[ovs-dev] [PATCH v6 12/12] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

to the proper NUMA socket. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 73 +-- lib/netde

[ovs-dev] [PATCH v6 04/12] dpif-netdev: Do not keep refcount for ports.

Only the main thread will delete ports after pausing every other thread. There's no need to keep count. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/

Re: [ovs-dev] [PATCH v5 10/12] dpif-netdev: Fix reconfigure_pmd_threads().

urning prematurely from the function without >>restarting the pmd threads). >> >>Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> >>Tested-by: Ilya Maximets <i.maxim...@samsung.com> >>Acked-by: Ilya Maximets <i.maxim...@samsung.com> >>

[ovs-dev] [PATCH v6 02/12] dpif-netdev: Proper error handling in do_add_port().

This fixes multiple error path mistakes in do_add_port, none of which has been a problem in practice so far. This change will make it easier for a following commit to return in case of error. Also, this removes an unneeded special case for tunnel ports. Signed-off-by: Daniele Di Proietto

[ovs-dev] [PATCH v6 05/12] dpif-netdev: Remove useless dpif-dummy/delete-port appctl.

It is only used in the testsuite and it can be replaced by a dpctl command. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/

[ovs-dev] [PATCH v6 08/12] dpif-netdev: Change pmd thread configuration in dpif_netdev_run().

Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 140 ++-- lib/dpif-provider.h | 3 +- 2

[ovs-dev] [PATCH v6 03/12] dpif-netdev: Keep count of elements in port->rxq[].

This will ease deleting a port with no open rxqs. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 24 +--- 1 file chan

[ovs-dev] [PATCH v6 01/12] netdev-dpdk: Consistent variable naming.

netdev_rxq':'rxq' 'struct netdev_rxq_dpdk':'rx' Also, 'dev->up.' is replaced by 'netdev->', where 'netdev' was already defined. Suggested-by: Ilya Maximets <i.maxim...@samsung.com> Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsun

[ovs-dev] [PATCH v6 00/12] Reconfigure netdev at runtime

): dev->requested_n_rxq is now accessed while holding the appropriate mutex. * Fixed some outdated comments about rx queue configuration. Daniele Di Proietto (12): netdev-dpdk: Consistent variable naming. dpif-netdev: Proper error handling in do_add_port(). dpif-netdev: Keep count of elements in

Re: [ovs-dev] [PATCH v2] ovs-thread: Do not always end quiescent state in ovs_thread_create().

On 25/03/2016 08:46, "Ben Pfaff" <b...@ovn.org> wrote: >On Thu, Mar 24, 2016 at 11:31:09AM -0700, Daniele Di Proietto wrote: >> A new thread must be started in a non quiescent state. There is a call >> to ovsrcu_quiesce_end() in ovsthread_wrapper(), to enforce

Re: [ovs-dev] [PATCH] ovs-thread: Do not end quiescent state in ovs_thread_create().

On 24/03/2016 09:18, "Ben Pfaff" <b...@ovn.org> wrote: >On Wed, Mar 23, 2016 at 06:02:58PM -0700, Daniele Di Proietto wrote: >> A new thread must be started in a non quiescent state. There is a call >> to ovsrcu_quiesce_end() in ovsthread_wrapper(), to enforce

[ovs-dev] [PATCH v2] ovs-thread: Do not always end quiescent state in ovs_thread_create().

hreaded the main thread is put in an active state Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- v2: The first call to ovs_thread_create must put the main thread in an active state, so: * Document this in ovs-rcu.h. * Keep the call to ovsrcu_quiesce_end() in ovs_thread_c

Re: [ovs-dev] [PATCH] ovs-thread: Do not end quiescent state in ovs_thread_create().

Oops, I forgot my signoff. Here it is: Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> On 23/03/2016 18:02, "Daniele Di Proietto" <diproiet...@vmware.com> wrote: >A new thread must be started in a non quiescent state. There is a call >to ovsrcu_quiesc

[ovs-dev] [PATCH] ovs-thread: Do not end quiescent state in ovs_thread_create().

A new thread must be started in a non quiescent state. There is a call to ovsrcu_quiesce_end() in ovsthread_wrapper(), to enforce this. ovs_thread_create(), instead, is executed in the parent thread, and calling ovsrcu_quiesce_end() there will cause the parent to end its (possible) quiescing

Re: [ovs-dev] [PATCH v4 00/12] Reconfigure netdev at runtime

.@samsung.com> wrote: >There are few style problems in patches. I've sent corresponding mails. >Other than this: > Tested-by: Ilya Maximets <i.maxim...@samsung.com> > Acked-by: Ilya Maximets <i.maxim...@samsung.com> > >On 17.03.2016 02:39, Daniele Di Pr

[ovs-dev] [PATCH v5 12/12] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

to the proper NUMA socket. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 71 +- lib/netde

[ovs-dev] [PATCH v5 10/12] dpif-netdev: Fix reconfigure_pmd_threads().

of returning prematurely from the function without restarting the pmd threads). Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/

[ovs-dev] [PATCH v5 09/12] dpif-netdev: Document locking discipline for non_pmd_mutex.

This just documents what the current code already assumes. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 1 + 1 file changed, 1 insertion(+)

[ovs-dev] [PATCH v5 11/12] netdev: Add reconfigure request mechanism.

rx and tx queues in netdev-dpdk. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/netdev-bsd.c | 1 + lib/netdev-dpdk.c | 1 + lib/netdev-dummy.c

[ovs-dev] [PATCH v5 08/12] dpif-netdev: Change pmd thread configuration in dpif_netdev_run().

Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 140 ++-- lib/dpif-provider.h | 3 +- 2

[ovs-dev] [PATCH v5 01/12] netdev-dpdk: Consistent variable naming.

netdev_rxq':'rxq' 'struct netdev_rxq_dpdk':'rx' Also, 'dev->up.' is replaced by 'netdev->', where 'netdev' was already defined. Suggested-by: Ilya Maximets <i.maxim...@samsung.com> Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsun

[ovs-dev] [PATCH v5 03/12] dpif-netdev: Keep count of elements in port->rxq[].

This will ease deleting a port with no open rxqs. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/dpif-netdev.c | 24 +--- 1 file chan

[ovs-dev] [PATCH v5 06/12] dpif-netdev: Wait an RCU grace period before freeing ports.

RCU is more in line with other cmap users. * We might want to allow port removal and queue reconfiguration without stopping completely all the pmd threads. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c

[ovs-dev] [PATCH v5 02/12] dpif-netdev: Proper error handling in do_add_port().

This fixes multiple error path mistakes in do_add_port, none of which has been a problem in practice so far. This change will make it easier for a following commit to return in case of error. Also, this removes an unneeded special case for tunnel ports. Signed-off-by: Daniele Di Proietto

[ovs-dev] [PATCH v5 04/12] dpif-netdev: Do not keep refcount for ports.

Only the main thread will delete ports after pausing every other thread. There's no need to keep count. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/

[ovs-dev] [PATCH v5 05/12] dpif-netdev: Remove useless dpif-dummy/delete-port appctl.

It is only used in the testsuite and it can be replaced by a dpctl command. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maximets <i.maxim...@samsung.com> --- lib/

[ovs-dev] [PATCH v5 07/12] ofproto-dpif: Call dpif_poll_threads_set() before dpif_run()

An upcoming commit will make dpif_poll_threads_set() record the requested configuration and dpif_run() apply it, so it makes sense to change the order. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> Tested-by: Ilya Maximets <i.maxim...@samsung.com> Acked-by: Ilya Maxim

[ovs-dev] [PATCH v5 00/12] Reconfigure netdev at runtime

outdated comments about rx queue configuration. Daniele Di Proietto (12): netdev-dpdk: Consistent variable naming. dpif-netdev: Proper error handling in do_add_port(). dpif-netdev: Keep count of elements in port->rxq[]. dpif-netdev: Do not keep refcount for ports. dpif-netdev: Remove useles

Re: [ovs-dev] [PATCH 07/10] netdev-dummy: Queue packets only to one rx queue.

On 22/03/2016 18:38, "Ben Pfaff" <b...@ovn.org> wrote: >On Wed, Mar 23, 2016 at 01:13:06AM +, Daniele Di Proietto wrote: >> >> On 22/03/2016 17:57, "Ben Pfaff" <b...@ovn.org> wrote: >> >> >On Mon, Mar 14, 2016 at 06:18:17PM

Re: [ovs-dev] [PATCH] netdev-dpdk: Fix crash when changing the vhost-user port.

March 22, 2016 12:42 PM >> To: dev@openvswitch.org; Daniele Di Proietto <diproiet...@vmware.com> >> Cc: Dyasly Sergey <s.dya...@samsung.com>; Ben Pfaff <b...@ovn.org>; >>Flavio >> Leitner <f...@sysclose.org>; Traynor, Kevin <kevin.tray...@intel.com>

Re: [ovs-dev] [PATCH 07/10] netdev-dummy: Queue packets only to one rx queue.

On 22/03/2016 17:57, "Ben Pfaff" wrote: >On Mon, Mar 14, 2016 at 06:18:17PM +0300, Ilya Maximets wrote: >> This is unused functionality and it will harm to >> multiqueue implementation. >> >> Signed-off-by: Ilya Maximets > >This change doesn't make any

[ovs-dev] [PATCH v4 00/12] Reconfigure netdev at runtime

e been merged. * Fixed netdev_dpdk_set_config(): dev->requested_n_rxq is now accessed while holding the appropriate mutex. * Fixed some outdated comments about rx queue configuration. Daniele Di Proietto (12): netdev-dpdk: Consistent variable naming. dpif-netdev: Proper error handling in do

Re: [ovs-dev] [PATCH v3 02/11] dpif-netdev: Keep count of elements in port->rxq[].

t it, I'd prefer we tackle this problem by simplifying/clarifying the behavior of the rxq/txq variables in netdev-dpdk Thanks > >Best regards, Ilya Maximets. > > >On 16.03.2016 01:29, Daniele Di Proietto wrote: >> This will ease deleting a port with no open rxqs. >>

[ovs-dev] [PATCH v4 11/12] netdev: Add reconfigure request mechanism.

rx and tx queues in netdev-dpdk. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/netdev-bsd.c | 1 + lib/netdev-dpdk.c | 1 + lib/netdev-dummy.c| 1 + lib/netdev-linux.c| 1 + lib/netdev-provider.h | 27 ++- lib/netdev-v

[ovs-dev] [PATCH v4 07/12] ofproto-dpif: Call dpif_poll_threads_set() before dpif_run()

An upcoming commit will make dpif_poll_threads_set() record the requested configuration and dpif_run() apply it, so it makes sense to change the order. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- ofproto/ofproto-dpif.c | 4 ++-- 1 file changed, 2 insertions(+), 2 del

Re: [ovs-dev] [PATCH 00/10] PMD Testsuite.

Hi Ilya, thanks for the series, I think it's going to ease development of dpif-netdev a lot. I have been thinking about doing the same for a while and I've also sketched out some code, which is very similar to this :-) (I've merged the first patch). The interesting thing about this tests for me

[ovs-dev] [PATCH v4 12/12] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

calling netdev_set_multiq(), can detect this and call reconfigure(). This mechanism can also be used to: * Automatically match the number of rxq with the one provided by qemu via the new_device callback. * Provide a way to change the MTU of dpdk devices at runtime. Signed-off-by: Daniele Di

[ovs-dev] [PATCH v4 04/12] dpif-netdev: Do not keep refcount for ports.

Only the main thread will delete ports after pausing every other thread. There's no need to keep count. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 50 ++ 1 file changed, 14 insertions(+), 36 del

[ovs-dev] [PATCH v4 01/12] netdev-dpdk: Consistent variable naming.

netdev_rxq':'rxq' 'struct netdev_rxq_dpdk':'rx' Also, 'dev->up.' is replaced by 'netdev->', where 'netdev' was already defined. Suggested-by: Ilya Maximets <i.maxim...@samsung.com> Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/n

[ovs-dev] [PATCH v4 05/12] dpif-netdev: Remove useless dpif-dummy/delete-port appctl.

It is only used in the testsuite and it can be replaced by a dpctl command. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 33 - tests/bridge.at | 4 ++-- 2 files changed, 2 insertions(+), 35 deletions(-) diff --git

[ovs-dev] [PATCH v4 08/12] dpif-netdev: Change pmd thread configuration in dpif_netdev_run().

Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 140 ++-- lib/dpif-provider.h | 3 +- 2 files changed, 83 insertions(+), 60 deletions(-) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 3

[ovs-dev] [PATCH v4 03/12] dpif-netdev: Keep count of elements in port->rxq[].

This will ease deleting a port with no open rxqs. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 24 +--- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 9c30dad..a2281b8

Re: [ovs-dev] [PATCH v3 11/11] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

Thanks for your detailed comments, replies inline. I'll send v4 in a minute. On 16/03/2016 06:12, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >On 16.03.2016 01:30, Daniele Di Proietto wrote: >> This introduces in dpif-netdev and netdev-dpdk the first use f

Re: [ovs-dev] [PATCH v3 10/11] netdev: Add reconfigure request mechanism.

On 16/03/2016 06:12, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >On 16.03.2016 01:30, Daniele Di Proietto wrote: >> A netdev provider, especially a PMD provider (like netdev DPDK) might >> not be able to change some of its parameters (such as MTU, or number

Re: [ovs-dev] [PATCH 00/10] PMD Testsuite.

even by the non pmd dummy class (even though it's probably not that useful) What do you think? Thanks, Daniele On 17/03/2016 06:58, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >Hi, Daniele. > >On 17.03.2016 03:53, Daniele Di Proietto wrote: >> Hi Ilya,

Re: [ovs-dev] [PATCH 01/10] netdev: New field 'is_pmd' in netdev_class.

Thanks for the patch, I had almost the exact same change in one of my local branches! Applied to master On 14/03/2016 08:18, "Ilya Maximets" wrote: >Made to simplify creation of derived classes. > >Signed-off-by: Ilya Maximets >--- >

Re: [ovs-dev] [PATCH v3 10/11] netdev: Add reconfigure request mechanism.

On 16/03/2016 06:12, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >On 16.03.2016 01:30, Daniele Di Proietto wrote: >> A netdev provider, especially a PMD provider (like netdev DPDK) might >> not be able to change some of its parameters (such as MTU, or number

[ovs-dev] [PATCH v4 02/12] dpif-netdev: Proper error handling in do_add_port().

This fixes multiple error path mistakes in do_add_port, none of which has been a problem in practice so far. This change will make it easier for a following commit to return in case of error. Also, this removes an unneeded special case for tunnel ports. Signed-off-by: Daniele Di Proietto

[ovs-dev] [PATCH v4 09/12] dpif-netdev: Document locking discipline for non_pmd_mutex.

This just documents what the current code already assumes. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 0268b9b..597e1d7 100644 --- a/lib/dpif-netdev.c +++

[ovs-dev] [PATCH v4 06/12] dpif-netdev: Wait an RCU grace period before freeing ports.

RCU is more in line with other cmap users. * We might want to allow port removal and queue reconfiguration without stopping completely all the pmd threads. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c

[ovs-dev] [PATCH v4 10/12] dpif-netdev: Fix reconfigure_pmd_threads().

of returning prematurely from the function without restarting the pmd threads). Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 78 ++- 1 file changed, 54 insertions(+), 24 deletions(-) diff --git a/li

[ovs-dev] [PATCH v3 09/11] dpif-netdev: Fix reconfigure_pmd_threads().

of returning prematurely from the function without restarting the pmd threads). Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 78 ++- 1 file changed, 54 insertions(+), 24 deletions(-) diff --git a/li

[ovs-dev] [PATCH v3 03/11] dpif-netdev: Do not keep refcount for ports.

Only the main thread will delete ports after pausing every other thread. There's no need to keep count. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 50 ++ 1 file changed, 14 insertions(+), 36 del

[ovs-dev] [PATCH v3 04/11] dpif-netdev: Remove useless dpif-dummy/delete-port appctl.

It is only used in the testsuite and it can be replaced by a dpctl command. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 33 - tests/bridge.at | 4 ++-- 2 files changed, 2 insertions(+), 35 deletions(-) diff --git

[ovs-dev] [PATCH v3 11/11] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

calling netdev_set_multiq(), can detect this and call reconfigure(). This mechanism can also be used to: * Automatically match the number of rxq with the one provided by qemu via the new_device callback. * Provide a way to change the MTU of dpdk devices at runtime. Signed-off-by: Daniele Di

[ovs-dev] [PATCH v3 02/11] dpif-netdev: Keep count of elements in port->rxq[].

This will ease deleting a port with no open rxqs. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 24 +--- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 9c30dad..a2281b8

[ovs-dev] [PATCH v3 07/11] dpif-netdev: Change pmd thread configuration in dpif_netdev_run().

Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 140 ++-- lib/dpif-provider.h | 3 +- 2 files changed, 83 insertions(+), 60 deletions(-) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 3

[ovs-dev] [PATCH v3 08/11] dpif-netdev: Document locking discipline for non_pmd_mutex.

This just documents what the current code already assumes. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c index 0268b9b..597e1d7 100644 --- a/lib/dpif-netdev.c +++

[ovs-dev] [PATCH v3 10/11] netdev: Add reconfigure request mechanism.

rx and tx queues in netdev-dpdk. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/netdev-bsd.c | 1 + lib/netdev-dpdk.c | 1 + lib/netdev-dummy.c| 1 + lib/netdev-linux.c| 1 + lib/netdev-provider.h | 27 ++- lib/netdev-v

[ovs-dev] [PATCH v3 01/11] dpif-netdev: Proper error handling in do_add_port().

This fixes multiple error path mistakes in do_add_port, none of which has been a problem in practice so far. This change will make it easier for a following commit to return in case of error. Also, this removes an unneeded special case for tunnel ports. Signed-off-by: Daniele Di Proietto

[ovs-dev] [PATCH v3 06/11] ofproto-dpif: Call dpif_poll_threads_set() before dpif_run()

An upcoming commit will make dpif_poll_threads_set() record the requested configuration and dpif_run() apply it, so it makes sense to change the order. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- ofproto/ofproto-dpif.c | 4 ++-- 1 file changed, 2 insertions(+), 2 del

[ovs-dev] [PATCH v3 05/11] dpif-netdev: Wait an RCU grace period before freeing ports.

RCU is more in line with other cmap users. * We might want to allow port removal and queue reconfiguration without stopping completely all the pmd threads. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c

[ovs-dev] [PATCH v3 00/11] Reconfigure netdev at runtime

ate mutex. * Fixed some outdated comments about rx queue configuration. Daniele Di Proietto (11): dpif-netdev: Proper error handling in do_add_port(). dpif-netdev: Keep count of elements in port->rxq[]. dpif-netdev: Do not keep refcount for ports. dpif-netdev: Remove useless dpif-dummy/delete

Re: [ovs-dev] [PATCH v2 11/11] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

Hi Ilya, thanks for your comments, replies inline I'll post a v3 soon On 15/03/2016 06:43, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >Few comments inline. > >Best regards, Ilya Maximets. > >On 03.03.2016 04:33, Daniele Di Proietto wrote: >> This intr

Re: [ovs-dev] [PATCH v2 11/11] netdev-dpdk: Use ->reconfigure() call to change rx/tx queues.

the one provided by qemu >> via the new_device callback. >>* Provide a way to change the MTU of dpdk devices at runtime. >> >>Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> >>--- >> lib/dpif-netdev.c | 70 +-- >> lib/ne

Re: [ovs-dev] [PATCH v2 09/11] dpif-netdev: Fix reconfigure_pmd_threads().

Hi Ilya, On 15/03/2016 06:43, "Ilya Maximets" <i.maxim...@samsung.com> wrote: >comment inline. > >Best regards, Ilya Maximets. > >On 03.03.2016 04:33, Daniele Di Proietto wrote: >> This commit change reconfigure_pmd_threads() to interact with the ports >

Re: [ovs-dev] [PATCH] INSTALL.DPDK: remove issue with QEMU v2.4.0 and dpdkvhostuser

On 09/03/2016 06:03, "Mauricio Vasquez B" wrote: >The patch mentioned in the commit e73b7508fb58 ("INSTALL.DPDK: Mention >issue with QEMU v2.4.0 & dpdkvhostuser") is present in DPDK v2.2.0, >then this issue is not valid anymore. > >Signed-off-by:

Re: [ovs-dev] [PATCH] netdev-dpdk: fix mbuf leaks

Thanks for clarifying, I've back ported this up to branch-2.4 On 09/03/2016 17:46, "Yuanhan Liu" <yuanhan@linux.intel.com> wrote: >On Thu, Mar 10, 2016 at 01:33:03AM +, Daniele Di Proietto wrote: >> Thanks for the patch! >> >> Were you able to hit t

Re: [ovs-dev] [Fix branch-2.4 travis builds] travis: Use container infrastructure.

Thanks for fixing this On 09/03/2016 19:55, "dev on behalf of Joe Stringer" <dev-boun...@openvswitch.org on behalf of j...@ovn.org> wrote: >Hi Andy, thanks for fixing this. > >On 9 March 2016 at 18:05, Andy Zhou <az...@ovn.org> wrote: >> From: Daniel

Re: [ovs-dev] [PATCH RFC 1/1] netdev-dpdk: NUMA Aware vHost

t;Ben Pfaff" <b...@ovn.org> wrote: >How does this numa library relate to lib/ovs-numa.[ch]? > >On Thu, Mar 10, 2016 at 01:22:42AM +, Daniele Di Proietto wrote: >> Thanks for the patch, I'll put this in the use case list for >> my series if I need to resend it! &g

Re: [ovs-dev] [PATCH] netdev-dpdk: fix mbuf leaks

Thanks for the patch! Were you able to hit this bug in your setup or did you just find this by code inspection? I'm asking because I'm wondering whether we should backport the fix. In any case I've applied this to master and added your name to the AUTHORS file, thanks! On 07/03/2016 17:50,

Re: [ovs-dev] [PATCH RFC 1/1] netdev-dpdk: NUMA Aware vHost

Thanks for the patch, I'll put this in the use case list for my series if I need to resend it! It would be nice to get the numa socket information without linking OVS with libnuma, maybe using some DPDK api. From a quick look I didn't find any way, but maybe you know a better way. Some

Re: [ovs-dev] [PATCH] netdev-dpdk: Fix memory leak in netdev_dpdk_vhost_destruct().

Thanks, I applied this to master and branch-2.5 On 03/03/2016 04:50, "Flavio Leitner" wrote: >On Thu, 03 Mar 2016 11:30:06 +0300 >Ilya Maximets wrote: > >> Fixes: 4573fbd38fa1 ("netdev-dpdk: Add vhost-user multiqueue support") >> Signed-off-by: Ilya

[ovs-dev] [PATCH v2 04/11] dpif-netdev: Remove useless dpif-dummy/delete-port appctl.

It is only used in the testsuite and it can be replaced by a dpctl command. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- lib/dpif-netdev.c | 33 - tests/bridge.at | 4 ++-- 2 files changed, 2 insertions(+), 35 deletions(-) diff --git

Re: [ovs-dev] [PATCH 07/11] dpif-netdev: Change pmd thread configuration in dpif_netdev_run().

On 01/03/2016 09:11, "Kavanagh, Mark B" wrote: >Hi Daniele, > >One minor comment below. > >Thanks, >Mark > >>--- >> lib/dpif-netdev.c | 137 >>++-- >> lib/dpif-provider.h | 3 +- >> 2 files changed, 82 insertions(+),

Re: [ovs-dev] [PATCH v4] netdev_dpdk.c: Add QoS functionality.

Hi Ian, I have one comment and some minor style nitpicks inline, but other than that I think the patch is good to go. If you agree with the comments would you mind fixing those and rebasing? Thanks On 19/02/2016 09:26, "Ian Stokes" wrote: >This patch provides the

Re: [ovs-dev] [PATCH 00/11] Reconfigure netdev at runtime

On 29/02/2016 11:07, "Ben Pfaff" <b...@ovn.org> wrote: >On Sun, Feb 28, 2016 at 12:13:12PM -0800, Daniele Di Proietto wrote: >> Currently we treat set_multiq() calls specially in netdev and >>dpif-netdev: >> every pmd thread must be stopped a

[ovs-dev] [PATCH 00/11] Reconfigure netdev at runtime

, but the idea is to use it also at least for: * Changing the MTU at runtime * Automatically detecting the number of rx queues for a vhost-user device The first commits also clean up some code in dpif-netdev and make the ports cmap fully RCU protected. Daniele Di Proietto (11): dpif-netdev

Re: [ovs-dev] [PATCH v3] INSTALL.DPDK: Add notes regarding vhost multiq configuration.

I applied this to master and branch-2.5, thanks On 24/02/2016 09:30, "dev on behalf of Ian Stokes" wrote: >Linux kernel network devices in a guest should have the number of >multi-purpose channels configured when used with DPDK

Re: [ovs-dev] [PATCH RFC v4] netdev-dpdk: vhost-user: Fix sending packets to queues not enabled by guest.

Thanks Ilya and Flavio, I fixed a clang warning and applied this to master and branch-2.5 On 24/02/2016 08:17, "Flavio Leitner" wrote: >On Wed, 24 Feb 2016 17:14:43 +0300 >Ilya Maximets wrote: > >> Currently virtio driver in guest operating system

Re: [ovs-dev] [PATCH] netdev-dpdk: Do not add vhost-user ports with '/' or '\' in name.

Thanks for the review Flavio, I applied this to master and brach-2.5 On 22/02/2016 20:29, "Flavio Leitner" <f...@sysclose.org> wrote: >On Mon, 22 Feb 2016 11:43:23 -0800 >Daniele Di Proietto <diproiet...@vmware.com> wrote: > >> This check prevents an obvious

Re: [ovs-dev] [PATCH 2/2] tests/dpdk/ring_client: extend range of supported dpdkr ports

Thanks! I applied the following incremental (minor style fixes) and push this to master diff --git a/tests/dpdk/ring_client.c b/tests/dpdk/ring_client.c index 089657a..8a43c64 100644 --- a/tests/dpdk/ring_client.c +++ b/tests/dpdk/ring_client.c @@ -99,10 +99,11 @@ usage(const char *progname)

Re: [ovs-dev] [PATCH 1/2] lib/netdev-dpdk: make device name parsing more robust

Thanks! I applied the following incremental (minor style fixes) and push this to master diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c index d0c3d66..f94e0d4 100644 --- a/lib/netdev-dpdk.c +++ b/lib/netdev-dpdk.c @@ -649,10 +649,11 @@ dpdk_dev_parse_name(const char dev_name[], const char

Re: [ovs-dev] [PATCH v2 2/2] tests/dpdk/ring_client: extend range of supported dpdkr ports

Apologies for the delay in the review. I think this and the previous patch could benefit from using str_to_uint(). Would you mind making that change and resubmitting? I'll be happy to apply it then. Thanks, Daniele On 25/01/2016 11:55, "dev on behalf of Mauricio Vasquez B"

Re: [ovs-dev] [PATCH 3/3] lib/netdev-dpdk: increase ring name length for dpdkr ports

Apologies for the delay and thanks for the patch and the review! I applied this to master On 25/01/2016 07:43, "dev on behalf of Aaron Conole" wrote: >Mauricio Vasquez B writes: >> A ring

Re: [ovs-dev] [PATCH 3/3] dpif-netdev: Destroy pmd threads only if pmd-cpu-mask changed.

Thanks for the patch I have a few comments inline On 08/02/2016 07:30, "Ilya Maximets" wrote: >Since 5f2ccb1c0d3b ("dpif: Allow adding ukeys for same flow by >different pmds.") there is the possibility to reassign queues among >pmd threads without restarting them and

Re: [ovs-dev] [PATCH 2/3] dpif-netdev: Move rxq management into functions.

Thanks for the patch and the review! I applied this to master On 18/02/2016 12:20, "Flavio Leitner" wrote: >On Mon, 08 Feb 2016 18:30:30 +0300 >Ilya Maximets wrote: > >> Signed-off-by: Ilya Maximets >> --- >>

Re: [ovs-dev] [PATCH 1/3] dpif-netdev: Reload each thread only once in do_add_port.

Thanks for the patch and the review! I applied this to master On 18/02/2016 12:19, "Flavio Leitner" wrote: >On Mon, 08 Feb 2016 18:30:29 +0300 >Ilya Maximets wrote: > >> While adding of pmd interface with multiple queues several queues >> may be

Re: [ovs-dev] [PATCH v2] dpif-netdev: Add dpif-netdev/pmd-rxq-show appctl command.

Thanks Ilya and Flavio, I applied this to master On 11/02/2016 21:08, "Flavio Leitner" wrote: >On Mon, 08 Feb 2016 10:38:47 +0300 >Ilya Maximets wrote: > >> This command can be used to check the port/rxq assignment to >> pmd threads. For each pmd

Re: [ovs-dev] [PATCH 2/3] netdev-dpdk: Do not add vhost-user ports with '/' in name.

On 22/02/2016 08:57, "Ben Pfaff" <b...@ovn.org> wrote: >On Fri, Feb 05, 2016 at 11:40:11AM -0800, Ben Pfaff wrote: >> On Thu, Feb 04, 2016 at 03:42:34AM +, Daniele Di Proietto wrote: >> > >> > >> > On 03/02/2016 14:47, "Ben Pfaf

[ovs-dev] [PATCH] netdev-dpdk: Do not add vhost-user ports with '/' or '\' in name.

for a bridge name, since that name is used as part of a socket as well. Signed-off-by: Daniele Di Proietto <diproiet...@vmware.com> --- v2: * Do not check for '..', as this doesn't really create a problem * Document restriction in INSTALL.DPDK.md * Also check for backward slash * Drop next patch that u

Re: [ovs-dev] [PATCH v2] INSTALL.DPDK: Update details of XL710 restrictions for DPDK 2.2.

Thanks for the patch, applied on master and branch-2.5 On 11/02/2016 17:36, "dev on behalf of Flavio Leitner" wrote: >On Tue, 9 Feb 2016 14:48:47 + >Ian Stokes wrote: > >> DPDK 2.2 removes restrictions

Re: [ovs-dev] [PATCH] INSTALL.DPDK.md: Correct mergeable buffers parameter.

Thanks for the patch, I applied this on master and branch-2.5 On 16/02/2016 04:21, "dev on behalf of Flavio Leitner" wrote: >On Tue, 16 Feb 2016 12:13:12 + >"Stokes, Ian" wrote: > >> > -Original

Re: [ovs-dev] [PATCH RFC v2] netdev-dpdk: vhost-user: Fix sending packets to queues not enabled by guest.

Would it be safe to assume that the enabled queues are sequential? In this case we could just play with 'real_n_txq' instead of keeping a mapping and the patch would be simpler. I'm not sure if that's a reasonable assumption though. Also, on my system, with qemu-2.5.0, vring_state_changed() is

Re: [ovs-dev] [PATCH v8 0/5] Convert DPDK configuration from command line to DB based

On 09/02/2016 09:48, "Traynor, Kevin" wrote: > > >> -Original Message- >> From: Aaron Conole [mailto:acon...@redhat.com] >> Sent: Friday, January 29, 2016 5:57 PM >> To: dev@openvswitch.org >> Cc: Flavio Leitner ; Panu Matilainen

Re: [ovs-dev] [PATCH V3] netdev-dpdk: Add Jumbo Frame Support.

On 08/02/2016 03:23, "Kavanagh, Mark B" wrote: >> >>Hi Mark, >> > >Hi Daniele, > >Thanks for the review! Responses inline below. > >Cheers, >Mark > >>This patch besides adding Jumbo Frame support also cleans up >>the mbuf initialization (by changing the macros,

Re: [ovs-dev] [PATCH v3] netdev_dpdk.c: Add QoS functionality.

Thanks for the patch and the review. I agree with everything Flavio pointed out, a few more comments below On 10/02/2016 11:54, "Flavio Leitner" wrote: >On Mon, 1 Feb 2016 20:47:25 + >Ian Stokes wrote: > >> This patch provides the modifications

Re: [ovs-dev] [PATCH v8 0/5] Convert DPDK configuration from command line to DB based

On 05/02/2016 04:15, "Traynor, Kevin" <kevin.tray...@intel.com> wrote: > >> -Original Message- >> From: dev [mailto:dev-boun...@openvswitch.org] On Behalf Of Daniele Di >> Proietto >> Sent: Friday, February 5, 2016 2:22 AM >> To: Fl

Re: [ovs-dev] [PATCH v3] netdev_dpdk.c: Add QoS functionality.

On 05/02/2016 13:18, "Ben Pfaff" wrote: >On Thu, Feb 04, 2016 at 09:30:59AM +, Stokes, Ian wrote: >> Just wondering is there any feedback for the V3 patch below for QoS >>functionality? >> >> This patch has been in and out of review since October, I think I've >>reworked it

Re: [ovs-dev] [PATCH] netdev-dpdk: Fix dpdk_watchdog failure to quiesce.

Patrik, thanks for debugging this issue, I've added you to AUTHORS. Thanks for the patch Kevin, I updated a comment (below) and pushed this up to branch-2.4. On 05/02/2016 09:07, "Kevin Traynor" wrote: >Fix issue whereby vhost_thread is waiting for dpdk_watchdog

<    3   4   5   6   7   8   9   10   11   12   >