ b/FAQ
> index ab1c1cc..bd4d37e 100644
> --- a/FAQ
> +++ b/FAQ
> @@ -158,28 +1
On 27 January 2013 19:58, Ethan Jackson wrote:
> This version of the series implements enough of the flow based
> tunneling that we can begin testing. There are still some things
> which need to be done (described in ofproto/tunnel.c) which we will
> need to complete before making a release.
As
On 28 January 2013 12:44, Ben Pfaff wrote:
> On Mon, Jan 28, 2013 at 10:43:23AM -0500, Ed Maste wrote:
>> On 27 January 2013 19:58, Ethan Jackson wrote:
>> > This version of the series implements enough of the flow based
>> > tunneling that we can begin testing.
On 27 January 2013 19:58, Ethan Jackson wrote:
> diff --git a/lib/netdev-linux.c b/lib/netdev-linux.c
> index f0f1dc2..a06aef3 100644
> --- a/lib/netdev-linux.c
> +++ b/lib/netdev-linux.c
> @@ -2430,6 +2430,7 @@ netdev_linux_change_seq(const struct netdev *netdev)
>
On 29 January 2013 00:12, Ben Pfaff wrote:
> It looks like Ethan was able to fix the problems--he posted a new
> version of patch 11 that says it fixes FreeBSD test failures.
I didn't receive the new version of patch 11 (maybe Gmail decided it
was a duplicate), but I grabbed it from the archive a
On 31 January 2013 05:49, YAMAMOTO Takashi wrote:
> From: YAMAMOTO Takashi
>
> -if (!IFM_TYPE(ifmr.ifm_current) & IFM_ETHER) {
> +if (!(IFM_TYPE(ifmr.ifm_current) & IFM_ETHER)) {
> VLOG_DBG_RL(&rl, "%s: doesn't appear to be ethernet",
> netdev_get_name(netdev
On 31 January 2013 05:49, YAMAMOTO Takashi wrote:
> From: YAMAMOTO Takashi
>
> on some platforms, including NetBSD-6, openssl is available but
> its pkg-config files (*.pc) are not.
FreeBSD as well (this is a common issue in the FreeBSD ports tree). Thanks.
_
On 27 February 2013 20:03, Ben Pfaff wrote:
> Its alleged convenience just doesn't outweigh the syntactical ugliness, and
> so it didn't have any users.
Not entirely true: there's a consumer in lib/netdev-bsd.c. I don't
have any time right now to look at it, but will try if nobody gets to
it fir
On 8 March 2013 16:06, Ed Maste wrote:
> On 27 February 2013 20:03, Ben Pfaff wrote:
>> Its alleged convenience just doesn't outweigh the syntactical ugliness, and
>> so it didn't have any users.
>
> Not entirely true: there's a consumer in lib/netdev-bsd.c.
On 15 March 2013 17:30, Ben Pfaff wrote:
> I promised some time ago to take a look at fixing the behavior of the
> Makefiles with non-GNU make, but it doesn't realistically seem that I will.
>
Thanks. Even if we addressed the current issues I wouldn't be surprised to
find new failures appearing
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
> on some platforms, including NetBSD-6, openssl is available but
> its pkg-config files (*.pc) are not. according to Ed Maste,
> this is the case for FreeBSD as well.
That's correct. This should be a nice convenience for anyo
> Ed, do you want to look over any of these changes? They affect code
> for which you are a primary author, so if you would prefer to see
> anything done a different way, etc., for maintainability purposes then
> it would be valuable to get that in review.
Yep - I'm looking at them now. I've rev
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
> this will be used for get_stats and set_etheraddr
>
> Signed-off-by: YAMAMOTO Takashi
> ---
> lib/netdev-bsd.c | 15 ++-
> 1 file changed, 14 insertions(+), 1 deletion(-)
>
> diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
> index ba
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
>
> Signed-off-by: YAMAMOTO Takashi
Acked-by: Ed Maste
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
>
> Signed-off-by: YAMAMOTO Takashi
Acked-by: Ed Maste
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Yamamoto-san,
Thanks for this work. I'm happy that it will result in BSD systems as
a group getting more OVS test coverage.
The patch set builds and passes unit tests for me on FreeBSD, with the
exception of the missing #if defined() guard described in a previous
email.
I have only one overall
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
> where interface renaming is not supported (NetBSD), remember both of
> our netdev name and the correspoinding kernel name separately.
> the latter is necessary to talk with kernel using interface names.
> eg. ifioctls, bpf
Acked-
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
>
> Signed-off-by: YAMAMOTO Takashi
> ---
> lib/netdev-bsd.c | 47 +++
> 1 file changed, 43 insertions(+), 4 deletions(-)
>
...
> +#elif defined(__NetBSD__)
> +struct netdev_dev_bsd *netdev_dev =
> +
On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
>
> Signed-off-by: YAMAMOTO Takashi
> ---
> lib/netdev-bsd.c | 60
> +---
> 1 file changed, 53 insertions(+), 7 deletions(-)
On 24 April 2013 11:56, Ben Pfaff wrote:
> On Wed, Apr 24, 2013 at 11:28:45AM -0400, Ed Maste wrote:
>> On 22 April 2013 09:20, YAMAMOTO Takashi wrote:
>> > +/*
>> > + * note: UINT64_MAX means unsupported
>> > + */
>>
>> Is this checked
> +static int
> +netdev_bsd_get_next_hop(const struct in_addr *host, struct in_addr *next_hop,
> +char **netdev_name)
> +{
> +#if defined(__NetBSD__)
Acked-by: Ed Maste
This is close to being usable on FreeBSD as well, with the exception
that we don't
FreeBSD does not have EAI_ADDRFAMILY or EAI_NODATA and thus failed to build
after commit 3cbb5dc7e89df2b40bb6f715873cf2b6b25a7054 "socket-util: Use
getaddrinfo() instead of gethostbyname() for thread safety."
Signed-off-by: Ed Maste
---
lib/socket-util.c | 4
1 file changed, 4
As documented in netdev-provider.h for the get_stats method.
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c | 26 +-
1 file changed, 13 insertions(+), 13 deletions(-)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
index e3461b8..7ab9d3e 100644
--- a/lib/netdev-bsd.c
#x27;ve also tested this on Ubuntu and it builds fine there.
>
> Signed-off-by: Kyle Mestery
Acked-by: Ed Maste
Builds fine for me on FreeBSD with this change.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 9 May 2013 18:35, Ben Pfaff wrote:
> This gets rid of the only per-instance data in "struct netdev", which
> will make it possible to merge "struct netdev_dev" into "struct netdev" in
> a later commit.
The four patches generally look good to me, and when I apply the whole
set they pass the uni
On 10 May 2013 11:55, Ben Pfaff wrote:
> Thanks, I found the problem with valgrind, and here is an incremental
> fix:
>
> diff --git a/lib/netdev.c b/lib/netdev.c
> index 9590f83..415cdb4 100644
> --- a/lib/netdev.c
> +++ b/lib/netdev.c
> @@ -322,8 +322,10 @@ void
> netdev_close(struct netdev *ne
On 10 May 2013 13:12, Ed Maste wrote:
> Ahh, of course. This fixes the segfaulting tests for me, but the unit
> tests now get stuck in "tunnel - input."
Oh, and of course the reason ovs-vsctl is stuck is that ovs-vswitchd segfaulted.
#0 0x00080164b560 in strncpy () from /
This is an incremental to patch v3 of
netdev: Factor restoring flags into new "struct netdev_saved_flags",
implementing the equivalent change for netdev-bsd.
Signed-off-by: Ed Maste
---
With this, the first patch now passes unit tests for me on FreeBSD.
lib/netdev-
On 10 May 2013 14:38, Ben Pfaff wrote:
> On Fri, May 10, 2013 at 02:15:50PM -0400, Ed Maste wrote:
>> This is an incremental to patch v3 of
>> netdev: Factor restoring flags into new "struct netdev_saved_flags",
>> implementing the equivalent change for netdev-bsd
This file presumably started out life as a copy of netdev-linux.c, and
some indentation was not updated after s/linux/bsd/.
Signed-off-by: Ed Maste
---
There are still a few cases that seem not to match style; I adjusted these
because Ethan noticed them in commit 4b60911.
lib/netdev-bsd.c | 18
stering them from netdev_dummy_register() and making that function
> idempotent, so that calling it twice under Linux has no additional effect.
> This commit implements that solution.
>
> CC: Ed Maste
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
Looks good to me, and unit t
On 21 May 2013 10:59, Ben Pfaff wrote:
> On Tue, May 21, 2013 at 05:49:54PM +0900, y...@mwd.biglobe.ne.jp wrote:
>> From: YAMAMOTO Takashi
>>
>> mostly ride on the existing FreeBSD support.
>>
>> Signed-off-by: YAMAMOTO Takashi
>
> I'd like to refine this patch a little in the following way (whi
Signed-off-by: Ed Maste
---
configure.ac | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/configure.ac b/configure.ac
index 1e23289..91f3835 100644
--- a/configure.ac
+++ b/configure.ac
@@ -63,8 +63,9 @@ AC_CHECK_MEMBERS([struct stat.st_mtim.tv_nsec, struct
Introduced in commit 666afb55e84e9118812de81a75655ec9567b7a5b.
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
index 1301aad..87ac79f 100644
--- a/lib/netdev-bsd.c
+++ b/lib/netdev-bsd.c
On 23 May 2013 00:05, Ben Pfaff wrote:
> I have all of these patches ready to apply, after Ed reviewed my changes
> to patch 2.
Apparently I didn't review them well enough; sorry for letting some
minor nits slip through.
___
dev mailing list
dev@openvsw
netdev_bsd_get_next_hop currently lacks an implementation on FreeBSD, so
its arguments are unused; mark them so.
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c | 5 +++--
1 file changed, 3 insertions(+), 2 deletions(-)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
index 87ac79f..0fb0057
On 12 June 2013 07:04, Murphy McCauley wrote:
> (Sorry this isn't an actual reply and is missing context -- I wasn't on the
> list when it was originally posted.)
>
> Simon and I have been in touch about this, and I thought I'd share my
> findings for what they're worth.
>
> The problem is from
On 19 June 2013 16:17, Ben Pfaff wrote:
> This series has two purposes:
>
> * Add a basic thread support library to the tree.
>
> * Get rid of all calls to C library functions that POSIX describes
> as inherently unsafe in a multithreaded program, and then add a
> make-time che
On 19 June 2013 16:17, Ben Pfaff wrote:
> +#if STRERROR_R_CHAR_P
> +/* GNU style strerror_r() might return an immutable static string, or it
> + * might write and return 'buffer', but in either case we can pass the
> + * returned string directly to the caller. */
> +s = strerror_r(
On 19 June 2013 17:22, Ben Pfaff wrote:
> On Wed, Jun 19, 2013 at 05:20:11PM -0400, Ed Maste wrote:
>> On 19 June 2013 16:17, Ben Pfaff wrote:
>> > This series has two purposes:
>> >
>> > * Add a basic thread support library to the tree.
>> >
On 19 June 2013 17:54, Ben Pfaff wrote:
> Oh, that's odd. C11 appears to say that the parentheses are optional
> with _Atomic in this case (you can use _Atomic(...) as a type
> specifier or _Atomic by itself as a type qualifier) but I'll change it
> since that fixes the problem.
Ahh, it looks li
On 19 June 2013 18:48, Ben Pfaff wrote:
> I pushed a fix for this issue (just s/ptb.bufsize/BUFSIZE/) and the
> ovs-atomic-c11.h issue you also mentioned to the "threads" branch at
> https://github.com/blp/ovs-reviews/commits/threads
Thanks Ben, with those two changes your threads branch passes u
X API, where they are available. This commit adds a
>> macro-based wrapper, DEFINE_PER_THREAD_DATA, that takes advantage of the
>> GCC extension where it is available and falls back to the POSIX API
>> otherwise. (I'm not aware of any compilers that implement the C11 feature,
>
On 20 June 2013 20:10, Ben Pfaff wrote:
> Can you confirm that the autoconf test needs #include but
> the actual program doesn't? It looks funny.
Yes. It looks like it should be added in the actual program; right
wow we end up with the definition leaking from another header.
__
In clang's attempt to be GCC compatible it ignores, but does not fail
on, unknown -W options. Configure then ends up added the option to
WARNING_FLAGS, and the warning is then emitted for every file
compiled.
>From config.log:
configure:7388: checking whether clang accepts -Wno-override-init
con
ation with -Werror, clang does fail on unknown -W options. This
> commit adds -Werror during configure's warning tests, which should cause
> the not-understood warnings to be detected that way.
>
> Reported-by: Ed Maste
> Signed-off-by: Ben Pfaff
Than
On 10 July 2013 19:04, Ben Pfaff wrote:
> static void format_log_message(const struct vlog_module *, enum vlog_level,
> - enum vlog_facility, unsigned int msg_num,
> + enum vlog_facility,
> const char *mes
On 10 July 2013 19:03, Ben Pfaff wrote:
> (The usual reason for ppoll() is to ensure atomicity of changing the signal
> mask and blocking, but time_poll() does not need atomicity. Therefore,
> even a userspace emulated ppoll() would work here in a race-free manner, if
> the kernel does not suppor
On 10 July 2013 19:03, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> lib/fatal-signal.c |5 -
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.or
On 10 July 2013 19:04, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 10 July 2013 19:03, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
t; multithreaded process. In a single-threaded process, it can just block
> the signal handler, but in a multithreaded process this does not prevent
> signal delivery to threads other than the one running format_backtrace().
>
> Signed-off-by: Ben Pfaff
Acked
np (FreeBSD)? The thread names will then
appear in gdb or lldb, and in top(1) (on FreeBSD at least).
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 10 July 2013 19:03, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 10 July 2013 19:03, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 10 July 2013 19:04, Ben Pfaff wrote:
> This is harder to implement once threads are introduced.
>
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch
On 10 July 2013 19:03, Ben Pfaff wrote:
> The worker process implementation isn't thread-safe and, once OVS
> itself is threaded, it doesn't make much sense to have a worker
> process anyway.
>
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelis
On 10 July 2013 19:04, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On 10 July 2013 19:03, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
(I rebased on master and had a trivial conflict to resolve.)
X-CudaMail-Whitelist-To: dev@openvswitch.org
___
dev mailing list
dev@openvswitch.org
h
On 10 July 2013 19:03, Ben Pfaff wrote:
> This should make the "timeval" module thread-safe, except for its
> calls into vlog (because vlog is not yet thread-safe).
>
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
X-CudaMail-Whitelist-T
On 12 July 2013 17:54, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
>
Acked-by: Ed Maste
with one comment,
diff --git a/lib/latch.c b/lib/latch.c
> +void
> +(latch_wait)(const struct latch *latch, const char *where)
>
...
> diff --git a/lib/latch.h b/lib/latch.h
> +#de
int f(int x, const char *file, int line);
>
> ...
>
> f(1);/* Expands to a call like f(1, __FILE__, __LINE__); */
>
> However it's somewhat confusing, so this commit stops using that trick.
>
> Reported-by: Ed Maste
> Signed-off-by: Ben Pfaff
Acked-by:
On 30 July 2013 20:31, Joe Stringer wrote:
> This implementation was derived from FreeBSD:
> http://code.google.com/p/freebsd-head/source/browse/sys/libkern/crc32.c
The canonical FreeBSD web repo location for this file would be:
http://svnweb.freebsd.org/base/head/sys/libkern/crc32.c
(It doesn't
Signed-off-by: Ed Maste
---
lib/bfd.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/lib/bfd.c b/lib/bfd.c
index 74b27c4..6f86f26 100644
--- a/lib/bfd.c
+++ b/lib/bfd.c
@@ -15,6 +15,7 @@
#include
#include "bfd.h"
+#include
#include
#include
#include
-
Avoid relying on a non-portable implementation detail for atomic_flag
tests. Per the standard, the only way to obtain the value of the flag
is via the return value from atomic_flag_test_and_set.
Signed-off-by: Ed Maste
---
tests/test-atomic.c | 4 ++--
1 file changed, 2 insertions(+), 2
Commit 1514b275 introduces the use of atomic_read_explicit for
once-only initializers. When using C11 atomics this becomes
atomic_load_explicit, and the first argument needs to be non-const --
with FreeBSD HEAD's in-tree clang 3.3 the build fails for me with:
In file included from lib/bfd.c:34:
./
On 20 August 2013 14:10, Ben Pfaff wrote:
> Without this patch, I get Clang warnings that I don't understand for each
> of the non-static functions in dirs.c:
>
> warning: control reaches end of non-void function
This is a clang bug; in your example the call to bar() is omitted by
clang, and
On 20 August 2013 14:10, Ben Pfaff wrote:
> C11 says that atomic_load() requires a non-const argument, and Clang
> enforces that. This fixes warnings with FreeBSD that uses
> the Clang extensions.
Acked-by: Ed Maste
___
dev mailing
On 20 August 2013 14:10, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
Acked-by: Ed Maste
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Signed-off-by: Ed Maste
---
On FreeBSD pthread_set_name_np's prototype is provided by pthread_np.h.
As I believe it is the only platform to provide the "set_name" (with an
underscore) variant I hope it's fine to use the existing autoconf macro
HAVE_PTHREAD_SET_NAME_NP.
l
On 5 December 2013 18:42, Joe Stringer wrote:
> On systems that provide /proc/cpuinfo similar to Linux on x86, this
> should allow us to choose a better default value for the number of
> upcall handler threads---in particular, it avoids counting hyper-thread
> cores. If /proc/cpuinfo cannot be par
>
> +/* On Linux IA64 systems, glibc avoids using syscalls for clock_gettime().
>
I assume you mean x86-64 or one of the equivalent names here, not IA64.
> +#ifndef __LP64__
> +#define CACHE_TIME 1
> +#endif
>
This will disable the syscall mitigation for 64-bit FreeBSD as well, which
is not (ye
The original developer of the BSD port (Gaetano Catalli) also implemented a
threaded version of the userspace datapath to increase performance. As
with the original port, the threading work was done against Open vSwitch
1.1.0pre2. Giuseppe Lettieri and I have ported it forward, and it's
functiona
Signed-off-by: Ed Maste
---
lib/route-table-bsd.c |4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/route-table-bsd.c b/lib/route-table-bsd.c
index 1c29071..eb4a168 100644
--- a/lib/route-table-bsd.c
+++ b/lib/route-table-bsd.c
@@ -104,7 +104,7
This fixes a warning on FreeBSD.
Signed-off-by: Ed Maste
---
lib/socket-util.c |1 +
1 file changed, 1 insertion(+)
diff --git a/lib/socket-util.c b/lib/socket-util.c
index 314761d..7c40ab8 100644
--- a/lib/socket-util.c
+++ b/lib/socket-util.c
@@ -30,6 +30,7 @@
#include
#include
FreeBSD's system compiler is a somewhat old version of GCC that produced
a spurious warning about a potential unitialized variable use.
Signed-off-by: Ed Maste
---
lib/netdev-bsd.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/netdev-bsd.c b/lib/netdev-bsd.c
On 9 August 2012 11:56, Ben Pfaff wrote:
> I'm curious about the performance improvement. You mentioned a 10x
> performance improvement. How much did CPU usage increase as part of
> that? We already have users who complain when CPU usage spikes to 100%;
> I'm not sure that users will be happy
> We've had good improvements in a similar issue in dpif-linux by
> switching to use "epoll". Can kqueue under FreeBSD do something
> similar?
Yes, epoll and kqueue basically provide the same functionality. What
do you think about incorporating something like libevent to abstract
that?
-Ed
> There may be some confusion about numbering. OpenFlow says that the
> port number of the "local port" is 65534 (OFPP_LOCAL). The datapaths,
> on the other hand, use port number 0 for the "local port". There is
> supposed to be translation going on at the interface between the
> datapath and th
On 14 August 2012 21:21, Ed Maste wrote:
>> There may be some confusion about numbering. OpenFlow says that the
>> port number of the "local port" is 65534 (OFPP_LOCAL). The datapaths,
>> on the other hand, use port number 0 for the "local port". There is
>>> There may be some confusion about numbering. OpenFlow says that the
>>> port number of the "local port" is 65534 (OFPP_LOCAL). The datapaths,
>>> on the other hand, use port number 0 for the "local port". There is
>>> supposed to be translation going on at the interface between the
>>> datap
In the flow hash special ports are stored using OpenFlow constants.
For example the "local port" is stored as 0xfffe (OFPP_LOCAL).
Signed-off-by: Ed Maste
---
lib/dpif-netdev.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netd
On 15 August 2012 19:12, Ben Pfaff wrote:
> upcall->packet is allocated with malloc(), via ofpbuf_new(), but nothing
> ever frees it.
>
> Found by valgrind.
>
> CC: Ed Maste
> Signed-off-by: Ben Pfaff
> ---
> lib/dpif-netdev.c |1 +
> 1 files chan
On 15 August 2012 19:12, Ben Pfaff wrote:
> This is easy enough, so it seems worthwhile now that FreeBSD is starting
> to make more use of the "userspace switch".
>
> CC: Ed Maste
> Signed-off-by: Ben Pfaff
> ---
> lib/dpif-netdev.c | 33 +++--
t.
>
> Found by valgrind.
>
> CC: Ed Maste
> Signed-off-by: Ben Pfaff
> ---
> lib/dpif-netdev.c |2 ++
> 1 files changed, 2 insertions(+), 0 deletions(-)
>
> diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
> index 7fa2720..63b59a3 100644
> --- a/li
On 16 August 2012 11:42, Ben Pfaff wrote:
> On Thu, Aug 16, 2012 at 10:22:21AM -0400, Ed Maste wrote:
>> On 15 August 2012 19:12, Ben Pfaff wrote:
>> > This is easy enough, so it seems worthwhile now that FreeBSD is starting
>> > to make more use of the "userspace
On 15 August 2012 19:09, Ben Pfaff wrote:
> On Wed, Aug 15, 2012 at 10:00:34PM +0000, Ed Maste wrote:
>> In the flow hash special ports are stored using OpenFlow constants.
>> For example the "local port" is stored as 0xfffe (OFPP_LOCAL).
>>
>> Signed-off-by:
I've updated the snapshot of the threaded userspace datapath path with
improvements Giuseppe and I have done over the last couple of weeks.
The current patch is available here:
http://people.freebsd.org/~emaste/openvswitch/threaded-20120820.diff
This patch contains a number of minor improvements
Signed-off-by: Ed Maste
---
As suggested by Ben Pfaff in the 'Threaded userspace datapath' thread.
This looks like a useful utility function refactoring, independent of
any outcome of that discussion.
lib/fatal-signal.c | 4 +---
lib/process.c | 4 +---
lib/signals.c | 4
Signed-off-by: Ed Maste
---
There should probably be an OS check to set only the appropriate environment
variables. I'm not sure if there's a generic autotest way to do that while
building atlocal from atlocal.in though (vs. a switch on $(uname) in the
generated file). Perhaps
On 1 October 2012 17:19, Ben Pfaff wrote:
> On Mon, Oct 01, 2012 at 09:11:31PM +0000, Ed Maste wrote:
>> Signed-off-by: Ed Maste
>
> Applied, thanks.
>
>> There should probably be an OS check to set only the appropriate environment
>> variables. I'm not sure i
Signed-off-by: Ed Maste
---
tests/atlocal.in | 28 +++-
1 file changed, 19 insertions(+), 9 deletions(-)
diff --git a/tests/atlocal.in b/tests/atlocal.in
index c23f8e9..c736df4 100644
--- a/tests/atlocal.in
+++ b/tests/atlocal.in
@@ -33,12 +33,22 @@ if test $HAVE_PYTHON
On 9 October 2012 17:18, Ben Pfaff wrote:
> On Tue, Oct 09, 2012 at 01:54:28PM -0700, Ethan Jackson wrote:
>> > Does the 'argv' code in command-line.c actually malfunction on ESX?
>> > It'd be nice to just leave it in, if not. It's not really tied to
>> > having the Linux datapath, it's orthogona
On 9 October 2012 18:07, Ben Pfaff wrote:
> On Tue, Oct 09, 2012 at 06:03:32PM -0400, Ed Maste wrote:
>>...
>>
>> Interesting - I didn't notice this code before.
>>
>> On FreeBSD we have setproctitle(const char *fmt, ...) for this
>> functionality, w
1 - 100 of 142 matches
Mail list logo
