Great, thanks. Looks good.
I'll leave it to whoever reviews the series as a whole to push this.
On Jun 26, 2014 6:36 PM, "Ryan Wilson 76511" wrote:
> Crap, its late in the day and I can't think / type apparently. Yes 0.04
> million is what I meant.
>
> And I ran 2 more tests in the meantime wi
Crap, its late in the day and I can't think / type apparently. Yes 0.04 million
is what I meant.
And I ran 2 more tests in the meantime with and without the patch and I got a
0.03 and 0.04 million PPS increase, respectively. Nonetheless, the increase is
fairly consistent over 5 different tests.
.4 million or .04 million? There's a big difference.
On Jun 26, 2014 6:24 PM, "Ryan Wilson 76511" wrote:
> Its between 0.2 - 0.6 million PPS increase after running 3 tests with
> and without this patch. So I went with the average of 0.4 :)
>
> And we actually use these annotations elsewhere in
Thanks, I agree with that decision.
On Jun 26, 2014 6:22 PM, "Gurucharan Shetty" wrote:
> On Tue, Jun 24, 2014 at 10:15 AM, Gurucharan Shetty
> wrote:
> >>
> >> There's a lot of weirdness around python and encodings that varies from
> >> version to version and from one Python configuration to th
Its between 0.2 - 0.6 million PPS increase after running 3 tests with and
without this patch. So I went with the average of 0.4 :)
And we actually use these annotations elsewhere in netdev_dpdk_send() where we
measure size of packets and dropped packets, so it would be nice to add these
annotat
On Tue, Jun 24, 2014 at 10:15 AM, Gurucharan Shetty wrote:
>>
>> There's a lot of weirdness around python and encodings that varies from
>> version to version and from one Python configuration to the next. Did
>> you try this on a non-XenServer system too, say Ubuntu? Also you might
>> want to t
That's pretty impressive. Is the performance consistent enough to be sure,
then?
In either case I don't object to the patch.
On Jun 26, 2014 6:17 PM, "Ryan Wilson" wrote:
> Since dropped packets due to large packet size or lack of memory
> are unlikely, it is best to add OVS_UNLIKELY annotations
This patch refactors dpdk_queue_flush() to reuse code in
dpdk_queue_pkts().
Signed-off-by: Ryan Wilson
---
lib/netdev-dpdk.c | 37 -
1 file changed, 16 insertions(+), 21 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 72add87..084e8cd
Since dropped packets due to large packet size or lack of memory
are unlikely, it is best to add OVS_UNLIKELY annotations to these
conditions.
With DPDK fast path forwarding test, this increased throughtput
from 4.12 to 4.16 million packets per second.
Signed-off-by: Ryan Wilson
---
lib/netdev-
This patch fixes a bug where rte_pktmbuf_alloc() would fail and
packets which succeeded to allocate memory with rte_pktmbuf_alloc()
would not be sent and leak memory.
Also, as a byproduct of using a local variable to record dropped
packets, this reduces the locking of the netdev's mutex when
multi
The current timestamp should be set every time the queue is flushed.
Thus, if DRAIN_TSC timer cycles have passed since the last timestamp,
the send queue should be flushed again.
Signed-off-by: Ryan Wilson
---
lib/netdev-dpdk.c |8 ++--
1 file changed, 2 insertions(+), 6 deletions(-)
di
Per style requirements, functions local to a module should not be
inlined.
Signed-off-by: Ryan Wilson
---
lib/netdev-dpdk.c |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index dfa7d77..0aee14e 100644
--- a/lib/netdev-dpdk.c
+++ b/lib
I'll submit as 3 separate patches.
The reducing locking is really a byproduct of the bug fix, so its
unnecessary to add performance measurements to that.
I'll add performance measurements for the OVS_UNLIKELY annotations (if
they do in fact help as I predict).
Ryan
On 6/26/14 5:48 PM, "Ethan Ja
On Fri, Jun 27, 2014 at 01:50:25AM +0200, Thomas Graf wrote:
> Reduces the dependency on autoconf from 2.64 to 2.63 to ease building
> on older platforms. There is only a few macros missing and they can
> be provided easily.
>
> A handful of tests needed modification. The difference in quoting
> b
> This patch reduces locking when dpdk_do_tx_copy() by totalling
> the dropped packets in a local variable before adding to the
> netdev's dropped stats field.
>
> This patch also fixes a bug where rte_pktmbuf_alloc() would fail
> and packets which succeeded to allocate memory with rte_pktmbuf_allo
This patch reduces locking when dpdk_do_tx_copy() by totalling
the dropped packets in a local variable before adding to the
netdev's dropped stats field.
This patch also fixes a bug where rte_pktmbuf_alloc() would fail
and packets which succeeded to allocate memory with rte_pktmbuf_alloc()
would n
The current timestamp should be set every time the queue is flushed.
Thus, if DRAIN_TSC timer cycles have passed since the last timestamp,
the send queue should be flushed again.
Signed-off-by: Ryan Wilson
---
lib/netdev-dpdk.c |8 ++--
1 file changed, 2 insertions(+), 6 deletions(-)
di
This patch refactors dpdk_queue_flush() to reuse code in
dpdk_queue_pkts().
Signed-off-by: Ryan Wilson
---
lib/netdev-dpdk.c | 37 -
1 file changed, 16 insertions(+), 21 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index 72add87..084e8cd
On packet recv OVS CB is initialized in multiple function.
Following patch moves all these initialization to
ovs_vport_receive().
This patch also save a check in execute actions.
Signed-off-by: Pravin B Shelar
---
datapath/actions.c | 3 ---
datapath/datapath.c | 6 ++
datapath/datapath.h |
Reduces the dependency on autoconf from 2.64 to 2.63 to ease building
on older platforms. There is only a few macros missing and they can
be provided easily.
A handful of tests needed modification. The difference in quoting
behaviour between 2.63 and later require the m4_define() to be
manually un
On Thu, Jun 26, 2014 at 07:00:16AM -0700, Gurucharan Shetty wrote:
> Till now, we were initializing 'unix_epoch' through time_init().
> But if there was a call directly to xgettimeofday(), we would
> miss the initialization causing overflows. This would cause
> 'ovs-ofctl benchmark' to produce wron
On Wed, Jun 25, 2014 at 06:22:30PM +0900, Simon Horman wrote:
> On Mon, Jun 23, 2014 at 05:14:52PM -0700, Ben Pfaff wrote:
> > On Mon, Jun 16, 2014 at 11:29:30AM +0900, Simon Horman wrote:
> > > Handle modify and delete commands in OpenFlow1.4 flow monitor requests.
> > > These commands are not yet
On Wed, Jun 25, 2014 at 04:13:38PM -0700, Jesse Gross wrote:
> On Wed, Jun 25, 2014 at 11:02 AM, Ben Pfaff wrote:
> > Signed-off-by: Ben Pfaff
> > ---
> > lib/dpif.c | 17 +
> > 1 file changed, 17 insertions(+)
>
> Acked-by: Jesse Gross
Thank you for the reviews. I applied
On Wed, Jun 25, 2014 at 06:02:45PM +1200, Joe Stringer wrote:
> This allows us to remove some of the sleeps from the testsuite.
>
> Signed-off-by: Joe Stringer
> ---
> v2: Add udpif_run(), manage connection from main thread.
> v1: First post.
Acked-by: Ben Pfaff
I realised that branch-2.3 also needed this fix, so I also pushed it there.
On 27 June 2014 10:46, Joe Stringer wrote:
> On 27 June 2014 09:51, Ben Pfaff wrote:
>>
>> Assuming you did a test build,
>>
>> Acked-by: Ben Pfaff
>>
>
> Yes, I checked again before pushing.
>
> Thanks, applied to ma
Thanks, also received private ack from Pravin.
I pushed this to master.
On 26 June 2014 21:55, Jarno Rajahalme wrote:
> Acked-by: Jarno Rajahalme
>
> On Jun 25, 2014, at 8:24 PM, Joe Stringer wrote:
>
> > These two cases used to be treated differently for IPv4/IPv6, but they
> > are now iden
On 27 June 2014 09:51, Ben Pfaff wrote:
>
> Assuming you did a test build,
>
> Acked-by: Ben Pfaff
>
Yes, I checked again before pushing.
Thanks, applied to master.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Thu, Jun 26, 2014 at 09:50:39AM +, Alessandro Pilotti wrote:
> The user space code that you are seeing is based on our port (the one
> we submitted some months ago to this ML, before Guru's Windows
> patching work started on master), that's why you see all those
> commits.
OK.
> The good n
On Tue, Jun 24, 2014 at 06:40:02PM -0700, Alex Wang wrote:
> Signed-off-by: Alex Wang
>
> ---
> PATCH -> V2:
> - Use readdir_r() instead of readdir() for reentrency.
> - Address review comments from Thomas Graf.
> - Add dummy interface for WIN32 case.
I think that this should be dummied not just
If someone wants to change the queue size, maybe bigger, this provides a
formula to allow this.
There are advantages and also some disadvantages on using a larger queue.
Doubling the size does not work since 4096 is invalid and will fail
compilation. Unless they
researched this carefully, they m
On Wed, Jun 25, 2014 at 11:41:41AM +1200, Joe Stringer wrote:
> Various recent commits have introduced build failures on FreeBSD. This
> patch fixes them.
>
> Signed-off-by: Joe Stringer
Assuming you did a test build,
Acked-by: Ben Pfaff
___
dev mail
On Thu, Jun 26, 2014 at 7:08 AM, Polehn, Mike A wrote:
> (n*32<4096) is the formula for calculating the buffer size. The maximum size
> is not
> 4096 but 4096 - 32 = 4064. I ran into this issue on a different project
> where I needed to
> use the largest buffers size available. Often 2**n is us
On Tue, Jun 24, 2014 at 11:53:03AM -0700, Gurucharan Shetty wrote:
> vlan splinters is to workaround buggy network drivers of Linux.
> Skip the test for Windows.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvsw
On Tue, Jun 24, 2014 at 11:53:02AM -0700, Gurucharan Shetty wrote:
> The unit test, "OFPST_TABLE reply - OF1.2" in ofp-print.at
> sends a very large hex string as an argument to 'ovs-ofctl ofp-print'.
> The length of the hex string exceeds the maximum command line length
> in Windows. With this com
On Tue, Jun 24, 2014 at 11:53:01AM -0700, Gurucharan Shetty wrote:
> Also, fflush(stderr) when we raise a signal. The test
> this commit is changing would fail otherwise.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
___
dev mailing list
de
On Tue, Jun 24, 2014 at 11:53:00AM -0700, Gurucharan Shetty wrote:
> By default, three digit exponents are printed on Windows.
> Many unit tests in Open vSwitch expect two digit exponents.
>
> Signed-off-by: Gurucharan Shetty
Interesting, this portability issue is new to me.
Acked-by: Ben Pfaff
On Tue, Jun 24, 2014 at 11:52:59AM -0700, Gurucharan Shetty wrote:
> rename() on an existing destination file fails on Windows. This commit
> worksaround that problem.
>
> There are two tests that test it. But both of them use the ovsdb-server's
> --run option for the test and it does not exist in
On Tue, Jun 24, 2014 at 11:52:58AM -0700, Gurucharan Shetty wrote:
> Commit ebed9f78(ovsdb-server: Improve message for "add-db" of
> database already open.) improved the error message seen when
> opening an already opened database on Linux. For Windows,
> we still need to look for the lockfile erro
On Tue, Jun 24, 2014 at 11:52:57AM -0700, Gurucharan Shetty wrote:
> interface-reconfigure.at mostly tests functionality for Xenserver
> and the tests use some POSIX only features. Skip them for Windows.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
__
On Thu, Jun 26, 2014 at 01:27:00PM -0700, Ben Pfaff wrote:
> On Tue, Jun 24, 2014 at 11:52:56AM -0700, Gurucharan Shetty wrote:
> > sflow feature needs to be investigated for Windows. Right now
> > test-sflow related tests do not pass because of LOOPBACK_INTERFACE
> > constraints for 'agent'. Add a
On Tue, Jun 24, 2014 at 11:52:56AM -0700, Gurucharan Shetty wrote:
> sflow feature needs to be investigated for Windows. Right now
> test-sflow related tests do not pass because of LOOPBACK_INTERFACE
> constraints for 'agent'. Add a TODO item and skip the tests.
>
> Signed-off-by: Gurucharan Shett
On Tue, Jun 24, 2014 at 11:52:51AM -0700, Gurucharan Shetty wrote:
> "xargs echo" introduces "^M" character inbetween the
> records on Windows. Workaround it.
>
> Signed-off-by: Gurucharan Shetty
I like this better anyway.
Acked-by: Ben Pfaff
___
dev
On Tue, Jun 24, 2014 at 11:52:54AM -0700, Gurucharan Shetty wrote:
> On Windows ECONNRESET is WSAECONNRESET.
>
> Also, "unix" connections are done through TCP sockets.
> For the 'refuse-connection' test, the error message for Windows
> is WSAECONNRESET instead of EPIPE.
>
> Signed-off-by: Gurucha
On Tue, Jun 24, 2014 at 11:52:53AM -0700, Gurucharan Shetty wrote:
> msys has a set of rules which triggers an automatic conversion of
> arguments into something else to suit Windows requirements. Sometimes
> this also causes unwanted conversions. Details of the rules is here:
> http://www.mingw.or
On Tue, Jun 24, 2014 at 11:52:52AM -0700, Gurucharan Shetty wrote:
> On Windows, when a peer terminates without calling a close
> on socket fd, the server ends up printing "connection dropped"
> warning messages. We probably don't want those warning messages
> when the error is WSAECONNRESET.
>
>
On Tue, Jun 24, 2014 at 11:52:47AM -0700, Gurucharan Shetty wrote:
> On Windows, we create "unix sockets" by creating TCP sockets
> and hiding the TCP port number in files. When we close the
> pstream session, we need to delete the file.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
On Tue, Jun 24, 2014 at 11:52:50AM -0700, Gurucharan Shetty wrote:
> For this particular test, we pass the PKIDIR through a
> ovsdb-tool transact and msys does not convert the path style.
> (On Windows, we have to pass the directory in the form C:/foo/bar.pem.)
>
> So get the Windows style path th
On Tue, Jun 24, 2014 at 11:52:49AM -0700, Gurucharan Shetty wrote:
> On Windows, we really do not have any "socket" files created.
> We instead create a regular file with a TCP port number written
> inside it.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
On Tue, Jun 24, 2014 at 11:52:48AM -0700, Gurucharan Shetty wrote:
> We do not support that option in Windows.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/de
Sorry I overlooked your response.
When I use dump-flows with " -O OpenFlow13", I see as expected.
# ovs-ofctl -O OpenFlow13 dump-flows br0
OFPST_FLOW reply (OF1.3) (xid=0x2):
cookie=0x0, duration=1647.125s, table=0, n_packets=0, n_bytes=0,
dl_dst=aa:bb:cc:dd:ee:ff actions=goto_table:1
Thanks. Y
On Thu, Jun 26, 2014 at 8:45 PM, Jarno Rajahalme wrote:
>
> On Jun 26, 2014, at 5:32 AM, Nirmalanand Jebakumar
> wrote:
>
>> Hello,
>>
>> I'm running OVS version 2.1.2 on RHEL 6.5.
>> When I try to add a flow with the goto_table keyword, it fails with the
>> below message:
>> # ovs-ofctl add-fl
On Fri, Jun 27, 2014 at 02:03:31AM +0900, YAMAMOTO Takashi wrote:
> > Signed-off-by: Flavio Leitner
>
> is "Implementation Details" an appropriate section for this?
> otherwise looks good to me.
Good question. I am all ears. To me 'Implementation Details'
is a bit better than 'VLANs'.
> Acked-
On Fri, Jun 27, 2014 at 02:03:31AM +0900, YAMAMOTO Takashi wrote:
> > Signed-off-by: Flavio Leitner
>
> is "Implementation Details" an appropriate section for this?
> otherwise looks good to me.
>
> Acked-by: YAMAMOTO Takashi
Fine by me.
___
dev mail
> Signed-off-by: Flavio Leitner
is "Implementation Details" an appropriate section for this?
otherwise looks good to me.
Acked-by: YAMAMOTO Takashi
> ---
> FAQ | 4
> 1 file changed, 4 insertions(+)
>
> diff --git a/FAQ b/FAQ
> index c2520fc..3470983 100644
> --- a/FAQ
> +++ b/FAQ
> @@
On Thu, Jun 26, 2014 at 04:32:06PM +, Polehn, Mike A wrote:
> There is a very good reason for putting constants on the left hand side of a
> compare statement.
> For example:
> if (NULL = x)
> will be a compiler error, while the following will compile and need debugging:
> if (x = NULL)
>
>
I'll rebase, study and correct to OVS coding style and repost.
There is a very good reason for putting constants on the left hand side of a
compare statement.
For example:
if (NULL = x)
will be a compiler error, while the following will compile and need debugging:
if (x = NULL)
Although I try
On Jun 13, 2014, at 10:34 AM, Ben Pfaff wrote:
> On Mon, Jun 09, 2014 at 11:53:54AM -0700, Jarno Rajahalme wrote:
>> Reduce the number of goto statements by returning via a new helper
>> fill_range_wc() when no match is found.
>>
>> Signed-off-by: Jarno Rajahalme
>
> I slightly lean toward ma
Thanks for the review! Pushed to master,
Jarno
On Jun 13, 2014, at 10:29 AM, Ben Pfaff wrote:
> On Mon, Jun 09, 2014 at 11:53:53AM -0700, Jarno Rajahalme wrote:
>> miniflow_and_mask_matches_flow_wc() fills in the masks in flow
>> wildcards, so a separate step to that effect is no longer neede
Pushed to master with changes as suggested.
Defined inline functions shown below, be32 variants were obsolete by earlier
changes, so did not add them now.
+static inline uint32_t flow_u32_value(const struct flow *flow, size_t index)
+{
+return ((uint32_t *)(flow))[index];
+}
+
+static inline
Solid suggestions as always :-) Pushed to master with the changes,
Jarno
On Jun 25, 2014, at 11:33 AM, Ben Pfaff wrote:
> On Wed, Jun 25, 2014 at 04:15:54AM -0700, Jarno Rajahalme wrote:
>> Factor out the priority vector code from the classifier.
>>
>> Making the classifier use RCU instead o
On Jun 26, 2014, at 5:32 AM, Nirmalanand Jebakumar
wrote:
> Hello,
>
> I'm running OVS version 2.1.2 on RHEL 6.5.
> When I try to add a flow with the goto_table keyword, it fails with the
> below message:
> # ovs-ofctl add-flow br0 "table=0,dl_dst=aa:bb:cc:dd:ee:ff,
> actions=goto_table:1"
>
Till now, we were initializing 'unix_epoch' through time_init().
But if there was a call directly to xgettimeofday(), we would
miss the initialization causing overflows. This would cause
'ovs-ofctl benchmark' to produce wrong results.
This commit fixes it by having a separate initialization functi
Hi,
Are you interested in acquiring *CHARON-VAX User’s Email list *for your
marketing campaigns?
*Titles like – IT Decision Makers, CEO, CFO, IT Engineers and many more.*
We provide the top Decision Makers Name, Email, Title, Phone and company
details across North America, EMEA, ASIA Pacif
(n*32<4096) is the formula for calculating the buffer size. The maximum size is
not
4096 but 4096 - 32 = 4064. I ran into this issue on a different project where
I needed to
use the largest buffers size available. Often 2**n is used for queue sizing,
but is not the
case for the PMD driver.
M
Hello,
I'm running OVS version 2.1.2 on RHEL 6.5.
When I try to add a flow with the goto_table keyword, it fails with the
below message:
# ovs-ofctl add-flow br0 "table=0,dl_dst=aa:bb:cc:dd:ee:ff,
actions=goto_table:1"
ovs-ofctl: none of the usable flow formats (OXM,OpenFlow11) is among the
allow
On Thu, Jun 26, 2014 at 01:00:33PM +0900, YAMAMOTO Takashi wrote:
> > On Wed, Jun 25, 2014 at 11:07:31AM +0900, YAMAMOTO Takashi wrote:
> >> Signed-off-by: YAMAMOTO Takashi
> >
> > Multicast snooping works in the same way (separated by vlans), but
> > I am afraid there is no special term for it.
Signed-off-by: Flavio Leitner
---
FAQ | 4
1 file changed, 4 insertions(+)
diff --git a/FAQ b/FAQ
index c2520fc..3470983 100644
--- a/FAQ
+++ b/FAQ
@@ -520,6 +520,10 @@ A: Open vSwitch has two kinds of flows (see the previous
question), so
dumps datapath flows for only the specifi
Acked-by: Jarno Rajahalme
On Jun 25, 2014, at 8:24 PM, Joe Stringer wrote:
> These two cases used to be treated differently for IPv4/IPv6, but they
> are now identical.
>
> Signed-off-by: Joe Stringer
> ---
> datapath/flow_netlink.c | 13 +++--
> 1 file changed, 3 insertions(+), 10 d
Hi Ben,
Thanks for reviewing our repos.
The user space code that you are seeing is based on our port (the one we
submitted some
months ago to this ML, before Guru's Windows patching work started on master),
that's why
you see all those commits.
The good news is that thanks to the way we appr
69 matches
Mail list logo