Is dpif-netdev support mega-flow ?
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Future patches will change the recirc action implementation to not
using recursion. The stack depth detection is no longer necessary.
Signed-off-by: Andy Zhou
---
datapath/actions.c | 63 -
datapath/datapath.c | 6 ++---
datapath/datapath.h |
Since kernel stack is limited in size, it is not wise to using
recursive function with large stack frames.
This patch provides an alternative implementation of recirc action
without using recursion.
A per CPU fixed sized, 'deferred action FIFO', is used to store either
recirc or sample actions en
dp_netdev_free() must free 'dp->upcall_rwlock', but when upcalls are disabled
(if the datapath is being freed upcalls should be disabled) 'dp->upcall_rwlock'
is taken and freeing it causes an assertion to fail.
This commit takes makes sure that the upcalls are disabled and releases
'dp->upcall_rwl
make distcheck runs a superset of make check and will additionally
catch failures in adding new files to the Makefile. It will also test
installation and uninstallation of the package.
Signed-off-by: Thomas Graf
---
.travis/build.sh | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --g
On 8/29/14, 4:33 PM, "Pravin Shelar" wrote:
>On Fri, Aug 29, 2014 at 4:06 PM, Daniele Di Proietto
> wrote:
>> This series introduces in dpif-netdev an exact match cache for
>>frequently
>> used flows. It uses the RSS hash (currently implemented only for
>>netdev-dpdk)
>> to search the miniflow
On Fri, Aug 29, 2014 at 4:06 PM, Daniele Di Proietto
wrote:
> This series introduces in dpif-netdev an exact match cache for frequently
> used flows. It uses the RSS hash (currently implemented only for netdev-dpdk)
> to search the miniflow in an hash table.
>
> While there might be still some tun
On Fri, Aug 29, 2014 at 04:21:52PM -0700, Jarno Rajahalme wrote:
> I changed the comment accordingly and pushed the series. Thanks for
> the reviews, and have a nice long weekend!
You're welcome. You too!
___
dev mailing list
dev@openvswitch.org
http://
On Aug 29, 2014, at 2:31 PM, Ben Pfaff wrote:
> On Fri, Aug 29, 2014 at 02:10:42PM -0700, Jarno Rajahalme wrote:
>> Seq objects would be really hard to use if they did not provide
>> acquire-release semantics. Currently they do that via
>> ovs_mutex_lock()/ovs_mutex_unlock(), respectively. Doc
Since lookups in the classifier can be pretty expensive, we introduce this
(thread local) cache which simply compares the miniflows of the packets
Signed-off-by: Daniele Di Proietto
---
lib/dpif-netdev.c | 447 +-
1 file changed, 374 insertions
This series introduces in dpif-netdev an exact match cache for frequently
used flows. It uses the RSS hash (currently implemented only for netdev-dpdk)
to search the miniflow in an hash table.
While there might be still some tuning left to do, these patch provides
significant performance improveme
These function are used to stored the packet hash. 'netdev-dpdk' automatically
set this value to the RSS hash returned by the NIC. Other 'netdev's set it to 0
(which is an invalid hash value), so that callers can compute the hash on their
own.
If DPDK support is enabled, struct dpif_packet's membe
Oops, seems there was no tomorrow!
Pushed, both patches, finally. Sorry for the delay :-)
Jarno
On Aug 12, 2014, at 6:01 PM, Jarno Rajahalme wrote:
> Will do, tomorrow :-)
>
> Jarno
>
>> On Aug 12, 2014, at 1:58 PM, Ben Pfaff wrote:
>>
>> I'm happy enough with these patches. Jarno, do
Hi,
I have sent a v4 for review.
All the code is from nithin only. I have only merged the patches and rebased it.
a. Since i am NOT the author and did review the changes, hence i have kept my
name in Acked-by.
b. As per the CONTRIBUTING guidelines i have added my name to Signed-off-by
(because
Thanks for the suggestions
I applied all of them except (as discussed offline):
- some style fixes (CodingStyle shows that operator && and || should be
after a newline)
- the removal of dp_netdev_queu_batches() Œmax_batches¹ parameter. Perhaps
we can remove
it in the future, but now it is neede
On Fri, Aug 15, 2014 at 7:07 AM, Polehn, Mike A wrote:
> Adds documentation on how to run IVSHMEM communication through VM.
>
I think INSTALL.DPDK is getting rather large and hard to understand
with all details.
so I dropped "Alternative method to get QEMU, download and build from
OVDK" section.
W
On Wed, Aug 6, 2014 at 6:07 PM, Daniele Di Proietto
wrote:
>
>
> On 8/6/14, 3:30 PM, "Pravin Shelar" wrote:
>
>>On Wed, Jul 30, 2014 at 8:51 AM, Daniele Di Proietto
>> wrote:
>>> DPDK rings must have a power-of-two size.
>>>
>>> Signed-off-by: Daniele Di Proietto
>>> ---
>>> lib/netdev-dpdk.c |
On Fri, Aug 15, 2014 at 7:03 AM, Polehn, Mike A wrote:
> Separates loop list process size from the endpoint DPDK queue size.
> Corrected DPDK queue size to be a power of 2 which allows dpdkr interface to
> be created.
> Increased queue size to improve zero loss data rate.
> Changed NIC queue size
In this patch we add a context structure for collecting all the parameters
passed from usersapce in one place. The idea is to reduce the number of
parameters being passed to the netlink command handler functions.
It can be argued that not all functions require all the arguments, but this
approach
In this patch, we add support for the GET_DP netlink command to dump
the datpaaths. The userspace workflow to get this to work is the same
as on Linux. dpif-linux.c initiates a dump start by writing a netlink
message, and after that continues to read data from the kernel while
the kernel has data.
I didn't realize earlier that version in a netlink message was a
UINT8. So, fixing that here.
Also, some of the commands don't pass a valid DP value. Hence adding
a field to identify such commands.
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribu
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this change to Nithin Raju
---
datapath-windows/ovsext/Datapath.h | 58 +++---
1 file changed, 48 insertions(+), 10 deletions(-)
diff --git a/datapath-windows/ovs
Thanks for the review. Pushed.
On Fri, Aug 29, 2014 at 2:32 PM, Pravin Shelar wrote:
> Can you add some commit msg.
>
> On Fri, Aug 29, 2014 at 1:22 PM, Andy Zhou wrote:
>> Signed-off-by: Andy Zhou
>> ---
>> datapath/actions.c | 45 +++--
>> 1 file chang
On Fri, Aug 29, 2014 at 1:41 PM, Andy Zhou wrote:
> On Thu, Aug 28, 2014 at 3:40 PM, Pravin Shelar wrote:
>> On Wed, Aug 27, 2014 at 4:13 AM, Andy Zhou wrote:
>>> Since kernel stack is limited in size, it is not wise to using
>>> recursive function with large stack frames.
>>>
>>> This patch pro
On Fri, Aug 29, 2014 at 02:10:44PM -0700, Jarno Rajahalme wrote:
> We can use a normal bool and rely on the mutex_lock/unlock and an
> atomic_thread_fence for synchronization.
>
> Also flip the return value of ovsthread_once_start__() to match the
> one of ovsthread_once_start().
>
> Signed-off-b
On Fri, Aug 29, 2014 at 02:10:43PM -0700, Jarno Rajahalme wrote:
> barrier->count is used as a simple counter and is not expected the
> synchronize the state of any other variable, so we can use atomic_count,
> which uses relaxed atomics.
>
> Ditto for the 'next_id' within ovsthread_wrapper().
>
Can you add some commit msg.
On Fri, Aug 29, 2014 at 1:22 PM, Andy Zhou wrote:
> Signed-off-by: Andy Zhou
> ---
> datapath/actions.c | 45 +++--
> 1 file changed, 19 insertions(+), 26 deletions(-)
>
> diff --git a/datapath/actions.c b/datapath/actions.c
>
On Fri, Aug 29, 2014 at 02:10:42PM -0700, Jarno Rajahalme wrote:
> Seq objects would be really hard to use if they did not provide
> acquire-release semantics. Currently they do that via
> ovs_mutex_lock()/ovs_mutex_unlock(), respectively. Document the
> behavior so that it is safer to rely on th
On Fri, Aug 29, 2014 at 02:11:26PM -0700, Jarno Rajahalme wrote:
>
> On Aug 28, 2014, at 9:42 AM, Ben Pfaff wrote:
>
> > On Fri, Aug 22, 2014 at 01:58:18PM -0700, Jarno Rajahalme wrote:
> >> Avoiding the atomic read may help if a function using
> >> ovsthread_once_start() is ever called in a loo
We can use a normal bool and rely on the mutex_lock/unlock and an
atomic_thread_fence for synchronization.
Also flip the return value of ovsthread_once_start__() to match the
one of ovsthread_once_start().
Signed-off-by: Jarno Rajahalme
---
v3: Simplified.
lib/ovs-thread.c | 14 ++---
On Aug 28, 2014, at 9:42 AM, Ben Pfaff wrote:
> On Fri, Aug 22, 2014 at 01:58:18PM -0700, Jarno Rajahalme wrote:
>> Avoiding the atomic read may help if a function using
>> ovsthread_once_start() is ever called in a loop, as the new
>> 'maybe_not_done' can be kept in a register. The atomic read
Seq objects would be really hard to use if they did not provide
acquire-release semantics. Currently they do that via
ovs_mutex_lock()/ovs_mutex_unlock(), respectively. Document the
behavior so that it is safer to rely on that elsewhere.
Signed-off-by: Jarno Rajahalme
---
lib/ovs-thread.c |
barrier->count is used as a simple counter and is not expected the
synchronize the state of any other variable, so we can use atomic_count,
which uses relaxed atomics.
Ditto for the 'next_id' within ovsthread_wrapper().
Signed-off-by: Jarno Rajahalme
---
v2: Do not seq_read again after seq_chang
On Fri, Aug 29, 2014 at 01:15:53PM +, Niels van Adrichem wrote:
> For some experiments I did around April/May for a paper I wrote I
> changed some sourcecode to ofproto-dpif-xlate.c to have the FastFailover
> Group Table look at the BFD status directly.
> We did this because we found that altho
On 08/29/14 at 01:53pm, Ben Pfaff wrote:
> On Fri, Aug 29, 2014 at 01:27:49PM -0700, Ben Pfaff wrote:
> > On Fri, Aug 29, 2014 at 07:56:26PM +0200, Thomas Graf wrote:
> > > Enable build notifications to bu...@openvswitch.org
> > >
> > > Co-authored-by: Ben Pfaff
> > > Signed-off-by: Thomas Graf
On Fri, Aug 29, 2014 at 01:27:49PM -0700, Ben Pfaff wrote:
> On Fri, Aug 29, 2014 at 07:56:26PM +0200, Thomas Graf wrote:
> > Enable build notifications to bu...@openvswitch.org
> >
> > Co-authored-by: Ben Pfaff
> > Signed-off-by: Thomas Graf
> > ---
> > v2: Add secure token as email recipient
>
On Thu, Aug 28, 2014 at 3:40 PM, Pravin Shelar wrote:
> On Wed, Aug 27, 2014 at 4:13 AM, Andy Zhou wrote:
>> Since kernel stack is limited in size, it is not wise to using
>> recursive function with large stack frames.
>>
>> This patch provides an alternative implementation of recirc action
>> wi
On Tue, Aug 19, 2014 at 5:11 PM, Daniele Di Proietto
wrote:
> Since lookups in the classifier can be pretty expensive, we introduce this
> (thread local) cache which simply compares the miniflows of the packets
>
> Signed-off-by: Daniele Di Proietto
I have few suggestions in attached patch, can
Hi Ben,
Thank you so much for the reply.
No problem @ all, i can merge the review comments in original series of 4
patches.
I'll send a v4 soon.
Regards,
Ankur
From: Ben Pfaff
Sent: Friday, August 29, 2014 1:30 PM
To: Ankur Sharma
Cc: Eitan Eliahu; Sam
On Fri, Aug 29, 2014 at 07:11:41PM +, Ankur Sharma wrote:
> I have talked to ben and he is fine with the approach of handling
> the review comment in another patch in the same series. But yes
> ideally we should try to keep the review comment fix in the same
> patch.
Is this just a matter of s
On Fri, Aug 29, 2014 at 07:56:26PM +0200, Thomas Graf wrote:
> Enable build notifications to bu...@openvswitch.org
>
> Co-authored-by: Ben Pfaff
> Signed-off-by: Thomas Graf
> ---
> v2: Add secure token as email recipient
>
> Ben: I can't really test this change without commit access to the mai
On Fri, Aug 29, 2014 at 12:38 PM, Pravin Shelar wrote:
> On Fri, Aug 15, 2014 at 11:18 AM, Jesse Gross wrote:
>> The checksum of ICMPv6 packets uses the IP pseudoheader as part of
>> the calculation, unlike ICMP in IPv4. This was not implemented,
>> which means that modifying the IP addresses of
Signed-off-by: Andy Zhou
---
datapath/actions.c | 45 +++--
1 file changed, 19 insertions(+), 26 deletions(-)
diff --git a/datapath/actions.c b/datapath/actions.c
index 7f25553..c8951f0 100644
--- a/datapath/actions.c
+++ b/datapath/actions.c
@@ -688,7 +68
> > ---
> > */
> > static __inline VOID *
> > -GetStartAddrNBL(const NET_BUFFER_LIST *_pNB)
> > +GetStartAddrNB(const NET_BUFFER *_pNB)
>
> Generally, we have not used '_' prefix for function parameter names. We
> use them for macros etc, but not for functions. Do you prefer it this way? I
> a
hi Sam/Alin,
Thanks for working on this.
In general, it would have been better if the two changes:
1. handle NBL with multiple NBs
2. Optimizing the parsing code to handle NBs rather than NBLs.
could have been split into different review. I was not a big fan of split up
changes into multiple rev
hi Sam/Alin,
Thanks for working on this.
In general, it would have been better if the two changes:
1. handle NBL with multiple NBs
2. Optimizing the parsing code to handle NBs rather than NBLs.
could have been split into different review. I was not a big fan of split up
changes into multiple rev
On Fri, Aug 29, 2014 at 12:50:24PM -0700, Jarno Rajahalme wrote:
>
> On Aug 29, 2014, at 9:36 AM, Jarno Rajahalme wrote:
>
> >
> > On Aug 28, 2014, at 9:32 AM, Ben Pfaff wrote:
> >
> >> On Fri, Aug 22, 2014 at 01:58:17PM -0700, Jarno Rajahalme wrote:
> >>> barrier->count is used as a simple c
On Aug 29, 2014, at 9:36 AM, Jarno Rajahalme wrote:
>
> On Aug 28, 2014, at 9:32 AM, Ben Pfaff wrote:
>
>> On Fri, Aug 22, 2014 at 01:58:17PM -0700, Jarno Rajahalme wrote:
>>> barrier->count is used as a simple counter and is not expected the
>>> synchronize the state of any other variable, s
On Fri, Aug 15, 2014 at 11:18 AM, Jesse Gross wrote:
> The checksum of ICMPv6 packets uses the IP pseudoheader as part of
> the calculation, unlike ICMP in IPv4. This was not implemented,
> which means that modifying the IP addresses of an ICMPv6 packet
> would cause the checksum to no longer be c
Hi Ankur,
I guess the distinction is a little subjective. Since you are making code
changes based on review comments I thought you become a co-author. Perhaps your
contribution is not enough to qualify you as one. In any case, it would be good
to have some consensus and wait for Samuels respons
Hi Saurabh,
I am not a co-Author.
As per CONTRIBUTING guidelines even if i am submitting a patch on someone
else's behalf i should add my name in Signed-off by.
I have categorically mentioned in all the patches that the change is attributed
to nithin.
Kindly, let me know if there is any confusi
>
> I have talked to ben and he is fine with the approach of handling the review
> comment in another patch in the same series. But yes ideally we should try
> to keep the review comment fix in the same patch.
>
I would go with the reviewers preference. No one likes to block reviews
unnecesaril
I would really like to discourage this practice of - "Lets commit this code and
we will take your comments in separate reviews". I just replied with a similar
comment to one of Samuel's review where he took care of comments in a separate
review. This does not make for a high quality code that we
Hi Sam,
I have talked to ben and he is fine with the approach of handling the review
comment in another patch in the same series. But yes ideally we should try to
keep the review comment fix in the same patch.
As i can see nithin has handled most of your comments. If you think something
critic
Per review comment, in this patch we refactor the code to create a
OvsSetupDumpStart() which can be leveraged by dump functions in the
future. I have not refactored the code that continues the dump
operation primarily since it is not final yet. Once the netlink set
APIs are in place, we can refacto
Also renamed the validateDp boolean to a more meaningful name.
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this to Nithin Raju
---
datapath-windows/ovsext/Datapath.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this to Nithin Raju
---
datapath-windows/ovsext/Datapath.h | 10 ++
1 file changed, 6 insertions(+), 4 deletions(-)
diff --git a/datapath-windows/ovsext/Datapath.h
b/datapath-windows/ov
A review comment was that we should check for device operation
rather than whether the dump context while deciding if the operation
is a dump start or a dump continue.
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this to Nithin Raju
---
da
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this to Nithin Raju
---
datapath-windows/ovsext/Datapath.h | 17 +
1 file changed, 9 insertions(+), 8 deletions(-)
diff --git a/datapath-windows/ovsext/Datapath.h
b/datapath-win
In this patch, we add support for the GET_DP netlink command to dump
the datpaaths. The userspace workflow to get this to work is the same
as on Linux. dpif-linux.c initiates a dump start by writing a netlink
message, and after that continues to read data from the kernel while
the kernel has data.
In this patch we add a context structure for collecting all the parameters
passed from usersapce in one place. The idea is to reduce the number of
parameters being passed to the netlink command handler functions.
It can be argued that not all functions require all the arguments, but this
approach
I didn't realize earlier that version in a netlink message was a
UINT8. So, fixing that here.
Also, some of the commands don't pass a valid DP value. Hence adding
a field to identify such commands.
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribu
Signed-off-by: Nithin Raju
Signed-off-by: Ankur Sharma
Acked-by: Ankur Sharma
---
Please attribute this to Nithin Raju
---
datapath-windows/ovsext/Datapath.h | 53 +++---
1 file changed, 44 insertions(+), 9 deletions(-)
diff --git a/datapath-windows/ovsext/Data
Hi Samuel,
>
> >All NET_BUFFERs of a NET_BUFFER_LIST must go through the pipeline:
> >extract, find flow, execute. Previously, only the first NET_BUFFER of a
> >NET_BUFFER_LIST was going through this pipeline, which was erroneous.
>
> comments here.
>
> It might make the reading clearer, and st
Enable build notifications to bu...@openvswitch.org
Co-authored-by: Ben Pfaff
Signed-off-by: Thomas Graf
---
v2: Add secure token as email recipient
Ben: I can't really test this change without commit access to the main
repo. If you want to test this before committing, push this combined
On Fri, Aug 29, 2014 at 07:48:06PM +0200, Thomas Graf wrote:
> On 08/29/14 at 10:39am, Ben Pfaff wrote:
> > CC: Thomas Graf
> > Signed-off-by: Ben Pfaff
> > ---
> > INSTALL |2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/INSTALL b/INSTALL
> > index 15e93c5..7f6
On Fri, Aug 29, 2014 at 12:21:49PM +0200, Thomas Graf wrote:
> Without the explicit wide type, the shift operation may be performed
> on a int which will result in implementation defined behaviour on a
> system with more than 32 CPUs.
>
> Signed-off-by: Thomas Graf
Thanks, I applied this to mast
On Fri, Aug 29, 2014 at 12:20:21PM +0200, Thomas Graf wrote:
> tc_fill_rate() takes a 64bit int, casting kbits_rate from int
> to uint64_t avoids a possible overflow when translating from
> kbits to bytes.
>
> Signed-off-by: Thomas Graf
Thanks, I applied this to master, branch-2.[3210], and bran
On 08/29/14 at 10:39am, Ben Pfaff wrote:
> CC: Thomas Graf
> Signed-off-by: Ben Pfaff
> ---
> INSTALL |2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/INSTALL b/INSTALL
> index 15e93c5..7f670dd 100644
> --- a/INSTALL
> +++ b/INSTALL
> @@ -568,7 +568,7 @@ is linked to
Thanks for the reviews!
Series apart from the liv/ovs-thread changes pushed to master.
Jarno
On Aug 28, 2014, at 12:22 PM, Ben Pfaff wrote:
> On Fri, Aug 22, 2014 at 01:58:29PM -0700, Jarno Rajahalme wrote:
>> Neither 'miss_config', 'n_missed', nor 'n_matched' is used to
>> synchronize the s
On Fri, Aug 29, 2014 at 01:50:21AM +0200, Thomas Graf wrote:
> Describe the steps required to setup use of travis-ci for any
> GitHub ovs repository.
>
> Signed-off-by: Thomas Graf
Applied, thanks!
___
dev mailing list
dev@openvswitch.org
http://openvs
CC: Thomas Graf
Signed-off-by: Ben Pfaff
---
INSTALL |2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/INSTALL b/INSTALL
index 15e93c5..7f670dd 100644
--- a/INSTALL
+++ b/INSTALL
@@ -568,7 +568,7 @@ is linked to travis-ci.
Instructions to setup travis-ci for your GitHub r
This is the change produced by:
travis encrypt --repo openvswitch/ovs bu...@openvswitch.org --add
I don't know whether it is minimal.
CC: Thomas Graf
Signed-off-by: Ben Pfaff
---
.travis.yml | 10 +-
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/.travis.yml b/.trav
On Fri, Aug 29, 2014 at 9:37 AM, Jesse Gross wrote:
> On Thu, Aug 28, 2014 at 11:30 AM, Pravin B Shelar wrote:
>> OVS tunnel compat code depends on this function pointer to
>> handle GSO packet. Currently we do not initialize for all
>> GRE GSO packets. Following patch fixes that.
>>
>> Signed-of
On Aug 28, 2014, at 11:28 AM, Ben Pfaff wrote:
> On Fri, Aug 22, 2014 at 01:58:21PM -0700, Jarno Rajahalme wrote:
>> The atomics here do not synchronize the state of any other variables,
>> so we can use relaxed atomics.
>>
>> cfm_should_process_flow() is rearranged to set the megaflow mask bit
On Fri, Aug 29, 2014 at 10:12:09AM -0700, Ben Pfaff wrote:
> On Wed, Aug 20, 2014 at 12:14:56PM -0700, Srini Seetharaman wrote:
> > > I'm going to let this one sit for a few days to see if anyone else has
> > > comments.
> >
> > Thanks for the update.
> >
> > For others looking at this patch, her
On Wed, Aug 20, 2014 at 12:14:56PM -0700, Srini Seetharaman wrote:
> > I'm going to let this one sit for a few days to see if anyone else has
> > comments.
>
> Thanks for the update.
>
> For others looking at this patch, here is the relevant discussion on
> the ovs-discuss:
> http://openvswitch.o
On Fri, Aug 29, 2014 at 9:30 AM, Jesse Gross wrote:
> On Fri, Aug 15, 2014 at 11:18 AM, Jesse Gross wrote:
>> The checksum of ICMPv6 packets uses the IP pseudoheader as part of
>> the calculation, unlike ICMP in IPv4. This was not implemented,
>> which means that modifying the IP addresses of an
On Thu, Aug 28, 2014 at 11:30 AM, Pravin B Shelar wrote:
> OVS tunnel compat code depends on this function pointer to
> handle GSO packet. Currently we do not initialize for all
> GRE GSO packets. Following patch fixes that.
>
> Signed-off-by: Pravin B Shelar
Acked-by: Jesse Gross
_
On Aug 28, 2014, at 9:32 AM, Ben Pfaff wrote:
> On Fri, Aug 22, 2014 at 01:58:17PM -0700, Jarno Rajahalme wrote:
>> barrier->count is used as a simple counter and is not expected the
>> synchronize the state of any other variable, so we can use atomic_count,
>> which uses relaxed atomics.
>>
>>
On Fri, Aug 15, 2014 at 11:18 AM, Jesse Gross wrote:
> The checksum of ICMPv6 packets uses the IP pseudoheader as part of
> the calculation, unlike ICMP in IPv4. This was not implemented,
> which means that modifying the IP addresses of an ICMPv6 packet
> would cause the checksum to no longer be c
Hi Sam,
I was wondering if we could go ahead and acknowledge Nithin's change. I know
you have some reservations but this work should be considered as a bring up
which effort to enable other developers to continue with their own tasks. The
current situation is that we have a circular dependency
Thanks Alin, I am planning to post a patch for the event notification so we can
follow the same logic for the packet miss path.
Eitan
From: Alin Serdean [mailto:aserd...@cloudbasesolutions.com]
Sent: Friday, August 29, 2014 7:14 AM
To: Eitan Eliahu; dev@openvswitch.org; Nithin Raju; Saurabh Shah
On Fri, Aug 29, 2014 at 04:06:48AM +, Samuel Ghinet wrote:
>
> This patch includes the file renaming and accommodations needed for the file
> renaming to build the forwarding extension for Hyper-V.
>
> This patch is also a follow-up for the thread:
> http://openvswitch.org/pipermail/dev/2014-
On Fri, Aug 29, 2014 at 04:02:18AM +, Samuel Ghinet wrote:
> I see that the automake.mk file in the repository has tabs instead of
> spaces., while the patch believes the original automake.mk has spaces.
> I'll re-create the patch, to also transform the tabs of automake.mk into
> spaces.
Mak
On 08/26/14 16:54, Thomas Graf wrote:
On 08/26/14 at 01:13pm, Alexei Starovoitov wrote:
I think it's important distinction. In-kernel OVS is not OF.
It's a networking function that has hard-coded packet parser,
N-tuple match and programmable actions.
There were times when HW vendors were using O
Hi Eitan,
I think we could use the same design.
I can respin the patch on Monday if you can wait.
Alin.
De la: Eitan Eliahu [mailto:elia...@vmware.com]
Trimis: Friday, August 29, 2014 6:30 AM
Către: Alin Serdean; dev@openvswitch.org; Nithin Raju; Saurabh Shah
Subiect: RE: [ovs-dev] [PATCHv2] Ch
btw,
in case you wondered about this:
> @@ -863,7 +863,7 @@ OvsPartialCopyToMultipleNBLs(PVOID ovsContext,
> if (prevNbl == NULL) {
> firstNbl = newNbl;
> } else {
> -NET_BUFFER_LIST_NEXT_NBL(prevNbl) = nbl;
> +NET_BUFFER_LIST_NEXT_NBL(prevNb
Hello,
Below is part 3 of the patch - main refactor of the ingress path
(OvsStartNBLIngress / OvsExtSendNBL)
Thank you!
Samuel Ghinet
=
addressing issues:
o) clearer variable names
o) merge OvsExtSendNBL and OvsStartNBLIngress into OvsExtSendNBL
(there is no reason to h
Hello,
Part 2 of the patch is below:
===
---
datapath-windows/ovsext/OvsBufferMgmt.c | 39 +
1 file changed, 20 insertions(+), 19 deletions(-)
diff --git a/datapath-windows/ovsext/OvsBufferMgmt.c
b/datapath-windows/ovsext/OvsBufferMgmt.c
ind
Hello guys,
This is the updated NBL patch, which have taken into account Nithin and
Saurabh's suggestions.
I have decided that the refactor of OvsStartNBLIngress - some points suggested
by Saurabh - should be separate commit(s) / patch(es).
The patches will be tested on monday, but it would be u
Dear OVS devs,
For some experiments I did around April/May for a paper I wrote I
changed some sourcecode to ofproto-dpif-xlate.c to have the FastFailover
Group Table look at the BFD status directly.
We did this because we found that although the BFD status flags changed
in time, interface status r
Nithin,
I had expected you modify the original patch with my suggestions, not add a new
patch on top of it, by which to refactor the original patch.
So this patch is: Not acked-by: Samuel Ghinet
Regards,
Sam
Date: Fri, 29 Aug 2014 01:15:21 -0700
From: Ni
Nithin,
This patch goes: Not acked-by: Samuel Ghinet
Regards,
Sam
Date: Fri, 29 Aug 2014 01:15:20 -0700
From: Nithin Raju
To: dev@openvswitch.org
Subject: [ovs-dev] [PATCH 8/9 v2] datpath-windows: fix the dp index
check inValidateNetlinkC
Nithin,
This patch goes: Not acked-by: Samuel Ghinet
Regards,
Sam
Date: Fri, 29 Aug 2014 01:15:19 -0700
From: Nithin Raju
To: dev@openvswitch.org
Subject: [ovs-dev] [PATCH 7/9 v2] datapath-windows: add clarifications
to markers in dump state
Nithin,
This patch goes: Not acked-by: Samuel Ghinet
Regards,
Sam
Date: Fri, 29 Aug 2014 01:15:18 -0700
From: Nithin Raju
To: dev@openvswitch.org
Subject: [ovs-dev] [PATCH 6/9 v2] datapath-windows: Check for device
operation in OvsGetDpCmdHandler
Nithin,
This patch goes: Not acked-by: Samuel Ghinet
Regards,
Sam
Message: 1
Date: Fri, 29 Aug 2014 01:15:17 -0700
From: Nithin Raju
To: dev@openvswitch.org
Subject: [ovs-dev] [PATCH 5/9 v2] datapath-windows: add description to
members of OVS_US
Nithin,
I had given some suggestions, which you agreed upon, but you did not apply them
to this patch.
Also, note the typo in the comment:
> In this patch, we add support for the GET_DP netlink command to dump
> the datpaaths
So this patch goes: Not acked-by: Samuel Ghinet
Regards,
Sam
__
Nithin,
You have in this patch 3:
> UINT32 devOp;
while in patch 5:
> UINT32 devOp;/* Device operation of the userspace call. */
I had expected you make only a single patch, with the latest version.
Also, I had suggested you specify the #define-s in Datapath.c as comments.
So this patch
Nithin,
You have in this patch 2:
> BOOLEAN validateDp;
while in patch 8:
> BOOLEAN validateDpIndex;
I had expected you make only a single patch, with the latest version.
Also you have ignored some of my suggestions.
So this patch goes: Not acked-by: Samuel Ghinet
Thanks,
Sam
_
1 - 100 of 120 matches
Mail list logo
